More Roads for the Money
PhD Thesis
Iman Mirzadeh
Department of Civil and Architectural Engineering School of Architecture and the Built Environment KTH, Royal Institute of Technology SE-100 44 Stockholm SWEDEN
2014
© Iman Mirzadeh TRITA-TSC- 14-003 ISBN 978-91-87353-41-3
Abstract: In order to keep the quality of the road networks at an acceptable level, large amount of investments for rehabilitation and maintenance activities are necessary in addition to investments in new and reconstructed roads. Therefore, an efficient allocation of road investment funds is of great economic importance. Decreasing the overall contractors’ bid prices and lowering the financial/technical risks imposed to the Swedish Transport Administration are two key strategies to achieve more value of money for the roads. An overall decrease in bid prices can be achieved either by lowering the risks involved in the contracts or by applying more effective hedging strategies. This Thesis aims at developing a framework to evaluate the financial and technical risks regarding asphalt pavement projects from the transport administration and contractors perspectives. Moreover, it enables valuation of different hedging strategies such as long term material/fuel contracts with suppliers. A significant part of the costs associated with asphalt pavements is related to the cost of oil products such as bitumen, fuel oil and transportation fuel. Moreover, the cost of energy has been usually subjected to high fluctuations. However, the financial risk regarding the energy price is not reflected in the discount rate suggested by the Swedish transport administration. Therefore, application of the common range (i.e. 0-8%) for discount rate sensitivity analysis regarding road projects may lead to under-estimation of the financial risk. It is observed that the proper range for discount rate sensitivity analysis of asphalt pavement has to be between -20% and 30%. However, the financial risk regarding the cost of asphalt pavements, due to the presence of Price Adjustment Clauses (PACs) (for bitumen), is shared between the Swedish Transport Administration and contractors. Additionally, the presence of PACs, similar to other governmental support mechanisms such as loan and revenue guarantees, results in asymmetric project’s payoff function which cannot be assessed by traditional methods (e.g., discounted cash flow (DCF)). In order to overcome this issue, an option pricing framework is suggested which can assess the project value and reflect the financial risks. Furthermore, the suggested framework can evaluate the project value under different features of highway projects such as limited liability of the public private partnership (PPP) firm and government revenue guarantees. It was observed that although the application of PACs significantly decreased the risk profile regarding the construction cost for the contractors it imposed a significant financial risk to the road administration and ultimately to the tax payers. Therefore, the value of PACs should be considered in the decision making process regarding large projects. As a general principle, a systematic risk should be borne by the party which either is best placed to manage it or has the possibility to minimize its impacts. Contractors and suppliers in large projects can sometimes be the best parties to minimize the financial and technical risks. They might have the possibility to depot the construction supply at the beginning of the project and by doing so hedge against the financial risk regarding the fuel/material cost escalations. Moreover, contractors may have the possibility to lower the financial and technical risks by implementing better solutions. For instance, application of preventive maintenance activities (i.e. thin asphalt layer) by smoothing the surface can lower the exerted dynamic loads and hence increase the pavement life span and decrease the technical and financial risk. i
Keywords: Road & Highways, Life Cycle Cost, Options Pricing, Pavement Design, Risk Assessment, Maintenance.
ii
Preface
The work presented in this Thesis has been carried out with financial supports from the Swedish Transport Administration (Trafikverket). I would like to express my gratitude to my supervisor Professor Bjorn Birgisson for his guidance and support throughout the whole process of this work. Mr. Måns Collin and Dr. Susanna Toller are acknowledged for their valuable feedback and help regarding this project.
Iman Mirzadeh Stockholm, Oct ’14
iii
iv
Publications
This Thesis is based on the following journal papers: I.
Butt, A., Mirzadeh, I., Toller, S., & Birgisson, B. (2014) Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives, International Journal of Pavement Engineering, 15:4, 290-302.
II.
Mirzadeh, I., Butt, A., Toller, S., & Birgisson, B. (2013). Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements, Structure and Infrastructure Engineering. doi:10.1080/15732479.2013.837494.
III.
Mirzadeh, I., & Birgisson, B. (2013). Accommodating energy price volatility in life cycle cost analysis of asphalt pavements, accepted for publication in: Journal of Civil Engineering and Management.
IV.
Mirzadeh, I., & Birgisson, B. (2014). Evaluation of Highway Projects under Price Adjustment Clauses Based on an Option Pricing Framework, submitted for publication in: Journal of Construction Engineering and Management.
V.
Khavassefat, P., Mirzadeh, I., & Birgisson, B. (2014). A Life Cycle Cost Approach on Minimization of Roughness-Related Damages on Flexible Pavements, submitted for publication in: Journal of Infrastructure Systems.
In Paper I the author of this Thesis was involved in the literature review and the framework development. In Paper V, he was the corresponding author and responsible for all the aspects regarding cost analysis, project management and pavement design.
Other relevant publications: i.
Mirzadeh, I., Butt, A.A., Toller, S. and Birgisson, B. (2012). A Life Cycle Cost Approach based on the Calibrated Mechanistic Asphalt Pavement Design Model. European Pavement and Asset Management, EPAM, 5–7 Sep, Malmö, Sweden.
ii.
Butt, A.A., Mirzadeh, I., Toller, S. and Birgisson, B. (2012), "Feedstock Energy of Bitumen and Efficient Electricity Production", ISAP 2012 international Symposium on Heavy Duty Asphalt Pavements and Bridge Deck Pavements, 2325 May, Nanjing, China.
v
vi
Contents
1. Introduction …………………………………………………………………………...1 2. Background…….…..………………..………………………………………………....5 3. Methodology……………..…………………………………………………………….8 3.1. Summary of the scientific tools …………………………….……………………….....8 3.2. Framework…………………………………………………...………………………....9 4. Results and Discussions Regarding the Appended Papers……………………..25 4.1. LCA Framework regarding asphalt pavements ……..………………………………...25 4.2. LCCA Based on the fundamental cost contributors……… ………………………......25 4.3. Accommodating energy price volatility in LCCA of asphalt pavements...….………....27 4.4. Evaluation of highway projects based on an option pricing framework………………..29 4.5. Estimating the surface deterioration cost in flexible pavements………………………..32 5. Summary and Conclusions ...…………………………………………………………37 6. Recommendations for Further Research …………………………………………….39 References ….………………………………………………………………………………41
vii
viii
1. Introduction In order to keep the quality of the roads at an acceptable level, large amount of investments for rehabilitation and maintenance activities are necessary in addition to investments in new and reconstructed roads. Therefore, an efficient allocation of road investment funds is of great economic importance. The price of energy has been identified as one of the most important factors affecting the cost of construction projects (e.g., Hastak & Shaked, 2000; Baloi & Price, 2003; Jaafari, 2001). Furthermore, the increase in cost of oil products has been the main cause of rises in highway construction cost (e.g., Wilmot & Cheng, 2003). In the mid-2000s highway construction showed a sharp increase in prices (e.g., Zhou & Damnjanovic, 2011). Consequently, construction projects in 2006 were valued two times more than similar projects in 1997 (Pandit et al. 2009). This was believed to be related to the dramatic increase in the price of crude oil and its products such as bitumen, fuel oil and transportation fuel. Evidently, a direct correlation between the oil price and the cost of asphalt pavement projects exists (Gallagher & Riggs, 2006; Damnjanovic & Zhou, 2009). In traditional LCC studies evaluation of future costs is usually done by predicting the effect on costs from likely trends in interest rate and inflation [(Eisenberger & Remer, 1977) (Walls & Smith, 1998) (FHWA, 2003) (Chan, Keoleian, & Gabler, 2008) (Zhang, 2008) (Santos & Ferreira, 2013) (Mandapaka et al., 2012)]. The prediction of inflation is sometimes based on broad indices, such as Gross Domestic Product chain deflator or intermediate indices, such as Consumer Price Index (CPI) or even narrower (e.g. highway construction cost index) (FHWA, 2002). The use of Producer Price Index (PPI) for non-residential construction was suggested by MDOT, (2012) in order to predict current and future costs. Recognizing that the broader indices may not adequately reflect the development of future costs related to road construction, the Swedish Transport Administration adopted a road construction price index entitled the ‘Vägindex’ (STA, 2012). However, the implementation of Vägindex has recently been criticized by the Swedish National Audit Office (NAO) based on the following reasons (NAO, 2010): 1) cost trends in the road construction sector are not comparable with that of the general economy, 2) the use of Vägindex does not adequately reflect the cost fluctuations resulting from the relatively high price trends in the construction industry, 3) the widespread use of Vägindex is likely to reduce the awareness of potential future costs at the Swedish Transport Administration. These concerns are illustrated in Figure 1, which shows that the Vägindex approximately followed the Swedish CPI for the last 25 years, but failed to adequately capture the dramatic increase in the price of oil products during this period. A significant part of the costs associated with asphalt pavements is related to the cost of oil products such as bitumen, fuel oil and transportation fuel. However, the calculation of Highway Construction Indices is usually based on the bidding prices regarding the awarded contracts. Moreover, due to the presence of the price adjustment clauses (for bitumen) the financial risk regarding the energy price is not reflected on Vägindex. Therefore, application of Vägindex may lead to false estimation of project costs for the transport administration.
1
Figure 1. The normalized indices, regarding Swedish road construction, CPI and crude oil price for Sweden. The Swedish Transport Administration is one of the largest public sector clients in Sweden, purchasing goods and services to an annual amount of nearly SEK 40 billion. The procurement process is the Swedish Transport Administration’s frequent key activity. Moreover, the Swedish parliament has urged the Swedish transport administration to increase the efficiency and improve the conditions for effective competition in procurement process. In order to develop better solutions and increase the competition for large infrastructure projects, the Swedish Transport Administration has intended to transfer a large part of responsibilities to contractors. The design-build projects with long-term maintenance responsibility (Totalentreprenader), which as of 2014 were 34% of the contracts, are to be increased to 50% of the projects by 2018. However, based on the NAO audit (NAO, 2012) implementation of the new design-build contracts has not led to more value of money, despite the expectation of the Parliament, government and the Swedish Transport Administration. In order to hedge against the financial and technical risks in projects with longer responsibility periods, contractors usually have incorporated a premium in bid prices (Damnjanovic & Zhou, 2009). Therefore, effective reduction regarding these premiums is equivalent to more value of money from the tax payers’ perspective. This can be achieved either by lowering the risks or by applying more effective hedging strategies. Transferring responsibilities and therefore risks to the contractors is been used as a mechanism for risk reduction by transport administrations. However, this will only be economically beneficial if the amount of risks counterweight the risk premiums. Currently, due to extensive application of PACs (for bitumen) in Sweden, the financial risk regarding the price of commodities is highly imposed to the Swedish Transport Administration. The capital asset pricing model (CAPM) is the common method for calculating the projects value under risk. It is based on the modern portfolio theory suggested by Markowitz (1952)
2
and further developed by (Yeo & Qiu, 2003, Lee et al. 2009, Fama & French 2004 and Campbell 1993). The discount rate in this model is presented as the sum of the risk free rate and a risk premium. Hence, the determination of the discount rate is an assessment of how much to increase the risk free rate to account for all the corresponding risks in the project (Baker & Fox, 2003). However, there are certain problems regarding the application of traditional approaches such as DCF and CAPM for highway projects: (1) in the CAPM method a higher risk in the project results in a higher discount rate which corresponds to lower discounted values. Whereas a lower value regarding revenues is reasonable, a lower value regarding cost due to higher risk is not relevant. (2) DCF and CAPM methods do not explicitly capture the uncertainty regarding costs and revenues during the operation phase of the projects (Ashuri et al. 2012). (3) Certain features of highway projects such as limited liability of the PPP firm, government revenue guarantees, price adjustment clauses (PACs) and long term supply contracts with material/energy suppliers result in contingent cash flows. These contingent cash flows creates asymmetric payoffs, which cannot be assessed by traditional methods such as DCF or CAPM. Options are a class of derivatives with asymmetric (or non-linear) payoffs. It is suggested that the option pricing method can be applied in assessment of infrastructure projects’ features, such as financial viability (e.g., Ho & Liu, 2002), governmental revenue guarantees (Chiara, Garvin & Vecer, 2007; Cheah & Liu, 2006; Galera & Solino, 2010), enhancing the contracts’ risk allocation (Quiggin, 2005; Nombela & Rus, 2004) and managerial flexibility (Majd & Pindyck, 1987; Zhao & Tseng, 2003; Yiu & Tam, 2006). Despite the high interest among researchers real option analysis has not been widely used by practitioners. Based on a survey on Fortune 1000 companies in the US, it was observed that only 14.3% of the companies were using real options (Block, 2007). A similar survey on Canadian firms, reported that only 16.8% of the firms reported implementing real options (Baker, Dutta & Saadi, 2011). The probable reason is that the financial principles are not fully applicable in engineering practice (Neufville et al. 2006). For example, pricing of a financial option deals with statistical data such as the asset’s volatility that, while meaningful in finance, have no clear equivalent in engineering. Moreover, the current use of real options has focused on a project’s upside potential (i.e. managerial flexibility) (e.g. Trigeorgis, 1996; Triantis, 2003; Chiara and Kokkaew, 2009). However, evaluating a project’s downside potential (i.e. risks) is more consistent with the risk-averse nature of the investors (Espinoza & Morris, 2013). In this aspect, studies such as evaluating revenue guarantees (e.g. Ashuri et al. 2012) and long term contracts for material procurement (e.g. Ng, Björnsson & Chiu, 2004) can be of a great interest from the practitioners’ perspective. In the last decades real option models have been implemented for assessment of highway projects’ features, such as financial viability and government revenue guarantees. However, to the authors’ knowledge, the effect of price adjustment clauses on the project’s value has not been studied.
3
1.1. Objectives This Thesis aims at developing a framework to enable valuation of asphalt pavement projects under financial and technical risks from the Swedish Transport Administration and contractors perspectives. To reach the research aim the work was performed through the following steps: •
Setting the system boundaries for calculating feedstock and expended energies regarding asphalt pavement projects (Paper I).
•
Developing of a transparent life cycle cost (LCC) framework based on the cost of energy and time associated with road construction which enables robust LCC analyses regarding asphalt pavements (Paper II).
•
Enhancement of the LCC framework in order to reflect the financial risks related to asphalt pavement projects (Paper III).
•
Refining the existing option pricing techniques to enable valuation of different types of highway projects in the presence price adjustment clauses (Paper IV).
•
Application of the LCC framework in order to evaluate the benefit of preventive maintenance activities (i.e. thin asphalt layers) regarding lowering the dynamic impact from the vehicles, increasing the pavement life, decreasing the project cost and limiting projects’ risk profile (Paper V).
1.2. Scope and Limitations The LCC framework has focused on the construction and rehabilitation of asphalt pavements. The activities/items that are mutual for the design alternatives such as routine maintenance, traffic costs during the design life, the amount of earth work and construction of the unbound material are not included in the study. A design alternative for the decided road alignment is defined in terms of the initial asphalt layer thickness, the overlay thickness and the rehabilitation frequency.
4
2. Background Life Cycle Cost Analysis (LCCA) is a methodology for systematic economic evaluation of a project for its defined life time (ISO15686-5, 2008). The minimum expected total cost over the design life of an infrastructure project has been the most common criterion used in design optimizations [(Frangopol & Liu, 2007) (Christensen, 2012) (Safi et al., 2013)]. LCCA is growing in importance as a tool for providing cost estimation over the life time of road projects. It provides a system based framework for identifying current costs and key factors affecting projected future costs and therefore holds the promise to help designers, road administrations, and contractors with choosing the most economically efficient investment strategies for roads. It is therefore extremely important that the LCC system include the key variables that drive future costs and thus provide a framework for reducing the risk of under- or over estimating the future costs for maintenance and rehabilitation activities. The prediction of future costs can be done by application of a discount rate. The social discount rate is widely used for public infrastructure projects. Different approaches such as social rate of time preferences (SRTP), marginal social opportunity cost of capital (SOC), weight average (WA) and shadow price of capital (SPC) have been applied for selecting the social discount rate. Evans & Sezer (2005) estimated discount rates based on the social rate of time preferences for countries such as USA, UK, Germany, Japan, France and Australia. Their result ranged between (3-5%) which clearly contradicted with the official discount rate at the time. Azar (2007) suggested that the social discount rate for the US is 5.66%, with a 95% confidence interval ranging from 5.62 to 5.71%. However, Lally (2008) argued that this underestimates the confidence interval on the discount rate primarily through ignoring uncertainty surrounding the expected return on risky assets. Percoco (2008) suggested a discount rate for Italy which was 1.2 to 1.3% lower than the official discount rate (5%). Due to the lack of an agreed approach for selecting the discount rate for the evaluation of public projects, many transport agencies across and within countries have used different discount rates for their public projects (e.g. Ferreira & Santos 2013). However, due to the high amount of energy related costs in asphalt pavement projects, the inflation regarding this sector is usually different from the rest of economy. Therefore, the use of social discount rate for asphalt pavements can be questioned. When uncertainties are involved in a project the discount rate can be adjusted based on capital asset pricing model (CAPM). The CAPM is widely used in calculating the discount rate regarding public private partnership (PPP) or built operate transfer (BOT) projects (e.g. Ashuri et al. 2012, Yeo & Qiu, 2003, Lee et al. 2009, Fama & French 2004, and Campbell 1993). The CAPM is based on the modern portfolio theory developed by Markowitz (1952). In the CAPM model the expected return is presented in terms of the risk free rate, e.g. interest rate arising from government bonds, plus a risk premium. Therefore, the calculation of the discount rate is a qualitative assessment of how much to increase the discount rate over the risk free rate to account for the projects’ risk (Baker & Fox, 2003). Although NPV is considered as the superior technique among other traditional methods (e.g. internal rate of return (IRR) and accounting rate of return (ARR)) it has some drawbacks such as failing to explicitly capturing the value of management’s flexibility (Trigeorgis, 1996). Moreover, it cannot assess the value of government support mechanisms (e.g. debt guarantees) and the negotiation possibility (Ho &Liu, 2002). 5
Options are a class of derivatives with asymmetric payoffs. An option provides the holder with the right to buy (i.e. call option) or sell (i.e. put option) a quantified underlying asset at a fixed price at or before the expiration date. Financial option pricing was established by Louis Bachelier in 1900 and matured with the seminal Nobel Prize winning work by Black, Scholes & Merton (Black & Scholes, 1973; Merton, 1973). Derived from financial option theory, real option pricing developed by Myers (1977), applies to non-financial assets (e.g. infrastructure projects). It is suggested that the option pricing method can be applied in proper assessment of infrastructure projects’ features, such as financial viability (Ho & Liu, 2002), government revenue guarantees (Chiara, Garvin & Vecer, 2007; Cheah & Liu, 2006; Galera & Solino, 2010), enhancing the contracts’ risk allocation (Quiggin, 2005; Nombela & Rus, 2004) and managerial flexibility (Majd & Pindyck, 1986; Zhao & Tseng, 2003; Yiu & Tam, 2006). Despite the high interest among researchers real option analysis has not been widely used by practitioners. Based on a survey on Fortune 1000 companies in the US, it was observed that 14.3% were using real options (Block, 2007). A similar survey of Canadian firms, reported that only 16.8% of the firms reported implementing real options (Baker, 2011). The probable reason is that the financial principles are not fully applicable in engineering practice (De Neufville et al. 2006). For example, pricing of a financial option deals with statistical data such as the asset’s volatility that, while meaningful in finance, have no clear equivalent in engineering. Moreover, the current use of real options has focused on a project’s upside potential (i.e. managerial flexibility) (e.g. Trigeorgis, 1996; Triantis, 2003; Chiara & Kokkaew, 2009). However, evaluating a project’s downside potential (i.e. risks) is more consistent with the risk-averse nature of the investors (Espinoza & Morris, 2013). In this aspect, studies such as evaluating revenue guarantees (e.g. Ashuri et al. 2012) and long term contracts for material procurement (e.g. Ng, Björnsson & Chiu, 2004) can be of a great interest from the practitioners’ perspective. Different models such as continuous-time approach suggested by Black-Scholes, (1973), discrete-time approach (e.g. binomial (e.g. Cox et al. 1979) and trinomial (e.g. Boyle, 1988)) and Monte-Carlo options pricing (e.g. Boyle, 1977) have been used for valuation of options. The Black-Scholes formula can be applied for estimating the price of European options. The model can simplistically calculate the value of an option based on five inputs (i.e. time to maturity, initial price of the underlying asset, exercise price, risk free rate of return and volatility). Some drawbacks regarding the application of this approach are: (1) the variables should follow standard normal distributions, (2) the cost/price processes should be continuous, (3) Due to lack of transparency in the model, its application for large infrastructure projects is problematic (Triantis, 2003). The binomial option pricing model, based on a risk neutral approach, provides a numerical discrete-time model for the valuation of options. Similar to the Black-Scholes approach, binomial model also requires advance financial knowledge. Moreover, it is difficult to apply the model if there are more than one type of uncertainty involved in the analysis. On the contrary, the binomial model can be applied for valuing different types of options including American options (Martins & Cruz, 2013). Monte-Carlo simulations are effective when dealing with several sources of uncertainties. However, this method should be applied in cases in which it is difficult, if not impossible, to use other more accurate methods.
6
3. Methodology
3.1. Summary of the scientific tools Life Cycle Cost Analysis (LCCA) Life cycle cost Analysis (LCCA) is a methodology for systematic economic evaluation of a project during its defined lifetime (ISO15686-5, 2008). LCCA is applied to compare alternatives which have the same level of service over the design life. The minimum expected total cost over the design life of an infrastructure project has been the most common criterion used in design optimizations (Frangopol & Liu, 2007; Safi, Sundquist, Karoumi, & Racutanu, 2013).
Life Cycle Assessment (LCA) Life Cycle Assessment (LCA) has been used as a comprehensive tool for quantifying and assessing environmental impacts of a product during whole (or part) of its life cycle. By providing the environmental profile of an asset, LCA enables measures to reduce resource used and environmental burdens.
Calibrated Mechanics based pavement design (MC) model The Calibrated Mechanics based (CM) model is based on the mechanistic empirical pavement design procedure developed by Wang, Birgisson & Roque (2007) which is calibrated by Gullberg et al. (2012) for the Swedish conditions. This approach allows for prediction of crack initiation and growth in asphalt pavements for a predefined loading history during a design life. The analyses are based on the principles of viscoelastic fracture mechanics. The calculation of the design thickness for an asphalt pavement is based on a threshold called dissipated creep strain energy limit. This threshold is defined as the amount of damage a mixture can endure before healable micro-cracks coalesce into non-healable macro-cracks.
Options pricing Derived from financial option theory, real option pricing developed by Myers (1977), applies to non-financial assets (e.g. infrastructure projects). It is suggested that the option pricing method can be applied in proper assessment of infrastructure projects’ features, such as financial viability, governmental revenue guarantees, enhancing the contracts’ risk allocation and managerial flexibility.
7
3.2. Framework LCC of asphalt pavements includes different cost components which are affected by national and global economy. Traditional LCC studies adjust future costs according to interest rate and inflation at the national economy level. However, tradable goods such as crude oil, fuel oil, transportation fuel and sometimes electricity have a global market and their future prices can be independent of the country’s economic indicators such as inflation rate. Therefore, in this framework the costs are divided into energy related and time related costs. Time related costs are those affected by the national economy such as labor/equipment and road user costs. Energy related costs are those affected by the global economy such as bitumen cost and expended energy costs regarding production, construction and material transportation. The costs of construction and rehabilitation are divided into energy and time related components (Figure 2). The time related components were those concerning labor and equipment for construction and rehabilitation activities together with the delay cost for the road users. Energy related costs are separated into crude oil energy and expended energy. The crude oil energy is the energy stored in the material which in the context of this Thesis represents the value of crude oil. The expended energy is the amount of the energy that is spent during material production, construction and rehabilitations of the asphalt road. The expended energy in the refinery and the asphalt plant was related to bitumen, aggregate and asphalt mixture production.
Figure 2. LCC framework for asphalt pavements. Agency costs (white), user costs (grey).
8
Regarding the construction site, the expended energy was attributed to laying, compacting, milling and resurfacing. Furthermore, the expended energy regarding the user cost during maintenance and rehabilitation activities was defined in terms of extra amount of fuel used by the vehicles in the work zone. The amount of expended energy for the transportation was related to all required material distribution from the refinery to the construction site. The amount of fuel used and depreciation of the machineries such as trucks, pavers and rollers were included in the framework. The amount of depreciation cost was included in the labor and equipment costs. The energy prices in Europe are affected by a number of factors. Prices of primary fuels such as oil, coal and gas will influence the price of secondary energy sources such as electricity and road fuels. Moreover, they can be affected by the other primary fuel prices. Coal and gas, specifically, can act as substitute in electricity generation. However, the price of oil is the main driver of energy prices (DECC, 2012). Products refined from crude oil, such as transportation fuels, gas oil and fuel oil, are obviously closely linked to the price of crude oil and follow oil price fluctuations (DECC, 2012). Other affecting factors on the oil products price are the rates of fuel duty and Value Added Tax (VAT). There has been a delay between the changes in the oil price and that of the other energy sources (Figure 3). The delay was shortest for the oil products such as fuel oil. Whereas, the delay is longer for electricity price, as the prices of coal and gas have to change before the price of electricity.
Figure 3. Fluctuation of different energy source prices in respect to oil price (price index 2005 = 100). In order to relate construction costs to a limited (minimum) number of fundamental cost indices, the crude oil price inflation index was chosen as the inflation index for the energy related costs in this framework.
9
The calculation regarding the energy cost is performed for crude oil and expended energy components (Table 1). The crude oil energy represents the value of crude oil price in bitumen and modifiers. Therefore, the costs of bitumen and modifiers are defined as the sum of the corresponding costs for crude oil and the expended energy for production and transportation. The cost of energy regarding material and processes are expressed based on the quantity of the involved energy sources and the corresponding energy prices (Table 1). Table 1. Energy related variables regarding LCC framework
Crude oil (MJ)
Energy source A (MJ)
Energy source B (MJ)
Cost of Energy (€)
Modifier
qc1 qc2
-
-
E1 E2
Aggregate production
-
qa3
qb3
E3
Bitumen production
-
qa4
qb4
E4
Modifier production
-
qa5
qb5
E5
Asphalt production
-
qa6
qb6
E6
Transportation
-
qa7
qb7
E7
Laying asphalt
-
qa8
qb8
E8
Compacting asphalt
-
qa9
qb9
E9
User's energy loss
-
qa10
qb10
E10
Bitumen
The unit cost of energy for each item (Table 1), which is related to crude oil and the other energy sources can be calculated based on Equation 1. The calculations can be extended in case of having more energy sources involved.
Ek = qck × Pc + qak × Pa + qbk × Pb + ...
[1]
Where: qck, qak, qbk are the quantities (MJ) regarding crude oil, energy source A and energy source B. Pc, Pa, Pb are the unit prices (€/MJ) regarding crude oil and energy sources A and B at the base year. The amount of energy used related to asphalt pavement works can be found in Life Cycle Inventory (LCI) data bases and several Life Cycle Assessment (LCA) publications, including LCA report by Stripple (2001) in Sweden, Energy Conservation in Road Pavement Design in Europe (ECRPD, 2009), EcoInvent (Swiss center for Life Cycle Inventories, 2011), the asphalt inventory produced by the Athena Institute in Canada (Athena Institute, 2006), the U.S. Life Cycle Inventory (USLCI) produced by the National Renewable Energy Laboratory (National Renewable Energy Laboratory, 2011). Different sources represent different local condition
10
and technologies. The amount of different sources of energy regarding the pavement works presented by Stripple (2001) is used in this paper since it represents the Swedish condition. The costs of time related variables regarding the labors and road users’ delay are presented in Table 2. The value of time included the hourly wage of the labors together with the amount of hourly depreciation regarding the equipment. The hourly depreciation was defined as the initial price of the equipment divided by the useful life of the equipment (hours). The calculation regarding the user delay cost should be done separately for the personal cars and heavy vehicles as they should be treated with different value of time.
Table 2. Time related variables regarding LCC framework
Labor & equipment
Road users
Item
Time Spent (Hr.)
Value of time (€/Hr.)
Transportation
t1
CL1
T1
Laying asphalt
t2
CL2
T2
Compacting asphalt
t3
CL3
T3
Milling and resurfacing
t4
CL4
T4
User's delay cost
t5
Cu
T5
Cost of time (€)
Whenever the time value of money is considered, the LCC is the sum of all costs in the lifecycle discounted at an interest rate i and an inflation j to the year zero (Eisenberger & Remer, 1977). The Net Present Value (NPV) of a cost C at the yth year should be calculated according to Equation 2. The traditional discount rate (r) is calculated according to Equation 3 (Amini et al., 2012).
1+ j NPV= C × 1+ i y
[2]
Where, C is the sum of energy and time related costs; y is the number years after the construction; j is the yearly inflation rate and i is the rate of interest.
= r
1+ i −1 1+ j
[3]
Based on the above concept NPV of a construction or a rehabilitation activity is reformulated as: n n y y 1 NPV = ∑Ek × (1 + jC ) + ∑ Tk × (1 + j ) × k =0 k =0 1+ i
11
y
[4]
Where: Ek and Tk are the cost of energy and time for item k and jC is the yearly oil price inflation rate. The NPV can also be calculated form Equation 5 based on the discount rate (FHWA 2002).
1 NPV= C × 1+ r
y
Since C consists of energy related costs (CE) and time related costs (CT), the discount rate should reflect inflations both regarding CE and CT. The inflation regarding CE and CT are defined as oil price inflation and general inflation respectively. For a portfolio consisting of the investments a and b where portion α of the wealth is placed in a and remainder (1- α ) in b the equivalent rate of return and its standard deviation are expressed in Equation 6-7 (Strong, 2008). r = α ⋅ ra + (1 − α ) ⋅ rb
= σ
[6]
[7]
α 2σ a2 + (1 − α ) 2 σ b2 + 2 βα (1 − α )σ aσ b
Where, ra and rb are the return on investments and σa, σb are their standard deviations. β is the correlation coefficient between the rates of return of the two investments. Since β indicates the degree to which asset’s expected return is correlated with broader market outcomes, it is simply an indicator of an asset’s vulnerability to systematic risk. A positive value for β indicates that there is a positive correlation between ra and rb. A zero value would indicate that ra and rb are completely independent. Furthermore, a negative value indicates that there is an inverse relationship between them. It was shown by Mirzadeh et al. (2013) that crude oil price inflation can be used as the inflation index for the energy related items. Consequently, the oil price discount rate (r1) can be used as the discount rate for the energy related costs. The general discount rate was chosen as the discount rate for the time related items (r0). Assuming α as the portion of the energy related costs Equations 6-7 can be rewritten as: [8]
r = α ⋅ r1 + (1 − α ) ⋅ r0 = σ
[9]
α 2σ 12 + (1 − α ) 2 σ 02 + 2 βα (1 − α )σ 1σ 0
Where, r is defined as the equivalent discount rate. r1 is the discount rate for energy related items based on the interest rate and the oil price inflation and σ1 is its standard deviation. r0 is the traditional discount rate based on interest rate and general inflation and σ0 is its corresponding standard deviation. The portion of energy related costs ( α ) can vary between zero to one. In this context Beta is chosen to describe the correlated volatility of r1 in relation to r0 (Equation 10). 12
β=
Co v(r1 , r0 ) Var (r0 )
[10]
By inserting the interest rate, the general inflation and the oil price inflation Equation 7 can be rewritten as: 1 1 1+ i )+ r = α × (1 + i ) × ( − −1 1 + jc 1 + j 1 + j
[11]
Therefore, NPV based on the equivalent discount rate can be obtained by inserting r into Equation 5. The NPV of the alternatives are only comparable if they have the same life span. In the case of comparing alternatives with different life span equivalent annual cost (EAC) is often used as a decision support tool. The EAC is the cost per year of owing an asset over its defined life span. The EAC is calculated by multiplying the total net present value (TNPV) by the annuity factor ( At , r ) (Mirzadeh et al. 2013): EAC= TNPV × At , r= TNPV ×
r 1 − (1 + r ) − d
[12]
Where, TNPV is the sum of all the NPVs regarding construction, maintenance and rehabilitation activities during the life span and d is the design life of the asphalt pavement. The portion of energy related cost regarding each project may vary for the initial construction and future maintenance and rehabilitation. Therefore, α should be calculated based on the discounted energy and time related cost. However, the equivalent discount rate itself is a function of α . In order to solve this problem a flowchart for evaluating the discount rate for asphalt pavements is presented in Figure 4.
