Jun 30, 2012 - Figure 22 Test Section #18 â Next Generation Diamond Grinding . ...... This treatment serves as a good bench mark to compare all other ...... Presentation to 2006 Airfield Operations Area Expo and Conference, Milwaukee,.
Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool – Phases 1 & 2 Final Report OTCREOS7.1-16/ OTCREOS9.1-21 June 30, 2012 Musharraf Zaman, PhD, PE Principal Investigator Douglas D. Gransberg, PhD, PE Co-Principal Investigator Caleb Riemer, PhD, PE Research Associate Dominique Pittenger, PhD Research Associate Bekir Aktaş, PhD Research Associate The University of Oklahoma School of Civil Engineering and Environmental Science Three Partners Place, Suite 150 201 David L. Boren Blvd. Norman, Oklahoma 73019-5300 (405) 325-6092 i
DISCLAIMER The contents of this report reflect the views of the authors, who are responsible for the facts and accuracy of the information presented herein. This document is disseminated under the sponsorship of the U.S. Department of Transportation University Transportation Centers Program, in the interest of information exchange. The U.S. Government assumes no liability for the contents or use thereof. i
TECHNICAL REPORT DOCUMENTATION PAGE 1. REPORT NO.
2. GOVERNMENT ACCESSION NO.
3. RECIPIENTS CATALOG NO.
4. TITLE AND SUBTITLE
5. REPORT DATE
Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool – Phases 1 & 2
30 June 2012
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT
6. PERFORMING ORGANIZATION CODE
Musharraf Zaman, Douglas Gransberg, Caleb Riemer, Dominique Pittenger, and Bekir Aktas 9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. WORK UNIT NO.
The University of Oklahoma, Three Partners Place, Suite 150, 201 David L. Boren Blvd Norman, Oklahoma 73019-5300
11. CONTRACT OR GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Oklahoma Transportation Center, 202 West Boyd Street, Room 107 Norman, OK 73019
Final June 2008 to June 2012
OTCREOS7.1-16 & 9.1-21 14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The objective of the study is to build on the research done overseas and conduct a comparative field evaluation of various methods used to restore pavement skid resistance by retexturing the existing surface with either a surface treatment, chemical treatment or a mechanical process. In Phase 1, 16 field test sections were constructed on State Highway 77 between Oklahoma City and Norman. Monthly microtexture and macrotexture measurements were taken over a period of 22 months. The field data was reduced to create deterioration models based on loss of both micro and macrotexture over time. The models were then used to calculate effective service lives for each treatment which was then used as input for a life cycle cost analysis. A new lifecycle cost analysis model for pavement preservation treatments based on equivalent uniform annual cost rather than net present value was developed and is used to process the pavement texture change data. This will allow pavement managers to have the required information to be able to make rational engineering design decisions based on both physical and financial data for a suite of potential pavement preservation tools. Each treatment alternative has been evaluated under the same conditions over the same period of time by an impartial research team. The project continued for a third year under the Phase 2 OTCREOS9.1-21 contract. A pavement preservation treatment toolbox for a total of 23 different treatments was developed and furnished to ODOT for use by its division maintenance engineers in the state-wide pavement preservation program.
17. KEY WORDS
18. DISTRIBUTION STATEMENT
Pavement management, skid resistance, life cycle cost analysis, retexturing, maintenance, pavement preservation 19. SECURITY CLASSIF. (OF THIS REPORT)
20. SECURITY CLASSIF. (OF THIS PAGE)
ii
21. NO. OF PAGES
228
22. PRICE
iii
Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool – Phases 1 & 2 Final Report OTCREOS7.1-16 & 9.1-21 June 30, 2012 Musharraf Zaman, PhD, PE Principal Investigator Douglas D. Gransberg, PhD, PE Co-Principal Investigator Caleb Riemer, PhD, PE Research Associate Dominique Pittenger, PhD Research Associate Bekir Aktaş Research Associate Under the Supervision Of Arnulf P. Hagen, PhD Technical Director, Oklahoma Transportation Center Oklahoma Transportation Center (OTC) 2601 Liberty Parkway, Suite 110 Midwest City, Oklahoma 73110 June 2012
iv
TABLE OF CONTENTS Table of Contents ............................................................................................................ v List of Figures ................................................................................................................. vii List of Tables ...................................................................................................................xi Executive Summary ....................................................................................................... xii Introduction ..................................................................................................................... 2 Importance of Pavement Surface Texture ....................................................................... 9 Life Cycle Cost Analysis ................................................................................................ 38 Field Test Results........................................................................................................ 103 Analysis of Project ....................................................................................................... 127 Conclusions................................................................................................................. 133 Recommendations ...................................................................................................... 135 Implementation/Technology Transfer .......................................................................... 136 References .................................................................................................................. 141 APPENDIX - Pavement Surface Texture Maintenance Guide .................................... 150 Table of Contents ............................................................................................................ v INTRODUCTION ............................................................................................................. 6 Pavement Maintenance Background............................................................................... 7 Pavement Texture Background ..................................................................................... 11 PAVEMENT TEXTURE MEASUREMENT .................................................................... 12 DETERIORATION MODELING ..................................................................................... 19 COST ANALYSIS .......................................................................................................... 22 REFERENCES .............................................................................................................. 49 APPENDIX A Pavement Preservation Treatment Descriptions..................................... 52 v
APPENDIX B Primary Capabilities and Functions of Pavement Preservation Treatments ...................................................................................................................................... 65 APPENDIX C Pavement Preservation Treatment Values (2010) .................................. 67 APPENDIX D Skid Number Cost Index Analysis (2010) ............................................... 68
vi
LIST OF FIGURES Figure 1: Pavement Surface Microtexture and Macrotexture (Pidwerbesky et al. 2006)11 Figure 2: Pavement Friction Model (Hall, 2006) ............................................................ 12 Figure 3 Pavement Texture Definitions (Evans 2002) ................................................... 14 Figure 4 Map of Research Test Sections ..................................................................... 20 Figure 5 Test Section #1 –Calupave Seal .................................................................... 21 Figure 6 Test Section #2 – Open Graded Friction Course w/ Fog Seal........................ 21 Figure 7 Test Section #3 – Open Graded Friction Course ............................................ 22 Figure 8 Test Section #4 – 1″ Mill and Inlay .................................................................. 22 Figure 9 Test Section #5 – Fog Seal ............................................................................. 23 Figure 10 Test Section #6 – Asphalt Penetrating Conditioner ....................................... 23 Figure 11 Test Section #7 – Asphalt Planer w/ Asphalt Penetrating Conditioner .......... 24 Figure 12 Test Section #8 – Chip Seal (5/8″) w/ Fog Seal ............................................ 24 Figure 13 Test Section #9 – Chip Seal (5/8″) ................................................................ 25 Figure 14 Test Section #10 – Chip Seal (3/8″) .............................................................. 25 Figure 15 Test Section #11 – Shotblasting (Blastrac) ................................................... 26 Figure 16 Test Section #12 – Shotblasting (Skidabrader) w/ Fog Seal ......................... 26 Figure 17 Test Section #13 – Shotblasting (Skidabrader) ............................................. 26 Figure 18 Test Section #14 – Novachip ........................................................................ 27 Figure 19 Test Section #15 – Shotblasting (Blastrac) ................................................... 27 Figure 20 Test Section #16 – Shotblasting (Skidabrader) ............................................. 28 Figure 21 Test Section #17 – Shotblasting (Blastrac) w/ Densifier................................ 28 Figure 22 Test Section #18 – Next Generation Diamond Grinding ............................... 29 Figure 23 Test Section #19 – Diamond Grinding .......................................................... 29 Figure 24 Test Section #20 – E-Krete ........................................................................... 29 Figure 25 Test Section #21 – Chip Seal (Single Size)................................................... 30 Figure 26 Test Section #22 – Chip Seal (Single Size) w/ Calupave .............................. 30 Figure 27 Test Section #23 – Microsurface ................................................................... 31 Figure 28 ODOT Skid Trailer......................................................................................... 32 Figure 29 Sand Circle Tests and Outflow Meter/Sand Circle Testing in Progress ........ 33 Figure 30 Circular Track Meter...................................................................................... 35 vii
Figure 31 RoboTex Device (top) with Example Output (bottom) ................................... 36 Figure 32 Change in Wheelpath Surface Characteristics Over Time ............................ 38 Figure 33 Proactive Approach versus Reactive Approach (Davies and Sorenson 2000) ...................................................................................................................................... 40 Figure 34 Probability Distribution Graphic and Equations (FHWA 1998) ...................... 47 Figure 35 Computation of NPV using probability and simulation (FHWA 1998) ............ 48 Figure 36 Research Approach ...................................................................................... 51 Figure 37 Pavement Preservation EUAC LCCA Model Logic ....................................... 55 Figure 38 Analysis Period in EUAC for Unequal-Life Alternatives................................. 56 Figure 39 Service Life as Analysis Period ..................................................................... 57 Figure 40 Anticipated Service Life ................................................................................. 59 Figure 41 Two Methods, Same Preferred Alternative .................................................. 60 Figure 42 Do-Nothing Scenario vs. Fill-the-Gap Scenario............................................. 61 Figure 43 1” Hot Mix Asphalt Mill & Inlay Microtexture Deterioration Model .................. 64 Figure 44 Open Graded Friction Course Macrotexture Deterioration Model ................. 65 Figure 45 Open Graded Friction Course Microtexture Deterioration Model .................. 66 Figure 46 5/8” Chip Seal Macrotexture Deterioration Model ......................................... 67 Figure 47 5/8” Chip Seal Microtexture Deterioration Model........................................... 67 Figure 48 Pavement Retexturing - Shotblasting Microtexture Deterioration Model ....... 68 Figure 49 Pavement Retexturing - Abrading Microtexture Deterioration Model ............ 68 Figure 50 LCCA Step 1: Establish Alternatives and Each Service Life ......................... 69 Figure 51 LCCA Step 2: Determine Activity Timing ...................................................... 70 Figure 52: LCCA Step 3: Determine Agency Costs, Construction................................. 71 Figure 53 LCCA Step 3: Determine Agency Costs, Maintenance ................................. 71 Figure 54 LCCA Step 3: Determine User Costs ........................................................... 72 Figure 55 LCCA Step 3: Determine User Costs ........................................................... 73 Figure 56 LCCA Step 3: Determine User Costs (Continued) ....................................... 73 Figure 57 Illustration of EUAC Calculation for Chip Seal............................................... 74 Figure 58 LCCA Step 4: Calculate EUAC, Asphalt Pavement Treatments ................... 75 Figure 59 EUAC Comparison, Asphalt Pavement Treatments ...................................... 76 Figure 60 LCCA Step 4: Calculate EUAC, Concrete Pavement Treatments ................. 77 viii
Figure 61 EUAC Comparison, Concrete Pavement Treatments ................................... 78 Figure 62 LCCA Step 4: Calculate EUAC, Terminal State ............................................ 79 Figure 63 LCCA Step 4: Compare EUAC, Terminal State ............................................ 80 Figure 64 LCCA Step 5: Service Life Sensitivity, Terminal – Microtexture-Based......... 82 Figure 65 LCCA Step 5: Service Life Sensitivity, Terminal – Macrotexture-Based ....... 82 Figure 66 LCCA Step 5: Pavement Life Extension Sensitivity, Terminal State ............. 83 Figure 67 LCCA Step 5: Service Life Sensitivity Analysis, Microtexture-Based ........... 84 Figure 68 LCCA Step 5: Discount Rate Sensitivity Analysis, Microtexture-Based ....... 85 Figure 69 LCCA Step 5: Service Life Sensitivity Analysis, Empirical Data-Based......... 85 Figure 70: LCCA Step 5: Discount Rate Sensitivity Analysis, Asphalt Pavement PPTs ...................................................................................................................................... 86 Figure 71 LCCA Step 5: Discount Rate Sensitivity Analysis, Concrete Pavement PPTs ...................................................................................................................................... 86 Figure 72 Stochastic EUAC Input Screen, Continuous Mode. ...................................... 88 Figure 73 Stochastic EUAC Input Screen, Terminal Mode............................................ 89 Figure 74 Chip Seal Service Life Probability Distributions, Continuous (left) and Terminal (right) Modes. ................................................................................................. 90 Figure 75 Stochastic EUAC LCCA Approach. (Adapted from FHWA 1998).................. 92 Figure 76 PPTs Stochastic EUAC Probability Distributions........................................... 92 Figure 77 Summary Statistics for Hot Mix Asphalt EUAC Distribution........................... 93 Figure 78 EUAC Simulation Output. .............................................................................. 99 Figure 79 EUAC Model Output Converted to NPV-Form. ........................................... 100 Figure 80 Correlation Error Check in Stochastic Model............................................... 101 Figure 81 OGFC Test Sections (#2 and #3), Macrotexture Deterioration.................... 105 Figure 82 5/8” Chip Seal Test Sections (#8 and #9), Macrotexture Deterioration ....... 105 Figure 83 Chip Seal (3/8”) Actual Macrotexture Graph ............................................... 108 Figure 84 Chip Seal (3/8”) Extrapolated Macrotexture Graph ..................................... 109 Figure 85 Fog Seal Texture Combination Graph ......................................................... 113 Figure 86 Mill and HMA Inlay Texture Combination Graph ......................................... 114 Figure 87 Shotblasting (Blastrac) Texture Combination Graph ................................... 115 Figure 88 Deterioration Models for Chip Seal (5/8″) .................................................... 116 ix
Figure 89 Impact of Service Life Method on Chip Seal EUAC ................................... 123 Figure 90 Deterministic and Performance-Based Chip Seal EUACs on Stochastic Empirical-Service-Life Cumulative Curve. ................................................................... 124 Figure 91: Test Sections Macrotexture Results and Differences of Two Test Methods in November 2009 ........................................................................................................... 128 Figure 92: Average Macrotexture Results and Differences between Two Test Methods Over 24 Months ........................................................................................................... 129 Figure 93: Macrotexture Percentage Differences between Seconds in Outflow Meter Test ............................................................................................................................. 130 Figure 94: Theoretical Curves of Outflow meter and Sand Circle Tests ...................... 131 Figure 95 Macrotexture Measurements Using Different Techniques .......................... 133
x
LIST OF TABLES Table 1 Oklahoma Pavement Preservation Test Sections .............................................. 5 Table 2 Pavement Preservation Guidelines (Geiger 2005) ............................................. 7 Table 3 Asphalt and Concrete Test Sections ............................................................... 19 Table 4 ODOT Survey Responses ................................................................................ 53 Table 5 Pavement Treatment Service Life Estimation................................................... 62 Table 6 Treatment Service Life Based on Extrapolated Field Data ............................... 69 Table 7 Average Pavement Treatment Cost ................................................................. 70 Table 8 Treatment Agency Costs Used in EUAC LCCA. .............................................. 88 Table 9 Goodness-of-Fit for PPTs. ................................................................................ 91 Table 10 Stochastic EUAC Results, Continuous Mode. ................................................ 93 Table 11 Stochastic EUAC Sensitivity Analysis. ........................................................... 95 Table 12 Stochastic EUAC Results, Terminal Mode. .................................................... 96 Table 13 Comparable EUAC and PV Rankings, Continuous State ............................... 97 Table 14 EUAC and PV Results (no user costs), Terminal State at 6yrs ...................... 98 Table 15 Average MTD values for chip seal test sections. .......................................... 102 Table 16 External Validation Model Results. ............................................................... 103 Table 17 Summary of Regression-Based Treatment Deterioration Models ................ 106 Table 18 TNZ P/17 Performance Specification Comparison ....................................... 111 Table 19 Life Cycle Cost Analysis Results for Field Treatments ................................. 119 Table 20 Deterministic LCCA Results. ........................................................................ 120 Table 21 Deterministic and Stochastic EUAC Comparison. ........................................ 120 Table 22 Performance-Based Service Life Values for PPT Field Tests ...................... 121 Table 23 Performance-Based Stochastic EUAC Results. ........................................... 122 Table 24 LCCA Sensitivity to LCCA Selection. ........................................................... 125 Table 25 Skid Number Cost Index for Pavement Preservation Treatments ................ 126
xi
EXECUTIVE SUMMARY The objective of the study is to build on the research done in Australia (ARRB, 2001) and New Zealand (TNZ, 2002) and conduct a comparative field evaluation of various methods used to restore pavement skid resistance by retexturing the existing surface with either a surface treatment, chemical treatment, mechanical process or some combination of two of the previous three options. The goal is to assemble the technical engineering data for each treatment coupled with an economic analysis of the costs and benefits associated with each treatment. This will allow pavement managers to have the required information to be able to make rational engineering design decisions based on both physical and financial data for a suite of potential pavement preservation tools. Each treatment alternative will have been evaluated under the same conditions over the same period of time by an impartial research team.
Pavement preservation is the embodiment of infrastructure stewardship. Its central theme is using pavement technology to “keep good roads good” (Galehouse 2003). There is a wide range of funding for pavement preservation and maintenance programs within the U.S., ranging (on average) from a low of $15.0 million to a high of $1.7 billion per year (Tighe and Gransberg 2010). In those agencies, like the Oklahoma Department of Transportation (ODOT) that are on the low end of the funding spectrum, the need for an aggressive pavement preservation program is critical to getting as much value out of each maintenance dollar as possible (Galehouse 2003, NHI 2007). Pavement preservation is inherently sustainable as it seeks to minimize the amount of natural resources consumed over a pavement’s life cycle (Geiger 2005). Therefore, focusing on pavement preservation rather than reactive maintenance and repair furnishes a broad foundation on which to build ODOT’s pavement sustainability program.
The research has already attracted much favorable interest in Oklahoma and nationally. Consequently, both ODOT engineers and members of the highway industry have come forth and requested that the project be expanded to add additional test sections, new testing methods, and extended to cover a longer period of observation and analysis to xii
significantly enhance its impact. Therefore, the research described in this report builds on the work that was completed in the first year of project OTCREOS 7.1-16 (Phase 1). As part of this research project, a series of test sections were constructed on existing asphalt and concrete pavement sections on State Highway 77H (Sooner Road) between Norman and Oklahoma City, Oklahoma. Each test section is ¼ miles (400 meters) long and one lane wide. Each section has been retextured with a different type of pavement preservation process. There are a total of 23 different treatments that are covered in the research. Of these, 14 were installed during the summer of 2008 as part of OTCREOS7.1-16 (Phase 1), which is now complete. The remaining nine sections were installed during the summer of 2009 as a part of OTCREOS9.1-21 (Phase 2). Because the two projects are integrated, it is difficult and confusing to attempt to discuss them as separate projects. Additionally, the test sections installed as part of Phase 1 were sampled on a monthly basis for macrotexture and skid along with the Phase 2 test sections. Surface friction and pavement macrotexture were measured on each test section before construction of the treatments and continue to be measured on a monthly basis for three years after application. Thus, changes in both skid resistance and pavement macrotexture are recorded over time, and each treatment’s performance can then be compared to all other treatments in the same traffic, environment, and time period. Additionally two new methods for quantitatively characterizing the test section surfaces are added in the Phase 2 project: a dynamic friction tester (DFT) and ASTM E2157 Circular Track Meter (CTM).
ASTM E1845 can be used to measure macrotexture,
along with a high speed laser to determine mean profile depth at highway speed. With high speed data collection and the methodology for texture deterioration modeling, a pavement management engineer now has the tools to better incorporate surface texture safety criteria into the pavement management system (PMS). Adding these to the TNZ T/3 sand circle test and the ASTM STP 583 volume outflow meter implemented in Phase 1, the test sections were evaluated with four different methods to measure macrotexture and provide definitive guidance to ODOT pavement managers as to the strengths and weaknesses of each method over an extended period in field conditions.
xiii
By conducting a long-term study of various methods to restore pavement skid resistance by retexturing on the full spectrum of FHWA-approved pavement preservation treatments, this project furnishes ODOT with the technical engineering data for each treatment coupled with an economic analysis of the costs and benefits associated with each treatment. This gives ODOT pavement managers the required information to make rational engineering decisions based on physical and financial information for the use of potential pavement preservation tools, evaluated under the same conditions, over the same period, by an impartial investigator.
The project’s major deliverable is a pavement surface texture maintenance guide that can be used by ODOT pavement managers to restore surface texture and skid resistance to various types of pavements throughout the state. It constitutes a surface retexturing “toolbox” that contains both the technical engineering information and the economic analysis of each treatment’s efficacy. The idea is not to identify the “best” method but rather to quantify the benefits of all the treatments in a manner that then allows a pavement engineer to select the right pavement preservation “tool” for the specific issue that they need to address and satisfy the fundamental definition of pavement preservation: “put the right treatment, on the right road, at the right time.”
This project demonstrates a robust partnership between the Oklahoma Transportation Center (OkTC), the University of Oklahoma (OU), the Oklahoma Department of Transportation (ODOT), and members of the pavement preservation industry from six states. This project is not a competition between products. It is the start of an encyclopedia of pavement preservation comparative analyses, and projects of this nature could be instituted throughout the U.S. to provide the unique local performance information that only long-term field testing can generate. It is demonstrating the benefits of pavement preservation materials and means and methods in a manner that will not only be of value to ODOT and other Oklahoma public agencies but also to the rest of the nation. All the test sections were donated by either ODOT or the pavement preservation industry, and the fact that over $400,000 worth of pavement preservation xiv
treatments were donated as well as the in-kind contributions by ODOT in providing traffic control, skid testing, and engineer’s time, shows the importance of pavement preservation research.
The project has drawn both national and international interest. Caleb Riemer, PhD, PE, was appointed to the FHWA Pavement Preservation Expert Task Group (PPETG), a rare honor for such a young person. Dominique Pittenger, PhD, was the only MS student of nine University Transportation Center graduate research assistants recognized by the Transportation Research Board (TRB) as the top graduate researchers in the nation. The other nine were doctoral candidates. She went on to write and win a TRB grant to support her doctoral work in extending this project’s results and methodology to airport pavements. Also, she was named the 2010 OTC Student of the Year, which is a very prestigious award. She was recently appointed to the TRB Committee on Pavement Preservation (AHD18). As a result of PPETG presentations made in New Orleans and Reno by Drs. Riemer and Pittenger, the Southeastern Pavement Preservation Partnership held its 2011 conference in Oklahoma City so that the OTC project could act as the conference centerpiece and possibly spawn similar research in other states.
Drs. Gransberg and Zaman were invited to submit and present a paper on the project at the 1st International Conference of Pavement Preservation sponsored by TRB and the Federal Highway Administration (FHWA). After that event Dr. Gransberg was appointed as a peer of Natural Sciences and Engineering Research Council of Canada, that nation’s version of the US National Science Foundation, and he was subsequently invited to present this OTC project at an international airport pavement conference in Toronto and to deliver a guest lecture at the University of Waterloo in Ontario. He made invited presentations in August 2010 at the Universities of Canterbury and Auckland in New Zealand. He was recently invited to serve on a panel that is evaluating the adequacy of the Ministry of Transportation in Ontario Canada's chip seal performance specification.
xv
Sarah Brockhaus and Rande Patterson were awarded grants to conduct research using the project’s data by the Oklahoma Asphalt Paving Association in 2011 and 2012, respectively. The award funded a trip for Sarah to collect data and work in Dr. Susan Tighe’s green pavements lab at the University of Waterloo in Canada. The award funded a trip for Rande Patterson to Washington DC to talk with other researchers in the pavement preservation field at the TRB Annual Meeting. Finally, Bekir Aktas, an ABD doctoral candidate in pavement engineering at Sulieman Demirel University in Isparta, Turkey, was awarded a year-long research-abroad fellowship by his government to join Drs. Gransberg and Zaman’s research team on this and other OU projects in the area of pavement preservation. Finally, articles on this project have been featured in Roads and Bridges and the Journal of Pavement Preservation.
To briefly summarize, this OkTC project has gone beyond its expectations and made Oklahoma the clear leader in pavement preservation research at a period in our nation’s economy where DOTs must wring every last bit of value out of their pavement management budgets. It has spun off research partnerships with engineering schools in Canada and Turkey, which will furnish a very rich environment for Oklahoma students to study pavement preservation in the future. The OkTC project’s student research assistants have won two national awards, two state awards, and three appointments to national technical committees. The project was selected by the Western Association of State Highway and Transportation Officials (WASHTO) as its candidate for the AASHTO recognition as the nation’s most influential state-funded research project and as a result of that recognition, this OkTC project was featured in a special informational report developed by the US Congress to demonstrate the return on investment in statefunded transportation research.
xvi
INTRODUCTION Pavement skid resistance is perhaps the most important engineering component of the road from a safety standpoint. Slippery pavements are the result of several causes, chief of which is the loss of pavement surface micro and macrotexture. A European study found that increasing the pavement’s macrotexture not only reduced total accidents under both wet and dry conditions, but also reduced low speed accidents (Roe et al., 1998). As a result, pavement managers must not only manage the structural condition of their roads but also their skid resistance (Gee 2007, NCHRP 1989). In fact, it is possible for a structurally sound pavement to be rendered unsafe from a loss of skid resistance due to polishing of the surface aggregate, or in the case of chip seals, flushing of the binder in the wheel paths (Patrick et al. 2000). This results in a safety requirement to modify the pavement surface to restore skid resistance. Many of the possible tools for restoring skid resistance, like chip seals, are also used for pavement structural preservation. Thus, it seems that maintenance of adequate pavement skid resistance is also a pavement preservation activity (Moulthrop 2003). This intersection of two requirements creates a technical synergy that a state like Oklahoma can leverage to stretch its pavement maintenance budget if it has the necessary technical and financial information to assist decision-makers in selecting the appropriate surface treatment tool for a given situation.
PROBLEM There is a wealth of information on skid resistance in the literature (Voight 2006, Patrick et al. 2000, Roe et al. 1998, Henry 2000). However, most of the previous research has been in the safety realm developing the relationship between skid resistance and crashes. There is also a wealth of information on pavement surface treatments (NCHRP 1989). However, a majority of previous studies has been in the laboratory and focused on the material science aspect. Very little substantive work has been done in the field regarding surface treatment performance, and most of the research in this area is focused on short-term performance (Owen 1999). The FHWA Long Term Pavement 2
Performance Program (LTPP) collects friction data as part of its standard protocol (Titus-Glover and Tayabji 1999). However, the LTPP data largely relates to pavement mix design criteria and while it includes data for chip seals, it does not collect data for any of the other potential pavement preservation treatments. Additionally, the typical research project only examines a single surface treatment. Also, making it more difficult for DOT pavement managers, much of the published research is commercial in nature and while completely valid, contains a strong inherent bias toward showing the given product in its best light (ARRB 2001, Vercoe 2002, Bennett 2007). Finally, with three exceptions, all completed by the authors of this proposal (Gransberg 2009, Gransberg and Pidwerbesky 2007, Gransberg and Zaman 2005), virtually no research in this area has addressed the economic aspects of pavement retexturing in conjunction with the engineering aspects. Thus, the gap in the body of knowledge is the lack of engineering data correlated with a comparative economic analysis of different alternatives to restore skid resistance on a long-term basis.
BACKGROUND Transportation agencies in the United States have procedures in place to identify and rectify skid resistance problems. However, the procedures are often empirical and tend to be reactive rather than proactive in nature. This is not the case in some other countries. For example, Austroads, the Australia/New Zealand equivalent to AASHTO, developed and has been successfully using a set of procedures to literally manage pavement macrotexture for the past three decades (Austroads 2005). Austroads sees macrotexture as furnishing enhanced drainage to combat hydroplaning during wet periods as well as enhancing skid resistance. As such, they implemented an aggressive macrotexture-oriented program as part of their pavement management system. Therefore, it is not necessary to develop new procedures for the industry and the transportation agencies in this country. This project seeks to “customize” the Austroads model to suit American needs. The Austroads “Procedure to Identify and Treat Sites with Skidding Resistance Problems” uses the following five steps: 1.
“Identify [possible] treatment [alternatives]
2.
Cost works and carry out economic evaluation 3
3.
Shortlist schemes in priority order
4.
Carry out short-term measures, if required
5.
Program longer term measures” (Austroads 2006).
From the aforementioned discussion, it is evident that pavement managers in Australia and New Zealand have not only the engineering technical data that they need to generate a set of technically feasible options for rectifying a loss of skid resistance, but they also have the economic data required to be able to place those alternatives in the context of a limited maintenance budget. It should be noted that this approach does not merely involve selecting the lowest cost alternative. Instead Austroads requires a life cycle cost analysis to accompany all public works and as a result, selects treatment alternatives on a basis of the lowest life cycle cost not the lowest construction cost (Austroads 2005). As a result, a treatment alternative with a higher initial cost but which effectively extends the service life of a pavement for a longer period can be selected, and the long-term benefits to the agency’s multi-year budgets can be accrued.
Additionally, Austroads advocates the use of both short and long-term measures. For example, a given pavement may lose its skid resistance during the winter months where it is climatically impossible to install a bituminous surface treatment due to low ambient air temperatures. Austroads has a machine called the ultra-high pressure watercutter that can literally go out in a limited area such as a slippery superelevated curve or a freeway ramp and restore pavement macrotexture in any weather (Vicroads 2003). This would be a short-term measure. The long-term measure might involve installing microsurfacing or a new chip seal in the summer when the climatic conditions allow it. Both treatments would be included in the life cycle cost analysis used to justify the retexturing project.
OBJECTIVES The primary objective of this study was to build on the research done in Australia and New Zealand. The OTCREOS9.1-21 Phase 2 builds on the OTCREOS7.1-16 Phase 1 work to allow a more comprehensive comparative field evaluation of various methods 4
used to restore pavement skid resistance by retexturing the existing surface with either a surface treatment or a mechanical process, as described in this report. The goal is to furnish pavement managers with the technical engineering data for each pavement preservation treatment coupled with a life cycle cost analysis of the costs and benefits associated with each treatment. This will allow ODOT pavement managers to have the required information to be able to make rational engineering design decisions based on both physical and financial data for a suite of potential pavement preservation tools, which were evaluated under the same conditions over the same period of time by an impartial research team. This project represents a true collaboration between ODOT, industry and university. Considering its importance, both ODOT and industry have committed significant resources toward this study.
SCOPE The project’s scope is best illustrated in the tabular form shown in Table 1. The table shows that the scope covers all three treatment types for asphalt pavements and two types for concrete. Table 1 Oklahoma Pavement Preservation Test Sections Asphalt Test Sections Surface Treatment
• Fog seal • Microsurfacing • ODOT Standard 3/8” chip seal
Chemical Treatment
Mechanical Treatment
• E-Krete pavement surface
• Pavement retexturing using
stabilizer • Asphalt penetrating conditioner with crack seal
• Pavement retexturing using
shotblasting (48” width) abrading (72” width)
• Pavement retexturing using
• ODOT Standard 5/8” chip
abrading (72” width) with fog seal
seal • ODOT Standard 5/8” chip seal with a fog seal • Single size ½” chip seal • Novachip • Open Graded Friction Course • Open Graded Friction Course with a fog seal • Calupave • Single Size Chip Seal with Calupave • 1” Hotmix Asphalt mill-inlay
• Pavement retexturing using a flat headed planing (milling) technique with asphalt penetrating conditioner
5
Concrete Test Sections Surface Treatment
Chemical Treatment • Pavement retexturing using shotblasting treated (48” width) with Nanolithium densifier
Mechanical Treatment • Pavement retexturing using shotblasting (48” width) • Pavement retexturing using abrading (72” width) • Diamond grinding • “Next Generation” diamond grinding
The test sections demonstrate how each of the 23 treatments performs when subjected to the same environment, same traffic, and same testing protocol over the same period. The 12 asphalt surface treatments and the concrete diamond grinding are commonly used by ODOT divisions across the state. The remaining ten test sections are treatments that are new to Oklahoma and represent an opportunity to expand the tools available to ODOT pavement managers.
Table 2 is taken from the FHWA Memorandum on Pavement Preservation (Geiger 2005). It shows that pavement preservation treatments (shaded in green) are applied to extend the service life of a given road, NOT to enhance its capacity or structural strength. The treatments shown in Table 1 literally cover the spectrum of pavement preservation. They range from Shotblasting, which merely restores macro and microtexture and uses no new raw materials, to a 1” (2.5cm) mill and inlay. The FHWA rules state that overlays that are less that 1.5” thick are not deemed to enhance structural capacity (Geiger 2005). The economic importance of understanding the difference between pavement preservation and pavement maintenance is huge. Pavement preservation projects are eligible for Federal-aid funding, while maintenance projects must be entirely funded by the state. Therefore, this project personifies the OkTC theme of “economic enhancement through infrastructure stewardship.” The results of this project will effectively expand the number of pavement preservation tools that can be used by ODOT and every single one of those tools brings federal aid eligibility along with it.
6
Table 2 Pavement Preservation Guidelines (Geiger 2005)
Pavement Preservation
Pavement Preservation Guidelines Type of Activity Increase Increase Capacity Strength New Construction X X Reconstruction X X Major Rehabilitation X Structural Overlay X Minor Rehabilitation Preventive Maintenance Routine Maintenance Corrective (Reactive) Maintenance Catastrophic Maintenance
Reduce Aging X X X X X X
Restore Serviceability X X X X X X X X X
Table 2 also has “corrective/reactive maintenance” highlighted in yellow. Some of the treatments shown in Table 1 can also be used for this category. For example, shotblasting, skidabrading, and microsurfacing can all be applied to a structurally sound pavement that has become unsafe due to loss of skid resistance. However, if used for this purpose, federal funding will not be authorized.
TECHNOLOGY TRANSFER Technology transfer has occurred continuously during this project. Two major technology transfer successes include the following. First, Drs. Caleb Riemer and Bekir Aktas identified and quantified significant limitations associated with the two macrotexture measurement test procedures. This information was presented at the TRB-sponsored 1st International Conference on Pavement Preservation, Newport Beach, California, April 2010. Additionally a paper was written documenting the issues, which has been published in the 2011 Transportation Research Record (TRB). As a result of this finding and the team’s discussions with Hydrotimer, the manufacturer of the Volume Outflow Meter (ASTM STP 583), Hydrotimer has designed and manufactured a version of the meter that is accurate to a tenth of a second instead of the nearest second. This will permit this important ASTM performance test to be used on a much broader spectrum of pavement surfaces.
Second, Dr. Dominique Pittenger developed new algorithms to compute life cycle costs for pavement preservation treatments as her master’s thesis and developed it to include 7
stochastic capability in her doctoral dissertation. Her discussions with FHWA’s Christopher Newman and Tashia Clemons, whose office is the proponent of the FHWA life cycle cost analysis manual at the FHWA Expert Task Group meeting and related meetings, resulted in an abridged version of her thesis being accepted to become a chapter in the next update of that vital policy document. FHWA’s acceptance of her approach using equivalent uniform annual cost analysis represents a huge change to the long-standing FHWA policy requiring the use of net present worth analysis. Considering the fact that a project life cycle cost analysis is a prerequisite to an application for federal-aid highway funding, this transfer of technology represents a contribution that will potentially impact the entire nation’s highway construction budget.
Technology transfer has occurred at a number of additional levels. At the local level, Caleb Riemer, PhD, PE served as both a maintenance engineer and resident engineer in ODOT’s Division 3. He has already implemented the use of several new treatments, including the shotblasting and E-Krete in his division. He has written specifications as a result and has shared those along with emerging test results with the other ODOT divisions during routine state-wide maintenance meetings.
