Resource and Energy Economics 20 Ž1997.

1–15

A comparative cost analysis of biodiesel,
compressed natural gas, methanol, and diesel for
transit bus systems
Nicolas B.C. Ahouissoussi a , Michael E. Wetzstein
b

b,)

a
World Bank Resident Mission, Cotonou, Benin
Department of Agricultural and Applied Economics, UniÕersity of Georgia, Athens, GA 30602, USA

Abstract
A comparison of operating costs for buses used in a transit system are investigated
considering four alternative fuels: biodiesel, compressed natural gas, methanol, and diesel.
Rust’s ‘nested fixed point’ maximum likelihood estimation algorithm is used in this
comparison. The algorithm considers both tangible costs such as fuel, maintenance, and
infrastructure, and intangible costs associated with different levels of bus engine operating
reliability under alternative fuels. Using data on actual monthly mileage and time of engine
rebuilds under the four alternative fuels, the Rust algorithm is employed assuming that an
optimal maintenance strategy is adopted for each alternative fuel type. Results indicate that
although biodiesel and biodiesel blends have higher total costs than diesel fuel, they have
the potential of competing with CNG and methanol as fuels for urban transit buses.q 1997
Elsevier Science B.V.
JEL classification: Q40; Q42; Q48
Keywords: Biodiesel; Natural gas; Methanol; Diesel; Buses; Alternative fuels

)
Corresponding author. Department of Agricultural and Applied Economics, University of Georgia,
Athens, GA 30602, USA. Tel.: Ž 706 . 542-0758; fax: Ž 706 . 542-0739; e-mail:
mwetzstein@agecon.uga.edu

0928-7655r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved.
PII S 0 9 2 8 - 7 6 5 5 Ž 9 6 . 0 0 0 1 5 - 2

However. engine replacement. equipment replacement. and 47% in total hydrocarbon emissions. cost comparisons incorporating infrastructure. In general. For a comprehensive comparison of these alternative fuels. and low-sulfur diesel fuel. Comparing these . and maintenance costs – can be misleading and can result in erroneous conclusions. the primary focus of this study is to provide a complete cost comparison for operating a transit bus fleet on compressed natural gas ŽCNG. For example. Thus. Ahouissoussi. With the addition of engine and fuel system modification costs for alternativefueled buses. estimates of the total costs are required Žboth tangible and intangible. Given this potential of improving air quality. Differences in maintenance costs among these alternative-fueled buses are estimated with Rust’s nested fixed point maximum likelihood estimation algorithm. A comparison of just the fuel cost per mile – without consideration of infrastructure. as well as intangible costs associated with bus reliability. and maintenance costs..B. and maintenance costs. Reductions include 31% in particulate matter. varying by type of fuel used. While the market for cleaner burning renewable fuels is becoming increasingly important. however. along with reliability. there is a significant gap in the literature comparing these alternative fuels. also vary by fuel type. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 1. A review of the literature found no previous empirical work that has examined the issue of economic competitiveness of alternative vehicle fuels based on total life-cycle costs.. engine replacement. Introduction Current regulatory policies place the alternative fuel industry at a critical junction. methanol. firms and governmental agencies promoting a particular alternative fuel for urban bus transit have an opportunity for significantly influencing its adoption. Emissions from alternative-fueled bus engines consistently indicate lower emissions of reactive hydrocarbons. reduces diesel engine air pollution. 21% in carbon monoxide.C. A specific objective is the development of a dynamic model of bus engine replacement for determining the present value of total fleet operating costs for these alternative fuels. including unobservable intangible costs. M. biodiesel.. such promotion requires a comprehensive analysis. the National Soydiesel Development Board reported that biodiesel used in a 20r80 blend with petroleum diesel.2 N. including an economic evaluation on the relative comparative costs of operating an urban bus transit fleet on alternative fuels. and particulate matter than diesel engines.E. a cost comparison among the fuels may be determined by considering the present value of the total fleet operating cost over the fleet’s life cycle. based on the assumption that an optimal maintenance strategy is adopted for each type of alternatively fueled bus. A small policy change by government or product promotion by industry may determine which alternative fuels will be widely adopted. carbon monoxide. are important factors in energy policy issues. Fuel is a major cost. The Rust algorithm estimates bus maintenance costs. other tangible costs such as infrastructure. along with a catalytic converter. as regulated by the Environmental Protection Agency ŽEPA. for different fuel types.

