Report#:DOE/EIA-0554(99)

Introduction

Macroeconomic Activity

International Energy

Household Expenditure

Residential Demand

Commercial Demand

Industrial Demand

Transportation Demand

Electricity Market

Oil and Gas Supply

Natural Gas Transmission & Distribution

Petroleum Market

Coal Market

Renewable Fuels

The NEMS Transportation Demand Module estimates energy consumption across the nine Census Divisions and over ten fuel types.  Each fuel type is modeled according to fuel-specific technology attributes applicable by transportation mode.  Total transportation energy consumption is the sum of energy use in eight transport modes:  light-duty vehicles (cars, light trucks, industry sport utility vehicles and vans), commercial light trucks (8501-10,000 lbs), freight trucks (>10,000 lbs), freight and passenger airplanes, freight rail, freight shipping, mass transit, and miscellaneous transport such as mass transit. Light-duty vehicle fuel consumption is further subdivided into personal usage and commercial fleet consumption.

Key Assumptions

Macroeconomic Sector Inputs

Macroeconomic sector inputs used in the NEMS Transportation Demand Module (Table 18) consist of the following:  gross domestic product (GDP), industrial output by Standard Industrial Classification code, personal disposable income, new car and light truck sales, total population, driving age population, total value of imports and exports, and the military budget.  The share of total vehicle sales that represent light truck sales is assumed to approach forty-six percent by 2020.

Table 18.   Macroeconomic Inputs to the Transportation Module (Millions)

Light-Duty Vehicle Assumptions

The light duty vehicle Fuel Economy Module includes 59 fuel saving technologies with data specific to car and light truck including incremental fuel efficiency improvement, incremental cost, first year of introduction, and fractional horsepower change.  These assumed technology characterizations are scaled up or down to approximate the differences in each attribute for 6 EPA size classes of cars and light trucks (Tables 19 and 20).

Table 19.   Standard Technology Matrix For Cars

Table 20.  Standard Technology Matrix For Trucks

The vehicle sales share module holds vehicle sales shares by import and domestic manufacturers constant within a vehicle size class at 1997 level from the National Highway Traffic and Safety Administration data.30

The fuel economy module utilizes 59 new technologies for each size class and origin of manufacturer (domestic or foreign) based on the cost-effectiveness of each technology and an initial availability year. The discounted stream of fuel savings is compared to the marginal cost of each technology.  The fuel economy module assumes the following:

• All fuel saving technologies have a 4-year payback period.

• The real discount rate remains steady at 8 percent.

• Corporate Average Fuel Efficiency standards remain constant at 1998 levels.

• Expected future fuel prices are calculated based on an extrapolation of the growth rate between fuel prices 3 years and 5 years prior to the present year.  This assumption is founded upon an assumed lead time of 3 to 5 years to significantly modify the vehicles offered by a manufacturer.

• Degradation factors (Table 21) used to convert Environmental Protection Agency-rated fuel economy to actual “on the road” fuel economy are based on application of a logistic curve to the projections of three factors:  increases in city/highway driving, increasing congestion levels, and rising highway speeds.31, 32, 33, 34   Degradation factors are also adjusted to reflect the percentage of reformulated gasoline consumed.

Table 21. The Average Length of Time Vehicles Are Kept Before They are Sold to Others (Months)

• The vehicle miles traveled (VMT) module forecasts VMT as a function of the cost of driving per mile, income per capita, ratio of female to male VMT, and age distribution of the driving population (Figure 4).  The ratio of female to male VMT is assumed to asymptotically approach 100 percent by 2010. VMT per driver by age group was also assumed to be more uniformly distributed to younger and older age groups. AEO99 assumed that all age group VMT per driver approaches one-half of the difference between itself and the maximum VMT per driver age category.  Total VMT is calibrated to Federal Highway Administration VMT data. 34, 35 The fuel price elasticity rises from -0.05 to -0.2 as fuel prices rise above reference case levels in each year.

• The share of light truck sales is assumed to reach a maximum of 46 percent of total sales by 2010. However, the light truck share will gradually decline to 41 percent if fuel prices rise to approximately $1.50/gal. The size class sales shares will also gravitate to 26 percent for subcompacts, 40 percent for compacts, 23 percent for mid size, and 10 percent for luxury if fuel prices exceed reference case levels approximately $1.50/gal.