As previously stated, the scale and breadth of this project has drawn national and international attention. The research team has made 34 presentations in 48 months and published (or submitted, awaiting publication) 25 journal and proceedings papers. A recent technology transfer workshop was held by Dr. Doug Gransberg at ODOT detailing pavement preservation treatment (chip seal) performance and specifications. Two technology transfer workshops were held in Norman as a part of a secondary education initiative to expose high school science teachers to engineering testing. Again, Caleb Riemer was the instructor at both. Dominique Pittenger led a day long workshop for top performing Oklahoma high school students in July 2010, which included a look at the test procedures used by pavement engineers in the field on this OkTC project. Finally, research partnerships have been developed with the University of Waterloo in Canada and Suliyeman Demirel University in Turkey. An OU research assistant traveled to Canada last fall, and the project is benefitting from the work done 8
by a Ph.D. student from Turkey who is spending a year working with the OU pavement preservation research team.
REPORT ORGANIZATION The body of the report is organized in four major sections, following the three primary areas in which the project is organized. Those sections are as follows: •
The science of pavement surface micro and macrotexture: Covers the information necessary to understand the field test results.
•
Field test protocol and methodology: Describes the procedures used in the research and features the results of the treatments installed in 2008 and 2009.
•
Life cycle cost analysis: Presents a new model based on EUAC for calculating pavement preservation life cycle costs that differs from the classic model advocated by the FHWA and presents the life cycle cost output for the treatments installed in 2008 and 2009.
•
Field test results: The analysis of the four years of Phase 1 (OTCREOS7.1-16) and Phase 2 (OTCREOS9.1-21) research.
IMPORTANCE OF PAVEMENT SURFACE TEXTURE Engineers must use every possible tool during design, construction and operations to make the road as safe as possible. The design/construction engineer has control over the geometry of the road, both in horizontal and vertical alignments, the speed of travel, the signage of the roadway system and the material properties of the surface course. As the pavement deteriorates over time, the maintenance engineer can also control the characteristics of the pavement surface by selecting various pavement preservation and maintenance treatments. Ultimately, the physics of the moving vehicle will determine if the engineers who have been involved in the road’s service life will determine whether or not the road can be safely traveled. Once the road is built, the only facet of the road that is truly controllable is its surface. No other factor of the complex three-dimensional equation that determines whether a moving vehicle will be able to safely remain on the surface of the road can be changed without a large relative commitment of resources to
9
affect the desired change. As a result, the maintenance engineer’s mission must be to preserve the road’s structural capacity and to ensure that its surface frictional characteristics are sufficient to safely pass traffic for which it was originally designed.
Roadway crashes are complex events that are the result of one or more contributing factors. Such factors fall under three main categories: driver-related, vehicle-related, and highway condition-related (Noyce et al. 2005). This project addresses solutions for highway condition-related crashes.
It does so by quantifying the rate at which 23
different pavement preservation treatments deteriorate over time. The comparative knowledge of treatment deterioration rates is essential for maintenance engineers to select the appropriate treatment for a given pavement condition problem.
Surface Texture Surface texture is the primary physical characteristic that can be measured after a traffic accident (Manion and Tighe 2007). One author posits that the factors that cause loss of skid resistance can be grouped into two categories: •
mechanical wear and polishing action rolling or braking
•
accumulation of contaminants (Neubert 2006)
These two categories directly relate to the two physical properties of pavements that create the friction that produces a pavement’s skid resistance. The first is called microtexture and it consists of the natural surface roughness of the aggregate as shown in Figure 1.
10
Figure 1: Pavement Surface Microtexture and Macrotexture (Pidwerbesky et al. 2006) Microtexture is lost due to mechanical wear of the aggregate’s surface as it is polished by repetitive contact with vehicle tires and gets smoother. The second is macrotexture and relates to the resistant force provided by the roughness of the pavement's surface. Macrotexture is reduced as the voids between the aggregate and either the cement or binder in the pavement’s surface is filled with contaminants. This can happen in three possible ways: 1. Transient macrotexture loss from icing or mud tracked onto the surface 2. Persistent macrotexture loss from flushing or bleeding in the asphalt binder 3. Localized macrotexture loss from accumulation of tire rubber deposits from braking or skidding.
The skid resistance of a highway pavement is the result of a “complex interplay between two principal frictional force components—adhesion and hysteresis” (Hall 2006). There are other components such as tire shear, but they are not nearly as significant as the adhesion and hysteresis force components. Figure 2 shows these forces and one can see that the force of friction (F) can be modeled as the sum of the friction forces due to adhesion (FA) and hysteresis (FH) per Equation 1 below: F = FA + FH
(1)
11
Figure 2: Pavement Friction Model (Hall, 2006) Relating Figures 1 and 2, one can see that the frictional force of adhesion is “proportional to the real area of adhesion between the tire and surface asperities” (Hall 2006), which makes it a function of pavement microtexture. The hysteresis force is “generated within the deflecting and visco-elastic tire tread material, and is a function of speed” making it mainly related to pavement macrotexture (Hall 2006). Thus, if an engineer wants to improve skid resistance through increasing the inherent friction of the physical properties of the pavement that engineer should seek to improve both surface microtexture and macrotexture. This idea is confirmed in a 1984 study of the effect of rubber deposits on airport runway pavements that stated: “Rubber buildup alters the texture properties of the runway as well as the frictional coefficient” (Lenke et al. 1986).
In Australia and New Zealand extensive work has been done to manage macrotexture to control crash rates. In North America extensive work has been done to manage skid number, or microtexture, to control crash rates. Generally, US agencies believe that if an engineer could control wet weather related crashes then all crashes would be reduced. Therefore, most studies regarding crash rates versus surface characteristics, whether macrotexture, skid number, or microtexture, primarily focus on the reduction of
12
wet weather crashes (NCHRP 1989). To better understand exactly how to manage the surface characteristics over time, a thorough definition of each characteristic must be established in order to see the role each plays in contribution to safe travel.
Skid number is a critical component when analyzing road safety, making it one of the most widely studied surface characteristics. Skid number can be measured in a number of ways with the common method being ASTM E 274 skid tester equipped with either with a smooth tire or a ribbed tire. Other common methods are the SCRIM device, the grip tester device, and the mu-meter devices. For this research, the ASTM E 274 skid tester with a ribbed tire serves as the primary way of obtaining the skid number. The testing apparatus is towed behind a vehicle at the desired speed, 40 mph in this project which is the standard for ODOT skid testing. Water is then applied in front of the tire just before the tire’s brakes force the tire to lock up. The resultant force is then measured and converted into a skid number value (ASTM 2000).
Surface texture is separated into three components microtexture, macrotexture, and megatexture as shown in Figure 3. Each has a varying range of texture depth and influences to the pavement tire interactions. Microtexture is the grittiness of the surface; it is a function of the aggregate’s geology and its ability to withstand polishing. The force normally associated with microtexture is “adhesion” which occurs from the shearing of molecular bonds formed when the tire rubber is pressed into close contact with pavement surface (Noyce et al. 2005).
13
Figure 3 Pavement Texture Definitions (Evans 2002)
Macrotexture can also be broken into two separate physical components: hysteresis and drainage. Good texture depth assists with drainage, preventing the formation of a water sheet across the surface with the resulting risk of hydroplaning. While hysteresis is the mechanical deformation of tire with the surface, good texture depth is needed to enable this mechanical deformation to occur, which then releases energy through heat (Cairney 2005).
The PIARC describes macrotexture as a surface roughness quality
defined by the mixture properties, i.e. shape, size, and gradation of aggregate, of asphalt paving mixtures and the method of finishing or texturing used on concrete paved surfaces, such as dragging, tining, or grooving. Macrotexture’s range is set at 0.5mm to 50mm, and it predominately controls the stopping ability on the roadway surface at speeds greater than 45 mph.
Megatexture is on a much larger scale than either microtexture or macrotexture and is in essence the irregularities of the road such as potholes, rutting, joints, cracks, raveling, and skin patches. These have a small effect on stopping ability. However, megatexture plays a significant role in how the driving public perceives the road, via the resultant road noise or poor ride quality. Another surface characteristic that accompanies microtexture, macrotexture, and megatexture, is ride quality or roughness measured by the International Roughness Index (IRI) number.
14
Many DOTs use
roughness measurements to portray the overall integrity of their roadway system (Li et al. 1974, Moulthrop et al. 1996, Zaniewski and Mamlouk 1999). This project does not address megatexture or roughness. Its focus is on micro and macrotexture, the variables of interest for road surface safety.
Surface Texture as a Pavement Preservation Activity Microtexture, macrotexture, megatexture, and roughness, taken together, summarize the universe of roadway surface defects that a maintenance engineer must address. If the road surface begins to ravel, rut, or develop base failures, megatexture will increase. If the road is found to be losing its skid resistance, measured microtexture will be found to be decreasing. If a section of road begins to experience crashes due to hydroplaning, its macrotexture will have decreased. The same is true if a chip sealed road does not retain its aggregate. All these scenarios are the direct result of the surface characteristics of the highway and can be used by the maintenance engineer to identify and select an appropriate pavement preservation or maintenance treatment for a given problem.
Pavement preservation’s focus is placing “the right treatment, on the right road, at the right time” (Galehouse et al. 2003). If a road’s megatexture or IRI become unacceptable, it is too late to attempt to “preserve” the pavement. These measures are indicative of inadequate structural capacity and will require major maintenance or reconstruction treatments to rectify the loss in ride quality and to recapture the pavement’s structural capacity.
Hence, understanding the relationships between
microtexture and macrotexture deterioration over time, allows the engineer to establish “trigger points” that permit sufficient time to schedule pavement preservation before the pavement’s structural capacity is permanently compromised.
A good example of a trigger point comes from the New Zealand Transport Agency (NZTA). NZTA uses macrotexture measurements as one of the key performance indicators (KPI) on its national highway network (Manion and Tighe 2007). Much of that
15
network is surfaced with a chip seal and the rainy climate found in New Zealand demands that pavement engineers manage macrotexture as a means to furnish the requisite surface drainage for safety. NZTA has established that if the average macrotexture of a road drops below 0.9mm (0.04 in) on roads with posted speed limits greater than 70 km/hr (43.5 mph) that pavement preservation by resealing is no longer an option and prescribes the removal and replacement of the surface course. With this failure criterion in mind, NZTA maintenance engineers have then developed individual trigger points based on local conditions that allow the programming of a pavement preservation seal before the macrotexture loss becomes critical (Pidwerbesky et al. 2006).
What pavement preservation often overlooks are other constraints a maintenance engineer also faces. These constraints are primarily budgetary in nature but also can include political or construction timelines. When an engineer determines that a roadway surface needs a treatment to address a surface characteristic deficiency, selecting the appropriate treatment is a critical decision. Currently, the engineer must rely on practical empirical data, gathered through experience in the field to estimate the impact of each treatment option and how long each treatment will last. The treatment’s service life is important due to factors such as the availability of current and/or future funding or the timing of the next major reconstruction project. This information has historically been estimated through laboratory testing or left to the judgment of the engineer. The goal of this research is to standardize the pavement preservation treatment selection process, to assist the pavement maintenance engineers in making treatment selection decisions by quantifying the engineering properties of commonly available surface treatments, and applying that data to create short term deterioration models for each treatment based on actual field testing.
RESEARCH METHODOLOGY AND PROTOCOLS Previous research projects provided guidance for setting up a research project of this magnitude. One of the largest research projects in the field of surface-tire interaction occurred on Wallops Island, Virginia (Yager et al. 2000). 16
It was designed to
characterize the various testing methods and machines used to determine the skid number, microtexture, and macrotexture. In doing so, they set up a large number of pavement test sections that furnish a wide variety of surface treatments. By the end of the process, Yager et al. captured a number of important lessons on how to create test sections that support the technical objectives of the research (Yager et al. 2000).
The goal of this project is to create simple deterioration models for a wide variety of pavement preservation treatments to assist the maintenance engineer with selecting the most appropriate treatment given their situational needs. To do this a uniform test section for each treatment was developed. Conversely, the project needed to eliminate as many ancillary factors as possible.
For all the asphalt test sections, it was
determined that a stretch of a 4-lane state highway (SH77) would furnish a satisfactory location to build the test sections. It had an adequate length to install all the test sections in the same lane of traffic. It also facilitated safety and ease of testing, as active traffic could flow at normal speed while testing was being conducted under a single standard lane closure. The sections were placed in the outside, south-bound lane of travel with gaps between each to avoid major turning motions at intersections and driveways and to act as control sections. The length was predetermined to be 1320 feet (402 meters) which allowed three skid measurements per section but reduced the expense to the contractors donating and installing each section. To ensure uniformity the sections were also designed as full lane-width sections to not inadvertently create an uneven driving surface.
The test section locations were determined before the
treatments were applied and a baseline test of micro and macrotexture was completed for each section.
Test Section Development In 2005, the Federal Highway Administration issued a memorandum that standardized the terminology for pavement preservation projects (Table 2, Geiger 2005). This document described those practices that are eligible for federal funding. The essence of that document was to restrict pavement preservation treatments to those that do not enhance or restore structural integrity. Subsequent guidance refined the definition to 17
allow thin overlays up to 1.5 inches (3.7 cm) thick (Gee 2007). Thus, authorized asphalt pavement preservation treatments range from minimal treatments such as shotblasting that merely restores microtexture and fog seals, to significant treatments like thin overlays on asphalt pavements. A similar range of possible treatments exists in concrete pavements which run from shotblasting through grinding and grooving to white-topping.
From the pavement preservation treatments approved by the FHWA, the researchers selected a series of pavement preservation treatments that spanned the spectrum of possible treatments, as well as cover some that were not highlighted by the FHWA but were available. The asphalt and concrete sections are shown in Table 3.
All sections included in Phase 1 were constructed in the summer of 2008, and all sections included in Phase 2 were constructed in the summer of 2009. All the asphalt sections were constructed in the southbound outside lane of Oklahoma State Highway 77H and the concrete sections were constructed on United States Highway 77, two in the southbound outside lane, one in the southbound inside lane, one in the northbound outside lane, and one in the northbound inside lane. A map of the locations is shown in Figure 4.
The construction of each test section was orchestrated by both the researchers and the contractor donating the material, time and labor. Test sections for the most part were donated by participating contractors; however a conscious effort was made on the part of the researchers to incorporate a broad spectrum of pavement preservation treatments. Individual test section descriptions follow.
18
Table 3 Asphalt and Concrete Test Sections Asphalt Test Sections Phase 1 (Completed Summer 2008)
Phase 2 (Completed Summer 2009)
Chip Seal (5/8″)
E-krete
Chip Seal (5/8″) with Fog Seal
Chip Seal (Single Size)
Chip Seal (3/8″)
Chip Seal (Single Size) With Calupave
Open Graded Friction Course (OGFC)
Calupave Seal
OGFC with Fog Seal
Novachip
1″ Mill and Inlay
Microsurfacing
Asphalt Pentrating Conditioner (APC) APC with Asphalt Planer Fog Seal Shotblasting (Skidabrader) Shotblasting (Skidabrader) with Fog Seal Shotblasting (Blastrac) Concrete Test Sections Phase 1 (Completed Summer 2008)
Phase 2 (Completed Summer 2009)
Shotblasting (Skidabrader)
Shotblasting (Blastrac) with Densifier
Shotblasting (Blastrac)
Diamond Grinding Next Generation Diamond Grinding
19
Figure 4 Map of Research Test Sections Test Section #1 was constructed in Phase 2 to replace a Phase 1 section that was unexpectedly withdrawn. This section was constructed by Calumet Specialty Products in accordance with company specification and included the application of a low penetration fog seal called “Calupave.” This acts and looks like a standard fog seal. 20
However, it appeared much darker and held its color much longer than the traditional fog seal. It is shown in Figure 5.
Figure 5 Test Section #1 –Calupave Seal Test Section #2 shown in Figure 6 is an Open Graded Friction Course (OGFC) with a Fog Seal. Researchers applied fog seal to determine if it would result in a longer service life than the untreated OGFC (Test Section #3, shown in Figure 7). OGFC exhibits a negative macrotexture which means that it is porous and allows water to flow through the surface, thus increasing drainage. Both OGFC sections were installed by a contractor in accordance to ODOT standard specifications and were part of Phase 1.
Figure 6 Test Section #2 – Open Graded Friction Course w/ Fog Seal
21
Figure 7 Test Section #3 – Open Graded Friction Course Test Section #4 shown in Figure 8 is a 1″ Mill and Inlay, installed by contract forces. Construction requires the removal of 1″ of existing surface material and replacement of that material with new hot mix asphalt. In this case, the researcher used a standard ODOT mix design: S4 PG (64-22 OK). This treatment serves as a good bench mark to compare all other treatments. It is commonly used by ODOT maintenance engineers so it was included in Phase 1.
Figure 8 Test Section #4 – 1″ Mill and Inlay Test Section #5 is a standard fog seal placed in accordance with ODOT specifications. This treatment consists of a small amount of emulsion applied to an oxidized surface. The fog seal was SS-1 oil diluted to a ratio of 5:1 water to oil and applied to the surface at a rate of 0.1gal/sy across the entire lane of the test section. It was included in Phase 1 and is shown in Figure 9.
22
Figure 9 Test Section #5 – Fog Seal Test Sections #6 and #7 were installed by the same contractor in accordance with company specification: JLT Inc. Test section #6 incorporated a product called Asphalt Penetrating Conditioner (a proprietary product of JLT Corp.), which works as a rejuvenator for asphalt pavements. It is supposed to penetrate the asphalt, react with the bitumen and then harden, for the purpose of sealing and extending life for an older oxidized pavement. This research was only focused on determining the effects this product had on texture and therefore, did not test any penetration. Test Section #7 was placed using a special machine developed by the contractor to plane the surface of the roadway, removing any humps, ruts, or inconsistencies.
After that process was
complete, the contractor then used the Asphalt Penetrating Conditioner to seal the surface. Both of these test sections, shown below in Figure 10 and Figure 11, were included in Phase 1.
Figure 10 Test Section #6 – Asphalt Penetrating Conditioner
23
Figure 11 Test Section #7 – Asphalt Planer w/ Asphalt Penetrating Conditioner Test Section #8 and #9 are identical chip seals using ODOT specification 3C (5/8″) gradation for aggregate and the same rate and type for binder. The only difference between the sections is that Test Section #8 had fog seal applied after construction. A close up of the texture can be seen in Figure 12 and Figure 13. Test Section #10 is a chip seal using ODOT specification 2 (3/8″) gradations for aggregate and the same rate and type for binder, its texture is shown in Figure 14. The oil for all chip seal sections, Cationic High Float Rapid Set – 2P, was provided by Ergon. All chip seal sections were installed by an ODOT chip seal crew.
Figure 12 Test Section #8 – Chip Seal (5/8″) w/ Fog Seal
24
Figure 13 Test Section #9 – Chip Seal (5/8″)
Figure 14 Test Section #10 – Chip Seal (3/8″) The next three test sections, #11, #12, and #13, all consist of a technique called shotblasting. It consists of shooting steel balls, called shot, at the roadway surface at high speeds. The impact is designed to remove excess bitumen from the road and expose and retexture existing aggregate. The potential of this product lies in the fact that the application process has no temperature constraints and requires no aggregate or binder material. Test Section #11, shown in Figure 15, was installed by Blastrac using a single directional shot pattern and 4′ wide head using company specifications and recommended rates. Test Sections #12 and #13, shown in Figure 16 and Figure 17, respectively, were shotblasted by Skidabrader using a multidirectional shot pattern and 6′ wide head using company specifications and recommended rates. The only difference between the two sections is that section #12 had a fog seal applied after the shotblasting application. An ODOT contractor installed the fog seal in accordance with
25
ODOT specifications, as described in the preceding fog seal section (Test Section #5, Figure 9). .
Figure 15 Test Section #11 – Shotblasting (Blastrac)
Figure 16 Test Section #12 – Shotblasting (Skidabrader) w/ Fog Seal
Figure 17 Test Section #13 – Shotblasting (Skidabrader) Test Section #14, shown in Figure 18, is an example of an ultra-thin bonded wearing course, in this case it is called Novachip. This section was supplied by Haskell Lemon 26
Construction and Hall Brothers. It is a 3/4" thick mixture of well-graded aggregate, polymerized Performance Grade binder, and mineral fillers that is mixed at an asphalt plant and delivered to the project. It is then placed using a material transfer device and a spray paver.
Figure 18 Test Section #14 – Novachip Test Sections #15, #16, and #17 are shotblasting sections on a concrete roadway surface. Test Sections #15 and #17, shown in Figure 19 and Figure 21, respectively, were shotblasted by Blastrac using a single directional shot pattern and 4′ wide head, in accordance with company specifications and recommended rates. The only difference between the two sections is Section #17 had a nanolithium densifier, supplied by Calumet Lubricants, applied to the surface after shotblasting for the purpose of determining the effect on service life of the shotblasted surface. Test Section #16, shown in Figure 20, was installed by Skidabrader using a multidirectional shot pattern and 6′ wide head.
Figure 19 Test Section #15 – Shotblasting (Blastrac) 27
Figure 20 Test Section #16 – Shotblasting (Skidabrader)
Figure 21 Test Section #17 – Shotblasting (Blastrac) w/ Densifier Test Sections #18 and #19 were installed by Penhall, Inc in accordance to company specifications and recommended rates, and consist of next generation diamond grinding and traditional diamond grinding, respectively. Both sections were placed on a concrete roadway surface.
Diamond grinding is a process where a drum-size cylinder with
diamond tipped discs is pressed into the pavement while the cylinder is spinning, grinding the surface of the roadway and adding texture.
With the next generation
diamond grinding technique, grooves are cut into the pavement surface by systematically placing larger diameter discs around the spinning cutter drum. These treatments can be seen in Figure 22 and Figure 23.
28
Figure 22 Test Section #18 – Next Generation Diamond Grinding
Figure 23 Test Section #19 – Diamond Grinding Test Section #20 is a proprietary product called E-Krete, an acrylic and polymer modified concrete mix, produced and installed by Polycon. This product is essentially a Portland cement based slurry seal which is placed on the road at a thickness of 1/8″. This product can then be dragged, like in this test section, or tined to produce the desired texture. This treatment can be seen in Figure 24.
Figure 24 Test Section #20 – E-Krete 29
Test Sections #21 and #22 are both chip seal that incorporate the same aggregate and the same rate and type of emulsion. The only difference between these sections and Test Sections #8 and #9 is that these sections utilized a more uniform gradation of aggregates, referred to as single size. Test Section #22, shown in Figure 26, varies from Test Section #21, shown in Figure 25, because it was treated with Calupave, the same low penetration fog seal material that was used in Test Section #1.
Figure 25 Test Section #21 – Chip Seal (Single Size)
Figure 26 Test Section #22 – Chip Seal (Single Size) w/ Calupave The final test section (#23) is a microsurfacing process, shown in Figure 27. Microsurfacing is a mixture of well-graded aggregate, polymerized asphalt emulsion, fillers, additives, and water that is mixed on the project.
30
Figure 27 Test Section #23 – Microsurface These 23 test sections exhibit a wide range of possible pavement preservation techniques. They were constructed in two separate phases and all sections were tested monthly.
Skid Resistance and Macrotexture Measurements Each test section was tested on a monthly basis to gather sufficient data to develop a deterioration model. There were five tests performed on each section: 1. Skid resistance was measured by ODOT’s skid trailer, 2. Macrotexture was measured using a New Zealand sand circle test, 3. Macrotexture was measured using a Hydrotimer outflow meter, 4. Macrotexture was measured using a Circular Track Meter, 5. Macrotexture was measured using the RoboTex device (selected test sections).
Microtexture Measurements Microtexture of the pavement surface (skid resistance) was measured by ODOT’s skid tester (ASTM E274). ODOT’s skid trailer (Figure 28) used a ribbed tire to produce three skid measurements per test section each month. The skid numbers were then averaged to eliminate any irregularities due to slight variations in the test location to provide the average skid number for a given test section. Tests were all conducted on the outside wheel path.
31
Figure 28 ODOT Skid Trailer Macrotexture Measurements Four testing methods were used to measure the macrotexture of the pavement surface. They included the TNZ T/3 sand circle, the volume outflow meter, the Circular Track Meter (CTM) and the RoboTex. The purpose of taking different types of macrotexture measurements is to allow a back-check by relative readings to be conducted and thus improve the accuracy of the discrete engineering property data collected, as well as to enhance reproducibility. This also allows an opportunity to compare and contrast various macrotexture measurements with the skid number from each section and develop a two-tiered deterioration model with failure criteria for both friction and macrotexture.
The first test used in this study to measure macrotexture, the New Zealand sand circle, was completed on a monthly basis using the NZTA TNZ T/3 standard. Three sand circles were taken on the outside wheel path and averaged together to eliminate any irregularities caused due to slight variations in the test location. The NZTA sand circle test was chosen due to the unreliability of the ASTM sand patch as demonstrated in an early California DOT study (Doty 1974) and a recent study conducted for the Texas DOT (Gransberg 2007). Figure 29 shows the test being conducted in the field. The TNZ T/3 testing procedure feeds the TNZ P/17 performance specification which can then be 32
used as a metric to judge the success or failure of the surface treatments in their first 12 months based on a field-proven standard (TNZ 2002). A recently completed pavement surface texture research project in Texas proved the validity of both the test procedure and the performance specification for use in the US (Gransberg 2007).
Figure 29 Sand Circle Tests and Outflow Meter/Sand Circle Testing in Progress The test is a volumetric test, preformed by placing a known volume of sand, in this case 45 ml, which is then spread by revolving a straight edge in a circle until the sand is level with the tops of the surface aggregate and can no longer be moved around (TNZ 1981). Once the known volume has been spread in a circle on the surface of the roadway and can no longer be moved, two measurements are taken to determine the average diameter of the circle, this value is then inserted into Equation 2. 57,300 Mean Texture Depth (mm) =
Diameter (mm)2
33
(2)
The surface texture is inversely proportional to the diameter of the circle produced on the surface. This testing protocol is relatively simple but has limitations: it is susceptible to operator inconsistency, environmental issues with rain and wind, and roadway imperfections, such as abnormal aggregate heights on the surface of the road. A wind shield is used to shelter the circle from winds and prevent loss of test sand during the test. The second test used in this study for determining macrotexture is an outflow meter as specified by ASTM E2380/E2380M - Standard Test Method for Measuring Pavement Texture Using an Outflow Meter. It is also shown in Figure 29. This method does not measure the texture depth directly; it quantifies the connectivity of the texture as it relates to the drainage capability of the pavement through its surface and subsurface voids. The technique is designed to measure the ability of the pavement surface to relieve pressure from the face of vehicular tires and thus is an indication of hydroplaning potential under wet conditions. A faster outflow time indicates a thinner film of water may exist between the tire and the pavement, thus more macrotexture is exposed to indent the face of the tire and more surface friction is available to the tire. The outflow meter test is based on a known volume of water, under a standard head of pressure, which is then allowed to disperse through the gaps, or macrotexture, between a circular rubber ring and the road surface. The time it takes to pass the known volume of water, i.e. the outflow time, is measured. Mean texture depth was calculated using the following equation. 3.114 Mean Texture Depth (mm) =
Outflow Time (sec)
+
0.636
(3)
Four outflow meter tests were taken and averaged together to eliminate any irregularities due to slight variations in the test location to provide the macrotexture measurement for a given test section.
The third test used in this study for determining macrotexture is the CTM, shown in Figure 30, and provided by The Transtec Group. Measurements were taken on an 34
annual basis in 2009 – 2011, providing three (average) measurements for each test section using ASTM E 2157 - Standard Test Method for Measuring Pavement Macro Texture Properties Using the Circular Track Meter. The CTM consists of a charge coupled device (CCD) laser– displacement sensor that is mounted on a rotating arm. The displacement sensor follows a 284-mm circular track to provide the measurement based on Equation 4, where MPD is mean profile depth, in mm.
Mean Texture Depth (mm) = 0.947 MPD + 0.069
(4)
Seven measurements were taken in the right wheel path, then averaged together to eliminate any irregularities due to slight variations in the test location to provide the macrotexture measurement for a given test section.
Figure 30 Circular Track Meter
35
The fourth test used in this study for determining macrotexture was the RoboTex (meets ISO 13473-3 Class DE device requirements and calibration per ASTM E 797), shown in Figure 31, and provided by The Transtec Group. Measurements were taken on an annual basis in 2009 – 2011, providing three (average) measurements for each test section.
The RoboTex procedure involves measuring MPD in transverse and
longitudinal directions. The output is graphs MPD in increments of distance.
Figure 31 RoboTex Device (top) with Example Output (bottom) There are limitations to these testing methods. Skid testing and the outflow meter have temperature limitation due to the water required for testing. This made testing very difficult and sometimes impossible when the temperature at the test site locations fell below freezing. Another limitation discovered through this research involved the timing mechanism for the Hydrotimer Outflow meter. The technology transfer stemming from this discovery is discussed in detail in the Analysis section.
The sand circle and hydrotimer tests are slow and cumbersome and require forward planning, good time management and adequate traffic control, and are not suited for 36
routine monitoring of the road surface texture over a large road network (Austroads 2005). Other testing methods have recently become available to overcome these obstacles and maintain network wide data for both types of safety criteria; for microtexture, the ASTM E274 skid tester with ribbed tires serves as a texture index, and for macrotexture the ASTM E1845 can be used along with a high speed laser to determine mean profile depth at highway speed (ASTM 2009).
The micro and macrotexture data over time from each of the measurement methods was then compiled and reduced. The output is similar to the output shown in Figure 32. It shows how the micro and macrotexture of a given road is gradually lost during the pavement preservation treatment’s life cycle. This curve portrays the rate of deterioration of a given treatment. When coupled with an appropriate failure criterion, this allows the engineer to estimate the treatment’s actual service life. From the analysis, an actual deterioration rate can be used to furnish a realistic period of analysis to be used in life cycle cost analysis (LCCA).
For instance, NZTA uses 0.9mm of macrotexture as the failure criterion for macrotexture. The test section data shown in Figure 32 has not reached that point after 9 months of service. With high speed data collection and the above-described methodology for texture deterioration modeling, a pavement management engineer now has the tools to better incorporate surface texture safety criteria into the pavement management system (PMS). Regressing the data would yield an equation that would allow the extrapolation of the macrotexture curve and the ability to find when it crossed the 0.9mm line. This would provide an approximate service life on which to base the life cycle cost calculation. Thus, this method furnishes a means to compare a marginally higher cost pavement preservation alternative to a lower cost one and then select the treatment that has the lowest life cycle cost not merely the one that has the lowest initial cost, as ODOT must do today.
37
Test Section 3 - Open Graded Friction Course 6.0
40
35 5.0
4.0 25
3.0
20
15
Skid Number (mu)
Mean Texture Depth (mm)
30
2.0 10 1.0 5
0.0 Sep-08
Oct-08
Dec-08
Feb-09
Mar-09
May-09
0 Jul-09
Month Sand Patch
Hydrotimer Flowmeter
Skid Number
Figure 32 Change in Wheelpath Surface Characteristics Over Time
LIFE CYCLE COST ANALYSIS Life cycle cost analysis (LCCA) provides a measure of a pavement preservation treatment’s sustainability (Chehovits and Galehouse 2010). Sustainability has become an issue as state transportation agencies are increasingly challenged with “high user demand, stretched budgets, declining staff resources, increasing complexity, more stringent accountability requirements, rapid technological change and a deteriorating infrastructure” (FHWA 2007). According to FHWA, “[State transportation] agencies are focusing on maintenance and rehabilitation of existing infrastructure to a greater extent than ever before” (FHWA 2002) and it is expected that pavement preservation will become the core of all future highway programs (FHWA 1998). Oklahoma is certainly one of those agencies for which these statements apply, as evidenced by its latest Asset Preservation Plan that states: “The preservation of our existing transportation system is an absolutely critical part of the Department’s mission” (ODOT 2010). 38
Preservation is especially critical in Oklahoma due to its relatively small transportation budget and correspondingly fragile maintenance budget (Riemer et al. 2010).
Unlike the reactive nature of traditional pavement maintenance programs, pavement preservation is a proactive approach to treating pavements before they fall into disrepair (Geiger 2005). In other words, preservation keeps good roads good (Galehouse et al. 2003). It costs significantly more to fix a pavement in poor condition than it does to maintain one in fair condition (Stroup-Gardiner and Shatnawi 2008, TRIP 2001, AASHTO 2001) because it “requires extensive and disruptive work” (FHWA 1998). The result is “a gradual decline in the number of miles an agency could treat each year and a decrease in the overall condition of the pavement network” (Peshkin et al. 2004), as well as an increase in the “potential for accidents, injuries and fatalities among the motorists and road workers” (FHWA 1998). Theoretically, this proactive approach could reduce the amount of “costly, time consuming rehabilitation and reconstruction projects” and “provide the traveling public with improved safety and mobility, reduced congestion and smoother, longer lasting pavements” (Geiger 2005).
A pavement preservation program, according to the FHWA, consists primarily of three components: preventive maintenance, minor rehabilitation (non-structural) and some routine maintenance implemented for the purposes of slowing deterioration and restoring serviceability of a pavement (Geiger 2005). Pavement preservation does not include corrective (reactive) or catastrophic maintenance, which only serve to restore serviceability (Geiger 2005). Figure 33 shows how pavement preservation differentiates itself from corrective maintenance with its goal of extending the life of a pavement. It is not expected to increase strength or capacity like construction, reconstruction or rehabilitation (Geiger 2005).
39
Figure 33 Proactive Approach versus Reactive Approach (Davies and Sorenson 2000) LCCA and Pavement Preservation The theoretical benefits resulting from pavement preservation is of supreme interest as state transportation agencies search for answers, but the information that would facilitate the widespread implementation of preservation programs is lacking (FHWA 1998, Gransberg et al. 2010). As with any paradigm shift, there is a learning curve. According to the Transportation System Preservation Research, Development, and Implementation Roadmap, January 2008, there is a “need for a comprehensive, large scale Research & Development program” in the area of preservation. The following is an excerpt: “Preservation practices can extend service life and can provide better, safer, and more reliable service to users at less cost. These points reflect common sense and intuitive conclusions, but many aspects of preservation actions or their effect on service have not been demonstrated quantitatively. The tools for pavement and bridge preservation exist, but guidelines for their application are often limited. Research, development and implementation have historically focused on
40
construction and rehabilitation activities and not on the topics of preservation and maintenance.” (FHWA 1998)
But as budget shortfalls and infrastructure needs increase to reach critical levels in the next two decades (FHWA 2007), states are in search of how to spend their limited funds more effectively (AASHTO 2001). “The core of transportation decision making is the evaluation of transportation projects and programs in the context of available funding” (Sinha 2007).
Economic analysis is a vital component to the new paradigm that is
Transportation Asset Management, and specifically, Pavement Preservation, and its application has long been promoted by the FHWA to “highway project planning, design, construction, preservation, and operation” (FHWA 2005), for cost-effectiveness evaluation and accountability (FHWA 2007).
“Considering the annual magnitude of
highway investments, the potential savings from following a cost-effective approach to meeting an agency’s performance objectives for pavements are significant” (Peshkin et al. 2004), thus, allowing agencies to stretch the budget to address sustainability needs in infrastructure and enhance stewardship.
Economic analysis is critical and encouraged at all levels of transportation, to include long term, midterm, and short term, although “models, methods and tools to construct and analyze economic tradeoffs are still being developed”
(FHWA 2007). The long
term, or planning level, uses include evaluation at the design-level for the purpose of assisting engineers in the selection of the most cost-effective design (FHWA 2007). The midterm uses include evaluation at the network-planning level for the purposes of determining cost-effectiveness in budget allocation (FHWA 2007).
The short term,
which is the programming implementation level, and the focus of this research, includes project-specific evaluation for the purpose of assisting pavement managers in pavement treatment selection (FHWA 2007).
To be effective, every programming framework should include a mechanism for assessing the cost effectiveness of alternatives considered for implementation (Sinha 2007). Life cycle cost analysis (LCCA) is an engineering economic analysis tool that 41
could be useful (FHWA 2002), although it is not commonly being employed by frontline pavement managers to determine the most cost effective pavement preservation treatment alternative for a given project (Gransberg et al. 2010, Bilal et al. 2009, J. Hall et al. 2009, Monsere et al. 2009, Cambridge et al. 2005).