nontoxic. nitrogen oxide ŽNOx. including CNG and methanol. 2. Ahouissoussi. 1993. Other sources for methanol production include coal. will encourage the adoption and use of alternativefueled vehicles and associated technologies. In the past. as a means of controlling urban ozone. more recent research has emphasized the potential of plant oils as diesel fuel extenders or replacements ŽGavett. Biodiesel is a clean-burning. McIntosh et al.N. natural gas is a fossil fuel composed primarily of methane. 1993. and rapeseed. carbon monoxide. 1980. biodiesel often is blended with diesel at a ratio of 20r80. Extracted from underground reservoirs. because it offers the potential for reducing exhaust emissions. a number of other alternative fuels. M.. Fuels produced via chemical and thermal processes are referred to as biodiesel fuels. particularly in urban areas.. Methanol is yet another alternative fuel produced from both fossil and renewable domestic resources. sunflower. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 3 maintenance cost estimates with actual observed costs reveals the sensitivity of total cost to changes in maintenance costs. and helium. Alternative fuels Federal regulatory policies resulting from the implementation of the National Energy Policy Act ŽNEPA. CNG is another fuel with the potential of meeting tighter vehicle-emission requirements. renewable. along with other hydrocarbons including ethane.B. One alternative is biodiesel fuel. . Methyl and ethyl esters derived from soybean and rapeseed have properties much closer to conventional diesel fuel ŽZiejewski et al. propane. From an environmental standpoint. 1993.... These nonpetroleum diesel fuel substitutes can be obtained from oilseed crops such as soybean. Clark and Wagner.C. 1984. This model identifies the potential competitiveness of biodiesel compared with CNG and methanol under alternative prices. nitrogen. and domestically produced fuel that can be used neat or in blend with petroleumderived diesel ŽHolmberg et al. biodiesel blended with diesel can significantly reduce emissions of particulates. residual oil. form as a diesel substitute or potentially blended with diesel. This is particularly true for the urban bus market. and inert gas such as carbon dioxide. petroleum refining was controlled primarily for gasoline yield and quality. thus. The majority of the methanol produced in the United States is from natural gas resources. however.. and particulate matter ŽPM-10. It can be used in neat Ž100%. biodegradable.. and biomass. will be competing with biodiesel fuel.E. and butane.. Although extensive research interest during the past decade centered upon ethanol and methanol processing technologies and policies. and unburned hydrocarbons. Petroleum-derived diesel is used as the base fuel in this study. Recent EPA regulatory activities are aimed at reducing emissions of volatile organic compounds ŽVOC. For economic and engine-compatibility reasons. Interest in using natural gas as a transportation fuel has increased in recent years.

Given an objective of minimizing unexpected engine failure. Given this optimal timing. the model estimates what the marginal maintenance cost per month must be to obtain this optimal timing.05% by weight on the sulfur content and a minimum cetane index of 40 for diesel fuel used in on-road vehicles. and i t s 0 otherwise. then the probability of other parts failing increases. . . the stochastic process is shown as  i t . u 1 . Letting r denote the expected cost of a rebuilt bus engine and cŽ x t . along with engine rebuild costs. u 1 . For a determination of this maintenance cost. for a bus with relatively high mileage. the optimal solution may be to rebuild the entire engine. fuel system costs and miles between rebuilds. u 1 . Basically. . each of which has its own stochastic failure as a function of accumulated use. and so it might be optimal to rebuild the engine. . if a particular engine component fails when an engine has relatively high mileage. Diesel fuel now faces significant fuel-quality and engine-emissions requirements. . a policy of preventive periodical engine rebuilding is cost effective. and t s 1. . if i t s 0 Ž 1. Unobserved state variables are incorporated by assuming that unobserved costs  e t Ž0. u 1 . and intangible costs that include loss of time for the bus driver and passengers. Applying Rust’s algorithm. T. In contrast. x t 4 .4 follow a specific stochastic process. where i t s 1 if rebuild occurs at time t.. Ahouissoussi. s ½ yr y c Ž 0. e t Ž1. and the optimal preventive maintenance cost for each alternative-fueled bus will depend on the tradeoff between the value of unused life and the cost of failure. q e t Ž 1 . If a particular component fails when a bus has relatively low mileage. However. this method assumes that transit authorities have developed a procedure for optimally determining when a bus should be rebuilt. M.B.. An optimal maintenance model for bus engines Generally.i t . differ by fuel type. . 3. The differences in miles between rebuilds and rebuild costs across alternative-fueled buses suggest that the stream of maintenance costs varies by fuel type. Current EPA regulations set a maximum limit of 0. a bus engine can be regarded as a portfolio of individual components. the utility function is u Ž x t .4 N. q e t Ž i t . Thus. then it may be optimal to replace or repair just this failed component. provides a consistent method for determining the differences in maintenance cost among these alternative-fueled buses. 1994. This has changed over the last few years. Bus failures on the road are expensive in terms of both tangible costs such as towing. 1987. the cost of engines is comparable among the alternative fuels considered in this analysis. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 the quality of diesel fuel varied widely depending on the demand for gasoline ŽNational Biodiesel Board. if i t s 1 yc Ž x t . the expected per period maintenance cost. Application of the Rust algorithm ŽRust. where u is a parameter to be estimated.E.C. q e t Ž 0 . x t denotes mileage since the last rebuild at month t.