13
Figure 4. The flowchart to calculate the equivalent discount rate. The design inputs such as the thickness of the asphalt layer for construction are to be obtained from the Calibrated Mechanics based model (MC). The MC model is based on the mechanistic empirical pavement design model developed by Wang, Birgisson & Roque (2007) which is calibrated by Gullberg et al. (2012) for Swedish conditions. The initial α is calculated based on undiscounted energy related and time related costs. The equivalent discount rate is evaluated as a function of α , general inflation, oil price inflation and interest rate. The new α is then calculated based on NPV for energy related and time related items. The iteration continues as long as the difference between α i+1 and α is larger than 1%. Once α satisfied the required condition, r can be used as the discount rate for the asphalt pavement project. A sensitivity analysis is to be performed in order to assess the effect of expected variation of the discount rate on the life cycle cost of the project. The sensitivity analysis should be done based on the historical values regarding interest rate, general inflation and oil price inflation.
14
To justify the application of a preventive maintenance, the equivalent annual costs (EACs) in the presence and the absence of the maintenance activity should be compared. In order for the preventive maintenance to be economically feasible, its benefit in increasing the pavement life should surpass the activity cost. The risk profile regarding cost of the alternatives can be assessed by applying a binomial lattice. In economics, the binomial lattice is a random walk model to capture the uncertainty regarding a variable that changes overtime (Copeland & Antikarov 2001). The financial risk regarding the cost of construction and maintenance activities is evaluated during the construction period (T). The analysis starts at the first node with the total discounted cost (C0) regarding the alternative. After each time step the cost are increased by applying an upward multiple (u) or decreased by application of a downward multiple (d) (Figure 5a). The probability regarding the upward movement is p and the probability for the downward movement is 1-p (Hull 2008). The parameters regarding the binomial lattice are the equivalent standard deviation (σ), the time step (∆t) and the discount rate (r). The jump factors and the probability for the upward movement can be calculated by Equation 13.
u = eσ
∆t
u = e −σ
∆t
p=
e r ∆t − d u−d
[13]
Figure 5. (a) Binomial lattice to model cost uncertainty; (b) randomly generated paths along the binomial lattice The last node can be reached by 0 ≤ m ≤ T upward and 0 ≤ T − m ≤ T downward movements along the binomial lattice (Figure 5b). The probability for having exact m upward movements on the last node for n number of time steps can be calculated by the probability mass function (Equation 14). Pr (total cost at the end of analysis period = C0 × u m d T − m ) =
n! p m (1 − p)T − m k !(n − k )!
[14]
By calculating the probability regarding each node on the last layer the probability density function (PDF) for the cost of each alternative can be obtained.
15
Highway contracts are arrangements in which a private party commits to providing a range of services (e.g. design, construction, finance, maintenance and operation). In exchange for the provided service, the private party might receive either a share of tolls on the highway for limited period of time or simply the fixed contract price. Based on the amount of activities and risks assumed by the private party, highway contracts are classified into traditional bid-build, design-build, design-build-finance and design-build-finance-operate (FHWA, 2010). In tradition public contracts (design-bid-build and design-build) the project costs regarding the initial construction and future maintenance and rehabilitations (M&R) are compensated by the road agency (Figure 6a). The costs regarding the contract items are listed in the tender document and the total estimated cost plus the expected profit is presented as the bid price. However, the material/fuel cost fluctuations usually are compensated according to price adjustment clauses (PACs). Design-build-finance-operate projects, which form a public private partnership (PPP), could have different compensation means such as availability payments and pass-through tolls. In toll roads the concessionaire receives its revenue directly from the road users by collecting tolls, however, in contracts with availability payments the concessionaire recieves its revenue from the road authority based on particular project milestones (e.g. delivering the project within a defined deadline) or performance standards. The financing arrangement is one of the most important characteristics of a design-build-finance-operate project. While tradition design-bid-build projects are usually financed within an existing firm, design-build-finance-operate projects are financed by forming a leveraged PPP firm which is independent from the projects developers. Different parties involved in these projects’ financing deals include promoters, construction contractors, road authority, lenders and insurance providers (Figure 6b). The developers and the government are the key stakeholders of the PPP project. The equity value is one of the most important indicators in financial assessment of a design-build-financeoperate project from both government and developers’ perspectives. The bankruptcy condition of a PPP firm is determined based on the asset value and debt value of the firm.
16
Lenders
Debt service
Debt
Const. Costs Const. Costs Road Agency
M&R Costs
M&R Costs Contractors/ Project Developers
Road Agency
PACs
Tolls
Tax
Tax
Levy
Guarantees
PPP firm
PACs
Levy
Tolls
Yield Equity
Users / Tax Payers
Users / Tax Payers
Sponsors/ Investors
Figure 6. (a) Financial arrangement of traditional design-bid-build contracts; (b) Financial arrangement of design-build-finance-operate projects. When exercising a financial call option, the owner of the option purchases the underlying asset, however, in PPP projects, the exercise of an option is analogous to the bankruptcy or completion of the project. In PPP projects in order to protect the equity holders the bankruptcy condition is determined by the lenders. Generally, the bankruptcy condition can be triggered if the borrower is unable to meet the repayment schedules. However, in PPP highway projects, since there are no revenues before the completion of the project, the repayments might be delayed until the project completion. Consequently, in the absence of any other bankruptcy condition, the project will not be bankrupted in the construction phase. Nevertheless, in large projects with long construction periods, lenders may impose other conditions to trigger the bankruptcy during the construction period (e.g. the upper limit of cost overrun). If the condition is triggered, unless the developer can justify the cost overrun or arrange other funding sources (e.g. government rescues or new equity injection), the project will be bankrupted. For example, in credit agreement of Channel Tunnel project the project value (R) and the debt value (D) were estimated by lenders to ensure that the project is greater than the debt value in any time (t) (Ho, 2001). The project value represents the total discounted amount of revenues in the operation phase of the project, whereas the debt value reflects the summation of discounted construction, maintenance and rehabilitation (M&R) costs at the end of construction period (Figure 7).
17
Figure 7. The project cash flow from the construction to the end of concession period. Based on a discrete binomial option pricing model (Cox et al. 1979), Ho & Liu, 2002 suggested an option valuation framework for two risk variables. The first node of the binomial pyramid represents the current state regarding project value (R) and debt value (D). After a time increment, four nodes will be generated from each node in the previous layer. The new values regarding R and D at each node should be obtained by applying the jump amplitudes as shown in Figure 8. The jump amplitudes regarding project and debt values should be obtained by applying a fixed pseudo probability, p = 0.5 (Hull, 1997) (Equations 15-16).
= uR exp[(r − 0,5σ R 2 )∆t + σ R ∆t ]
[15]
= d R exp[(r − 0,5σ R 2 )∆t − σ R ∆t ]
[16]
= uC exp[(r − 0,5σ C 2 )∆t + σ C ∆t ]
[17]
= dC exp[(r − 0,5σ C 2 )∆t − σ C ∆t ]
[18]
where σ R and σ C are assumed to be the volatilities regarding revenues and construction costs.
18
Figure 8. A two-step Binomial pyramid for calculating the project value The total discounted costs and revenues at the first node can be obtained by Equation 19-21. = C0
tE
T
∑ (e( gc −r )i ⋅ CC) + ∑ (e( gc −r )i ⋅ MC)
[19]
=i t0=i T
= R0
tE
∑ e −(WACC − g
R
)i
⋅ RI
[20]
i =T
WACC = (
E D × Re ) + ( × Rd ) × (1 − Tr ) E+D E+D
[21]
where CC and MC are the yearly construction and M&R costs based on the prices at the first construction year; T is the construction time; RI is the expected revenue at the first year; gC and gR are the yearly growth rates regarding costs and revenues; r is the risk free rate of return; E is the market value of firm’s equity; D is the market value of the firm’s debt; Re is the cost of equity; Rd is the cost of debt; Tr is the tax rate and WACC is the weighted average cost of capital.
19
For projects’ financing schemes, in which the developers do not take part in the project finance, WACC should be assumed as zero. The option payoff represents the project value at the time of exercising the option. In traditional contracts the exercise of the option is generally equivalent to the project completion. However, in Design-build-finance-operate projects the option exercise could also be equivalent to bankruptcy of the PPP firm. The equity payoff of the firm is dependent on the project value (R) and the debt value (D). As long as the project value is greater than the debt value the payoff will be calculated by deducting the debt value from the project value. However, if the project value is lower than the debt value, the firm is considered to be bankrupted. Consequently, due to the limited liability of the equity holder, the equity payoff will be zero. Therefore, the equity payoff of the PPP firm is identical to the payoff from an American call option (Ho & Li, 2002). Under this bankruptcy threshold, the payoff upon the project completion can be obtained by Equation 22. It should be noted that the debt value is a function of construction, maintenance and operation (M&O) costs. The payoff upon the project completion for design-bid-build and design-build contracts can be simply obtained by Equation 23. RT ,i − CT ,i ⋅ e rL .T
if
RT ,i − CT ,i ⋅ e rL ⋅T ≥ 0
FT,i =
[22] if
0
RT ,i − CT ,i ⋅ e rL ⋅T < 0
FT,i = RT ,i − CT ,i ⋅ e rL .T
[23]
where RT,i and CT,i are the project value (e.g. discounted revenues) and total cost (i.e. total costs regarding construction and M&O) at the ith node on the last binomial layer and rL is the loan interest rate. The payoff ( Ft-∆t, i ) on the nodes in the previous layers can be found by backward calculations according to Equation 24. Fb
if
Rt-∆t, i - Ct-∆t, i ⋅ e −∆t ⋅rL ≥ 0
Ft-∆t, i =
[24] 0
Fb =
1 4e r ∆t
if
Rt-∆t, i - Ct-∆t, i ⋅ e −∆t ⋅rL < 0
{(1+ρ)×[Ft, i +Ft, i+3 ]+(1-ρ)×[Ft, i+1+Ft, i+2 ]}
[25]
where ρ is the correlation coefficient between the risk variables (i.e. costs and revenues). While a positive value for ρ indicates that there is a positive correlation between the revenues and costs, a zero value would indicate that they are completely independent. The back calculation should continue to find the payoff on the first node of the pyramid which is equivalent to the equity payoff of the firm. Depending on the type of contract and the degree of governmental supports in adverse circumstances Equation 24 should be modified. For 20
example, if it is costly for the government to replace the developer, even in adverse conditions it might rescue the project. Therefore, it can be assumed that the only possible time for the project bankruptcy is at the end of construction time. Consequently, the payoff on the nodes for the layers before T, can be obtained by back-calculating Fb according to Equation 25. Price volatility of construction material and supplies such as bitumen and fuel can result in substantial difficulties for contractors in preparing realistic bids. Often, bidders cannot receive firm price quotes from material suppliers for the project duration. This may lead to price speculations and inflated bid prices due to possible price increases. Price adjustment clauses (PACs) are contractual mechanisms that allow contractors to be at least partially protected against material or fuel price increases that may occur between the contract award and the execution of the work. Moreover, they should be based on an index which is not prone to manipulations by contractors or material suppliers. It was shown that more than 50% of the total cost of asphalt pavements is related to the cost of oil products (Mirzadeh et al. 2013). Additionally, the fluctuation related to oil price has been much more than the general inflation. Consequently, high volatility regarding the energy price has imposed tremendous financial risk to highway projects. This financial risk is shared between contractors and transportation agencies. The impact of this financial risk for each party depends on the level of price adjustment clauses offered by the transport administration. Determination of the value of these price adjustment clauses and their impact on the project financial viability is of great interest for developers, contractors, transport administrations and ultimately tax payers. In order to evaluate financial viability of a project under price adjustment clauses, the corresponding volatility of the project cost needs to be determined. This volatility should reflect the cost fluctuations regarding energy, labor and equipment. Based on the modern portfolio theory introduced by Markowitz (1952) the equivalent rate of return and the volatility of a combined investment can be derived. Moreover, it was shown by Mirzadeh et al. (2013) that oil price inflation jc can be used as the inflation index for the energy related items. Consequently, oil price volatility can be used as the volatility regarding the energy items. Therefore, the equivalent inflation and the volatility regarding the cost of highway projects are calculated according to Equations (26-28). j p = α ⋅ j c + (1 − α ) ⋅ j = σ
β=
[26]
α 2σ o2 + (1 − α ) 2 σ le2 + 2 βα (1 − α )σ oσ le
[27]
Cov (J C ,CPI) Var (CPI)
[28]
where α is the portion of energy related items; jp is the equivalent inflation index; σo is oil price volatility; σle is volatility regarding the cost of labor and equipment based on CPI; β is the correlation coefficient and j is the inflation based on CPI. In the presence of PACs for bitumen or fuel, α should be limited to the portion of energy related costs which are not subjected to PACs. The jump amplitudes regarding debt value
21
(Equations 17-18) should be adjusted replacing σo by the equivalent volatility (σ) which is calculated from Equation (27). The construction cost is obtained by Equation 29.
Ct =
FC
if
(1 − z ) ⋅ FC ≥ Ct
Ct
if
(1 − z ) ⋅ FC ≤ Ct ≤ (1 + z ) ⋅ FC
FC
if
(1 + z ) ⋅ FC < Ct
[29]
Price adjustment clauses usually include a ceiling on upward and downward cost changes. They should be responsive to significant changes in the selected index rather than being triggered by insignificant price fluctuations. The equivalent inflation index (Equation 26) represents the fluctuation regarding the cost of highway projects. Furthermore, it is not susceptible to manipulations by contractors and suppliers. Therefore, it can be used as the index regarding application of PACs. The ceiling for the upward cost changes is similar to a call option in which the contractor has a right, to claim a subsidy from the government (Figure 9a). Moreover, the floor option on the downward movements is equivalent to a put option in which the government has a right, but not an obligation to receive the cost differences. Road agencies in PPP projects usually provide some commitments to support the developers against risks involved in future revenues, material/energy prices and currency exchange rate. It is essential for both developers and the government to know the value of these guarantees and their impact on the project value. The uncertainty over the future levels of demand for traffic on the completed road is one of the greatest sources of risk in PPPs, especially toll road projects (Brandao & Saraiva, 2008). A revenue guarantee is a contract in which the government promises to annually pay the revenue shortfall accumulated in the project financial auditing intervals (Chiara & Kokkaew, 2013). The fair price of such contract is the value of n European put option with maturity time at the end of each time step (Dailami et al. 1999). Consequently, the project value ( Rgt ) under the government minimum revenue guarantee (MRG) at each time step can be obtained by Equation 30. In this case Rt in Equations 22-24 should be replaced by Rgt . Rvt + X t (GR − Rvt )
if
GR − Rvt ≥ 0
Rgt =
[30] Rvt
if
GR − Rvt < 0
where Rvt is the total summation of discounted revenues at the time t; X t is the percentage of the revenue shortfall which is to be compensated by the government and GR is the guaranteed level of the revenue. In order to provide a right for the government to claim a portion of excess revenue due to a higher actual traffic demand, MRG options may be combined with traffic revenue cap (TRC). In this case, the developers share the excess revenue with the government once the revenue exceeds a pre-specified ceiling level (MandriPerrott, 2006). Thus, the combined impact of MRG and TRC contracts should be considered by application of Equation 31. The combination of MRG and TRC options is equivalent to a collar option in which the MRG and TRC act as put and call options.
22
Rgt =
Rvt + X (GR − Rvt )
if
GR ≥ Rvt
Rvt
if
GR < Rvt ≤ (1 + K ) × GR
Rvt − [ Rvt − (1 + K ) × GR ] × H
if
[31]
Rvt > (1 + K ) × GR
where K is the maximum portion of revenue above GR that developers can claim entirely and H is the portion of the revenue above (1 + K ) × GR that developers can entirely claim.
Figure 9. (a) The payoff regarding a PAC; (b) the payoff regarding government revenue guarantee The put option provides the developers a protection against lower revenue. However, the call option gives the government a right to collect a portion of revenue beyond a certain level. Line á-b-c-ć in Figure 9b denotes the payoff for the collar option (i.e. the combined MRG and TRC options).
23
24
4. Results and Discussions Regarding the Appended Papers
4.1. LCA Framework regarding Asphalt Pavements (Paper I) In Paper I, a general framework regarding LCA of asphalt pavements was suggested. In this paper, a model was developed to calculate the amount of expended and feedstock energy regarding asphalt pavement projects. The asphalt production was shown to be the most energy-consuming process, both regarding the electricity and the fuel consumption. High temperatures are usually required for drying aggregates, melting bitumen and additives for the mixing and the storage of asphalt mixtures. The second highest energy intensive process was material transportation. According to the sensitivity analysis, change in the transport distances highly affects the energy consumption of the system. The asphalt mixture consists of almost 95% by weight aggregate which means that it is highly favorable having the aggregate quarry site and the asphalt plant close to each other. Otherwise, the transportation of aggregates to the asphalt plant could be subjected to large amount of energy use and fuel combustion emissions. Furthermore, it was observed that the energy production method had a large impact on the results. Therefore, it should be considered in LCA studies regarding asphalt pavements. Since the asphalt production is a highly energy-consuming process the additives’ potential to decrease expended energy could be extremely beneficial.
4.2. LCCA Based on the Fundamental Cost Contributors (Paper II) Paper II presents a transparent framework for asphalt pavement LCCA by applying energy and time as a basis for calculations which underlined main financial risks pertaining to asphalt paving works. By applying the energy and time units as basis for cost calculations, the framework imposed separate inflation indices on the energy and the labor/equipment related costs. The unit price of different asphalt mixtures are calculated based on the corresponding distances regarding the case study. The results were compared with the average bidding prices for the awarded contracts (STA, 2010) (Figure 10). It was observed that for most of the mixtures, the calculated costs well reflected the prices regarding the laid asphalt. However, the price regarding ABT11 was much higher than the calculated price. This is believed to be related to high amount of ABT11 used in filling the potholes which due to the small amount of work resulted in higher unit prices. The effect of transportation distances on the calculated costs was studied in a sensitivity analysis.
25
Figure 10. Average bidding prices of laid asphalt for awarded contracts (STA, 2010) versus calculated cost with the suggested model. It was observed that more than 50% of the total cost regarding construction and rehabilitation of asphalt pavements were related to energy (Figure 11). A sensitivity analysis was conducted in order to study the effect of transportation distances on the total cost of the laid asphalt. It was observed that by tripling the transportation distances, the calculated cost of laid asphalt for all the studied mixes increased by 35%. Thus, the transportation distance has an important effect on the total cost of laid asphalt. The contribution of the cost drivers had also changed by increasing the transport distances. Due to the increase in transportation distances, the cost contributions related to labor/equipment and diesel have increased. Consequently, contributions of other cost drivers have decreased. However, the contribution of the energy sources which closely follow the oil price fluctuation was still more than 50% of the total cost for all the analyzed mixtures.
26
Figure 11. Contribution of the different cost drivers in total cost of the laid asphalt for different asphalt mixtures.
4.3. Accommodating energy price volatility in LCCA of asphalt pavements (Paper III) The objective of the paper was to present a method to determine discount rate for asphalt pavements which can reflect the financial risk regarding cost escalations for construction, maintenance and rehabilitation activities. The suggested discount rate was defined as a function of interest rate, general inflation, oil price inflation and the portion of energy related costs during the life time of the pavement. Furthermore, it can be applied for cost evaluations regarding both public and PPP projects.
Figure 12. The average equivalent discount rate based on the historical data for Sweden. The variation of discount rates for different portion of energy related costs ( α ) based on the historical data regarding interest rate, inflation and oil price inflation for Sweden is presented in Figure 12. Volatilities regarding equivalent discount rates have been higher than that of traditional discount rate. It was observed that the portion of energy related cost had a great impact on the volatility regarding equivalent discount rate. It was observed that during 1981-1994 the Swedish economy faced a relatively high interest rate, inflation and an overall deflation in the oil price. This situation caused the calculated equivalent discount rates to be higher than the traditional discount rate (Figure 13) and β in this period was 0.2. Beta coefficient in this period was positive which means that cost of asphalt pavements followed the general inflations trend. However, during 1970-1980 and 1999-2012 the high oil price inflation caused the equivalent discount rate to be negative and β coefficients were -2.3 and -3.7. The negative β coefficients indicate that the cost of asphalt pavements did not 27
follow the general inflation trend and implied a rapid increase in the cost of asphalt pavements’ construction and rehabilitation.
Figure 13. The average equivalent discount rates in different time periods. The cumulative probability distribution functions for discount rates with α equal to 0% and 50% are plotted in Figure 14. It was observed that the common range, that is widely used in sensitivity analysis regarding discount rate for infrastructure projects (i.e. 0-8%), is relevant for traditional discount rate ( α =0). However, the same probability requires a range between -20% and 30% for α equal to 50%.
Figure 14. Normal cumulative distributions for discount rates with α equal to 0% and 50% for Sweden during 1970-2012. To hedge against this risk, especially for contracts without price adjustment clauses, contractors usually have incorporated premiums in bid prices (Damnjanovic & Zhou, 2009). Contractors can allocate this risk in four following ways: (1) adding the risk to the price of 28
individual items in the contract, (2) adding the risk to the overall estimated price of the project, (3) dividing the risk and spreading it on some particular items in the bid and (4) spreading the risk on a portfolio of projects. Contractors in projects with longer responsibility periods are more concerned about the future prices. Therefore, they usually get engaged in hedging against commodity (material/fuel) risk with suppliers. The contractor risk premium, especially for smaller contractors, might be larger than the suppliers’ premium. In this case the transport administration could consider including the price adjustment clauses to cover the risk of the future price increases. It can be done by hedging with purchasing options for the future consumption of the fuel and material at the predetermined fixed prices. Furthermore, this will encourage more contractors to take part in the bidding and consequently could result in overall lower bid prices. This information can be used by the road authorities in the context of managing risks on an individual project level as well as the network level. At the project level results provide a basis for making decisions to either retain or transfer the commodity (material/fuel) price risk. For example having high values regarding α and energy price volatility the road administration should consider offering some level of price adjustment clauses. On the network level, the results provide a measure for minimizing portfolio risks by diversifying risk retention and transfer policies. For instance, by implementing price adjustment clauses for a portion of the projects, the portfolio risk can be minimized.
4.4. Evaluation of Highway Projects Based on an Option Pricing Framework (Paper IV) The project value under different level of government revenue and PACs is studied from contractors’ and the transport administration’s perspectives. Furthermore, the amount of financial risk borne by the road agency under the government guarantees and PACs was evaluated. Although in this Thesis the government revenue guarantees and PACs were analyzed, the model is flexible and can be easily extended to include other forms of guarantees regarding loan, exchange rate and equity. The discounted construction cost (C0), assuming 3% yearly cost inflation, is calculated as 443 MEURO. The portion of energy related items for this project was assumed to be 60%. Moreover, the volatility regarding the oil price during 1970-2013 was about 25%. By assuming 5% volatility regarding the cost of labor and equipment and a beta equal to zero, the project cost volatility was 15% according to Equation 27. Furthermore, the jump amplitudes for the binomial pyramid were calculated as 1.08 and 0.92. Consequently, the project equity value based on the suggested option pricing framework was F1, 1 = 93.5 MEURO. Since the calculated equity value was lower than the invested equity, the investment in the project in the absence of PACs or government guarantees was not feasible from the shareholders’ perspective. However, the project NPV 23.9 MEURO which suggests that project based on a DCF analysis was 700 ⋅ e−WACC ⋅3 - 450 = should be undertaken. The main problem with this result is that the DCF method neglects the uncertainty regarding the project costs and revenues. As discussed previously, the high volatility regarding the price of oil products is the main source of uncertainty for the project
29
costs. The impact of the energy price volatility on the project equity value was shown to be substantial (Figure 15).
Figure 15. The impact of energy price volatility and the portion of energy related costs (α) on the project value The energy price volatility imposes high financial risk to infrastructure projects. The risk is more substantial for projects with longer contract duration. Nevertheless, it can also be significant for traditional public contracts with longer construction period. For example, many contractors in the U.S were affected by sharp increases in the price of oil products such as fuel and bitumen (Gallagher & Riggs, 2006). Consequently, in order to compensate the contractors for their losses, 40 states used fuel PACs and 42 states implemented bitumen PACs (ASHTO, 2009). Different trigger values have been applied in each state (e.g. the Washington State Department of Transportation (DOT) applied the trigger value of 10% for fuel cost adjustments). However, the trigger levels were determined arbitrarily and were not based on a risk analysis. In Sweden price fluctuations regarding bitumen are entirely reimbursed by the Swedish Transport Administration. PACs are believed to have positive effects such as increasing the number of bidders, decreasing the bid prices and the number of bid withdrawals, enhancing both market stability and reliability in the supply chain and increasing the transparency regarding the contractors profit margin (Newcomb, Lenz & Epps, 2013). On the contrary, they may impose high financial risks to the transport administration. The value of the PACs under different threshold is presented in Figure 16a. By raising the uncertainty regarding the energy price, which was shown to be the main deriver of asphalt pavement costs, the value of PAC significantly increased. Similarly, the value of revenue guarantees increased for a higher level of uncertainty regarding the revenues (Figure 16b). However, offering government supports as collar options balanced
30
the risks and benefits for the two parties and lowered the value of the government support from the transport administration perspective.
Figure 16. (a) Evaluation of PACs with different trigger value (b) Evaluation of revenue guarantees The impact of different PACs on the project cost histogram for the contractors is shown in Figure 17 in which the cost histograms are fitted with lognormal distributions. Application of collar PACs was shown to have a substantial impact on limiting the amount of financial risk regarding the escalation of material/fuel prices for the contractors. Moreover, decreasing the trigger value for the PAC limited the amount of the financial risk. Nevertheless, by applying PACs the transport administration has transferred a substantial amount of financial risk to itself. This could result in a total project cost which is much higher than the anticipated cost by the agency.
31
Figure 17. The impact of PACs with different trigger values on the project cost histogram (which is fitted with a log-normal distribution) for the contractors. In the absence of PACs, increases in material or energy prices could result in extensive losses for contractors. Contractors in projects without PACs, usually include a premium in the bid price to account for the financial risk regarding the material and fuel price inflations (Damnjanovic & Zhou, 2009). They usually allocate this risk in different manners such as adding the risk to prices of certain or all of the contract items or spreading the risk on a portfolio of projects. Moreover, contractors in larger projects sometimes get involved in long term material/fuel contracts with suppliers. The values of these agreements are identical to those of PACs which can be analyzed by the suggested model.
4.5. A Life Cycle Cost Approach on Minimization of Roughness-Related Damages on Flexible Pavements (Paper V)
The effect of integrating the pavement surface deterioration on pavement service life is depicted in Figure 18. Four different scenarios (i.e. four different rate of change of DLC) for pavement surface deterioration have been considered. In Figure 18 each colored line corresponds to one scenario with the input data documented in the legend. The dotted grey line corresponds to the case in which no dynamic loads have been taken into account. The pavement structure fails when the number of ESALs on the road exceeds the allowed number of axles. Therefore, in Figure 18 whenever the colored lines (i.e. allowed number of axles) intersect with the dashed line, the pavement service life is terminated. The yearly growth of DLC results in higher magnitude of traffic loads exerted on the pavement. Thus, the amount of horizontal tensile strain at the bottom AC layer, as well as vertical compressive strain on top of the subgrade increases dramatically. This results in less allowed number of standard axles on the pavement. One may see in Figure 18 that the additional dynamic loads can decrease the predicted pavement service life up to 10 years.
32
Figure 18 : Service life reductions as a result of applying the additional dynamic loads It was observed that having a medium to high DLC, the total energy (i.e. the sum of the expended and feedstock energy) was lower for the alternative with the application of the preventive maintenance (Figure 18a). Similarly, the alternatives with application of preventive maintenance had a lower relative LCC. However, the implementation of the preventive maintenance was more economically efficient for the cases with higher traffic level (Figure 18b). It was observed that surface roughness increased during the life span of asphalt pavements. Moreover, increases in the surface roughness resulted in higher DLCs which decreased the pavement life. Thus, implementation of preventive maintenance activities by reducing the surface roughness can moderate the impact of dynamic loads induced from road-vehicle interactions. Furthermore, reduction in the surface roughness can also decrease vehicles fuel consumption. Construction and rehabilitation of roads are energy intensive processes. Additionally, high volatility regarding the energy price has imposed high financial risks to asphalt pavement projects. Therefore, application of the preventive maintenance by increasing the pavement service life and decreasing the portion of energy related costs have lowered the amount of financial risk regarding the cost asphalt pavements (Figure 19). The application of a 20 mm thin asphalt layer had a lower amount of energy related costs compare to a more traditional thicker rehabilitation. Therefore, the application of the preventive maintenance reduced the portion of energy related costs by 6-7% for all the cases.
33
Figure 18: (a) The amount of energy savings for different DLC rate of change; (b) The costs for alternatives with preventive maintenance (Cp) in relation to alternatives without preventive maintenance (C0). Furthermore, the amount of fuel consumption by the vehicles is directly affected by the road surface roughness. The vehicle fuel consumption is a function of rolling resistance of the vehicle tires. The rolling resistance forces are highly influenced by the pavement surface conditions. Therefore, the pavement surface deterioration during its life span increases rolling resistance forces which results in higher vehicle fuel consumption. In the calibrated HDM 4 model (Zaabar and Chatti, 2010) the fuel consumption is directly related to IRI. The IRI value can be calculated directly from a given displacement PSD (Johannesson and Rychlik, 2012; Sun et al., 2001).
34
Figure 19. Asphalt pavement cost profile regarding alternatives with and without preventive maintenance for the studied cases.
In order to develop better solutions, increase the competition and to transfer the responsibility to the party best placed to minimize the risks, the STA has implemented contracts with functional specifications. Contractors in design-bid-build projects are responsible for a certain guarantee period after the project completion. Currently, 34% of road projects in Sweden are design-bid-build and this portion is to be increased to 50% by 2018.
In traditional bid-build contracts the lower LCC due to application of the preventive maintenance can be of interest for transport administrations. However, in the design-bidbuild contracts the application of the preventive maintenance can be of a great interest from the transport administration and contractors’ perspectives. By lowering the total cost of the project this can help the contractors to offer lower bids in order to win more contracts. Moreover, the application of the preventive maintenance can lower the contractor’s risk profile.
35
36
5. Summary and Conclusions This Thesis focused on financial evaluation of asphalt pavement projects from the transport administration and contractors’ perspectives. A transparent LCC framework was presented in which energy and time units were used as the basis for calculations. Comparison between the framework results and the average price of the laid asphalt for different Swedish asphalt mixtures showed that the total calculated costs well reflected the actual purchase price of the corresponding asphalt pavement. The sensitivity analysis regarding the transportation distances showed that the transportation distance has a high impact on the total cost of asphalt pavements. Therefore, it is believed that the transportation distance is one of the most important factors regarding the high variation in the price of laid asphalt pavements in Sweden. Based on the values regarding inflation, oil price inflation and interest rate and the portion of energy related costs in the asphalt pavement project an approach was suggested for calculating the discount rate. The lower discount rate, especially the negative discount rate in the recent years, implies that the future costs can be more expensive than projected due to discounting. This highlights the importance of lowering the amount of required maintenance and rehabilitation by increasing the initial construction quality. It was observed that the common range (i.e. 0-8%) for discount rate sensitivity analysis regarding infrastructure projects is not suitable for highway projects. By assuming 50% of the costs related to the energy items, the discount rate sensitivity analysis should cover a range between -20% and +30%. The high amount of volatility in the energy price imposed a significant financial risk to highway projects. However, depending on the level of PACs, this financial risk is shared between the transport administration and contractors. Form the contractors’ perspective this financial risk is much higher in fixed-price unit-based contracts with longer responsibility periods. Different features included in highway contracts such as limited liability of the PPP firm, government support mechanisms (e.g. PACs and revenue guarantees) result in asymmetric project value payoff functions which cannot be properly assessed by traditional methods such as DCF. To overcome this issue, an option pricing framework is developed which enables financial assessment of different types of road projects in the presence of different government support mechanisms. It was observed that energy price fluctuation has been the main contributor regarding the significant changes in the cost of highway projects. Therefore, the volatility regarding the cost of highway projects was calculated as a function of energy price volatility and the portion of energy related costs. Furthermore, an equivalent inflation index which is not prone to manipulations by either contractors or material suppliers was suggested as the index regarding the application of PACs. From the tax payers perspective the application of PACs has some advantages and drawbacks. On the one hand, application of PACs reduces the financial risk for contractors. This could especially benefit smaller contractors that have larger premiums compare to suppliers and encourage them to take part in the bidding. Consequently, increases in the number of participating contractors in the bidding can potentially decrease bid prices. On the other hand, a financial risk which is substantial for large projects will be transferred to the transport administration and ultimately to the tax payers.
