risk assessment concluded that the absence of CP on water storage tanks, together with the ...... liquid phase on the external surface, consumes the positive current associated with the pit tip ...... corrosion will require removal of the defective area and careful repair welding. ...... concentration that drives the anodic dissolution.
INDUSTRIAL CORROSION AND CORROSION CONTROL TECHNOLOGY
H.M. Shalaby A. Al-Hashem M. Lowther J. Al-Besharah (Editors)
Published By Kuwait Institute for Scientific Research
INDUSTRIAL CORROSION AND CORROSION CONTROL TECHNOLOGY 1996
Sponsored by the Kuwait Institute for Scientific Research (KISR), the Kuwait Foundation for the Advancement of Science (KFAS), the Kuwait National Petroleum Company (KNPC), the Kuwait Oil Company (KOC), Ministry of Electricity and Water (MEW), Kuwait University (KU), Ministry of Oil (MO), the Gulf Cooperation Council-General Secretariat (GCC), Kuwait Chemical Society (KCS), Organization of Arab Petroleum Exporting Countries (OAPEC), and Petrochemical Industries Company (PIC).
INDUSTRIAL CORROSION AND CORROSION CONTROL TECHNOLOY Proceedings of the 2nd Arabian Corrosion Conference Kuwait, October 12-15, 1996
Editors H.M. Shalaby, A. Al-Hashem, M. Lowther and J. Al-Besharah
PUBLISHED BY KUWAIT INSTITUTE FOR SCIENTIFIC RESEARCH P.O. BOX 24885, 13109 SAFAT, KUWAIT
Published by Kuwait Institute for Scientific Research P.O. Box 24885, 13109 Safat, Kuwait Publication Number: KISR 4890
Copyright ® 1996 by Kuwait Institute for Scientific Research The papers were reviewed for their technical contents. Editing was restricted to matters of format, general organization and retyping. The editors assume no responsibility for the accuracy, completeness or usefulness of the information disclosed in this book. Unauthorized use might infringe on privately owned patents of publication right. Please contact the individual authors for permission to reprint or otherwise use information from their papers This book was printed in Kuwait The 2nd Arabian Corrosion Conference
FOREWORD The 2nd Arabian Corrosion Conference was held in the state of Kuwait during the period 1215 October, 1996 under the auspices of H.H. Sheikh Saad Al-Abdullah Al-Salem Al-Sabah, Kuwait’s Crown Prince and Prime Minister. The present conference was scheduled to be held in Kuwait during 27-30 April, 1991, however, it was postponed due to the events that encompassed Kuwait and the Gulf region in 1990-1991. The 1st Arabian Corrosion Conference was held in Kuwait during 4-8 February, 1984. It was attended by over 300 scientists and engineers, representing 26 countries. The conference proceedings were published in two volumes by Pergamon Press under the title “Corrosion: Industrial Problems, Treatment and Control Techniques”. The conference provided a forum for the exchange of ideas between scientists and engineers from the region with their counterparts from the industrialized countries. The patronage of the present conference, the organizing bodies, and the emphasis on industrial corrosion and corrosion prevention reflect the keen interest of the countries in the region in actively combating corrosion problems. This also reflect the recognition of the economic impact resulting from the corrosion of materials. Kuwait and the other Arab countries rely heavily on the utilization of metallic materials in their oil-based industries. Seawater derived from the Arabian Gulf is used in water desalination and as an industrial cooling media. The salinity of the Arabian Gulf seawater is very high when compared to other seawater bodies. The Arabian Gulf countries are located in an arid environmental zone where the temperature during the summer months could reach 50oC and the humidity during the autumn season could become 80% in some of the Gulf states. All these factors contribute to the enhancement of the rate of corrosion of metals and/or cause unpredictable service failures. The program of the present conference includes a field visit to one of Kuwait’s modern refineries and a trip to one of Kuwait’s oil fields. The success of the conference is perhaps difficult to assess. However, the quality of the papers in this volume provides some indication.
The Editors
v
PREFACE The technical program of the present conference includes five plenary lectures and fifty three scientific presentations from about twenty two countries. A number of honorary speakers, carefully selected from high ranking officials and policy makers, were also invited to address the conference. The honorary speakers are expected to provide an overview of the magnitude of corrosion related problems in the Middle East as well as the avenues of linkage between corrosion science and industrial applications. The conference papers were carefully selected to include a blend of fundamental and applied research, and industrial experience. Such a blend was thought to be essential for providing the participants from both industry and academia with a chance to become familiar with the challenges facing each group and the preventive actions to meet them. The papers were refereed in terms of scientific and technical content and format in accordance with internationally accepted standards. The papers in the proceedings are grouped in the following sections for quick reference:
• • • • • • • • •
Plenary Lectures Oil Field Corrosion Corrosion in Refinery and Petrochemical Industries Seawater Corrosion Corrosion in the Building Industry Fundamental Aspects Corrosion Protection and Monitoring Corrosion Management Novel Techniques
The plenary papers are mostly reviews covering important topic related to the objectives of the conference. The remaining papers cover various topics of major importance to corrosion in general and particularly to the oil-based and desalination industries. A good number of papers delt with corrosion protection and new techniques for corrosion monitoring. The task of editing this volume was facilitated by the efforts of the International Advisory Committee and the Scientific Committee for the conference who reviewed all the papers. The editorial board gratefully acknowledge these efforts; the cooperation, time and effort of all authors ; and the management of the Kuwait Institute for Scientific Research for allocating the required resources to prepare the manuscript of this volume.
The Editors
vi
TABLE OF CONTENTS Foreword............................................................................................................................................v Preface...............................................................................................................................................vi Organizing Committees......................................................................................................................xi Acknowledgement.............................................................................................................................xii PLENARY LECTURES Corrosion Management V. Ashworth........................................................................................................................................1 The Deterministic Prediction of Damage D.D. Macdonald...............................................................................................................................17 Relevance of Laboratory Corrosion Tests in Corrosivity Assessment and Materials Selection: Case Studies R.D. Kane.........................................................................................................................................37 Corrosion of Condensers in Multi Stage Flash Evaporation Distillers A.M. Shams El Din...........................................................................................................................49 Correct Materials Selection for Desalination -The Key To Plant Reliability J.W. Oldfield....................................................................................................................................67 OIL FIELD CORROSION Corrosivity Prediction for Co2/H2s Production Environments S. Srinivasan and R.D. Kane.............................................................................................................89 Testing of Drilling Fluids Formulated From Tabuk Formation Clays M.N.J. Al-Awad, A.S. Dahab and M.E. El-Dahshan........................................................................111 Preventing Sulfate Scale Deposition in Oil Production Facilities C.J. Hinrichsen, M.J. McKinzie, S. He, J. Oddo, A.J. Gerbino, A.T. Kan, and M.B. Tomson...........127 Concerns Over the Selection of Biocides for Oil Fields and Power Plants: A Laboratory Corrosion Assessment J. Alhajji and M. Valliappan...........................................................................................................135 Evaluation of Microbially Influenced Corrosion Risks and Control Strategies in Seawater and Produced Water Injection Systems, Kuwait P.F. Sanders, M. Salman and K. Al-Muhanna.................................................................................149 Hydrogen Degradation of Steel - Diffusion and Deterioration M. Farzam...................................................................................................................................................165 Control Strategies for Thermophilic Sulphate-Reducing Bacteria P.F. Sanders, H.M. Lappin-Scott and C.J. Bass...............................................................................179 Corrosion Evaluation of Austenitic and Duplex Stainless Steels in Simulated Hydrogen Sulphide Containing Petrochemical Environments K. Saarinen and E. Hamalainen......................................................................................................191
vii
Damage of Pump Linkages and Tool Joints Caused by Crack Corrosion A. Kinzel.........................................................................................................................................201 Analysis of Soils Possibility to Give Rise to Pipe Metal Stress Corrosion Cracking V.G. Antonov and S.A. Loubenski....................................................................................................209 A Mysterious Downhole Corrosion Failure in an Oil Well A. Husain and A. Hasan..................................................................................................................215 CORROSION IN REFINERY AND PETROCHEMICAL INDUSTRIES Methodologies for Assessment of Crude Oil Corrosivity in Petroleum Refining S. Tebbal and R.D. Kane.................................................................................................................225 New Nickel Alloys Solve Corrosion Problems of Various Industries D.C. Agarwal and W.R. Herda....................................................................................................................233 Macro-Micro Segregation Bands (MMB) as a Main Factor Influencing Steel Applicability for the Petroleum Industry A. Mazur.........................................................................................................................................245 Fluid Catalytic Cracking Interstage and High-Pressure Cooler Corrosion S.M. Halawani................................................................................................................................255 Assessment of Cracks in a High Pressure Multilayered Reactor for its Fitness for Purpose A.M. Askari, M.I. AL-Kandari and P.K. Mukhopadhyay.................................................................263 Polythionic Acid Stress Corrosion Cracking of Incoloy 800: Case Study and Failure Analysis M.S. Mostafa and S.A. Hajaj...........................................................................................................273 Corrosion of Tube Heaters in Refineries: Symptoms and Cures A. Attou , A. Rais and H. Smamen...................................................................................................283 SEAWATER CORROSION Super Duplex Grade UNS S32750 for Seawater Cooled Heat Exchangers P.A. Olsson and M.B. Newman.......................................................................................................289 Evaluation of Aluminum Alloy 5083 Weldments to Stress Corrosion Cracking in Seawater A. Saatchi, M.A. Golozar and R. Mozafarinia.................................................................................301 Cavitation Corrosion Behavior of Some Cast Alloys in Seawater A. Al-Hashem, P.G. Caceres and H.M. Shalaby..............................................................................311 Microbiologically Induced Corrosion of a Stainless Steel Pipe H.H. Lee, M. Ali and K. Al-Omrani................................................................................................323 A Laboratory Study of Service Failure of Al-Brass Tubes in Arabian Gulf Seawater H.M. Shalaby, W.T. Riad and V.K. Gouda......................................................................................329 CORROSION IN THE BUILDING INDUSTRY Corrosion of Reinforced Concrete Structures and the Effects of the Service Environment S. Al-Bahar and E.K. Attiogbe........................................................................................................341
Corrosion of Concrete in Seawater viii
M. Pakshir and S. Esmaili...............................................................................................................353 Concrete Quality and its Effect on Corrosion of Steel Reinforcement E.K. Attiogbe and S. Al-Bahar........................................................................................................361 The Effect of the Type of Copper on its Corrosion Behavior in Kuwait’s Soft Tap Water H.M. Shalaby and F.M. Al-Kharafi.................................................................................................371 FUNDAMENTAL ASPECTS Corrosion Behavior of Vanadium in Aqueous Solutions W.A. Badawy, F.M. AI-Kharafi and M.H. Fath-Allah......................................................................383 The Effect of UV Irradiation on Passive Films Formed on Type 304 and 316 Stainless Steels M.S. Al-Rifaie, C.B. Breslin, D.D. Macdonald and E. Sikora..........................................................395 Kinetics of High Temperature Corrosion of a Low Cr-Mo Steel in Aqueous NaCl Solution W.A. Ghanem, F.M. Bayyoum and B.G. Ateya................................................................................407 Corrosion and Passivation Behaviour of Aluminium and Aluminium Alloys: Mechanism of the Corrosion Process F.M. AI-Kharafi, W.A. Badawy and A.S. El-Azab............................................................................417 The Susceptibility of Molybednum and Vanadium-Bearing Austenitic Stainless Steel Weldments to Intergranular Corrosion M.K. Karfoul..................................................................................................................................431 Effect of Crystallization on the Corrosion Behavior of Amorphous FeCr9P6C3Si0.2 Alloy in 1 M H2SO4 F. Hajji, S. Kertit, J. Aride and M. Ferhat.......................................................................................441 CORROSION PROTECTION AND MONITORING Experience With VOC-Compliant Waterborne and High Solids Coatings in Corrosive Environments P Kronborg Nielsen........................................................................................................................449 Anticorrosive Film-Forming Nonpolluting Products Achieved in Romania R. Serban, N. Moga and E. Stockel.................................................................................................461 Cathodic Protection Under Disbonded Coatings of 56 Inch Gas Pipeline Along the Kangan-Shiraz M. Pakshir......................................................................................................................................471 Synergistic Effect Existing Between and Among a Phosphonate, Zn2+, and Molybdate on the Inhibition of Corrosion of Mild Steel in a Neutral Aqueous Environment S. Rajendran, B.V. Apparao and N. Palaniswamy...........................................................................483 Evaluation of Corrosion Inhibitors for Carbon Steel, Monel 400 and Stainless Steel 321 in a Monoethanolamine Environment Under Stagnant and Hydrodynamic Conditions J. Carew, H. Al-Sumait, A. Abdullah and A. Al-Hashem..................................................................493 Laboratory Evaluation of the Effects of Ozone on Corrosion Rates and Pitting of Engineering Alloys S. Nasrazadani...............................................................................................................................501 A Critical Comparison of Corrosion Monitoring Techniques Used in Industrial Applications M.S. Reading and A.F. Denzine......................................................................................................511
ix
Detection, Localization and Monitoring of Stress Corrosion Cracking, Hydrogen Embrittlement and Corrosion Fatigue Cracks During Service Conditions Using Acoustic Emission L. Giuliani......................................................................................................................................521 Electrochemical Monitoring of Aerobic Bacteria and Automation of Biocide Treatments L. Giuliani......................................................................................................................................533 Corrosion Monitoring for Integrity of Pipeline G.L. Rajani.....................................................................................................................................543 Power and Desalination Plants: Pumps, Corrosion and Maintenance H. Hosni, N.J. Paul and A. Masri...................................................................................................555 CORROSION MANAGEMENT Impact of Metallic Corrosion on the Kuwait Economy Before and After the Iraqi Invasion: A Case Study F. Al-Matrouk, A. Al-Hashem, F.M. AL-Kharafi and M. EL-Khafif.................................................567 Corrosion Problems in a Steam Condensate System and Treatment of Condensate for Recovery G.L. Rajani.....................................................................................................................................581 Improved Cathodic Protection of Above Ground Storage Tank Bottoms: MAA Refinery Experience A.K. Jain, L. Cheruvu and M.E. Al-Ramadhan................................................................................597 Impact on Ship Strength of Structural Degradation Due to Corrosion M.A. Shama....................................................................................................................................615 NOVEL TECHNIQUES Contact Electric Resistance (CER) Technique for Monitoring of Process Plants and for Solving Practical Corrosion Problems K. Saarinen and T. Saario..............................................................................................................627 Design of Radio Frequency Methods for Corrosion Processes Monitoring Yu.N. Pchel’nikov, Z.T. Galiullin and A.S. Sovlukov.......................................................................637 A New, Rapid Corrosion Rate Measurement Technique for All Process Environments A.F. Denzine and M.S. Reading......................................................................................................647 Assessing Corrosion of Thick Marine Paints by Surface Corrosion Potential Mapping (SCM) and AC Impedance Spectroscopy (EIS) A. Husain........................................................................................................................................657 Optics and Lasers in Corrosion Laboratory K. Habib and F. Al-Sabti................................................................................................................669
Author Index...................................................................................................................................677 Subject Index..................................................................................................................................679
x
ORGANIZING COMMITTEE Jasem Al-Besharah Khaled Al-Muhailan Abdulhameed Al Hashem Hamdy M. Shalaby Abbas Ali Khan Hussain Shareb Jamal Al-Hajji Khaled Shehab Khalifa Al-Feraij Abdel Monem Bedair Mohammad Ashkanani Mohammad Al-Rasheed Mohammed Al-Qalaf Abdul Khaliq Mustafa Khawla Al-Rifaee
Chairman Rapporteur Coordinator Member Member Member Member Member Member Member Member Member Member Member Member
KISR KFAS KISR KISR KFAS OAPEC KU KNPC MEW PIC KOC GCC KCS KISR MO
INTERNATIONAL ADVISORY COMMITTEE Ahmed M. Shams El Din John Oldfield Russel D. Kane Digby D. MacDonald
Member Member Member Member
UAE UK USA USA
Chairman Rapporteur Member Member Member Member Member Member Member Member
KISR KISR KISR KISR KU KOC KOC KOC KNPC KNPC
SCIENTIFIC COMMITTEE Hamdy M. Shalaby Abdulhameed Al Hashem Khalid Habib Adel Hussein Waheed Badawi Afkar Hussain Emad Al Naser Eman A. Razzak Al-Shayji Lakshmipati Cheruvu Fahed Al-Otaibi
xi
ACKNOWLEDGEMENT The Organizing Committee was deeply honored by the patronage of H. H. The Crown Prince and Prime Minister Sheikh Saad Al-Abdullah Al-Salem Al-Sabah, which reflects his keen interest in science and technology. The Committee was also grateful for the financial support of the Kuwait Institute for Scientific Research, Kuwait Foundation for the Advancement of Science, Kuwait National Petroleum Company, Kuwait Oil Company, Ministry of Electricity and Water, Kuwait University, Ministry of Oil, the Gulf Cooperation Council, Kuwait Chemical Society, Organization of Arab Petroleum Exporting Countries, and Petrochemical Industries Company. The Committee would also like to extend its deep appreciation for the effort and time put forth by the distinguished honorary speakers, the members of the International Advisory Committee, and the Scientific Committee. We would like to thank our colleagues, the members of the working committees, at the Kuwait Institute for Scientific Research and the chairmen and cochairmen of the sessions, who provided unlimited assistance at times when it was really needed. Finally, we feel deeply indebted to the authors of papers and participants for their valuable contribution to the success of the conference
Jasem Al-Besharah Chairman, Organizing Committee
xii
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CORROSION MANAGEMENT V. Ashworth Global Corrosion The White House, Victoria Road, Shifnal, England, TF11 8AF
ABSTRACT The consequences of corrosion are often very costly. Little surprise, therefore, that a substantial engineering effort is directed towards its prevention and control. By contrast, little consideration seems to be directed towards making anti-corrosion effort cost-effective. This paper addresses the problem of ensuring value-for-money corrosion engineering and the possible limitation of unnecessary corrosion control activities. Corrosion in itself is not important, but the consequences of corrosion failure may well be. So the first step in corrosion management is a corrosion risk assessment to evaluate the risk associated with failure in any item. This is not an evaluation of the risk of failure alone, but of the consequences should that failure occur. Given an assessment of risk, a strategy of corrosion management can be constructed. This might involve lifetime corrosion control for items identified as producing a high risk. A less rigorous, but monitored, level of protection might be adopted for medium risk items, whilst no action at all may be considered necessary in the case of low risk items. Thus, resources are distributed according to the risk. Once a strategic approach has been defined, the tactics of corrosion management may be determined. These will include not only the specific corrosion control activity or activities that will be used in any given case, but also any monitoring and inspection requirements that are necessary. The object is to maintain corrosion within acceptable limits at minimum cost in all parts of the facility and throughout the facilitiy’s life. Key Words: Risk, probability, monitoring, inspection, corrosion management
INTRODUCTION The purpose of industry is to make a profit from the production of supplies and artefacts. In an increasingly competitive world, there is continuing pressure on prices. If the selling price is under pressure, profitability may only be maintained or increased by cutting costs. Any factor that serves to increase costs represents a tax on profits. Corrosion is one certain consequence of using engineering materials. Commonly, corrosion will be modest, but not always. In the hydrocarbon production and processing industries and the chemical industry, for example, the exception almost becomes the rule. Since corrosion brings a cost, it impacts profits.
1
Plenary Lectures
THE COST OF CORROSION The first formal attempt to assess the cost of corrosion to a nation was made in the UK in 1970 [1]. Since that time, similar studies have been published, in Australia [2], the US [3] and elsewhere. One surprising outcome is that the cost of corrosion to an industrialized nation is relatively constant at approximately 3.5% of the gross domestic product (GDP). To put the matter in context, this is substantially higher than the cost of fires which in the UK is put at ~0.5% GDP. Sedriks [4] reported the experience of the Dupont Company in the period 1968-71. After examining 685 plant failures, it was concluded that 55% were due to corrosion and 45% to mechanical failure. This may be regarded as a remarkable outcome given that, by the standards of the time, Dupont was corrosion aware and the greater proportion of the material that failed was stainless steel. The Dupont experience was mirrored by that of Britoil in the UK during the period 1978-88 [5]. As Table 1 shows, 33% of the failures that were analysed were attributed to corrosion. Table 1. Analysis of Oilfield Failures [5] Type Corrosion (all forms) Fatigue Mechanical damage/overload Brittle fracture Fabrication defects (not welding) Welding defects Other
Frequency (%) 33 18 14 9 9 7 10
Not infrequently, corrosion hits the headlines because some particularly dramatic failure occurs resulting in the loss of life and property. At Flixborough Works in the UK, a chemical explosion related to a corrosion failure resulted in 28 fatalities, 36 serious injuries, virtual destruction of the plant and damage to some 2000 third party properties [6]. In Guadalajara, in the early 1990's, stray current corrosion of a water pipe produced a failure that caused erosion-corrosion of an adjacent gasoline line [7]. The leaking gasoline caught fire, producing an explosion in which tens of local inhabitants were killed. The accumulated corrosion failures at Dupont and Britoil were potentially costly and the two accidents were certainly so. That cost is ultimately borne by the community, but in the short term it falls on the industry concerned. There is a growing awareness in industry of the costs of corrosion. This engenders a desire for more effort and expenditure on corrosion prevention and control. Industry finds willing allies in meeting this goal from the companies that sell anticorrosion materials and systems. Their products and services are not free.
2
Ashworth
THE COST OF CORROSION PREVENTION AND CONTROL Expenditure on corrosion prevention and control is no less a tax on profits than the cost of corrosion itself. It is, therefore, entirely appropriate to ask if this expenditure is necessary. An accountant can usually produce figures to illustrate the impact of a corrosion failure on profit. Some tangible, and less tangible, inputs to the calculation are given in Table 2. Table 2. Cost of Corrosion Failure Safety hazards Loss of capital plant or equipment Fire/explosion Loss of production capacity Loss of product quality Maintenance/repair/replacement Loss of stored/entrained product
Pollution clean-up costs Increased insurance premiums Loss of consumer confidence Alienation of workforce Increased scrutiny by statutory bodies Public image Accountants
By contrast, the accountant is rarely moved to make an assessment of the cost of not having a failure. If a plant and equipment operate without breaking down, everybody is usually well satisfied. It is rare to question whether the cost of achieving that performance has been excessive or even worthwhile. Table 3 lists some sources of possible over-spending in endeavouring to avoid corrosion failures. Table 3. Costs of Over-Protection Unnecessarily expensive materials Overdesign of metal sections Excessive weight Excessive inhibitor consumption Excessive monitoring Excessive inspection
Excessive data handling Overdesigned CP systems Over-operated CP systems Excessive replacement stocks Premature retirement of equipment Over specified protective coatings
The items that relate to evident over-engineering in this list may be readily understood. However, two areas, monitoring and inspection, are worth singling out because they are so often regarded as a good thing, i.e., they have intrinsic merit. This is far from the case. Industrial corrosion monitoring is commonly excessive both in terms of the extent of monitoring and the sophistication of the equipment used. We need to remind ourselves that corrosion monitoring has never controlled any corrosion. This author believes that corrosion monitoring should be used almost exclusively in a process control function, the process, in this case, being corrosion. Thus, a monitoring device should only be used in circumstances where the output from it can validly be used to adjust some corrosion controlling function, e.g., inhibitor injection or cathodic protection (CP) output. It follows that probes should not be installed where such action cannot be taken, nor should they be installed where probes 3
Plenary Lectures
installed elsewhere fulfil essentially the same function. In practice, these rules are observed more in their breach than in their application. Likewise, for reasons that are well understood, corrosion probes provide precise information on what is happening on the probe and, often, comparatively little about what is happening on the pipe or vessel wall. The value of the probe is that it detects change and prompts review and, possibly, action. Very often simpler means of detecting change than the use of corrosion-measuring devices will serve the same function at less cost, e.g., reference electrodes, pH probes, dissolved oxygen meters and moisture meters in the gas phase. They also have the merit, where relevant, of permitting continuous readout which allows the identification of the precise 'upset' that produces corrosion. Inspection is similarly open to over-engineering. How often is inspection carried out because we have the opportunity ? It is remarkable that upon shutdown, the internal inspection of tanks and vessels will often be considered mandatory. Yet pipelines or pipework that carry the same fluids are not inspected. It has been pointed out [8] that the cost of inspection is high, often equivalent to 2-6% of the invested capital. That is a significant tax on profits. What is often overlooked is that inspection can often be potentially dangerous and may even produce conditions conducive to corrosion, e.g., when sulphuric acid tanks are opened up for inspection. If corrosion costs money and corrosion control costs money, how do we target the optimum approach that strikes the correct balance between ignoring corrosion and seeking to control it ? The answer lies in
• assessing the risk of corrosion failure on an item-by-item basis • developing a lifetime corrosion management plan for each item to contain corrosion at an acceptable level of risk This latter implies the need for risk modification, and sometimes, a defined degree of corrosion control activity. The goal is to maintain corrosion at an acceptable level. This begs the question: What is acceptable ? RISK It is a very dangerous game to talk about risk, largely because it is an ill-defined subject, and yet, everybody has a perception of what it is. What is clear is that risk is bad. We always associate risk with the likelihood of an undesirable or catastrophic event occurring. Thus, we talk of the risk of climbing, flying or crossing the road, but never of the risk of a traffic-free journey to work. Moreover, not everybody sees a particular risk in the same way. For example, we know that individuals are prepared to take a greater risk if they feel that they have some control over the process, or if the risk is associated with some activity considered to be beneficial [9]. That is why climbers do not appear to recognise the risk of climbing that is so self-evident to the rest of us. They feel a measure of personal control and perceive a personal benefit. In short, risk is subjective. This is a worrying matter if, before we can proceed to a corrosion management plan, we need to make a corrosion risk assessment. It is necessary to remove a little fuzziness. The most appropriate definition of risk is
4
Ashworth
Risk = Probability x Consequence (1) Despite its formality, this is not a very precise equation as we shall see. It does, however, indicate that risk is not simply a reflection of the probability of something bad happening. Probability Probability has the appearance of precision because it is a mathematical quantity. It derives from the stochastic nature of the frequency of the occurrence of events. Given sufficient failure data, a classic probability may be calculated to reflect the likelihood of a particular event occurring. In using the probability, it is important to be sure of its validity. Consider above-ground pipelines. The probability of failure, taking the overground pipeline population as a whole, is much less than the probability of failure of a small diameter (150250 mm) line. Considerable error can arise from using the former probability in the latter case. Nevertheless, given valid and relevant failure data, a useful quantitative probability can be assessed. Very commonly the failure data from which probability is calculated do not exist. The so-called Bayesian technique [10] can then be used to compute a probability. The technique uses prior knowledge (e.g., failure rates in similar, but not identical, circumstances elsewhere and the view of experts) and refines it steadily as specific information becomes available with plant operation. In the limit, of course, when the specific information database builds up sufficiently, the classic approach becomes more reliable. Probability has one other unfortunate characteristic: uncertainty. The probability of an occurrence may be low, but it can happen tomorrow. Consequences The consequences element in Eq. 1 relates to the perceived magnitude of the loss if the failure occurs. This is a very subjective matter since different people rate the various consequences of an individual event differently, and many even disagree about the consequences that derive from that event. Nevertheless, it is possible for these individuals to list the potential consequences and to rate each consequence on a scale of 1-10 to compute a consequence for Eq. 1. The number that emerges is entirely subjective. We see that the probability we use in Eq. 1 may hide a degree of uncertainty because of a lack of failure data, and it may include educated guesswork. Similarly, the consequence is a subjective valuation of the consequences of a failure. The outcome is a value for risk which, although numerical, is not exact. CORROSION RISK ASSESSMENT The foregoing does not seem to suggest that any form of risk assessment is likely to be productive. Yet the experience is that, in the case of corrosion, it can be helpful and rewarding. In carrying out a corrosion risk assessment it is axiomatic that corrosion does not matter, but its consequences do. The assessment then aims to combine objective estimates of the possibility of a corrosion failure with the operating company's view of the level of
5
Plenary Lectures
undesirability of the consequences of it occurring. It will be seen below that the probability component of the risk is rendered quantitative and that the consequence component remains subjective, but reflects accurately the perceptions and ambitions of the people that own and run the plant. Subjective it may be, arbitrary it is not. Methodology The methodology of conducting a corrosion risk assessment has been discussed in detail elsewhere [11]. Only a brief outline is presented here. If we take the example of a corrosion risk assessment in a refinery or chemical plant, it may be as coarse or as refined as the operator wishes. First, the plant is divided into systems, e.g., the gas sweetening unit. Second, each system is broken down into items. These usually comprise individual components, e.g., a vessel, a heat exchanger, a pump or a specifically identified length of pipework associated with the system. An item may be more widely defined in a coarse corrosion risk assessment or more closely defined in a fine analysis. In the latter case, for example, it may involve considering a vessel as a number of discrete items according to the known variation in fluid composition with height. Equally, it may be necessary to single out non-stressed relieved welds as separate items and, from an internal corrosion point of view, each dead leg. What follows is an outline approach to corrosion risk assessment which has proved to be successful. Other methods are available that operate somewhat differently [12,13], but aim to achieve the same objective. Life Factor The aim of the corrosion risk assessment is to assign a risk number to each item using a risk equation similar to equation (1). There are a variety of ways to deal with the probability element. The experience of the author's company is that it is best dealt with by assigning a life factor (L) that relates to the residual corrosion life of the item. The residual corrosion life is the anticipated time required for corrosion at the predicted rate, or rates, to lead to failure to perform the required mechanical duty. Given information on the materials of construction, the exposure environment, and the relevant circumstances (e.g., temperature, pressure, flow, heat transfer, and stress), the morphology of corrosion can be predicted with confidence. Where uniform attack is expected, maximum penetration rates can be calculated using conventional corrosion engineering practices, including public domain algorithms [14,15,16], in-house database information and, if necessary, modelling. If localized corrosion is expected (e.g., pitting attack or one of the cracking modes of failure), probabilistic analyses of failure [17 ] are more useful. For risk assessment purposes, the estimate of residual life in years is transposed to a dimensionless L. For example, an anticipated time to failure shorter than the time to the next shutdown would be assigned an L = 3. Anticipated lives beyond that point would attract L = 2, except where the residual life is put at >10 years in which case L = 1. Of course, this breakdown is arbitrary and the individual cut-offs, and the relative scoring, can be selected to match the requirements of the plant owners.
6
Ashworth
Consequence Factors The point has been made that the consequence of a corrosion failure are more important than the failure itself. Thus, the consequences that bear on plant operators' minds include:
• • • • • • • •
Safety, Production, Emergency repair, Operability, Environment, Third party interests, Customer perception, and Public perception.
Adverse effects on any, or all, of these may often flow from an isolated failure. The consequence factor (C) is a numerical assessment of the perceived consequences of a corrosion failure. The number is arrived at using structured group discussions with plant management, operations personnel, maintenance engineers, loss prevention officers etc. It elicits a subjective assessment. However, because individuals work towards a consensus in a group, and the methodology of subsequent analysis is rigorous, the rankings produced accurately reflect, in a quantitative way, the operating aspirations of the company concerned. Thus, the C numbers provide the relative importance attached to any consequence. Since the risk numbers that finally emerge are not absolutes but reflect perceived risk in a relative manner, the subjectivity of the consequence analysis is not only permissible, but desirable. The risk assessment becomes plant specific. That is, identical plants operated by different companies or in different locations will produce different risk assessment results. There are two elements in establishing the value of C:
• The individual consequence, and • The events that can lead to that consequence. Some consequences will always be regarded by company personnel as more undesirable than others; to that extent, the staff can develop a point loading to be applied to each. This gives a consequence rating (F); a typical set to emerge in one case is given in Table 4. Table 4. Typical Consequence Rating (f) Consequence risk of safety to personnel or public loss of production pollution loss of produce quality loss of consumer confidence
Rating (F) 10 9 3 1 1
7
Plenary Lectures
It should be noted that the company in question is not concerned about contamination of the product by corrosion or any consequent alienation of the customer. This is a typical response from a primary producer; quite different numbers would have arisen in an assessment made in a food or pharmaceutical factory. The events that lead to a given consequence produce an event rating (P). It is clear that a number of events which might occur in a process plant, may lead to the same consequence. The company staff are able to identify and rank these events according to their perceived undesirability, as shown in Table 5. In this case, the table relates to two plants owned by the same company in which one uses the product of the other. Table 5. Typical Event Rating (P) Consequence Safety
Outage
Pollution Quality
Event Crack in a toxic line or equipment Pinhole in a melt line Crack in a flammable line or equipment Crack in other HP line or equipment Pinhole in a flammable line or equipment Pinhole in other HP line or equipment Other cracks Pinhole in toxic line or equipment Other leaks Falling objects Plant no. 1 - no standby Plant no. 2 (HP) - no standby Plant no. 2 (LP) - no standby Plant no. 1 - standby Plant no. 1 - non-critical - no standby Plant no. 2 - standby Plant no. 2 - non-critical Plant no. 1 - non-critical - standby Marine Atmospheric Final product (colour only) Intermediate product
Rating (P) 10 10 9 7 7 6 5 4 2 1 10 9 7 5 5 4 3 3 10 5 10 5
It is not uncommon when considering safety, for staff to take into account the inventory of a system. Thus, they will commonly regard a crack in a system or item with a high inventory and, therefore, a high potential for damage, as more significant than one where the inventory is small. Different values of P may then arise according to the volume of an unisolatable part of the system. Any item included in the unisolatable part attracts the P value for that part. The values of F and P are combined to yield C: 8
Ashworth
C = ∑Cx =
x=n
∑x = 1
fn (Fx, Px)
(2)
Where the subscript x refers to each of the consequences, e.g., safety, pollution etc. in turn. The Risk Equation The risk equation must reflect the operating company's perception of risk associated with various forms and rates of failure. The equation, which is derivative of Eq. 1, produces a numerical assessment of risk (R) and takes the form: R = fn (L, C)
(3)
The shape of the function linking the life and consequence factors is determined by a formalized heuristic procedure. The function is modified through a series of computer iterations with the effect on the value of risk being assessed after each iteration. Allocation of Risk Classes The numerical value of risk can be calculated for each item in the plant. The higher the value of R for any item, the greater the risk and the more attention that must be focused on the local corrosion situation. In practice, the spread of numerical risk values amongst all the items within a plant usually proves to be a discontinuous spectrum. That is, the risk numbers tend to fall into clusters with distinctive breaks between. This is an inevitable consequence of data like those recorded in Tables 4 and 5. It permits a convenient reduction of the numeric data into risk classes (e.g., low, medium and high). Such a sub-division aids communication of the outcome of the corrosion risk assessment either on a narrative basis or as colour coded P and ID's. It also assists with establishing a corrosion management programme. Risk Modification The fact that, in assessing a new or existing plant, areas of high risk have been identified, does not mean that the risk must be tolerated. The aim should be to moderate the risk and to move to a more acceptable condition. Some methods of corrosion risk assessment [12] do not proceed as far as a risk equation or a risk number, but rather consider separately the perceived severity of the probability (L in this case) and the consequence (C). This produces a risk matrix as shown in Table 6. Table 6. Risk Matrix Consequence H M L
H H HM M
Probability M HM M ML
L M ML L
9
Plenary Lectures
Risk categories can then be devised as shown in Table 7: Intelligent risk modification aims to move towards Zone 3. This is not to aim at zero failure but to achieve, by good management, a tolerable level of risk. Table 7. Risk Categories High consequence High probability 1
High consequence Low probability 4
Consequence Low consequence Low consequence High probability Low probability 2 3 Probability Using the risk equation approach, the aim is to concentrate resources on areas of high risk in order to reduce the risk number (i.e., modify the risk). Some care has to be taken here. It will often be the case that a high risk number will place the specific item in Zone 1 of Table 7. Clearly, for these items, it is important to move towards Zone 3 by means of corrosion control activities that reduce the risk number. Somewhat lower risk numbers may fall into either Zone 2 or Zone 4. Indeed, the same risk number may apply to either a low consequence/high probability situation or a high consequence/low probability. The former simply represent failures that will be an irritation; pinholing in a seawater cooling line. The latter are certainly more serious; pinholing in a dry flammable gas line, for example. In the case of Zone 2, the high probability means that sufficient data were available to assess the probability fairly accurately. By contrast, in the case of Zone 4, the reverse is true, and there may be considerable uncertainty. If the probability (in our case, L) has been calculated using tried and tested tools, then identical risk numbers that derive from high probability/low consequence and low probability/high consequence events are equally reliable. Where the L calculation has used limited data, an uncertain algorithm or stochastic techniques, that level of reliability is absent. Thus, in moving from Zone 1 towards Zone 3, it is often better to achieve Zone 2 rather than Zone 4. Equally, it may be more important to move from Zone 4 than to move from Zone 2. In general, the consequences of failure are usually not amenable to modification; thus, risk can only be modified by changing the probability. For that reason, and to introduce security, risk modification needs to pay attention to moving to situations where the probability is known or can be determined with some degree of confidence.
10
Ashworth
CORROSION MANAGEMENT Corrosion risk assessment is not an end in itself. It identifies areas where corrosion may be safely ignored and where it must be attended to. It even provides the pointer to where resources will be spent with greatest reward. Thus, it provides the evidence that permits the construction of a cost-effective corrosion management programme. The objectives of a programme relate to the whole life of a facility and are to
• Maintain corrosion within predetermined acceptable limits at minimum cost, • Develop and facilitate rapid access to, records showing the corrosion status of each item within the facility in order to form a basis for future corrosion management decisions, and to provide assurance for managers, owners and statutory bodies, and • Ensure that corrosion upsets are quickly identified and appropriate remedial action is implemented, if necessary, to minimize the consequences of any failure. There is no universal corrosion management programme. Targeting these objectives is a unique exercise for every facility. However, the philosophy of corrosion management is common to them all. An overall strategy for corrosion management must first be agreed upon, and then the tactics become self-evident. Strategic Considerations The corrosion risk assessment will have produced a risk ranking for all items of a plant. This will enable a strategy for corrosion management to be set down. Table 8 illustrates a strategy that might be drawn up for an industrial facility. Table 8. A Corrosion Management Strategy Assessed Risk High
Medium Low
Alternative Corrosion Management Options Corrosion prevention, or corrosion control for life, or corrosion control to meet planned maintenance or planned replacement Corrosion control for life, or planned maintenance No action, replace if required
It will be noted that corrosion prevention, or careful corrosion control, is dictated by a high risk classification. By contrast, a low risk classification justifies no corrosion controlling action. A medium risk requires some action. Thus, corrosion management involves a spectrum of activity from no action to considerable action according to the risk. However, taking no action, or taking action, is not corrosion management unless the decision to follow the particular course has been based on an assessment of risk. Action where it is not needed, like inaction where it is, represents a waste of resources and a tax on profits.
11
Plenary Lectures
It will be recalled that corrosion risk assessment is carried out by dividing the plant into systems and items. Ultimately, the output relates to individual items. It is possible for an item within a system to have a high risk classification whilst other items in the same system belong to a lower risk class. The decision must then be made whether to apply corrosion control to the system in order to preserve the item, or to ensure corrosion prevention for the item (say, by the use of more corrosion-resistant material) and avoid dealing with the system. The application of the broad strategy does allow, and requires, some flexibility in the tactics adopted. Tactical Considerations The complete elimination of the chance of corrosion failure, i.e., corrosion prevention, in a high risk area is rarely possible in an existing plant. Invariably, it would require a significant engineering change, for example, replacement of existing materials by corrosionresistant alloys or modification of the process (e.g., addition of gas dehydration). Even with new plants, such proposals might raise major design and engineering problems, not to mention cost. It is much more likely that active corrosion control will be adopted with the objective of extending the time to failure of an item beyond the planned life of the plant, or up to some planned maintenance shutdown. The adoption of this tactic requires that:
• the performance targets for the corrosion control are defined, and • procedures are put in place to ensure the targets are met. The performance target may be set in terms of an allowable rate of metal penetration. This approach will most commonly be adopted when uniform corrosion is anticipated. Alternatively, limits may be set on some parameter that is an indication of fluid corrosivity, e.g., electrode potential in anodic and cathodic protection systems, dissolved oxygen in oilfield water injection or boiler feedwater, pH, temperature, or dewpoint. Irrespective of which approach is adopted, it will be necessary to obtain on-line information to make adjustments as required. Thus, corrosion monitoring is necessary, and it then forms an essential segment of the corrosion management plan. Table 9 lists some monitoring techniques and indicates how they may be used in corrosion management. Table 9. Corrosion Monitoring in Corrosion Management Corrosion Control Strategy
Examples of Adjustments and Activities Based on Data Monitoring Inhibition of crude oil On-line probes (e.g., Adjust inhibitor dosage, pipelines coupons, electrical resistance change inhibitor type, probes) discontinue inhibition De-oxygenation of boiler O2 probes Adjust oxygen scavenger, feed-water check pump seals, etc. Impressed current CP Potential Adjust system output Anodic protection of Potential Adjust system output sulphuric acid plant 12
Monitoring Technique
Ashworth
Dehydration of process gas
On-line probes, moisture detection
Temporary inhibition, overhaul dehydrator
The key to effective corrosion management is information since it is on the basis of that information that on-going adjustments to corrosion control are made. Information is valid data. Thus, to make effective corrosion management decisions on a day-to-day basis, the monitoring data must be valid. This is not simply a requirement for the probes to be operating correctly. It requires that they be placed in the most appropriate places, i.e., at those points where the corrosion controlling activity might be expected to work, but where it might equally be expected to be least effective, e.g., remote from the inhibitor injection point. In many cases specially designed traps are introduced into a plant so that corrosion probes may be inserted. These often produce their own microenvironment, atypical of the plant itself, and with little hope of effective entry for an inhibitor. Data from a probe in such a location are unlikely to be relevant to corrosion management elsewhere in the system. Invalid data leads to ineffective corrosion management. Keeping Track In any facility the means of corrosion management will vary from place to place. In one location a corrosion resistant alloy may be used; in another, CP allied to coating may be employed, whilst elsewhere no action may be taken because the consequences of any failure are regarded as unimportant. In short, no corrosion management action is taken that does not contribute positively to meeting the objective of containing risk whilst maintaining the level of action at the minimum necessary. It is important to ensure that the targets are being met. Overshooting the target will involve excessive corrosion control costs, whilst undershooting the target may lead to a situation that cannot economically be recovered. Corrosion monitoring is not appropriate for the purpose since it rarely provides evidence of the metal loss from a pipe or vessel wall. That is, aggregating the output from probes over time does not give any indication of the loss of a section. The value of corrosion probes is that we rapidly develop experience so that we can be reasonably sure that when the probes read a given value, we are on target, and that a change in reading requires consideration of an adjustment to the corrosion controlling activity. Reference electrodes, pH probes, moisture meters etc. often fulfil the same function. Thus, corrosion probes and the like, do not provide quantitative performance assessment. That can only come from inspection and nondestructive testing (NDT). These activities are part of corrosion management since they provide reassurance, identify wasteful corrosion control activity, and permit reassessment of the corrosion management programme. The same critical approach that was adopted in setting up the corrosion control strategy must be applied to the inspection strategy. That is, the resources must be applied according to the risk. If we have attempted to modify the risk by instituting some corrosion control activity, we should be tracking the success, or excess, of the activity in our inspection programme. Thus, inspection is not based on convenience, inspecting because an item is accessible (at shutdowns, for example). It should be based on the premise that if the consequences of failure are to be avoided and the cost of control is to be minimized, inspection is necessary. There must, therefore, be a clear connection between the risk assessment output, the corrosion management strategy, the tactics of corrosion management and the inspection programme. 13
Plenary Lectures
The key point approach to NDT is particularly effective. Here a limited number of points, in areas where validation is required, that are regarded as typical and extreme, are identified for NDT inspection at regular intervals. This provides, on a temporal basis, a readout of the progress of corrosion which will validate, or otherwise, the targeting achieved by, say, inhibitor injection. Similarly, during internal inspection, the risk assessment will have identified particular areas of concern, e.g., tube/tube sheet assemblies, tube baffles, and non-stress relieved welds, which must become the focus of activity. Again, this will confirm whether the corrosion control is adequate or perhaps is insufficient or excessive. The data that are produced from the inspection activities must be valid and limited in volume so as not to deter analysis or hide anomalies. Thus, it is important to restrict key points and inspections to critical positions and to limit the frequency of inspection and survey work. The time to the next inspection or survey should be indicated by the outcome of the current work. That is, a lifetime fixed interval programme will usually prove wasteful; inspection and survey should be carried out on an as-needed basis. A valuable template giving an approach to the re-classification of in-service inspection is to be published in 1996 [18] and has been reviewed in reference [12]. Review From time to time a corrosion management programme should be reviewed at both the strategic and tactical level. In human affairs, things change. The management of a facility will always be alive to current market trends, competitors activities, interest rate movements and so on. Inevitably, it may be necessary to revise the management objectives from time to time. Since the corrosion management programme was constructed to meet the objectives of an earlier plant management plan, it will be necessary to review the programme and possibly to alter it. Likewise, the pace of technological change is rapid compared to the anticipated lifetime of most facilities. Thus, newer, more effective, cheaper means of achieving the same ends may emerge, and indeed, it may be possible to adopt them in place of existing tactics within the corrosion management programme. Thus, the programme is not a fixed blueprint, but a means to an end that must be reviewed and revised to meet the current management objective. One objection that is raised to corrosion management planning comes from the corrosion engineers themselves. They draw attention to the fact that by fixing permissible rates of metal loss, the lifetime of the facility is effectively determined. Further, that management will often, at a later stage, decide to extend the required operating life. There is then a mismatch. The argument seems to be that corrosion management planning should ignore the present requirements and anticipate the future requirements of the management. This is an extremely wasteful approach. Certainly there is a possibility that a mismatch will occur and will need to be overcome. That will be achieved at some cost. That cost must be attributed to the decision to go for life extension and is, therefore, a natural consequence of that extension. It needs to be included in the cost benefit analysis of extension, not hidden in lifetime overspending in anticipation that life extension might be required. It may not be. ILLUSTRATIONS
14
Ashworth
Two recent examples illustrate how corrosion risk assessment provided important results for the clients. In the first instance, the assessment of a petrochemical complex in the Middle East found the plant to be extremely well engineered from the corrosion standpoint. It was constructed predominantly in carbon steel, with excursions into more exotic metallurgy only where the process conditions demanded it. However, the assessment highlighted, somewhat to the client's surprise, the cooling water system as a high risk area. By using a closed system with secondary cooling by seawater and specifying high quality primary water with corrosion inhibitor injection, the designers had judged that it was possible to construct the majority of the cooling system in carbon steel. Certain that the primary heat exchangers were, however, constructed in a stainless steel due to the aggressivity of the process fluid, the corrosion risk assessment identified modes whereby the quality of the cooling water could be adversely affected (e.g., by leakage of seawater at secondary plate exchangers). Failure to maintain cooling water within specification would very rapidly lead to stress corrosion cracking of one of ten critical stainless steel process exchangers, failure of any one of which would halt production. In view of this, Global Corrosion put forward recommendations for modest on-line monitoring of cooling water quality. Tied to this was the setting up of a formal action plan to be followed in the event a sudden deterioration in water quality should be detected. The client accepted and implemented these recommendations but, unfortunately, not before one failure of the type predicted occurred. The second example derives from an installation, also in the Middle East. The corrosion risk assessment concluded that the absence of CP on water storage tanks, together with the prevailing soil conditions, would result in high tank bottom corrosion rates. Since an adequate supply of water was essential to maintain production, the assessment concluded that the prospective failure of the tanks constituted a high risk and it was strongly recommended that CP be installed. In the event the client was reluctant to accept and act upon the outcome of the report. Just under a year later the raw water tank perforated due to soil-side corrosion. The resulting loss of water caused a two week interruption in production, prompted a belated decision to install CP and engendered in the client a heightened appreciation of the benefits of corrosion risk assessment and the need for effective corrosion management. CONCLUSIONS Corrosion cannot be ignored for it will not go away. However, there is little merit in controlling corrosion simply because it occurs, and none in ignoring it completely. The consequences of corrosion must always be considered. If the consequence of corrosion can be lived with, it is entirely proper to take no action to control it. If the consequences are unacceptable, steps must be taken to manage it throughout the facility’s life at a level that is acceptable. To manage is not simply to control. Good corrosion management aims to maintain, at a minimum life cycle cost, the levels of corrosion within predetermined acceptable limits. This requires that, where appropriate, corrosion control measures be introduced and their effectiveness ensured by judicious, and not excessive, corrosion monitoring and inspection. Good corrosion management serves to support the general management plan for a facility. Since the latter changes as market 15
Plenary Lectures
conditions, for example, change, the corrosion management plan must be responsive to that change. The perceptions of the consequences and risk of a given corrosion failure may change as the management plan changes. Equally, some aspects of the corrosion management strategy may become irrelevant. Changes in the corrosion management plan must, inevitably, follow. REFERENCES 1. T.P. Hoar (Chairman), Report of the Committee on Corrosion and Protection, HMSO, London, 1971 2. B.W. Cherry and B.S. Skerry, Corrosion in Australia - the Report of the Australian National Centre for Corrosion Prevention and Control Feasibility Study, Monash University, 1983. 3. L.H. Bennett, National Bureau of Standards Special Publication 511.1, NBS, Washington, 1978. 4. A.J. Sedriks, Corrosion of Stainless Steels, Wiley, 1979, p. 7. 5. Kermani, An Overview of Wet H2S Attack: Types, Causes and Problems, in Papers of the Conference on Wet H2S Attack on Steels, Institute of Mechanical Engineers, London, 1996. 6. F. Lees, Loss Prevention in the Process Industries, Vol 2, p863, Butterworth (1989) 7. J.M. Malo, V. Salinas and J. Uruchurtu, Materials Performance 33, 8, 1994, p. 63. 8. C. Edeleanu and J.G. Hines, Materials Performance 29, 12, 1990, p. 68. 9. P. Slovic, Science 236, 17 April 1987, p. 280. 10. M.E. Giuntini, Proceedings of Fourth Space Logistics Symposium, Florida, November 1992. 11. V. Ashworth and W.R. Jacob, Proc. Corrosion 32, Australasian Corrosion Association, 1992. 12. B. Spalford, Carbon steel equipment in wet H2S service, Papers of Conference on Wet H2S Attack on Steels, Institution of Mechanical Engineers, London, 1996. 13. Private communication, Shell-Expro, UK 14. C. de Waard, V. Lutz and D.E. Milliams, Corrosion 47, 1991, 976. 15. F.A. Posey and A.A. Palko, Corrosion 35, 38 (1979) 16. J.W. Oldfield, G.L. Swales and B. Todd, Proc. 2nd BSE/NACE Corrosion Conference, Bahrain, 1981. 17. M. Akashi, Proc. Conference of Life Prediction of Corrodable Structures, NACE, 1991. 18. EEMUA publication 179, A Working Guide for Carbon Steel Equipment in Wet H2S Service (to be published in 1996)
16
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
THE DETERMINISTIC PREDICTION OF DAMAGE D.D. Macdonald Center for Advanced Materials The Pennsylvania State University 517 Deike Building, University Park, PA 16802, USA
ABSTRACT As our industrial and infrastructural systems (refineries, power plants, pipelines, etc.) age, considerable economic incentive develops to avoid unscheduled outages and to extend operation beyond the design lifetime. The avoidance of unscheduled outages is of particular interest, because the failure of even a minor component can result in the complete shutdown of a facility. For example, the unscheduled shutdown of a 1000 Mwe nuclear power plant may cost the operator between US $1 million and US $3 million per day, depending upon the cost of replacement power and other factors. However, if component failures could be accurately predicted, maintenance could be performed during scheduled outages, the cost of which has already been built into the price of the product. With regard to life extension, the successful extension of operation beyond the design life translates into enhanced profits and the avoidance of costly licensing and environmental impact assessments associated with the development and construction of a new facility. In this case, as well, the key to successful operation is the ability to avoid downtime, and hence, to maintain production. Eventually, the frequency and severity of unscheduled outages will render operation uneconomic, and at that point, replacement of the facility is necessary. In order to develop effective inspection and maintenance scheduling and life extension technologies, it is first necessary to predict the evolution of damage into the future as a function of various system variables. The only effective prediction technologies are those based on determinism, in which the system behavior is described in terms of natural laws. In this paper, the deterministic prediction of damage, via damage function analysis (DFA), which provides a robust technology for estimating the damage function at future times, is described. The application of DFA to the prediction of pitting damage is illustrated by reference to pitting damage in condensing heat exchangers. Key Words: Corrosion damage, determinism, prediction, pitting corrosion.
INTRODUCTION Corrosion is a major cause of component failure, and hence, in the occurrence of unscheduled downtime, in complex industrial systems. In particular, the various forms of localized corrosion, including pitting corrosion, crevice corrosion, stress corrosion cracking (SCC), and corrosion fatigue (to name the common forms) are particularly deleterious because they frequently occur without any outward sign of damage, and because they often result in sudden and catastrophic failures. Thus, the development of effective corrosion damage prediction technologies is essential for the successful avoidance of unscheduled downtime and for the successful implementation of life extension strategies.
17
Plenary Lectures
Corrosion damage is currently extrapolated to future times using damage tolerance analysis (DTA). In this strategy, known damage is surveyed during each subsequent inspection, and the damage is extrapolated to the next inspection period allowing for a suitable safety margin. We have argued [1] that this strategy is inaccurate and inefficient, and that in many instances it is too conservative. Instead, we argue that damage function analysis (DFA) is a more effective method for predicting the progression of damage, particularly when combined with periodic inspection. DFA is based upon the deterministic prediction of the rates of nucleation and growth of damage, with particular emphasis on the compliance of the embedded models with natural laws. Although corrosion is generally complicated mechanistically, a high level of determinism has been achieved in various treatments of both general and localized corrosion. The application of DFA is illustrated by reference to the development of damage due to pitting corrosion of stainless steels in condensing heat exchangers. Deterministic models have been developed for both the nucleation and the growth of damage, and these models have allowed us to calculate the damage function as a function of exposure time and system conditions. FUNDAMENTAL CONCEPTS In this paper, I outline a deterministic method for predicting the damage function for pitting corrosion in condensing heat exchangers [1,2]. This method is considered to be potentially superior to empirical (including stochastic and probabilistic) techniques, because it is mechanistically-based and hence provides analytical relationships between the damage function (i.e., number of pits versus pit depth presented in the form of a histogram [1]) and the damaging variables (e.g., chloride concentration and combustion parameters). Accordingly, deterministic methods are expected to be more efficient at using databases, because a lesser need exists to establish the damage function/damaging variable relationships empirically. Any deterministic model must account for the fact that localized corrosion involves nucleation and growth phenomena which occur sequentially for a single site but that tend to occur in parallel for an ensemble of pits. Furthermore, the model must account for the experimental observation that the parameters that characterize the breakdown event are distributed, due to the fact that the population of sites on any real surface is not homogeneous. Outlined below is one model that satisfies these (and many other) conditions related to the nucleation and growth of damage resulting from localized corrosion. While the model may not be complete (or even correct), it is deterministic in that the distribution function and the relationships between the model parameters and the damage function are analytic and follow from the natural laws. In illustrating this technology, I have chosen to discuss the prediction of the damage function for pitting corrosion, because this form of attack is almost ubiquitous in condensing heat exchangers. Furthermore, pitting corrosion displays most of the features of all forms of localized attack, including an induction time and the autocatalytic development of the damage. The algorithm developed in this study to estimate the damage functions for condensing heat exchangers contains five modules as outlined in Fig. 1. Also indicated are the parameters that propagate from one module to the next. The output of the algorithm can be specified in three forms: 18
Macdonald
• For a specified probability of failure, the algorithm estimates the damage function as a function of exposure time and computes the number of pits with lengths exceeding the condenser wall thickness to predict the service life. • For a specified probability of failure and design life, the algorithm calculates the wall thickness to ensure acceptable performance. • For a specified wall thickness and design life, the algorithm calculates the failure probability. MODEL INPUTS
Duty Cycle
Chloride Concentration
Condensate Temperature
Flue Gas Composition
Condensate Chemistry Model pH.[Cl-]* Mixed Potential Model Ecorr.pH. [Cl-] Pit Nucleation Model N(t).Ecorr.pH. [Cl-] Pit Growth Model N(tobs) vs. n(u) Damage Function Model
Service Life
Figure 1.
Wall Thickness Specifier
Failure Probability
Structure of the algorithm for the prediction of damage function (*parameters propagated from one model to the next)
Below I describe the various modules in this algorithm; however, due to the limited space available, I outline only the principles of these modules. 19
Plenary Lectures
The Condensed Chemistry Module (CCM) The composition of the flue gas will differ from burner to burner. With this in mind, we developed a generalized condensate model for the condensate environment. This model assumes the flue gas to be a mixture of CO, CO2, H2S, NO, NO2, SO2, SO3, and H2O. The relative proportions of these components may vary widely from furnace to furnace, depending on the nature of the ambient air, the air/gas ratio, and the impurities of the gas. The goal of the condensed chemistry module (CCM) is to calculate the pH and the composition of the condensate on the condenser surface. The pH is a key parameter in controlling the rates of pit nucleation and pit growth. The concentrations of species in the liquid layer determine the ionic conductivity of the solution, which has great impact on the pit growth rate. The module employs an equilibrium model along with mass balance and charge balance constraints, and computes ion activity coefficients using the extended Debye-Huckel theory. It is assumed that the condensed liquid film is in equilibrium with the ambient environment, so that equilibrium calculations are applicable. The details of this module are described in the literature [3]. A typical gas-fired heat exchanger is schematically shown in Fig. 2a [4]. The temperature ranges from approximately 308oK in the cold end to 353oK in the hot end, depending on the design of the heat exchanger. Typical values of the pH and chloride concentration in these different zones are given in Fig. 2b [4]. It is shown that the condensed liquid phase is enriched in chloride to the extent of approximately 150 ppm in the hot end. Acidification of the condensed thin liquid layer also occurs, in that pH values as low as 2.7 and 3.3 are found at the hot end and the cold end, respectively. In Fig. 2c, the computed pH for a typical composition of the flue gas and the chloride content of the condensate are presented. The calculation shows a variation in pH from 2.93 to 3.32 from the hot end down to the cold end. Recognizing the wide range of operating conditions and designs of condensing heat exchangers, it is concluded that good agreement is observed between the experimental data and theoretical prediction.
Cooling Air Exhaust
Heat Exchanger Simulator Flue Gas Flow Zone 1 Zone 2
Zone 3
Zone 4
T = 353-326oK
Zone 5
T = 308-326oK
Cooling Air
Figure 2a. Schematic diagram of a typical heat exchanger in a gas-fired furnace [4]
20
Macdonald
Figure 2b. Characteristics of flue-gas condensate from different zones [4]
Figure 2c. Calculated pH in the condensate as a function of temperature and chloride concentration (PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm)
The Mixed Potential Module (MPM) The mixed potential module (MPM), which is based on the Wagner-Traud hypothesis [5] for free corrosion processes, was developed to calculate the corrosion potentials of alloys in corrosive environments. The theory outlined here is essentially identical to that developed by Macdonald et al. for calculating corrosion potentials for stainless steel components in the heat transport circuits of boiling water reactors (BWRs) [6,7]. The theory is based on the physical condition that charge must be conserved in the system.
21
Plenary Lectures
Because electrochemical reactions transfer charge across a metal/solution interface at a rate measured by the partial currents, charge conservation demands that Σ iR/O,j (E ) + icorr (E ) = 0 j=1
(1) where iR/O,j is the partial current density due to the j-th redox couple in the system, and icorr is the corrosion current density of the substrate. The currents are written as functions of the potential E to emphasize the fact that the partial currents depend on the potential drop across the metal/solution interface. Indeed, the solution to Eq. 1 provides the quantity that we seek (i.e., the corrosion potential). Note that in deriving Eq. 1, the surface of the alloy is assumed to be equally accessible to all reactions in the system.
Figure 3. Calculated and measured corrosion potentials as a function of temperature and oxygen partial pressure for alloy A129-4C (the solution composition for the experimental measurement: [HCl] = 200 ppm, [HF] = 40 ppm, [H2SO4] = 20 ppm, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm) Experimental polarization curves and kinetic data for the reduction of oxygen on Alloy Al29-4C were used as input to the MPM. The module then calculates the corrosion potential of the alloy under the service condition, as shown in Fig. 3. The calculation indicates that if the solution is saturated with oxygen, the corrosion potential is only weakly dependent upon case of an operating condensing heat exchanger, the corrosion potential varies from -290 mV to -460 mV as the temperature increases from 300o to 370oK. The Pit Nucleation Module (PNM) As a result of an intensive effort over the past decade to develop an understanding of the breakdown of passive films, we have derived theoretical distribution functions for passivity 22
Macdonald
breakdown that are in good agreement with experimental data [8-11]. We derived these distribution functions from our point defect model for the growth and breakdown of passive films, by assuming that breakdown occurs when a critical concentration of cation vacancies accumulates locally at the metal/passive film interface, such that decohesion occurs between the barrier layer and the metal substrate. Subsequently, localized dissolution and/or mechanical instability leads to rupture of the film [8]. We further assumed that the breakdown sites are normally distributed with respect to the diffusivity of cation vacancies within the film [9-11]. The full derivation of these distribution functions can be found in the literature [8-11]. From the point defect model [8], the breakdown voltage and the induction time for a single pit nucleation site on the surface are given by Vc =
4.606RT ⎡ J ⎤ 2.303RT log o m−χ / 2 − log(a x ) ⎣ J uˆ ⎦ χα F αF
and
(2)
−1
⎡ ⎛ χα FΔV ⎞ ⎤ tind = ξ ′⎢ exp − 1 +τ ⎣ ⎝ 2RT ⎠ ⎥⎦
(3)
respectively, where χ is the film stoichiometry ( MO χ / 2 ), α is the dependence of the potential drop across the film/solution interface on applied voltage, ax is the activity of halide ion in the solution, τ is a relaxation time (τ ~0), Δ V = V-Vc, V is the applied voltage (or the corrosion potential under open circuit conditions), (4)
o J = aˆ D
and N ˆa = χ K⎡ v ⎤ ⎣Ω⎦
1+
χ 2
[
exp − ΔGs /RT o
]
(5)
The parameters u and Jm are defined in the original derivation [8]. Assuming that the breakdown sites are normally distributed in terms of the cation vacancy diffusivity, dN = dD
1 2πσ D
[
exp −(D − D) / 2σ D 2
2
]
(6)
where D is the mean diffusivity and σ D is the standard deviation, the point defect model yields the following expressions for the distributions in the breakdown voltage and induction time [9-11]: 2 2 γˆD dN =− ⋅ e−(D− D) / 2 σ D (7) 2πσ D dVc and 23
Plenary Lectures
ˆ dN ⎡ −ξu χ / 2 ⎤ −(D −D )2 / 2 σ D2 e − γV = ⋅ χ /2 e 2 dt ind ⎢⎣ 2πσ D aˆ ⎥⎦ a x (t ind − τ )
where
γˆ = χFα/2RT
(8)
(9)
Eqs. 7 and 8 contain four important system parameters: D and σ D which describe the transport properties of cation vacancies in the passive film, and α and β which appear in the expression for the dependence of the potential drop across the film/solution interface on applied voltage and pH
φf/s = α V + βpH + φf/so
(10)
All four parameters can be determined by independent experiments: D from electrochemical impedance spectroscopy, σ D from passivity breakdown induction time measurements [8], and α and β from film growth data, much of which exists in the literature. Accordingly, Eqs. 7 and 8 represent analytical distribution functions for the nucleation of pitting attack, provided that the assumptions in the model hold. Our previous work [9-11] has demonstrated the quantitative nature of these expressions for representing experimental distribution in Vc and tind, for those cases where sufficient experimental data are available for analysis. The Pit Growth Module (PGM) The pit growth module (PGM) computes the growth rate of an individual pit. Figure 4 shows schematically a typical pit that develops on the surface of a metal in contact with a thin liquid layer. The module developed in this study calculates the pit growth rate for a pit of cylindrical geometry. The details of the theoretical approach can be found in the literature [12,13], except that we developed a transmission line analog of the external environment from which the distributions in the electrostatic potential and the current in the external environment, as a function of distance away from the pit, can be estimated. z
Gas
Thin electrolyte film
h a0
r
L
Metal
24
Macdonald
Fe + n 4 H 2 O − Fe
( OH ) (n 2 − n
Cr + n 5 H 2 O − Cr ( OH
) (n
Ni + n 6 H 2 O − Ni ( OH
) (n 2 − n
H2O − H
+
+ OH
4
)−
4
3−n
5
)−
6
)
5
6
+ n4H
+
+ 2e
−
+ n5H
+
+ 3e
−
+ n6H
+
+ 2e
−
−
−
Figure 4. Schematic diagram of a pit on the surface of the condensing heat exchanger The principle of the transmission line approach is shown in Fig. 5, which yields the following equations for the distributions in the electrostatic potential (φs ) and the current in the external environment d 2φs 1 dφ s ρ − φ =0 2 + dr r dr hZs s
(11)
d 2 I 1 dI ρ − I= 0 2 − dr r dr hZs
(12)
where ρ is the resistivity ( Ω⋅ cm ) of the solution, Zs is the specific impedance ( Ω⋅ cm 2 ) of the external surface, h is the electrolyte film thickness, and r is the radial distance from the C center of the cylindrical pit. Note that Zs is a function of distance [Zs(r) =- (φ s − φm )/iN , where i CN is the net cathodic current density]. The value of Zs(r) was determined iteratively when solving Eq. 11 by substituting for the net current density the following expression
e−(φ s −φ s )/b a − e(φ s −φ s )/b c i = +i φ 1 1 -(φ s −φ se )/b a 1 ⎛⎜⎝φ s −φ se⎞⎟⎠ /b c p ( s ) + e − e il,r i o il,f e
e
C N
(13)
where the first term on the right-hand side is the generalized Butler-Volmer equation for the oxygen electrode reduction 2H 2 O ⇔ O 2 + 4H+ + 4e−
(14)
and the second is the polarization current of the substrate, both of which are functions of the potential difference across the interface. The parameters φse , io, il,f, il,r, ba, and bc in Eq. 13 are the negative of the equilibrium potential for Reaction 14, io is the exchange current density, il,f and il,r are the limiting current densities for Reaction 14 in the forward and reverse directions, respectively, and ba and bc are the corresponding Tafel constants. Note that the signs of the exponents in the first term in Eq. 13 are opposite to those normally defined because we have written the current in terms of the electrostatic potential in the solution with respect to the metal. We used the finite difference method to solve Eqs. 11 and 12 for φs (r) and i c(r) , respectively.
25
Plenary Lectures
The distribution of the electrostatic potential within the pit is obtained by solving Laplace's equation, assuming that the environment within the pit confine is electrically neutral, ∇φ= 0 2
(15)
The solution to Laplace's equation (Eq. 15) yields the following expression, assuming that the potential variation in the radial direction is negligible compared to that in the longitudinal direction: Z
Electrolyte film
Flue gas (O2) r
r+dr
Crevice
Metal Figure b (a) Element of electrolyte film on the metal surface. I
Rdr
φS
I-dI
Z(r)/dr φm (b) Element of transmission line for calculating current and potential distributions radially from the crevice mouth.
Figure 5. Transmission line model for thin electrolyte film on the metal surface
φs (z) = (φ s0 − φ s−L ) +φ s0 z L
(16)
where φs−L is the electrochemical potential at the pit tip, L is the pit depth, and z is a negative quantity. We also apply the Butler-Volmer equation to the electrodissolution reaction occurring at the pit bottom to yield the electrochemical potential at the pit tip as [14] ⎛ i 00 A ⎞ ⎟ ⎝ I0 ⎠
φs−L = φ s00 + ba1n ⎜
(17)
where φs00 is the (negative of the) standard electrochemical potential for the dissolution of the metal, i 00 is the standard exchange current density, ba is the anodic Tafel constant for metal dissolution, and At is the effective active surface at the pit tip. The model outlined above is a variant of the Coupled Environment Fracture Model (CEFM) that we developed some time ago [15] for describing crack growth in stainless steel piping in nuclear power reactor heat
26
Macdonald
transport circuits. Thus, following our previous work [15], Eqs. 11, 12, 16, and 17 are solved for the unknowns φs (r) , i c(r) , and Io, such that charge conservation, expressed as I0 + ∫ i C dS= 0 N
(18)
s
is obeyed, where I0 is the (positive) current exiting the pit mouth, and dS is an increment of the external surface (dS = 2π rdr ). Because the cathodic current due to oxygen reduction predominates on the external surface, the second term on the left side of Eq. 18 is negative. Once I0 is known, then the pit growth rate is calculated using Faraday's law: dL M I0 = dt 2ρ m Z FA
(19)
where ρm is the density of the metal (g/cm3), M is the composition-averaged atomic weight of the alloy, and Z is the composition-weighted oxidation state of the metal dissolving at the pit tip. Finally, the pit length is calculated as a function of time using the recursive formula:
L(t) = L(t − 1) +
dL Δt dt
(20)
where L(t-1) is the depth of the pit calculated from the previous time (t-1), and Δt is the increment in time. The Damage Function Module (DFM) By combining Eqs. 8 and 20 for a fixed density of potential breakdown sites (No, number/cm2), it is possible to estimate the pitting damage function. Thus, if one observes the system at time tobs, then the number of pits that nucleate over the time increment Δt at tind is ΔN , as determined from Eq. 8. However, these pits will have grown to a depth L(t), as given by Eq. 20, at the time the system is examined. By moving the increment Δt from t = τ to t = t obs , the damage function is then generated in the form of the number of pits versus the depth of the pits. If this procedure is repeated for different observation times, a family of damage functions is generated that extends to greater depths with increasing tobs. By specifying the surface area of interest, it is possible to define the service life as the time taken for one or more pit to grow to the critical length, which in this case corresponds to the wall thickness of the condensing heat exchanger. The number of pits with lengths exceeding the critical dimension is simply calculated as L max
N L ≥L crit = S∑ N(L) ⋅ ΔL
(21)
L crit
where N(L) is the density of pits per unit surface area and per unit increment in pit length (number/cm3) in the damage function, S is the surface area of interest (cm2), L is the pit
27
Plenary Lectures
length, ΔL is the increment of the pit length in the damage function, and Lcrit is the critical dimension. The service life is simply the time at which N L ≥L crit = 1. DISCUSSION The procedure outlined above for estimating damage functions for localized corrosion is currently being developed to explore the impact of corrosion on condensing heat exchangers in domestic and industrial gas-fired furnaces. The practical problem lies in selecting the most cost-effective alloys for the condensing stages of heat exchangers, because of the highly competitive nature of the furnace manufacturing business. Consequently, little room exists for over-designing furnaces by employing highly-alloyed, costly materials to fabricate the condensing sections. Therefore, selection of materials with adequate pitting resistance, and of acceptable cost, is of prime concern to furnace manufacturers and users alike. It is evident, then, that the design and materials specifications for condensing heat exchangers would greatly benefit from the development of a deterministic method for predicting localized corrosion damage functions. This, in turn, could reduce the cost of the alloy by decreasing the required database, through the availability of deterministic relationships between the damage function and important environmental variables (including pH, [Cl-], and gas composition). In this study, I present predictions of the model in comparison with experimental damage functions measured on Type 304L stainless steel by G. Stickford, B. Hindin, and A.K. Agrawal of The Battelle Columbus Laboratories. For the experimental data, the damage functions are measured on condensing heat exchanger tubes after a given number of cycles, at the hot end (temperature ranging from 326o to 353oK) and at the cold end (temperature ranging from 308o to 326oK). Each cycle consists of 240 second with the burner on and 480 sesond with burner off, which represent the dry and wet conditions, respectively. It was shown in a previous study [4], that no significant difference exists in the damage functions between the hot end and the cold end; the damage functions are, therefore, plotted without distinction between the hot ends and cold ends. Based on this experimental finding, the model is constrained to the case where the surface is covered at all times by a condensing liquid phase (wet condition). However, I choose the appropriate temperature at the hot end to calculate the damage functions in order to avoid underestimating the damage. The experimental data reported by Battelle were measured at three levels of chloride concentration (3, 26, and 225 ppm) on a number of different candidate alloys. I present in this study only the damage functions for Type 304L stainless steel, as shown in Figs. 6, 7, and 8, as a function of chloride concentration. Not surprisingly, fewer pits were observed at the lower chloride concentration (3 ppm, Fig. 6). At higher chloride levels (26 and 225 ppm), the number of pits increased substantially (Figs. 7 and 8) and led to perforation of the wall in shorter time, thus reducing the service life. However, the experimental data show some inconsistencies, which are due to the fact that different tubes were used to determine the damage functions in each case. The chemical composition, the metallurgical history, and the surface state may vary from tube to tube. Because, the kinetics of the cathodic oxygen reaction on Alloy Al29-4C are considered to be essentially identical to those on Type 304L stainless steel, the parameters for Al29-4C were chosen for calculating the corrosion potential used in estimating the damage functions 28
Macdonald
(Table 1). Calculated damage functions are presented in Figs. 9 through 12 for chloride levels of 3, 10, 26 and 225 ppm, respectively. The calculations clearly indicate the progressive nature of the nucleation and growth of pits on the alloy surface. It is predicted that at lower chloride concentrations (3 ppm), fewer pits exist on the surface of the steel, while at higher chloride concentrations (26 to 225 ppm), the number of pits increases substantially, leading to the majority of the pits perforating the wall thickness in a short period of time. The predicted service life is presented as a function of chloride concentration in Fig. 13, in comparison with the experimental data. The calculations indicate that the service life of a condensing heat exchanger is highly sensitive to the chloride level in the condensate, especially at the lower chloride concentration (3 ppm). The principal effect of increasing chloride is to accelerate pit nucleation, so that, in the limits of very high chloride concentration, (~>100 ppm) in the condensate, the failure time is dominated by pit growth. Because the pit growth rate is dominated by the conductivity of the external environment (i.e., the condensate film), for any given pH and oxygen concentration, and because the conductivity is dominated by non-chloride species, the failure time becomes constant at sufficiently high chloride levels. This corresponds to the situation where the entire service life is determined by the time required for the pits that nucleate on initial exposure of the alloy to condensate and grow through the condenser wall. Noting that the service life for the case shown in Fig. 13 is calculated to decrease from 1.55 x 108 s (4.92 years) for a chloride concentration of 3 ppm to 2.34 x 107 s (0.74 year) for a chloride concentration of 225 ppm, it is evident that the time required for an active pit to perforate the wall is about three-quarters of a year, corresponding to an average pit growth rate of 0.7 mm/year. Clearly, then, the increase in the service life on lowering the chloride concentration is due almost entirely to an increase in the initiation time, and it would seem that substantial service lives for this alloy can only be obtained if nucleation becomes the dominant phase in the development of damage. Finally, due to the fact that different tubes are used in determining the damage functions, the experimental data are rather scattered. In recognition of this observation, relatively good agreement is claimed between the experimental data and the theoretical prediction. Table 1. Values for Parameters Used in Calculating Damage Functions Parameter
χ
Ω ΔG AO −1 φf/sO O Δ Gs τ ε α
β ξ
(Passive film stoichiometry) (Mole volume of passive film) (Gibbs energy of Cl- absorption) (Constant) (Gibbs energy of cation vacancy formation) (Relaxation time) (Electric field strength) (Constant) (Constant) (Critical areal concentration of vacancies)
Value 3 30 -60 -0.375 20 0 1 x 106 0.25 -0.001 1 x 1016
Units cm3/gm cation kJ/mol V kJ/mol s V/cm V No/cm2
29
Plenary Lectures
Jm D
σD
(Vacancy flux in metal phase) (Standard deviation in cation diffusivity) (Standard deviation in cation diffusivity)
0.12 x 107 vacancy 1.0 x 10-18 vacancy
Vacancies/cm2.s cm2/s
0.5 D
cm2/s
The influence of the oxygen partial pressure on the development of damage functions has been calculated, and is shown in Fig. 14. The calculations indicate that oxygen has a great impact on the service life of heat exchangers. This is because oxygen, in the condensed liquid phase on the external surface, consumes the positive current associated with the pit tip dissolution process, thereby driving the growth of the pit. By decreasing the partial pressure of oxygen from 10-4 atm to 10-8 atm, the service life of a heat exchanger having the characteristics assumed in this work could be extended from 3.02 x 107 s (1 year) to 1.05 x 109 s (approximately 30 years).
Figure 6. Measured pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 3 ppm, temperature ranging from 308o to 353oK, pit counting interval = 2.54 x 10-3 cm. a) Tobs = 4.32 x 106 s; b) Tobs = 8.63 x 106 s; c) Tobs = 1.73 x 107 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line
Figure 7. Measured pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 26 ppm, temperature ranging from 308o to 353oK, pit counting interval = 2.54 x 10-3 cm. a) Tobs = 4.32 x 106 s; b) Tobs = 8.63 x 106 s; c) Tobs = 1.73 x 107 s; d) Tobs = 3.46 x 107 s; e) Tobs = 5.18 x 107 s; tube thickness = 5.34 x 102 cm, as indicated by the dashed line
Figure 8. Measured pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 225 ppm, 30
Macdonald
temperature ranging from 308o to 353oK, pit counting interval = 2.54 x 10-3 cm. a) Tobs = 4.32 x 106 s; b) Tobs = 8.63 x 106 s; c) Tobs = 1.73 x 107 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line
Figure 9. Calculated pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 3 ppm, temperature = 350oK, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 104 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm, pit radius = 1.00 x 10-4 cm, initial pit depth = 1.00 x 10-3 cm, pit counting interval = 2.54 x 10-3 cm, a) Tobs = 6.30 x 107 s; b) Tobs = 1.58 x 108 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line
Figure 10. Calculated pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 10 ppm, temperature = 350oK, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm, pit radius = 1.00 x 10-4 cm, initial pit depth = 1.00 x 10-3 cm, pit counting interval = 2.54 x 10-3 cm, a) Tobs = 3.15 x 107 s; b) Tobs = 4.10 x 107 s; c) Tobs = 4.41 x 107 s d) Tobs = 5.36 x 107 s e) Tobs = 6.30 x 107 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line 31
Plenary Lectures
Figure 11. Calculated pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 26 ppm, temperature = 350oK, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm, pit radius = 1.00 x 10-4 cm, initial pit depth = 1.00 x 10-3 cm, pit counting interval = 2.54 x 10-3 cm, a) Tobs = 4.32 x 106 s; b) Tobs = 8.63 x 106 s; c) Tobs = 1.73 x 107 s d) Tobs = 2.52 x 107 s e) Tobs = 2.84 x 107 s; f) Tobs = 2.99 x 107 s; g) Tobs = 3.15 x 107 s; h) Tobs = 3.46 x 107 s; i) Tobs = 5.19 x 107 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line
Figure 12. Calculated pitting damage functions for Type 304L stainless steel heat exchanger tubes under condensing conditions: [Cl-] = 225 ppm, temperature = 350oK, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, 32
Macdonald
PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm, PO2 = 1.00 x 10-4 atm, pit radius = 1.00 x 10-4 cm, initial pit depth = 1.00 x 10-3 cm, pit counting interval = 2.54 x 10-3 cm, a) Tobs = 1.58 x 107 s; b) Tobs = 2.21 x 107 s; c) Tobs = 2.52 x 107 s d) Tobs = 3.15 x 107 s; tube thickness = 5.34 x 10-2 cm, as indicated by the dashed line
Figure 13. The measured and calculated service life for Type 304L stainless steel heat exchanger tubes as a function of the chloride concentration (parameters are identical to that for Figs. 6-8 for experimental data and Figs. 9-12 for calculation)
33
Plenary Lectures
Figure 14.
The calculated service life for Type 304L stainless steel heat exchanger tubes as a function of the partial pressure of oxygen (temperature = 350oK, PCO = 2.60 x 10-5 atm, PCO2 = 6.80 x 10-2 atm, PSO2 = 9.37 x 10-9 atm, PSO3 = 2.00 x 10-9 atm, PH2S = 1.20 x 10-4 atm, PNO = 1.74 x 10-8 atm, PNO2 = 2.67 x 10-8 atm)
SUMMARY AND CONCLUSIONS A deterministic model has been developed to predict the damage functions for condensing heat exchangers in gas-fired furnaces. The model incorporates calculations for the condensed chemistry environment, the electrochemical corrosion potential of the alloy, and mechanistic treatments of the nucleation and growth of pits. The model predicts that the chloride concentration in the condensed liquid layer has great impact on the service life of the condensing heat exchanger, particularly at low chloride concentrations. At high chloride concentrations, the service life of the condensing heat exchanger is predicted to be relatively independent of the chloride concentration, corresponding to the dominance of pit growth in determining the failure time. The service life for the condensing heat exchanger with Type 304L stainless steel tubes is predicted to decrease from 1.55 x 108 s (4.92 years) for a chloride concentration of 3 ppm to 2.34 x 107 s (0.742 year) for a chloride concentration of 225 ppm. The model predicts that the service life of the condensing heat exchanger also depends strongly on the oxygen content in the flue gas; by decreasing the oxygen partial pressure from 10-4 atm to 10-8 atm, the service life of the condensed heat exchanger can be extended from 3.02 x 107 s (1 year) to 1.05 x 109 s (approximately 30 years). Recognizing the scattered nature of the experimental data, I conclude that the algorithm developed in this work provides estimates of the service life that are in good agreement with the available experimental data, even though no a priori fit of the experimental data to the model was made. ACKNOWLEDGEMENTS The author gratefully acknowledges the support of this work by the Gas Research Institute (GRI) through Contract No. 5090-260-1969, and G. Stickford, B. Hindin, and A.K. Agrawal at The Battelle Columbus Laboratories for supporting the experimental damage functions used in this study. REFERENCES 1. D.D. Macdonald and M. Urquidi-Macdonald, "The Corrosion Damage Functions: Interface between Science and Engineering," 1992 Whitney Award Address, NACE, Nashville, Tenessee, submitted to Corrosion, 1992. 2. R. Razgaitis, J.H. Payer, S.G. Talbet, B. Hindin, E.L. White, D.W. Locklin, R.A. Cudnik, and G.H. Stickford, Condensing Heat Exchanger Systems for Residential/Commercial Furnaces and Boilers, Phase II, Battelle Report to DOE/BNL, BNL Report No. 51943, October, 1985.