i 1 . and Lf .: T L1 Ž x 1 . . i T < x 0 . Lf . Ž3. .i 0 . Using the discretized mileage data. Also.i ty1 . . L2 . and Ž5. s Ł P Ž x t < x ty1 . It is hypothesized that the unobservable state variables  e t Ž0. M . . ease of interpretation. u 1 . . u 1 . should be interpreted as an unobserved component of cost associated with rebuilding the bus engine. u 2 . . A linear functional form for cŽ x t . Ž4. . s ½ g Ž x tq1 .0. u . x t . the remaining structural parameters Ž u 1 . This involves discretizing the state variable x t Žmileage. . u . using the likelihood function L1. x T . because neither the location nor the scale of these observed costs is identifiable without further information. . shown by Eq.i 0 . Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 5 Letting monthly mileage Ž x tq1 y x t . r . where i tm is the engine rebuild decision in month t for bus m. The estimation procedure consists of three stages corresponding to each of the likelihood functions: L1 . M. into a certain number of intervals Žsay n. . The full likelihood. x T . ts1 Ž 3. g. . so cŽ x t . . with maximum likelihood using the nested fixed point algorithm. Notice that e t Ž0. . s Ł P Ž x t < x ty1 . .B. . u 2 .i 1 . if i t s 0 Ž 2. u . is shown by Eq. . . . .. will not affect the choice probabilities.4 obey an independently and identically distributed bivariate process. e t Ž1. x tm 4 Ž t s 1. and its satisfactory fit with the data. . u . u 2 . i t . is chosen because of its computational tractability. . . Ž 4. .: T Lf Ž x 1 .E. are calculated using L2 and the estimates of u 2 as . it is implicitly assumed that the stochastic process  x ti . ts1 The first stage is to estimate the parameters u 2 of the transition probability P Ž x tq1 < x t . e ti 4 is independently distributed across buses. and e t Ž1. The specification does not include a constant term because subtracting a constant term from the utility function Ž1. i T < x 0 . the distribution g reduces to a multinomial distribution corresponding to monthly mileage. if i t s 1 g Ž x tq1 . of a specified length.i ty1 . In Stage 2.C. Tm . Ž 5. u 2 . and x is the mileage since the last rebuild of bus m in month t. . have a subjective parametric density function. m s 1. should be interpreted as an unobserved component of maintenance costs for the bus in period t. i T < u . Data required consist of  i tm . u 1 . . . implies a transition density of the form p Ž x tq1 < x t . u 2 . s Ł P Ž i t < x t .i 1 . s 0. The most that can be identified is the value of change in maintenance costs as a function of mileage. . . u 2 . The approach is to estimate the unknown parameters. can be normalized by setting cŽ0.. r. Ahouissoussi. . P Ž i t < x t . .N. x T . u s Ž u 1 . with normalized mean and variance.. .i t . . ts1 T L2 Ž x 1 . . u 2 . . L1 and L2 are partial likelihood functions.

yields an annual loss of 256 hours.6 N. Fifteen days of paid vacation Ž120 hours. Data Unfortunately. Service life for the building is 30 years. d Labor Žper hr. and a third laborer refuels the buses. bus engine plus fuel systemf Units Total cost Ž$.04 478.C.80 1620 3240 6963 10 905 0. Five-day work week. The Denver Regional Transport District refuels approximately 280 buses per night in a three-lane refueling system. Diesel tank is an FTP-3. Annual miles driven Žper bus.95.298 0. 36 578 92 000 40 600 3 4 300 22.89 19. . size b Total infrastructure cost Total infrastructure cost per bus Refueling cost: Labor costs per lane per day c Supervisor Žper hr. and eight paid sick leave days Ž64 hours. which results in $19. e Labor costs per day for 3 lanes Overhead multiplier Total labor costs per bus per year Ž365 days.95 539. Ahouissoussi. a Three lanes can service 360 buses at full capacity. gal. $r1000 ft 2 per lane a Tankage. f The diesel engine and fuel system are the base.000 Source: Colorado Institute for Fuel and High Altitude Engine Research.. data on monthly mileage and time of rebuilds for alternativefueled buses are very limited.rmonth. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 initial starting values. d Hourly rate is $16 plus benefits of 29%.67 1r3 3 3 2 3942 866 276 000 162 400 438 400 1461 61. 20 000-gal.q 0. employs the initial consistent estimate of u computed in Stages 1 and 2 to produce efficient maximum estimates of u using Lf . M.95 actual hourly labor cost based on the same formula used for supervisor’s rate. Actual hourly rate is then $16wŽ2080r1824.89.B. A tank capacity of 26 666 gallons is required per lane.29x s$22. 4. Two laborers drive the buses to the lane..5 times the tank size. Annual refueling cost per bus Cost per mile Bus capital data: Incremental first cost.84 0. c Refueling labor includes one supervisor per three lanes and three laborers per one lane. Fuel usage per bus Ž$rgal. so the incremental cost is the additional cost of methanol or CNG fuel engine and fuel system. b Tank is $10 600 plus 1. The final estimation ŽStage 3. nine paid holidays Ž72 hours.. 20 000-gallon tank with dimensions of 11 ft diameter by 28 ft tall. eight hours per day yields 2080 hours per year. Most experiments on alternative fuels were conTable 1 Diesel bus cost summary Item Unit cost Ž$.E. e Hourly rate is $13. Infrastructure cost per lane: Building cost.

