Estimating the Effects of Pavement Condition on Vehicle Operating Costs (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3 This chapter summarizes the findings from the review of published literature on the relationship between pavement conditions and vehicle operating costs. The review included Federal Highway Administration (FHWA) and National Coop- erative Highway Research Program (NCHRP) reports, and other relevant domestic and foreign reports and publications. Road user costs represent a portion of the transportation cost. These costs include vehicle operating costs, travel time delay, safety, comfort and convenience, and environmental impacts. Figure 2-1 presents the different components of road user costs (Bennett and Greenwood, 2003b). Vehicle operating costs are the costs associated with owning, operating, and maintaining a vehicle and include fuel consumption, oil and lubrication, tire wear, repair and maintenance, depreciation, and license and insurance. VOC components modeled include fuel and oil consump- tion, repair and maintenance costs, tire wear, and vehicle depreciation. Each of these cost components are typically modeled separately and summed to obtain overall vehicle operating costs. Common to many of these relationships is a road roughness factor used to describe the condition of the road. One such roughness measure is the international roughness index (IRI) developed as part of the World Bank Highway Development and Management System (HDM) standards studies (Sayers et al., 1986). Road roughness is a broad term describing the range of irregularities from surface texture through road unevenness. To better characterize the influence of road roughness on vehicle operating costs, the total texture spectrum was subdi- vided into the four categories defined in Figure 2-2 (Sandberg and Ejsmont, 2002). This categorization of roughness allowed a better evalu- ation of the surface factors influencing fuel consumption. As with fuel consumption, road roughness influences repair and maintenance costs, tire replacement, and the market value of vehicles. Barnes and Langworthy (2004) estimated vehicle operating costs using data from various sources. Fig- ure 2-3 shows the reported costs of fuel, tire replacement, and repair and maintenance expressed as a percentage of total operating expenses. Fuel cost is shown to be the pri- mary cost component followed by maintenance and repair costs, and tire wear. Overview of Existing VOC Models Based on the literature review, the major models that have been developed in various countries were identified. The most relevant models include: • The World Bank’s HDM 3 and HDM 4 VOC models (Bennett and Greenwood, 2003a, 2003b); • Texas Research and Development Foundation (TRDF) VOC model (Zaniewski et al., 1982); • MicroBENCOST VOC model (McFarland et al., 1993); • Saskatchewan VOC models (Berthelot et al., 1996); • British COBA VOC module (British Department of Trans- portation, 1993); • Swedish VETO model (Hammarström and Karlsson, 1991); • Australian NIMPAC VOC module (National Association of Australian State Road Authorities, 1978); • ARFCOM model of fuel consumption (Biggs, 1988); • New Zealand NZVOC (Bennett, 1989); and • South African VOC models (du Plessis, 1989). Most of the present VOC models have benefited from the World Bank’s HDM research to some extent. Fig- ures 2-4 and 2-5 outline the chronological development of these models. As shown in Figure 2-4, the basis of HDM research dates back to a study by de Weille (1966) for the World Bank, which led to the development of the Highway Cost Model (Becker, 1972) and subsequently to the most recent HDM 4 module. Figure 2-5 highlights the VOC research conducted in the United States, which was primarily initiated by Winfrey (1969) C h a p t e r 2 Vehicle Operating Cost Models

4followed by Claffey (1971). These initial efforts laid the founda- tion for an assembly of VOC data and estimation models in the American Association of State Highway and Transportation Officials (AASHTO) Red Book by 1977. In 1982, Zaniewski et al. (1982) developed new VOC models as part of the TRDF study. The TRDF models considered vehicle technology at that time and the effect of pavement roughness on vehicle oper- Source: adapted from Bennett and Greenwood (2003b) Figure 2-1. Components of road user costs. 1 mm = 0.04 in, 1 m = 3.3 ft. PIARC: Permanent International Association of Road Congresses Source: adapted from Henry (2000) and Sandberg and Ejsmont (2002) Friction Tire/Vehicle Tire Wear Rolling Resistance PIARC (1987) Micro-texture Macro-texture Mega- texture Roughness 10 0.005 0.05 0.5 5 50 0.5 5 50 mm m 0.001 0.01 0.1 1 10 0.1 1 Figure 2-2. Ranges in terms of texture wavelength and their influence on pavement–tire interactions. Source: after Barnes and Langworthy (2004) 0 4 8 12 16 20 Fuel Maint./Repair Tires R el at iv e to T ot al C os t ( % ) Cost Component Mean Standard Deviation Figure 2-3. Relative vehicle operating costs for trucks. ating costs addressed in the Brazil HDM study (Chesher and Harrison, 1987). These models were incorporated into the MicroBENCOST VOC models, which was intended to replace the AASHTO Red Book models. It should be noted that IRI was not an accepted roughness index at that time. More recently, a user-friendly model for personal com- puters, Vehicle/Highway