37
Decreasing the overall contractors’ bid prices and lowering the financial and technical risks imposed to the Swedish Transport Administration are two key strategies to achieve more value of money for roads. An overall decrease in bid prices can be achieved either by lowering the risks involved in the contract or by applying more effective hedging strategies. As a general principle, a systematic risk should be borne by the party which either is best placed to manage it or has the possibility to minimize its impacts. Contractors and suppliers in large projects sometimes are the best parties to manage the financial and technical risk. They might have the possibility to depot the construction supply at the beginning of the project and by doing so they can hedge against the financial risk regarding the fuel/material cost escalations. Moreover, contractors may have the possibility to lower the financial and technical risks by implementing better solutions. For instance, application of preventive maintenance activities (i.e. thin asphalt layer) by smoothing the surface can lower the exerted dynamic loads and hence increase the pavement life span and decrease the technical/financial risks.
38
6. Recommendations for Further Research This PhD Thesis evaluated different risks involved in road projects (e.g. financial, technical) from the transport administration and contractors’ perspectives. The impact of the risks on each party depends on the type of contract and the level of governmental supports offered by the transport administration. So far, the impacts from the financial and technical risks have been assessed separately. However, financial and technical risks in highway projects are usually correlated. This suggests that the technical risk in an asphalt pavement project is more severe in the presence of high financial risk regarding material/fuel cost escalations and vice versa. Recommendations for future works are listed in the following: • Valuation of different highway contracts under the presence of financial and technical risks based on a Monte Carlo option pricing model. • Extending the suggested option pricing framework to enable valuation of other forms of guarantees in highway projects regarding loan, exchange rate and equity.
39
References Amini, A.A., Mashayekhi, M., Ziari, H., & Nobakht, S. (2012). Life cycle cost comparison of highways with perpetual and conventional pavements. International Journal of Pavement Engineering, 13, 553–568. Ashuri, B., Kashani, H., Molenaar, K. R., Lee, S., & Lu, J. (2012). Risk-neutral pricing approach for evaluating BOT highway projects with government minimum revenue guarantee options. Journal of Construction Engineering and Management, 138(4), 545-557. Athena Institute (2006). A life cycle perspective on concrete and asphalt roadways: Embodied primary energy and global warming potential. Ottawa, ON. Azar, S. A. (2007). Measuring the US social discount rate. Applied Financial Economics Letters, 3(1), 63-66. Baker, H. K., Dutta, S., & Saadi, S. (2011). Management views on real options in capital budgeting. Journal of applied Finance, 21(1), 18-29. Baker, R., & Fox, R. (2003). Capital investment appraisal: a new risk premium model. International Transactions in Operational Research, 10(2), 115-126. Baloi, D., & Price, A. D. (2003). Modelling global risk factors affecting construction cost performance. International Journal of Project Management, 21(4), 261-269. Black, F., & Scholes, M. (1973). The pricing of options and corporate liabilities. The journal of political economy, 637-654. Block, S. (2007). Are “real options” actually used in the real world?. The Engineering Economist, 52(3), 255-267. Boyle, P. P. (1977). Options: A Monte-Carlo approach. Journal of financial economics, 4(3), 323-338. Butt, A., Mirzadeh, I., Toller, S., & Birgisson, B. (2014). Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives, International Journal of Pavement Engineering, 15:4, 290-302. Christensen, P. (2012). Infrastructures and life-cycle cost–benefit analysis. Structure and Infrastructure Engineering, 8, 507–516. Chan, A., Keoleian, G., & Gabler, E. (2008). Evaluation of lifecycle cost analysis practices used by the Michigan Department of Transportation. Journal of Transportation Engineering, 134, 236–245. Cheah, C. Y., & Liu, J. (2006). Valuing governmental support in infrastructure projects as real options using Monte Carlo simulation. Construction Management and Economics, 24(5), 545554.
Chiara, N., Garvin, M. J., & Vecer, J. (2007). Valuing simple multiple-exercise real options in infrastructure projects. Journal of Infrastructure Systems, 13(2), 97-104. Chiara, N., & Kokkaew, N. (2009). Risk analysis of contractual flexibility in BOT negotiations: a quantitative approach using risk flexibility theory. International Journal of Engineering and Management, 1(1), 71-79. Cox, J. C., Ross, S. A., & Rubinstein, M. (1979). Option pricing: A simplified approach. Journal of financial Economics, 7(3), 229-263. De Neufville, R., Scholtes, S., & Wang, T. (2006). Real options by spreadsheet: parking garage case example. Journal of infrastructure systems, 12(2), 107-111. Eisenberger, I., & Remer, D. (1977). The role of interest and inflation rates in life-cycle cost analysis (NASA Deep Space Network Progress Rep. No. 42–43, pp. 105–109). Energy Conservation in Road Pavement Design (ECRPD) (2009). Maintenance and utilization WP6 – Life cycle evaluation. Centrum dopravniho vyzkumu, Lisenska, Czech Republic. Evans, D. J., & Sezer, H. (2005). Social discount rates for member countries of the European Union. Journal of Economic Studies, 32(1), 47-59. Fama, E. F., & French, K. R. (2004). The capital asset pricing model: theory and evidence. Journal of Economic Perspectives, 25-46. Federal Highway Administration (FHWA) (2002). Life-cycle cost analysis primer. Washington, DC. Federal Highway Administration (FHWA) (2003). Economic analysis primer. Washington, DC. Ferreira, A., & Santos, J. (2013). Life-cycle cost analysis system for pavement management at project level: Sensitivity analysis to the discount rate. International Journal of Pavement Engineering. doi:10.1080/10298436.2012.719618. Frangopol, D., & Liu, M. (2007). Maintenance and management of civil infrastructure based on condition, safety, optimization, and life-cycle cost. Structure and Infrastructure Engineering, 3, 29–41. Galera, A. L. L., & Solino, A. S. (2010). A real options approach for the valuation of highway concessions. Transportation Science, 44(3), 416-427. Gallagher, J. and Riggs, F. (2006) Material price escalation: allocating the risks. Construction Briefings, Thomson/West, Eagan. Gullberg, D., Birgisson, B., & Jelagin, D. (2012). Evaluation of a novel calibrated-mechanistic model to design against fracture under Swedish conditions. Road Materials and Pavement Design, 13(1), 49-66.
Hastak, M., & Shaked, A. (2000). ICRAM-1: Model for international construction risk assessment. Journal of Management in Engineering, 16(1), 59-69. Ho, S. P., & Liu, L. Y. (2002). An option pricing-based model for evaluating the financial viability of privatized infrastructure projects. Construction Management & Economics, 20(2), 143-156. Hull, J. (2009). Options, futures and other derivatives. Pearson education. ISO15686-5 (2008). Building and constructed assets service life planning, Part 5: Life-cycle costing. Stockholm: Swedish Standard Institute. Jaafari, A. (2001). Management of risks, uncertainties and opportunities on projects: time for a fundamental shift. International journal of project management, 19(2), 89-101. Lally, M. (2008). Measuring the US social discount rate: reply to Azar. Applied Financial Economics Letters, 4(4), 283-285. Lee, H., Lee, S., Lee, J., Chon, k. (2009). The Revenue-Beta Model for Private Participation in Infrastructure. Journal of the Eastern Asia Society for Transportation Studies, 8. Majd, S., & Pindyck, R. S. (1987). Time to build, option value, and investment decisions. Journal of financial Economics, 18(1), 7-27. Mandapaka, V., Basheer, I., Sahasi, K., Ullidtz, P., Harvey, J., & Sivaneswaran, N. (2012). Mechanistic-empirical and life cycle cost analysis for optimizing flexible pavement maintenance and rehabilitation. Journal of Transportation Engineering, 138, 625–633. Markowitz, H. (1952). Harry M. Markowitz. Portfolio selection, Journal of Finance, 7(1), 7791. Martins, J., Marques, R. C., & Cruz, C. O. (2013). Real options in infrastructure: revisiting the literature. Journal of Infrastructure Systems. doi: 10.1061/(ASCE)IS.1943-555X.0000188. Michigan Department of Transportation (MDOT) (2012). Pavement design and selection. Lansing, MI. Mirzadeh, I., Butt, A., Toller, S., & Birgisson, B. (2013). Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements, Structure and Infrastructure Engineering. doi:10.1080/15732479.2013.837494. Mirzadeh, I., & Birgisson, B. (2014). Accommodating energy price volatility in life cycle cost analysis of asphalt pavements. Submitted to: Journal of Civil Engineering and Management. Myers, S. C. (1977). Determinants of corporate borrowing. Journal of financial economics, 5(2), 147-175. National Renewable Energy Laboratory (2011). U.S. life cycle inventory database. Golden, CO.
Ng, F. P., Björnsson, H. C., & Chiu, S. S. (2004). Valuing a price cap contract for material procurement as a real option. Construction management and economics, 22(2), 141-150. Nombela, G., & De Rus, G. (2004). Flexible-term contracts for road franchising. Transportation Research Part A: Policy and Practice, 38(3), 163-179. Pandit, D., Anderson, S., Damnjanovic, I., & Nejat, A. (2009). Assessment of strategies and methods to reduce construction costs while maintaining Quality. In Proceedings of 1st International Conference on Transportation Construction Management, Orlando, FL. Percoco, M. (2008). A social discount rate for Italy. Applied Economics Letters, 15(1), 73-77. Quiggin, J. (2005). Public–private partnerships: options for improved risk allocation. Australian Economic Review, 38(4), 445-450. Safi, M., Sundquist, H., Karoumi, R., & Racutanu, G. (2013). Development of the Swedish bridge management system by upgrading and expanding the use of LCC. Structure and Infrastructure Engineering, 9(12), 1240-1250. Santos, J., & Ferreira, A. (2013). Life-cycle cost analysis system for pavement management at project level. International Journal of Pavement Engineering, 14, 71–84. Stripple, H. (2001). Life cycle assessment of road, A pilot study for inventory analysis. Goteborg: IVL Swedish Environmental Research Institute. Strong, R. A. (2008). Portfolio Construction, Management, and Protection [With Stock-Trak]. CengageBrain. com. Swedish National Audit Office (NAO) (2012). Procurement of roads and railways by the Swedish Transport Administration: does it lead to high productivity? RiR 2012:14. Stockholm. Swedish Transport Administration (STA) (2010). Bitumen bound layers. Publication number: 2010:093. Borlange. Swedish Transport Administration (STA) (2012). Impact analysis of rehabilitation and reconstruction of roads. Stockholm. Swiss Centre for Life Cycle Inventories (2011). EcoInvent. Dubendorf. Triantis, A. J. (2003). Real options. Handbook of modern finance, Logue, D. and. Seward, J., New York, Research Institute of America, pp. D1-D32. Trigeorgis, L. (1996). Real options: Managerial flexibility and strategy in resource allocation. MIT press. UK Department of Energy and Climate Change (DECC) (2012). Industrial energy prices, quarterly energy prices. London.
Wang, J., Birgisson, B., & Roque, R. (2007). Implementation of the Florida Cracking Model into the AASHTO 2002 Mechanistic -Empirical Pavement Design. Journal of Transportation Research Record, 86-96. Walls, J., & Smith, M.R. (1998). Life cycle cost analysis in pavement design. Washington, DC: FHWA. Wilmot, C. G., & Cheng, G. (2003). Estimating future highway construction costs. Journal of Construction Engineering and Management, 129(3), 272-279. Yeo, K., & Qiu, F. (2003). The value of management flexibility: a real option approach to investment evaluation. International Journal of Project Management, 21, 243-250. Yiu, C. Y., & Tam, C. S. (2006). Rational underpricing in bidding strategy: a real options model. Construction Management and Economics, 24(5), 475-484. Zhang, H., Keoleian, G., & Lepech, M. (2008). An integrated life cycle assessment and life cycle analysis model for pavement overlay systems. Paper presented at the first international symposium on life-cycle civil engineering, Varenna, Italy, 907–912. Zhao, T., & Tseng, C. L. (2003). Valuing flexibility in infrastructure expansion. Journal of infrastructure systems, 9(3), 89-97. Zhou, X., & Damnjanovic, I. D. (2011). Optimal Hedging of Commodity Price Risks in Highway Contracts. Transportation Research Record: Journal of the Transportation Research Board, 2228(1), 19-25.
Paper I Butt, A., Mirzadeh, I., Toller, S., & Birgisson, B. (2014) Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives, International Journal of Pavement Engineering, 15:4, 290-302.
This article was downloaded by: [Kungliga Tekniska Hogskola], [Iman Mirzadeh] On: 24 March 2014, At: 06:17 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
International Journal of Pavement Engineering Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/gpav20
Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives a
a
b
a
Ali Azhar Butt , Iman Mirzadeh , Susanna Toller & Björn Birgisson a
Division of Highway and Railway Engineering, KTH – Royal Institute of Technology, SE-10044Stockholm, Sweden b
Division of Environmental Strategies Research (fms), Department of Urban Planning and Environment, KTH – Royal Institute of Technology, SE-10044Stockholm, Sweden Published online: 24 Aug 2012.
To cite this article: Ali Azhar Butt, Iman Mirzadeh, Susanna Toller & Björn Birgisson (2014) Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives, International Journal of Pavement Engineering, 15:4, 290-302, DOI: 10.1080/10298436.2012.718348 To link to this article: http://dx.doi.org/10.1080/10298436.2012.718348
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
International Journal of Pavement Engineering, 2014 Vol. 15, No. 4, 290–302, http://dx.doi.org/10.1080/10298436.2012.718348
Life cycle assessment framework for asphalt pavements: methods to calculate and allocate energy of binder and additives Ali Azhar Butta*, Iman Mirzadeha1, Susanna Tollerb2 and Bjo¨rn Birgissona3 a
Division of Highway and Railway Engineering, KTH – Royal Institute of Technology, SE-10044 Stockholm, Sweden; Division of Environmental Strategies Research (fms), Department of Urban Planning and Environment, KTH – Royal Institute of Technology, SE-10044 Stockholm, Sweden
b
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
(Received 22 June 2012; final version received 31 July 2012) The construction, maintenance and disposal of asphalt pavements may lead to considerable environmental impacts, in terms of energy use and emissions during the life of the pavement. In order to enable quantification of the potential environmental impacts due to construction, maintenance and disposal of roads, an open life cycle assessment (LCA) framework for the asphalt pavements is presented in this paper. Emphasis was placed on the calculation and allocation of energy used for binder and additives at the project level. It was concluded from this study that when progressing from LCA to its corresponding life cycle cost, the feedstock energy of the binder becomes highly relevant as the cost of the binder will be reflected in its alternative value as fuel. Regarding additives like wax, a framework for energy allocation was suggested. The suggested project level LCA framework was demonstrated in a limited case study of a Swedish asphalt pavement. It was concluded that the asphalt production and transporting materials were the two most energy-consuming processes, emitting most greenhouse gases depending on the fuel type and electricity mix. Keywords: life cycle assessment; feedstock energy; asphalt binder additives; mass – energy flows
1.
Introduction
Roads are assets to society and an integral component in the development of the transportation infrastructure. According to the Swedish National Road Administration, Sweden has a road network consisting of streets, state and municipal roads and private roads which sums up to be about 0.6 million km. Large amounts of aggregates and asphalt are produced all over the world to fulfil the material requirements for the construction of the roads. To produce asphalt mixtures, considerable amounts of materials including bitumen, aggregates and additives are required. About 3.5 billion tonnes of aggregates are produced annually in Europe (Koziol et al. 2008). In 2009, Sweden, having almost 1.3% of Europe’s population, produced 84.5 million tonnes of aggregates which is almost 2.4% of EU’s production (SGU 2009). USA produced 1.91 billion metric tonnes of aggregate in the year 2010 (Willett 2011). For the construction of the roads, large amount of hot-mix asphalt (HMA) is produced all over the world (Table 1). Energy is needed for the production of the materials and if utilisation of the energy is not properly done, this may lead to more wastes and higher costs. The impact on the environment cannot be neglected either. Therefore, new techniques are being developed to make the production processes efficient with less consumption of energy. Energy savings are driving the development of warm mix asphalt technologies due to both the environmental and
*Corresponding author. Email:
[email protected] q 2012 Taylor & Francis
economic reasons and therefore subject of research (Chowdhury and Butto 2008, D’Angelo et al. 2008, Edwards et al. 2010, Das et al. 2012). Waxes such as Fischer Tropsch (FT) paraffin, Sasobitw and Montan are some of the common ones investigated in different studies in the literature for lowering the mixing temperatures of the asphalt mixtures. However, the energy and resources needed to produce such additives may not be an offset by the lowered processing temperature when used in the asphalt mixtures. It is not yet known whether the use of additives like waxes is lowering the total use of energy and emissions when a life cycle perspective is taken into consideration. Life cycle thinking is becoming popular in different fields of research as it is being recognised that resource depletion and the emissions of different potentially harmful substances are often a result from activities in different life cycle stages of a product’s life. Life cycle assessment (LCA) is a versatile tool to investigate the environmental aspect of a product, a service, a process or an activity by identifying and quantifying the related input and output flows utilised by the system and its delivered functional output in a life cycle perspective (Baumann and Tillman 2003). Ideally, it includes all the processes associated with a product from its ‘cradle-raw material extraction’ to its ‘grave-disposal’. LCA studies emerged in the late twentieth century aimed at optimising energy consumption due to restraints on the excess use of
International Journal of Pavement Engineering Table 1. The amount of HMA production in the EU and other countries in million tonnes (Sivilevicˇius and Sˇukevicˇius 2009).
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
1998– 2007 period Country
Min
Avg
Max
2007
USA Japan Turkey Australia Norway Canada (Ontario) European Union
465 54.9 9.5 6.5 3.9 11 238.8
504 64.27 14.06 7.07 4.67 13.24 284.5
545 71.4 22.2 7.7 5.9 14.5 341.6
500 54.9 22.2 7.7 5.9 13.2 304.1
energy in the industries. Soon raw material consumption studies were taken into account to get more information about the inputs and eventually the development studies started which also took into account the output flows (Ecobilan). In late 1990s and early 2000s, International Organisation for Standardisation (ISO) released ISO 14040 series on LCA, which included the general framework (ISO 14040 1997), goal and scope definition and inventory assessment (ISO 14041 1998), impact assessment (ISO 14042 2000) and interpretation (ISO 14043 2000). This laid the basis for the LCA methodology and is being used by and large in different fields of research. Several researchers have studied the effects on the environment due to the construction, maintenance and disposal of the roads (Stripple 2001, Birgisdo´ttir 2005, Zhang et al. 2008, Huang et al. 2009, Santero et al. 2010a, 2010b). Such research enables effective measures to be identified to reduce the resource use and the environmental loads from the roads, for example by suggesting changes in the technical procedure or choice of the materials. The impact of the traffic is considered to be more significant from an environmental point of view than the construction, operation and maintenance of the road’s lifetime (Stripple 2001). This may be true because of high fuel consumption and emissions from the vehicles, if the use phase is included in the life cycle study of a road. At the network level when the type of infrastructure is being decided, impact from the traffic must be considered. However, if the use phase (impact from the vehicles) is included in a life cycle study at the project level, the impacts due to the other life cycle phases will generally be unnoticed, and thus ignored due to the large impact from the vehicles. For the roads, the large use of aggregates and energy for bitumen production may cause considerable emissions during the production phase of the road’s life cycle. Park et al. (2003) reported that the most energy intensive process in a road’s life cycle is the manufacturing of construction materials, which in their study consumed 1525.8 tonnes of oil equivalents per 1 km and four lane highways. The authors stated that the construction and demolition consume more energy than the maintenance/ repair. This conclusion, however, is probably a result from
291
assuming a relatively low number of maintenance cycles. In the European project named Energy Conservation in Road Pavement Design, Maintenance and Utilisation, it was concluded that the construction of a new road (20 years design life) consumes huge amounts of energy (9384.7– 9986.3 GJ/km for motorways) and 92.4– 92.9% of energy comes from the asphalt production (ECRPD 2009). In the maintenance phase of the motorways, it was reported that by using hot-in-place recycling methods, 27.5 –29.2% of energy could be saved when compared with the hot method of recycling in the asphalt plant. Different tools are being developed to calculate and quantify the environmental and economic burdens from the roads. PaLATE, the Excel-based pavement LCA tool for environmental and economic effects was developed in 2003 by the University of California, Berkeley (Horvath 2004). It can analyse the life cycle stages including the construction and maintenance of the pavement in regard to the environmental and economic aspects. ROAD-RES is another tool developed by Birgisdo´ttir (2005). It has been divided into two main parts: the construction and the disposal. It looks particularly into the residues from waste incineration usage against the virgin materials. A number of other softwares and models have been developed such as Federal Highway Administration’s (FHWA) RealCost, Caltrans’ Cal B/C and FHWA’s IMPACTS, but most look at the life cycle cost (LCC) analysis or neglect the pavements life cycle perspective (Santero et al., 2010b). Several previous LCA’s of the roads have been focused on comparing the asphalt and the concrete pavements (Santero et al. 2010a, 2010b). The results consistently indicate that an asphalt pavement implies a larger use of energy but lower emissions than the concrete pavements. It was reported by Horvath and Hendrickson (1998) that the HMA pavement consumed 40% more energy but produced fewer emissions as compared with a continuous reinforced concrete pavement. The comparison was made with an economic input – output LCA approach, and no feedstock energy was taken into account. However, the results are in line with other LCAs in which feedstock energy is included (Ha¨kkinen and Ma¨kela¨ 1996, Stripple 2001). In Ha¨kkinen and Ma¨kela¨ (1996), a stone mastic asphalt pavement was reported to consume almost twice the non-renewable energy compared with a doweled jointed plain concrete pavement (JPCP), whereas the concrete pavement produced 40– 60% more CO2 emissions depending on the maintenance schedules. Similar results were reported by Stripple (2001) who compared a doweled JPCPs and two asphalt pavements (hot- and cold-mix asphalt). There is potential energy embedded in the bitumen, which may be referred to as feedstock energy, which is not used as an energy resource (ISO 14041 1998). There are authors who include the feedstock energy of the bitumen in the life cycle studies while others do not include without any rationale. In the literature, the procedure to calculate
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
292
A.A. Butt et al.
the feedstock energy of bitumen and the theories behind the decision to include it are seldom explicitly described. So far the most common source being referred regarding the feedstock energy of the bitumen is Garg et al. (2006) and VTT. According to ISO 14040 standards, feedstock energy is required to be reported, but it does not discuss the repercussions of combusting bitumen (Santero 2010). A general approach to calculate feedstock energy in bitumen is therefore missing. What is also missing in the previous life cycle studies is taking into account of the additives that are used widely nowadays to enhance the properties of asphalt mixtures. Most common examples of such materials are polymers and waxes. In this study, a new method for accounting for feedstock energy is presented, as well as a method for accounting for warm asphalt mixture additives and polymers. 1.1 Aim The objective of this study was to enable improvements of the asphalt pavement LCAs by describing methods to consider feedstock energies and warm mixture additives and polymers. A general framework for LCA of the asphalt pavement was suggested, and methods to calculate the feedstock energy and quantify the mass and energy flows of the additives, such as waxes and polymers, were developed and demonstrated. This study does not cover a road network, rather it focuses on an asphalt pavement project only and therefore does not include the traffic.
Figure 1. Processes involved in the construction of the road and selected processes to be considered in the suggested framework.
2. LCA framework 2.1 Conceptual model The life cycle of a road can be divided into several stages: extraction of the raw materials, processing the construction materials, construction, operation, maintenance, demolition, recycling and waste treatment. Figures 1 and 2 show the unit processes to be considered in the road LCAs. Material transport, construction and maintenance equipment, and machinery are present at each unit process in the road system, and each unit process is based on the design considerations. Construction, maintenance and end of life of a flexible pavement (asphalt road) have been considered for the development of the LCA framework. The use phase is, however, important to be considered at the network level when decisions of constructing a road, bridge or a tunnel are being made. On the other hand, if the fuel usage and emissions from the traffic are included in a life cycle study at a project level, it will overshadow all the other phases in a road’s life cycle. The LCA framework developed in this study focuses on a project level; therefore, the use phase was not considered. Processes discussed and used regarding the construction of the asphalt pavement in the suggested LCA framework are (i) processing in the mixing plant,
Figure 2. Processes involved in the maintenance of the road and selected processes to be considered in the suggested framework.
(ii) paving of the asphalt layer, (iii) compaction of the asphalt layer and (iv) transportation of the materials (Figure 1). The impact on the environment from the maintenance of a road depends on the number of maintenance cycles considered. Processes discussed and used regarding the maintenance of the asphalt pavement in the suggested LCA framework are (i) milling of the wearing course and repaving and (ii) rehabilitation (rebuilt) (Figure 2). In a cradle-to-grave LCA, end of life of the system’s function should be considered. Processes discussed and used regarding the end of life of the asphalt pavement in the suggested LCA framework are (i) recycling materials and (ii) burial in the sub-grade.
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
International Journal of Pavement Engineering 2.2 System boundaries Figure 3 shows the framework for the LCA of an asphalt pavement. This study focused on the energy consumption and emissions produced in the asphalt production and handling of the asphalt mixtures and their components. The starting point of the study was the use of the raw materials and the end point was the end of life of the pavement. Quantification of the environmental loads and energy was part of the life cycle procedure. In general, the land area use, ground water level variation, biological resource, etc. were not considered for the development of the framework. This was due to the reason that these complex vast studies were to be looked in detail separately. These research areas may be linked in the future to the framework described in this paper. 2.2.1 Land area Use of the land is an important aspect in the planning phase at the network level. Given the objective of this study the land area by definition becomes a default setting. Scope of this paper was limited to the project level. The road location was pre-defined, i.e. it was already decided where to lay or construct a road. The land area usages for some other purposes like building construction were not considered. 2.2.2
Base layer
Topography of the land is very important when a planner and later the road designer are preparing the layout of a road. As a result of an architectural layout, local geology and ‘cut and fill’ operations, the road thickness changes along the
length of the road. For this study, the base layer, sub-base layer and the drainage were not part of the scope. These are complex areas that must be looked into separately. The thickness of the asphalt layer was considered to remain constant which implies that the load-carrying capacity was constant along the length of the road. 2.2.3 Traffic Traffic plays a vital role in the road engineering as it implies loading in a road design process. Loading and expected number of load cycles are major part of the road design. In the design phase of the road, traffic is automatically incorporated in the system definition in the form of equivalent single axle loads (ESALs). Emissions from traffic were not taken into consideration due to the reasons defined earlier in the introduction and LCA framework section. 2.2.4
Utilities
There are many other resources used in the system which are not visible as direct inputs but still play an important role in the system. According to this definition, energy also becomes a utility. So to avoid this misunderstanding or confusion, the energy was considered separately and anything other than energy, which played a key role in the system but did not end up in the output, was defined as other utilities. Water used in the construction and maintenance of the road can be linked to the suggested framework. 2.2.5 Raw material Raw materials for the road construction include crude oil, aggregates and additives. Extraction of crude oil, transportation to a refinery and extraction of bitumen from fractional distillation of crude oil was not inventoried for this study. Crude oil refining was described briefly in order to show how complex the system is and how difficult energy allocation becomes for bitumen and oil waxes for such complicated systems. Aggregate extraction from predetermined quarry sites was considered. Method to calculate the energy– mass flows for additives such as polymers and waxes will be described in the later section. 2.2.6
Figure 3. The asphalt pavement LCA framework showing the input – output flows in the system.
293
Energy
The second law of thermodynamics sets the limits to the conversion of thermal energy into the mechanical/electric energy. Therefore, there is a quality difference between both of these energies. The second form of energy can always be completely (100%) converted to the first form, but it is not true vice versa as this will depend on the starting and final temperatures of the energy conversion process. The conversion factor depends on the efficiency of the processes and the primary resources used. For the electricity production, the difference between the electricity mixes
294
A.A. Butt et al.
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
may have a large impact on the conversion factor. In Finland, the electricity efficiency is 0.48, i.e. 2.08 MJ of the energy raw material is required to give 1 MJ of electricity (Ha¨kkinen and Ma¨kela¨ 1996). In Sweden, Stripple (2001) reported 2.23 MJ of the energy raw material was used to produce 1 MJ of electricity. In the suggested framework, the electricity and the fuel energy have been kept separate. If the electricity and the fuel energy are to be cumulated to get equivalent thermal energy (ETE), conversion factors are needed to be known. 3. Production of additives Naturally occurring and synthetic waxes are organic high molecular weight compounds, mainly linear in structure, which are insoluble in water but soluble in organics. Petroleum, coal, animal or plant-derived waxes are the most common waxes used in different industries. Petroleum wax, for example is derived from crude oil (Figures 4 and 5). Crude oil is the natural resource that is extracted from the ground and transported to the refinery. Products such as fuel oils (diesel, gasoline, etc.) and bitumen are obtained in the refinery processes. In the refinery process, different distillate products are obtained by the fractional distillation of crude oil (Figure 4). Percentage weight of a certain product obtained after distillation is highly dependent on the origin of the crude oil. As an example, it has been seen that the Venezuela crude oil is far more enriched in bitumen than the North Sea crude oil. In the refinery process, the light gases (methane and ethane) are normally separated in the oil fields whereas the heavier gases (propane and butane) are topped off in the refinery; raw gasoline is naphtha, gas oil is raw diesel and vacuum gas oils (VGOs) are raw lube oils. Waste streams, unwanted distillates and residual oil that cannot be economically processed are blended into a range of heavy fuel oil (HFO) products. Bitumen or HFO can either be used as a fuel or can be processed by cracking or de-asphalting to form other fuel products. Bitumen, in particular, is referred to as a dirty fuel because of being highly impure organic material. Therefore, combustion of bitumen implies environmental risks if not used in the plants with proper flue gas cleaning. Similarly, the processing of
bitumen to lighter products will also require much energy and generate by-products that are of environmental concern if not properly managed. This means that burning or processing of bitumen is associated with extra environmental costs as compared with alternative fuels and feedstocks. VGOs or raw lube oil derived from paraffinic crude oils contains waxy parts of the crude oil. In a LCA, the raffinate distillation process can be looked upon in two ways. If the lube oil is the required product, then the wax becomes a byproduct, whereas if the wax is the main product, the whole life cycle energy and emissions from the base oil production should be allocated to the wax. Through a de-waxing process of the raffinate, the wax and the base oil (lube oil) are separated (Figure 5). The raffinate and solvent are fed to the mixer and the mixture is chilled. The wax then crystallises and is filtered out as the slack wax which is further de-oiled to produce the raw wax. Depending on its melting point, the raw wax can be used as a binder additive. If refined, it can be used in the shoe polish and waterproofing of paper. A method is required to show and calculate the energy and mass flows, and the emissions to air for the different products, e.g. Montan wax, FT paraffin wax, styrenebutadiene-styrene polymer if they are to be included in the LCAs. Process energies (electricity and fuel) then need to be known at each level of the Montan wax production and refining (Figure 6). Generally in an asphalt mixture, bitumen constitutes about 4 –6% by weight of the asphalt mixture and wax additive is almost 2– 3% by weight of the bitumen. As wax additive is used in such a small quantity, the benefit of using wax additive may be worth the energy consumed to produce it. However, such comparisons have not yet been published. 4.