nine paid holidays Ž72 hours.B.24 0.5 laborers per one lane. eight hours per day yields 2. $r1000 ft 2 per lane a Tankage.5 times more fuel than diesel. Five dual CNGrdiesel buses Ždiesel buses converted so they can also use CNG.. The Denver Regional Transport District refuels approximately 280 buses per night in a three-lane refueling system. and eight paid sick leave days Ž64 hours.5 3 2 9185 19 867 12. 20 000-gal. Regional Transportation District. and five methanol buses.N. a Three lanes can service 360 buses at full capacity.824. so the incremental cost is the additional cost of methanol fuel engine and fuel system. gal.25 gallons per 10.C.ryear. Infrastructure cost per lane: Building cost. Annual miles driven Žper bus. That experiment.29x s$22. which lasted from June 1989 through December 1993. b Tankage is based on 2. were added in 1991. e Hourly rate is $13. Five-day work week. resulting in short time intervals and little if any data collection.89 19. yields an annual loss of 256 hours.69 1r3 7. f Lubrizol is added at 6. d Hourly rate is $16 plus benefits of 29%. Ahouissoussi.95 1258. g The diesel engine and fuel system are the base. given that methanol buses require on average 2. Service life for the building is 30 years.20 3775 7550 11 722 195 21 102 0. bus engine plus fuel system g Units Total cost Ž$.080r1..42 276 000 2 681 480 2 957 480 9858 61.89. q0.080 hours per year. which results in $19. f Annual refueling cost per bus Cost per mile Bus capital data: Incremental first cost. e Labor costs per day for 3 lanes Overhead multiplier Total labor costs per bus per year Ž365 days. size b Total infrastructure cost Total infrastructure cost per bus Refueling cost: Labor costs per lane per day c Supervisor Žper hr. c Refueling labor includes one supervisor per three lanes and 7. Colorado. gal. Lubrizol Ž$rgal.708 29 900 Source: Colorado Institute for Fuel and High Altitude Engine Research.E. initially included five diesel buses which could use both diesel and biodiesel fuel.95..04 1197. Fifteen days of paid vacation Ž120 hours. Actual hourly rate is then $16wŽ2.ryear.59 15.. Summary assumes 2.5 times as much fuel.000 gallons of methanol. Fuel usageper bus Ž$rgal. M. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 7 ducted for demonstration purposes only.5 times diesel tankage. 29 801 92 000 268 148 3 10 300 22. d Labor Žper hr. .5 additional labor hours for fueling based on 2. One exception is an experiment by the Denver.95 actual hourly labor cost based on the same formula used for supervisor’s rate. The 15 Table 2 Methanol bus cost summary Item Unit cost Ž$.

M. so the incremental cost is the conversion cost of a diesel bus to CNG and additional cost of the fuel system. Two laborers drive the buses to the lane.95 501. Actual hourly rate is then $16wŽ2080r1824. d Labor Žper hr.95 actual hourly labor cost based on the same formula used for supervisor’s rate. combined with the application of Rust’s Table 3 CNGrdiesel dual bus cost summary Item Unit cost Ž$.ryear. e Labor costs per day for 8 lanes Overhead multiplier Total labor costs per bus per year Ž365 days. e Hourly rate is $13. stops per mile. 10% for engineering. Infrastructure cost per lane: Building cost. engineering plus development and permitting for contingency. gal. a Eight lanes can service 300 buses at full capacity. Annual miles driven Žper bus.95. eight hours per day yields 2080 hours per year.89. f Maintenance costs per bus per year Energy cost of compressors per bus per year Annual refueling cost per bus Cost per mile Bus capital data: Incremental first cost.E.. This unique data set.89 478. c Refueling labor includes one supervisor per three lanes and three laborers per one lane. Ahouissoussi.40 1r8 3 8 2 9767 10 764 1.431 35 000 Source: Colorado Institute for Fuel and High Altitude Engine Research.80 4014 8028 4306 400 467 14 940 0. Five-day work week. which results in $19.8 N. $r1000 ft 2 per lane a Fueling facility b Total infrastructure cost Total infrastructure costrbus Refueling cost: Labor costs per lane per day c Supervisor Žper hr. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 buses Žthree fleets comprised of five buses each. The buses were fueled on site. and a third laborer refuels the buses. They were maintained under the same preventive maintenance program.77 0. Service life for the building is 30 years.B. 34 691 92 000 8 300 22. Fuel usage per bus Ž$rgal. resulted in 146 monthly observations for the analysis. f Fuel usage is in equivalent gallons. The three fleets of buses were exposed to similar operating conditions such as scheduled speeds.28 365 736 000 2 320 000 3 056 500 10 188 22. Fifteen days of paid vacation Ž120 hours. g The CNGrdiesel bus engine is the same as the diesel bus.C. and 10% installed cost. and passenger loading. d Hourly rate is $16 plus benefits of 29%.q0. 5% for development and permitting..29x s $22. .89 19. b Estimated installed cost is $1 700 000 plus 10% for contractor’s markup. bus engine plus fuel system g Units Total cost Ž$. yields a loss of 256 hours. and eight paid sick leave days Ž64 hours.. nine paid holidays Ž72 hours. traffic conditions.