Figure 4. VMT per Driver by Age Group

Commercial Light-Duty Fleet Assumptions

With the current focus of transportation legislation on commercial fleets and their composition, the Transportation Demand Module has been redesigned to divide commercial light-duty fleets into three types of fleets:  business, government, and utility.  Based on this classification, commercial light-duty fleet vehicles vary in survival rates and duration in the fleet, before being combined with the personal vehicle stock (Table 22). Sales shares of fleet vehicles by fleet type also remain constant over the forecast period.  Automobile fleets are divided into the following shares:  business (87.39%), government (7.42%), and utilities (5.19%).  Light truck fleets are divided into the following shares:  business (83.50%), government (14.1%), and utilities  (2.40%)36, 37.  Both car (23.70%) and light truck (28.57%) fleet sales are assumed to be a constant fraction of total car and light truck sales.

Table 22. Car and Light Truck Degradation Factors

Alternative-fuel shares of fleet sales by fleet type are initially set according to historical shares (business (0.36%), government (2.21%), utility (2.64%))38, 39 then compared to a minimum constraint level of sales based on legislative initiatives, such as the Energy Policy Act and the Low Emission Vehicle Program.38, 39, 40, 41  Size class sales of alternative-fuel and conventional vehicles are held constant at anticipated levels (Table 23).42 Individual sales shares of alternative-fuel fleet vehicles by technology type are assumed to remain at anticipated levels for utility, government, and for business fleets in accordance with the technology shares applied from EIA surveys36, 37 (Table 24).

Table 23. Commercial Fleet Size Class Shares by Fleet and Vehicle Type 1992

Table 24. Anticipated Purchases of Alternative-Fuel Vehicles by Fleet Type and Technology Type (Percentage)

Annual VMT per vehicle by fleet type stays constant over the forecast period based on the Oak Ridge National Laboratory fleet data.

Fleet fuel economy for both conventional and alternative-fuel vehicles is assumed to be the same as the personal new vehicle fuel economy and is subdivided into three size classes.

The Light Commercial Truck Module

The Light Commercial Truck Module of the NEMS Transportation Model is constructed to represent trucks that weight 8501 lbs. to 10,000 lbs.  These vehicles are assumed to be used for commercial freight purposes.

The primary source of data for this model is the microdata file of the 1992 Truck Inventory and Use Survey (TIUS), which provides numerous details on truck stock and usage patterns at a high level of disaggregation.  The data derived from this source are used to allocate and sort the summary truck data presented in the Federal Highway Administration’s annual publication of highway statistics, which constitute the baseline from which the NEMS forecast is made (Figure 5).  TIUS data are also used to distribute estimated sales of trucks, obtained from the Macroeconomic Model, among the affected models according to their weight class (Figure 6).  Finally, the TIUS microdata set is used to construct a characterization of these Light Commercial

Figure 5. Distribution of FHWA Single-Unit Truck Stocks

Figure 6. Distribution of Light Truck Sales

Trucks, comprising their average annual miles of travel, fuel economy, and distribution among several aggregate industrial groupings chosen for their correspondence with output measures currently being forecast by NEMS (Tables 25 and 26).  It is expected that projected growth in industrial output will provide a useful proxy for the growth in demand for the services of light commercial trucks.

Over the forecast period 1997-2020 VMT for light commercial trucks is a function of industrial output for agriculture, mining, construction, trade, utilities, and personal VMT. Forecasted fuel efficiencies are assumed to increase at the same annual growth rate of light-duty trucks (<8500 lbs.).

Table 25.   Anticipated Annual Miles, by Major Use (1992 TIUS) (Aggregated for NEMS)

Table 26. Average Miles Per Gallon: Biweighted Mean Iterated

Alternative-Fuel Vehicle Technology Choice Assumptions

The alternative-fuel vehicle (AFV) technology choice module utilizes a discrete choice specification, which uses vehicle attributes as inputs and forecasts vehicle sales shares among the following 15 light-duty technologies:  gasoline internal combustion engine (ICE), direct injection diesel ICE, ethanol flex, ethanol neat, methanol flex, methanol neat, electric dedicated (uses only electricity), diesel electric hybrid, compressed natural gas (CNG), CNG bi-fuel, LPG, LPG bi-fuel, fuel cell gasoline, fuel cell methanol, and fuel cell liquid hydrogen.43 Direct injection gasoline and gasoline electric hybid technologies are included in the conventional gasoline ICE technologies.