Feedstock energy calculation
The asphalt components are not consumed during the pavements life, and therefore it may be argued that feedstock energy should not be included in a life cycle study of the asphalt pavements. It could be considered as borrowed from the nature. According to Van Oers et al. (2002), a certain functional element from a natural resource which can be recycled and has an economical reserve is considered as
Figure 4. Flow diagram of a crude oil refinery with a conventional lube line (rough numbers of carbon atoms per molecule are shown beside each separate distillate).
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
International Journal of Pavement Engineering
Figure 5.
Flow diagram of de-waxing process of the raffinate in which the wax and the base oil are separated.
Figure 6.
Block diagram of Montan wax extraction (Clariant) and refining (Steinera 1987).
borrowed. The feedstock energy consideration becomes important only when the asphalt is combusted to extract energy. The asphalt mixture is placed on the ground in form of a pavement, and once the asphalt pavement serves its function during its design life, it could be recycled, reused or else buried in the ground as the sub-grade. The feedstock energy remains unused as the asphalt materials although relocated but still are placed back into the nature in the form of asphalt mixture. When progressing from an LCA to its corresponding LCC, however, the feedstock energy contents of the binder become highly relevant as the cost of the binder
295
will be reflected in its alternative value as fuel. Therefore, it may be argued that the energy used within the asphalt materials should be reported although it is not consumed. What is missing in the literature is the method to calculate the feedstock energy in the asphalt pavements. Bitumen performance and properties as a binder could be investigated in the laboratory study, which is then used as the inputs in the road design procedures. The main component of HFO is the residual oil, i.e. the heaviest or the bottom product that comes out from a crude oil refinery if bitumen is not being produced. Energy contents of HFO
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
296
A.A. Butt et al.
Figure 7. Heating values relative to gravity and sulphur content (notes on Heavy Fuel Oil 1984; heating values converted to MJ/tonnes).
and that of the bitumen are basically the same if they are of the same density. Considering this fact, HFO equivalence can be used to find the energy value for the bitumen. There are good correlations available between density and energy contents for HFO (notes on Heavy Fuel Oil 1984). Hence, if the sulphur content and the density of the bitumen are obtained from the laboratory tests, the higher heating value (includes condensation of water in the flue gases) or lower heating value (LHV, all combustion products leave the system as gases except ashes) could be read from Figure 7. Normally, LHV is closest to the actual energy yield in most of the cases (NPC 2007). This heating value may therefore be seen as the feedstock energy which is the inherent energy. Bitumen has a high energy content of 40.2 MJ/kg (Garg et al. 2006) but using bitumen as a fuel results in very high emissions (Faber 2002, Herold 2003) and high energy costs. It has also been reported in number of previous studies that bitumen has a low expended energy (energy used throughout the production of a material) of approximately 0.4– 6 MJ/kg (Zapata and Gambatese 2005). Aggregate is considered to have no feedstock energy.
5.
In Sweden, resources such as fossil fuels, nuclear and hydropower, wind and biofuel are used to produce an electricity mix. The mix varies every year resulting in a different ETE value for every electricity mix produced. The conversion factor to get ETE consumed has been denoted as ‘X’ in the formulas depending on what electricity mix is being used. If a 100% feed enters the process, then equations will be as follows: ETE consumed per tonne product ¼ 100 £
X·E1 þ H 1 ; Y1 ð1Þ
ETE by-product calculation per tonne product ¼ 100 £
X·E1 þ H 1 ; 100 2 Y 1
ð2Þ
Method to calculate mass –energy flows
A framework to calculate the mass and energy consumption has been developed (Figure 8). There are different terms used: P stands for process, B means by-product, Y refers to yield, E is electricity and H is heating.
Figure 8. Framework to calculate mass balance and energy consumption of a process.
International Journal of Pavement Engineering Mass – distributed ETE ¼ ðX·E1 þ H 1 Þ £
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
ðX·E1 þ H 1 Þ £
ð100 2 Y 1 Þ : 100
Y1 100
Table 2. Energy data used for bitumen and aggregate.
and ð3Þ
Equations (1) and (2) are the result of typical economic calculations, whereas Equation (3) takes no position to allocation. Being faced with LCA data from product sheets, it is not always clear which distribution principles have been used. Standards normally recommend allocation to mass but this is no universal solution. The equation chosen must be based on the questions asked. As an example, one can also look at different scenarios in a process. If the final yield (Y3) is the required product (wax), the energy flow accumulates and may be allocated to the final product only. This way the by-product (Y2 2 Y3) could be considered having no energy allocation. 6.
Case study
6.1 Goal and scope The suggested framework for the asphalt pavement was applied on a theoretical case in which a typical Swedish asphalt pavement was assumed to be constructed as part of the Norra La¨nken (the North Link) project in Stockholm, Sweden. The functional unit (FU) for the case study was defined as the construction of 1 km flexible pavement per lane for the nominal design life. Energy, fuel and electricity were calculated as MJ/FU, whereas emissions and materials were calculated as tonne/FU. 6.2
297
Inventory analysis
The asphalt pavement design was based on the design life of 18 years for the ESALs of 7.5 million and a reliability of 85%. The resulting thickness of the asphalt pavement was 0.165 m and the width of the lane was 3.5 m. The density of asphalt was assumed to be 2.4 tonne/m3. The asphalt mix design consists of 4.5% bitumen and 95.5% aggregate by weight of the asphalt. Waxes and polymers were not considered, since these will not be used in the Norra La¨nken project. The 70/100 bitumen had sulphur content of 3% and specific gravity of 1.02 at 15.68C (608F). The feedstock energy of the bitumen was determined based on the information shown in Figure 7. The material energy and emission data-set for bitumen and aggregate can be read from Tables 2 and 3, respectively. For the impact assessment, only greenhouse gases (GHGs) were considered and their contribution to the environmental impact category of the climate changes. Swedish electricity mix was calculated based on the data from IEA (2008), whereas the raw material data and GHGs were calculated from Baumann and Tillman (2003; Table 4). Sensitivity analyses were done regarding the transport distances and the electricity production mix. The different choices regarding
Material
Type
Bitumena
70/100
Energy per tonne of material (MJ/tonne) 39,213 Electricity used per tonne of material (MJ/tonne)
Fuel used per tonne of material (MJ/tonne)
252 21.19
1060 16.99
Bitumenb 70/100 Aggregateb Crushed a
Feedstock energy calculated based on (Notes on Heavy Fuel Oil 1984; Figure 7). b Expended energy (Stripple 2001).
Table 3. Emissions to air in grams per tonne of bitumen and aggregates produced. Emissions to air (g/tonne of material)
Bitumena
Aggregate
CO2 N2O CH4
173,000 0.106 0.035
1537 0.058 0.529
a
Emissions from bitumen were assumed to be the same as reported by Stripple (2001).
both the framework design and the case specific system boundaries were done in cooperation with the asphalt industry and the construction companies in order to increase the relevance and the quality of the assessment. The main processes considered for this case study were the emissions and the energy used during the transportation of the materials, the asphalt mixing, paving and compaction. The data for the processes listed above is presented in Tables 5– 8. 6.3
Impact assessment and interpretation
The feedstock energy of the bitumen (2408 GJ) was almost 30 times higher than the process or the expended energy (82 GJ) to produce it (Table 9). The production energy of aggregate was 51 GJ. As no additives were considered and aggregates do not have any feedstock energy, the feedstock energy of the asphalt was the same as for the bitumen. The results of the electricity and the fuel were shown separately as without the conversion factor (Section 2.2.6.); the energy from the electricity and the fuel could not be accumulated. The asphalt production in the plant was the most energy-consuming process, both regarding the electricity and the fuel consumption due to the fact that the asphalt requires heating of the materials before mixing. High temperatures usually are required to dry the aggregates, melt the bitumen and additives, for the mixing and the storage of the asphalt mixtures. The second highest energy intensive process was the transportation of the materials as considerable amount of diesel is burned to transport the asphalt. Due to the localisation assumption done in this case study, a
24.94 1.431
Material
9181
Electricity/heat
Hot mix b
2726 8.67 48.63 536.55 17.71 135.62 20.35 28.27 35.25 237.58 176.72 17.23 10.89 5.03
Units
39,213 Amount per tonne of asphalt
Swedish mix Eldningsolja 1
kWh/tonne litre/tonne
Emissions to airc
Units
CO2 N2O CH4
g g g
8.3 6.8 Amount per tonne of asphalt 19,392 0.430 0.757
a
Feedstock energy. NCC (Jonas Ekblad). It has been assumed that the emissions from the production and combustion of Eldningsolja 1 are same as diesel.
b c
Table 6. Data set for the paver and the compactor (Stripple 2001).
Energy Speed Effective capacity Paving time (efficiency) Number of passes
The emissions from waste combustion (1.44%) were assumed to be equal to biofuel.
32.92 2.51 0.80 69.41 5.18 302.17 10.04 29.05 7.16 1.07 101.35 7.62 30.17 3.11 38.43 425.35 5.06 104.76 167.84 34.51 8.91 0.0638 0.0243 0.0176 0.9624 0.0268 0.0267 0.5508 1.49 0.58 0.40 42.6 6.04 46.1 1.33 Hard coal Oil Fuel gas Nuclear Biofuel Hydro Wind
Energy per tonne of asphalt (MJ/tonne)
Type
Asphalta
Summa
14.95 1.786 1.502 4.369 1.024 1.103 0.203 0.0267 0.0322 0.006 0.341 1.02 0.0069 0.001 19.66 0.166 0.418 9.113 10,517 1.361 0.209 159,396 112,302 2443.64 1,077,365 3977 5397.3 618 0.002 0.0005 7.36 £ 1025 3.3411 0.0005 0.0021 0.0002
4109.04 1334.71 988.03 1535.14 670.66 482.07 60.90
CH4 (kg) N2O (kg) CO2 (kg) Wood (kg) Water (kg) U in ore (kg) Hard coal (kg) Natural gas (Nm3) Limestone (kg) Lignite (kg) Oil (kg) Swedish mix (%)
Cu in ore (kg)
Table 5. Asphalt mixing process.
Paving/ rolling
Electricity produced 1 TJ
Emissions to air Resources consumed
Swedish electricity mix calculation based on IEA (2008) and the corresponding resource consumption and GHGs calculation based on Baumann and Tillman (2003).
A.A. Butt et al.
Table 4.
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
298
Units MJ/m2 m/h m2/h min/h
Paver Compactor (Dynapac F16) (Dynapac CC421) 0.5940 240 1300 50
0.7988 4000 791 50
1
6
relatively low amount of energy was used for transporting the asphalt and aggregates. However, as shown in the sensitivity analysis, different assumptions regarding the transport distances can totally change the results. Paving and compaction, on the other hand, do not require much energy, but this depends on what system boundaries have been defined. If the production energy of the equipments used to pave and compact the road are considered, the results might be quite different than what can be seen. Regarding GHGs, almost 51 tonnes of CO2, 0.9 kg of N2O and 2 kg of CH4 were produced per FU (Table 10). Using the data of 100-year global warming potential (GWP; Solomon et al. 2007; Table 11), these emissions correspond to almost 52 tonnes CO2-equiv. in terms of global warming contribution. The asphalt production was the most important process regarding these emissions, whereas transporting materials and bitumen production were also relatively important. According to the sensitivity analysis, change in the transport distances largely affects the energy consumption of the system (Tables 12 and 13). The asphalt consists of almost 95% by weight of aggregate which means that the aggregate quarry site and the asphalt plant may not be very far from each other or else, one of the most energy consuming processes will be the transportation of
International Journal of Pavement Engineering Table 7.
299
Transportation of materials by distribution trucks with 14 tonnes load capacity including weight of the vehicle.
Transport material
From
To b
Binder Aggregate Asphalt
Distancea (km)
Material quantity (tonne)
Tonne-kilometre (tkm)
100 5 50
63 1324 1386
12,474 13,236 138,600
c
Refinery Quarry sitec Mixing plant
Mixing plant Mixing plant Construction sited
a
Distance will double as loaded trucks will roll to the required site and unloaded when coming back. Nyna¨shamn: Bitumen refinery. c Arlanda: aggregate quarry site and asphalt mixing plant. d Norra La¨nken: road construction site.
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
b
Table 8. Emissions from vehicles, paver and compactor (Stripple 2001). Emissions to air
Units
CO2 N2O CH4
g g g
Table 9.
Amount per MJ energy used (g/MJ) 79 0.0016 0.00005
aggregates to the asphalt plant. With an increase in the distance of 95 km between aggregate quarry site and asphalt plant, the fuel energy increased from 11 to 226 GJ/FU. Similarly, the distance between the construction site and the asphalt plant will also alter the results significantly. Increasing the distance between the asphalt mixing plant and the construction site also resulted in an increase in the transportation energy from 118 to 177 GJ/FU. Thus, in case of the sensitivity analysis, the transportation energy
Results of the case study.
Feedstocks energy (TJ)
Bitumen Aggregate Asphalt
2.4 0 2.4 Energy consumed per tonne of material (MJ/tonne)
Total energy (GJ/FU)
Bitumen production Aggregate production Asphalt production
252.00 21.19 29.88
15.72 28.05 41.41
Bitumen production Aggregate production Asphalt production
1060.00 16.99 242.00
66.11 22.49 335.41
Item Electricity consumption
Fuel consumption
Transport bitumen to the asphalt plant Transport aggregate to the asphalt plant Transport asphalt to the construction site Laying asphalt
3.86
Compacting asphalt
2.27
Table 10. Total emissions to air from different processes of road construction in tonnes/FU. Emissions to air
10.63 11.28 118.15
CO2
N2O
CH4 26
Bitumen production Aggregate production Asphalt production Paving Compacting Transportation
10.79 2.03 26.88 0.31 0.18 11.06
6.61 £ 10 7.61 £ 1025 5.96 £ 1024 6.18 £ 1026 3.64 £ 1026 2.24 £ 1024
2.20 £ 1026 7.01 £ 1024 1.05 £ 1023 1.93 £ 1027 1.14 £ 1027 7.00 £ 1026
S (tonnes)
51.25
9.13 £ 1024
1.76 £ 1023
consumption was much higher than the asphalt production energy, making the transportation energy to be highest on the energy chain. According to the sensitivity analysis of the electricity production assumptions, the production may have a large impact on the results. Electricity is used for heating in Table 11. GWP for the GHGs (Solomon et al. 2007). Greenhouse gas (GHG)
Formula
Carbon dioxide Methane Nitrous oxide
CO2 CH4 N2O
100-year GWP 1 25 298
300 Table 12.
Transportation of materials by distribution trucks with 14 tonnes load capacity including weight of the vehicle.
Transport material
From
To
Binder Aggregate Asphalt
Refinery Quarry site Mixing plant
Mixing plant Mixing plant Construction site
a
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
A.A. Butt et al.
Distancea (km)
Material quantity (tonne)
Tonne-kilometre (tkm)
150 100 75
63 1324 1386
18,711 264,726 207,900
Distance will double as loaded trucks will roll to the required site and unloaded while coming back.
most of the asphalt plants in countries where electricity is relatively low cost. However, it might have a high environmental impact due to the resources used for its production. This should not be neglected. The sensitivity analysis was done by comparing the process energy at an asphalt plant which uses Swedish electricity mix from 2008 (IEA), and the asphalt plant which produces the electricity from an electricity generator running on diesel. The efficiency of the generator is around 33%. Hence, 3 MJ of diesel energy is used to produce 1 MJ of electricity resulting in the excess amount of emissions. Almost 26 times more emissions per tonne of asphalt produced were reported if the electricity used in the asphalt plant was generated using a diesel generator (Table 14). This may be significantly exasperated if the heating in an asphalt plant is also carried out using electricity rather than fuel. From this case study, it was found that the asphalt production and transportation of materials were the two most energy-consuming processes, and they had an impact on the environment in regard to emissions. This conclusion largely depends on what fuel and electricity mix has been used. Hence, if the environmental improvements are to be made, transport distances and efficient electricity proTable 13. Sensitivity analysis by changing transportation distances.
Item Fuel consumption
Transport bitumen to the asphalt plant Transport aggregate to the asphalt plant Transport asphalt to the construction site
Total energy consumption (GJ/FU) 15.95 225.66 177.22
Table 14. CO2 emissions from Swedish electricity mix and a power plant run on diesel.
duction are important aspects to be considered. However, it should be noted that the direct emissions from the asphalt material such as dust, solvents and gases have not been considered in this case study. Furthermore, the asphalt pavement is a part of a larger system in which surrounding constructions and the traffic may also have an impact on the results.
7.
Conclusions
In this study, an open LCA framework was suggested for quantifying energy and environmental loads during the construction, maintenance and end-of-life phases of a given asphalt pavement. A method to calculate the feedstock energy of bitumen was developed, and a method to quantify the mass – energy flows of warm asphalt and polymer bitumen additives was described. If the production data of additives are available, an energy –mass flow of any asphalt additive can be calculated based on the method suggested. Such calculations for waxes and polymers should be valuable in order to determine the life cycle benefits from using such additives. However, this would require information on the electricity and the fuel usage. Regarding the feedstock energy in the binder, it is highly relevant for the LCC as the cost of the binder will be reflected in its alternative value as fuel. For LCAs, it may be used to quantify the resource energy. From this case study, it could be concluded that the asphalt production was a highly energy-consuming process. Hence, the use of additives should be further studied in order to determine their potential to decrease energy use through lowering the mixing temperatures. Transportation of the materials plays a very important role in terms of the energy consumption and emissions. It is highly favourable to have the quarry site, the asphalt plant and the construction site not far from each other to avoid excess energy use and fuel combustion emissions. It is also highly favourable to use the electricity that has been produced in an efficient way.
Emissions to air from asphalt production (g/tonne asphalt)
CO2
Acknowledgements
Electricity mix Electricity generator
274 7082
All the authors specially like to thank Mr Ma˚ns Collin for his ideas and help in development of this work, as well as Akzo-Nobel and Mr Mats Norell, and Dr Jonas Ekblad at NCC.
International Journal of Pavement Engineering Notes 1. 2. 3.
Email:
[email protected] Email:
[email protected] Email:
[email protected]
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
References Baumann, H. and Tillman, A.M., 2003. The Hitch Hiker’s guide to LCA, an orientataion in LCA methodology and application. Go¨teborg: Studentlitteratur. Birgisdo´ttir, H., 2005. Life cycle assessment model for road construction and use of residues from waste incineration. Thesis (PhD), Institute of Environment and Resources, Technical University of Denmark, DTU. Chowdhury, A. and Butto, J., 2008. A review of warm mix asphalt, Report Number: SWUTC/08/473700-00080-1, Texas Transportation Institute, December, Texas, US. Clariant, Montan Wax Extraction [online]. Available from: http://pa.clariant.com/pa/e2wtools.nsf/lookupDownloads/ W320GB_0503.pdf/$FILE/W320GB_0503.pdf [Accessed 10 May 2011]. D’Angelo, J., et al., 2008. Warm mix asphalt: European practice, Report number: FHWA-PL-08-007, Federal Highway Administration, February, Washington DC. Das, P., Tasdemir, Y., and Birgisson, B., 2012. Evaluation of fracture and moisture damage performance of wax modified asphalt mixtures. International Journal on Road Materials and Pavement Design, 13, 142– 155. Ecobilan, LCA History [online]. Available from: https://www.ecobilan.com/uk_lca02.php [Accessed 5 November 2010]. ECRPD, Energy Conservation in Road Pavement Design, Maintenance and Utilisation, 2009. WP6 – Life Cycles Evaluation, Centrum dopravniho vyzkumu, Lisenska, Czech Republic, November. Edwards, Y., Tasdemir, Y., and Butt, A.A., 2010. Energy saving and environmental friendly wax concept for polymer modified mastic asphalt. Materials and Structures, 43 (Supplement 1), 123– 131. Faber, J., 2002. Towards small scale use of asphalt as a fuel: an application of interest to developing countries, Chemiewinkel Rapport C102, University of Groningen, The Netherlands. Garg, A., Kazunari, K., and Pulles, T., 2006. 2006 IPCC Guidelines for the National Greenhouse Gas Inventories, Intergovernmental Panel on Climate Change. Ha¨kkinen, T. and Ma¨kela¨, K., 1996. Environmental impact of concrete and asphalt pavements, environmental adaption of concrete, Research Notes 1752, Technical Research Center of Finland. Herold, A., Comparison of CO2 emission factors for fuels used in greenhouse gas inventories and consequences for monitoring and reporting under the EC emissions trading scheme. European Toxic Center on Air and Climatic Change, ETC/ACC Technical paper 2003/10. Horvath, A., 2004. Pavement life-cycle assessment tool for environmental and economic effects, PaLATE [online]. Available from: http://www.ce.berkeley.edu/, horvath/ palate.html [Accessed 6 April 2011]. Horvath, A. and Hendrickson, C., 1998. Comparison of environmental implications of asphalt and steel-reinforced concrete pavements. Transportation Research Board, 1626, 105– 113. Huang, Y., Bird, R., and Heidrich, O., 2009. Development of a life cycle assessment tool for construction and maintenance
301
of asphalt pavements. Journal of Cleaner Production, 283– 296. IEA, International Energy Agency, Electricity/Heat in Sweden in 2008 [online]. Available from: http://www.iea.org/stats/ electricitydata.asp?COUNTRY_CODE¼SE [Accessed 29 August 2011]. ISO 14040, 1997. Environmental management – life cycle assessment – principles and framework. Geneva: International Organisation for Standardisation. ISO 14041, 1998. Environmental management – life cycle assessment – goal and scope definition and inventory analysis. Geneva: International Organisation for Standardisation. ISO 14042, 2000. Environmental management – life cycle assessment – life cycle impact assessment. Geneva: International Organisation for Standardisation. ISO 14043, 2000. Environmental management – life cycle assessment – life cycle interpretation. Geneva: International Organisation for Standardisation. Koziol, W., Kawalec, P., and Kabzinski, A., 2008. Production of Aggregates in European Union 2008 [online]. Available from: http://www.min-pan.krakow.pl/Wydawnictwa/GSM2443/ koziol-kawalec-kabzinski.pdf [Accessed 18 May 2011]. Notes on Heavy Fuel Oil, 1984. American Bureau of Shipping, ABS, publication 31 in January 1984. NPC, 2007. Working Document of the National Petroleum Council (NPC) Global Oil and Gas Study (18th July 2007) [online]. Available from: http://www.npc.org/ study_topic_papers/8-stg-biomass.pdf [Accessed 13 June 2012]. Park, K., et al., 2003. Quantitative assessment of environmental impacts on life cycle of highways. Journal of Construction Engineering and Management, 129 (1), 25 – 31. Santero, N., 2010. How to consider bitumen feedstock energy. Pavement life cycle assessment workshop. Feedstock Energy in Bitumen [online]. Available from: http://www.ucprc. ucdavis.edu/P-LCA/pdf/04_feedstock_web.pdf [Accessed 29 August 2011]. Santero, N., et al., 2010a. Environmental life-cycle assessment for asphalt pavements: issues and recommended future directions. ISAP, 2010. Nagoya, Japan. Santero, N., Masanet, E., and Horvath, A., 2010b. LCA of Pavements: a critical review of existing literature and research. Portland Cement Association, Skokie, Illinois, USA. SGU, Grus, Sand Och Krossberg 2009 [online]. Periodiska publikationer 2010:2. Available from: http://www.sgu.se/ dokument/service_sgu_publ/perpubl_2010-2.pdf [Accessed 25 May 2011]. Sivilevicˇius, H. and Sˇukevicˇius, Sˇ., 2009. Manufacturing technologies and dynamics of hot-mix asphalt mixture production. Journal of Civil Engineering and Management, 15 (2), 169– 179. Solomon, S., et al., 2007. Intergovernmental Panel on Climate Change (IPCC), Fourth Assessment Report (AR4), Working Group 1 (WG1), Chapter 2, Changes in Atmospheric Constituents and in Radiative Forcing, Table 2.14, p. 212. Steinera, R., 1987. Operating characteristics of special bubble column reactors. Chemical Engineering and Processing: Process Intensification, 21 (1), 1 – 8. Stripple, H., 2001. Life cycle assessment of road: a pilot study for inventory analysis. Second Revised Edition in March. Go¨teborg, Sweden: IVL Swedish Environmental Research Institute.
302
A.A. Butt et al.
Downloaded by [Kungliga Tekniska Hogskola], [Iman Mirzadeh] at 06:17 24 March 2014
Van Oers, L., et al., 2002. Abiotic resource depletion in LCA; Improving characterisation factors abiotic resource depletion as recommended in the new Dutch LCA handbook. RWSDWW report 2002-061, CML-Industrial Ecology, Leiden. VTT, Measurements made by VTT communities and infrastructure. Finland. Willett, J., 2011. USGS Releases 2010 Production Estimates from Rock Products (25th January 2011) [online]. Available from: http://rockproducts.com/index.php/news-late/10189-usgs-
releases-2010-production-estimates.html [Accessed 18 May 2011]. Zapata, P. and Gambatese, J., 2005. Energy consumption of asphalt and reinforced concrete pavement materials and construction. Journal of Infrastructure Systems, 11 (1), 9–20. Zhang, H., Keoleian, G., and Lepech, M., 2008. An integrated life cycle assessment and life cycle analysis model for pavement overlay systems. Life-Cycle Civil Engineering, 907– 912.
Paper II Mirzadeh, I., Butt, A., Toller, S., & Birgisson, B. (2013). Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements, Structure and Infrastructure Engineering. doi:10.1080/15732479.2013.837494.
This article was downloaded by: [Kungliga Tekniska Hogskola] On: 24 September 2013, At: 07:33 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Structure and Infrastructure Engineering: Maintenance, Management, Life-Cycle Design and Performance Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/nsie20
Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements a
a
b
Iman Mirzadeh , Ali Azhar Butt , Susanna Toller & Björn Birgisson
a
a
Division of Highway and Railway Engineering, KTH, Royal Institute of Technology , SE-10044 , Stockholm , Sweden b
Swedish Transport Administration , Stockholm , Sweden Published online: 24 Sep 2013.
To cite this article: Iman Mirzadeh , Ali Azhar Butt , Susanna Toller & Björn Birgisson , Structure and Infrastructure Engineering (2013): Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements, Structure and Infrastructure Engineering: Maintenance, Management, Life-Cycle Design and Performance To link to this article: http://dx.doi.org/10.1080/15732479.2013.837494
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Structure and Infrastructure Engineering, 2013 http://dx.doi.org/10.1080/15732479.2013.837494
Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements Iman Mirzadeha*, Ali Azhar Butta1, Susanna Tollerb2 and Bjo¨rn Birgissona3 a
Division of Highway and Railway Engineering, KTH, Royal Institute of Technology, SE-10044 Stockholm, Sweden b Swedish Transport Administration, Stockholm, Sweden
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
(Received 29 April 2013; final version received 17 August 2013; accepted 19 August 2013) A life cycle costing system should include the key variables that drive future costs in order to provide a framework for reducing the risk of under- or overestimating the future costs for maintenance and rehabilitation activities. In Sweden, price of oil products is mostly affected by the global economy rather than by the national economy. Whereas the price index of oil products has had a high fluctuation in different time periods, the cost fluctuation related to labour and equipment has been steady and followed the consumer price index (CPI). Contribution of the oil products was shown to be more than 50% of the total costs regarding construction and rehabilitation of asphalt pavements in Sweden. Consequently, it was observed that neither Swedish road construction price index (Va¨gindex) nor CPI has properly reflected the price trend regarding the asphalt pavement construction at the project level. Therefore, in this study, a framework is suggested in which energy- and time-related costs are treated with different inflation indices in order to perform a better financial risk assessment regarding future costs. Keywords: road and highways; life cycle cost; maintenance; rehabilitation; user cost
1.
Introduction
Roads provide one of the most significant functions of infrastructures, and they play a key role in the development of the society all around the world. Despite being a crucial part of national economies, modern road network systems require a high level of national investments. In order to keep the quality of the roads at an acceptable level, a large amount of investments for rehabilitation and maintenance activities are necessary in addition to investments in new and reconstructed roads. Therefore, an efficient allocation and use of road investment funds are of great economic importance. Life cycle costing (LCC) is a methodology for systematic economic evaluation of a project for its defined lifetime (ISO15686-5, 2008). The minimum expected total cost over the design life of an infrastructure project has been the most common criterion used in design optimisations (Frangopol & Liu, 2007; Safi, Sundquist, Karoumi, & Racutanu, 2012). LCC is growing in importance as a tool for providing cost estimation over the lifetime of a road project. LCC provides a system-based framework for identifying current costs and key factors affecting projected future costs and therefore holds the promise to help designers, road administrations and contractors with choosing the most economically efficient investment strategies for roads. It is therefore extremely important that the LCC system includes the key variables that drive future costs and thus provides a framework for reducing
*Corresponding author. Email:
[email protected] q 2013 Taylor & Francis
the risk of under- or overestimating the future costs for maintenance and rehabilitation activities. The Swedish Transport Administration (STA) has required LCC for all road-related investments as of 2012. In traditional LCC analyses, the prediction of the costs in a project is done through the estimation of current or future costs by predicting the effect on costs from likely trends in interest rate and inflation (Chan, Keoleian, & Gabler, 2008; Eisenberger & Remer, 1977; Federal Highway Administration [FHWA], 2003; Mandapaka et al., 2012; Santos & Ferreira, 2013; Walls & Smith, 1998; Zhang, Keoleian, & Lepech, 2008). The prediction of inflation is sometimes based on broad indices, such as gross domestic product chain deflator or intermediate indices, such as the consumer price index (CPI) or even narrower highway construction-related cost indices (e.g. FHWA, 2002). The use of producer price index for nonresidential construction was suggested by Michigan Department of Transportation (MDOT, 2012) in order to predict current and future costs. Recognising that the broader indices may not adequately reflect the development of future costs related to road construction, the STA (2012) adopted a road construction price index entitled the ‘Va¨gindex’. However, the use of the Va¨gindex was criticised by the Swedish National Audit Office (NAO, 2010) based on the following reasons: (1) cost trends in the road construction sector are not comparable with that of the general economy, (2) the
2
I. Mirzadeh et al. the decided road alignment is defined in terms of the initial asphalt layer thickness, the overlay thickness and the rehabilitation frequency.
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
2.2
Figure 1. The normalised indices, regarding Swedish road construction, CPI for Sweden and crude oil price.
use of Va¨gindex does not adequately reflect the cost increases resulting from the relatively high price trends in the construction industry and (3) the widespread use of Va¨gindex is likely to reduce the awareness of potential future costs at the STA. These concerns are illustrated in Figure 1 that shows that although the Va¨ gindex approximately followed the Swedish CPI for the last 25 years, it failed to adequately capture the dramatic increase in the price of oil during this period. A significant part of the costs associated with asphalt pavements is related to the cost of oil products such as bitumen, fuel oil and transportation fuel. Therefore, application of one generalised inflation index such as Va¨gindex on the total cost may lead to a false estimation of future costs. This study by applying the energy and time units as a basis for cost calculations imposed separate inflation indices on the energy-related costs and the labour/road user-related costs. Therefore, it can accommodate the trends in the energy price and increase the awareness among the road authorities and contractors regarding the cost of future maintenance and rehabilitation activities. The objective of this study was to present an LCC framework that is based on the cost of energy and time associated with road construction in order to better reflect the current and future costs associated with construction, rehabilitation and maintenance of roads.
2.