the 2. This lower coefficient of variation implies less uncertainty regarding timing of rebuilds. The coefficients . allows a cost comparison of alternative fuels based not only on fuel cost and usage. which is approximately twice the refueling cost for diesel buses. methanol. thus. using the nested fixed point algorithm.B..5 additional laborers for bus refueling. compared with $6963 and $11 722 for diesel and methanol buses. compared with approximately $10 000 per bus for both methanol and CNGrdiesel buses ŽTables 2 and 3. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 9 algorithm. Total infrastructure cost per bus is only $1461 for diesel and biodiesel ŽTable 1.5 factor increase in fuel consumption for methanol buses relative to diesel requires 4. However.x. the actual cost of fuel is lowest for CNGrdiesel buses at $4306. Ahouissoussi. this higher cost is potentially offset by the considerably lower coefficient of variation for the CNGrdiesel bus. The significantly higher infrastructure costs for CNGrdiesel buses as compared with those for diesel are due to the requirement of eight refueling lanes instead of three. associated with operating costs are computed by maximizing the full likelihood function. Lf wEq. repair. engine-rebuild.506. and 0. M. However. u 1 . Maximum likelihood results Estimates of the unknown parameters Ž r. 5. it is possible that CNGrdiesel buses may have lower costs due to fewer unexpected required rebuilds. Model results for diesel. Although this assumption may favor diesel and biodiesel. and in-service failure costs for the total operational life of the transit buses. CNGrdiesel buses are rebuilt at approximately half the time and mileage intervals of diesel and methanol buses. resulting in larger tankage. but also on maintenance. methanol.N. Further.E. respectively. and CNGrdiesel. Bus capital costs are the additional fuel system and engine conversion costs required for methanol and CNGrdiesel bus conversion.220 for diesel. All operating cost coefficients are significantly different from zero at the 1% significance level. fixed costs for alternative fuel facilities are incremental to diesel facility fixed costs. 0. respectively. it is realistic given current transit operations. Methanol buses require 2. Summary statistics indicate a large variation in mileage at time of engine rebuild.5 times more fuel. particularly for diesel and methanol buses. and CNGrdiesel alternative fuels are presented in Table 4. resulting in higher CNGrdiesel maintenance costs.C. The additional six refueling lanes for CNGrdiesel buses primarily explain that alternative’s 37% higher annual refueling cost per bus compared with diesel buses. Annual refueling cost per bus is $21 102 for methanol buses. Ž3.522. and to storing pressurized fuel. The coefficients of variation for mileage at rebuild are 0. The costs of alternativefueled buses are based on the assumption that a regional transportation district already has diesel bus refueling and maintenance facilities. methanol tanks are over six times more expensive than those used for diesel fuels. Again. Therefore.

B. ) ) ) 3. ) ) ) 5.80. 0.10. These average costs are considerably lower than the data obtained from the Denver Regional Transportation District on explicit average maintenance costs per month of $41. for diesel and biodiesel. M.16.38 Ž2. the proportion of this maintenance can be increased from zero.01 levels.49 9500 2047 31. ) y29.38 for methanol. respectively.80 for CNGrdiesel compared with $31.84 0. RC Scale parameter b . Assuming this base level of maintenance is the same for all alternative-fueled buses.21 Ž0. respectively. and CNGrdiesel buses. and $71.90 for diesel and biodiesel.05. and 0. and 30% significance levels. then total monthly maintenance cost is incremented every 5000 miles by the estimates in Table 4.10 N. divided by the rebuild cost coefficient Ž r . and cngrdiesel buses Summary statistics Structural coefficients: Operating costs a . Maintenance marginal cost is only $1. u 1 Rebuild costs. c Diesel and biodiesel Methanol CNGrdiesel 3..80 Notes: Numbers in parentheses are standard errors of estimates. a Operating costs include maintenance costs. including such items as brake adjustment and replacement or repair of individual components. $419. for diesel. and $5.40. Ahouissoussi. ) ) y45. insurance costs.59 for CNGrdiesel buses.32. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 Table 4 Marginal cost estimation results for diesel and biodiesel. methanol. This results in an average monthly maintenance cost of $28. associated with rebuild costs are significantly different from zero at the 10%. methanol. b Scale parameter is the actual rebuild cost Ž RC . ) ) ) 4. Marginal cost is the incremental change in total monthly maintenance cost. These marginal cost estimates indicate that the Denver Regional Transportation District perceives average monthly maintenance costs to increase for every 5000 accumulated miles on the buses.94. This wide variation in mileage at rebuild implies higher marginal cost associated with determining optimal preventive maintenance. respectively.28 6500 1284 4. s Marginal cost Žper 5000 miles. and triple asterisks Ž ) . 5%.62. Single.74 6500 2025 1.001u 1 s . denote significance at the 0. $128. c Marginal cost s 0.34 15. r Log-likelihood Rebuild costs. and CNGrdiesel-fueled buses. y22.06 Ž1. methanol.E.32. and loss of ridership and goodwill costs due to unexpected breakdowns.89 Ž1. The large variation in marginal cost can be explained by the relatively large rebuild cost and variation in mileage at rebuild of methanol compared with CNGrdiesel buses. This is the change from a base level of initial maintenance cost which is composed of routine maintenance. Lacking information on the exact proportions of these explicit costs that represent some base level of maintenance.64 Ž0.64. starting with zero as the base level.40.55 Ž7.84 for methanol and $4.C. which will provide .34 for diesel. double.