34
Macdonald
3. D.D. Macdonald, M. Urquidi-Macdonald, S.D. Bhakta, N. Khalil, and H. Yashiro, Development of Analytical Methods for Predicting Damage Functions for Pitting Corrosion in Condensing Heat Exchangers, Final report to the Gas Research Institute, GRI No 5090-260-1969, January, 1992. 4. G.H. Stickford, B. Hindin, S.G. Talbert, A.K. Agrawal, M.J. Murphy, R. Razgaitis, J.H. Payer, R.A. Cudnik, and D.W. Locklin, Technology Development for CorrosionResistant Condensing Heat Exchanger, Final Report to the Gas Research Institute, GRI-85/0282NTIS PB86-172038, October, 1985. 5. C. Wagner and W. Traud, Z. Electrochem. 44, 1938, p. 391. 6. D.D. Macdonald, Corrosion 48, 1992, p. 194. 7. D.D. Macdonald, Proc. 5th Int. Symp. Environ. Degrad. Mat. Nucl. Power Systs: Water Reactors, Monterey, California, NACE, August, 1991. 8. L.F. Lin, C.Y. Chao, and D.D. Macdonald, J. Electrochem. Soc. 128, 1981, p. 1194. 9. D.D. Macdonald and M. Urquidi-Macdonald, Electrochim. Acta 31, 1986, p. 1079. 10. D.D. Macdonald and M. Urquidi-Macdonald, J. Electrochem. Soc. 134, 1987, p. 41. 11. D.D. Macdonald and M. Urquidi-Macdonald, J. Electrochem. Soc. 136, 1989, p. 961. 12. D.D. Macdonald, M. Urquidi-Macdonald, C. Liu, S. Bhakta, N. Khalil, and H. Yashiro, Proc. Int. Gas Res. Conf., Orlando, Florida, November, 1992. 13. D.D. Macdonald, M. Urquidi-Macdonald, and C. Liu, Paper No. 173, CORROSION 93, New Orleans, Louisiana, March, 1993. 14. D.D. Macdonald and M. Urquidi-Macdonald, Corros. Sci. 32, 1991, p. 51.
35
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
RELEVANCE OF LABORATORY CORROSION TESTS IN CORROSIVITY ASSESSMENT AND MATERIALS SELECTION: CASE STUDIES R.D. Kane CLI International, Inc. 14503 Bammel-N. Houston, Suite 300, Houston, TX USA
ABSTRACT Laboratory tests are a convenient means for simulating service environments for the purpose of evaluating both corrosiveness of the environment and material performance. Such tests can provide substantial information upon which engineering decisions can be made. These decisions generally will be made with greater confidence and a more efficient materials design. Measurements can be made under more controlled and reproducible conditions than are possible in field or plant situations. The consequences of process changes can be evaluated in advance of the actual situation. However, when conducting tests in simulated service environments, it is most important that the experimental methods be designed with the specific intent to provide meaningful, representative, correlative results. This presentation will review the various aspects that must be included when designing, conducting, and interpreting corrosion tests in simulated service environments. Several case studies will also be presented that involve petroleum applications, including the selection of corrosion-resistant alloys in sour petroleum production, inhibition of multiphase flow lines, assessment of crude oil corrosiveness, and corrosion under insulation. Key Words:
Laboratory testing, correlation, oil and gas, multiphase production, corrosion under insulation, electrochemistry
INTRODUCTION There are many types of laboratory corrosion tests utilized for various purposes. These tests can range from simple glassware immersion tests involving freely corroding, nonstressed coupons to highly sophisticated exposure tests involving dynamic replenishment, heat and/or mass transfer and high pressures maintained only through autoclave, flow loop or pilot plant operations. Furthermore, they may also involve ancillary techniques which include a variety of DC or AC electrochemical methods and mechanical loading configurations. To extend the predictive capabilities of the tests results, modeling techniques and statistical experimental design programs can be employed that allow for identification or verification of mechanisms or for establishment of linkages between the results of the laboratory test and the intended service application while minimizing the number of tests required. LABORATORY TESTING: A BASIC AND IMPORTANT TOOL In many areas of technical activity, emphasis is now being placed on reducing the overall time, funding and effort allocated for testing, research, development and engineering. In 37
Plenary Lectures
some cases, corrosion testing appears as only an afterthought. At the same time, there are pressures to provide systems with higher reliability and reduced operating costs. In simpler words, these engineering concepts can be referred to in four basic terms: Better, Cheaper, Faster, Safer [1]. One of the most important aspects that must be realized is that these are not necessarily mutually inclusive concepts. Usually, substantial scientific and technological developments are required before these terms can come together in a complex engineering system. Laboratory testing is one of the basic tools available to investigate complex interactions of variables that exist in real service applications. Data can be developed which both have applied engineering significance and provide insight into fundamental relationships in engineered systems that can in turn be used as stepping stones to achieve quantum leaps in efficiency, reliability and safety [2]. Furthermore, when compared to the costs generally associated with corrosion (which can be around 10% of revenues [3]), and those saved through application of corrosion testing, there can often be a cost reduction of between one and two orders of magnitude and, in some cases, more. NEED FOR SIMULATED-ENVIRONMENT TESTS The need for simulation varies greatly depending on the purpose of the test. For example, corrosion tests are usually conducted for three reasons [4]: screening-comparison of the response of two or more materials or material conditions relative to a particular form of corrosion (e.g., general corrosion, local pitting or crevice corrosion, and stress corrosion cracking); qualification-verification that the material has a required conformance to composition and that the metallurgical or fabrication processes have resulted in a microstructure that will provide adequate corrosion performance; evaluation-assessment of the influence of process changes (e.g., temperature, additives, inhibitors, and product purity). In many cases, relatively simple, standardized tests found in NACE, ASTM, and ISO documents can be conducted that are useful for these purposes. These are relatively simple environments often involving combinations of acids and salts that can be handled in standard glassware. [5-7]. However, a limitation common to many standardized tests is the difficulty in obtaining a direct correlation between the performance in such tests and actual in-service performance. To expand the applicability of standardized tests, substantial development work is often required which establishes such correlations. This work usually includes one or more of the following: review of service or failure records on a range of material conditions or over a range of process variables, analysis of field or in-plant tests in actual service environments, and use of laboratory exposure tests conducted under simulated service conditions. In many situations, the laboratory provides the most convenient avenue since it yields firsthand data that allows the engineers to make the fine distinctions in performance required to select the most cost-effective corrosion mitigation or control methods, and thereby gives the maximum cost benefit. THE REASONABLE WORST-CASE SCENARIO In the study of corrosion in simulated environments, the concept of a reasonable worstcase scenario has been a guiding light for those involved. Being reasonable includes two aspects: reasonable levels of corrosive severity and reasonable involvement of expense and time. It is often necessary to refine the set of exposure conditions by a process of 38
Kane
prioritization so that only the most important variables related to the operable corrosion mechanisms are taken into account and thereby minimize testing costs. Keeping these two aspects in mind, one must first define and preserve the active in-service mechanism(s) of corrosion. Once this has been completed, one must identify the aggravating and mitigating factors that combine in the actual service condition to determine the severity of corrosion. SIMULATED ENVIRONMENT TEST PROCEDURES Several case studies are presented herein which highlight a few of the basic, yet important, concepts which provide the link between the laboratory and field conditions. Case Study No. 1: Simulation of Conditions of Deaeration or Aeration One of the most important and universally applicable situations that must be evaluated when conducting laboratory tests in simulated service environments is the need to produce a reasonable representation of the level of deaeration or aeration found in the actual environment. The main reason for the importance of this effect is that corrosivity, in many service applications, can change dramatically with changes in oxygen content. Aeration accelerates anodic corrosion processes with a concomitant increase in localized corrosion activity (i.e., pitting, crevice attack and stress corrosion cracking). Examples of oxygen effects can be seen in applications such as seawater injection, the use of heavy brine completion fluids in oilfield operations and desalination. As shown in Fig. 1, the corrosion rate of steel increases by an order of magnitude going from 10 ppb to just 100 ppb [8]. It only takes very low levels of oxygen contamination (about 1% of normal atmospheric saturation levels) to greatly accelerate corrosion. Furthermore, due to the sensitivity of corrosion reactions even in low levels of aeration, oxygen contamination can produce excursions to higher corrosion rates that have prolonged effects [9]. The increase in localized anodic attack produced by aeration can be illustrated by its interaction with other species such as chlorides and sulfides. An example, in terms of susceptibility to stress corrosion cracking (SCC), is the interaction between dissolved oxygen and chlorides in elevated temperature applications involving alkaline phosphate treated boiler water [10]. As the availability of oxygen increases above the 0.1 ppm (100 ppb) level, the tolerance for chloride is reduced resulting in a dramatic increase in susceptibility to SCC.
Figure 1. Corrosion rate vs temperature for various oxygen levels
39
Plenary Lectures
The situation in aqueous sulfide-containing environments is even more complicated since oxygen can result in the formation of elemental sulfur, acids, and in some cases, polysulfide species. These can synergistically interact with the oxygen effects mentioned previously to produce quite severe limitations on the corrosion and SCC performance of even very highly alloyed materials. Such conditions can be found in applications which involve pumping of sour oilfield brines, injection of wastewater and flue gas desulfurization. A comparison of the minimum required pitting resistance equivalent (PRE) for conditions involving a simulated sour oilfield service can be seen for identical situations (0.7 kPa H2S, 138kPa CO2, 2 meq/L HCO3, 30,000 ppm Cl-, 65oC) except one is aerated and one deaerated [11]. For the deaerated condition, the minimum pitting resistance equivalent [PRE = Cr + 3.3Mo + 11N + 1.5(W + Cb)] is only 12. This would indicate successful use of materials with > 12 Cr. However, this same environment under aerated conditions yields a minimum PRE value of 30. Under evaporative conditions, this can increase still further. Understanding the conditions of aeration in the service application is necessary to reproduce similar conditions in the laboratory corrosion test. For example, most geochemical systems naturally contain less than 10 ppb oxygen. By comparison, mechanical deaeration techniques usually will not go below 100 ppm. Multiple vacuum, ultra-low oxygen inert gas purge cycles and prolonged gas purges are usually required to get below 50 ppb oxygen. In some cases, oxygen scavengers must be used to obtain complete deaeration. However, these must be used carefully because they may, in some cases, add other chemical species into the environment that can complicate electrochemical measurements. Case Study No. 2: Simulation of Corrosion in Multiphase Environments There are many factors that need to be considered when conducting corrosion assessments in multiphase environments. These include important factors related to the dynamic or flowing nature of the fluids which determine the mode of flow [12] and the kinetic shear forces that are imparted by the flowing fluids on the pipe wall. There have been several major studies involving very sophisticated simulations of three-phase flow. These studies are particularly capital intensive and costly since major investments must be made in the handling, pumping and disposal facilities required for such tests. However, there are no real alternatives for investigating questions involving the direct effects of flow regime such as measuring the shear forces developed by particular flow regimes and operating conditions and the movement of inhibitors [13]. On a more practical basis, more simple yet reasonable approximations of multiphase flow conditions can be obtained using pseudo-three phase systems such as the flow loop shown in Fig. 2 [14]. These systems provide for the establishment of three phase conditions (gas/oil/water) in a reservoir autoclave. Under these conditions, the primary corrodents are dissolved gases (e.g., CO2, H2S and sometimes O2) and an aqueous brine or condensed water phase. Facilities for replenishment of both the gas and liquid phases must be considered depending on the exact nature of the environment. Simulation of the affects of a flowing environment is usually based on modeling the shear stress produced in service on the metal surface by the flowing liquid containing the dissolved gases using the equations given in Table 1 [15]. The main assumption utilized in this approximation is that the major contribution to the wall shear stress is usually made by the liquid phase. In most cases, this
40
Kane
technique is valid since the contribution of liquid phase density and viscosity on the resultant shear stress predominates over that of the gas phase.
Figure 2. MAPS™ - Multiphase Autoclave Pipeline Simulator Of significance in most flowing multiphase systems is the handling of slug flow which is the predominate flow regime for horizontal and near horizontal flow applications. The main attribute of slug flow is the very high shear stresses and accompanying high turbulence in the region of the flow just ahead of the moving slug [16]. This effect results in levels of shear stress much greater than those produced by the bulk fluid. It has been proposed that this is the location where excessive corrosion is generated as a result of the effect of locally high shear stress and turbulence on both corrosion and inhibitor films. Investigations have recently focused on techniques such as flow loops and jet impingement to reproduce accurate simulations of such highly turbulent conditions for assessment of corrosion resistance and inhibitor performance [17]. Another major effect that must be addressed in multiphase systems is the potential role of the oil phase as a possible mitigation factor in terms of reducing the corrosion rate [18]. The properties of the hydrocarbon/liquid phase significantly influence the severity of the environment with respect to weight loss corrosion (see Fig. 3). In a typical case, oil/water mixtures remain relatively non-corrosive under flowing conditions of up to about 30% water cut resulting mostly from the preferential wetting and persistence of the oil phase on the metal surface. However, depending on the nature of the liquid hydrocarbon phase, some cases become corrosive with very low water cuts (0.1 ppm) increase in the Cu2+ content of the distillate. This occurs mainly in the high temperature cells of the distiller. The general belief [76] is that the vapor is largely contaminated with CO2 resulting from the high temperature decomposition of the HCO -3 ion of the brine [95]. The formed carbonic acid, H2CO3, is assumed to be sufficiently strong at high temperatures to attack the copper tubes. Apparently this explanation is an over simplification of a more complex process. At the temperatures of the first heat gain cells, the solubility of CO2 in the vapor is both limited and transient. However, neither copper nor its alloys displace hydrogen from acid solutions. If H2CO3 is to exert a corroding action it must do so by dissolving the cuprous and cupric oxides formed on the metal surface. Practical experience confirms this conclusion. Condenser tubes undergoing overhaul acquire black coloration as a result of air oxidation. When returning to service, the distillate exhibits above-normal copper-content and has to be dumped back into
59
Plenary Lectures
the sea. Considerable time (up to 60 hours [75]) elapses before the copper concentration in the distillate returns to permissable limits. For copper tubes to continue corroding, their surface has to reoxidize. This occurs through the oxygen (air) dissolved in the flashing brine. As this is limited, the rate of corrosion falls to low values. Higher corrosion rates point to the presence of tiny leaks in cell gaskets. A vacuum leak test (or pressure leak test) on the distiller will confirm this assumption. Vapor-side corrosion of copper condenser tubes can be serious if seawater is polluted by ammonia or hydrogen sulphide. What has been said regarding water-side corrosion applies to vapor-side corrosion too. OPERATION PROCESSES AFFECTING CORROSION IN MULTI STAGE FLASH (MSF) DISTILLERS In running an MSF plant, certain processes are carried out the aim of which is to overcome an operational difficulty and/or to improve production efficiency. These processes affect the corrosion of distiller components in variety of ways. In some instances, the role of a certain parameter can be beneficial at one level and detrimental at another. The successful operation of the distiller depends, therefore, on a proper understanding and careful execution of the process. The following operations deserve special mention: Chlorination of Seawater Chlorination of seawater at the intake is carried out to discourage marine organisms from entering the distiller and to prevent bifouling. There is no single chlorination procedure that applies to all units. The adoption of a certain course of action evolves from trial and error and depends on factors such as geographic location, bioactivity, temperature, and water purity. For example, in Umm Al Nar (Abu Dhabi) chlorination is carried out continuously to a level of 0.4 ppm. This ensures a residual chlorine content of about 0.25 ppm in the condenser tubes. To prevent organisms from building up resistance, shock dosing at 1 ppm is achieved for 1 hour. once a week [78]. Chlorine reacts with water to produce hydrochloric and hypochlorous acids : Cl2 + H2O = HCl + HClO. Both acids are directly neutralized by seawater alkalinity. The hypochlorous ion, ClO-, is a strong oxidizing agent which raises the free corrosion potentials of alloys to higher values. This raises the susceptibility of normal stainless steels to pittings- and crevice corrosion, particularly in regions of stagnation [79]. Similarly, chlorination of seawater interferes with the process of ferrous sulphate dosing. Chlorination is stopped before, during and for some time after dosing to allow the iron-oxide film the chance to build up properly. On the other hand, chlorine injection destroys pollution by hydrogen sulphide, and neutralizes, to some extent, ammonia polluted seawater. Ferrous Sulphate Dosing Copper-base condenser tubes have little resistance to erosion-corrosion and to corrosion by hydrogen sulphide-polluted seawater. Both types of attack are greatly inhibited by ferrous sulphate dosing [80]. The ferrous ion readily hydrolyzes and oxidizes to colloidal FeOOH, which adheres strongly to tube surface. The iron oxide film improves the naturally formed copper oxide and offers better protection. Two to three hours before sulphate dosing, chlorination of the seawater is stopped. The sulphate is injected at a rate of 3 ppm for one hour per day for one month at tube inlets. Chlorination can be resumed after two hours. During the second and third months, ferrous sulphate dosing is reduced to 2 and 1 ppm for 60
Shams El Din
one hour per day, successively. During the whole period of treatment, sponge ball treatment is reduced to one or two times per week for one hour. The success of sulphate dosing is recognized through the uniform coloration of the tube plates and tubes with a light- to deepbrown film, which does not peel off when scratched with a nail. A number of factors interfere with film formation. These involve waters flowing at high rates or carrying large amounts of sand, slime or deposits; excessive or continuous ball cleaning; chlorination directly before, during or after sulphate dosing and/or the presence of entrapped, polluted dead waters. Scaling in Condenser Tubes Scale formation in the condenser tubes is one of the major problems encountered during the operation of MSF distillers. Scaling impairs heat transfer, causes tube blockage and can induce corrosion. Scale formation is an inherent product of the composition of seawater, which contains HCO -3 , Ca2+ and Mg2+ ions. The thermal decomposition of the HCO -3 leads to the deposition of calcium carbonate and magnesium hydroxide. Two alternative techniques are employed to prevent (retard) scale formation. The first involves the controlled acidification of the makeup water to convert the HCO -3 ion into CO2. Either sulphuric or hydrochloric acids is used; the first is preferred on the basis of cost considerations. An addition of 100-200 ppm sulphuric acid is normally used, and the evolved CO2 is stripped off either in a separate degassing tower or in the MSF vacuum deaerator. Unless very carefully carried out, acidification can lead to serious corrosion of the condenser tubes and the deaerator system. To overcome this problem under-acidification has been proposed [81]. Laboratory experiments have shown that 80% of the acid required to completely neutralize the bicarbonate ions is sufficient to prevent scale formation for a long time [82]. Another technique to prevent corrosion involves the neutralization of excess acid. Following degassing, the brine pH is brought back to ~7.5 through the controlled addition of caustic soda. This technique is elaborate and expensive [83]. The second approach for scale prevention is the use of antiscaling agents. These are polymeric, surface-active substances which adsorb on active centers of CaCO3 and Mg(OH)2 crystallites. This inhibits them from forming a continuous layer. A small amount of the inhibitor (2-5 ppm) is needed to retard scale formation, and the method is known as a threshold treatment [84]. A large variety of antiscalants is available on the market, and the nature of the scale depends on the compound used. Polyphosphates, for example, give rise to dense, fluffy, grayish-brown deposits [85]. Maleic anhydride polymers, on the other hand, yield thin, hard scales [86]. A hybrid technique for scale control has been suggested [87]. It involves the use of lessthan-stiochiometric quantities of a mineral acid together with small quantities of a threshold antiscalant. Under-acidification eliminates the problem of acid corrosion of tubes, while the removal of the largest part of the water’s alkalinity allows lower concentrations of the antiscalant to be used. This represents a sizable reduction in cost. Sponge Ball Cleaning The retardation of scale formation in the condenser tubes through dosing of antiscaling agents is usually coupled with the technique of sponge ball cycling (Tapproge ball cleaning). The rubber balls are slightly larger than the interior diameter of the tubes and are forced 61
Plenary Lectures
through by the pressure of circulating water. As they travel through the tubes, the balls wipe out the tube’s insides and prevent scale crystallites from building a solid layer. As the balls pass randomly through the tubes, they are cycled for enough time to ensure the cleaning of the maximum number of tubes. There is no standard procedure for ball cycling. This varies between one or more cycles per day and continuous cycling. The duration of a cycle also differs from one plant to another. On the market are sponge balls with different surface hardness. The choice of the appropriate type depends on the seawater purity, type and concentration of the additive used, nature of the deposit and distiller top temperature. A trial to optimize sponge ball cleaning was recently published [88]. Sponge balls can affect tube corrosion in many ways. Soft balls and/or short treatment times might be ineffective and allow sand, silt and scale deposition which lead to crevice attack. The same result might be noted when the number of balls is insufficient. On the other hand, balls that are too hard can induce erosion-corrosion by stripping the protective film from the metal surface. Also, it is not uncommon for balls to get stuck inside the tubes, promoting crevice attack. Finally, as mentioned above, ball cleaning damages the fresh iron hydroxide film resulting from ferrous sulphate dosing.
Acid Cleaning Antiscaling agents do not inhibit completely the formation of alkaline scales; they only retard their growth. Even when their use is coupled with sponge ball cleaning, a scale film continues to grow inside the condenser tubes. Due to their bad thermal conductivity, a situation is eventually reached when the gained output ratio of the distiller drops below a preset value. The operation of the distiller becomes impractical (noneconomical) and an acidwash of the distiller is necessary. The washing process is carried out by circulating warm (~65°C) fresh seawater through the water boxes and condenser tubes of the cells and the brine heater. Enough inhibited acid is added to the water to bring its pH value to 1.8-2.0. For copper-base condensers either hydrochloric or sulphuric acids can be used. The pH of the water is monitored, and its value rises as result of reaction with the scale. Extra acid is added to bring the pH to its lower value and the process is repeated until no further increase in pH is recorded for a long time (usually two hours). The acid water with the accumulated sludge is discharged, and the distiller is flushed clean with fresh seawater. The use of an improper corrosion inhibitor [89] or insufficient quantities of a suitable one leads to general attack on the tube material. The same washing procedure described above applies to condensers with titanium tubes as well. However, because of the high tendency of titanium to absorb hydrogen, weak organic acids, e.g., citric [90] or sulfamic [91] acids are used instead. A few organic compounds marketed under trade names, e.g., Galvane® (ICI) and IBIT [91], are said to retard acid attack on titanium. Distiller Outage As is clear by now, seawater is an extremely aggressive medium which attacks all of the metallic components of an MSF unit. The fact that distillers operate for long times with little damage is due to two main reasons: 62
Shams El Din
• the exploitation of certain material properties manifested during operation; • the application of protective and preventive measures during operation. To the first reason belongs:
• The ability of stainless steels to withstand the corrosive action of flowing seawater, while being quickly and severely attacked by the same water under stagnation, and • The readiness by which copper-base alloys develop protective surface films which offer some resistance to the action of flowing water. Preventive and protective measures that can taken during operation are mentioned in detail in previous sections. These involve degassing to remove oxygen (depolarizer) and carbon dioxide (acidity) from the brine, ferrous sulphate dosing to increase the protection of copper tubes, chlorination to discourage marine organisms from gaining access to the inside of the distiller, use of antiscalants and ball cleaning to minimize scale formation and the application of cathodic protection to prevent tube failure. Accordingly, as long as these measures are observed and the distiller is operating, the corrosion of its components is largely under control. This well balanced system of protection is lost once the distiller is shut down. Upon opening the water boxes and the flash chambers, air (oxygen) fills the entire unit, and accelerates ongoing corrosion. On the other hand, brine stagnation inside condenser tubes causes the settling of sand and silt, and occasionally a few sponge balls. This initiates crevice corrosion. Stagnation also promotes the pitting corrosion of stainless steel components. Unless the shutdown is for a short time (i.e., a maximum of two days) precautions against attack should be applied. These involve the draining of the stagnant brine from the distiller, followed by a thorough flushing with potable or distilled water. The washings should likewise be drained out. During long outages, all inlets and outlets should be left open to remove humidity and speed up the drying of the distiller’s inside. The various components of the unit should be examined and the necessary corrective measures taken. REFERENCES 1. T.H. Rogers, Marine Corrosion, New York: John Wiley, 1968. 2. F.L. LaQue, Marine Corrosion, New York: John Wiley, 1975. 3. D. Pecker and I.M. Bernstein, Handbook of Stainless Steels, New York: McGrawHill, chapter 14, 1977. 4. Inco, Copper-Nickel and Other Alloys for Desalination Plant, London: INCO Europe Ltd., 1981. 5. European Federation of Corrosion Publications, No. 3, General Guidelines for Corrosion Testing of Materials for Marine Applications, London: The Institute of Metals, 1989. 6. European Federation of Corrosion Publications, No. 10, Marine Corrosion of Stainless Steels, London: The Institute of Materials, 1993. 7. Nickel Development Institute, Guidelines for Selection of Nickel Stainless Steels for Marine Environments, Natural Waters and Brines, Toronto: NIDI, vol. 2, 1987. 8. Nickel Development Institute, Materials for Saline Water, Desalination and Oilfield Brine Pumps, NIDI, 1988.
63
Plenary Lectures
9. B. Todd, Proceedings 25th Annual Conference of Metallurgists, Toronto, 1986. 10. J.V. Dawson and B. Todd, BCIRA J., 1987, p. 1. 11. A.M. Shams El Din, Corrosion Resistant Materials for Desalination Plants, Abu Dhabi, Internal Report to WED, 1991. 12. A.M. Shams El Din, Desalination 93, 1993, p. 499. 13. T. Hodgkiess, Desalination 93, 1993, p. 445. 14. J.A. Carew, M. Abdel-Jawad and Y. Al-Wazzah, Desalination 95, 1994, p. 53. 15. Titanium Engineering Alloys, Product Promotion Brochure, 1988. 16. F.L. LaQue and W.C. Stewart, Corrosion 8, 1952, p. 259. 17. J.M. Popplewell, R.J. Hart and J.A. Ford, Corrosion science 13, 1973, p. 295. 18. R.F. North and M.J. Pryor, Corrosion science 8, 1968, p. 149; 1969, 9, p. 509. 19. S. Kido and T. Shinohara, Desalination 22, 1977, p. 369. 20. T.G. Temperley, Desalination 31, 1979, p. 353. 21. K. Hill, Desalination 34, 1980, p. 325. 22. K.R. Fröhner, Desalination 21, 1977, p. 147. 23. B. Wallen and T. Andersson, ACOM 2, 1987. 24. P.T. Gilbert, Materials performance 21, 2, 1982, p. 47. 25. G. Odone, A. DeMaio and F. Fioravanti, J. IDA 1, 2, 1985. 26. C. Lakshmipati, Corrosion Maintenance, 1984, p. 305. 27. M.W. Joseph, F.W. Hammond and T.S. Lee, Corrosion 86, Houston, 1986, Paper No. 226. 28. R. Cigna, A. DeMaio, L. Giuliani and G. Gusmano, Desalination 38, 1981, p. 269. 29. G.A. Gehring, C.K. Kuester and J.R. Maurer, Corrosion 80, Houston, 1980, Paper No. 32. 30. T. Atsumi, A. Ogiso, K. Nagata and S. Sato, 10th Intern. Cong. Met. Corros., Madras, India, 1987. 31. T. Atsumi, A. Ogiso, K. Nagata and S. Sato, Sumito Light Metal Technical Reports 29, 4,1988, p. 257. 32. S. Sato, K. Nagata and S. Yamauchi, Corrosion 81, Ontario, Canada, 1981, Paper No. 195. 33. G.L. Bailey, Journal Institute of Metals 79, 5, 1951p. 243. 34. K.D. Efird, Corrosion 33, 1, 1977, p. 3. 35. D.B. Anderson and F.A. Badia, ASME J. Engineering for Power, 1973. 36. D.B. Anderson, Corrosion 81, NACE, Toronto, 1981, Paper No. 197. 37. S. Sato and K. Nagata, Somitomo Light Metal Technical Reports 19, 1978, p. 83. 38. V.K. Gouda and W.T. Riad, 9th Europ. Cong. Corrosion, Utrecht, Holland, 1989, p. P1-197. 39. A.M. Shams El Din, Wsia J. 11, 2, 1984, p. 1. 40. A.M. Shams El Din, WED, Abu Dhabi, unpublished results. 41. W.E. Heaton, Br. Corros. J. 13, 1978, p. 57. 42. T. Sydberger and U. Lotz, J. Electrochem. Soc. 129, 1982, p. 276. 43. J.M. Popplewell and E.A. Thiele, Corrosion 80, Houston, 1980, Paper No. 30. 44. J.A. Ellor and G.A. Gehring, Corrosion 86, Houston, 1986, Paper No. 225. 45. J.N. Al-hajji and M.R. Reda, Corrosion science 34, 1993, p. 163. 46. A.M. Beccaria, G. Poggi, P. Traverso and M. Ghiazza, Corrosion science 32, 1991, p. 1263. 64
Shams El Din
47. K. Habib, Desalination 89, 1992, p. 41. 48. B.C. Syrett, Corrosion 80, Houston, 1980, Paper No. 33. 49. G.A. Gehring, R.L. Foster and B.C. Syrett, Corrosion 83, Anaheim, 1983, Paper No. 76. 50. Trent Tube, Stainless Steel Pipe and Tube Alloy Handbook, Product Promotion Brochure, 1992. 51. K.R. Fröhner, Desalination 21, 1977, p. 147. 52. A.J. Sedriks, Corrosion 45, NACE, 1989, p. 510. 53. T. Hodgkiess and A. Asimakopoulos, Desalination 38, 1981, p. 247. 54. T. Hodgkiess, W.T. Hanbury and M.N. Hejazian, Desalination 44, 1983, p. 223. 55. H.P. Hack, Materials performance 22, 1983, p. 24. 56. T.S. Lee, R.M. Kain and J.W. Oldfield, Materials performance 23, 1984, p. 9. 57. T. Rogne and J.M. Drugli, Corrosion 86, NACE, 1986, Paper No. 230. 58. H. Al Zahrani, S. Somuah and N.M.A. Eid, 12th Inter. Symp. Desalination and Water Reuse, Malta, 1991, vol. 4, p. 349. 59. B. Wallen and T. Andersson, ACOM 2, 1987, p. 1. 60. G.A. Gehring, C.K. Kuester and J.R. Maurer, Corrosion 80, NACE, 1980, Paper No. 32. 61. G.A. Gehring and J.R. Maurer, Corrosion 81, NACE, 1981. 62. G.A. Gehring and R.J. Kyle, Corrosion 82, NACE, 1982, Paper No. 60. 63. J.A.S. Green, B.W. Gamson and W.F. Westerbaan, Desalination 22, 1977, p. 359. 64. A.R. Morris, Desalination 31, 1979, p. 387. 65. S. Kido and T. Shinohara, Desalination 22, 1977, p. 369. 66. T. Fukuzuka, K. Shimogori, H. Satoh and F. Kamikubo, Desalination 31, 1979, p. 389. 67. K. Shimogori, H. Satoh, F. Kamikubo and T. Fukuzuka, Desalination 22, 1977, p. 403. 68. P.D. Simon, Corrosion 83, NACE, 1983, p. 60. 69. J.P. Fulford, R.W. Schutz and R.C. Lisenbey, Joint ASME/IEEE Power Generation Conference, Miami Beach, Florida, 1987, Paper No. 78 - JPGC-Pwr-F. 70. J.I. Lee, P. Chung and C.H. Tsai, Corrosion 86, NACE, 1986, p. 259. 71. T. Moroishi and H. Miyuki, Titanium 80, vol. 4, 4th Intern. Conf. on Titanium, Kyoto, 1980, p. 2713. 72. K. Kohsaka, K. Kitaoka, Y. Masuyama, M. Oshiyama, M. Yamamoto and K. Kashida, Seawater Desalination Group, Development Committee of Japan Titanium Group, Product Promotion Brochure, May 1984, and April, 1986. 73. A.M. Shams El Din, T.M.H. Saber and A.M. Taj El Din, Paper presented before the IDA Congress on Desalination and Water Sciences, Abu Dhabi, 1995. 74. R. Heaton and T.A. Douglas, Desalination 41, 1982, p. 71. 75. E.A. Al-Sum, Sh. A. Aziz, A. Al-Radif, M.S. Said and O. Heikal, Proc. IDA and WRPC World Conf. Desal. and Water Reuse, Yokohama, 1993, vol. I, p. 501. 76. A.H. Khan, Desalination Processes and Multi-Stage Flash Distillation Practice, Elsevier, 1986, p. 441. 77. A.M. Shams El Din nd R.A. Mohammed, Desalination 99, 1994, p. 73. 78. A.M. Shams El Din, B. Makkawi and Sh.A. Aziz, Desalination 97, 1994, p. 373. 79. B. Wallen, ACOM 4, 1989, 1990. 65
Plenary Lectures
80. T.W. Bostwick, Corrosion 17, 1961, p. 12. 81. Office of Saline Water, Res. Develop. Report, 1968, Vol. 559. 82. J.W. McCutchan, UCLA, Dept. Eng. Rept., 1967, No. 67-1. 83. M.N. Elliot, Desalination 6, 1969, p. 87. 84. K.S. Spiegler and A.D.K Laird, Principles of Desalination, 2nd ed., Part B., New York: Academic Press, 1980, p. 672. 85. A.M. Shams El Din, Desalination 61, 1987, p. 89. 86. A.M. Shams El Din, Desalination 69, 1988, p. 147. 87. F. Butt, F. Rahman, A. Al-Abdallah, H. Al-Zahrani, A. Maadhah and M. Amin, Desalination 54, 1985, p. 307. 88. F. Al-Bakeri, F. and H. El Hares, Desalination 94, 1993, p. 133. 89. T.M.H. Saber, A.M. Tag El Din and A.M. Shams El Din, Br. Corros. J. 27, 1992, p. 139. 90. A.M. Shams El Din, H.A. El Shayeb and F.M. Abd El Wahab, J. Electroanalyt. Chem. 214, 1986, p. 567. 91. Japan Titanium Society, Product Promotion Brochure, September 1991.
66
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CORRECT MATERIALS SELECTION FOR DESALINATION: THE KEY TO PLANT RELIABILITY J.W. Oldfield Cortest International 23 Shepherd Street Sheffield, S3 7BA, UK
ABSTRACT Large scale desalination really began with the multi-stage flash (MSF) plant built by Weir Westgarth in Kuwait in 1960. This and similar plants elsewhere used mainly carbon steel and brass alloys in their construction. Corrosion problems with these early plants led to costly shutdowns and maintenance. As knowledge of the requirements of materials has grown there has been a steady upgrading of materials and this, together with improved control of operation has resulted in much better reliability and performance. This paper presents a state of the art review of factors which are important in the selection and use of materials in MSF and reverse osmosis (RO) desalination plants. Economics play an important part in materials selection and are given consideration in general terms. Key Words: Desalination, MSF, RO, corrosion, materials.
INTRODUCTION Large scale desalination really began with the multi-stage flash (MSF) plant built by Weir Westgarth in Kuwait in 1960. This and similar plants built elsewhere used mainly carbon steel and brass alloys for their construction. Experience in these early plants, particularly when acid-dosing was used as an antiscalant, was that corrosion problems were often extremely severe leading to costly shutdowns and maintenance. As knowledge of the requirements of materials to meet the conditions in MSF plants has grown, there has been a steady upgrading of materials, and this, together with improved control of operation, has resulted in much better reliability and performance. This paper presents a state of the art review of the factors which are important for selection and use of materials in MSF and reverse osmosis (RO) desalination processes. As the risk of corrosion is always present in desalination, corrosion plays a major part in determining material selection. Corrosion is a process involving the reaction of a material with its environment, and this paper reviews materials for usiing incomponents in the environments encountered in MSF and RO processes. In MSF plants these environments are sea water, deaerated sea water and brine, distillate and incondensible gases; in RO plants they are sea water, brackish waters and brines. Economics play an important part in materials selection and are given consideration in general terms. For general reference an appendix is included giving typical compositions of materials used in desalination plants. 67
Plenary Lectures
BEHAVIOUR OF MATERIALS IN SEA WATER AND BRINE ENVIRONMENTS General Considerations Sea water is the most corrosive of the natural environments that materials have to withstand, but it is much less corrosive than many environments encountered in industry, such as mineral acids. This situation has important implications for desalination materials in that there is a wide range of corrosion resistant alloys readily available, which have been developed for the chemical and process industries, and which are resistant to sea water. However, many of these materials are much more expensive than those used in industries which have traditionally handled sea water, such as shipping and power plant. In these industries, carbon steel, cast iron, copper base alloys and standard grades of stainless steel have been the usual choice for sea water applications. The desalination industry followed traditional practice and whilst in many cases this proved satisfactory, in others it did not. Upgrading, therefore, largely involved economic decisions-as the better materials were always available, the main problem was to decide what level of cost and performance was acceptable. Corrosion of Carbon Steel in Sea Water Sea water is a complex environment consisting of a mixture of inorganic salts, dissolved gases and organic compounds [1]. However, it also supports living matter in the form of both macro-organism (e.g., fish, shellfish, seaweed etc) and micro-organisms. All of these can have an influence on corrosion processes. Sea water is a well buffered solution, so the pH remains fairly constant at about 8. This means that most corrosion processes are dependent on the presence of oxygen. For carbon steels immersed in sea water, the rate of corrosion is mainly dependent on the oxygen content and the temperature of the sea water, the composition of the steel has relatively minor influence even when small amounts of alloying elements are added. Table 1 gives data on several steels exposed in the Pacific Ocean near Panama. Table 1. Corrosion of Steel in Sea Water [2] Steel
Nominal Composition
Carbon Copper-Bearing
0.2 5% C 0.22% C 0.31% Cu 2% Ni 0.6% Cu
Corrosion Rate (mm/yr) 1 year 16 years 0.15 0.075 0.15 0.077
Low Alloy Cu0.15 0.076 Ni Chromium Steel 0.08% C 3% Cr 0.05 0.110 Nickel Steel 5% Ni 0.16 0.080 Grey Cast Iron 0.24 0.147 Samples immersed below minimum low tide. Tidal flow 0.3 m/sec. Water temperature 15.5-32.2°C. Surfaces pickled before exposure.
68
Oldfield
In most engineering applications sea water flows over the metal surfaces so, it is important to have corrosion data under these conditions. Figure 1 [3] shows how flow rapidly increases corrosion of carbon steel in sea water. This is mainly due to the increase in the mass flow of oxygen to the corroding surface. In applications where continuous flow is required, a corrosion rate of about 1 mm/yr can be assumed for carbon steel and cast iron.
Figure 1. Effect of velocity on corrosion of carbon steel in sea water
Figure 2. Estimates of corrosion of carbon steel in deaerated sea water Deaeration, by reducing the oxygen content, would be expected to reduce corrosion, and 69
Plenary Lectures
in general, this is found to be the case (Fig. 2). However, in MSF plants, where deaeration occurs as part of the process, two other factors have an important effect. These are temperature and flow rate. Few test data have been measured under controlled oxygen, flow and temperature conditions, but Oldfield and Todd [4] have modelled the process and compared calculated and actual test data. Comparison with measured data [5] show that even with good deaeration, high corrosion rates can be experienced at high flow rates and high temperatures. At low corrosion rates, the measured and calculated corrosion rates are in good agreement, but at higher rates the calculated values are significantly above the measured values. This was attributed to the buildup of rust scales on the steel. Although these are not protective, they do provide a barrier to oxygen access to the surface and reduce the rate of attack. Corrosion of Copper Base Alloys in Sea Water Copper alloys have been used traditionally in marine engineering for heat exchanger tubing and for cast and wrought components in pumps and valves. These alloys form good protective films in sea water, and provided these films are undamaged, corrosion is slight. However, the protective films are susceptible to damage by fast flowing sea water, and this is an important factor in selecting these alloys for applications involving flow-for example, heat exchanger tubing. Table 2 gives data on some commonly used heat exchanger alloys. Table 2. Jet Impingement Tests on Copper Base Alloys Alloy Admiralty Brass Aluminium Brass 90/10 Cupronickel 7 0/30 Cupronickel 66/30/2/2 CuNiFeMn
Depth of Attack (mm). (jet velocity = 5.5 m/s) 0.60 0.13 0.06 0.03 0.025 (jet velocity: 12 m/sec)
In deaerated sea water copper alloys can still form protective films and show higher resistance to corrosion than in natural water. This is because the potential of copper alloys in sea water is much higher than the potential at which hydrogen can evolve; thus, in the absence of oxygen, corrosion is negligible. However, as in the case of carbon steel, low oxygen levels with high flow rates can cause impingement attack as shown by Anderson [6], but the rate of attack was much lower than in aerated sea water. Further data on corrosion of heat exchanger tubing alloys in deaerated sea water is given in Fig. 3 [7]. These data indicate that at low temperatures and oxygen levels, corrosion of the three alloys tested is acceptably low. However, at high temperatures with high oxygen levels, although the cupronickels continue to show low corrosion rates, aluminium brass is attacked. Thus, in MSF plants, where aluminium brass tubing is used, it is confined to the lower temperature recovery stages with the cupronickels being used in the higher temperatures areas.
70
Oldfield
Figure 3. Effect of oxygen content on corrosion of Cu base alloys in deaerated sea water Corrosion of Stainless Steels in Sea Water When chromium is added to steel, there is a marked increase in corrosion resistance, and at about 12% chromium, the alloys form a protective passive film and are referred to as stainless steels. Although the iron-chromium ferritic stainless steels are used commercially, the greatest tonnage of stainless steels are the austenitic iron-nickel-chromium alloys. These have better ductility and weldability than the ferritic alloys, and as they usually contain about 18% chromium, they have better corrosion resistance. In sea water, general attack on these alloys is negligible; however, they are prone to localised attack due to their high chloride content. This attack is particularly severe in crevices such as occur with overlapping surfaces, under seals and gaskets, and in similar areas. Resistance to this type of attack is improved by adding molybdenum to the alloys. The most commonly used grade of stainless steel in marine environments contains about 17% chromium, 12% nickel and 2.5% molybdenum and is commonly referred to by its US AISI designation, Type 316. Localised attack is stimulated by the difference in oxygen level within the pit or crevice and that in the surrounding area. In deaerated sea water, where the oxygen level is low, the risk of localised pitting and crevice corrosion of stainless steels is greatly reduced. Table 3 gives data on some stainless steels in aerated and deaerated sea water. These show that pitting in deaerated sea water and high chloride-containing brines is much less severe than in aerated sea water. Another characteristic of stainless steels is their ability to remain passive even in very fast flowing sea water. Table 4 gives data on some materials in fast flowing sea water. Carbon steel and cast iron are rapidly attacked. The copper base alloys suffer erosioncorrosion, but the stainless steel and nickel-base alloys are virtually unattacked. Table 3. Stainless Steels in Aerated and Deaerated Sea Water and Brine Alloy
Environment
Type 316 North Atlantic Ocean Type 316 Sea water (100°C, 25 ppb 02) Type 304 Sea water (100°C, 25 ppb 02) Type 316 130 g/l Cl- (pH 7, 8.25 ppb 02) 13-4 CrNi 130 g/l Cl- (pH 7, 8.25 ppb 02) P = Pitting G = General corrosion
Velocity (m/s) 0 0 0 40 40
Max Depth of Attack (mm) 2.400 P (486 days) 0.170 P (547 days) 0.600 P (547 days) 0.027 G (/year) [8] 0.180 G (/year) [8]
Table 4. Corrosion in Fast Flowing (35-42 m/s) Natural Sea Water Alloy Carbon Steel Cast Iron
Corrosion Rate (mm/yr) 4.50 13.20 71
Plenary Lectures
Gunmetal (85/5/5/5 CuSnZnPb) Ni Aluminium Bronze Type 316SS Ni Cu Alloy 400
1.32 0.97 2C > 2G > 2E > 2B and for the sour tests (sulphide added at 220 ppm), with water C (mixed 2H/effluent/aquifer), the biocide ranking was 2H > 2G > 2D > 2F > 2A CASE HISTORY 3: SEAWATER INJECTION SYSTEM, NORTH KUWAIT This is a conventional seawater injection system and, in this case, dynamic trials were undertaken by injecting alternate biocides directly into the pre-fouled, recirculating SBMTs. One SBMT was used for each combination biocide treatment. Formulations of the biocides are shown in Table 3. Biocides were dosed to the other 5 SBMTs over a 4 week period, steadily increasing the dose from 50% to 100% of the manufacturers recommended dose concentration, retaining the recommended frequency and dosage time. In all cases, a pronounced saw-tooth effect was evident, with bacterial numbers recovering rapidly once the biocide was removed from the system. This saw-tooth effect is regularly seen in such trials and in field monitoring of water injection systems. Biocides tend to kill only a proportion of the bacteria present in biofilms, and the survivors rapidly grow to re-form the active biofilm. This confirms the need to dose biocide frequently in order to keep sessile bacterial populations at a low level. Biocide Regime A Both biocides were each dosed up to 700 ppm of the product for 3 hours once per week (i.e., two biocide treatments each week). Biocide 3A.1 was an aldehyde mixture that did not foam, whilst 3A.2 was a mixture of aldehydes and amines that did form foam when agitated.
160
Sanders et al.
At half the recommended dose (350 ppm), biocide 3A.1 did not perform well, but biocide 3A.2 reduced bacterial numbers to ca 103 per stud. At three-quarters of the recommended dose (525 ppm), biocide 3A.1 was more effective, particularly against SRB, but biocide 3A.2 was only marginally more effective. At 90% of the recommended dose (630 ppm), biocide 3A.1 reduced numbers to 104 per stud while biocide 3A.2 reduced numbers to 102-103 per stud. At 700 ppm, biocide 3A.1 reduced bacterial numbers to 103 per stud, and biocide 3A.2 reduced them to 101 to 102 per stud. Bacterial numbers appear to recover rapidly (within a few days) after all the biocide doses. After the 75% dose of biocide 3A.2, however, numbers did not recover to their original level: 104-105 cells of each bacterial type were present compared to 106 per stud in the early stages of the trial. The data indicate that 3A.2 (containing an amine) was particularly effective at 630 ppm and above, and that 3A.1 above 525 ppm was able to support the effects of 3A.2. Continued treatment at 630-700 ppm with this biocide regime would be likely to maintain (and probably substantially reduce) sessile bacterial populations under these conditions. Biocide Regime B These biocides were dosed only once per week with the recommended concentration being 1000 ppm; 3B.1 was an aldehyde and amine blend dosed for 4 hours, 3B.1 was a mixed glutaraldehyde and THPS blend, again dosed for 4 hours. Biocide 3B.1 was strongly foaming whilst biocide 3B.2 did not foam when agitated. This once-per-week treatment gave good kill of all 3 bacterial types at 750 ppm (106/stud was reduced to 102-103 per stud), but within a few days the initial bacterial population levels were reestablished. There appears to be no significant difference in the data from the 1000 ppm dosing regime as compared to the 750 ppm dosing regime for GAB and GAnB (numbers were reduced to 101-103 per stud). Data from the Regime B trial indicated that one biocide dose per week is insufficient, and that 750 ppm for 4 hours is as effective as 1000 ppm for 4 hours. Biocide Regime C The biocides were dosed twice per week. Biocide 3C.1 was a fatty amine (250 ppm, 4 hours) and biocide 3C.2 was an aldehyde/QAC mixture (250 ppm, 4 hours). Both biocides formed foam when agitated, so they would not be suitable for dosing upstream of a deaerator tower. Biocide 3C.1 was very effective against GAB, GAnB and SRB;: even at 125 ppm, bacterial numbers were reduced from 106 to 102-103. Above 187 ppm, 3C.1 reduced numbers even further, in some cases to below 10 cells per stud. Biocide 3C.2 was, however, less effective. Even at 250 ppm, numbers were reduced by only 1 or 2 orders of magnitude. Despite the relatively poor performance of biocide 3C.2, however, there was a general downwards trend in bacterial numbers over the course of the trial and continuation with such a regime would probably lead to effective microbiological control. The most effective biocide in this case appeared to be 3C.1. Biocide 3C.2 could be improved by increasing the dosing concentration or changing to an alternative type. Biocide Regime D This vendor chose to submit only one chemical for testing, at a recommended dosing rate of 600 ppm for 4 hours once per week. The biocide was a complex blend of formaldehyde, 161
Oil Field Corrosion
QAC and THPS. It formed foam when agitated, and thus, might only be dosed downstream of a deaerator tower. At 300 ppm, bacterial numbers were reduced to 103 per stud, but at 450 ppm they were reduced to only 104. At 540 - 600 ppm, bacterial numbers were reduced to between 102 and 103 per stud, but only transiently. There was no apparent downwards trend in the data, and effective microbiological control did not seem to have been imposed by this regime, probably due to the infrequency of dosing. Biocide Regime E Regime E biocides were dosed twice per week. Biocide 3E.1 was a mixture of aldehydes, dosed at 200 ppm for 4 hours, while biocide 3E.2 was an aldehyde/QAC blend, again dosed at the recommended 200 ppm for 4 hours once per week. Biocide 3E.1 did not foam and thus could be dosed upstream of deaerator towers, whilst biocide 3E.2 did form foam and so would only be suitable for dosing downstream of such a tower. This biocide regime was particularly ineffective (apart from some transient reduction in SRB numbers early on). No consistent reduction in bacterial numbers was observed over the course of the treatment. Even if the concentration of the products were increased, effective control could not be guaranteed. The ranking derived from these trials is given in Table 4.
Table 4. Ranking of the Different Biocide Regimes Most Effective
Least Effective
Regime C Regime A Regime B Regime D Regime E
Biocide 3C-1 effective at 50% dose; biocide 3C.2 less effective even at 100% dose Effective at 90% dose Effective at 50% dose but needs twice per week Effective at 90% dose but needs twice per week Ineffective at 100% dose
CONCLUSIONS It is clear from this trial that one biocide treatment per week is insufficient to achieve good kill of biofilm bacteria. The data clearly show that two doses per week of an effective chemical can give good control of a well established biofilm. Such dosing would also minimize the buildup of biofilm in the first place if implemented from the startup of the system. In order to assess the suitability of any biocides for the system, other factors must be taken into account: 1. System demand. Sufficient residual biocide must be present at the end of the water distribution system to exert control there. A higher dose may be required at the main treatment plant to ensure that the minimum effective concentration is maintained at the system extremities, if there is a significant
162
Sanders et al.
biocide demand. System biocide demand must be determined as soon as the plant is in operation. 2. Environmental impact. Certain of the biocide components may have an adverse environment impact, and this must be assessed. 3. Safety considerations. All biocides are harmful but some may contain, for example, carcinogens, and these may be banned from use in Kuwait. 4. Physical properties. Foaming tendency, for example, is critical in deciding where in the system particular biocides can be dosed. Other factors such as precipitation in the formation should also be considered. 5. Consistency of supply. Selected products must be continuously available so that there are no periods when biocide cannot be supplied. 6. Chemical reliability. All products supplied in bulk should conform to the samples tested without any changes in composition. Random testing of supply for fingerprinting should be implemented to ensure consistency. 7. Local service base. Biocide vendors should ideally have a permanent base in Kuwait to deal with technical questions and to provide a local testing and advisory service. REFERENCES 1. 2. 3. 4. 5.
6.
7.
8. 9.
A.K. Tiller, Aspects of microbial corrosion, In Corrosion Process, R.N. Parkins Ed., Applied Science, 1982, pp. 115-160. W.A. Hamilton, Sulphate-reducing bacteria and anaerobic corrosion, Annual Reviews of Microbiology 39, 1985, pp. 195-217. G. Kobrin (Ed), A Practical Manual on Microbiologically Induced Corrosion, NACE, 1993. G.H. Booth, Microbial Corrosion, M & B Monograph CE/1, Mills & Boon, 1971. P.F. Sanders and D.L. Robinson, Corrosion control using continuous residual chlorine, In Microbial Corrosion (Proceedings of the 2nd European Federation of Corrosion Workshop on Microbially Induced Corrosion), European Federation of Corrosion, Publication No. 8, Institute of Materials. C.A.C. Sequeira and A.K. Tiller Eds., 1992, pp. 198-209. P.F. Sanders, Monitoring and control of sessile microbes: Cost effective ways to reduce microbial corrosion, In Microbial Corrosion 1, Elsevier Applied Science, C.A.C. Sequeira and A.K. Tiller Eds., 1988, pp. 191-223. W.J. Georgie, P.I. Nice and S. Maxwell, Selection, optimisation and monitoring of biocide efficiency in the Statfjord water injection systems, In UK Corrosion ’91, 1991. I. Ruseska et al., Biocide testing against corrosion causing oilfield bacteria helps control plugging, Oil and Gas Journal 80, 1982, pp. 253-264. P.F. Sanders and L. Latifi, On-site evaluation of organic biocides for cost-effective control of sessile bacteria, In Proceedings of the Second International Conference on Chemistry in Industry, American Chemical Society, Vol. I, Paper O-30, 1994, pp 242-257.
163
Oil Field Corrosion
10. Review of current practices for monitoring bacterial growth in oilfield systems. CCEJV 1987. Document number 001/87. Corrosion Control Engineering Joint Venture, CCEJV, UK. 11. J.W. Costerton and G.G. Geesey, Microbial contamination of surfaces, In Surface Contamination (1),. K.L. Mittal Ed., Plenum Pub., 1979, pp. 211-222. 12. P.F. Sanders, Rapid methods for detecting microbial corrosion, In Proceedings of UK Corrosion ’92, Volume 3, Session D. Institute of Corrosion, 1992. 13. Review of current practices for monitoring bacterial growth in oilfield systems, 1987, Document No. 001/87, Corrosion Control Engineering Joint Venture CCEJV/NACE. 14. P.F. Sanders, Control of microbiologically induced corrosion using field and laboratory methods. International Biodeterioration 24, 1988, pp. 239-246. 15. P.F. Sanders and J.F.D. Stott, Assessment, monitoring and control of microbiological corrosion hazards in offshore oil production systems. NACE Corrosion ‘87. Paper No. 367, 1987.
164
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
HYDROGEN DEGRADATION OF STEEL - DIFFUSION AND DETERIORATION M. Farzam Faculty of Petroleum Eng. University of Petroleum Industry, Ahwaz, Iran
ABSTRACT Following aqueous surface electrochemical reactions and hydrogen reduction, atomic hydrogen will diffuse through steel, to reside at the microstructural defects with possible catastrophic failures to follow. Carbon steel, low alloy steel or high strength low alloy steel under no-load, static or cyclic loading will be affected by the presence of hydrogen. Such an effect is mainly influnced by temperature, alloying, residual and applied stress, and partial pressure of hydrogen. Permeation tests were conducted by Devanathan's electrochemical method and a newly invented electro-vacuum method. Experiments illustrated that as the microstructure changed the diffusion constant, D, changed. Ingress of hydrogen increased with the reduction of voltage and pH. Thickness would not have a realistic effect, but cold work reduced D. The difference in D measured by the two methods of permeation was due to the variation in material, apparatus design and operation. Hydrogen degradation during stress corrosion and corrosion fatigue tests is one of the two processes of crack propogation. Immersion, dynamic polarization, stress corrosion (dynamic and static), corrosion fatigue tests, cathodic protection and fractographic studies were conducted in seawater and sour (hydrogen sulfide generated by sulfate reducing bacteria) environments. In all, it was concluded that with regards to Nernest's equation, as the partial pressure of hydrogen increased, the hydrogen concentration gradient (Fick's first law) increased and D increased. Furthermore, the speed of hydrogen diffusion increased. In the sour environment, the overtaking mechanism of failure was found to be hydrogen embrittlement rather than anodic dissolution. Key Words:
Hydrogen degradation, hydrogen diffusion, diffusion constant, corrosion fatigue, stress corrosion cracking
INTRODUCTION The sources of hydrogen production in the industry are many: water dissociation
+ H 2 O → H + OH
(1)
Fe-H2O reaction
2Fe+ 3 H 2 O
(2)
effect of pH and bacteria
+ H +e → H
(3)
165
Oil Field Corrosion
After atomic hydrogen production at the cathodic side of a corrosion cell, hydrogen will diffuse into steel residing at the lattice interstitial sites, grain boundaries, vacancies, impurities and alloying elements. The bonding energy of such interactions is different (see Fig. 1 and Table 1). Trap sites are either low energy reversible (e.g., interstitial solutes 3-15 KJ/mol [1]) or high energy irreversible (e.g., TiC particles 96 KJ/mol [1]) It is believed that steels with high energy traps will release hydrogen at about 2500C and at higher temperatures. Generally, hydrogen will not saturate and therefore is less likely to damage the austenitic stainless steels, but carbon steel, and ferritic or martensitic steels will be degraded, and finally cracked. Table 1. Trap Energy [1] Trap Interstitial Solutes (C, N) Si Atom Ti Atom Vacancy Y Vacancy Dislocation Elastic Stress Field Dislocation Core (Screw) Dislocation Core (Mixed) 1/2 H2 (Vapour /Liquid) Grain Boundary Free Surface AlN Interface Fe3C Interface TiC Interface
EB (KJ/mol)
NT (m-3)
~3-15 >20 26 46 126 20 20 - 30 59 29 ~59 70 - 95 65 84 96
1025 1027 1027 2.5 mm/year (100 mpy)). The combination of hydrogen sulfide and carbon dioxide is more aggressive than hydrogen sulfide alone and is frequently found in oil field environments. Once again, the presence of even minute quantities of oxygen can be disastrous. In all cases, increased velocity would be expected to increase the corrosion rate of steel [2]. The following ferrous corrosion products would form with H2S and CO2 in the presence of oxygen and low solid water. For corrosion, they are the only products of concern: Fe + H2S → FeS + H2 (Sour corrosion)
(1)
Fe + H2O + CO2 → FeCO3 + H2 (Sweet corrosion)
(2)
4Fe + 3O2 → 2Fe2O3 (Oxygen corrosion)
(3)
The iron sulfide produced by reaction (Eq. 1) generally adheres to the steel surfaces as a black powder or scale. The scale tends to cause local acceleration of corrosion because the iron sulfide is cathodic to the steel; this usually results in deep pitting during O2 reduction reactions along this FeS layer. Oxygen will provide a high electrochemical potential because it is a strong, rapid oxidizing agent. This means that it will easily combine with electrons at the cathode, and allow the corrosion to proceed at a rate limited primarily by the rate at which oxygen can diffuse to the cathode. The primary concern in this case would be when the H2S partial pressure is ≥ 0.05 psi. As a first approximation, calculations of the partial pressure of both carbon dioxide and hydrogen sulfide are useful for the present case study in predicting the degree of corrosivity of the oil well (i.e., No. W-41): Partial pressure of CO2 = 1500 x 0.07 = 105 psi Partial pressure of H2S = 1500 x 0.001 = 1.65 psi Using the CO2 partial pressure as a yardstick to predict corrosion, it was found that the CO2 partial pressure was above 30 psi, which usually indicates the onset of corrosion, however, in this study, the volume fraction of CO2 over that of H2S was < 200. This indicates that the presence of small concentrations of H2S play a prominent role in the corrosion mechanism involved in the reactions (Eqs. 1-3). If the given volume fraction is > 200, then carbon dioxide will be the controlling factor for the reactions; this does not apply to the present case and its oil field environment. According to a computer modeling program [3] used in this study to estimate the corrosion rate on the pitted tubing, and based on the analysis of the production data in addition to the operation conditions given in Appendix A, the following two cases were predicted:
221
Oil Field Corrosion
1. The production of oil and formation water with H2S and CO2 under deaerated conditions (i.e., oxygen is excluded) would result in a corrosion rate resulting in tubing perforation after more than two years of exposure, and 2. Changing the production conditions to include oxygen in addition to H2S and CO2 would result in a worst-case corrosion rate leading to tubing failure in about eight days (10 days in service to actual failure). Based on the examination and this preliminary computer analysis from CLI International, Inc., it was assumed that case 2 is likely to model the situation involved in the tubing failure. However, according to on-site field inspection of the oil well and observation of the huge amount of gas pressure released from the wellhead, when the pressure valve was opened, the theoretical assumption based on the combined effects of oxygen and corrosive gases could not be accurate because oxygen had no chance to diffuse into the well tubing under such a high-pressure release of gas at the outlet. Moreover, there was no chance for oxygen to be drawn into the well through leaky valves or the expansion of joints in this solidly designed oil well tubing part. Therefore, the only solution was to look for other causes of damage that would have enhanced the corrosivity of the gases (i.e., CO2 and H2S) and could have contributed to the acceleration of the rate of corrosion attack. On observing the case history of the oil well (see Appendixes A and B) and by comparing the performance of this oil well with respect to the nearest oil wells located within one kilometer in the same oil field, the following suspected factors can be assumed: The failure of the tubing was initiated due to differential acid concentration cell (case 1) that had been established beneath crevices that were hydrocarbon in nature, and being strengthened with a very high propagation rate of corrosion attack caused by one or more of the following suspected factors:
• Failure in the electrical submersible pump materials (ESP) as indicated in • • • • •
Appendix B, Electrical power problems (i.e., a rectifier power supply problem), Galvanic coupling effects between the materials of the ESP and the internal surface of the tube, Failure including doglegs, splits, and leakage of the well casing materials at a certain depth adjacent to the well tubing promoted after the burning of the well during the Iraqi invasion, and Faulty electrical ground connection (ground electrical contact was made to the well tubings), and Stray current effect from either the ESP or other exterior effects.
CONCLUSIONS 1. Metallographic examination, SEM and mechanical testing of the tubing did not show any major differences between the as-received and the failed tubing in terms of mechanical properties or the presence of second phase particles and inclusions. The minor additions of Cu, Cr and Si did not contribute much to the poor corrosion properties of the steel tubing; however, the presence of higher contents of nonhomogenous MnS in both tubing specimens may also indicate the possibility of
222
Husain and Hasan
MnS inclusion caused by variable degrees of extrusion during pipe manufacturing. Manganese sulfide inclusions are known to be detrimental to corrosion resistance. 2. Oxygen was not the most likely cause of the perforation. 3. A solution to combat the tubing corrosion problem is to use a more expensive approach by carefully evaluating and selecting corrosion-resistant alloys or to inhibit corrosion with chemicals to minimize corrosive attack on the steel tubing. 4. Internal coating of tubing using powder coating systems (e.g., Tuboscope Vetco International) or fiberglass reinforcements have proven to be successful in many oil industry and downhole applications. REFERENCES 1. F.W. Smith, Structure and Properties of Engineering Alloys, New York, MacGrawHill, 1993. 2. NACE. Corrosion control in petroleum production. National Association of Corrosion Engineers, Publication No. TPC 5, Houston, Texas, 1979. 3. Private communication with Dr. R.D. Kane during his visit to Kuwait, 1995.
APPENDIX A Case History for Well No. W-41 Industry: A Kuwaiti oil production Specimen location: Downhole tube from Well No. W-41 Specimen orientation: Vertical Years/days in service: 10 days Salinity: 80,000 to 90,000 ppm Average water cut: 73%-Temperature: 116oF (47oC). Total pressure: 1250-1500 downhole CO2 Concentration: 7 vol% H2S Concentration: 0.11 vol% Failed samples given: Two failed tubing samples with tube specifications J55-2 3/8 in. O.D. Sample designation: Sample No. 3 from Joint No. 33. Downhole tubing depth: ± 1008 ft Failure condition: One perforated hole Sample designation: Sample No. 4 from Joint No. 60. Downhole tubing depth: 1832 ft Condition: 4 perforated holes Engineering component connection: Electrical submersible pump (ESP) at 2,258 ft Oil well fluid level: static 527 ft. Dynamic 558 ft
223
Oil Field Corrosion
APPENDIX B .History of the Premature Failure of Downhole Equipment and Tubings in Well No. W-41. Visual Inspection Findings of Pulled Equipment Failure Failure Run Life No. Date (months) 1 Jan. 2.5 90 2 May 3.5 90 3 Oct. 5 92
___
burnt
leaking
___
___
shorted
leaking
damaged
corroded
burnt
___
___
___
___
pump Joint. #29
burnt with 2 holes burnt Ext. corr. erosion corrosion.
___
good
___
good
#30 #59
External corrosion Erosion corrosion
___
Ext. Corr. left in hole
5
Jul. 93 Jan. 94
6
___
6
good
7
Nov. 94
10
good (red alloy)
8
Nov. 94
0.5
good (red alloy)
9
Jan. 95
1.5
10
Feb. 95
10 days
11
Mar. 95
40 days
good (carbon steel) good (carbon steel) good (carbon steel)
12
Jul. 95
42 days
13
Sept. 95
Failed tubing Jts (Times Failed) Depth of tubing Section Failed Frequently
29 (1)
one hole one hole
Iraqi Invasion DDI while RIH
Separator
good
2
___
Protector
3
6
___
Motor
Jan. 90
good (carbon steel) good (carbon steel) 30 (1)
good
hole & good Ext. Corr. (Red Alloy) (carbon steel) reused shorted good +big hole stuck (Red Alloy) (carbon (carbon steel) steel) good good good (carbon (carbon (Red Alloy) steel) steel) good good good (carbon (Carbon (red alloy) steel) steel) wire boltsred alloy loose (carbon steel) good good 1/16 inch hole (carbon (red (red alloy) steel) alloy) good red red alloy (carbon alloy steel) 33 37 39 (1) (1) (1)
900 ft- 1000 ft
Cable
Remarks
Pump
4
PSI
Failed Tubing Joint Joint No. of No. Holes ___ ___
not R.C. work damage ___ ___
shorted (red alloy)
DDI while RIH pump Joint
___
F.C. burned & parted good
___
good
___
good
#59 3 holes DDI while ? &1 RIH hole (KSRC) #33 1 hole KISR < &4 hole
___
___
___
one hole
hi Volt hi Amp
#37, #39, #50
___
#50, #58, #59
58 (1)
59 (3)
5 holes 3 holes 7 holes
red alloy 50 (2)
60 (1)
N-80 grade tubing N-80 grade tubing 63 pup (1) joint (2)
1800 ft- 1925 ft
Production data for Well No. W-41: Operating Conditions: -Average production rate = 2,750 BFPD At subsurface (downhole) -Average water cut = 73% Temp.:116°C -Salinity = 80,000- 90,000 ppm Pressure:1250-1500 psi -Setting depth of: CO2 Conc.:7 Vol.% Electrical Submersible Pump (ESP) = 2,258 ft. H2S Conc.:0.11 Vol% Dynamic fluid level = 558 ft. Static fluid level = 527 ft. -GOR = 635
224
DDI while RIH
At Surface 85°C 110 psi 15 Vol.% 1.1 Vol.%
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
METHODOLOGIES FOR ASSESSMENT OF CRUDE OIL CORROSIVITY IN PETROLEUM REFINING S. Tebbal and R.D. Kane CLI International, Inc. 14503 Bammel N. Houston, Suite 300 Houston, Texas 77014-1149, USA
ABSTRACT Lower quality “opportunity” crudes are now processed in most refineries, and the source of the crudes may vary daily. These feedstocks, if not properly handled, can result in reduction in service life of equipment as well as costly failure and downtime. Analytical tools are needed to predict their high-temperature corrosivity toward distillation units. Threshold levels of total sulfur and total acid number (TAN) have been used for many years as rules of thumb for predicting crude corrosivity. However, it is now realized that they are not accurate in their predictive ability. Crudes with similar compositions and comparable with respect to process considerations have been found to be entirely different in their impact on corrosion. Naphthenic acid content, sulfur content, velocity, temperature, and materials of construction are the main factors affecting the corrosion process. Despite progress made in elucidating the role of the different parameters on the crude corrosivity process, the main problem is in calculating their combined effect, especially when the corroding stream is such a complex mixture. The TAN is usually related directly to naphthenic acid content. However, discrepancies between analytical methods and interference of numerous components of the crude itself lead to unreliable reported content of naphthenic acid. The sulfur compounds, with respect to corrosivity, appear to be related more to their decomposition at elevated temperature to form hydrogen sulfide than to their total content in the crude. This paper reviews the present situation regarding crude corrosivity in distillation units, with the aim of indicating the extent of available information, and areas where further research is necessary. Key Words: Naphthenic acid corrosion, crude oil corrosion, high temperature corrosion
INTRODUCTION The quality of crude oils processed around the world is worsening. The higher densities of crudes and higher contents of sulfur, acids and other impurities found in crude oils have increased the likelihood of corrosion failures in the processing plants. Moreover, it is now realized that many of the old rules of thumb, namely threshold levels of sulfur and total acid number (TAN) or neutralization number, do not appear to be accurate in their predictive ability. Analytical tools are needed to predict their high-temperature corrosivity toward distillation units. Sulfur at a level of 0.2% and above is known to be corrosive to carbon and low alloy steels at temperatures from 230°C (450°F) to 455°C (850°F). When sulfur is the only contaminant, McConomy curves [1], with the help of correction factors, are used to predict the relative corrosivity of crude oils and their various fractions as well as the effect of
225
Corrosion in Refinery and Petrochemical Industries
operational changes on corrosion rates already experienced in the field. However, when naphthenic acids are present, crude corrosivity prediction becomes more complex. Crudes with similar compositions and comparable with respect to process considerations have been found to be entirely different in their impact on corrosion. MANIFESTATION OF NAPHTHENIC ACID CORROSION Naphthenic acids are organic acids present in many crude oils from around the world, especially those from California, Venezuela, the North Sea, Western Africa, India, China, and Russia. They have the following generic chemical formula: R(CH2)nCOOH, where R is a cyclopentane ring and n is typically greater than 12. Naphthenic acid corrosion is differentiated from sulfidic corrosion by the nature of the corrosion (i.e., pitting and impingement) and by its severe attack at high velocities in crude distillation units. Crude feedstock heaters, furnaces, transfer lines, feed and reflux sections of columns, atmospheric and vacuum columns, heat exchangers, and condensers are among the type of equipment subject to this type of corrosion [2]. Damage is in the form of unexpected high corrosion rates on alloys that would normally be expected to resist sulfidic corrosion. Isolated, deep pits in partially filmed areas and/or impingement attack in essentially film-free areas is typical of naphthenic acid corrosion [3,4]. In many cases, even very highly alloyed materials (i.e., 12 Cr, AISI 316 and 317, and in some severe cases even 6% Mo stainless alloys) have been found to exhibit sensitivity to corrosion under these conditions. PARAMETERS AFFECTING NAPHTHENIC ACID CORROSION TAN (or neutralization number), sulfur content, velocity, degree of vaporization, temperatures, and alloy composition (i.e., Cr and Mo) were found to be the main factors affecting the corrosion process. Total Acid Number The TAN was related in early studies to the naphthenic acid corrosion rate, and its threshold was believed to be around 0.5 mg KOH/g [3]. Analysis of crudes’ TAN distribution as a function of the true boiling point showed where the acids concentrated in the refinery and resulted in better correlation of TAN with corrosion experienced in the field [5]. However, correlating the TAN of specific cut to their corrosivity was still far from being reached. Even naphthenic acid weight percent did not correlate to experienced corrosion [6]. A standard laboratory test was developed and showed promising crude corrosivity prediction. An index called the naphthenic acid corrosion index (NACI) was calculated from the exposure of a carbon steel coupon at 500°F (260oC) for 48 hours in the fraction to be tested [7]. The index was the ratio of the corrosion rate of the coupon in mills per year (mpy) to the weight of its corrosion film in milligrams per square centimeter. It was postulated that a calculated ratio under 10 indicated sulfidic corrosion while a higher index indicated naphthenic acid attack. However, this index was developed from static test results, and correlations to high velocity might not be that simple and need to be verified. Moreover, carbon steel was the material used in these tests while most of the refineries process naphthenic crudes with a minimum alloy of 5 Cr.
Sulfur Content 226
Tebbal and Kane
Most crude oil feedstocks vary greatly in both the amount of sulfur and the type of sulfide species present. It is believed that the sulfur content does not reflect the true effect of sulfur. A more important factor may be the capability of these sulfur compounds to form H2S during heating in the refining process [8]. At low temperatures, certain sulfur compounds in the crude may reduce the severity of naphthenic acid corrosion [9]. In this case, the sulfide film may offer some degree of protection from the acidic corrosion. At higher temperature conditions, the presence of naphthenic acids was found to increase the severity of sulfidic corrosion. Presumably, the presence of these organic acids disrupted the sulfide film, thereby promoting sulfidic corrosion on alloys that would normally be expected to resist this form of attack (i.e., 12 Cr and higher alloys). Figures 1 and 2 show the effect of chromium and molybdenum content as well as TAN and sulfur on the corrosion rate of crude fractions tested in the laboratory at a velocity of about 3 m/s (10 ft/s) and temperature of 370°C (700°F) for three days. In the vacuum heater feeder line (VHFL) cut the 1.5 Cr and 5 Cr showed the same corrosion rate, but in the long resid (LR) fraction the 5 Cr and 9 Cr behaved similarly. In general, the corrosion rate decreased steadily with increase in Cr and Mo. However, both TAN and sulfur content do not correlate in any way to corrosion rate. In the VHFL, the fraction with medium TAN and sulfur was the most corrosive to low alloy steels. In the LR, the fraction with the lowest TAN and sulfur was the most corrosive. In this case, blending a high TAN crude to lower the acid content did not provide the expected results and aggravated corrosion. These cases are common and refinery experience shows that prediction of corrosion is more complex than it is believed to be. Temperature Naphthenic acid corrosion occurs primarily in high velocity areas of crude distillation units in the 220 to 400°C (430 to 750°F) temperature range. No corrosion damage is found at temperatures above 400°C (750°F) probably, because of the formation of coke at the metal surface. The corrosion rate of all alloys of importance to the distillation units increases with increases in temperature. Velocity The flow regime and the degree of vaporization have a significant effect on both sulfidic corrosion and naphthenic acid corrosion. The higher the acid content generally, the greater the sensitivity to velocity. In fact, in some cases, it appears possible to obtain very high corrosion rates even at very low levels of naphthenic acid content (i.e., TAN ≈ 0.3) and low sulfur content when combined with high-temperature and high velocity. Materials of Construction The normal materials of construction used in crude distillation units are carbon steel , 5 Cr, 9 Cr, 410SS, and 316SS [10]. If only sulfur is present and the temperature is above 288°C (550°F), 5 Cr or 12 Cr cladding is recommended for crudes over 1% sulfur when no operating experience is available [2]. If hydrogen sulfide is evolved, 9 Cr minimum is preferred. In contrast to high-temperature sulfidic corrosion, low-alloy steels containing up to 12% Cr provide no benefits over carbon steel in naphthenic acid service [1]. With 316 SS (with 2.5% Mo minimum) or better, with 317 SS with a higher Mo content (3.5% minimum), cladding of the vacuum column is recommended when TAN is above 0.5 mg KOH/g and in an atmospheric column when the TAN is above 2.0 mg KOH/g [2]. 227
Corrosion in Refinery and Petrochemical Industries
120
Corrosion Rate (mpy)
VHFL I (TAN = 0.35, S = 4.17%) VHFL II (TAN = 1.64, S = 1.06%) 80
VHFL III (TAN = 0.54, S = 2.09%)
40
0 0
5
10
15
20
25
Cr + Mo (%)
Figure 1. Effect of Cr and Mo content on the general corrosion rate of alloys in three vacuum heater feeder line oil cuts
50
Corrosion Rate (mpy)
LR I (TAN = 0.35, S = 0.6%, H2S = 2.0%) 40
LR II (TAN = 2.35, S = 0.9%, H2S = 2.0%)
30
20
10
0 0
5
10
15
20
25
Cr + Mo (%)
Figure 2. Effect of Cr and Mo content on the general corrosion rate of alloys in two long resid oil cuts MITIGATION METHODS Mitigation of naphthenic acid corrosion includes blending, inhibition, and materials upgrading [11]. Blending is the most preferred method and is accomplished by diluting a high TAN crude with a low TAN one, thus reducing the acid content to a level which corresponds to an acceptable corrosion attack. Injection of corrosion inhibitors may provide 228
Tebbal and Kane
adequate and economic protection if it is closely monitored and used for specific fractions that are known to be particularly severe, or if it fluctuates with feedstock quality. When possible, upgrading the construction materials to a higher chrome and/or molybdenum alloy is the best solution for long term reliability. CRUDE CHEMISTRY AND CORROSIVITY CORRELATION Most of the laboratory studies and refinery experience so far have shown that crude corrosivity prediction is very complex and that further studies are needed to correlate the chemistry of crudes to refinery corrosion. Unfortunately, crude chemistry (sulfur and TAN), process variables (temperature and velocity), and failure analysis (sulfidic and/or naphthenic acid corrosion) are far from being assessed uniformly throughout the industry. To be able to achieve correlations among refineries and between laboratories and plants, the measurement method for each parameter needs to be defined precisely. Naphthenic Acid Distribution Naphthenic acid content is generally expressed in terms of TAN. ASTM D974 is a colorimetric method., with reproducibility of 15% and interference from inorganic acids, esters, phenolic compounds, sulfur compounds, lactones, resins, salts, and additives such as inhibitors and detergents. ASTM D664 is a potentiometric method with reproducibility of 20 to 44% depending on the end point (i.e., buffer or inflection), type of oil (i.e., used or fresh), and titration mode (i.e., automatic or manual), and the same interfering impurities as ASTM D974. Both ASTM methods do not differentiate between naphthenic acids, phenols, carbon dioxide, hydrogen sulfide, mercaptans, and other acidic compounds present in the oil. In addition, the two methods were compared [5], and D664 yielded numbers that were 30 to 80% higher than D974. Thus, prediction of crude corrosivity based on TAN alone could be misleading. Additionally, for assessment of plant corrosion effects, the naphthenic acid content needs to be determined for each cut in order to predict exactly where the acids will concentrate during the distillation of the crude. The isolation and analysis of naphthenic acids from crude oil may be performed adequately with methods such as UOP 565 (potentiometric) and UOP 587 (colorimetric), chromatographic separations, or other available analytical techniques [12,13]. The relative abundance of naphthenic acid and its average molecular weight (i.e., boiling point) may be determined. In addition, the assays of crude must be current [2]. Once steam flooding or other recovery method is begun in an oil field, the specific gravity and the organic and sulfur content of the crude can change. Fire flooding, when used in some fields, tends also to increase the naphthenic acid content. Hydrogen Sulfide Evolution with Temperature Sulfur is the most abundant element in petroleum other than carbon and hydrogen. It may be present as elemental sulfur, hydrogen sulfide, mercaptans, sulfides, or polysulfides. The total sulfur content is generally analyzed with the ASTM D4294 method using x-ray fluorescence. Halides and heavy metals interfere with this method. The capability of these sulfur compounds to form H2S during heating in the refining process, rather than their total content, is believed to correlate to corrosion in the plants [8]. However, a standard procedure for determining hydrogen sulfide evolution with temperature is not currently available.