Framework development
2.1 Scope and limitations This study as an LCC study at the project level has focused on the construction and rehabilitation of the asphalt pavement. The activities that are mutual for the design alternatives such as the routine maintenance, the traffic cost during the lifetime of the asphalt pavement, the amount of earth work and construction of the unbound material are not part of this study. A design alternative for
Model development
The LCC of asphalt pavements includes different cost components, which are affected by national and global economy. Traditional LCC studies adjust future costs according to interest rate and inflation at the national economy level. However, tradable goods such as crude oil, fuel oil, transportation fuel and sometimes electricity have a global market, and their future prices can be independent of the country economic indicators such as inflation rate. Therefore, in this framework, the costs are divided into energy- and time-related costs. Time-related costs are those affected by the national economy such as labour/equipment and road user costs. Energy-related costs are those affected by the global economy such as crude oil cost as a part of bitumen cost and expended energy regarding production, construction and transportation of material. The costs of construction and rehabilitation are divided into energy- and time-related components regarding the agency and user costs (Figure 2). The time-related components were those concerning labour and equipment for construction and rehabilitation activities together with the delay cost for the road users during the rehabilitation activity. Energy-related costs are separated into crude oil energy and expended energy. The crude oil energy is the energy stored in the material which in this study represents the value of crude oil. The expended energy is the amount of the energy which was spent during the material production, construction and rehabilitations of the road. The expended energy in the refinery and asphalt plant was expressed for the bitumen, aggregate and asphalt mixture production. Regarding the construction site, the agency-related expended energy was attributed to laying, compacting, milling and resurfacing. Furthermore, the user-related expended energy regarding the construction site was defined for the extra amount of fuel used by the vehicles in the work zone. The amount of the expended energy for the transportation was related to all the required material distribution from the refinery to the construction site. The amount of the fuel used and depreciation of the machineries such as trucks, pavers and rollers were included in the framework. The amount of the depreciation was included in the labour and equipment costs.
2.2.1
Energy inflation index
The energy prices in Europe are affected by a number of factors. Prices of primary fuels such as oil, coal and gas will influence the price of secondary fuels such as electricity and road fuels. Moreover, they can themselves be affected by
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Structure and Infrastructure Engineering
Figure 2.
3
LCC framework for asphalt pavements. Agency costs (white), user costs (grey).
the other primary fuel prices. Coal and gas, specifically, can act as substitute in electricity generation. However, the price of oil is the main driver of energy prices (UK Department of Energy and Climate Change [DECC], 2012). Products refined from crude oil, such as transportation fuels, gas oil and fuel oil, are obviously closely linked to the price of crude oil and follow oil price fluctuations (DECC, 2012). Other affecting factors on the oil products price are the rates of fuel duty and value added tax. There has been a delay between the changes in the oil price and that in the other energy sources (Figure 3). The delay was shortest for the oil products such as fuel oil.
Although the delay is longer for electricity price, the prices of coal and gas have to change before the price of electricity. In order to relate construction costs to a limited (minimum) number of fundamental cost indices, the crude oil inflation index was chosen as the inflation index for the energy-related costs in this framework.
2.2.2
LCC based on energy and time
The calculation regarding the energy cost is done for crude oil energy and expended energy components (Table 1). The crude oil energy represents the value of crude oil in bitumen and modifier. Therefore, the costs of bitumen and Table 1.
Energy-related variables regarding LCC framework. Energy Energy Crude source source Cost of oil (MJ) A (MJ) B (MJ) energy (e)
Figure 3. Fluctuation of different energy source prices in respect to oil price (price index 2005 ¼ 100).
Bitumen Modifier Aggregate production Bitumen production Modifier production Asphalt production Transportation Laying asphalt Compacting asphalt User’s energy loss
qc1 qc2 – – – – – – – –
– – qa3 qa4 qa5 qa6 qa7 qa8 qa9 qa10
– – qb3 qb4 qb5 qb6 qb7 qb8 qb9 qb10
E1 E2 E3 E4 E5 E6 E7 E8 E9 E10
4
I. Mirzadeh et al.
modifier are defined as the sum of the corresponding costs for crude oil energy and the expended energy cost regarding the production and transportation. The cost of energy regarding materials and processes is expressed based on the quantity of the involved energy sources and the corresponding energy prices. The unit cost of energy for each item (Table 1), which is the total cost related to crude oil and the other energy sources, can be calculated via Equation (1), while the calculations can be repeated in case of having more energy sources involved:
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Ek ¼ qck £ Pc þ qak £ Pa þ qbk £ Pb þ · · ·;
ð1Þ
where qck, qak and qbk are the quantities (MJ) regarding crude oil, energy source A and energy source B. Pc, Pa and Pb are the unit prices (e/MJ) regarding crude oil, energy source A and energy source B at the base year. The amount of energy used related to asphalt pavement works can be found in life cycle inventory databases and several life cycle assessment (LCA) publications, including LCA report by Stripple (2001) in Sweden, Energy Conservation in Road Pavement Design (ECRPD, 2009) in Europe, EcoInvent (Swiss Centre for Life Cycle Inventories, 2011), the asphalt inventory produced by the Athena Institute (2006) in Canada and the US Life Cycle Inventory produced by the National Renewable Energy Laboratory (2011). Different sources represent different local condition and technologies. The amount of different sources of energy regarding the pavement works presented by Stripple (2001) is used in this study as it represents the Swedish condition. The costs of time-related variables regarding the labour cost and road users’ delay cost are presented in Table 2. The value of time included the hourly wage of the labours together with the amount of hourly depreciation regarding the equipment. The hourly depreciation was defined as the initial price of the equipment divided by the useful life of the equipment (hours). The calculation regarding the user delay cost has to be done separately for the personal cars and heavy vehicles as they should be treated with different values of time. Whenever the time value of money is considered, the life cycle cost is the sum of all costs in the life cycle Table 2.
Time-related variables regarding LCC framework.
Item Labour and equipment Transportation Laying asphalt Compacting asphalt Milling and resurfacing Road users User’s delay cost
Time spent (h)
Value of time (e/h)
Cost of time (e)
t1 t2 t3 t4
CL1 CL2 CL3 CL4
T1 T2 T3 T4
t5
Cu
T5
discounted at an interest rate i and an inflation j to the year 0 (Eisenberger & Remer, 1977). The net present value (NPV) of a cost C at the yth year can be calculated according to: 1þj y NPV ¼ C £ ; ð2Þ 1þi in which y is the number of years after the construction; j is the yearly general inflation rate and i is the rate of interest; while the conventional discount rate (r) is calculated according to Amini, Mashayekhi, Ziari, & Nobakht (2012): r¼
1þi 2 1: 1þj
ð3Þ
Based on the above concept, NPV of a construction or a rehabilitation activity is reformulated as: " # n n X X 1 y y y NPV ¼ Ek £ ð1 þ jc Þ þ T k £ ð1 þ jÞ £ ; 1þi k¼0 k¼0 ð4Þ where Ek and Tk are the cost of energy and time for item k, while jc is the yearly crude oil inflation rate. It is obvious that the rate of interest and inflation should be 0 for the initial construction activity. The NPV can also be calculated based on the discount rate (FHWA, 2002): y 1 NPV ¼ C £ : ð5Þ 1þr Using Equations (2)– (4), the equivalent discount rate (req) is calculated as follows: 1=y ðE þ TÞ £ ð1 þ iÞy r eq ¼ 21: ð6Þ E £ ð1 þ jc Þy þ T £ ð1 þ jÞy In the case of comparing alternatives with different life spans, equivalent annual cost (EAC) can be used in order to find the cost-effective alternative. Actually, EAC is the cost per year of possessing and maintaining a property over its life span (Equation (7)): r eq EAC ¼ TNPV £ ; ð7Þ 1 2 ð1 þ r eq Þ2d where TNPV is the total net present value (sum of all the NPVs regarding construction and rehabilitation activities) and d is the design life of the pavement.
2.2.3
Traffic simulation model
The road user cost in this framework is related to added vehicle operating cost and delay costs in the work zones resulting from the construction, maintenance or rehabilitation activity. Other costs in the work zone such as accident costs, safety-related costs and environmental costs are not considered. The Swedish National Road and Transport
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Structure and Infrastructure Engineering Research Institute traffic model which is based on AIMSUN micro-simulation software was applied for the traffic analysis (Wennstrom, 2010). As long as the traffic volume is lower than the work zone capacity, the traffic flow is disturbed either by lowering the traffic capacity or by setting a speed limit (New Jersey Department of Transportation [NJDOT], 2001). As the traffic flow exceeds the road capacity, a queue starts to build up in the upstream of the work zone. Vehicles approaching to the work zone should stop and go through creep along the length of the queue under the forced flow condition (Walls & Smith, 1998). Consequently, the total delay and the energy loss in the work zone increase with an exponential rate. The economic impact of traffic congestion can be categorised as the effects on freight and service delivery, business operation, business scheduling, international connections and workers travel (Weisbrod & Fitzroy, 2007). The impact of the work zone on the business operations was shown to be the dominant part of user costs (Yu & Lo, 2007). Hence, the value of time regarding road users should reflect those effects.
2.2.4 Financial uncertainties In infrastructure risk assessment, uncertainties are defined as aleatory and epistemic (Ang & De Leon, 2005). In order to compute the costs related to energy and time components for any future activity such as maintenance or rehabilitation, the aleatory uncertainties regarding variables such as inflation rates, interest rate, traffic level and currency fluctuation should be considered. The epistemic uncertainty due to the possible inability of the model not fully representing the constructional cost components also needs to be considered. Any tradable input (e.g. fuel) even if it is not imported will have an international price. Therefore, its price, measured in local currency, will vary inversely with the exchange rate (Gray & Irwin, 2003). However, the effect of the currency fluctuation would be more critical in the public –private partnerships and private financial initiative projects for either the contractors or the investors on the overseas projects. The contractors might source materials and equipment globally and receive part or all of their revenues in terms of the host country’s currency [e.g. in China’s infrastructure projects, foreign companies receive almost all of their revenues in Renminbi (Wang, Tiong, Ting, & Ashley, 2000)].
3.
Case study
The framework was applied on a case study regarding the construction and rehabilitation of 1 km flexible Swedish asphalt pavement located at Norra La¨nken in Stockholm. The total energy and time variables were calculated for a
5
functional unit that was defined as construction and rehabilitation of 1 km asphalt pavement per lane for the 25-year design life. The analysed pavement profile consisted of 50 mm wearing course above a 100-mm binder course and a 500-mm granular unbound foundation layer (Gullberg, Birgisson, & Jelagin, 2012). The wearing course was Dense Asphalt Concrete (ABT11) which is a dense-graded mixture with a maximum aggregate size of 11 mm. The binder course was assumed to be Gravel Asphalt (AG22) which is an asphalt-bound foundation layer with a maximum aggregate size of 22 mm (Gullberg et al., 2012). The traffic level for both directions was assumed to be 13 million Equivalent Single Axel Loads (ESALs) during the design life. A rehabilitation activity was assumed at the 15th year consisted of a 75-mm overlay layer (ABT11). The required thickness regarding the initial design and the rehabilitation was obtained by the calibrated mechanicsbased (MC) model (Gullberg et al., 2012). The MC model is based on the mechanistic empirical pavement design model developed by Wang, Birgisson, and Roque (2007). The MC model is calibrated for the Swedish condition. In order to study the validity of the model on Swedish mixtures, more mixtures were analysed. The mixture properties such as binder content, aggregate sieve size and void content can be found in STA (2010).
3.1 Refinery and asphalt plant The crude oil energy content was assumed to be 46,300 MJ/tonne. The amount of fuel oil and electricity consumption regarding the material production based on Stripple (2001) and ECRPD (2009) is presented in Table 3. The data regarding bitumen production from ECRPD (2009) which was much higher than from Stripple (2001) represented a Czech producer (Table 4).
3.2 Transportation Calculations regarding the amount of energy used and spent time were done based on the corresponding transportation Table 3.
Energy used in the asphalt plant. Fuel oil consumption (MJ/tonne)
Bitumen production Aggregate production Asphalt production a
Electricity consumption (MJ/tonne)
ECRPD (2009)
Stripple (2001)a
ECRPD (2009)
Stripple (2001)a
4000 5 268
1060 17 242
12 6 18
252 21 30
Stripple data were used for this study because it represents Swedish condition.
6 Table 4.
I. Mirzadeh et al. Asphalt pavement machineries energy consumption and depreciation.
Construction machinery Asphalt paver, Dynapac F16 Asphalt compactor, Dynapac CC122 Distribution truck, max load 14 tonne Long-distance truck, max load 32 tonne a
Energy consumption (MJ/m2)a
Price (e/unit)
Depreciation (e/h)
0.017 0.011 0.34 0.38
0.594 0.388 11.9 13.3
200,000 50,000 130,000 200,000
6.8 1.7 4.5 6.8
Expended energy (Stripple, 2001).
Table 5.
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Fuel consumption (l/m2)a
Material transportation with 14 tonnes capacity trucks.
Transport materiala
From– to
Distance (km)b
Binder Aggregate Asphalt
Refineryc –mixing plantd Quarry sited – mixing plant Mixing plant – construction sitee
100 5 50
regarding milling was assumed to be 1.56 MJ/m2 or 0.0044 l diesel/m2 (Stripple, 2001). 3.4
a
Average trucks speed assumed to be 50 km/h. Norra La¨nken: road construction site. c Nyna¨shamn: bitumen refinery. d Arlanda: aggregate quarry site and asphalt mixing plant. e Distance will double as loaded trucks will roll to the required site and unloaded when coming back. b
distances (Table 5). The energy used by trucks at the maximum load and empty on return is presented in Table 1. The amount of the depreciation was calculated for the assumed 10 years useful life of trucks (Table 4). 3.3 Laying, compacting and milling The required time and corresponding energy consumption for laying, compacting and milling were calculated for the construction area of 3500 m2/km/lane. The effective capacity was assumed as 1300 m2/h for pavers and 790 m2/h for compactors. The amount of energy consumption and depreciation regarding the paver and the compactor is presented in Table 4. The energy consumption
Traffic
The road is chosen to be a two-lane directional. The outer lane was assumed to be closed during the maintenance activity. The maximum allowable speeds under normal condition and under the work zone condition were assumed to be 110 and 50 km/h. The rehabilitation was assumed to be performed during both day and night (24 h) The hourly distribution of traffic was assumed to be 1% during the night hours (0 –7, 18– 24), 12% for the peak hours (7 –9, 18 –24) and 5% during the midday hours. The amount of energy loss by vehicles and the delay time for the users related to variation in average annual daily traffic (AADT) for both lanes and different heavy vehicle ratio (HVR) are presented in Figure 4. The value of time was assumed as 60 e/h for passenger cars and 500 e/h for the heavy vehicles. 4. Results and discussion 4.1 Analysis excluding user costs The calculation of the energy costs was done based on the price of energy sources as 0.01 e/MJ for crude oil, 0.038 e/MJ for diesel, 0.024 e/MJ regarding electricity price for
Figure 4. (a) AADT impact on the total delay during the rehabilitation activity and (b) AADT impact on the energy loss during the rehabilitation activity.
Structure and Infrastructure Engineering Table 6.
The summary of energy consumptions and corresponding costs for the construction and the rehabilitation activity.
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Item
Crude oil (GJ/km)
Fuel oil (GJ/km)
Electricity (GJ/km)
Diesel (GJ/km)
Cost of energy (e/km)
2956 – – – – – – –
– 20 67 305 – – – –
– 25 16 38 – – – –
– – – – 64 2 8 –
30,148 922 1388 5489 2410 78 307 –
1750 – – – – – – – –
– 10 40 152 – – – – –
– 12 10 19 – – – – –
– – – – 32 5 2 8 24
17,849 443 833 2744 1222 206 78 307 904
Construction Bitumen Aggregate production Bitumen production Asphalt production Transportation Laying asphalt Compacting asphalt User’s energy loss Rehabilitation Bitumen Aggregate production Bitumen production Asphalt production Transportation Milling Laying asphalt Compacting asphalt User’s energy loss
Table 7.
7
The summary of time spent and corresponding costs for the construction and the rehabilitation activity.
Construction
Labour and equipment
Rehabilitation
Road users Labour and equipment Road users
Item
Time spent (h/km)
Value of time (e/h)
Unit cost of time (e/km)
Transportation Laying asphalt Compacting asphalt User’s time loss Transportation Laying asphalt Compacting asphalt User’s time loss
215 13 133 – 109 13 133 53
40 43 37 – 40 43 37 60 – 500
8604 559 4921 – 4360 559 4921 10,980
industries and 0.015 e/MJ for residual oil as of 2012 (Table 6). Time-related costs were calculated based on the value of time for labour/equipment and road users (Table 7). The value of time has been reflected in terms of 35 e/h average wage for labours and corresponding depreciation regarding equipment from Table 4. The contribution of the main cost drivers into the total cost is presented in Figure 5. It was observed that the corresponding value of crude oil in the bitumen price was about 50% of the total cost for both construction and rehabilitation activity. The amount of energy used corresponding to the fuel oil made an almost 10% contribution to the total cost. The cost contribution related to labour/equipment was shown to be about 15% regarding transportation and 10% regarding compacting for both construction and rehabilitation activity. The total contribution of energy sources which closely follow the crude oil price such as the corresponding value of crude oil in bitumen price, fuel oil and diesel was shown to be about 65% of the total cost of the laid asphalt. The effect of discount rate on the results has been studied in the LCC papers regarding infrastructures (e.g. Amini et al., 2012; Christensen, 2012; Ferreira & Santos,
2012; Safi et al., 2012; Santos & Ferreira, 2013). However, in this study, the discount rate has been expressed in terms of the interest rate and inflation indices. The sensitivity of the EAC to interest rate was studied for a range of assumed crude oil inflation rates where the general inflation rate for the time related cost was assumed to be 3% (Figure 6).
Figure 5. Contribution of the different cost drivers in total cost of the laid asphalt for different asphalt mixtures.
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
8
Figure 6.
I. Mirzadeh et al.
Sensitivity of the EAC to interest rate and the energy price inflation.
Figure 7. Average bidding price of laid asphalt for the winner contractor according to the data from STA (2010) versus calculated cost with the model.
The average crude oil inflation rate for the last 10 years has been about 20% (Figure 1). By assuming a low crude oil inflation rate (3 – 15%), the increase in the rate of interest generally increased the EAC. Assuming medium range (15 – 20%) of crude oil inflation, there was no significant impact on EAC when the interest rate was lower than the crude oil inflation. However, assuming that the crude oil inflation is more than 20%, the EAC had been lowered as the effect of the increased interest rate has been cancelled with the high inflation rate. It was also observed that by assuming the current range of interest rate in Sweden for recent years
Table 8.
Price distribution for different Swedish asphalt mixtures.
Binder content % Void content % Average bidding price (e/tonne)a Price standard deviation a
which has been less than 10%, the assumption of the crude oil inflation rate has highly affected the result (Figure 6). In the current LCC studies, the discount rate for the future maintenance and rehabilitation activities was assumed to be 3 – 4% which corresponds to 7% longterm interest rate and 3% long-term inflation rate (e.g. Chan et al., 2008; Mandapaka et al., 2012; Safi et al., 2012; The University of Chicago Facility Services, 2011). However, more than 50% of rehabilitation cost of the asphalt pavement was shown to be closely related to crude oil price. The calculated discount rate based on 7% interest rate, 3% inflation rate and 20% inflation for the price of oil products for the last 20 years was 2 4%. This means that the future cost in this period due to discounting has been more expensive. Distribution of unit prices of laid asphalt pavements in Sweden is a function of different parameters such as corresponding transportation distances for each project, establishment cost and quality of the existing pavement. It was observed that the price distributions of asphalt mixtures had high standard deviations (Table 8). The unit price of different asphalt mixtures is calculated based on the corresponding distances regarding the case study. The results are compared with the average bidding prices for the winner contractor according to the data from STA from 2010 in order to evaluate the model (Figure 7). It was observed that for most of the mixtures, the calculated prices were good representatives of the prices regarding the laid asphalt. However, the price
ABB16
ABB22
ABT11
ABT16
AG16
AG22
5 4 54 31
4.8 3.5 55 35
6.2 2.5 95 49
6 2.5 52 36
5 5 62 37
4.6 5 40 31
Average bidding price regarding the winner contractor according to the data from STA (2010) which are inflated for 2012.
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
Structure and Infrastructure Engineering
Figure 8. Contribution of the different cost drivers in total cost of the laid asphalt for different asphalt mixtures for the triple transportation distances.
regarding the ABT11 was much higher than the calculated price. This could be related to the high amount of ABT11 used in filling the potholes which due to the small amount of work resulted in a higher unit price. The effect of transportation distances on the calculated costs was studied in a sensitivity analysis. A sensitivity analysis was conducted in order to study the effect of transportation distances on the total cost of the laid asphalt. It was observed that by tripling the transportation distances, the calculated cost of laid asphalt for all the studied mixes increased about 35%. Thus, the transportation distances have an important effect on the total cost of laid asphalt mixture. The contribution of the cost drivers had also changed by the increase in the transport distances. Due to the increase in transportation distances, the cost contributions related to labour/ equipment and diesel have increased (Figure 8). Consequently, the contributions of other cost drivers on the total have decreased. However, the contribution of the energy sources which closely follow the oil price fluctuation still was more than 50% for all the analysed mixtures. 4.2
5. Conclusions A transparent framework was presented for asphalt pavement LCC by applying energy and time as a basis for calculations which underlined the main financial risks pertaining to asphalt paving works. By applying the energy and time units as the basis for cost calculations, the framework imposed separate inflation indices on the energy-related costs and the labour/road user-related costs. The framework is capable of reflecting the price trends in the energy sector. Therefore, it can increase the awareness for the future costs related to maintenance and rehabilitation activities. The prices of different energy sources in Europe during the recent years more or less have followed the fluctuation of the oil price. Therefore, the crude oil inflation index was chosen as the inflation index for all energy sources. Furthermore, the country general inflation index was used for the time-related items. The following conclusions can be drawn from this LCC study: . Comparison between the framework results and the
.
.
Analysis regarding the road user costs
The anticipated traffic level for the rehabilitation activity was much less than the work zone capacity. Therefore, the corresponding costs were related to lowering the speed limit. The hourly time value in Sweden has been usually assumed 10– 15 e for passenger cars and 20– 30 e for heavy vehicles (Safi et al., 2012; Wennstrom, 2010). In this study, higher hourly time value was assumed in order to take into account the effect of the traffic disruption on business operations. The amount of user cost related to delay was 25% of the rehabilitation cost (Table 7). The energy loss of the corresponding vehicles was shown to be 3% of the rehabilitation cost. However, the contribution of the user cost would have been much more if the traffic level had exceeded the work zone capacity (Figure 4).
9
.
.
average price of the laid asphalt for different Swedish asphalt mixtures showed that the total calculated costs well reflect the actual purchase price of the corresponding asphalt pavement. The sensitivity analysis regarding the transportation distances showed that the transportation distances have a high impact on the total cost of asphalt pavement. Therefore, it is believed that the transportation distance is one of the most important factors regarding the high variation in the price of laid asphalt pavements in Sweden. The average yearly increase in the crude oil price has been around 20% for the last 20 years (Figure 1). However, the average yearly increase in this period has been around 5% for Va¨gindex and 3% for CPI. Contribution of the oil products was shown to be more than 50% of the total costs regarding construction and rehabilitation of asphalt pavement in Sweden. Therefore, neither Va¨gindex nor CPI has properly reflected the price trend regarding the asphalt pavement construction at the project level. In Sweden, price of oil products is mostly affected by the global economy rather than by the national economy. Moreover, despite the price index of oil products which had a high fluctuation in different time periods, the cost fluctuation related to labour and equipment has been steady and followed the CPI. Therefore, energy- and time-related costs are in this framework treated with different inflation indices in order to perform a better financial risk assessment regarding the future costs. The calculated discount rate based on the suggested framework for the last 20 years was 2 4%, which means that the future costs in this period have been
10
I. Mirzadeh et al. more expensive due to discounting. Therefore, by assuming a similar pattern for the coming future, it is highly beneficial to minimise future costs. This can be done by increasing the material quality to have a better performance regarding cracking and rutting in order to require less rehabilitation in the future. The amount of future user costs regarding the energy used by vehicles can also be decreased by using pavements with a lower rolling resistance.
Notes
Downloaded by [Kungliga Tekniska Hogskola] at 07:33 24 September 2013
1. 2. 3.
Email:
[email protected] Email:
[email protected] Email:
[email protected]
References Amini, A.A., Mashayekhi, M., Ziari, H., & Nobakht, S. (2012). Life cycle cost comparison of highways with perpetual and conventional pavements. International Journal of Pavement Engineering, 13, 553– 568. Ang, A., & De Leon, D. (2005). Modeling and analysis of uncertainties for risk-informed decisions in infrastructures engineering. Structure and Infrastructure Engineering, 1, 19 – 31. Athena Institute (2006). A life cycle perspective on concrete and asphalt roadways: Embodied primary energy and global warming potential. Ottawa, ON: Author. Chan, A., Keoleian, G., & Gabler, E. (2008). Evaluation of lifecycle cost analysis practices used by the Michigan Department of Transportation. Journal of Transportation Engineering, 134, 236– 245. Christensen, P. (2012). Infrastructures and life-cycle cost – benefit analysis. Structure and Infrastructure Engineering, 8, 507– 516. Eisenberger, I., & Remer, D. (1977). The role of interest and inflation rates in life-cycle cost analysis (NASA Deep Space Network Progress Rep. No. 42 – 43, pp. 105– 109). Energy Conservation in Road Pavement Design (ECRPD) (2009). Maintenance and utilization WP6 – Life cycle evaluation. Centrum dopravniho vyzkumu, Lisenska, Czech Republic. Federal Highway Administration (FHWA) (2002). Life-cycle cost analysis primer. Washington, DC: Author. Federal Highway Administration (FHWA) (2003). Economic analysis primer. Washington, DC: Author. Ferreira, A., & Santos, J. (2012). Life-cycle cost analysis system for pavement management at project level: Sensitivity analysis to the discount rate. International Journal of Pavement Engineering. doi:10.1080/10298436.2012.719618. Frangopol, D., & Liu, M. (2007). Maintenance and management of civil infrastructure based on condition, safety, optimization, and life-cycle cost. Structure and Infrastructure Engineering, 3, 29 – 41. Gray, P., & Irwin, T. (2003). Exchange rate risk (No. 262). The World Bank Group, Washington, DC, USA. Gullberg, D., Birgisson, B., & Jelagin, D. (2012). Evaluation of a novel calibrated-mechanistic model to design against fracture under Swedish conditions. Road Materials and Pavement Design, 13, 49 – 66.
ISO15686-5 (2008). Building and constructed assets service life planning, Part 5: Life-cycle costing. Stockholm: Swedish Standard Institute. Mandapaka, V., Basheer, I., Sahasi, K., Ullidtz, P., Harvey, J., & Sivaneswaran, N. (2012). Mechanistic-empirical and life cycle cost analysis for optimizing flexible pavement maintenance and rehabilitation. Journal of Transportation Engineering, 138, 625– 633. Michigan Department of Transportation (MDOT) (2012). Pavement design and selection. Lansing, MI: Author. National Renewable Energy Laboratory (2011). U.S. life cycle inventory database. Golden, CO: Author. New Jersey Department of Transportation (NJDOT) (2001). Road user cost manual. Trenton, NJ: Author. Safi, M., Sundquist, H., Karoumi, R., & Racutanu, G. (2012). Development of the Swedish bridge management system by upgrading and expanding the use of LCC. Structure and Infrastructure Engineering, 8, 1 – 11. Santos, J., & Ferreira, A. (2013). Life-cycle cost analysis system for pavement management at project level. International Journal of Pavement Engineering, 14, 71 – 84. Stripple, H. (2001). Life cycle assessment of road, A pilot study for inventory analysis. Goteborg: IVL Swedish Environmental Research Institute. Swedish National Audit Office (NAO) (2010). Cost analysis in large road investments, RiR 2010:25. Stockholm: Author. Swedish Transport Administration (STA) (2010). Bitumen bound layers. Publication number: 2010:093. Borlange: Author. Swedish Transport Administration (STA) (2012). Impact analysis of rehabilitation and reconstruction of roads. Stockholm: Author. Swiss Centre for Life Cycle Inventories (2011). EcoInvent. Dubendorf: Author. The University of Chicago Facility Services (2011). Life cycle cost calculations. Chicago, IL: Author. UK Department of Energy and Climate Change (DECC) (2012). Industrial energy prices, quarterly energy prices. London: Author. Walls, J., & Smith, M.R. (1998). Life cycle cost analysis in pavement design. Washington, DC: FHWA. Wang, J., Birgisson, B., & Roque, R. (2007). Development of a windows-based top-down cracking design tool for Florida based on the energy ratio concept. Transportation Research Record, 2037, 86 – 96. Wang, S., Tiong, R., Ting, S.K., & Ashley, D. (2000). Evaluation and management of foreign exchange and revenue risks in China’s BOT projects. Construction Management and Economics, 18, 197– 207. Weisbrod, G., & Fitzroy, S. (2007). Defining the range of urban congestion impacts on freight and the consequences for business activity. In presentation at the TRB annual conference, Washington, DC, 179: 208– 7. Wennstrom, J. (2010). Estimation of delay costs caused by road works. Department of Science and Technology, University of Linko¨ping, Norrkoping, Sweden. Retrieved from: http://liu. diva-portal.org/smash/get/diva2:436257/FULLTEXT01.pdf Yu, W., & Lo, S. (2007). Time-dependent construction social costs model. Construction Management and Economics, 23, 327– 337. Zhang, H., Keoleian, G., & Lepech, M. (2008). An integrated life cycle assessment and life cycle analysis model for pavement overlay systems. Paper presented at the first international symposium on life-cycle civil engineering, Varenna, Italy, 907– 912.
Paper III Mirzadeh, I., & Birgisson, B. (2013). Accommodating energy price volatility in life-cycle cost analysis of asphalt pavements, accepted for publication in: Journal of Civil Engineering and Management.
Accommodating Energy Price Volatility in Life-Cycle Cost Analysis of Asphalt Pavements Iman Mirzadeh- Bjorn Birgisson Division of Highway and Railway Engineering, KTH- Royal Institute of Technology, SE10044 Stockholm, Sweden
[email protected] [email protected]
•
ABSTRACT.
Energy price is related to more than half of the total life cycle cost of asphalt pavements. Furthermore, the fluctuation related to price of energy has been much higher than the general inflation and interest rate. This makes the energy price inflation an important variable that should be addressed when performing life cycle cost (LCC) studies regarding asphalt pavements. The present value of future costs is highly sensitive to the selected discount rate. Therefore, the choice of the discount rate is the most critical element in LCC analysis during the life time of a project. The objective of the paper is to present a discount rate for asphalt pavement projects as a function of interest rate, general inflation and energy price inflation. The discount rate is defined based on the portion of the energy related costs during the life time of the pavement. Consequently, it can reflect the financial risks related to the energy price in asphalt pavement projects. It is suggested that a discount rate sensitivity analysis for asphalt pavements in Sweden should range between -20 and 30%.