Performance Predictor (HPP), was developed for highway designers, planners, and strate- gists to estimate fuel consumption and exhaust emissions related to modes of vehicle operations on highways of vari- ous configurations and traffic controls (Klaubert, 2001). This model simulates vehicle operations by evaluating the vehicle external loads or propulsive demands determined by longitudinal and lateral accelerations, positive and neg- ative road grades, rolling resistance, and aerodynamic drag for various transmission gears. Table 2-1 summarizes the essential features of the existing VOC models; Appendixes A, B, and C present the detailed equations and relationships of these VOC models. In sum- mary, most of the recent available VOC models were devel- oped in countries other than the United States. Most of the existing models are derived from previous models as a means

5 for improving them. The most recent VOC model found in the literature is HDM 4 (Bennett and Greenwood, 2003b). Evaluation of the Existing Models This section summarizes the findings of the evaluation of available models. More details are presented in Appendixes A (fuel consumption models), B (tire wear models), and C (repair and maintenance models). The model evaluation and selection were based on the practicality and statistical soundness of the model. The practicality evaluation of the models considered the following factors: (1) Ease of use, (2) Availability of appropriate input data, (3) Ability of the model to incorporate pavement-surface conditions as currently being measured, and (4) Reasonableness and applicability to US conditions. The statistical soundness evaluation of the models consid- ered the theoretical validity and accuracy of the models. In this regard the following factors were assessed: (1) Data reliability, (2) Original sample size, (3) Model assumptions, (4) Model formulation, (5) Estimation techniques, (6) Goodness-of-fit of the model, (7) Estimated standard error of the predictions, and (8) Statistics of the parameters. Source: Bein (1993) De Weille 1966 Highway Cost Model 1971 MIT, TRRL & LCPC Kenya, India & Caribbean 1971-1986 TRRL & CRRI Brazil Study 1975-1984 TRDF & TRRL HDM 3 VOC 1987 VETO NITRR NZVOC PMIS CB-Roads HDM 4 VOC 1994-2000 TRDF VOC 1980 -82 COBA Background Work Major VOC Model Other VOC Model Figure 2-4. World Bank VOC models development. Source: Bein (1993) Figure 2-5. VOC models development in the United States.

6Empirical and Mechanistic VOC Models VOC models can be categorized as empirical and mecha- nistic models. Mechanistic models are theoretically formu- lated so that they encompass the main physical parameters according to basic laws of physics/mechanics. Empirical mod- els rely more on “blind” mathematical correlations between different parameters to produce a model whose applicability can be limited by the specific data used in its development. In light of the relative strengths and weaknesses of both model- ing approaches, a hybrid mechanistic–empirical approach is often used. The development of empirical models is data-intensive and requires frequent updating and re-calibration to account for the changes in prices, and vehicle and road parameters. Also, most of the empirical models make use of classical regression assumptions (i.e., normality, independence, constant variance). However, many of the response vari- ables (i.e., vehicle operating costs) do not follow these classical assumptions. These models would produce appro- priate results when the input data are within the ranges used for developing the models but could produce erro- neous results if the input data were outside these ranges. Nevertheless, these models have the advantage of requiring less input data and therefore are more suitable for those applications where limited data are available. Mechanistic–empirical models are based on mathemati- cal representations of the mechanical relationship between vehicle and road conditions. The accuracy of these formula- tions depends on the validity of the theoretical assumptions and the calibration process. The calibration of these models is generally less data-intensive than for empirical models. This type of model is capable of predicting the cost for a wide vari- ety of scenarios, where the appropriate input data are avail- able. However, extensive input data are often necessary to obtain reliable results. To address the data issues, analytical approaches—such as making valid assumptions in the absence of full data, using default values, using composite or weighted values, and conducting scenarios and sensitivity analysis— were used. Feature VOC Models HDM 3 COBA9 VETO NIMPAC ARFCOM TRDF, MicroBENCOST HDM 4 Empirical – – – Mechanistic – – – Level of Aggregation Simulation – – – – Project Level Network Level – – Vehicle Operation Uniform Speed Curves – Speed Change – – Idling – – Typical Vehicles Default User Specified – – – Modern Truck – – Road–Related Variables Gradient Curvature Superelevation – – – Roughness Pavement Type – – Texture – – – – Snow, Water – – – – Wind, Temperature – – – – Absolute Elevation – – VOC Components Fuel, Oil, Tires, Repair/Maintenance, Depreciation Interest – – Cargo Damage – – – – – – Overhead – – – – Fleet Stock – – – – – Exhaust Emissions – – – – – Table 2-1. Categories of VOC models (empirical versus mechanistic).