The FHWA states the
following purpose for LCCA use:
“LCCA is an analysis technique that builds on the well-founded principles of economic analysis to evaluate the over-all-long-term economic efficiency between [mutually exclusive] competing alternative investment options. It does not address equity issues. It incorporates initial and discounted future agency, user, and other relevant costs over the life of alternative investments. It attempts to identify the best value (the lowest long-term cost that satisfies the performance objective being sought) for investment expenditures.” (FHWA 1998)
LCCA can become quite complex, so an analyst should be judicious about the level of detail included (FHWA 1998).
The analysis can be simplified by including only
differential costs, i.e. omitting those that cancel out, as well as disregarding those costs that contribute minimal or no impact on the final results (FHWA 1998). The FHWA offers “LCCA Principles of Good Practice” in its Life Cycle Cost Analysis in Pavement Design, Interim Technical Bulletin released in 1998. “Good Practice” is that constant dollars and real discount rate be used for the purposes of discounting future costs (i.e. omit inflation and effects).
FHWA recommends a rate between 3-5% be used in
analyses, which is consistent with the OMB Circular A-94. Other “LCCA Principles of Good Practice” are integrated with the LCCA procedures/methodology.
LCCA Procedures/Methodology The following are LCCA procedures, as excerpted from the FHWA Life Cycle Cost Analysis Primer (FHWA 2002) and the Interim Technical Bulletin (FHWA 1998): 1. Establish design alternatives [and analysis period] 2. Determine [performance period and] activity timing 3. Estimate costs [agency and user] 42
4. Compute [net present value] life cycle costs 5. Analyze results 6. Reevaluate design strategies (FHWA 2002, FHWA 1998)
Design Alternatives and Analysis Period The first step in the procedure involves establishing strategies, i.e. associated rehabilitation and maintenance activities of each alternative expected over the analysis period (FHWA 1998). The analysis period can be selected by various methods when alternatives have differing performance periods for the purposes of comparing all alternatives over a “common period of time”, which is an engineering economic analysis principle (White et al. 2010). The general suggestion is that the analysis period be a standard length, such as 35-40 years, and long enough to allow “at least one major rehabilitation activity” for each design alternative (FHWA 2002). The net present value (NPV) method is the preferred analysis method, with the equivalent uniform annual cost (EUAC) being used only as a re-statement of the NPV (FHWA 1998).
Performance Period and Activity Timing The second step involves determining the performance period (i.e. cash flow diagram) for an alternative, which is the period that covers one life cycle of that alternative. It is generally determined by the analyst’s judgment based on experience and historical data (FHWA 1998). Activity timing includes the determination of maintenance and other activity frequency associated with a specific alternative strategy, as established in Step 1 (FHWA 1998). The performance period determination has a significant effect on the LCCA output, and should be considered with care (FHWA 1998).
Agency and User Costs Estimation The third step involves determining or estimating agency and user costs for each of the competing alternatives.
Agency and user costs are determined for each of the
competing alternatives and future costs are “discounted” to determine the NPV. Agency costs are those costs directly incurred by the agency, such as costs for project supervision and administration, materials, labor and traffic control for the initial 43
installation, as well as any associated rehabilitation and maintenance costs required over the life cycle of the alternative. These costs are generally based on current and/or historical costs.
According to the FHWA, salvage value is the value associated with each alternative determined at the point of analysis terminal and involves any residual value (value attributed to the reclaimed materials) or any serviceable life (value attributed to alternative “life” that exists after analysis terminal) and should be attributed to alternatives appropriately for the purposes of analysis (FHWA 1998). Sunk costs, which are costs occurring pre-analysis, should not be included in the analysis unless they specifically apply to the alternatives that are to be compared (FHWA 1998).
User costs relate to costs incurred by the traveling public in both work-zone and nonwork-zone phases for a given extent of road for which alternatives are being compared (FHWA 1998). Generally, the user costs incurred during non-work-zone phases are disregarded in LCCA due to a lower likelihood of difference among alternatives (FHWA 2002).
Differing [work zone] user costs among alternatives are pertinent to the
analyses, and generally include “[time] delay, vehicle operating, and crash costs incurred by the users of a facility” (FHWA 1998).
“User costs are heavily influenced by current and future roadway operating characteristics. They are directly related to the current and future traffic demand, facility capacity, and the timing, duration, and frequency of work zone-induced capacity restrictions, as well as any circuitous mileage caused by detours. Directional hourly traffic demand forecasts for the analysis year in question are essential for determining work zone user costs.” (FHWA 1998)
It is suggested that “different vehicle classes have different operating characteristics and associated operating costs, and as a result, user costs should be analyzed for at least three broad vehicle classes: Passenger Vehicles, Single-Unit Trucks, and 44
Combination Trucks” (FHWA 1998). Unit costs are generally translated into monetary terms (for the purposes of analysis) and can be ascertained from various sources, and those costs escalated with the use of the transportation component of the Consumer Price Index (CPI) (FHWA 1998). Delay costs are calculated by multiplying the unit of “wait” time attributed to each alternative’s work-zone timings by the monetary unit (FHWA 1998). Vehicle operating costs (VOC) are calculated by multiplying the vehiclerelated cost factors attributable to each alternative’s work-zone timings by the monetary unit (FHWA 1998). Crash costs are calculated by multiplying the number of specific types of crashes by their respective monetary unit (FHWA 1998). User costs as a result of detours are typically assigned a cents-per-mile rate, such as that used by the Internal Revenue Service for mileage allowance (FHWA 1998).
Compute Life-Cycle Costs (NPV) As excerpted from FHWA’s LCCA Interim Technical Bulletin: “Economic analysis focuses on the relationship between costs, timings of costs, and discount rates employed. Once all costs and their timing have been developed, future costs must be discounted to the base year and added to the initial cost to determine the NPV for the LCCA alternative.
The basic NPV
formula for discounting discrete future amounts at various points in time back to some base year is shown in Equation 5 below: NPV = Initial Cost + ∑Rehab Costs [1÷ ((1 + i)n)]
(5)
Where: i = discount rate, n = year of expenditure and [1÷ ((1 + i)n)] = PV formula”
(FHWA 1998).
Analyze Results LCCA has two possible computational approaches: deterministic and probabilistic (FHWA 1998). The deterministic approach involves using discrete input values and a single output value and has been the traditional LCCA type used in transportation decision making (FHWA 2002). A sensitivity analysis should be conducted so that the 45
analyst may determine the level of variability of a given input value relative to the output (FHWA 1998). For example, an analyst chooses a 4% discount rate to do the LCCA, which results in output (a preferred alternative). The sensitivity analysis will allow the analyst to conduct a What if scenario to determine if choosing a 5% discount rate would result in different output (different preferred alternative).
The sensitivity analysis is
limited in application because it is unable to analyze simultaneous variability (FHWA 2002).
A deterministic-type LCCA is less complex than a probabilistic type and can be adequate, and therefore appropriate, when uncertainty is not expected to have a material effect on the outcome of the economic analysis (FHWA 2003). However, if uncertainty could materially alter the outcome, the deterministic approach is not recommended because of its inability to effectively analyze simultaneous variability (FHWA 2003). The stochastic approach using probabilistic methods involves analyzing input value probability based on the full range of What if scenarios allowed by sensitivity analysis by providing a distribution of PV results (FHWA 2002).
It is generally
accompanied with a risk analysis, which unlike sensitivity analysis, does allow the analyst to determine the level of certainty with regard to simultaneous variability in all input parameters (FHWA 1998). Stochastic LCCA specifically addresses and quantifies the uncertainty associated with a transportation project decision and contributes to decision validation (FHWA 1998).
The normal distribution is commonly assumed to be the appropriate distribution for data in stochastic analysis and is further justified by the central limit theorem (Montgomery 2009). The central limit theorem states that most data are “approximately normal” (or approximately symmetric in nature) and grow more normalized as sample size increases (Montgomery 2009). It also states that it should be applicable in most cases where possible values of a given input are identically distributed and do not “depart radically from the normal” (or are not extremely skewed) (Montgomery 2009).
There is
no standard rule for sample size and the central limit theorem. Some smaller samples
46
can be approximately normal, while other cases may need larger sample sizes to fit the normal distribution.
A triangular distribution is a continuous distribution that contains user-defined values for (a) the minimum value, (b) the maximum value and (c) the most likely value for an input variable (Figure 34). It is commonly used for variables that have limited sample data but can be reasonably estimated.
This type of distribution may be appropriate for
service life input values. No significant research has been done to quantify the actual service lives of PPTs; therefore, service life data is limited. This type of distribution has also been suggested for discount rate input values of 3, 4 and 5%, the discount range suggested by the FHWA (1998).
Figure 34 is an example of what a triangular
distribution would look like for a thin overlay pavement treatment with an expected service life of 8 to 14 years, with the most likely value being around 9 years (FHWA 1998).
Figure 34 Probability Distribution Graphic and Equations (FHWA 1998) FHWA provides guidance for conducting stochastic LCCA (FHWA 1998), which includes three steps. First, the analyst must decide whether to treat the input values deterministically or probabilistically.
Secondly, the input data must be fitted to the
47
appropriate probability distribution.
Lastly, risk analysis should be conducted.
The
stochastic LCCA calculation for NPV is illustrated in Figure 35.
Figure 35 Computation of NPV using probability and simulation (FHWA 1998) Step 1: Deterministic and Probabilistic Input Value Determination Step one of stochastic LCCA requires the analyst to identify which input values have associated uncertainty and will have a material effect on the outcome (FHWA 2004). Only those values should be treated probabilistically to simplify the analysis (FHWA 2004).
The deterministic approach does not allow probabilistic values, but the
probabilistic approach allows deterministic values. If a pavement manager expected service life to contribute uncertainty that would affect outcome, then it should be incorporated into the analysis in a probabilistic manner (represented as a probability distribution). If material cost was not expected to have an impact, then it could be treated deterministically (represented as a point estimate). If the pavement treatment was expected to incur maintenance costs comparable to those of other alternatives and would not have a material effect on the output, those costs could be ignored altogether (FHWA 1998).
Input value distributions can be subjectively defined based on the
pavement manager’s judgment or objectively defined based on historical or current data from sources like bid tabulations and pavement management systems. Ultimately, the analyst must use judgment to properly assess and address uncertainty in analyses. 48
Step 2: Selecting Appropriate Probability Distributions The second step of a stochastic analysis is to “fit” a given data set to the “best” theoretical probability distribution (FHWA 1998). This is commonly accomplished with statistically-based goodness-of-fit tests, such as Anderson-Darling (A-D) and chisquared (χ2) tests (Lomax 2007). Software is available that can execute the task in seconds (Pallisade 2011). Determining the appropriate probability distribution for given data is a critical step to ensure output validity, because the LCCA is based upon the theoretical probability distribution, not the actual data (Lomax 2007, Tighe 2001).
Step 3: Risk Analysis The third step of a stochastic approach is risk analysis (FHWA 1998), which is based on probability theory. It can be defined as a “systematic use of available information to determine how often specified events may occur and the magnitude of their consequences” (Palisade 2011). Like sensitivity analysis, risk analysis seeks to expose uncertainty associated with input parameters. Risk analysis differentiates itself from sensitivity analysis because it combines “probabilistic descriptions of uncertain input parameters with computer simulation to characterize the risk associated with future outcomes” (FHWA 1998). It also allows the analyst to assess variability in all input parameters simultaneously (FHWA 1998).
Risk analysis can be conducted on a
deterministic basis or a stochastic basis, although employing deterministic risk analysis (like triangular distribution) results in oversimplification and reduced accuracy (Palisade 2011). A Monte Carlo simulation satisfies the conditions of a probabilistic, quantitative risk analysis and is the type used in this research.
Monte Carlo simulation is a
“computerized mathematical technique that allows people to account for risk in quantitative analysis and decision making” (Palisade 2011).
Reevaluate Design Strategies LCCA results should be coupled with other decision-support factors such as “risk, available budgets, and political and environmental concerns” (FHWA 2002). The output from an LCCA should not be considered the answer, but merely an indication of the cost 49
effectiveness of alternatives (FHWA 1998). Considering cost effectiveness without also considering treatment effectiveness (and vice versa), or the economic efficiency of a treatment, may not provide the whole picture either and may result in not selecting the “best” alternative (Bilal et al. 2009).
LCCA Research Approach As pavement preservation becomes increasingly vital to sustainability in infrastructure, it has become apparent that economic analysis, specifically LCCA, could be an essential tool in assisting ODOT pavement managers in the selection of cost-effective alternatives that may yield extended service lives of Oklahoma pavements. Figure 36 shows that the LCCA research is the synthesis of three independent sources of information: •
A comprehensive literature review,
•
A survey of Oklahoma Department of Transportation (ODOT) pavement managers
•
Pavement preservation treatment field trial data.
Additionally, a pilot study was conducted to validate the models created for this portion of the research. The surveys were deployed in an effort to define the current processes used by ODOT pavement managers when making decisions regarding pavement treatment selection, and the literature review was conducted to determine the current processes used by other state highway agencies. The survey data, combined with literature review and the field trial data associated with this project, was also collected for the purpose of defining the input values and other parameters with regard to costs (user, construction, etc.) and time (analysis period, service life, etc.) associated to the specific trial pavement preservation treatments.
50
Phase 1 Evaluate Current Decision-Making Processes
Task 1 Evaluate SHA processes
Task 2 Evaluate ODOT processes
Research Methods • •
•
Task 3 Evaluate current LCCA methods
Key Outcomes •
Literature review ODOT survey Document content analysis
•
LCCA use determination
LCCA method
Phase 2 Strategies, Methods & Tools Development Task 4 Develop deterministic algorithm
Task 5 Develop deterioration models
Research Methods • •
Task 6 Develop stochastic algorithm
Key Outcomes
LCCA step-by-step analysis LCCA application analysis
• • •
Deterministic LCCA EUAC model Field trial deterioration models Stochastic LCCA EUAC model
Phase 3 Application and Validation Task 7 Case Studies field trial data selection
Task 8 Calculate EUACs
Research Methods •
Case study output analyses
Task 9 Compare EUAC Results
Task 10 Tasks 11-13 Explore Validate EUAC Models relationships and sensitivities revealed by LCCA
Key Outcomes • • •
Figure 36 Research Approach 51
Final Deterministic LCCA EUAC model Final Stochastic LCCA EUAC model Final Report for LCCA for Pavement Preservation Treatments
This research aims to develop, complete and report economic and life cycle cost analysis using deterministic and stochastic engineering economic analysis so that models (one for concrete and the other for asphalt pavement preservation treatments) may be created that can produce standardized results relevant to pavement managers when comparing pavement preservation alternatives.
Research Instruments This research employed the following: 1. Analysis of current state transportation agencies decision-making processes 2. Survey of ODOT pavement managers (Table 4) 3. Case study analysis from field trials
Surveys for this study were developed in accordance with the methodology specified by Lehtonan and Pahkinen (2004).
The responses indicated that initial cost plays a
primary role when deciding which pavement treatment to employ and that long-term cost or cost-effectiveness of a treatment selection is not considered, i.e. LCCA is not conducted. The survey also yielded information about other decision making factors, as well as the types of preservation and maintenance treatments typically employed in Oklahoma, and each treatment’s cost range, productivity range and typical service life range based on factors such as average annual daily traffic (AADT), percent truck traffic and pavement condition.
The survey data, combined with literature review and the field trial data associated with this project, was collected for the purpose of defining the input values and other parameters with regard to costs (user, construction, etc.) and time (analysis period, service life, etc.) associated with specific field trial PPTs for use in both deterministic and stochastic EUAC models.
52
3
1 3
Safety Traffic volume
53 3 2 4 3 5
4 3 5 4 1
Service life of PT Availability of PT Availability of trained crew to install PT Weather constraints for PT Past experience with PT’s effectiveness
No
3
2
5
5
4
1
5
3
1
Judgment
3
2
4
3
5
3
2
2
2
1
Judgment
4
5
$1.77/SY)
Most common PPTs (average service life, unit cost) HMA (10 years, $70/ton); chip seal (5 years,
4
1
Existing surface condition
3
1
2
Judgment
2
Initial cost of pavement treatment (PT)
(Ranking: 1-most important, 5- least or not important)
Importance of decision-making factors
Current decision-making methodology
LCCA used in current decisions
1
Response #
ODOT Survey Responses
2
3
4
5
2
2
2
1
1
Table 4 ODOT Survey Responses
LCCA Procedures/Methodology, Equivalent Uniform Annual Cost Method Maintenance funding is authorized on an annual basis making comparing alternatives on an annual cost basis more closely fit the funding model than using NPV which would assume availability of funds across the treatment’s entire service life. Since pavement managers typically consider several alternatives with varying services lives based on available funding rather than technical superiority, the FHWA LCCA method based on NPV creates more problems than it solves.
EUAC has been deemed appropriate for this application (Thoreson et al. 2012, Gransberg and Scheepbouwer 2010, Bilal et al. 2009, Sinha and Labi 2007). Because the EUAC method seemingly bypasses the analysis-period selection process and neutralizes the sensitivity issue based on a non-treatment-relevant basis, it was selected for the model created for this research. The analyst does not have to select an analysis period, which is problematic because of the adjust-to-fit requirements and other cited issues, such as the requirement to understand underlying engineering economic theory garnered from economist experience.
The application of the EUAC LCCA model developed for this research is demonstrated in subsequent sections using extrapolated field data. Input values are based upon field trial, vendor and ODOT survey data, literature review results and bid tabulations. The output was manually verified.
The EUAC model ranking results were compared to
present value alternative ranking results for verification. The model was validated when it produced standardized results. The model’s logic is shown graphically in the flow chart found in Figure 37.
The LCCA procedures, as excerpted from the FHWA Life Cycle Cost Analysis Primer (FHWA 2002) and the Interim Technical Bulletin (FHWA 1998) and discussed in the previous section of this writing, will serve as a guide for the development of the EUAC model.
54
Professional Judgement
Establish Treatment Alternatives
PQI AADT % Truck Traffic Macrotexture
Is year of next expected rehabilitation or reconstruction known?
Microtexture
Other
No
Yes
Determine Service Life for each alternative
Set Analysis Period to year of expected R/R
Determine Activity Timing
Determine Agency Costs (Initial Construction, maintenance, etc)
Determine User Costs (CCI, AADT, type of road)
Calculate EUAC for each alternative
Analyze Results (Sensitivity Analysis)
Reevaluate Results (risk, budget, political & environmental concerns)
Figure 37 Pavement Preservation EUAC LCCA Model Logic
55
LCCA Step 1: Establish [pavement treatment] alternatives [and analysis period] According to the survey deployed for this study, ODOT pavement managers consider the pavement quality index (PQI), percent truck traffic and AADT, among other things, when establishing alternatives. Instead of choosing an analysis period consistent with the NPV method, each alternative’s service life is selected consistent with the EUAC method. The AP for each treatment alternative is equal to its own anticipated service life (ASLalt = APalt), as illustrated in Figure 38.
Alternative A Time ‘A’ Alternative B Time ‘B’ Time 0
Figure 38 Analysis Period in EUAC for Unequal-Life Alternatives According to the FHWA’s LCCA Primer, “allowing analysis periods to vary among design alternatives would result in the comparison of alternatives with different total benefit levels, which is not appropriated under LCCA” (FHWA 2002). Although it may appear on the surface that using EUAC is violating the engineering economic principle of “analyzing alternatives over a common period of time” due to the differing-servicelives-as-analysis-period input, it in fact is not, due to the relationship that exists between EUAC and the PW, least-common-multiple method.
In other words, EUAC can be
considered a “covert” PW least-common-multiple method due to the fact that both calculations for a given data set are proportionately equal, allowing the substitution of EUAC (White et al. 2010).
Furthermore, when one adjusts alternatives’ residual
value(s) in accordance with FHWA’s “good practices” (FHWA 1998), the output, or preferred alternative should be the same regardless of EUAC or PW method employed (as demonstrated in subsequent sections of this writing). EUAC also accommodates the short term, as well as the continuous nature of pavement treatments and does not 56
unnecessarily truncate lives (Lee 2002). EUAC allows a straightforward comparison of each pavement treatment alternative based on a cost-versus-service-life function.
LCCA Step 2: Determine [service life and] activity timing The next step is to enter service-life length for each alternative (shown in Figure 39). This step should be exercised with care due to the sensitivity of the parameter (FHWA 1998). The benefit of using the EUAC model and methodology described and demonstrated herein is that it leverages pavement-manager experience.
Replacing
analysis period input with treatment-relevant input, such as service life and pavement extension, allows the pavement manager to intuitively analyze LCCA results.
Figure 39 Service Life as Analysis Period This section disseminates the methodology for determining the appropriate service life for inclusion in model calculations based upon several critical components. First, the timeframe for the next pavement rehabilitation/reconstruction should be considered (Lee 2002), and is accommodated with the continuous and terminal modes of the EUAC model.
Second, service life values based upon actual local performance can yield 57
superior results to empirical values (Bilal et al. 2009, Reigle and Zaniewski 2002), and is accommodated by the anticipated service life feature. This section will demonstrate how the EUAC model is enhanced and justified by inclusion of these features.
EUAC Model - Continuous and Terminal Modes The pavement manager will be faced with one of two scenarios in the short-term implementation level of decision making: the year of the next expected rehabilitation or reconstruction will either be known (i.e. terminal state (Lee 2002)) or it will not (i.e. continuous state (Lee 2002)). In the terminal state, the pavement manager generally chooses the “do nothing” option. In other words, the pavement manager usually defers maintenance because the pavement is scheduled to be rehabilitated or reconstructed according to the work plan. Therefore, the decision essentially is to ignore pavement preservation on a given pavement knowing that it will be “fixed” in the near future. This permits the reprogramming of those funds to preserving other pavements in the network. There may be occasions, however, when a pavement manager must employ treatment in the interim for safety. To avoid the common “mistake” associated with employing the EUAC method, the analyst must consider the encroachment (White et al. 2010).
In the instant case, when the year of the next expected rehabilitation or
reconstruction is known (i.e. terminal state (Lee 2002)), it must be addressed for engineering economic principles adherence. The model for this OkTC research, as illustrated in Figure 40, has been built to accommodate the terminal state and engineering economic principles.
58
Figure 40 Anticipated Service Life In the continuous mode (when the next rehabilitation/reconstruction is unknown), the “Years until next Rehabilitation/Reconstruction” cell is left blank and the calculations are based on each alternative’s respective service life, as shown in Figure 39. However, as illustrated in the Figure 40, when the next rehabilitation/reconstruction is expected in, for example, five years, “5” is entered into the cell.
The model automatically sets the
analysis period to five years and truncates each service life accordingly, which is notated as the anticipated service life. The residual value is considered to be “0” since the alternatives are not expected to serve in the pavement-treatment capacity at the time of the rehabilitation or reconstruction (Lee, 2002).
Pavement preservation
treatments and the associated pavement-life-extending properties are currently being researched (Peshkin et al. 2004).
If the terminal state exists and the “Years until next Rehabilitation/Reconstruction” cell contains a value that creates a gap for one or more alternatives, the model is built to ignore the gap and calculate all alternatives with the EUAC method (Figure 41). This 59
situation, although rare due to the “do nothing” preference and very short-term nature of the terminal state, may not explicitly adhere to the specific “common period of time” engineering economic principle, but does not warrant it because the gap will most likely be filled with another “do nothing” option. In this scenario, if the analyst were to choose the shortest-life alternative to set the analysis period and the other longer-life alternatives were adjusted to fit in accordance with FHWA straight-line-depreciation-like method, the LCCA should still yield the same preferred alternative as the EUAC method. This type of calculation would likely favor the shortest-life alternative (Hall et al. 2003).
Do nothing Alt. B RV = SLD Alt. A
Alt. A
Alt. B RV = 0
RV=0 Alt. B
Alt. B Short life Next R/R Next R/R
Time 0
Time 0
EUAC, terminal state
NPV, Shortest Life AP
Figure 41 Two Methods, Same Preferred Alternative The only other analysis period setting option that would accommodate this scenario would be to set all alternatives lives to the next expected rehabilitation/reconstruction, which would require filling the gap. This option would be less palatable than shortestlife and EUAC methods because it would require a repetition of the shortest-life alternatives in an effort to fill the gap. Couple the issues associated with filling the gap and the fact that “do nothing” will be the choice selected if possible, filling the gap may result in faulty output.
If the next expected rehabilitation/ reconstruction period is
expected in seven years, leaving a one-year gap for a six-year service life alternative, if the gap is filled with a repetitive treatment, keeping in mind that the residual value will equal zero, then the long-life alternatives may be favored. This would be especially 60
problematic when the gap will realistically be filled with “do nothing”, as illustrated in Figure 42.
Alt. A2 RV = 0
do nothing Alt. A2 Alt. A1
Alt. A1 Alt. B1 Alt. B1 RV = 0
RV = 0 Alt. B1
Alt. B1
Next Expected Time 0
Time 0
Next Expected
Rehabilitation/
Rehabilitation/
Reconstruction, Time ‘t’
Reconstruction, Time ‘t’
Figure 42 Do-Nothing Scenario vs. Fill-the-Gap Scenario On the other hand, the analyst could figure the “do nothing” and use the seven-year period for the six-year alternative, i.e. a one-shot investment in which the six-year alternative would be figured over the seven-period, but the treatment may require some cost in that gap year.
All analysis-period selection methods, when applied to this
scenario, have inherent issues, so one must decide which method would yield the best information for the pavement manager.
EUAC would tend to be the compromise
between treating the short life alternative like a one-shot option or repeating it to fill the gap. The shortest-life method would adhere to the “common period of time” engineering economic principle while EUAC would overtly not. But because the same preferred alternative is being yielded from both methods, and for the purpose of having a consistent model, and with all of the previously-cited issues with the analysis period, using EUAC, even in this rare situation, would appear to be a covert short-life method and provide the pavement manager with relevant decision-making information based on cost, service life and the real possibility of “do nothing” during this state.
61
Essentially, when the EUAC model is used for decision making regarding pavement treatments in the continuous mode, it functions similar to present value in leastcommon-multiple mode.
Furthermore, “instead of employing a rule of thumb for
establishing [an analysis period]”, one should consider the nature of the investment when establishing an analysis period (White et al. 2010). Since ODOT maintenance budgets are annual in nature, then it follows that pavement preservation and maintenance treatment LCCA should use EUAC. The screen captures shown in Figure 39 and Figure 40 come from the LCCA model developed specifically for this project and are currently available for use by ODOT maintenance engineers to assist in pavement treatment selection analysis.
EUAC Model – Incorporating Deterioration Model Information The purpose of this section is to introduce the performance-based LCCA developed within this research, which incorporates performance-based service life value determination based on micro and macrotexture deterioration data. This section will demonstrate the methodology using actual field data from the Phase 1 portion of this project. Five pavement preservation treatments from the field tests serve as the selected alternatives in Table 5. They range from a minimal treatment, shotblasting, to the heaviest treatment, the overlay.
Table 5 Pavement Treatment Service Life Estimation
Pavement Preservation Treatment 1" Hot Mix Asphalt Mill & Inlay (HMA) Open Graded Friction Course (OGFC) 5/8" Chip Seal Pavement Retexturing, Abrading Pavement Retexturing, Shotblasting
ODOT/Lit. Review Service Life (years) 10 (avg) 10 (avg) 5 (avg) 2 (avg) 2 (avg)
Linear regression was applied to the field trial microtexture and macrotexture data to approximate the deterioration rate and extrapolate the remaining service life of each treatment as summarized and illustrated in the following figures. compared to failure criteria found in the literature. 62
These were then
Service life was determined by
identifying the time it took each treatment to deteriorate to each failure criterion. The failure criterion for macrotexture was 0.9mm, which is consistent with TNZ P/12 performance specification (TNZ 2002). The failure point considered for microtexture was a skid number less than 25.
The resulting approximate service life for each
alternative was compared to the ODOT survey and literature review results (StroupGardiner and Shatnawi 2008, FHWA 2005).
This methodology for estimating pavement preservation treatment service from local field data reduces the uncertainty of selecting the service life from experience or the literature, and literally enhances the analog because the data is taken in the same environment that the selected treatment must function (Bilal et al. 2009; Reigle and Zaniewski 2002). Figure 43 illustrates the deterioration based on microtexture for the 1” Hot Mix Asphalt mill and inlay (HMA).
It is important to note that these models are based on limited data gathered in the field because long term data does not exist and serve only to demonstrate the methodology. As more deterioration data is gathered over time, the models will be based upon more information that will yield a clearer idea of the nature of each treatment-type’s deterioration behavior, as further explained in the Results section. This study shows the value of long-term pavement preservation field research and the need to have the combination of both skid resistance and macrotexture available to the maintenance engineer when pavement preservation treatments are selected.
63
SN
Mill & Inlay (HMA), Microtexture Deterioration Data y = 55.371x -0.0127 R2 = 0.9054
55.6 55.4 55.2 55.0 54.8 54.6 54.4 54.2 54.0 53.8
Series1 Power (Series1)
0
1
2
4
3
6
5
7
8
Months
Mill & Inlay (HMA), Extrapolated Microtexture Deterioration Model 56.0
SN
55.0 54.0
Series1
53.0
Power (Series1)
52.0 51.0 0
20
40
60
80
100
120
140
Months
Figure 43 1” Hot Mix Asphalt Mill & Inlay Microtexture Deterioration Model The HMA’s loss of microtexture was plotted over time and the linear regression for the change in microtexture over time was conducted. The equation shown in the upper right-hand corner of the figure was derived and the coefficient of determination (R2) was calculated to be 0.9054. The regression equation was then used to calculate the deterioration rate beyond the available data. These values were added to the actual data points to extrapolate the curve out to 120 months (i.e. 10 years) as shown in Figure 43. Based upon this procedure and a failure criterion of 25, it appears that the HMA will not fail due to a loss of skid resistance over this period. Since this treatment is HMA, macrotexture does not apply due to the inherent properties of the material, which is not designed to provide measurable macrotexture. 64
Open graded friction course (OGFC) is a treatment whose service life can be determined using analysis of both macro and microtexture. The extrapolation methodology described above was used for this and the other treatments. As illustrated in Figure 44 and Figure 45, OGFC falls below the critical level for macrotexture at around 64 months with a coefficient of determination of 0.90, but would retain sufficient skid resistance at the time it fails due to loss of macrotexture.
mm
OGFC Macrotexture Deterioration Data 4 3.5 3 2.5 2 1.5 1 0.9 0.5 0 0
y = 3.259x -0.3095 R2 = 0.9039 Series1 Power (Series1)
2
4
6
8
10
12
14
16
Months
OGFC Extrapolated Macrotexture Deterioration Model 4 3.5 3 mm
2.5
Series1
2
Power (Series1)
1.5 1
0.9
0.5 0 0
10
20
30
40
50
60
70
80
Months
Figure 44 Open Graded Friction Course Macrotexture Deterioration Model
65
SN
OGFC Microtexture Deterioration Data y = 26.767x 0.1301 R2 = 0.9471
40.0 35.0 30.0 25.0 25.0 20.0 15.0 10.0 5.0 0.0
Series1 Power (Series1)
0
2
4
6
8
10
Months
OGFC Extrapolated Microtexture Deterioration Model 60.0 50.0
SN
40.0 Series1
30.0
Power (Series1)
25.0
20.0 10.0 0.0 0
20
40
60
80
100
120
140
Months
Figure 45 Open Graded Friction Course Microtexture Deterioration Model As illustrated in Figure 46 and Figure 47, the 5/8” Chip Seal falls below the failure criteria for macrotexture and microtexture at around 21 months and 46 months respectively.
66
5/8" Chip Seal Macrotexture Deterioration Data
5/8" Chip Seal Extrapolated Macrotexture Deterioration Model
y = -0.004x 2 - 0.0012x + 2.5216 R2 = 0.9687
3
3
2.5
2.5 2 Series1
1.5
mm
mm
2 Poly. (Series1)
Series1
1.5
1
Poly. (Series1)
1
0.9
0.9
0.5
0.5
0
0 0
2
4
6
8
10
12
14
0
16
5
10
Months
15
20
25
Months
Figure 46 5/8” Chip Seal Macrotexture Deterioration Model
5/8" Chip Seal Microtexture Deterioration Data
5/8" Chip Seal Extrapolated Microtexture Deterioration Model y = -5.5434Ln(x) + 46.099 R2 = 0.9191
50.0
50.0
40.0
30.0
SN
SN
40.0 Series1
25.0
Log. (Series1)
20.0
30.0
Series1
25.0
Log. (Series1)
20.0 10.0
10.0 0.0
0.0
0
2
4
6
8
10
12
0
Months
10
20
30 Months
Figure 47 5/8” Chip Seal Microtexture Deterioration Model
67
40
50
Shotblasting Extrapolated Microtexture Deterioration Model
Shotblasting Microtexture Deterioration Data y = 51.304x -0.0321 R2 = 0.6764
54.0 53.0 51.0
Series1
50.0
Power (Series1)
SN
SN
52.0
49.0 48.0 47.0 0
2
4
6
8
10
54.0 53.0 52.0 51.0 50.0 49.0 48.0 47.0 46.0 45.0 44.0
12
Series1 Power (Series1)
0
10
20
30
Months
40
50
60
70
Months
Figure 48 Pavement Retexturing - Shotblasting Microtexture Deterioration Model
Abrading Microtexture Deterioration Data 50.0
Abrading Extrapolated Microtexture Deterioration Model
y = 43.813x -0.0587 R2 = 0.3536
50.0 40.0
30.0
Series1
25.0
Power (Series1)
20.0
SN
SN
40.0
30.0
Series1
25.0
Power (Series1)
20.0 10.0
10.0
0.0
0.0 0
2
6
4
8
0
10
10
20
30
40
Months
Months
Figure 49 Pavement Retexturing - Abrading Microtexture Deterioration Model
68
50
60
70
The analysis of field data, coupled with literature and ODOT survey information, can be used to select an appropriate service life value as shown in Table 6 and gauge the sensitivity for each treatment being evaluated.
Table 6 Treatment Service Life Based on Extrapolated Field Data
Pavement Preservation Treatment 1" Hot Mix Asphalt Mill & Inlay (HMA) Open Graded Friction Course (OGFC) 5/8" Chip Seal Pavement Retexturing, Abrading Pavement Retexturing, Shotblasting
Microtexture > 10 > 10 3.8 >5 >5
Service Life (years) Macrotexture ODOT & Lit. Review Minimum 10 10 N/A 5.3 years 10 5.3 1.8 5 1.8 N/A 2 2 2 2 N/A
The selected alternatives and the corresponding minimum service life values from Table 6 were entered into the model, as well as other items required for LCCA as illustrated by Figure 50.
Figure 50 LCCA Step 1: Establish Alternatives and Each Service Life
69
After determining service life, Step 2 of the LCCA process requires determining activity timing. For this demonstration, maintenance includes crack seal and 2%-of-total-area patching with a three-year frequency for all asphalt treatments (Figure 51).
Figure 51 LCCA Step 2: Determine Activity Timing LCCA Step 3: Estimate Agency & User Costs The average cost for treatments and maintenance shown in Table 7 came from the ODOT survey and was verified by field trial donor data, literature review results (StroupGardiner and Shatnawi 2008, FHWA 2005, Bausano et al. 2004), and ODOT bid tabulations. It is used as shown in Figure 52 and Figure 53.