KEYWORDS: Road & Highways, Life Cycle Cost, Pavement Design, Discount Rate
i
Introduction
The price of energy has been identified as one of the most important factors affecting the cost of construction projects (Hastak & Shaked, 2000; Baloi & Price, 2003; Jaafari, 2001). Furthermore, the increase in cost of oil products has been the main cause of rise in highway construction cost (Wilmot & Cheng, 2003). In the mid-2000 highway construction showed a sharp increase in prices (Zhou & Damnjanovic, 2011). Consequently, construction projects in 2006 were valued two times more than similar projects in 1997 (Pandit et al. 2009). This was believed to be related to the dramatic increase in the price of crude oil and its products such as bitumen, fuel oil and transportation fuel. Evidently, a direct correlation between the oil price and the cost of asphalt pavement projects exists (Gallagher & Riggs, 2006; Damnjanovic & Zhou, 2009). In a recent study conducted by the authors the crude oil price was closely related to more than 50% of the cost of asphalt pavement during its life time (Mirzadeh et al. 2013). Different approaches such as real option models (e.g. Garvin & Cheah 2004; Zhao et al. 2004; Chiara et al. 2007; Brandao & Saraiva 2008; Cui et al. 2004; Vassallo 2006), the neural network modelling (e.g. Sonmez & Ontepeli, 2009, Baalousha & Çelik, 2011, Wilmot & Mei 2005; Hegazy & Ayed, 1998; and Sodikov, 2005) and discounted cash flow (DCF) are used for cost evaluation of infrastructure projects. Due to its simplicity, DCF analysis which provides net present value (NPV) of a project is the most common approach among the practitioners (Yeo & Qiu, 2003). Moreover, it has been used by the Swedish transport administration for evaluating the infrastructure projects. The present value of future costs and benefits are highly sensitive to discount rate. Therefore, the choice of the discount rate is the most critical element in any evaluation of costs and benefits during the life time of a project. Often, decision makers are faced with the problem of either investing more initially or saving money for maintenance and rehabilitation. The extent to which discounted future costs exceed the initial costs depends directly on the discount rate. The social discount rate is used for public infrastructure projects. Different approaches such as social rate of time preferences (SRTP), marginal social opportunity cost of capital (SOC), weight average (WA) and shadow price of capital (SPC) have been applied for selecting the social discount rate. Evans and Sezer (2005) estimated discount rates , based on the social rate of time preferences, for countries such as USA, UK, Germany, Japan, France and Australia. Their result ranged between (3-5%) which clearly contradicted with the official discount rate at the time. Azar (2007) suggested that the social discount rate for the US is 5.66%, with a 95% confidence interval ranging from 5.62 to 5.71%. However, Lally (2008) argued that this underestimates the confidence interval on the discount rate primarily through ignoring uncertainty surrounding the expected return on risky assets. Percoco (2008) suggested a discount rate for Italy which was 1.2 to 1.3% lower than the official discount rate (5%). Due to the lack of an agreed approach for selecting the discount rate for the evaluation of the public projects, many transport agencies across and within countries have used different discount rates for their public projects (e.g. Ferreira & Santos 2013). Moreover, the social discount rate reflects the general inflation in each country. However, due to high amount of energy related costs in asphalt pavement projects, the inflation regarding this sector is usually different from the rest of economy. Therefore the use of social discount rate for asphalt pavements can be questioned. The other common method which calculates the discount rate for a specific project is capital asset pricing model (CAPM). The CAPM is widely used in calculating the discount rate regarding public private partnership (PPP) or built operate transfer (BOT) projects (e.g. Ashuri et al. 2012, Yeo & Qiu, 2003, Lee et al. 2009, Fama & French 2004, and Campbell 1993). The CAPM is based on the modern portfolio theory developed by Markowitz (1952). In the CAPM model the expected return is presented in terms of the risk free rate, e.g. interest rate arising from government bonds, plus a risk premium. Therefore, the calculation of the discount rate is a qualitative assessment of how much to increase the discount rate over the risk free rate to account for the projects’ risk (Baker & Fox, 2003). This is means a higher risk regarding the project gives a higher discount rate. Consequently, the NPV decreases due to a higher discount rate. A lower NPV regarding revenues due to 2
higher risk is reasonable. However, a lower NPV regarding cost due to higher risk is not relevant. In order to deal with this problem, the authors used an approach based on the modern portfolio theory to calculate the discount rate. In modern portfolio theory the return on a portfolio is the proportion-weighted combination of the constituent assets' returns (e.g. Sharpe, 1964; Elton et al., 2007; Melicher & Norton, 2010). It was shown that more than 50% of the total LCC cost of asphalt pavement is related to oil products (Mirzadeh et al. 2013). Additionally, the fluctuation related to oil price has been more than the general inflation and interest rate (Figure 1). Consequently, calculation of the discount rate cannot be done based on a homogeneous inflation. However, based on the modern portfolio theory an equivalent discount rate for asphalt pavement projects can be defined as the sum of two separate discount rates.
The objective of the paper is to present a discount rate for asphalt pavement costs as a function of interest rate, general inflation and oil price inflation. The discount rate is defined based on the portion of the energy related costs during the life time of the pavement. It reflects the financial risks related to the costs of asphalt pavement projects. Furthermore, it can be applied for costs calculation regarding both public and PPP projects.
Figure. 1. The yearly average oil price inflation, lending interest rate (World Bank, 2012) and inflation for Sweden.
Methodology
In traditional LCC of asphalt pavements the calculation of the discount rate has been generally based on the interest rate and inflation. However, the fluctuation regarding the energy price has been more than the general inflation. The price of oil products was shown to be related to more than 50% of the costs related to asphalt pavement life cycle. In order to capture the financial risk related to cost of oil products the cost items were divided into energy related and time related items. Energy related items are those affected by the crude oil price such as bitumen, modifier, fuel oil and transportation fuel (Table 1). The price of bitumen and modifiers is expressed as the sum of the costs related to crude oil and production in the refinery. The amount 3
of energy for the transportation was related to the required material distribution from the refinery to the construction site. Regarding the construction site, the agency-related expended energy was attributed to laying, compacting, milling and resurfacing. Furthermore, the user-related expended energy was defined for the extra amount of fuel used by the vehicles in the work zone. The unit cost of energy for each activity can be calculated via Equation 1. Time related costs are defined as those affected by the general inflation such as labor and equipment costs (Table 2). The value of time included the hourly wage of the labors together with the amount of hourly depreciation regarding the equipment. The hourly depreciation was defined as the initial price of the equipment divided by the useful life of the equipment (hours). The calculation regarding the user delay cost has to be done separately for the personal cars and heavy vehicles since they should be treated with different values of time. The road user costs are defined as those related to energy loss and delay for the road users during maintenance and rehabilitation activities. The amount of energy spent (MJ/tonne), time spent (hr. /km), unit price of energy (€/MJ) and unit value of time (€/hr.) for different Swedish mixes are presented by Mirzadeh et al. (2013).
Table 1. Energy related variables
Crude oil (MJ)
Energy source A (MJ)
Energy Cost of source Energy B (MJ) (€)
Bitumen
qc1
-
-
CE1
Modifier
qc2
-
-
CE2
Aggregate production
-
qa3
qb3
CE3
Bitumen production
-
qa4
qb4
CE4
Modifier production
-
qa5
qb5
CE5
Asphalt production
-
qa6
qb6
CE6
Transportation
-
qa7
qb7
CE7
Laying asphalt
-
qa8
qb8
CE8
Compacting asphalt
-
qa9
qb9
CE9
User's energy loss
-
qa10
qb10
CE10
n
= ∑ (qck × Pc + qak × Pa + qbk × Pb + ...) CE
(1)
k =1
Where, CE is the unit cost of energy related items for a construction/rehabilitation activity. q ck , q ak and q bk are the quantities (MJ) regarding crude oil, energy source A and energy source B. P c , P a and P b are the unit prices (€/MJ) regarding crude oil, energy source A and energy source B at the base year. n
= CT ∑ Vk × tk
(2)
k =1
4
Where, CT is the unit cost of time related items for a construction/rehabilitation activity. t k is the amount of time spent for each item. V k is the value of time for each item. Table 2. Time related variables
Labor & equipment
Road users
Item
Time Spent (Hr.)
Transportation
t1
V1
CT1
Laying asphalt
t2
V2
CT2
t3
V3
CT3
t4
V4
CT4
t5
Vu
CT5
Compacting asphalt Milling and resurfacing User's delay cost
Value Cost of time of time (€/Hr.) (€)
The most common method used to compare past and future cash flow with those of today is the NPV method. Costs regarding initial construction, maintenance and rehabilitation occur at different times. Therefore, it is essential to use a discount rate in order to reflect the time value of money. The time value of money results from inflation and opportunity cost. Inflation is the rise in the general level of prices of goods and services in an economy over a period of time. Opportunity cost is equivalent to the benefit, the cash could have achieved, had it been spent differently or invested. The NPV for a construction/rehabilitation cost (C) at the yth year can be calculated form Equation 3 based on the discount rate (r) (FHWA 2002). 1 NPV= C × 1+ r
y
(3)
The traditional discount rate (r) is calculated according to Churchill & Panesar (2013): = r
1+ i −1 1+ j
(4)
Where, i and j are the rate of interest and inflation. Since C consists of energy related costs (CE) and time related costs (CT), the discount rate should reflect inflations both regarding CE and CT. The inflation regarding CE and CT are defined as crude inflation and general inflation respectively. For a portfolio consisting of the investments a and b where portion α of the wealth is placed in a and remainder (1-α) in b the equivalent rate of return and its standard deviation are expressed in Equations 5-6 (Strong, 2008). r = α ⋅ ra + (1 − α ) ⋅ rb
= σ
(5)
α 2σ a2 + (1 − α ) 2 σ b2 + 2 βα (1 − α )σ aσ b
(6)
5
Where, r a and r b are the return on investments and σ a , σ b are their standard deviations. β is the correlation coefficient between the rates of return of the two investments. Since β indicates the degree to which asset’s expected return is correlated with broader market outcomes, it is simply an indicator of an asset’s vulnerability to systematic risk. A positive value for β indicates that there is a positive correlation between r a and r b . A zero value would indicate that r a and r b are completely independent. Furthermore, a negative value indicates that there is an inverse relationship between them.
It was shown by Mirzadeh et al. (2013) that crude oil inflation can be used as the inflation index for the energy related items. Consequently, the crude oil discount rate (r 1 ) can be used as the discount rate for the energy related costs. The general discount rate was chosen as the discount rate for the time related items (r 0 ). Assuming α as the portion of the energy related costs Equations 5-6 can be rewritten as: r = α ⋅ r1 + (1 − α ) ⋅ r0 = σ
(7)
α 2σ 12 + (1 − α ) 2 σ 02 + 2 βα (1 − α )σ 1σ 0
(8)
Where, r is defined as the equivalent discount rate. r 1 is the discount rate for energy related items based on the interest rate and the oil price inflation and σ 1 is its standard deviation. r 0 is the traditional discount rate based on interest rate and general inflation and σ 0 is its corresponding standard deviation. The portion of energy related costs (α) can vary between zero to one. In this context β is chosen to describe the correlation between r 1 and r 0 (Equation 9). β=
Co v(r1 , r0 ) Var (r0 )
(9)
By inserting the interest rate, the general inflation and the oil price inflation Equation 7 can be rewritten as: 1 1 1+ i r = α × (1 + i ) × ( − −1 )+ 1 + jc 1 + j 1 + j
(10)
The NPV based on the equivalent discount rate can be obtained by Equation 3. The NPV of the alternatives are only comparable if they have the same life span. In the case of comparing alternatives with different life span equivalent annual cost (EAC) is often used as a decision support tool. The EAC is the cost per year of owing an asset over its defined life span. The EAC is calculated by multiplying the total net present value (TNPV) by the annuity factor ( At , r ) (Mirzadeh et al. 2013): EAC= TNPV × At , r= TNPV ×
r 1 − (1 + r ) − d
(11)
Where, TNPV is the sum of all the NPVs regarding construction, maintenance and rehabilitation activities during the life span and d is the design life of the asphalt pavement. The portion of energy related cost regarding each project may vary for the initial construction and future maintenance and rehabilitation. Therefore, α should be calculated based on the discounted energy and time related cost. However, the equivalent discount rate itself is a function of α. In order to solve this problem a flowchart for evaluating the discount rate for asphalt pavements is presented in Figure 2. 6
Figure 2. The flowchart to calculate the equivalent discount rate. The design inputs such as the thickness of the asphalt layer for construction are to be obtained from the Calibrated Mechanics based model (MC). The MC model is based on the mechanistic empirical pavement design model developed by Wang, Birgisson & Roque (2007) which is calibrated by Gullberg et al. (2012) for the Swedish conditions. The initial α is calculated based on undiscounted energy related and time related costs. The equivalent discount rate is evaluated as a function of α, general inflation, crude inflation and interest rate. The new α is then calculated based on NPV for energy related and time related items by substituting r into Equation 9. The iteration continues as long as the difference between α i+1 and α is larger than 1%. Once α satisfied the required condition, r can be used as the discount rate for the asphalt pavement project. A sensitivity analysis is to be performed in order to assess the effect of expected variation of the discount rate on the life cycle cost of the project. The sensitivity analysis should be done based on the historical values of interest rate, general inflation and oil price inflation.
7
Results and Discussion The variation of the discount rates for different assumed values for α based on the historical data regarding interest rate, inflation and oil price inflation for Sweden is presented in Figure 3. Expected range of change regarding the equivalent discount rates, has been higher than the traditional discount rate. Depending on the portion of energy related items, volatility of the discount rate has excessively increased. It was observed that during 1981-1994 the Swedish economy faced a relatively high interest rate, inflation and an overall deflation in the oil price. This situation caused the equivalent discount rates to be higher than the traditional discount rate (Figure 4) and beta in this period was 0.2. Beta in this period was positive which means r 1 and r 0 changed in the same direction. However, during 1970-1980 and 1999-2012 the high oil price inflation caused the equivalent discount rate to be negative and β coefficients were -2.3 and -3.7. The negative β coefficients indicate that r 1 was moving in the opposite direction of r 0 . The discount rate in these periods implied a rapid increase in the cost of asphalt pavement construction and rehabilitation which did not correspond to the general inflation.
Figure 3. The average equivalent discount rates based on the historical data for Sweden.
8
Figure 4. The average equivalent discount rates in different time periods. The empirical cumulative distribution function (CDF) of discount rates regarding energy related items (r 1 ) and time related items (r 0 ) during 1970-2012 in Sweden was plotted against the normal CDF in order to test if they follow a normal distribution (Figure 5). The midpoint probability plotting positions was used where the ith sorted values were plotted against the midpoint in the jump of the empirical cumulative distribution function (CDF) on the y axis. It was observed that the historical values regarding the both discount rates have followed normal distributions.
Figure 5. Normal probability distribution plot test for discount rates regarding energy related items (r 1 ) and time related items (r 0 ) for Sweden during 1970-2012. 9
The NPV for the future cost items is the multiplication of their current cost by the discount factor. The discount factor is a function of the discount rate and the year of the analysis (Equation 11). In sensitivity analysis regarding the current LCC studies, the discount rate usually is assumed to be positive (e.g. Ferreira & Santos, 2013, Christensen, 2009, Vacheyroux & Corotis, 2013, Zhou, Banerjee & Shinozuka, 2010, Churchill & Panesar, 2013). However, if the inflation is higher than the interest rate from the Equation 4 the discount rate should be negative. This behavior was observed for the equivalent discount rate in certain time periods such as 1970-1980 and 1999-2012 in Sweden (Figure 4). The high amount of crude oil price inflation caused a negative equivalent discount rate in these periods. The effect of the discount rate on the discount factor is presented in Figure 6.
Figure 6. Discount factor as a function of time and discount rate. A positive discount rate causes the discounted value for the future costs to be lower. However, a negative discount rate makes the value of the discounted costs to be higher. This causes different interpretations of an LCC study. Having a positive value for the discount rate may imply that it is worth to invest less for the initial construction and have a higher level of maintenance and rehabilitation in the future. However, a negative discount rate supports the argument to invest more in the initial construction to lower the amount of the future costs. The cumulative probability distribution function for discount rates with α equal to 0% and 50% based on a Swedish data regarding 1970-2012 is plotted in Figure 7. It was observed that the common range, that is widely used in sensitivity analysis regarding discount rate for infrastructure projects (i.e. 0-8%), is relevant for traditional discount rate (α=0). However, the same probability requires a range between -20% and 30% for α equal to 50%.
10
Figure 7. Normal cumulative distributions for discount rates with α equal to 0% and 50% based on the Swedish data regarding 1970-2012. The normal probability Distribution Function (PDF) for discount rates with different portion of energy related items during 1970-2012 in Sweden is depicted in Figure 8. The probability distribution regarding r 1 with a standard deviation equal to 0.3 was shown to be a wide distribution compare to r 0 with a standard deviation equal to 0.03. The standard deviation for the equivalent discount rate which is a combination of energy and time related discount rates has changed depending on its α. Furthermore, the shape of the probability distribution for the equivalent discount rate has been more similar to r 1 for a higher α and more similar to r 0 for a lower α. This volatility has imposed a high level of financial risk to the highway projects. This financial risk is primarily subjected to the longer projects. On the other hand, it also exists during the contract time for construction contracts. The length of the contract highly affects this risk. The longer the contract the more significant this risk becomes. If the contract duration is short the high volatility should not affect the prices as the contractor has the material/fuel available before the beginning of the contract. However, the impact of volatility can be significant even for modest volatility measures for long duration contracts. The sharp increase in the level of material and fuel prices has been experienced during the last years. Consequently, a large number of contractors in U.S. were affected by escalating material prices (bitumen, fuel) (Gallagher & Riggs, 2006). According to the results of a survey conducted by AASHTO in 2008, 40 states used fuel price adjustment clauses, and 42 states used bitumen price adjustment clauses. However, even when transportation agencies use price adjustments, the trigger levels are only arbitrarily defined and are not considered from the risk analysis perspective. For example, the Washington state department of transportation (DOT) applied the trigger value of 10% for fuel and bitumen cost adjustments (AASHTO, 2009). By considering the high level of risks (Figure 8) the amount of the adjustments might not be sufficient.
11
Figure 8. Normal distributions for discount rates with different portions of energy related items during 19702012. To hedge against this risk, especially for contracts without price adjustment clauses, contractors usually have incorporated a premium in the bid price (Damnjanovic & Zhou, 2009). Contractors can allocate this risk in four following ways: (1) adding the risk to the price of individual items in the contract, (2) adding the risk to the overall estimated price of the project, (3) dividing the risk and spreading it on some particular items in a bid and (4) spreading the risk on a portfolio of projects. Contractors in longer contracts are more concerned about the future prices. Therefore, they usually get engaged in hedging against commodity (material/fuel) risk with suppliers. The contractor risk premium, especially for smaller contractors, might be larger than the suppliers’ premium. In this case the transportation agencies could consider including the price adjustment clauses to cover the risk of the future price increases. This can be done by hedging with purchasing options for the future consumption of the fuel and material at the predetermined fixed prices. Furthermore, this will encourage more contractors to take part in the bidding and consequently could result in overall lower bidding prices.
The equivalent discount rate reflected the financial risk regarding the energy price volatility. This information can be used by the road authorities in the context of managing risks on an individual project level as well as the network level. At the project level results provide a basis for making decisions to either retain or transfer the commodity (material/fuel) price risk. For example having high values regarding α and the expected change in the energy price the road agency should consider price adjustment clauses. On the network level, the results provide a measure for minimizing portfolio risks by diversifying risk retention and transfer policies. For example, exposing only a portion of projects to risk by implementing price adjustment clauses, the portfolio risk can be minimized. For instance, an increase in the energy related prices causes only a part of the projects to be subjected to losses that are compensated by gains on the rest of the portfolio.
12
Summary and conclusions
This paper presented a discount rate for asphalt pavement projects as a function of interest rate, general inflation and crude inflation which can be defined for each project based on the portion of the energy related items during the defined life time. This so called “equivalent discount rate” reflected the impact of energy price fluctuation on the costs and the associated financial risks related to asphalt pavement works. A flowchart was suggested for calculating the discount rate for an asphalt pavement project. The equivalent discount rate was studied by analyzing the historical values regarding the interest rate, general inflation and oil price inflation in Sweden during 1970-2012. The range of α was assumed to be between 50%-70%. In periods between 1970-1980 and 1999-2012 the average value for discount rate has been -4% and -3% respectively. The beta coefficient in these periods has been -2.3 and -3.6. The negative beta was related to high oil price inflation which changed the behavior of the equivalent discount rate. The standard deviation of the equivalent discount rate was 0.15 and 0.20 for α equal to 50% and 70% respectively. This was much higher than the standard deviation for the traditional discount rate in the same period. The high volatility for the discount rate was caused by the high volatility in the oil price. The following conclusions are drawn from this study:
• By evaluating the historical values regarding inflation, oil price inflation and interest rate, during specific time periods, the equivalent discount rate was shown to be much lower than the traditional discount rate (α=0). Therefore, applying an average traditional discount rate may lead to under-estimation of the project costs. • The lower discount rate, specially the negative discount rate in the recent years, implies that the future costs can be more expensive than projected due to discounting. This highlights the importance of lowering the amount of required maintenance and rehabilitation by increasing the initial construction quality. • It was observed that the common range (i.e. 0-8%) for discount rate sensitivity analysis regarding infrastructure projects is not suitable for highway projects. By assuming 50% of the costs related to the energy related items, the discount rate sensitivity analysis should cover a range between -20% and +30%. • Both the equivalent discount rate and its volatility are dependent on the amount of contribution of the energy related costs. Therefore, the discount rate should be defined based on α for each highway project. • The high amount of volatility in the oil price imposed increased financial risk to the highway projects. This financial risk may highly affect the fixed-price unit-based contracts with longer time period which the owner transferred the risk to the contractor. • By incorporating the financial risks imposed by the energy price, this paper helps the contractors and road authorities with a more transparent estimation of the total LCC cost and the associated risks related to asphalt pavement projects.
13
References
AASHTO (2009). AASHTO Subcommittee on Construction. Survey on the Use of Price Adjustment Clauses. http://fhwicsint01.fhwa. dot.gov/programadmin/contracts/aashto.cfm. Ashuri, B., Kashani, H., Molenaar, R., Lee, J., Lu, J. (2012). Journal of Construction Engineering and Management, 138:545.557. Azar, S. A. (2007). Measuring the US social discount rate. Applied Financial Economics Letters, 3(1), 63-66. Baalousha, Y., & Çelik, T. (2011). An integrated web-based data warehouse and artificial neural networks system for unit price analysis with inflation adjustment. Journal of Civil Engineering and Management, 17(2), 157-167. Baker, R. and Fox, R. (2003) Capital investment appraisal: a new risk premium model. International Transactions in Operational Research, 10, 115–26. Baloi, D., & Price, A. D. (2003). Modelling global risk factors affecting construction cost performance. International Journal of Project Management, 21(4), 261-269. Brandao, L. E. T., & Saraiva, E. (2008). The option value of government guarantees in infrastructure projects. Construction Management and Economics, 26(11), 1171-1180. Campbell, John Y. (1993). Intertemporal Asset Pricing without Consumption Data. American Economic Review 83(3): 487-512. Chiara, N., Garvin, M., & Vecer, J. (2007). Valuing Simple Multiple-Exercise Real Options in Infrastructure Projects. Journal of Infrastructure Systems, 13:2, 97-104. Christensen, P. (2009). Life-cycle cost-benefit (LCCB) analysis of bridges from a user and social point of view, Structure and Infrastructure Engineering, 5:1, 49-57, DOI: 10.1080/15732470701322818. Churchill, C. & Panesar, D. (2013) Life-cycle cost analysis of highway noise barriers designed with photocatalytic cement, Structure and Infrastructure Engineering, 9:10, 983-998, DOI: 10.1080/15732479.2011.653574. Cui, Q., Bayraktar M., Hastak, M., Minkarah, I. (2004). Use of Warranties on Highway Projects: A Real Option Perspective. Journal of Management in Engineering, 20:3, 118-125. Damnjanovic, I., & Zhou, X. (2009). Impact of crude oil market behavior on unit bid prices: the evidence from the highway construction sector. Construction Management and Economics, 27(9), 881-890. Elton, E. J., Gruber, M. J., Brown, S. J., & Goetzmann, W. N. (2007), Modern Portfolio Theory and Investment Analysis (7th ed.). New York: John Wiley & Sons.
14
Evans, D. J., & Sezer, H. (2005). Social discount rates for member countries of the European Union. Journal of Economic Studies, 32(1), 47-59. Fama, E., & French K. (2004). The Capital Asset Pricing Model: The Capital Asset Pricing Model: Theory and Evidence. Journal of Economic Perspectives, 18:3, 25-46. Ferreira, A., & Santos, J. (2013). Life-cycle cost analysis system for pavement management at project level: Sensitivity analysis to the discount rate. International Journal of Pavement Engineering. doi:10.1080/10298436.2012.719618. Federal Highway Administration (FHWA) (2002). Life-cycle cost analysis primer. Washington, DC. Gallagher, J. and Riggs, F. (2006) Material price escalation: allocating the risks. Construction Briefings, Thomson/West, Eagan. Garvin, M., and Cheah, C. (2004). Valuation technics for infrastructure investment decisions. Construction Management and Economics, 22, 373-383. Gullberg, D., Birgisson, B., & Jelagin, D. (2012). Evaluation of a novel calibrated-mechanistic model to design against fracture under Swedish condition. International Journal on Road Material and Pavement Design. Hastak, M., & Shaked, A. (2000). ICRAM-1: Model for international construction risk assessment. Journal of Management in Engineering, 16(1), 59-69. Hegazy, T., Ayed, A. (1998). Neural Network Model for Parametric Cost Estimation of Highway projects. Journal of Construction Engineering and Management, 124:3, 210-218. Jaafari, A. (2001). Management of risks, uncertainties and opportunities on projects: time for a fundamental shift. International journal of project management, 19(2), 89-101. Lally, M. (2008). Measuring the US social discount rate: reply to Azar. Applied Financial Economics Letters, 4(4), 283-285. Lee, H., Lee, S., Lee, J., Chon, k. (2009). The Revenue-Beta Model for Private Participation in Infrastructure. Journal of the Eastern Asia Society for Transportation Studies, 8. Markowitz, Harry. (1952). Portfolio Selection. Journal of Finance. 7:1, pp. 77–99. Melicher, R. W. & Norton, E. A. (2010). Introduction to Finance: Markets, Investments, and Financial Management (14th ed.). Danvers: John Wiley & Sons. Mirzadeh, I., Butt, A., Toller, S., & Birgisson, B. (2013). Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements, Structure and Infrastructure Engineering. doi:10.1080/15732479.2013.837494. Pandit, D., Anderson, S., Damnjanovic, I., & Nejat, A. (2009). Assessment of strategies and methods to reduce construction costs while maintaining Quality. In Proceedings of 1st International Conference on Transportation Construction Management, Orlando, FL. Percoco, M. (2008). A social discount rate for Italy. Applied Economics Letters, 15(1), 73-77. 15
Sharpe, W. (1964). Capital asset prices: a theory of market equilibrium under conditions of risk. The Journal of finance. 3:8, 425-442. Sodikov, J. (2005). Cost Estimation of Highway Projects in developing Countries: Artificial Neural Network Approach. Journal of the Eastern Asia Society for Transportation Studies, 6, 1036-1047. Sonmez, R., & Ontepeli, B. (2009). Predesign cost estimation of urban railway projects with parametric modeling. Journal of Civil Engineering and management, 15(4), 405-409. Strong, R. A. (2008). Portfolio Construction, Management, and Protection [With Stock-Trak]. CengageBrain. com. Vacheyroux, G., & Corotis, R. (2013) Strategies of investment in the management of urban bridges: a life-cycle approach illustrated for Paris, Structure and Infrastructure Engineering, 9:11, 1080-1093, DOI: 10.1080/15732479.2012.665464. Vassallo, J. (2006). Traffic Risk Mitigation in Highway Concession Projects. Journal of Transport Economics and Policy, 40:3, 359–381. Wang, J., Birgisson, B., & Roque, R. (2007). Implementation of the Florida Cracking Model into the AASHTO 2002 Mechanistic -Empirical Pavement Design. Journal of Transportation Research Record, 8696. Wilmot, C. G., & Cheng, G. (2003). Estimating future highway construction costs. Journal of Construction Engineering and Management, 129(3), 272-279. Wilmot, C., & Mei, B. (2005). Neural Network Modeling of Highway Construction Costs. Journal of Construction Engineering and Management, 131:7, 765-771. World Bank. Development Data Group (Ed.). (2012). World Development Indicators 2012. World Bank-free PDF. Yeo, K., & Qiu, F. (2003). The value of management flexibility: a real option approach to investment evaluation. International Journal of Project Management, 21, 243-250. Zhao, T., Sundararajan, S., Tseng, C. (2004). Highway Development Decision –Making under Uncertainty: A Real Options Approach. Journal of Infrastructure Systems, 10:1, 23:32. Zhou, Y., Banerjee, S., & Shinozuka, M. (2010) Socio-economic effect of seismic retrofit of bridges for highway transportation networks: a pilot study, Structure and Infrastructure Engineering, 6:1-2, 145-157, DOI: 10.1080/15732470802663862.N Zhou, X., & Damnjanovic, I. D. (2011). Optimal Hedging of Commodity Price Risks in Highway Contracts. Transportation Research Record: Journal of the Transportation Research Board, 2228(1), 19-25.
16
Paper IV Mirzadeh, I., & Birgisson, B. (2014). Evaluation of Highway Projects under Price Adjustment Clauses Based on an Option Pricing Framework, submitted for publication in: Journal of Construction Engineering and Management.
Evaluation of Highway Projects under Price Adjustment Clauses Based on an Option Pricing Framework Iman Mirzadeha- Bjorn Birgissona b a
Division of Highway and Railway Engineering, KTH- Royal Institute of Technology, SE10044 Stockholm, Sweden b School of Engineering and Applied Science, Aston University, Aston Triangle Birmingham, B4 7ET, UK
[email protected] [email protected]
•
ABSTRACT.
Different features included in highway projects such as limited liability of the PPP firm and government supports mechanisms (e.g. price adjustment clauses (PACs)) result in asymmetric project value payoff functions which cannot be properly assessed by traditional methods such as Discounted Cash Flow (DCF) analysis. To overcome this issue, an option pricing framework is developed which enables financial assessment of different types of road projects in the presence of different government support mechanisms. However, the paper has focused on evaluation of highway projects under the presence of PACs. By providing a framework for calculating the value of government support mechanisms and long term material/energy contracts this model can help road agencies and contractors with choosing effective hedging strategies.
KEYWORDS: Road & Highways, Option pricing, Public-private-partnership, Government supports
i
1. Introduction The Swedish Transport Administration (STA) is one of the largest public sector clients in Sweden, purchasing goods and services to an annual amount of nearly SEK 40 billion. The procurement process is the transport administration’s frequent key activity. Moreover, the Swedish parliament has urged the STA to increase the efficiency and improve the conditions for effective competition in procurement process. In order to develop better solutions and increase the competition for large infrastructure projects, the STA has intended to transfer a large part of responsibilities to contractors. The design-build projects with long-term maintenance responsibility (Totalentreprenader), which as of 2014 was 34% of the contracts, are to be increased to 50% of the projects by 2018. However, based on the NAO audit (NAO, 2012) implementation of the design-build contracts has not led to more value of money, despite the expectation of the Parliament, government and the STA. In order to hedge against the financial risks in projects with longer responsibility periods, contractors usually have incorporated a premium in bid prices (Damnjanovic & Zhou, 2009). Therefore, effective reduction regarding these premiums is equivalent to more value of money from the tax payers’ perspective. This can be achieved either by lowering the risks or by applying more effective hedging strategies. Transferring responsibilities and therefore risks to the contractors is been used as a mechanism for risk reduction by transport administrations. However, this will only be economically beneficial if the value of risks counterweights the risk premiums. Currently, due to extensive application of PACs (for bitumen) in Sweden, the financial risk regarding the price of commodities is highly imposed to the STA. Therefore, a proper evaluation of this financial risk should be of a great interest for the transport administration. Financial risks in road projects are shared between the transport administration and contractors based on the contract type and the level of governmental supports (e.g. PACs and revenue guarantees). These risks are substantial in projects with long contract periods since market and political conditions may change during the construction and operation period. The financial risk during the construction period is mainly related to material price escalations whereas the financial risk during operation can be related to the revenue stream. Uncertainty regarding future traffic demands is identified as the main risk regarding toll road projects (e.g. Ho and Liu 2002, Zhao et al. 2004, Garvin & Cheah 2004, Chiara et al. 2007, Brandao & Saraiva 2008, Cuttaree 2008). It is evident that improper consideration of uncertainties could lead to financial failure of public private partnership (PPP) projects (Li & Zou, 2011). Facilities such as English Chanel Tunnel or Tokyo Bay Bridge are examples of unsuccessful PPP projects due to underestimation of the uncertainties (De Neufville et al. 2003, Flyvbjerg et al. 2003). The most common approach for evaluating highway projects is discounted cash flow (DFC) analysis (Cheah & Garvin 2009). In this approach the annual net cash flows are discounted back to the beginning of the project by implementing a discount rate. The other common method for evaluating highway projects is the capital asset pricing model (CAPM) which is based on the modern portfolio theory suggested by Markowitz (1952) and further developed by (Yeo & Qiu, 2003, Lee et al. 2009, Fama & French 2004 and Campbell 1993). The discount rate in this model is presented as the sum of the risk free rate and a risk premium. Hence, the determination of the discount rate is an assessment of how much to increase the risk free rate to account for all the corresponding risks in the project (Baker & Fox, 2003). However, there are certain problems regarding the application of traditional approaches such as DCF and CAPM for the highway projects: (1) in CAPM method a higher risk in the project results in a higher discount rate which leads to a lower project value. Whereas a lower value regarding revenues is reasonable, a lower value regarding cost due to higher risk is not relevant. (2) The DCF 2
and the CAPM methods do not explicitly capture the uncertainties regarding costs and revenues during the operation phase of the project (Ashuri et al. 2012). (3) Certain features of highway projects such as limited liability of the PPP firm, government revenue guarantees, price adjustment clauses (PACs) and long term supply contract with material/energy suppliers result in contingent cash flows. These contingent cash flows create asymmetric payoffs, which cannot be assessed by traditional methods such as DFC or CAPM. Options are a class of derivatives with asymmetric payoffs. It is suggested that the option pricing method can be applied in proper assessment of different infrastructure projects’ features, such as financial viability (Ho & Liu, 2002), government revenue guarantees (Chiara, Garvin & Vecer, 2007; Cheah & Liu, 2006; Galera & Solino, 2010), enhancing the contracts’ risk allocation (Quiggin, 2005; Nombela & Rus, 2004) and managerial flexibility (Majd & Pindyck, 1986; Zhao & Tseng, 2003; Yiu & Tam, 2006). Despite the high interest among researchers real option analysis has not been widely used by highway owners and construction practitioners. Based on a survey of Fortune 1000 companies in the US, it was observed that 14.3% were using real options (Block, 2007). A similar survey of Canadian firms, reported that only 16.8% of the firms implemented real options (Baker, 2011). The probable reason is that the financial principles are not fully applicable in engineering practice (Neufville et al. 2006). For example, pricing of a financial option deals with statistical data such as the asset’s volatility that, while meaningful in finance, have no clear equivalent in engineering. Moreover, the current use of real options has focused on the project’s upside potential (i.e. managerial flexibility) (e.g. Trigeorgis, 1996; Triantis, 2003; Chiara and Kokkaew, 2009). However, evaluating a project’s downside potential (i.e. risks) is more consistent with the risk-averse nature of the investors (Espinoza & Morris, 2013). In this aspect, studies such as evaluating revenue guarantees (e.g. Ashuri et al. 2012) and long term contracts for material procurement (e.g. Ng, Björnsson & Chiu, 2004) can be of a great interest from the practitioners’ perspective. In the last decades real option models have been implemented for assessment of some highway projects’ features, such as financial viability and government revenue guarantees. However, to the authors’ knowledge, the effect of price adjustment clauses on the project’s value has not been studied. The aim of this paper is to refine the existing option pricing techniques to enable valuation of different types of highway projects in the presence of price adjustment clauses and long term contract with suppliers.