229
Corrosion in Refinery and Petrochemical Industries
Figure 3. Schematic diagram of a rotating autoclave
Figure 4. Schematic diagram of a jet impingement apparatus Wall Shear Stress Fluid flow velocity has long been used as the parameter for comparing flow among refineries and between laboratory and field. However, this concept was found to lack predictive capabilities and was replaced by data related to fluid flow parameters such as wall shear stress and Reynolds number [14]. Wall shear stress, rather than velocity, is the parameter directly proportional to corrosion through the removal of normally protective films. The wall shear stress in the field is proportional to (1) the density and viscosity of the liquid and vapor in the pipe at temperature, (2) the degree of vaporization in the pipe, and (3) the pipe’s diameter. Wall shear stress in the laboratory depends on the geometry and dimensions 230
Tebbal and Kane
of the laboratory apparatus. Figure 3 shows a schematic diagram of a rotating autoclave with a condenser for the return of light components, and Fig. 4 is a schematic diagram of a jetimpingement laboratory setup. Both of these apparatus are used in the laboratory to simulate corrosion in the field at high velocities. Table 1 compares the shear stress level between the field and the laboratory. The results show that the wall shear stress changes drastically with the degree of vaporization, and that identical fluid flow in the laboratory and the field do not correspond to the same level of shear stress. Table 1. Wall Shear Stress (in Pascals) Calculated at Different Velocities (VGO cut of 0.7 Specific Gravity and 0.55 Centistokes Viscosity at 700oF) Velocity
3 m/s (10 ft/s)
Rotoclave Jet Impingement
18 4
DV = 0% 46 DV = 30% 32 DV = 70% 14 DV = Degree of vaporization
6 m/s 16 m/s (20 ft/s) (50 ft/s) Laboratory Setup 58 N/A 13 80 Plant 158 691 111 514 47 286
33 m/s (100 ft/s)
66 m/s (200 ft/s)
N/A 298
N/A 1049
2830 2080 862
10291 5148 2210
Table 2. Effect of Velocity on General and Pitting Corrosion Rates Alloy General Corrosion Rate (mpy) Pitting Corrosion Rate (mpy) at 10 ft/s at 200 ft/s at 10 ft/s at 200 ft/s 5 Cr 21.8 25.5 0.0 201.1 9 Cr 20.3 24.2 0.0 191.8 317 SS 3.29 6.09 0.0 28.7 Laboratory Testing Laboratory studies are directed at the simulation of field conditions under controlled and reproducible conditions. CLI International, Inc. is currently involved in a major multiclient effort to provide more systematic understanding and a methodology for handling crude corrosivity and naphthenic acid corrosion issues. The interpretation of laboratory results and their correlation to the plant need to be analyzed carefully. The temperature of fluid and specimens in laboratory studies most likely are equal. However, in furnaces and heat exchangers of crude distillation units, temperature differences between the stream and the metal skin may be as high as 85 to 100°C (150 to 200°F) [1]. Rates of corrosion found in laboratory testing may correspond to the maximum corrosion rates found in the field. This usually results from the short test duration in the laboratory. The corrosion rate is usually high initially and then decreases with time because of the formation of protective films. The laboratory corrosion rates may also be much lower than those experienced in the field if the composition of the test solution changes with time as a result of degradation of its corrosive components. The type and rate of corrosion may be easily calculated in the laboratory. Table 2 shows the effect of velocity on the general and localized corrosion rates of three alloys after 231
Corrosion in Refinery and Petrochemical Industries
three days of exposure. Differences in flow velocity on the general corrosion of the three alloys was minimal. However, a significant increase in the pitting corrosion rate with increases in flow from 10 ft/s to 200 ft/s was found especially for the 5 Cr and 9 Cr alloys. Corrosion rates in the field are evaluated by on-line monitoring tools which indicate only general corrosion rates unless the equipment is inspected for pitting and/or impingement. CONCLUSIONS It has clearly been proven through extensive laboratory and plant studies that predicting crude corrosivity by using the general rules of total acid and sulfur content is not reliable especially with the wide range of crude oil feedstocks being processed today. Naphthenic acid content and distribution in side cuts, hydrogen sulfide evolution with temperature, wall shear stress, temperature at the metal surface, and materials of construction are the main factors affecting the crude corrosivity process. The exact mechanisms which are operating are not precisely known at this point, and much research and testing is necessary to build a more comprehensive understanding. Based on the complexity of the situation and the current level of understanding, each case must be dealt with on an individual basis until a more comprehensive methodology for assessment can be developed. It is possible, however, to provide a practical assessment of the plant corrosion process by establishing a more comprehensive database from both laboratory and field experience where the various parameters affecting naphthenic acid corrosion can be more extensively and unambiguously defined and quantified. This information will serve as a firm basis for materials selection decisions, feedstock blending requirements and plant operating conditions. REFERENCES 1. L. Garverick, Ed., Corrosion in the Petrochemical Industry, ASM International, 1994. 2. R.A. White and E.F. Ehmke, Materials Selection for Refineries and Associated Facilities, NACE, Houston, Texas, 1991. 3. W.A. Derungs, Corrosion 12, 12, 1956, p. 617t. 4. J. Gutzeit, Materials Performance 16, 10, 1977, p. 24. 5. R.L. Piehl, Materials Performance, January 1988, and Paper No. 196, Corrosion/87, NACE. 6. E. Babian-Kibala, et al., Naphthenic acid corrosion in a refinery setting, NACE Conference, Corrosion/93, Paper No. 631, 1993. 7. H.L. Craig, Naphthenic Acid corrosion in the refinery, Paper No. 333, Corrosion/95, NACE. 8. R.L. Piehl., Corrosion, June 1960, p. 305t. 9. Heller, Materials Protection, September 1963. 10. F. Blanco and B. Hopkinson, Experience with naphthenic acid corrosion in refinery distillation process units, NACE Conference, Corrosion/83, Paper No. 99, 1983. 11. G.L. Scattergood and R.C. Strong, Naphthenic acid corrosion: An update of control methods, NACE Conference, Corrosion/87, Paper No. 197, 1987. 12. Tseng-Pu Fau, Energy and Fuels 5, 3, 1991, p. 371. 13. I. Dzidic, et al., Analytical Chemistry 60, 13, July 1, 1988, p. 1318. 14. K.D. Effird, et al., Experimental correlation of steel corrosion in pipe flow jet impingement and rotating cylinder laboratory tests, Corrosion/93, Paper No. 81, NACE. 232
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
NEW NICKEL ALLOYS SOLVE CORROSION PROBLEMS OF VARIOUS INDUSTRIES D.C. Agarwal1 and W.R. Herda2 1
VDM Technologies 11210 Steeplecrest, # 120, Houston, Texas 77065, USA 2
Krupp-VDM GmbH P.O. Box 1820, 58778 Werdohl, Germany
ABSTRACT The materials of construction for the modern chemical process and petrochemical industries not only have to resist uniform corrosion caused by various corrodents, but must have sufficient localized corrosion and stress corrosion cracking resistance as well. These industries have to cope with both the technical and commercial challenges of rigid environmental regulations, the need to increase production efficiency by utilizing higher temperatures and pressures, using more corrosive catalysts, and at the same time possess the necessary versatility to handle varied feed stock and upset conditions. Over the past 30 years improvements in alloy metallurgy and a fundamental understanding of the role of various alloying elements has led to newly developed Ni-Cr-Mo and Ni-Mo alloys, which not only extend the range of usefulness of existing alloys by overcoming their limitations, but are also cost-effective and open new avenues of applications. This paper presents the various metallurgical, thermal stability and corrosion resistance characteristics of some newer alloys along with actual case histories, where these alloys have solved specific problems. Key Words:
Nickel alloys, chemical process industries, petrochemical applications, localized corrosion, uniform corrosion, acid corrosion, Alloy 59, Alloy 31, Alloy 33, Alloy B-2, Alloy B-4
INTRODUCTION At one time one, of the major factors in any material selection was initial cost with little thought given to maintenance and the cost associated with lost production due to unscheduled equipment downtime. In today's economic environment, increased maintenance costs and downtime have placed a greater emphasis on the need for reliable, safe and versatile performance of process equipment. Prior to the 1950's, the alloy choices to combat corrosion were very limited. The latter half of the 20th century saw a phenomenal growth in the development of new nickel-based alloys due to improved melting and thermo-mechanical processing innovations, a better fundamental understanding of the role of various alloying elements and their effect on both corrosion and physical metallurgy behavior. Even though the standard austenitic stainless steels (Alloy 304L and 316L) have been and continue to be the workhorse of many industries, their vulnerability to localized corrosion and chloride stress corrosion cracking (SCC) has been a major problem in 233
Corrosion in Refinery and Petrochemical Industries
many chemical processes. The knowledge that chromium and molybdenum improved the localized corrosion resistance and increased nickel enhanced the chloride SCC resistance led to different alloys with varying nickel, chromium, molybdenum and iron contents. The knowledge that in certain low nickel containing alloys nitrogen could be added to impart certain unique mechanical, metallurgical and corrosion characteristics was used to come up with a completely new 6% Mo alloy class which was very cost-effective and in certain cases approached or equaled the corrosion resistance of the more expensive high nickel containing alloys. Two alloys of this 6% Mo stainless steel class, with varying chromium and nickel contents, were developed to bridge the performance gap between the standard austenitic stainless steel and the very high performance nickel-based alloy of the Ni-Cr-Mo family such as Alloy C-276, Alloy C-22 and the most recent and advanced development, Alloy 59. Table 1 gives some of the alloy groupings of the materials of construction used today, whereas Table 2 lists the effects of the various alloying elements in the Ni-Cr-Mo-containing alloys. As one moves upward from Type 304 stainless steel metallurgy (Group I of Table 1) to higher alloys groupings such as Groups II, IV and VI, corrosion resistance improves, as evidenced by their chemical composition and the higher calculated pitting resistance equivalent (PRE) shown in Table 3. This higher PRE accounts for the improved corrosion resistance of the various alloys in a variety of environments. For interested readers, a list of references is provided for further in depth reading to gain a better understanding of these newer developments [1-16]. Table 1. Some Austenitic Corrosion Resistant Alloys for Combating Aqueous Corrosion Alloy Group I II III IV V VI
Generic Description Iron-based 18-8 austenitic SS alloys High performance austenitic SS alloys Ni-based general purpose alloys 6% Mo superaustenitic SS alloys Ni-based special alloys Nickel based high performance alloys
VII
Chromium based high performance wrought super austenitic SS * Newer developments in the last 6 years.
Typical Alloys 304, 316, 317 904L, 20, 28, 825 200, 400, 600, 800 1925hMo, 31*, 254SMo, B-2, B-3*, B-4* G-3, G-30, 625, C-276, C-4, C-22, 59*, 686* 33*
C-FAMILY Ni-Cr-Mo ALLOYS Alloy C, the oldest alloy of this family (now obsolete), was superseded by Alloy C-276 in the early 1960's, due to improvements in melting technology. Between 1983 and 1994, three new alloys of this family were introduced to the market place: Alloy C-22 in the mid 1980's, Nicrofer 5923hMo-Alloy 59-in the late 1980's, and Alloy 686 in the early 1990's. Alloy 59 has the highest pitting resistance equivalent and the lowest iron content (Table 3), which provides for improved corrosion resistance over other alloys in a variety of standard laboratory environments, as shown in Table 4. Eliminating tungsten and reducing the iron content to very low levels, formulated an alloy with superior thermal stability characteristics, as shown in Table 5. This data clearly shows the detrimental effects of tungsten on the thermal stability of the various alloys, all of which 234
Agarwal and Herda
contain tungsten, except Alloy 59. Not only is the uniform corrosion behavior and the thermal stability improved, its localized corrosion resistance as measured by the standard ASTM G48 test method in 10% FeCl3 is also enhanced. Table 6 clearly shows the beneficial effects of higher PRE due to the highest Cr plus Mo content in Alloy 59. Further details on Ni-Cr-Mo alloys are provided elsewhere [1-3,10,11,14,15,16]. Applications of Alloy 59 Due to its highest PRE and these unique corrosion-resistant properties, Alloy 59 has found a number of successful applications, where other nickel-based alloys have been either inadequate or marginal in performance. Some of these applications are described below. Pollution Control Combustion gases from burning fossil fuels or from waste incineration of municipal or hazardous waste contain acidic pollutants such as SO2, SO3, HCl and Nox, which must be scrubbed before the gases are released into the atmosphere. The equipment most frequently employed to achieve this is a wet scrubber. Operating conditions in critical sections of wet scrubbing systems can be extremely severe: for example, in condensates chloride levels may approach 100,000 ppm at pH values below 1 and temperatures of 80°C. Laboratory and field testing has shown Alloy 59 to be one of the very few metallic materials able to withstand such aggressive corrosive conditions. In one lignite fired power station in Germany, 40 month field test rack data showed Alloy 59 to be the only alloy free of any localized attack. Table 2. Alloying Elements and Their Major Effects Alloying Element Ni
Cr Mo
W N
Cu Ti, Cb, Ta Fe
Main Features and Benefits Provides metallurgical compatibility to various alloying elements. Improves thermal stability and fabricability. Enhances corrosion in mildly reducing and alkali media, and improves chloride SCC. Provides resistance to oxidizing corrosive media. Enhances localized corrosion resistance. Provides resistance to reducing (nonoxidizing) corrosive media. Enhances localized corrosion resistance and chloride SCC. Provides solid solution strengthening. Behaves similar to Mo, but is less effective. Is detrimental to thermal stability. Provides solid solution strengthening. Austenitic stabilizer-economical substitute for nickel. Enhances localized corrosion resistance, thermal stability and mechanical properties. Improves resistance to seawater. Enhances resistance to H2SO4 and HF containing acid environments. Carbon stabilizers. Improves HAZ corrosion resistance. Provides matrix for metallurgical compatibility to various alloying elements. Enhances resistance to oxidizing corrosive
235
Corrosion in Refinery and Petrochemical Industries
media. Reduces cost by replacing nickel, and enhances scrap utilization.
Table 3. Typical Chemical Composition of Some Ni-Cr-Mo Type Alloys UNS # Alloy Ni Cr Mo Fe Others S30400 304 8 18 72 18 S31603 316L 12 17 2.3 66 24 N08904 904L 25 21 4.8 48 Cu N08020 20 38 20 2.4 34 Cu, Cb N08825 825 40 22 3.2 31 Cu N08028 28 31 27 3.5 36 Cu, Cb N08926 1925hMo 25 21 6.5 46 Cu, N-0.2 N08031 31** 31 27 6.5 32 Cu, N-0.2 N06985 G-3 48 23 7 20 Cu, Cb R20033 33** 31 33 1.6 32 Cu, N-0.4 N06625 625 62 23 9 3 Cb N10276 C-276 57 16 16 5 W N06022 C-22 57 22 13 3 W N06686 686** 56 21 16 2 W N06059 59** 59 23 16 1 * PRE = Pitting Resistance Equivalent = % Cr + 3.3 (% Mo) + 30 N ** Recent Alloy Developments
PRE*
37 29 32 38 48 54 45 50 52 69 65 74 76
Table 4. Comparison of Some Ni-Cr-Mo Alloys in Various Boiling Corrosive Environments
Alloy Media C-276 ASTM 28A 168 ASTM 28B 55 Green Death 26 10% HNO3 19 65% HNO3 750 10% H2SO4 23 50% H2SO4 240 1.5% HCl 27 10% HCl 239 10% H2SO + 1% HCl 87 10% H2SO4 + 1% HCl (90°C) 41 * To convert to mm/y, multiply by 0.0254. 236
Uniform Corrosion Rate (mpy)* Alloy Alloy C-22 686 36 60 7 12 4 8 2 52 231 18 1000 4 3 >1000 4 5 >1000 17 * To convert to mm/y, multiply by 0.0254 ** Alloys C-276, 22 and 686 - Heavy pitting attack with grains falling due to deep intergranular attack. *** Alloy 59 - No pitting attack. Table 6. Critical Pitting and Crevice Corrosion Temperature Per ASTM G-48 Alloy
Critical Pitting Corrosion Critical Crevice Corrosion Temperature (oC) Temperature (oC) 316 15 85* >85* 686 >85* >85* 59 >85* >85* * At temperatures exceeding 85°C, 10% FeCl3 chemically breaks down.
PRE 24 32 29 32 45 38 52 50 54 65 69 74 76
The municipal incinerator of Essen-Karnap in Germany originally had a scrubber with a rubber lining, installed in 1987. After some 20,000 hours of service, the rubber lining failed, when liquid permeated it and attacked the underlying carbon steel substrate. A decision to install a metal lining was taken end of 1991. Following extensive laboratory and field tests in other plants belonging to the same owners, Alloy 59 was selected for this lining. In 1992, 55 tons of alloy 59 were used for this project. Examination after two years of operation revealed no detectable loss of thickness or any localized attack of Alloy 59. A further advantage is that the quantity of deposits retained on the lining of the absorber was reduced by a factor of one 237
Corrosion in Refinery and Petrochemical Industries
thousand, which significantly reduced periodic cleaning costs. Many hundreds of tons of Alloy 59 have been ordered in recent years for flue gas desulphurization systems of both power stations and incinerators throughout the world [14]. Synthesis of Acrylates and Methacrylates One process for the synthesis of acrylic or methacrylic esters involves reacting the corresponding acids with fatty alcohols, in the presence of para-toluene sulphonic acid as a catalyst. The reaction temperature is 130°C, and the reaction is carried out under oxidizing conditions. Heating is by an internal steam coil. Following rapid failure of the material previously used for the steam heating coil, a series of plant tests was made with alloys including 904L, 28, G-3, 625, C-276, 31, and 59. The only alloy, which showed no pitting or crevice corrosion, and a corrosion rate of less than 0.01 mm/yr, was Alloy 59. A steam heating coil made of Alloy 59 was installed in 1993 and has operated without any problems ever since. Aluminum Refining When aluminum scrap is remelted, the molten metal is protected from oxidation by a layer of sodium and potassium chlorides. During the refining process this salt layer becomes contaminated with ammonium chloride. These chloride salts then have to be purified and recovered. This is done by dissolving them in water, and then recrystallizing the solution. In one European plant, the solution thus obtained contains 20-25% NaCl, 6-8% KCl and 5-8% NH4Cl. The pH is in the range 4.5 to 6. The evaporator operates at a temperature of 107°C. The initial plant was built in rubber-lined steel, and failed rapidly due to cracking of the rubber lining and subsequent corrosion of the underlying carbon steel. A plant test in 1994 with Alloy 59 showed that after some 3800 hours of operating time, no corrosion could be detected. The recystallization plant has since been rebuilt in Alloy 59. Metals Processing In a copper plant, the SO2-rich gas from the flash furnace is scrubbed with a solution of 5% H2SO4 at a temperature of 45°-60°C. The acid produced has a concentration of typically around 50-55% H2SO4 and a temperature of about 75°C. The chloride and fluoride contents of this acid are both high, at about 7000 ppm. Tests were carried out using both Alloy 59 and Alloy 31. Corrosion rates for both alloys were below 0.013 mm/yr with no localized corrosion. Following these tests, Alloy 31 was purchased for the scrubber internals handling the produced acid, and Alloy 59 for the induced draft fans. These have been in successful operation for the last two years with no detectable corrosion. Since then another order has been placed. Citric Acid Production Citric acid is produced in one plant by reacting calcium citrate with 95%-98.5% sulphuric acid at 95°C-97°C. A pilot installation of Alloy 254SMo failed rapidly. A three month test with Alloy 59 gave a corrosion rate of 0.05 mm/yr. The first of four reactors made of Alloy 59 was installed in 1990, and continues to operate well with no problems. Effluent Treatment Effluent from an acetic acid derivatives plant is cooled in Alloy C-276 plate heat exchangers. These require frequent replacement. Initial tests suggested that Alloy 59 might be a better alternative, so more extensive testing was carried out. The test conditions, and the
238
Agarwal and Herda
corrosion rates observed during testing led to the selection of Alloy 59 to replace Alloy C-276 for the new effluent treatment plant. Fine Chemicals Production At one major chemical company, the production of fluorinated organic chemicals requires a halogen exchange reaction in which one fluorine atom is substituted for chlorine in the molecule. This reaction is carried out at about 100°C in the presence of ammonium fluoride and a proprietary catalyst. Because of the severely corrosive conditions, extensive tests were made with Alloy 59 and with other alloys of the Ni-Cr-Mo and Ni-Mo family. The lowest corrosion rate was exhibited by Alloy 59. A 2600 US-gallon reactor (9.8m3) (Fig. 1) was built to ASME code requirements and has been giving excellent performance over the past 24 months. It is expected that a life cycle cost analysis will show Alloy 59 to be at least 50% cheaper than the next best candidate alloy belonging to the Ni-Mo family, Alloy B-2. Due to the excellent performance of Alloy 59, another bigger ASME vessel has now been ordered by the same chemical company (4,000 gallons capacity).
Figure 1. ASME code vessel made of Alloy 59 producing chlorinated and fluorinated chemicals Alloy 625 was giving only three years life in a column in a fine chemicals plant. The operating conditions were a temperature of 140°C and a medium consisting of 83.1% water, 14.3% sodium bisulphate, 0.34% sodium sulphate, 0.02% acetone, 0.46% isopropanol, 0.06% copper sulphate, 0.04% DCNB, and 1.5% various organics. Tests were carried out both at the inlet to the column, and at the foot of the column. Based on the results of these tests, an inquiry was issued for a new column to be built in Alloy 59. 239
Corrosion in Refinery and Petrochemical Industries
There are many other applications of Alloy 59 which continue to find increasing usage and specification in the various industries throughout the world. This alloy is covered under appropriate ASTM, AWS and ASME specifications. 6% Mo ALLOYS These alloys, such as Alloy 1925hMo, were derived from alloy 904L metallurgy by increasing the molybdenum content from 4.5% to 6.5% and fortification with 0.2% nitrogen. This addition of nitrogen provided added benefits of improved localized corrosion resistance, thermal stability and mechanical properties. These alloys are readily weldable with over-alloyed filler metals, such as Alloy 625, Alloy C-276 or Alloy 59 to compensate for segregation of molybdenum occurring in the interdentritic regions of weldments. A higher chromium-nickel version of these alloys known as Alloy 31 further improves the corrosion resistance characteristics in a variety of media. Its corrosion resistance in sulfuric acid in medium concentration range is superior to even that of Alloy C-276 and Alloy 20 (Table 7). However, one must be careful, when specifying this alloy for higher concentrations and temperatures. At 80% concentration and temperatures above 80°C, Alloy 31 exhibits active behavior. The 6Mo alloys have found extensive usage in pulp and paper, phosphoric acid, copper smelters, sulfuric acid production, pollution control, rayon production, specialty chemicals production, marine and offshore applications, heat exchangers using seawater and brackish water as coolant, pickling baths and many other applications. These alloys are covered under appropriate ASTM and ASME specifications. More details on these 6Mo alloys are presented elsewhere [4-8]. Some applications of 6Mo alloys are presented in Tables 8 and 9. Table 7. Corrosion Resistance in Sulfuric Acid
60°C H2SO4 Alloy Alloy Alloy Alloy (%) 20 C-276 31 20 8) differed from its behavior in acidic or neutral solutions (pH 1-8). Open-circuit potential measurements revealed that the steady-state potential was a linear function of the solution’s pH. The slope of the linear relation changed from 0.039 V/decade for solutions of pH < 8 to -0.058 V/decade for solutions of pH > 8. Electrechemical impedance spectroscopy, which represent an effective method for studying corrosion phenomena, has shown that the metal undergoes active dissolution in aqueous media. In solutions of pH > 8, the corrosion behavior of the metal can be simulated to a Randles equivalent circuit model. In oxygen-saturated solutions, the electrode’s surface is covered with a thin oxide film. The interaction of this film with the ambient electrolyte depends on the solution’s pH. Polarization measurements have shown that the rate of corrosion in acidic solutions is not affected by the prevailing gas. In alkaline solutions, the removal of air or oxygen from the ambient electrolyte leads to a decrease in the corrosion rate. Key Words: Corrosion, electrochemistry, impedance, passivity, polarization, vanadium
INTRODUCTION Vanadium is an important transition metal due to its use and the use of its alloys as structural materials not only in metallurgical applications but also in nuclear reactors [l]. The metal ions are used extensively in redox flow batteries [2-5]. Unlike many transition metals, vanadium shows active behavior [6-10]. The electrode potential of vanadium and the effect of oxygen on this potential were the subjects of the very early studies concerning this metal [11,12]. The active behavior of vanadium and its anodic dissolution in aqueous solutions was investigated [13,14]. It was found that the rate of metal dissolution is independent of the hydrogen ion concentration or the nature of the anions present. The rate determining step was + a monovalent adsorbed intermediate (V ads) [14]. Investigations of the anodic behavior in acidic solutions containing different anions and cations have shown that the metal undergoes 2+ active dissolution in all acidic solutions except those containing Ba at a pH > 4, and that the active dissolution of the metal occurs through a monovalent intermediate [15]. The behavior of the metal in both acidic and basic media was found to obey the Tafel approximation of the Butler-Volmer equation over a wide range of potentials [16]. Extensive studies have been 383
Fundamental Aspects
carried out in glacial acetic acid and have shown oxide film formation in the presence of sodium borate and water [17,18,19]. The present investigation was aimed at throwing more light on the electrochemical behavior of vanadium in aqueous solutions and on the effect of solution pH and oxygen on the corrosion and passivation processes occurring at the electrode/electrolyte interface. EXPERIMENTAL PROCEDURE Massive, cylindrical, spectroscopically pure, vanadium rods (Alderich-Chemie) were used as working electrodes; they were mounted in glass tubes of appropriate internal diameter 2 with an epoxy resin leaving a front surface area of 0.302 cm to contact the electrolyte. An all glass three electrode cell with a large surface area, Pt, counter electrode and a Ag/AgCl/Cl (3 M KCl) reference electrode was used. The solutions were prepared from analytical grade reagents and triple distilled water. The buffer solutions covering a pH range of 1-13 were prepared according to Clark and Lub's series [20]. The pH of each solution was controlled before each experiment, and the electrodes were mechanically polished with successive grades of emery papers down to 1200 grit, then wiped with a smooth cloth and washed with triple distilled water. In this way, the electrodes acquired reproducible, bright silvery surfaces. After polishing and rinsing, the electrodes were immersed directly in the test solutions. Electrochemical impedance spectroscopy (EIS) investigations were carried out using the IM5d-AMOS system (Zahner Elektrik GmbH & Co., Kronach, Germany). The 5 input signal was usually 10 mV peak to peak in the frequency domain of 0.1-10 Hz. Frequencies down to 0.1 MHz were also investigated. Polarization measurements were performed using an EG&G (Princeton Applied Research) model 273A Potentiostat/Galvanostat interfaced to an IBM PS3 computer. The potentiostatic measurements were traced using programs that enable ohmic drop compensation. The steady-state potential, ESS, measurements were controlled separately using a high impedance value voltmeter (Keithley type 197A Autoranging Multiplier, England). All measurements were carried out at constant room temperature of 22oC. The potentials were measured against the Ag/AgCl/Cl- reference, and then refered to the normal hydrogen electrode (nhe). The gases used for deaeration or solution saturation were purified and dried before being bubbled in the electrolytic cell. The gas was bubbled for at least 20 min. in the test solution prior to each experiment. The details of the experimental procedures were as described elsewhere [10,21]. RESULTS AND DISCUSSION Steady-State Potential Measurements The potential of the vanadium electrode was traced over a period of 180 min. in naturally aerated aqueous solutions covering a pH range from 1 to 13. In all solutions, the electrode potential became more positive with time. A typical example of the variation of the electrode potential with immersion time in the buffer solutions is presented in Fig. 1A. In this figure, the results in solutions of pH’s 2, 7 and 12 are presented. The steady-state potential, Ess , measured after 180 min. of electrode immersion in each solution is plotted against the pH of the solution, and is presented in Fig. 1B. The results reveal that ESS is pH-dependent over the
384
Badawy et al.
whole pH range. The Ess versus pH relation is linear and can be presented by an empirical equation of the form: Ess = a - b pH
(1)
where a is a constant representing the value of ESS extrapolated from the linear relation at pH = 0. As can be seen from Fig. 1B, the ESS versus pH relation has an inflection between pH = 8 and pH = 9 which means that the slope, b, changes. In the basic medium, i.e., pH > 8, the slope of the linear relation is very close to the value of 0.059 V/pH, which was calculated from the Nernst equation for a pH indicator electrode with one electron electrode process at o 25 C according to: o
E = E - 0.059 pH n
(2)
o
E is the standard pH-independent electrode potential and n is the number of electrons involved in the electrochemical process. Therefore, the value of n for basic solutions (pH > 8) is equal to 1. In acidic and neutral solutions (pH < 8), on the other hand, the slope of the Ess versus pH relation is < 0.059 V/pH. A value of 0.039 V/pH was calculated. This can be + explained by the interaction of both H and OH ions with the electrode’s surface which is essentially covered with a passive film or an adsorbed O2 film as will be discussed later. In regions where the OH ions are inaccessible to the electrode surface, solvation of the electrode occurs, and deviation from the n = 1 process is observed [22]. The vanadium electrode can be considered to be a pH indicator electrode taking into account the inflection in the ESS versus pH relation and the values of a and b in each pH in the range 1-8 Ess = 0.180 - 0.039 pH
(3)
Ess = 0.380 - 0.058 pH
(4)
and in solutions of pH’s > 8
The steady-state potential values obtained according to Eqs. 3 and 4 are quite different 2+ from the standard electrode potentials assigned for the systems V/V (Eo = -1.19 V) and 2+ 3+ o V /V (E = - 0.26 V) [22]. This supports the notion that the electrode process cannot be represented by a simple equilibrium relationship such as that given for the simple redox 2+ 2+ 3+ equilibria of the metal and its ions, e.g., the V/V or V /V Systems. This can be understood on the basis that the solution does not contain significant concentrations of those ionic species, and that the electrode’s surface is covered by a thin oxide film in aqueous solutions. Consistent with this is the dependence of the steady-state potential on the prevailing gas and the stirring conditions of the solution.
Effect of Oxygen on the Steady-State Potential 385
Fundamental Aspects
The electrode’s potential was traced in naturally aerated, oxygen-saturated and oxygenfree solutions of different pHs. Oxygen was removed from the solution by bubbling N2 or H2 at least 20 min. before electrode immersion. A typical example of the results in solutions of pH = 12 is presented in Fig. 2. The results reveal that the electrode’s potential is sensitive to the oxygen concentration in the solution. Under all conditions, the potential became more positive with immersion time until it reached a steady-state value. The use of either N2 or H2 to remove oxygen from the solution did not produce any remarkable difference; the steadystate potential lay, in both cases, in approximately the same range (≈ -385 mV (nhe) in pH = 12). In oxygen-saturated solutions, the Ess shifted in the positive direction by ≈ 100 mV. This shift can be attributed to the presence of a thin oxide film on the electrode’s surface. The same trend was observed over the whole pH range from 1 to 13.
Figure 1. (A) Variation of the electrode potential Figure 2. with time of the vanadium electrodes in naturally aerated solutions of different pHs (o) pH = 2 (∗) pH = 7 (Δ) pH = 12 (B) Steady-state potential (Ess) vs. pH for the vanadium electrode in naturally aerated buffer solutions
Effect of the prevailing gas on the electrode potential of vanadium in solutions of pH 12 (o) naturally aerated (∗) O2-saturated (Δ) N2-saturated (Δ) H2-saturated
The sensitivity of metals with active/passive transitions towards oxygen is well known, especially for those which do not show active dissolution like niobium, tantalum and titanium [10,21,24]. The passivation of vanadium in aqueous solutions was discussed very early by 386
Badawy et al.
Muthman and Frauenberger [11]. Later, it was suggested that the passivity is due to the presence of a gaseous film [12]. Unlike many transition metals, vanadium has an active corrosion behavior with a limited tendency for passivation. The passivation behavior of the metal was explained earlier by the presence of a chemisorbed oxygen film [25]. Dry or moist o oxygen did not tarnish the polished metal’s surface, and even hot water (60-85 C) had no effect on the surface’s brightness. The decrease of the steady-state potential on oxygen removal by bubbling of N2 or H2 in the test solution can be considered as an indication of the presence of a thin oxide film on the electrode’s surface. The oxide film formed represents a non-stoichiometric oxidation state of the metal which is responsible for the observed behavior of the metal in each solution. Removal of oxygen by bubbling of N2 or H2 in the solution leads to dissolution of the formed oxide, and hence, a decrease of its thickness leading to a shift of the steady-state potential in the negative direction (Fig. 2), and the following equilibrium state is shifted to the left: 2V + xO2 → 2VOx
(5)
The value of x determines the stoichiometric factor n of Eq. 2. In acidic solutions, where an excess of hydrogen ions are present, the interaction between the non-stoichiometric oxide film and the hydrogen ion takes place and the electrode’s potential is determined by the hydrogen ion concentration according to: +
-
V-Ox + 2x H + 2xe → V + x H2O
(6)
The value of x in this case, and hence, the value of n of Eq. 2, can be calculated from the slope of the first segment of the steady-state potential/pH relation (Fig. 1B), i.e., in the pH range of 1-8. slope = -0.039 V = - 0.059 n
where n = 2x i.e., n = 1.5
In basic solutions, the interaction between the electrode’s surface and the solution occurs through OH ions according to: -
V- Ox + 2OH → VO1+X + H2O + 2 e
-
(7)
The lower oxides of vanadium are basic and very unstable [25]; therefore, they cannot protect the metal from corrosion as in the case of the valve metals with active/passive transitions.
Open-Circuit Impedance Measurements EIS is a powerful tool for investigating electrochemical and corrosion systems, since it is essentially a steady-state technique that is capable of accessing relaxation phenomena with 387
Fundamental Aspects
relaxation times that vary over several orders of magnitude and permits single averaging, within a single experiment to obtain highly precise levels. The open-circuit impedance of vanadium electrodes was traced for 180 min. after electrode immersion in the test solutions. Typical data for pH’s of 2, 7 and 12 are presented as Bode plots in Fig. 3.
(A) in solution of pH 2
(B) in solution of pH 7
(C) in solution of pH 12 Figure 3. Bode plots of the vanadium electrodes in naturally aerated solutions at different time intervals from electrode immersion (⎯) 15 min. (…) 60 min. (----) 130 min. Bode plots are recommended as standard impedance plots since the phase angle, θ, is a sensitive parameter for indicating the presence of additional time constants in the impedance spectra [10,26-28]. It employs frequency as an independent variable, so that a more precise comparison between experimental and calculated impedance spectra can be made [29-31]. The use of the log versus log format enables equal representation of all experimental data over the whole frequency domain. 388
Badawy et al.
The EIS spectra in Fig. 3 contain only one capacitive contribution represented by the linear variation of the electrode’s impedance, Z, with the frequency, f, [26-28]. In acidic and neutral solutions (pH = 1-8), there is a part of the spectrum where the phase angle is independent of frequency (Fig. 3a and b at f < 1 Hz). Such behavior is explained by the assumption of frequency dispersion and surface inhomogeneity [29]. For such systems, the electrode impedance is given by: Rp Z = ⎯⎯⎯⎯⎯ α I + (sCRP)
(8)
where RP is the polarization resistance which is considered to be a pure charge transfer resistance, C is the electrode capacitance and α is a fit parameter ( 0 < α < 1 ) that is correlated to the angle of rotation of the center of the capacitive semicircle, φ, below the real axis: φ = (1- α) π/2 s = j ϖ where j =
(9)
−1 and ω = 2 π f.
The value of the fit exponent α corresponds to the extent of dispersion and is attributed to surface inhomogeneity [29,30]. A nonlinear concentration of metal ions will occur, since preferential charge transfer takes place at active sites. The impedance spectra of Fig. 3 show that there is an active dissolution of the metal, as can be identified by the decrease of the polarization resistance with immersion time in each solution. The rate of dissolution is limited and is dependent of the solution’s pH. The impedance spectra of the electrodes in solutions of pHs of 2, 7 and 12 after 180 rnin. of electrode immersion are collectively presented in Fig. 4. The results show clearly that the behavior of vanadium in basic solutions is different from its behavior in acidic or neutral solutions. In solutions of pH 12, a phase maximum at a frequency of ≈1 Hz was observed. To be sure that there was no second time constant at lower frequencies, impedance measurements down to frequencies of 1 MHz were taken. A typical example of these measurements for solutions of pH 12 is presented in Fig. 5. The behavior of the electrode in basic solutions is very similar to an ideally corroding system with a single time constant corresponding to the corrosion reaction rate determining step. This behavior can be simulated by a simple equivalent circuit model of the Randles type which consists of a parallel combination of a resistor, RP , representing the polarization or charge transfer resistance and a capacitor, C , representing the capacitance of the electrode/electrolyte interface. This parallel combination is in series to a small resistor, Rs , equivalent to the electrolyte resistance. The data in Fig. 5 were subjected to a procedure of data fitting [31] to fit the experimental data in this figure to the electronic model described. Good agreement was obtained between the experimental and theoretical data for values of RP = 10.22 kΩ, C = -2 170.4 μF cm , and Rs = 42 Ω. The fitting procedure for the results of solutions of pH of 12
389
Fundamental Aspects
0
had around 2% mean error in the absolute impedance and ≈1.2 mean error in the phase angle. Deviations of the phase angle from ideal behavior were found to be related to the polishing of the electrode’s surface. The impedance data presented show that the vanadium electrode, although it showed active dissolution, it had a great tendency towards oxygen and the surface was covered with a thin film of non-stoichiometric oxide or a mixture of oxides of varying valences. The instability of such oxides explain the corrodibility of vanadium’s surface. The increased rate of corrosion with increases in the pH of the solution was varied by measuring the corrosion currents and polarization resistance in each solution. The values of these parameters in pHs 2, 7 and 12, are presented in Table 1.
Figure 4. Bode plots of vanadium electrode after Figure 5. 130 min. immersion in naturally aerated solutions of pH 2 (…), pH 7 (----) and pH 12 (⎯)
Bode plot of the vanadium electrode in naturally aerated solution of pH = 12 in the -3 5 frequency range 10 to 10 Hz
Table 1. Values of the Polarization Resistance, RP , Corrosion Current, icorr, and Corrosion Potential Ecorr, of the Vanadium Electrode in Naturally Areated Solutions of Different pHs pH 2 7 12
RP 2 (kΩcm ) 5.012 2.880 1.652
icorr -2 (μAcm ) 0.407 1.239 3.930
Ecorr (mV) -215 -407 -597
The effect of the prevailing gas on the impedance behavior of the vanadium electrode was also investigated. An example of these measurements in solutions of pH 12 is presented in Fig. 6. After 3 hrs of electrode immersion, the impedance behavior in H2 or N2 saturated solutions is quite similar. In air or oxygen-saturated solutions, the measured polarization resistance is lower. Polarization measurements have shown that the corrosion current 390
Badawy et al.
decreases in the presence of an inert gas in basic solutions. In acidic solutions, on the other hand, the corrosion rate is not much affected by the inert gases. These results are summarized in Table 2. The results of polarization experiments are in good agreement with the + explanation based on Eqs. 6 and 7. In acidic solutions, the interaction of the excess H ions with the surface film leads to the removal of this film without any appreciable change in the rate of corrosion of the metal by changing the gas. In basic solutions, the formation of the non-stoichiometric, unstable, basic oxides is responsible for the increased rate of corrosion of the metal. The formation of the oxide film and its stoichiometry is dependent of the presence of air or O2 in the solution. Removal of air or oxygen from a basic solution shifts the equilibrium of Eq. 5 to the left, and hence, decreases the rate of corrosion as can be seen from the values of the corrosion currents presented in Table 2. The presence of the surface film is confirmed by capacitance measurements. In all solutions, the electrode capacitance showed an approximately constant value within a wide potential range (-100 - +100 mV from the -2 steady-state potential). In acidic solutions, an average capacitance value of 25 μFcm was measured. This value is higher than the reported value of the Helmholtz capacitance (17 -2 μFcm ) [32]. The higher capacitance value can be attributed to the presence of adsorbed electroactive species on the surface film. The concentration of these electroactive species at the electrode’s surface is constant in the potential range where the electrode capacitance is potential-independent.
Figure 6. Effect of the prevailing gas on the impedance characteristics of the vanadium electrode in solutions of pH 12 (- -) oxygen, (⎯) nitrogen, (----) naturally aerated, (....) hydrogen Table 2. Corrosion Currents of Vanadium in Solutions of pHs 2, 7 and 12 Saturated with Different Gases Gas
-2
icorr (μA-cm )
2
Rp (kΩcm ) 391
Fundamental Aspects
Air Oxygen Nitrogen Hydrogen
pH 2 0.407 1.057 1.578 1.046
pH 7 1.239 2.939 1.967 1.082
pH 12 3.930 17.50 1.710 0.928
pH 12 1.652 1.596 2.352 2.072
Effect of Temperature on the Corrosion Behavior of Vanadium To study the effect of temperature on the corrosion behavior of vanadium, an all glass, double walled cell was used with the same arrangement of counter, reference and working electrodes. The measurements were made in naturally aerated solutions of pHs 2, 7 and 12. In all solutions, the general trend was an increase in the rate of corrosion with increasing temperature. Polarization measurements were taken at each temperature, and the corresponding corrosion current, iCorr , which represents the rate of corrosion, was obtained. A plot of log icorr versus 1/T obeys the familiar Arrhenius equation [33]. d logicorr = Ea 2 dT RT
(10)
where Ea is the activation energy which is given by Ea = NA εa
(11)
εa is the energy relative to the ground state energy which an atom or molecule must have in order to react, i.e., the activation energy per molecule, whereas Ea is the molar activation energy. Figure 7 presents the Arrhenius plots obtained in solutions of pH 2, 7 and 12. In the solutions investigated, almost parallel Arrhenius plots were obtained, which means that the activation energy of the corrosion process lies in the same range without regard to the solution’s pH. Calculation of the activation energy of the corrosion process in each solution gave the values presented in Table 3. The values given in Table 3 show that the activation -1 energy of the corrosion process is less than 40 kJmol , which supports the view that the dissolution of the metal is a one-electron charge transfer process [34]. This supports the mechanism suggested by Armstrong and Henderson [14], and the corrosion reaction may be presented by V(s)
⎯ ⎯→
V(I) + e 2+
-
(12) +
-
H2O + V(I)ads. ⎯fast (13) ⎯ ⎯→ VO +2H +3e This reaction is enhanced in basic media, and hence, the rate of corrosion increases. In accordance with this, the calculated activation energy in solutions of pH 12 is slightly lower.
Table 3. Activation Energy of the 392
Badawy et al.
Corrosion of Vanadium in Naturally Aerated Solutions of pHs 2, 7, and 12 pH 2 7 12
-1
Ea, (kJ mol ) 34.8 36.4 30.6
Figure 7. Log icorr vs 1/T relations for the corrosion behavior of vanadium in naturally aerated solutions of pH 2 (o), pH 7 (∗) and pH 12 (Δ) CONCLUSIONS The steady-state potential of vanadium is sensitive to the solution’s pH and can be used for pH calculations. The rate of corrosion of the metal in basic media decreases with the removal of air or oxygen. In acidic or neutral solutions, the prevailing gas has no significant effect on the rate of the corrosion process. Activation energy calculations support a oneelectron transfer step as the rate-determining corrosion process. ACKNOWLEDGMENT This work has been supported by Kuwait University, Research Grant No. SCO59. The financial support of the research administration is gratefully acknowledged. REFERENCES 1.
W.J. Tomlinson, R. Rushton, R. Cindery and S. Palmer; J. Less Common Met. 132, 1987, p. 1. 2. E. Sum and M. Skyllas-Kazacos, J. Power Sources 15, 1985, p. 179. 3. E. Sum, M. Rychcik and M. Skyllas-Kazacos, J. Power Sources 16, 1985, p. 35. 4. M. Rychcik and M. Skyllas-Kazacos, J. Power Sources 19, 45, 1987, p. 45. 5. M. Kazacas and M. Skyllas-Kazacos, J. Electrochem. Soc. 136, 1989, p. 2759 6. T. Hurlen and W. Wilhelmsen, Electrochim. Acta 31, 1986, p. 1139. 7. W.Wilhelmsen and T. Hurlen, Electrochim. Acta 32, 1987, p. 85. 8. S. Hornkjol, Electrochim. Acta 33, 1988, p. 337. 9. W.A. Badawy, J. Appl. Electrochem. 20, 139, 1990, p. 139. 10. W.A. Badawy and K.M. Ismail; Electrochim. Acta 38, 1993, p. 2231. 11. W. Muthman and F. Fraunberger, Sitzbl. Bayr. Akad. Wiss., 1904, p. 201. 12. G.C. Schmidt, Z. Phys. Chem. 106, 1923, p. 105. 393
Fundamental Aspects
13. R. Kammel, T.Kishi, T.Takei and H. Winterhager, Metalloberflaesche 24, 1970, p. 335. 14. R.D. Armstrong and M. Henderson, J. Electroanal. Chem. 26, 1980, 381. 15. S. Homkjol and I.M. Homkjol, Electrochim. Acta 36, 1991, p. 571. 16. A. Deschanvers and G. Nouet, Bull. Soc. Chim. Fr. 718, 1975, p. 1589. 17. A.G. Keil and R.E. Salomon, J. Electrochem. Soc. 112, 643, 1965, p. 643 and 115, 1968, p. 628. 18. A.G. Keil and R.E. Salomon, J. Electrochem. Soc. 115, 1968, p. 628. 19. R.G. Keil and K. Ludwig, J. Electrochem. Soc. 118, 864, 1971, p. 864. 20. G.D. Fasman, Practical Handbook of Biochemistry and Molecular Biology, CRC Press Inc., Boca Raton, Florida, 1989. 21. F.M. AI-Kharafi and W.A. Badawy, Electrochim. Acta 40, 1995, p. 2623. 22. A. Belanger and K. Vijh Ashok, J. Electrochem. Soc. 121, 225,1974, p. 225. 23. P.W. Atkins, Physical Chemistry, 5th Ed., Oxford University Press, Oxford, 1994. 24. W.A. Badawy, A. Felske and W.J. Plieth, Electrochim. Acta 34, 1989, p. 1771. 25. M. Pourbaix Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon Press, London, UK, 1966, pp. 234-245. 26. A.J. Brock and G.C.Wood, Electrochim. Acta 12, 395, 1967, p. 395. 27. D.D. Macdonald, S. Real, S.I. Smedley and M. Uraquidi-Macdonald; J. Electrochem. Soc. 135, 2410, 1988, p. 2410. 28. C.M.A. Brett, Corros. Sci. 33, 1992, p. 203. 29. K. Juttner, W.J. Lorenz, M.W. Kendig and F. Mansfeld, J. Electrochem. Soc. 135, 1988, p. 322. 30. K. Juttner, Electrochim. Acta 35, 1990, p. 1501. 31. W.A. Badawy, S.S. El-Egamy and K.M. Ismail, British Corros. J 28, 1993, p. 133. 32. J.O.M. Bockris and A.K.N. Reddy, Modern Electrochemistry, 2nd ed., Chap. 7, Plenum Press, New York, USA, 1977. 33. R.G. Martimer, Physical Chemistry, The Benjamin/Cummings Publishing Company Inc., Redwood City, California, USA, 1993. 34. G.A. Wright, J. Electrochem. Soc. 114, 1967, p. 1263.
394
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
THE EFFECT OF UV-IRRADIATION ON PASSIVE FILMS FORMED ON TYPE 304 AND 316 STAINLESS STEELS M.S. Al-Rifaie, C.B. Breslin, D.D. Macdonald and E. Sikora Center for Advanced Materials, The Pennsylvania State University, 517 Deike Building, University Park, Pennsylvania, 16802, USA
ABSTRACT The effect of monochromatic ultraviolet (UV) light on the passive films formed on Types 304 and 316 stainless steels (SS) is described. Under UV irradiation 304SS and 316SS specimens, in neutral and acidic solutions, exhibited an increased resistance to localized corrosion (pitting). This resistance to localized corrosion was gauged by an increase in induction time, an increase in breakdown potential, and some significant changes in the current noise at constant potentials. All these changes indicate that shining UV light on a metal specimen can sometimes decrease its susceptibility to pitting. It was observed that the extent of photoinhibition of localized attack (PILA) depends on the nature of the passive film, the period of illumination, and the incident photon energy. It was also observed that the PILA effect can, in some cases, last for over 200 hours after the illumination has been removed. Increased pitting resistance was observed with higher energy incident photons. The minimum apparent incident photon energy corresponds to the band gap of the metal specimen (375 nm for SS). The optimum illumination period observed in these experiments was approximately 5 hours. In alkaline solutions a much decreased PILA effect was observed, this decrease in PILA was attributed to the formation of a precipitate layer. This precipitate layer in turn interfered with the incident photon interaction with the barrier layer. Finally, a possible explanation of PILA is given within the framework of the Point Defect Model (PDM). Key Words: Photoinhibition, stainless steel, localized corrosion, pitting, UV, PDM, passive films
INTRODUCTION The initiation and propagation of pits and the corresponding breakdown of the passive film that forms on metals and alloys is of great fundamental interest in electrochemistry and corrosion science. Numerous efforts, ranging from the addition of inhibitors [1] to the alloying of base metals [2-5], have been made in an attempt to prolong the effective life of the passive film. Recently, it has been shown that irradiating immersed electrodes with ultraviolet (UV) light can inhibit localized corrosion. The first observation of this kind was made when polycrystalline nickel, in chloride-containing solution, was irradiated with white light [6]. Other observations supporting photoinhibition of localized attack (PILA) include similar effects on 304 stainless steel (SS) [7] and even pure iron under UV irradiation [8-9]. The purpose of this study is to examine the effects of UV irradiation on the photoinhibition of 304SS and 316SS in chloride-containing solutions. These results could help shed light on
395
Fundamental Aspects
how PILA affects SSs and lay the groundwork for photoinhibition as a new form of corrosion control. EXPERIMENTAL PROCEDURE Test specimens were prepared from Types 304 and 316 SS rods, which were covered with lacquer, mounted in a PVC holder and then embedded in a two-component epoxy resin. The exposed surface, approximately 0.8 cm2 in area, was polished mechanically with successively finer grades of SiC paper and 0.05 μ alumina powder to a mirror finish. The chemical composition of the SS samples used is shown clearly in Table 1. Table 1. Chemical Composition of 304 SS and 316 SS (in Wt %) SS Type C 304 0.08 316 0.08
Mn 2 2
Cr 18 16
Ni 8 12
P S 0.04 0.003 0.04 0.003
Si 1 1
Mo Trace 2
Fe Bal. Bal.
The electrochemical cell consisted of a three-electrode PTFE cell equipped with a quartz window to allow irradiation of test electrodes. A saturated calomel electrode (SCE) was used as the reference electrode, and a platinum wire, coiled inside the cell, was used as the auxilliary electrode. All test solutions were prepared from Analar-grade reagents and deionized water, and were deoxygenated with nitrogen. The pH of the solution was adjusted to 7.5 with NaOH, or alternatively, buffered to pH 7.5 with a 0.15 mol dm-3 H3BO3/0.007 mol dm-3 Na2B4O7 solution. B
The working electrodes were irradiated at wavelengths between 300 and 425 nm using a 150 W UV-enhanced Xe lamp (Oriel Model 6254) and a 1/8 monochromator (Oriel Model 77250). The incident power density at 300 nm was 0.4 mW cm-2, giving a photon flux of 6.04 x 1014 cm-2. The photon flux was maintained at approximately this value at each wavelength by adjusting the light intensity at the surface. Electrochemical tests were carried out using a Solartron/Schlumberger Electrochemical Interface (Model 1286). In potentiodynamic polarization tests, the working electrodes were polarized at a rate of 0.1 mV s-1 in the anodic direction up to the breakdown potential. In illumination experiments, the electrodes were illuminated continuously throughout the potential scan. The breakdown potential was recorded as the potential at which the current exceeded 80 μA cm-2. In current-time measurements, the electrodes were initially polarized at a potential in the passive region for a 30-minute period, and then the potential was stepped to an appropriate point where metastable pitting could be observed for the non-illuminated specimens. The current transients were then recorded as a function of time, using a Keithley Model 576 data acquisition unit at a sampling rate of 90 mS. Additional experiments involved polarizing the working electrodes, under illumination for periods of up to 15 hours, and then determining the breakdown potential using the potentiodynamic polarization method. The exact same polarization periods were used for the illuminated and non-illuminated electrodes.
396
Al-Rifaie et al.
RESULTS AND DISCUSSION Figure 1 shows typical anodic polarization curves for type 304 SS in a neutral 0.5 mol dm-3 NaCl solution (unbuffered) under conditions of illumination and non-illumination. Figure 1 clearly shows that when the sample was illuminated, an increase in pitting resistance resulted. This increase in pitting resistance was indicated by a shift of both the breakdown potential and the initial metastable pitting potential towards the more noble direction.
Figure 1. Potentiodynamic polarization curves for type 304 SS in neutral 0.5 mol dm-3 NaCl under: (a) non-illumination; and (b) illumination at 300 nm The effect of illumination on the breakdown potentials of 304 SS and 316 SS can perhaps be seen more clearly from the data shown in Table 2, which shows averages of the breakdown potentials for 304 SS and 316 SS as a function of chloride concentration. In each case, an average increase of about 60 ± 40 mV in the breakdown potential can be observed upon illumination. Another observation that can be made from the data is that the breakdown potentials of the illuminated 304 SS specimens approach those of the 316 SS specimens in the dark. This observation could suggest a comparable degree of passivity enhancement between alloying and illumination under these conditions. An even greater increase in the breakdown potential, approximately 150 ± 50 mV, was observed on prior illumination of the specimens at 300 nm for periods exceeding 5 hours. In these experiments, the electrodes were polarized in a 0.1 mol dm-3, NaCl buffered solution at +250 mV (SCE) for various periods of time under illumination and non-illumination. The specimens were then polarized from +250 mV (SCE) in the anodic direction at a rate of 0.1 397
Fundamental Aspects
mV s-1 up to the breakdown potential. The displacement in the breakdown potential was calculated by subtracting the average breakdown potential for the specimens polarized in the dark from the average breakdown potential for the specimens polarized in the light. Each experiment was repeated three times. This data are shown graphically in Fig. 2, where the average displacement in the breakdown potential, ΔEb is shown as a function of the prior illumination period. Table 2. Breakdown Potential Values for 304 SS and 316 SS Under Light and Dark Conditions [Cl-] mol dm-3 304 SS Eb (dark) 304 SS Eb (light) 316 SS Eb (dark) 316 SS Eb (light) 0.025 355 ± 7 mV (3) 420 ± 10 mV (3) 430 ± 10 mV (3) 490 ± 11 mV (4) 0.5 275 ± 18 mV (20) 350 ± 20 mV (22) 330 ± 20 mV (17) 395 ± 20 mV (17) 2.0 160 ± 8 mV (4) 210 ± 15 mV (5) 230 ± 12 mV (5) 290 ± 10 mV (4) Eb in mV vs. SCE Light = 300 nm The numbers in parentheses indicate the number of times the experiment was repeated
Figure 2. Displacement in the breakdown potential, ΔEb, as a function of the illumination period for 316 SS at +250 mV (SCE) in a buffered 0.1 mol dm-3 NaCl solution
398
Al-Rifaie et al.
Further evidence for photoinhibition of pitting attack was obtained from current-time measurements where the current decay transients were monitored as a function of time for illuminated (300 nm) and non-illuminated 316 SS. These data are shown in Fig. 3. It is evident from Fig. 3 that illumination causes a delay in the onset of metastable pitting, even after approximately 110 minutes, the illuminated film was still able to repassivate while the dark film had started breaking up after only 25 minutes. Therefore, illumination seemed to postpone metastable pitting by at least a factor of 3, which in itself is quite astounding. One of the questions that this study aims to answer is how prior illumination affects the passive film on 304 SS and 316SS. In order to quantify the permanent nature of the photoinhibition effect, the pitting susceptibility of 316SS was studied at various periods of time with prior illumination. The specimens were illuminated at 300 nm for 80 minutes under polarizing conditions in a neutral 0.5 mol dm-3 NaCl solution. The specimens were then immersed under open-circuit conditions (dark) in a borate buffer solution (pH of 7.5), and removed at selected intervals. The breakdown potential was determined in a neutral 0.5 mol dm-3 NaCl solution using the potential scan method. Identical experiments were carried out in the dark; the specimens were polarized in the chloride solution (dark) for 80 minutes, removed and immersed in the borate solution. The breakdown potential was determined at selected intervals. Data collected in this manner for periods up to 350 hours are shown in Fig. 4, where the breakdown potential is plotted against the immersion period following illumination or polarization.
399
Fundamental Aspects
Figure 3. Current-time decay profiles for 316 SS polarized at 285 mV (SCE) in a neutral 0.025 mol dm-3 NaCl solution under: (a) dark, and (b) light (300 nm) conditions
Figure 4. Breakdown potential of illuminated and non-illuminated 316SS in 0.5 mol dm-3 NaCl as a function of the immersion period in a borate buffer solution (dark) following polarization or polarization and illumination at 300 nm. (a) on a linear scale; and (b) on a semi-log scale A clear difference between the breakdown potentials measured for the illuminated and non-illuminated specimens can be seen for immersion periods up to about 220 hours, indicating that the photoinhibition effect persists over this period of time. The gradual increase in breakdown potential (i.e., a shift towards the more noble direction) may be attributed to a crystallization process or chromium-enrichment in the passive film. A similar trend was observed for 304SS. The influence of solution pH, and thus the nature of the passive film, on the extent of photoinhibition was studied by polarizing and illuminating the electrodes in solutions of varying acidity. A 0.5 mol dm-3 NaCl solution was used as the test solution; the pH was adjusted to the desired value by the addition of NaOH or HCl. All irradiation experiments were carried out at 300 nm. The breakdown potentials were determined from polarization measurements for specimens polarized in the dark and under conditions of continuous 400
Al-Rifaie et al.
illumination. Each experiment was carried out at least three times. The displacement in the breakdown potential, ΔEb, was calculated as the difference between the light and dark breakdown potentials. The average displacements in the breakdown potentials for 304SS and 316SS are shown as a function of pH in Fig. 5; the degree of scatter in the average displacements was ± 30 mV. An essentially constant increase in the breakdown potentials, of approximately 60 mV, was observed on illumination, except for those specimens polarized in the alkaline solutions, where no apparent photoinhibition effect was detected. However, it was found that the photoinhibition effect was partially restored under these alkaline conditions (pH of 10) by the addition of a 0.01 mol dm-3 EDTA solution to the test solution. The pH in this region was adjusted with NaOH and maintained at 10 on addition of the complexing EDTA agent. The average displacements in the breakdown potentials for both 304SS and 316SS at a pH of 10 on addition of the EDTA were 45 and 50 mV, respectively. The presence of EDTA at other pH values did not enhance the photoinhibition effect. This seems to suggest that the precipitated layers formed in alkaline environments are photoelectrochemically inactive, but that the addition of a chelating agent hinders the formation of this layer, allowing the photons to reach the barrier layer.
Figure 5. Displacement in the breakdown potential, ΔEb, as a function of the pH of a 0.5 mol dm3 NaCl solution, on illumination of: (a) 304SS; and (b) 316SS at 300 nm 401
Fundamental Aspects
The effects of variations in the photon energy on the breakdown potential displacement (a measure of photoinhibition) are shown graphically in Fig. 6. A constant photon flux was maintained at each wavelength. A neutral 0.5 mol dm-3 NaCl solution was used as the test solution. A total of twenty experiments were carried out for each of the SSs under conditions of non-illumination in order to obtain adequate reference breakdown potentials. The amount of scatter in the breakdown potentials, under these conditions, was on the order of ± 20 mV. The mean value of the breakdown potential calculated for 304SS in the dark was 272 mV (SCE), while that for 316SS in the dark was 327 mV (SCE). Displacements in the breakdown potential, ΔEb, on illumination were calculated as the difference between the breakdown potentials in the light and the dark. It can be seen from Fig. 6 that the degree of photoinhibition depended on the photon energy, with the photoinhibition effect decreasing with wavelengths exceeding 375nm.
Figure 6. Displacement in the breakdown potential, ΔEb, measured in a 0.5 mol dm-3 NaCl solution, as a function of the incident light wavelength on illumination of: (a) 304SS; and (b) 316SS The induction periods for 316SS specimens polarized at +285 mV (SCE) in a 0.025 mol dm NaCl solution can be plotted as a function of varying photon energy as shown in Fig. 7. Figure 7 also shows the induction periods measured for identical experiments carried out in the dark. These data points are plotted at each wavelength so that the increase in the induction period on illumination is evident. The induction periods were measured as the time -3
402
Al-Rifaie et al.
between the application of the polarizing potential and the first metastable pitting events in which the current exceeded 500 nA. It is clear from this figure that the induction time was reduced slightly with decreasing photon energy.
Figure 7. Measured induction periods as a function of the incident light wavelength for 316SS polarized at +285 mV (SCE) in 0.025 mol dm-3 NaCl. Induction periods for non-illuminated specimens are also shown for reference (i.e., dark triangles at each wavelength) In previous papers [6,7], PILA has been interpreted using the PDM for the growth and breakdown of passive films [10] as the photo-quenching of the electric field within the barrier layer. The PDM has already been developed to give theoretical expressions for the breakdown potential, Eb, and the induction time, tind. It is proposed that on illumination, incident photons (with energies in excess of the band gap of the metal) generate electron-hole pairs that are separated by a steep potential gradient in a manner which quenches the electric field. Once the electric field is decreased, the theoretical expressions predict a higher breakdown potential and a larger induction time. Therefore, Figs. 1-7 are indeed consistent with a PDM interpretation of the photoinhibition effect since they predict that incident photon energy increases pitting resistance. It is not clear, at least from the experimental evidence shown here, how the electric field itself is quenched. The data presented in Fig. 5 suggests that the formation of a precipitate layer on the metal specimens polarized in the alkaline solutions screens the barrier layer from the incident 403
Fundamental Aspects
photons, and thus inhibits the generation of electron-hole pairs (and subsequent quenching of the electric field). This is supported by the experiments in which the PILA effect was partially restored on addition of the complexing agent, EDTA. This complexing agent may hinder the formation of the precipitate layer, and hence, facilitate the photon-barrier layer interaction. Also, the fact that the PILA effect remains unaffected (in neutral and acidic solutions), or increases (in alkaline solutions), in the presence of EDTA suggests that it is unlikely that photo-induced reactions, involving oxidized iron, at the film/solution interface account for the photoinhibition effect. The semipermanent nature of PILA is evident in this study (as shown in Figs. 4 and 7) and other studies [8,9]. The PILA effect remaining after the irradiation has been removed seems to indicate that the photoinhibition effect cannot be explained solely by the changing of the electronic structure of the film. Consequently, it is postulated that the suppression of the electric field strength also modifies the vacancy distribution. If we assume that the thickness of the barrier layer (L) is 3 nm, and the cation vacancy diffusivity is on the order of 10-19 cm2s-1 [12], we can calculate an approximate relaxation time (t=L2/D). The relaxation time of the vacancy structure is thus estimated to be around 9 x 105 s. If we then compare this value with the period of 220 hours (7.92 x 105 s, from Fig. 4), there is good agreement between the two values, suggesting that the photoinhibition effect persists until the vacancy structure relaxes. Figure 2 also supports this idea in the sense that longer illumination periods lead to an increased PILA effect by altering the vacancy structure more dramatically, and therefore, requiring a larger relaxation time with increasing illumination time. Another possible explanation of PILA is that the UV irradiation leads to a chromium enrichment of the passive film. Since the PILA effect on pure iron [8,9] and nickel [6] cannot be explained by chromium enrichment of the passive film; a more likely possibility is one in which both the electric field is quenched and the passive film is enriched in chromium simultaneously. CONCLUSIONS The results of this work show that photoinhibition of pitting corrosion can be achieved for SSs on illumination with UV light. Increases in both the breakdown potential and induction period, and a decrease in the frequency of metastable pitting events were observed upon irradiation. It was also found that this PILA effect depended on the photon energy (with photons having energies above the band gap of the specimen being more effective), the illumination period, the pH of the test solution, and the nature of the resulting passive film. It appeared that the precipitate layer formed on passivation of 304SS and 316SS in alkaline solutions (pH > 10) adversely affected the interaction of the incident photons with the barrier layer. The addition of EDTA, a complexing agent, partially restored the PILA effect, which seems to support the hypothesis that the precipitate layer is indeed hindering the photon-barrier layer interaction. These observations can be explained within the framework of the PDM in terms of the generation of electron-hole pairs and consequent photo-quenching of the electric field. This in turn modifies the vacancy structure, leading to an enhancement in the pitting resistance of the specimens, that remains effective for some 220 hours. ACKNOWLEDGEMENT
404
Al-Rifaie et al.
The authors gratefully acknowledge the support of this work by the Electric Power Research Institute under Contract No. RP8041-07, and by the US Department of Energy/Basic Sciences Division through Grant No. DE-FG02-91ER45461. REFERENCES 1. M. Ohi, H. Nishihara and K. Aramaki, Corrosion 50, 1994, p. 226. 2. R.G. Wendt, W.C. Moshier, B. Shaw, P. Miller and D.L. Olson, Corrosion 50, 1994, p. 819. 3. B.A. Shaw, G.D. Davis, T.L. Fritz, B.J. Rees and W.C. Moshier, Journal of the Electrochemical Society 138, 1991, p. 3288. 4. A.J. Sedriks, Corrosion of Stainless Steels, The Electrochemical Society, Princeton, New Jersey, 1979. 5. P.M. Natishan, E. McCafferty and G.K. Hubler, Journal of the Electrochemical Society 135, 1988, p. 321. 6. S.J. Lenhart, M. Urquidi-Macdonald and D.D. Macdonald, Electrochimical Acta 32, 1987, p. 1739. 7. E. Sikora, M.W. Balmas, D.D. Macdonald and R.C. Alkira, Corrosion Science, in press. 8. P. Schmuki and H. Bohni, Journal of the Electrochemical Society 139, 1992, p. 1908. 9. P. Schmuki and H. Bohni, Electrochimical Acta 40, 1995, p. 775. 10. D.D. Macdonald, Journal of the Electrochemical Society 139, 1992, p. 3434.
405
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
KINETICS OF HIGH TEMPERATURE CORROSION OF A LOW Cr-Mo STEEL IN AQUEOUS NaCl SOLUTION W.A. Ghanem1, F.M. Bayyoumi1 and B.G. Ateya2 1 2
Central Metallurgical Research and Development Institute, Helwan, Egypt.
Corresponding Author Department of Chemistry, Faculty of Science, Cairo University, Cairo, Egypt.
ABSTRACT The kinetics of corrosion of a low Cr-Mo steel alloy were studied over a temperature range of 75250 C in 1 m NaCl in the absence and in presence of various levels of contamination with CuCl2. We measured corrosion rates, weights of corrosion product (magnetite) film and total (integral) weight loss of the alloy over exposure times of 1-480 hours. The corrosion rate decreased rapidly with time, before it leveled off at longer time periods, indicating the formation of a protective corrosion product film. The ability of the alloy to retain an adherent corrosion product (magnetite) film was expressed in terms of a retention coefficient. This increased with temperature and exposure time, and decreased with the level of contamination with CuCl2. The effect of temperature was attributed to the improvement of the crystallinity of the corrosion product. On the other hand, the effect of the CuCl2 was attributed to the electro-deposition of Cu and its impregnation within the corrosion product, which became less adherent. The free corrosion potential was affected by the presence of the CuCl2 in a fashion compatible with the Wagner-Traud theory of mixed potential. 0
Key Words: Kinetics, corrosion, steel, high temperature, sodium chloride solution, copper chloride.
INTRODUCTION The corrosion of steel in high temperature aqueous environments is encountered in many industrial applications, e.g., boiling water reactors [1], desalination plants [2], high temperature aqueous fuel cells [3], and steam generators [4]. In such environments, contaminants, which are present in the aqueous media at trace levels, are concentrated by several orders of magnitude to form highly corrosive solutions [5]. Due to the active nature of iron, it corrodes in high temperature deaerated water and steam giving rise to the formation of ferrous (Fe2+) species which change to ferrous hydroxide, Fe(OH)2, and magnetite, Fe3O4 [6-10]: i.e., 3 Fe + 6 H2O = 3Fe(OH)2 + 3 H2
(1)
3 Fe(OH)2 = Fe3O4 + H2O + H2
(2)
The overall reaction is represented by: 3 Fe + 4 H2O = Fe3O4 + 4 H2
(3) 407
Fundamental Aspects
Some of the resulting magnetite adheres to the surface in the form of a film which affects the kinetics of any subsequent corrosion of the steel. The rest of the resulting magnetite spalls off the surface into the electrolyte. The qualities of the adherent magnetite film depend on the temperature, composition of the environment and exposure time. The objectives of this paper are to study the kinetics of corrosion of a low Cr-Mo steel in high temperature NaCl solution, and the mechanism of protective film formation during this process. Particular attention is given to the effect of contamination of the electrolyte with CuCl2 on the integrity of the protective film, and hence, on the subsequent corrosion. The effects of CuCl2 concentration, temperature and exposure time on the adherence of the magnetite film were also determined. EXPERIMENTAL PROCEDURE All measurements were performed in an autoclave fabricated from 316 type stainless steel. The autoclave consisted of two parts. A (Teflon) PTFE cell was machined to fit tightly into the autoclave, to accommodate the electrolyte. Further details are given elsewhere [11]. The cell was preheated for about 5 hours to obtain thermal stability [12,13]. The autoclave was placed in a furnace connected with the temperature regulator to the heating source. A NiCr thermocouple was used to regulate the temperature. It was inserted into a stainless steel tube coated with a layer of PTFE. The corrosion rate measurements were taken on coupons (2 x 5 x 0.2 cm) made of a low Cr-Mo steel of the following composition (wt%): 2.3 Cr; 1.0 Mo, 0.46 Mn, 0.2 Si, 0.13 C, 0.015 P, 0.007 S and Fe balance. They were annealed at 900oC for 1 hour in an argon atmosphere and furnace-cooled. Their microstructure revealed fine dispersed carbide in a matrix of ferrite. Before use, they were polished successively down to 600 grit using silicon carbide paper, rinsed with ethyl alcohol and distilled water, and then dried in the air. Three electrolytes were used: (I) 1 molal (m) NaCl, (II) 0.999 m NaCl + 5 x 10-4 m CuCl2 and (III) 0.9 m NaCl + 5 x 10-2 m CuCl2. They were deaerated by boiling the electrolyte under reflux for 15 miutes to give an oxygen content of < 0.1 ppm (as estimated polarographically). The volume of the electrolyte in each test was 200 ml. The tests lasted for various durations, i.e., 1, 3, 6, 12, 24, 48, 96 and 480 hours and were performed in triplicate. Two of the three specimens were subjected to successive descaling by immersing for 20 minutes in a 20% ammonium citrate solution at 80oC [14] to dissolve the corrosion products. They were then rinsed with distilled water, dried and weighed until a constant weight was obtained. We calculated the integral weight (ΔWi ) i.e., the weight of the alloy which dissolved up to a particular time, and the weight of the corrosion product film (ΔWf ) which remained adherent per square centimeter of the area after a particular time. After each test, the solution was found to contain some solid (spalled) corrosion products. The working electrode consisted of a rectangular sheet about 9 cm long, 0.5 cm wide and 0.2 cm thick. It was insulated with PTFE in such a way that an area of 1 cm2 was exposed at its end. The other end was threaded and connected to a stainless steel rod of 0.3 cm diameter through a stainless steel connector. A graphite rod of 0.5 cm diameter and about 10 cm length was used as a counter electrode, and an Ag/AgCl was used as a reference electrode [12,15]. RESULTS AND DISCUSSION 408
Ghanem et al.
Corrosion Rate Figure 1 (a-c) illustrates the variation of the corrosion rate with time of immersion at various temperatures in electrolyte I, electrolyte II and electrolyte III. Figure 1d compares the behavior in the three electrolytes at 2500C. They clearly reveal same significant features. During short time periods, the corrosion rate decreased rapidly with the time of immersion before it tended to level off after longer time periods. This behavior is characteristic of protective film formation [16]. There was a strong detrimental effect of CuCl2 on the ability of the magnetite film to protect the substrate alloy. As the temperature and/or exposure time increased, the corrosion rate decreased. The present work reveals that the mechanism of corrosion changes after a transition time, τ, the magnitude of which, generally, decreases as the temperature of the test increases. At and beyond this transition time, an adherent layer of the corrosion product was shown to protect the substrate alloy by acting as a diffusion barrier [11,17], thus reducing the rate of corrosion. Before this transition time, the alloy corrodes more freely with a higher rate of corrosion. It was found that, at a given temperature, increasing the concentration of CuCl2 increased the transition time, τ [17]. Retention Coefficient The retention coefficient is introduced here to give a quantitative expression of the ability of the alloy to retain an adherent corrosion product film on its surface under a corrosive environment. It is defined as the ratio of the weight of the adherent (magnetite) film, ΔWf (adh.), at a particular time of immersion to the total weight of the (magnetite) film which would form if the integral weight loss of the alloy were to be totally consumed in forming the film material (magnetite), i.e., ΔWf (total). The later value is related to the integral weight loss ΔWi by a chemical factor (CF), which in the present case is given by the ratio of the molecular weights of Fe3O4 and 3 Fe i.e., CF = 232/168 = 1.38. Thus, the retention coefficient is given by Retention coefficient ϕ = ΔWf (adh.) /ΔWf
(total)
(4)
The retention coefficient was determined at various temperatures, CuCl2 concentrations and time intervals. Figure 2 (a-d) illustrates the variation of the retention coefficient, ϕ , with the time of immersion for the three electrolytes at various temperatures. The curves clearly reveal that ϕ increased as the temperature increased, and decreased as the concentration of CuCl2 increased. Reaction 2, which is called the Schikorr reaction [18], has been extensively studied [19,20]. Ferrous hydroxide, Fe(OH)2, decomposes rapidly above 1000C [21] , but relatively slowly at lower temperatures. Robertson [22] stated that the corrosion of steel in hot water is controlled by the dehydration of the hydroxide phase (Eq. 2), which proceeds when the metal/solution interface becomes saturated with Fe(OH)2. Consequently, two factors affecting reaction 2,
• The saturation of Fe(OH)2 at the metal/solution interface which is time dependent, and
• The temperature, which enhances the reaction in the forward direction and affects the solubility [23] and crystallinity of the magnetite film [17]. 409
Fundamental Aspects
This explains the higher ϕ values obtained at higher temperatures, that longer exposure times, and hence, the increased efficiency of the film in retarding corrosion. On the other hand, the presence of CuCl2 in the electrolyte decreased ϕ . This has previously been shown [17] by xray diffraction and scanning electron microscopy to be due to the electro-deposition of Cu and its impregnation within the magnetite film, which then becomes less adherent.
Figure 1. Variation of the corrosion rate with the time of immersion at different temperatures in: (a) sol. I, 1 m NaCl; (b) sol. II, 0.999 m NaCl + 0.0005 m CuCl2; (c) sol. III, 0.9 m NaCl + 0.05 m CuCl2.; and (d) at 250oC in different solutions
410
Ghanem et al.
Figure 2. Variation of the retention coefficient, ϕ , with the time of immersion at different temperatures in: (a) sol. I, 1 m NaCl; (b) sol. II, 0.999 m NaCl + 0.0005 m CuCl2; (c) sol. III, 0.9 m NaCl + 0.05 m CuCl2; and (d) at 250oC in different solutions Comparing the results in Figs. 1 and 2, it can be concluded that, in most cases, as the retention coefficient decreases, the corrosion rate increases. In other cases, both the retention coefficient and the corrosion rate increase in the same direction. This result indicates that the film formed, though retained on the alloy surface, is unable to protect it from subsequent corrosion. Potential-Time Curves Figure 3 illustrates the time variation of the free corrosion potential, Ecor , of the alloy at different temperatures in electrolytes I, II and III. It is seen that Ecor shifts toward the noble direction as the concentration of CuCl2 increases. This is in agreement with the results of Lin et al [23]. The increase in temperature above 750C shifted the values of Ecor in electrolytes I and II closer to each other than they were at 750C. In electrolyte III, increasing the temperature shifted the free corrosion potential to more noble values. A comparison of these free corrosion potential values with the equilibrium potentials of the hydrogen evolution (H2O/H2) and copper reduction (Cu/Cu++) reactions is in order to identify the cathodic half reactions. Table 1 lists the values of the equilibrium potentials of both systems at various temperatures in electrolytes I, II and III. Note that these Ecor values are considerably negative (cathodic) with respect to the reversible equilibrium potentials of the Cu/Cu++ or the H2O/H2 electrode systems [6]. The approximate values for Cu/Cu++ system in electrolytes II and III are calculated at various temperatures using the Nernst equation, i.e., Cu2+ + 2e
Cu
E = E0 Cu/CuCl2 + 2.303 RT/2F log [Cu2+]
(5) (6)
The values of E0 were obtained from Latimer [24]; the activity of the Cu2+ species was taken equal to its concentration. The values of E (H2O/H2) at different temperatures were taken from Pourbaix diagrams [6, 25,26]. Consequently, under the potentials shown in Fig. 4, the cathodic half cell reaction involves both the reduction of water i.e., reaction 7 2 H2O + 2 e →
H2 + 2 OH-
(7)
and the electro-deposition of Cu according to reaction 5, while the anodic reaction involves the dissolution of the iron, i.e., reaction 8 Fe + H2O
→
Fe(OH)+ + H+ + 2e
(8)
Since the concentration of Cu2+ is rather small in electrolyte II, the time behavior of Ecor is not significantly different from that in electrolyte I at the higher temperatures i.e., 125, 175 and
411
Fundamental Aspects
2500C. Alternatively, in presence of higher concentration of CuCl2, the rate of reaction 5 is greatly enhanced leading to an increase in the corrosion rate. The results of Figure 3 can be explained within the domain of the Wagner-Traud theory of mixed potential [27,28], shown schematically in Fig. 4, which illustrates the effect of a significant increase in the rate of the cathodic reaction on the corrosion rate (Icor) and the corrosion potential (Ecor). For the sake of simplicity, we neglect the changes in the anodic polarization curves of reaction 5 brought about by adding CuCl2. Upon changing the cathodic half cell reaction from reaction 7 to reaction 5, Fig. 4 shows a significant increase in the corrosion current, Icor , and a significant shift in the mixed (free corrosion) potential towards more noble values. Both phenomena were confirmed by the experimental measurements shown in Figs. 1 and 3.
412
Ghanem et al.
Figure 3. Time variation of the free corrosion potential for the alloy at different temperatures in: electrolyte I (1 m NaCl); II (0.999 m NaCl + 0.0005 m CuCl2), and III (0.9 m NaCl + 0.05 m CuCl2) Table 1. Approximate Values of the Electrode Potential, V (NHE) of the Cu/Cu2+ Calculated at Various Temperatures Using the Nernst Equation in Electrolytes II and III and of the H2O/H2 Systems in Electrolytes I, II and III [6, 25,26]
Temperature 750C 1250C 1750C 2500C
Electrolyte I H2O/H+ - 0.442 - 0.551
Electrolyte II Cu/Cu2+ H2O/H+ + 0.312 - 0.405 + 0.309 + 0.305 + 0.299 - 0.484
Electrolyte III Cu/Cu2+ H2O/H+ + 0.329 - 0.361 + 0.327 + 0.326 + 0.325 - 0.415
Figure 4. Schematic representation of the effect of CuCl2 on the mixed potential of iron in the corrosive medium CONCLUSIONS Inspection of the results presented reveals the following conclusions:
413
Fundamental Aspects
1.
At short times, the corrosion rate decreases rapidly with the time of immersion before it tends to level off at longer times. This behavior is characteristic of protective film formation
2.
The retention coefficient is introduced to give a quantitative expression for the ability of the alloy to produce an adherent corrosion product film under the corrosive environment. Increasing the temperature enhances the ability of the alloy’s surface to retain the film. The concentration of CuCl2 has an opposite effect.
3.
The results of potential-time curves measured in different electrolytes reveal that Ecor shifts in the noble direction to an extent that increases with the concentration of CuCl2. This is compatible with the Wagner-Traud theory of mixed potential. The increase in temperature above 750C shifts the values of Ecor in electrolyte I and II closer to each other than they are at 750C.
ACKNOWLEDGMENT The authors express their warm gratitude to Prof. A.A. Abdul Azim, the former chairman of CMRDI, for valuable discussions. REFERENCES 1. B.C. Syrett, Materials Performance 30, 8, 1992, p. 52. 2. O. Osborn and F.H. Coley, in High Temperature High Pressure Electrochemistry in Aqueous Solutions, Houston, NACE 4, 1976, p. 7. 3. F.T. Bacon, in High Temperature High Pressure Electrochemistry in Aqueous Solutions, Houston, NACE 4, 1976, p. 24 4. M.J. Wootten, G. Economy, A.R. Pebler and W.T. Linsay. Jr., Materials Performance 17, 2, 1978, p.30. 5. C.B. Ashmore, M.H. Hurdus, A.P. Mead, P.J B. Silver, L.Tomlinson and D.J. Finnigan, Corrosion 44, 1988, p. 334. 6. M. Pourbaix, in Atlas of Electrochemical Equilibria in Aqueous Solutions, London, Pergamon Press, 1974, p. 305. 7. J.E. Castle and G.M.W. Mann, Corrosion Science 6, 1966, p. 253. 8. J. Robertson, Corrosion Science 29, 1989, p. 1275. 9. J. Jelinek, P. Neufield, Corrosion 38, 1982, p. 98. 10. G. Butler, H.C.K. Ison and A.D. Mercer, British Corrosion Journal 6, 1971, p. 23. 11. F.M. Bayyoumi, M.Sc. Thesis, Cairo University, 1995. 12. M.H. Lietzke, R.S. Greeley, W.T. Smith and R.W. Stoughton, Journal of Physical Chemistry 64, 1960, p. 652. 13. D.D.G. Jones and H.G. Masterson, in Advances in Corrosion Science and Technology, Vol. 1, New York, Plenum Press, 1970, p. 1. 14. F.A. Champion, in Corrosion Testing Procedures, London, Chapman and Hall, 1964, p. 192. 15. D.D. Macdonald, A.C. Scott and P. Wentrcek, Journal of the Electrochemical Society 126, 1979, p. 908 16. U.R. Evans, in The Corrosion and Oxidation of Metals: Scientific Principles and Practical Applications, New York, St. Martin’s Press , 1960, p. 819. 17. W.A. Ghanem, F.M. Bayyoumi and B.G. Ateya, Corrosion Science, in press, 1996. 414
Ghanem et al.