The rebuild costs for CNGrdiesel are equivalent to those for diesel. Three alternative fuel prices Ž$0.1406. Present value analysis A complete comparison of cost differentials among the alternative fuels requires consideration of all costs. but the methanol’s rebuild cost is approximately 46% higher than the rebuild cost for the diesel. are $38 296 and $43 727 for methanol and CNGrdiesel buses. respectively. the fuel costs are the same as for diesel. methanol.8887 for 20%. are 0. and 100% biodiesel blend. represent a major component of the overall total.B.75. Considering only the fuel cost.9766. Fuel costs for diesel. Rebuild costs Žlisted in Table 4.9297.1406. the interval between CNGrdiesel rebuilds is only half of that for diesel. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 11 information on the sensitivity of the total cost of alternative fuels to changes in this base level of maintenance.00. the equivalent gallon of CNG in one million BTUs of CNG is 7. methanol. 60%. the fuel efficiency estimates of biodiesel blend fuels compared with diesel Žas estimated by the Colorado Institute for Fuel and High Altitude Engine Research. M. and $0.1123.. cS . CNGrdiesel-fueled buses offer the lowest cost per mile.9916.C.50. however. rebuild. These additional costs above diesel base costs are accrued every ten years over the life of a bus. Denver. and operating costs. and $3. an equivalent gallon costs $0. respectively. 21. It is assumed that a2 diesel has a heat content value of 140. The interval between engine rebuilds for methanol is in close association with diesel. considering cost changes every 5000 miles Žgiven estimates of marginal cost detailed in Table 4.. and 0. For example. 0.E. An equivalent gallon is the number of BTUs in a gallon of diesel fuel. and 10 months for diesel. given that the engines are rebuilt on average every 20. with diesel and methanol a distant second and third. thus.. For biodiesel. as the current thin market for biodiesel may not reflect the long-run equilibrium price. at a price of one dollar per million BTUs of CNG. 0.2812. minus infrastructure cost for diesel Ž$1461. Incremental bus capital costs. however. 6. and CNGrdiesel-fueled buses. and CNGrdiesel buses are the mileage per month times the cost per mile Žfrom Tables 1–3. include fuel system costs and engine conversion costs. c I . plus methanol’s bus capital cost Ž$29 900. per equivalent gallon for CNGrdiesel were considered. Operating cost is composed of both maintenance and fuel costs. 35%.296 for methanol. Monthly maintenance cost is based on mileage per month.4218. equals startup costs of $38. Ahouissoussi. $0. respectively. respectively ŽTable . Startup costs.N. Based on this gross heating value. $2. for biodiesel were considered. total infrastructure cost per bus for methanol Ž$9858. including startup. Three prices per gallon Ž$1. These costs are comprised of incremental refueling infrastructure costs and incremental bus capital costs ŽTables 1–3.600 BTUs per gallon.