2. Framework Highway contracts are arrangements in which a private party commits to provide a range of services (e.g. design, construction, finance, maintenance and operation). In exchange for the provided service, the private party might receive either a share of tolls on the highway for limited period of time or simply the fixed contract price. Based on the amount of activities and risks assumed by the private party, highway contracts are classified into traditional design-bid-build, design-build, design-buildfinance and design-build-finance-operate (FHWA, 2010). In tradition public contracts (design-bidbuild and design-build) the project costs regarding the initial construction and future maintenance and rehabilitations (M&R) are compensated by the road agency. The costs regarding the contract items are listed in the tender document in which the total estimated cost plus the expected profit is presented as the bid price. However, the material/fuel cost fluctuations usually are compensated according to price adjustment clauses (PACs). Design-build-finance-operate projects could have different compensation means such as availability payments and pass-through tolls. In toll roads the concessionaire receives 3
its revenue directly from the road users by collecting tolls. However, in contracts with availability payments the concessionaire recieves its revenue from the road authority based on particular project milestones (e.g. delivering the project within a defined deadline) or performance standards. The financing arrangement is one of the most important characteristics of a design-build-finance-operate project. These projects are usually financed by forming a leveraged PPP firm which is independent from the projects developers. Different parties involved in these projects’ financing deals include promoters, construction contractors, road authority, lenders and insurance providers (Figure 1b). The government and the developers are the key stakeholders of the project. The equity value is one of the most important indicators in financial assessment of a design-build-finance-operate project from both government and developers’ perspectives. The bankruptcy condition of a PPP firm is determined based on the asset value and debt value of the firm.
Lenders Debt service
Debt
Const. Costs
Road Agency
Const. Costs M&R Costs
Contractors/ Project Developers
Road Agency
PACs
Tolls
Tax
M&R Costs Guarantees
PPP firm
PACs
Tax
Levy
Levy
Tolls
Yield
Equity
Users / Tax Payers
Users / Tax Payers
Sponsors/ Investors
Figure 1. (a) Financial arrangement of traditional design-bid-build contracts; (b) Financial arrangement of design-build-finance-operate projects. When exercising a financial call option, the owner of the option purchases the underlying asset. However, in PPP projects, the exercise of an option is analogous to the bankruptcy or the completion of the project. In PPP projects in order to protect the equity holders the bankruptcy condition is determined by the lenders. Generally, the bankruptcy condition can be triggered if the borrower is unable to meet the repayment schedules. However, since there are no revenues before the completion of the project, the repayments might be delayed until the project completion. Consequently, in the absence of any other bankruptcy condition, the project will not be bankrupted in the construction phase. Nevertheless, in large projects with long construction periods, lenders may impose other conditions to trigger the bankruptcy during the construction period (e.g. the upper limit of cost overrun). If the condition is triggered, unless the developer can justify the cost overrun or arrange other funding sources (e.g. government rescues or new equity injection), the project will be bankrupted. For example, in credit agreement of Channel Tunnel project the project value (R) and the debt value (D) were estimated by lenders to ensure that the project value is greater than the debt value 4
at any time during the concession period (Ho, 2001). The project value represents the total discounted amount of revenues in the operation phase of the project, whereas the debt value reflects the summation of discounted construction and M&R costs (Figure 2). It should be noted that the project value in traditional projects is the bidding price for the awarded contract together with any other governmental reimbursements during which occur during the contractor’s responsibility period.
Figure 2. The project cash flow from the construction to the end of concession period. Based on a discrete binomial option pricing model (Cox et al. 1979), Ho & Liu, 2002 suggested an option valuation framework for two risk variables. The first node of the binomial pyramid represents the discounted project value (R) and debt value (D). After a time increment four nodes will be generated from each node in the previous layer. The new values regarding R and D at each node should be obtained by applying the jump amplitudes as shown in the Figure 3. The jump amplitudes regarding project and debt values should be obtained by applying a fixed pseudo probability, p = 0.5 (Hull, 1997) (Equations 1-4).
= uR exp[(r − 0,5σ R 2 )∆t + σ R ∆t ]
[1]
= d R exp[(r − 0,5σ R 2 )∆t − σ R ∆t ]
[2]
= uC exp[(r − 0,5σ C 2 )∆t + σ C ∆t ]
[3]
= dC exp[(r − 0,5σ C 2 )∆t − σ C ∆t ]
[4]
where σ R and σ C are assumed to be the volatilities regarding revenues and construction costs.
5
Figure 3. A two-step Binomial pyramid for calculating the project value The total discounted costs and revenues at the first node can be obtained by Equation 5-7. C0 =
tE
T
∑ (e( g −r )i ⋅ CC) + ∑ (e( g −r )i ⋅ MC) c
[5]
c
=i t0=i T
= R0
tE
∑ e −(WACC − g
R
)i
⋅ RI
[6]
i =T
WACC = (
E D × Re ) + ( × Rd ) × (1 − T ) E+D E+D
[7]
where CC and MC are the yearly construction and M&R costs based on the prices at the first construction year; RI is the expected revenue at the first year; T is the construction time, g C and g R are the yearly growth rates regarding costs and revenues; r is the risk free rate of return; E is the market value of firm’s equity; D is the market value of the firm’s debt; R e is the cost of equity; R d is the cost of debt; T is the tax rate and WACC is the weighted average cost of capital regarding PPP projects. For projects’ financing schemes, in which the developers do not take part in the project finance, WACC should be assumed zero.
6
The option payoff represents the project value at the time of exercising the option. In traditional contracts the exercise of the option is generally equivalent to the project completion. However, in Design-build-finance-operate projects the option exercise could also be equivalent to bankruptcy of the PPP firm. The equity payoff of the firm is dependent on the project value (R) and the debt value (D). As long as the project value is greater than the debt value the payoff will be calculated by deducting the debt value from the project value. However, if the project value is lower than the debt value, the firm is considered to be bankrupted. Consequently, due to the limited liability of the equity holder, the equity payoff will be zero (Figure 4a). Therefore, the equity payoff of the PPP firm is identical to the payoff from an American call option (Ho & Li, 2002). Under this bankruptcy threshold, the payoff upon the project completion can be obtained by Equation 8. It should be noted that the debt value is a function of construction, maintenance and operation (M&O) costs. The payoff upon the project completion for traditional contracts, in which the private party is not involved in project’s funding, can be simply obtained by Equation 9. RT ,i − CT ,i ⋅ e rL .T
if
RT ,i − CT ,i ⋅ e rL ⋅T ≥ 0
FT,i =
[8] if
0
RT ,i − CT ,i ⋅ e rL ⋅T < 0
FT,i = RT ,i − CT ,i ⋅ e rL .T
[9]
where R T,i and C T,i are the project value (i.e. discounted revenue) and total cost (i.e. total costs regarding construction and M&O) at the ith node on the last binomial layer and r L is the loan interest rate. The payoff ( Ft-∆t, i ) on the nodes in the previous layer can be found by backward calculations according to Equation 10. Fb
if
Rt-∆t, i - Ct-∆t, i ⋅ et ⋅rL ≥ 0
Ft-∆t, i =
[10] 0
Fb =
1 4er ∆t
if
Rt-∆t, i - Ct-∆t, i ⋅ et ⋅rL < 0
{(1+ρ)×[Ft, i +Ft, i+3 ]+(1-ρ)×[Ft, i+1+Ft, i+2 ]}
[11]
where ρ is the correlation coefficient between the risk variables (i.e. costs and revenues). While a positive value for ρ indicates that there is a positive correlation between the project’s revenues and costs, a zero value would indicate that they are completely independent. The back-calculations should continue to find the payoff on the first node of the pyramid which is equivalent to the equity payoff of the firm. Depending on the type of contract and the degree of governmental supports in adverse circumstances Equation 10 should be modified. For example, if it is costly for the government to replace the developer, even in adverse conditions it might rescue the project. Therefore, it can be assumed that the only possible time for the project bankruptcy is at the end of construction time. Consequently, the payoff on the nodes for the layers before T, can be obtained by back-calculating Fb according to Equation 11.
7
Price volatility of construction material and supplies such as bitumen and fuel can result in substantial difficulties for contractors in preparing realistic bids. Often, bidders cannot receive firm price quotes from material suppliers for the project duration. This may lead to price speculations and inflated bid prices due to possible price increases. Price adjustment clauses (PACs) are contractual mechanisms that allow contractors to be at least partially protected against material or fuel price increases that may occur between the contract award and the execution of the work. Moreover, they should be based on an index which is not prone to manipulations by contractors or material suppliers. It was shown that more than 50% of the total cost of asphalt pavement is related to oil products (Mirzadeh et al. 2013). Additionally, the fluctuation related to oil price has been much more than the general inflation. Consequently, high volatility regarding the energy price has imposed tremendous financial risk to highway projects. This financial risk is shared between contractors and the transport administration. The impact of this financial risk for each party depends on the level of price adjustment clauses offered by the transportation agency. Determination of the value of these price adjustment clauses and their impact on the project financial viability is of great interest for developers, contractors, transport administrations and ultimately tax payers. In order to evaluate financial viability of a project under price adjustment clauses, the corresponding volatility of the project cost needs to be determined. This volatility should reflect the cost fluctuations regarding energy, labor and equipment. Based on modern portfolio theory introduced by Markowitz (1952) the equivalent rate of return and the volatility of a combined investment can be derived. Moreover, it was shown by Mirzadeh et al. (2013) that oil price inflation j c can be used as the inflation index for the energy related items. Consequently, oil price volatility can be used as the volatility regarding the energy items. Therefore, the equivalent inflation and the volatility regarding the cost of highway projects are calculated according to Equations (12-14). j p = α ⋅ j c + (1 − α ) ⋅ j
= σ
β=
[12]
α 2σ o2 + (1 − α ) 2 σ le2 + 2 βα (1 − α )σ oσ le
[13]
Cov (J C ,CPI) Var (CPI)
[14]
where, α is the portion of energy related items, j p is the equivalent inflation index, σ o is oil price volatility, σ le is volatility regarding the cost of labor and equipment based on consumer price index (CPI), β is the correlation coefficient and j is the inflation based on CPI. In the presence of PACs for bitumen or fuel, α should be limited to the portion of energy related costs which are not subjected to PACs. The jump amplitudes regarding debt value (Equations 3-4) should be adjusted replacing σ o by the equivalent volatility (σ) which is calculated from Equation (14). The construction cost is obtained by Equation 15. FC Ct =
Ct FC
if if if
(1 − z ) ⋅ FC ≥ Ct (1 − z ) ⋅ FC ≤ Ct ≤ (1 + z ) ⋅ FC (1 + z ) ⋅ FC < Ct
[15]
Price adjustment clauses usually include a ceiling on upward and downward cost changes. They should be responsive to significant changes in the selected index rather than being triggered by insignificant price fluctuations. Moreover, they should be based on an index which is not prone to manipulations by contractors or material suppliers. The equivalent inflation index (Equation 12) represents the fluctuation regarding the cost of highway projects. Furthermore, it is not susceptible to manipulation by contractors and suppliers. Therefore, it can be used as the index regarding implication 8
of PACs. The ceiling for the upward cost changes is similar to a call option in which the developer has a right, to claim a subsidy from the government. Moreover, the floor option on the downward movements is equivalent to a put option in which the government has a right, but not an obligation to receive the cost difference. Road agencies usually provide some commitments to support the developers against risks involved in future revenues, material/energy prices and currency exchange rate. It is essential for both developers and the government to know the value of these guarantees and their impact on the project value. The uncertainty over the future levels of demand for traffic on the completed road is one of the greatest sources of risk in PPPs, especially toll road projects (Brandao & Saraiva, 2008). A revenue guarantee is a contract in which the government promises to annually pay the revenue shortfall accumulated in the project financial auditing intervals (Chiara & Kokkaew, 2013). The fair price of such contract is the value of n European put option with maturity time at the end of each time step (Dailami et al. 1999). Consequently, the project value ( Rgt ) under the government minimum revenue guarantee (MRG) at each time step can be obtained by Equation 18. In this case Rt in Equations 8-10 should be replaced by Rgt . Rvt + X t (GR − Rvt )
GR − Rvt ≥ 0
if
Rgt =
[18] Rvt
if
GR − Rvt < 0
where Rvt is the total summation of discounted revenues at the time t; X t is the percentage of the revenue shortfall which is to be compensated by the government and GR is the guaranteed level of the revenue. In order to provide a right for the government to claim a portion of excess revenue due to a higher actual traffic demand, MRG options may be combined with traffic revenue cap (TRC). In this case, the developers share the excess revenue with the government once the revenue exceeds a prespecified ceiling level (Mandri-Perrott, 2006). Thus, the combined impact of MRG and TRC contracts should be considered by application of Equation 17. The combination of MRG and TRC options is equivalent to a collar option in which the MRG and TRC act as put and call options. Rvt + X (GR − Rvt ) Rgt =
Rvt Rvt − [ Rvt − (1 + K ) × GR ] × H
if
GR ≥ Rvt GR < Rvt ≤ (1 + K ) × GR
if if
[17]
Rvt > (1 + K ) × GR
where K is the maximum portion of revenue above GR that developers can claim entirely and H is the portion of the revenue above (1 + K ) × GR that developers can entirely claim.
9
Figure 4. (a) The payoff regarding a PAC; (b) the payoff regarding government revenue guarantee The put option provides the developers a protection against lower revenue. However, the call option gives the government a right to collect a portion of revenue beyond a certain level. Line á-b-c-ć in Figure 4b denotes the payoff for the collar option (i.e. the combined MRG and TRC options).
2.3. Application of the framework The risk variables should be determined depending on the type of project. The common risk variables are construction cost, energy price volatility, portion of energy related costs and project value. While in design-build-finance-operate projects the traffic demand can be the most important risk variable, in traditional design-bid-build contracts without price adjustment clauses the construction cost can be the major risk variable. The jump amplitudes regarding asset and debt value should be calculated based on Equations (1-4). Furthermore, total discounted construction costs and revenues at the year zero should be obtained by Equations (5-7). Depending on the contract type and the presence of PACs, the calculated construction cost at each node should be adjusted based on Equation 15. Similarly, the revenue at each binomial node needs to be modified by Equation 17 in order to accommodate the effect of the revenue guarantee. The number of time steps for the binomial pyramid should be chosen considering accuracy and computing time. At the last binomial layer, each node represents a possible scenario regarding the costs and revenues at the end of construction period. While the payoff at the last layer should be calculated based on Equation 8, the payoff on other nodes can be obtained by backward calculation according to Equation 10. Consequently, by computing the payoff for each time layer recursively, the payoff on the last node (i.e. pyramid apex) should be obtained. This payoff for design-build-finance-operate projects represents the equity value of the firm. However, for the other types of contract, it represents the present value of the project.
3. Numerical example This numerical example demonstrates the implementation of the suggested model in the financial valuation of highway projects under price adjustment clauses and government revenue guarantees. The hypothetical case is a part of E18 Highway in Finland which includes 51 km of motorway, 12 km of other public roads and 27 km of private roads. The motorway includes eight interchanges and seven tunnels with the total length of 5 km. The project followed a shadow toll payment process which the 10
payment was based on vehicle volumes on the road segments. The concession period was 18 years, including a construction period of 3 years and a maintenance period of 12 years. The cost is estimated to be 450 MEURO based on the prices at the first year of construction. The developers have invested 100 MEURO into the PPP firm and the rest of the funding is assumed to be a bank loan with 9% interest rate per annum. Moreover, the WACC of the firm is assumed to be 13%. The total value of the project was estimated approximately as 700 MEURO. The project value under different level of government revenue and PACs is studied from the transport administration and the developer’s perspective. Furthermore, the amount of financial risk borne by the road agency under the government guarantees and PACs needs to be clarified. Although here we analyzed the case of government revenue guarantees and PACs, the model is flexible and can be easily extended to include other forms of guarantees regarding loan, exchange rate and equity. The discounted construction cost (C 0 ), assuming 3% yearly cost inflation, is calculated as 443 MEURO. The portion of energy related items for this project is assumed to be 60%. Moreover, the volatility regarding the oil price during 19702013 was about 25%. By assuming 5% volatility regarding the cost of labor and equipment and a beta equal to zero, the project cost volatility was 15% according to Equation 13. Furthermore, the jump amplitudes for the binomial pyramid are calculated as 1.08 and 0.92. Consequently, the project equity value was F1, 1 = 93.5 MEURO. Since the calculated equity value was lower than the invested equity, the investment in the project in the absence of PACs or government guarantees is not feasible from the shareholders’ perspective. However, the project NPV based on a DCF analysis 23.9 MEURO which suggests that project should be undertaken. The main is 700 ⋅ e−WACC ⋅3 - 450 = problem with this result is that the DCF method neglects the uncertainty regarding the project costs and revenues. As discussed previously, the high volatility regarding the price of oil products is the main source of uncertainty for the project costs. The impact of the energy price volatility on the project equity value was shown to be substantial (Figure 5).
Figure 5. The impact of energy price volatility and the portion of energy related costs (α) on the project value
The energy price volatility imposes high financial risk to infrastructure projects. The risk is more substantial for projects with longer contract duration. Nevertheless, it can also be significant for 11
traditional public contracts with longer construction period. For example, many contractors in the U.S were affected by sharp increases in the price of oil products such as fuel and bitumen (Gallagher & Riggs, 2006). Consequently, in order to compensate the contractors for their losses, 40 states used fuel PACs and 42 states implemented bitumen PACs (ASHTO, 2009). Different trigger values have been applied in each state (e.g. the Washington State Department of Transportation (DOT) applied the trigger value of 10% for fuel cost adjustments). However, the trigger levels were determined arbitrarily and were not based on a risk analysis. In Sweden price fluctuations regarding bitumen are entirely reimbursed. PACs are believed to have some positive effects such as increasing the number of bidders, decreasing the bid prices and the number bid withdrawals, enhancing both market stability and reliability in the supply chain and increasing the transparency regarding the contractors profit margin (Newcomb, Lenz & Epps, 2013). On the contrary, they impose high financial risks to the transport administration. The value of the PACs under different threshold is presented in Figure 16a. By raising the uncertainty regarding the energy price, which was shown to be the main deriver of asphalt pavement costs, the value of PAC significantly increased. Similarly, the value of revenue guarantees increased for a higher level of uncertainty regarding the revenues (Figure 6b). However, offering the government support as a collar option balanced the risks and benefits for the two parties and lowered the value of the government support from the transport administration perspective.
Figure 6. (a) Evaluation of PACs with different trigger value (b) Evaluation of revenue guarantees The impact of different PACs on the project cost histogram for the contractors is shown in Figure 7 where the cost histograms are fitted with lognormal distributions. Application of collar PACs was shown to have a substantial impact on limiting the amount of financial risk for the developers regarding the escalation of material/fuel prices. Moreover, decreasing the trigger value for the PAC limited the amount of this financial risk. Nevertheless, by applying PACs the transport administration has transferred a substantial amount of financial risk to itself. This could result in a total project cost which is much higher than the anticipated cost by the agency.
12
Figure 7. The impact of PACs with different trigger values on the project cost histogram (which is fitted with a log-normal distribution) for the developers.
In the absence of price adjustment clauses (PACs), an increase in the material or energy prices could result in extensive losses for contractors. Contractors in projects without PACs, usually include a premium in the bid price to account for the financial risk regarding the material and fuel price inflations (Damnjanovic & Zhou, 2009). They usually allocate this risk in different manners such as adding the risk to prices of certain or all of the contract items or spreading the risk on a portfolio of projects. Moreover, contractors in larger projects sometimes get involved in long term material/fuel contracts with suppliers. The values of these agreements are identical to those of PACs which can be analyzed by the suggested model.
4. Summary and Conclusions
Different features included in highway contracts such as limited liability of the PPP firm, government supports mechanisms (e.g. PACs) result in asymmetric project value payoff functions which cannot be properly assessed by traditional methods such as DCF. To overcome this issue, an option based framework is developed which enables financial assessment of different types of highway projects in the presence of different government support mechanisms. However, the paper has focused on evaluation of highway projects under the presence of PACs. It was observed that fluctuation of energy price has been the main contributor regarding the significant changes in the cost of highway projects. Therefore, the amount of volatility regarding the cost of highway projects was calculated as a function of energy price volatility and the portion of energy related costs. Furthermore, an equivalent inflation index which is not prone to manipulations by either contractors or material suppliers was suggested as the index regarding the application of PACs. From the tax payers perspective the application PACs has some advantages and drawbacks. On the one hand, application of PACs reduces the financial risk profile for developers (or contractors) (Figure 7). This could specially benefit smaller contractors that have larger premiums compare to suppliers and encourage them to take part in the bidding. Consequently, increases in the number of participating contractors in the bidding can potentially decrease bid prices. On the other hand, a financial risk which is substantial for large projects will be transferred to the transport administration and ultimately to the tax payers. However, a risk should be 13
borne by the party best placed to minimize its impacts. The developers (or contractors) and suppliers in large projects may be the best parties which can differ the financial risk regarding the material/fuel cost escalations. They might have the possibility to depot the construction supply at the beginning of the project. Moreover, the developers may have the possibility to lower the financial risk profile (e.g. lowering the portion of energy related items in the initial construction and M&R activities). The following conclusions are drawn from this paper: •
The portion of energy related costs had a significant impact on the total cost of highway projects. Thus, any design strategy (e.g. using bitumen additives and alternative waste material) which can lower the amount of energy related costs can significantly limit highway projects’ risk profile.
•
It was observed that the application of PACs can significantly lower the risk profile regarding the construction cost for the contractors. However, by implementing PACs this financial risk will be imposed to the road administration and ultimately to the tax payers. Therefore, the value of PACs should be considered in decision making process regarding large projects.
•
Contractors and suppliers should be encouraged to get engaged in the application of long term material/fuel supply contracts. These contracts will share the financial risk between contractors and suppliers and eventually can partially replace the PACs. The value of these contracts is identical to PACs which can be assessed by the suggested framework.
•
By providing a framework for calculating the value of government support mechanisms and long term material/energy contracts this model can help road administrations and contractors with choosing effective hedging strategies.
Bibliography
AASHTO (2009). AASHTO Subcommittee on Construction. Survey on the Use of Price Adjustment Clauses. http://fhwicsint01.fhwa. dot.gov/programadmin/contracts/aashto.cfm. Ashuri, B., Kashani, H., Molenaar, K. R., Lee, S., & Lu, J. (2012). Risk-neutral pricing approach for evaluating BOT highway projects with government minimum revenue guarantee options. Journal of Construction Engineering and Management, 138(4), 545-557. Baker, H. K., Dutta, S., & Saadi, S. (2011). Management views on real options in capital budgeting. Journal of applied Finance, 21(1), 18-29. Block, S. (2007). Are “real options” actually used in the real world?. The Engineering Economist, 52(3), 255-267. Brandao, L. E. T., & Saraiva, E. (2008). The option value of government guarantees in infrastructure projects. Construction Management and Economics, 26(11), 1171-1180. 14
Cheah, C., and Garvin, M. J. (2009). “Application of real options in PPP infrastructure projects: Opportunities and challenges.” Policy, finance and management for public-private partnership, A. Akintoye, and M. Beck, eds., Wiley-Blackwell, Chichester, West Sussex, UK. Cheah, C. Y., & Liu, J. (2006). Valuing governmental support in infrastructure projects as real options using Monte Carlo simulation. Construction Management and Economics, 24(5), 545-554. Chiara, N., Garvin, M. J., & Vecer, J. (2007). Valuing simple multiple-exercise real options in infrastructure projects. Journal of Infrastructure Systems, 13(2), 97-104. Chiara, N., & Kokkaew, N. (2009). Risk analysis of contractual flexibility in BOT negotiations: a quantitative approach using risk flexibility theory. International Journal of Engineering and Management, 1(1), 71-79. Chiara, N., & Kokkaew, N. (2013). Alternative to Government Revenue Guarantees: Dynamic Revenue Insurance Contracts. Journal of Infrastructure Systems, 19(3), 287-296. Cox, J. C., Ross, S. A., & Rubinstein, M. (1979). Option pricing: A simplified approach. Journal of financial Economics, 7(3), 229-263. Cruz, C. O., & Marques, R. C. (2013). Flexible contracts to cope with uncertainty in public–private partnerships. International Journal of Project Management, 31(3), 473-483. Dailami, M., Lipkovich, I., & Van Dyck, J. (1999). A computer simulation approach to risk management in infrastructure project finance transactions. POLICY RESEARCH WORKING PAPERS-WORLD BANK WPS. De Neufville, R., Scholtes, S., & Wang, T. (2006). Real options by spreadsheet: parking garage case example. Journal of infrastructure systems, 12(2), 107-111. Espinoza, D., & Morris, J. W. (2013). Decoupled NPV: a simple, improved method to value infrastructure investments. Construction Management and Economics, 31(5), 471-496. Fama, E. F., & French, K. R. (2004). The capital asset pricing model: theory and evidence. Journal of Economic Perspectives, 25-46. Federal Highway Administration (FHWA) (2010). Public-private partnership concessions for highway projects: A primer. Washington, DC. Galera, A. L. L., & Solino, A. S. (2010). A real options approach for the valuation of highway concessions. Transportation Science, 44(3), 416-427. Garvin, M. J., & Ford, D. N. (2012). Real options in infrastructure projects: theory, practice and prospects. Engineering Project Organization Journal, 2(1-2), 97-108. Ho, S. P. (2001) Real options and game theoretical valuation, financing and tendering for investments on build-operate-transfer projects. Ph.D. thesis, University of Illinois at Urbana-Champaign.
15
Ho, S. P. (2006). Model for financial renegotiation in public-private partnership projects and its policy implications: game theoretic view. Journal of Construction Engineering and Management, 132(7), 678-688. Ho, S. P., & Liu, L. Y. (2002). An option pricing-based model for evaluating the financial viability of privatized infrastructure projects. Construction Management & Economics, 20(2), 143-156. Hull, J. C. (1997) Options, Futures, and Other Derivatives, 3rd Edition, Prentice-Hall, Englewood Cliffs, NJ. Jaillet, P., Ronn, E. I., & Tompaidis, S. (2004). Valuation of commodity-based swing options. Management science, 50(7), 909-921. Jeerangsuwan, T., Said, H., Kandil, A., & Ukkusuri, S. (2013). Financial Evaluation for Toll Road Projects Considering Traffic Volume and Serviceability Interactions. Journal of Infrastructure Systems. Lee, H., Lee, S., Lee, J., Chon, k. (2009). The Revenue-Beta Model for Private Participation in Infrastructure. Journal of the Eastern Asia Society for Transportation Studies, 8. Li, J., & Zou, P. X. (2011). Fuzzy AHP-based risk assessment methodology for PPP projects. Journal of Construction Engineering and Management, 137(12), 1205-1209. Majd, S., & Pindyck, R. S. (1987). Time to build, option value, and investment decisions. Journal of financial Economics, 18(1), 7-27. Mandri-Perrott, C. (2006). Mobilizing private capital and management into infrastructure development. The World Bank, Washington, DC. Newcomb, D., Lenz, R., & Epps, J. (2013). Price Adjustment Clauses. Texas Department of Transportation (TxDOT). Research and Technology Implementation Office (RTI). (No. 0-6799) Ng, F. P., Björnsson, H. C., & Chiu, S. S. (2004). Valuing a price cap contract for material procurement as a real option. Construction management and economics, 22(2), 141-150. Nombela, G., & De Rus, G. (2004). Flexible-term contracts for road franchising. Transportation Research Part A: Policy and Practice, 38(3), 163-179. Quiggin, J. (2005). Public–private partnerships: options for improved risk allocation. Australian Economic Review, 38(4), 445-450. Trigeorgis, L. (1996). Real options: Managerial flexibility and strategy in resource allocation. MIT press. Tiong, R. L. (1996). CSFs in competitive tendering and negotiation model for BOT projects. Journal of Construction Engineering and Management, 122(3), 205-211. Yeo, K., & Qiu, F. (2003). The value of management flexibility: a real option approach to investment evaluation. International Journal of Project Management, 21, 243-250. Yiu, C. Y., & Tam, C. S. (2006). Rational underpricing in bidding strategy: a real options model. Construction Management and Economics, 24(5), 475-484. 16
Zhao, T., & Tseng, C. L. (2003). Valuing flexibility in infrastructure expansion. Journal of infrastructure systems, 9(3), 89-97.
17
Paper V Khavassefat, P., Mirzadeh, I.,* & Birgisson, B. (2014). A Life Cycle Cost Approach on Minimization of Roughness-Related Damages on Flexible Pavements, submitted for publication in: Journal of Infrastructure Systems.
*Corresponding author
A Life Cycle Cost Approach on Minimization of Roughness-Related Damages on Flexible Pavements Parisa Khavassefata, Iman Mirzadeha*, Bjorn Birgissona b a
Division of Highway and Railway Engineering, KTH- Royal Institute of Technology, SE10044 Stockholm, Sweden b School of Engineering and Applied Science, Aston University, Aston Triangle Birmingham, B4 7ET, UK
*Corresponding author
•
ABSTRACT.
Due to the surface deterioration the impact form dynamic loads increases gradually during the life span of flexible pavements. A surface deterioration model, based on yearly measurements performed in Swedish road network, has been utilized. The results are discussed for three different case studies with different traffic regimes. It was indicated that the predicted pavement service life decreased considerably when the extra dynamic loads, as a result of pavement surface deterioration, has been taken into account. Furthermore, the effect of performing a preventive maintenance (i.e. thin asphalt layer) has been studied based on an LCC framework. The application of the preventive maintenance was shown to be effective, especially for high deterioration rates.