18. G. Schikorr, Z. Anorg. Allg. Chem. 212, 1933, p. 533. 19. U. R. Evans and J.N. Wanklyn, Nature 162, 1948, p. 27. 20. B. McEnaney and D.C. Smith., Corrosion Science 18, 1978, p. 591. 21. F.J. Shipko and D.L. Douglas, Journal of Physical Chemistry 60, 1956, p. 1519. 22. J. Robertson, Corrosion Science 32, 1991, p. 443. 23. C.C. Lin, F.R. Smith, N. Ichikawa and M. Itow, Corrosion 48, 1992, p. 16. 24. W.M. Latimer, in The Oxidation States of the Elements and Their Potentials in Aqueous Solutions, 2nd ed., New York, Prentice Hall, 1961. 25. H.E. Townsend, Jr., Corrosion Science 10, 1970, p. 343. 26. V. Ashworth and P.J. Boden, Corrosion Science, 10, 1970, p. 709. 27. C. Wagner and W. Traud, Z. Elektrochem. 44, 1938, p. 391. 28. D.D. Macdonald, Corrosion 48, 1992, p. 194.
415
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CORROSION AND PASSIVATION BEHAVIOUR OF ALUMINIUM AND ALUMINIUM ALLOYS MECHANISM OF THE CORROSION PROCESS F.M. Al-Kharafi, W.A. Badawy and A.S. El-Azab Department of Chemistry, Faculty of Science Kuwait University, P.O. Box 5969 Safat, 13060 Kuwait
ABSTRACT Aluminum and Al-alloys represent technologically and industrially important materials. The electrochemical behavior of these materials in different solutions represents a major subject of investigation. The corrosion characteristics of naturally passivated Al, Al-Cu, Al-6061 and Al-7075 were studied in nitric acid and nitric acid containing chloride solutions. The effect of the concentration of anions on the corrosion behavior of these materials was traced. Electrochemical impedance spectroscopy (EIS) is a powerful tool in studying corrosion and passivation problems. Besides polarization techniques, the method has been applied successfully to investigate the corrosion behavior of Al and Al-alloys. The Al-6061 alloy was found to be the most corrosion resistant. In all cases, the naturally occurring passive film was too thin to impart complete passivity. Equilibrium occurred between barrier film dissolution and surface passivation especially in dilute solutions (< 0.1 M HNO3). The electrode/electrolyte interface was fitted to a parallel resistor/capacitor combination. The barrier film formed on Al or Al-6061 behaved like a perfect dielectric whereas that formed on Al-Cu or Al-7075 alloys deviated from the ideal capacitor behavior. X-ray photoelectron spectroscopy (XPS) experiments have shown that Al-Cu alloys contain remarkable amounts of Cu on the material surface. Scanning electron microscopy (SEM) investigations have shown that the presence of Cu on the alloy surface initiates flawed regions which are responsible for the increased corrosion rate of the Cu-containing alloys. Key Words: Aluminium, aIuminium alloys, corrosion, electrochemistry, impedance, passivation
INTRODUCTION Due to the technological importance of Al and Al-Alloys and their increased industrial applications, the electrochemical behavior of these materials represents an important subject for many investigators [1-10]. Investigations have been conducted to optimize the anodic polarization and passive film growth [11-15]. The effect of anions like Cl and the mechanism of their attack on the metallic surface has been a major research subject [5,10,16,17]. The corrosion and passivation behavior of aluminum and its alloys has been subjected to intensive investigation [18-21]. The use of nitric acid and nitric acid/phosphoric acid mixtures in the surface finishing of aluminum and aluminum alloys, especially in the household industry, required detailed information about the electrochemical behavior of these materials in this medium [17,22-25].
417
Fundamental Aspects
Several techniques have been used to study the corrosion and passivation behavior of metals and alloys. Electrochemical impedance spectroscopy (EIS) is becoming a well established method for investigating electrochemical systems in which the solid/electrolyte interface plays the main role [26]. One of its important aspects is the direct matching that often exists between experimental impedance data and data obtained from discrete electrical components which represent physical processes taking place in the system under investigation [27]. Such electronic components are usually termed electronic models or equivalent circuits. EIS and other polarization techniques are very useful in studying the corrosion and passivation behavior of Al and its alloys. The polarization resistance, RP , which represents the major indication of kinetic facility [28], and electrode capacitance, C , which represents the main source for calculating the barrier layer thickness and its dielectric properties [29] are the main factors used to describe and control the corrosion and passivation of these materials [22-25]. In the present investigation, the electrochemical behavior of mechanically polished, naturally passivated Al, Al-Cu, Al-6061 and Al-7075 was investigated in nitric acid and nitric acid containing chloride solutions. An electronic model for the barrier layer/electrolyte interface was described. The effect of chloride ion concentration on the corrosion and passivation behavior of the metal and its alloys in nitric acid solutions was studied. X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) were used to investigate the material’s surface. EXPERIMENTAL PROCEDURE Commercial-grade aluminum and aluminum alloys (Al-Cu, 6061 and 7075) were used as electrodes. The mass spectroscopic analysis of these materials is presented in Table 1. Electrodes in the form of cylindrical rods were mounted into glass tubes of appropriate internal diameter with an epoxy resin leaving an exposed surface area of 0.50 , 0.21 , 0.20 and 0.21 cm2 for Al, Al-Cu, Al-6061 and Al-7075, respectively, to contact the test solution. The electrolytic cell was an all glass, three electrode cell with a large surface area Pt counter electrode and Ag/AgCl/Cl- (3 M KCl) reference electrode. The electrolytic solutions were prepared using analytical grade reagents and triply distilled water. All measurements were carried out at a constant room temperature of 25oC. The potentials were measured against the Ag/AgCl/Cl- (3M KCI) reference {Eo = 0.1970 V(nhe)}. Before each experiment, the electrode was mechanically polished with successive grades of emery paper down to 3/0, and then with a smooth cloth and washed with triply distilled water. In this way, the electrode’s surface acquired a reproducibly bright appearance. For comparison, some experiments were carried out after chemical etching of the electrode’s surface to be sure that the mechanical polishing had no effect on the alloy’s structure. The electrodes were chemically etched in a 80oC heated mixture of phosphoric, acetic and nitric acids for 5 minutes [22,23]. The impedance data obtained for both mechanically polished and chemically etched electrodes showed almost the same trend with slightly higher impedance values (5-10% higher) for the chemically etched surface at different time intervals of electrode immersion in the test solution. EIS measurements were performed using the IM5d-AMOS system (Zahner Elektric GmbH & Co., Kronach, Germany). All experiments involved single frequency measurements in the frequency domain of 0.1-105 Hz. To check the presence of another time 418
Al-Kharafi et al.
constant at lower frequencies, some experiments were conducted over a bandwidth of 1 mHz - 105 Hz. The input signal’s amplitude was usually 10 mV peak to peak. Polarization measurements were carried out using an EG&G (Princeton Applied Research) Model 273A potentiostat/galvanostat interfaced to an IBM PS/3 computer. The XPS experiments were carried out using an ESCA-Lab 200 (VG instruments). The surface was etched as required by argon ion bombardment. In each spectrum, the XPS peaks of C 1S, O 1S, Al 2P and Cu 2P1 and 2P3 were traced. The electrode’s surface was examined by SEM before and after immersion in the test solution. The details of experimental procedures were as described elsewhere [22-25]. Table1. Mass Spectrometric Analysis of the Different Electrode Materials in Mass % Alloy
Al
Cu
Mg
Al Al6061 Al-Cu Al7075
99.23 97.09 93.43 90.93
0.043 0.201 4.80 1.17
0.217 1.40 0.229 2.21
Si 0.038 0.601 0.047 0.272
Fe
Mn
Ni
Zn
pb
Sn
Ti
Cr
0.164 0.193 0.499 0.124
0.001 0.012 0.024 0.067
0.010 0.010 0.012 0.007
0.027 0.029 0.025 4.95
0.001 0.000 0.721 0.000
0.003 0.000 0.006 0.000
0.006 0.016 0.015 0.024
0.001 0.248 0.001 0.046
RESULTS AND DISCUSSION Corrosion Behaviour in Nitric Acid Solution, Equivalent Circuit for the Electrode/Electrolyte Interface The impedance behavior of the different electrodes was investigated in 0.1 M HNO3. The mechanically polished electrodes were left in 0.1 M HNO3 until a steady state was reached, and then the impedance data were recorded. Although Bode plots for impedance data presentation are always recommended as standard impedance plots [26], they sometimes lead to no indication of features hidden at high frequency [10]. In such cases, the Nyquist plot format is more favorable. For data fitting procedures, Bode plots are always used since all experimental data are equally represented, and the phase angle is very sensitive for indicating the presence of additional time constants in the impedance spectra. In our experiments, both formats were used. Typical Nyquist plots of Al-7075 alloy electrodes taken at different time intervals from the steady state are presented in Fig. 1. Bode plots as a function of immersion time in the test electrolyte of Al-6061 are presented in Fig. 2. For all electrodes, the impedance Nyquist plot at any time interval consists of two semicircles (two phase maxima in the Bode plot). A high frequency semicircle, which is due to the interaction between the electrode surface and the electrolyte, is associated with a high field conduction mechanism through the oxide film and its thickness, and a low frequency loop which is concerned with the relaxation processes occurring in the barrier layer either in the bulk or at the surface, which is typical of passivated surfaces. Below the assigned low frequency of the experiment (i.e., 0.1 Hz), no reproducible data could be obtained. In the very low frequency range (0.1-100 mHz), the structural changes of the interfacial region were faster than the measurements and no reliable data could be obtained [30-33]. The diameter of the high frequency semicircle changed with the time of immersion in the electrolyte. For all alloy electrodes, the diameter decreased with immersion time which reflects a decrease in the polarization resistance of the barrier layer, Rp, and its
419
Fundamental Aspects
thickness, δ. Pure Al electrodes showed continuous increases of diameter with immersion time in nitric acid solutions (Fig. 3). The increase in the diameter with time indicates oxide film thickening which means continuous passivation of Al in nitric acid solutions as was observed before [22]. The barrier layer thickness was calculated from the impedance data according to C = I/ 2 π f Zim
(1)
C = A ε ε0 / δ
(2)
where C is the electrode’s capacitance, f is the frequency, Zim is the electrode’s impedance, A is the electrode’s area, ε0 is the permittivity of free space (8.85 x 10-14 Fcm-l), δ is the barrier layer thickness and ε is the oxide film dielectric constant taken as 8.4 [33] considering that the barrier layer consists mainly of A1-203. The calculated barrier layer thickness after a long period of electrode immersion (≈ 4 hours) in the test solution ranges between 0.2 and 0.6 nm for all electrodes irrespective the barrier layer thickening or thinning that occurred at the electrode/electrolyte interface. This thickness is about one-tenth of the thickness of the barrier layer occurring on Al or Al alloys in neutral solutions (pH = 7) [6,34]. The presence of such a barrier layer on Al or its alloys after long immersion times in nitric acid solution (≈ 4 hours in 0.1 M solution) indicates the remarkable passive behavior of these materials in these electrolytes.
Nyquist plots of Al-7075 electrode at different time intervals of immersion in 0.1 M HNO3 (steady state potential = -258 mV vs. Ag/AgCl/Cl- [3 M KCl]) (⎯) 45 min (steady state), (…) 75 min, (---) 135 min, (- - -) 250 min. 420
Bode plots of Al-6061 electrode at different time intervals of immersion in 0.1 M HNO3 (steady state potential = -532 mV vs. Ag/AgCl/Cl- [3 M KCl]) (⎯) 45 min (steady state), (…) 75 min, (---) 135 min, (- - -) 250 min.
Al-Kharafi et al.
The polarization and impedance data are in good agreement. Figure 4 presents the Tafel polarization curves of the four materials investigated after reaching the steady state in 0.1 M HNO3. The values of the polarization resistance, Rp , corrosion current, icorr , and steady state potential, Ecorr , for the different electrodes as obtained from these measurements are presented in Table 2. Figure 5 presents the impedance data of the electrodes after 250 minutes of electrode immersion under the same conditions. Taking into consideration the polarization resistance and corrosion current data given in Table 2, the stability order of the investigated materials after reaching the steady state in 0.1 M HNO3 (i.e., 45 minutes of electrode immersion) follows the sequence: Al-6061 > Al > Al-7075 > Al-Cu After a long immersion time in the test solution (i.e., 4 hours), Al attained of comparable stability or became even more passive than the Al-6061 alloy and the order changed to: Al > Al-6061 > Al-7075 > Al-Cu
Nyquist plots of aluminium electrodes at different time intervals of immersion in 0.1 M HNO3 (steady state potential = 650 mV vs. Ag/AgCl/Cl- [3 M KCl]) (⎯) 45 min (steady state), (…) 75 min, (---) 135 min, (- - -) 250 min.
Tafel polarization curves of Al-Cu (1), Al-7075 (2), Al-6061 (3) and Al (4) after 45 min of electrode immersion in 0.1 M HNO3
This is clearly reflected in the Nyquist plots of Fig. 5. The change in the order of stability of Al and Al-6061 after long immersion times in the nitric acid solution is due to the 421
Fundamental Aspects
observed passivation of aluminum from the moment of immersion in nitric acid (Fig. 3). The polarization resistance of the Al-electrode increased from 230 Ωcm2 after 45 minutes of electrode immersion in 0.1 M HNO3 solution at 25oC, to 372 Ωcm2 after 250 minutes of electrode immersion in the same solution under the same conditions. Table 2. Values of RP , icorr and Ecorr (vs. Ag/AgCl/3M Cl-) for the Different Electrodes Measured in 0.1 M HNO3 at the Steady State ( ≈ 45 min from electrode immersion) Electrode Al-6061 Al Al-7075 Al-Cu
2
Rp, (Ω cm ) 502 374 276 269
icorr, (μA cm2) 113.5 121.2 169.6 196.1
Ecorr, (mV) -500 -778 -271 -111
The data presented in this section show that Al-6061, either from the moment of immersion in nitric acid or after a long period of immersion, represents the most stable alloy in this solution of the alloys investigated. The results reveal that the presence of the small amount of Mg (1.4%) improves the passivation behavior of the aluminum alloy. The mass spectrometric investigation of the alloy showed that it contains 1.40% Mg and 0.60% Si. Such a combination in a heat-treatable wrought alloy leads to the formation of a Mg2Si phase, which is the basis for precipitation hardening. Either in solid solution or as submicroscopic precipitate, Mg2Si has a negligible effect on electrode potential. The alloy is normally used in a heat-treated form; therefore, no detrimental effects derive from the major alloying element or from the minor components like Cr and/or Zn which are usually added to control the grain structure. Copper additions which increase strength in the alloy are limited to very small amounts 0.2% in this alloy (Al-6061), to minimize its effects on corrosion resistance [35]. Increasing the copper content decreases the corrosion resistance of the alloy, as can be seen for Al-7075 (1.17% Cu) and Al-Cu (4.80% Cu) in Table 2. The electrochemical system can be represented by a theoretical model consisting of a parallel combination of resistor, Rp , and capacitor, C , in series with the electrolyte resistance, RS, [22,23]. Other equivalent circuit models including capacitive features and inductive features are successful in describing the electrochemical behavior of Al or its alloys in the very low frequency regions, (i.e., f < 0.1 Hz) [10,19,36]. Since it is necessary to compare the electrochemical behavior of Al and the investigated alloys, it is useful to reduce the theoretical model to the least number of components which can describe the dielectric properties of the oxide film. The capacitor/parallel resistor model, investigated in the high frequency region (f > 0.1 Hz) is suitable for such investigation. At high frequencies the resistance of the inductive features becomes included in the polarization resistance, RP , which is equivalent to the corrosion resistance, Rcorr , of the material. The capacitive features of the high frequency semicircle are related to the barrier layer itself [30,33]. The impedance data of the different electrodes were correlated to the model described above. A procedure of data fitting with minimum error was used in which a fitting program was applied to fit the experimental data to the computer-generated data. The program used 422
Al-Kharafi et al.
enables data fitting in the required range of frequency. For the measurements presented, it was necessary to fit experimental data of the high frequency semicircle to the computergenerated data of the proposed model. For data fitting procedures, Bode plots are always recommended as standard impedance plots [26,37]. Figure 6 presents the experimental Bode plots for Al-6061 after ≈ 4 hours of electrode immersion in 0.1 M HNO3 (dotted line) correlated to the computer-generated data of RS = 25.2 Ω, Rp = 1.39 kΩ and C = 2.36 μF according to the data fitting program. The data fitting of Al-6061 gives a mean error in the absolute impedance of 1.4% and a mean deviation in the phase angle, θ, of 1.00. The procedure of data fitting was applied to other electrode materials and also to impedance spectra taken at different time intervals of electrode immersion in the test solution. The Al and Al-6061 impedance data represent the best fitt to the theoretical model after reaching the steady state, whereas Al-Cu showed the largest deviation. The absolute impedance and phase angle deviation for Al-Cu electrodes after 4 hours of electrode immersion in 0.1 M HNO3 solution are 3.1% and 1.6o, respectively. The small deviation of the absolute impedance values obtained with Al-6061 and pure Al indicate that the barrier layer on these materials approaches ideal capacitor behavior.
Nyquist plots of Al-Cu (⎯), Al7075 (…), Al (----) and Al-6061 (- -) after 250 min of electrode immersion in 0.1 M HNO3. The values of Rp for each electrode in 2 Ωcm are 122, 142, 372 and 234, respectively
Computer fitted data of RS = 25.2 Ω , Rp = 1.39 kΩ and C = 2.3 μF (⎯) to experimental Bode plot of Al-6061 after 250 min of electrode immersion in 0.1 M HNO3 (••)
Effect of Chloride Ion Concentration In this series of experiments the effect of chloride ion concentration on the corrosion and passivation behavior of Al, Al-Cu, Al-6061 and Al-7075 was investigated. The electrodes 423
Fundamental Aspects
were mechanically polished and investigated in 0.1 M HNO3 solutions containing different concentrations of Cl- ions ranging between 3.5 mM and 0.35 M. A typical example of the data from these investigations is presented as Nyquist plots for the Al-Cu alloy in Fig. 7. For all investigated materials in all measurements, two semicircles were recorded. A high frequency semicircle and a low frequency inductive loop. The diameter of the high frequency semicircle depended on the concentration of Cl- (Fig. 7). It decreased as the concentration of Cl- ions increased. This means that the natural passivity of Al or Al alloys decreases in the presence of chloride ions, as was reported for aluminum. Chloride ions attack the base metal by dissolving the passive film at defective areas [16,37,38]. The decrease of polarization resistance and passive layer thickness with increasing Cl- ion concentrations means that the native barrier layer is too thin to impart complete passivity. As the concentration of Clincreases, the extent of surface attack increases and Cl- spreads laterally beneath the original native film leading to the loss of its passive characteristics with the formation of a nonprotective, oxyhalide layer on the metallic surface [16,24,25]. At very low concentrations of Cl- (i.e., 3.5 mM), the rate of barrier layer removal is very low and is exceeded by the rate of passive film repair that occurs in nitric acid solutions; hence, no remarkable attack can be observed. In this case, oxide film thickening occurs as was indicated by the increase of 1/C versus t relations [17,24,25]. At higher concentrations of Cl- (i.e., > 35 mM), the rate of barrier film removal is compensated for by the rate of passive film formation, and equilibrium is attained in which an approximate constant value of δ can be calculated. The remaining barrier layer thickness and its polarization resistance depends on the electrode material. Figure 8 presents the effect of Cl- ions with the same concentration (i.e., 35 mM) on the different electrode materials. As can be seen from the Nyquist plots of Fig. 8, Al-6061 has better corrosion resistance against Cl- than the other alloys investigated and even better resistance than aluminum itself. The Al-Cu and Al-7075 alloys are much affected by the presence of chloride ions. A remarkable decrease in the corrosion resistance of both alloys with the increase of the concentration of Cl- was recorded. This behavior can be attributed to the alloy’s constituents. The presence of Mg in the Al-6061 alloy improved the corrosion characteristics of the alloy in chloride media, whereas the presence of Cu increased the corrosion rate of the alloy. The values of corrosion resistance of the different materials after reaching the steady state in 0.1M HNO3 containing 35 mM Cl- took the order, Rp Al-6061 > Rp Al > Rp Al-7075 > Rp Al-Cu. The values in kΩ-cm2 for the sequence are 0.754, 0.382, 0.157 and 0.097 for Al-6061, Al, Al-7075 and Al-Cu, respectively. The presence of Cu on the alloy’s surface was confirmed by XPS measurements. Figure 9 presents the XPS spectra for Al, Al-Cu and Al-6061. In all spectra, the characteristic peaks of aluminum (Al 2P at 75.5 eV and Al 2S at 120.0 eV), oxygen (O 1S at 532.5 eV) and carbon (C 1S at 285.5 eV) were recorded. The XPS of Al-Cu (Fig. 9b) contains additional copper peaks (Cu 2P3 at 932.5 eV and Cu 2Pl at 952.5 eV) which indicate the presence of Cu on the electrode’s surface even after 3 hours of electrode immersion in the test solution. The XPS of Al-6061 did not show a pronounced Mg peak, i.e., the characteristic sharp Mg XPS peak (Mg 1S at 1305 eV) is not present [39]. This means that Mg is present more likely in the form of Mg2Si in the Al-6061 bulk and not on the alloy’s surface. The similarity between the XPS spectra for Al and Al-6061 (compare Fig. 9a and b) explains the close corrosion behavior of both materials and supports the conclusion that the barrier layer on both materials consists of a stable Al2O3 film. The presence of Cu on the Al-Cu surface is responsible for 424
Al-Kharafi et al.
the higher rates of corrosion recorded for this alloy. It initiates cathodic areas or flawed regions which leads to the observed decrease in the corrosion resistance after long immersion times in the test electrolyte. The presence of flawed regions on Al-Cu was confirmed by SEM. Figure 10 presents the scanning electron micrographs of Al and Al-Cu surfaces before immersion in the test solution (Fig. 10a and b, respectively) and of an Al-Cu surface after 3 hours of immersion in 0.1 M HNO3 containing 0.35 M Cl-.
Nyquist plots of Al-Cu electrodes after 250 min of electrode immersion in 0.1 M HNO3 containing different concentrations of chloride ions (⎯) 0.35 M Cl , (…) 0.035 M Cl and (---) 3.5 mM Cl-
Nyquist plots of Al-Cu (⎯), Al7075 (----), Al (- - -) and Al-6061 (…) electrodes after 250 min of immersion in 0.1 M HNO3 + 35 mM Cl- solution. The values of Rp 2 for each electrode in Ωcm are 91, 114, 630 and 336, respectively
Mechanism of the Corrosion/Passivation Process The mechanism of corrosion of Al and its alloys is based on the dissolution of Al atoms from the active sites or flawed regions on the naturally passivated material. The dissolved atoms are gradually removed through the formation of hydroxide with increased coordination from 1 to 3 to form Al(OH)3. The formed hydroxide sticks to the surface in neutral solutions, and hence, a decrease in the corrosion rate takes place which gives the remarkable passive behavior of Al and its alloys in neutral solutions [40]. In acid and alkaline solutions, the formed Al(OH)3 reacts in a purely chemical manner to form soluble species which go in solution leaving bare active sites which in turn lead to the observed increase in the corrosion rate in these media [25,411. The mechanism of the corrosion process represents an + irreversible coupled reaction, the anodic part of which is the reduction of H , water or dissolved oxygen leading to hydrogen evolution and OH- formation in the vicinity of the active regions according to:
425
Fundamental Aspects
- cathodic reaction: H+ + e- → H H2O + e- → H + OH-
(3) (3’)
H + H2O + e- → H2 + OH-
(4)
1/2 O2 + H2O → OHads. + OH-
(5)
OHads..+ e- → OH-
(6)
Al + OH- → Al(OH)ads + e-
(7)
Al(OH)ads. + OH- → Al(OH)2ads. +e-
(8)
Al (OH)2ads. + OH- → AI(OH)3ads. + e-
(9)
- anodic reaction:
In the presence of Cl- ions, metal dissolution occurs through the attack of Cl- according to Al + Cl- → AlClads. + e-
(10)
AlClads.. + Cl- → AlCl2ads. + e-
(11)
AlCI2ads.. + Cl- → AlCl3ads. + e-
(12)
and the most appropriate cathodic counterpart is reaction (3) followed by reaction (4). The presence of cathodic areas enhances the corrosion process which was observed with the Cucontaining Al-alloys (i.e., Al-Cu or Al-7075). The presence of Cu on the material surface increases the ratio of cathodic/anodic areas leading to an increase in the corrosion rate. The natural tendency of Cu or Zn to form oxyhalide complexes is also an additional effect which causes the loss of the protective properties of the naturally occurring barrier layer in the chloride solution. In chloride-free nitric acid solutions, the formed Al(OH)3 can be oxidized to the stable A12O3 passive film. The oxidation power of nitric acid depends on its concentration [22-24]. This explains the passive behavior of Al and Al-6061 in nitric acid solutions. The presence of Cu on the surface of the Cu-containing alloys increases the cathodic areas, and hence, increases the tendency for galvanic corrosion to occur, which explains the comparatively high rates of corrosion of these alloys after long immersion time in nitric acid solutions (Table 2).
426
Al-Kharafi et al.
X-ray photoelectron survey spectra of naturally passivated Al(a), AlCu(b) and Al-6061 (c) after 3 h of electrode immersion in 0.1 M HNO3
. SEM micrographs of mechanically polished Al(A), AlCu (B) and Al-6061 (C) after 3 h of immersion in 0.1 M HNO3 solution containing 0.35 M Cl
CONCLUSIONS The corrosion and passivation behavior of Al, Al-6061 Al-7075 and Al-Cu, is dependent of both the alloying element and the corrosive medium. In nitric acid and nitric acid containing Cl- ions solutions, Al-6061 has the highest corrosion resistance. This alloy with its 1.40% of Mg. and 0.60% of Si behaves like a perfect dielectric. The corrosion behavior of which is most likely similar to aluminum, especially after long periods of immersion in nitric acid or nitric acid containing chloride solutions. The presence of Cu on the surface of Cucontaining alloys initiates flawed regions in the barrier layer which are responsible for the higher corrosion rates of these alloys. ACKNOWLEDGEMENT 427
Fundamental Aspects
The financial support of Kuwait University, Research Grant No. SC060, is gratefully acknowledged. REFERENCES 1. M. Heine, D. Keir and M. Pryor; J. Electrochem. Soc. 113, 1965, p. 24. 2. J. Painot and J. Augustynski, Electrochim. Acta 20, 1975, p. 747. 3. D.M. Drazic, S.K. Zecevic, R.T, Atanososki and A.R. Despic, Electrochim. Acta 28, 1983, p. 968. 4. Y. Fukuda and T. Fukushima, Electrochim. Acta 28, 1983, p. 47. 5. W.A. Badawy, M.M. Ibrahim, M.M. Abou-Romia and M.S. El- Basiouny, Corrosion 42, 1986, p. 342. 6. W.A. Badawy, M.S. El- Basiouny and M.M. Ibrahim, Ind. J. Technol. 24, 1986, ,p. 1. 7. M.G. Khedr and A.M.S. Lashien, J. Electrochem. Soc. 136, 1989, p. 968. 8. G. Burri, W. Luedi and O. Haas, J. Electrochem. Soc. 136, 1989, p. 2167. 9. C.B. Breslin and W.M. Carroll, Corros.Sci. 34, 1993, p. 327. 10. C.M.A. Brett, I.A.R. Gomes, J.P.S. Martins, J. Appl. Electrochem. 24, 1994, p. 1158. 11. M. Elboujdaini, E. Ghall, R.G. Barradas and M. Glrgis, Corros. Sci. 30, 1990, p. 855. 12. N. Khalil and J.S.L. Leach, Electrochim. Acta 31, 1986, p. 1279. 13. V. Surganov, P. Morgan, J.G. Nielsen, G. Gorokh and A. Mozalev, Electrochim. Acta 32, 1987, p. 1125. 14. V.P. Parkhutik, J.M. Albella, Yu. E. Nlakushok, 1. Montero, J.M. Martinez Duart and V.L Shershulskii, Electrochim.Acta 35, 1990, p. 955. 15. V.P. Parkhutik, V.T. Belov and M.A. Chemyckh, Electrochim. Acta 35, 1990, p.961. 16. A.A. Mazhar, W.A. Badawy and M.M. Abou-Romia, Surface and Coating Technology 29, 1986, p. 335. 17. F.M. Al-Kharafi and W.A. Badawy, Proceedings of the 186th Electrochemical Society Meeting, Miami, Florida, USA, October 1994. 18. T. Hurlen, H. Lian, O.S. Odegerd and T. Valand, Electrochim. Acta 29, 1984, p. 679. 19. T. Hurien and A.T. Haug, Electrochim.Acta 29, 1984, p. 1133 and p. 1161. 20. W.C. Moshier, G.D. Davis and J.S.Aheam, Corros. Sci. 27, 1987, p.785. 21. L. Tomcsanyi, K. Varga, I. Bartik, G. Horayi and E. Maleczki, Electrochim. Acta 34, 1989, p. 855. 22. W.A. Badawy and F.M. Al-Kharafi, B.Electrochem. 11, 1995, p. 505. 23. F.M. Al-Kharafi and W.A. Badawy, Ind. J. Chem. Technol., In press. 24. F.M. AI-Kharafi and W.A.Badawy, Electrochim. Acta 40, 1995, p.1811. 25. W.A. Badawy and F.M. Al-Kharafi, Corros.Sci., Accepted. 26. D.D. Macdonald, S. Real, S.I. Smedley and M. Uraquidi-Macdonald, J.Electrochem. Soc. 135, 1988, p. 2410. 27. W.A. Badawy and Kh.M. Ismall, Electrochim. Acta 38, 1993, p. 2231. 28. J. Koryta, J. Dvorak and L. Kavan, Principles of Electrochentistry, Chap. 5, J. Wiley & Sons, Chichester, 1995. 29. J.M. Albella, I. Montero and J.I. Martinz-Durat, Thin Solid Films 125, 1985, p. 57. 30. H.J.W. Lendednk, W.V.D. Linden and J.H.W. De Wit; Electrochim. Acta, 38, 1993, p. 1989.
428
Al-Kharafi et al.
31. S.E. Feres, M.M. Stefenel, C. Mayer and T. Chierche, J. Appl. Electrochem. 20, 1990, p. 996. 32. C.M.A. Brett; J. Appl. Electrochem. 20, 1990, p. 1000. 33. C.M.A. Brett; Corros. Sci. 33, 1992, p. 203. 34. S. Srinivasan and C.K. Mital, Electrochim. Acta 39, 1994, p. 2633. 35. E.H. Hollingworth and H.Y. Hunsicker, Corrosion Resistance of Aluminium Alloys in Metal Handbook, 9th ed., Vol. 2, 1990, Americam Society for metals, Metals Park, Oh, USA, pp. 204-236. 36. J. Bessone, C. Mayer, K. Juttner and W.J. Lorenz, Electrocim. Acta 28, 1983, p. 171. 37. P.L. Cabat, F.A. Centellas, J.A. Garrido, E. Perez and H. Vidal, Electrochim. Acta 36, 1991, p. 179. 38. W.M. Carroll and C.B. Breslin, Br. Corros. J. 26, 1991, p. 255. 39. E. Adem, VG Scientific XPS Handbook, 1st ed., VG Scientific Ltd., England, 1989. 40. W.A. Badawy and F.M. AI-Kharafi and A.S. El-Azab, J. Appl. Electrochem. (Submitted). 41. F.M. Al-Kharafi and W.A. Badawy, Corrosion (Submitted).
429
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
THE SUSCEPTIBILITY OF MOLYBEDNUM AND VANADIUM-BEARING AUSTENITIC STAINLESS STEEL WELDMENTS TO INTERGRANULAR CORROSION M.K. Karfoul College of Chemical Engineering and Petroleum Ba’ath University, Syria
ABSTRACT The development of the chemical, fertilizer, petrochemical, refining and energy industries depends, in most cases, on resolving the problems associated with the use and maintenance of stainless steels. The most important problem faced with the use of stainless steels is intergranular corrosion (IGC). This subject has attracted the attention of many research organizations for several years and still concerns many to date. Key Words: Austentic stainless steel, weld metal, intergranular corrosion
INTRODUCTION Nowadays, the metal manufacturing industries produce chromium and nickel-bearing austenitic stainless steels which are highly resistant to intergranular corrosion (IGC). This level of resistance is achieved by stabilizing the characteristics of the steels by adding titanium or niobium, or by decreasing the percentage of carbon in the steel to a very low level or both [1]. In most cases, these stainless steels are used in welded mechanical equipment which are also repaired by welding. Such uses, however, might lead to IGC (sensitization), especially when the steel remains at temperatures of 500-700°C during the welding process. Within this temperature range, depletion of chromium can occur at the grain boundaries in the heat affected zone. Welding or repeated welding promotes such a phenomenon. Therefore, it is deemed desirable to have a stainless steel alloy that resists sensitization during welding in order to prevent IGC. One of the basic methods used to stabilize the austenitic chromiumnickel stainless steel weldments is through the addition of titanium or niobium. After welding, steel weldments passes through the temperature range of 900-1000°C for a period of time. This allows the formation of carbide stabilizers such as Ti and Nb carbides [2-4], resulting in lowering the carbon content in the solid solution. However, when such an alloy reaches the critical temperature range for sensitization, i.e., 500-750°C, chromium carbide is formed (i.e., Cr23C6) along the grain boundaries. This results in dispersed and unconnected areas depleted in chromium. Therefore, the grain boundaries become less resistance to IGC than those that are stabilized with Ti or Nb. The percentage of these latter alloying elements in the steel should depend on the amount of carbon present in the steel.
431
Fundamental Aspects
This matter appears simple, but many complications, however, occur upon the addition of Ti or Nb. Titanium has a higher affinity for reaction with carbon than does chromium. Ti is also more reactive to oxygen than chromium or the residual elements such as Mn and Si are. Therefore, during electric arc welding, a titanium-stabilized steel electrode tends to react with oxygen and be consumed completely. To avoid such complications, inert gas is used during welding of such electrodes. However, this increases the cost of the components to be welded. Alternatively, titanium can be replaced by niobium with its lower oxidation rate as a stabilizer for the steel weld metal. When mechanical components are operated under severe corrosive conditions and/or are exposed for an extended period to a temperature in the range of 500-600°C, it is recommended that the amount of Nb to be 10-12 times greater than the percentage of carbon in the steel. In addition to the amount of Nb which is required in such weld metal, the amount of azote and carbon need to be calculated [4,5]. Therefore, more Nb is required than is mentioned above, and this leads to a loss of steel toughness and the appearance of hot short cracks in the weld metal during the welding process [6]. The sensitivity of welding materials to IGC increases with increases in carbon, titanium and niobium. When using a welding electrode with a maximum carbon percentage of 0.06, the sensitivity of weld metal may be increased. However, this does not negate the weld metals sensitivity to IGC, especially during solid welding processes or applied on different layers. This is related to the presence of large percentage of carbon in the welding material the migration of carbon from the base metal or the degradation of carbon containing materials that cover the welding electrode [7]. The chromium-nickel steels with niobium added had the tendency to crack during cooling. This action appears more than in castings, solid welding joints and during the welding of thick plates. This damage increases with the increase in acidity of the slag during welding [8] (especially for electrode materials containing cilium) which produces hot short cracks. Getting rid of compounds containing cilium in materials covering electrodes is impossible. The negative effect of niobium on hot short cracking is associated with the small dissociation of niobium in iron especially at the low eutectic melting temperature of Fe-Nb [9,10]. This eutectic effect cannot be avoided in practice unless it is associated with the amount of niobium and carbon. Niobium usually increases the ferrite phase in steels, especially if it is present in steel in a ratio of 1:10 with respect to carbon. Therefore, due to the inhomogeneous concentrations in weld metal, microscopic cracks appear in pure austenitic areas. This is expected since niobium activate cracking in weldments of pure austenite. One may thus conclude that stabilizing the chromium-nickel steels with niobium is associated with many technological difficulties. Molybdenum is a more positive stabilizing additive with this respect. Mo is known to posses a large degree of dissociation in iron, chromium and nickel. The eutectic melting point of Mo is not that different from the melting point of the original metals of Fe, Ni and Cr. Metals that contain Mo do not exhibit hot short cracks. Molybdenum is also known to promote the ferrite phase and has the affinity to react with carbon, but to a lesser degree than Nb with carbon. Therefore, Mo aids in raising the resistance of weld metals of chromiumnickel stainless steels to hot cracks. 432
Karfoul
Molybdenum plays a role in softening the microstructure of weld metals of pure austenite to hot cracks [11]. Molybdenum also increases the resistance of weld metals to corrosion, and increases its surface negativity [12.3]. The positive effect of Mo is exhibited in improving the resistance of weld metal to hot cracks and improving its technological soundness [14] or increasing the IGC resistance of weldments [15] in the presence of special electrodes for welding the austenitic stainless steel with the use of the common stabilizing elements. Vanadium possesses a great ability to dissociate in iron. It is strongly reactive with carbon, vanadium carbides are formed such as V4C3 and remain stable at higher temperatures. Therefore, vanadium is considered to be a stabilizer for carbon as carbides in the grains of the chromium-nickel stainless steel. Vanadium also plays a major role in promoting ferrite formation [16]; thus increasing the resistance of weldments to hot cracks. The effect of vanadium in reducing the IGC of chromium-nickel steels is not clear. The information available with respect to the subject is lacking. Therefore, the objective of this research was to evaluate the addition of vanadium to steel weldments with molybdenum, rather than of titanium and niobium, as to its resistance to IGC. EXPERIMENTAL PROCEDURE In this research, stainless steel specimens were prepared by electric arc welding of two sheets, 500 mm in length and 5 mm thick, made of chromium-nickel steels stabilized with titanium. These sheets were composed of: 16.5% Cr, 9.6% Ni, 0.66% Ti, 0.1%C and the balance was Fe. The welding system was chosen in a way to ensure minimum interaction with the base metal. The operational angle of the welded surfaces was 90°. The distance between the two plates was 2 mm, and the welding current and potential were 160 A and 25 V, respectively. The welding speed was 0.17 cm/s, and the diameter of the welding electrodes was 5 mm. Using this welding system, a single pass was applied. the base metal was protected by a copper sheet that was placed underneath the two welded plates. Stainless steel welding rods were used. They were composed of 0.06-0.08 wt.% C, 18.4 wt.% Cr, 11.00-11.38 wt.% Ni, 2.17-2.37 wt.% Mo, 1.35 wt.% Ti, 0.35 wt.% Si, 0.029 wt.% P, and 0.014 wt.% S. The proper chemical composition of the weld metals was achieved by adding the necessary elements to the substance covering the welding electrode. The chemical and phase compositions of the weldments [15] was based on physical and mathematical methods. The chemical and phase compositions of the weldments are shown in Table 1. Table 1 shows that the first three components of the weldment had a fixed composition except for their Mo content. The percentage of Mo was increased to 2, 3, and 6% to study the effect of the alloy on the weld metal’s resistance to IGC. In the forth composition, half of the Mo was replaced with vanadium, the percentage of Mo was 3 wt%. The percentage of Cr was kept at 17 in order to allow for the formation of a ferrite phase in the weld metal. In addition, a fifth composition was also tried with the same Mo and V percentages as in the fourth composition, but with 20 wt% Cr to maximize the ferrite phase in the weld metal.
433
Fundamental Aspects
Table 1a. The Chosen Chemical and Phase Compositions of Weldments Chemical Composition (wt%) Composition No C Mn Cr Ni Mo V 1 0.1 2.0 20 10.5 2 2 0.1 2.0 20 10.5 3 3 0.1 2.0 20 10.5 6 4 0.1 2.0 17 10.5 3 3 5 0.1 2.0 20 10.5 3 3
Phase Composition S(%) Grain Diameter (μm) 4 20 5 20 10 20 4 20 10 20
Table 1b. Actual Experimental Chemical Compositions of Weldments
Composition No. 1 2 3 4 5
C 0.11 0.11 0.11 0.11 0.11
P 0.03 0.028 0.027 0.033 0.03
Chemical Composition (wt %) S Si Mn Cr Ni Mo 0.018 0.13 1.83 20.8 10.55 1.9 0.018 0.16 1.74 20.0 10.96 3.1 0.018 0.20 0.81 19.7 9.84 6.38 0.018 0.20 1.79 16.6 10.50 2.95 0.018 0.20 1.80 19.5 10.85 2.77
V 2.85 1.86
Table 1c. Chemical and Phase Compositions of Weldments
Composition No. 1 2 3 4 5
δ (%) 4.7 4.4 8.2 6.0 9.0
Phase Composition Grain Diameter (μm) 21 17 19 17 17
The amount of the ferrite phase on the weld metal specimens was determined magnetically by an α-phase meter with ± 5% error, as is shown in Fig. 1. These same specimens were also used to study the weld metals resistance to IGC. RESULTS A metallurgical microscope was used to measure the austenite grain diameter of the weld metals, as shown in Fig. 2. After being welded, the specimens were prepared (Fig. 1) and heat-treated at a temperature range of 500-800°C for different time periods as shown in Table 2. Then, the specimens were quenched with water. To determine the susceptibility of the specimens to IGC after the heat treatment, they were all immersed in Strauss solution for 24
434
Karfoul
hrs. Strauss solution is made up of 100 gr of CuSO4 5H2O + 0.1l H2SO4 + 1l of distilled water + thin copper sheets.
Figure 1. The Location of the points where the δ ferrite phase was measured in the welded metal
Composition No. 1
Composition No. 2
Composition No. 4
Composition No. 3
Composition No. 5 435
Fundamental Aspects
Figure 2. Microstructures of the weld metals for the different studied compositions Table 2. Time in Minutes of the Different Heat-Treatment Temperatures Studied 550
600
650
700
750
800
1
Heat Treatment Temperature (°C) -
25 50
3 5 10 25
10 25
10
25
3
500
55 110 300 500
5 30 50 10 300 10 30 50 110 300 50 10 25 50 100 300 500 25 50 100 300 500
0.5 1 3 5 10 5 10 30 50 110 300 500
5 10
2
1 3 5 10 100 5 30 50 300 10 30 50 110 300 500 10 25 50 100 300 500 25 50 110 300 500
5 10 30 50 90 -
5 10 30 50 100 -
4
50 100 300 500
5
50 100 300 500
10 25 50 100 300
60
25 50 110 300 500
50 110 110 300
After being boiled for 24 hours. in Strauss solution, the specimens were bent to a 180° angle. If no cracking was observed in the bent samples, they were considered to be resistant to IGC. DISCUSSION The aqueous solution of copper sulfate and sulfuric acid was chosen because it attacks only the regions of the specimens with chromium contents of < 12%. This negates the use of weight-loss or other methods to check whether or not such specimens are susceptible to IGC. On the other hand, results obtained using acetic acid may require the determination of the weight loss of specimens because IGC might not be well defined. This is because in this case, the acid attacks regions containing more than 12% Cr and removes grains from the 436
Karfoul
metal surface. Therefore, weight -loss measurements are not representative of intergranular corrosion. It can be seen in Figs. 3, 4 and 5 that the addition of 6 wt% Mo to the weldments increased greatly the IGC resistance of the austenitic stainless steel weld metal. It also increased annealing of the microstructure of the weld metal, resulting in increasing the austenitic grain boundary surface area and thus not allowing the precipitation of Cr23C6 and preventing the depletion of Cr around the grains. The addition of Mo also aids the formation of δ ferrite phase in the microstructure of the weld metal and thus decreases the IGC susceptibility of the metal. It was also observed from Figs. 6 and 7 that the replacement of half of the Mo by V in the weld metal of the composition No 3 did not change the resistance of the alloy to IGC, indicating similarity of V and Cr. However, the increase in the amount of δ ferrite in the microstructure of the weld metal tended to increase the resistance of Cr-Ni weld metals to IGC, as shown in Fig. 7.
Figure 3. Response of composition No. 1
Figure 4. Response of composition No. 2
Figure 5. Response of composition No. 3
Figure 6. Response of composition No. 4
437
Fundamental Aspects
Figure 7. Response of composition No. 5 CONCLUSIONS The results of tests conducted on the Cr-Ni austenitic stainless steel showed that the addition of either Mo or V to the weld metal possess similar effects on the sensitivity of the weld metal to I.G.C. The addition of Mo and V tends to shift the Rolason curves down and to the right, indicating an increase in the resistance of the weld metal to IGC. REFERENCES 1.
E.C. Bain, R.H. Aborn and J.J.B Rutherford, The nature and prevention of intergranular corrosion in austenitic stainless steels, Transactions of the American Society for Steel Treating, Vol. XXI, No. 1, 1933, pp. 481-509.
2.
V.V. Levitin and V.I. Cirashikova, precipitation of carbides on the grain boundaries during the production of austenitic steel, Metallovedmie and Thermocheskoe, No. 8, 1960, pp. 20-25. V. Chihal, N. Lehka and J.K. Malik, Probleme der interkristalline und der hesser tinienkorrosion van schweibvindunger der mit niob stabilisierten korrosiosbestandiger stahle. Metalloberflache angewandte elekrochemie, Iq. 26, heft. 12m 1972, pp. 453-558. W.O. Binder, C.M. Brown and S.R. Frank, Resistance to sensitization of austenitic chromium-nickel steels of 0.03% max carbon content, Transactions of the American Society for metals (ASM), Vol. 41, 1994, pp. 1301 - 1371. R.A. Mulford, E.L. Hall and C.L. Brint, Sensitization of austenitic stainless steels. II Commercial purity alloys, Corrosion 39, 4, April 1983. I.Z. Kogan, Niobium in welding electrodes, Heat resisting, No. 4, 1951, pp. 1-3. H.J. Rocha, Die sensibilisiertes austenisches stahle durch chromkarbide, Zeitchrift fur Schweisstechnik, No. 3, 1962, pp. 98-106. B.I. Medovar, Welding of heat resisting austenitic stainless steel and alloys, Moscow, Machinostrorkue, 1966, pp. 202 G.L. Petrov, V.N. Zemzin and F.G. Goncherovsky, welding of heat resisting stainless steels, Moscow, Petersburg, Machgas, 1963, pp. 53
3.
4.
5. 6. 7. 8. 9.
438
Karfoul
10. M.Ck. Sharshorov, Hot cracks during the welding of heat resistant alloys, Moscow, Machinostructure, 1973, pp. 183. 11. B.A. Movchan, Edges of crystals in cast metalls and alloys, Keef, Teschnika, 1970, p. 151. 12. F.F. Ckhimushin, Stainless steel, Metallorg, Moscow, 1967, p. 351. 13. F.R.G. Patent, No. 148342, 29, 35/30; February 1973, Welding electrods. 14. M.K. Karfoul, Syrian patent No. 4405, 13 December 1992, “Welding electrodes for stainless steel weldments”. 15. M.K. Karfoul, Intergranuler corrosion of austenitic weld metal type 18-10 with molybdenum, The Sixth Middle East Corrosion Conference, Conference Proceeding, NACE, Vol. 1, pp. 313 - 328. 16. E. Houdermont, Handbuch der Soonder Stahlkunde, Springer-verlag, Berlin/Gohingen/Heidd, 1956, Russian Translation, Moscow, 1960, p. 1064.
439
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
EFFECT OF CRYSTALLIZATION ON THE CORROSION BEHAVIOR OF AMORPHOUS FeCr9P6C3Si0.2 ALLOY IN 1M H2SO4 F. Hajji1,2, S. Kertit1, J. Aride1 and M. Ferhat2 1
Laboratoire de Physico-Chimie des Matériaux associé l'AUPELF.UREF (LAF502), Ecole Normale Supérieure de Takaddoum, B.P. 5118, Rabat, Morocco 2
Laboratoire de Chimie-Physique Générale, Faculté de Sciences, Rabat, Morocco
ABSTRACT The crystallization of amorphous FeCr9P6C3Si0.2 alloy was investigated by x-ray diffraction (XRD), differential thermal analysis (DTA) and scanning electron microscopy (SEM). Specific differential heat curves of the FeCr9P6C3Si0.2 alloy exhibited three exothermic peaks indicating that the crystallization of the amorphous alloy occurred through the formation of three kinds of metastable crystalline phases. The deterioration of corrosion resistance by the crystallization of the amorphous FeCr9P6C3Si0.2 alloy was studied by electrochemical methods to correlate the corrosion behavior with the increase in the heterogeneity of the alloy. As soon as the stage formed in the amorphous matrix, the anodic current density increased. The current density in the active and passive regions increased continuously in sulfuric acid solution during the crystallization processes. Chemical heterogeneity, based on the formation of precipitation segregation and other compositional fluctuations, seemed to be responsible for the deterioration of the corrosion resistance. Keys Words: Amorphous alloys, corrosion, crystallization, potentiodynamic measurements
INTRODUCTION The corrosion behavior of amorphous alloys was first studied in 1974 [1]. It was then extensively reviewed [2-6] for simple binary alloys such as Fe80B20 as well as commercial, multicomponent systems. Amorphous iron-based alloys prepared by quenching from the liquid state contain a large amount of various metalloid elements that stabilize the amorphous structure in the solid state. Among the metalloid elements, phosphorus is the most effective at concentrating chromium in the passive film [7], and hence, the passive film formed on the amorphous alloys contained phosphorus and a small amount of chromium consisting exclusively of hydrated chromium oxyhydroxide. This is partly responsible for the extremely high corrosion resistance of these amorphous alloys [8-9]. Amorphous alloys containing P were almost always more corrosion resistant than alloys containing B, Si or C. However, the alloys containing P became much less corrosion resistant [10,11] after treatment at the crystallization temperature, at which point P migrated into the grain boundaries and caused severe intergranular corrosion. The chemical heterogeneity seemed to form a high density of weak points in the passive film with respect to corrosion as well as localized corrosion attack. The heat treatment of amorphous alloys gives rise to the formation of various metastable crystalline phases (MS) in the amorphous matrix before the formation of stable crystalline 441
Fundamental Aspects
phases [12]. In this work, amorphous FeCr9P6C3Si0.2 alloy was used. Mainly potentiodynamic polarization experiments were conducted to understand the electrochemical corrosion of the alloy. The effect of heat treatment on the behavior of the amorphous alloy in aerated sulfuric acid media was investigated. EXPERIMENTAL PROCEDURE Amorphous FeCr9P6C3Si0.2 alloy ribbons 2 mm in width and 10 μm in thickness were produced by a rapid quenching technique (melt-spinning). The number attached to a respective element in the alloy formula denotes the nominal content in atomic percentage. After isothermal heat treatment of the alloy in an evacuated quartz tube at various constant temperatures at a gas pressure of 3.10-6 torr, diffractometric measurements were made using a diffractometer, with Co Kα radiation at a scanning speed of 8o/minute. Differential thermal analysis (DTA) at a heating rate of 10 K/minute was carried out to confirm the multistage crystallization which produced multiple exothermic peaks. The crystallization process was also examined by a scanning electron microscopy (SEM) after undergoing isothermal heat treatment. The electrochemical experiments were performed with an Amel potentiostat system. Anodic polarization curves for the alloy were measured potentiodynamically with a potential sweep rate of 1 mV/s. The electrochemical measurements were conducted in unstirred, aerated 1 M H2SO4 solution which was prepared using a reagent. The alloy was not mechanically polished. Both sides of the ribbons were immersed in the test solutions. All the experiments were carried out at room temperature. Polarization studies were conducted in a simple electrolytic cell at three electrode using platinum counter electrodes and a saturated calomel (reference) electrode (SCE). RESULTS Differential Thermal Analysis (DTA) Measurements The DTA curves for the amorphous FeCr9P6C3SiO.2 alloy exhibited three exothermic peaks as shown in Fig. 1. The first peak was at about 448°C, the second peak was at 489°C and the final peak was at 578°C. This indicates that the crystallization of the amorphous alloy occurred through the formation of three kinds of MS. In the case of heat treatment at 300 and 400°C, the pace of the thermograph was not modified. However, the DTA curves for the alloy treated at 500°C for 1, 2 and 3 hours exhibited only one exothermic peak. In this case, the crystallization was only partial. But, for the alloy treated at 500°C for 8 hours or at 600, 700 and 800°C for 1 hour, crystallization was total. X-Ray Diffraction (XRD) Measurements The x-ray diffraction (XRD) patterns for the amorphous FeCr9P6C3SiO.2 alloy and a series of isochronal heat treatments are shown in Fig. 2. It is evident that the as-quenched state was typical of the amorphous state, and no crystalline phases were observed. With annealing at 300 and 400°C the alloy stayed amorphous. When the temperature was increased, the x-ray pattern became totally crystalline. As shown in Fig. 2, XRD patterns revealed that the crystallized FeCr9P6C3SiO.2 alloy consisted of many phases. XRD confirmed the presence of α-Fe as an fcc phase with a = 2,866 Å in the ribbons annealed at 442
Hajji et al.
500°C. Because of the very complicated nature of the diffraction patterns and the intense diffraction lines of the α-Fe phase, an accurate analysis of the lattice constants of the other phases was not possible. The crystallization of various amorphous metal-metalloid alloys has been studied by many authors. When the metalloid concentration is not exceedingly high, such as at 28 atomic percentage, the crystallization of amorphous alloys generally takes place as follows: the heat treatment of the amorphous alloy gives rise to the precipitation of a MS in the amorphous matrix. This phase contains a large amount of the main metallic component of the alloy and thus it has the same crystal structure as the main metallic component. The amorphous phase then disappears by the formation of two or three MS, through transformation diffusion of various elements and by recrystallization and/or decomposition of the metastable phases.
Figure1. Differential thermal analysis curves Figure of amorphous FeCr9P6C3Si0,2 alloy before and after isothermal heat treatment with a heating rate of 10 K/minute
2.
X-ray diffraction patterns FeCr9P6C3Si0,2 alloy before and after isothermal annealing at different temperatures and times 443
Fundamental Aspects
Figure 3. Breaking faces of the alloy FeCr9P6C3Si0,2 thermally treated at 300°C for 1 hour (a), 500°C for 1 hour (b) and 800°C for 1 hour (c), Micrograph of the shining face of the alloy FeCr9P6C3Si0,2 after annealing at 500°C for 1 hour (d), 800°C for 3 hours (e) and mate face at 800°C for 3 hours (f). Micrograph of the alloy FeCr9P6C3Si0,2 after annealing at 500°C for 1 hour (g). x-ray cartography of the shining face of the alloy FeCr9P6C3Si0,2 after annealing at 800°C for 1 hour (h) Scanning Electron Microscopy (SEM) Scanning electron microscopy (SEM) images obtained on breaking faces that have been thermally treated are shown on Fig. 3. For samples annealed at 300 or 500°C for 1 hour, these alloys formed in a disordered grain stacking. They seemed to have a spherical form, and their size increased with increasing temperature (Fig. 3a and 3b). On the other hand, when the temperature and the annealing time increased, the alloy presented a mixture of two aspects: a granular aspect and a column-like aspect (Fig. 3c). The observed crystallization was probably due to an intrinsic heating up of the alloy during the annealing process. Annealing of the alloy did not change the state of the external surface of the alloy. Indeed the state of the surface stayed amorphous, as shown in the SEM micrograph (Fig. 3d). In the case of the alloy annealed at 800°C for 3 hours, the state of the surface of the shining face (Fig. 3e) seemed to be formed by a granular stacking of a quasi-spherical form, even if the mate face 444
Hajji et al.
always stayed amorphous (Fig. 3f). In order to clarify this phenomenon, note that the micrograph in Fig. 3g, obtained for a sample annealed at 500°C for 1hour, presents a mixture of two different crystallographic aspects: an amorphous aspect (alloy surface) and a crystal aspect (inside the alloy). An x-ray cartograph giving the distribution of the Fe, Cr, P and C elements of a sample annealed at 800°C for 1 hour showed good homogeneity of the shining surface (Fig. 3). This confirms that the surface of the sample was not affected by the temperature. These results could be explained by the fact that increases in the annealing temperature and time led to an increase in the molecule mobility inside the alloys before their freezing into a crystal state. Therefore, the grain size increased with annealing temperature and time. Nevertheless, this increased mobility did not lead to a perfect atomic order which would have filled the entire volume of the alloy, mainly the external surface as shown by the SEM results. The inside of the alloy was then formed by a stacking of micro-crystallites separated by a granular boundary which are probably disordered. Furthermore during the growth of such microcrystals, depending on the annealing temperature and time, it is possible that some impurities formed around the grains with compositions quite different from those of the crystallites.
Figure 4. Change in the anodic polarization curves of amorphous FeCr9P6C3Si0,2 alloy before and after annealing in 1 M H2SO4 with 1 hour of heat treatment at 300°C, 400, 500, 600, 700 and 800°C Electrochemical Measurements 445
Fundamental Aspects
Figure 4 shows changes in the anodic polarization curves of the alloy measured in 1 M H2SO4 according to the temperature of heat treatment at different time periods. The anodic polarization curve of the untreated alloy is also shown in Fig. 4 for comparison. The various anodic parameters determined from these curves are given in Table 1. The curve for the amorphous alloy exhibits a typical active-passive transition (Fig. 4), and passivation in 1 M H2SO4 solution. When the alloy was amorphous with the annealing (300 and 400°C), the activation current density began to increase slowly. However, as soon as the first crystalline phase was formed in the amorphous matrix, the speed of the activation current density began to increase. The corrosion current density for activation (Ic) and for passivation (Ip) was dependent on the annealing temperature. Before annealing, the activation current density (Ia) was 34 μA/cm², and after annealing (Ic) became 40, 64, 726, 13793, 32307 and 47984 μ A/cm² at 300, 400, 500, 600, 700 and 800°C for 1 hour, respectively. The current density in the active and passive regions continuously increased during all the stages of nucleation. The crystallized alloy also exhibited a wide passivity range. The superior corrosion resistance of the amorphous alloy decreased with the crystallization of the alloy in 1 M H2SO4 solution. Table 1.
Electrochemical Parameters of the Amorphous FeCr9P6C3Si0,2 Alloy Before and After Isothermal Heat Treatment at 1 Hour in 1 M H2SO4 Solution
Amorphous 300°C 400°C 500°C Ecor(mVvsS.C.E) -310 -310 -315 -365 34 40 64 726 Ia (μA/cm²) 15 13.2 27 222 Ip (μA/cm²)
600°C 700°C -375 -382 13793 32307 1034 462
800°C -380 47984 161
DISCUSSION The present results revealed clearly that the formation of a crystalline phase in the amorphous matrix increased the anodic current density. The anodic current density of FeCr9P6C3Si0.2 alloy in the active and passive regions continuously increased during the growth of metastable phases crystallites. Therefore, the crystallization of the amorphous alloys led to the appearance of chemical heterogeneity with a subsequent increase in the current density in the active region. The appearance of chemical heterogeneity also increased the current density in the passive region. It has been assumed [13-16] that the passive film was not essentially uniform but contained weak points (micropores) which were responsible for the apparent passive current density in aggressive solutions. The micropores could be formed on heterogeneous sites of the underlying alloy surface as well as on the phases which are relatively difficult to passivate. Accordingly, the formation of chemically heterogeneous sites in the alloys by heat treatment increased the passive current density. The question raised is whether or not the chemical heterogeneity, in comparison with structural heterogeneity, is a dominant factor in decreasing the corrosion resistance. The present authors [17] have shown that the rapidly quenched, single phase alloys showed significantly high corrosion resistance in comparison with the corresponding ordinary crystalline alloys. Naka et al. [11] have reported that the passive current density of the Fe-10Cr-13P-7C alloy increased by two orders of magnitude due to the formation of the solid solution phase in the amorphous matrix. 446
Hajji et al.
Therefore, rapidly quenched single phase alloys show significantly high corrosion resistance. In contrast, heat treatment inevitably induces solid state diffusion and hence results in various compositional fluctuations such as precipitation, segregation, and other composition gradients. These compositional fluctuations may act as dominant active surface sites with respect to corrosion. CONCLUSIONS It can be concluded that crystallization of this amorphous alloy FeCr9P6C3Si,0.2 did not considerably alter its excellent corrosion resistance as long as the alloy remained a single phase solid solution. This suggests that structure may not be a dominant factor in determining the corrosion resistance of this amorphous alloy. When the crystallization was complete, the corrosion resistance of the alloy deteriorated significantly. REFERENCES 1. M. Naka, K. Hashimoto and T. Masumoto, J. Japan. Inst. Metals 38, 1974, p. 835. 2. Y. Waseda and K.T. Aust. J. Mater. Sci. 16, 1981, 2337. 3. R.B. Diegle, N.R. Sorensen, T. Tsuru and R.M. Latanision, in Treatise on Materials Science and Technology (Edited by J. Scully), Vol. 23, p. 63, Academic Press, London (1983). 4. K. Hashimoto, in Amorphous Metallic Alloy (Edited by T.E. Luborsky), p.471, Butterworth, London (1983). 5. M.D. Archer, C.C. Corke and B. H. Harji, Electrochim. Acta 32, 1987, p. 13 6. P.C. Searson, P.V. Nagarkar and R.M. Latanision, in Modern Aspects of Electrochemistry (Edited by R.E. White, J.O. Bockris and B.E. Conway), No. 21, pp.121-161. Plenum Press, New York (1990). 7. K. Hashimoto, M. Naka, K. Asami and T. Masumoto, Corros. Eng. 27, 1978, p. 279. 8. K. Asami, K. Hashimoto, T. Masumoto and S. Shimodaira, Corros. Sci. 16, 1976, p. 909. 9. K. Hashimoto, K. Asami, K. Asami, and T. Masumoto, Corros. Eng. 28, 1979, p. 271. 10. R.B. Diegle and D.M. Lineman, Interim Technical Report No. 0NR-00014-77-C-0488-3 to the Office of Naval Research. 11. M. Naka, K. Hashimoto and T. Masumoto, Corrosion 36, 1980, p. 679. 12. T. Masumoto and R. Maddin, Acta Metall. 19, 1971, p. 725. 13. K. Hashimoto, K. Asami and K. Teramoto, Corro. Sci. 19, 1979, p. 3 14. K. Hashimoto and K. Asami, Corros. Sci. 19, 1979, p. 251. 15. K. Hashimoto and K. Asami, Passivity of Metals, Proceedings, 4th Intern. Symp. on Passivity of Metals, (Edited by R.P. Frankenthal and J. Kruger), the Electrochemical Society, Princeton, New Jersey, p.749 (1978). 16. K. Sugimoto and Y. Sawada, Corros. Sci. 17, 1977, p. 425. 17. M.Naka, K. Asami, K. Hashimoto and T. Masumoto, Proceedings, 4th International Conference on Titanium (1980).
447
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
EXPERIENCE WITH VOC-COMPLIANT WATERBORNE AND HIGH SOLIDS COATINGS IN CORROSIVE ENVIRONMENTS P Kronborg Nielsen HEMPEL Coatings, Lyngby, Denmark
ABSTRACT Waterborne and high solids paints-VOC compliant paints are becoming more important in coating specifications for environmental reasons. They contain between 0% and 35% organic solvents (VOC) per litre, compared to standard paints, where VOC often contributes 60% or more. High solids paints, such as low solvent epoxies, have been used successfully in, for example, submerged areas. However, petrochemical installations are still often coated with standard paint types. In recent years, official tests, practical demonstrations and case histories have shown that waterborne and low solvent coating systems protect steel in aggressive environments on a par with standard paints, especially in cases when high solid epoxy primers are combined with waterborne acrylic top coats-the hybrid system. Calculated per square meter applied and considering the longer maintenance free intervals, the cost of high solids and hybrid systems is not excessive. Painting contractors and specifiers can thereby meet upcoming VOC-legislation on sound economic and technical bases. Key Words:
Waterborne coatings, high solids coatings, corrosion, VOC, coating specification, petrochemical industry
INTRODUCTION For environmental reasons, coating specifiers and painting contractors are today constrained to using low volatile organic content (VOC) coatings for an increasing number of painting jobs. The low VOC coatings are paints with a reduced content of organic solvents, and they are present in the market in the form of waterborne paints and coatings with high solids. In less corrosive environments, the long-term performance of the low VOC coatings is now recognized to be fine. Their performance is often even above that of comparable standard coatings such as alkyd, acrylic and epoxy [1]. However, in corrosive environments, experience especially with waterborne coatings, is still limited. Lately, though, the offshore market and independent testing laboratories have shown that coating systems with low VOC and waterborne paints are well suited for severe corrosive environments. These systems may reduce VOC emission by 70% or more, and the cost on an applied-square-meter basis is not excessive. Thus, the new coating technology is in full accordance with the principle of BATNEEC, the Best Available Technology Not Entailing Excessive Cost, which is statutory for all environmentally directed developments. EFFECT OF VOC's
449
Corrosion Protection and Monitoring
The VOC in paints are organic solvents, and solvents are necessary to facilitate production and application. But once the paints are applied, solvents are only a nuisance. They are inflammable, and have a negative influence both on man and nature. A predisposed painter's long-term continuous exposure to organic solvents will have a negative effect on his
• • • •
Respiratory system, Nervous system, Capacity for reproduction, and Skin. Various governments and civil councils have, therefore, introduced health and safety measures to protect painters, such as the Control of Substances Hazardous to Health (COSHH) regulations in force in the UK. When organic solvents evaporate, they are decomposed by the ultraviolet radiation from the sun. The decomposed molecules are highly reactive and easily form compounds with the exhaust from automobiles and industrial air pollution. These chemical reaction products will affect the local environment, and eventually cause smog and reduce metabolism in human beings, animals, and plants. In Europe and the USA, VOC emission is being addressed by various legislative measures, e.g., organic solvents from painting processes by an European Union Directive [2] to be implemented into a law. The directive requests a solvent management and reduction plan, and sets limits on emissions, but it is aimed at the user of the paint, i.e., the painting shops. Upcoming British and existing American laws address the solvent content in the paint itself. The aim of both types of regulation is to reduce the overall VOC emissions. It is expected that the actions laid down in the European Union Directive-once introduced-will reduce the solvent emissions in European Union member states by 30% in 1999 (compared with 1985). Paint manufacturers today have products in their assortment that can meet these regulations, among which are the waterborne and the high solids (low VOC) products for the protective coating of steel structures. The user-the painting contractor-may address an emission directive by installing an abatement system in his plant, but this could be a costly solution. A better way is to modify working procedures and adjust the equipment to handle waterborne and/or high solids paints, and to train the applicators accordingly. COATING SYSTEMS WITH REDUCED SOLVENT EMISSIONS The satisfactory performance of paint coatings for the protection of steel structures against corrosion is determined by
• The choice and formulation of the products used in differently classified environments, and • The standard of workmanship and execution of the contract. Agreement between the client and the contractor as to the specifications to be applied is essential to the satisfactory execution of the work. The paint producer may also introduce specifications, or they can be made in accordance with national standards, such as BS 5493, 450
Kronborg Nielsen
DS/R 454, or DIN 55928/5. There has, however, been a long-felt wish to standardize coating specifications and workmanship. Therefore, at the time of writing (1996), the secretariat of the International Organization for Standardization (ISO) has established working groups which have already presented a series of working drafts on, among other topics, surface preparation, classification of environments and protective paint systems (ISO 12944 [3]). The various corrosiveenvironments classifications in ISO 12944-2 [3] are given in Table 1. Industrial and coastal areas are in Category 4, and Category 5M covers marine and aggressive areas. ISO/CD#1 12944-5 (Table 2) contains a wide selection of paint systems suited for each of these environments. Their performance has been confirmed after long experience and/or series of successful trials. In order to make the ISO standard suitable for the future, systems with low solvent emissions are also included. The paint systems are water dilutable, contain high solids, or may be combined (hybrids). In Figs. 1 and 2 are typical examples of the emission from the three types of systems included in Categorys 4 and 5M. They are compared with normal standard systems for the same environments. Table 1. ISO 12944-2 Classification of Environments Corrosivity Category C1: Very low
C2: Low
C3: Medium
C4: High
Examples of Typical Environments in a Moderate Climate Exterior -
Atmospheres with low pollution and dry climate. Mostly rural areas. Urban and industrial atmospheres, moderate sulphur dioxide pollution. Moderate coastal climate. Industrial and coastal areas.
Interior Inside, heated buildings with neutral atmospheres, and relative humidity below 60%, e.g., offices, shops, schools and hotels. Unheated buildings where condensation may occur, e.g., depots and sports halls. Production rooms with high humidity and some air pollution, e.g., food processing plants, laundries, breweries, and dairies. Chemical processing plants, and boat yards over seawater. -
C5: Very high Industry and areas high humidity (industry) and aggressive atmosphere. C5M: Very high Marine coastal, offshore, areas with (marine) high salinity. Table 2. Selected ISO/CD#1 12944-5 Coating Systems for Corrosion Categories C4 and C5M, Aggressive Industrial and Marine Areas Corrosivity Category &
Paint System
Dry Film Thickness
Solvent Emission
451
Corrosion Protection and Monitoring
Number
(g/sqm) (μm) 66 Chlorinated rubber primer 60 66 Chlorinated rubber primer 60 72 C4 - 5 Acrylic intermediate 60 60 72 Acrylic Top Coat 240 276 Standard solvent-borne system Ethyl zinc silicate primer 67 80 C4 - 16 Waterborne epoxy intermediate 5 60 60 5 Waterborne epoxy top coat 77 200 Hybrid system 13 High solids epoxy primer 80 13 C4 - 7 High solids epoxy intermediate 80 80 13 High solids epoxy top coat 240 39 High solids system 2 Waterborne zinc epoxy 40 3 Waterborne acrylic intermediate 60 0.5 C4 - 10 Waterborne acrylic top coat 50 50 0.5 Waterborne acrylic top coat 200. 6 Waterborne system 67 Ethyl zinc silicate primer 80 52 Epoxy intermediate 60 43 C5 - 10x 1) Epoxy intermediate 50 50 40 Polyurethane top coat 240 175 Standard solvent-borne system 40 21 High solids zinc epoxy primer 150 24 C5M - 7x High solids intermediate 50 0.5 Waterborne acrylic top coat 2) Hybrid system 240 ∼46 C5M - 3 High solids epoxy primer 24 150 150 24 High solids epoxy top coat 300 48 High solids systems 40 2 Waterborne zinc epoxy 70 6 Waterborne epoxy intermediate 70 6 C5M - 6x Waterborne epoxy intermediate 60 0.6 Waterborne acrylic top coat 3) Waterborne system 240 ∼14 Corrosivity Categories: C4: Industrial and coastal areas (ISO/WD 12944-2) C5M: Marine areas with high salinity and corrosive areas All systems are claimed in ISO/CD#1 12944-5 to have an expected medium durability in the respective corrosion categories. Notes: 1). In ISO/CD#1 12944-5 this system is a 5 coat system 2). In ISO/CD#1 12944-5 the top coat is chlorinated rubber 3). In ISO/CD#1 12944-5 the top coat is polyurethane
452
Kronborg Nielsen
Figure 1. Solvent emission from coating systems for industrial and coastal areas
Figure 2. Solvent emission from coating systems for marine and aggressive areas The paint systems mentioned are only examples of many possible combinations having the same performance. However, a considerable decrease in emissions is possible when one of the waterborne or high solids/low VOC systems is selected. 453
Corrosion Protection and Monitoring
EXPERIENCE WITH WATERBORNE AND LOW VOC COATINGS Application High solids paints are normally applied without problems if the application is carried out with heavy-duty airless spraying equipment, e.g., pumping at least 45:1. On the other hand, waterborne coatings are more delicate because of their nature and their weather window, i.e., limited by temperature and relative humidity during application. The most frequently observed mistakes are as follows
• Mixing waterborne paint with thinners which results in clogging spray equipment. Preventive action includes cleaning spraying equipment carefully with thinner followed by fresh water before the waterborne paint is introduced.
• Applying waterborne paint on cold steel and/or at low temperatures which results in insufficient curing and poor resistance. Preventive action includes painting indoors in ventilated, heated facilities (ambient temperatures of 5°C are sufficient for waterborne acrylics), or making covers with heating, if possible. If not possible, low VOC systems should be used.
• Exposing newly waterborne painted objects to frosty weather which results in the cracking of coating films. Preventive action includes keeping coated objects away from frost for at least 24 hours after the application of waterborne paints.
• Using waterborne paint indoors in areas without ventilation which results in runners and slow drying. Prevention action includes allowing sufficient ventilation to extract the water liberated during the application; for 20 l of paint more than 10 l of water have to be removed in the form of vapor. The ventilation requirement is at least 75 m3 air/l paint at 20°C and 50% relative humidity. The above errors can be overcome by changing the painting procedures and by training the applicators. Performance Paint manufacturers, specifiers and societies use a number of accelerated test methods to predict the lifetime performance properties of coatings. In particular, the corrosion resistance of coating systems is important. Some of the test methods used are
• Salt spray test (ISO 7253, ASTM B-117), • Continuous condensation test (ISO 6270), and • Prohesion chamber cyclic test (No ISO yet). The salt spray test has been used for a number of years. However, its inability to demonstrate a direct relationship between the resistance of organic coatings to the action of salt spray and resistance to corrosion in other (natural) exterior environments is now acknowledged. Actually, for a number of years, the salt spray test averted the introduction of waterborne coatings. The low dry-film thicknesses of systems with these coatings (dft 50 -100 μm) fail quickly in the salt spray test, but perform well at exterior exposure sites, as demonstrated by Andrews et al. [4]. Salt spray tests are, however, valuable in comparison situations for high dft, high solids, and solvent-borne systems. 454
Kronborg Nielsen
In the prohesion chamber cyclic test, the coated panel is dried between the salt spray and the ultraviolet radiation cycles. Thereby more reliable test results, i.e., results with correlations to genuine exterior situations, are obtained. Waterborne and low VOC coating systems have performed satisfactorily in a number of accelerated cyclic tests in comparison with standard high solvent systems. The results have been confirmed by exterior exposure. PERFORMANCE IN CORROSION CATEGORY 4: INDUSTRIAL AND COASTAL AREAS Waterborne coatings have been tested and applied since 1984 on dry cargo containers for carriage by sea, and the obtained experience has only been positive [1]. Today more than 3000 containers are in service, either with waterborne systems only (zinc epoxy/epoxy/acrylics), or with a hybrid system made of solvent-borne (zinc) epoxy prime coat(s) followed by waterborne acrylic top coats. In Kuwait, the Kuwait Institute for Scientific Research has compared 11 coating systems in two laboratory cyclic tests and at five exposure sites in Kuwait [5]. The sites selected for the outdoor panel exposure had different environmental parameters of the industrial area and were at varying distances from the Arabian Gulf. Inorganic zinc silicate (IOZ)/epoxy/polyurethane systems and waterborne acrylic systems performed better than other systems, both in the laboratory (3000-hour test) and at the sites (two year exposure). Also in the Gulf area, the petrochemical market has discovered that the exterior of storage tanks can be advantageously finished with an acrylic waterborne top coat. The traditional system of zinc epoxy/epoxy/polyurethane is occasionally being replaced by zinc epoxy/high solids epoxy/waterborne acrylic on tank farms. Although the gloss retention of the polyurethane top coat is slightly superior, the new system has three distinct advantages:
• Cost. Calculated per square meter, the IOZ/high solids epoxy/waterborne coating system is 10-15% cheaper than an IOZ/epoxy/polyurethane system (dft 225 μm).
• Application. A one-pack product like a waterborne acrylic is easier to apply than a two pack polyurethane. Additionally, pot-life problems are avoided.
• Lower VOC-emission. VOC emission is reduced by 70% (Fig. 3). The change from polyurethanes to waterborne acrylics has also been introduced in the UK. On the exterior of tanks at a major oil terminal near the coast of Southhampton, a test program has been concluded, with the result that a number of tank externals were primed in 1995 with high solids (low VOC) epoxies and finished with waterborne acrylics [6]. Also in the UK, waterborne acrylics have been used, among others, for
• British Gas' gasholders. Many of the gasholders seen near the major cities are now maintained with waterborne coatings with excellent performance results.
• Maintenance of ships' interiors. Maintenance of bulkheads, deckheads, engine rooms, etc., is normally carried out with alkyd paint onboard or during dry-docking. However, the Royal Fleet Auxiliary (RFA) recognized in 1990 that a switch to
455
Corrosion Protection and Monitoring
waterborne coatings has several advantages over solvent-borne systems. First there is no fire risk. Fire and explosion risks from paint are eliminated, during both application and storage. Additionally, waterborne paints have low flame spread characteristics once they are applied. Second, no thinner is required, water is the diluent and is also used for the cleaning of application tools. Third, there is less odor and less inconvenience. Waterborne coatings are popular for use indoors because they allow other trades to work in the immediate vicinity, so painters do not have to work night shifts. Fourth, waterborne coatings have good protective properties. The performance of the coatings in respect to gloss and color retention, and protection is highly satisfactory. The number of RFA-ships using waterborne coatings is now nearly 25.
Figure 3. Solvent emission from coating systems for marine and aggressive areas CORROSION CATEGORY 5: MARINE AND AGGRESSIVE AREAS In 1992, the three major Norwegian petrochemical and offshore operators, Statoil, Saga Petroleum and Norsk Hydro, introduced a prequalification test for coatings used on offshore structures. The prequalification test [7] included, among other things, coating systems for structural steel, exteriors of vessels and tanks, piping (not insulated), valves, and steel in noncorrosive areas; all decks; and submerged steel. The tests used for the prequalification and the acceptance criteria are listed in Table 3. High solids/low VOC coating types like reinforced polyester and solvent-free epoxies have been specified for offshore decks and have performed very satisfactorily over the years. The testing confirmed their good performance. Similarly, the good experience with solvent-free or low solvent epoxies in submerged areas has been verified. 456
Kronborg Nielsen
Table 3. Prequalification Testing by Statoil R-SP-630 Test Salt spray
Method ISO 7253
Duration 8000 h
Acceptance Criteria Max disbonding 5 mm (ISO 4628). Blistering: not visible (ISO 4628).
Remarks
Adhesion: 2,0 MPa (ISO 4624) and maximum 50% reduction from original value. Condensation chamber Weatherometer
Cathodic disbonding
ISO 6270
8000 h
ASTM 623-89
2000 h
ASTM G8
28 d
Max disbonding 5 mm
Overcoatable without mechanical pretreatment. Only for non submerged coatings. Only for submerged coatings.
Among systems tested for topsides are those mentioned in Table 4. In general all three systems performed equally overall in the prequalification tests. The IOZ/vinyl topside system in Table 4 is a system that has been used since the 1960s, especially by American-owned offshore operators, and is still performing well in, for example, the Gulf of Mexico and the North Sea. During the 1980s the IOZ/HB epoxy/PU system gradually took over. Important reasons were the lower cost of the paint, and the possibility of applying coatings in higher film thicknesses, thereby reducing the number of coats and, as a result, the application costs. The change was not caused by environmental pressure.