and maintenance costs associated with alternative-fueled bus transit systems. Estimated total cost is fuel cost plus an estimate of maintenance and opportunity costs. However.334 0.342 0.B.E.571 0.C.318 0. Biodiesel $1.1406rgal.75rgal. Biodiesel at only a 20% blend is competitive with methanol.044 0.216 0.660 0.454 0. Results from accounting for these other costs in a present value analysis are listed in the last four columns of Table 5.418 0.653 0. Blends 20% 35% 60% 100% $2.664 1.131 0.588 0.413 0.578 0.483 0.492 0.606 0.381 0.841 0.417 0.321 0.217 0.401 0.562 0.315 0. b Estimated cost is based on baseline maintenance costs of zero and 50% of actual explicit maintenance costs.446 0.00rgal.435 0. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 Table 5 Fuel cost and present value per mile of estimated total costs and actual explicit operating costs over a 30-year life cycle with an annual 5% discount rate a Estimated cost b Alternative fuel Fuel cost Zero 50% Diesel 0.289 0.12 N.4218rgal.394 0. 5.369 0. M.384 0.429 0. The salvage value of the engine is implicit in the replacement parameter r.445 0.380 0.267 0.2812rgal.366 0.441 0.008 0. At the end of 30 years.087 0. this fuel cost does not consider the differences in infrastructure.383 0.213 0.355 0.288 0.393 CNGrdiesel $0.314 0.325 0. $0.312 0.437 0.513 0.297 0.603 0.396 0.568 0.. Present value is calculated over a 30-year Ž360 months.365 0. Ahouissoussi. Opportunity costs are lost ridership and goodwill due to unexpected breakdowns. refueling.357 0. engine and fuel system.406 0.316 0.573 0.447 0. the infrastructure salvage value is assumed to be zero.569 0.389 0. life cycle of the refueling infrastructure. Blends 20% 35% 60% 100% $3.423 0. otherwise it is the highest priced fuel per mile.436 0.292 0.652 a Explicit operating cost is the sum of maintenance and fuel costs.354 0.449 0.312 0.390 0.487 0.310 0.323 0.454 0.365 0.200 0.655 0.50rgal.326 0.374 0.372 0.329 0. Blends 20% 35% 60% 100% Explicit cost Double mileage 0. The parameter r estimates the difference between .214 Methanol 0.319 0.439 0.484 0. $0.

this trajectory does not exist. Ž6. and CNGrdiesel. and 35 rebuilds. Based on Eq. however. iqj R q Ý isRMqMq1 1 120 q Ž1qd . all alternative-fueled buses are at the relatively same point of requiring another rebuild. Note that the last rebuild month Ž R = M . and CNGrdiesel. Again. Ahouissoussi. resulting from the relatively high level of explicit maintenance costs for methanol. 357. Specifically. i Ž 6. Thus. Thus. there is generally less than a 3% difference in per mile costs for all of the alternative-fueled buses. the present value of estimated total costs Ž PVC . and 10 months. represents the methanol trajectory off the rebuild cycle. and ending at 360 months. is only around 1%. and CNGrdiesel. For diesel and biodiesel. Their rebuild cycles of 20 and 10 are factors of 360. limited model sensitivity is apparent when varying maintenance costs.C. methanol’s rebuild cycle of 21 is not. methanol. is calculated as M PVC s cS q Ý R is1 js0 q cI 360 ci j Ý Ž1qd . the third term on the right-hand side of Ž6. is calculated and contrasted with the present value of total operating cost estimated solely on monthly mileage and time of rebuild. and R is 17.E. the difference in cost per mile Žcolumns four and five in Table 5. These per mile cost figures can be misleading because they do not consider variation in the intensity of bus utilization. at month 360. respectively. An exception is methanol-fueled buses. For diesel and biodiesel. M is 20. As indicated in Tables 1–3. 16. For comparison. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 13 the scrap value and the cost of installing a new engine or rebuilding the existing one. where c i j denotes monthly operating cost in the ith month after rebuild for the jth rebuild. the present value per mile of actual explicit operating costs Žcolumn five in Table 5.B. calculations of the present value per mile of estimated total costs with a 5% annual discount rate are presented in Table 5. methanol. where the difference is approximately 7%. is 340. Even in the case where estimated maintenance cost for CNGrdiesel is less than 8% of its associated explicit cost. respectively. respectively. M. The term RM q M q 1 represents the start of a trajectory off the rebuild cycle.. and 350 for diesel and biodiesel. Limited model sensitivity from this variation in maintenance costs is evident. When comparing these two levels of baseline maintenance cost. and thus their overall per mile costs should decline relative to diesel as the mileage per . 1 Ž1qd . a comparison of baseline maintenance costs of zero and 50% of actual explicit maintenance costs is provided in columns three and four of Table 5. Monthly discount rate is denoted as d . starting at 358 Ž RM q M q 1. As an illustration of the sensitivity of operating costs. Variables M and R represent the number of months in an engine rebuild cycle and number of rebuilds. 240 ci Ž1qd . 21.N. methanol and CNGrdiesel-fueled buses have higher fixed costs relative to diesel buses.