Listed in Table 27 are a few examples of the input variables that correspond to the vehicle attributes used in the analysis.  With the exception of vehicle fuel economy, vehicle price, and vehicle range, all other attributes are exogenously set, based on offline analysis.44, 45

Table 27. Alternative-Fuel Vehicle Attribute Inputs For Three Stage Logit Model

Vehicle attributes vary by six EPA size classes for cars and light trucks,  and fuel availability varies by Census Division.  It is assumed that the logit model coefficients can be used for both estimates for future sales shares of both cars and light trucks separately.  Vehicle prices are assumed to follow exponential curves of economies of scale in production dependent upon the volumes and cost curves which vary by AFV technologies.  Where applicable, AFV fuel efficient technologies are calculated relative to conventional gasoline miles per gallon.  It is assumed that many fuel efficiency improvements to conventional vehicles will be transferred to alternative-fuel vehicles.  Specific individual alternative-fuel technological improvements are also handled dependent upon the AFV technology type, cost, research and development, and availability over time.  Commercial availability estimates are assumed values according to a logistic curve based on the initial technology introduction date and were constructed in cooperation with the Office of Energy Efficiency and Renewable Energy of the Department of Energy (DOE).  Coefficients summarizing consumer valuation of vehicle attributes were derived from a stated preference survey conducted in the U.S.46 and are assumed to be representative of the United States. Initial AFV vehicle stocks are set according to EIA surveys.36, 37 A fuel switching algorithm based on the relative fuel prices for AF compared to gasoline is used to detrmine the percentage of total VMT represented by AF in bi-fuel and flex-fuel alcohol vehicles. An upper limit of 50 percent and a lower limit of 25 percent is assumed.

Freight Truck Assumptions

The freight stock truck module converts industrial output in dollar terms to an equivalent measure of volume by using a freight adjustment coefficient.47, 48  These freight truck adjustment coefficients vary by NEMS Standard Industrial Classification (SIC) code, gradually diminishing their deviation over time and are estimated from historical freight data. Freight truck load factors (ton-miles per truck) by SIC code are constants formulated from historical load factors39.  All freight trucks are subdivided into  medium, and heavy-duty trucks.  Freight truck fuel efficiency growth rates relative to fuel prices are tied to historical growth rates by size class and are also dependent on the maximum penetration, introduction year, fuel trigger price (based on cost-effectiveness), and fuel economy improvement of the technologies including alternative fuel technologies (Table 28).49  VMT freight estimates by size class and technology are based on matching freight needs as measured by the growth in industrial output by SIC code to VMT levels associated with truck stocks and new vehicles.  Fuel consumption by freight trucks is regionalized according to the State Energy Data Report distillate regional shares.50

Initial freight trucks are obtained by the Federal Highway Administration (FHWA) and are distributed by Truck and Inventory Use Survey (TIUS) shares.

Table 28. Diesel Technology Characteristics for the Freight Truck Model

Freight and Transit Rail Assumptions

The freight rail module receives industrial output by SIC code measured in real 1987 dollars and converts these dollars into an adjusted volume equivalent.  Specific NEMS coal production from the Coal Module is also used to adjust coal rail travel. Freight rail adjustment coefficients, which are used to convert dollars into volume equivalents, remain constant and are based on historical data.47, 48   Initial freight rail efficiencies are based on the freight model from Argonne National Laboratory.52  The distribution of rail fuel consumption by fuel type remains constant and is based on historical data (Table 28).51  Regional freight rail consumption estimates are distributed according to the State Energy Data Report 1994.50