Table 4. Systems Tested for Topsides Standard IOZ/Vinyl System Zinc silicate primer Vinyl tie coat
IOZ/HB Epoxy/PU System
60 m Zinc silicate primer 25 m Epoxy tie-coat
IOZ/HB Epoxy/WB Acrylic System 60 m Zinc silicate primer 60 m 25 m Epoxy tie-coat 25 m 457
Corrosion Protection and Monitoring
3x Vinyl interm/finish 215 m HB epoxy, LTC* 165 m Polyurethane 50 m Total 300 m Total 300 m Solvent emission: 450 g/m2 Solvent emission: 257 g/m2 Paint price/m2, index: 100 Paint price/m2, index: 86 *LTC: Low temperature curing (-10°C - 20°C)
HB epoxy, LTC* 165 m WBorne acrylic 50 m Total 300 m Solvent emission: 217 g/m2 Paint price/m2, index: 83
In the 1990s, with greater focus on environmental issues, systems employing low VOC products are becoming more important. Also the isocyanate curing agent in the polyurethanes is being put under surveillance (officially and unofficially) in, for example, Great Britain and Norway. Therefore, the IOZ/HB epoxy/WB acrylic described in Table 4 is a step towards top side systems formulated with respect for environmental concerns both for the applicator and his surroundings. It is worth noticing that the IOZ/HB epoxy/WB acrylic top-coat system has both the lowest emission and the lowest cost. Since 1992 a major Norwegian offshore operator, Amoco Norway, has employed this zinc silicate/HB epoxy (amide type)/waterborne acrylic system both for offshore maintenance painting and for new construction in manufacturing units. Occasionally, zinc-rich epoxies are also used as prime coats. The operator decided to change to the low VOC system after a thorough evaluation of the anticorrosive properties, and the reduced solvent, low molecular amine and isocyanate exposure of the applicators. After the initial adjustment of procedures, the result has been very positive from environmental, operational, and economical points of view [8]. The low VOC, high solids epoxy mastic has for a long time been used for maintenance on ships' topsides, superstructures and exposed steel structures, especially on power-tool cleaned surfaces. They have replaced traditional chlorinated rubber and alkyd systems, and their success is again due to the possibility of applying high film systems in a few coats; a comparably lower cost per square meter when applied at the same dry film thickness with recognized better protective performance [9]. The epoxy mastic may be top coated with waterborne acrylic coatings to obtain gloss and better color retention while keeping the solvent emission down. An exceptional coating test object is situated on the Thames near the Tower Bridge: The HMS Belfast. This World War II warship was painted with waterborne acrylics in 1993, and the surface condition is excellent [10,11]. SUMMARY Legislation is gradually forcing painting contractors and shipyards to employ paint systems with low solvent emissions (i.e., low VOC). Among the low VOC paints on the market are high solids epoxies and waterborne acrylics. The introduction of these coatings in less aggressive environments is already in place. However, the use of waterborne coatings in particular has been more cautious in highly corrosive environments, e.g., ships' topsides, petrochemical installations, and offshore. The reluctance is mainly originating from limited experience with their long-term performance. However, official tests, practical demonstrations and case histories in aggressive areas have
458
Kronborg Nielsen
now shown that waterborne acrylic finishes combined with high solids/low VOC prime coats are as fully resistant to corrosion as any alternative. Similarly, high solids epoxies, polyurethanes and polyesters have also demonstrated their performance in corrosive environments. Therefore, on sound economic and environmental bases, coating contractors, specifiers and shipyards can meet the upcoming VOC legislation with environmentally acceptable and resistant coating systems. REFERENCES 1. S. Nysteen, Surface Coatings International, July 1994, p. 311. 2. European Union, Proposal for a council directive (EEC) on the limitations of the emissions of organic compounds due to the use of organic solvents in certain processes and industrial installations, April 1994. 3. ISO 12944. Secretariat of ISO/TC 35/SC 14, N76, N37 and N94. 4. J. Andrews et al., Cleveland Society for Coatings Technology Technical Committee, Journal of Coatings Technology, October 1994, p. 49. 5. J. Carew et al., Materials Performance, December 1994, p. 24. 6. I. Walker, Petroleum Review, November 1994, p. 520. 7. Statoil, Norway, Specification for purchase, Surface Preparation and Protective Coating Doc. no. R-SP-630, 1992. 8. T.M. Ege and H. Erikstein, Maling offshore: Bruk av vanntynnbare toppstr ksmalinger i Nordsj en. Hvilke krav og erfaringer har man? (Painting offshore: Demands and experience with waterborne top coats in the North Sea), Overflatedagar 95, Paper A-5, Oslo, Norway, Teknologisk Institutt, November 1995 (In Norwegian). 9. P.K. Nielsen, J.H. Hansen, Ecology and economy in the development and use of heavy-duty protective coatings for Steel, Corrosion Asia/94, Paper No. 1130, Singapore, National Association of Corrosion Engineers, September 1994. 10. Lloyds List, 29 September 1994, p. 14. 11. D. Woodyard, Lloyds List, 19 April 1991, p. 5.
459
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
ANTICORROSIVE FILM-FORMING NONPOLLUTING PRODUCTS ACHIEVED IN ROMANIA R. Serban, N. Moga and E. Stockel Anticorrosive Protection, Paints and Varnishes Research Institute (ICEPALV), 49A Theodor Pallady Av. 74585, Bucharest, Romania
ABSTRACT Present law stipulations concerning environmental protection require that modern film-forming products should produce an as little pollution as possible. To achieve this, water should be used as a thinner to produce waterborne anticorrosive paints for metal and emulsion paints for anticorrosive protection of concrete. As concerns waterborne products, ICEPALV is researching products applied by electrophoresis, and offering licenses for anaphoretic primers; and has developed air-drying waterborne primers and paints, based on epoxy and alkyd resins. As concerns emulsion paints, ICEPALV has researched and developed acrylic, acrylo-styrene, vinyl-acrylic, and vinylic as well as decorative plasters for concrete protection. ICEPALV is also researching anticorrosive emulsion paints, based on meth(acrylic) copolymer latices with vinyl esters of heavily monocarboxylic acid containing 10 carbon atoms (VeoVa 10). Key Words: Metal, concrete, anticorrosive protection, waterborne products, emulsion paints
INTRODUCTION Since 1980, the ICEPALV research in the field of film-forming materials has been directed by the need to protect the environment. The use of water as a thinner in the development of waterborne products and emulsion paints is one of the ways to achieve this purpose. Water Thinnable Products ICEPALV is concerned with developing this important group of products used especially in the building machines industries (i.e., bodies, cases, wheels and other accessories), but also in other industries, as for example, for the anticorrosive protection of metallic buildings. As concerns the application of these products, they are used simultaneously in both older application systems (i.e., immersion, flow-coating, and spraying) as well as in newer ones (i.e., anionic and cationic electrodeposition, and autophoresis) [1]. The main waterborne products developed by ICEPALV in Romania are presented in Table 1. Emulsion Paints Concrete corrosion is a complex physicochemical process. Corrosion may be spoken about as an electrochemical phenomenon only in the case of reinforced concrete; otherwise, it is only about the support isolation from the corrosive medium. Therefore, nowadays especially, the outdoor painting of buildings is a necessity, having both an aesthetic and an 461
Corrosion Protection and Monitoring
anticorrosive protection function. Emulsion paints are more and more often used for these purposes, and their advantages are well known. A whole series of putties, primers, decorative plaster and paints based on acrylic, acrylostyrene, acrylo-vinyl, vinyl, etc. latices for concrete protection have been developed and launched on the market by ICEPALV. Among these the most important are those mentioned in Table 2. Table 1. The Main Waterborne Products Developed by ICEPALV in Romania Product
Series
Grey primer Brown primer Grey primer Grey, red A-B primer Colourless, black, khaki primer Grey anticorrosive primer Dark-black primer
7100 7101 7003 7004 7004
alkyd-phenol alkyd-phenol alkyd-phenol alkyd-phenol alkyd-phenol
Application Method flow-coating spraying immersion electrophoresis electrophoresis
7003
epoxy-ester
immersion
7206
acrylic
flow-coating, immersion
7005
polybutadiene electrophoresis
7210
acrylic
Anticorrosive grey primer Black primer
Binder
Recommended Uses Building-machines industry Building-machines industry Building-machines industry Automotives industry Building-machines industry Automotives and electrodomestic industry The protection of gasoline tanks and of the other accessories in the automotive industry Automotives industry
immersion, Phosphatized plates spraying protection Air drying primer 7352 waterborne by brush, roller Building-machines industry alkyd or spraying Air drying primer 7702 waterborne by brush, roller Building-machines industry alkyd or spraying Air drying primer 7752 waterborne by brush, roller Building-machines industry alkyd or spraying Beige primer 7385 epoxy-ester electrophoresis Automotives industry Black primer 7325 epoxy-ester electrophoresis Automotives industry Air-drying paint 7351 epoxy by brush, roller Building-machines industry or spraying Air-drying paint 7751 epoxy by brush, roller Building-machines industry or spraying Colourless, 8771 vinyl-polyby brush or roller Protection of paint spray removable varnish acetate latex booth glass surfaces White, removable 8772 vinyl-polyby brush or roller Protection of paint spray paint acetate latex booth metal surfaces Table 2. The Main Emulsion Products Developed by ICEPALV in Romania 462
Serban et al.
Product Soaking primer
Series 8440
Putty for concrete
8640
LUCICRIL-half-glossy paint for outdoors DASIROM-half-matt paint for outdoors Half-matt paint for outdoors VEPAROM-matt paint for indoors VEPATIM-matt paint for indoors Matt paint for indoors
8513
Matt paint for indoors
8213
STROP- decorative plaster
8411
8415 8427 8430 8630 8426
Latex Acrylo-styrene hydrosol
Recommended Uses To fill in the pores and to increase adherence (concrete, masonry) Vinyl-maleic To putty the concrete before copolymer finishing Pure acrylic Outdoor finishing of buildings copolymer (wood, concrete, masonry) Acrylo-styrene Outdoor finishing of buildings copolymer (concrete, masonry) Acrylo-styrene-maleic Outdoor finishing of buildings copolymer (concrete, masonry) Acrylo-styrene Indoor finishing of buildings copolymer (concrete, masonry) Vinyl-maleic Indoor finishing of buildings copolymer (concrete, masonry) Acrylo-styrene-maleic Indoor finishing of buildings copolymer (concrete, masonry) Vinyl-polyacetate Indoor finishing of buildings homopolymer (concrete, masonry) Acrylo-styrene Decorative finishings for outdoor copolymer buildings (concrete, masonry)
NEW RESEARCH TRENDS Waterborne Epoxy-Ester Primer Series 7301 ICEPALV has lately ended the research concerning waterborne epoxy-ester primer series 7301. It is used for the anticorrosive protection of wheels and some automotive accessories, and it is applied as a first coat over zinc phosphate pretreated iron plate by Bonder 125 technology. Application is carried out by anionic type electrophoresis. EXPERIMENTAL PROCEDURE The epoxy-ester, which represents the binder of this primer, is produced from an epoxy resin of diglycidil ether of the A bisphenol type, with a molecular weight of 900-1000, which in the first stage is partially fatty acids reaction, with some of the hydroxyl groups left are esterified with COOH groups from the tricarboxyl adduct and hydrolyzed according to the following scheme:
463
Corrosion Protection and Monitoring
O
OH
unsaturated fatty OH acids
OH
OH
OH
245 °C catalyst NaOH
O
COOH
HOOC
COO
CH
OOC
CH
HOOC
OH 120-150°C
OH
OH
HOOC
O OC
HOOC
COO Epoxy resin M=900-1000
COOH
HOOC
COO
COOH
COOH
HOOC
Tricarboxyl adduct
COOH COO
The tricarboxyl adduct is previously achieved from fatty acids, maleic anhydride and water, according to the reaction: COOH COOH CH COOH HC
C
CH2
C
COOH
O H2O O
O HC
C
HC
C
O
O
COOH
O
HOOC Unsaturated fatty acids
Hydrolized succinic adduct
125°C
Succinic adduct 220°C COOH
O
HC
C
HC
C
O
CH
COOH
CH
COOH
Hydrolized Diels-Alder adduct
O
Diels-Alder adduct
Finally, the epoxy-ester is so reactioned to provide the achievement of an epoxy-ester with free functional groups: hydroxyl and carboxyl, which exhibit a good water solubility, increase the system’s stability, as well as the salt spray resistance by increasing film adherence in an alkaline medium (due to the hydroxyl polar groups and being inert to alkalies). The primer also contains cosolvents (for spreading and adherence) and various additives: wetting and dispersing agents, antifoaming agents, antioxidants, antibacterial, etc., which provide high quality films. The pigments and extenders were so selected to resist the alkaline medium, and to provide a high corrosion resistance, a good hiding power for the support and a migration speed in an electric field similar to that of the film-forming.
464
Serban et al.
RESULTS AND DISCUSSION The electrophoretic primer developed is a slightly thixotropic, grey fluid, with a medium viscosity (under 100 P), with a nonvolatile matter content of about 40%. It is neutralized with an alkaline base. It has an alkaline pH, being water soluble and sensitive to low temperatures (under 10°C). It is not flammable and presents low toxicological hazards compared to the classical products. Characteristics of the Electrodeposition Bath By diluting the primer with demineralized water, an electrodeposition bath is achieved having a nonvolatile matters content of about 12.5%. The following parameters should be kept constant: pH, conductivity, content of cosolvents and free fatty acids, degree of neutralization and free acidity, and pigment/binder ratio. The formulation is so balanced to provide physical stability for the system expressed by an adequate settling curve, as can be seen in Fig. 1. 0 -10 Settling degee (%)
-20 -30 -40 -50 -60 -70 -80 -90 -100 0
2
4
6
8 10 12 14 16 18 20 22 24 Time (hours)
Figure 1. Evolution of settling degree over time Application Conditions
• • • • •
Electric voltage (V) Medium density of anode current (A/m2) Application time (sec) Bath temperature (°C) Film curing(drying) is carried out in the oven 180°C
140 - 220 max. 20 60 -360 25-28 for 30 minutes at a temperature of
From Fig. 2, the variation of film thickness according to application time, at the application voltage (180 V), as well as at breakdown voltage (50 V) can be seen. Film Characterization 465
Corrosion Protection and Monitoring
Between 20 and 30 µm films are achieved with a uniform appearance, free of surface faults (i.e., pinholing, cratering). The mechanical characteristics are very good:
• • • •
Cross-cut adherence (mm) Erichsen elasticity (mm) Impact resistance (1 kg/cm) Flexibility (mm)
1 4 min. 30 min. 1
The films corresponded from the point of view of corrosion resistance, so:
• Salt spray resistance (hours)
192 6 4 192 absent 21 absent
Blistering (note), (min.) Rust spreading (mm, max.) • Water resistance by immersion (hours) Blistering • Humidity resistance (days) Surface alteration
The throwing power was determined on 24-cm samples, and their values are presented in Fig. 3. In conclusion, from the short presentation of the epoxy-ester electrophoretic primer series 7301, it may be noticed that this product, used for the anticorrosive protection of some parts and units in the automotive industry, corresponds to the present requirements of the Romanian industry. 35
Thickness (um)
30 25 20 15 50 V (max. 25 minutes)
10
180 V (max. 5 minutes) 5 0 0
5
10
15
20
25
30
Time (minutes)
Figure 2. Variation of film thickness with the application time
466
Serban et al.
Throwing power (um)
30
20
10
0 0
3
6
9
12
15
18
21
24
Distance (cm)
10 20 30 40 50 60 70 80 90 100 Figure 3. Evolution of throwing power with distance Anticorrosive Primer Paints Based on VEOVA 10 (Meth) Acrylate Latices ICEPALV has been researching anticorrosive primer paints based on VeoVa 10 (meth)acrylate latices. Various studies [2,3] demonstrated that the incorporation of VeoVa monomers in (meth)acrylic copolymers improves the chemical and, especially, the water resistance of the latex films. The bulky aliphatic entity gives the copolymer a high hydrofobicity, an excellent UV resistance and also a good alkali resistance by protecting it from saponification [4]. EXPERIMENTAL PROCEDURE The latices were obtained by the copolymerisation of VeoVa 10 and 2 ethylhexyl acrylate (which also contributes to good water repellency) as soft monomers with methylmethacrylate as a hard monomer. Performing the polymerization essentially in the absence of colloids and in the presence of a minimum quantity of surfactant with phosphate groups (e.g., organic ester phosphate: REWOPHAT E 1027 - REWO, Germany), whilst at the same time carefully adding the paint formulation additives (i.e., coalescing agents, thickener, and dispersant) and pigments and mineral fillers, films possessing intrinsically good barrier properties can be achieved. The best results were obtained with monomer compositions falling in the shaded area of Fig. 4. In the primer paint formulations zinc, phosphate was introduced. It proved to be an anticorrosive and nontoxic pigment. It appears to pack in the film in a manner which presents a high resistance to the passage of water molecules and salts and an anticorrosive efficiency similar to zinc chromate, in long-term exposure tests. The primer paints were formulated at two different PVC: 20% and 30%, at basic pH. The evolution of the viscosities was essentially unchanged after 8 months of storage, thus demonstrating the good stability of these primers. They are being kept under observation. 467
Corrosion Protection and Monitoring
RESULTS AND DISCUSSION Some Mechanical Properties Adhesion was evaluated both on concrete (by a pull-off test) and on metal (by a cross-cut test). The improvement of adhesion property is directly related to the amount of VeoVa 10 from the copolymer composition indifferent to the PVC (Fig. 5). Due to the good metal adherence, Erichsen elasticity could be evaluated. It showed the same evolution according to the quantity of VeoVa 10 from the latex as the former case. The paints with lower PVC, have higher elasticities (Fig. 6). The thickness of the analyzed primer paint films was about 100µm. Water Resistance of Latices Concerning the water-spot resistance test (100 µm latex films, 24 hours), Fig. 7 demonstrates clearly, the positive contribution of the increase of proportion of VeoVa 10 from the copolymer composition at Tg = constant = 15°C. It can be noticed from Fig. 7, that the VeoVa latices have superior values when compared to acrylo-styrene latices, which have about 2. Water Vapor Permeability of Primer Paints The studies of Geelhaar and Melan [5] have shown that an emulsion film absorbs 10 to 100 times greater amounts of moisture than normal solvent cast films from alkyds, and reactive and crosslinked polymers. Continuous films produced by solvent-borne coatings, absorb water in the polymer matrix by a very slow diffusion process. The water absorption of emulsion films is via microcapillaires between coalesced particles. From Fig. 8, it can be noticed that primer paints based on VeoVa 10 (meth)acrylate latices have lower water vapor permeabilities than acrylo-styrene emulsion paints, at the same PVC level (e.g., 2.9 g/100cm2/100µm/day at PVC = 20% and 3.5g/100cm2/100µm/day at PVC = 30%). It can be also noticed that an increase of the VeoVa quantity from the latex up to 55-60% leads to a great decrease in permeability. As was expected, if the PVC increases, water vapor permeabilities also increases. Corrosion Resistance The primer paints studied were applied on panels of concrete with a thickness of about 100µm, and after 7 days of drying they, were salt-spray tested in order to simulate a marine atmosphere, one of the harshest climates. After 400 hours of exposure, the films with 50-60% VeoVa 10 in latex were unchanged. The others exhibited about 10% blistering. The test is continuing. The same type of panels were also exposed outdoors in an industrial climate, and after 8 months of exposure, the films were unchanged. In conclusion, the primer paints based on VeoVa 10 (meth)acrylate latices, containing at least 50-60% VeoVa 10 monomer and stabilized by a phosphate based surfactant and without colloids, provide good qualities as an anticorrosive protection for concrete even in a marine environment. In addition, they exhibit resistance to flash rusting and early rusting as well as good metal adherence, qualities which convinced ICEPALV to keep on testing these primer paints for the anticorrosive protection of metals, too.
468
Serban et al.
Adhesion 10
8
6 PVC= 20 % 4 PVC= 30 % 2
0 20
30
40
50
60
70
80
VeoVa 10 concentration (%)
Figure 4. Optium copolymer compositions
Erichsen elasticity (mm)
Tg = 15 oC; 0 - bad, 10 - excellent Figure 5. Influence of VeoVa 10 concentration on primer paint adhesion
10 Water spot rezistence
8 7
8
6 6
5 PVC= 20 % 4
4
PVC= 30 % 2
3 0
2 20
30
40
50
60
70
VeoVa 10 concentration (%)
Tg = 15°C Figure 6. Influence of VeoVa 10 concentration on primer paint elasticity on metal
80
20
30
40
50
60
70
80
VeoVa 10 concentration (%)
Tg = 15 oC; 0 - film completely white, 10 - film unaffected Figure 7. Water spot resistance of latices
469
Corrosion Protection and Monitoring
W a te r v a p o u r p e rm e a b ility (g / 0 .1 m m / 100cm / day)
2
P VC = 2 0 % P VC = 3 0 %
1 .6
1 .2
0 .8
0 .4
0 10
20
30
40
50
60
70
Ve o Va 1 0 c o n c e n tra tio n (% )
Tg = 15 oC Figure 8. The effect of VeoVa 10 concentration on the water vapour permeability of primer paints CONCLUSIONS ICPALV is concerned with developing some new electrophoretic products of the latest generation (i.e., cataphoresis) which together with other measures (e.g., the use of galvanized sheet, waxes and protection products, and good and severe services) will increase the storage life of the new types of automotive and metallic surfaces, in general. As concerns emulsion paints, ICEPALV is concerned with developing new products with higher durability for concrete protection and new anticorrosive primer and emulsion paints for metal protection. REFERENCES 1. 2. 3. 4. 5.
470
C. Robu, XV FATIPEC Congress, Amsterdam, June 1980. M. Slincks and M.F. Daniel, PPCJ April, 1995, pp. 28-29 M. Slincks and M.S. Sonderman, XXI FATIPEC Congress, Amsterdam, 14-18 June 1992. C. Bondy and M.M. Coleman, JOCCA, 1970, 53, p. 555. H. Geelhaar and M. Melan, 13e AFTPV Congressbook 147, 1979.
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CATHODIC PROTECTION UNDER DISBONDED COATINGS OF 56 INCH GAS PIPELINE ALONG THE KANGAN-SHIRAZ M. Pakshir Department of Materials Science and Engineering Shiraz University, Shiraz, Iran
ABSTRACT The present study investigates the disbonding phenomenon according to the British Gas Method PS/CW6 and the ASTM-G8 standard. This investigation showed that the best protection for disbonded buried pipelines under cathodic protection would be achieved if the applied potential was kept -950 to -1000 mV with respect to the Cu/CuSO4 half-cell rather than the usual value of -850 mV (Cu/CuSO4) which is currently used. Key Words: Pipeline, coating, soil, disbonding, cathodic protection
INTRODUCTION Underground corrosion of buried steel is a major problem in the oil and gas industries [1]. A good practice in modern pipeline corrosion control work comprises the use of good coatings, in combination with cathodic protection as the main lines of defence [2], and better current distribution is accomplished by using an insulating coating [3]. Therefore, attempts to control pipeline corrosion rely on the use of coating materials with the reasoning that if the pipeline metal is isolated from contact with the scratches from soil, no corrosion could occur. However, the insulating coatings must be free of any defects such as scratches or pinholes. Pitting associated with such defects in the coating, i.e., Holidays, and subsequent disbonding have been observed on pipelines which are nominally cathodically protected [4,5]. Fessler [6] suggested that stress corrosion cracking on buried, coated pipeline tends to occur under disbonded coatings near small pinholes or holidays. It was deduced that a disbonded coating acted as a shield to the cathodic protection current and caused potential gradients under the coatings [7]. Aqueous displacement as a possible mechanism of the cathodic disbondment of protective organic coatings was suggested by Evans [8] in terms of the ability of an alkaline solution to creep over the metal surface and displace the organic coating. Bolger and Micheals [9] argued that displacement of coatings from a metal’s surface is promoted by pH values far removed from the isoelectric point of the surface oxide so that the oxide has a greater affinity for the water than for the organic materials.
EXPERIMENTAL PROCEDURE 471
Corrosion Protection and Monitoring
Soil Analysis Since the types of soil differ along the buried pipeline, four types of soil were taken from different sites as follows:
• • • •
Site 1 - 76 km Kangan - Shiraz Site 2 - 298 km Site 3 - 151 km Site 4 - 36/900 The soil samples obtained from each site were taken from areas adjacent to the pipeline and from the same depth as the pipeline was buried, to evaluate the chemical characteristics of the soil. A soil-distilled water slurry was obtained by mixing 50 g of soil with l00 ml of distilled water. The pH of the soil was determined using a special pH electrode cup for soil. The slurry was placed in the electrode cup and the pH value was read directly from the meter. Conductivity tests were conducted on the slurry using the Wenner four-electrode method, a procedure similar to ASTM G57-78. The concentration of soluble materials was evaluated from the slurry filtrate by saturation and flame photometric techniques. Table 1 shows the soil analysis for different sites along the 56 in. pipeline. Soil Type 4 with a pH of 8.44 and an Ec of 1 1.52 was chosen as a saline alkaline-type soil which is representative of the southern part of Iran. The results were compared with soil taken from the Northern part of the country, which is acidic. It was decided to make an acidic soildistilled water slurry using the same procedure (Table 2). Table 1. Characteristics of Saline Alkaline-Type Soil NO3
CO3
HCO3
SO4
Cl
Na
K
Mg
Ec
Ca
pH
Soil Type
(mmhos/cm) 33
0
146.5
71
1036
15.5
18.5
68
516
2.15
7.56
1
26
0
148.2
106.5
1728
75.5
20.5
49.5
560
2.70
7.43
2
37
0
341.6
461.5
960
980
17
113.7
320
4.87
7.79
3
6.4
0
1753
770.5
1536
1006.5
105
121.5
528
11.52
8.44
4
Ec (mmhos/cm)
pH
Soil
Table 2. Characteristics of Acidic-Type Soil NO3
CO3
HCO3
Cl
SO4
Na
K
Mg
Ca
Type 188
0
0
Tested Material
472
2912.5
4467
1398.6
868
754
540
28.42
4
acidic
Pakshir
Test specimens were cut and flattened, and then they were stamped with an identifying number from the existing new pipeline used in the Shiraz gas industry according to ASTMA53 grade B having the following composition: 0.30% C, 1.20 % Mn, 0.05 % P, 0.06 % S Each specimen was hand brushed with a wire brush, and soaked in a solvent to remove the cutting oil. Test Procedure In order to investigate the potential gradient under the disbonded coatings experimentally, special suitable cells were constructed ( Fig. 1). Steel samples 200 x 160 mm were coated with a cold tape used in Shiraz gas industries called NITO. The tape consisted of a polyethelene backing and a thermoplastic adhesive. For better adhesion, NITO primer was used. The technique for measuring the potential under the disbonded coatings was based on British gas method PS/CW6 in which backside electrodes were prepared by inserting a 500 mm length of a 5 mm outside diameter PVC tubing filled with agar and KCI. They were then filled with a sintered glass plug through four predrilled holes in the back of the steel plate. The holes were positioned in a line so as to give eight positions 5 mm apart at distances from 10 to 40 mm from the center of the holiday.
Figure 1. Experimental apparatus On the coated side of the specimen, a PVC pot (150 mm in diameter and 160 mm high) was attached by silicon rubber to the coating. This pot contained 1.5 l of prepared simulated 473
Corrosion Protection and Monitoring
solution. A platinum wire was used as the counter electrode. The steel plate was connected to a voltage supply by a wire tapped into it and insulated. The holiday potentials were controlled by a voltage regulator. The voltage was measured using a reference electrode and a digital multimeter. The holiday potentials were set at values of -780, -920 and - 1200 mV (SCE). In order to extract solutions from beneath the disbonded coating, a piece of polyethylene tubing was inserted through the steel in a fation similar to that used for the electrodes. A thinner length of tubing was then be inserted through the first hole to allow solution to be extracted with a syringe. The pH of the extracted solution was measured using a pH paper, and the chloride solution was measured using a Ag/AgCl microprobe. However, at distances further than 10 cm from the holiday, little success was achieved in extracting any crevice solution due to the thinness of the layer of electrolyte. Along the crevice, most of this liquid was absorbed into the adhesive. In this region, the pH values of the solution were measured directly from moisture that was found on the steel underneath the backing after the coating had been removed. Polarization Study In order to obtain the potential-pH diagram, a polarization study was carried out. Specimens 1 x 1 cm in dimensions were cut from the original pipe and mounted in a specimen holder so that 1 cm2 of the steel was exposed, and the polished specimen was placed in a corrosion cell with a platinum counter electrode and a lugging probe. The cell was filled with an already made. simulated solution of pH 8.44. To simulate the alkaline environments found beneath the disbonded coating, the pH of the solution was increased by adding various amounts of NaOH. RESULTS AND DISCUSSION Using a backside electrode as a special technique, the potentials under the disbonded coating were measured as a function of distance away from the holiday (Figs. 2 and 3). Also, using a catheter arrangement, the pH and the concentration of the chloride solution under the disbonded coating could be determined (Tables 3 and 4). In this investigation, three holiday potentials were chosen: the potential of -780 mV (SCE) was chosen since it represents the minimum cathodic current density in order to polarize the pipe to 850 mV Cu/CuSO4, and the holiday potential of -1500 mV (SCE) since it represents the overprotection potential. A test temperature of 40°C was used to simulate the conditions of the hottest part of the soil. As can be seen from Tables 3 and 4, the concentration of chloride ions in the crevice did not vary significantly from the bulk solution, and also, the concentration of chloride ions in the crevice was not a function of the holiday potential. Since the solution pH beneath the disbonded coating was thought to ncrease, polarization tests were carried out in the alkaline range of pH. Two distinct points can be seen in Figs. 4 and 5: the interaction of polarization current with the potential axis. i.e., where the current density was zero and was representative of the corrosion potential (Ecorr); and the start of the decrease in current density while the potential increased and was representative of the 474
Pakshir
protection potential (Ep). Therefore, in the potential-pH diagram, the immunity-general corrosion boundary represented by the corrosion potential points and the general corrosioncomplete passivation boundary could be represented by protection potential points (Figs. 6 and 7).
Potential vs. distance variation for alkaline-saline soil
Potential vs. distance variation for acidic type soil
Table 3. Experimental Results for Alkaline-Saline Soil and the Holiday Potentials of -0.780, -0.920 and -1.5 V Cl (ppm)
Crevice Tip/pH
Time (hour)
805.2 798.5 752.1 794.3
10.48 11.02 10.63 10.83
40 -0.604 -0.603 -0.604 -0.707
30 -0.606 -0.605 -0.709 -0.714
20 -0.603 -0.710 -0.716 -0.719
10 -0.715 -0.724 -0.731 -0.736
0 -0.780 -0.780 -0.780 -0.780
513 862 1103 1223
791 812.3 783.4 800.8 751.5 818.1 759.5 791.2
11.08 10.54 10.96 10.61 10.41 10.52 11.02 10.75
-0.611 -0.611 -0.610 -0.673 -0.612 -0.611 -0.610 -0.743
-0.608 -0.607 -0.674 -0.680 -0.605 -0.607 -0.722 -0.747
-0.616 -0.675 -0.685 -0.691 -0.609 -0.717 -0.785 -0.828
-0.704 -0.728 -0.751 -0.778 -0.725 -0.823 -0.907 -0.991
-0.920 -0.920 -0.920 -0.920 -1.500 -1.500 -1.500 -1.500
363 605 770 863 143 325 297 335
Distance from Holiday (mm)
Holiday Potential (Volt)
-0.780
-0.920
-1.580
Table 4. Experimental Results for an Acidic-Type Soil and the Holiday Potential of -0.780, -0.920 and -1.5 V (cm/CuSO4) 475
Corrosion Protection and Monitoring
Cl (ppm)
Crevice Tip/pH
2850.5 2920.2 2860.4 2898.7 29.20.2 2870.8 2932.7 2946.5 2873.6 2950.2 2873.8 2932.6
10.33 10.28 10.47 10.86 10.52 10.63 10.21 10.44 10.73 10.96 10.85 10.91
40 -0.605 -0.605 -0.604 -0.655 -0.600 -0.601 -0.600 -0.628 -0.611 -0.612 -0.611 -0.686
30 -0.609 -0.607 -0.665 -0.673 -0.603 -0.604 -0.632 -0.641 -0.608 -0.608 -0.687 -0.701
Polarization curve for alkalinesaline soil
476
Time (hour)
Distance from Holiday (mm) 20 -0.608 -0.676 -0.681 -0.692 -0.610 -0.630 -0.647 -0.652 -0.614 -0.695 -0.724 -0.750
10 -0.690 -0.704 -0.711 -0.715 -0.680 -0.708 -0.729 -0.744 -0.701 -0.807 -0.891 -0.945
0 -0.780 -0.780 -0.780 -0.780 -0.940 -0.920 -0.920 -0.920 -1.500 -1.500 -1.500 -1.500.
623 1038 1320 1482 421 719 911 1016 215 359 455 505
Holiday Potential (Volt)
-0.780
-0.920
-1.500
Figure 5. Polarization curve for acidic soil
Pakshir
. Pourbiax diagram extracted from polarization curve for alkalinesaline
urbiax diagram extracted from larization curve for acidic soil
As can be seen from Figs. 4 and 5, a significant shift of the crevice potentials to more positive values occurred as the distance from the holiday along the crevice increased. Also, the largest potential drop occurred near the holiday. The potential gradient appeared to decrease at distances further along the holiday. The major differences between the three potentials applied is the disbondment time, i.e., as the holiday potential became negative, the disbondment time decreased. For example, for an alkaline-saline soil, it took two months to disbond the coating at a holiday potential of -780 mV (SCE), while at a potential of -1500 mV (SCE), it only took 14 days. The crevice tip potential for the three applied holiday potentials varied between -673 and -734 mV (SCE), and the pH of the tip of the crevice varied between 10.41 and 11.08. Therefore, neither the crevice potential nor the crevice tip pH were a function of the applied holiday potentials. Polarization tests enabled an experimental potential-pH diagram to be constructed for the steel exposed to conditions which simulated those formed beneath the disbonded coatings (Figs. 6 and 7). Superimposed on these diagrams are the crevice tip environments which were determined from the corresponding cathodic disbondment test. Hence, if the -crevicetip potential lay between -673 and -734 mV (SCE), and the crevice-tip pH lay between 10.41 and 11.08, then according to Fig. 6, only a small part of the blackened area lay within the general corrosion area, whereas a large part of it was in the complete passivation area. Hence, maintaining the steel exposed along the crevice in a passive state depended only on a continued high PH. Consequently, the condition of the crevice tip determined the corrosion behavior of the metal beneath the disbonded coating. In another words, by knowing the condition of the crevice tip on the potential-pH diagram, one can estimate the corrosion and noncorrosion condition beneath the disbonded coating.
477
Corrosion Protection and Monitoring
Figures 8 and 9 show that the condition required for the occurrence of corrosion beneath a disbonded coating is for the crevice potentials to be within the general corrosion range of about -695 to -845 mV (SCE). Therefore, as cathodic disbondment occurs, the crevice potentials place the steel at some distance along the crevice into a region of general corrosion. This can be seen by the surface morphology of the specimens after the disbonding test ( Figs. 10-15). As can be seen from Fig. 10, all the disbonded surfaces were in the region of general corrosion, but when the holiday potential was kept at -920 mV (SCE), the first 5 mm of the disbonded surface was in the immunity region, from 5 to 20 mm was in the general corrosion region and at distances > 20 mm from the holiday, the surface was in the passive state. These situations can also be predicted by Figs. 8 and 9.
. Potential distance curve for alkaline-saline soil
478
. Potential distance curve for acidic soil
Pakshir
Figures 10-12. Surface morphology after disbondment for alkaline-saline type soil at various holiday potentials
479
Corrosion Protection and Monitoring
Figures 13-15. Surface morphology after disbondment for acidic type soil at various holiday potentials From Fig. 11, it can be seen that in the first 20 mm from the holiday, the disbonded surface was covered by an oxide layer which represents the general corrosion and suggests that the metal underneath the disbonded coating was in the passive state. When the holiday potential was kept at 1500 mV (SCE), the potential-pH diagram predicted that up to 18 mm of the disbonded surface would be immune to corrosion, and then would show general corrosion. In Fig. 9, the curve predicts the regions of corrosion, passivation and immunity under the disbonded coating, i.e., for an acidic soil it predicts that at a holiday potential of 780 mV (SCE), the surface would be in the general corrosion region, at a holiday potential of 480
Pakshir
-920 mV (SCE), the first 4 mm of the disbonded surface would be in the immunity region, and between 4 and 20 mm, the metal would be in the general corrosion region and then in the passivated region. This prediction can be shown by the surface morphology examination shown in Figs. 13, 14, and 15.
6. Distance from holiday vs. time curve for alkaline-saline soil
7. Distance from holiday vs. time curve for acidic soil
CONCLUSIONS For an alkaline-saline soil, the occurrence of corrosion beneath a disbonded coating requires that the crevice potentials be within the general corrosion range of -695 to -845 mV (SCE). The crevice potentials for a holiday potential of -920 mV lie solely within the perfect passivation region beyond a distance of about 5 mm from the holiday. However, during the growth of the disbondment area and the consequent movement of the crevice potentials to their final values, the crevice potentials must be in the region of general corrosion. For the crevice potential obtained for a holiday potential of -1500 mV (SCE), the potentials must lie within the general corrosion region from about 18 to 28 mm from the holiday. Important conclusions which can be drawn from this investigation is that the applied potential of -780 mV (SCE) is not satisfactory, and a holiday potential of -1500 mV (SCE) is not representative of an instantaneous off potential which suggests that overprotection is undesirable. However, when the applied potential is -920 mV (SCE), only small parts around the holiday are in the general corrosion region while the rest of the surface is in the protected state. Therefore, this potential is recommended for buried pipeline. Also from Figs. 16 and 17, it can be concluded that for both types of soil, the disbonded mechanisms follow similar patterns, although the disbonding rate is much slower for an acidic soil.
REFERENCES
481
Corrosion Protection and Monitoring
2. 3. 4. 5. 6. 7. 8. 9.
1. H. Azad, M.Sc. thesis, School of Engineering, Shiraz University, 1994. A.W. Peabody, NACE, Control of Pipe Line Corrosion, 1976. H.H. Uhlig, Corrosion and Corrosion Control, John Wiley and Sons Inc, 1971. C.G. Manger and R.C. Robinson, Materials Performance 20, 7, 1981, p. 46. B.W. Cherry and A.N. Gould, Pitting Corrosion of Nominally Protected Land Base Pipelines, Materials Performance, Aug. 1990. R.R. Fessler, Oil and Gas Industry 74, 7, 1976, p. 81. R.P. Fessler, Sixth Symposium on Pipe Line Research, American Gas Association, Arlington, Virginia, P-R-1, 1960. V.R. Evans, Corrosion and Oxidation of Metals, St. Martin's Press, New York, 1960. J.C. Bolger and A.S. Micheals, Interface Conversion for Polymer Coatings, Weiss and Cheever, eds., Elsivier, New York, 1968.
482
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
SYNERGISTIC EFFECT EXISTING BETWEEN AND AMONG A PHOSPHONATE, Zn2+, AND MOLYBDATE ON THE INHIBITION OF CORROSION OF MILD STEEL IN A NEUTRAL AQUEOUS ENVIRONMENT S. Rajendran1, B.V. Apparao2 and N. Palaniswamy3 1
2
Department of Chemistry, G.T.N. Arts College, Dindigul - 624 001, Tamil Nadu , India
Department of Chemistry, Regional Engineering College, Warangal - 506 004, Andhra Pradesh, India 3
Corrosion Science and Engineering Division, Central Electrochemical Research Institute, Karaikudi - 630 006, Tamil Nadu, India
ABSTRACT The synergistic effect existing between and among the sodium salt of ethyl phosphonic acid (EPA), Zn2+, and molybdate on the inhibition of corrosion of mild steel in a neutral aqueous environment containing 60 ppm Cl- was evaluated by the classical weight-loss method. The formulation consisting of 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ had 99% inhibition efficiency. The mechanistic aspects of corrosion inhibition are discussed, in a holistic way, based on the results obtained from a potentiostatic polarization study, the x-ray diffraction (XRD) technique, and UV-visible diffused reflectance, FTIR and luminescence spectra. Key Words: Mild steel, neutral environment, corrosion inhibition, synergistic effect, ethyl phosphonate-zinc-molybdate
INTRODUCTION Molybdates are among the most broadly applied inhibitors, chiefly because of their efficacy towards both ferrous and nonferrous metals and their very low order of toxicity [1]. Molybdate can be used as corrosion inhibitor alone or in combination with other synergists like nitrite [2], metallic cations like Ni2+, Mn2+, Zn2+ [3], azoles like benzotriazole and tolyltriazole [4], chromate [5], amine phosphonates [6], inorganic phosphates [7], citrate and calcium [8]. Even though several papers [1-14] have discussed the use of molybdate as corrosion inhibitor, the mechanistic aspects of corrosion inhibition have not been studied in detail. The present work evaluates the synergistic effect existing between and among molybdate, Zn2+ and ethyl phosphonate by the weight-loss method. The mechanistic aspects of corrosion inhibition were studied, in a holistic way, based on the results obtained from a potentiostatic polarization study, the x-ray diffraction (XRD) technique, UV-visible diffused reflectance, FTIR and luminescence spectra. EXPERIMENTAL PROCEDURE 483
Corrosion Protection and Monitoring
Preparations of the Specimens Mild steel specimens (0.02 to 0.03% S, 0.03 to 0.08% P, 0.4 to 0.5%Mn, 0.1 to 0.2% C and the rest iron) of the dimensions 1 x 4 x 0.2 cm were polished to a mirror finish and degreased with trichloroethylene for use in the weight-loss method and surface examination studies. For the potentiostatic polarization studies, a mild steel rod encapsulated in Teflon with an exposed cross section 0.5 cm in diameter was used as the working electrode. Weight-Loss Method Mild steel specimens, in triplicate, were immersed in 100 ml of the solutions containing various concentrations of the inhibitors for a period of seven days. The weights of the specimens before and after immersion were determined using a Mettler balance, AE-240. Potentiostatic Polarization Study This study was carried out in a three-electrode cell assembly connected to a bioanalytical system (BAS-100 A) electrochemical analyzer, provided with an IR compensation facility, using mild steel as the working electrode, platinum as the counter electrode and a saturated calomel electrode as the reference electrode. Surface Examination Study The mild steel specimens were immersed in various test solutions. After two days, the specimens were taken out and dried. The nature of the film formed on the surface of the metal specimens was analyzed by various surface analysis techniques. FTIR Spectroscopic Study The FTIR spectra were recorded using a Perkin-Elmer 1600 FTIR spectrophotometer. UV-Visible Diffused Reflectance Spectroscopy The UV-visible diffused reflectance spectra were recorded using a Hitachi U-3400 spectrophotometer. X-Ray Diffraction Technique The XRD patterns were recorded using a computer-controlled x-ray powder diffractometer, JEOL JDX 8030, with CuKα (Ni-filtered) radiation (λ = 1.5418 A). Luminescence Spectroscopy The luminescence spectra were recorded by Hitachi 650-10 S fluorescence spectrophotometer equipped with a 150 W xenon lamp and a Hamamatsu R 928 F photomultiplier tube. RESULTS AND DISCUSSION Analysis of the Results of the Weight-Loss Method The corrosion rates of mild steel in a neutral aqueous environment containing 60 ppm Clin the absence and presence of inhibitors at various concentrations, obtained by the weight loss method are given in Table 1. The corrosion inhibition efficiencies of various systems are also given in Table 1.
484
Rajendran et al.
Table 1. Corrosion Rates of Mild Steel in a Neutral Aqueous Environment (Cl- = 60 ppm) in the Absence and Presence of Inhibitors, and the Inhibition Efficiencies Obtained by the Weight-Loss Method
2+
SI. EPA Zn MoO4 No. (ppm) (ppm) (ppm) 1 2 300 3 50 4 300 50 5 300 50 50 6 300 50 100 7 300 50 200 8 300 50 300 9 300 300 10 50 300 11 300 Inhibitor system: EPA + Zn2+ + MoO42-
2-
Corrosion rate (mdd) 15.54 15.39 19.11 6.22 13.98 11.66 9.32 0.16 0.16 1.58 3.11
Inhibition Efficiency (%) 1 -23 60 10 25 40 99 99 90 80
It is evident from Table 1 that ethyl phosphonic acid (EPA) by itself is not a good inhibitor and Zn2+ is corrosive. Interestingly the formulation consisting of 300 ppm EPA and 50 ppm Zn2+ had a 60% inhibition efficiency. This indicates the synergistic effect between EPA and Zn2+. When various concentrations of molybdate were added to the above system, the inhibition efficiency increased at 300 ppm MoO42-. The formulation consisting of 300 ppm EPA, 50 ppm Zn2+ and 300 ppm MoO42- had a 99% efficiency. It was found that the formulations consisting of 300 ppm EPA and 300 ppm MoO42-, and also MoO42- (300ppm)Zn2+ (50 ppm) showed a synergistic effect. Analysis of the Potentiostatic Polarization Curves The potentiostatic polarization curves of mild steel immersed in various environments are given in Fig. 1. It is observed that, when molybdate was added to chloride or EPA or Zn2+ or EPA-Zn2+, the corrosion potential shifted to the anodic side. The formulation consisting of 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ shifted the corrosion potential to -390 mV vs. SCE. This indicates that this formulation acts as a mixed inhibitor. This is further supported by the fact that the anodic and cathodic Tafel slopes shifted almost equally (28 mV/decade). Analysis of the FTIR Spectra The FTIR spectrum of pure EPA (KBr) is given in Fig. 2a. The FTIR spectrum (by the multiple internal reflection (MIR) technique) of the film formed on the surface of the metal specimen immersed in an environment consisting of 60 ppm Cl-, 300 ppm MoO42- and 300 ppm EPA is given in Fig. 2b. It is found that the P-O stretching frequency [15-17] of the phosphonic acid decreased from 1071.7 cm-1 to 1018.6 cm-1. This suggests that the oxygen atom of the phosphonic acid coordinated to Fe2+ on the metal surface [18,19].
485
Corrosion Protection and Monitoring
The FTIR spectrum (MIR) of the film due to the environment consisting of 60 ppm Cl-, 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ reveals that the P-O stretching frequency decreased from 1071.7 cm-1 to 1018.4 cm-1. This suggests that in this case also, the oxygen atom of the phosphonic acid coordinated to Fe2+ on the metal surface. Furthermore, the peak at 1456 cm-1 was due to ZnO2 [20]. This may be explained by the fact that Zn(OH)2 formed on the cathodic sites [19] converts into ZnO2. Analysis of the UV-Visible Reflectance Spectra The UV-visible reflectance spectra of the films formed on the surface of metal specimens immersed in various test solutions are given in Fig. 3. The spectrum of the film due to the environment containing 60 ppm Cl- and 300 ppm MoO42- shows a peak at 320 nm (Fig. 3a). This may be due to a complex formed between the iron and molybdate. The spectrum of the film formed on the surface of the metal immersed in the environment, consisting of 60 ppm Cl-, 300 ppm MoO42- and 50 ppm Zn2+ is given in Fig. 3b. The wavelength transition at 550 nm indicates the presence of oxides of iron (band gap = 1.239/0.55 = 2.25 eV) on the metal surface [21] having semiconducting property [22]. The peak at 320 nm may be due to an iron-molybdate complex.
Figure 1. Potentiostatic polarization curves (a) Cl- 60 ppm
(e) Cl- 60 ppm + EPA 300 ppm
(b) Cl- 60 ppm + Zn2+ 50 ppm
(f) Cl- 60 ppm + EPA 300 ppm + Zn2+ 50 ppm
(c) Cl- 60 ppm + MoO42- 300 ppm
(g) Cl- 60 ppm + EPA 300 ppm + MoO42300 ppm
(d) Cl- 60 ppm + MoO42- 300 ppm + Zn2+ 50 ppm
(h) Cl- 60 ppm + EPA 300 ppm + MoO42300 ppm + Zn2+ 50 ppm
486
Rajendran et al.
Figure 2. FTIR spectra
Figure 3. UV-visible reflectance spectra
(a) Pure EPA
(a) Cl- 60 ppm + MoO42- 300 ppm
(b) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm
(b) Cl- 60 ppm + MoO42- 300 ppm + Zn2+ 50 ppm
(c) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm + Zn2+ 50 ppm
(c) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm (d) Cl- 60 ppm + EPA 300 ppm + MoO42300 ppm + Zn2+ 50 ppm
The spectrum of the film due to the environment consisting of 60 ppm Cl-, 300 ppm MoO42- and 300 ppm EPA (Fig. 3c) does not show any wavelength transition at 550 nm indicating the absence of any oxides of iron on the metal surface. The peak at 320 nm is due to an iron-molybdate complex formed on the metal surface. The reflectance spectrum of the film due to the environment containing 60 ppm Cl-, 300 ppm MoO42-,300 ppm EPA and 50 ppm Zn2+ (Fig. 3d) has a peak at 320 nm due to an ironmolybdate complex. Absence of a wavelength transition at 550 nm indicates the absence of oxides of iron on the metal surface. Analysis of the X-Ray Diffraction Patterns The XRD patterns of the film formed on the surface of the metal specimens immersed in various test solutions are given in Fig. 4. The XRD pattern of the film due to the environment consisting of 60 ppm Cl- and 300 ppm MoO42- is given in Fig. 4a. The film consisted of Fe2(MoO4)3 (2θ = 14.1°, 22.6°, 30.63° and 31.88°) [23]. The peaks due to iron appear at 2θ = 44.5°, 64.8° and 82.2°. The film formed on the surface of the metal specimen immersed in the environment containing 60 ppm Cl-, 300 ppm MoO42- and 50 ppm Zn2+ contained Fe2(MoO4)3 (2θ = 22.0°, 26.7°, 45.5°, 66.1°) [23], ZnO2 (2θ = 45.5° and 66.1°) [24] and γ-FeOOH (2θ = 60.9°) [25]. The iron peaks appear at 2θ = 44.6°, 65.0° and 82.3°.
487
Corrosion Protection and Monitoring
Figure 4. XRD patterns
Figure 5. Luminescence spectra
(a) Cl- 60 ppm + MoO42- 300 ppm
(a) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm
(b) Cl- 60 ppm + MoO42- 300 ppm + Zn2+ 50 ppm
(b) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm + Zn2+ 50 ppm
(c) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm (d) Cl- 60 ppm + EPA 300 ppm + MoO42- 300 ppm + Zn2+ 50 ppm The film due to the environment containing 60 ppm Cl-, 300 ppm MoO42- and 300 ppm EPA consisted of Fe2MoO4 (2 θ = 34.9°) [26]. The peaks due to iron appear at 2θ = 44.5°, 64.9° and 82.2°. The film formed on the surface of the metal immersed in the environment containing 60 ppm Cl-, 300 ppm MoO42-, 300ppm EPA and 50 ppm Zn2+ consisted of Fe2(MoO4)3 (2θ = 22.7°) [23], ZnMoO4 (2θ = 30.5°) [27] and ZnO2 (2θ = 41.1°) [24]. The iron peaks appear at 44.6°, 65.0° and 82.4°. Analysis of the Luminescence Spectra The emission spectrum (λex = 300 nm) of the film formed on the surface of the metal immersed in the environment containing 60 ppm Cl-, 300 ppm EPA and 300 ppm MoO42- is given in Fig. 5a. This spectrum may be due to an Fe2+-EPA complex and Fe2MoO4. The emission spectrum (λex = 300 nm) of the film due to the environment consisting of 60 ppm Cl- , 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ is given in Fig. 5b. This spectrum may be due to an Fe2+-EPA complex and Fe2(MoO4)3 in the presence of ZnMoO4 and ZnO2. Mechanism of the Inhibition of Corrosion The results of the weight loss method show that the formulation consisting of 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ had an inhibition efficiency of 99%. The polarization study revealed that this system acts as a mixed inhibitor. The FTIR spectra indicate that the protective film consisted of an Fe2+-EPA complex and ZnO2. The UVvisible reflectance spectra show that the film did not contain any oxides of iron. The XRD patterns show that the protective film consisted of Fe2(MoO4)3, ZnMoO4 and ZnO2. The film was found to be luminescent. In order to explain these observations in a holistic way, the following mechanism of inhibition of corrosion is proposed. 1. When the environment consisting of 60 ppm Cl-, 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ is prepared, there is formation of an Zn2+-EPA complex and a Zn2+-MoO42complex in solution. 2. When the metal is immersed in this environment, the Zn2+-EPA complex and the Zn2+MoO42- complex diffuse from the bulk of the solution to the surface of the metal.
488
Rajendran et al.
3. On the surface of the metal, the Zn2+-EPA complex is converted into an Fe2+-EPA complex in the local anodic sites, since the latter is more stable than the former. Zn2+-EPA + Fe2+ ---> Fe2+ -EPA + Zn2+
(1)
4. Similarly, the Zn2+-MoO42- complex is converted into an iron-molybdate complex, namely, Fe2(MoO4)3 Zn2+-MoO42- + 2 Fe3+ ---> Fe2(MoO4)3 + 3 Zn 2+
(2)
(Formation of an Fe3+-EPA complex and an Fe2+-MoO42-complex to some extent cannot be ruled out) 5. The released Zn2+ on the metal surface forms Zn(OH)2 in the local cathodic regions. Zn2+ + 2 OH- ---> Zn(OH)2
(3)
This may be converted into ZnO2 6. ZnMoO4 also forms on the metal surface. CONCLUSIONS 1. A synergistic effect was noticed between MoO42- and Zn2+; MoO42- and EPA; and MoO42, EPA and Zn2+. 2. Molybdate shifted the corrosion potential of the Zn2+, EPA or EPA-Zn2+ system to the anodic side. 3. The formulation consisting of 300 ppm EPA and 300 ppm MoO42- had a 99% inhibition efficiency. The protective film consisted of an Fe2+-EPA complex and Fe2MoO4. This film was found to be luminescent. 4. The formulation consisting of 300 ppm EPA, 300 ppm MoO42- and 50 ppm Zn2+ had 99% inhibition efficiency. The protective film consisted of an Fe2+-EPA complex, Fe2(MoO4)3, ZnMoO4 and ZnO2. This film was found to be luminescent. ACKNOWLEDGEMENT S. Rajendran wishes to thank the University Grants Commission, India, for awarding him a fellowship; and Mr. Ranjit Soundararajan, the Correspondent, Prof. S. Ramakrishnan, the Principal, and Prof. P. Jayaram, HOD, Chemistry Department, GTN Arts College, Dindigul, for their encouragement. REFERENCES 1. M.S. Vukasovich and J.P.G. Farr, Materials Performance, May 1986, p. 9. 2. A.Y. Al-Borno, R.A. Haleem, A. Al-Shatti, A. Abdulla and T.H. Mustafa, Technical Report No. 2132, Kuwait Institute for Scientific Research, Kuwait, 1986. 3. M.S. Vukasovich and D.R. Robitaille, J. Less-Common Metals 54, 1977, p. 437. 489
Corrosion Protection and Monitoring
4. C. O'Neal, Jr., R.N. Borger, Materials Performance 15, 1976, p. 9. 5. J.I. Bregman, US Patent 3,024,201, 1962. 6. T.C. Breske, Materials Performance 16, 1977, p. 17. 7. H. Leidheiser , Jr., Corrosion 36, 1980, p. 339. 8. J.P.G. Farr and M. Saremi, Surface Technology 17, 1982, p. 19. 9. D.B. Alexander and A.A. Moccari, Corrosion 49, 1993, p. 921. 10. M.R. Reda and J.N. Alhajji, Journal of the University of Kuwait (Science) 20, 1993, p. 171. 11. A. Vonkoepper, G.A. Emerle, K. Nishio and B.A. Metz, Materials Protection Performance 12, 1973, p. 23. 12. Y.J. Qian and S. Turgoose, British Corrosion Journal 22, 1987, p. 268. 13. A. Hussain, K. Habib and R. Jarman, Proceedings 7th European Symposium on Corrosion Inhibitors, Ferrara, Italy, 1, 1990, p. 621. 14. A. Al-Borno, Proceedings 7th European Symposium on Corrosion Inhibitors, Ferrara, Italy, 1, 1990, p. 583. 15. R.M. Silverstein, G.C. Bassler and T.C. Morrill, Spectrometric Identification of Organic Compounds, New York, John Wiley and Sons, 1981. 16. K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, New York, Wiley-Interscience, 1986. 17. A.D. Cross, Introduction to Practical Infrared Spectroscopy, London, Butterworths Scientific Publication, 1960. 18. L. Horner and C.L. Horner, Werkstoff und Korrosion 27, 1976, p. 223. 19. S. Rajendran, B.V. Apparao and N. Palaniswamy, Proceedings 8th European Symposium on Corrosion Inhibitors, Ferrara, Italy, 1, 1995, p. 465. 20. R.A.Nyquist and R.O. Kadel, Infrared Spectra of Inorganic Compounds, New York, Acadamic Press, 1971. 21. C. Sanchez, K.D. Sieber and G.A. Somorjai, Journal Electroanalytical Chemistry 252, 1988, p. 269. 22. S.M. Wilhelm and N. Hackerman, Journal Electrochemical Society 128, 1981, p. 1668. 23. JCPDS Nr. 200526. 24. JCPDS Nr. 130 311 25. M. Favre and D. Landolt, Corrosion Science 34, 1993, 1481. 26. JCPDS Nr. 251403. 27. JCPDS Nr. 251024.
490
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
EVALUATION OF CORROSION INHIBITORS FOR CARBON STEEL, MONEL 400 AND STAINLESS STEEL 321 IN A MONOETHANOLAMINE ENVIRONMENT UNDER STAGNANT AND HYDRODYNAMIC CONDITIONS J. Carew, H. Al-Sumait, A. Abdullah and A. Al-Hashem Materials Application Department Kuwait Institute for Scientific Research P.O.Box 24885, Safat, 13109, Kuwait
ABSTRACT Four organic based corrosion inhibitors have been evaluated for carbon steel, Monel 400 (UNS No 400) and stainless steel 321 (UNS No. 32100) in fresh monoethanolamine (MEA) environments saturated with an H2/CO2 gas mixture at 40oC. The test solutions were prepared from fresh MEA solutions with and without the addition of 250 ppm of each inhibitor. Initial screening of the inhibitors was performed using the wheel test to determine the corrosion rates of the three alloys with and without inhibitors. The rotating disc electrode (RDE) method was used to determine the effectiveness of the four organic inhibitors under hydrodynamic conditions. It was found that flow conditions tended to increase the effectiveness of some corrosion inhibitors with respect to stagnant conditions. The weight-loss and electrochemical tests conducted under hydrodynamic conditions indicated that the quaternary ammonium-based inhibitor was the most effective of the three alloys in the different MEA solutions. Key Words: Corrosion inhibitors, carbon steel, monel 400, stainless steel 321, monoethanolamine, weight-loss, rotating disc electrode
INTRODUCTION A large variety of corrosive conditions is encountered in the different industries. The costs of corrosion, and correspondingly, the savings gained through the use of appropriate corrosion mitigation techniques is considerable. Corrosion inhibitors are one of the main methods used to reduce corrosion problems in metallic installations all over the world. One of the most expensive and corrodible installations in chemical plants and refineries is the gas purification system. The most economical and effective method of protection is the addition of inhibitors to the closed circulation circuit of the system. Because carbon steel, UNS No 400 and stainless steel 321 (SS321) are the major alloys of construction in monoethanolamine (MEA) gas treating systems, both inorganic- and organic-based inhibitors have been utilized to reduce the corrosion rate in such an environment. However, due to environmental regulations and toxicity considerations, the use of inorganic inhibitors is declining and that of organic-based inhibitors is rising. The ability to evaluate and screen corrosion inhibitors for this type of application is important in order to choose an appropriate inhibitor.
493
Corrosion Protection and Monitoring
It is well known that flow velocity exerts a great influence on the corrosion rate of metallic materials [1,2]. Despite this knowledge the importance of the flow rate is not sufficiently considered in evaluation tests of corrosion inhibitors. The use of circulating loops to enable the examination of a wide range of flow rates (laminar and turbulent) is sometimes considered too expensive and time consuming because of the long sequence of operations needed [3]. Several authors [2,4,5] suggest simulating the turbulent flow present in many systems by using rotating disc electrodes (RDEs). In this technique, cylindrical electrodes are rotated at different speeds to measure the effect of flow at various velocities. This work was undertaken to screen four corrosion inhibitors in fresh monoethanolamine saturated with 85% CO2/15% H2 using the weight-loss and RDE methods. The aim of this investigation was to determine which corrosion inhibitor produces the lowest corrosion rate for carbon steel, UNS No 400 and SS321 in MEA solution by correlating the data obtained by the weight-loss and RDE methods. EXPERIMENTAL PROCEDURE Materials The materials tested were carbon steel (ASTM A283 grade B) Monel 400 and SS321. All the alloys were supplied by the Kuwait National Petroleum Company (KNPC) and represent the materials of construction of the plant. Inhibitors Tested Four corrosion inhibitors were tested under stagnant conditions by the weight-loss test, and under flow conditions by the RDE technique. The inhibitors were labeled A, B, C and D for simplicity and are shown in Table 1. Table 1. Type of Commercial Inhibitors Tested Inhibitor Type A B C D
Chemical Family Amides Aklylamino acid and ethylene glycol Amines Quaternary ammonium compound
Weight-Loss Method The procedure for the weight-loss method is essentially like that given in ASTM standard G-31 (1990). The specimens were in the form of coupons in dimensions of 40 x 20 x 1 mm. Before exposure to the test environment, the surfaces of the coupons were successively ground with 180, 400 and 600 grit silicon carbide papers, washed with detergent and dried. The weight of each specimen was determined accurately on an electronic balance. Prior to immersion in the test medium, the coupons were decreased in acetone and dried by hot air and hung with nylon thread. The test vessels were 800 ml Pyrex glass fitted with a gas inlet and immersed in thermostatically controlled water baths. The volume of the test solution was 400 ml. Each cell contained 2 coupons of carbon steel. During the course of the
494
Carew et al.
test, coupons were removed after 2 and 4 weeks. After weight-loss determination, the corrosion rate was calculated from the weight loss data as (mpy) from the following formula according to ASTM G-31 (1990). Corrosion rate = (K x W)/(A x T x D)
(1)
where K = constant = 3.45 X 106, W = mass loss in g, A = area in cm2, T = time of exposure in hours, and D = density in g/cm3. Rotating Disc Electrode (RDE) For the RDE measurements, the system Model 616 RDE, by EG&G PARC, was used, mounting and rotating the cylindrical electrodes at 200, 1000, and 3000 rpm. Some tests were conducted under stagnant conditions for comparison. The corrosion rates for the steel electrodes were determined at each speed selected, by using a potentiostat/galvanostat M000odel 273 A by EG&G PARC through the linear polarization resistance (LPR) measurement method. The electrochemical cell used consisted of the working electrode, a graphite counter and a saturated calomel electrode (SCE) as a reference electrode. The test was conducted with 400 ml of fresh MEA solution. Tests were conducted in 4 uninhibited MEA solutions as well as solutions inhibited ones with addition of 250 ppm of each inhibitor. The solution was continuously purged with a gas mixture of 85% CO2 and 15% H2 ,and the temperature was maintained at 40°C. RESULTS Weight-Loss Method Carbon Steel. Figure 1 shows the corrosion rate of this alloy in fresh (18% H2O) MEA solution in the presence of four organic-based inhibitors at a temperature of 40°C for a period of 4 weeks. To determine the most effective inhibitor in the MEA solution for this alloy, it was decided to rank the inhibitors by comparing the corrosion rates of the alloy in the absence and presence of inhibitors. In other words, the corrosion inhibitor that reduced the corrosion rate of carbon steel to the lowest value would be ranked as the best, and the one with the highest corrosion rate would be ranked as the worst. The corrosion rate of carbon steel in fresh MEA and in the absence of any corrosion inhibitor was considered to be the reference point (blank conditions). Therefore, the ranking of the inhibitors in terms of their corrosion performance for carbon steel in fresh MEA solution was as follows (Fig. 1): D>C>A>B Monel 400. Figure 2 illustrates the corrosion rates of this alloy in fresh MEA solution in the absence and pressure of the four inhibitors at 40°C and after four weeks of immersion. The ranking of these inhibitors in terms of their corrosion protection to Monel 400 in the fresh MEA-solutions is as follows: A>D>B>C
495
Corrosion Protection and Monitoring
Figure 1. Corrosion rate of carbon steel in fresh MEA solution as a function of inhibitor type at a temperature of 40°C
Figure 2. Corrosion rate of Monel 400 in fresh MEA solution as a function of inhibitor type at a temperature of 40°C Stainless steel 321. Figure 3 shows the corrosion rate of this alloy with and without inhibitors at 40°C for 4 weeks. The ranking of those inhibitors for this alloys was as follows: D > C> A>B
496
Carew et al.
Figure 3. Corrosion rate of stainless steel in fresh MEA solution as a function of inhibitor type at a temperature of 40°C Rotating Disc Electrode Technique Carbon Steel. Figure 4 shows the corrosion rate of carbon steel in fresh MEA solution with and without corrosion inhibitors under hydrodynamic conditions at 40°C. At a rotational speed of 200 rpm the inhibitors were ranked as follows: D > A> B> C However, at high speeds, the inhibitors were not as effective as under blank conditions. Monel 400. Figure 5 shows the corrosion rate of Monel 400 in fresh MEA with and without corrosion inhibitors under three different hydrodynamic velocities. The corrosion protection performance varied from one inhibitor to another, as well as from one speed to another. None of the inhibitors were effective for Monel 400 under hydrodynamic conditions. Stainless Steel 321. Figure 6 shows the corrosion rate of this alloy in fresh MEA with and without corrosion inhibitors under hydrodynamic conditions at 40°C. The inhibitors were ranked in terms of their corrosion protection as follows: C>B>A DISCUSSION This investigation was carried out to evaluate the relative performance of 4 organicbased inhibitors for the CO2 removal system of one of the refineries in Kuwait using MEA solution. The 3 main alloys that comprise such a system are carbon steel, Monel 400 and SS321. The two methods used in the evaluation process were the weight-loss and RDE methods representing stagnant and flow conditions, respectively. The 4 inhibitors were studied to asses their effect on the general corrosion of carbon steel, Monel 400 and SS321 in CO2 saturated fresh MEA solutions under stagnant and hydrodynamic conditions. Under stagnant conditions, the corrosion rates of carbon steel in fresh MEA (Fig. 1) was slightly more than 0.2 mpy for blank conditions. The addition of inhibitors A, C and D tended to lower the corrosion rate of carbon steel to an acceptable level. However, inhibitor B was found to enhance the corrosion rate of carbon steel to an acceptable level. This behavior might be attributed to the nonuniform distribution of the inhibitor film on the surface of the carbon steel specimen.
497
Corrosion Protection and Monitoring
Figure 4. Corrosion rate of carbon steel in fresh MEA solution as a function of inhibitor type and rotational speeds at a temperature of 40°C
Figure 5. Corrosion rate of Monel 400 in fresh MEA solution as a function of inhibitor type and rotational speeds at a temperature of 40°C
Figure 6. Corrosion rate of stainless steel in fresh MEA solution as a function of inhibition type and rotational speeds at a temperature of 40°C The corrosion rate of Monel 400, as shown in Fig. 2, was surprisingly high for such type of nickel-based alloy under stagnant blank conditions. The addition of any of the four inhibitors at the recommended dosage level reduced the corrosion rate quite dramatically especially for inhibitors A and D.
498
Carew et al.
Figure 3 shows the corrosion rate of SS321 under stagnant conditions which was quite low in the uninhibited media. The addition of inhibitors C and D reduced the corrosion rate of SS321 to very low levels. However, the addition of inhibitors A and B increased the corrosion rate of this alloy. This behavior was observed for SS304 and SS316 in identical media in previous studies [6,7]. Under hydrodynamic conditions, the corrosion rate of carbon steel, as shown in Fig. 4, indicates that under blank conditions, the flow velocity tended to decrease the corrosion rate of this alloy, in comparison to stagnant condition. This observation has been reported by many authors [1,2,4,5]. However, the dissolution rate of the steel cylinders in the inhibited MEA solutions were some what independent of the rotation velocity. This behavior may be interpreted by assuming the formation of a thick surface layer on the carbon steel electode. This layer strongly hindered either the anodic or the cathodic reaction or both on the surface of the steel electrode. Such a phenomenon was also reported by Zucchi et al.[2]. The corrosion rate of Monel 400 under flow conditions (Fig. 5) was lower under uninhibited conditions for the three different velocities than for samples with inhibitors A, B, C, and D. According to Fig. 5, Monel 400 did not seem to be affected by the different rotational speeds in the uninhibited MEA solutions. In other words, the passive oxide layer on the surface of Monel 400 was sufficient to resist destruction at up to 3000 rpm. The increase in the corrosion rate of this alloy upon the addition of the 4 inhibitors could be attributed to the removal or nonuniform formation of inhibitor film under hydrodynamic conditions. Figure 6 shows the corrosion rate of SS321 under hydrodynamic conditions in the uninhibited and inhibited MEA solutions. The addition of inhibitors A, B and C tended to lower the corrosion rate of SS321 under flow conditions. Inhibitor C seemed to be the most effective one in reducing the corrosion rate of SS321. Based on the results obtained by the weight-loss and RDE methods, the inhibitor that seemed to be the most effective in reducing the overall general corrosion rate for the three alloys was inhibitor D (quaternary ammonium compound). Generally, the inhibition mechanism of this compound is due to its ability to adsorb into the metal or alloy surface to form a protective film. Organic inhibitors can adsorb on to a metal surface by hydrogen bonding, by electron donation from the nitrogen atom, or by interaction of the dipole with the surface charge [8, 9, 10]. CONCLUSIONS The most effective corrosion inhibitor that may be used for carbon steel, Monel 400 and SS321 as determined under laboratory conditions, is the quaternary ammonium compound. The corrosion rate of the three alloys in the inhibited fresh MEA solutions varied with respect to flow conditions. The corrosion rate of carbon steel under such conditions was independent of the rotational speed, slightly increased for Monel 400 and decreased with respect to SS321. ACKNOWLEDGMENT The authors would like to acknowledge the in-kind support of KNPC’s - Shuaiba Refinery of this research work.
499
Corrosion Protection and Monitoring
REFERENCES 1. E. Heitz, Corrosion ‘90, Paper No. 1, NACE, Houston, Texas, USA., 1990. 2. F. Zucchi, G. Trabanelli, and G. Brunoro, Effect of flow velocity on corrosion inhibition of steel in HCl, in Progress in the Understanding and Prevention of Corrosion, Vol.2, J. M. Costa and A. D. Mercer, eds.,The Institute of Materials, p. 845, 1993. 3. J.L. Dawson, C.C. Shih, R.G. Miller and J.W. Palmer, Corrosion 90, Paper No. 14, NACE, Houston, Texas, USA., 1990. 4. A. Mazanek and H. Bala, Corrosion Science 28, 5, 1988, p. 513. 5. D.C. Silverman, Corrosion prediction from circuit models application to evaluation of corrosion inhibitors, in electrochemical impedance - Analysis and Interpretation, J. Scully, D. Silverman, and M. Kendig Editors, ASTM Publication, Philadelphia, Pensylvania,1993, p. 192. 6. M. Islam, A. Abdullah, W. Riad and G. Mansi, An investigation of corrosion and its control in the PIC monoethanolamine carbon dioxide removal unit”. Kuwait Institute for Scientific Research, Report No. KISR 1446, Kuwait 1984. 7. M. Islam, A. Abdullah, W. Riad , R. Al-Taib and G. Mansi, An investigation of corrosion and its control in the PIC MEA carbon dioxide removal unit laboratory studies”, Kuwait Institute for Scientific Research, Report No. KISR 1603, Kuwait 1984. 8. A. Moccari and D. D. Macdonald, Corrosion 41, 5, 1985, p. 263, 9. J.S. Robinson, Corrosion Inhibitors, 1979. 10. C.C. Nathan, Corrosion Inhibitors, NACE publications, Houston Texas, USA, 1973.
500
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
LABORATORY EVALUATION OF THE EFFECTS OF OZONE ON CORROSION RATES AND PITTING OF ENGINEERING ALLOYS S. Nasrazadani Department of Materials Engineering Isfahan University of Technology, Isfahan, 84156, Iran
ABSTRACT Cyclic polarization experiments were performed on 1018 steel, yellow brass, and 6061-T6 aluminum in ozonated and non-ozonated tap water under stagnant conditions to evaluate pitting and corrosion tendencies of these engineering metals. Results show that the corrosion rate of 1018 carbon steel in tap water under stagnant conditions increased about three-fold with the injection of ozone (0.05-0.1 ppm). No considerable changes in the corrosion rate occurred for yellow brass and aluminum when ozone was added under similar conditions. Study of the pitting behavior of the materials also demonstrated that ozone did not increase pitting when injected into the test solution for yellow brass. But in the case of 1018 carbon steel and aluminum, pitting was more pronounced. Key Words:
Ozone, pitting, corrosion rates, electrochemical testing, engineering alloys, laboratory evaluation
INTRODUCTION It is now a well known fact that the injection of ozone into the cooling water circulated in cooling towers can help to prevent biofouling due to the biocidal property of ozone [1-4]. Ozone is a strong oxidizer that is unstable at standard temperature and pressure [5]. Recently ozone has attracted a great deal of interest for its application as a corrosion inhibitor for open recirculating cooling towers. Some of the advantages of ozone versus multichemical water treatment systems currently in use include: (1) minimal on-site chemical inventory, (2) no toxic discharge, (3) water conservation, and (4) cost effectiveness. However, there are potential disadvantages for ozone applications that include: (1) possible corrosiveness to metallic components and (2) incompatibility with common inhibitors. Literature data indicate conflicting results on the effect of ozone on the corrosion of metals. Pryor and Buckay [6] summarized some case histories of both success and failure of ozone application in cooling towers of different industries. Matsudaira et al. [7] reported about two to three-fold increases in the corrosion rate with increasing ozone concentrations for carbon steel under stagnant and slow flow states in deionized water, and five to six-fold increases for brass under similar conditions. Walton [8] performed similar experiments as Matsudaira et al. [7] and reported a decrease in corrosion rates for mild steel and aluminum with increasing ozone concentrations. Obviously there are contradictions in these two cases as well as in a number of other investigations. Furthermore, to the best of our knowledge no research has been done to investigate the pitting tendencies of cooling tower materials when exposed to ozonated aqueous solutions. The objectives of this study were to
501
Corrosion Protection and Monitoring
evaluate the effect of ozone on the corrosion rates and pitting tendencies of 1018 carbon steel, yellow brass, and aluminum. EXPERIMENTAL PROCEDURE The cyclic polarization technique was used to study the pitting behavior and corrosion rates of 1018 carbon steel, yellow brass, and aluminum 6061-T6. This technique enables both pitting tendency evaluation as well as corrosion rate measurement of a metallic specimen in a given metal-solution system. The setup consists of a test cell, a potentiostat (VersastatTM), and an ozone generator using an ultraviolet lamp. The test cell contained the test solution (tap water), and the test specimen which was prepared according to procedures outlined in ASTM G5-87 [9]. Dry air was passed through the ozone generator and into the cell at a constant flow rate. The specimen was soaked in the solution for one hour before the cyclic polarization scan to attain a steady-state condition. The specimen was then polarized from -0.25 V with respect to the opencircuit potential (oc) to 0.7 V, and then reverse scan to -0.05 V (oc). All potentials referred to in this paper are with respect to saturated calomel electrode (SCE). The pH of the solution before and after the scan was measured. Also, the ozone concentration in the solution before and after the polarization scan was determined by either the indigo trisulfonate method or the DPD method (supplied by the Hach Company) whenever applicable. The corrosion rates of the materials were then determined by the EG&G Corrosion Measurement and Analysis software (MODEL 352 SoftCorrTM II) which uses a nonlinear least-squares fit algorithm to fit the data to the Stern-Geary equation [10] to obtain parameters for the calculation of corrosion rates. A metallograph (Versamet UNITRON 5327) was employed to document the surface conditions of the test samples before and after each test. RESULTS AND DISCUSSION The concentrations of ozone were maintained within the range of 0.05-0.1 ppm for all experiments involving ozonation. The pH of the test solutions were found to remain relatively unchanged in the region of 7.3 throughout the course of the scan for either case with or without ozone. 1018 Carbon Steel A comparison of the corrosion rates of 1018 carbon steel in ozonated and non-ozonated tap water (Table 1) indicates an increase of about three times in the corrosion rates from about 16.5 mpy to about 56.5 mpy with the injection of about 0.1 ppm of ozone. These numbers agree with the findings of Matsudaira et al. [7] who reported about two to three-fold increases in the corrosion rate with increasing ozone concentrations for carbon steel under stagnant and slow flow states in deionized water. The open-circuit potential of steels exposed to ozonated water shifted toward the cathodic direction by about 0.1 V. Such a slight shift in cathodic direction for a lightly ozonated system agrees with the results of Strittmatter et al. [11]. The general shape of the cyclic polarization plots (Fig. 1) in the absence of ozone indicates the presence of knees at pitting potential (Epit) where the corrosion current increases drastically. Such Epit was not seen for the steel sample tested in ozonated water. In addition, extrapolation of the reverse scan shows that the protection potential (Epro) is lower than the pitting potential (Epit) meaning that the tendency for pits to form is great. Indeed, photographs of the 1018 carbon steel specimens surfaces tested in ozonated and non-ozonated solutions before and after the scan (Fig. 2a-d) show that extensive pitting had occurred on the steel samples tested in ozonated solutions. 502
Nasrazadani
Figure 1 also indicates that pitting occurred on steel at a much lower potential with the introduction of ozone. This can be explained by the gentler anodic slopes of the plots, and the absence of the kneel as observed in Fig. 1 for steel samples tested in ozonated solutions. However, inspection of the photographs taken on those specimens with ozonation (Fig. 2d) shows no differences in the extent of pitting compared to those without ozonation (Fig. 2b). One can conclude that ozone does not increase the extent of pitting in carbon steel, but instead, accelerates the formation of pits. Table 1. Corrosion Rates of 1018 Carbon Steel Experiment Number Dry Air Only
1 2 3 4
Ozonated Dry Air
1
Corrosion Rates (mpy) 9.57 8.39 22.80 25.47 -------Avg. 16.50 33.94 2
3 4
Open-Circuit Potential (V vs. SCE) -0.3530 -0.4640 -0.3970 -0.4140
-0.5080 59.77
64.11 68.26 -------Avg. 56.50
-0.5520 -0.5320 -0.5090
503
Corrosion Protection and Monitoring
Figure 1. Cyclic polarization plots of 1018 carbon steel exposed to ozonated (0.1 ppm) and nonozonated solutions
504
Nasrazadani
Figure 2. Micrographs of 1018 carbon steel surfaces exposed to non-ozonated (a and b) and ozonated (c and d) solutions before (a and c) and after (b and d) cyclic polarization experiments (magnification 50x) Yellow Brass Corrosion rates for yellow brass remain relatively unchanged in the low range between one and four mpy (Table 2) for both cases in ozonated and non-ozonated conditions. The overlay of both ozonated and non-ozonated cyclic polarization plots (Fig. 3) shows no significant shift in the corrosion potential (Ecorr). These results contradict the findings of Matsudaira et al. [7] who reported five to six-fold increases for the corrosion rates of brass. In either ozonated and nonozonated conditions, the general shape of the plots is almost identical implying that ozone does not change the extent of pitting for yellow brass. The small area of the hysteresis loop also indicates that pitting is not as serious a problem as in carbon steel. Photographs of yellow brass test specimens (Fig. 4a-d) confirm this conclusion. Figure 4 shows no real differences in the number of pits for cases with and without ozonation. Table 2. Corrosion Rates of Yellow Brass Experiment Number Dry Air Only
Corrosion Rate (mpy)
1 2 Avg.
Ozonated Dry Air
1
3.02 3.86 -----3.44 1.12
2
-0.0023 3.42
Avg.
Open-Circuit Potential (V vs. SCE) -0.0018 -0.0048
-0.0017
-----2.27
505
Corrosion Protection and Monitoring
Figure 3. Cyclic polarization plots of yellow brass exposed to ozonated (0.1 ppm) and nonozonated solutions
506
Nasrazadani
Figure 4. Micrographs of yellow brass surface exposed to non-ozonated (a and b) and ozonated (c and d) solutions before (a and c) and after (b and d) cyclic polarization experiments (magnification 50x) Aluminum 6061 As in the case of yellow brass, the corrosion rates for aluminum remain relatively low and without much variation between 0.78 mpy to 1.12 mpy for both ozonated and non-ozonated conditions (Table 3). The overlay of the plots (Fig. 5) also shows no significant differences in the general shape or shift in the Ecorr under either condition. However, the area of the hysteresis loop is larger in the case of the ozonated scan which indicates that pitting has increased. Indeed, a comparison of the micrographs taken before and after the scan (Fig. 6 a-d) shows that pitting is more extensive under ozonated conditions. Table 3. Corrosion Rates of 6061 Aluminum Experiment Number Dry Air Only
Corrosion Rate (mpy)
1 2 Avg.
Ozonated Dry Air
1
1.07 0.79 -----0.93 0.45
2
Open-Circuit Potential (V vs. SCE) -0.4570 -0.4560
-0.0023 1.12
Avg.
1.12 -----0.90
-0.4450 -0.4900
507
Corrosion Protection and Monitoring
Figure 5. Cyclic polarization plots of aluminum 6061-T6 exposed to ozonated (0.1 ppm) and non-ozonated solutions
508
Nasrazadani
Figure 6. Micrographs of aluminum 6061-T6 surfaces exposed to non-ozonated (a and b) and ozonated (c and d) solutions before (a and c) and after (b and d) cyclic polarization experiments (magnification 50x) CONCLUSIONS Laboratory electrochemical cyclic polarization experiments showed that ozone increases the corrosion rate of carbon steel in municipal water under stagnant condition. As for aluminum, the corrosion rate remained relatively unchanged. The corrosion rate of yellow brass also did not show much significant variation under both ozonation and non-ozonated conditions. A study of the cyclic polarization plots and micrographs taken before and after the cyclic polarization scans showed that ozone does not increase the number of pits, but instead, accelerates the formation of pits for carbon steel. For yellow brass, ozone neither increases or decreases pitting as there was no shift or change in the corrosion potential and the general shape of the polarization curves. This result was confirmed by microscopic examination, where no differences were observed in yellow brass under either ozonated or non-ozonated conditions. As for aluminum, ozone did not increase the number of pits formed. This result was also confirmed by the micrographs and the increase in size of the hysteresis loop of the cyclic polarization plots. REFERENCES 1. H.B. Edwards, Ozone: An alternate method of treating cooling tower water, Journal of the Cooling Tower Institute 8, No. 2, 1987. 2. J.T. Echols and S.T. Mayne, Cooling-water cleanup by ozone, Chemical Engineering, May 1990. 3. J.T. Echols and S.T. Mayne, Cooling tower management using ozone instead of multichemicals, ASHRAE Journal, June 1990. 4. K. Stopka, Our experiences with 14 ozone-treated cooling towers, in: Ozone Treatment of Water for Cooling Applications, ed. R. Rice, Proc., IOA Seminar, Cincinnati, Ohio, June 1981. 5. R. Rice and A. Netzer, eds., Handbook of Ozone Technology and Applications, Ann Arbor MI, Ann Arbor Science Publisher, 1982. 6. A. Pryor and M. Buckay, Historical perspective of cooling tower ozonation, Industrial Water Treatment, October 1990, pp. 26-32. 7. M. Matsudaira, M. Suzuki, and Y. Sato, Dissolved ozone effect on corrosion of metals in water, Material Performance, NACE, Nov. 1981. 8. J.R. Walton, The effect of ozone on the corrosion rate of metals, 6th Ozone World Congress, May 1983. 9. ASTM G5-87, Standard Reference Test Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements, ASTM, 1987. 10. M. Stern and A.L. Geary, Electrochemical polarization: a theoretical analysis of the shape of polarization curves, Journal of the Electrochemical Society 104, 56, 1957. 11. R.J. Strittmatter, Bo Yang, and D.A. Johnson, Application of ozone in cooling water systems, NACE, Paper No. 347, 1992.