This competitiveness is underscored by the low infrastructure cost and lack of enginerfuel system cost of biodiesel relative to methanol and CNGrdiesel. cost associated with the risk of technology failure is potentially minimized with biodiesel. which is over three times the base diesel price.. Biodiesel does not require additional infrastructure over current diesel facilities. For example. However. nevertheless.14 N.00 per gallon are competitive with the other alternative fuels. a 35–60% blend at $2. biodiesel fuel can comply with regulatory emission standards. biodiesel fuels at prices as high as $3. restrictions in undertaking a project with a larger number of buses would primarily be concerned with the fuel cost differential between petroleum diesel and biodiesel fuels. Such an expanded project would increase the accuracy in any comparative fuel cost differences. This results in per mile cost estimates listed in the last column of Table 5. Ahouissoussi. Such an expanded project would diminish problems associated with a small sample. thereby possibly inflating the cost per mile of diesel and biodiesel buses compared with the alternatives.75 per gallon. As expected. This produced a lower than expected mean value of mileage at rebuild and months between rebuilds. As diesel is blended with biodiesel and the cost of biodiesel rises. and around a 35% blend at $3. As indicated in Tables 1–3. For example.B. potentially resulting in a large loss if the technology does not meet expectations.C. Especially for methanol.65. CNGrdiesel has a significantly lower cost per mile compared with methanol. Conversely. Assuming a 35% blend. converting a fleet of buses to methanol or CNG fuel requires substantial investment costs relative to diesel and biodiesel. it is still over 70% more expensive than diesel. it is still within 7% of the cost per mile for a methanol-fueled bus. the cost per mile increases to around $0.50. methanol-fueled buses experience a reduction of approximately 9%. However. Thus. These higher costs are predominantly irreversible once incurred. whereas CNGrdiesel buses experience a more modest decline of approximately 3%. biodiesel requires no additional infrastructure over current diesel facilities and only minor modifications in engine tuning. In contrast. the irreversible sunk cost associated with both CNGrdiesel and methanol-fueled buses restricts the expansion of these fuel systems from relatively small pilot projects resembling the Colorado project used for this analysis. The threshold at which biodiesel is competitive with CNGrdiesel on a cost-per-mile basis is between a 60–100% blend at $1. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 month increases. consider doubling the average monthly mileage under the assumption of zero baseline maintenance costs.E. The results presented in Table 5 would support a significantly broadened pilot project involving a larger number of buses over a longer time duration. in the sample of diesel bus engines. the higher fixed costs do result in a lower per mile cost as utilization increases.5% difference. . M.00. thus. per mile costs for diesel and biodiesel-fueled buses experience relatively minimal reductions in per mile costs Žless than a 0. one of the engines required a rebuild at only 12 150 miles. In all of these scenarios it is not surprising that diesel buses reflect the lowest cost per mile.

Seminar III. and butyl soybean esters as a renewable fuel for diesel engines.S. Jefferson City. DC.N. 1994. Biodiesel recycles carbon rather than pumping it from petroleum wells. Gavett. ethyl.. Biodiesel alert ŽNBB. Acknowledgements This research was partially funded by the USDA Office of Energy and New Uses. Econometrica 55..R.S. C. leading to economies of size. .. Energy balance of on-farm production and extraction of vegetable oil for fuel in the United States’ Island Northwest. National Biodiesel Board. Implications The findings of this analysis show that biodiesel is competitive with CNGrdiesel and methanol fuels. Alternative fuels for direct injection diesel engines. 1987. Rust. Washington. Jefferson City.. M. E. Washington. L. Vegetable Oil As A Diesel Fuel.Z. ARM-NC-28. Missouri. However. energy.. G. Agricultural Research Service. M.. Paper presented at the ASAE National Energy Symposium. Gavett.. J.. Thus. DC. Holmberg. S. ARM-NC-28. Optimal replacement of GMC bus engines: An empirical model of Harold Zurcher. biodiesel may also be very cost competitive compared with the methanol and CNG alternative fuels. Ahouissoussi. Merrill.E.. Kaufman. Ziejewski.. Overview of energy consumption in U. and environmental protection perspectives..E. and thus making any alternative fuel more competitive in the commercial marketplace. 1993.E. Withers. biodiesel is a fuel that does not contribute to the greenhouse effect. Peoria ŽUS Department of Agriculture..V. biodiesel is less competitive compared with petroleum diesel fuel.N. 1984. In the present situation of liquid fuel supply and at current crude oil prices. Kansas City. National Soydiesel Development Board standards for biodiesel ŽAmerican Biofuels Association. MO. E. R. W. 1993. S. Vegetable Oil As A Diesel Fuel. 999–1033..M. Agricultural Research Service. Wetzsteinr Resource and Energy Economics 20 (1997) 1–15 15 7. As suggested by the results of this study.C. 1993. there is no great incentive to find replacements for liquid fossil fuels. Due to its structural nature..L. P. Biodiesel represents one of the best alternatives as a renewable fuel for diesel engines from economic. agriculture. Seminar III.. Energy in Agriculture 3. Smith. Washington. References Clark.. Methyl.. Incentives will be necessary for further industry development. DC. and the National Biodiesel Board. Pratt. K.J. Peoria ŽUS Department of Agriculture. 1980.B.C. Wagner. compelling environmental or socioeconomic benefits must exist to warrant incentives for promoting alternative fuels. McIntosh. 155–166. MO.

Sign up to vote on this title
UsefulNot useful