Freight Domestic and International Shipping Assumptions

The freight domestic shipping module also converts industrial output by SIC code measured in dollars, to a volumetric equivalent by SIC code.47, 48  These freight adjustment coefficients are based on analysis of historical data and remain constant throughout the forecast period.  Domestic shipping efficiencies are based on the freight model by Argonne National Laboratory.  The energy consumption in the freight international shipping module is a function of the total level of imports and exports.  The distribution of domestic and international shipping fuel consumption by fuel type remains constant throughout the analysis and is based on historical data.  Regional domestic and international shipping consumption estimates are distributed according to the State Energy Data Report residual oil regional shares.50

Air Travel Demand Assumptions

The air travel demand module calculates the ticket price for travel as a function of fuel cost.  Similar to the light-duty vehicle module, the air travel fuel price elasticity rises from -0.04 to -0.2 if jet fuel prices exceed reference case levels. A demographic index based on the propensity to fly was introduced into the air travel demand equation.53, 54  The propensity to fly was made a function of the age and sex group distribution over the forecast period.55, 56  The air travel demand module assumes that these relationships between the groups and their propensity to fly remain constant over time.  International revenue passenger miles are calculated as a percentage of domestic revenue passenger miles based on an extrapolation of historical data, which asymptotically approaches 56 percent by 2020.57 The revenue ton miles of air freights are based on merchandise exports and gross domestic product.

Aircraft Stock/Efficiency Assumptions

The aircraft stock and efficiency module consists of a stock model of both wide and narrow body planes by vintage.  The shifting of passenger load between narrow and wide body aircraft occurs at a constant historical annual 1-percent rate.58  The available seat-miles per plane, which measure the carrying capacity of the airplanes by aircraft type, remain constant and are based on holding the seat-miles and the number of planes constant within an aircraft type.58 The difference between the seat-miles demanded and the available seat-miles represents newly purchased aircraft.  Aircraft purchases in a given year cannot exceed historical annual growth rates, a constraint that sets an upper limit on the application of new aircraft to meet the gap between seat-miles demanded and available seat-miles.  With a constraint on new aircraft purchases, it is assumed that when the gap exceeds historical aircraft sales levels, planes that have been temporarily stored or retired will be brought back into service.  Technological availability, economic viability, and efficiency characteristics of new aircraft are based on the technologies listed in the Oak Ridge National Laboratory Air Transport Energy Use Model. (Table 31)59  Fuel efficiency of new aircraft acquisitions represents, at a minimum, a 5-percent improvement over the stock efficiency of surviving airplanes.59  Maximum growth rates of fuel efficiency for new aircraft are based on a future technology improvement list consisting of an estimate of the introduction year, jet fuel price, and an estimate of the proposed marginal fuel efficiency improvement.  Regional shares of all types of aircraft fuel are assumed to be constant and are consistent with the State Energy Data Report estimate of regional jet fuel shares.

Legislation

Energy Policy Act of 1992 (EPACT)

Fleet alternative-fuel vehicle sales necessary to meet the EPACT regulations were derived based on the mandates as they currently stand and the Commercial Fleet Vehicle Module calculations.  Total projected AFV sales are divided into fleets by government, business, and fuel providers (Table 29).  Although inclusion of the business fleet is dependent upon a rulemaking by the Secretary of Energy, the assumption is that fuel displacement goals set in EPACT can only be reached by inclusion of the business fleet.  It is assumed that business fleet EPACT mandates do not take effect until the year 2002 based on the late mandated schedule of proposed rulemaking.

Table 29. EPACT Legislative Mandates for Percentage AFV Purchases by Fleet Type, Year

Because the commercial fleet model operates on three fleet type representations (business, government, and utility), the federal and state mandates were weighted by fleet vehicle stocks to create a composite mandate for both. The same combining methodology was used to create a composite mandate for electric utilities and fuel providers based on fleet vehicle stocks.36, 37 Fleet vehicle stocks by car and light truck were disaggregated to include only fleets of 50 or more (in accordance with EPACT) by using a fleet size distribution function based on The Fleet Factbook and the Truck and Inventory Use Survey.38, 39  To account for the EPACT regulations which stipulate that “covered” fleets (which refers to fleets bound by the EPACT mandates) include only fleets in the metropolitan statistical areas (MSA’s) of 250,000 population or greater, 90 percent of the business and utility fleets were included and 63 percent were included for government fleets.40   EPACT covered fleets were to only include those fleets that could be centrally fueled, which was assumed to be 50 percent of the fleets for all fleet types, and only fleets of 50 or more that had 20 vehicles or more in those MSA’s of 250,000 or greater population; it was assumed that 90 percent of all fleets were within this category except for business fleets, which were assumed to be 75 percent.40