Table 7 Average Pavement Treatment Cost Asphalt Asphalt Asphalt Asphalt Concrete Concrete
PAVEMENT TREATMENTS ODOT Standard 5/8" chip seal Open Graded Friction Course (OGFC) 1" Hot Mix Asphalt mill/inlay (HMA) Crack Seal Pavement Retexturing using shotblasting (48" Width) Pavement Retexturing using abrading (72" Width)
70
Average Cost/SY 1.77 3.75 4.00 0.34 3.00 2.00
Figure 52: LCCA Step 3: Determine Agency Costs, Construction
Figure 53 LCCA Step 3: Determine Agency Costs, Maintenance
71
User costs should be included in LCCA if installation time and frequency for the pavement preservation treatments being considered differ greatly (Bilal et al. 2009). Therefore, user costs were calculated for this demonstration. The initial construction installation time is represented by days, to two significant digits, to capture the differences between alternatives. Production rates came from the ODOT survey and vendor data. According to ODOT, the traffic volume for the section of Sooner Road containing the asphalt test sections is 16,900 (AADT); the traffic volume for the section of Highway 77 containing the concrete test sections is 19,500 (AADT). For the given road sections, ODOT anticipates a 2% traffic growth every two to three years for the remaining pavement design life of ten years. Traffic is expected to be comprised of 92% vehicle and 8% truck traffic split 50/50. The routes are classified as “5: Urban Minor Arterial or Collector”. The May 2010 Construction Cost Index (CCI) is 8677 (ENR 2010). These user costs input were entered into the model as illustrated by Figure 54, Figure 55 and Figure 56.
Figure 54 LCCA Step 3: Determine User Costs
72
Figure 55 LCCA Step 3: Determine User Costs
Figure 56 LCCA Step 3: Determine User Costs (Continued)
73
LCCA Step 4: Compute Life Cycle Costs The EUAC model created for this research calculates life cycle cost for each alternative based on the EUAC method. As illustrated in Figure 57, all incurred costs expected throughout the service life of an alternative are brought to a base year, summed, and then annualized according to the treatment’s service life as proposed in this research, determined by field data and pavement manager professional judgment. It uses the following EUAC calculation (Equation 6): EUAC (i%) = [∑P] * [i(1+i)n ÷ (1+i)n – 1)]
(6)
Where: i = discount rate P = present value n = pavement treatment anticipated service life
Figure 57 Illustration of EUAC Calculation for Chip Seal The discount rate selected for the demonstration of the model is 4%.
Project length is
one lane-mile. LCCA Calculations, Continuous Mode – Treatments on Asphalt Pavement No value was entered in the “Years until next Rehabilitation/Reconstruction” because of the analysis assumes the continuous state, so the “Service Life” is equal to the 74
“Anticipated Service Life” for all alternatives. The life cycle cost calculations were conducted to determine the EUAC of each alternative as shown in Figure 58. The results were manually verified (see the validation section found at the end of this section for more detail). Although the HMA alternative has the highest initial cost ($28,160), it has the lowest EUAC ($4,696) when service lives are set to the minimum value, as demonstrated in Figure 59.
Figure 58 LCCA Step 4: Calculate EUAC, Asphalt Pavement Treatments 75
Figure 59 EUAC Comparison, Asphalt Pavement Treatments LCCA Calculations, Continuous Mode – Treatments on Portland Cement Concrete Pavement The same LCCA steps are applied to demonstrate treatment comparison for Portland cement concrete pavement using the information from the service life and cost Table 6 and Table 7, and all other parameters remaining the same, with the exception of the exclusion of a maintenance strategy on these two-year treatments (Figure 60 and Figure 61). Based on the given data, the pavement retexturing using abrading alternative would be preferred because it has a lower initial cost and EUAC.
76
Figure 60 LCCA Step 4: Calculate EUAC, Concrete Pavement Treatments
77
Figure 61 EUAC Comparison, Concrete Pavement Treatments LCCA Calculations, Terminal Mode To demonstrate the EUAC LCCA model’s ability to accurately depict all possible modes and adhere to engineering economics principles (White et al. 2010), the following will demonstrate the terminal mode.
This mode is required when the next major
rehabilitation or reconstruction for a given pavement is known.
The model should rarely be operated in terminal mode due to a pavement manager’s propensity to “do nothing” when the next rehabilitation/reconstruction is known. However, if “do nothing” is not an option for safety reasons, the model can be used to determine the preferred alternative in this short term period. Although it can yield the same preferred alternative as NPV regardless of analysis period selected (see validation section at the end of this section), it can be sensitive to the analysis period selection depending on the input data. In an analysis-period-sensitive situation, the EUAC will function like NPV when setting the analysis period consistent with the shortest-life alternative. For example, if the next expected rehabilitation/reconstruction is in six years, “do nothing” would be the pavement manager’s first choice. 78
If “do
nothing” is not an option, then “6” would be entered into the appropriate cell, truncating the longer service lives of the OGFC and HMA, but leaving a one-year gap filled implicitly with “do nothing” for the chip seal, as illustrated in Figure 62. The residual value equals zero at the time of the encroachment, so the OGFC and the HMA will have analysis periods of six years. Figure 63 shows the preferred alternative to be chip seal. It would also be the intuitive choice because it would efficiently fill the gap. The model was validated (see validation section) in both continuous and terminal modes by comparing EUAC model output with NPV model output (White et al. 2010).
Figure 62 LCCA Step 4: Calculate EUAC, Terminal State
79
Figure 63 LCCA Step 4: Compare EUAC, Terminal State LCCA Step 5: Analyze the Results Deterministic (empirical and performance-based) and stochastic methods are used to expose sensitivity in LCCA and support decision making. This section demonstrates those methods for the purpose of presenting a methodology for extracting critical decision-making information.
Service life and discount rate values are generally
sensitive parameters (FHWA 1998), so both are analyzed for sensitivity.
Analyze LCCA Results – Deterministic Methods The deterministic, performance-based analysis is a relatively new concept that can be used to assess pavement preservation treatment’s economic efficiency that derives cost as a function of service life (Pittenger et al. 2011, Bilal et al. 2009, Reigle and Zaniewski 2002).
The service life value is based upon actual performance data rather than
assumption.
Incorporating performance data into analyses may contribute to
determining the optimal preservation timing (Peshkin et al. 2004) so that a pavement manager can install the “right treatment to the right road at the right time” (Galehouse et 80
al. 2003).
Both macrotexture and microtexture (deterministic, performance-based)
values are routinely collected by ODOT and are extrapolated to estimate service life duration on the basis of localized performance (Pittenger et al. 2011). These parameters are used to verify the sensitivity of the service life value in the LCCA, as demonstrated in Figure 64.
The analyst can compare the performance-based output to the empirical data-based output (based upon ODOT expected service life values, as stated in Table 5), which will serve as the sensitivity analysis. When a service life of 3.8 years (microtexture extrapolation) is entered for chip seal, its LCC becomes equal to the LCC of the HMA alternative. However, it is not preferred when a 1.8-year service life value extrapolated from the macrotexture data is used, as illustrated in Figure 64 and Figure 65. This situation illustrates the value of exposing the service life sensitivity versus obscuring it with NPV so that the pavement manager can intuitively analyze the results. With these EUAC results, the pavement manager can use judgment to determine whether or not the 3.8-year chip seal, coupled implicitly with “do nothing” for 2.2 years, can realistically fill the 6-year window.
81
Figure 64 LCCA Step 5: Service Life Sensitivity, Terminal – Microtexture-Based
Figure 65 LCCA Step 5: Service Life Sensitivity, Terminal – Macrotexture-Based
82
It is also prudent to analyze the effect of pavement-life extending possibilities on alternative ranking by those alternatives whose service lives were truncated by the next expected rehabilitation/reconstruction. This is accomplished by changing the value in the next expected rehabilitation/reconstruction cell to the expected or possible service life value for given alternatives. In the instant case, when the cell is set to “10”, which is consistent with the expected life of HMA and OGFC, the preferred alternative does not change, meaning there is no sensitivity to the pavement-life extension parameter. This calculation is identical to the one conducted in Figure 66. If, on the other hand, the pavement-life extension parameter is sensitive, the pavement manager may ascertain the effect by intuitively adjusting the R/R cell until the preferred alternative changes, within the expected limits of service life for alternatives.
Figure 66 LCCA Step 5: Pavement Life Extension Sensitivity, Terminal State
83
Figure 67 LCCA Step 5: Service Life Sensitivity Analysis, Microtexture-Based In the above example where the chip seal and HMA options result in the same EUAC (Figure 30), essentially, the pavement manager could choose either option on the basis of LCC at a 4% discount rate. However, the discount rate is generally a sensitive parameter and the sensitivity should be analyzed on a case-by-case basis (FHWA 1998). A deterministic sensitivity analysis (Figure 68) reveals that this LCCA is sensitive to the discount rate. Selecting a discount rate less than 4% results in the HMA having a lower EUAC ($4,580) than chip seal ($4,643). When using a discount rate higher than 4%, the HMA has a higher EUAC ($4,814) than chip seal ($4,749). This means that the selection of a discount rate could effectively drive the pavement preservation design decision between these three alternatives. Therefore, the classic notion that alternatives can be compared equitably using the same discount rate (Walls and Smith 1998) is disproven for application in the pavement preservation treatment LCCA (Gransberg and Scheepbouwer 2010). This will not always be the case based on the relative differences between alternatives’ installation costs (Bilal et al. 2009, Reigle and Zaniewski 2002). However, it shows that if a low discount rate is used, it favors the higher cost, longer lived alternative. The converse is true if a high discount rate is used.
84
Figure 68 LCCA Step 5: Discount Rate Sensitivity Analysis, Microtexture-Based When all service lives are set to ODOT expected values (deterministic – empirical analysis), the chip seal becomes the preferred alternative, while the OGFC moves to 2nd place (Figure 69).
This proves that using field data-derived deterioration curves and
performance-based failure criteria provides a more accurate result than the empirical values for service life in use for the current FHWA-approved LCCA process.
Figure 69 LCCA Step 5: Service Life Sensitivity Analysis, Empirical Data-Based 85
Based on this data, the service life parameter is sensitive because an alternative’s service life and cost are directly correlated in LCCA. By changing the service life input value of chip seal from 1.8 years to 3.8 years and then to 5 years, its rank changes from 3, to tied with HMA, to 1, respectively. The discount rate, in the instant case, is not sensitive. In other words, using a different discount rate will not change the preferred alternative, as illustrated by Figure 70. Discount Rate Sensitivity 10,000
EUAC
8,000 5/8" Chip Seal
6,000
OGFC 4,000
HMA
2,000
20
15
11
9. 5
8. 5
7. 5
6. 5
5. 5
4. 5
3. 5
2. 5
0. 25 0. 75 1. 25 1. 75
0
Discount Rate
Figure 70: LCCA Step 5: Discount Rate Sensitivity Analysis, Asphalt Pavement PPTs A deterministic sensitivity analysis conducted for the concrete pavement treatments reveals that the LCCA is not sensitive to discount rate (Figure 71).
16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0
Shotblasting
18
12
10
9
8
7
6
5
4
3
2
1. 5
Abrading
1
0. 50
EUAC
Discount Rate Sensitivity
Discount Rate
Figure 71 LCCA Step 5: Discount Rate Sensitivity Analysis, Concrete Pavement PPTs
86
Analyze LCCA Results – Stochastic Methods If uncertainty is expected to impact LCCA results, stochastic analysis is the appropriate approach (FHWA 1998). This study followed FHWA guidance for conducting stochastic LCCA (FHWA 1998), as outlined in the previous section.
Several steps were
considered for model construction, including whether to treat the input values deterministically or probabilistically, fitting the initial costs and discount rate with the appropriate distribution and developing the algorithm for risk analysis. The first step in stochastic analysis is to determine which input values to treat deterministically and which to treat probabilistically based on the uncertainty expected of each parameter (Tighe 2001, FHWA 1998). For this study, the initial costs, service life and discount rate were treated probabilistically and the rest were treated deterministically.
The following equation (7) will constitute the sum of present values: ∑P = PPTP + DV + FCD-P + UCD-P
(7)
where: PPTP = Expected initial cost of PPT (probabilistic) DV = Deterministic value of other agency costs (such as equipment, traffic control, striping) FCD-P = Expected value of future (maintenance) costs (probabilistic discount rate and service life components, deterministic cost) UCD-P = Expected value of user costs (probabilistic discount rate and service life components, deterministic cost).
The expected initial costs of treatments (PPTp) were probabilistically determined from ODOT and Oklahoma Department of Central Services (DCS) bid tabulations for the calendar years 2010-2011 and are consistent with ODOT survey and literature and are listed in Table 8. Prices from projects with similar quantities and in similar geographic locations were used so as to not inadvertently skew output (Tighe 2001).
87
Table 8 Treatment Agency Costs Used in EUAC LCCA. Agency Costs ($ per Square Yard)
PPT Chip Seal
Deterministic Value 1.77
Probabilistic Value Mean (Standard Dev.) 1.82 (0.07)
3.75 4.00
3.88 (0.39) 3.78 (0.49)
OGFC 1” HMA Mill & Inlay
Treatment agency costs for deterministic LCCA were found in literature (StroupGardiner and Shatnawi 2008, FHWA 2005, Bausano et al. 2004) and ODOT survey (Table 8). Activity timing for future maintenance included a three-year frequency for crack seal and 2%-of-total-area patching.
The second step in the stochastic process is to conduct goodness-of-fit tests for the stochastic parameters determined in the first step.
A triangular distribution is a
continuous distribution that contains user-defined values for the minimum value, the maximum value and the most likely value for an input variable. It is commonly used for variables that have limited sample data but can be reasonably estimated (FHWA 1998), which is the case for service life. In the continuous mode, 4-5-6 years were used for the chip seal service life distribution parameters. 8-10-12 years were used in the OGFC and HMA distributions, consistent with literature and ODOT survey. This common pavement management scenario is illustrated in Figure 72.
Figure 72 Stochastic EUAC Input Screen, Continuous Mode. 88
In the continuous mode, the “Years until next Rehabilitation/Reconstruction” is unknown, so that field is left blank. Therefore, the service life value is equal to the anticipated service life value. The service life distribution is also noted in the entry screen as a triangular distribution and the low, most likely and high values are noted in parentheses. The mechanics of continuous and terminal modes in the EUAC model in stochastic mode are consistent with those in the deterministic mode, as discussed in the previous section.
The terminal mode is used when the next rehabilitation or reconstruction year is known. If the next rehabilitation is expected in 5 years, then that value is entered into the “Years until next Rehabilitation/Reconstruction” field, as shown in Figure 73.
Figure 73 Stochastic EUAC Input Screen, Terminal Mode. Based upon the rehabilitation year (RY), the terminal mode automatically truncates the service life values of the alternatives to provide the anticipated service life (ASL). Both the HMA and OGFC service lives are truncated because the low value of their triangular distributions is 8, which is greater than RY, so their EUACs will be figured deterministically with an ASL of 5 years. The chip seal service life will also be truncated at 5 years; however, the 5-year value falls within its distribution, so it will be treated probabilistically. 89
For the triangular distributions, if: RY ≤ SLlow, then ASL = RY (deterministic); RY > SLlow and RY ≤
SLhigh, then ASL = RY as triangular max value
(probabilistic); RY ≥ SLhigh, then ASL = SL (probabilistic). The chip seal’s terminal service life distribution will have parameters of 4-5-5, instead of its continuous distribution of 4-5-6, as shown in Figure 74.
Figure 74 Chip Seal Service Life Probability Distributions, Continuous (left) and Terminal (right) Modes. The future maintenance costs and user costs are deterministic values in the analyses, but are discounted probabilistically so that the discounting period and discount rate are consistent within each iteration. This study used the previous 30 years of discount rate data from the Federal Reserve (2011), best fitted to the Weibull distribution (μ: 4.92, sd: 2.76). Next, initial PPT costs were considered. The best fit for the OGFC bid tabulation data is the normal theoretical distribution, which has a chi-squared (χ2) value of 5.94, followed by the extreme value distribution (χ2 =11.71) (Table 9).
90
Table 9 Goodness-of-Fit for PPTs. Bid Tabulation Data Fit ($/Square Yard)
PPT
# of Data Points
Chip Seal
25
OGFC
36
1” HMA Mill & Inlay
222
Distribution Type Normal Extreme Value Normal Extreme Value Normal Gamma
Chi-Square Value 7.60
Mean 1.817
Standard Deviation 0.0704
5.20
1.812
0.0799
5.94
3.88
0.391
11.71
3.88
0.391
9.51
3.78
0.49
15.7
3.78
0.49
The extreme value distribution was also the best fit for chip seal, while the gamma distribution best described the HMA data.
In all cases, the χ2 values are minimal,
meaning the theoretical distributions closely approximate the data. In addition, the best fit distributions returned the same (similar) means and standard deviations as the normal distributions, consistent with the central limit theorem that states data normalizes with increasing number of data points (Montgomery 2009).
Therefore, the EUACs
yielded from the normal distribution of bid tabulations is sufficient and will be used in this chapter.
The third step of stochastic analysis involves risk analysis. The algorithm developed for Monte Carlo simulation is based upon the EUAC equation and is illustrated in Figure 75. It incorporates the costs, service life and discounting methods discussed in the previous two steps to produce the sampling distribution that provides the expected EUAC for each alternative being considered.
91
Figure 75 Stochastic EUAC LCCA Approach. (Adapted from FHWA 1998) The model in continuous mode returns EUAC distributions for the three alternatives, as shown in Figure 76.
The curves show that the chip seal and OGFC have similar
variability, which is less than the HMA. Additionally, the chip seal has a lower life cycle cost, as demonstrated by its curve being to the left of the other two alternatives.
Figure 76 PPTs Stochastic EUAC Probability Distributions.
92
Numeric values associated with these distributions are shown in Table 10 and Figure 76. Based on these values, it would appear that chip seal would be the most cost effective option in this specific case. Table 10 Stochastic EUAC Results, Continuous Mode. EUAC ($/lane-mile), Stochastic Method (Continuous Mode)
Mean Standard Deviation th
5 Percentile th
95 Percentile
Chip Seal
OGFC
HMA
3,738
4,546
4,410
414
426
758
3,095
3,856
3,254
4,452
5,275
5,726
Stochastic analysis offers a wealth of decision-assisting information, such as percentile statistics.
Percentiles can provide the pavement manager relative probabilities
associated with each pavement treatment alternative being considered. For example, the mean EUAC of HMA ($4,410) falls around the 95th percentile of the chip seal EUAC range, as shown in Figure 77.
This means that the central tendency, or bulk, of
possible HMA EUACs would be higher than the expected chip seal EUAC 95% of the time.
Figure 77 Summary Statistics for Hot Mix Asphalt EUAC Distribution. 93
However, the HMA EUAC range does reach the lower end of the chip seal range. The summary statistics reveal that the chip seal mean of $3,738 lies around the 20th percentile of the HMA range, so HMA should be expected to be less than the mean value of chip seal 20% of the time.
The deterministic EUAC values and subsequent rank for HMA, OGFC and chip seal were $4,696 (3), $4,460 (2) and $3,651 (1), respectively, as discussed in the previous section (Figure 69).
If the pavement manager was only considering the HMA and
OGFC options, then the stochastic and deterministic outcomes would be different because the stochastic results show HMA having the lower LCC, based on actual bid tabulations.
When considering the deterministic and stochastic HMA values, the
stochastic analysis reveals that the EUAC should not be expected to exceed $4,696 70% of the time, while the chip seal is at 45% probability.
The stochastic approach also offers a comprehensive sensitivity analysis based on regression. The sensitivity analysis reveals that the chip seal is most sensitive to the service life parameter, the OGFC is most sensitive to discount rate and the HMA is most sensitive to the agency cost parameter, as shown in Table 11. The pavement manager may take the following into consideration: if chip seal was not expected to realize its’ full service life due to traffic conditions, the pavement manager may select the HMA treatment, which is not as sensitive to the parameter. However, if commodity volatility (agency costs) were expected to increase, the chip seal may be the better option because it is less sensitive to the cost parameter than HMA. Although both treatments contain the same volatile commodity (asphalt binder), the sensitivities reflect the different methods of procurement. The HMA agency costs are based upon monthly ODOT bid prices, whereas the chip seal is based upon DCS bid prices that are “lockedin” for some time period. The stochastic analysis offers this type of critical information that the deterministic analysis cannot.
94
95
Inlay
2.76
2.76
OGFC
1” HMA Mill &
2.76
0.771
0.808
0.688
Coefficient
Regression
584.55
601.11
285.21
EUAC ($)
Change in
(%)
Deviation
Chip Seal
PPT
Net
Standard
Discount Rate
0.816
0.816
0.410
(Years)
Deviation
Standard
-0.302
-0.313
-0.657
Coefficient
Regression
Service Life
Sensitivity Analysis
-228.77
-233.14
-272.06
in EUAC ($)
Net Change
Standard
8.37/ton
9.30/ton
0.07/SY
Deviation ($)
0.558
0.484
0.277
Coefficient
Regression
Cost
422.98
360.30
114.61
in EUAC ($)
Net Change
Table 11 Stochastic EUAC Sensitivity Analysis.
The model in terminal mode returns the EUAC distributions as shown in Table 12. The OGFC and HMA values are substantially higher than the values returned in the continuous mode (Table 9.3), and the chip seal and HMA EUAC distribution curves do not overlap, except for the extreme (tail) values that have little associated probability.
Table 12 Stochastic EUAC Results, Terminal Mode. EUAC ($/lane-mile), Stochastic Method (Terminal Mode) Chip Seal
OGFC
HMA
3,962
7,191
7,015
362
827
942
5th Percentile
3,399
5,898
5,540
95th Percentile
4,580
8,622
8,627
Mean Standard Deviation
In the continuous mode example, the decision would require pavement manager judgment, facilitated by percentiles and sensitivity analysis, to justify a decision due to the overlapping EUAC distributions. In the terminal case, the chip seal appears to be the best option.
The stochastic results reveal that the impact of terminal mode on the chip seal probability associated with the $3,651 (deterministic EUAC) dropped from 45% to 20%, meaning the chip seal is 25% more likely to exceed that value. There is a greater than 95% chance that HMA will exceed its deterministic value. This demonstrates the impact of the rehabilitation year on service life and justifies the inclusion of the terminal mode in the EUAC model for upholding engineering economic principles (White et al. 2010).
LCCA Step 6: Reevaluate the Results Armed with field data, expertise and the LCCA results for a continuous or terminal scenario, a pavement manager would have enough information to make decisions about which treatment to employ.
This information should be coupled with other
decision-support factors such as “risk, available budgets, and political and environmental concerns” (FHWA 2002).
The output from an LCCA should not be 96
considered the answer, but merely an indication of the relative cost effectiveness of alternatives (FHWA 1998) and a rough measure of each treatment’s sustainability as compared to the other options.
LCCA Model Validation The purpose of this section is to present the methods used to validate the EUAC model created for this research. The model was validated in the continuous and terminal modes, as well as the deterministic and stochastic modes.
EUAC Model Validation - Continuous Mode To verify the model in continuous mode, EUAC and NPV were calculated to demonstrate that all should yield the same preferred alternative when gaps and residual values are addressed as discussed and cited in the previous sections. The standard analysis period was set to twenty years, consistent with an FHWA case study on project-level planning (FHWA 2005). Twenty years is also the least common multiple of lives.
User costs were omitted for simplification.
All methods returned the same
ranking, as illustrated in Table 13, in support of validating EUAC as a valid pavement preservation LCCA method.
Table 13 Comparable EUAC and PV Rankings, Continuous State PAVEMENT TREATMENTS
Agency Costs
Analysis Period
Rank
EUAC ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
3,408 4,150 4,367
5 10 10
1 2 3
Present Value - Shortest Life ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
15,172 20,463 21,343
5 5 5
1 2 3
Present Value - Longest Life ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
30,344 33,663 35,423
10 10 10
1 2 3
Present Value - Standard Period & LCM ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
60,688 67,326 70,846
20 20 20
1 2 3
97
EUAC Model Validation - Terminal Mode Table 14 illustrates that in terminal state, EUAC can yield the same results as PV when service
lives
are
truncated
rehabilitation/reconstruction.
to
accommodate
the
next
expected
The situation to avoid would be to fill the 1-year gap
before terminal with another chip seal.
Although there is no change in preferred
alternative when this was calculated in the following Table, the PV values are close and could result in faulty output in other situations because although a pavement manager would not realistically fill a 1-year gap with another 5-year chip seal, the PV method may default to and obscure that calculation in the LCCA.
Table 14 EUAC and PV Results (no user costs), Terminal State at 6yrs Agency Costs
Analysis Period
Rank
EUAC ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
3,408 5,553 5,889
5 6 6
1 2 3
Present Value - Shortest Life ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
15,172 29,111 30,871
5 5 5
1 2 3
Present Value - Rehab year, Fill the gap for Chip Seal ODOT Standard 5/8" chip seal (5-yr) OGFC (10-yr) 1" Hot Mix Asphalt mill/inlay (10-yr)
27,633 29,111 30,871
6 6 6
1 2 3
PAVEMENT TREATMENTS
EUAC Model Validation – Deterministic and Stochastic Modes The deterministic and stochastic modes of the EUAC models proposed in this research were internally validated in a number of ways (Pallisade 2011, FHWA 2001). First, model calculations were manually verified by calculator (Pallisade 2011, FHWA 2001). Secondly, the same input values were entered into the FHWA RealCost Model and the EUAC models to verify that both yielded the same preferred alternatives. Lastly, the stochastic EUAC model was checked for errors (Pallisade 2011).
98
Manual Verification of EUAC Models For the deterministic EUAC model, output of $4,696 was verified by calculator for the 1” HMA treatment in Chapter 4 using the EUAC equation when the agency and user costs total is $38,092, service life is 10 years and discount rate is 4%. The stochastic EUAC model can be validated using the same method and equation used for the deterministic model validation by selecting iterations of the simulation (Pallisade 2011, FHWA 2001), like 1-7 of 10,000 iterations shown in Figure 78.
Figure 78 EUAC Simulation Output. For example, iteration 1 was calculated based upon a 10.15 year service life using a 2.92 value for discount rate and $23,691.15 for agency costs. Future maintenance costs of $3,049 are expected to occur at years 3, 6 and 9, with the discounted values being $2,797, $2,566 and $2,353 respectively.
Manually calculating the EUAC based
on these values returns the output value listed in iteration 1 (less rounding error), thus, manually validating the stochastic EUAC model.
As previously stated, there are no models available for comparing alternatives on the basis of stochastic EUAC. The only readily-available model is the FHWA RealCost Model, so that was used to validate the proposed models output. However, RealCost does not provide a stochastic EUAC for comparison, so the output from the stochastic EUAC model was converted to NPV-form (Figure 79) to serve as a basis for the next 99
phase of internal validation found in the next paragraph that compares overall probabilistic results. After the conversion, a few iterations were selected to manually verify the model, as shown in Figure 79.
Figure 79 EUAC Model Output Converted to NPV-Form. Iteration 1 was calculated with a service life value of 10.97 years. Future maintenance costs are calculated to occur every 3 years, so 3 costs should be calculated given the service life. Using the PV equation to discount future costs (PV = FV * (1/(1+i)^n)), the $3,049 expected to occur at year 3 is $2,711.73 based on the 3.98 discount rate in iteration 1. PV of $3,049 expected at year 6 is $2,411.76, and year 9 is $2,144.98. These future costs are added to the present cost of $26,412.61 to get a NPV of $33,681, verified by the HMA output column. For iteration 2, only include the $3,049 maintenance cost for year 3 ($2,696.26) and year 6 (2,384.32) for the less-than-nineyear service life at a discount rate of 4.18 to get an NPV of 30,937. Convergence occurred at 650 iterations with final error of 1.55%.
EUAC Models Provide Same Rankings NPV and EUAC (continuous mode) models should return the same ranking (White et al. 2010), which is demonstrated in the previous section.
100
Stochastic EUAC Model Error Check Lastly, the stochastic EUAC model was checked for errors (Pallisade 2011). First, the correlation feature was checked for error by verifying that the random values were consistent for the correlated items of asphalt binder (μ=1.25, sd= 0.13; in $/SY) and diesel fuel (μ=0.063, sd=0.008; in $/SY), since both are refinery products and their prices rise and fall together. In Figure 80, iteration 1 contains a $1.14 value for asphalt binder pulled on the lower end of its distribution and a $0.05 value for diesel fuel pulled from the low end of its distribution, and so on.
Figure 80 Correlation Error Check in Stochastic Model. Next, the simulation setting for initial seed setting was changed from “choose randomly” to “fixed” so that the simulation would run the same iterations and produce the exact same output (Pallisade 2011). If the model in the “fixed” setting returns different output with each new simulation, this is an indication of model error. The stochastic EUAC model returned same results for two different simulations and it was concluded that the model was free from simulation error.
The following study was used to externally validate the EUAC PPT LCCA deterministic and stochastic models: “Evaluate TxDOT Chip Seal Binder Performance Using Pavement Management Information System and Field Measurement Data San Antonio District,” (Gransberg 2008, Gransberg 2007).
Twelve Farm-to-Market roads in the
TxDOT San Antonio district served as test sections and were comprised of six hot applied and six emulsion chip seals sections. Macrotexture measurements using the Transit New Zealand (TNZ) T/3 Sand Circle test described in the previous section were 101
gathered over a three-year period commencing in 2005. Treatment costs were also gathered for comparing the two chip seal types: $0.92/SY (average) for the hot applied and $0.82/SY for the emulsion. The study found that the emulsion chip seal was more cost effective than the hot applied chip seals for the test section roads on a cost-index basis.
To validate the EUAC PPT models proposed in this research, the macrotexture and cost data resulting from the test sections were processed using the same methodology previously discussed. LCCA calculations were made via EUAC models to determine if the research models yielded the same preferred alternative reported in the study.
The methodology for developing deterioration models to approximate service life based upon MTD values is as previously described in this report.
MTD values were
interpolated between the months in which measurements were taken (Table 15), then extrapolated beyond December 2007 to estimate service life. The hot applied chip seals (R2= 0.8936) were only expected to have 75% of the service life expected of the emulsion chip seal (R2=0.8521), which were 3 and 4 years, respectively. Because the emulsion chip seal has the longer service life and lower initial cost, the models should result in the emulsion having the lower LCC, consistent with the study.
Table 15 Average MTD values for chip seal test sections. Testing Date
Emulsion
Hot Applied
June 2005
3.35
3.36
August 2005
3.12
2.91
December 2005
3.46
2.97
March 2006
3.14
2.84
May 2006
2.79
2.16
July 2006
2.41
1.69
November 2006
2.27
1.61
February 2007
2.26
1.55
May 2007
2.17
1.49
September 2007
1.98
1.45
102
December 2007
1.90
1.36
The service life value for each alternative was entered into the probabilistic model as a deterministic value to produce consistent results for the purpose of validating the model. The San Antonio cost data was entered, keeping the means consistent with the average values provided by the study. Table 16 shows that the output from both deterministic and stochastic EUAC models was similar, and both show the same preferred alternative of emulsion chip seal on the basis of LCC, externally validating both models.
Table 16 External Validation Model Results. Model EUAC ($/lane-mile) Deterministic
Probabilistic μ, (SD)
Emulsion Chip Seal
1,590
1,589 (39)
Hot Applied Chip Seal
2,334
2,337 (42)
PPT
In summary, the EUAC model was validated by comparing EUAC model output with NPV model output (White et al. 2010).
Both models (methodologies) yielded the same
decision support. This illustrates the point that using different analysis periods corresponding with the differing service lives of alternatives in a LCCA does not remove the “fairness” nor does it result in differing benefits. It does, however, bypass the commonly problematic analysis period selection, associated adjust-to-fit requirements and well-cited sensitivity issues for that parameter.
The service life selection and
corresponding sensitivity can be obscured by the analysis period in present value calculations, but are exposed by this EUAC model. This allows for the sensitivity to be moved from the analysis period parameter, which may be arbitrary and uncontrollable, to the service life parameter, which allows the pavement manager to intuitively adjust and account for service life selection and sensitivity based on professional judgment.
FIELD TEST RESULTS Furnishing micro and macrotexture deterioration models based on field test data will allow maintenance engineers to use them to calculate a rational value for the service life 103
of pavement preservation treatments under consideration for a given project (Austroads 2000, TNZ 2005). Once the models are proven, they can then be used in conjunction with rationally developed failure criteria to permit LCCA as an additional treatment selection decision parameter.
DETERIORATION MODEL DEVELOPMENT Tests for both microtexture and macrotexture for this project were conducted on a monthly basis on each pavement preservation test section. The test sections associated with Phases 1 & 2 of the project will continue to be tested on a quarterly basis. The deterioration models were based upon the best available information: Phase 1 test sections contain approximately three years worth of field data, while Phase 2 sections contain two years. It is important to note that more data resulting from long term results (deterioration rate to failure) will enhance the deterioration model. This section shows that the appropriate amount of data needed to create a highly reliable model, like those shown in the following figures (Figures 81 and 82), varies between pavement preservation treatments.
The collected field data was analyzed using time series analysis (Chatfield 2004) and the results are shown in Table 17. The coefficient of determination (R2) value is used to determine whether a deterioration model is reliable for a given texture (George et al. 1989). Unlike laboratory research, the researchers engaged in field research are unable to exercise total control over all the variables. Therefore, based on the definition of the R2 value and the reported experience of a similar study of skid resistance, the team determined that if the independent variable (time) could account for 50% or more of the variation in the dependent variable (texture), then the given deterioration model added value to the pavement preservation management process (Chatfield 2004, Meier 2006). Thus, a model whose R2 value is above 0.50 the model is considered reliable and if the R2 is above 0.70 the model is highly reliable. This standard can be used when determining which treatments in Table 17 have a reliable model.