509
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
A CRITICAL COMPARISON OF CORROSION MONITORING TECHNIQUES USED IN INDUSTRIAL APPLICATIONS M.S. Reading1 and A.F. Denzine2 1
Cortest Instrument Systems, Inc., Willoughby, Ohio USA 2
Cortest International, Inc., Chardon, Ohio USA
ABSTRACT For over thirty years, corrosion coupons (weight-loss coupons) have been used in industrial processes to measure corrosion rates. These simple and reliable devices provide an accurate measure of corrosion rate in any environment, but require an extended time base before meaningful (and therefore historical) information can be obtained. Electrical resistance (E/R) probes provide the means to electrically transmit and record corrosion rate data based on the correlation of the increase in the electrical resistance of the probe element to the metal loss of the element. Although useable in any process environment, this technique also requires an extended time base before data is available. The linear polarization resistance (LPR) technique provides instantaneous corrosion rate data. However, it is applicable only in conductive media. Newer techniques provide the combined benefit of the rapid response of LPR probes to the universal applicability of E/R probes and coupons. In this paper, the limitations of all current corrosion monitoring techniques will be critically evaluated. For each technique, application limitations will be described in terms of corrosive media characteristics (i.e., temperature, pressure, conductivity, H2S concentration, pH, and phase) as well as probe characteristics (i.e., response time, sensitivity, alloy availability, and means of insertion). The result is a definitive guide to the selection of the proper corrosion monitoring technique for any industrial application. Key Words: Corrosion monitoring, new, rapid techniques, MICROCORTM, LPR, electric resistance
SYNOPSIS The electrical resistance (E/R) and linear polarization resistance (LPR) methods for online corrosion monitoring are compared with a new technique based on measurement of inductive resistance. Particular emphasis is given to the breadth of applicability of techniques, and the speed with which they measure changes in corrosion rate. Comparative test data show the new technique to have both the rapid response of LPR plus the universal applicability of E/R. INTRODUCTION Electrical resistance (E/R) and linear polarization resistance (LPR) are the industry's standard technique for on-line, instrumented, corrosion monitoring. The former is universally applicable, but response time is measured at best in days or weeks [1,2]. The latter gives
511
Corrosion Protection and Monitoring
rapid response, but is restricted to continuous electrolyte corrodents, and provides only semiquantitative data [1]. Newer techniques [2] such as thin-layer activation, ultrasonic transducers, the "field signature-method", electrochemical noise, and AC-impedance are more costly, some prohibitively so. They are also extremely slow to respond, or give complex and ambiguous data requiring expert analysis and interpretation. All have specialized applications, and none show promise as a general monitoring technique that overcomes the limitations of the more traditional methods. Inductive resistance measurements do, however, offer the response speed of LPR coupled with the universal applicability and absolute measurement capability of E/R. This paper reviews the performance characteristics of all three techniques. Electrical Resistance Measurements E/R, the most widely used of instrumented corrosion monitoring methods, measures the increase in electrical resistance of a metal sensor as its cross-section is reduced by corrosion. It will operate in almost any environment, and it provides an absolute measure of metal loss [3]. Only sulfide environments present difficulties for the E/R technique; electrically conducting iron sulfide corrosion products cause underestimates of metal loss. In extreme cases, sensors will indicate a metal gain [4] which, being obviously erroneous, is less serious than moderate instances showing corrosion rates that are lower than the actual values. This seriously limits the utility of the technique in sour oil and gas production systems, and refineries processing sour crude. Figure 1 shows comparative weight-loss and electrical resistance data in sour service.
Figure 1. Effect of iron sulphide on E/R sensors 512
Reading and Denzine
A second, and more fundamental drawback is the lengthy response time of this technique. Sensors that are sufficiently long-lived and rugged enough to be used in industrial environments have very low electrical resistance, as illustrated in Table 1. Removal of sufficient metal for a measurable resistance increase to occur can take days, or even weeks [2,3]. Response time is a function of sensor thickness, absolute resistance, and corrosion rate. Typical manufacturers' data relating response time and probe-life to corrosion-rate and element thickness are shown in Fig. 2. Evidently the most sensitive senors will take almost five days to register a corrosion rate change of 1 MPY. Perhaps this is more clearly illustrated in Fig. 3a and 3b, which show data from electrical resistance sensors of varying thicknesses in stagnant, potable, water showing a corrosion rate of about 5 MPY for carbon steel. Figure 3a also illustrates the increasing scatter experienced as the sensor resistance decreases, this is dramatically illustrated when comparing data from the 40 mil wire-loop and 20 mil cylindrical sensors. Table 1. E/R Element Resistance
Figure 2. E/R response time/probe life 513
Corrosion Protection and Monitoring
(a)
(b) Figure 3. E/R data scatter Compensation for temperature effects in E/R sensors is accomplished by measuring the resistance ratio of the corroding sensor to a second, noncorroding sensor housed within the probe’s body. The exposed and protected sensors are equally influenced by temperature, and
514
Reading and Denzine
consequently, ratio changes result exclusively from metal loss at the exposed sensor. Figure 4 illustrates the effect of small, rapid, temperature perturbations on sensor readings. Although the resulting changes in probe readings are significant, they recover in less than two minutes, and adequate temperature compensation is demonstrated.
Figure 4. Temperature compensation of MICROCORTM and E/R corrosion probes The effect of larger, slower, temperature changes is shown in Fig. 5. A temperature drop of 60°C over 30 hours, produces serious deviations in sensor readings. Corrosion rates derived from any two consecutive data points vary between 0-12 MPY, and relatively sharp changes in corrosion rate are indicated at 20, 30 and 60 hours. Both effects are purely artifacts of sustained temperature differences between the exposed and protected sensors. This is explainable by the large initial temperature differences between the process fluid and the exterior, ambient environment, coupled with the high heat exchange capabilities of the rear portion of the probe and its associated entry fitting. Irrespective, a trend of 3.5 MPY has taken 120 hours to establish due to temperature perturbations. Such undesirable effects are common when using E/R sensors [1,4]. Linear Polarization This technique is based on the fact that the current (Δi) required to produce a small shift (ΔV) in the potential of a current-corroding electrode is proportional to the corrosion current density (icor), and hence, to the corrosion rate (R). The essentials of the technique are embodied in derivations of the Stern-Geary [5] equation and Faraday's laws: icor = Δi.C1 / 2.3 ΔV; R = icor . C2
(1)
C1 is a constant comprising a function of the Tafel rate constant (βa, Bβc), and C2 is a constant involving dimensional factors, equivalent weight, density, time conversions, and the Faraday’s constant. 515
Corrosion Protection and Monitoring
Figure 5. E/R effects of temperature variation LP gives rapid data (5-15 minutes), a direct measure of rate, and a resolution in excess of 0.5 MPY. It has been widely used throughout the industry for more than 35 years. However, it is restricted to use in continuous electrolytes [1]. Multiphase, or discontinuous, corrosive fluids such as emulsions, condensing vapors, thin moisture films, and condensates prove difficult, or impossible, to monitor using this technique. Furthermore, increasingly resistive electrolytes produce IR-drop errors of ever increasing magnitude [6]. Despite improved electrode designs and AC polarization corrections [6] for IR-drop errors, unambiguous and straightforward measurements are generally restricted to natural waters, brines, cooling waters, seawater, potable waters, and some inorganic solutions. Iron sulfide corrosion products will eventually cause electrical shorts between the electrodes of a LP probe. Although this is a less severe constraint than it is for the E/R method, the presence of significant quantities of sulfide in the corrosive environment will eventually cause a malfunction of LP sensors. Redox couples with potentials close to the corrosion potential will cause overestimation of the corrosion rate as measured by LP. Part of the polarizing current (Δi) will be supplied by oxidation/reduction of the redox species, as opposed to a sole contribution from the corrosion reaction. In systems where the electrodes are passive, such errors can be very large, since almost all of the polarizing current can come from spurious redox couples. Although not widely reported, this phenomenon can cause vast overestimates in the corrosion rate. An example of this process involves polysulphide anions in wood pulp digestion fluids [7]. The most serious inaccuracy of the LP technique results from the use of an average value for the βaβc/(βa + βc) factor in the Stern-Geary relationship. All commercially available measurement instrumentation incorporates an empirical, average, constant. In practice, βa and βc values may vary by a factor of ten [7,8] or more, and resultant βaβc/(βa + βc) values 516
Reading and Denzine
may vary by a factor of three for commonly encountered alloy/solution environments. Consequently, corrosion rates, as measured by the LP technique, have an uncertainty of ±150%, unless each system is initially calibrated using full DC polarization curves. This does not negate the use of LP as a relative measure of corrosion, unless apparent changes in rate are accompanied by changes in mechanism, and concomitant changes in β values, such as might occur in going from acid corrosion under activation control to oxygen-induced corrosion under diffusion control. Data from LP sensors is approximate, possibly bordering on being qualitative only. Inductive Resistance This resembles the E/R technique, insofar as it provides an absolute measure of metal loss, and is applicable in almost any environment. It is dissimilar in that a small change in metal mass produces a large change in measurement signal, making it possible to measure small changes in corrosion rate in minutes rather than days. This patented technology (MICROCORTM) measures the changes in the inductive resistance of a coil embedded within the metal/alloy sensing element as the mass of the sensing element decreases due to corrosion. The sensing element, having a high magnetic permeability, greatly intensifies the magnetic field surrounding the coil, which in turn causes considerable magnification of the inductive resistance of the coil. Inductive resistances equivalent to 1-5 Ω can be developed in such a sensor, as opposed to 2-60 mΩ for E/R sensors of similar geometry, yielding improvements in both resolution and response time of 100-2500 times that found with E/R techniques. This resolution and response time is not decreased by temperature noise, since the thermal coefficients of magnetic permeability are several orders of magnitude lower than the equivalent parameters for electrical resistivity. Although temperature compensation is required, the same principle can be applied as with E/R sensors, and this is sufficient to almost totally eliminate the spurious effects of temperature. Figure 4 shows the temperature compensation efficiency of an inductive resistance probe compared with a conventional E/R probe. At this time, no extensive testing has been undertaken to ascertain the influence of corrosion product film on the effectiveness of the measurement. Preliminary indications, from tests conducted in deaerated acid solution, show the technique to be unaffected by magnetic films. This being the case, it is anticipated that iron sulfide corrosion products will not adversely affect the measurement. Trials of the measurement technique in sulfide environments are planned for the near future. Figure 6, shows comparative data for a LP probe, a cylindrical (10 mil) E/R probe, and a MICROCORTM probe (10 mil). This data was obtained in potable water, alternating between stagnant and flowing conditions, and finally in an acidified state. Both the LP and MICROCORTM data show sharp discontinuities between the various corrosive conditions, the changes being apparent within 15 minutes of implementation. The E/R probe, at the lower corrosion rates, indicates some corrosion, but the data scatter is too great to meaningfully assign any values within the time periods for which individual corrosion conditions were maintained. At the higher corrosion rates, approximate rates can be obtained from the data within 10-20 hours.
517
Corrosion Protection and Monitoring
Figure 6. MICROCORTM, E/R, LPR comparison CONCLUSIONS The MICROCORTM technique represents a fundamental breakthrough. It offers the possibility of measuring changes in corrosion rate rapidly and accurately in any corrosive
518
Reading and Denzine
environment. These combined attributes cannot be found in any of the other techniques currently employed for on-line corrosion monitoring. REFERENCES
2. 3. 4. 5. 6. 7. 8.
1. S.W. Dean, ASTM STP908, Corrosion Monitoring in Industrial Plants, 1986, pp. 197218. M.W. Joosten, K.P. Fisher, R. Strommen and K.C. Lunden, Materials Performance, April 1995, pp. 44-48. G.L. Cooper, ASTM STP908, Corrosion Monitoring in Industrial Plants, 1986, pp. 237250. G.R. Cameron and L.G. Coker, ASTM STP908, Corrosion Monitoring in Industrial Plant, 1986, pp. 251-265. M. Stern and A.J. Geary, Journal of the Electrochemical Society 104, 1957, p. 56. K.M. Lawson, N.G. Thompson, and M. Islam, NACE International Corrosion Conference, Paper No. 310, 1994. R.A. Yeske, ASTM STP908, Corrosion Monitoring in Industrial Plants, 1986, pp. 266288. K.M. Lawson, N.G. Thompson, M. Islam and M.J. Schofield, British Journal of NonDestructive Testing, 35, 6, June 1993, pp. 319-329.
519
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
DETECTION, LOCALIZATION AND MONITORING OF STRESS CORROSION CRACKING, HYDROGEN EMBRITTLEMENT AND CORROSION FATIGUE CRACKS DURING SERVICE CONDITIONS USING ACOUSTIC EMISSION L. Giuliani ATEL s.r.l., Pomezia, Italy
ABSTRACT Acoustic emission (AE) is a nondestructive technique originally developed to locate possible defects on nuclear vessels during hydrotesting. The development of a different signal analysis and of an interpretation code based on characteristics of environmental crack growth allows the use of such a simple technique to inspect plants during service, and to detect and locate active defects. A number of industrial applications for detection of stress corrosion cracks, of embrittled zones and of corrosion fatigue cracks will be presented. AE can be further used to evidence of the effectiveness of electrochemical treatments aimed at stopping stress corrosion cracking (SCC). Key Words: Acoustic emssion, nondestructive techniques, plant inspection, active defects, stress corrosion cracks, fatigue cracks
INTRODUCTION Among localized corrosion forms, those caused by interplay between a specific corrosion medium and mechanical stresses (either applied or residual) can provoke potentially dangerous cracks. In most cases, such cracks are generated on the internal wall of a vessel, a pipe or another plant component and cannot be detected unless an outside leak is produced or a complete internal inspection is carried out during a plant shutdown. Acoustic emission (AE) is the only nondestructive technique that can reveal the presence of active cracks, i.e., those defects growing under service conditions, and therefore, is especially suited to detect, locate and monitor the occurrence and propagation of stress corrosion, hydrogen embrittlement and corrosion fatigue, the so-called environmental cracking. The Technique and the Theory AE was initially developed to locate metal defects on relatively large structures (typically nuclear vessels) during the final hydrotesting. The theory of fracture mechanics assumes that a defect, when subjected to a suitable load, can progress in a discontinuous mode, and therefore, it would be possible to detect its presence during the increase of an applied load such as pressure. Because of this theoretical approach, AE has been mostly used in hydraulic tests, and localization techniques based on a large number of sensors have been developed. Also when applied to a plant component after a certain period of service, e.g., as in the requalification of pressure vessels, the AE test has been consistently associated with an increase of pressure. To detect an AE signal from a far away source requires a very high gain 521
Corrosion Protection and Monitoring
of the amplification chain; a fact that severely limits the use of AE in a noisy environment. Unfortunately, the results obtained with AE as a defect localization system have been very poor for a number of reasons, among which the use of the fracture mechanics concepts might be considered major. A much more successful application of AE can be obtained when the electrochemical approach to environmental fracture is adopted. The basic concept of electrochemical theory is the existence of surface conditions (quantified and described by the electrochemical potential) which can nucleate and propagate brittle cracks, while the mechanical factor (external and/or local stresses) only determines the rate of crack propagation. Crack propagation itself depends on a number of phenomena (e.g., local corrosion, acidification, formation and rupture of protective oxides or films, hydrogen diffusion, and dislocation movement), most of which are totally or largely insensitive to the extremely small variation in local stress induced by a pressure increase. Accordingly, AE must be applied only when such critical conditions exist and for a length of time consistent with the specific type of brittle cracking. For all practical purposes, the electrochemical theory suggests the use of AE during the normal service of a given plant component to detect whether active cracks are present. This requires, in turn, the capability of distinguishing a signal generated by a brittle crack extension from all possible signals generated by concomitant, but not interesting, phenomena (noise). Acoustic Emission Energy The parameter that permits distinguishing between crack activity and noise is the energy transported by the signal. An AE signal is a damped oscillating wave, and its energy is proportional to the square of the maximal amplitude. For each signal, it is necessary to measure the maximal amplitude, to calculate the squared value and to add all contributions in order to generate an energy output. Other types of energy calculations (e.g., the envelope of the signal and the integral) are not suitable; therefore, the calculation of the maximal amplitude squared has been called true energy analysis (TEA). The main reason for favouring this parameter is the large difference in value between the energy released by a crack the moment it grows compared with the energies associated with noise signals. In order to measure the maximal amplitudes of AE signals, it is very important to avoid signal deformation caused by excessive gain in the amplification chain; in other words, the range of amplitudes to be measured should be known. The TEA Code and Environmental Cracking Interpretation of AE data to detect and locate environmental cracks is possible on the basis of a set of criteria which form the TEA Code. Obviously such criteria are related to the specific cracking phenomenon and take into account how crack propagation mechanisms are expected to generate AE signals. A brief discussion of environmental cracking phenomena can help in identifying basic evaluation criteria for AE signals. In the case of stress corrosion cracking (SCC), the crack nucleation and growth processes can be simplified as follows:
• Occurrence of localized corrosion on points of the metal surface not covered by a passivation layer, or rupture of such a passivating layer; • Formation of a geometry able to concentrate mechanical stress and specific chemical compounds (mainly H+ and Cl-); 522
Giuliani
• Brittle extension of the crack; • Formation of a passive layer on the new surface exposed to the environment. Although this simplified description of the SCC mechanism can explain the possibility of quite a large number of oscillations, when the energy content of the two processes (corrosion and brittle fracture) are considered, it is apparent that crack extension should give a stepwise jump increase in the energy-time plot because the energy involved is much higher and is released suddenly. Considering that an SCC mechanism can generate a large number of cracks which will progress at different rates (due to local disuniformity of metallurgy and stress), it can be assumed that several jumps should be obtained on the energy-time plot, with the magnitude of the individual jumps comprising a relatively limited range. The first criterion for SCC crack detection is the presence of jumps in the energy graph. The second criterion is the occurrence of several jumps of similar magnitude, without a fixed or processrelated frequency. A third criterion is that such jumps are much more frequent when the metal-solution interface conditions are suitable for SCC, and the jumps disappear when such conditions are modified. For hydrogen embrittlement, a general mechanism for low to medium-strength steels (UTS up to 60 kg/mm2) can be summarized as follows:
• Hydrogen atoms are produced at the metal surface either by electrolytic reduction of hydrogen ions or water molecules, or by reaction with hydrogenated compounds; • Hydrogen atoms enter the metallic matrix (especially in the presence of promoters such as H2S, CN-, As and Sb compounds) and diffuse to regions of high local stresses such as those around hard inclusions (e.g., aluminates and aluminosilicates). • A brittle area is created around the inclusion, and this results in metal detachment; • The stress relaxation following the formation of the crack promotes a diffusion of hydrogen away from the cracked area to adjacent inclusions. On the basis of such a simplified mechanism, relatively few, but highy energetic, signals are generated at the moment the embrittled metal is cracked; cracks can be more localized because they depend on the metallurgical defects of the steel, and they will occur at longer time intervals when compared to SCC cracks. More isolated energy jumps are expected. Again, there is no relation between crack signal frequency and any other variation in the plant’s operating conditions. The corrosion fatigue cracking rate is directly related to the stress variation frequency, and therefore, for practical purposes, AE signals are much more frequent and increase in number with the length of the crack. Energy jumps are more regularly spaced but different in magnitude because the amount of fractured surface in the elemental crack growth step is different. In the case of a noisy environment, corrosion fatigue cracks can be evidenced by peaks in the plot of energy/oscillation against time. For all three phenomena, there are locations where cracking is more likely to occur: welds, junctions, geometrical discontinuities and the like. It is, in general, always possible to compare AE signals obtained near one of these preferential locations with signals obtained
523
Corrosion Protection and Monitoring
from locations where the probability of cracking is very low. Further, it is possible to utilize signal attenuation to locate zones or regions of the structure where cracks are active. Finally, based on the described interpretation method, continuous monitoring becomes feasible as a few transducers independently listen to detect possible nucleation of brittle cracks. AE Confirmation A major problem of conducting an AE test during hydrotesting is the well known Kaiser effect: if a defect emits at a given pressure, it will remain silent when the test is repeated unless the previous pressure is exceeded. This effect hinders the repetition of the AE test (apart from practical and economical difficulties) in order to verify the presence of a defect. The application of AE to in-service inspection completely overcomes this problem because defect propagation is a continuously repeated process. It is possible, therefore, to confirm the existence of active defects in any given location by repeating the AE inspection at any time.
RESULTS Some typical results of AE inspection are presented in order to evidence the usefulness of the energy calculation. Corrosion Fatigue A carter of a large CO2 compressor for a urea plant was inspected to detect corrosion fatigue cracks during operation. One transducer was positioned near a carter weld for less than 20 minutes and then moved to another weld. The amount of noise associated with the compressor is enormous, and in every position on the carter, the number of oscillations is nearly constant and in excess of 50,000/minute. The calculation of energy, however, shows a completely different picture. Figure 1a refers to a position where no crack was present and shows how the energy content of the AE signals is constant and very low. Figures 1b and 1c report results obtained on a weld where a crack had been previously revealed by penetrant dye and on a weld that had passed the penetrant dye test, respectively. In both cases, the amount of emitted energy was much higher and discontinuous in time, while no difference at all can be noticed in the amount of oscillations when compared to Fig. 1a. Another crack was detected on the second weld after grinding and repetition of the penetrant dye test. Figure 2 gives the amount of emitted energy per minute in the three locations, plotted in a semilogarithmic scale: it is easy to see that cracks emit energies which are orders of magnitudes higher than noise. Hydrogen Cracking A high-temperature reactor containing hydrogen was inspected to detect whether there was an active process of cracking under the prevailing operating conditions. By positioning one transducer at several locations on the vessel, one zone near a manhole was found emitting highly energetic signals. Figure 3 gives a typical result obtained from the defective zone; it should be compared to the zero emission at the other locations on the vessel. It is evident that
524
Giuliani
(a)
(b)
(c) Figure 1. Results from a carter of a CO2 compressor of a urea plant at (a) a position where no crack was present; (b) a weld where a crack had been previously revealed by penetrant dye; and (c) a weld that had passed the penetrant dye test 525
Corrosion Protection and Monitoring
Figure 2. The amount of emitted energy per minute at three locations 526
Giuliani
Figure 3. Typical results obtained from a defective zone in a high-temperature reactor containing hydrogen 527
Corrosion Protection and Monitoring
Figure 4. Results obtained on a Horton sphere containing LNG (the three plots refer to reptitions of the test on the weld) 528
Giuliani
Figure 5. A plot of the total emitted energy against time during nucleation and initial propagation of cracks in a stainless steel pipe through which hot CaCl2 was flowing
Figure 6. A plot of the energy derivative ΔE/Δt as a function of time during the period of SCC crack propagation in Monel pipes
529
Corrosion Protection and Monitoring
Figure 7. An AE signal before treatment
Figure 8. An AE signal after treatment the discontinuous mode of emission of the hydrogen cracking phenomenon, coupled with the high frequency of signals, indicate the presence of a relatively large number of cracking sites. At low temperatures, hydrogen cracking is a much slower process and longer times must be allowed. Figure 4 shows the results obtained on a Horton sphere containing LNG. The three plots refer to the same weld on which the test was repeated three times over two days. Again, comparison should be with zero emissions of nearby welds. Stress Corrosion Cracking AE signals become very frequent in the case of SCC because this cracking mechanism produces a very large number of cracks. Figure 5 is a plot of the total emitted energy against time during nucleation and initial propagation of cracks in a stainless steel pipe through which hot CaCl2 was flowing: the discontinuous emission of high energy signals, which appear as jumps in the energy-time plot, is clear evidence of SCC. Monel pipes, subjected to the same solution, did not show any noticeable emission. A plot of the energy derivative ΔE/Δt as a function of time during the period of crack propagation (Fig. 6) gives an idea of the large amount of energy and the discontinuous mode of emission associated with SCC cracks. As stated before, the electrochemical theory says that SCC cracks can progress only in a limited range of potentials. If it is possible to modify such potentials, then cracks cannot propagate any longer. Examples of SCC control through electrochemical potential shift have been described for a gasoline distillation column of carbon steel, a PVC reactor of stainless steel and a 70,000 m3 530
Giuliani
gas holder of carbon steel [1]. AE signals can be used as evidence in real time of the effectiveness of a treatment aimed at potential modification: when cracks stop propagating, AE signals decrease to zero (Figs. 7 and 8). CONCLUSIONS AE is an invaluable, nondestructive technique when used to detect active cracks in industrial plants under service conditions because there is no alternative technique able to provide the same real time information. In many industrial plants, under normal service conditions, unless unexpected substantial overload occurs, the only active cracks are those that grow because of a specific environment, e.g., those generated by SCC, hydrogen embrittlement or corrosion fatigue. Such cracks do not progress because of any pressure increase as the external load is only one of the factors determining their growth, and in many cases, not the most important one, when internal stresses are considered. The use of AE for in-service inspection is based on proper analysis of signals and on proper correlation with the emitting phenomenon: the TEA Code is an example of such an analysis. The possibility of using AE while a plant is in service is the basis for its economical use. Indeed, AE should be used to locate those areas where active cracks are present and should be subjected to inspection during plant shutdown. Such a pre-check substantially reduces the need for extensive inspection with magnetic particles, penetrant dyes and the like, thereby saving time and money. Furthermore, the information on the actual presence of active defects can be used to increase the accuracy of normal NDT tests on the suspected areas for a better evaluation of defects. Finally, if the electrochemical approach to environmental cracking is adopted, it is often possible to find treatments which can minimize or stop the cracking phenomenon; in such cases, AE will indicate immediately whether or not such treatment is effective. REFERENCES 1. C. Caneva, L. Giuliani, and A. Pampallona, Stress corrosion cracking inhibition monitoring by means of electrochemical and acoustic emission measurements, International Conference on Monitoring, Surveillance and Predictive Maintenance of Plants and Structures, Taormina, Italy, 16-18 October 1989.
531
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
ELECTROCHEMICAL MONITORING OF AEROBIC BACTERIA AND AUTOMATION OF BIOCIDE TREATMENTS L. Giuliani Atel srl Via Valle Caia, 00040 Pomezia, Italy
ABSTRACT The recent development of electrochemical probes to monitor the first settlement of aerobic bacteria on metal surfaces allows an optimization of biocide treatments in cooling water circuits. Practical application to a steel mill cooling water circuit using contaminated seawater is described. The reduction in chemical injection resulting from such monitoring favors the choice of a more expensive but more efficient biocide compared to the common chlorine or hypochlorite. Key Words: Electrochemical probes, aerobic bacteria, biocide treatment, cooling water
INTRODUCTION Aerobic bacteria present in cooling waters tend to attach themselves to heat exchange surfaces under favorable conditions. The growth of bacteria colonies creates a microbiological fouling of the surface which can support slime formation and higher forms of life. The detrimental effects of biofouling are related to a reduction in the heat transfer coefficient, an increased power consuption to maintain constant flow of water and to the possibility of increased corrosion, especially on those materials which are sensitive to localized modifications of the metal-solution interface. The formation and growth of microbiofouling layers are highly dependent on water quality, especially in terms of amount of organic matter, plant process conditions, contamination and season of the year. It is, therefore, practically impossible to forecast what might happen in any specific heat exchanger or steam condenser. There are a variety of measures for biofouling control, but the most widely used one is injection of a biocide, although also continuous mechanical cleaning or heat shocks are often considered. The most popular biocide is chlorine, due to cost considerations; next to it, there are its compounds such as hypochlorite and chlorine dioxide. Most of the other biocides appear to be too expensive for large cooling water circuits. Chlorine tends to react with all the organic matter present in water by formating potentially dangerous substances, and for this reason, there is a growing concern over its ecological impact. Legislation already limits the amount of residual chlorine allowed in effluent water, and in some cases, the length of time during which discharge is permitted. Beside environmental problems, chlorination has some operational difficulties of its own whether chlorine gas is used or local electrochemical generation is preferred. It would be of interest also from an economical point of view to reduce a biocide treatment to the minimum compatible with the design fouling factor. 533
Corrosion Protection and Monitoring
BIOFOULING MONITORING As microbiofouling is a complex function of many variables, it is of obvious interest to detect when bacteria start attaching themselves to the metallic surface and how effective the biocide treatment is in removing them. Development of biofouling sensors has been oriented toward detection of the physical presence of a film (with optical means) or the measurement of the heat exchange coefficient. Those methods, however, have produced cumbersome and/or unreliable equipment, which are not really suited for long-term monitoring. A simple and effective method of monitoring has instead been developed by studying the electrochemical modifications produced by bacteria at the metal-solution interphase. It has been experimentally shown that:
• Bacteria colonies on metallic surfaces act like catalysts for oxygen reduction. The increase in the oxygen reduction rate can be easily measured with an electrochemical setup including a probe and an electronic instrument. • The readings of the instrument can be related to the percent of surface covered by bacteria. An electrochemical biofouling probe can be used for continuous monitoring of the cooling water and will detect the formation of the very first layer of bacteria by showing a sigmoid increase of signal from a baseline (no bacteria present on the probe surface) to a maximum (full coverage of the probe surface by bacteria colonies). Injection of chlorine results in a drastic drop of the probe signal, and the attainment of the original baseline value indicate complete cleaning of the metallic surface. Typical results in different locations are shown in Fig. 1 and clearly indicate the variance of the phenomenon especially in terms of incubation times. PILOT PLANT APPLICATION In Livorno, Italy, ENEL runs a pilot plant reproducing a closed cooling tower circuit. An electrochemical bioprobe was tested for more than 18 months in 1992-1993 and biocide has been added as a function of the probe readings, with positive results. Bacteria counts in the pit water used for this pilot plant have indicated that the probe starts giving a signal of biofilm formation when bacteria are in the range 104-105/cm3. POWER PLANT COOLING WATERS Biofouling monitoring was applied to a number of power station cooling water circuits, including well water and river or estuarine waters. Results from five different locations indicated the immediate response of the electrochemical probe to the onset of biofilms, with the only exception being one location where ferro-bacteria rather then aerobic bacteria were present. A typical signal of the bioprobe, located in Tihange-2, Belgium, is reported in Fig. 2. In another location, Dragenbos, Belgium, water comes from wells and has a high carbonate content; thus, a periodic sulphuric acid injection is made to decrease scaling tendency. A typical plot of the bioprobe signal against time reveals that the probe is sensitive to both acid and unnecessary chlorine injections, showing a sharp peak for the injection duration. When, however, a biofilm develops, the signal increases continuously and decreases when chlorine is injected (Fig. 3). It is precisely this behavior of the probe signal that indicates the effectiveness of the monitoring system: 534
Giuliani
Figure 1. Typical results obtained at different locations using the electrochemical biofouling probe
535
Corrosion Protection and Monitoring
Figure 2. A typical signal of the bioprobe which was located in Tihange-2, Belgium
536
Giuliani
Figure 3. Effect of chlorine injection on the signals of the bioprobe
537
Corrosion Protection and Monitoring
• When no biofilm is present, an injection of oxidant, such as chlorine will produce a sharp increase of the signal, lasting for as long as the oxidant-rich water flows through the probe;
• When biofilm is present, the biocide injection will produce a sharp decrease in the probe signal. INDUSTRIAL APPLICATIONS Since September 1993, a biofouling monitoring system has been in operation in the Ilva Steel Mill, located in Taranto, Italy. The plant derives its cooling water mainly from an internal sea (i.e., Mare Piccolo) which has only limited water exchange with the open sea. The warm and biologically contaminated waters of Mare Piccolo are ideal for mussel farming, and indeed, one of the farms is not more than 200 meters away from the plant’s water intake. The simplified seawater cooling circuit is shown in Fig. 4 from the two intake channels. After the filtering grids, the water is divided into four basins with different flow rates (average values: 2 x 28,000; 27,000 and 41,000 m3/h for a total average of 120,000 m3/h). Water first cools the steam condensers of the ILVA power plants, and then goes to two basins from which the individual cooling water streams of the steel plant are fed. The initial four basins are connected to the two intake channels by tunnels and a periodic observation of the amount of biofouling on the tunnel walls is carried out. Some monitoring panels are used to determine the biofouling rate and an average value of 60 kg/m2/year is obtained, with surges up to 10-15 kg/m2/month during springtime. The seawater is treated with chlorine dioxide by injecting this gaseous biocide into the initial intake channels before the filtering grids. Initial tests were made with a biofouling monitoring system including one probe, located on the bank of one of the intake channels and fed with untreated water, and another probe, positioned in a bypass on one of the mill cooling towers. The first probe indicates the incubation and growth of the biofilm as a function of the actual water conditions (i.e., temperature, season, etc); this probe is cleaned when it reaches its maximum value. The second probe is used to guarantee that no biofilm is formed from the treated water and is never cleaned. Typical results from the first probe are reported in Fig. 5 and show how the phenomenon is cyclically repeated on cleaned surfaces with different incubation times. A typical result from the second probe is given in Fig. 6: the probe signal remains low for a certain period of time after starting the biocide injection, then raises to a maximum where it remains until treatment is discontinued. Due to the length of the circuit, the probe will see the beginning and the end of the treatment with a delay, as is clearly shown in Fig. 6. After such initial tests, a fully automatic monitoring system has been installed, and its use has greatly helped in defining the proper injection schedule as a function of environmental variables. Although local management is very cautious on this subject, a decrease of 50% in injection has already been achieved, while progressive tests are still going on. BIOGUARD INSTRUMENTATION Based on the positive results obtained with electrochemical bioprobes, an automatic instrumentation system has been designed to operate the biocide dosing pump automatically. 538
Giuliani
The instrumentation includes three independent water circuits (i.e., pump, flow meter, and bioprobe) and three meters to read the probes. An additional electronic circuit generates a signal to start the injection pump when anyone of the three bioprobes exceeds a preset threshold, and a signal to stop the pump when all three bioprobes fall below another threshold. Water is fed to the Bioguard instrumentation through a flanged 1"-pipe connection and is discharged at the exit of the bioprobe manifold. The whole instrumentation is contained in a steel cabinet, with an IP 65 protection rating. There are no external displays or knobs as the meter panel is accessible only after opening the cabinet.
539
Corrosion Protection and Monitoring
Figure 4. A simplified seawater cooling circuit at Ilva, Taranato, Italy
Figure 5. Typical results from the first probe used at the Ilva Steel Mill, showing how the phenomenon is cyclically repeated on cleaned surfaces with different incubation times
540
Giuliani
Figure 6. A typical result from the second probe used at the Ilva Steel Mill, showing the changes in the probe signal as a function of time of the biocide treatment CONCLUSIONS The scientific observation that bacteria promote the oxygen reduction reaction on metallic surfaces has been developed into an automatic instrumentation system to control biocide injection into cooling waters. Extensive testing of this technique and debugging of some initial unexpected results have resulted in a reliable system to minimize biocide injection and to keep heat exchanger tubes clean from biofouling. The basic concept of this biofouling monitoring system is to inject the biocide only to remove those bacteria which get attached to metallic surfaces as opposed to the current practice of continuous or intermittent water sterilization. This new concept guarantees that the minimum amount of biocide is injected, with a resultant positive effect on costs and pollution. The reduction in the quantity of biocide can also promote the use of more sophisticated products which have so far not been economically competitive on the basis of extended dosages. REFERENCES 1. A. Mollica et al., Interaction between biofouling and oxygen reduction rate on stainless steel in seawater, 6th International Congress on Marine Corrosion and Fouling, Athens, 1984. 2. A. Mollica and G. Ventura, Electrochemical monitoring of biofilm growth and corrosivity in Seawater: Effect of intermittent chlorination, 8th International Congress on Marine Corrosion and Biofouling, Taranto, 1992. 3. L. Giuliani, On-line monitoring of aerobic bacteria to optimise chlorine injection, U.K., Corrosion 1993, Vol. 2, London, 1993.
541
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CORROSION MONITORING FOR INTEGRITY OF PIPELINE G.L. Rajani Engineering & Corrosion Services CII-3, Ansari Nagar, New Delhi 110 029, India
ABSTRACT Pipelines are generally externally protected against soil-side corrosion by a combination of protective coatings and impressed current cathodic protection. For the protection of the internal area of the pipelines, depending upon the corrosive tendency of the process fluids, corrosion control measures are adopted in the form of inhibitor dosing. Considering the inherent hazard posed in the event of the failure of pipelines, corrosion monitoring of the external and internal surface is necessary for pipelines in spite of the fact that proper corrosion control measures like protective coating and cathodic protection have been implemented. Monitoring is required to ensure that all sections of pipelines are protected against corrosion (external and internal) and also to assess the performance level of various control measures, and to thereby ensure that the corrosion prevention systems, e.g., the cathodic protection and protective coating, are performing adequately as intended and affording complete protection. Various causes of pipeline failures are discussed, and the methods for ensuring the safety and integrity of a system are described in detail. The capabilities and limitations of various techniques in use for monitoring the effectiveness of corrosion protection systems and for assessing the health of pipelines are discussed based on practical experience and established practice in the industry. Recommendations for selection of a particular method and program of monitoring to ensure the safe operation of a pipeline are also given. Key Words: Pipeline integrity, cathodic protection, protective coatings, corrosion control
INTRODUCTION A pipeline is generally externally protected against soil-side corrosion by a combination of protective coating and impressed current cathodic protection. For protection of the internal area of a pipeline, depending upon the corrosive tendency of the process fluids, corrosion control measures can be adopted in the form of inhibitor dosing. External corrosion protection coating is made up of an inner layer of fusion-bonded epoxy and an outer layer of yard-applied, hot, extruded polyethylene (PE) coating with a total thickness of 4 mm, or coal tar enamel coating (4-5 mm thick). Impressed current cathodic protection consists of the required rectifier station with permanent anode groundbeds located at predesigned intervals along the pipeline. Considering the inherent hazard posed in the event of the failure of pipelines, corrosion monitoring of the external and internal surface is a necessity for pipelines in spite of the fact that proper corrosion control measures, like protective coatings and cathodic protection have 543
Corrosion Protection and Monitoring
been implemented. Monitoring is required to ensure that all the sections of pipelines are protected against corrosion (external or internal) and also to assess or monitor the performance of various control measures, and to thereby, ensure that the corrosion prevention systems, e.g., the cathodic protection system and coating, are performing adequately as intended and affording complete protection. Gas sweetening, in the case of sweetened natural gas high-pressure pipelines, does not preclude the chance of internal corrosion due to fact that all gases carry a considerable amount of condensate (i.e., slug). This condensate contains water. Condensate settles in low lying areas of pipeline, and in the presence of oxygen, leads to corrosion of the internal surface of the pipeline. Also, in the case of a crude oil, the internal surface of a pipeline containing saltwater can suffer pitting corrosion. Many other factors like turbulence (leading to erosion-corrosion), deposit attack, and impingement attack also lead to internal corrosion. All of these conditions necessitate routine corrosion monitoring. It is well known that all coatings, however superior, deteriorate with time due to ageing, bacterial attack, water absorption and soil stress (which is very high in Indian soil) due to climatic conditions and alternate wetting and drying. The degradation of a coating is reflected in the fall in its electrical insulation property and mechanical damage which brings the pipeline into direct contact with the soil, which is inherently corrosive due to presence of water, Cl, SO4 and other ions. The corrosion, thereby, proceeds by either of the several mechanisms described above. The degradation of a coating has a direct influence on cathodic protection, since the lower the electrical insulation property of the coating, the smaller the spread of cathodic protection from each rectifier station due to the increase in the cathodic protection current requirement to prevent corrosion. Further direct contact of soil and water electrolytes with the metallic surface of a pipeline of areas of mechanical damage will lead to an increased demand for cathodic protection current. This increase, which is 50-100-fold, always falls below the current available in the system. Consequently the localized areas where the coating has been damaged no longer receive cathodic protection, and thus suffer active corrosion which can lead to pipe wall perforation within one year depending on the aggressiveness of the soil. Attack on weld seams is characterized by bursts/rupture over several meters along the longitudinal weld. Thus, the degradation of the coating in combination with the presence of inherent holidays leads to active corrosion by soil electrolytes. The corrosivity of the soil is directly related to the soil’s resistivity. A lower soil resistivity means a more aggressive soil. Soil resistivities may vary from 300 Ω-cm (i.e., extremely corrosive soil) to 10,000 Ωcm (i.e., least corrosive soil) along the pipeline. Even in areas of mechanical damage or degraded of coating in the form of tearing off by soil stress, disbonding, large holidays which have brought soil into direct contact with the pipeline’s surface, no corrosion will occur so long as the cathodic protection maintains a minimum pipe-to-soil potential of - 850 mV when measured with reference to a saturated copper-copper sulfate reference cell. In soils containing sulfate reducing bacteria (SRB), the minimum potential requirement is -950 mV (with respect to copper-copper sulphate reference cell) for adequate protection. Most cathodic protection systems are designed to cater to the increase in current required due to the degradation of the coating or a fall in the spread of protection (i.e., the influence of each T/R unit), but even with all of these design
544
Rajani
considerations, many sections of pipelines are without adequate cathodic protection, which leads to active corrosion from the soil side due to unexpected coating damage or disbonding during operation of the pipeline. PIPELINE CORROSION MONITORING Based on the above phenomenon and to enhance the health and integrity of a pipeline, government regulations in Europe and the USA, as well as in India, have made the monitoring of corrosion and various corrosion prevention measures mandatory at fixed regular intervals to safeguard against major hazards due to pipeline failure. The various tasks undertaken for assessing the integrity of a pipeline are as under:
• • • • • • •
Soil resistivity tests, C.P. performance monitoring including interference tests if required, Coating resistance tests, Test coupons and corrosion probes for internal corrosion, Corrosion detection pigging (intelligent pigging survey), Iron count, and Ultrasonic thickness measurement. Integrity Monitoring Integrity monitoring alone cannot prevent pipeline leakage or failures. It is only one of the complex procedures and precautions adopted for pipelines to ensure that safety and environmental hazards are controlled and minimized. Integrity surveys should complement the routine monitoring of the pipeline corrosion protection system, and the choice of survey technique should be technically sound and cost-effective. No single method is available that will provide all the information required, and some methods may not be suitable for specific applications. It is essential that integrity surveys be directed toward identifying points where loss of wall thickness, and thereby strength, has occurred, or may occur in the future because of external corrosion or physical damage. Internal corrosion in white oil lines is negligible, and therefore not considered. Generally, intelligent pigging techniques will provide data for wall thickness loss of more than 10%. Apart from third-party damage, or construction, or metallurgical failure, which are unlikely to be detected by corrosion monitoring, the most important aspect to determine is the areas where corrosion has occurred or is likely to occur. In areas where the coating is sound, the possibility of corrosion is minimal, and it is extremely unlikely that third-party contact with the pipeline has caused significant defect. In areas where the coating is damaged, it is possible that corrosion is occurring or that it may occur. It is also possible that the line has been gouged, scarred, or significantly deformed by third-party activity. Therefore, it follows that if points of coating damage are located, and CP effectiveness is assessed, then repairs can be made and the corrosion minimized. Where coating disbonding occurs, it is also possible for corrosion to occur. This type of defect may show up as a small coating defect with protected potential at the point of exposure. The corrosion potential and extent of disbonding is not evident using above-
545
Corrosion Protection and Monitoring
ground survey techniques and is only likely to be found by intelligent pigging if noticeable metal loss has occurred. As the pipeline ages, this cause of corrosion may become more prevalent. COATING SURVEYS There are a number of coating evaluation survey techniques available for burned pipelines. The main types of surveys are as follows:
• Pearson surveys (the AC technique)
• Electromagnetic current attenuation surveys (CATS) (AC technique) • DC voltage gradient survey (DCVG) A DC voltage gradient survey (Fig. 1) is similar to a Pearson survey but uses a pulsed DC signal applied between the pipeline and the earth. The operator uses two probes, spaced about 2 m apart, and advances along the line. When a defect is approached the potential gradient will become steeper, and by following the potential gradient until the gradient becomes zero, the operator will be above the coating defect. Precise defect location is established by locating a null, that is, by positioning the probes in the vicinity of the defect so that the indicator remains on zero. At this point, the defect is exactly at the midpoint between the probes. Probe separation is practical down to 300 mm, which gives defect location to within 150 mm. The needle always points in the direction of the defect. Once the defect is accurately located, the potential gradient from remote earth to the defect location is measured using the formula: % IR = OL/RE x 100 PS
(1)
where OL/RE is the Potential gradient from the defect’s location to remote earth in mV, and PS is the pipe-to-soil potential in mV. It is then possible to establish the severity of the defect. As can be seen from the formula, the severity takes into account the size of the defect size and the available cathodic protection current. From the % IR drop alone, it is not usually possible to determine the size of the defect accurately because of variations in soil conditions and pipeline depth, but as a guide, 6% IR represents a defect with bare steel being in contact with approximately 10 cm2 of soil at a depth of 1 m. The following criteria are used to establish the severity of the coating defect from Eq. 1.
• Less than 15% IR would not normally require coating repair work and would be protected satisfactorily by the cathodic protection system, • Between 15 and 50% IR, requires general maintenance (within 4 years), and • More than 50% IR requires immediate coating repair/recoating. CP Surveys The only practical measurement of CP effectiveness is the pipe-to-soil potential. The accepted criterion is a pipe-to-soil potential more negative than - 850 mV referred to a copper-copper sulfate (Cu-CuSO4) reference electrode, or - 950 mV where SRB may be 546
Rajani
present. Because the pipelines pass through a wide variety of soil conditions, and because it is not practical to test for microbiologically influenced corrosion, the criterion of - 950 mV has been adopted. Providing this criterion is met, corrosion on the pipeline is expected to be mitigated. The conventional method of monitoring has been to measure the pipe-to-soil potential at approximately 1 km intervals with the current switched on. Providing all the measurements have been more negative than - 950 mV, as given in international standards, it has been accepted that the CP is satisfactory over the entire length of the pipeline. While pipe-to-soil potentials measured at 1 km intervals may be acceptable, the potentials between test points can vary significantly if there are distinct coating faults. The pipeline may thus be unprotected at these locations. To locate these defects, pipe-to-soil potential measurements should be taken along the entire length of the pipeline. This entails taking measurements at 1.5-5.0 m intervals, which is known as close-interval potential logging survey (CPL). With the advance of technology, instrumentation and equipment for recording, large amounts of data have now made it practical to conduct detailed surveys to identify areas of unprotected pipe and provide an indication of where coating faults exist. This is particularly important at faults identified by intelligent pigging, faults which may not be severe enough to warrant immediate repair.
Figure 1. Principle of the DC voltage gradient method CPL Survey A CPL survey (Fig,. 2 ) is a very important tool for checking the performance level of a total cathodic protection system including the integrity of or extent of damage to the corrosion protection coating. This method is exhaustive, and therefore, falls under the intensive survey methods, and is undertaken less frequently than routine monitoring. Any
547
Corrosion Protection and Monitoring
deficiency in the external corrosion protection system which was undetected with routine monitoring is always detected with this method.
Figure 2. Close order potential logging (PL) survey technique to monitor pipe to soil potentials When conducting a CPL survey, the power sources normally are switched synchronously, and a polarized potential is taken at a maximum interval of 5 m. The typical switching period is 3 seconds off and 12 seconds on (4 cycles per minute). There is now sufficient concern that ignoring the errors caused by IR drop will lead to an increase in corrosion on pipelines This concern is international. Some countries are now making it mandatory for IR drop-free potentials to be measured to prove that CP criteria are being met. In fact in Germany, a protection criterion of -850 mV as the instant off’ value is from very early date. Where a close-interval polarized potential survey is conducted, the IR drop at the intermediate test points can be determined and the corresponding ‘on' potential can be used as a criterion for routine repeat monitoring. It is the measurement of the ‘off’’ potential that determines the effectiveness of the cathodic protection system. Areas of the potential profile on the pipeline that fall outside the selected ‘off’ potential criteria (i.e., -850 mV Cu/CuSO4) are not being provided with adequate cathodic protection, and could therefore be corroding. These regions could be suffering from
• Coating defect, • Stray current effects, and/or • Soil variance. Likewise, if the pipe-to-soil potential is too high (i.e., -1200 mV in the 'off’ condition Cu/CuSO4 limit) then the pipe could be overprotected and accelerated coating degradation can be expected from the excessive alkali generated. Lateral Close Order Potential Survey 548
Rajani
This technique is identical to the CPL survey with the addition of a second half-cell positioned 10 m laterally at right angles to the pipe (see Fig. 3). The potential differences are measured from the half-cell over the pipeline to the test post via the training cable (V). The potential difference is also measured between the two reference electrodes (U). Readings are taken with the cathodic protection systems on and off giving U = UON - UOFF and V = VON - VOFF readings from which a fault index, U/V, is calculated. This will indicate directly the severity of the coating defect. The ratio cancels out the effects of soil resistivity and attenuation. This technique has been extensively used in Germany.
Figure 3. Lateral half-cell close order potential logging (CPL) survey Advantages The advantages of lateral CPL are the same as for the standard CPL method. The additional advantage is that the lateral half-cell enables a less subjective interpretation of the straight overline graph. Disadvantages The disadvantages of lateral CPL are similar to those of standard CPL method, with the addition that the lateral technique is even more labor-intensive, requiring a minimum team of three, one of whom must be very experienced. Also, the technique is very slow, with rates of 1-1.5 km/day being typical. CORROSION DETECTION BY A COMBINATION OF DCVG AND CPL SURVEY CPL is used to identify regions of low potential, findings which imply coating faults. These regions are investigated in more detail by the DCVG technique, which monitors, in finer detail, the voltage gradient generated in the soil by the passage of cathodic protection current to a coating fault.
549
Corrosion Protection and Monitoring
The DCVG technique, like CPL, is directly related to the cathodic protection applied on the pipeline. Other coating fault techniques like the Pearson technique and C-scan have no such relationship to CP and lack the DCVG's ability to detect coating faults, especially in a complex piping network. A coating defect survey using the combined method is usually carried out as follows : 1. Conduct a DCVG survey and identify all coating faults, 2. Determine the defect size as gauged by the percentage of IR drop. 3. Using the CPL technique, measure the on and instant off potentials. The trailing cable is also utilized to obtain the pipe to remote earth potential for the IR percentage calculation. 4. Perform a soil resistivity survey to correlate the corrosive soil conditions at coating defect locations. A coating can have many faults and potential less negative than - 950 mV, and if the soil’s resistivity is high, no corrosion will occur. Based on the survey results, a comparison of the three coating defect survey methods (i.e., the Pearson, C-Scan, and DCVG) on buried coal tar enamel-coated pipeline indicated that DCVG detected nine defects, all of which when excavated were found to be actual defects, and all of which when excavated were found to be accurate. The Pearson technique detected three defects, while the C-Scan detected only two defects. The DCVG technique will also identify the percentage of defect value so that only major defects need be excavated. The combined CPL + DCVG survey can be carried out at a speed of 5-8 km/day. The DCVG technique has both value and direction for the defect and can pinpoint the exact spot where the current either leaves or enters the pipe. Cathodic Protection System Monitoring Cathodic protection monitoring is an important part of cathodic protection system maintenance and operation. Cathodic protection performance monitoring at fixed, regular intervals ensures that all the structures are receiving the desired protection against corrosion, and that the total system is performing as intended in the design stage. In other words, cathodic protection monitoring is aimed at the following two objectives :
• Assessing the level of cathodic protection by measuring the structure-toelectrolyte potentials at various locations or points of the structure, and • Inspecting and assessing the performance of the major cathodic protection equipment, e.g., the DC power source, T/R units, anodes, cables, junction boxes, and insulating joints/flanges. The various activities that are performed in regular monitoring are broadly classified as
• • • • • • •
Potential measurements at fixed point, Measurement, by CPL, Protective current density measurement, Interference measurement wherever suspected, Coating performance tests, T/R unit operation checks (DC power source) and inspection, Individual anode performance checks,
550
Rajani
• Insulation joint checks, • Casing/carrier short tests, and • Junction boxes. POTENTIAL MEASUREMENT METHODS Fixed-Point Potential Measurement This involves measurement of the structure-to-electrolyte potentials at discrete points on the structure with reference to a portable reference electrode (Ag/AgCl or Cu/CuSO4) placed on the electrolyte as close as possible to the protected structure. Measurements are made with the cathodic protection current both on and off condition. For systems working with sacrificial anodes, potentials are always measured in on condition. For onshore pipelines, fixed-point potential monitoring consists of measuring the pipeto-soil potential at permanent test stations located at 1 km intervals. For offshore pipelines, such monitoring consists of measurements at shore-based test points and diver-assisted measurement at bracelet anodes. For underwater structures, measurement is achieved with permanently monitored anodes or by a diver taking measurements at selected points on the structure. For tank bottoms, potential measurements are made at permanent test points keeping a reference electrode close to the tank rim, and taking both the on and off potentials. Close Order Potential Surveys Various methods for assessing the effectiveness of cathodic protection and thereby to ensure that every part of pipeline is adequately protected has been discussed. Transformer/Rectifier Operations: Checks and Inspection Monitoring should be carried out in two parts :
• Routine inspection (once every 15 days) • Annual inspection (once per year) Individual Anode and Anode Groundbed Inspection and Performance Test Inspection should take place once every 6 months, preferably during the peak summer and post-monsoon periods. The following parameters should be inspected/monitored:
• Any disturbance in the soil compaction over the anode bed. This would be attributable to washing away of soil during monsoon, 3rd party construction activity etc. • Anode groundbed resistance (to be maintained around 1.0 ohm). • Proper functioning of each individual anode so that each one of them is drawing almost equal current. • In case the soil over the groundbed is very dry, arrangement should be made to sprinkle water. Moist soil is required for functioning of anode bed. Insulation Joint Test Insulation joints should be carried out to check the effectiveness of the insulating flange. 551
Corrosion Protection and Monitoring
Interference Tests Cathodic protection interference is a source of great problems, and unless detected during routine monitoring or a follow-up investigative survey, e.g., CPL, will lead to costly pipeline failures, as has been experienced by many pipeline owners the world over. The main forms of interference are as follows:
• Interference caused by another cathodic protection system operated by a different pipeline owner, termed cathodic protection interference, and
• Stray current interference caused by DC railway traction, welding, etc., which forces a high magnitude of stray DC current into the earth, a part of which enters the pipeline and leaves the pipeline to return to earth, areas of pipelines where the current discharges (i.e., leaves the pipeline to return to earth) are areas of concern where proper levels of CP must be maintained (Fig. 4). The detection of such interference and its mitigation are two of the most important tasks during routine monitoring and investigative surveys undertaken to assess the performance level of the CP system. MONITORING PROGRAM FOR A CATHODIC PROTECTION SYSTEM AND INVESTIGATIVE SURVEYS The cathodic protection monitoring program should be as per Table 1. Table 1. Monitoring Program for a Cathodic Protection System Parameter Pipe-to soil potential (at test points)
Frequency Once every 3 months or once/month at locations of under protection,interference or any other abnormality
Potential measurement -- do -(at cased crossings)
Insulating flanges
552
Once/year
Remarks Integrity of each test point should be checked and any defects/ missing components should be replaced. Measurements should be taken at each test point using a Cu/CuSO4 half cell. Measure casing and career pipe-to-soil potential values at each cased crossing. Acceptable values casing to soil-natural potential. Carrier pipe to soil -850 mV (min), Cu/CuSO4 On/off potentials should be taken on either side of insulating joint and recorded to check performance of flange joint for CPcurrent flow, if any
Rajani
Figure 4. Stray-current corrosion
Table 2. Required Investigative Surveys Survey Instant off CPL (at approximately 1.5-5.0 m intervals) to detect coating holidays, efficacy of CP system, and existence of interference
Frequency Initially and subsequently once in every 5 years
Soil resistivity survey
Initially and once every 8-10 years
Coating evaluation conductance test, line current survey as part of CPL survey
Initially and once every 5 years
Pigging
Once/year
Intelligent pig survey for assessment of the internal/external metal loss/damage and plan for major repairs/ replacements
Initially and subsequently once in 5 years
Investigative Surveys Investigative surveys as given in Table 2 are required to be carried out initially after taking over the pipelines, and subsequently at the frequencies given. These surveys are very useful in identifying any major shortcomings which are likely to go undetected during routine monitoring. SELECTION OF MONITORING SYTEMS Each of the various survey techniques described, provides information which, when combined, provides a clear understanding of the condition of the pipeline. This must be followed by a thorough economic study leading to the implementation of a remedial program. Exposure of Test Coupons for Internal Corrosion Test coupons exposed at different locations should be removed periodically for inspection and determination of the corrosion rate on the fluid side. The iron count should also be analyzed periodically in fluid to check the pickup of iron from the pipeline’s surface. These tests will assess the internal condition of the pipeline. RECOMMENDATIONS It is recommended that pipeline be monitored externally to assess the CP performance level and the condition of the coating by methods discussed to safeguard the line against external corrosion damage by soil/water electrolyte and internally by fluid-side corrosion. 553
Corrosion Protection and Monitoring
The choice of technique should be made after taking into consideration the pipeline’s operating parameters, the type of CP, the pipeline’s characteristics, and the end result that is desired from a particular technique. All the results of corrosion monitoring, whether routine or follow-up (i.e., investigative surveys), must be immediately checked and reviewed by an experienced corrosion engineer. The comments and any corrective measures suggested by the specialist must be implemented as early as possible. REFERENCES 1. Harvey, Maintaining integrity on buried pipelines, Materials Performance, August 1994. 2. Leeds, A Critical Review of the Techniques Used to Delineate Pipeline Coating Defects as a Precursor to Refurbishment.
554
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
POWER AND DESALINATION PLANTS: PUMPS, CORROSION AND MAINTENANCE H. Hosni, N.J. Paul and A. Masri Water and Electricity Department, Abu Dhabi, UAE
ABSTRACT In the Emirate of Abu Dhabi, there are many power and desalination plants spread over several locations. Some of these plants are large and designed to produce 100 MG of water and 1000 MW per day. Seawater for desalination and cooling is brought through large diameter pipelines and huge pumps that work nonstop for years. There are different pumps to handle cooling water, fire water, brine, distillate, potable water, chemicals and crude for such large plants. These pumps are made of several alloys and metals, such as Ni-resist, stainless steel, and copper alloys, and corrosion failure has occurred due to galvanic, stress and other factors. Maintenance by repair, coating and cathodic protection are the subject matter of this paper. Key Words: Power plants, desalination plants, pumps, maintenance
INTRODUCTION In power and desalination plants where a variety of fluids are transported and stored, hundreds of pumps are utilized which are manufactured by a dozen of international companies. The fluids that are handled are:
• Water such as seawater, brine, distillate, potable water, steam LP and HP, makeup water, • Chemical dosing such as chlorine, hydrazine, caustic, acid, and • Fuel oil such as crude, diesel, and light oil. The pumps are manufactured by leading companies like KSB, Kubota, and Thermomechanica, and have been performing for several years. Some of the pumps have been running nonstop for years, and as is the case with many pumps, they have exhibited the following corrosion problems in some of their components.
• • • •
Cavitation of the impeller, Cracks in the casing, elbow etc., Pitting in the shaft, and Disintegration of the diffusers Some of the aforesaid problems are normal, but the repeated failure due to cracking seems to be unexplainable.
555
Corrosion Protection and Monitoring
• Is it due to a design failure ? • Is it a material or casting defect ? • Is it a stress associated with manufacturing or due to operation ? These questions arise because the problems are not repeated in all pumps from the same manufacturer or in the same part of the pump from different manufacturers. If the failure is stress-induced; is it due to an operational problem ? Some pumps have shown problems since their installation. In that case, are the problems due to installation problems ? Concrete foundations have failed only in some pumps. Does this suggest that a design defect has caused unusual vibration and stresses ? In addition, Ni-resist which is supposed to be the ideal material for seawater and brine applications, but has been defected by heavy general corrosion. Should we opt for a material like stainless steel ? Some stainless steels have also suffered cavitation problems in seawater. Then, should we select a coating or cathodic protection (CP) ? After twenty years of handling various problems, the following case studies have been selected to review and illustrate an appropriate course of preventive action. SEAWATER SUPPLY PUMPS A sketch of a typical seawater supply pump is given in Fig. 1. The material of fabrication is also given. Generally, the casing, bell impellers, etc., are made of Ni-resist and the shaft is made of stainless steel.
Figure 1. Schematic of a seawater supply pump 556
Hosni et al.
Whenever the pumps were taken out for maintenance, a lot of barnacles was found grown on parts of the pumps, in spite of chlorination (Fig. 2). The general place of failure in most of the pumps is the shaft which is pitted; then strong vibrations sets in resulting in the failure of concrete foundations. Does the small imbalance, combined with high velocity cause the concrete to fail or was it the other way around ? In either case, the concrete foundations have been restored, and the failure pattern of the pump was studied. Also, it is a matter of concern as to how the barnacles grew on the components in swift water flows with a high velocity.
Figure 2. A photograph of a seawater supply pump without CP. In one of the power stations where the pumps were cathodically protected by impressed current, the pumps did not show any failure (Fig. 3). Therefore, thirty four of such pumps from different manufacturers will be fitted with CP by the impressed current method. The anodes will be placed on the outerside of the pumps. The pumps which had CP did not encourage barnacle growth, while the ones which did not have CP were covered with barnacles. This matter needs further explanation. As for the internal components, the pumps without CP showed extensive damage at the web which is now being repaired with belzona. The success of the treatment with belzona is not, however, assured in all cases. In another case where the pump had CP, no failure was exhibited at any location, but a pump without CP had heavy cavitation at the stainless steel impeller (Fig. 4). In both cases, the casings did not suffer from damage, regardless of their differences in design. It is, therefore, a matter of judgment in deciding what is warranted before any conclusion can be drawn. 557
Corrosion Protection and Monitoring
Figure 3. A photograph of seawater supply pumps with CP
558
Hosni et al.
Figure 4. Two photos of seawater supply pumps: (a) without CP; and (b) with CP
Figure 5. The brine recirculation pump THE BRINE RECIRCULATION PUMP A drawing of the brine recirculation pump is given in Fig. 5, and the material specifications are provided in Table 1. The casing, delivery elbow, and diffuser are made of GGG, the suction bell, main pump shaft, and drive shaft are made of stainless steel (1.4404). The barrel is made of mild steel with rubber coating on the inside and paint on the outside. The majority of failure is inside the barrel at the mid-level with cavitation pattern and large pittings; the delivery elbow has cracks extending to several feet and holes at the bottom where turbulence and velocity is highest (Figs. 6 and 7).
559
Corrosion Protection and Monitoring
Table 1. Material Specifications for Pumps Serial No. 1.
Pump Designation Brine recirculating
2.
Seawater supply pump
3.
Seawater recirculation pump
4.
560
Distillate pump
Component Material Suction bell mouth 1.4404 Impeller Pump shaft 1.4406 Driveshaft Delivery elbow, GGG-Ni.Cr.-Nb20-2 Diffuser, riser pipe, Discharge cover Barrel MS 37.2, Rubber-lined Delivery elbow, column GGG- NiCr-Nb-202 pipe, diffuser Driveshaft, pump shaft 1.4406 Bell mouth 1.4404 Casing halves GGG-NiCr-Nb-202 Pump shaft Impeller Shaft sleeve Suction bell mouth Diffuser Delivery elbow, bearing spider, riser pipe Pump shaft, driveshaft, impeller Barrel
1.4406 1.4404 1.4136 G-Cu Sn10 G-Cu Al I0Ni 1.4404 MS 37.2
Hosni et al.
Figure 6. A photograph of external corrosion on a brine pump without CP
Figure 7. A photograph of a brine pump internal components with zinc anode CP One of the pump casings was internally lined at the bottom with type 316 stainless steel (to cover a huge hole); and this clad stayed in position for 10 years without any defect or failure. But, is the cost of cladding the whole casing internally with stainless steel justified ? Also, the corrosion encountered at the mid-position was not solved. Originally, the pump casing was internally coated with elastomeric product by the manufacturer, but the coating failed in a very short period of time. Later on, a thin rubber lining was tried in one of the pumps; but also failed. Therefore, belzona was used to patch the defective locations, and it is staying well. However, coating the entire inner surface with this product would be quite expensive. Thick rubber lining followed by vulcanization could be tried as an alternative, but handling equipment of this size would require a very large autoclave. In one of the pumps, internal CP has been implemented using sacrificial anodes specially designed to take care of turbulence and anode detachment. Caging was also installed in the case of anode detachment. Such an arrangement is shown in (Figs. 6 and 7). The zinc anodes, each weighing 10 Kg, have been found to last at least two years, and the pump is working well without additional coating maintenance. It appears, therefore, that CP controls such corrosion problems to a large extent. The pump barrel normally is in a pool of water as the subsoil groundwater seeps into the concrete pit. There is severe corrosion externally at the water level. To stop this external
561
Corrosion Protection and Monitoring
corrosion in one of the pumps, magnesium anodes were installed in the pool of water to afford protection for the portion submerged in brackish water. Cracking in the delivery elbow is a matter of great concern (Fig. 8). Some of the pumps have been stitch-welded using castoline welding rods. In some cases, the repairs have been successful, while in one or two cases cracks have extended in spite of extreme care being taken during welding. Is this a material failure or an operational problem ? Two pumps made by the same manufacturer, but fixed at different job sites, have exhibited the same type of failure. Thus, the failure cannot be due to an operational problem.
Figure 8. Cracks in the delivery elbow SEAWATER CIRCULATION PUMPS These are pumps which are used only during the winter season. When the water temperature rises above 25°C, the pumps are kept idle (Figs. 9 and 10), which could be for a period of at least six months/year. This standby situation causes a lot of corrosion on the inner side of the casings. Several longitudinal cracks were also observed and were attributed to casting defects. The pump’s casing is in two parts, and the defect lines can be seen as deep-cut grooves and cracks in the pictures. Several coating compounds, such as chestertone, have been applied, but without total success. Some of the pumps have been coated inside with belzona. All of these coating compounds can fill the cracks, but they may not be able to impart mechanical strength during operation and withstand turbulence. Therefore, alternative protection measures have been thought out. Welding may be helpful but it may cause additional stresses in the absence of heat treatment. In most of the cases, only the bottom portion of the casing is affected. Would it be sufficient to coat only the bottom and leave the top as it is, and thus reduce the cost of 562
Hosni et al.
repair by 50% ? In this case, since the shaft is made of stainless steel, the top portion could act in an anodic manner when the pump is idle. It is not known whether the anodic effect would be uniformly distributed, or if it would affect the flange end and cause more leakage. Could an additional anode be placed inside ? Would this introduce more turbulence ? Could stub anodes be inserted for impressed current CP ? What would be the current requirements ? Is there a provision in casting for the placement of such anodes ? All these considerations are shown in Fig 11.
Figure 9. A photograph of a seawater circulation pump
Figure 10. A photograph of a distillate pump (foreground)
The calculations of CP for a seawater circulation pump is as follows: The total area of coating for the internal parts is 14,800 cm for one-half, and the area of the stainless steel shaft is 11,756 cm . If the top half is not coated and is left to act as an anode for the shaft, an area ratio of 1:12 would be available. Thus, the top half could serve as an anode, but it would corrode at the rate of 8 kg/A year. The current requirement for the shaft would be 176 A at the rate of 150 mA/m of steel area. Therefore, about 1.5 kg of the upper casing would be consumed as anode. If this corrosion occurs uniformly, it is not a problem. However, failure pattern showed the situation to be otherwise. If an impressed anode is installed as a stub, it could easily take care of the pump for several years. Ti/MMO anodes are already being used in CP of seawater pumps. They last
563
Corrosion Protection and Monitoring
for over 10 years and could easily work in this case. A schematic diagram of such installation is shown in Fig. 12.
Figure 11. Coating and welding at defective spots
564
Hosni et al.
Figure 12. CP by impressed current system Table 2. Supplier, Part Function, Type of Part and Repair Method Supplier KSB
Function Brine recirculation pump Brine blow down Seawater supply
Thermo mechanica
Sea water recirculation Circulating water pump Cooling water pump T.W. Pump Brine recirculation
Type SEZT 1100
SELT 700-730
Distillate pumps Kubota
Diffuser GGG NiCr Nb 202 Discharge elbow Diffuser GGG NiCr no discharge elbow
SEZT 11001080
Diffuser discharge elbow GGG NiCr Nb 202
RDL 900 760 BVA 1800
Bottom-half erosion GGG NiCr Nb 202 Webs and pump cone
SEL - 1200
Discharge elbow
95 C2 Ex78
Brine blow down
Sea water supply pumps
Part Failing and Specification
Discharge elbow St.35.8 Diffuser impeller Column pipes Discharge elbow
Discharge elbow ASTMA 439 D2 Diffuser, upper lower column pipe 90C1 PPA95
Impeller ASTM A743CF3M lower column pipe stainless steel
32C1E x A 53
Impeller
Seawater supply S.W. Recirculating Brine recirculating
ASTMA436D2 column pipe discharge elbow Upper and lower casing NiResist Barrel, impeller 316L
Repair Done & Conclusion Belzona not good Hicoat Elastometric Welding and super metal metalock Belzona SMAW weld using (unsuitable material) Castoline x HD 2245 or Gricast 31 Belzona supermetal 1111 Welding (stitch) and bolt
Weld/corrocote Change to stainless steel (Weld with castoline x HD 2245 or Gricast 31) ASTMA 439 D2 cold repair in some cases Blasting and two coats of ARC 858 to 700 micron Cavitation machined and protection coating MIG welding using metrode ER 317L
Replace or belzona Chesterton ARC 858 Coating SMAW welding
DISTILLATE PUMPS Distillate pumps are very different in design and operation. They handle water moving from the distiller to the storage tanks through the ring main. Most of the components are made of stainless steel.
565
Corrosion Protection and Monitoring
Some of the pumps have been stifled and plenty of red deposits have clogged in. The red deposits are copper carried away from the tube bundles. Copper collecting on stainless steel could cause both galvanic corrosion and underside deposit pitting corrosion. Stainless steel, once affected, could continue to corrode. Repairing by welding of the stainless steel shaft, and machining and heat treatment are a matter of metallurgical skill. Cold repair is not recommendable for stainless steel (see Table 2). Could copper deposition be achieved at a different location before the distillate reaches the pump ? Could galvanic steel in a sump, be placed far away from the pump so that copper deposit could be filtered off or sumped out? These are some methods to be considered. CONCLUSIONS Although there are dozens of larger pumps and hundreds of smaller pumps in operation everyday, only four different types are taken into consideration. The others handle chemicals and other corrosive fluids, and corrosion failure is attributed to other reasons as well. In discussed pumps, the occurrence of failure is typically at a particular location irrespective of the material used. For example, in the seawater supply pumps, the most affected component is the impeller while in the brine pump, it is the diffuser and the delivery elbow that have repeated and acute failures. Cold repair and welding does not help in any way. It has been recommended to use stainless steel instead of GGG or Ni-resist. But the critical pitting potential of stainless steel at the temperature of brine is a matter to be considered. In the case of distillate and town water pumps, copper deposition is to be circumvented. In the case of seawater circulation pumps, only the lower casings have failed although the top and bottom are of the same material. The failure is possibly enhanced by casting defects and operational conditions. Cold repair is of no use. CP was successfully tried in most of the pumps.
566
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
IMPACT OF METALLIC CORROSION ON THE KUWAIT ECONOMY BEFORE AND AFTER THE IRAQI INVASION : A CASE STUDY F. Al-Matrouk1, A. Al-Hashem1, F.M. Al-Kharafi2 and M. El-Khafif 3 1
Kuwait Institute for Scientific Research P. O. Box 24885, 13109 Safat, Kuwait 2
Kuwait University P. O. Box 5469, 13060 Safat, Kuwait 3
Central Bank of Kuwait P. O. Box 526, 13006 Safat, Kuwait
ABSTRACT An estimation of the cost of metallic corrosion in the State of Kuwait was determined for the base years 1987, before the invasion of Kuwait, and 1992 after liberation of the country. The typical cost of corrosion was represented by the data for 1987, and that for the year 1992 was for an abnormal year since massive reconstruction and renovation of the basic industries in Kuwait were made. Information was obtained via questionnaires distributed to the major industrial sectors in Kuwait, official correspondence, and personal communication with key people in the different economic sectors of the country. For 1987, as the base year, the total cost of corrosion was estimated to be 5.2% of the gross national product (GNP). Approximately 17.4% of the total cost (0.9% GNP) was avoidable, i.e., amenable to reduction by presently available corrosion-control technology. The estimated value of resources such as materials, labor, energy and technical capabilities ascribed to corrosion-related activities in 1987 were Total cost to Kuwait Avoidable costs Unavoidable costs
KD 322 million KD 56 million KD 266 million
This study used a modified version of the input/output model of the Battle Columbus Laboratories (BCL) and the National Bureau of Standards (NBS) of the US. The model quantitatively identifies corrosion-related changes in resources, capital stock, and replacement lives of capital stock for all sectors of the economy. The use of this model is well suited for estimating the total direct and indirect costs of corrosion. Key Words: Metallic corrosion, Kuwait, total cost, avoidable cost, unavoidable cost, input/output model
INTRODUCTION In the 1970s, several countries began to study the economic consequences of corrosion. The findings of these major investigations indicated that the range of the estimated cost of corrosion as a percentage of the gross national product (GNP) was 1.5-5.6 [1]. The
567
Corrosion Management
investigations approached the problem from two different directions. One of the first, and probably one of the most important, studies published was the Hoar report [2] in the UK which was based on statistics gathered from various industries experiences.. A different approach was taken in the Bennett Report for the US [1], where the input-output analysis, as developed by Leontief [3], was used. Based on this model, the entire USA economy, new matrices, capital/output-matrix, capital life matrix, capital replacement matrix and new demand vectors (personnel consumption and government) were constructed for the economy in two worlds: World II (corrosion-free world), and World III (best practice world). The differences between these two worlds and the existing world (World I) yield the total costs of corrosion. The present study adopted modified US approach, as it assesses the economic effects of corrosion in a systematic framework. In addition, it estimates the direct and indirect cost of corrosion for every sector of the economy. Characteristics of Kuwait Economy Traditionally, Kuwait’s economy isheavily dependent on one sector as the main source of income. The hydrocarbon sector is the dominant natural resource, as is reflected in its contribution of more than half of the gross domestic product (GDP) in 1992, while the agricultural and nonoil manufacturing sectors accounted for less than 4% of the GDP in the same year. The public sector dominates the economic activities not only in terms of ownership and management, but also in the employment of Kuwaiti manpower i.e. more than 90% of employed Kuwaities are employed by government. The private sector contributes only 25% of GDP verss 75% for the public sector. Finally, the economy has an open policy in term of the flow of goods, capital and labor in and out of the country [4]. MODEL SPECIFICATIONS The assessment of the cost of corrosion requires a methodology capable of reflecting the production processes of the various sectors in the economy and the interrelationships between these sectors. One of the most suitable techniques is the input/output (I/O) analysis. This techniques is capable of estimating the direct cost of corrosion and also the indirect costs of inputs used to produce the inputs applied directly in the process of corrosion prevention and control [5] and [6]. The model used in the study is quite detailed, it consists of 14 economic sectors, each of which is represented by a production function consisting of the respective inputs from each sectors plus value added. As a result, relatively detailed industry corrosion cost data was incorporated into the model for simulation exercises. This paper uses two I/O tables for the State of Kuwait. The first is the 1987 tables produced by the Central Statistical Office (CSO) of the Ministry of Planning (MOP) [7]. The second set of tables reflects the economy in 1992 and is an update for the 1987 tables using the CSO's national accounts figures for 1992 [8]. The I/O tables of 1987 and 1992 have been reconciled and aggregated into 14 economic sectors. Structure of the I/O Model The I/O model relies on the fact that total sectoral output is divided between intermediate demand and final demand. It can be presented by the following equation: X = AX+F 578
(1)
Al-Matrouk et al.
where X = vector (14 x 1) of gross output A = technical coefficient matrix (14 x 14) F = vector (14 x 1) of final demand To include the impact of corrosion on investment within the analytical framework, investment has to be incorporated into the I/O inverse matrix and the basic model has to be modified. This modification assumes a fixed investment output ratio, and can be reformulated in the following form: X = (I - A - N)-1 FD
(2)
FD = PC + GC + E - M
(3)
and where N is a diagonal matrix representing the investment output ratio and FD is a stipulated final demand (final demand less gross investment), PC, i.e., private consumption, GIC, i.e, Government intermeridate consumption, E Gross investment,.and M export and import respectively. Equation 2 forms the modified I/O model which will be adjusted for World II and World III to estimate the total, avoidable and non-avoidable costs of corrosion at the sectoral level. This formulation assumes that the economy is growing at its long term growth rate. Model Assumptions The desirable model assumes that the economy operates at full employment or full capacity. The rational behind this is that it allows us to measure the full potential capacity of the economy, and thus, the full cost of corrosion. However, due to the special characteristic of the labor market in the State of Kuwait, and consequently the difficulty of obtaining a measure of full employment, this assumption was relaxed. Instead, the two I/O tables, i.e., for 1987 and 1992 were used in the exercise of estimating the cost of corrosion. The choice of these two years was based on the availability of the I/O tables, as well as the state of the economy during those years. In 1987, the economy was operating near full capacity, and therefore, it is possible to assume that it was a case of full employment. On the other hand, in 1992, one year after liberation, the economy was operating for below capacity. Therefore, the estimation of the cost of corrosion in these two year will provide us with two estimates; one for the economy when operating at or near full employment and the second when the economy is operating well below full capacity. The main assumptions are implicit in the developed model are:
• • • • •
Linearity of the production process or constant return to scale, Homogeneous product, Inelasticity of substitution, Steady growth, and Average technology.
579
Corrosion Management
In addition, the modified I/O tables contain two rows: one for corrosion’s social savings and one for value-added. These rows are below the "A" matrix, and therefore, do not directly affect the inverse of the matrix. Sectoral Aggregation of the I/O Model The original I/O tables for 1987 and 1992 disaggregate the economy into 34 and 32 sectors, respectively. The aggregation process took into consideration the homogeneity of the products in the aggregated sectors. The aggregation process was based on three conditions:
• The aggregated sectors are not significantly affected by corrosion, • The aggregated sectors are small in size relative to the economy as a whole, and • The outputs of all sectors aggregated in one new sector had some common characteristics and maintained a certain degree of homogeneity. Adjustment for Cost of Corrosion First, World I, reflects the real world as it existed in the year of the I/O table. World II is a hypothetical world that represents the economy as it would be in the complete absence of corrosion. World III is another hypothetical world in which the best economical known corrosion prevention practices known are used by all economic agents. The total cost of corrosion is obtained by subtracting the sectoral outputs of world I from the sectoral outputs produced from the simulation of the model in world II. The difference between the sectoral outputs of world I and the simulated outputs of world III represents the avoidable cost of corrosion. Consequently, the unavoidable cost of corrosion is the total cost minus the avoidable cost of corrosion. The I/O methodology produces the cost of corrosion, either avoidable or unavoidable, in terms of direct and indirect costs. Data and Assumptions The process of adjusting the I/O model for World II and World III requires data and information on the impact of corrosion in three specific areas:
• Inputs required to produce one Kuwaiti Dinars worth of output for each sector, • Replacement life of capital in each sector, or sectoral investment output ratio, and a • Stipulated final demand for the output of each sector. The methodology applied in the gathering of the data and information, required for the model's adjustment, depended on three sources:
• Data compiled from a survey specifically designed for industries in the State of Kuwait, • Judgment of experts in the field of metallic corrosion in Kuwait, and • Experience of other countries and previous studies. The first choice for data compilation was a questionnaire or the survey. However, if the information for a specific sector or industry was not available from the survey or was of poor quality, the analysis then depended on the second source, i.e., the judgment of the experts. In the instances where the first and second sources were not sufficient, the study made use of the third source to fill any gaps. This did not eliminate the possibility of using all three sources 580
Al-Matrouk et al.
together. On the contrary, in many cases, sectoral data were compiled from all three sources. The questionnaire was the main source of information four important sectors: petroleum refining; other chemical products; basic metal products; and agriculture, livestock and fishing. As for the remaining 10 sectors, the analysis relied on the judgment of the experts in metallic corrosion in the State of Kuwait, as well as, the literature, namely the US study [9] and the Egyptian study [10]. The elements of corrosion that can affect the model coefficients and parameters are
• • • • • •
Replacement of equipment and buildings, Excess capacity and redundant equipment, Loss of product, Maintenance and repair, Corrosion control measures, and Other cost elements
The effects of the remaining elements of cost of corrosion on the model can be represented by the input produced by the various sectors in the economy, and consumed, as intermediate demand, by other sectors. World II and World III World II is a world free of corrosion, while World III is a world where the best corrosion practice is applied universally. The comparison between WORLDII and World I, the world as it exists, provides the total cost of corrosion. On the other hand, the comparison between World III and World I indicates the avoidable cost of corrosion. The latter is this amenable to reduction by presently available corrosion protection and control technology. This does not imply a gold- plated world. It only implies that resources are used and allocated efficiently in the process of controlling corrosion. The categorization of total costs of corrosion into avoidable and unavoidable (i.e., the difference between total cost and avoidable cost) can provide guidance in the allocation of priorities to research and development, legislative, and/or educational efforts. Avoidable cost is influenced by the transfer of technology. On the other hand, unavoidable cost is not influenced by the presently known technology, but requires technological progress for further reduction in costs to be realized. Adjustment for World III After adjusting the "A" matrix to reflect the total cost of corrosion in World II, the avoidable sector costs of corrosion were estimated as percentages of the total costs. The criterion for this ratings was a qualitative comparison of producing sectors as to their incentive or need to apply the best corrosion practices available and their flexibility in applying them (i.e.,corrosion practice). The ratings range between 0.5 and 3.0. The percentage of avoidable cost out of the total costs ranges from 5% for the sectors with the low rating of 0.5, to 35%, for sectors with the high rating of 3.0. A low rating indicates that the sector is not very sensitive to corrosion, and therefore, avoidable cost as a percentage of the total cost of corrosion will be relatively small.
