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 Industrial Demand Module estimates energy consumption by energy source (fuels and feedstocks) for 9 manufacturing and 6 nonmanufacturing industries.  The manufacturing industries are further subdivided into the energy-intensive manufacturing industries and nonenergy-intensive manufacturing industries.  The distinction between the two sets of manufacturing industries pertains to the level of modeling.  The energy-intensive industries are modeled through the use of a detailed process flow accounting procedure, whereas the nonenergy-intensive and the nonmanufacturing industries are modeled with substantially less detail (Table 13).  The Industrial Demand Module forecasts energy consumption at the four Census region levels; energy consumption at the Census Division level is allocated by using the SEDS25 data.

The energy-intensive industries (food and kindred products, paper and allied products, bulk chemicals, glass and glass products, hydraulic cement, blast furnace and basic steel products, and primary aluminum) are modeled in considerable detail.  Each industry is modeled as three separate but interrelated components consisting of the Process Assembly (PA) Component, the Buildings Component (BLD), and the Boiler/Steam/Cogeneration (BSC) Component.  The BSC Component satisfies the steam demand from the PA and BLD Components.  In some industries, the PA Component produces byproducts that are consumed in the BSC Component.  For the energy-intensive industries, the PA Component is separated into the major production processes or end uses.  

Petroleum refining (Standard Industrial Classification 2911) is modeled in detail in a separate module of NEMS, and the projected energy consumption is included in the manufacturing total.  Forecasts of refining energy use and oil and gas lease and plant fuel and fuels consumed in cogeneration (Standard Industrial Classification 1311) are exogenous to the Industrial Demand Module, but endogenous to the NEMS modeling system.

Key Assumptions

The NEMS Industrial Demand Module primarily uses a bottom-up process modeling approach.  An energy accounting framework traces energy flows from fuels to the industry’s output.  An important assumption in the development of this system is the use of 1994 baseline Unit Energy Consumption (UEC) estimates based on analysis of the Manufacturing Energy Consumption Survey 1994.26  The UEC represents the energy required to produce one unit of the industry’s output.  The output may be defined in terms of physical units (e.g., tons of steel) or in terms of the dollar value of output.

The module depicts the seven most energy-intensive manufacturing industries (apart from petroleum refining, which is modeled in the Petroleum Market Module of NEMS) with a detailed process flow approach.  The dominant process technologies are characterized by a combination of unit energy consumption estimates and “technology possibility curves.”  The technology possibility curves indicate the energy intensity of new and existing stock relative to the 1994 stock over time.  Rates of energy efficiency improvements assumed for new and existing plants vary by industry and process.  These assumed rates were developed using professional engineering judgments regarding the energy characteristics, year of availability, and rate of market adoption of new process technologies.

Table 13.   Industry Categories

Process/Assembly Component

The Process/Assembly (PA) Component models each major manufacturing production step for the energy-intensive industries. The throughput production for each process step is computed as well as the energy required to produce it.

Within this component, the UEC is adjusted based on the technology possibility curves for each step.  For example, additions to waste fiber pulping capacity are assumed to require only 93 percent as much energy as does the average existing plant (Table 14).  The technology possibility curve is a means of embodying assumptions regarding new technology adoption in the manufacturing industry and the associated increased energy efficiency of capital without characterizing individual technologies.  It is unlikely that new technology is employed in all new capacity additions.  Many facilities will only partially incorporate the technology or will need time to debug the operating aspects of the newly installed capacity.  To some extent, all industries will increase the energy efficiency of their process and assembly steps.  The reasons for the increased efficiency are not likely to be directly attributable to changing energy prices but due to other exogenous factors.  Since the exact nature of the technology improvement is too uncertain to model in detail, the module employs a technology possibility curve to characterize the bundle of technologies available for each process step.

Fuel shares for process and assembly energy use in six of the energy-intensive manufacturing industries27 are adjusted for changes in relative fuel prices.  The six industries are food, paper, chemicals, glass, cement, and steel.  In each industry, two logit fuel-sharing equations are applied to revise the initial fuel shares obtained from the process-assembly component.  The resharing does not affect the industry’s total energy use-only the fuel shares.  The methodology adjusts total fuel shares across all process stages and vintages of equipment to account for aggregate market response to changes in relative fuel prices. 

Table 14.   Coefficients for Technology Possibility Curve

The fuel share adjustments are done in two stages. The first stage determines the fuel shares of electricity and nonelectricity energy.  The latter group excludes boiler fuel and feedstocks.  The second stage determines the fossil fuel shares of nonelectricity energy.   In each case, a new fuel-group share, NEWSHR_i, is established as a function of the initial, default fuel-group shares, DEFLTSHR_j and fuel-group prices indices, PRCRAT_i.  The price indices are the ratio of the current year price to the base year price, in real dollars.  The formulation is as follows:

The coefficients b_j are all assumed to be 0.2.   

The form of the equation results in unchanged fuel shares when the price indices are all 1, or unchanged from their 1997 levels.  The implied own-price elasticity of demand is about -0.1.

Byproducts produced in the PA Component serve as fuels for the BSC Component.  In the industrial module, byproducts are assumed to be consumed before purchased fuel.

Buildings Component

The total buildings energy demand by industry for each region is the product of the building UEC and regional industrial employment.  Building UEC’s were derived by first estimating energy requirements for building lighting, air conditioning, and space heating, where space heating was further divided to estimate the amount provided by direct combustion of fossil fuels and that provided by steam (Table 15).  Energy consumption in the BLD Component for an industry is assumed to grow at the same rate as regional employment for that industry.