KEYWORDS: Dynamic loads; Life Cycle Cost; Surface deterioration; Longitudinal Profile; Pavement Design, Maintenance and Rehabilitation
i
1. Introduction
The effect of dynamic loads exerted on the pavement has often been neglected in most of pavement design procedures (e.g. PMS Objekt ( Swedish Transport Administration (STA), 2013)). While the dynamic pavement responses (i.e. displacement, stress and strain) are significantly different from the static responses, it is crucial that dynamic loads be integrated into pavement design procedure (Monismith et al. 1988; Potter et al., 1997, Sun and Kennedy 2002). Due to various reasons (e.g. climate effects and traffic) the surface roughness deteriorates during pavement service life. The road surface roughness directly affects the magnitude of the dynamic loads on the pavement. Thus, a rougher surface amplifies the magnitude of the loads exerted on the pavement. This can decrease the pavement service life and consequently increase the Life Cycle Cost of the road. Deterioration models have been implemented since early times along with pavement management models in order to give the status of the pavement condition in each stage during the structure service life ( e.g. Collop and Cebon, 1995; Kropáč and Múčka, 2008; Liu et al., 1998; Liu, 2000). Collop and Cebon (1995) and Liu (2000) utilised a deterministic approach in order to model the pavement surface deterioration during its service life. In the deterministic approach, the pavement surface unevenness updates after several axles loadings by calculating the permanent deformation at each point on the surface. Once the deterministic road profile unevenness is updated the magnitude of the dynamic loads exerted on the pavement can be calculated accordingly. The pavement community has been keen on using International Roughness Index (IRI) as a single number indicator of road surface status. However it has become evident that IRI has many shortages. Some of the limitations of using IRI are discussed by Múčka and Granlund (2012) and Kropáč and Múčka (2005). For example, the dynamic loads induced at different profiles with identical IRI can be quite different. Moreover IRI does not capture mega-texture (5-50 cm) which affects the ride quality. Based on the random nature of the road surface unevenness (Dods and Robson, 1973), the longitudinal profile can be defined with Power Spectral Density (PSD). Due to the lack of randomness (e.g. existence of joints cracking) the PSD cannot always be a correct statistical description of the road roughness. However, it can be used as an average classification of the pavement roughness (Andren, 2006). Moreover when it comes to vehicle-pavement interaction, the PSD as a more extensive definition of the surface roughness is more suitable parameter to be used for obtaining the exerted dynamic loads on the pavement. Khavaessefat et al. (2014) proposed a deterioration function based on the observed road roughness evolution in the Swedish road network. The deterioration function is derived based on the average gradient of PSD for 60 km road sections in Sweden. Based on the deterioration function the yearly response of the pavement can be obtained. The deterioration function developed by Khavassefat et al. (2014) is used in the current paper for obtaining the yearly pavement response during its service life. The pavement surface deterioration can also have an impact on the vehicle fuel consumption. In a comprehensive study, Zhang et al. (2010) included the effect of pavement roughness on vehicle fuel economy using the calibrated HDM4 fuel consumption model 2
(Zaabar and Chatti, 2010). The model is calibrated with the US conditions and requires the IRI for calculating the fuel consumption. Their method for including the effect of roughness was based on a linear relationship between IRI and vehicle fuel consumption, in which they showed that by increasing IRI from 1.2 to 2.4 m/km, the fuel economy increased from 53.5 L/100 km to 56.0 L/100 km. Furthermore, a life cycle assessment (LCA) study by Wang et al. (2012) evaluated the potential energy savings from treating distressed pavements. They evaluated fuel economy changes induced by changing the rolling resistance. Their results showed that when a life-cycle scope is considered, rehabilitating a rough pavement segment with high traffic volume has a great potential to reduce fuel consumption and GHG emissions. Nevertheless, the potential benefits for lower volume roads take much longer to accrue. In a recent study by Swedish National Road and Transport Research Institute (VTI), the traffic energy change due to the improvement of road surface was evaluated (VTI, 2012). They observed that decreasing IRI for roads with rather high heavy vehicle ratio can lower the energy consumption. It is reported by several studies (e.g. NCRHP, 2014; Newcomb, 2009; Chou et al. 2008; Sandberg et al. 2011) that application of thin overlays can be more favourable compare to thicker dense-graded layers. The thin layers allow pavement managers to overlay more lane-kilometre with the same tonnage. Therefore, thin overlays are often shown to have lower life-cycle costs compare to other pavement maintenance techniques (Newcomb 2009). Moreover, it was observed that thin overlays have improved the road smoothness and the riding quality over a rather long period of time (NCRHP, 2014). For instance, Ohio department of transportation determined that thin overlays can perform as long as 16 years before the roughness values fall to the point they were at before the maintenance activities (Chou et al. 2008). Furthermore, thin asphalt layers are shown to have other benefits such as noise reduction, higher skid resistance (at low and medium speed) and better rut resistance (Sandberg et al. 2011). The increase of roughness during the pavement life intensifies the impact of dynamic loads on the pavement. Therefore, performing preventive maintenance activities by lowering this impact can potentially increase the pavement life and consequently lower the LCC of the project. The benefit of pavement preservations on lowering vehicles’ fuel economy, user cost and environmental impacts has already been addressed in the literature. This paper aims to evaluate the benefit of preventive maintenance activities (i.e. thin asphalt layers) regarding lowering the dynamic impact from the vehicles, increasing the pavement life, decreasing the project LCC and limiting projects’ risk profile. Methodology 2.1. Surface deterioration model Longitudinal road profile is one of the most important measures for pavement management system since it affects the ride quality, safety, and vehicle operation cost to a great extent (FHWA, 1998). The pavement surface roughness increases during its service life which eventually results in higher amplitudes of exerted dynamic loads from the vehicles to the pavement. The important factors affecting surface deterioration according to Kropáč, and Múčka (2008) are listed as follows: 3
1) Surface changes caused by seasonal effects; 2) Longitudinal profile variation as a result of traffic loading; 3) Presence of local obstacles such as potholes, cracks, ruts. In a study by Khavassefat et al. (2014) the general surface roughness evolution trends in Sweden was studied in detail. In the interest of obtaining the general trend of pavement surface deterioration, the displacement PSD of longitudinal profile unevenness of 35 road sections with a total length of approximately 60 km has been analysed. The location of the analysed sections is shown in Figure 1. The analysed road sections are measured by the STA on yearly basis. The data was collected during 11 years (2001-2011). The road surface unevenness is measured by a moving vehicle with a constant speed which collects the data at every 10 centimetres.
Figure 1 : Geographical location of the analysed road sections
The evolution of displacement PSD for each road section is studied by obtaining the yearly gradient of displacement PSD and averaging the gradient for all of the selected road sections. The average gradient of the displacement PSD as a function of wavenumber and wavelength regarding 35 Swedish road sections which are normalized by their amplitudes is shown in Figure 2. In principal the deterioration related to the surface occurs in short wavelengths while the deterioration regarding the pavement structure reflects in longer wavelengths (Kropáč and Múčka, 2008). The general trend in Figure 2 is shown with a red line. It was observed that the deterioration rate of shorter wavelengths is generally higher in comparison with longer wavelengths (Figure 2).
4
Figure 2 : Mean gradient value normalized by the amplitude
A general linear trend regarding evolution of displacement PSD was observed for the analysed data. Based on physical deterioration models, it is a viable argument that an exponential model may be a better physical representative of deterioration process. However, due to frequent maintenance activities on the road network, there was no evidence to support this hypothesis. Therefore, a linear model based on pure observations was considered in this study (cf. Khavassefat et al., 2014). Based on the linear evolution hypothesis Equation 1 was proposed as a prognostic model for pavement surface deterioration as function of wavenumber, k x and long-term time (i.e. years), tl : Sx (k x , = tl ) A(k x ) ⋅ tl + Sx0 (k x ) ,
(1)
where A(k x ) is the observed evolution function as shown in Figure 2 and Sx0 (k x ) is the initial displacement PSD (i.e. at tl = 0 ). Due to the interaction of the vehicle suspension system and the pavement surface unevenness dynamic loads are exerted to pavement structures. In this study a quarter-car model has been used in order to quantify the dynamic loads on the pavement. In the present study the DLC is calculated for a vehicle model with constant moving speed. The effect of varying speed on dynamic loads on the pavement has been studied by Sun and Lou (2007). It was shown by them that acceleration and deceleration of the moving vehicles affects the DLC to a great extent. However in the current study for the sake of simplification the vehicle speed is considered constant. The schematic view of the vehicle model and the input parameters are shown in Figure 3. By ignoring the possible jumps which occurs when the vehicle loses contact with the road surface, the response of the quarter car model can be obtained as follows (Bogsjö et al., 2012):
5
h f (k x ) =
mswl 2 (kt + iwv ct ) kt − ((ks + iwv cs )wv 2 ms / −mswv + ks + iwv cs ) − mtwv 2 + iwv ct mswv 2 ), × (1 + k s − mswv 2 + iwv cs
(2)
where w= kx × v v
Figure 3 : Quarter car model with 2 degrees of freedom (Quarter car data from Cebon (2000))
In this paper the simulation regarding evolution of exerted dynamic loads on the pavement has been performed based on an average longitudinal road profile in Sweden. The average PSD is chosen according to Andrén (2006) with split lines approximation method proposed by Dodds (BSI, 1972) as follows :
C k − w1 for k x < k x 0 Sx (k x ) = x− w 2 for k x ≥ k x 0 C k x
(3)
The discontinuity between the two straight lines is set to 0.82 cycle/m, which according to Andrén (2006), gives the minimum least square residual error for the Swedish road network. The values of the exponents are chosen to be 3 and 1.4 and the value of C is 3e8 which again is the optimal value for the Swedish road network (cf. Andrén, 2006). The displacement PSD of average Swedish roads with the aforementioned parameters is depicted in Figure 4.
6
Figure 4 : Displacement PSD of an average Swedish road section
Based on Equation (1-2), the evolution of Dynamic Load Coefficient (DLC) for an average road section in Sweden is presented in Figure 5. The DLC is calculated as the ratio of Root Mean Square (RMS) value of the dynamic tyre loads to the vehicle static load. The magnitude of DLC is directly related to the vehicle model properties, road roughness and the moving speed of the vehicle. While the surface roughness deteriorates according to the model presented in Equation (1), the vehicle model parameters, are kept constant (see Figure 5). Each coloured line in Figure 5 is related to a moving vehicle with a constant speed. The sensitivity of the initial DLC as well as the rate of change of DLC to the vehicle speed has been shown in Figure 6. One may see that by increasing the speed of the vehicle, both the DLC and the rate of change of DLC increases.
Figure 5 : Evolution of DLC as a result of surface deterioration (The general trend of an average road in Swedish road network)
7
Figure 6 : Evolution of initial DLC (D0) with the speed of moving vehicle
Mechanistic empirical design procedure In the present paper in order to calculate the service life of the pavement, PMS Objekt (STA, 2013) has been used. PMS Objekt is a Mechanistic Empirical (ME) design method calibrated for the Swedish condition and material. In the design procedure, two main modes of failure, i.e. fatigue cracking and sub-base rutting, are considered. In addition to the previously mentioned distress modes, PMS Objekt can estimate the damage due to frost heave. In the model the horizontal tensile strain at the bottom of the asphalt layer is used in order to estimate the number of allowed axle loadings before fatigue cracking failure occurs. In a similar approach, the vertical compressive strain on top of the subgrade is used for calculating the number of allowed axle loadings before rutting failure. The STA suggests the fatigue failure criterion as follows (STA (STA), 2011): N till ,bb ≥ N ekv N till .bb =
365 , ni ∑ i =1 N bb ,i
N bb ,i = f s
(4)
m
2.37 ⋅ 10−12 ⋅ 1.16(1.8Ti + 32)
e 4bb ,i
where: N ekv
Equivalent number of standard axles
N bb ,i Number of allowed standard axles during climate period i
ε bb,i
Horizontal tensile strain at the bottom of AC layer for climate period i
ni
Length of climate period i
T
Pavement temperature in ºC for climate period i
fs
Correction factor for cracked pavement; f s = 1.0 for new constructions
8
Fatigue cracking is usually the critical failure mode according to PMS Objekt. It should be pointed out that top-down cracking (i.e. cracks that initiates on the surface of the pavement and propagate downwards) is more common in Europe in comparison with fatigue cracking and rutting (COST 333, 1999). However, since PMS Objekt is calibrated with field measurements without considering the cause of crack development, top-down cracking is also captured empirically in the model (Gullberg et al., 2012). 2.3 Life Cycle Cost Analysis In traditional LCC studies regarding highway projects the net present value is calculated as the sum of construction costs and the discounted future costs (e.g. maintenance and rehabilitation costs). The discounting regarding the future costs is done by application of a discount rate which is a function of interest rate and inflation. Therefore, the net present value (NPV) regarding a construction or a maintenance activity can be calculated according to (FHWA, 2002): 1 NPV= C × 1+ r = r
y
1+ i −1 1+ j
(5) (6)
where C is the cost of the activity; r is the discount rate; y is the year of application; i is the interest rate and j is the rate of inflation. The sum of NPVs regarding construction, maintenance and rehabilitation activities can be used as an indicator for comparing different alternatives. However, equivalent annual cost (EAC) can be used when comparing alternatives with different life span: = TNPV × EAC
r 1 − (1 + r ) − d
(7)
where TNPV is the Total Net Present Value (i.e. sum of all the NPVs regarding construction and rehabilitation activities) and d is the project life span. The inflation index regarding highway projects is usually based on either the general inflation index or a highway construction index (FHWA, 2003). The calculation of the highway construction index is based on bidding prices regarding the awarded contracts. Moreover, due to the application of price adjustment clauses regarding bitumen in Sweden, the fluctuation of the bitumen price is not reflected in the calculation of the highway construction index. It was shown that fluctuations regarding the price of oil products (e.g. bitumen, fuel oil and transportation fuel) have been higher than those of the general inflation and the highway construction index (Mirzadeh et al. 2013). Therefore, any LCC analysis based on these indices cannot reflect the financial risk regarding asphalt pavement projects from the transport administration and ultimately tax payers’ perspectives. In their LCC framework, Mirzadeh et al. 2013 divided the costs into time related and energy related items. The time related costs were those concerning labour and equipment for construction and rehabilitation activities. Energy related costs were separated into crude oil energy and expended energy. The crude oil 9
energy is the energy stored in the material which represents the value of crude oil in bitumen and additives. The expended energy is the energy which was spent during material production, construction, maintenance and rehabilitations of roads. Furthermore, it was shown that the oil price inflation index can be used as the inflation index regarding energy related costs. Based on the modern portfolio theory introduced by Markowitz (1952) and further developed by (Yeo & Qiu, 2003, Lee et al. 2009, Fama & French 2004) the equivalent inflation regarding asphalt pavement projects (jeq) and its standard deviation σ can be defined as: jeq = α ⋅ jc + (1 − α ) ⋅ j σ =
(8)
α 2σ c2 + (1 − α ) 2 σ l2 + 2 βα (1 − α )σ cσ l
(9)
where jc is the oil price inflation; α is the portion of energy related costs which can vary between zero to one; σc is the standard deviation regarding the oil price; σl is the standard deviation regarding the general inflation. In this context β describes the correlated volatility of jc in relation to j (Equation (10)). β=
Cov (jc , j) Var (j)
(10)
A positive value for β indicates that there is a positive correlation between jc and j. A zero value would indicate that they are completely independent and finally, a negative value indicates that there is an inverse relationship between them. In order to justify the application of a preventive maintenance, the EACs in the presence and the absence of the maintenance activity should be compared. In order for the preventive maintenance to be economically feasible, its benefit in increasing the pavement life should surpass the activity cost. The risk profile regarding cost of the alternatives can be assessed by applying a binomial lattice. In economics, the binomial lattice is a random walk model to capture the uncertainty regarding a variable that changes overtime (Copeland & Antikarov 2001). The financial risk regarding the cost of construction and maintenance activities are evaluated during the construction period (T). The analysis starts at the first node with the total discounted cost (C0) regarding the alternative. After each time step the cost is increased by applying an upward multiple (u) or decreased by application of a downward multiple (d) (Figure 7a). The probability regarding the upward movement is p and the probability for the downward movement is 1-p (Hull 2008). The parameters regarding the binomial lattice are the equivalent standard deviation (σ), the time step (∆t) and the discount rate (r). The jump factors and the probability for the upward movement can be calculated as follows: σ ∆t = u e= u e −σ ,
∆t
= p ,
10
e r ∆t − d u−d
(11)
Figure 7 : (a) Binomial lattice to model cost uncertainty; (b) randomly generated paths along the binomial lattice
The last node can be reached by 0 ≤ m ≤ T upward and 0 ≤ T − m ≤ T downward movements along the binomial lattice. The probability for having exact m upward movement on the last node for n number of time steps can be calculated by the probability mass function: T −m Pr (total cos t at the end of analysis period =C0 × u m d= )
n! p m (1 − p )T − m (12) k !(n − k ) !
By calculating the probability regarding each node on the last layer the probability density function (PDF) for the cost of each alternative can be obtained. Case study In order to account for the effect of dynamic loads on the pavement service life three different road sections with different traffic regimes have been analysed. The pavement structure was designed for the scenarios listed in Table 1 using PMS Objekt 5.0 (STA, 2013).The structure for all three cases consisted of 50 mm wearing course above a layer of binder course and a 500 mm granular unbound foundation layer. The wearing course was a Dense Graded Asphalt mixture (ABT 11) with the maximum aggregate size of 11 mm, and the binder course was a Gravel Asphalt (AG 22) with the maximum aggregate size of 22 mm, (Gullberg et al., 2012). The preventive maintenance activity consisted of 10 mm milling of the existing pavement and applying a 20 mm overlay (ABT 11). The required thickness of the binder course is obtained by the ME model for a design period of 25 years. It has to be pointed out that for calculating the required thickness of the binder course, PMS Objekt removes 20 mm from the wearing course during the design procedure. The details on mixture properties can be found in a relevant report by STA (2011). The dynamic loads have been added to the static standard axle loads, assuming a linear evolution of DLC (see Figure 4) over the pavement service life. Generally, the pavement surface is fairly smooth after construction. Therefore, the initial DLC is chosen to be 8% which corresponds to a very good road surface (Cebon, 2000; ISO, 1996). In Figure 5 one may see that the average rate of change for the DLC of a selected quarter car vehicle varies 11
from 0.3 to 0.8 percent depending on the constant speed of the moving vehicle. Thus, the rate of change for the analysis regarding different case studies is selected with respect to this range. Maximum horizontal tensile strain at the bottom of AC layer as well as maximum compressive vertical strain at top of the subgrade is calculated for each pavement section and climate period using multilayer elastic theory. Afterwards by applying Equation (2) the yearly maximum allowed standard single axles are calculated for each scenario. Table 1 : Traffic information on selected road sections Region
AADT
% Trucks
ESALs [Millions]
Case A
Stockholm
7800
10
10.9
Case B
Stockholm
32000
6
27.5
Case C
Skåne
15000
1
2,1
* 1% yearly traffic growth has been assumed for all three cases Table 2 : Pavement structure configuration of road sections
Bituminous surface layer
Case A Thickness [mm] 50
Case B Thickness [mm] 50
Case C Thickness [mm] 40
Bituminous bearing layer
115
145
75
Unbound base
80
80
80
Crushed rock sub-base
200
200
200
Rock bedding
-
-
-
Material
The total energy and time variables were calculated for a functional unit that was defined as construction and rehabilitation of 1 km asphalt pavement per lane for the 25 years design life. Calculations of energy related costs were done based on the price of energy sources as 0.01 €/MJ for crude oil, 0.038 €/MJ for diesel, 0.024 €/MJ regarding electricity price for industries and 0.015 €/MJ for residual oil as of 2012. The information regarding transportation distances, the value of time for labour/equipment and the amount of time and energy spent for construction and rehabilitation can be found in Mirzadeh et al. (2013).
Results and discussions In Figure 8 the effect of integrating the pavement surface deterioration on pavement service life is depicted. Four different scenarios (i.e. four different rate of change of DLC) for pavement surface deterioration have been considered. In Figure 8 each coloured line corresponds to one scenario with the input data documented in the legend. The dotted grey line corresponds to the case in which no dynamic loads have been taken into account. According to Equation (4) the pavement structure fails when the allowed number of axles is lower than the number of ESALs on the road. Therefore, in Figure 8 whenever the coloured
12
lines (i.e. allowed number of axles) intersect with the dashed line, the pavement service life is terminated. The yearly growth of DLC results in higher magnitude of traffic loads exerted on the pavement. Thus, the amount of horizontal tensile strain at the bottom AC layer, as well as vertical compressive strain on top of the subgrade increases dramatically. This results in less allowed number of standard axles on the pavement. One may see in Figure 8 that the additional dynamic loads can decrease the predicted pavement service life up to 10 years.
Figure 8 : Service life reduction as results of applying the additional dynamic loads for different cases
The application of the preventive maintenance was evaluated for the worse case scenarios regarding each case. It was observed that having a medium to high DLC, the total energy (i.e. the sum of the expended and feedstock energy) was lower for the alternative with the application of the preventive maintenance (Figure 9a). Similarly, the alternatives with application of preventive maintenance had a lower relative LCC. However, the implementation of the preventive maintenance was more economically efficient for the cases with higher traffic level (Figure 9b). It was observed that surface roughness increased during the life span of asphalt pavements (Figure 7). Moreover, increases in the surface roughness resulted in higher DLCs which decreased the pavement life. Thus, implementation of preventive maintenance activities by reducing the surface roughness can moderate the impact of dynamic loads induced from road-vehicle interactions. Furthermore, reduction in the surface roughness can also decrease vehicles fuel consumption. Construction and rehabilitation of roads are energy intensive processes. Additionally, high volatility regarding 13
the energy price has imposed high financial risks to asphalt pavement projects. Therefore, application of the preventive maintenance by increasing the pavement service life and decreasing the portion of energy related costs have lowered the amount of financial risk regarding the cost asphalt pavements (Figure 10). The application of a 20 mm thin asphalt layer had a lower amount of energy related costs compare to a more traditional thicker rehabilitation. Therefore, the application of the preventive maintenance reduced the portion of energy related costs by 6-7% for all the cases.
Figure 9 : (a) The amount of energy savings for different DLC rate of change; (b) The costs for alternatives with preventive maintenance (Cp) in relation to alternatives without preventive maintenance (C0).
In order to develop better solutions, increase the competition and to transfer the responsibility to the party best placed to minimize the risks, the STA has implemented contracts with functional specifications. Contractors in design-bid-build projects are responsible for a certain guarantee period after the project completion. Currently, 34% of road projects in Sweden are design-bid-build and this portion is to be increased to 50% by 2018. In traditional bid-build contracts the lower LCC due to application of the preventive maintenance can be of interest for transport administrations. However, in the design-bid-build contracts the application of the preventive maintenance can be of a great interest from the transport administration and contractors’ perspectives. By lowering the total cost of the project this can help the contractors to offer lower bids in order to win more contracts. Moreover, the application of the preventive maintenance can lower the contractor’s risk profile. Furthermore, the amount of fuel consumption by the vehicles is directly affected by the road surface roughness. The vehicle fuel consumption is a function of rolling resistance of the vehicle tyres. The rolling resistance forces are highly influenced by the pavement surface conditions. Therefore, the pavement surface deterioration during its life span increases rolling resistance forces which results in higher vehicle fuel consumption. In the calibrated HDM 4 model (Zaabar and Chatti, 2010) the fuel consumption is directly related to IRI. The IRI value can be calculated directly from a given displacement PSD (Johannesson and Rychlik, 2012; Sun et al., 2001). 14
Figure 10 : Asphalt pavement cost profile regarding alternatives with and without preventive maintenance for the three cases.
Summary and conclusions In this paper, the effect of surface deterioration on the exerted dynamic loads and consequently the pavement service life was studied through an LCC framework. The pavement surface deteriorates during its lifespan which eventually results in intensified exerted dynamic loads on the road. In this paper, the pavement surface deterioration is simulated based on the average yearly gradient of displacement PSD regarding road sections in Sweden. Furthermore, the yearly DLC is calculated based on the response of the quarter car model with constant speed. Increases in the exerted dynamic loads on the pavement can ultimately decrease the pavement service life. The impact of a decrease in the pavement life was reflected in terms of higher EACs and consequently higher total LCCs for the pavement. However, application of a preventive maintenance by smoothing the surface can lower the exerted dynamic loads. The benefits regarding the application of preventive maintenance activities were reflected on the LCCs and the financial risk profiles for the studied scenarios. The following conclusions are drawn from this paper: •
It was observed that incorporation of the impact from exerted dynamic loads into the asphalt pavement design framework significantly decreased the life length of the 15
• •
pavements. However, due to a certain level of empiricism in the design framework these impacts need to be further studied. The application of the preventive maintenance was shown to be economically efficient especially for higher deterioration rates. The implementation of the preventive maintenance (i.e. thin asphalt layer) was proven to decrease the financial risk regarding the cost of asphalt pavements. This strategy should be of a great interest from the road administrations and contractors’ perspectives specialty in design-bid-build contracts with longer responsibility periods.
16
References Abaza, K., Ashur, S., Al-Khatib, I., 2004. Integrated Pavement Management System with a Markovian Prediction Model. Journal of Transportation Engineering 130, 24–33. doi:10.1061/(ASCE)0733-947X(2004)130:1(24) Andren, P., 2006. Power spectral density approximations of longitudinal road profiles. International Journal of Vehicle Design 40, 2–14. Antikarov, V., & Copeland, T. (2001). Real options: A practitioner’s guide. New York. Bogsjö, K., Podgórski, K., Rychlik, I., 2012. Models for road surface roughness. Vehicle System Dynamics 50, 725–747. BSI, 1972. Proposals for Generalised Road Inputs to Vehicles. British Standard Institution (BSI 72/34562), London, England. Cebon, David, 2000. Handbook of Vehicle-Road Interaction. Taylor & Francis. Chou, E. Y. J., Datta, D., & Pulugurta, H. (2008). Effectiveness of Thin Hot Mix Asphalt Overlay on Pavement Ride and Condition Performance (No. FHWA/OH-2008/4). Collop, A.C., Cebon, D., 1995. A Model of Whole-Life Flexible Pavement Performance. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science 209, 389–407. doi:10.1243/PIME_PROC_1995_209_170_02 COST 333, 1999. Development of new bituminous pavement design method. Final Rep., Directorate General Transport, Brussels, Belgium. Dodds, C.J., Robson, J.D., 1973. The description of road surface roughness. Journal of Sound and Vibration 31, 175–183. doi:10.1016/S0022-460X(73)80373-6 Fama, E. F., & French, K. R. (2004). The capital asset pricing model: theory and evidence. Journal of Economic Perspectives, 25-46. FHWA, 1998. Roughness trends of flexible pavements. Federal Highway Administration (FHWA) (2003). Economic analysis primer. Washington DC. Federal highway administration (2002). Life-cycle cost analysis primer. Washington DC. Gullberg, D., Birgisson, B., Jelagin, D., 2012. Evaluation of a novel calibratedmechanistic model to design against fracture under Swedish conditions. Road Materials and Pavement Design 13, 49–66. doi:10.1080/14680629.2011.651838 Hammarström, U., Eriksson, J., Karlsson, R., & Yahya, M. R. (2012). Rolling resistance model, fuel consumption model and the traffic energy saving potential from 17
changed road surface conditions. Swedish National Road and Transport Research Institute (VTI), 748A. Hull, J. C. (2008). Options, futures, and other derivatives, Prentice Hall, Upper Saddle River, NJ. ISO, 1996. Mechanical Vibration. Road Surface Profiles. Reporting of Measured Data. B S I Standards. Johannesson, P., Rychlik, I., 2012. Modelling of road profiles using roughness indicators. Department of Mathematical Sciences, Chalmers University of Technology and Göteborg University 27. Khavassefat, P., Jelagin, D., Birgisson, B, 2014. The Non-Stationary Response of Flexible Pavements to Moving Loads. Submitted to International Journal of Pavement Engineering. Kropáč, O., Múčka, P., 2008. Deterioration Model of Longitudinal Road Unevenness Based on its Power Spectral Density Indices. Road Materials and Pavement Design 9, 389– 420. doi:10.1080/14680629.2008.9690125 Lee, H., Lee, S., Lee, J., Chon, k. (2009). The Revenue-Beta Model for Private Participation in Infrastructure. Journal of the Eastern Asia Society for Transportation Studies, 8. Liu, C., Herman, R., McCullough, B.F., 1998. Pavement deterioration, rate of dynamic force, and ride quality. Transportation Research Record: Journal of the Transportation Research Board 1643, 14–19. Liu, Chiu, 2000. Vehicle-road interaction, evolution of road profiles and present serviceability index. Road Materials and Pavement Design 1, 35–51. doi:10.1080/14680629.2000.9689883 Markowitz, Harry. (1952). Portfolio Selection. Journal of Finance. 7:1, pp. 77–99. Mirzadeh, I., Butt, A.A., Toller, S., Birgisson, B., 2013. Life cycle cost analysis based on the fundamental cost contributors for asphalt pavements. Structure and Infrastructure Engineering 0, 1–10. doi:10.1080/15732479.2013.837494 Monismith, C.L., Lysmer, J., Sousa, J., Hedrick, J.K., 1988. Truck Pavement Interactions - Requisite Research (SAE Technical Paper No. 881849). SAE International, Warrendale, PA. Newcomb, D. E. (2009). Thin Asphalt Overlays for Pavement Preservation. National Asphalt Pavement Association, Lanham, Md. (No. 135). Potter, T.E.C., Cebon, D., Cole, D.J., Collop, A.C., 1997. Road damage due to dynamic tyre forces, measured on a public road. International Journal of Heavy Vehicle Systems 3. Sandberg, U., Kragh, J., Goubert, L., Bendtsen, H., Bergiers, A., Biligiri, K. P., ... & Vansteenkiste, S. (2011). Optimization of Thin Asphalt Layers–State-of-the-Art Review. 18
ERA-NET Road report, http://www. highways.gov.uk / knowledge_compendium /assets/documents/Portfolio/ Optimization_of_Thin_Asphalt_Layers_Report. pdf. Sun, L., Zhang, Z., Ruth, J., 2001. Modeling Indirect Statistics of Surface Roughness. Journal of Transportation Engineering 127, 105–111. doi:10.1061/(ASCE)0733947X(2001)127:2(105) Sun, Lu, Luo, F., 2007. Nonstationary Dynamic Pavement Loads Generated by Vehicles Traveling at Varying Speed. Journal of Transportation Engineering 133, 252–263. doi:10.1061/(ASCE)0733-947X(2007)133:4(252) STA (STA ), 2011. Trafikverkets Krav Beskrivningstexter för Bitumenbundna lager i vägkonstruktioner. TRVKB 10 Bitumenbundna lager. STA (STA), 2011. Trafikverkets tekniska krav Vägkonstruktion. TRVK Väg. STA (STA), 2013. PMS Objekt [WWW Document]. PMS Objekt. URL http://www.trafikverket.se/Foretag/Bygga-och-underhalla/Vag/Tekniskadokument/Vagteknik/PMS-Objekt/ (accessed 9.19.14). Wang, T., Lee, I. S., Kendall, A., Harvey, J., Lee, E. B., & Kim, C. (2012). Life cycle energy consumption and GHG emission from pavement rehabilitation with different rolling resistance. Journal of Cleaner Production, 33, 86-96. Watson, D. E., & Heitzman, M. (2014). Thin Asphalt Concrete Overlays. Transport Research Board of the national academies (No. Project 20-05, Topic 44-07). Yeo, K., & Qiu, F. (2003). The value of management flexibility: a real option approach to investment evaluation. International Journal of Project Management, 21, 243-250. Zaabar, I., Chatti, K., 2010. Calibration of HDM-4 Models for Estimating the Effect of Pavement Roughness on Fuel Consumption for U. S. Conditions. Transportation Research Record: Journal of the Transportation Research Board 2155, 105–116. doi:10.3141/2155-12
19