Low Emission Vehicle Program (LEVP)

The LEVP, which began in California, which was originally instituted in New York and Massachusetts, has now been rolled back to begin in 2003 at the original 10 percent mandate for California, Massachusetts and New York.  The following Zero Emission Vehicle (ZEV) sales percentage numbers (Table 30) come from the California Air Resources Board.60 All of the ULEV sales were assumed to meet the ULEV air standards with reformulated gasoline and a heated catalytic converter.

Table 30. Original and Revised California Low Emission Vehicle Program Legislatively Mandated Alternative-Fuel Vehicle Sales (Percentage)

Table 31. Future New Aircraft Technology Improvement List

The AFV sales module compares these legislatively mandated sales to the results from the AFV logit market-driven sales shares.  The legislatively mandated sales serve as a minimum constraint to AFV sales (Table 32).

Table 32. EPACT Alternative-Fuel fleet Sale Estimates

Climate Change Action Plan

There were four programs implemented from the Climate Change Action Plan (CCAP) transportation policies—reform Federal subsidy for employer-provided parking, adopt a transportation system efficiency strategy, promote telecommuting, and develop fuel economy labels for tires.  The combined effect of the Federal subsidy, system efficiency, and telecommuting policies was a reduction in VMT of 1.6 percent in 2010, representing a decline in consumption of approximately 270 trillion Btu which increases to 2.45 percent VMT reduction and a decline in fuel consumption of 470 trillion Btu by 2020.  The fuel economy tire labeling program improved fuel efficiency by 4 percent among vehicles that switched to low rolling resistance tires, and resulted in a reduction in fuel consumption of 1 trillion Btu by 2010.  Total reductions of carbon emissions from CCAP reach 6.5 million metric tons per year by 2010.

Advanced Technology and 1999 Technology Cases

In the advanced technology case, the light-duty vehicle assumptions for alternative fuel vehicles are presented in Table 33 and are based on the yearly U.S. Department of Energy Office of Energy Efficiency and Renewables Office of Transportation Technologies (OTT) Program Analysis62.  The conventional fuel saving technology characteristics come from a study by the American Council For an Energy Efficient Economy.61 In the advanced technology case, fuel efficiency improvements from new technology more than offset the increasing travel in each transportation mode.  As a result, the total energy consumption in the transportation sector was 7.9 percent lower (2.90 quadrillion Btu) than in the reference case by 2020.

Table 33. Alternative-Fuel Large Car Vehicle Assumptions Relative to Conventional Gasoline Vehicle, 2020 (Thousands)

The 1999 technology case assumes that new fuel efficiency technologies are held constant at 1999 levels over the forecast.  As a result, the energy use in the transportation sector was 7.5 percent higher (2.77 quadrillion Btu) than in the reference case by 2020. Both cases were run with only the transportation demand module rather than as a fully integrated NEMS run. Consequently, no potential macroeconomic feedback on travel demand, or fuel economy was captured.

Freight trucks in the advanced technology case were constructed in accordance with the assumptions from a Department of Energy (DOE) study.49 The following technologies were made commercially available within the forecast period: advanced drag reduction, turbocompound diesel engine, heat engine CLE-55, and reduced empty weight technologies.  Additionally, shorter market penetration periods, and technology prices were made cost-effective at $6/MMBtu for diesel fuel, instead of the range of $8-28.60/MMBtu in the AEO99 reference case.

The air model assumptions for the advanced technology case were also constructed to replicate the assumptions in the DOE interlab study.43  Aircraft load factors were increased to 69% for domestic and 72% for international travel.  Efficiency improvements were approximately 51% higher than the 1997 levels for new aircraft by 2020, which is the equivalent of a 1.8% annual growth rate.

Table 34. High Technology Matrix For Cars

Table 35. High Technology Matrix For Trucks