Model's Modifications 581
Corrosion Management
Mdifications were made for the two I/O tables, i.e., 1987 and 1992. The corrosion saving parameters used in the construction of the World II and World III models, were more relevant to the 1987 I/O table. Therefore, these parameters were used directly in the development of World II and World III of the 1987 table. As for the 1992 model, it is assumed that the Kuwaiti economy employed more efficient capital after the liberation, and consequently, it is likely that the economy's corrosion-saving parameters are higher than their levels before the invasion. Since the information available does not warrant reasonable estimates of these parameters for the period following the liberation, the analysis considered two scenarios for the I/O model in 1992. The first scenario assumed a 10% increase in the corrosion-saving parameters over the levels used in the 1987 model, and the second scenario assumed a 20% increase. In addition, Kuwait’s I/O tables aggregate the economy's investment in four sectors: other manufactures, construction, transportation and storage, and other commercial services. To account for the impact of corrosion on investment, and to calculate the investment-saving parameters, the analysis considered the three elements that could affect investment: replacement of equipment and buildings, excess capacity, and loss of product. COST OF CORROSION IN THE STATE OF KUWAIT The final results of the study with the model modified to represent World II and World III, allow the quantification of the savings that can be achieved if corrosion did not exist (i.e. the total cost of corrosion), as well as the savings that can be achieved if the best known corrosion control practices were applied in every sector of the economy. These savings are estimated on two levels: the macro level, which shows the cost to the economy as a whole, and the micro-level which illustrate the costs to the various sectors in the economy. Estimation of Total Costs of Corrosion The total cost of corrosion can be estimated by comparing the GDP of the economy as it exists with the GDP of a hypothetical economy where corrosion does not exist. Table 1 summarizes the results of estimated total, avoidable and unavoidable costs of corrosion in 1987 and 1992. The significance of the year 1987 is that the economy was operating near its full capacity. On the other hand, the analysis of the year 1992 assumes that the economy employed new, efficient capital after the liberation, and therefore, is more efficient with respect to corrosion, which implies a reduction in the savings related to the cost of corrosion. The analysis considers two scenarios for the 1992 model: the first scenario which assumes a 10% improvement in the corrosion-saving parameters over their levels in 1987; and a resultant reduction in the total cost of corrosion, and the second scenario which considers a 20% improvement in the parameters. The main results are summarized as follows: 1. The total cost of corrosion in 1987 amounted to KD 322 million, representing 5.2% of the GDP. The avoidable cost was about KD 56 million, representing 0.9% of the GDP. This is the cost that could have been saved if the economy had applied the best corrosion practices available. The unavoidable cost of corrosion is estimated to have been KD 266 million or 4.3% of GDP. 2. The costs of corrosion in 1992 were lower than those estimated for 1987. There are three factors behind this decrease: the economy in 1992 was smaller than in 1987, the 582
Al-Matrouk et al.
1992 model assumed a more efficient economy with respect to the corrosion-saving parameters, and therefore a lower cost of corrosion per unit of output’ and the economy in 1992 operated below capacity, which means that only the more efficient equipment was used in production processes. 3. The first scenario in 1992 suggests that the total cost of corrosion was KD 160 million, or 2.92% of the 1992 GDP. The cost that could have been saved if the best corrosion practices had been applied, amounted to KD 40 million, representing 0.73% of the GDP. 4. The second scenario, which assumed a 20% improvement in the corrosion-saving parameters over their levels in 1987, had total and avoidable costs of corrosion estimated at KD 140 million and KD 38 million respectively, representing corresponding percentages of the GDP of 2.56% and 0.69%. Table 1. Total Avoidable and Unavoidable Costs of Corrosion 1987 Total Cost 321,579 TKD % of GDP 5.21% Avoidable Costs 55,929 TKD % of GDP 0.91% Unavoidable Costs 265,650 TKD % of GDP 4.3% TKD = thousand Kuwaiti dinars
1992 Scenario I 160,033 2.92% 39,991 0.73% 120,042 2.19%
1992 Scenario II 140,303 2.56% 37,945 0.69% 120,358 1.87%
Estimation Costs of Corrosion at the Sectoral Level The model allowed estimations for all elements of the cost of corrosion (i.e., total, avoidable and unavoidable costs) at the sectoral level, and disaggregates it into direct and indirect costs. To identify and quantify the relative measures of the impact of corrosion on the various sectors in the economy, the results obtained from the simulation of the I/O model were used to develop industry indicators. There are eight main indicators that can be used to evaluate the relative position of each sector in the economy: 1. 2. 3. 4. 5. 6. 7. 8.
Total direct cost in thousands of Kuwaiti Dinars, Total direct cost per one KD.1 of value added, Total direct and indirect cost in thousands Kuwaiti Dinars, Total direct and indirect cost per KD.1 of value added, Avoidable direct cost in thousands of Kuwati Dinars, Avoidable direct cost per KD.1 of value added, Avoidable direct and indirect cost in thousands of Kuwaiti Dinars, and Avoidable direct and indirect cost per KD.1 of value added.
Each of these indicators is significant and plays an important role in comparing the various sectors. Tables 2, 3 and 4 list the eight indicators for the 14 sectors in the economy in 1987, 1992 scenario I, and 1992 scenario II. The main results can be summarized as follows: 1. The oil sectors, petroleum refining, and crude petroleum, are the most important sectors in the Kuwaiti economy. In 1987, while these two sectors represented 47% of
583
Corrosion Management
2.
3.
4. 5.
6.
7.
8.
9.
584
the economy, they were responsible for only 6% of the total cost of corrosion (both direct and indirect) estimated for that year. The total cost of corrosion in the oil sectors amounted to KD 19.2 million, of which KD 3.1 million could be saved if the best corrosion control practices available were applied. In terms of the cost permit of value added, the total and avoidable costs of corrosion represented 2.2% and 0.4% of the value added to the oil sectors, respectively. In other words, 22 fils and 4 fils for every KD 1 of output. In 1992, the oil sector constituede 38% of the country’s GDP. The first scenario for 1992, suggests that the total cost of corrosion in the oil sector is estimated to be KD 13.8 million, or 8.6% of the total cost of corrosion in the economy, of which KD 2.7 million could be considered to be avoidable or reducible. This latter amountaccounts for 6.7% of the sum of the avoidable costs in the economy. For every one KD.1 of output in the oil sectors, 26 fils are attributed to corrosion of which 5 fils can be saved by employing the best corrosion control practices available. The comparison between the total cost in scenario I and II for 1992 indicates that with a 10% improvement in the corrosion-saving (i.e, cost) parameters, the oil sectors alone could save up to KD 1.7 million in corrosion-related expenses. Among the 14 sectors, the other commercial service sector (sector 13) and the other government, social and household service sector (sector 14) bear the largest portion of the total corrosion cost. In 1987, these two sectors contributed more than 70% and 62% of the total and avoidable costs in the economy. The total cost and the avoidable cost for the other commercial service sector amounted to KD 118.2 million and KD 20.7 million respectively. The costs of corrosion to the other government, and social and household services sectors are estimated to be KD 106.9 million and KD 14.0 million respectively. The 1992 simulations show similar but lower results. The difference between the first and second scenario indicates that a 10% improvement in the corrosion-saving (i.e.,cost) parameters could result in KD 5.8 million and KD 8.0 million reductions in the total cost of corrosion for the commercial and government sectors, respectively. As for the cost per unit of value added, the cost indicators for these two sectors appears to be below average. These results suggest that the potential savings in these two sectors are high, but the effort required to achieve them could be above average. Some of the factors behind the high total cost of corrosion in the two sectors could be the diversity of the sectors, in addition to the possibility that the participants in these sectors do not consider corrosion to be a problem or are not aware of corrosion as a cost. The sectors with the highest cost per KD.1 of value added, are as follows: nonmetallic products (sector 6) basic metal products (sector 7) construction (sector 10) and other manufacturers (sector 8). The total cost of corrosion for all of these sectors together was 15% of the economy’s total cost in 1987, and 25% in 1992 (scenario I). The avoidable cost as a percentage of the economy’s total avoidable cost was estimated to be 23% in 1987 and 41% in 1992 (scenario I).
Al-Matrouk et al.
Table 2. Cost of Corrosion at the Sectoral Level Sector 1.Agriculture, Livestock & Fishing 2.Crude Petroleum, N.Gas & Other Mining 3.Food, Beverages & Tabacco 4.Petroleum Refining 5.Other Chemical, Products (Except Pet. Ref.) 6.Non-Metal Products 7.Basic Metal Products 8.Other Manufactures 9.Electricity, Gas & Water 10.Construction 11.Hotels & Resturantes 12.Transport & Storage 13.Other Commercial Services 14.Other Govt., Social & Household Ser. Average
Total Cost Direct Savings Dirrect & Indirect Savings 435 22080 730 7340 1883 473 165 6650 2415 3915 1570 5904 27196 21070 101827
0.0150 0.0096 0.0167 0.0130 0.0504 0.0136 0.0381 0.0569 -0.0794 0.0250 0.0276 0.0321 0.0198 0.0163 0.0165
1761 8786 3585 10480 3252 8292 673 6452 -6125 34394 7258 17627 118203 106940 321579
0.0605 0.0038 0.0820 0.0186 0.0871 0.2384 0.1558 0.0552 0.2014 0.2196 0.1276 0.0958 0.0861 0.0827 0.0521
Avoidable (Reducible Direct Savings Direc S 22 6624 73 2202 377 71 49 665 845 587 78 886 2720 2107 17306
0.0008 0.0029 0.0078 0.0030 0.0101 0.0020 0.0114 0.0057 -0.0278 0.0038 0.0014 0.0048 0.0020 0.0016 0.0028
585
18 142 34 167 63 208 36 113 -119 930 110 422 20665 1399 5592
Corrosion Management
Table 3. Cost of Corrosion at the Sectoral Level - 1992 Scenario I Sector 1.Agriculture, Livestock & Fishing 2.Crude Petroleum, N.Gas & Other Mining 3.Food, Beverages & Tabacco 4.Petroleum Refining 5.OTHER CHEMICAL, PRODUCTS (Except Pet. Ref.) 6.Non-Metal Products 7.Basic Metal Products 8.Other Manufactures 9.Electricity, Gas & Water 10.Construction 11.Hotels & Resturantes 12.Transport & Storage 13.Other Commercial Services 14.Other Govt., Social & Household Ser. Average
586
Total Costs Direct Savings Direct & Indirect Savings
Avoidable (Reducible Direct Savings Direc S
320 15449 445 3451 1331 259 171 6485 5424 4611 1068 5177 20134 24630 88954
16 4635 44 1035 266 39 51 648 1898 692 53 777 2013 2463 14632
0.0135 0.0086 0.0150 0.0117 0.0454 0.0122 0.0343 0.0512 -0.0715 0.0225 0.0248 0.0289 0.0178 0.0147 0.0162
1016 7335 2145 6428 3458 4227 749 6179 1178 28228 2969 9684 56641 29794 160033
0.0429 0.0041 0.0725 0.0218 0.1179 0.2001 0.1499 0.0488 -0.0155 0.1377 0.0691 0.0540 0.0501 0.0177 0.0292
0.0007 0.0026 0.0015 0.0035 0.0091 0.0081 0.00103 0.0051 -0.0250 0.0034 0.0012 0.0043 0.0018 0.0015 0.0027
11 154 20 113 94 106 65 607 99 890 50 265 940 580 3999
Al-Matrouk et al.
Table 4. Cost of Corrosion at the Sectoral Level Sector 1.Agriculture, Livestock & Fishing 2.Crude Petroleum, N.Gas & Other Mining 3.Food, Beverages & Tabacco 4.Petroleum Refining 5.OTHER CHEMICAL, PRODUCTS (Except Pet. Ref.) 6.Non-Metal Products 7.Basic Metal Products 8.Other Manufactures 9.Electricity, Gas & Water 10.Construction 11.Hotels & Resturantes 12.Transport & Storage 13.Other Commercial Services 14.Other Govt., Social & Household Ser. Average
Total Costs Direct Savings Direct & Indirect Savings
Avoidable (Reducible Direct Savings Direc S
285 13732 395 3068 1183 230 152 5764 4821 4098 949 4602 17897 21893 79070
14 4120 40 920 237 34 46 576 1687 615 47 690 1790 2080 12896
0.0120 0.0076 0.0134 0.0104 0.0403 0.0109 0.0305 0.0455 -0.0635 0.0200 0.0221 0.0257 0.0158 0.0130 0.0144
905 6383 1907 5724 3001 3796 716 6674 1053 26163 2641 8797 50794 21749 140303
0.0382 0.0036 0.0645 0.0194 0.1023 0.1798 0.1433 0.0527 -0.0139 0.1277 0.0614 0.0491 0.0450 0.0130 0.0256
0.006 0.0023 0.0013 0.0031 0.0081 0.0016 0.0091 0.0046 -0.0222 0.0030 0.0011 0.0039 0.0016 0.0012 0.0024
SUMMARY AND CONCLUSIONS
587
10 142 18 104 93 96 59 619 87 861 45 255 873 527 3794
Corrosion Management
The objective of this paper was to quantify the costs and economic effects of corrosion in the State of Kuwait, and to identify the sectors which suffer the most from corrosion. The results indicate that corrosion has a significant impact on the economy, and that considerable savings can be realized through the application of more advanced and new corrosion control techniques. The main results and conclusions are 1. When the economy operates at or near full capacity, corrosion can cost the State of Kuwait more than 5.2% of its GDP ( i.e.,KD 322 million in 1987). Given the present technology, there is room for corrosion-related savings (i.e.,avoidable costs) of up to KD 56 million, or about 1% of the GDP, by applying the best known corrosion control practices. Furthermore, a 10% improvement in the corrosion-savings (i.e.,cost) parameters, could save the economy up to KD 20 million, or 0.4% of GDP. 2. On the sectoral level, the estimates for the total cost of corrosion in the oil sector (i.e.,crude petroleum and petroleum refining) was KD 19.2 million in 1987, accounting for 6% of the economy’s total corrosion cost, and KD 13.8 million in the first scenario for 1992, representing 8.6% of the economy’s total corrosion cost. The avoidable cost, or the saving that could be achieved by applying the best corrosion control practices available in these sectors ranges from KD 3.1 million in 1987 to KD 2.7 million in the first scenario for 1992. 3. The commercial services sector, and the government, social and household services sector are responsible for the largest share of the cost of corrosion in the economy. These service sectors accounted for 70% (KD 118.2 million, of which KD 20.7 million was avoidable) of the economy’s total cost for corrosion in 1987, and 62% in the first scenario for 1992 (KD 106.9 million of which KD 14.0 million was avoidable). The potentials for savings in the service sectors are high but they are not easy to achieve. 4. The results also show that the nonmetallic products sector, basic metal products sector, construction sector, and other manufacturers sector have the highest costs per KD.1 of value added. This suggests that corrosion control efforts would be more costeffective in these sectors than in the rest of the economy. 5. Finally, it should be mentioned that the estimates represented in this study should not be considered as absolutes, but rather as being the best available indicators of the economic effect of corrosion. They provide a benchmark against which the relative impact of other factors affecting the economy can be compared and assessed. This can help in the development of a program for corrosion control which prioritize the actions needed and the resource reallocation required. GENERAL RECOMMENDATIONS Due to the great economic losses incurred by corrosion to the nation’s economy, the following recommendations are suggested to aid in reducing such cost; 1. All major industries in Kuwait should have at least one experienced corrosion engineers with knowledge in corrosion protection and monitoring. These corrosion engineers in staff should also assist in material selection during the design, commissioning and operation of plants. 2. Institutional links should be developed between research institutions, such as the Kuwait Institute for the Scientific Research and Kuwait University,and the major 588
Al-Matrouk et al.
3.
4.
5.
6.
industries in Kuwait to provide consultations and long-term solutions to common and more frequent corrosion problems. If knowledge on some corrosion problems is lacking within the country, then international consultants should be contacted to assist in the diagnosis and remediation of such corrosion problems. Short training courses on the subject of corrosion and failure investigation of real industrial components should be held annually at the research and academic institutions of the country for the concerned employees in plants and factories. Research and academic institutions should focus on applied research and development in the area of corrosion and corrosion protection, research which reflects the actual needs of the key industries in Kuwait. This can be achieved by strengthening the institutional links between the institutions and the major industries in the country. Standard methods and specifications should always be followed by the maintenance group of the major industrial plants in the country when applying the different corrosion protection methods. A campaign to increase the awareness of corrosion losses and how such losses can be reduced should be directed towards the commercial services sector, and the government, social and household services sectors since they are responsible for the largest share of the cost of corrosion in the economy.
REFERENCES 1. L.H. Bennett ed., Economic Effects of Metallic Corrosion in the United States, US Department of Commerce, US Government Printing Office, Washington, DC, USA, 1978. 2. T.P. Hoar, Report of the Committee on Corrosion and Protection, Department of Trade and Industry, London, U.K, 1971. 3. Leontief, Input-Output Economics , Oxford University Press, New York, 1986. 4. The World Bank, Kuwait Privatization Strategy, Vol. III, Macroeconomic Environment, Labor Market and Legal Framework, Report submitted to the Ministry of Finance, Kuwait, Confidential Document, 1993. 5. V. Blumer-Thomas, Input-Out Analysis in Developing Countries,: John Wiley and Sons Inc, New York, 1982. 6. R.E. Miller and P.D. Blair, Input-Output Analysis: Foundations and Extensions prentice Hall Inc., Englewood Cliffs, New Jersey, U.S.A. 1985. 7. National Accounts and Input-Output Tables, Central Statistical Office - Ministry of Planning, State of Kuwait, 1987. 8. F. Al-Matrouk and J. Haieih, Economic Guide for the Selection of Industrial Project on Sectoral Level in the State of Kuwait, Kuwait Institute for Scientific Research, KISR Report, August 1995. 9. National Bureau of Standards, Economic Effects of Metallic Corrosion in U.S.A, Special Publications 511, 1978 - 1, 2.. 1.Egyptian Academy of Scientific Research Technology, Economic Effects of Metallic Corrosion in Egypt, 1983.
589
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CORROSION PROBLEMS IN A STEAM CONDENSATE SYSTEM AND TREATMENT OF CONDENSATE FOR RECOVERY G.L. Rajani Engineering & Corrosion Services CII-3, Ansari Nagar, New Delhi 110 029, India
ABSTRACT In refinery, petrochemical, fertilizer, chemical and process industries; power plants and metallurgical plants, control of corrosion damage by steam from medium and high pressure boilers is of paramount importance to prevent corrosion leaks in condenser, condensate and steam lines. The presence of parts per billion levels of aggressive impurities like O2, CO2 and Na may lead to catastrophic corrosion failures in vacuum condenser, condensate and steam lines, and may ultimately contaminate the condensate with ionic impurities and corrosion products. This results in wastage of condensate/steam and energy, and costly condensate polishing operations. In this age of energy conservation consciousness, steam and condensate lines are required to be perfectly protected by controlling the quality of steam so that condensate with its sensible heat is returned to the boiler to the maximum extent. Sources of corrosives are the boiler feedwater, ingress of gases through flanges and atemperator in the post-boiler section and condenser leakage. Erosion-corrosion, cavitation and impingement are other common causes of leakages in steam condensate circuitry. Steam purity is primarily maintained through boiler feedwater (BFW) conditioning, e.g., effective deaeration of BFW, and chemical treatments through the use of oxygen scavengers and volatile amines and effective monitoring of steam-condensate quality. In other instances, along with boiler water conditioning, neutralizer and filming amines are injected into steam in the post-boiler section. Varieties of volatile amine and filming amine combinations are available to prevent corrosion in steam-condensate systems. Steam used for heating in process equipment in the hydrocarbon industry becomes contaminated with oil and corrosion products. This contamination and leaks in the vacuum condenser require that the unit be polished to remove various contaminants before the condensate is returned to the deaerator. Polishing is carried out with the help of activated carbon, a microfilter and ion-exchange. Control of the sodium carryover in the steam prevents sodium-induced cracking problems in the welds of a steam-condensate circuit. The selection of carbon steel piping and fitting the schedule to limit the velocity of steam/condensate, as well as the provision of effective steam traps are the engineering measures required to prevent erosion-corrosion, cavitation and impingement. This paper describes, by means of a case history, the causes of corrosion damage in steamcondensate lines, and its prevention through chemical treatment, monitoring and design control, condensate return and condensate polishing. Key Words. Steam condensate, boiler, aggressive impurities, corrosion failure, chemical treatment
INTRODUCTION 581
Corrosion Management
Steam is primarily used in industries for heating process fluids, for catalytic reforming and for distillation operations, beside being used for the generation of power. Steam condensate is either discarded or recovered and reused. Turbine condensate is generally clean, and is recovered completely and reused. The recovery of process condensate for reuse depends on various factors such as process contamination, economics of recovery and availability of spent condensate. In some processes like reformer, delayed coker unit fractionator steam is consumed, and hence, it is not recoverable. With the increasing emphasis on energy conservation, industries attempt to maximize recovery of as much condensate as possible along with its sensible heat for reuse in steam generation. The condensate contains dissolved solid impurities and dissolved gases. The presence of oxygen and carbon dioxide causes corrosion in the steam-condensate system. Sodium in excess of 20 ppb in steam has been found to attack welds at bends and elbows in steam lines. In the plant, this post-boiler corrosion leads to cost penalties by requiring repairs to be made to the steam-condensate system, with resultant losses of production and energy (through steam leaks and deposition). A 1.5 mm diameter hole in a steam line causes 30 kg/hour and 65 kg/hour steam loss in 28 kg/cm2 and 60 kg/cm2 boilers, respectively, resulting in corresponding losses of Rs. 1.0 lakh and Rs. 2.50 lakhs per annum. The losses due to steam leaks of various sizes are given in Table 1. Table 1. Steam Leaks versus Cost Pressure (kg/cm2)
1.5
3
7 14 28 60
8 15 30 65
33 62 119 248
Diameter of Leak ( mm) 6 12 Steam Wasted (Kg/hr) 132 247 476 -
530 987 -
25
-2110 -
Cost in Rupees Lakh per anum 7 0.29 1.18 14 0.54 2.23 28 1.08 4.28 60 2.34 8.93 Note : Cost of steam (assumed) US $ 1 KD (Kuwaiti Dinnar) 1 KD (Kuwaiti Dinnar) 1000
4.75 8.90 17.14 Rs. 450/MT (12.50US$/MT) Rs. 36.0 Rs. 100.0 (≈ 110) Rs. 1.0 Lakh
19.08 35.53 -
76.0 -
In post-boiler corrosion, the return of corrosion products along with the condensate can result in the deposition of metal oxides on boiler heat-transfer surfaces which then bind the boiler water slugs. These deposits have poor thermal conductivity, which leads to energy 582
Rajani
losses. Several chemical treatments are used to control oxygen and carbon dioxide corrosion. A combination of chemicals offers the best protection. Controlling the sodium level in steam prevents sodium-induced corrosion problems in steam-condensate circuitry [1]. When condensate is returned to the deaerator, the sensible heat in the condensate increases the temperature of the feedwater which results in savings of energy and a lower makeup water requirement. Because of impurities and contamination, the condensate requires proper treatment-cumpolishing before being returned for reuse. Polishing units are operated when contaminants and impurities exceed acceptable limits. TYPES OF CORROSION The types of corrosion generally encountered in steam condensate circuitry are [2]
• • • •
• Oxygen attack Carbon dioxide attack, Erosion-corrosion, Cavitation, and Sodium attack on welds. Corrosion due to Dissolved Oxygen (O2) Boiler-quality carbon steel material is generally used in boiler systems. A thin protective coating of iron oxide (magnetite Fe3O4) forms at boiler temperatures, which prevents corrosion. However, dissolved oxygen present in the system even in traces, can destroy the protective coating resulting in rapid pitting-type corrosion. Oxygen pitting is often accompanied by oxide deposits (rust hematite Fe2O3) or blisters near the pitted area. Oxygen is a strong depolarizer of corrosion cells. This electrochemical reaction is quick because of the high temperature in the boiler. Very often corrosion products are transferred to other areas, including the boiler drum, where they cause insulating deposition. Figure 1 shows typical pitting-type corrosion in a pressured hot-water system due to dissolved oxygen. Apart from the feedwater, oxygen can enter the system through atmospheric leakage at flanged piping and other sections under vacuum (such as the condenser which is kept at a partial vacuum). Another source of oxygen is the leakage of oxygen-laden water through condensate pump seals and surface condensers by the direct injection of quench water to cool the condensate to prevent pump cavitation and the direct injection of boiler feedwater into the atemperator. Corrosion due to Dissolved Carbon Dioxide (CO2) Carbon dioxide dissolves in condensate to form carbonic acid which is corrosive. Carbonic acid corrosion is often seen as deep channelling, grooving or thinning at the bottoms of pipes, and as penetration in threaded areas. It may also appear as uniform corrosion in condensate lines where the flow is continuous. Figure 2 shows a condensate line before and after CO2 control treatment.
583
Corrosion Management
Figure 1. Pitting-type corrosion cells in mild steel piping caused by dissolved oxygen
Figure 2. Condensate tube In addition to the ingress of air into the steam-condensate system through the vacuum condenser and pump packing glands, carbon dioxide results from the breakdown of carbonate and bicarbonate alkalinity in feedwater under boiler conditions, as can be seen in the following reactions, especially in low and medium pressure boilers. 2 NaHCO3 + heat = Na2 CO3 + CO2 + H2O Na2 CO3 + H2O + Heat = 2 NaOH + CO2
(1) (2)
CO2 + H2O = H2CO3
(3)
H2CO3 = H+ + HCO3 (causes low-pH condensate)
(4)
In most cases, the bulk of the carbon dioxide comes from the breakdown of feedwater alkalinity (Eqs. 1-4). However, CO2 can also come from process contamination of the condensate or decomposition of certain organic compounds. In the steam and condensate system, the combination of oxygen and carbon dioxide may be devastating. Corrosion proceeds more rapidly in such a situation, and the final reaction releases CO2 which makes the process self propagating:
584
Fe + 2H2CO3 = Fe ( HCO3 )2 + H2
(5)
4Fe (HCO3)2 + O2 = Fe2O3 + 4 H2O + CO2
(6)
Rajani
Erosion-Corrosion Erosion-corrosion occurs in the steam condensate lines due to local turbulence such as in elbows and tees. This type of corrosion can also occur if the steam trap is defective and unseparated condensate mists enter into the steam. Erosion-corrosion can be prevented by preventing carryover of solids in steam and by installing correct steam traps. Cavitation When condensate flashes from a low diameter pipe into a high diameter pipe wall, cavitation can occur. It can also occur on condensate pump impellers. Cavitation damage can be identified by a honeycomb-type of pits with smooth rounded edges. Cavitation can be prevented by installing impingement plates and selecting resistant materials for pump impellers. Sodium Attack When the sodium content in steam is increased, welds at elbows and tees are attacked. The attack is from surface cracks to deep cracks. The only way to prevent sodium attack is to use high-purity water and all volatile boiler water conditioning chemicals.
CHEMICAL TREATMENT TO CONTROL CORROSION IN STEAM CONDENSATE SYSTEM After boiler corrosion control through the use of various chemicals is well established, four types of treatment can be implemented: oxygen scavengers, neutralizing inhibitors, filming inhibitors and combinations of these. Oxygen Scavengers Sodium sulphite is used as an oxygen scavenger for low and medium pressure boilers. Sodium sulphite reacts with oxygen to form sodium sulphate as shown below: Na2SO3 + 1/2 O2 = Na2SO4
(7)
If, rust is formed, sodium sulphite will react with the rust to form a protective magnetite film 3Fe2O3 + Na2 SO3 = 2Fe3 O4 + Na2SO4
(8)
Each part per million of dissolved oxygen requires approximately 8 ppm of sulphite. The addition of a catalyst such as cobalt will enhance the reaction of sulphite with oxygen. Figure 3 shows the rate of the reaction of uncatalyzed and catalyzed sulphite. Hydrazine (N2H4) is the most popular and effective oxygen scavenger used conventionally in the steam and condensate lines of high pressure boilers. It protects condensate lines from oxygen pitting. The reaction with oxygen is as follows:
585
Corrosion Management
N2 H4 + O2 = N2 + 2H2O
(9)
Neither the reactants nor the products are solids. Since Hydrazine adds no solids, it is quite attractive for use in steam lines. In addition to being an excellent oxygen scavenger, hydrazine also passivates active iron and copper surfaces, rendering them, less susceptible to corrosion. N2 H4 + 6Fe2 O3 = 4Fe3O4 + N2 + 2H2O Reactive oxide film Protective oxide film N2 H4 + 4 CuO = 2 Cu2O + N2 + 2H2O
(10a)
(10b)
The addition of a catalyst to the hydrazine mixture ensures completion of the oxygen scavenging and metal-passivating reactions (Fig. 4). Organic Volatile Oxygen Scavengers These are non-hydrazine, non-sulphite organic compounds that effectively remove oxygen in boiler water system’s condensate circuitry. They react in a volatile manner with steam, and thus effectively protect the boiler drum, steam and condensate lines from oxygen corrosion. The rate of oxygen scavenging is the same or faster than with catalyzed hydrazine. The functions of this class of compound are similar to hydrazine. These compounds have been introduced because hydrazine has been proven to be highly carcinogenic and is banned in advanced countries. Organic volatile oxygen scavengers are mainly proprietary chemicals and are manufactured by reputable water treatment chemical companies. Neutralizing Type Inhibitors The neutralizing type of amines react with the carbonic acid in the condensate to form neutral amine salts, thus raising the pH. The most popular of such amines are morpholine, cyclohexylamine and diethylaminoethanol. Although not an amine, ammonia can also be used.
586
Rajani
Figure 3. Reaction rates for catalyzed and uncatalyzed sodium sulphite for a water temperature of 200C
Figure 4. Reaction rates for catalyzed and uncatalyzed hydrazine for a water temperature of 200C The distribution ratio (DR) of an amine is the most important factor and is expressed as DR = Amine in vapor phase (steam) Amine in water phase (condensate)
(11)
Amines with a DR > 1.0 have more amine in the vapour phase than in the water phase, whereas amines with a DR < 1.0 have more amine in the water phase than in the vapor phase. The degree of protection offered by these amines varies depending on the DR.
587
Corrosion Management
Table 2 shows the DR of ammonia and other commonly used neutralizing amines. These amines are dosed at the outlet of the deaerator to maintain the pH of condensate at 8.5 (minimum) to prevent corrosion due to CO2. Neutralizing-type amines can reduce concentrations of iron and copper that return to the boiler in condensate. R - NH2 + H2 CO3 = R-NH3 + HCO3
(12)
Table 2. Distribution Ratio for Neutralizing Amines Sr. No. 1. 2. 3. 4. 5.
Neutralizing Amine Morpholine Diethylaminoethanol Diethylisopropanolamine Cyclohexylamine Ammonia
Typical Distribution Ratio (DR) 0.4 : 1 1.7 : 1 1.7 : 1 4:1 10 : 1
Most Applicable Condensate System Short Medium Medium Long or branched Long or branched
DR = amine in vapor phase (steam) amine in water phase (condensate) Filming Type Inhibitor The filming amines form a barrier film on the condensate lines that prevents corrosion from reaching the metal’s surface. This provides protection from corrosion caused by oxygen and by low pH. Actually, they are not independent of pH; severe swings into the highly acidic or alkaline ranges can strip the film from condensate lines. Two popular filming amines are octadecylamine (ODA) and ethoxylated soya amine (ESA) which offer protection against both oxygen and carbon dioxide corrosion. They form a mono-molecular layer on the surface and protect the system against corrosion. The use of ODA is an older technology and has some drawbacks, which include: the inability to effectively form a film over existing pits, relatively poor distribution in long or branched systems, and action as a boiler-sludge binder if overcycled or overfed. ESA minimizes these pitfalls. Also, it is compatible with other boiler water treatment compounds and is more soluble in water, making it easier to feed to the system. If a turbine is there, filming amine is fed downstream of the turbine. COMBINATION TREATMENTS Neutralizing-amine blends that have multiple volatilities such as morpholine and cyclohexylamine combinations have been used to provide low-pH protection throughout the entire post-boiler system. Adding catalyzed hydrazine to such a neutralizing combination provides protection against oxygen pitting as well. Neutralizing amines and ESA form an excellent combination to protect low-pH and oxygen-caused corrosion. Catalyzed hydrazine and ESA provide protection against oxygen pitting and acidic corrosion. The combination of catalyzed hydrazine, neutralizing amines, and ESA can offer the optimum post-boiler treatment by minimizing both forms of corrosion. Such a multicomponent system affords 588
Rajani
built-in backup protection, should there be condensate contamination or a system upset. The optimum blend is also directly injected into the post-boiler section to prevent corrosion of the steam-condensate circuitry. Proprietary formulations such as ELOGUARD are available. ELOGUARD is multicomponent formulation which forms filming layer and conditions for oxide film and offers protection to the entire system pre-boiler, it is nontoxic and provides protection over a wide pH range. ELOGUARD is a polyamine formulation that has the following components:
• Film forming polyamine, • Neutralizing polyamine, and • Dispersing polymer. This replaces the use of oxygen scavenger, neutralizing amine and phosphate. A few industries have used it and found it to be effective in low and medium pressure boilers. However, its performance in high pressure boilers has yet to be established. The treatment chemicals are injected into the deaerator feed drum under suction from the Boiler feed water pumps (Fig. 5). TREATMENT OF PROCESS AND TURBINE CONDENSATE FOR RECOVERY AND REUSE The main types of condensate are turbine condensate and process condensate. The turbine condensate is from the turbines of compressors and turbogenerators, whereas process condensate is generated by the condensate of steam in various heat exchangers and reboilers. It consists of LP condensate and MP condensate. There is a possibility of hydrocarbon leakage from the process side to the steam side in process heat exchangers. Turbine condensate can pickup corrosion products, if the specified pH is not maintained. There is also a possibility of ingress of dissolved solids when the vacuum condenser leaks. In the case of refineries and petrochemicals, process condensate is contaminated with oil and other impurities. Steam is consumed in some processes and hence is not recoverable. In some industries, steam is used for tracing pipelines distributed over extensive areas (such as on tank farms and in refineries). In these cases, it is not economical to recover such condensate. Condensate from catalytic reforming contains huge amounts of methanol and is discarded. In view of the considerable amount of sensible heat present in condensate, it is necessary to recover as much condensate as possible and return it to the boiler after treatment. Return of condensate will result in a reduction in the makeup water required. Hence, the return of condensate will reduce the treatment cost as well as save energy. Depending on the feed impurities present in the condensate, treatment of condensate is required before its reuse in high pressure boilers. Treatment of Process Condensate Tables 3 and 4 show typical analyses of contaminated condensate from a petrochemical industry. The contaminated condensate contains hydrocarbons as well as other impurities such as ammonia, dissolved mineral impurities and heavy metals. Oil can be removed from process condensate by two method. One method is the use of coalescers that incorporate oleophilic resins which help to coagulate oil droplets into larger globules. Coalescers produce treated condensate with < 1 mg/1 of oil. The other method is the use of a special 589
Corrosion Management
grade of granular activated carbon (AC). Primary and secondary AC units are employed. The primary AC unit
590
Rajani
Figure 5. Schematic arrangement for chemical injection 591
Corrosion Management
removes oil by a process of straining, and the secondary AC unit removes the residual oil by a process of adsorption. After a period of use (once a year), the flow is modified such that the primary unit becomes the secondary unit and vice versa. The AC in the secondary unit is replaced with fresh AC, and the primary AC unit now retains its charge of old AC. Table 3. Process Condensate Impurities (Typical) Parameter
Unit
pH Sodium, as Na Calcium as CaCO3 Magnesium as MgCO3 Chloride as Cl Sulphate as SO4 Silica as SiO2 Iron as Fe Copper as Cu Ammonia Oil Total dissolved solids
Polished Condensate Quality Required
ppb ppb ppb
Process Condensate (Petrochemical gas Cracker unit) 8.5-9.5 40 55 35
ppb ppb ppb ppb ppm ppb ppm ppm
35 80 30 20 0.1 0.06-1.3 20 270
5 max. 5 max. 5 max. 0.003 < 1.0 25 max.
6.5-7.0 5 max. -
Table 4. Process Condensate Impurities (Typical) Parameter pH Total Fe Total Cu Total Al Silica as SiO2 Hydrocarbons Conductivity TDS
Process Condensate 8.0-8.50 1 mg/l 5 mg/l 4 mg/l
Polished Condensate Quality Required 7.0 0.005 mg/l 0.003 mg/l 0.03 mg/l < 0.01 mg/l NIL < 0.20 micromho/cm < 0.05 mg/l
An ion-exchange condensate polisher contains a strong acid cation exchanger with a mixed-bed polisher unit for the removal of dissolved impure solids in mineral acids. The cation exchanger removes the bulk of the ammonia or amines, and converts them into dissolved strong acid. The strong based anion unit in the mixed-bed polisher have acid adsorption capacities and function better than a single mixed-bed unit. The cation unit also removes almost all suspended solids. Typical flow diagrams for the treatment of process condensate are shown in Figs. 6 and 7. The unit can be put on a bypass circuit and depending
592
Rajani
upon the impurities, can be used in operation. The unit is attached to a cooler to cool the condensate to 40-500C before it is fed to polishing unit.
593
Corrosion Management
Figure 6. Schematic arrangement for filters
Figure 7. Schematic arrangement for turbine condenser
594
Rajani
In the case of condensate containing colloidal iron, oil and/or other impurities, the treatment scheme consists of
• • • • •
AC filters (primary and secondary), Oleophilic resin bed, Course micron filter, Polished fine micron filter, and Mixed-bed (cation and anion) resins To avoid frequent fouling and increases in pressure, two types of micron filters are required: a course micron filter for < 10 μm and a polished micron filter for < 1.0 μm for removal of the iron. The AC filter will remove condensate oil from levels of 15 ppm to < 10 ppm. The oleophilic resin bed can be used for condensate oil contents of condensate up to 100 ppm. Generally no regeneration is carried out for AC filter. Instead the AC is replaced every 9-12 months (i.e., whenever exhausted). TURBINE CONDENSATE Turbine condensate from thermal power plants is generally pure and does not have contaminants, except for corrosion products in the form of copper and nickel oxides, the concentration being maximum during start up. The turbine condensate of thermal power stations with ratings of 210 MW and below can be returned to the boiler without any treatment; however, feedwater requirements for boiler for super power stations ( 500 MW and above) are stringent, and hence, cannot be reused without treatment. For this type of situation, a condensate polisher is provided for the removal of impurities in feedwater before its reuse in the boiler system. Cartridge filters effectively remove suspended and colloidal impurities present in turbine condensate. The condensate, free from suspended and colloidal particles, is then fed to a strong acid cation exchanger followed by a mixed-bed polisher for the removal of dissolved impurities. Again the polisher is put on a bypass circuit. CONCLUSIONS The corrosion in a steam condensate system can be controlled by proper chemical treatment of the boiler feedwater and effective boiler water conditioning and monitoring. Erosion-corrosion can be prevented by preventing carryover and installing efficient steam traps. Cavitation can be controlled by providing impingement plates and selecting resistant materials for the pump impeller. The chemical treatment program controls corrosion in steam condensate and reduces energy losses due to steam leaks; it also reduces production losses. The contaminated condensate is treated before its reuse in boiler system, especially in high pressure boilers, where stringent quality standards are necessary. As much condensate as possible should be recovered to reduce the feedwater requirement, which will result in savings in both fuel and energy costs as well as in the cost of the treatment of makeup water. REFERENCES
1. P.F. Pelosi and C.J. Cappablanca, Corrosion Control in Steam and Condensate lines, Chemical Engineering, June 1985. 595
Corrosion Management
2. H. Uhlig, Corrosion Handbook, John Wiley & Sons Inc., New York.
596
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
IMPROVED CATHODIC PROTECTION OF ABOVE GROUND STORAGE TANK BOTTOMS: MAA REFINERY EXPERIENCE A.K. Jain, L. Cheruvu and M.E. Al-Ramadhan Kuwait National Petroleum Co., MAA Refinery, Kuwait
ABSTRACT A number of above ground storage tank bottom failures due to soil-side corrosion were experienced and the causes were investigated. The existing deep ground bed-based cathodic protection (CP) system was found to be inadequate due to the large scale drain of CP current to other structures and the nonuniform distribution of current to all the areas of tank bottom. A number of ways to improve CP system’s performance were considered, and two types of improved CP systems were implemented while a third type is under implementation. Use of correct monitoring practices is critical and adequate facilities have been built for monitoring the true performance of the CP system. This paper describes the implementation and operating experience of the improved CP systems including the total CP system revamping plan for a large refinery tank farm. Key Words.
Cathodic Protection, tank bottom, tank farm, monitoring, current density, current distribution, interference, reference electrode.
INTRODUCTION The Mina Al-Ahmadi (MAA) Refinery in Kuwait has over 170 above ground storage tanks (AST) in the refinery tank farm located on the Arabian Gulf shore. The refinery has been in existence since the early fifties, and major expansion and modernization were carried out in 1984-85 converting it to a mega-refinery. The MAA tank farm stores petroleum feed, intermediates and products at cryogenic, and ambient to hot service temperatures. Hence, a variety of operating conditions exist. During the expansion programs, a large number of ASTs and associated piping facilities were added leading to severe congestion of the tank farm. A series of tank bottom leaks led engineers to thoroughly scrutinize the existing cathodic protection (CP) system for its effectiveness in controlling soil-side corrosion. The monitoring practices were also evaluated for their applicability in the given environment. The severe soil-side corrosion on some tanks within a short life of 7-12 years underscores the prevailing highly corrosive environment and the urgent need to improve the CP system to mitigate the corrosion. EXISTING CATHODIC PROTECTION (CP) SYSTEM Most of the tanks provided with CP operate at ambient temperature, while some store hot products at 90oC. The tanks have two types of foundation. The first type is directly erected 597
Corrosion Management
on compacted soil with a 3 in. thick sand-bitumen mix pad. The second type is made on annular reinforced cement concrete ring wall 1.5 ft width x 5 ft depth on which the tank shell rests. The tank bottom sits on 3 in. thick sand-bitumen mix which is laid on compacted earth. The soil side of the tank bottom was generally sandblasted and painted with coal tar epoxy paint. The tank farm is located on the Arabian Gulf shore where the water table is very shallow (1-3 m) and an abundance of chlorides is found in the soil. A severely corrosive environment is, therefore, present. Most of the tank bottoms are protected by deep ground bed-based impressed current cathodic protection (ICCP) systems. The ground bed consists of a string of 13 high-siliconchromium iron cast anodes located in a hole drilled to a depth of 200 ft with calcined petroleum coke backfill. The deep ground bed is powered by a direct current (DC) transformer rectifier (T/R) power source. A typical schematic diagram is given in Fig. 1. The negative drain cables are connected to tanks in a variety of manners, sometimes directly and sometimes through pipework. No electrical isolation is provided for the tanks or to any of the systems. The CP system is, therefore, an integrated type in which all of the underground structures are provided with simultaneous CP using a number of ground beds and T/R, working in parallel circuit. Permanent zinc reference electrodes were provided under the tank bottoms for the tanks built during the refinery modernization plan and expansions, while other tanks had no monitoring reference electrodes. The tank periphery is generally covered with 1 1.5 m of sand-bitumen layer and no soil pots (test stations) exist for the monitoring of potentials close to the tank bottom. EXTERNAL CORROSION Routine tank inspection in accordance with API-653 [1] revealed that many tank bottoms are suffering from severe soil-side corrosion attack. The attack was very severe in the case of tanks storing hot products (80-90oC). Severe corrosion of annular plates and adjoining sketch plates leading to perforations and paper thinning was found in a short life of 8 years. The corrosion rates are estimated to be in the range of 45-50 mpy. An investigation into the existing CP system resulted into following conclusions:
• Most of the tank bottoms are not protected by the existing deep ground bed CP system.
• The current output from deep ground bed anodes is diverted to the massive underground structures installed during the refinery expansion. The presence of sand-bitumen mix below the tank bottom only aided in large scale diversion of CP current to other structures by offering higher resistance to CP current.
• The monitoring of protective potentials by conventional protective potential measurements near the periphery is not adequate. Also large IR, deep in the soil are included in the reading unless measurements are taken very close to the tank bottom.
598
Jain et al.
Figure 1. Schematic diagram of the existing deep ground bed cathodic protection 599
Corrosion Management
• The presence of foreign matter in the soil such as oil, bitumen and vegetation, significantly influences the accuracy of CP potentials measurements. Considering these factors, the main objectives of the CP system revamping were laid down as
• • • • •
Uniform distribution of CP current to the entire tank bottom, Avoidance of CP current drainage to other structures, Minimum interference, CP system life comparable to tank bottom life, i.e., 25-30 years, Adequate monitoring facilities for true evaluation of the CP system’s performance, • Easy implementation, and • Reasonable cost. A number of options were considered for the CP system’s improvement. The first choice to be made was between the remote and close ground bed. The remote ground bed was considered nonworkable due to the congestion of the tank farm, difficulties involved in the electrical isolation of the tanks and interference problems from other structures. A sacrificial anode bed-based CP system was considered inadequate due to the limited driving potential, limitations on the maximum achievable life and the lack of controls. A close distributed ICCP system was, therefore, the only workable option that could achieve the listed objectives. A number of ground bed designs are possible for a close-distributed ICCP system. On one extreme are the grid type of ground beds located under the bottom which offer excellent current distribution, while on the other extreme are the periphery-based anodes which offer fair current distribution. Placing anodes under the bottom by angular drilling or by direct installation inside the tank during overhaul falls in between the above two extremes. A few published case histories are available on situations where this type of CP system has been installed. Kroon [2] has discussed various types of CP system designs with estimated cost comparisons. Horizontally installed anodes under the bottom, the so called crow foot CP system have been implemented in an Arctic environment [3] with good success. Garrity [4] has described case histories of angle-drilled CP systems resulting in satisfactory performance. Considering the size and complexity of the tank farm, and the difficulty with tank shutdowns, it is readily apparent that no single prescription can be made for all the tanks. In fact, depending upon the particular conditions of a tank, the optimal option has to be selected out of a number of feasible alternatives while simultaneously considering time, resources and cost constraints. Therefore, the following five types of close-distributed CP system designs were selected for implementation:
• • • • •
Perimeter ground bed, Angle-drilled ground bed, Under tank, directly buried anode bed, Grid type ground bed, and Loop type ground bed. The major criteria for selecting a particular CP ground bed out of the above are 600
Jain et al.
• • • • •
Tank size, viz small, medium, or large, Tank type, viz single bottom, double bottom, or with liner, Tank service, viz hot or ambient, Tank location, viz isolated or congested, and Tank status, viz in-service, under rebottoming, or under major overhaul. While a detailed description of each CP system will be discussed under Implementation, a brief summary is presented in Table 1. The applicability of the particular CP system to tanks according to the aforesaid criteria is given in Table 2. Selection of a particular CP system for a given tank is done by short listing the applicable CP systems from Table 2, and then selecting the most economical system that can protect the given tank. CATHODIC PROTECTION (CP) SYSTEM DESIGN While designing a CP system, optimization was carried out for system reliability, uniform protection spread and cost. The main design highlights are
• Current density of 30 mA/m2 for ambient temperature service and 50 mA/m2 for hot service (80-90oC) tanks with 25% extra for possible drainage,
• Electrical isolation of tanks from break out piping to minimize current drain, • Uniform spread of CP current by distributing the anodes (as per Eq. 2below), • Mixed metal oxide coated titanium anodes with armored tail cables for system longevity,
• 30 year design life, • Resistors for control of each anode, and shunts for current measurement, • Permanent Cu/CuSO4 reference electrodes under the bottom for monitoring the potential at the center and annulus, and
• Soil pots adjacent to the bottom rim for monitoring. The uniform spread of protection was calculated using following approach [5]: Spread Efficiency
E = 2.Ps / (Ps + Pr)
(1)
Minimum depth of anode
da = D / E.n
(2)
Where E = spread efficiency, Ps = surface soil resistivity, Pr = bulk soil resistivity at a depth equal to the tank radius, n = number of anodes, D = tank diameter, and da = minimum depth of anodes from tank bottom for uniform protection spread. Table 1. Merits and Limitations of Selected Close-Distribution Cathodic Protection (CP) 601
Corrosion Management
Table 2. Applicability of Cathodic Protection (CP) Systems to Tanks 602
Jain et al.
Since in MAA the water table is very shallow, the Pr is normally negligible as compared to the PS hence resulting in a spread efficiency of 2. The depth of each anode below the tank bottom and the number of anodes are chosen to exceed the limit imposed by Eq. 2 for uniform distribution of CP current to the entire tank bottom, with adequate overlap. A higher number of anodes are normally installed for redundancy to cover incidental anode failure during the design life of the system. IMPLEMENTATION OF THE IMPROVED CATHODIC PROTECTION (CP) SYSTEM Perimeter-Based Cathodic Protection (CP) System This system was chosen for two tanks, numbered TK-101 and TK-204, 59ft and 100 ft in diameter, respectively. These tanks were newly built and located in an isolated area away from the refinery complex. The CP system consisted of a number of high-silicon Cr-canistered anodes located around the tank periphery in a vertically drilled hole at 8 m depth with 3 m stand off from the tank. The schematic CP system layout is given in Fig. 2. Although Cu/CuSO4 was the preferred reference electrode, zinc was already installed under the bottom. Resistors and shunts were included in each anode circuit for individual anode monitoring and current balancing. The system was satisfactorily commissioned in May 1994 (even though the tanks remained empty) and an instant off potential of -1.400 V was achieved at a current density of 1.18 mA/m2. Once the tanks were filled with product, protective potentials of -1.030 to 1.340 V instant off were achieved at a current density of 4.33 mA/m2. Detailed protective potentials are given in Tables 3 and 4.
603
Corrosion Management
Figure 2. Schematic diagram of a perimeter cathodic protection (CP) system Table 3. Protective Potentials, TK-101 Perimeter Cathodic Protection (CP) System
604
Jain et al.
The system has operated satisfactorily since that time with a very low current density of 1.34 mA/m2. Under Tank Cathodic Protection (CP) System Under the tank CP system was the answer for the tanks located in the refinery tank farm and that come-up for rebottoming work. This CP system has been implemented for 3 tanks. TK-820, a 192 ft diameter tank is a hot tank operating at 80-90oC. It was built in 1984. The tank was originally provided with a deep ground bed CP system which failed to provide adequate protection. Severe soil-side corrosion was observed on the annular plates and adjoining sketch plates resulting in paper thinning and perforations (Fig. 3). Hence, the tank was partially rebottomed in 1994. Making use of this opportunity, an under-bottom CP system was installed by excavating around and drilling inside the tank for anode installation. The CP system consisted of a number of mixed metal oxide coated titanium anodes located inside the tank in a vertically drilled hole at a depth of 3.5 m with a 5 m stand off distance from the shell. The water table was observed at a depth of 2 m. The monitoring facilities installed were 3 Cu/CuSO4 permanent reference electrodes and 12 test stations at the tank’s periphery, very close to the shell. The control facilities included resistors and shunts for each anode. The tank was electrically isolated from the connected piping by installing flange isolation half kits. A schematic drawing is given in Fig. 4. The CP system was commissioned in October 1994, and with the tank empty, protective potentials of -0.89-1.2 V were achieved at a current density of 6.8 mA/m2. Increase in current demand was observed when the tank was filled with hot product, and protective potentials of 1.0-1.4 V were achieved at a current density of 13 mA/m2. The system has been operating satisfactorily since that time. The details of the protective potentials are given in Table 5. Monitoring of the earthing system of the tank has revealed that there is no significant interference. Similar CP systems have been installed for two other tanks, TK-504 with an 80 ft diameter and TK-753 with a 168 ft diameter. The results available from these tanks show satisfactory performance of the CP system. Protective potentials of -1.05-1.20 V have been achieved at a current density of 3.4 mA/m2, under empty tank conditions. The tanks are yet to be commissioned. Table 4. Protective Potentials, TK-204
605
Corrosion Management
Figure 3. Severe soil-side corrosion and perforation of a tank bottom plate (TK-61-821)
Table 5. Protective Potentials for the TK-820 Under Tank Bottom Cathodic Protection (CP) System
606
Jain et al.
Figure 4. Schematic diagram of an under tank bottom cathodic protection (CP) system 607
Corrosion Management
Angle-Drilled Cathodic Protection (CP) System In order to accomplish CP system up gradating in a reasonable time, it is imperative that the CP system upgrade be carried out while the tanks are in service. The angle-drilled CP system is, therefore, the answer for most of the tanks. Around 25 tanks are scheduled to be upgraded by November 1998, while the rest are to be upgraded by 2001. The angle-drilled CP system consists of a number of mixed metal oxide coated titanium anodes located under the bottom of the tank at a vertical depth of 5 m. Angle drilling at 45o is envisaged to be carried out from the tank periphery which shall allow on-stream anode installation. A greater drilling challenge is offered for Cu/CuSO4 reference electrode installation which would require horizontal drilling up to the tank center 2-3 ft from the bottom of the tank. Other facilities would include resistors and shunts for each anode control and test stations at tank periphery. A schematic diagram of the angle-drilled CP system is given in Fig. 5. During drilling, the safety and stability of the tank is of prime concern, and a laser-based monitoring system is envisaged to determine the impact during the progress of drilling. The angle-drilled anodes are reported to have a good success record and have resulted in better current distribution to the tank bottom. A number of case histories are described by Garrity [4] in which conventional deep ground bed and shallow ground bed CP systems could not effectively protect the bottom, while subsequent angle-drilled CP system installations led to fully protected tank bottoms. Grid-Loop Type Cathodic Protection (CP) System The recent trend, especially in the US, is to install a secondary containment plastic liner under the tank bottom for environmental protection. The presence of such a liner precludes the deep, shallow or angle-drilled CP system methods, as the liner prevents the flow of CP current to the bottom. Similarly for double-bottom tanks, the new bottom cannot be protected by the above types of CP systems. In these cases, the anode must be located between the bottom and the liner in the first case and between the tank bottoms in the second case [6]. The grid type CP system consists of mixed metal oxide coated titanium ribbon anodes laid in the form of a closely spaced grid. The most common ribbon size is 12.5 mm wide x 0.5 mm thick. A titanium conductor bar and multiple power feed points are installed to minimize circuit resistance, to promote uniform ground bed potential and to provide a redundancy of power supply points to the ground bed. A schematic diagram of a grid type CP system is given in Fig. 6. The main advantages of a grid type CP system are its ability to provide CP in the presence of a liner, sandwich type construction requiring little space, extended life expectancy, and excellent current distribution. The ribbon anode spacing is critical to ensuring uniform protection of the tank bottom. Considerable variation in opinions and practices is found. Simulation experiments have been carried out and it has been recommended that the ribbon anode spacing not exceed four times their distance from the tank bottom [7].
608
Jain et al.
Figure 5. Typical schematic diagram of an angle-drilled cathodic protection (CP) system 609
Corrosion Management
Figure 6. Typical schematic diagram of a grid type cathodic protection (CP) system
Figure 7. Typical schematic diagram of a loop type cathodic protection system 610
Jain et al.
Some successful accounts of the use of the grid type CP system are available in the published literature. McJones [8] described two case histories where two 150 ft diameter hot service (260oF) tanks have been satisfactorily protected with the grid type CP system. In this case, the tanks are reported to be adequately protected with a 34 mA/m2 current. The loop type CP system offers some of the same advantages as the grid type CP system. The loop type CP system consists of a series of fully preassembled concentric loops of mixed metal oxide coated titanium anodes with piggyback wire (Fig. 7). These loops are laid in coke breeze backfill under the bottom with around 5 feet spacing. The major advantages over the grid type system are greater spacing between loops, easier construction, and complete absence of field joints thus enhancing anode reliability. However, more space is required for the anode bed construction. None of the tanks in MAA have secondary containment liners. The use of grid-loop type CP is planned for three new tanks. CONCLUSIONS Deep ground bed-based CP systems could not provide effective CP for tank bottoms in the MAA refinery due to the large scale diversion of the CP current to other underground structures. No single type of CP system could be prescribed for the CP revamping of a big, congested refinery tank farm. Optimal CP systems has to be selected from a number of short listed options depending upon each tank’s location, service, type, size and status with consideration given to time, resources and cost constraints. Uniform distribution of CP current to the entire tank bottom should be the primary consideration for any CP system design. The traditional monitoring of protective potentials near or around the tank periphery is not adequate and may even result in wrong conclusions being drawn. Permanent soil pots free from contaminants located immediately next to the tank bottom and permanent reference electrodes installed under the tank bottom are the minimum essential for true evaluation of CP system performance. The observed protective potentials must be corrected for IR drop in the soil, and it is felt that apart from the -850 mV with respect to Cu/CuSO4 criterion, the other two criteria given in NACE-RP-01-69-92 and NACE-RP-01-93 [9,10] should be applied from time to time. An improved perimeter-based CP system has been installed for two tanks, and adequate protection has been achieved with a 1.34 mA/m2 current. An improved under-tank CP system has been installed for three tanks, and adequate protection has been achieved with a 13 mA/m2 current for the hot (90oC) tank and a 3.4 mA/m2 current for other tanks. Onstream installation of an angle-drilled CP system is planned for 25 tanks in the next two years, while grid and loop type CP systems are planned for implementation for 3 new tanks. REFERENCES 1. API-653, Tank Inspection, Repair, Alteration and Reconstruction, American Petroleum Institute. 2. D.H. Kroon, Cathodic protection of above ground storage tank bottoms, Conference on Above Ground Storage Tanks, MTI & NACE, Paper No. 29, January 1992, pp. 29/1 to 29/9.
611
Corrosion Management
3. T. Barletta et al., Taps storage tank bottoms fitted with improved cathodic protection, Oil and Gas Journal, Oct. 23, 1995, pp. 89-93. 4. K.C. Garrity, Cathodic protection of external tank bottoms, Material Performance, April 1988, pp. 32-35. 5. J. Morgan, Cathodic Protection, 2nd edition, Houston, Texas, NACE, January 1993, pp. 237-238. 6. API-651, Cathodic protection of above ground petroleum storage tanks, American Petroleum Institute, Washington, DC, April 1991. 7. L. Koszewski, Impressed current CP for new and existing tank bottoms with secondary containment, Conference on Above Ground Storage Tanks, MTI & NACE, Paper No. 40, January 1992, pp. 40/1-40/4. 8. S. McJones, Retrofitting storage tanks with double bottoms and improved CP for hot service, Conference on Above Ground Storage Tanks, MTI & NACE, January 1992, Paper No. 33, pp. 33/1 to 33/10. 9. NACE-RP-0193-93, External Cathodic Protection of On Grade Metallic Storage Tank Bottoms, National Association of Corrosion Engineers. 10. NACE RP-0169-92, Control of external corrosion on underground or submerged metallic piping system, NACE.
612
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
IMPACT ON SHIP STRENGTH OF STRUCTURAL DEGRADATION DUE TO CORROSION M.A. Shama Naval Architecture and Marine Engineering, Faculty of Engineering Alexandria University, Egypt
ABSTRACT This paper addresses the problem of corrosion of ship structures. The main environmental and operational factors affecting the initiation, spread and rate of corrosion of ship structures are given. The main problems of general and local corrosion of cargo ships, bulk carriers and oil tankers are briefly discussed. The methods commonly used to control and prevent the initiation and spread of corrosion are briefly indicated. The consequences of corrosion associated with the degradation of the strength of ship structures are stressed with particular emphasis on the buckling strength and stiffness of plating panels. Parabolic and exponential models are assumed to represent the variation of the rate of wastage with time of the buckling strength. The effect of uniform corrosion on the section modulus and the inertia of stiffeners having different geometrical configurations are presented. The impact of material wastage due to corrosion of the deck structure on the ship section geometrical characteristics is examined. The effect of corrosion on the magnitude of the factor of safety and the probability of failure is presented. Structural degradation problems associated with corrosion of HTS are clarified. It is shown that ship section modulus, inertia, buckling strength and the flexural rigidity of panels of plating could be significantly reduced in ships experiencing normal rates of wastage due to corrosion. The deterioration of the geometrical characteristics of stiffeners could vary between 15% and 38% depending on the section configuration. Key Words: Corrosion rates, structural degradation, structural failure, material wastage
INTRODUCTION Failure statistics reveal that corrosion is the most common defect in steel vessels and is the dominant cause of structural failures for ships older than eight years. Corrosion of ship structures results mainly from age, inadequate maintenance, the chemical or corrosive action of cargoes, local wear of steel plating and sections, and improper structural design features among other factors. A corroded steel plate is not only thinner but is also more brittle and is thus more prone to initiating fatigue cracks. High stress concentration induces microscopic cracks in the highly stressed parts of the steel structure. These cracks propagate into the coating and act as pockets where corrosion action begins. It is, therefore, necessary to eliminate any critical defects prior to service and to prevent noncritical defects from growing to critical size during service. Local pitting can cause serious thinning of plates and sections at critical, highly stressed areas. The cause of local pitting can be water condensation, lack of cathodic protection, material defects, faulty preparation and treatment of the material, etc.
615
Corrosion Management
The deterioration of a HTS plate due to general or pitting corrosion will be more significant than that of a mild steel plate having the same strength. Therefore, ships with greater use of HTS will experience serious degradation problems in terms of buckling and loss in fatigue strength. This will certainly impair the expected service life of the vessel. Protection of the hull structure against corrosion is, therefore, crucial for the prolonged life of ships. Higher values of wasted material may be accepted in certain areas for ships built to scantlings higher than those required by classification societies. Such increased scantlings are often used by owners to account for loss of material due to corrosion and to minimize the amount of material replacement required throughout the life of the ship. This paper, therefore, addresses the corrosion problem of ship structures with particular emphasis on the impact of material wastage due to corrosion on the strength and stiffness of stiffeners, plate panels and the geometrical characteristics of ship sections. FACTORS AFFECTING CORROSION Corrosion of ship structures represents one of the main causes of structural failure [1-3], see Fig. 1, and is generally affected by the following factors: design, fabrication, protective coating, operation, maintenance and repair, and environment. 78%
corrosion 8%
design & workmanship
5%
fatigue
4%
excessive wave load
5%
vibration 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Figure 1. Main factors affecting hull structural failure The main environmental factors influencing the initiation of corrosion are seawater salinity, temperature, pollution, marine fouling, humidity, and presence of oxygen. The main operational factors participating in the initiation and spread of corrosion are type of cargo, cargo residues, speed of flow, mechanical abrasion, frequency of tank washing, presence of stray current, and coating failure. The main causes of coating failure are numerous, among them are poor coating specifications, high stress concentration, poor penetration-resistance of the coating surface, low coating thickness, painting on moist
616
Shama
surface, contamination between coats, insufficient curing time, and poor handling of coated surfaces. RATES OF CORROSION Corrosion occurs at a rate independent of plate thickness. The rate of corrosion of ship structures varies with the ship’s age and type, the area of operation, and section of the ship, i.e., bottom, vertical or top. Several factors affect the rate of corrosion, such as type of corrosion, type of cargo carried, presence or absence of sulphur, operating profile, characteristics of the environment, treatment of the steel before painting, presence of water, frequency and extent of inspection and maintenance, effectiveness of the corrosion control system, condition of coating, and orientation of the surface. The mean values of corrosion rates for the different ship types are given in Table 1. These values are average values and are subjected to several sources of uncertainty. Corrosion rates, however, could be significantly reduced by controlling the main factors causing corrosion. Good flexible coatings could be very useful against stress-induced corrosion. Inert gas systems are effective in reducing the rate of corrosion in tankers. Cathodic protection, based on sacrificial or impressed current system is also used as a reliable and cost-effective measure against corrosion of steel ships. The proper specifications of coatings in the building stage may represent a crucial factor for reducing corrosion of hull structure. Table 1. Mean Corrosion Rates for Different Ship Types Ship Type General Cargo Annual Rate (mm/year)
0.09
Oil Tankers 0.1
Ore Carriers 0.12
Bulk Carriers 0.17
CONSEQUENCES OF CORROSION Corrosion of ship structures has a deleterious effect on maintenance and repair costs, hull girder and local strength and service life [4,5]. The impact of corrosion on the structural strength and stiffness of ship hull girders and local structural members are as follows:
• • • •
Reduction of local and hull girder scantlings and load carrying capacity, Increase of local and hull girder stresses and stress concentration, Reduction of hull girder and local safety factors, Reduction of local and hull girder flexural rigidity, buckling strength and fatigue strength of structural connections
DESIGN CONSIDERATIONS General
617
Corrosion Management
Deficient hull girder and local strength may result from errors in design, material, fabrication and operation [1]. The main factor causing strength deficiency is structural deterioration by corrosion. Deterioration by corrosion of welded steel connections may be excessive around fillet welds. The effect of the quality of maintenance on the rate of strength deterioration is shown in Fig. 2. Figure 3 shows the effect of corrosion on the shape of the SN curve. It is clear that uncontrolled corrosion of a ship’s structure could lead to a significant reduction of the fatigue strength [6]. Bottom plating sustains a major portion of the hull bending moment in addition to the local loads induced by hydrostatic pressure. Its strength may be reduced significantly by general and localized corrosion [7]. Deck plating comprises a highly stressed portion of the ship’s hull girder and is of critical importance to the longitudinal strength of the vessel. Corrosion of the deck structure is expected to occur because of the frequent deck washing, water condensation, local deck deformations, and mechanical abuse from deck cargo. Therefore, the deck plating should be carefully examined for cracks, leaks, signs of excessive corrosion, wear and tear, and buckling [8].
Ro B R(t)
Over Maint. Good Maint. A C
SAFE ZONE
Rm Under Maint. Ro = Original strength Rm = Minimum strength Time, t
FAILURE ZONE T
Figure 2. Variation of R(t) with time The reduction of ship section geometrical characteristics due to material wastage of deck structure due to corrosion is shown in Fig. 4. A 30% reduction in the sectional area of the deck plating could cause a 15% reduction in ship section modulus. The variation of the critical buckling strength with time for a panel of plating is shown in Fig. 5 for an assumed parabolic model for the material wastage due to corrosion, and in Fig. 6 for an exponential model. It is shown that the critical buckling strength could be reduced by more than 30%
618
Shama
when the thickness of the plate panel experiences the normal wastage due to corrosion. The effect of uniform corrosion of an angle section 200 x 100 x 13 mm in size on the section modulus and inertia is shown in Fig. 7 and for an OBP section 430 x 20 mm in size is shown in Fig. 8. The section modulus and inertia could be reduced by more than 25% for normal values of material wastage. Figure 9 shows the deterioration of the section modulus with time for three different section configurations. It is shown that the deterioration of the section modulus and inertia vary between 15% and 38% depending on the section configuration. This may have a significant impact on the magnitude and distribution of flexural and warping stresses over the web and flange of the section, particularly for asymmetrical sections [9].
Fatigue In Air (No Corrosion) Cathodically Protected (Controlled Corrosion) Log o Fatigue In Sea Water (Uncontrolled Corrosion)
Log N Figure 3. Effect of corrosion on fatigue curve
619
Corrosion Management
0.25
dI = reduction of second moment of area of ship section dZ = reduction of ship section modulus alpha = % reduction of deck area e = shift of N.A.
0.2 0.15 0.1 0.05
e / yo 0 10%
20%
30%
40%
50%
alpha
Figure 4. Variation of ship section characteristics with deck deterioration
Rt /
1 0.9 Ro 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Parabolic Model, Rt = Ro ( 1 - b.t ) b = 0.0006 b = 0.0007 Ro = original buckling strength of plate Rt = deteriorated buckling strength of plate 0
5
10
15
20
Time , Years Figure 5. Variation of buckling strength with time: Parabolic model
Rt
1 0.9 /0.8Ro 0.7 0.6 0.5 a 0.4 0.3 0.2 0.1 0
Exponential Model, Rt = Ro.Exp.(- a t ) = 0.02 a = 0.005
Rt = Deteriorated buckling strength a = 0.01 Ro = Original buckling strength of the plate 0
5
10
15
20
Time , Years F Figure 6. Variation of plate buckling strength with time: Exponential model
620
Shama
1
Io = Original Section Inertia Zo = Original Section Modulus
0.9
610 x 13 0.8
200 x 100 x 13
0.7
0
Z / Zo
1.5
I / Io
Material 3 Wastage , mm
Figure 7. Effect of material wastage on Z/Zo and I/Io
Figure 8. Effect of uniform wastage on Z/Zo and I/Io
Figure 9. Variation of Z/Zo with time
The increased flexibility of ship structures due to the extensive use of HTS or due to the improper distribution of material over the ship section may cause local breakdown of protective coatings and initiation of corrosion. The effect of material wastage for a panel of plating on the flexural rigidity is shown in Fig. 10 for three different plate thickness. It is clear from Fig. 10 that the flexural rigidity could be significantly reduced when the plate panel experiences normal wastage due to corrosion. General Cargo Ships For general cargo ships, corrosion occurs at several places such as the deck structure, the lower part of the holds or twin deck frames, the lower part of transverse bulkheads, the aft end of double bottom tanks and certain areas of tank tops where water is trapped due to ineffective drainage systems [1]. Figure 11 shows the distribution of corrosion failure among the different structural elements of the deck structure. Corrosion may aggravate and accelerate crack and buckling failures of the deck structure.
621
Corrosion Management
Figure 10. Effect of plating wastage on flexural rigidity
Figure 11. Corrosion failures of deck structure
For transversely framed ships, severe buckling of bottom plating within the midship region may result from general and local corrosion. Figure 12 shows the distribution of structural failure due to corrosion among the bottom structural elements. Such failure can seriously impair the hull girder and local strength of the midship section. Structural failures of the side shell plating due to corrosion may account for 66% of the corrosion failures of the side shell structure. Transverse bulkheads between cargo holds experience excessive corrosion in the lower boundaries and in the way of bilge wells. High wastage occurs in ballast tanks and in inaccessible structural areas where inspection and maintenance are difficult, such as the top parts of transverse bulkheads. Ease of access to the various parts of cargo and ballast tanks is necessary if efficient inspection, maintenance and monitoring is required. Bulk Carriers The structural failures of bulk carriers due to corrosion have been recognized and reported in several papers [10]. The strakes of side shell plating between wind and water are highly susceptible to severe uniform corrosion and localized pitting. This area is generally subjected to high shear stresses [11,12]. High material wastage of these areas could have an adverse effect on the shear buckling strength of the side shell plating, particularly for HTS. Hopper, topside and double bottom tanks, ballast holds and void spaces are subjected to excessive material wastage by corrosion. The tops of transverse bulkheads and the underside of the deck structure are also subjected to severe corrosion. Oil Tankers Hull fractures and failures in oil tankers are generally attributed to detail design aspects in areas of stress concentration at bracket toes and longitudinal connections to transverse web frames [1,13]. The stress concentration factors at these connections are relatively high and could represent a serious hazard when material wastage due to corrosion reaches unacceptable values. The rate of corrosion of oil tankers is significantly influenced by the following main factors: tank washing, tank contents, tank atmosphere when empty, and presence of inert gas. All parts of ballast tanks and transverse bulkheads between the cargo tanks and ballast tanks are subjected to severe corrosion. The lower and top strakes of transverse and longitudinal bulkheads are generally subjected to excessive corrosion. The deterioration of deck longitudinals by corrosion may be more rapid and extensive than that of deck plating. Wastage may reach 50% in some parts. The use of HTS allows lighter scantlings and results in a more flexible structure. Greater flexibility of the primary structure results in more load being shed to secondary structures. Thus, the strength of the ship is more dependent on the integrity of the structure as a whole. Excessive flexibility may also lead to fatigue problems. HTS structures exposed to stress cycling at higher stress levels will have a shorter fatigue life than the equivalent mild steel
622
Shama
structure. Therefore, the deterioration of strength due to corrosion is far more significant for HTS than for mild steel. EFFECT OF CORROSION ON THE PROBABILITY OF FAILURE It is evident that the deterioration of ship’s structural elements due to corrosion will have direct impacts on the critical buckling strength, fatigue strength, factor of safety and probability of structural failure. Figure 13 shows the variation with material wastage of the factor of safety and the critical buckling strength of a panel of plating. The factor of safety for flexural buckling is reduced by about 30% when the plate thickness experiences a normal rate of corrosion.
Figure 12. Failure of bottom structure due to corrosion Figure 13. Variation of γ and σcr with wastage
623
Corrosion Management
Figure 14. Effect of poor maintenance on the probability of failure (Pf)
Figure 15. Variation of the probability of failure (Pf) with time The effect of strength deterioration due to poor maintenance and excessive corrosion on the shape of the density function of strength and the probability of failure, Pf, is shown in Fig. 14. It is evident that the probability of failure increases significantly with increasing deterioration due to corrosion. Figure 15 shows the increase with time of "Pf" for fatigue and buckling strength of a ship’s structural member subjected to normal rates of corrosion. CONCLUSIONS The main conclusions drawn from this investigation are 1. For normal rates of material wastage experienced in ship structures due to corrosion, the geometrical characteristics of ship sections and section modulus and inertia of stiffening members are seriously impaired. The reduction could vary between 15% and 38% depending on the section configuration. The buckling strength and flexural rigidity of panels of plating could also be reduced by 30%. 2. Proper maintenance strategies and methods are crucial elements in reducing corrosion of steel ships. 3. Cracks induced by high stress concentration propagate into the coating and act as pockets where corrosion action begins. It is, therefore, necessary to eliminate any critical defects prior to service and to prevent noncritical defects from growing to critical size during service. 4. In order to prevent/reduce the initiation of pitting corrosion, it is necessary to eliminate any critical defects prior to service and to prevent noncritical defects from growing to critical size during service. 5. A good inspection, maintenance and monitoring system for assessing and protecting hull structures from corrosion represents a main factor for reducing strength deficiencies and extending ship life. 6. High quality surface preparation and protective coatings significantly reduce the rates of general and local corrosion.
624
Shama
7. The increased flexibility of ship structures due to the extensive use of HTS or due to the improper distribution of material over the ship section may cause local breakdown of protective coatings with subsequent initiation of corrosion. REFERENCES 1. M.A. Shama, Ship structural failures: Types, causes and environmental impacts, AEJ, July, 1995. 2. Y.Akita, Considerations for prevention of hull failure, NK Technical Bulletin, 1983. 3. M.A. Shama, Ship casualties: Types, causes and environmental impacts, AEJ, April, 1995. 4. ABS, "Guide For Vessel Life Extent", 1990. 5. Ship repair and conversion technology, Life extent: The problems and prospects, No.2, 1989. 6. SSC-318, SSC-346, Fatigue characterization of fabricated ship details for design, 1983, 1990. 7. Int. chamber of ship, Guide. manual for the insp. and condition assess. for tanker structures, 1986. 8. ABS, "Guide For Vessel Condition Assessment And Reconditioning", 1991. 9. M.A Shama, Stress analysis and design of fabricated asymmetrical sections, Schiffstechnik, September 1976. 10. ABS, Bulk Carriers: A Guide for concern, 1993. 11. M.A. Shama, Shear stresses in bulk carriers due to shear loading, JSR, SNAME, September 1975. 12. M.A. Shama, Analysis of shear stresses in bulk carriers, Computers and Structures 6, 1976. 13. SSC-312, Investigation of internal corrosion and Corrosion control alternatives in commercial tank ships,1981.
625
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
CONTACT ELECTRIC RESISTANCE (CER) TECHNIQUE FOR MONITORING OF PROCESS PLANTS AND FOR SOLVING PRACTICAL CORROSION PROBLEMS K. Saarinen and T. Saario VTT Manufacturing Technology, P.O. Box 1704 FIN - 02044 VTT, Finland
ABSTRACT At the moment both the scientific understanding of corrosion processes and the engineeing practices for corrosion control in process plants can benefit considerably from the development of in situ on-line instruments for the characterization of surface films on construction materials. In this paper, a new in situ contact electric resistance (CER) technique for the measurement of the electric resistance of surface films on metals is described in detail. The CER technique has been applied in several different research areas. These include optimization of the chrome-plating process, prevention of localized corrosion of stainless steel, investigation of the effect of inhibitors in crevice environments and monitoring of the kinetics of oxide growth in high-temperature water. Key Words: On-line monitoring, surface film properties, corrosion
INTRODUCTION The role of surface films in corrosion, and more generally, in all surface-related phenomena has become more recognized as new in situ techniques have evolved, and relevant information has become available. For example, the properties of the surface film on Inconel 600 have been targeted [1,2] as the main area of research for mitigating stress corrosion cracking in steam generators. This work outlines the scope of industrial and scientific problems which have been successfully encountered using the newly developed [3,4] contact electric resistance (CER) technique. With the CER technique, the electric resistance of the surface film on a material is monitored in situ. Electric resistance is a basic physical property of a material, and can be used to extract a multitude of information on surface films. When used in industrial processes for monitoring purposes, the measured electric resistance is positively correlated to important variables such as onset of pitting corrosion [5], stress corrosion crack growth rate [6], efficiency of particular gas mixtures in the oxidation of metals, and reduction of oxides or the eddy current detectability of surface cracks in steam generator tubes [7].
THE CONTACT ELECTRIC RESISTANCE (CER) TECHNIQUE 627
Novel Techniques
In general, the electric resistance (R) of a material depends on the cross-sectional area of the sample (A), the thickness of the sample (l), and the specific resistivity of the material (r), according to the equation:
R = ρ⋅
l A
(1)
In the CER method [3], normally two identical specimen surfaces are brought into contact and then disconnected with a user-selectable frequency. When the surfaces are disconnected, they interact with the electrolyte or gas. When the surfaces are in contact, the resulting change in the contact resistance (R) is measured. Direct current is led through the surfaces, and the resulting voltage is recorded. The contact resistance (R) is then calculated from the voltage using Ohm’s law. The resulting resistance is the resistance of the surface layer in the thickness, i.e., in the z-direction. In another development of the CER method one of the two opposing identical specimens is replaced by a probe made of iridium metal. This allows a more detailed analysis of the electric properties of the surface layer [3]. Experiments with a further modification of the iridium probe technique have shown that the resistance of the surface layer in the x-ydirection (i.e., the plane of the layer) is radically different from the resistance in the zdirection. The design of the CER-instrument developed for measurements in high temperature high-pressure process environments is shown in Fig. 1. In field applications, another design can be used which fits into a 50 mm diameter penetration. The maximum temperature achievable with the CER technique is above 1000oC in a gaseous environment. In liquid environments, the maximum pressure and temperature are limited by the specification of the pressure container typically to 20 MPa and 360oC. The specimen surfaces have been standardized to have a diameter of 2.0 mm, although other forms, such as tubes [8] can be used as well. The actual contact area depends on the surface roughness and on the contact displacement (load) used. For the surface pretreatment normally used, with the typical contact displacement of about 2 μm, the load directed to the specimen surfaces is roughly 2 N/mm2, and the actual contact area is 38% of the nominal surface area. If the thickness of the surface layer is known, the specific resistivity (ρ) of the surface layer can be calculated from Eq. 1. The resolution of the standard measurement system is better than 10-7 Ω. The movement is produced by a step motor, and it is mechanically reduced to provide a displacement control accuracy of about 10-9 m. MEASUREMENTS IN AQUEOUS ENVIRONMENTS Some examples of the applications of the CER technique in various industrial and basic science problems follow. Basic Science Figure 2 shows an example of a potentiodynamic scan for chromium in 1 M H2SO4 at room temperature [9]. In addition to the polarization current, the electric resistance of the surface film was monitored as a function of the potential. The polarization currents of Cr are seen to correlate with the electric resistance of the surface film. This suggests that the 628
Saarinen and Saario
reactions occurring on Cr are controlled by electron or electron hole transfer across the film. In the range -200...0 mV, the resistance had increased to a level of 1 Ω. This corresponds to a resistivity of 105 Ωcm, if the thickness of the film is assumed to be 10 Å.