104
Open Graded Friction Course y = -0.499ln(x) + 2.9017 R² = 0.765
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0
5
10
15
20
25
30
Mean Texture Depth (mm)
Mean Texture Depth (mm)
4.0
Open Graded Friction Course with Fog Seal 5.0 y = -0.625ln(x) + 3.2829 R² = 0.7701
4.0 3.0 2.0 1.0 0.0 0
5
10
Months Sand Circle
15
20
25
30
35
Months
Log. (Sand Circle)
Sand Patch
Log. (Sand Patch)
Figure 81 OGFC Test Sections (#2 and #3), Macrotexture Deterioration
4.0
y = -0.636ln(x) + 3.8985 R² = 0.8512
3.0 2.0 1.0 0.0 0
5
10
15
20
25
30
35
Mean Texture Depth (mm)
Mean Texture Depth (mm)
Chip Seal (5/8")
Chip Seal (5/8") with Fog Seal 4.0 y = -0.386ln(x) + 2.8431 R² = 0.8962
3.0 2.0 1.0 0.0 0
5
10
20
25
Months
Months Sand Circle
15
Sand Patch
Log. (Sand Circle)
Figure 82 5/8” Chip Seal Test Sections (#8 and #9), Macrotexture Deterioration 105
Log. (Sand Patch)
30
35
Table 17 Summary of Regression-Based Treatment Deterioration Models Macrotexture Deterioration TS#
1 2 3
Treatment Calumet CaluPave Fog Seal Open Graded Friction Course w/Fog Seal Open Graded Friction Course
Logarithmic Equation
2
R Value
Microtexture Deterioration Power Equation
2
R Value
y = -0.073ln(x) + 1.0557
0.56
y = 47.285x
-0.045
0.38
y = -0.625ln(x) + 3.2829
0.77
y = 28.597x
0.08
0.56
y = -0.499ln(x) + 2.9017
0.77
y = 29.67x
0.0724
0.49
4
1” Mill and HMA Inlay
y = 0.038ln(x) + 0.4864
0.23
y = 56.652x
-0.046
0.47
5
Fog Seal
y = -0.034ln(x) + 0.95
0.21
y = 41.347x
0.0418
0.45
y = -0.084ln(x) + 1.4699
0.52
y = 34.135x
0.0839
0.70
y = -0.077ln(x) + 1.1715
0.56
y = 32.82x
y = -0.386ln(x) + 2.8431
0.90
y = 36.399x
-0.071
0.38
6 7 8
JLT Asphalt Penetrating Conditioner JLT Asphalt Planer w/Conditioner ODOT 5/8” Chip Seal w/Fog Seal
0.0974
0.79
9
ODOT 5/8” Chip Seal
y = -0.636ln(x) + 3.8985
0.85
y = 44.065x
-0.115
0.71
10
ODOT 3/8” Chip Seal
y = -0.509ln(x) + 2.4198
0.85
y = 54.729x
-0.162
0.77
11
Blastrac Shotblasting
y = 0.0028ln(x) + 1.3458
0.001
y = 52.16x
12
Skidabrader Shotblasting w/Fog Seal
y = -0.064ln(x) + 1.7158
0.34
y = 49.868x
-0.023
0.12
13
Skidabrader Shotblasting
y = -0.054ln(x) + 1.7522
0.21
y = 51.286x
-0.031
0.27
14
NovaChip
y = -0.115ln(x) + 1.4947
0.79
y = 43.901x
-0.042
0.48
y = -0.03ln(x) + 0.7103
0.28
y = 49.115x
-0.008
0.03
y = -0.075ln(x) + 1.1123
0.56
y = 39.681x
0.01
0.02
-0.004
0.02
15 16
Blastrac Shotblasting (PCC substrate) Skidabrader Shotblasting (PCC substrate)
-0.031
0.27
17
Skidabrader Shotblasting w/Nanolithium densifier
y = 0.032ln(x) + 0.5437
0.63
y = 43.78x
18
Next Generation Diamond Grinding
y = 0.0786ln(x) + 1.7303
0.26
y = 44.871x
19
Diamond Grinding
y = -0.048ln(x) + 1.1316
0.31
y = 44.14x
20
Polycon E-Krete
y = -0.139ln(x) + 1.1635
0.73
y = 28.564x
0.0569
0.10
y = -0.34ln(x) + 3.0577
0.75
y = 49.636x
-0.13
0.76
y = -0.287ln(x) + 2.3926
0.83
y = 47.891x
-0.151
0.41
y = -0.048ln(x) + 1.1105
0.42
y = 45.468x
-0.02
0.17
21 22 23
ODOT Single Size Chip Seal ODOT Single Size Chip Seal w/CaluPave Microsurface
106
-0.097
-0.023
0.39 0.05
There are certain attributes within the equation of the model that must be explained to better understand what one could expect from the model’s predictions.
For the
macrotexture deterioration models, it was determined that logarithmic functions best modeled deterioration. This determination stems from comparing the coefficients of determination of several different regression equations for the list of pavement preservation treatments and selecting the function that produced the highest R2 value.
It is important to ensure that the regression output is realistic. First, the sign in front of the equation should be negative, indicating deterioration (i.e., the microtexture decreases over time). Next, the constant at the end of the equation should roughly equal the initial macrotexture for the given section.
For example, in Table 17, the
equation for the macrotexture deterioration model for the Chip Seal (3/8″) test section is y = -0.509 ln(x) + 2.4198. Therefore, the equation shows that the initial macrotexture should be approximately 240 mm, the rate of deterioration is -0.509 ln(x) and the R2 value is 0.85.
The power function was selected for microtexture modeling because it produced the best R2 values. Within these regression equations, the coefficient in front of the variable is now the initial value of the microtexture index value, (i.e. the initial skid number), and the exponent is the rate of or deterioration. Again using the microtexture deterioration model for the Chip Seal (3/8″), the equation is y = 54.729 x-0.162. Therefore, the initial measurement should be about 55, the rate of deterioration is x-0.162, and the R2 value is 0.79. This method helps determine the likelihood that a given pavement preservation treatment has a viable model to help predict its service life based on loss of texture (i.e., safety performance characteristics).
Test Section Evaluation The next step is to determine which pavement preservation techniques provide deterioration models that are reliable, and then to determine their predicted service lives based on their safety performance characteristics. This is accomplished by considering their time series analysis and extrapolating the resultant regression equation to estimate 107
service life of each treatment.
There are three possible outcomes for the different
pavement preservation treatments in the field trial: 1. Viable deterioration model 2. No model: R2 value is too low or equation constants are unrealistic 3. A constant model: Macro or microtexture measurements were roughly linear over time. The first possible outcome for the deterioration model is that the data produce a viable deterioration model. For example, the data from the Chip Seal (3/8″) test section, shown in Figure 83, produce a highly reliable deterioration model with an R2 value of 0.85. Chip Seal (3/8") 4.0
y = -0.50ln(x) + 2.419 R² = 0.847 3.5
Mean Texture Depth (mm)
3.0
2.5
2.0
1.5
1.0
0.5
0.0 0
5
10
15
20
25
30
35
Months Sand Circle
Log. (Sand Circle)
Figure 83 Chip Seal (3/8”) Actual Macrotexture Graph The classic definition for R2 states that it is a measure of “the proportion of variability in a data set that is accounted for by the statistical model” and describes “goodness of fit of the curve derived from the regression analysis” (Draper and Smith 1998). Based on this definition, the equation derived from the chip seal curve shown in Table 17 accounts for 85% of the variation in chip seal macrotexture loss. Said another way, the age of the chip seal is the major factor in its loss of macrotexture, and the regression 108
equations for each treatment will be able to reliably predict the loss of macrotexture over time to a + 15% degree of accuracy. To demonstrate, the 3/8” chip seal deterioration model is used to extrapolate the data beyond a known failure criteria of 0.9mm, shown in Figure 84. The model estimates the chip seal service life to be 20 months (TNZ 2002). The deterioration model output is validated when compared to the actual data (Figure 83) that shows failure occurs at month 20. This validates the literature and the case that deterioration models can be used to estimate service life (Noyce et al. 2005).
Extrapolated Macrotexture for Chip Seal (3/8") 3.0
Mean Texture Depth (mm)
2.5
2.0
1.5
1.0
0.5
0.0 0
5
10
15
20
25
30
35
40
Months Mean Texture Depth
Failure Line
Log. (Mean Texture Depth)
Linear (Failure Line)
Figure 84 Chip Seal (3/8”) Extrapolated Macrotexture Graph The chip seal models were externally validated by the use of the TNZ performance specification. The TNZ T/3 sand circle test monthly macrotexture data was chosen for this particular report because it can be directly compared with the Australian and New Zealand research upon which the project builds. One aspect of that research is a performance specification for chip seals that measures macrotexture after 12 months and compares it to a value that is derived from a deterioration model for a specific design life. A performance specification is defined as a measurement of “how the finished product should perform over time” (Chamberlain 1995). Specification of design life expectations is an effective means of determining long-term chip seal performance. 109
The entire TNZ specification is based on this deterioration model and is founded on the assumption that long-term chip seal service life is ultimately determined by the consequence of texture loss due to flushing (a condition also called bleeding in the US literature) (TNZ 2002).
The chip seal performance specification is New Zealand’s P/17, Notes for the Specification of Bituminous Reseals (TNZ 2002).
The philosophy behind the P/17
specification is that the texture depth after twelve months of service is the most accurate indication of the performance of the chip seal for its remaining life. The New Zealand specification also contends, “the design life of a chip seal is reached when the texture depth drops below 0.9 mm (0.035 inches) on road surface areas supporting speeds greater than70 km/h (43 mph)” (TNZ 2002). The deterioration models developed in New Zealand have directed the P/17 Specification to require the minimum texture depth one year after the chip seal is completed using Equation 8.
Td1 = 0.07 ALD log Yd + 0.9
(8)
Where: Td1 = texture depth in one year (mm) Yd
= design life in years
ALD = average least dimension of the aggregate (mm)
First, it must be stated that only the chip seal test sections shown in Table 17 can be evaluated by the model shown in Equation 8. Table 18 shows the output from that analysis. The TNZ failure criterion is 0.9mm. After 12 months of service, the 3/8” chip seal has failed the P/17 12- month criterion at all possible design lives, and at 22 months it drops below the ultimate failure criterion of 0.9mm. The opposite is true for the other chip seal sections. Thus, the chip seal models are validated. The table also shows that the standard treatment with a fog seal’s macrotexture is less than the ones without the fog seal. When viewed from the macrotexture standpoint, the fog seal marginally fills the new seal’s voids with bitumen and therefore intuitively, it must reduce
110
the macrotexture has observed in Table 18. The same is also true when comparing the two single size chip seals. Table 18 TNZ P/17 Performance Specification Comparison
Treatment
P/17 12-month Minimum Macrotexture at Given Design Life 4-year 5-year* 6-year
Before Seal Macrotexture
3/8 Chip Seal 1.17 1.29 1.37 0.94 No Fog Seal 5/8 Chip Seal 1.43 1.68 1.85 0.94 No Fog Seal 5/8 Chip Seal 1.43 1.68 1.85 0.94 w/Fog Seal Single Size 1.32 1.52 1.60 1.03 Chip Seal Single Size Chip Seal 1.32 1.52 1.60 0.92 w/CaluPave * Average design life for chip seals cited in ODOT survey
Actual 12-month Macrotexture
Current Macrotexture Months of Data MTD
1.07
32
0.66
2.18
32
1.68
1.93
32
1.45
2.49
24
2.29
2.15
24
1.50
The two OGFC sections also had highly reliable models based upon the R2 values (Table 17). There is no previously established performance failure criterion for OGFC macrotexture in the literature and the analysis shown in Table 18 is not used for this treatment. However, the analysis shown in Figure 44 for OGFC to determine service life as input to the LCCA model may be a possible solution. That analysis incorporates the NZTA ultimate failure criterion of 0.9mm as the point at which the service life ends. While that specific number comes from chip seal research and is not directly relatable to OGFC, it does indicate that given research methods can be used to measure how much macrotexture is required for OGFC to perform as designed. Then a performance failure criterion could be determined. That is beyond the scope of this project and is a future research need that may be considered by OTC and ODOT in the years to come.
In summary, the treatments in Table 17 that exhibit viable deterioration models are the chip seals, OGFC and NovaChip sections. Their R2 values show that time is the primary factor in surface texture loss.
The second possible outcome for a deterioration model is that it does not show any failure criteria or improvement, which would be considered a “no model” scenario. This 111
was observed in test sections that involved chemical treatments such as standard fog seal and penetrating asphalt conditioner. The two JLT penetrating pavement conditioner test sections show that time only accounts for roughly half of the variation in macrotexture loss. The difference between the two sections was whether or not the surface had been planed prior to applying the conditioner. The planing operation removed a small amount (~1/8”) of existing surface that theoretically “opened” the surface to permit a greater penetration of the conditioner. Data shows that the planer actually removed about 0.2mm of average macrotexture and as a result, the planed test section is literally hovering slightly above the 0.9mm failure criterion. Thus, if restoring macrotexture is the goal for a given asphalt pavement, then planing is not a good option for the maintenance engineer to select. However, these treatments have not exhibited deterioration, but rather an increase in microtexture due to the chemical being worn off the surface by traffic. It is important to note in dealing with chemical treatments that there will be an initial decrease in microtexture followed by a gradual increase back to the original values or slightly lower than original values. Figure 85, which shows the application of a chemical only affects the microtexture while the macrotexture remains constant. Coupling the use of chemical treatments with other treatments that increase texture, such as chip seals or shotblasting, can prove to be a very effective pavement preservation technique (Gransberg 2009).
112
Fog Seal 8.0
60.0 55.0
7.0 50.0 45.0 40.0 5.0 35.0 4.0
30.0 25.0
3.0 20.0 2.0
Skid Number (mu)
Macrotexture (mm)
6.0
15.0 10.0
1.0 5.0 0.0
Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09 Jul-09 Aug-09 Sep-09 Oct-09 Nov-09 Dec-09 Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 Nov-10 Dec-10 Jan-11 Feb-11 Mar-11 Apr-11
0.0
Month-Year Sand Circle
Skid
Figure 85 Fog Seal Texture Combination Graph The last possible outcome for a deterioration model is related to macrotexture or microtexture deterioration models where the texture was relatively constant and where the slopes of the texture trend lines are quite subtle. This occurred in a number of test sections during the research and may be because, with only two and three years worth of data for both phases, it is still too early in the treatments’ service lives to exhibit appreciable deterioration. For example, Figure 86 shows the asphalt mill and HMA inlay, which should last around 10 to12 years; therefore, any deterioration within the first 3 years would signal major problems (George et al. 1989). It is notable that the mill and inlay section has markedly higher microtexture than the original surface and has remained consistent and its service life will extend beyond this research project.
113
Mill and Inlay 8.0
60 55
7.0 50 6.0
45
35 4.0
30 25
3.0
Skid Number (mu)
Macrotexture (mm)
40 5.0
20 2.0
15 10
1.0 5
Jan-11
Feb-11
Dec-10
Oct-10
Nov-10
Sep-10
Jul-10
Aug-10
Jun-10
Apr-10
May-10
Mar-10
Jan-10
Feb-10
Dec-09
Nov-09
Oct-09
Sep-09
Aug-09
Jul-09
Jun-09
May-09
Apr-09
Feb-09
Mar-09
Jan-09
Dec-08
Nov-08
0 Oct-08
0.0
Month-Year Sand Circle
Skid
Figure 86 Mill and HMA Inlay Texture Combination Graph The Shotblasting test sections have also increased the macrotexture of the original surface and the macrotexture and microtexture have remained relatively constant for the duration of the project. Figure 87 shows an example of this based upon TS #11 (Shotblasting-Blastrac) data. Data from all of the Shotblasting sections (including those with fog seal and nano-lithium densifier) exhibit this behavior on both types of substrate and show that constant levels of texture can be maintained beyond the research project (May 2011 being the analysis cutoff date for this research project). These sections can also be used to increase and sustain higher texture levels.
114
Shot-Blasting (Blastrac) 60
8.0
55 7.0 50 6.0
45
35 4.0
30 25
3.0
Skid Number (mu)
Macrotexture (mm)
40 5.0
20 2.0
15 10
1.0 5 0
Aug-08 Sep-08 Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09 Jul-09 Aug-09 Sep-09 Oct-09 Nov-09 Dec-09 Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Jul-10 Aug-10 Sep-10 Oct-10 Nov-10 Dec-10 Jan-11 Feb-11 Mar-11 Apr-11
0.0
Month-Year Sand Circle
Skid
Figure 87 Shotblasting (Blastrac) Texture Combination Graph This research lays the ground work to begin to associate safety data, the level of both types of texture, with the currently used surface characteristics and traffic data to develop a more holistic view of pavement management. An example of how this process could be implemented is illustrated in Figure 88.
In this scenario, the
deterioration models, developed for a chip seal using 5/8″ aggregate, are used to determine trigger points.
This alerts pavement engineers to start the project
development process in order to implement a pavement preservation project before the pavement fails. Recognizing that every state DOT’s project development process is different, trigger point calculations would vary.
115
Figure 88 Deterioration Models for Chip Seal (5/8″) A hypothetical example with 12 months duration is used to illustrate the proposed process for integrating pavement texture into the pavement management process. The first step is to collect macro and microtexture data for the roads under management and develop the deterioration models as previously described. Next, failure criteria must be established to allow the computation of a rational trigger point. In this example, the macrotexture failure criterion of 0.9mm used in New Zealand is used (TNZ 2002). The pavement engineer then extrapolates the deterioration curve to the failure value, then 116
back calculates a macrotexture trigger value of 0.969mm. This is too precise to measure with the available test equipment. So a value of 1.0mm is chosen, which generates an alert 16 months prior to failure, giving the engineer a period greater than required project development and authorization period to initiate action to preserve this pavement. The same approach can be used with microtexture.
The example uses skid number criteria found in the literature to develop a skid number failure criterion. Minnesota, Wisconsin, Maine, and Washington have cutoff or acceptable value limits of 45, 38, 35, and 30, respectively (Noyce et al. 2005). These values are the trigger points used by each DOT, and therefore this example will use 30, the lowest published trigger value and 25 as the failure criterion to illustrate the proposed process. One can see that a trigger value for microtexture is reached early on in the life of the section, at month 29, yet this gives the pavement management engineer almost 9 years before ultimate failure to act. The deterioration also indicates that this particular treatment will likely fail due to macrotexture loss before it reaches the microtexture failure point. However, a chip seal in a different location that uses a less durable aggregate that is susceptible to polishing would have a deterioration model that predicted failure much earlier. If the failure occurred before the macrotexture trigger value, the pavement manager would start the project development process upon reaching the microtexture trigger point. By using safety trigger values in this manner and with the aid of a standard pavement management system (PMS), the pavement engineer should now be able to better prioritize which roads need to be preserved first and what treatments need to be placed, further ensuring that the “right treatment is placed on the right road at the right time” (Galehouse et al. 2003).
Testing apparatus have recently become available to maintain network wide data for both types of safety criteria; for microtexture, the ASTM E274 skid tester with ribbed tires serves as a texture index, and for macrotexture the ASTM E1845 can be used along with a high speed laser to determine mean profile depth at highway speed (ASTM 2009). Therefore, it appears the time-consuming macrotexture test procedure can be replaced with the high speed laser, making data collection both safer and more efficient 117
since it can be collected without the need to establish traffic control. With high speed data collection and the above-described methodology for texture deterioration modeling, a pavement management engineer now has the tools to better incorporate surface texture safety criteria into the PMS system.
Cost Analyses Results LCCA was conducted on all field treatments and the results are shown in Table 19. Treatment agency costs for deterministic LCCA were found in literature (SHRP 2011; NCHRP 634, Stroup-Gardiner and Shatnawi, 2008; FHWA, 2005; Bausano et al. 2004), product literature and ODOT survey (Table 4).
118
Table 19 Life Cycle Cost Analysis Results for Field Treatments TS#
Treatment
Cost (Average, $ per SY)
Service Life (Range, in Years)
EUAC (Average, $ per lane mile)
1
Calumet CaluPave Fog Seal
0.85
2-4
2,200
2
Open Graded Friction Course w/Fog Seal
4.25
8 - 12
4,900
3
Open Graded Friction Course
3.75
8 - 12
4,200
4
1” Mill and HMA Inlay
4.00
8 – 12
4,400
5
Fog Seal
0.65
2-4
1,700
6
JLT Asphalt Penetrating Conditioner
0.90
3-5
1,800
7
JLT Asphalt Planer w/Conditioner
1.50
3-5
2,900
8
ODOT 5/8” Chip Seal w/Fog Seal
2.60
4–6
4,500
9
ODOT 5/8” Chip Seal
1.77
4–6
3,400
10
ODOT 3/8” Chip Seal
1.77
4–6
3,400
11
Blastrac Shotblasting
2.25
3-7
3,600
12
Skidabrader Shotblasting w/Fog Seal
2.55
3-7
4,100
13
Skidabrader Shotblasting
1.75
3-7
2,800
14
NovaChip
4.00
8 - 10
4,200
2.25
3-7
3,600
1.75
3-7
2,800
3.15
3-7
4,900
15 16 17
Blastrac Shotblasting (PCC substrate) Skidabrader Shotblasting (PCC substrate) Skidabrader Shotblasting w/Nanolithium densifier
18
Next Generation Diamond Grinding
5.00
8 - 15
4,400
19
Diamond Grinding
4.00
8 - 15
3,500
20
Polycon E-Krete
7.00
4-6
11,100
21
ODOT Single Size Chip Seal
2.25
4–6
3,600
22
ODOT Single Size Chip Seal w/CaluPave
3.10
4–6
4,900
23
Microsurface
2.50
7 - 10
2,500
119
Comparison of Sensitivity Analysis Methods The deterministic and stochastic methods based upon empirical service life data provide different results. The deterministic EUACs for example treatments are in Table 20. Based upon current bid tabulation data, the most likely scenario value for chip seal has only 43% probability, whereas the HMA has a 70% probability, as shown in Table 21.
Table 20 Deterministic LCCA Results. EUAC ($/lane-mile), Deterministic Method with Sensitivity Analysis 3%
4%
5%
EUAC
EUAC
EUAC
PPT
(Service Life)
(Service Life)
(Service Life)
Chip Seal
$3,019 (6 YR)
$3,651 (5 YR)
$4,557 (4 YR)
OGFC
$3,771 (12 YR)
$4,460 (10 YR)
$5,117 (8 YR)
1” HMA Mill & Inlay
$3,963 (12 YR)
$4,696 (10 YR)
$5,413 (8 YR)
Table 21 Deterministic and Stochastic EUAC Comparison. EUAC ($/lane-mile), Deterministic EUAC with Corresponding Probability Deterministic
Deterministic
Deterministic
Best Case
Most Likely Case
Worst Case
EUAC
EUAC
EUAC
(Probability)
(Probability)
(Probability)
Chip Seal
$3,019 (12
>12 YR
10-12-12
>12 YR
N/A
10-12-12
1” HMA Mill & Inlay
The Monte Carlo simulation was set to “fixed” mode so that it would run the same iterations run for the stochastic EUAC values in Table 10, so that differences can only be attributed to the change in service life treatment. Performance-based service life values are listed in Table 23, along with corresponding EUACs. The different EUACs demonstrate LCCA sensitivity to deterministic and probabilistic treatment of service life input.
121
Table 23 Performance-Based Stochastic EUAC Results. EUAC ($/lane-mile), Stochastic-Method, Performance Based Deterministic Probabilistic μ (SD) μ (SD) μ (SD) (Macrotexture) (Microtexture) (Triangular) Chip Seal
$5,340 (282)
$2,712 (245)
$3,466 (774)
OGFC
$3,904 (688)
$3,904 (688)
$4,105 (710)
N/A
$3,839 (706)
$4,009 (727)
1” HMA Mill & Inlay
Stochastic LCCA: Empirical vs. Performance-Based Treatment of Probabilistic Service Life The chip seal stochastic EUAC value based upon the empirical 4-5-6 service life distribution was $3,738 (std. dev: $414), as noted in Table 27. When compared to the EUAC using the performance-based 3-5-9.1 service life distribution ($3,466, sd: $774; Table 23), it is apparent that the LCCA is sensitive to the service life parameter, as noted by the different EUAC values. The performance-based EUAC exhibits greater variability due to the wider range in possible service life values.
Service Life Treatment Method Impact on LCCA Figure 89 shows distributions based upon the four service life assumptions for stochastic analyses: (1) 4-5-6 empirical service life, shown by the black distribution, (2) 3-5-9 performance-based service life, shown by the blue distribution, (3) 3-year performance-based (macrotexture) deterministic service life, shown by the red distribution and (4) 9.1-year per performance-based (microtexture) deterministic service life, shown by the green distribution.
122
Figure 89 Impact of Service Life Method on Chip Seal EUAC The deterministic performance methods put the distributions on far ends of the chip seal EUAC spectrum, while the triangular (stochastic) distributions cover the middle portion. Also, the stochastic treatments (black and blue) differ from each other. This illustrates the deterministic-treatment effect, the stochastic-treatment effect and the performancebased treatment effect on the service life input parameter, and subsequently the LCCA output.
LCCA Sensitivity to LCCA Selection Method Chip
seal
deterministic
EUACs
(Table
20)
and
performance-based
EUACs
(deterministic treatment, Table 23) are charted on the empirically-based stochastic EUACs (Table 10) cumulative curve for chip seal in Figure 90.
123
Figure 90 Deterministic and Performance-Based Chip Seal EUACs on Stochastic Empirical-Service-Life Cumulative Curve. Although empirical and performance-based stochastic LCCA both result in chip seal being the preferred alternative (Table 24), the performance-based mean EUAC values do not fall within the 90% probability range of the empirical LCCA. This shows that although the ranking is the same and it may be the most cost effective choice, the output is different and empirical LCCA may miss the budgetary mark (Bilal et al. 2010, Reigle and Zaniewski, 2002).
LCCA results in rankings that order alternatives on the basis of cost effectiveness. This study has shown that LCCA results are sensitive to LCCA method, as shown in Table 24 by the different rankings based only on LCCA method selection.
124
Table 24 LCCA Sensitivity to LCCA Selection. EUAC Rankings, Various LCCA Methods Performance Based (TABLE 23)
Stochastic,
Deterministic
Deterministic
Empirical
(TABLE 20)
(TABLE 10)
MACRO
MICRO
Stochastic
Chip Seal
1
1
2
1
1
OGFC
2
3
1
3
3
1” HMA Mill & Inlay
3
2
N/A
2
2
(1 – Lowest EUAC; 3 – Highest EUAC)
The analysis reveals that chip seal would be the preferred alternative in all of the LCCA methods except for the deterministic performance-based (macrotexture) LCCA. If only the OGFC and HMA were being considered, it appears that the HMA would be the preferred alternative in all but the deterministic LCCA case. The assumed cost for OGFC is less than HMA, shown by its better deterministic analysis ranking. However, in the rest of the analyses which use actual bid tabulations, the HMA is better positioned, showing the superiority of using stochastic methods. The different rankings for alternatives based upon deterministic-based, stochastic-based and performancebased methods reveal the sensitivity of LCCA to LCCA method selection and demonstrate the need for consideration.
Skid Number Cost Index Analysis The final type of cost analysis conducted on the pavement preservation field trials is the skid number cost index analysis, as presented in the previous section. This technique allows the analyst to measure the “bang for the buck”. The alternative with the lower cost index number is the more cost effective option. The idea is to change the decision criterion from “minimize cost” to “maximize value” by having all the necessary decisionmaking information in one place. Treatment costs are from Table 19 and the skid number is from the average of microtexture measurements taken on the test sections
125
during the first 24 months of service. The results are listed in Table 25. The skid number cost indices are listed in ascending order. Table 25 Skid Number Cost Index for Pavement Preservation Treatments Average Skid Number (24 months)
Skid Number Cost Index (SNCI)
Fog Seal
45.2
101
Calumet CaluPave Fog Seal
39.9
150
JLT Asphalt Penetrating Conditioner
41.5
153
Skidabrader Shotblasting
48.0
257
JLT Asphalt Planer w/Conditioner
41.1
257
Skidabrader Shotblasting (PCC substrate)
39.8
310
ODOT 3/8” Chip Seal
39.1
319
Blastrac Shotblasting
48.7
325
Blastrac Shotblasting (PCC substrate)
48.2
329
ODOT 5/8” Chip Seal
34.1
365
Skidabrader Shotblasting w/Fog Seal
47.6
377
Microsurface
43.5
405
ODOT Single Size Chip Seal
38.3
414
Skidabrader Shotblasting w/Nanolithium densifier
43.4
511
1”Mill and HMA Inlay
51.6
546
ODOT 5/8” Chip Seal w/Fog Seal
33.2
551
ODOT Single Size Chip Seal w/CaluPave
35.1
622
Diamond Grinding
42.1
669
NovaChip
40.2
700
Open Graded Friction Course
34.8
759
Open Graded Friction Course w/Fog Seal
34.4
870
Next Generation Diamond Grinding
36.4
967
Polycon E-Krete
32.9
1498
Treatment
126
So given the information on skid change, macrotexture and the financial facts, the maintenance engineer can now determine if spending more per lane mile is justified by the increase in skid number. The seal coats have provided the most “bang for skid number buck” during this period due to their relatively low cost and ability to maintain skid, while the E-Krete option has provided the least due to its substantially higher cost. As with other cost analyses, the SNCI does not provide the answer, only insight into a treatment’s relative cost on the basis of performance. Other issues, like location and traffic level, should be considered when making the final decision.
ANALYSIS OF PROJECT The most important research finding deals with the limitations of the two macrotexture test procedures. The research team observed that each procedure seemed to be more accurate on certain treatments. Macrotexture depths and relative differences between both methods were observed in November 2009. In Figure 91, macrotexture values between the two methods are given as a percentage. As is seen in this figure, the largest relative difference between the test methods are on TS8, TS9, TS10, TS11, TS21, and TS22. Those test sections showed a difference between the two test methods that is greater than 30%. The test sections, TS8, TS9, TS10, TS21 and TS22, are all chip sealed surfaces, which have greater macrotexture depths than the other surface treatments. Thus, the outflow meter test outflow times are very low, which yields calculated macrotexture depths that are excessive. Chip seal surface smoothness that is not regular due to different aggregate dimensions, creates a situation where the bottom surface of the outflow meter test device cannot completely cover the road surface causing the water to flow out very quickly. In many cases the outflow time was one second or less. This creates a limitation within the equipment since the smallest measurable unit of time is one second. For these reasons, the calculated macrotexture depth differences are great when compared to the sand circle. TS11 consists of shotblasting on hot mix asphalt. This section has structural and capillary cracks on the road’s surface, which affect the outflow time by providing a channel for the water
127
to pass that is not related to macrotexture. This explains the high relative difference between the two methods on this test section
In other test sections, the relative macrotexture value differences between the two test methods were less than 30%. The lowest differences occurred in test sections 2, 3, 6, 7, 12, 13, 14, 16, 18, and 23. These test sections had macrotexture depths between 1.00 – 2.00 mm (0.04 – 0.08 in.) and the difference in the two test methods is less than 25%. In test sections 1, 5, 15, 17 and 20, macrotexture depths are less than 1.00 mm (0.04 in.) and the difference between the two methods is between 25% and 30%. In TS4 the macrotexture depth is less than 1.00 mm (0.04 in.) and the difference is 18.6%.
Figure 91: Test Sections Macrotexture Results and Differences of Two Test Methods in November 2009 Figure 92 illustrates the average macrotexture results and differences between the two test methods in the test sections over a total of 24 months. This graph’s results are similar to the results shown in Figure 91, the results for a single month. Therefore, 128
Figure 92 validates the Figure 92 trend showing differences between the two methods are high on surfaces where macrotexture depth is high (roughly greater than 1.5 mm (0.06 in.)) and low (roughly less than 1.00 mm (0.04 in.)). This leads to the conclusion that each method has its own inherent functional limitations. The outflow meter is not ideal for high macrotexture surfaces because it cannot measure outflow times less than one second. The sand circle’s limitation is for low macrotexture surfaces. The limitation here is the ability of the engineer to be able to reliably observe when all the voids have been filled and stop expanding the sand circle. On a totally smooth surface such as glass, the circle would be one grain of sand deep and could be theoretically expanded to infinity since there are no voids to fill. In fact, NZTA (TNZ 2005, TNZ 1981) specifies the functional limit of sand circles to be 300 mm (11.8 in.) in diameter or less. Any larger measurements are deemed to be unreliable. The results of these analyses indicate that neither test method is appropriate for all surfaces.
Figure 92: Average Macrotexture Results and Differences between Two Test Methods Over 24 Months Figure 93 shows the percentage difference in calculated macrotexture values versus outflow time. It shows that the relative change in macrotexture is very high in the initial seconds of the outflow meter test. For instance, if the outflow time were to be 0.1 second, which cannot be measured by the current device, then the calculated macrotexture is 31.7 mm (1.25 in.), and if outflow time is 1.0 second then macrotexture is calculated 3.75 mm (0.15 in.). The difference between those values is 88.20%. Since the device cannot measure outflow times of less than 1.0 seconds, the engineer will get 129
the same outflow time value across the range from 3.7mm to 31.8 mm (0.15 in. to 1.25 in.). Since macrotexture values decrease as the outflow time increases, this trend continues until the curve flattens out. For instance, the calculated macrotexture value changes 41.52 % between 1-2 seconds, 23.67% between 2-3 seconds, 15.50% between 3-4 seconds, and 11.01% between 4-5 seconds. If outflow time is more than 5 seconds, macrotexture changes per second of outflow time are less than 10 %. This leads to the conclusion that the 5th second of outflow time portrays a functional limiting point past which the calculated macrotexture values become more reliable. Taking this information, one can infer that the outflow meter method should not be used on surfaces that result in outflow times less than 5 seconds. This translates to a macrotexture value of 1.26 mm (0.05 in.) or more.
Figure 93: Macrotexture Percentage Differences between Seconds in Outflow Meter Test Macrotexture curves that are derived from outflow meter and sand circle methods are shown as a theoretical curve in Figure 94. It shows that across the initial 5 seconds in the outflow meter test the macrotexture curve is steep which means measurements will 130
be unreliable. Hence, if outflow time is less than 5 seconds then the sand circle method should be used for macrotexture measurements. The outflow meter and sand circle curves cross at 0.79 mm (0.03 in.) macrotexture value. This value is equal at the 20th second in outflow meter method and a sand circle with a diameter of 265 mm (10.4 in.). The sand circle diameter is large because the surface’s macrotexture values are low. This value is close the NZTA specified maximum diameter of 300mm (11.8 in.). The difficulty of creating a large circle during the testing, results in a testing error and reproducibility. The outflow meter method is faster and easier than the sand circle test and should be used on surfaces where the macrotexture value is less than 0.79 mm (0.03 in.).
Figure 94: Theoretical Curves of Outflow meter and Sand Circle Tests Determining macrotexture on pavement correctly and quickly is important for safety and economy in pavement preservation testing. This study investigated and compared two methods commonly used to determine macrotexture on pavement surfaces: the outflow meter and the sand circle test. The research and analysis results show that there are functional limitations in each method’s ability to accurately measure pavement 131
macrotexture. The outflow meter provides users with results measured in seconds. It is portable, practical on wet surfaces, inexpensive, and fast, but the measured outflow time can be inaccurate for pavement preservation treatments with high macrotextures. The opposite is true for the sand circle method which should be avoided on surfaces with low macrotexture. This results in the following recommendations for appropriate use of each test method: •
If macrotexture < 0.79mm (0.03 in.), use the outflow meter only.
•
If macrotexture > 0.79mm (0.03 in.) and < 1.26mm (0.05 in.), either test is appropriate
•
If macrotexture > 1.26mm (0.05 in.), use the sand circle test only.
Previous studies have been conducted to establish relationships of various test methods to measure macrotexture. However, those typically looked at a single surface treatment and as a result did not create an opportunity to observe the relative differences between two or more macrotexture measurement methodologies. The results discussed above are the first to give quantitative guidance to researchers and practitioners regarding trigger points where the two test methods become most appropriate for differing pavement surfaces. It is recommended that the macrotexture limitations for each test method should be contained in specifications for each test to ensure that those agencies that use these tests are made aware of each test’s functional limitations.
In addition to the two measurement methods listed in the previous section, two other quantitative measures of pavement texture were taken on selected test sections using the RoboTex and the CTM during Phase 2.
As shown in Figure 95, all four
measurement techniques provide similar MTD for test sections 3 (OGFC) and 4 (HMA). The MTD results are vastly different between measurement techniques on the Next Generation Diamond Grinding section. For the rest of the test sections, all but the outflow meter provided similar results. The greatest discrepancy between the outflow meter and sand circle results occurred in the chip seal sections, as discussed in the previous section.
Measurements yielded by RoboTex, CTM and sand circle were 132
similar for the chip seal sections (#9 and 21), further validating the sand circle test as the appropriate measure for treatments with high macrotexture.
Figure 95 Macrotexture Measurements Using Different Techniques The comparable results between the higher tech methods and the low tech method (the sand circle) show that the sand circle is an appropriate method for measuring macrotexture.
One study (NCAT 2004) found that the sand circle exhibited less
variability than the CTM method.
CONCLUSIONS This project has accomplished a number of research milestones. First, it developed a robust protocol for field test research that is both consistent and repeatable. The research produced a previously unpublished model, with deterministic and stochastic capability, for the LCCA of specific pavement preservation treatments. That EUACbased model is the major contribution to the body of knowledge in pavement economics. The research also developed a methodology for developing pavement preservation treatment-specific deterioration models and demonstrated how these provide a superior result to those based on empirical service lives. Finally, the research 133
demonstrated how the new model could be utilized to assist an ODOT maintenance engineer in selecting the most cost effective pavement preservation treatment for a given pavement management problem.