7 The World Bank updated the ARFCOM fuel consump- tion module in the HDM 4 model (Bennett and Greenwood, 2003b). Therefore, the HDM 4 model was selected for calibra- tion and field tests were conducted to collect fuel consumption data for calibration and validation purposes. Tire Wear Models The only mechanistic–empirical tire wear models are included in HDM 3 and HDM 4. The HDM 3 model adopted the slip energy concept to calculate the changes in tread wear. The HDM 4 model has been extended from the HDM 3 model to include horizontal curvature force and traffic interaction effects. The theoretical formulation of the HDM 4 tire wear model is based on the same assumptions as the fuel consump- tion model. Therefore, the model was selected for calibration using field test data for passenger cars and the data provided by the National Center for Asphalt Technology (NCAT) for trucks. Repair and Maintenance Models All the repair and maintenance models are empirical, except for the Swedish VETO model. The empirical models that were developed in other countries (such as the World Bank’s HDM 3 and HDM 4 models, the Saskatchewan models, and the South African model) could not be applied to US conditions, because these models were developed using data from different vehicle fleets (i.e., different vehicle characteristics and different technologies), different pavement conditions, as well as dif- ferent labor and parts costs. The effect of pavement conditions on repair and maintenance costs predicted by the (mechanis- tic) VETO model are far higher than the change in parts cost predicted using empirical models. Therefore, it was decided to update the results from the latest study in the United States (Zaniewski et al., 1982), which is the only model that could be applied for the United States, and to develop a new mechanistic– empirical repair and maintenance model. This was done using the data obtained from Michigan and Texas Departments of Transportation. Selection of Appropriate VOC Models Fuel Consumption Models The major mechanistic–empirical fuel consumption models are the HDM 3, the South African model, and the Australian model (ARFCOM) that was adapted in the HDM 4 study. The assessment of these models considered their appropriateness to model the key characteristics (i.e., forces opposing motion, internal vehicle forces, engine speed effect, driving in acceleration mode and transferability to different vehicles). Table 2-2 summarizes the results of the assessment. It can be seen that only the ARFCOM model satisfies all the listed requirements. The assessment of the models also considered the validity of the assumptions used in their formulation. The following issues were identified: • The South African model assumes that the fuel efficiency of the vehicle is independent from the driving mode. However, a number of studies that were conducted in the early 1980s in Australia to model fuel consumption found that the fuel efficiency increases in the acceleration case (Biggs, 1987) and that it is a function of tractive and engine power. • The HDM 3 model adopted a constant engine speed. How- ever, it is known that the engine speed is a function of the vehicle speed and the driving mode. • The ARFCOM model predicts the fuel consumption as a function of the input (engine) and output power. How- ever, the determination of the parameter values for the engine drag equation had low coefficients of determi- nation and high standard errors (Biggs, 1988). Also, two different equations in the engine speed model were for- mulated: one for a vehicle in top gear, the other for a vehi- cle in less than top gear. However, these equations lead to a discontinuous relationship between vehicle speed and engine speed when the vehicle shifts into top gear. Such discontinuities lead to inconsistent fuel consumption pre- dictions and should therefore be avoided (Biggs, 1988). Model Forces Opposing Motion1 Internal Vehicle Forces2 Engine Speed2 Excess Fuel due to Acceleration2 Transferability2 HDM 3 South African ARFCOM 1 Suitable when considering the effect of pavement conditions 2 Suitable when considering the effect of emerging vehicle technologies Table 2-2. Assessment of VOC models.