Figure 1. Design of the CER instrument for measurements at high-temperature high-pressure processes (a). (1) flange, (2) supporting frame, (3) step motor, (4) pull rod, (5) seal, (6) supporting frame, (7) stiff spring, (8) specimen holder, (9) specimen, (10) connecting wires, (11) soft spring. Specimen holder with fixed specimen (b). (1) specimen, (2) connecting wire, (3) insulation, (4) Zr holder, (5) Zr-ring, (6) Zr screw Steam Generator Tubes The most widely used technique for NDE of steam generator tubing is eddy current. This technique can reliably detect cracks grown in a sodium hydroxide environment only at depths greater than 50% through the wall. However, cracking caused by thiosulphate solutions has been detected and sized at shallower depths. The disparity has been proposed to be caused by the different electric resistance of the crack wall surface films and corrosion products in the cracks formed in different environments. Eddy current signals generated by a closed crack having corrosion products with low electric resistance can remain undetected or be undersized in comparison with a similar crack that has surface film and corrosion products with high electric resistance. Figure 3 shows the apparent size of a crack as measured by the eddy current technique presented as a function of the surface film electric resistance [7]. The detectability of tightly 629
Novel Techniques
closed cracks by the eddy current technique depends on the electric resistance of the surface films of the crack walls. The nature and resistance of the films which form on the crack walls during operation depends on the composition of the solution inside the crack and close to the crack location. During the cooling down of a steam generator, because of contraction and loss of internal pressurization, the cracks are rather tightly closed so that the exchange of electrolytes and thus changes in the film’s properties, becomes difficult. As a result, the surface condition prevailing at high temperature is preserved. If the environment is such that the films formed on the crack walls under operating conditions have low electric resistance, the eddy current technique will fail to indicate those cracks or will underestimate their size. However, if the electric resistance of the films is high, a tightly closed crack will resemble an open crack, and will be easily indicated and correctly sized by the eddy current technique. 40
30
10000
1000
surface film resistance
100
10
current density
10
1
R/Ohm
jCr / mAcm-2
20
0.1
0 0.01
-10 0.001
-20 -0.8
0.0001
-0.3
0.2
0.7
1.2
ECr / V (vs. SHE)
Figure 2. Cyclic voltammogram (sweep rate 0.1 mVs-1) of a Cr electrode in 1 M H2SO4 (T = 25oC), and the surface film resistance of Cr registered during the positive going potential scan [8] Paper Mill Thiosulphate is known to cause pitting of stainless steel used as a construction material in paper mills. The CER technique was used [5] to show that if thiosulphate was present in a simulated wet end environment of the paper mill in excess of 10 mg/l, it decreased the surface film electric resistance by more than one order of magnitude. Figure 4 shows, as a function of thiosulphate concentration, the ratio of the surface film electric resistance measured in the presence of thiosulphate to that measured without thiosulphate. The same technique was then used to find an efficient inhibitor to prevent the adverse effect of thiosulphate. The best inhibitor found was hydrogen peroxide at a 3:1 concentration ratio with respect to the concentration of thiosulphate. 630
Saarinen and Saario
Figure 3. The apparent crack size (in percentage of the real crack) as measured by the eddy current technique, presented as a function of the electric resistance of the surface film on the crack walls [7]
Figure 4. Ratio of the surface film resistance measured in the presence of thiosulphate to that measured when thiosulphate is absent, presented as a function of the concentration of thiosulphate [5] Electroplating
631
Novel Techniques
In electroplating, formation of an electrically insulating surface film during any of the typical pretreatment processes is potentially detrimental to the quality of the final plating layer. Field measurements with the CER-technique were performed at FinnAir’s aircraftengine maintenance department to establish a step-by-step follow-up of the company’s plating processes. Figure 5 shows an example of a laboratory measurement in which the effect of pretreatments on the surface film electric resistance of brass was investigated with the CER-technique. The wax removal bath was efficient, and the resistance decreased to a low level, where it stayed during the following wash with water. However, upon exposure to the grease removal bath, a high resistance film was immediately formed on the surface. Switching the potentiostat on (at time moment 12 minutes) and polarizing the specimen to -0.5 V (about 0.3 V below the open circuit potential) resulted in a gradual decrease of the resistance, indicating film reduction. Further polarization to -0.6 V resulted in a decrease of the resistance to that of the level in water before exposure to the grease removal bath. When the potential was then increased stepwise, the increase of resistance commenced at -0.45 V. The film was again reducted at -0.6 V, after which the polarization was switched off. As a result, the potential increased very quickly to about -0.2 V, and simultaneously the resistance started to increase very steeply. It is evident from the data shown in Fig. 5, that monitoring with the CER-technique can be a very efficient way for optimizing the pretreatments used for electroplating processes.
100
0
Grease removal wash
Brass
T = 45 °C
-0.1
1
-0.3
0.1
Water 75 °C
Wax removal
RESISTANCE / Ohm
-0.2
-0.4
-0.5
POTENTIAL (V vs. Ag/AgCl)
10
0.01 -0.6
0.001
-0.7 0
5
10
15
20
25
30
35
40
TIME / min
Figure 5. The effect of electroplating pretreatments on the surface film electric resistance on brass MEASUREMENTS IN GASEOUS ENVIRONMENTS
632
Saarinen and Saario
In situ monitoring of surface layer properties in high temperature gaseous environments can be of advantage both in materials research and in controlling industrial processes. High Temperature Oxidation Figure 6 shows an example of changes in the electric resistance of the surface layer on nickel at 300oC in a gaseous environment. When exposed to a reducing gas mixture of argon with 3% hydrogen, the measured electric resistance was rather low, a few milliohms. The proposed cause of this resistance is adsorption of argon and hydrogen on the surface. 100 Ar + 3% H2
Air
Ar + 3% H2
RESISTANCE / Ohm
10
1
NICKEL T = 300 0.1
0.01
0.001 70
80
90
100
110
120
130
140
150
TIME / min
Figure 6. An example of monitoring of the electric resistance of the surface layer on nickel in a gaseous environment at 300oC As the gas mixture was substituted by air, the electric resistance of the surface layer on the nickel probe increased by more than three orders of magnitude. The kinetics of the increase in resistance was very fast in the beginning, after which the growth was almost linear as a function of time. The increase in the resistance of the surface layer when exposed to air is proposed to indicate the formation of nickel oxide. When air was substituted with the same argon/hydrogen gas mixture as before, the electric resistance fell, achieving in a few minutes the level prevailing before exposure to air. The decrease of resistance when exposed to the reducing gas mixture is considered to indicate a reduction of the nickel oxide layer that formed during the preceding exposure to air.
Low Temperature Oxidation
633
Novel Techniques
Many processes, for example electroplating, are often preceeded by surface cleaning in air by sandblasting or other techniques. Another example is the electronics industry, where pretreatments are applied either to oxidize or to reduce the surface layers. In air exposures, the kinetics of the oxidation/reduction processes as well as the electric resistance of the oxides, is known to be very sensitive to variation in the air’s humidity. Figure 7 gives an example where oxidation of pure nickel was studied with the CER-technique in laboratory air at room temperature. 1000 NICKEL T = 25 oC LAB AIR
100
RESISTANCE / Ohm
10
1
0.1
0.01
0.001
0.0001 0
5
10
15
20
25
30
35
40
TIME / min
Figure 7. Kinetics of oxidation of nickel in laboratory air at room temperature. Three repetitions are shown The three repetitions show a considerable scatter both in oxidation kinetics and in the resistance level achieved. When similar repetitions were made in an environment with carefully controlled humidity, the scatter was almost absent. CONCLUSIONS The CER technique is applicable for in situ monitoring of the electric resistance of surface layers on metals and semiconductors. The technique is applicable both in gaseous and aqueous environments at both ambient and high temperatures. The CER technique can be used for in-line monitoring of industrial processes. Electric resistance of a surface layer is a material property, the measurement of which gives additional insight to a great majority of surface-related processes. REFERENCES
634
Saarinen and Saario
1. P.L. Andresen and F.P. Ford, Fundamental quantification of crack advance for life prediction in energy systems. Corrosion NACE´96, Proceedings, Research Topical Symposia, pp. 51-99. NACE International. 2. R.W. Staehle, Bases for objectives of testing for steam generator tubing. Corrosion NACE 96, Paper No. 122, NACE International. 3. T. Saario, Development and applications of the contact electric resistance technique, Dr.Tech. thesis, Helsinki University of Technology, Espoo, Finland, 1995. 4. T. Saario and J. Piippo, A new electrochemical technique for in situ measurement of electric resistance and semiconductor characteristics of surface films on metals, Materials Science Forum 185-188, 1995, pp. 621-628. 5. V.A. Marichev, T. Saario and V. Molokanov, Prevention of localised corrosion caused by thiosulphate in paper mill environments, 12th International Corrosion Congress: Corrosion Control for Low Cost Reliability, Vol. 2. Houston, Texas, USA, 19-24 September 1993, National Association of Corrosion Engineers (NACE), 1993, pp. 826833. 6. U. Ehrnstén, J. Lagerström, T. Saario, J. Piippo, P. Aaltonen, S. Tähtinen, T. Laitinen and H. Hänninen, Environmentally assisted cracking of Alloy 600 in PWR primary water. Accepted for publication in Eurocorr’96. 7. T. Saario and J.P.N. Paine, Effect of the surface film electric resistance on eddy current detectability of surface cracks in Alloy 600 tubes, Proceedings, Seventh International Symposium on Environmental Degradation of Materials in Nuclear Power Systems: Water Reactors, Vol. 1, August 6-10, 1995, Breckenridge, Colorado, pp. 327-337. 8. T. Saario, S. Tähtinen, J. Piippo and J.J.V. Kukkonen, In situ measurement of the effect of LiOH on the stability of fuel cladding oxide film in simulated PWR primary water environment. Proceedings, Seventh International Symposium on Environmental Degradation of Materials in Nuclear Power Systems: Water Reactors, Vol. 2, August 610, 1995, Breckenridge, Colorado. Vol. 2, pp. 1231-1245. 9. G. Fabricius, T. Laitinen, J. Piippo, T. Saario, K. Salmi and G. Sundholm, Electrochemical behaviour of chromium in sulphuric acid. Accepted for publication in Eurocorr’96.
635
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
DESIGN OF RADIO FREQUENCY METHODS FOR CORROSION PROCESSES MONITORING Yu.N. Pchel’nikov, Z.T. Galiullin and A.S. Sovlukov Institute of Control Sciences 65, Profsoyuznaya str., Moscow, 117806, Russia
ABSTRACT Methods for monitoring nondestructive defects (e.g., cracks and corrosion deposits) are proposed. Their application field includes corrosion process monitoring of pipelines and various technological installations including those that are coated by protective layers. These methods are based on probing monitored objects by radio frequency (RF) electromagnetic waves, i.e., by surface electromagnetic waves. Different practically adopted, slow-wave structure (SWS)-based RF-sensors have been designed for use in a real exploitation environment. Some peculiarities of SWS are described and their application for testing corrosive objects are considered. Key Words: Corrosion deposits, cracks, radio frequency monitoring, surface electromagnetic wave, slow wave structure
INTRODUCTION The corrosive processes of metal objects take an important place among factors that determine the reliability and durability of their exploitation. There are many reasons for the development of corrosion. Among them are stress-corrosion and ulcer corrosion, caused by sulfur-hydrogen cracks; a lot of objects’ failures are connected with hydrogen cracking of metals. The degree of that or other factors influencing the formation and development of corrosion depends on concrete exploitation conditions and the aggressiveness of different substances. Monitoring corrosion processes is thus of great importance for many industrial and transportation objects. Among them are oil and gas main-pipelines. They are subjected to stress-corrosion, in particular resulting in possible environmental pollution and loss of transported substances. Attempts to design measuring devices for testing crack development on metal surfaces have resulted in the creation of sensors based on various physical principles. These methods do not allow for proper decision on the problem of testing defects on objects covered by protective, nonmetallic coatings without destruction of their integrity. Such anticorrosive protective layers with thicknesses of several millimeters and more, do not hinder the receipt of information with the help of microwave and radio frequency (RF) testing techniques [1,2]. The methods proposed are directed at the monitoring of nondestructive defects (e.g., cracks and corrosion deposits) on pipelines and technological installations, including those
637
Novel Techniques
that are coated with protective isolation layers. These methods are based on the application of RF electromagnetic waves (1-1000 MHz) for the probing of monitored objects. SLOW-WAVE STRUCTURES POTENTIAL FOR CORROSION PROCESS MONITORING Great functional abilities are inherent to devices based on the use of surface (decelerated) electromagnetic waves. Electromagnetic wave deceleration (slowing down) of hundreds and thousands of times results in the RF-sensor’s size decrease, due to the miniaturization of sensitive elements, made as a piece of the slow-wave structure (SWS), for example, a meander-line or a helix (Fig. 1). Some positive peculiarities are inherent to SWS that allow it to perform corrosion monitoring tasks on a new qualitative level. Among these peculiarities are the following (Fig. 1):
a) meander line
b) helix line
Figure 1. Examples of radio frequency (RF) slow-wave structures (SWS) Slowing down provides an electromagnetic field power concentration near the SWS surface (Fig. 2). By the choosing the slowing-down value or the electromagnetic oscillation frequency, the RF-energy concentration area can be changed without any additional screens.
Figure 2. Radio frequency (RF)-power (P) distribution near planar slowwave structures (SWS) surface (x-distance to SWS, z-SWS longitudinal coordinate, τ-transverse distribution field coefficient) Surface wave specifics take place for various SWS-sensor forms (e.g., planar, circular etc.). It provides the ability to concentrate RF-energy in needed, monitored areas. RF-wave 638
Pchel’nikov et al.
deceleration depends essentially on environmental parameters, including electrophysical properties and distance to the metal surface. These dependencies can be used for the receipt of information on the different corrosion processes’ parameters. The slowing down index of the SWS-sensor is frequency-dependent as a rule. By choosing SWS construction, this dependence can be changed within broad limits, that provides the possibility of controlling the RF-field parameters. Deceleration and transverse RF-field structure depend significantly on the excited electromagnetic wave mode. Thus, the sensitivity of SWS-sensor to monitored and disturbed parameters can be controlled. By choosing the SWS-sensor conductors and/or excited wave mode in coupled systems, different magnetic or electric field energy concentration can be provided in the monitored areas. As a result, multifunctional monitoring devices can be designed. The peculiarities of SWS listed above allowed the design of many types of RF-sensors with different nonelectrical quantities. Among them are sensors for measuring gaps, linear and angular displacement, pressure, dielectric and metal thickness, diameter, and level [2]. Deceleration of a RF-wave’s phase velocity is realized both by rolling up the conductors, for instance into a helix, or by resonance elements sequent junction, or by a combination of these. The phase velocity of basic (zero) space harmonic deceleration takes place in this case, while the phase velocities of high-order harmonics may be less than most of light velocity also under the absence of the wave slowing down. According to the method of deceleration, SWS constructions are realized that can be conventionally divided to wave-guide SWS, designed on the basis of cavity or dielectric wave-guides, and two-conductors SWS. The latter contains two-conductor live transmission live, and one conductor or both of them are rolled up into a helix, meander or a line of some other type. The SWS-sensor may contain three or more conductors. SLOW RF-WAVE AND CORROSION SUBJECTED METAL SURFACE INTERACTION ANALYSIS Monitoring of the corrosion state of a metal surface, based on the use of SWS-sensors can be realized in the following way. On the monitored object ,1, (Fig. 3), disposed in corrosive environment ,2, outside dielectric coating layer ,3, for instance in the soil, a sensor ,4, is installed. In this sensor, the surface electromagnetic wave directed along the metal surface ,5, of the monitored object is excited. The sensor is connected with the transducer ,6, that provides measurement of the sensor’s informative parameter. By measuring as a function of time the changes in the surface wave deceleration index ,n, caused by the gradual removal of the conducting surface as a result of the corrosion process, the changes in the corrosive layer can be determined. The measured changes also reflects the increase in the distance ,d, between the surface and the sensor. The change in the deceleration index ,n, is an informative parameter that is transformed in the change in the sensor’s resonance frequency ,fr, measured with the help of sweep oscillator , or in the autogenerator’s frequency ,fg. In this case, the sensor is inserted in the
639
Novel Techniques
backward net of the generator [2]. If it is necessary, the number of the monitored parts of the object and sensors may be more than one.
Figure 3. Slow wave structures (SWS)-sensor and monitored object disposition The surface electromagnetic wave is excited in the sensor in such a way that the magnetic field’s energy is located within the coating layer of the object. As a result, there is practically no sensitivity to the coating layer’s parameters (its permittivity and losses). The magnetic field’s energy displacement is proved by the antiphase excitation of impedance conductors 7 and 8 of sensor 4 (Fig. 4). They have identical sizes and configurations in the form of 180o mirror imaging. The electric field’s energy is basically concentrated between the impedance conductors, and under such magnetic field displacement, this electric field energy is really absent within dielectric coating 3 outside the conductors. Therefore, the permittivity and conductivity of the layer do not influence the propagation coefficient ,g, of the decelerated wave, that is excited in the sensor. The availability of object 1’s monitored metal surface ,5, is accompanied by a decrease in the phase coefficient ,b, caused by the excitation of the electric current on the metal surface. Its amplitude is approximately proportional to exp.(-2pd/lS), where lS is the decelerated wave’s length, and its direction is opposite to that of the current basic component in impedance conductors [4]. The wave length , lS , is determined by the relation of wave length , l , in free space to the deceleration index , n. The magnetic field flow, excited by the current induced on surface 5, is subtracted from the magnetic flow excited by conductors 7 and 8, that results in the decrease in the sensor’s equivalent inductance. The decrease of the phase coefficient ,b, caused by the reduction of the equivalent 0.5 inductance is approximately proportional to [1 - exp.(-2pd/ls)] . Such dependence is explained with the exponential character of the decelerated wave’s field of distribution near the sensor’s surface and has significant influence on measurement sensitivity. Indeed, if the gap ,d, between the sensor and the monitored surface exceeds the area of the surface location of the wave energy, expressed as lS/2p, then measurement sensitivity is really low. That is the reason why the wave length ,lS, and admittable gap value ,d, between the sensor and the metal surface must be chosen with satisfaction of the next requirement. d < lS/2p
640
(1)
Pchel’nikov et al.
Figure 4. Slow-wave structures (SWS)-sensor with magnetic field energy displacement in a monitored area Here the wave length lS is approximately equal to half of the meander-line period , T , when the magnetic-type wave of the sensor is represented basically by plus the first and minus the first space harmonics, as it takes place in the case of the impedance conductor realization in the form of meander-lines (Fig. 5).
Figure 5. Meander-line conductors of a slow-wave structures (SWS) sensor The decrease of current induced on the metal surface is also caused by the appearance of a defect on this surface, for instance a crack that crosses the induced current’s route (Fig. 6). As it follows the helix SWS model analysis [4], the availability of longitudinal cracks in a screen surface, when they have a depth compared with surface wave field location area, results in a significant increase in the phase coefficient ,b, that is an increase in the deceleration index , n. In the case considered, such a crack (or other defect) can be alone and its influence is essentially lower. However, as the measurement results show, this influence is enough for crack detection and monitoring its development. 641
Novel Techniques
Figure 6. Surface with a crack defect The crack caused a decrease in the induced current on the monitored surface and, as a result, an increase in the phase coefficient b. This is transformed into an adequate sensor’s resonance frequency fr , testifying the availability of a crack or a defect of another kind. The change of resonance frequency fr with time follows the defect development. The increase in crack depth or corrosion depth on a monitored surface area is accompanied by a corresponding decrease in resonance frequency. The presence of defects, and in particular of cracks, results in the change not only in the imaginary part of the propagation coefficient ,g, the value of b ( g = -jb - a), but in its real part a too. This real part (i.e., attenuation coefficient) is increased because of additional losses caused by current route lengthening on cracks’ walls or on other defects.
Figure 7. Coupled arithmetic helices for SWS - sensor As is confirmed by the measurement results obtained with the help of the arithmetic helix-sensors (Fig. 7), the presence of dielectric or semiconductor media in a monitored area does not really influence on the measurement results. Analytically, it is confirmed by the fact that the surface wave phase coefficient ,b, in coupled-SWS under their antiphase excitation increases essentially the geometric deceleration index, determined as the relation of the length of impedance conductors to the length formed by these conductors SWS in the direction of the wave propagation.
642
Pchel’nikov et al.
Here, transverse coefficient ,t, characterizing surface wave field distribution practically equals the phase coefficient b (b = 2p/ ls). These coefficients are related by the expression 2
2
2
β = τ + k (ε - jσ/wεo)
(2)
where k = 2p/l is wave index for free space, w is angular frequency, eo is vacuum permittivity, e is relative permittivity, and s is the isolating layer material’s conductivity. The deceleration index , n , equals b/k; therefore, from the above expression for b it is seen that under the condition 2
n > 0.1(ε - jσ/wεo)
(3)
The typical parameters e and s for the isolating layer are insufficient in order to influence the value of b. DESIGN OF SLOW-WAVE STRUCTURES (SWS) SENSORS In most of the cases of practical interest, the surface of the monitored object has the form of a cylinder or sphere, which makes it difficult to monitor with the plane sensor. This drawback can be removed by designing a sensor with impedance conductors repeating the geometrical form of the monitored object. For instance, in order to monitor a pipeline’s corrosion state, a sensor made in the form of a dielectric channel 9 (Fig. 8) will be suitable. Conductors 7 and 8, in the form of rectangular or oval logarithmic helices with oppositely directed windings, are deposited on the inner and external surfaces of channel 9. Such helices form results in an increase in the sensitivity of the depth measurement of cracks directed along cylinder’s axis on the pipeline’s surface. Opposite directions of the helix conductors windings provide, as was mentioned above, magnetic field energy displacement into the isolating layer area and a high value for the slowing down index n. The latter results in the ability to use relatively low frequencies (0.1-10 MHz) and the absence of influence by the isolating layer’s parameters. The logarithmic form of the impedance conductors allows receiving uniform distribution of the decelerated wave’s field along these conductors, which provides the maximum value of sensitivity.
Figure 8. Slow-wave structures (SWS)-sensor for monitoring pipeline surface defects 643
Novel Techniques
When a relatively large area of a cylindrical object should be monitored, the sensor can be realized in the form of a channel. But in this case, impedance conductors 7, 8 on inner and outer surfaces of the channel 9 should have meander-line form (Fig. 9). These conductors are shifted in the longitudinal direction in half of the period T. The latter provides magnetic field displacement into the isolating layer , 3 , and the reception of a high value for the deceleration index , n. Here, the period T is chosen taking into account the needed depth of monitoring.
Figure 9. Slow-wave structures (SWS)-sensor for monitoring large pipeline surface defects SOME MEASUREMENT RESULTS Measurement results confirmed the ability to monitor corrosion processes with the use of SWS-sensors. In the experimental device, sensor 4, based on coupled logarithmic helices 7 and 8, was inserted in the backward circuit 10, connecting the output of amplifier 11 with its input (Fig. 10). Such a scheme transforms a change of the phase coefficient , b , into a change in the generator frequency , fg. Measurements were made with a piece of a real metal pipeline, having an inner diameter of 1.4 m. The cracks’ length did not exceed 100 mm, their width did not exceed 0.4 mm, and their depth did not exceed 6 mm in the middle. The sensor was installed close to the isolating film with a thickness of 3 mm. As it was ascertained when the sensor was installed over a nondamaged part of the sample, the output frequency was 10.3 MHz, and when the sensor was placed over the middle of one crack, the output frequency decreased to a value of 9.6 MHz. This proved the high sensitivity of the proposed method.
Figure 10. Slow-wave structures (SWS)-device autogenerator scheme 644
Pchel’nikov et al.
Other types of measurement schemes may be also used. In one of them (Fig. 11), sensor 4’s resonance frequency fr was measured by the dependence on corrosion process parameter. This scheme contained a sweep oscillator ,12, detector unit ,13, and a frequency counter ,14.
Figure 11. Slow-wave Structures (SWS) resonance device scheme CONCLUSION Many different metal surfaces can be monitored using the RF-methods proposed. The fields of application include both pipelines for product transport (e.g., oil, gas etc.) and various other technological installations. From the plurality of proposed sensors located in the monitored objects areas, all the necessary information about the metal surface quality may be sampled at rather distant points. REFERENCES 1. V.A. Viktorov, B.V. Lunkin and A.S. Sovlukov, Radiowave methods for technological processes parameter measurements, Moscow, Russia, Energoatomizdat Publ., 1989, p. 208 (in Russian). 2. Yu.N. Pchelnikov, A.V. Fadeev and V.V. Annenkov, Proc. of the Seventh Int. Conference on Vacuum Web Coating, Miami, Florida, USA, 1993. Pp. 26-42. 3. Yu.N. Pchelnikov, Radiotechnika i Elektronika. 33, 1988. pp. 2042-2045 (in Russian). 4. N.P. Kravchenko, L.N. Loshakov and Yu.N. Pchelnikov, Radiotechnika i Elektronika. 21, 1976, pp. 706-714 (in Russian).
645
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
A NEW, RAPID CORROSION RATE MEASUREMENT TECHNIQUE FOR ALL PROCESS ENVIRONMENTS A.F. Denzine1 and M.S. Reading2 1 2
Cortest International, Inc., Chardon, Ohio, USA
Cortest Instrument Systems, Inc., Willoughby, Ohio USA
ABSTRACT Three main corrosion monitoring techniques are used to measure process fluid corrosivity. For the majority of industrial applications, electrical resistance (E/R) probes and corrosion coupons (CC) are most widely used, since these techniques can be applied to nearly all process environments. These devices provide a basic measurement of metal loss, but require that a significant time base be used before an accurate corrosion rate can be determined. In conductive electrolytes, the linear polarization resistance (LPR) probe technique may be employed. This technique has the advantage of providing instantaneous measurement of the corrosion rate, but because of its electrochemical nature, cannot be applied in most process environments where monitoring is required. Newer techniques seek to provide both the instantaneous corrosion rate information offered by the LPR technique, as well as the universal applicability of the E/R technique. This paper describes a new corrosion monitoring technology which achieves these objectives. Under trial testing, sensitivities 400 times that of E/R technology have been apparent. For industrial monitoring situations, this technology now approaches the instantaneous corrosion rate measurement offered by the LPR technique. In addition, since the fundamental parameter measured by this new technique is metal loss, it has the advantage of being applicable to all process environments. Key Words:
MICROCOR, corrosion monitoring, rapid technique, electrical resistance, linear polarization resistance
INTRODUCTION Weight loss coupons, electrical resistance (E/R) probes, and linear polarization (LPR) probes have long been used to measure corrosion rates in industrial plants and equipment. Both weight loss coupons and (E/R) sensors require considerable metal loss to take place prior to producing a measurable response. Many days, or even weeks are required to sense moderate changes in corrosion rates with any certainty. Newer variants on the E/R technique apply excitation current to the actual pipe or vessel wall, measuring the resultant potential field pattern. They show promise for use in inaccessible locations, (e.g., subsea applications), but are two orders of magnitude slower in response than conventional E/R probes [1]. Months, and sometimes years, are required to obtain measurable results. Such historical data cannot be used in automated-feedback process control systems. 647
Novel Techniques
The LPR technique provides rapid response; even small changes in the corrosion rate can be detected within 5-15 minutes. However, LPR measurements are only possible in corrosive fluids that are aqueous electrolytes. In addition to this limited applicability, LPR provides relative data, frequently requiring expert interpretation. Generally, traditional monitoring methods are either slow to respond, or are of limited applicability and questionable accuracy. There is a need for a new monitoring method that incorporates the speed of LPR with the accuracy and universal applicability of E/R or weight loss measurements. Cortest has developed a patented technique in which changes in the mass of corroding metal elements are sensed as changes in the inductive resistance of the sensing element. Test data shows a response time approximately 200 times faster than that of a corresponding E/R probe. Corrosion rate changes on the order of 3-4 mpy are accurately detectable in less than 10 minutes. The following outlines initial test data comparing traditional monitoring techniques with this new technology, i.e., MICROCORTM. EXPERIMENTAL PROCEDURE Response to Rapid Changes in Corrosion Rate A recirculating system of tap water, flowing at 40 liters/minute was allowed to reach equilibrium at ambient temperature (24.3°C) and pressure. Commercially available E/R probes from three major manufacturers were installed in the system, together with LPR probes and MICROCORTM probes. The sensors were allowed to stand for 48 hours to establish a stable corrosion product film, after which time, probe readings were taken at 60 second intervals using real-time LabviewTM data acquisition software. After 180 readings (a period of three hours), during which baseline corrosion data was collected, acetic acid was introduced into the circulation loop at a concentration level of 0.015 g.mole.1-1. Data collection proceeded for an additional 255 readings, at which point the system was purged of acidified water, and replenished with conventional tap water at the same temperature. Data collection proceeded for an additional 130 readings in the replenished, less aggressive environment. Temperature Sensitivity E/R probes of both the cylindrical and wire-loop type were placed in a constant temperature bath at 24°C ± 0.5°C, together with the MICROCORTM probe. The probes were left for two days to ensure the establishment of thermal equilibrium and a stable corrosion product layer, at which point the probes were read at four-second intervals. After a suitable time period, the probes were rapidly transferred to another constant temperature bath maintained at 17°C ± 0.5%. Rapid data collection was maintained until the probe readings indicated a full recovery from the temperature excursion. The probes were then rapidly returned to the higher temperature bath, and the experiment was concluded by taking data points until the probe readings again stabilized.
648
Denzine and Reading
RESULTS Corrosion Rate Measurements Figure 1 shows comparative data from the MICROCORTM probe and a cylindrical element, E/R probe. To aid in comparison, probe life is shown in volts as opposed to the 01000 probe life units used for conventional E/R readings. For the purposes of comparison, two millivolts is equivalent to one probe life unit. The MICROCORTM probe clearly shows two sharp discontinuities corresponding to the acidification of the solution, and the subsequent purging of acid from the system. These changes approximate to an initial corrosion rate of 12 mpy changing to 31 mpy and, finally returning to a low value of 4 mpy. The E/R data shows no discernible trend, and gives no indication of extreme rate variations.
Figure 1. Responses of MICROCORTM and electrical resisrance (E/R) probes to rapid changes in corrosion rate Figure 2 shows the same MICROCORTM data compared with LPR measurements. The LPR data shows initial rates varying from 4-8 mpy, followed by a period where the rate varied from 11-29 mpy, and finally a gradual decline from 8 mpy to 5 mpy. 649
Novel Techniques
Figure 2. MICROCORTM/linear polarization resistance (LPR) comparison Slope variations in the MICROCORTM data are exactly coincident with corresponding changes in LPR rate data. This is most clearly seen in the period of high corrosion, where six distinct slope changes occur in the MICROCORTM data, and are accomplished by six corresponding variations in the LPR data. Changes in corrosion rate are detected as rapidly by MICROCORTM as by LPR, and the magnitude of those changes correlate well on a relative basis. Comparison of absolute readings for MICROCORTM and LPR probes also shows excellent correlation, considering the LPR instruments use an industry accepted average proportionality constant and were not precalibrated using DC polarization curves. Figure 3 shows the comparison of LPR data with that of E/R data during this same experiment. The E/R data shows excessive, but typical scatter and gives no indication of the extreme variations of corrosion rate; the E/R probes require vastly longer times to discern a measurable, average, corrosion rate. Temperature Sensitivity Figure 4 illustrates the results obtained by subjecting E/R probes and MICROCORTM probes to small, but rapid, temperature changes. The period for recovery on all probes tested was less than two minutes. However, in the case of the E/R probes, the initial deviation in readings was 1.5% of the full scale, whereas the MICROCORTM probe showed less than 0.05% deviation. An isolated temperature change will obviously do little to adversely effect the data quality of E/R measurements. However, frequent, small, temperature changes, typically encountered in industrial applications, will impose a noise level in excess of 1% of the full 650
Denzine and Reading
scale on the E/R data, which, in part accounts for the data scatter and the lengthy time lapse required to discern corrosion trends using this technique. The temperature compensation of the MICROCORTM probe, by contrast, is sufficiently rapid to produce no adverse effects in its theoretical resolution.
Figure 3. Electrical resistance(E/R) /linear polarization resistance comparison (LPR)
Figure 4. Temperature compensation of MICROCORTM and electrical resistance (E/R) corrosion probes DISCUSSION 651
Novel Techniques
Over the last four decades, manufacturers of E/R probes have introduced a wide variety of element designs. Some improve sensitivity, such as the thin-walled hollow wires; others to increase resistance to physical damage and increase element life, such as cylindrical elements. The spectrum of element designs that fall within the practical limits of normal use are either extremely insensitive, with concomitantly long response times, or comparatively sensitive, fragile, and short-lived. E/R probes have a useful life (span) equal to one-half of the element thickness (except for wire elements, where the span equals one-quarter of the wire diameter). The span is divided into 1000 probe units. Therefore, more robust (thicker) elements suffer from reduced sensitivity as each probe unit represents a greater amount in metal loss. In addition, the thicker elements have extremely low resistance resulting in a low signal-to-noise ratio. A 20 mil cylindrical element, for example, will take approximately 18 days to measure a change in corrosion rate of 1 mpy. At the other end of the spectrum, a 4 mil hollow wire can detect a corrosion rate change of 1 mpy in less than 48 hours. However, effective life of the 4 mil element will be only one-tenth of the 20 mil cylindrical element. The numbers given above assume perfect temperature compensation. Temperature compensation is, however, a system variable. Systems in which temperatures change extremely slowly, even if by large values (50-100°C), permit the temperature of the E/R probe's corroding and reference element to remain in equilibrium, and temperature compensation is very effective. Systems experiencing large, sudden temperature swings may require many hours for the sensor to recover thermal equilibrium, thus compromising much of the data gathered by the E/R probe. The worst situation, however, is the system experiencing small, but rapid temperature fluctuations. As observed in Fig. 4, such a system can impose a temperature noise of 1-1.5% of the full scale on E/R data, thereby concealing resistance changes due to corrosion effects. As a consequence of this system-dependent temperature noise, the true resolution and hence, response time of an E/R probe in any given system is nearly impossible to predict. The rule of thumb of most manufacturers, which seems to work quite well, requires 1% of the full scale (i.e., 10 probe units) change in probe reading to measure a real change in corrosion behavior [2,3,4,5]. These changes can clearly be obscured by temperature noise. The response times of various element types, at various corrosion rates, are illustrated in Fig. 5. As shown, low to moderate corrosion rates are measured in hundreds of hours. Higher corrosion rates provide faster response, but reduce sensor life to a few weeks. LPR, being a direct measure of rate, provides a response time measured in minutes. The only time lapse involved is that required for capacitive discharge at the electrode to take place, allowing the true polarizing current to be measured. However, LPR can only give an approximation of corrosion rate. The technique is based on the Stern-Geary relationship [6]. ΔV Δi
=
BaBc______ 2.3 (icor)(Ba + Bc)
(1)
where ΔV is the applied polarizing voltage that is set in the instrument, and Δi is the measured polarizing current. Provided the constant BaBc/(Ba+Bc) is known, the corrosion 652
Denzine and Reading
current (icor) may be calculated and hence the corrosion rate. Unfortunately, the Tafel constants Ba and Bc are system-dependent, and the constant BaBc/(Ba+Bc) may vary in the normal spectrum of electrolytes by a factor of 3. Since it is tedious, time consuming, and requires expensive equipment to precalibrate each system for the proportionality constant, all commercial instruments contain an average constant. As a result, the absolute value of corrosion rates measured by the LPR technique may be off by a factor of ±150%. However, relative measurements are reasonably accurate, provided changes in rate are not accompanied by basic changes in mechanism. This question of accuracy, coupled with the need for a continuous electrolyte as a corrosive fluid, poses a serious limitation to the applicability of LPR, in spite of its advantage of rapid response [7].
Figure 5. Response/time comparison of corrosion monitoring techniques Temperature compensation is required for MICROCORTM probes, as it is for E/R probes. However, the temperature effect relates to the thermal coefficient of the magnetic susceptibility of the corroding element material which is normally an order of magnitude less than the corresponding thermal coefficient of resistivity. Consequently, the temperature effects in MICROCORTM probes can be virtually elminated as can be seen in Fig. 4. The MICROCORTM data presented has been collected without any electronic or software-based data smoothing or filtering. Yet, accurate measurement of small changes in corrosion rate can be made in a fraction of one hour. More recent data shows improved sensitivity with reduced noise (Fig. 6).
653
Novel Techniques
Figure 6. Recent MICROCORTM data showing improved sensitivity with reduced noice CONCLUSIONS The MICROCORTM probe offers, for the first time, the possibility of making real-time corrosion rate measurements in corrosive environments that do not present a continuous electrolyte phase to the sensor. Such environments include many of the most critical in terms of the economic consequences of unchecked corrosion. Among the more important of these are oil and gas production and distribution systems, refining and petrochemical process-side operations, cathodically protected structures, boiler feedwaters and condensates, and all operations involving vapors and condensed moisture films. The rapid response of MICROCORTM lends this technique to automated on-line process system controls. Since the MICROCORTM technology is based on the measurement of properties related definitively to mass, it will provide absolute data, and therefore offers advantages over the more approximate LPR technique. The whole spectrum of cooling waters and aqueous solutions encountered throughout the chemical process industries can now be monitored accurately as well as rapidly. ________________________ MICROCORTM is a trademark of Cortest Instrument Systems, Inc. LabviewTM is a trademark of National Instruments, Inc.
REFERENCES 1. M.W. Joosten, K.P. Fisher, R. Strommen and K.C. Lunden, Materials Performance, April, 1995, pp. 44-48. 2. Rohrbach Cosasco Systems, Inc., Bulletin No. 200-B. 3. Cormon, Ltd., GL 003, p. 4. 4. Cortest Instrument Systems, Inc., Bulletin No. 4.2, January, 1992, p. 6. 5. Petrolite Corp., Bulletin ERD-11. 6. M. Stern and A.J. Geary, Journal of the Electrochemical Society 104, 1957, p. 56. 7. E. Tiefnig, Method and Apparatus for Determining Corrosivity of Fluids on Metallic Materials, United States Patent Application Serial #08/229, 449.
654
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
ASSESSING CORROSION OF THICK MARINE PAINTS BY SURFACE CORROSION POTENTIAL MAPPING (SCM) AND AC IMPEDANCE SPECTROSCOPY (EIS) A. Husain Materials Application Department Kuwait Institute for Scientific Research P.O. Box 24885-Safat 13109, Kuwait
ABSTRACT This study aimed at assessing the degradation properties of three different types of organic coatings in laboratory-simulated 3% NaCl solution. Multilayer coating systems of coal tar epoxy, modified epoxy and abrasion-resistant epoxy were investigated. A new electrochemical technique, surface corrosion potential mapping (SCM), was employed to detect the surface potential changes (variations in electrical conductivity) that occurred within the paint film. The surfaces of the painted steel samples were also characterized with electrochemical impedance spectroscopy (EIS) and SEM. The results indicated that SCM has the potential for use in the laboratory as a diagnostic tool for the detection of paint defects or commercial paint quality with a higher degree of surface roughness. Key Words: Corrosion potential mapping (SCM) technique, electrochemical impedance spectroscopy (EIS), paint defects, organic coating, coal tar epoxy, modified epoxy, and abrasionresistant epoxy
INTRODUCTION Electrochemical impedance spectroscopy (EIS) has been extensively used in the study of the corrosion performance of organic coatings [1-3]. The technique involves the application of a small AC signal across the coating/substrate system over a wide frequency range (normally 8 to 10 decades). The magnitude and phase response signal can be used to obtain information such as the capacitive and resistive behavior of the coating system. Continuous monitoring of the impedance spectra for a coated metal enables determination of the water uptake (in percentage of volume) by the coating system, and the protective properties of the coating system. The most up-to-date, and hence the most expensive, coating or insulation systems gradually lose their effectiveness, and deteriorate and scale off sooner or later. It is well known that small coating defects can be more serious than major coating defects. This is because the anodic current density is much greater at the smaller coating defect. Moreover, small coating defects can result in pipelines or tank bottoms developing holes within a relatively short period of time. Thus, corrosion diagnostics is the basic determining factor of the effectiveness of corrosion protection.
657
Novel Techniques
It is universally accepted [4] that when a material corrodes in an electrolyte, specific regions of the substrate act either as an anode or a cathode, and in the case of localized corrosion the anodic and cathodic sites occur at distinctly isolated, separate areas and have equipotential lines associated with them. The equipotential lines can be measured using nonpolarizable microelectrode scanning in close proximity to the substrate or corroding surface. The surface potential maps produced will show a maximum or a minimum above the center of an anodic or cathodic area. As a result, the detected potential peak height can be correlated to the degree of corrosion occurring on the sample surface. The earliest application of the scanning reference electrode technique (SRET) was carried out by Evans [5], and Isaacs and Vyas [6]. A computerized SRET was then developed by Bassi et al. [7], to study the corrosion behavior of alloy samples in saline solution. Isaacs and Ishikawa [8] further advanced the SRET technique so it could detect much smaller currents by vibrating the tip of a microelectrode in a potential gradient (viz. scanning vibrating electrode technique (SVET). However, surface scanning with these instruments was inadequate, and the leveling of the specimen was critical, due to the effort involved and time spent adjusting and controlling the distance of the probe with respect to the specimen’s surface, especially when trying to detect an optimum potential signal that required a scanning height of ≤ 80 μm so that, the probe would not scrape and ruin the specimen. ESSENCE OF THE SURFACE CORROSION POTENTIAL MAPPING TECHNIQUE (SCM) Under the newly developed surface corrosion potential mapping technique (SCM) design, these difficulties were overcome by employing the probe design shown in Fig. 1. The scanning probe’s glass tube body containing a saturated calomel electrode (SCE) is mounted in a small automatic device incorporating a Z stepper motor in conjunction with a micrometer, a sensor switch and cam arrangement. As a result, a fixed distance in the range of 45 to 80 μm between the tip of the probe and the specimen surface has been achieved during the course of potential mapping. Therefore, the probe’s sensitivity relative to imperfections or raised points (e.g., paint blisters, embedded dust particle deposits and scratches) on the sample’s surface was increased during scanning. In addition, the flow of current associated with this probe was restricted by the base sensor so that a large portion flowed into the probe tip. It should be emphasized that the whole probe compartment and its unique sensor does not directly scan along the x and y directions. Instead it responds to the mechanical stage, x y movement of the flat specimen surface, by oscillating up and down at a fixed distance from the specimen surface profile being tested. The transmission of the output signal on the photodetector was by means of an electronic feedback loop coupled to the electronic circuit voltage. The electronic circuit and sensor are in charge of the probe detection and scanning, and function to keep the signal in the null position in where it is directly linked to the surface topography. In the present work, commercial quality coatings such as coal tar epoxy (MP1), modified epoxy (MP2), and abrasion resistant epoxy (MP3) (all half-contaminated with a 3% NaCl fine mist), have been studied by EIS and SCM. 658
Husain
These coatings are several times thicker than those usually described in the literature. The average thickness of the MP1, MP2, and MP3 systems were 450, 500, and 1000 μm, respectively. Furthermore, most of the reported research does not take into account the actual service-life of the paint thickness.
Probe (Calomel Electrode)
(Calomel Electrode) Black Plastic Box "Proximity Sensor" Z-Direction movement of Probe
X-Direction movement of tray
Red Indicator Green Indicator Scanning Electrode
Y-Direction movement of tray
"SCM" Plastic component Diameter "1.5mm" Sample
Tip Dia. 80 micron
Secondary Electrode (Remote)
Sensor Socket To OP-amp.
Electrolyte
Levelling screw
Figure 1. Schematic representation of the SCM technique The primary purpose of this corrosion accelerated-rating test would be to measure the overall rate of corrosion degradation of a commercial-quality-painted steel in a given environment by a simple, repetitive, nondestructive technique. This study was also necessary in order to compare the surface corrosion potential behavior of these paint systems. The test was expected to locate micropores and tiny blisters situated in any area of the painted panels. Such detection cannot be achieved with any other standard technique. In this way, mechanistic information and patterns of performance behavior can be built up with SCM/EIS methods long before visible signs of failure occur. EXPERIMENTAL PROCEDURE AC Impedance Measurement The impedance measurements were performed with an AC impedance frequency response analyzer (FRA), manufactured by ACM instruments. The applied potential signal amplitude was in the range of 10-150 mV and had a nominal frequency range of 10 kHz to 10 MHz. In all cases, the panel testing was conducted in three identical transparent Plexiglas cells. The setup comprised the painted panel with a large active test area of (24 cm2) consisting of both the contaminated zone, the C area, (substrate sprayed with 3% NaCl fine mist), and the uncontaminated zone, the UC area, tested together as one large area. The electrolyte used was 0.5 M NaCl with only ambient aeration during the immersion periods,
659
Novel Techniques
which ranged from 1 to 53 days. The AC impedance test data were taken at the open circuit potential using three electrode arrangement. Surface Corrosion Potential Mapping (SCM) Measurements In this part of the study, an attempt was made to perform SCM on the ,UC, area of the paint surfaces for the same paint systems (MP1, MP2, and MP3) following 53 days of immersion in 3% NaCl solution. The test was performed on the UC area mainly because there was no defect on the surface visible to the naked eye. SCM scanning was conducted at a fixed distance of 10 mm from the borderline separating the C and UC areas of the painted panel. The total area scanned was 100 mm2. The total number of recorded potential scanning steps was 2000, that is, x = 50 readings, and y = 40 readings. The total time taken per SCM scan was 1 hour, corresponding to 2000 steps. The SCM test solution was 3% NaCl, made at the ambient temperature. An equilibration immersion time of 30 minutes for each sample was used prior to making the electrochemical measurements. A schematic drawing of the panel is also included for the 1-cm 2 scanned area. The location of the UC area scanned in the SCM is indicated on the sketch as A and B for each of the paint systems. Following the SCM, selected samples were examined using JOEL T200 scanning electron microscope (SEM). The samples were cut and prepared to the recommended size. They were first coated with carbon, and then gold coated to render the surface conductive before microscopical examination. RESULTS AND DISCUSSION Figure 2a shows AC impedance data in the form of Bode plots. The data is associated with the coal tar epoxy (MP1) paint system half-contaminated with 3% NaCl fine mist. During the initial stages of immersion, the paint system appeared to be intact. Also, it had little apparent porosity or delamination (blisters) on the C area at later stages of immersion. An almost purely capacitive response was observed. This can be seen in the Bode plots corresponding to 24, 48, and 120 hours of immersion. Apparently, there was hardly any change during this period, suggesting that the system was relatively stable. Also there was very little water uptake. Therefore, the anticipated corrosion occurring under this paint would be extremely low. In fact, no visible signs of deterioration occurred, even with the presence of a surface stimulating agent (in the form of NaCl) beneath the coating. The impedance and capacitance data in Fig. 3a and b for coal tar epoxy are representative of the responses of coating having little or no additional effects due to Faradic 9 (corrosion) processes. These values, when examined, would be greater than > 10 for paint -11 film resistance (Rpf) and less than 10 -12
2
Ωcm along with a capacitance
2
(Cc) value approaching 10 Fcm . Such trends in the impedance spectra are typical of data obtained for thick, commercial-quality paints. It has long been established that Cc increases as the level of water in the coating increases [9]. Thus, the rate of increase in Cc should be proportional to the amount of area delaminated by wet adhesion loss. This is indeed what is seen in the results depicted in Fig. 3a and b and the previously discussed complex impedance plots. Figure 3a and b also shows that the Rpf values were higher for the thinner organic coating (i.e., for MP1 with dft = 450 μm compared to that of MP2 with dft = 500 μm). This
663
Novel Techniques
means that the incidence of defects in the coating was independent of the coating thickness. The reason for this is due to the failure of the coating at a single point, allowing the entry of the electrolyte. This lowers the resistance considerably (as in the case of paint MP2, and to a lesser extent paint MP1). If water invaded the coating more generally, an increase in capacitance would remain at a constant rate without appreciably changing the resistance due to the increased dielectric constant of the water ( as in the case of paint film MP3). Surface Corrosion Potential Mapping (SCM) for MP1 After 53 Days of Immersion The results of the SCM plots for each paint system are shown in Fig. 4. A schematic drawing of the panel is also included for the 1-cm2 scanned area. The location of the UC area scanned in the SCM is indicated in the sketch as A and B for each of the paint systems. The blue end of the spectrum is cathodic and the red/yellow end is anodic when presented in colored map. The potential scales attached to each map, are of the surface potential value in mV. What is remarkable is the SCM plot result in Fig. 4a for coal tar epoxy (MP1) made on UC area. The SCM map clearly shows that the surface of this painted sample exhibited a surface film free of conductive paths or active anodic pores (i.e., no red/yellow spots shown on the colored map). Moreover, the SEM qualitative analysis did not detect any iron or show any sign of surface-deposited corrosion by-products. This confirms the deduction made that MP1 had low electrochemical surface activities with no active anodic sites. Surface Corrosion Potential Mapping (SCM) for MP2 After 53 Days Immersion The SCM maps of Fig. 4b for the modified epoxy (MP2) show evidence of active anodic paths indicating the presence of micropores dispersed in the paint film. The SCM map also indicates that most of the anodic active paths originated in greater density from the right-hand side of the scanned surface. This is the region nearest the interface separating the UC area from the C area of the paint film. The elongated shape of these active regions (depicted by red and yellow in the colored map), could have been induced by the potential imposed by the contaminated substrate across the C area, and may be caused by a lateral aqueous diffusion under the paint film moving towards the UC area to the left of the borderline. These diffusion effects were also confirmed with visual observation of many blisters of different sizes in the UC area to the left of the borderline. In order to substantiate such findings, a further set of SCM scanning tests were conducted on an area which was suspected to have developed a blister. The location of the blister was just below the first scanned area and to the left of the borderline (see schematic illustration in Fig. 4 marked B). The SCM map shown in Fig. 4b displays the presence of large active anodic regions located between x = 2 mm and x = 9 mm on the x-axis and between y = 2 and y = 6 mm on the y-axis. This scanned area was then carefully cut and prepared for SEM examination. The SEM photomicrograph (Fig. 4d) clearly indicates the presence of a surface blister. This SEM photomicrograph confirms the observation made earlier with SCM on the same area in Fig. 4b. A question may be raised in this particular situation about the causes of the surface potential recorded with the SCM technique. The answer to this can be found by examining the SEM photograph of the same blister (top of the blister) at a higher magnification. The
664
Husain
SEM micrograph shows the presence of many micropores, in addition to the deposits of corrosion products film detected with SEM.
SCM maps area on the panel 10 cm "UC" area SCM map for MP2 1 cm
A
1 cm
1 cm
15 cm MP2
B
"C" area
Blister Located in area "B"
665
Novel Techniques
Figure 4. The SCM map with schematic drawings made for each scanned location (A, B), (a) MP1, (b) MP2, (c) MP3, (d) SEM of a blister detected with SCM in (b), (e) SEM of MP3 in (c) Surface Corrosion Potential Mapping (SCM) for MP3 After 53 Days Immersion Figure 4c shows the SCM results obtained for the abrasion-resistant epoxy (MP3) paints. The map indicates the presence of active anodic sites, which are characteristic of microporous paint films. It also shows deterioration trends as observed with the modified epoxy (MP2) paint, but lesser than with MP2, and thus some decrease in the number of active sites. SEM photographs taken from both the UC area as well as the C area (see Fig. 4e), reveal similar surface morphologies. No distinctive effects were noticed on the contaminated paint substrate in area C. Therefore, the active anodic site detected with SCM appears to be due to active anodic voids created by the natural solidification of the glass flake pigmentation used with such paint systems. Therefore, the porous morphological image observed with SEM is the normal surface microstructure of this paint film and was not caused by paint immersion degradation with the 3% NaCl solution. The formation of such a porous spongy morphology was due mainly to the glass pigment volume concentration being higher than the binder or vehicle (i.e., exceeded CPVC). The SCM electrochemical data also showed that, although a reactive anodic surface area was present for paint system MP3, no delamination processes occurred. Therefore, in this case study, a good-adhesion metal coating resulted. CONCLUSIONS 1.
The SCM developed in this study has significantly better resolution than earlier versions of the scanning reference electrode methods.
2.
SCM is a useful and convenient procedure by which characterizations of surface corrosion properties of different coating systems may be compared before visible defects manifest themselves. It also shows that large changes in the electrical properties of a coating occur locally when micro-defects or -pores (natural voids) are present in the coated substrate.
3.
AC impedance measurement is an ideal technique when combined with SCM to monitor coating behavior. Observations of both the capacitance and resistance of the commercial-quality paint films can be followed independently over time using the procedure adapted in this study.
4.
The micropores are believed to be responsible for the observed surface potential recorded by the SCM technique. Therefore, the observed surface potential corresponded to the partial anodic and cathodic currents originating from the electrolyte substrate bridging through tiny pores on the top of the blisters.
5.
EIS and SCM were in good agreement in showing the coal tar epoxy (MP1) and to a lesser extent the abrasion resistant epoxy (MP3) to be substantially more effective than the modified epoxy (MP2) as marine paint systems.
REFERENCES 1. 666
F. Mansfeld, M.W. Kendig and S.S. Tasi, Corrosion 38, 9, 1982.
Husain
2. 3. 4. 5. 6. 7. 8. 9.
K. Haldky, L.M. Callow and J.L. Dawson, Br.Corrs.J., 15, 1, 1980, p. 20. H. Leidhesier, Corrosion, 39, 5, 1983, p.189. H.S. Isaacs, Localized Corrosion, NACE, Houston, 1974, p. 158. U.R. Evans, The Corrosion and Oxidation of Metals, London, Arnold Ltd., 1960, pp. 615-617, pp. 860-866. H.S. Isaacs and N. Vyas, Electrochemical Corrosion Testing, (eds. F. Mansfield and U. Bertocii), ASTM STP-727, 1981, p. 3. R.S. Bassi, M.G. Hocking and S. Vasantasree, Proc. of the Nat. Corr. Conf. U.K. Corrosion 90, Swindon Park, U.K.2nd Oct 1990, p. 249. H.S. Isaacs and Y. Ishikawa, NACE, Corrosion/83, Anaheim, CA, 18-20 Apr. (1983), p. 25. D.M. Brasher and A.H. Kingsbury, J. Appl. Chem. 4, 1954, pp. 62-72.
667
Industrial Corrosion and Corrosion Control Technology Shalaby, H.M. et al. (Editors) 1996 Kuwait Institute for Scientific Research. Printed in Kuwait
OPTICS AND LASERS IN CORROSION LABORATORY K. Habib and F. Al-Sabti Materials Application Department Kuwait Institute for Scientific Research P.O. Box 24885, Safat 13109, Kuwait
ABSTRACT In the present paper, the advantages and limitations of holographic interferometry as a nondestructive method of materials testing in the corrosion field are presented. Data on the corrosion current density and the thickness of the oxide layer of anodized aluminum samples in aqueous solution (0.5-3.125% H2SO4) are given. The corrosion current density and the thickness of the oxide layer of anodized aluminum samples were measured by holographic interferometry as a function of elapsed time. In general, the results show that as the thickness of the oxide layer of the aluminum samples increases, the corrosion current density increases as well. Also, the results indicate that the optical interferometry technique is a more conservative technique for measuring the corrosion current density of aluminum samples in sulfuric acid than the weight loss method. Key Words: Holographic interferometry, corrosion current density, aluminum, sulfuric acid, Al2O3 oxide film
INTRODUCTION In recent work conducted by the author, a novel technique for monitoring the mechanochemical behavior, i.e., stress corrosion cracking, corrosion fatigue, and hydrogen embrittlement, of metallic electrodes in aqueous solutions has been developed [1-5]. The technique incorporates holographic interferometry for measuring microscopic deformation and electrochemical techniques for determining the corrosion current of metallic samples. In addition, the author has recently reported mathematical models describing the cathodic deposition and anodic dissolution of metals in aqueous solutions by holographic interferometry [6,7]. Furthermore, the technique of holographic interferometry has been applied to build an optical corrosion-meter for measuring the corrosion of different alloys as well as to evaluate different epoxy-based coatings applied to protect metals against corrosion [8,9]. Also, holographic interferometry is currently being utilized to monitor the pitting corrosion of several metals in aqueous solutions and to measure the formation of an oxide layer at a metal surface [10,11]. As a result, one may suggest that the techniques of holographic interferometry have many useful applications yet to be explored in the corrosion field. The objective of the present work is to demonstrate the advantages and limitations of holographic interferometry in the field of corrosion.
669
Novel Techniques
EXPERIMENTAL PROCEDURE Metallic samples of a pure aluminum (99.7%) were used in this investigation. The aluminum samples were fabricated in a cylindrical form 8 cm in diameter and 0.15 cm thick. Then, all samples were polished and ground with silicon carbide papers until the finest grade was reached. In order to be sure that the aluminum samples had a scratch-free surface, the samples were etched with a chemical solution for 2 minutes at a temperature that ranged between 85 and 95oC. The etching solution was made of 3 g/l of sodium hydroxide + 30 g/l of trisodium phosphate. At the beginning of each test, the aluminum sample first was immersed in the acid solution. Then a hologram of the sample was recorded using off axis holography (Fig. 1) for the optical setup. In this study, a camera with a thermoplastic film was used to facilitate recording of the holographic interferometry of the samples without removing the samples from the solutions. The camera is an HC-300 thermoplastic recorder made by the Newport Corporation. For more details on the procedures of the experiment, the reader is encouraged to refer to the literature elsewhere [7,9]. During each experiment, the holographic interferograms were recorded as a function of time, in which each test lasted for less than 60 minutes. Then, the interferograms were converted to an orthogonal displacement, i.e., an anodization process. Thereafter, the displacement measurements were used to determine the thickness of the anodized films of the samples in 0.5-3.125 % H2SO4 solutions. Finally the corrosion current density of the aluminum samples was determined according to the following mathematical relationship [7]: J = F |Z| du MT where
J K |Z| M T u d
(1)
is the corrosion current density of the metal, is Faraday's constant, is the absolute number of electron charge, is the atomic weight of the sample material, is the time of the anodic current, is the orthogonal displacement of the metal surface, and is the density of the metal.
In order to compare the obtained data of corrosion current density by optical interferometry with another technique, the corrosion current density of the same aluminum samples was also measured by the weight loss method. As a result a comparison between the corrosion current density taken by the two different techniques was made. RESULTS AND DISCUSSION Figures 2a, b, c, d, e, and f show an example of progressive interferograms of an aluminum sample anodized in 3.125% H2SO4 solution as a function of time. Figure 2a represents a real-time interferogram of the sample at the beginning of the test. Eight fringes appeared on the photograph which indicates that the aluminum sample had anodized (oxidized) as soon as it was immersed in the solution. Figure 2b is the same interferogram after two minutes of elapsed time. Fourteen fringes were detected on the photograph. It is obvious from this photograph that there is general chemical oxidation, depicted by the 670
Habib and Al-Sabti
uniform interferometric pattern. Figure 2c is the same interferogram after five minutes. Here 22 fringes were recorded on the photograph. Figure 2d is the same interferogram after 7 minutes, where 29 fringes recorded on the photograph. Figure 2e is the same interferogram after 12 minutes, and 36 fringes were recorded on the photograph. Figure 2f is the same interferogram after 15 minutes, with 57 fringes recorded on the photograph. Each fringe (dark line) in Fig. 2 (dark line) is equivalent to an orthogonal displacement of 0.3 mm according to mathematical models reported elsewhere [7,8]. In other words, holographic interferometry can be used as a powerful tool in the field of electrochemistry.
Figure 1. Optical setup of off-axis holographic interferometry 671
Novel Techniques
Figure 2. Progressive interferogram of aluminum sample in 3.125% H2SO4 as a function of time: (a) at the beginning of the test; (b) after 2 minutes; (c) after 5 minutes; (d) after 7 minutes; (e) after 12 minutes; and (f) after15 minutes By using the data from interferograms such as those in Fig. 2, one can determine the relationship between the thickness of the anodized layer (film) and the corrosion current density of the aluminum samples in different solutions. The data obtained on the thickness of the anodized layer (film) and the corrosion current density of the aluminum samples in different solutions by optical interferometry (OI) and the weight loss (WL) method are given in Table 1. It is obvious from the data in Table 1 that as the thickness of the oxide layer of the aluminum samples increases, the corrosion current density, by OI increases as well, with respect to the increase of the acid concentration from 1.0 to 3.125% H2SO4, but not for 0.5% 672
Habib and Al-Sabti
H2SO4. This behavior is interpreted by Faraday’s law in which the increase of the electronic current reflects the increase of the mass gain or the increase of the mass loss on the metal surface. In this case, it is obvious that there was mass gain on the aluminum samples because of the anodization process. Furthermore, the data in the table show clearly that the corrosion current density values determined by WL are independent from the increase of the thickness of the oxide layer. In fact, the corrosion current density values measured by WL are approximately equal to 1.0 mA/cm2 for all the investigated samples. In addition, the corrosion data in the table show that the OI technique is a more conservative technique for measuring the corrosion current density, of the aluminum samples in sulfuric acid than the WL method. This is because the OI technique measures the corrosion current density by a volume change rather than a weight change as in the WL method. In other words, the corrosion current density values measured by OI are higher than those measured by WL by a magnitude of 101. This occurred because the mass gain of the Al2O3 on the sample surface had a different density than the substrate metal (Al). It is obvious that the corrosion current density values determined by both methods would be the same if there were mass loss or mass gain without any drastic changes in the surface chemistry of the samples. In such cases, it would not make any difference if there was a volume change or a mass change on the metal sample due to the fact that the density of the metal (surface and bulk) remained nearly constant. Table 1. The Corrosion Current Density of Aluminum Samples in Different H2SO4 Concentrations Solution Concentration (H2SO4%) 0.500 1.000 1.500 3.125
Film Thickness
J by OI
J by WL
(μm) 44.15 9.00 31.50 36.20
(mA/cm2) 39.4 7.3 25.7 35.5
(mA/cm2) 0.80 0.83 0.76 0.65
REFERENCES 1. K.J. Habib, Holographic interferometry of a polarized and loaded metallic electrodes in aqueous solution, Applied Optics 29, 13, 1990, pp. 867-869. 2. K.J. Habib, Initial behavior of corrosion fatigue/hydrogen embrittlement of metallic electrodes in aqueous solutions, Experimental Techniques of Physics 38, 5/6, 1990, pp. 535-538. 3. K.J. Habib, G. Carmichael, R. Lakes and W. Stwalley, Novel technique for measuring stress corrosion cracking of metallic electrodes in aqueous solutions: Theory and applications, Corrosion 49, 5,1993, pp. 354-362. 4. K.J. Habib, Initiation of stress corrosion cracking of Ti90Al6V4 wire in aqueous solution: Non-destructive monitoring by holographic interferometry, Optics and Lasers in Engineering 20, 1994, pp.81-85.
673
Novel Techniques
5. K.J. Habib, Non-destructive evaluation of metallic electrodes under corrosion fatigue conditions by holographic interferometry, Optics and Lasers in Engineering 23, 1995, pp. 65-70. 6. K.J. Habib, Holograhic interferometry in predicting cathodic deposition of metals in aqueous solution., Proc. SPIE 1230, 1990, pp. 293-296. 7. K.J. Habib, Model of holographic interferometry of anodic dissolution of metals in aqueous solution, Optics and Lasers in Engineering 18, 1993, pp. 115-120. 8. K. Habib, Non-destructive evaluation of an epoxy-based coating by optical interferometry techniques, Optics and Lasers in Engineering 22, 1995, pp. 213-219. 9. K.J. Habib, F. Al-Sabti and H. Al-Mazidi, Optical Corrosion-Meter, Proceedings SPIE, Vol. 2577, 1995, pp. 210-217. 10. K. Habib and F. Al-Sabti, Monitoring pitting corrosion by holographic interferometry, Proceedings 9th Asian-Pacific Corrosion Control Conference, Kaohsiung, Taiwan, Nov.510, 1995, pp. 949-954. 11. K. Habib, A. Husain and F. Al-Sabti, Anodization of metals in aqueous solutions by holographic interferometry, Proceedings CORROSION and PREVENTION 95, Nov.1215, 1995, Perth, Australia, pp. 11/1-12.
674
Author Index
675
Abdullah, A., 493 Agarwal, D.C., 233 Al-Awad, M.N.J., 111 Al-Bahar, S., 341, 361 Al-Kandari, M.I., 263 AI-Kharafi, F.M., 371, 383, 417, 567 Alhajji, J., 135 Al-Hashem, A., 311, 493, 567 Ali, M., 323 Al-Matrouk, F., 567 Al-Muhanna, K., 149 Al-Omrani, K., 323 Al-Ramadhan, M.E., 597 Al-Rifaie, M.S., 395 Al-Sabti, F., 669 Al-Sumait, H., 493 Antonov, V.G., 209 Apparao, B.V., 483 Aride, J., 441 Ashworth, V., 1 Askari, A.M., 263 Ateya, B.G., 407 Attiogbe, E.K., 341, 361 Attou, A., 283 Badawy, W.A., 383, 417 Bass, C.J., 179 Bayyoumi, F.M., 407 Breslin, C.B., 395 Caceres, P.G., 311 Carew, J., 493 Cheruvu, L., 597 Dahab, A.S., 111 Denzine, A.F., 511, 647 El-Azab, A.S., 417 El-Dahshan, M.E., 111 El-Khafif, M., 567 Esmaili, S., 353 Farzam, M., 165 Fath-Allah, M.H., 383 Ferhat, M., 441 Galiullin, Z.T., 637 Gerbino, A.J., 127 Ghanem, W.A., 407 Giuliani, L., 521, 533
Golozar, M.A., 301 Gouda, V.K., 329 Habib, K., 669 Hajaj, S.A., 273 Hajji, F., 441 Halawani, S.M., 255 Hamalainen, E., 191 Hasan, A., 215 He, S., 127 Herda, W.R., 233 Hinrichsen, C.J., 127 Hosni, H., 555 Husain, A., 215, 657 Jain, A.K., 597 Kan, A.T., 127 Kane, R.D., 37, 89, 225 Karfoul, M.K., 431 Kertit, S., 441 Kinzel, A., 201 Kronborg Nielsen, P., 449 Lappin-Scott, H.M., 179 Lee, H.H., 323 Loubenski, S.A., 209 Macdonald, D.D., 17, 395 Masri, A., 555 Mazur, A., 245 McKinzie, M.J., 127 Moga, N., 461 Mostafa, M.S., 273 Mozafarinia, R., 301 Mukhopadhyay, P.K., 263 Nasrazadani, S., 501 Newman, M.B., 289 Oddo, J., 127 Oldfield, J.W., 67 Olsson, P.A., 289 Pakshir, M., 353, 471 Palaniswamy, N., 483 Paul, N.J., 555 Pchel’nikov, Yu.N., 637 Rais, A., 283 Rajani, G.L., 543, 581 Rajendran, S., 483 Reading, M.S., 511, 647
Riad, W.T., 329
Shams El Din, A.M., 49
Saarinen, K., 191, 627 Saario, T., 627 Saatchi, A., 301
Sikora, E., 395 Smamen, H., 283 Sovlukov, A.S., 637 Salman, M., 149 Srinivasan, S., 89 Sanders, P.F., 149, 179 Stockel, E., 461 Serban, R., 461 Tebbal, S., 225 Shalaby, H.M., 311, 329, 371 Tomson, M.B., 127 Shama, M.A., 615 Valliappan, M., 135
676
Subject Index
AC impedance, 659 Acceleration of corrosion, 45 Acid number, 101 Acidic soil, 472 Acoustic emission, 522 Aggregates, 362 (see Concrete) Aggressive areas, 456 Algorithm for damage prediction, 19 Alloy alloy 59, 235 aluminium 5083-O, 302 aluminum 6061, 507 (see Corrosion rate) copper, 87 Alloying elements, 235 Aluminum oxide film, 671 refining, 238 Aluminum-brass, 337 Ammonia, 78 Angle-drilled cathodic protection, 609 Anode, 551 Anticorrosive primer paints, 467 Application industrial, 538 pilot plant, 534 Aqueous environments, 628 Aquifer, 158 Atmospheric residue desulphurization, 274 Austenitic alloys, 234 Bacteriostatic test, 151 Bandar Bushehr seawater, 302 Basic science, 628 Bicarbonates, 99 Biocide effects of, 143 planktonic, 151 sessile, 152 Biofouling monitoring, 534 Biological corrosion, 325 Brackish water high pressure piping, 84 high pressure pumps, 84 Brine heater, 79 Chlorination, 291 679
recirculation pump, 559 scaling tendency, 128 Brittle fracture, 264 Bromine seawater, 75 Bulk carriers, 622 Calculated pH, 21 Calculated pitting damage functions, 31 Carbon steel (see Corrosion rate) carbon steel, 502 MEA solution, 498 Carbonation corrosion, 343 depth, 346 Cargo ships, 621 Cast alloys, 312 Catastrophic failures plant operation, 246 Cathodic protection angle-drilled, 609 grid-loop type, 609 implementation, 603 perimeter-based, 603 system design, 601 system monitoring, 550 (see Monitoring) Cavitation (see Rates of mass loss) rates of mass loss, 313 surface damage, 317 Concrete cover, 346 (see Carbonation depth) Cementing materials, 364 Charpy-V Specimen, 248 Chemical admixture, 365 Chemical composition artificial seawater, 354 (see Seawater) cast alloys, 312 soft tap water, 374 stainless steel, 396 weldments, 434 Chemical treatment, 586 Chloride content, 346 ion concentration, 423 induced corrosion, 342 Chlorides, 98, 105 copper base alloys in seawater, 70
Chlorine, effect of, 78 Citric acid production, 238 Cleaning acid, 63 sponge ball, 62 Coastal areas, 455 Coating application, 454 performance, 454 systems, 450 surveys, 546 Coefficient, retention, 409 Combination treatments, 590 Combustible products, 286 Concrete aggregates, 362 curing, 366 materials, 362 mix characteristics, 364 placement, 366 quality, 346 Condensed chemistry module, 20 (see module) Condenser tubes stainless steel, 58 titanium, 59 Condensers, 81 Condensing heat exchanger, 24 Conditions, aeration, 39 deaeration 39 Consequence corrosion, 617 factors, 7 rating, 7 Consequences, 5 Constant rate slow deformation, 211 Constant strain stress tests, 304 (see Test) Contact resistance, 627 (see Techniques) Cooling waters, 534 Copper alloys, 87 Copper base alloys, 70 (see Corrosion) Corrosion category, 455 carbon dioxide, 75 carbonation, 343 chloride, 342 condenser tubes, 55
copper-base, 55 cost, 2 dealloying, 54 detection, 549 effect of velocity, 69 fatigue, 524 management, 11 mass-loss under insulation, 44 monitoring, 12 nomogram, 92 polluted waters, 57 potentials, 22 processes, 45 product water, 76 resistance, 468 risk assessment, 5 seawater, 68 stainless steel, 58, 71 titanium,, 59 tube plates, 53 tube supports, 59 under insulation, 44 vapor-side, 60 Corrosion rate aluminum 6061, 507 carbon steel, 503 temperature, 39 yellow brass, 505 Corrosivity correlation, 229 prediction, 93 Cost corrosion failure, 3 corrosion in the state of Kuwait, 572 corrosion, 2, 570 prevention and control, 3 over-protection, 3 CP surveys, 546 CPL survey, 547 Crack advance vs temperature, 168 Crack rate vs. stress-intensity factor, 253 Cracking, 277 Cracks in welded joints, 251 Crevice corrosion, 54, 56 seawater, 78 titanium, 73 Crude
chemistry, 229 oil type, effect of 102 Damage
Filming type inhibitor, 589 Fine chemicals production, 238 Flow vs. shear stress relationships, 42
680
analysis, 206 module, 27 parameters, 29 algorithm for prediction, 19 (see Algorithm) deterministic prediction, 17 Demisters, 81 Design considerations, 617 Devanathan cell vs electrochemical testing, 170 Differential thermal analysis, 442 Dissolved carbon dioxide vs. corrosion, 583 Dissolved oxygen vs. corrosion, 583 Distillate pumps, 565 Dosing biocide, 155 ferrous sulphate, 61 Downhole equipment, 224 Economy, Kuwait, 568 Effect of corrosion, 622 Effluent, 160 treatment, 238 Ejectors, 81 Electrical resistance, 512 (see Measurements) Electrodeposition bath, 465 Electroplating, 632 Electro-vacuum test, 172 Emulsion paints, 461 Environment classification, 451 Environmental cracking, 522 Environments aqueous, 628 gaseous, 633 Erosion-corrosion, 53, 57, 293, 585 Evaluation, bacterial, 155 Evaporator, 80 Event rating, 8 Exposure test, 554 geometry, 42 (see Geometry of exposure) External corrosion, 598 Factors affecting corrosion, 616 Failure analysis, 246 Failure cost, 3 (see Cost of corrosion failure) Failures vs. oilfield, 2 Feed, 256 Film characterization, 466
Flue-gas characteristics, 21 Fractography, 280 Fundamental concepts, 18 Furnaces vs. corrosion investigation, 284 Galvanic corrosion, 294 Gas composition, effect of, 97 velocity, effect of, 100 Gaseous environments, 633 Geometry of exposure, 42 Grid-loop type cathodic protection, 609 Groundbed inspection, 551 Guidelines for repair, 350 (see Repair) Halogen emission vs. vapor-side, 74 Hardness, 286 Heat exchanger, 77 condensing, 24 (see Condensing heat exchanger) Heat recovery, 79 rejection, 77 transfer, 293 High pressure cooler, 256 warm separator, 275 High temperature oxidation, 633 Holiday potentials, 476 Holographic interferometry, 671 Hull structural failure, 616 Hydrogen blistering, 167 cracking, 524 diffusion, 168 embrittlement, 268 Hydrogen sulfide vs. temperature, 229 Impedance measurements, 388 Impedance response, 376 Implementation of cathodic protection, 603 Incoloy 800, 276 Inductive resistance, 517 Industrial applications, 240, 538 areas, 455 Inhibition, 102 corrosion allowance, 106 Inhibitor vs. squeeze design, 132 (see Squeeze)
Insulation joint test, 552 Integrity monitoring, 545 Interaction analysis, 639 Interference tests, 552 Intergranular corrosion tests, 305 (see Test)
integrity 545 pipeline corrosion, 545 process, 638 MSF distiller, 51 Multi-stage flash desalination, 76 681
Jet impingement, 70 Kuwait economy, 568 Laboratory testing, 37 Lack of ductility, 247 Lattice trapping, 166 Life factor, 6 Linear polarization, 515 Localized corrosion, 217, 296 Low temperature oxidation, 634 Luminescence spectroscopy, 484 Management strategy, 11 Management corrosion, 11 Marine areas, 456 Mass spectrometric analysis, 419 Mass-loss corrosion under insulation, 44 Materials of construction, 227 Measurements ac impedance, 659 corrosion rate, 649 electrical resistance 512 surface corrosion potential, 660 Mechanism vs. corrosion inhibition, 489 Metallography, 277 Metals processing, 238 Microbially induced corrosion, 291 Mitigation methods, 228 Mix water, 160, 362 Mixed potential module, 21 Model specifications, 568 corrosivity prediction, 93 transmission line, 26 Module condensed chemistry, 20 (see Condensed) damage function, 27 (see Damage function) mixed potential, 21 pit growth, 24 (see Pit growth module) pit nucleation, 22 (see Pit nucleation module) Monel vs. MEA solution, 498 Monitoring biofouling, 534 cathodic protection, 550 corrosion, 12 (see Corrosion monitoring)
Naphthenic acid corrosion parameters, 226 distribution, 229 Neutralizing type inhibitors, 587 Ni resist cast iron, 88 Nickel base alloys, 88 Nitric acid vs. corrosion behavior, 419 Nondestructive testing, 277 Oil tankers, 622 Oilfield failures, 2 Organic volatile oxygen scavengers, 586 Over-protection cost, 3 (see Cost) Oxidation high temperature, 633 low temperature, 634 Oxygen cu base alloys, 71 effects of, 103, 386 scavengers, 586 Paper mill, 630 Parameters vs. damage functions, 29 Performance test, 551 Perimeter-based cathodic protection, 603 Phase compositions vs. weldments, 434 Pilot plant application, 534 Pipeline corrosion monitoring, 545 Piping, 81 Pit depth, 378 diameter, 378 growth module, 24 nucleation module, 22 Pitting corrosion, 323 damage functions for 304 stainless steel, 30 Planktonic vs. time-kill tests, 153 Polarization behavior, 374 resistance, 390 tests, 303 (see Test) Polluted seawater, 290 (see Seawater) Pollution Control, 235 Portland cement, 362
Potential measurement, 551 corrosion, 22 (see Corrosion potentials) Potentiostatic polarization, 484 Power plant cooling waters, 534 Prevention and control cost, 3 Probability of failure, 622 (see Effect of)
Scenario vs. worst-case, 38 Seawater artificial, 354 chlorination, 61 circulation pumps, 562 corrosion, 68 high pressure piping, 83
682
Probability, 5 Procedures vs. simulated environment, 39 Process monitoring, 638 operating, 60 Protection measures, 344 Pump blowdown, 82 brine recirculation, 559 distillate, 82, 565 linkage, 201 seawater circulation, 562 seawater supply, 556 Rates of corrosion, 617 Rates of mass loss vs. cavitation, 313 Reactor, 264 Rectifier operation, 551 Reduced solvent emissions, 450 Relationships vs. flow and shear stress, 42 Repair of corrosion damage, 344 Replenishment vs. environment, 44 Research shuttle laboratory, 194 Resistance of steel, 211 (see Constant rate) Retention coefficient, 409 (see Coefficient) Reverse osmosis vs. selection of materials, 83 Rheology, 117 Risk, 4 assessment, 5 (see Corrosion risk assessment) categories, 10 classes, 9 equation, 9 matrix, 9 modification, 9 Rotating cage, 46 disc electrode, 495 Saline alkaline soil, 472 Sand, effect of, 78 Saturation index, 129 Scaling, 61 Scanning electron microscopy, 444
high pressure pumps, 83 pollutants, 338 polluted, 290 pumps, 82 supply pumps, 556 recovery turbines, 83 Sectorial aggregation, 570 Segregation bands, 249 Selection of monitoring systems, 554 Service failure, 337 Simulated-environment tests, 38 Slow strain rate tests, 304 (see Test) Slow-wave structures, 638 Sodium attack, 585 Soil analysis, 472 Squeeze simulation, 130 Stainless steel, 87, 143 austentic, 87 duplex, 88 seawater, 71 superaustenitics, 88 vs. biological corrosion, 325 vs. chloride environment, 323 vs. MEA solution, 499 Steam condensate system, 586 (see Chemical treatment) leaks vs. cost, 582 generator tubes, 629 Strategic considerations, 11 Stress corrosion cracking, 56, 295, 531 Sulfur, 105 Sulfur content, 227 Sulfuric acid, 671 corrosion resistance, 240 Sulphate-reducing bacteria, thermophilic, 181 Sulphide assay, vs. spectrophotometric, 181 production, vs treatment chemicals, 182 production, vs. surface area, 181 production, vs. thermophilic bacteria, 182 Sulfides, effect of, 77
Surface corrosion potential mapping, 658 Surface examination, 484 Synthesis of acrylates and methacrylates, 237 Tactical considerations, 12 Techniques contact electric resistance, 627 surface potential mapping, 658 Temperature, 98, 227 Temperature sensitivity, 648, 650
boxes, 52, 79 resistance of latices, 468 thinnable products, 461 vapor permeability, 468 wash role, 258 Waterborne coatings, 454 (see Coatings) products, 462 Water-cement ratio, 364 683
Tensile test specimens vs. dimensions, 194
Weight loss Test hydrocarbon/liquid phase, 43 bacteriostatic, 151 method, 484, 494 constant strain stress corrosion, 304 Weldments, 434 insulation joint, 552 Worst-case scenario, 38 interference, 552 X-ray diffraction, 442 intergranular corrosion, 305 technique, 484 performance, 551 Yellow brass, 505 (see Corrosion rate) planktonic time-kill, 153 polarization, 303 simulated-environment, 38 time-kill, 151 Testing vs. laboratory, 37 Time-kill test, 151 Titanium, 73, 87 Tool joint, 204 Total acid number, 226 Transformer operation, 551 Transmission line model, 26 Trap energy, 166 Treatment of process condensate, 590 Troubleshooting, 258 Tube plates, 79 Turbine condensate, 590 , 595 (see Combination) Types of corrosion, 256, 583 oilfield treatment chemicals, 186 Vanadium vs. effect of temperature, 392 Vapor side behavior of materials, 74 corrosion, 60, 74 Vapor spaces, 80 Velocity, 99, 227 vs. type of flow, 106 Ven diagram vs. stress corrosion cracking, 280 Venting, 81 Wall shear stress, 230 Wash water specifications, 276 Water
684