Other contributions are as follows: •
EUAC-based LCCA models for both asphalt and concrete pavements using conspicuous service life and short-term cost efficiency analysis
•
Developed the concepts continuous mode LCCA for maintenance treatments based on individual service life as analysis period and terminal mode which automatically truncates service life due to next expected rehabilitation or reconstruction encroachment and corresponding sensitivity analysis
•
A model which can minimize analysis period selection issues and their associated sensitivity
•
Microtexture and macrotexture deterioration model data usage to reduce service life selection sensitivity based on local conditions
•
Rational method for determining the timing of treatment application by using the deterioration model for each treatment to identify trigger points
•
Quantifying the range of macrotexture where the sand circle and the outflow meter are most appropriate.
The Equivalent Uniform Annual Cost (EUAC) method was found to be the most efficient method to determine the cost effectiveness of treatment alternatives.
Specific
pavement-preservation LCCA adaptability issues were addressed, and subsequently the research contribution has made, by building LCCA asphalt and concrete models based on EUAC. The process is less complex, more consistent with investment level, more efficient, and provides relevant decision-making information for the pavement manager applicable to the short term window of pavement treatment operations based on treatment-relevant input. Maintenance funding is authorized on an annual basis making comparing alternatives on an annual cost basis more closely fit the funding model than using NPV which would assume availability of funds across the treatment’s entire service life. Since pavement managers typically consider several alternatives 134
with varying services lives based on available funding rather than technical superiority, the FHWA LCCA method based on NPV creates more problems than it solves.
Deterioration models for the chip seal, OGFC and NovaChip test sections were found to be highly reliable. Models for other treatments were not as highly correlated. This leads to the conclusion that factors other than treatment age have more impact on micro and macrotexture deterioration. The project did not attempt to determine what those factors were, but it can be speculated that they are the same as for any pavement, traffic, weather, etc.
Economic and engineering technical data gathered from pavement preservation field trials can be quantified and correlated to produce meaningful, standardized economic and life cycle cost analysis (LCCA) information that would assist pavement managers in selecting an alternative that would yield extended service lives of Oklahoma pavements. Life cycle cost analysis can be correlated with engineering field data to assist Oklahoma Department of Transportation (ODOT) pavement managers in determining the “right treatment” component of the “right treatment for the right road at the right time” (Galehouse et al., 2003) pavement preservation strategy and increase the effectiveness of budget expenditure resulting in decision making validation and justification and enhanced stewardship.
RECOMMENDATIONS There are limited performance measures that are globally applied to all PPTs. PPTspecific performance measures with associated failure criteria need to be developed to furnish metrics that describe treatment performance. Doing so would provide better data for deterioration models/LCCA service life input. Deterioration models could then be developed that are applicable to specific AASHTO climatic regions as well as for urban versus rural traffic. This analysis uses the NZTA ultimate failure criterion of 0.9mm as the point at which the service life ends. While that specific number comes from chip seal research and is not directly relatable to OGFC, it does indicate that given research to measure how much macrotexture is required for OGFC to perform as designed that a 135
performance failure criterion could be determined. These concepts are beyond the scope of this project and are future research needs that may be considered by OTC and ODOT in the years to come.
IMPLEMENTATION/TECHNOLOGY TRANSFER Technology transfer has occurred continuously during this project. It has occurred at a number of levels. First, at the local level, Caleb Riemer, PE is a maintenance engineer in Division 3. He has already implemented the use of several new treatments, including the shotblasting and E-krete. He has written specifications as a result and has shared those along with emerging test results with the other ODOT divisions during routine state-wide maintenance meetings. As previously stated, the scale and breadth of this project has drawn national and international attention. The research team has made 34 presentations in 48 months and published (or currently in submission) 25 journal and proceedings papers. Two technology transfer workshops were held in Norman as a part of a secondary education initiative to expose high school science teachers to engineering testing. Again, Caleb was the instructor at both. Dominique Pittenger, led a day long workshop for top performing Oklahoma high school students in July 2010 which included a look at the test procedures used by pavement engineers in the field on this OTC project. Finally, research partnerships have been developed with the University of Waterloo in Canada and Suliyeman Demirel University in Turkey. An OU research assistant traveled to Canada last fall, and the project is benefitting from the work done by a PhD student from Turkey who is spending a year working with the OU pavement preservation research team. Listed below are the products produced by the research team in the past two years.
JOURNAL AND PROCEEDINGS PAPERS The team has produced a total 25 publications (or submissions) in four years. They are as follows: 1. Gransberg, D.D., “Preseal Surface Texture as a Chip Seal Performance Predictor,” Pavement Preservation Journal, Foundation for Pavement Preservation, Summer 2008, p 13.
136
2. Gransberg, D.D., “Comparing Hot Asphalt Cement and Emulsion Chip Seal Binder Performance Using Macrotexture Measurements, Qualitative Ratings, and Economic Analysis,” 2009 Transportation Research Board, Paper #09-0411 January 2009. 3. Gransberg, D.D., “Surface Retexturing with Shotblasting is a Pavement Preservation Tool,” Roads and Bridges, April 2009, p.44. 4. Gransberg, D.D. and M. Zaman, “Qualitatively Describe Precoat Status to Track Performance of Chip Seals,” Pavement Preservation Journal, Foundation for Pavement Preservation, Fall 2009, pp. 37-38. 5. Gransberg, D.D., “Life Cycle Cost Analysis of Surface Retexturing with Shotblasting as a Pavement Preservation Tool,” Transportation Research Record 2108, Journal of the Transportation Research Board, National Academies, December 2009 pp. 46-52. 6. Gransberg, D.D. and E. Scheepbouwer, “Performance Based Construction Contracting: The US versus the World,” 2010 Transportation Research Board, Paper # 10-0093, National Academies, Washington, D.C., January 2010, pp. 7. Gransberg, D.D., E. Scheepbouwer and S.L. Tighe, “Performance-Specified Maintenance Contracting: The New Zealand Approach to Pavement Preservation,” Proceedings, 1st International Conference on Pavement Preservation, Transportation Research Board Newport Beach, California, April 2010, pp103-116. 8. Riemer, C., D.D. Gransberg, M. Zaman, and D. Pittenger, “Comparative Field Testing of Asphalt and Concrete Pavement Preservation Treatments in Oklahoma,” Proceedings, 1st International Conference on Pavement Preservation, Transportation Research Board, Newport Beach, California, April 2010, pp.447-460 9. Gransberg, D.D., and E. Scheepbouwer, “Infrastructure Asset Life Cycle Cost Analysis Issues,” 2010 Transactions, AACE, International, Atlanta, Georgia, June 2010, pp. CSC.03.01- CSC.03.8. 10. Gransberg, D.D., and S. Mueller, "Key Findings from NCHRP Synthesis 342: Chip Seal Best Practices," Pavement Preservation Journal, Foundation for Pavement Preservation, Spring 2010, pp. 18-19. 11. Aktas, B., D.D. Gransberg, C. Riemer, and D. Pittenger, “Comparative Analysis of Macrotexture Measurement Tests for Pavement Preservation Treatments, Transportation Research Record, Journal of the Transportation Research Board, National Academies, Issue 2209, 2011, pp 34-40.
137
12. Pittenger, D., “Sustainable Airport Pavement Practices,” 2011 Transportation Research Board, Journal of the Transportation Research Board, National Academies, Issue 2206, 2011, pp 61-68. 13. Pittenger, D., D.D. Gransberg, M. Zaman, and C. Riemer, “Life Cycle Cost-Based Pavement Preservation Treatment Design,” 2011 Transportation Research Board, Journal of the Transportation Research Board, National Academies, Issue 2235, 2011, pp 28-35. 14. Pittenger, D., D.D. Gransberg, M. Zaman and C. Riemer, “Stochastic Life Cycle Cost Analysis for Pavement Preservation Treatments,” 2012 Transportation Research Record, Journal of the Transportation Research Board, National Academies, Washington, D.C. (Accepted for publication 2012). 15. Gransberg, D.D., D.M. Pittenger and S.M. Tighe, “Microsurfacing Best Practices in North America,” accepted for publication for 2012 International MAIREPAV (Maintenance and Rehabilitation of Pavements and Technological Control) 7 Conference in The University of Auckland. 16. Pittenger, D.M. and D.D. Gransberg, “Performance-Based Life Cycle Cost Analysis: A Chip Seal Field Test Case Study,” accepted for publication for 2012 International MAIREPAV7 (Maintenance and Rehabilitation of Pavements and Technological Control) Conference in The University of Auckland. 17. Riemer, C., D.M. Pittenger, D.D. Gransberg, and M. Zaman, "Preservation of Concrete Pavement Using a Lithium Densifier and Shotblasting: A Life Cycle Cost Analysis," 2012 Transportation Research Record, Journal of the Transportation Research Board, Paper #12-0531. 18. Riemer, C. and D. Pittenger, “Modeling Pavement Texture Deterioration as a Pavement Preservation Management System Tool,” 2012 Transportation Research Record, Journal of the Transportation Research Board National Academies, Paper #12-0938. 19. Gransberg, D.D. and D. M. Pittenger, 2012. “A Framework For Microsurfacing Project Selection: The Key To Long-Term Performance,” 2012 Transportation Research Record, Journal of the Transportation Research Board National Academies Paper #12-0246. 20. Tighe, S.L. and D.D. Gransberg, “Sustainable Pavement Maintenance and Preservation Practices: A Review of Current Practice, “Proceedings, 8th International Conference on Managing Pavements Assets, Santiago, Chile (Accepted for publication in November 2011).
138
21. Gransberg, D.D. and D. M. Pittenger, “Quantifying the Whole Life Benefit of Preserving Concrete Pavements using Silicon Reactive Lithium Densifier and Shotblasting – A Promising New Technology. Research, Development, and Practice in Structural Engineering and Construction. Vimonsatit, V., Singh, A., Yazdani, S. (eds.) ASEA-SEC-1, Perth, November 28–December 2, 2012, RPS Publishers, Singapore. 22. Gransberg, D.D., and D.M. Pittenger, “Correlating Pavement Skid Data with Aggregate Imaging System (AIMS) Texture Indexes, 7th International Conference on Maintenance and Rehabilitation of Pavements, Auckland, New Zealand, August 28-30, 2012. (Abstract accepted December 2011 – full paper submitted March 2012). 23. Tighe, S. and D.D Gransberg, “Sustainable Pavement Maintenance Practices,” NCHRP Research Results Digest 365, Transportation Research Board, National Academies, Washington, D.C., 2011, ISBN: 978-0-309-22349-2; 27pp. 24. Gransberg D.D. Microsurfacing, NCHRP Synthesis 411, Transportation Research Board, National Academies, Washington, D.C.ISBN: 978-0-30914319-6, 2010, 124pp. 25. Pittenger, D.M. and C. Ramseyer, “Stochastic Pavement Preservation Treatment Life Cycle Cost Analysis Algorithm,” submitted to the 2012 International Journal of Pavement Engineering.
TECHNOLOGY TRANSFER A total of 34 presentations across a range of local, national, and international venues have been or will be made by the research team. They are listed below: 1. “Stochastic Life Cycle Cost Analysis for Pavement Preservation Treatments,”
2. 3.
4.
5. 6.
National Academy of Sciences’ Transportation Research Board Session 549, Washington DC, January 2012. “New Life Cycle Costing Model,” Southeast Pavement Preservation Partnership Conference, Oklahoma City, OK, May 2011. “Life Cycle Cost-based Pavement Preservation Treatment Design,” National Academy of Sciences’ Transportation Research Board Session 260, Washington DC, January 2011. “Evaluate Airport Pavement Maintenance/Preservation Treatment Sustainability Using Life-Cycle Cost, Raw Material Consumption and ‘Greenroads’ Standards,” National Academy of Sciences’ Transportation Research Board, Sessions 369 and 398 and International Discussion Panel, Washington DC, January 2011. “Comparative Analysis of Macrotexture Measurement Tests,” National Academy of Sciences’ Transportation Research Board, Washington DC, January 2011. “Life Cycle Cost-based Pavement Preservation Treatment Design,” Part II, Oklahoma Transportation Center/Oklahoma Department of Transportation 139
Transportation Research Day, Oklahoma City, OK, October, 2010. 7. “Comparative Analysis of Macrotexture Measurement Tests,” Part I, Oklahoma Transportation Center/Oklahoma Department of Transportation Transportation Research Day, Oklahoma City, OK, October, 2010. 8. “Life Cycle Cost-based Pavement Preservation Treatment Design,” Oklahoma Transportation Center’s First Annual Summer Symposium, Midwest City, OK, July 2010. 9. “Evaluate Airport Pavement Maintenance/Preservation Treatment Sustainability Using Life-Cycle Cost, Raw Material Consumption and ‘Greenroads’ Standards,” Oklahoma Transportation Center’s First Annual Summer Symposium, Midwest City, OK, July 2010.“Evaluate Airport Pavement Maintenance/Preservation Treatment Sustainability Using Life-Cycle Cost, Raw Material Consumption and ‘Greenroads’ Standards,” Oklahoma Asphalt Paving Association, 2010 Conference and Annual Meeting, Branson, MO, April 2010. 10. “Using Macrotexture to Measure Pavement Preservation Treatment Performance,” Invited guest lecture, Iowa State University, Ames Iowa, April 2010. 11. “Cost Engineering Applied to Pavement Preservation: The Value of Sustainable Infrastructure,” Invited guest lecture, North Carolina State University, Raleigh, North Carolina, April 2010. 12. “Cost Model for Sustainable Asphalt Pavements,” Oklahoma Asphalt Paving Association, Oklahoma City, OK February 2010. 13. “Performance Based Construction Contracting: The US versus the World,” Transportation Research Board, National Academies, Washington, D.C., January 2010. 14. “Project Delivery Method Issues of Different Transportation Modes: One Size Does Not Fit All,” Transportation Research Board, National Academies, Washington, D.C., January 2010. 15. “Comparative Field Testing of Asphalt and Concrete Pavement Preservation Treatments in Oklahoma,” Transportation Research Board, National Academies, Washington, D.C., January 2010. 16. “Pavement Preservation Through Retexturing: Status Report,” FHWA Pavement Preservation Expert Task Group, Reno, Nevada, December 2009. 17. Achieving Sustainability Through Integrated Project Delivery,” DBIA National Conference, Washington DC, November 2009. 18. “Cost Engineering Applied to Green Pavements: Measuring the Value of Sustainable Infrastructure,” Department of Civil Engineering, University of Waterloo, Waterloo, Ontario, Canada September, 2009. 19. “Airport Runway Rubber Removal: Preserving Pavements by Restoring Friction,” Canadian Airport Association Annual Conference, Toronto, Canada, September 2009. 20. “Chip Seal Binder Research in Texas: 15 Years of Progress,” Ergon Asphalt and Emulsions, Inc. Mid-Continent Sales Conference, Dallas, Texas, September 2009. 21. “Achieving Sustainability Through Integrated Project Delivery,” DBIA National Conference, Washington DC, October 2009. 140
22. “Chip Seal Binder Research in Texas: 15 Years of Progress,” Ergon Asphalt and
Emulsions, Inc. Mid-Continent Sales Conference, Dallas, Texas, October 2009. 23. “Sustainable Pavement Performance: Cost Engineering Justification,” Invited
Lecture, College of Engineering, University of Waterloo, Waterloo, Ontario, Canada, September, 2009. 24. “Building Good Roads and Keeping Them Good: A Case Study,” AACE, International, Oklahoma Section, Oklahoma City, Oklahoma, September, 2009. 25. “Performance Contracting Success in New Zealand,” AASHTO Subcommittee on Construction, Chicago, Illinois, August 2009. 26. “Pavement Preservation Research in Oklahoma,” University of Canterbury, Christchurch, New Zealand, June 2009. 27. “Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool” FHWA Pavement Preservation Expert Task Group, New Orleans, LA May 2009. 28. “Sustainable Pavement Preservation Treatments,” Fulton Hogan Technical Group Meeting, Christchurch New Zealand, April 2009 29. “Comparing Hot Asphalt Cement and Emulsion Chip Seal Binder Performance Using Macrotexture Measurements, Qualitative Ratings, and Economic Analysis,” Transportation Research Board, Washington, DC, January 2009. 30. “Life Cycle Cost Analysis of Surface Retexturing with Shotblasting as a Pavement Preservation Tool,” Transportation Research Board, Washington, DC, January 2009. 31. “Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool” OTC Research Day, Oklahoma City, OK October 2008. 32. “Chip Seal Best Practices Workshop” ODOT/OTC Workshop, Oklahoma City, OK, March 2012. 33. “Pavement Preservation Toolbox,” Fulton Hogan Performance Based Maintenance Workshop, Christchurch New Zealand, March 2012. 34. “Microsurfacing and Chip Seal Project Selection Criteria,” 16th Annual Transportation Engineering and Road Research Alliance Pavement Conference, St. Paul, MN February 2012.
TECHNOLOGY TRANSFER EVENTS Caleb Riemer has led the team’s technology transfer event effort. He developed a 2-day module based on the field and laboratory research and has delivered it once each year to the Oklahoma Science Teachers Engineering Program. Douglas Gransberg delivered a Chip Seal Technology Transfer Workshop this year at ODOT.
REFERENCES Abdul-Malak, M.-A.U., D.W. Fowler, and A.H. Meyer (1993). “Major Factors Explaining Performance Variability of Seal Coat Pavement Rehabilitation Overlays,” 141
Transportation Research Record 1338, Transportation Research Board, National Research Council, Washington, D.C., 1993, pp. 140–149.
American Association of State Highway and Transportation Officials (AASHTO), Pavement Management Guide, Executive Summary Report, AASHTO, Washington, D.C., 2001. Australian Road Research Board (ARRB), “Variation in Surface Texture with Differing Test Methods,” Unpublished report prepared for National Bituminous Surfacings Research Group by ARRB Transport Research, Melbourne, Australia, 2001. (ASTM E 1911) American Society for Testing and Materials, International, Standard Test Method for Measuring Paved Surface Frictional Properties Using the Dynamic Friction Tester, ASTM International, West Conshohocken, PA, 2009, http://www.astm.org, July 16, 2010. (ASTM E274) American Society for Testing and Materials, International, Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale Tire, ASTM International, West Conshohocken, PA, 2009, http://www.astm.org, July 16, 2010. (ASTM E 2157) American Society for Testing and Materials, International, Standard Test Method for Measuring Pavement Macrotexture Properties Using the Circular Track Meter, ASTM E2380/E2380M-09, ASTM International, West Conshohocken, PA, 2009, http://www.astm.org, July 16, 2010. (ASTM E 2380) American Society for Testing and Materials, International, Standard Test Method for Measuring Pavement Texture Drainage Using an Outflow Meter, ASTM E2380/E2380M-09, ASTM International, West Conshohocken, PA, 2000, http://www.astm.org, July 16, 2010. (ASTM E1845) American Society for Testing and Materials, International (ASTM), Standard Practice for Calculating Pavement Macrotexture Mean Profile Depth, ASTM E1845, 2009, ASTM International, West Conshohocken, PA, 2009, DOI:10.1520/E1845-09, http://www.astm.org, July 16, 2010. (ASTM E670-94) American Society for Testing and Materials, International (ASTM), Standard Test Method for Side Force Friction on Paved Surfaces Using the MuMeter, ASTM E670-94, ASTM International, West Conshohocken, PA, 2000, http://www.astm.org, October 3, 2007. Austroads, Guidelines for the Management of Road Surface Skid Resistance, Austroads Publication No, AP-G83/05, Sidney, Australia, 2005. Austroads, Austroads Provisional Sprayed Seal Design Method, Austroads, Sydney, Australia, 2000. 142
Bathina, M, “Quality Analysis of The Aggregate Imaging System (Aims) Measurements,” Master’s Thesis, Texas A&M University, College Station,Texas, 2005. Bausano, J.P., K. Chatti and R.C. Williams, “Determining Life Expectancy of Preventive Maintenance Fixes for Asphalt-Surfaced Pavements”, Transportation Research Record: Journal of the Transportation Research Board,No. 1866, TRB, National Research Council, pp. 1-8. Washington D.C., 2004. Beatty, T.L., D.C. Jackson, D.A. Dawood, and ten others, “Pavement Preservation Technology in France, South Africa, and Australia,” Federal Highway Administration Report # FHWA- PL-03-00, US Department of Transportation, Washington, DC, 2002. Bennett R., “Balstrac Shot Blasting Trial and Technical Assessment, Bendigo, Victoria,” Unpublished draft research report prepared by Geotest Civil Services, North Geelong, Victoria, Australia, pp. 4-9, August, 2007. Bilal, M., M. Irfan, and S. Labi, Comparing the Methods for Evaluating Pavement Interventions – A Discussion and Case Study, Transportation Research Board (TRB) Paper No. 09-2661, Washington, D.C., 2009. Cambridge Systematics, Inc., PB Consult and System Metrics Group, Inc., Analytical Tools for Asset Management, National Cooperative Highway Research Program (NCHRP) Report 545, Washington, D.C., 2005. Cairney, P. and E. Styles, “A Pilot Study of the Relationship between Macrotexture and Crash Occurrence,” Road Safety Research Report, CR223, ARRB Transport Research, Victoria, Australia, February, 2005. Chamberlain, W.P. Performance Related Specifications for Highway Construction and Rehabilitation, National Cooperative Highway Research Program Synthesis 212, Transportation Research Board, National Academies, Washington DC, 1995. Chatfield, C. The Analysis of Time Series: An Introduction. New York: Chapman & Hall/CRC., 2004. Clemen, R., 1996. Making Hard Decisions: An Introduction to Decision Analysis, Second Edition. Duxbury Press, London. Davies, R.M. and J. Sorenson, Pavement Preservation: Preserving Our Investment in Highways, January/February 2000· Vol. 63· No. 4, Retrieved from http://www.tfhrc.gov/pubrds/jan00/pavement.htm, August 28, 2008. Doty, R.N., “A Study of the Sand Patch and Outflow Meter Methods of Pavement Surface Texture Measurement,” Proceedings, ASTM 1974 Annual Meeting, Washington, D.C., 35pp, June 27, 1974.
143
Draper, N.R. and H. Smith, Applied Regression Analysis, Wiley-Interscience, New York, New York, p.121, 1998. Engineering News Record, “Construction Cost Index History,” McGraw-Hill Construction, Retrieved from http://enr.ecnext.com/coms2/article_echi100801constIndexHi, June 12, 2010. Federal Highway Administration (FHWA), Life Cycle Cost Analysis, Summary of Proceedings: FHWA Life Cycle Cost Symposium, Washington, DC, 1993. Federal Highway Administration, (FHWA), Pavement Preservation: A Road Map for the Future, Ideas, strategies and techniques for Pavement Preservation, Forum held Kansas City, MO, 1998. Federal Highway Administration (FHWA), Life Cycle Cost Analysis in Pavement Design, Interim Technical Bulletin.Washington, D.C., 1998. Federal Highway Administration (FHWA), Life Cycle Cost Analysis Primer, Office of Asset Management Washington, D.C., 2002. Federal Highway Administration (FHWA), Transportation Asset Management Case Studies, Economics in Asset Management: The Hillsborough County, Florida Experience, Office of Asset Management, Washington D.C., 2005. Federal Highway Administration (FHWA), Asset Management Overview, Office of Asset Management, Washington, D.C., 2007. The Federal Reserve, Economic and Research Data: Data Releases, 2011. Retrieved on July 15, 2011 from: http://www.federalreserve.gov/releases/h15/data.htm. Galehouse, L., J.S. Moulthrop, and R.G. Hicks, “Principles for Pavement Preservation: Definitions, Benefits, Issues and Barriers,” TR News, Transportation Research Board, pp. 4-9, 2003. Gee, K.W., “Preservation and Rehabilitation,” Proceedings, AEMA-ARRA-ISSA Joint Meeting, Bonita Springs Florida, p. 8, 2007. Geiger, D.R., “Pavement Preservation Definitions,” Federal Highway Administration Memorandum, Washington, D.C., p.2, September 12, 2005. George, K.P., A.S. Rjagopal and L.K. Lim, “Models for Predicting Pavement Deterioration,” Transportation Research Record 1215, Journal of the Transportation Research Board, National Academies, pp.1-7., 1989. Goodman, S., Y. Hassan, and A.O. Abd El Halim, The Road to Cost Effective Urban Pavement Friction Measurement, SURF 2004 5th Symposium on Pavement Surface Characteristics on CD, Toronto, Canada, 2004. 144
Gransberg, D.D., and E. Scheepbouwer, “Infrastructure Asset Life Cycle Cost Analysis Issues,” 2010 Transactions, AACE, International, Atlanta, Georgia, pp. CSC.03.01CSC.03.8., June 2010. Gransberg, D.D., “Life Cycle Cost Analysis of Surface Retexturing with Shotblasting as a Pavement Preservation Tool,” Transportation Research Record 2108, Journal of the Transportation Research Board, National Academies, pp. 46-52, December 2009. Gransberg, D.D., “Using a New Zealand Performance Specification to Evaluate US Chip Seal Performance,” Journal of Transportation Engineering, ASCE, Vol. 133 (12), pp 688-695, 2007. Gransberg, D.D. and B.D. Pidwerbesky, "Strip Sealing and Ultra-High Pressure Watercutting Technique for Restoring Skid Resistance on Low-Volume Roads: Life Cycle Cost Comparison,” Transportation Research Record 1989, Journal of the Transportation Research Board, National Academies, pp. 234-239, 2007. Gransberg, D.D., and D.M.B. James. Chip Seal Best Practices, National Cooperative Highway Research Program Synthesis 342, Transportation Research Board, National Academies, Washington, D.C, 2005. Gransberg, D.D. and M. Zaman, “Analysis of Emulsion and Hot Asphalt Cement Chip Seal Performance,” Journal of Transportation Engineering, ASCE, Vol.131 (3), pp. 229-238, 2005. Hall, J.W. “Guide for Pavement Friction,” American Association of State Highway and Transportation Officials, 2006, http://www.transportation.org/sites/design/docs/Friction%20Guide%20with%20AASH TO%20changes%20-%208-14-2007%20line%20nos.pdf, September 22, 2007. Hall, K.T., C.E. Correa, S.H. Carpenter, R.P. Elliott, “Guidelines for Life-Cycle Cost Analysis of Pavement Rehabilitation Strategies,” Proceedings, 2007 Transportation Research Board, Paper #07-0131, National Academies, Washington, DC, 19 pp, January 2007. Henry, J. J., NCHRP Synthesis of Highway Practice No. 291: Evaluation of Pavement Friction Characteristics, TRB, National Research Council, Washington, D.C., 2000. Lee, Jr. Douglass B., “Fundamentals of Life-Cycle Cost Analysis,” Transportation Research Record 1812, Paper No. 02-3121, Transportation Research Board (TRB), Washington, D.C., 2002. Lehtonen, R. and E. Pahkinen, Practical Methods for Design and Analysis of Complex Surveys, 2nd Edition, John Wiley and Sons, New York, 394pp., 2004.
145
Lenke, L.R. and R.A. Graul, “Development of Runway Rubber Removal Specifications Using Friction Measurements and Surface Texture for Control,” Chapter in The Tire Pavement Interface, American Society for Testing and Materials, STP929-EB, West Conshohocken, Pennsylvania, pp. 1-17, 1986. Li, Y., A. Cheetham, S. Zaghloul, K. Helali, and W. Bekheet, “Enhancement of Arizona Pavement Management System for Construction and Maintenance Activities,” Transportation Research Record: Journal of the Transportation Research Board, No. 1974, Transportation Research Board of the National Academies, Washington, D.C., pp. 26–36, 2006. Lomax, R., 2007. An Introduction to Statistical Concepts, Second Edition. Lawrence Erlbaum Associates, Publishers. Mahwah, New Jersey. Manion, M. and S.L. Tighe, “Performance-Specified Maintenance Contracts: Adding Value Through Improved Safety Performance,” Transportation Research Record: Journal of the Transportation Research Board 1990, Transportation Research Board of the National Academies, Washington, D.C., pp. 72-79, 2007. Massad, E., “Review of Imaging Techniques for Characterizing the Shape of Aggregates Used in Asphalt Mixes,” Paper presented at the 9th Annual Symp. of the International Center for Aggregate Research (ICAR), Austin, Texas, 2001. Meier, K. J. Applied Statistics for Public and Nonprofit Administration. Toronto: Thomson Wadsworth, 2006. Monsere, C.M., L. Diercksen, K. Dixon, and M. Liebler, Evaluating the Effectiveness of the Safety Investment Program (SIP) Policies for Oregon, SPR 651, Final Report, Oregon Department of Transportation Research Section., Portland, Oregon, 2009. Montgomery, Douglas C., 2009. Introduction to Statistical Quality Control, 6th Edition. John Wiley and Sons, Inc. Hoboken, New Jersey. Moulthrop, J., “Pavement Preservation: Protecting the Investment,” Presentation made at NEAUPG Annual Meeting, Wilkes-Barre, Pennsylvania, 2003. Moulthrop, J., T. Thomas, and B. Ballou, “Initial Improvement in Ride Quality of Jointed, Plain Concrete Pavement with Microsurfacing: Case Study,” Transportation Research Record: Journal of the Transportation Research Board, No. 1545, Transportation Research Board of the National Academies, Washington, D.C., pp. 3–11, 1996. National Center for Asphalt Technology, Evaluation of Circular Texture Meter For Measuring Surface Texture of Pavements, NCAT Report 04-05, Auburn University, Auburn, Alabama, 2004. National Cooperative Highway Research Program, Texturing of Concrete Pavements, NCHRP Report 634, Transportation Research Board, Washington, D.C., 2009. 146
National Cooperative Highway Research Program, Evolution and Benefits of Preventative Maintenance Strategies, Synthesis of Highway Practice No. 153, National Cooperative Highway Research Program, Transportation Research Board, Washington, D.C., 1989. National Highway Institute (NHI), Pavement Preservation Treatment Construction Guide, “Chapter 8 – Microsurfacing,” FHWA Washington, D.C., 2007, [Online] Available: http://fhwapap34.fhwa.dot.gov/NHI-PPTCG/chapter_8/unit3.htm, April 12, 2010. Neubert, T.W., “Runway Friction Measurement and Reporting Procedures,” Presentation to 2006 Airfield Operations Area Expo and Conference, Milwaukee, Wisconsin, p.3, 2006. Noyce, D.A., H.U. Bahia, J.M. Yambo, and G. Kim, “Incorporating Road Safety into Pavement Management: Maximizing Asphalt surface Friction for Road Safety Improvements,” Midwest Regional University Transportation Center, Madison, Wisconsin, 2005. Oklahoma Department of Transportation (ODOT), SFY-2010 through SFY-2013 Asset Preservation Plan, Oklahoma City, Oklahoma, 2010. Owen, M., Managing the Risk in a New Performance Based Environment, Conference on Asphalt Pavements for Southern Africa, Zimbabwe, 1999. Palisade Corporation, 2011. Palisade: Maker of the world's leading risk and decision analysis software, @RISK and the DecisionTools Suite. Palisade Corporation. Accessed January 4, 2011. http://www.palisade.com/. Patrick, J.E., P.D.Cenek, and M. Owen, Comparison of Methods to Measure Macrotexture, Proc. 1st Int’l Conf. World of Asphalt Pavements on CD, Australian Asphalt Pavement Association, Sydney, pp. 24-27, 2000. Peshkin, D.G., T.E. Hoerner, and K.A. Zimmerman, Optimal Timing of Pavement Preventive Maintenance Treatment Applications, NCHRP, Report 523 Transportation Research Board, National Research Council, Washington, D.C., 2004. Pidwerbesky, B.D., D.D. Gransberg, R. Stemprok, and J. Waters, Road Surface Texture Measurement Using Digital Image Processing and Information Theory, Land Transport New Zealand Research Report 290, 42pp, 2006. Pittenger, D.M. (2010). Life Cycle Cost-Based Pavement Preservation Treatment Design. Master's Thesis, University of Oklahoma, Norman, Oklahoma, p. 67,May 2010.
147
Reigle, J. and J. Zaniewski, “Risk-Based Life-Cycle Cost Analysis for Project-Level Pavement Management”, Transportation Research Record 1816, Paper No. 022579, 2002. Riemer, C., D.D. Gransberg, M. Zaman, and D. Pittenger, “Comparative Field Testing of Asphalt and Concrete Pavement Preservation Treatments in Oklahoma,” Proceedings, 1st International Conference on Pavement Preservation, Transportation Research Board, Newport Beach, California, pp.447-460, April 2010. Riemer, C., “Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool”, Dissertation, University of Oklahoma, Norman, OK, 2011. Roe, P.G., A.R. Parry, and H.E. Viner, “High and Low Speed Skidding Resistance: The Influence of Texture Depth,” TRL Report 367, Crowthrone, U.K., 1998. Roque, R., D. Anderson, and M. Thompson (1991). “Effect of Material, Design, and Construction Variables on Seal-Coat Performance,” Transportation Research Record 1300, Transportation Research Board, National Research Council, Washington, D.C., pp. 108–115. Sinha, K.C. and S. Labi, Transportation Decision Making: Principles of Project Evaluation and Programming, John Wiley & Sons, Inc., Hoboken, New Jersey, pp. 199-211, 2007. Speidel, D.J., “Airfield Rubber Removal,” Proceedings, 2002 Federal Aviation Administration Technology Transfer Conference, Atlantic City, New Jersey, pp. 1-7, 2002. Strategic Highway Research Program, Preservation Approaches for High-TrafficVolume Roadways, Transportation Research Board, Washington, D.C., 2011. Stroup-Gardiner, M. and S. Shatnawi, The Economics of Flexible Pavement Preservation, TRB 2009 Annual Meeting Paper, Transportation Research Board, Washington D.C., 2008. Thoreson, T., T. Martin, R. Hassain, M. Byrne, W. Hore-Lacy and G. Jameson, 2012. “Preliminary Methodology for Estimating Cost Implications of Incremental Loads on Road Pavements,” Austroads Research Report, Publication No. AP–R402-12, Sydney, Australia. Tighe, S. “Guidelines for Probabilistic Pavement Life Cycle Cost Analysis,” Transportation Research Record: Journal of the Transportation Research Board, No. 1769, TRB, National Research Council, pp. 28-38. Washington, D.C. 2001.
148
Tighe, S. and D.D. Gransberg, Sustainable Pavement Maintenance Practices, NCHRP Synthesis Report 41-04, Transportation Research Board, Washington D.C., pp.109, 2010. Titus-Glover, L. and S.D. Tayabji, Assessment of LTPP Friction Data LTPP Report # FHWA-RD-99-037, Federal Highway Administration, MacLean, Virginia, 1999. Transit New Zealand (TNZ), Notes for the Specification for Bituminous Reseals, TNZ P17, Wellington, New Zealand, 2002. Transit New Zealand (TNZ), Standard Test Procedure for Measurement of Texture by the Sand Circle Methods, TNZ T/3, Wellington, New Zealand, 1981. Transit New Zealand, Chipsealing in New Zealand, Transit New Zealand, Road Controlling Authorities, and Roading New Zealand, Wellington, New Zealand, 2005. Vercoe, J., “Chip Seal Texture Measurement by High Speed Laser,” Unpublished research report, Fulton Hogan, Christchurch, p.11, 2002. VicRoads Bituminous Sprayed Surfacing Manual (Draft), VicRoads Bookshop, Victoria, Australia, p.233, 2003. Voigt, G. F., “Moving Forward on Surface Texture Issues,” Better Roads Magazine, Vol 76(8), August 2006, pp.64-70, 2006. Walls, J. and M.R. Smith, “Life-Cycle Cost Analysis in Pavement Design — Interim Technical Bulletin,” Federal Highway Administration, FHWA-SA-98-079, pp. 28-39, 1998. White, J., K. Case, and D. Pratt, Principles of Engineering Economic Analysis, Fifth Edition, John Wiley & Sons, Inc., Hoboken, New Jersey, 2010. Yager, T.J., “Pavement Surface Properties,” Transportation Research Board, Report A2B07: Committee of Surface Properties-Vehicle Interaction, Millenium On-Line Publication, National Academies, Washington, D.C., 2000, http://onlinepubs.trb.org/onlinepubs/millennium/00087.pdf, September 26, 2009. Zaniewski, J. and M. Mamlouk, “Pavement Preventive Maintenance Key to Quality Highways,” Transportation Research Record: Journal of the Transportation Research Board, No. 1680, TRB, National Research Council, Washington, D.C., pp. 26-31, 1999.