Table 15.   Building Component Unit Energy Consumption (Trillion Btu/Thousand People Employed)

Boiler/Steam/Cogeneration Component

The steam demand and byproducts from the PA and BLD Components are passed to the BSC Component, which applies a heat rate and a fuel share equation (Table 16) to the boiler steam requirements to compute the required energy consumption.  

Table 16.   Logit Function Parameters for Estimating Boiler Fuel Shares

The boiler fuel shares are calculated using a logit formulation.  The equation is calibrated to 1994 so that the actual boiler fuel shares are produced for the relative prices that prevailed in 1994.  The equation for each manufacturing industry is as follows:

where the fuels are coal, petroleum, and natural gas.  The P_i are the fuel prices; a_i are sensitivity parameters; and the b_i are calibrated to reproduce the 1994 fuel shares using the relative prices that prevailed in 1994.  The byproduct fuels are consumed before the quantity of purchased fuels is estimated.  The boiler fuel shares are assumed to be those estimated using the 1994 MECS.28

Technology

The amount of energy consumption reported by the industrial module is also a function of vintage of the capital stock that produces the output.  It is assumed that new vintage stock will consist of state-of-the-art technologies that are more energy efficient than the average efficiency of the existing capital stock. Consequently, the amount of energy required to produce a unit of output using new capital stock is less than that required by the existing capital stock.  Capital stock is grouped into three vintages: old, middle, and new.  The old vintage  consists of capital in production  prior to 1995 and is assumed to retire at a fixed rate each year (Table 17).  Middle vintage capital is that which is added after 1994 but not including the year of the forecast.  New production capacity is built in the forecast years when the capacity of the existing stock of capital in the industrial model cannot produce the output forecasted by the NEMS Regional Macroeconomic Model.  Capital additions during the forecast horizon are retired in subsequent years at the same rate as the pre-1995 capital stock.

Table 17.   Retirement Rates

The energy intensity of the new capital stock relative to 1994 capital stock is reflected in the parameter of the technology possibility curve estimated for the major production steps for each of the energy-intensive industries.  These curves are based on engineering judgment of the likely future path of energy intensity changes (Table 14).  The energy intensity of the existing capital stock also is assumed to decrease over time, but not as rapidly as new capital stock.  The net effect is that over time the amount of energy required to produce a unit of output declines.  Although total energy consumption in the industrial sector is projected to increase, overall energy intensity is projected to decrease.

Cogeneration

Cogeneration (the generation of electricity and steam) has been a standard practice in the industrial sector for many years.  The cogeneration estimates in the module are based on the assumption that the historical relationship between industrial steam demand and cogeneration will continue in the future.  The data source is Form EIA-867, Annual Nonutility Power Producer Report, consisting of data from approximately 400 cogenerators for 1989-1994.

Legislation

Energy Policy Act of 1992 (EPACT)

EPACT and the Clean Air Act Amendments of 1990 (CAAA90) contain several implications for the industrial module.  These implications fall into three categories: coke oven standards; efficiency standards for boilers, furnaces, and electric motors; and industrial process technologies.  The industrial module assumes the leakage standards for coke oven doors do not reduce the efficiency of producing coke or increase unit energy consumption.  The industrial module uses heat rates of 1.25 (80 percent efficiency) and 1.22 (82 percent efficiency) for gas and oil burners respectively.  These efficiencies meet the EPACT standards.  The standards for electric motors call for a 10-percent efficiency increase.  The industrial module incorporates a 10-percent savings for state-of-the-art motors increasing to 20-percent savings in 2015.  Given the time lag in the legislation and the expected lifetime of electric motors, no further adjustments are necessary to meet the EPACT standards for electric motors.  The industrial module incorporates the necessary reductions in unit energy consumption for the energy-intensive industries.

Climate Change Action Plan

Several programs included in the Climate Change Action Plan (CCAP) target the industrial sector.  Note that the potential impacts of the Climate Wise Program are also included in the CCAP impacts.  The intent of these programs is to reduce greenhouse gas emissions by lowering industrial energy consumption.  The Department of Energy (DOE) program offices estimated that full implementation of these programs would reduce industrial electricity consumption by 79 billion kilowatthours and fossil energy consumption by 359 trillion Btu by 2010.  However, since the energy savings associated with the voluntary programs in the CCAP largely duplicate savings that would have occurred in their absence and since some programs were not fully funded, total CCAP energy savings were reduced.  The Annual Energy Outlook 1999 (AEO99) assumes that CCAP reduces electricity consumption by 41 billion kilowatthours and fossil energy consumption by 90 trillion Btu in 2010.  The fossil energy is assumed to be 85 percent natural gas and 15 percent steam coal. In this situation, carbon emissions would be reduced by about 7 million metric tons (1 percent) in 2010.

High Technology and 1999 Technology Cases

The high technology case assumes earlier availability, lower costs, and higher efficiency for more advanced equipment.29 Changes in aggregate energy intensity result both from changing equipment and production efficiency and from changes in the composition of industrial output. Since the composition of industrial output remains the same as in the reference case, aggregate intensity declines by only 1.4 percent annually even though the intensity declines for some individual industries doubles. In the reference case, aggregate intensity declines by 1.0 percent annually.

AEO99 also analyzed an integrated high technology case (consumption high technology), which combines the high technology cases of the four end-use demand sectors and the electricity high fossil technology case.

1999 technology case holds the energy efficiency of plant and equipment constant at the 1999 level over the forecast.  Both cases were run with only the Industrial Demand Module rather than as a fully integrated NEMS run, (i.e., the other demand models and the supply models of NEMS were not executed).  Consequently, no potential feedback effects from energy market interactions were captured.