149
APPENDIX - PAVEMENT SURFACE TEXTURE MAINTENANCE GUIDE
150
Pavement Surface Texture Maintenance Guide
Oklahoma Department of Transportation
Pavement Surface Texture Maintenance Guide
June 2012
Oklahoma Department of Transportation 200 N.E. 21st Street Oklahoma City, OK 73105
Acknowledgments This guidebook is the result of a robust partnership between the Oklahoma Transportation Center (OkTC), the University of Oklahoma (OU), the Oklahoma Department of Transportation (ODOT), and members of the pavement preservation industry from six states. Our sincere appreciation is extended to all persons who contributed information for the use in this guide.
Pavement Surface Texture Maintenance Pavement preservation is the embodiment of infrastructure stewardship. Its central theme is using pavement technology to “keep good roads good”. Pavement preservation projects are eligible for Federal-aid funding, while maintenance projects must be entirely funded by the state. There is a wide range of funding for pavement preservation and maintenance programs within the U.S. In those agencies, like the Oklahoma Department of Transportation (ODOT) that are on the low end of the funding spectrum, the need for an aggressive pavement preservation program is critical to getting as much value out of each maintenance dollar as possible. Pavement preservation is inherently sustainable as it seeks to minimize the amount of natural resources consumed over a pavement’s life cycle. Therefore, focusing on pavement preservation rather than reactive maintenance and repair furnishes a broad foundation on which to build ODOT’s pavement sustainability program.
The purpose of this guide is to provide a practical document that ODOT pavement managers can reference when restoring surface texture and skid resistance to various types of pavements throughout the state. It constitutes a surface retexturing “toolbox” that contains both the technical engineering information and the economic analysis of each treatment’s efficacy. The idea is not to identify the “best” method but rather to quantify the benefits of all the treatments in a manner that then allows the pavement engineer to select the right pavement preservation “tool” for the specific issue that needs to be addressed and satisfy the fundamental definition of pavement preservation: “put the right treatment, on the right road, at the right time.” This will effectively expand
the number of pavement preservation tools that can be used by ODOT and every single one of those tools brings federal aid eligibility along with it.
Pavement Surface Texture Maintenance Guide Purpose of this Guidebook The purpose of this guide is to serve as a tool to assist pavement engineers in a methodology for selecting the right pavement preservation “tool” for a given pavement issue. The guide combines an explanation of the various engineering and economic components of the methodology to help pavement engineers understand and implement a system that will work for them.
The goal of this guidebook is to furnish pavement managers with a methodology for quantifying the technical engineering data for pavement preservation treatments coupled with life cycle cost analysis of the costs and benefits associated with each treatment. This will allow ODOT pavement managers to have the required information to be able to make rational engineering design decisions based on both physical and financial data for a suite of potential pavement preservation tools. How to Use this Guidebook This guide presents the methodology for coupling engineering and economic data for informed decision making in a logical, step-by-step way. Measuring pavement surface texture, modeling deterioration and conducting economic analysis comprise the major components of the step-by-step process.
This guidebook is the result of two OkTC research projects, “Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool – Phases 1 & 2”, and can be referenced for a more in-depth understanding of the concepts discussed in this guide.
This guide should be supplemented with other sources of information,
regarding items such as standard testing procedures and software products.
TABLE OF CONTENTS TABLE OF CONTENTS .................................................. Error! Bookmark not defined. INTRODUCTION ............................................................................................................. 6 Pavement Maintenance Background............................................................................... 7 Pavement Texture Background ..................................................................................... 11 PAVEMENT TEXTURE MEASUREMENT .................................................................... 12 DETERIORATION MODELING ..................................................................................... 19 COST ANALYSIS .......................................................................................................... 22 REFERENCES .............................................................................................................. 49 APPENDIX A Pavement Preservation Treatment Descriptions..................................... 52 APPENDIX B Primary Capabilities and Functions of Pavement Preservation Treatments ...................................................................................................................................... 65 APPENDIX C Pavement Preservation Treatment Values (2010) .................................. 67 APPENDIX D Skid Number Cost Index Analysis (2010) ............................................... 68
INTRODUCTION Pavement skid resistance is perhaps the most important engineering component of the road from a safety standpoint. Slippery pavements are the result of several causes, chief of which is the loss of pavement surface micro and macrotexture. As a result, pavement managers must not only manage the structural condition of their roads but also their skid resistance (Gee 2007, NCHRP 1989). In fact, it is possible for a structurally sound pavement to be rendered unsafe from a loss of skid resistance due to polishing of the surface aggregate, or in the case of chip seals, flushing of the binder in the wheel paths (Patrick et al. 2000). Engineers must use every possible tool during design, construction and operations to make the road as safe as possible.
The
design/construction engineer has control over the geometry of the road, both in horizontal and vertical alignments, the speed of travel, the signage of the roadway system and the material properties of the surface course. As the pavement deteriorates over time, the maintenance engineer can also control the characteristics of the pavement surface by selecting various pavement preservation and maintenance treatments. Ultimately, the physics of the moving vehicle will determine if the engineers who have been involved in the road’s service life will determine whether or not the road can be safely traveled. Once the road is built, the only facet of the road that is truly controllable is its surface. No other factor of the complex three-dimensional equation that determines whether a moving vehicle will be able to safely remain on the surface of the road can be changed without a large relative commitment of resources to affect the desired change. As a result, the maintenance engineer’s mission must be to preserve the road’s structural capacity and to ensure that its surface frictional characteristics are sufficient to safely pass traffic for which it was originally designed.
This results in a safety requirement to modify the pavement surface to restore skid resistance. Many of the possible tools for restoring skid resistance, like chip seals, are also used for pavement structural preservation.
PAVEMENT MAINTENANCE BACKGROUND Transportation agencies in the United States have procedures in place to identify and rectify skid resistance problems. However, the procedures are often empirical and tend to be reactive rather than proactive in nature. This is not the case in some other countries. For example, Austroads, the Australia/New Zealand equivalent to AASHTO, developed and has been successfully using a set of procedures to literally manage pavement macrotexture for the past three decades (Austroads 2005). Therefore, it is not necessary to develop new procedures for the industry and the transportation agencies in this country. This guidebook provides a “customized” version of the Austroads model to suit American needs. The Austroads “Procedure to Identify and Treat Sites with Skidding Resistance Problems” uses the following five steps: 6.
“Identify [possible] treatment [alternatives]
7.
Cost works and carry out economic evaluation
8.
Shortlist schemes in priority order
9.
Carry out short-term measures, if required
10. Program longer term measures” (Austroads 2006).
Pavement managers in Australia and New Zealand have not only the engineering technical data that they need to generate a set of technically feasible options for rectifying a loss of skid resistance, but they also have the economic data required to be able to place those alternatives in the context of a limited maintenance budget. It should be noted that this approach does not merely involve selecting the lowest cost alternative. Instead Austroads requires a life cycle cost analysis to accompany all public works and as a result, selects treatment alternatives on a basis of the lowest life cycle cost not the lowest construction cost (Austroads 2005). As a result, a treatment alternative with a higher initial cost but which effectively extends the service life of a pavement for a longer period can be selected, and the long-term benefits to the agency’s multi-year budgets can be accrued.
Additionally, Austroads advocates the use of both short and long-term measures. For example, a given pavement may lose its skid resistance during the winter months where it is climatically impossible to install a bituminous surface treatment due to low ambient air temperatures. Austroads has a machine called the ultra-high pressure watercutter that can literally go out in a limited area such as a slippery superelevated curve or a freeway ramp and restore pavement macrotexture in any weather (Vicroads 2003). This would be a short-term measure. The long-term measure might involve installing microsurfacing or a new chip seal in the summer when the climatic conditions allow it. Both treatments would be included in the life cycle cost analysis used to justify the retexturing project.
On this premise, the guidebook will provide ODOT pavement engineers the tools needed to conduct a more comprehensive comparative field evaluation of various methods used to restore pavement skid resistance by retexturing the existing surface with either a surface treatment or a mechanical process. The treatments listed in Table 1 literally cover the spectrum of pavement preservation and will be the focus of this guidebook. See the Appendix for more information about each treatment. They range from Shotblasting, which merely restores macro and microtexture and uses no new raw materials, to a 1” (2.5cm) mill and inlay. The FHWA rules state that overlays that are less that 1.5” thick are not deemed to enhance structural capacity (Geiger 2005).
Table 1 Pavement Preservation Treatments
On Asphalt Substrate Surface Treatment
Chemical Treatment
Mechanical Treatment
• Fog seal • Microsurfacing • ODOT Standard 3/8” chip seal • ODOT Standard 5/8” chip seal • ODOT Standard 5/8” chip seal
• E-Krete pavement surface
• Pavement retexturing using
stabilizer • Asphalt penetrating conditioner with crack seal
shotblasting (48” width) • Pavement retexturing using abrading (72” width) • Pavement retexturing using abrading (72” width) with fog seal • Pavement retexturing using a flat headed planing (milling) technique with asphalt penetrating conditioner
with a fog seal
• Single size ½” chip seal • Novachip • Open Graded Friction Course • Open Graded Friction Course with a fog seal
• Calupave • Single Size Chip Seal with Calupave
• 1” Hotmix Asphalt mill-inlay
On Concrete Substrate Surface Treatment
Chemical Treatment
Mechanical Treatment
• Pavement retexturing using
• Pavement retexturing using
shotblasting treated (48” width) with Nanolithium densifier
shotblasting (48” width)
• Pavement retexturing using abrading (72” width)
• Diamond grinding • “Next Generation” diamond grinding
Table 2 is taken from the FHWA Memorandum on Pavement Preservation (Geiger 2005). It shows that pavement preservation treatments (shaded in green) are applied to extend the service life of a given road, NOT to enhance its capacity or structural strength.
Table 2 Pavement Preservation Guidelines (Geiger 2005) Pavement Preservation Guidelines Type of Activity
Pavement Preservation
New Construction Reconstruction Major Rehabilitation Structural Overlay Minor Rehabilitation Preventive Maintenance Routine Maintenance Corrective (Reactive) Maintenance Catastrophic Maintenance
Increase Capacity X X
Increase Strength X X X X
Reduce Aging X X X X X X
Restore Serviceability X X X X X X X X X
Table 2 also has “corrective/reactive maintenance” highlighted in yellow. Some of the treatments shown in Table 1 can also be used for this category. For example, shotblasting, skidabrading, and microsurfacing can all be applied to a structurally sound pavement that has become unsafe due to loss of skid resistance. However, if used for this purpose, federal funding will not be authorized.
Maintenance of adequate pavement skid resistance can be a pavement preservation activity (Moulthrop 2003). This intersection of two requirements creates a technical synergy that a state like Oklahoma can leverage to stretch its pavement maintenance budget if it has the necessary technical and financial information to assist decision-makers in selecting the appropriate surface treatment tool for a given situation.
The economic importance of understanding the difference between pavement preservation and pavement maintenance is huge. Pavement preservation projects are eligible for Federal-aid funding, while maintenance projects must be entirely funded by the state. This guidebook will effectively expand the number of pavement preservation tools that can be used by ODOT and every single one of those tools brings federal aid eligibility along with it.
PAVEMENT TEXTURE BACKGROUND Two common measurements used to assess surface texture are microtexture and macrotexture (Gransberg & James 2005, Roque et al. 1991).
Essentially,
microtexture is the quantitative measure of aggregate surface friction properties that contribute to skid resistance, while macrotexture is the quantitative measure of aggregate physical properties (size, shape and spacing) that contribute to drainability, whereby enhancing surface friction and skid resistance (Abdul-Malak et al. 1993).
Micro and macrotexture deteriorate over time due to traffic and
environmental
conditions.
Pavement
managers
assess
surface
treatment
performance (service life) by monitoring the deterioration rate until the surface reaches a certain threshold value that signals remedial action is required. Pavement preservation’s focus is placing “the right treatment, on the right road, at the right time” (Galehouse et al. 2003). If a road’s megatexture or IRI become unacceptable, it is too late to attempt to “preserve” the pavement. These measures are indicative of inadequate structural capacity and will require major maintenance or reconstruction treatments to rectify the loss in ride quality and to recapture the pavement’s structural capacity. Hence, understanding the relationships between microtexture and macrotexture deterioration over time, allows the engineer to establish “trigger points” that permit sufficient time to schedule pavement preservation
before
the
pavement’s
structural
capacity
is
permanently
compromised.
Benefit of Pavement Surface Texture Maintenance Methodology Implementation What pavement preservation often overlooks are other constraints a maintenance engineer also faces. These constraints are primarily budgetary in nature but also can include political or construction timelines. When an engineer determines that a roadway surface needs a treatment to address a surface characteristic deficiency, selecting the appropriate treatment is a critical decision. Currently, the engineer must rely on practical empirical data, gathered through experience in the field to estimate the impact of each treatment option and how long each treatment will last. The treatment’s service life is
important due to factors such as the availability of current and/or future funding or the timing of the next major reconstruction project. This information has historically been estimated through laboratory testing or left to the judgment of the engineer. The goal of implementing a surface texture management methodology is to standardize the pavement
preservation
treatment
selection
process,
to
assist
the pavement
maintenance engineers in making treatment selection decisions by quantifying the engineering properties of commonly available surface treatments, and applying that data to create short term deterioration models for each treatment based on actual field testing.
PAVEMENT TEXTURE MEASUREMENT Various measurement methods are available to gather sufficient data to develop deterioration models. Five tests are presented in this guide: 1. Skid resistance measured by ODOT’s skid trailer, 2. Macrotexture measured using a New Zealand sand circle test, 3. Macrotexture measured using a Hydrotimer outflow meter, 4. Macrotexture measured using a Circular Track Meter, 5. Macrotexture measured using the RoboTex device (selected test sections).
Microtexture Measurements Microtexture of the pavement surface (skid resistance) can be measured by ODOT’s skid tester (ASTM E274). ODOT’s skid trailer (Figure 1) uses a ribbed or smooth tire to produce skid measurements. The skid numbers can be averaged to eliminate any irregularities due to slight variations in the test location to provide the average skid number for a given section. Tests can be conducted on the outside wheel path.
Figure 1 ODOT Skid Trailer Macrotexture Measurements Four testing methods can be used to measure the macrotexture of the pavement surface. They include the TNZ T/3 sand circle, the volume outflow meter, the Circular Track Meter (CTM) and the RoboTex.
The first test used in this study to measure macrotexture, the New Zealand sand circle, was completed on a monthly basis using the NZTA TNZ T/3 standard. The NZTA sand circle test should be used due to the unreliability of the ASTM sand patch as demonstrated in an early California DOT study (Doty 1974) and a recent study conducted for the Texas DOT (Gransberg 2007).
Figure 2 shows the test being
conducted in the field. The TNZ T/3 testing procedure feeds the TNZ P/17 performance specification which can then be used as a metric to judge the success or failure of the surface treatments in their first 12 months based on a field-proven standard (TNZ 2002).
Figure 2 Sand Circle Tests and Outflow Meter/Sand Circle Testing in Progress The test is a volumetric test, preformed by placing a known volume of sand, in this case 45 ml, which is then spread by revolving a straight edge in a circle until the sand is level with the tops of the surface aggregate and can no longer be moved around (TNZ 1981). Once the known volume has been spread in a circle on the surface of the roadway and can no longer be moved, two measurements are taken to determine the average diameter of the circle, this value is then inserted into Equation 1. 57,300 Mean Texture Depth (mm) =
Diameter (mm)2
(1)
The surface texture is inversely proportional to the diameter of the circle produced on the surface. This testing protocol is relatively simple but has limitations: it is susceptible to operator inconsistency, environmental issues with rain and wind, and roadway imperfections, such as abnormal aggregate heights on the surface of the road. A wind shield is used to shelter the circle from winds and prevent loss of test sand during the test.
The second test used in this study for determining macrotexture is an outflow meter as specified by ASTM E2380/E2380M - Standard Test Method for Measuring Pavement Texture Using an Outflow Meter. It is also shown in Figure 2. This method does not measure the texture depth directly; it quantifies the connectivity of the texture as it relates to the drainage capability of the pavement through its surface and subsurface voids. The technique is designed to measure the ability of the pavement surface to relieve pressure from the face of vehicular tires and thus is an indication of hydroplaning potential under wet conditions. A faster outflow time indicates a thinner film of water may exist between the tire and the pavement, thus more macrotexture is exposed to indent the face of the tire and more surface friction is available to the tire. The outflow meter test is based on a known volume of water, under a standard head of pressure, which is then allowed to disperse through the gaps, or macrotexture, between a circular rubber ring and the road surface. The time it takes to pass the known volume of water, i.e. the outflow time, is measured. Mean texture depth was calculated using the following equation. 3.114 Mean Texture Depth (mm) =
Outflow Time (sec)
+
0.636
(2)
Determining macrotexture on pavement correctly and quickly is important for safety and economy in pavement preservation testing. There are functional limitations in each method’s ability to accurately measure pavement macrotexture (Aktas et al. 2011). The outflow meter provides users with results measured in seconds. It is portable, practical on wet surfaces, inexpensive, and fast, but the measured outflow time can be inaccurate for pavement preservation treatments with high macrotextures. The opposite is true for the sand circle method which should be avoided on surfaces with low macrotexture. This results in the following recommendations for appropriate use of each test method: •
If macrotexture < 0.79mm (0.03 in.), use the outflow meter only.
•
If macrotexture > 0.79mm (0.03 in.) and < 1.26mm (0.05 in.), either test is appropriate
•
If macrotexture > 1.26mm (0.05 in.), use the sand circle test only.
The third test that can be used to determine macrotexture is the circular track meter (CTM): ASTM E 2157 - Standard Test Method for Measuring Pavement Macro Texture Properties Using the Circular Track Meter. It is shown in Figure 3 and provided by The Transtec Group. The CTM consists of a charge coupled device (CCD) laser– displacement sensor that is mounted on a rotating arm. The displacement sensor follows a 284-mm circular track to provide the measurement based on Equation 3, where MPD is mean profile depth, in mm.
Mean Texture Depth (mm) = 0.947 MPD + 0.069
Figure 3 Circular Track Meter
(3)
The fourth test that can be used to determine macrotexture is the RoboTex (meets ISO 13473-3 Class DE device requirements and calibration per ASTM E 797), shown in Figure 4, and provided by The Transtec Group. The RoboTex procedure involves measuring MPD in transverse and longitudinal directions. The output is graphs MPD in increments of distance.
Figure 4 RoboTex (top) with Example Output (bottom).
There are limitations to these testing methods. Skid testing and the outflow meter have temperature limitation due to the water required for testing. This makes testing very difficult and sometimes impossible when the temperature at the test site locations falls below freezing. All of the macrotexture tests are slow and cumbersome and require forward planning, good time management and adequate traffic control, and are not suited for routine monitoring of the road surface texture over a large road network (Austroads 2005). Other testing methods have recently become available to overcome these obstacles and maintain network wide data for both types of safety criteria; for microtexture, the ASTM E274 skid tester with ribbed tires serves as a texture index, and for macrotexture the ASTM E1845 can be used along with a high speed laser to determine mean profile depth at highway speed (ASTM 2009).
The micro and macrotexture data over time from each of the measurement methods can be compiled and reduced. The output is similar to the output shown in Figure 5. It shows how the micro and macrotexture of a given road is gradually lost during the pavement preservation treatment’s life cycle. This curve portrays the rate of deterioration of a given treatment. When coupled with an appropriate failure criterion, this allows the engineer to estimate the treatment’s actual service life. From the analysis, an actual deterioration rate can be used to furnish a realistic period of analysis to be used in life cycle cost analysis (LCCA).
For instance, NZTA uses 0.9mm of macrotexture as the failure criterion for macrotexture. The test section data shown in Figure 5 has not reached that point after 9 months of service. With high speed data collection and the above-described methodology for texture deterioration modeling, a pavement management engineer now has the tools to better incorporate surface texture safety criteria into the pavement management system (PMS). Regressing the data would yield an equation that would allow the extrapolation of the macrotexture curve and the ability to find when it crossed the 0.9mm line. This would provide an approximate service life on which to base the life cycle cost calculation. Thus, this method furnishes a means to compare a marginally higher cost pavement preservation alternative to a lower cost one and then select the treatment that has the lowest life cycle cost not merely the one that has the lowest initial cost, as ODOT must do today.
Test Section 3 - Open Graded Friction Course 6.0
40
35 5.0
4.0 25
3.0
20
15
Skid Number (mu)
Mean Texture Depth (mm)
30
2.0 10 1.0 5
0.0 Sep-08
Oct-08
Dec-08
Feb-09
Mar-09
May-09
0 Jul-09
Month Sand Patch
Hydrotimer Flowmeter
Skid Number
Figure 5 Change in Wheelpath Surface Characteristics Over Time
DETERIORATION MODELING Simple deterioration models can be created for a wide variety of pavement preservation treatments to assist the maintenance engineer with selecting the most appropriate treatment given their situational needs. Linear regression can be applied to the field microtexture and macrotexture data to approximate the deterioration rate and extrapolate the remaining service life of each treatment as illustrated in the following figures. These are then compared to failure criteria. Service life is determined by identifying the time it takes each treatment to deteriorate to each failure criterion. This methodology for estimating pavement preservation treatment service from local field data reduces the uncertainty of selecting an empirical service life, and literally enhances the analog because the data is taken in the same environment that the selected treatment must function (Bilal et al. 2009; Reigle and Zaniewski 2002).
The failure criterion for macrotexture is 0.9mm, which is consistent with TNZ P/12 performance specification (TNZ 2002). The failure point considered for microtexture is a skid number less than 25. It is important to note that models based on limited data may not produce adequate results. As more deterioration data is gathered over time, the models will be based upon more information that will yield a clearer idea of the nature of each treatment-type’s deterioration behavior.
Figure 6 illustrates the
deterioration based on macrotexture for an example alternative: a 5/8” chip seal.
Chip Seal (5/8") Mean Texture Depth (mm)
4.0
y = -0.636ln(x) + 3.8985 R² = 0.8512
3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0
5
10
15
20
25
30
35
Months Sand Circle
Log. (Sand Circle)
Figure 6 Chip Seal Deterioration Model Based on Macrotexture Data
This lays the ground work to begin to associate safety data, the level of both types of texture, with the currently used surface characteristics and traffic data to develop a more holistic view of pavement management. An example of how this process could be implemented can be seen in Figure 7.
In this scenario, the deterioration models,
developed for a chip seal using 5/8″ aggregate, are used to determine trigger points.
This alerts pavement engineers to start the project development process in order to implement a pavement preservation project before the pavement fails.
Figure 7 Deterioration Models for Chip Seal (5/8″) A hypothetical example with 12 months duration is used to illustrate the proposed process for integrating pavement texture into the pavement management process. Failure criteria must be established to allow the computation of a rational trigger point. In this example, the macrotexture failure criterion of 0.9mm used in New Zealand is
used (TNZ 2002). The pavement engineer then extrapolates the deterioration curve to the failure value, then back calculates a macrotexture trigger value of 0.969mm. This is too precise to measure with the available test equipment. So a value of 1.0mm is chosen, which generates an alert 16 months prior to failure, giving the engineer a period greater than required project development and authorization period to initiate action to preserve this pavement. The same approach can be used with microtexture.
The example uses skid number criteria found in the literature to develop a skid number failure criterion. In this example, the trigger point value is 30 and the failure criterion is 25. One can see that a trigger value for microtexture is reached early on in the life of the section, at month 29, yet this gives the pavement management engineer almost 9 years to act before ultimate failure. The deterioration models indicate that this particular treatment will likely fail due to macrotexture loss before it reaches the microtexture failure point.
However, a chip seal in a different location that uses a less durable
aggregate (susceptible to polishing) would have a deterioration model that predicted failure much earlier. This shows the need to have localized deterioration data. If the failure occurred before the macrotexture trigger value, the pavement manager would start the project development process upon reaching the microtexture trigger point. By using safety trigger values in this manner, the pavement management engineer should now be able to better prioritize which roads need to preserved first, and coupled with standard pavement management systems, what treatments need to be placed, further ensuring that the “right treatment is placed on the right road at the right time” (Galehouse et al. 2003).
COST ANALYSIS Economic and engineering technical data gathered from pavement preservation treatments can be quantified and correlated to produce meaningful, standardized economic information that furnishes pavement managers measurable failure criteria to estimate extended service lives of Oklahoma pavements. Two types of cost analysis are
presented in this guide for use when making pavement treatment decisions: life cycle cost analysis (LCCA) and skid number cost index (SNCI) analysis.
Life Cycle Cost Analysis The LCCA model is based on the engineering economic theory of equivalent uniform annual cost (EUAC) and allows pavement managers to evaluate the cost effectiveness of competing pavement preservation treatment alternatives on either a deterministic or stochastic life cycle cost basis. The use of microtexture and macrotexture deterioration models provide local pavement condition data that correlate with service life and pavement life extension input values.
This allows the extent of variability in these
parameters to be exposed via sensitivity analysis and thereby enhance the credibility of results. Treatment-specific input values, such as expected service life and pavement life extension, allow the pavement manager to intuitively analyze the life cycle cost analysis results. EUAC allows the engineer to better relate treatment life cycle cost output to annual maintenance budgets and specifically determine the most cost effective pavement preservation treatment for a given project.
LCCA can become quite complex, so an analyst should be judicious about the level of detail included (FHWA 1998). An analyst can simplify the analysis by including only differential costs, i.e. omitting those that cancel out, as well as disregarding those costs that contribute minimal to no impact on the final results, keeping in mind that discounting might render costs so (FHWA 1998). The FHWA offers “LCCA Principles of Good Practice” in its Life Cycle Cost Analysis in Pavement Design, Interim Technical Bulletin released in 1998, such as in selecting a discount rate. “Good Practice” is that constant dollars and real discount rate be used for the purposes of discounting future costs (i.e. omit inflation and effects). FHWA recommends a rate between 3-5% be used in analyses, which is consistent with the OMB Circular A-94. Other “LCCA Principles of Good Practice” are integrated with the LCCA procedures/methodology.
The following are LCCA procedures, as excerpted from the FHWA Life Cycle Cost Analysis Primer (FHWA 2002) and the Interim Technical Bulletin (FHWA 1998) and adapted to the EUAC methodology established by the research (Pittenger et al. 2011): •
Step1: Establish design alternatives
•
Step 2: Determine [performance period and] activity timing
•
Step 3: Estimate costs [agency and user]
•
Step 4: Compute life cycle costs
•
Step 5: Analyze results
•
Step 6: Reevaluate design strategies.
Step 1: Design Alternatives and Analysis Period The first step in the procedure involves establishing strategies, i.e. associated rehabilitation and maintenance activities associated with each alternative expected over the analysis period (FHWA 1998). To aid in the selection of design alternatives, information has been included in the appendices. Appendix A lists the pavement preservation treatments investigated in the research project. Appendix B provides a table that shows various pavement preservation treatments and their capabilities and functions.
Step 2: Performance Period and Activity Timing The second step involves determining the performance period (i.e. cash flow diagram) for each alternative, which is the period that covers one life cycle of that alternative. The performance period determination has a significant effect on the LCCA output, and should be considered with care (FHWA 1998). Deterioration models establish the performance period for a given treatment in a general location. If deterioration model data is not available, Appendix C provides common performance period (service life) values for listed treatments. Activity timing includes the determination of maintenance and other activity frequency associated with a specific alternative strategy, as established in step 1 (FHWA 1998).
Continuous Mode EUAC accommodates the continuous, short-term nature of PPT application because the next expected rehabilitation/reconstruction of the pavement is commonly unknown, i.e. is not on the current work plan. The pavement manager must plan to continuously maintain, preserve or “do nothing” to the pavement in the undefined interim. Because encroachment is not expected in the continuous mode, material or mathematical adjustments to costs or service life lengths are not required.
Therefore, each
treatment’s service life input value will be equivalent to its anticipated service life (n), which is the value used in EUAC calculations in this model to determine life cycle cost. For example, if the service life of chip seal is estimated to be 5 years, then “5” is entered into the calculation for chip seal service life. Terminal Mode In the terminal scenario, the pavement manager generally chooses the “do nothing” option. In other words, the pavement manager usually defers maintenance because the pavement is scheduled to be rehabilitated or reconstructed according to the work plan. Therefore, the decision essentially is to ignore pavement preservation on a given pavement knowing that it will be “fixed” in the near future. This permits the reprogramming of those funds to preserving other pavements in the network. However, if the pavement must be treated in the interim for safety reasons, one must avoid the common “mistake” associated with employing the EUAC method by considering the encroachment upon (i.e. materially alter) treatment service lives. For example, if the service life of chip seal is estimated to be 5 years, and the reconstruction is expected in 3 years, then “3” is entered into the calculation for chip seal service life. This keeps the model in compliance with engineering economic principles (White et al. 2010).
Step 3: Agency and User Costs Estimation The third step involves determining or estimating agency and user costs for each of the competing alternatives.
Agency and user costs are determined for each of the
competing alternatives and future costs are “discounted” to determine the EUAC. Agency costs are those costs directly incurred by the agency, such as costs for project supervision and administration, materials, labor and traffic control for the initial
installation. Appendix C provides a reference document that includes average cost for treatments. It also includes any associated rehabilitation and maintenance costs required over the life cycle of the alternative. These costs are generally based on current and/or historical costs.
According to the FHWA, salvage value is the value associated with each alternative determined at the point of analysis termination and involves any residual value (value attributed to the reclaimed materials) or any serviceable life (value attributed to alternative “life” that exists after analysis termination).
It should be attributed to
alternatives appropriately for the purposes of analysis (FHWA 1998).
Sunk costs, which are costs occurring pre-analysis, should not be included in the analysis unless they specifically apply to the alternatives that are to be compared (FHWA 1998).
User costs relate to costs incurred by the traveling public in both work-zone and nonwork-zone phases for a given extent of road for which alternatives are being compared (FHWA 1998). Generally, the user costs incurred during non-work-zone phases are disregarded in LCCA due to a lower likelihood of difference among alternatives (FHWA 2002).
Differing [work zone] user costs among alternatives are pertinent to the
analyses, and generally include “[time] delay, vehicle operating, and crash costs incurred by the users of a facility” (FHWA 1998). It is suggested that “different vehicle classes have different operating characteristics and associated operating costs, and as a result, user costs should be analyzed for at least three broad vehicle classes: Passenger Vehicles, Single-Unit Trucks, and Combination Trucks” (FHWA 1998). User costs are generally translated into monetary terms (for the purposes of analysis) and can be ascertained from various sources, and those costs escalated with the use of the transportation component of the Consumer Price Index (CPI) (FHWA 1998). Delay costs are calculated by multiplying the unit of “wait” time attributed to each alternative’s work-zone timings by the monetary unit (FHWA 1998). Vehicle operating costs (VOC) are calculated by multiplying the vehicle-related cost factors attributable to each
alternative’s work-zone timings by the monetary unit (FHWA 1998). Crash costs are calculated by multiplying the number of specific types of crashes by their respective monetary unit (FHWA 1998). User costs as a result of detours are typically assigned a cents-per-mile rate, such as that used by the Internal Revenue Service for mileage allowance (FHWA 1998). Step 4: Compute Life-Cycle Costs Life cycle costs are determined by conducting LCCA on the EUAC basis (Figure 8).
Figure 8 EUAC Chip Seal Calculation and Service Life Diagram in EUAC. The calculation of EUAC involves the summation of all present and future costs (∑P), as described in Step 3 and illustrated in Figure 8. The discount rate (i) and the service life value (n) are inserted into the equation. For example, if the anticipated service life value for chip seal is 5 years, then an EUAC of $3,482 is calculated based on an initial cost of $12,792 and a discounted future maintenance cost of $3,049 (discount rate: 4%). EUAC is calculated for each alternative. The alternative with the lowest EUAC is considered the most cost effective alternative.
Step 5: Analyze Results LCCA has two possible computational approaches: deterministic (for deterministic LCCA) and probabilistic (for stochastic LCCA) (FHWA 1998), which are discussed in the next section. To summarize, the deterministic approach involves using discrete input values and a single output value (FHWA 2002).
A sensitivity analysis should be
conducted so that the analyst may determine the level of variability of a given input value relative to the output (FHWA 1998). For example, an analyst chooses a 4% discount rate to do the LCCA, which results in output (a preferred alternative). The sensitivity analysis will allow the analyst to conduct a “What if” scenario to determine if choosing a 5% discount rate would result in different output (different preferred alternative). This exercise must be repeated individually for other input variables to discover any effect on overall LCCA results. The sensitivity analysis is limited in application with regard to being unable to analyze simultaneous variability (FHWA 2002).
The probabilistic approach associated with stochastic LCCA involves analyzing input value probability based on the full range of “What if” scenarios allowed by sensitivity analysis by providing a “distribution of PV results” (FHWA 2002).
It is generally
accompanied with a risk analysis, which unlike sensitivity analysis, does allow the analyst to determine the level of certainty with regard to simultaneous variability in all input parameters (FHWA 1998).
Whether LCCA should be conducted deterministically or probabilistically (Step 5 of the FHWA LCCA Good Practices), depends on the level of input uncertainty (FHWA 1998). Much debate continues about the validity of LCCA output when inherent uncertainty is not addressed, hindering the agencies’ ability to justify decisions (FHWA, 1998). Transportation decision making is subject to scrutiny because of its public nature. Transportation agencies are charged with stewardship and therefore must provide justification for decision making and its inherent uncertainties. Major uncertainties in transportation projects are generally related to cost and performance (service life) and contribute to the complexity and difficulty associated with transportation decision
making. Uncertainty about future events, when quantified, is known as risk and can be incorporated into the decision-making process. A deterministic-type LCCA is less complex than a stochastic type and can be adequate, and therefore appropriate, when uncertainty is not expected to have a material effect on the outcome of the economic analysis (FHWA 1998). However, if uncertainty could materially alter the outcome, the deterministic approach is not recommended because of its inability to effectively analyze simultaneous variability (FHWA 2002). The probabilistic approach is used to address these issues.
Stochastic LCCA specifically addresses and quantifies the
uncertainty associated with a transportation project decision and contributes to decision validation (FHWA 1998).
Step 6: Reevaluate Design Strategies LCCA results should be coupled with other decision-support factors such as “risk, available budgets, and political and environmental concerns” (FHWA 2002). A decisionanalysis framework can offer insight as to relative differences between alternatives in the areas of “uncertainty, objectives and trade-offs”, but should not be expected to offer the answer (FHWA 1998). The analyst would still need to rely on judgment in the final decision-making phase (FHWA 1998).
Considering cost-effectiveness without also
considering treatment effectiveness (and vice versa), or the economic efficiency of a treatment, may not provide the whole picture either and may result in not selecting the “best” alternative (Bilal et al. 2009).
Figure 9 shows the model logic that incorporates these six steps.
-Professional Judgment -PCI -AADT -Microtexture -Macrotexture
Establish PPT Alternatives Is the year of the next rehabilitation or reconstruction known? No
Yes
Determine Service Life
Enter year of next
Determine Activity Timing
Determine Agency And User Costs
Calculate EUAC
Analyze Results
Reevaluate Results Figure 9 PPT EUAC LCCA Model Logic.
R/R
LCCA Example The following example incorporates the six LCCA steps and illustrates the stochastic approach,
the
deterministic
approach
and
the
performance-based
approach
(deterioration model-based approach). Several additional steps should be addressed in the stochastic model construction, including whether to treat the input values deterministically or probabilistically, fitting the initial input values with the appropriate distribution and developing the algorithm for risk analysis. The deterministic LCCA can be conducted on a calculator; the stochastic approach requires statistical software. Step 1: Input Value Consideration The first step in probabilistic analysis is to determine which input values to treat deterministically and which to treat probabilistically based on the uncertainty expected of each parameter (FHWA 1998).
For this study, the initial costs, service life and
discount rate will be treated probabilistically and the rest will be treated deterministically. The following equation (4) will constitute the sum of present values: ∑P = PPTP + DV + FCD-P + UCD-P
(4)
where: PPTP = Expected initial cost of PPT (probabilistic) DV = Deterministic value of other agency costs (such as equipment, traffic control, striping) FCD-P = Expected value of future (maintenance) costs (probabilistic discount rate and service life components, deterministic cost) UCD-P = Expected value of user costs (probabilistic discount rate and service life components, deterministic cost). The expected initial costs of treatments (PPTp) were probabilistically determined from ODOT and Oklahoma Department of Central Services (DCS) bid tabulations for the calendar years 2010-2011 and are consistent with ODOT survey and literature and are listed in Table 3. Prices from projects with similar quantities and in similar geographic locations were used so as to not inadvertently skew output (Tighe 2001). Activity timing for future maintenance included a three-year frequency for crack seal and 2%-of-totalarea patching.
Table 3 Treatment Agency Costs Used in EUAC LCCA. Agency Costs ($ per Square Yard) PPT Chip Seal
Probabilistic Value Mean (Standard Dev.) 1.82 (0.07)
Deterministic Value 1.77
OGFC 1” HMA Mill & Inlay
3.75 4.00
3.88 (0.39) 3.78 (0.49)
Step 2: Goodness-of-Fit Tests The second step in the stochastic process is to conduct goodness-of-fit tests for the stochastic parameters determined in the first step.
A triangular distribution is a
continuous distribution that contains user-defined values for the minimum value, the maximum value and the most likely value for an input variable. It is commonly used for variables that have limited sample data but can be reasonably estimated (FHWA 1998), which is the case for service life. In the continuous mode, 4-5-6 years were used for the chip seal service life distribution parameters. 8-10-12 years were used in the OGFC and HMA distributions, consistent with literature and ODOT survey. This common pavement management scenario is illustrated in Figure 10.
Figure 10 Stochastic EUAC Input Screen, Continuous Mode In the continuous mode, the “Years until next Rehabilitation/Reconstruction” is unknown, so that field is left blank. Therefore, the service life value is equal to the
anticipated service life value. The service life distribution is also noted in the entry screen as a triangular distribution and the low, most likely and high values are noted in parentheses.
The terminal mode is used when the next rehabilitation or reconstruction year is known. If the next rehabilitation is expected in 5 years, then that value is entered into the “Years until next Rehabilitation/Reconstruction” field, as shown in Figure 11.
Figure 11 Stochastic EUAC Input Screen, Terminal Mode Based upon the rehabilitation year (RY), the terminal mode automatically truncates the service life values of the alternatives to provide the anticipated service life (ASL). Both the HMA and OGFC service lives are truncated because the low value of their triangular distributions is 8, which is greater than RY, so their EUACs will be figured deterministically with an ASL of 5 years. The chip seal service life will also be truncated at 5 years; however, the 5-year value falls within its distribution, so it will be treated probabilistically. For the triangular distributions, if: RY ≤ SLlow, then ASL = RY (deterministic); RY > SLlow and RY ≤ SLhigh, then ASL = RY as triangular max value (probabilistic); RY ≥ SLhigh, then ASL = SL (probabilistic).
The chip seal’s terminal service life distribution will have parameters of 4-5-5, instead of its continuous distribution of 4-5-6, as shown in Figure 12.
Figure 12 Chip Seal Service Life Probability Distributions, Continuous (left) and Terminal (right) Modes The future maintenance costs and user costs are deterministic values in the analyses, but are discounted probabilistically so that the discounting period and discount rate are consistent within each iteration. This example used the previous 30 years of discount rate data from the Federal Reserve (2011), best fitted to the Weibull distribution (μ: 4.92, sd: 2.76). Next, initial PPT costs were considered. The best fit for the OGFC bid tabulation data is the normal theoretical distribution, which has a chi-squared (χ2) value of 5.94, followed by the extreme value distribution (χ2 =11.71) (Table 4).
Table 4 Goodness-of-Fit for PPTs Bid Tabulation Data Fit ($/Square Yard) # of Data Distribution Chi-Square PPT Points Type Value Mean Normal 7.60 1.817 Chip Seal 25 Extreme Value 5.20 1.812 Normal 5.94 3.88 OGFC 36 Extreme Value 11.71 3.88 Normal 9.51 3.78 1” HMA Mill & Inlay 222 Gamma 15.7 3.78
Standard Deviation 0.0704 0.0799 0.391 0.391 0.49 0.49
The extreme value distribution was also the best fit for chip seal, while the gamma distribution best described the HMA data.
In all cases, the χ2 values are minimal,
meaning the theoretical distributions closely approximate the data. In addition, the best fit distributions returned the same (similar) means and standard deviations as the normal distributions, consistent with the central limit theorem that states data normalizes with increasing number of data points (Montgomery 2009).
Therefore, the EUACs
yielded from the normal distribution of bid tabulations is sufficient and will be used in this chapter. Step 3: Risk Analysis The third step of stochastic analysis involves risk analysis. The algorithm developed for Monte Carlo simulation is based upon the EUAC equation and is illustrated in Figure 13. It incorporates the costs, service life and discounting methods discussed in the previous two steps to produce the sampling distribution that provides the expected EUAC for each alternative being considered.
Figure 13 Stochastic EUAC LCCA Approach. (Adapted from FHWA, 1998a) The model in continuous mode returns EUAC distributions for the three alternatives, as shown in Figure 14.
The curves show that the chip seal and OGFC have similar
variability, which is less than the HMA. Additionally, the chip seal has a lower life cycle cost, as demonstrated by its curve being to the left of the other two alternatives.
Figure 14 PPTs Stochastic EUAC Probability Distributions
Numeric values associated with these distributions are shown in Table 5. Based on these values, it would appear that chip seal would be the most cost effective option in this specific case. Table 5 Stochastic EUAC Results, Continuous Mode. EUAC ($/lane-mile), Stochastic Method (Continuous Mode) Mean
Chip Seal
OGFC
HMA
3,738
4,546
4,410
414
426
758
3,095
3,856
3,254
4,452
5,275
5,726
Standard Deviation 5th Percentile th
95 Percentile
Stochastic analysis offers a wealth of decision-assisting information, such as percentile statistics.
Percentiles can provide the pavement manager relative probabilities
associated with each pavement treatment alternative being considered. For example, the mean EUAC of HMA ($4,410) falls around the 95th percentile of the chip seal EUAC range, as shown in Figure 15.
This means that the central tendency, or bulk, of
possible HMA EUACs would be higher than the expected chip seal EUAC 95% of the time.
Figure 15 Summary Statistics for Hot Mix Asphalt EUAC Distribution
However, the HMA EUAC range does reach the lower end of the chip seal range. The summary statistics reveal that the chip seal mean of $3,738 lies around the 20th percentile of the HMA range, so HMA should be expected to be less than the mean value of chip seal 20% of the time.
The deterministic EUAC values and subsequent rank for HMA, OGFC and chip seal were $4,696 (3), $4,460 (2) and $3,651 (1), respectively. If the pavement manager was only considering the HMA and OGFC options, then the stochastic and deterministic outcomes would be different because the stochastic results show HMA having the lower LCC, based on actual bid tabulations.
When considering the deterministic and
stochastic HMA values, the stochastic analysis reveals that the EUAC should not be expected to exceed $4,696 70% of the time, while the chip seal is at 45% probability.
The stochastic approach also offers a comprehensive sensitivity analysis based on regression. The sensitivity analysis reveals that the chip seal is most sensitive to the service life parameter, the OGFC is most sensitive to discount rate and the HMA is most sensitive to the agency cost parameter, as shown in Table 6. The pavement manager may take the following into consideration: if chip seal was not expected to realize its’ full service life due to traffic conditions, the pavement manager may select the HMA treatment, which is not as sensitive to the parameter. However, if commodity volatility (agency costs) were expected to increase, the chip seal may be the better option because it is less sensitive to the cost parameter than HMA. Although both treatments contain the same volatile commodity (asphalt binder), the sensitivities reflect the different methods of procurement. The HMA agency costs are based upon monthly ODOT bid prices, whereas the chip seal is based upon DCS bid prices that are “lockedin” for some time period. The stochastic analysis offers this type of critical information that the deterministic analysis cannot.
2.76
2.76
2.76
Chip Seal
OGFC
1” HMA Mill & Inlay
PPT
Standard Deviation (%)
0.771
0.808
0.688
Regression Coefficient
Discount Rate
Sensitivity Analysis
584.55
601.11
285.21
Net Change in EUAC ($)
0.816
0.816
0.410
Standard Deviation (Years)
-0.302
-0.313
-0.657
Regression Coefficient
Service Life
-228.77
-233.14
-272.06
Net Change in EUAC ($)
8.37/ton
9.30/ton
0.07/SY
Standard Deviation ($)
Cost
0.558
0.484
0.277
Regression Coefficient
422.98
360.30
114.61
Net Change in EUAC ($)
Table 6 Stochastic EUAC Sensitivity Analysis.
The model in terminal mode returns the EUAC distributions as shown in Table 7. The OGFC and HMA values are substantially higher than the values returned in the continuous mode (Table 5), and the chip seal and HMA EUAC distribution curves do not overlap, except for the extreme (tail) values that have little associated probability. Table 7 Stochastic EUAC Results, Terminal Mode EUAC ($/lane-mile), Stochastic Method (Terminal Mode) Chip Seal
OGFC
HMA
3,962
7,191
7,015
362
827
942
5th Percentile
3,399
5,898
5,540
95th Percentile
4,580
8,622
8,627
Mean Standard Deviation
In the continuous mode example, the decision would require pavement manager judgment, facilitated by percentiles and sensitivity analysis, to justify a decision due to the overlapping EUAC distributions. In the terminal case, the chip seal appears to be the best option.
The stochastic results reveal that the impact of terminal mode on the chip seal probability associated with the $3,651 (deterministic EUAC) dropped from 45% to 20%, meaning the chip seal is 25% more likely to exceed that value. There is a greater than 95% chance that HMA will exceed its deterministic value. This demonstrates the impact of the rehabilitation year on service life and justifies the inclusion of the terminal mode in the EUAC model for upholding engineering economic principles (White et al. 2010).
Stochastic EUAC for PPTs adheres to engineering economic principles with the inclusion of the continuous and terminal modes.
Software can facilitate the risk
analysis, making it a practical tool at the implementation level. The stochastic approach can reveal sensitivities that the deterministic model conceals, and is therefore recommended when uncertainty is expected to impact LCCA results to allow pavement managers to understand and mitigate budget risk.
LCCA results are sensitive to LCCA method selection on all levels as demonstrated through the use of deterministic-based, stochastic-based and performance-based methods in this section. It also demonstrates the value of including performance-based service life input values. Deterministic LCCA vs. Stochastic LCCA The deterministic LCCA was conducted for three scenarios that a pavement manager might consider: the worst case scenario, the most likely scenario and best case scenario. These scenarios are based on the two most sensitive LCCA parameters: service life and discount rate. The EUAC for each treatment is in Table 8. Table 8 Deterministic LCCA Results. EUAC ($/lane-mile), Deterministic Method with Sensitivity Analysis
PPT Chip Seal
3% EUAC (Service Life) $3,019 (6 YR)
4% EUAC (Service Life) $3,651 (5 YR)
5% EUAC (Service Life) $4,557 (4 YR)
OGFC
$3,771 (12 YR)
$4,460 (10 YR)
$5,117 (8 YR)
1” HMA Mill & Inlay
$3,963 (12 YR)
$4,696 (10 YR)
$5,413 (8 YR)
The deterministic and stochastic methods provide different results. The most likely scenario value for chip seal has only 43% probability, whereas the HMA has a 70% probability, as shown in Table 9. Table 9 Deterministic and Stochastic EUAC Comparison. EUAC ($/lane-mile), Deterministic EUAC with Corresponding Probability
PPT
Deterministic Best Case EUAC (Probability)
Deterministic Most Likely Case EUAC (Probability)
Deterministic Worst Case EUAC (Probability)
Chip Seal
$3,019 (12
>12 YR
10-12-12
>12 YR
N/A
10-12-12
1” HMA Mill & Inlay
The service life parameter was treated deterministically and probabilistically for each of the performance measures for comparative analysis, as noted in Figure 16. Probabilistic modeling of service life based upon microtexture and macrotexture measurements was accomplished with the triangular distribution. The third value (5 for chip seal, 10 for OGFC/HMA) is consistent with ODOT expected values.
Figure 16 Performance-Based Triangular Service Life Distributions. The Monte Carlo simulation was set to “fixed” mode so that it would run the same iterations run for the stochastic EUAC values in Table 5, so that differences can only be attributed to the change in service life treatment.
Table 11 Performance-Based Stochastic EUAC Results. EUAC ($/lane-mile), Stochastic-Method, Performance Based Deterministic Probabilistic μ (SD) μ (SD) μ (SD) (Macrotexture) (Microtexture) (Triangular) Chip Seal
$5,340 (282)
$2,712 (245)
$3,466 (774)
OGFC
$3,904 (688)
$3,904 (688)
$4,105 (710)
N/A
$3,839 (706)
$4,009 (727)
1” HMA Mill & Inlay
Performance-based service life values are listed in Table 11, along with corresponding EUACs.
The different EUACs demonstrate LCCA sensitivity to deterministic and
probabilistic treatment of service life input.
Stochastic LCCA: Empirical vs. Performance-Based Treatment of Probabilistic Service Life
The chip seal stochastic EUAC value based upon the empirical 4-5-6 service life distribution was $3,738 (std. dev: $414), as noted in Table 5. When compared to the EUAC using the performance-based 3-5-9.1 service life distribution ($3,466, sd: $774; Table 11), it is apparent that the LCCA is sensitive to the service life parameter, as noted by the different EUAC values. The performance-based EUAC exhibits greater variability due to the wider range in possible service life values.
Service Life Treatment Method Impact on LCCA Figure 17 shows distributions based upon the four service life assumptions for stochastic analyses listed in Tables 5 and 11: (1) 4-5-6 empirical service life, shown by the black distribution, (2) 3-5-9 performance-based service life, shown by the blue distribution, (3) 3-year performance-based (macrotexture) deterministic service life, shown by the red distribution and (4) 9.1-year per performance-based (microtexture) deterministic service life, shown by the green distribution.
Figure 17 Impact of Service Life Method on Chip Seal EUAC.
The deterministic performance methods put the distributions on far ends of the chip seal EUAC spectrum, while the triangular (stochastic) distributions cover the middle portion. Also, the stochastic treatments (black and blue) differ from each other. This illustrates the deterministic-treatment effect, the stochastic-treatment effect and the performancebased treatment effect on the service life input parameter, and subsequently the LCCA output.
LCCA Sensitivity to LCCA Selection Method Chip seal deterministic EUACs (Table 8) and performance-based EUACs (deterministic treatment, Table 11) are charted on the empirically-based stochastic EUACs (Table 5) cumulative curve for chip seal in Figure 18.
Figure 18 Deterministic and Performance-Based Chip Seal EUACs on Stochastic Empirical-Service-Life Cumulative Curve. Although empirical and performance-based stochastic LCCA both result in chip seal being the preferred alternative (Table 10.5), the performance-based mean EUAC values
do not fall within the 90% probability range of the empirical LCCA. This shows that although the ranking is the same and it may be the most cost effective choice, the output is different and empirical LCCA may miss the budgetary mark (Bilal et al. 2010, Reigle and Zaniewski, 2002).
LCCA results in rankings that order alternatives on the basis of cost effectiveness. This study has shown that LCCA results are sensitive to LCCA method, as shown in Table 12 by the different rankings based only on LCCA method selection. Table 12 LCCA Sensitivity to LCCA Selection. EUAC Rankings, Various LCCA Methods Performance Based (TABLE 11) Stochastic, Deterministic Stochastic Deterministic Empirical (TABLE 8) (TABLE 5) MACRO MICRO Chip Seal 1 1 2 1 1 OGFC 2 3 1 3 3 1” HMA Mill & Inlay 3 2 N/A 2 2 (1 – Lowest EUAC; 3 – Highest EUAC)
The example reveals that chip seal would be the preferred alternative in all of the LCCA methods except for the deterministic performance-based (macrotexture) LCCA. If only the OGFC and HMA were being considered, it appears that the HMA would be the preferred alternative in all but the deterministic LCCA case. The assumed cost for OGFC is less than HMA, shown by its better deterministic analysis ranking. However, in the rest of the analyses which use actual bid tabulations, the HMA is better positioned, showing the superiority of using stochastic methods. The different rankings for alternatives based upon deterministic-based, stochastic-based and performancebased methods reveal the sensitivity of LCCA to LCCA method selection and demonstrate the need for consideration.
Lastly, the simulation output should not be blindly accepted. The pavement manager should check the output to see if it is reasonable considering the problem at hand. The
results provide much statistical information about competing alternatives and should be thoroughly analyzed (FHWA 1998).
This example demonstrated that LCCA results are sensitive to LCCA methods on all levels, as demonstrated through the use of deterministic-based, stochastic-based and performance-based methods. On the global level, it showed that the deterministic and stochastic LCCA produce different results. On the next level, it showed that even within the stochastic LCCA analysis, deterministic and stochastic treatment of service life produces different results. On the local level, it showed that stochastic treatment of service life based upon empirical and performance-based treatment produced different results. Therefore, the pavement manager should consider the LCCA method selection process when evaluating PPTs so as not to inadvertently skew output results. Stochastic EUAC provides another brick in the justification for pavement preservation.
SKID NUMBER COST INDEX ANALYSIS The other type of analysis presented in this guide is the skid number cost index analysis. This technique allows the analyst to measure the “bang for the buck.” In this case, the following equation can be used to calculate the Skid Number Cost Index for each treatment alternative (Gransberg and Zaman 2005). TCi SNCIi =
where: SNCIi
Average SNi
(5)
= Skid Number Cost Index of Treatment “i”
Ave SNi
= Average Skid Number of Treatment “i”
TCi
= Total Cost per Lane-mile of Treatment “i”
The alternative with the lower cost index number is the more cost effective option. Based on skid change, macrotexture and cost information, the maintenance engineer can now determine if spending more per lane mile is justified by an increase in skid number. The idea is to change the decision criterion from “minimize cost” to “maximize
value” by having all the necessary decision-making information in one place. An example of the SNCI analysis is shown in Table 13. The pavement engineer would need to decide whether paying six times more for the mill and inlay is justified by the 40% increase in skid.
Table 13 Skid Number Cost Index Analysis of Treatment Alternatives Treatment
Unit Price (July 2009)
Total Cost per Lane-Mile
Average Skid Number
Skid Number Cost Index
1” Mill and Inlay
$8.52/SY
$59,981
52.6
1,141.28
Emulsion Chip Seal
$1.51/SY
$10,630
37.6
282.45
1 SY = 0.84 SM; 1 Lane-mile = 5,890 SM
As with other cost analyses, the SNCI does not provide the answer, only insight into a treatment’s relative cost on the basis of performance. Other issues, like location and traffic level, should be considered when making the final decision.
Appendix D contains SNCI analysis based on the research project test sections. Costs are based on those listed in Appendix C.
In summary, this guidebook has presented methods used to restore pavement skid resistance by retexturing the existing surface with either a surface treatment or a mechanical process. It has furnished pavement managers with the technical engineering data for each pavement preservation treatment coupled with a life cycle cost analysis of the costs and benefits associated with each treatment. This will allow ODOT pavement managers to have the required information to be able to make rational engineering design decisions based on both physical and financial data for a suite of potential pavement preservation tools
REFERENCES Abdul-Malak, M.-A.U., D.W. Fowler, and A.H. Meyer (1993). “Major Factors Explaining Performance Variability of Seal Coat Pavement Rehabilitation Overlays,” Transportation Research Record 1338, Transportation Research Board, National Research Council, Washington, D.C., 1993, pp. 140–149. Aktas, B., D.D. Gransberg, C. Riemer, and D. Pittenger, “Comparative Analysis of Macrotexture Measurement Tests for Pavement Preservation Treatments, Transportation Research Record, Journal of the Transportation Research Board, National Academies, Issue 2209, 2011, pp 34-40. (ASTM E274) American Society for Testing and Materials, International, Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale Tire, ASTM International, West Conshohocken, PA, 2009, http://www.astm.org, July 16, 2010. (ASTM E 2157) American Society for Testing and Materials, International, Standard Test Method for Measuring Pavement Macrotexture Properties Using the Circular Track Meter, ASTM E2380/E2380M-09, ASTM International, West Conshohocken, PA, 2009, http://www.astm.org, July 16, 2010. (ASTM E 2380) American Society for Testing and Materials, International, Standard Test Method for Measuring Pavement Texture Drainage Using an Outflow Meter, ASTM E2380/E2380M-09, ASTM International, West Conshohocken, PA, 2000, http://www.astm.org, July 16, 2010. (ASTM E1845) American Society for Testing and Materials, International (ASTM), Standard Practice for Calculating Pavement Macrotexture Mean Profile Depth, ASTM E1845, 2009, ASTM International, West Conshohocken, PA, 2009, DOI:10.1520/E1845-09, http://www.astm.org, July 16, 2010. Austroads, Guidelines for the Management of Road Surface Skid Resistance, Austroads Publication No, AP-G83/05, Sidney, Australia, 2005. Bilal, M., M. Irfan, and S. Labi, Comparing the Methods for Evaluating Pavement Interventions – A Discussion and Case Study, Transportation Research Board (TRB) Paper No. 09-2661, Washington, D.C., 2009. Doty, R.N., “A Study of the Sand Patch and Outflow Meter Methods of Pavement Surface Texture Measurement,” Proceedings, ASTM 1974 Annual Meeting, Washington, D.C., 35pp, June 27, 1974. Federal Highway Administration (FHWA), Life Cycle Cost Analysis Primer, Office of Asset Management Washington, D.C., 2002. Federal Highway Administration (FHWA), Life Cycle Cost Analysis in Pavement Design, Interim Technical Bulletin.Washington, D.C., 1998.
The Federal Reserve, Economic and Research Data: Data Releases, 2011. Retrieved on July 15, 2011 from: http://www.federalreserve.gov/releases/h15/data.htm. Galehouse, L., J.S. Moulthrop, and R.G. Hicks, “Principles for Pavement Preservation: Definitions, Benefits, Issues and Barriers,” TR News, Transportation Research Board, pp. 4-9, 2003. Gee, K.W., “Preservation and Rehabilitation,” Proceedings, AEMA-ARRA-ISSA Joint Meeting, Bonita Springs Florida, p. 8, 2007. Geiger, D.R., “Pavement Preservation Definitions,” Federal Highway Administration Memorandum, Washington, D.C., p.2, September 12, 2005. George, K.P., A.S. Rjagopal and L.K. Lim, “Models for Predicting Pavement Deterioration,” Transportation Research Record 1215, Journal of the Transportation Research Board, National Academies, pp.1-7., 1989. Gransberg, D.D., “Using a New Zealand Performance Specification to Evaluate US Chip Seal Performance,” Journal of Transportation Engineering, ASCE, Vol. 133 (12), pp 688-695, 2007. Gransberg, D.D., and D.M.B. James. Chip Seal Best Practices, National Cooperative Highway Research Program Synthesis 342, Transportation Research Board, National Academies, Washington, D.C, 2005. Gransberg, D.D. and M. Zaman, “Analysis of Emulsion and Hot Asphalt Cement Chip Seal Performance,” Journal of Transportation Engineering, ASCE, Vol.131 (3), pp. 229-238, 2005. Montgomery, Douglas C., 2009. Introduction to Statistical Quality Control, 6th Edition. John Wiley and Sons, Inc. Hoboken, New Jersey. Moulthrop, J., “Pavement Preservation: Protecting the Investment,” Presentation made at NEAUPG Annual Meeting, Wilkes-Barre, Pennsylvania, 2003. National Cooperative Highway Research Program, Evolution and Benefits of Preventative Maintenance Strategies, Synthesis of Highway Practice No. 153, National Cooperative Highway Research Program, Transportation Research Board, Washington, D.C., 1989. Patrick, J.E., P.D.Cenek, and M. Owen, Comparison of Methods to Measure Macrotexture, Proc. 1st Int’l Conf. World of Asphalt Pavements on CD, Australian Asphalt Pavement Association, Sydney, pp. 24-27, 2000. Pittenger, D., D.D. Gransberg, M. Zaman, and C. Riemer, “Life Cycle Cost-Based Pavement Preservation Treatment Design,” 2011 Transportation Research Board,
Journal of the Transportation Research Board, National Academies, Issue 2235, 2011, pp 28-35. Reigle, J. and J. Zaniewski, “Risk-Based Life-Cycle Cost Analysis for Project-Level Pavement Management”, Transportation Research Record 1816, Paper No. 022579, 2002. Riemer, C., “Quantifying the Costs and Benefits of Pavement Retexturing as a Pavement Preservation Tool”, Dissertation, University of Oklahoma, Norman, OK, 2011. Roque, R., D. Anderson, and M. Thompson (1991). “Effect of Material, Design, and Construction Variables on Seal-Coat Performance,” Transportation Research Record 1300, Transportation Research Board, National Research Council, Washington, D.C., pp. 108–115. Tighe, S. “Guidelines for Probabilistic Pavement Life Cycle Cost Analysis,” Transportation Research Record: Journal of the Transportation Research Board, No. 1769, TRB, National Research Council, pp. 28-38. Washington, D.C. 2001. Transit New Zealand (TNZ), Standard Test Procedure for Measurement of Texture by the Sand Circle Methods, TNZ T/3, Wellington, New Zealand, 1981. Transit New Zealand (TNZ), Notes for the Specification for Bituminous Reseals, TNZ P17, Wellington, New Zealand, 2002. VicRoads Bituminous Sprayed Surfacing Manual (Draft), VicRoads Bookshop, Victoria, Australia, p.233, 2003. White, J., K. Case, and D. Pratt, Principles of Engineering Economic Analysis, Fifth Edition, John Wiley & Sons, Inc., Hoboken, New Jersey, 2010.
APPENDIX A PAVEMENT PRESERVATION TREATMENT DESCRIPTIONS This section alphabetically lists and briefly describes pavement preservation treatments that are available to ODOT and are endorsed by the FHWA (eligible for federal funding) for asphalt and Portland cement concrete pavements. These treatments have been investigated in the research project associated with this guidebook (Riemer et al. 2011) and conclusions specific to those test sections are included to provide some insight. However, the user must understand that individual product application results may vary and will be determined by many factors, such as means and methods, environment, budget, political concerns, traffic and product attributes. Treatment-specific sources should be referenced for more information.
Calupave Seal “Calupave” is a low penetration fog seal (shown in Figure A.1). Calumet Specialty Products provides this product. Installation should comply with Calumet specifications. Research project note (Riemer et al. 2011): This product acts and looks like a standard fog seal. However, it appeared much darker and held its color much longer than the traditional fog seal.
Figure A.1 – Calupave Seal
Open Graded Friction Course (OGFC) with Fog Seal Figure A.2 shows an Open Graded Friction Course (OGFC) with a Fog Seal. OGFC exhibits a negative macrotexture which means that it is porous and allows water to flow
through the surface, thus increasing drainage. It can be installed by local contractors and in accordance with ODOT standard specifications. Research project note (Riemer et al. 2011): Researchers applied fog seal to determine if it would result in a longer service life than the untreated OGFC. At the time of this writing, both sections are performing similarly.
Figure A.2 – Open Graded Friction Course w/ Fog Seal
Figure A.3 – Open Graded Friction Course
Hot Mix Asphalt (HMA), 1” Mill and Inlay Figure A.4 shows a 1″ Mill and (HMA) Inlay. Construction requires the removal of 1″ of existing surface material and replacement of that material with new hot mix asphalt. It can be installed by local contractors and in accordance with ODOT standard specifications.
Research project note (Riemer et al. 2011): a standard ODOT mix design (S4 PG (6422 OK)) was installed. This treatment served as a good bench mark to compare all other treatments.
Figure A.4 – 1″ Mill and (HMA) Inlay Fog Seal Fog seal consists of a small amount of emulsion applied to an oxidized surface and placed in accordance with ODOT specifications. Research project note (Riemer et al. 2011): The fog seal was a typical SS-1 oil diluted to a ratio of 5:1 water to oil and applied to the surface at a rate of 0.1gal/sy.
Figure A.5 – Fog Seal
Asphalt Penetrating Conditioner (ACP), with and without Asphalt Planer JLT Inc. provides the proprietary asphalt penetrating conditioner product, which is designed to rejuvenate (seal and extend the life of) asphalt pavements by penetrating the asphalt, reacting with the bitumen and then hardening. The asphalt planer technique involves using a special machine developed by JLT Inc. to plane the surface of the roadway, removing any humps, ruts, or inconsistencies. Asphalt Penetrating Conditioner can be applied to a recently-planed surface to seal the surface. Research project note (Riemer et al. 2011): The research only focused on determining the effects this product had on texture and therefore, did not test any penetration. Figures A.6 and A.7 show these treatments.
Figure A.6 – Asphalt Penetrating Conditioner
Figure A.7 – Asphalt Planer w/ Asphalt Penetrating Conditioner
Chip Seal, ODOT 3C (5/8″), with and without Fog Seal Figures A.8 and A.9 show ODOT 3C (5/8″) chip seal with and without fog seal. Research project note (Riemer et al. 2011): The binder for the chip seal sections, Cationic High Float Rapid Set – 2P, was provided by Ergon. All chip seal sections were installed by an ODOT chip seal crew.
Figure A.8 Test Section #8 – Chip Seal (5/8″) w/ Fog Seal
Figure A.9 – Chip Seal (5/8″) Chip Seal, ODOT #2 (3/8″) Figure A.10 shows ODOT #2 (3/8″) chip seal. Research project note (Riemer et al. 2011): The binder was Cationic High Float Rapid Set – 2P, provided by Ergon. It was installed by an ODOT chip seal crew. Failure occurred earlier than expected (before two years). It is hypothesized that the existing
asphalt pavement substrate contained enough rutting to make the nominal aggregate size ineffective.
Figure A.10 – Chip Seal (3/8″) Shotblasting, with and without Fog Seal Shotblasting consists of shooting steel balls, called shot, at the roadway surface at high speeds. The impact is designed to remove excess bitumen from the road and expose and retexture existing aggregate. The advantages of this product are that it has no temperature constraints and requires no aggregate or binder material. Figures A.11, A.12 and A.13 show a shotblasted surface. Research project note (Riemer et al. 2011): Blastrac shotblasted the surface using a single directional shot pattern and 4′ wide head using company specifications and recommended rates. Skidabrader shotblasted the surface using a multidirectional shot pattern and 6′ wide head using company specifications and recommended rates. An ODOT contractor installed the fog seal on one of the sections in accordance with ODOT specifications, as described in the preceding fog seal section (Figure A.5). .
Figure A.11 – Shotblasting (Blastrac)
Figure A.12 – Shotblasting (Skidabrader) w/ Fog Seal
Figure A.13 – Shotblasting (Skidabrader)
Figure A.14 shows Novachip, also known as an ultra-thin bonded wearing course (UTBWC). It is a 3/4" thick mixture of well-graded aggregate, polymerized Performance
Grade binder, and mineral fillers that is mixed at an asphalt plant and delivered to the project. It is then placed using a material transfer device and a spray paver.
Figure A.14 – Novachip/Ultra Thin Bonded Wearing Course
PORTLAND CEMENT CONCRETE SUBSTRATE Shotblasting, with and without Nanolithium Densifier For Shotblasting description, see shotblasting in previous figures. Nanolithium densifier is a concrete hardener that is designed to reduce wear/rutting due to abrasion. Research project note (Riemer et al. 2011): Blastrac shotblasted the surface using a single directional shot pattern and 4′ wide head using company specifications and recommended rates. Skidabrader shotblasted the surface using a multidirectional shot pattern and 6′ wide head using company specifications and recommended rates. Nanolithium densifier, supplied by Calumet Lubricants, was applied to one of the surfaces after Shotblasting (Figure A.17). The research found that when the concrete pavement surface was first retextured using shotblasting and then treated with the concrete hardener that it reduced wear due to abrasion and maintained safe skid numbers during the testing period.
Figure A.15 – Shotblasting (Blastrac)
Figure A.16 – Shotblasting (Skidabrader)
Figure A.17 – Shotblasting (Blastrac) w/ Densifier
Traditional Diamond Grinding and Next Generation Diamond Grinding Diamond grinding is a process where a drum-size cylinder with diamond tipped discs is pressed into the pavement while the cylinder is spinning, grinding the surface of the roadway and adding texture. Next generation diamond grinding involves cutting
grooves into the pavement surface by systematically placing larger diameter discs around the spinning cutter drum. These treatments can be seen in Figure A.18 and Figure A.19.
Figure A.18 – Next Generation Diamond Grinding
Figure A.19 – Diamond Grinding Research project note (Riemer et al. 2011): diamond grinding was provided by Penhall, Inc in accordance to company specifications and recommended rates. The following graph (Figure A.20) shows the difference in transverse macrotexture that Next Generation diamond grinding provides, while all others stayed relatively consistent with longitudinal measure.
Figure A.20 Longitudinal and Transverse Macrotexture Results.
E-Krete E-Krete is proprietary and produced by Polycon. It is essentially a Portland cement based slurry seal which is placed on the road at a thickness of 1/8″. This product can then be dragged, like in this test section, or tined to produce the desired texture. This treatment can be seen in Figure A.21.
Figure A.21 – E-Krete
Single Size Chip Seal The only difference between these sections and Test Sections #8 and #9 is that these sections utilized a more uniform gradation of aggregates (1/2” NMAS in this case), referred to as single size. Emulsion rate and type were the same. Test Section #22, shown in Figure A.23, varies from Test Section #21, shown in Figure A.22, because it was treated with Calupave, the same low penetration fog seal material that was used in Test Section #1.
Figure A.22 – Chip Seal (Single Size)
Figure A.23 Test Section #22 – Chip Seal (Single Size) w/ Calupave
The final test section (#23) is a microsurfacing process, shown in Figure A.24. Microsurfacing is a mixture of well-graded aggregate, polymerized asphalt emulsion, fillers, additives, and water that is mixed on the project.
Figure A.24 Test Section #23 – Microsurface
APPENDIX B PRIMARY CAPABILITIES AND FUNCTIONS OF PAVEMENT PRESERVATION TREATMENTS
Primary Capabilities and Functions of Preservation Treatments for Hot Mix Asphalt-Surfaced Pavements Prevention Preservation
Crack filling
Treatment
Crack sealing
Seal/ Waterproof Pavement
Rejuvenate Surface/ Inhibit Oxidation
Cold milling
Fill Surface Defects*
Fill Stable Ruts
Profile milling Shotblasting Shotblasting with lithium densifier
Rejuvenation Fog seal Scrub seal Slurry seal Portland cement slurry seal (E-Krete) Microsurfacing Chip seal Ultra-thin hot mix asphalt overlay Ultra-thin bonded wearing course Hot mix asphalt (
