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 Commercial Sector Demand Module generates forecasts of commercial sector energy demand through 2020.  The definition of the commercial sector is consistent with EIA’s State Energy Data System (SEDS). That is, the commercial sector includes business establishments that are not engaged in transportation or in manufacturing or other types of industrial activity (e.g., agriculture, mining or construction).  The bulk of commercial sector energy is consumed within buildings, however, street lights, pumps, bridges, and public services are also included if the establishment operating them is considered commercial.  Since most of commercial energy consumption occurs in buildings, the commercial module relies on the data from the EIA Commercial Buildings Energy Consumption Survey (CBECS) for characterizing the commercial sector activity mix as well as the equipment stock and fuels consumed to provide end use services.12

The commercial module forecasts consumption by fuel13 at the Census Division level using prices from the NEMS energy supply modules, macroeconomic variables from the NEMS Macroeconomic Activity Module (MAM), as well as external data sources (technology characterizations, for example).  Energy demands are forecast for ten end-use services14 for eleven building categories15 in each of the nine Census Divisions.  The model begins by developing forecasts of floorspace for the 99 building category and Census Division combinations.  Next, the ten end-use service demands required for the projected floorspace are developed. Technologies are then chosen to meet the projected service demands for the seven major end uses.16  Once technologies are chosen, the energy consumed by the equipment stock (both previously existing and purchased equipment) chosen to meet the projected end-use service demands is developed.17

Key Assumptions

The key assumptions made by the commercial module are presented in terms of the flow of the calculations described above.  Each section below will summarize the assumptions in each of the commercial module submodules:  floorspace, service demand, technology choice, and end-use consumption.  The four submodules are executed sequentially in the order presented, and the outputs of each submodule become the inputs to subsequently executed submodules.  As a result, key forecast drivers for the floorspace submodule are also key drivers for the service demand submodule, and so on.

Floorspace Submodule

Floorspace is forecast by starting with the previous year’s stock of floorspace and eliminating a certain portion to represent the removal of buildings.  Total floorspace is the sum of the surviving floorspace plus new additions to the stock derived from the Macroeconomic Activity Module’s floorspace projection.18

Existing Floorspace and Attrition

Existing floorspace is based on the estimated floorspace reported in the Commercial Buildings Energy Consumption Survey 1995 (Table 9).  Over time the 1995 stock is projected to decline as buildings are removed from service (floorspace attrition).  Floorspace attrition is estimated by a logistic decay function, the shape of which is dependent upon the values of two parameters: average building lifetime and gamma. Gamma controls the acceleration of the rate of retirement around the average building lifetime.  The current values for the average building lifetime and gamma are 59 years and 5.4, respectively.19

Table 9.   1995 Total Floorspace by Census Division and Principal Building Activity (Millions of Square Feet)

New Construction Additions to Floorspace

The commercial module develops estimates of projected commercial floorspace additions by combining the surviving floorspace estimates with the Data Resources, Inc. (DRI) total floorspace forecast from MAM. A total NEMS floorspace projection is calculated by applying DRI’s assumed floorspace growth rate within each Census Division and DRI building type to the corresponding NEMS Commercial Demand Module’s building types based on the CBECS building types shares. The NEMS surviving floorspace from the previous year is then subtracted from the total NEMS floorspace projection for the current year to yield new floorspace additions.20

Service Demand Submodule

Once the building stock is projected, the Commercial Demand module develops a forecast of demand for energy-consuming services required for the projected floorspace.  The module projects service demands for the following explicit end-use services:  space heating, space cooling, ventilation, water heating, lighting, cooking, refrigeration, personal computer office equipment, and other office equipment.21 The service demand intensity (SDI) is measured in thousand Btu of end-use service demand per square foot and differs across service, Census Division and building type.  The SDIs are based on a hybrid engineering and statistical approach of CBECS consumption data.22  Projected service demand is the product of square feet and SDI for all end uses across the eleven building categories with adjustments for changes in shell efficiency for space heating and cooling.

Shell Efficiency

The shell integrity of the building envelope is an important determinant of the heating and cooling loads for each type of building.  In the NEMS Commercial Demand Module, the shell efficiency  is represented by an index, which changes over time to reflect improvements in the building shell.  This index is dimensioned by building type and Census Division and applies directly to heating.  For cooling, the effects are computed from the index, but differ from heating effects, because of different marginal effects of shell integrity and because of internal building loads.  In the AEO99 reference case, shell improvements for new buildings are up to 24 percent more efficient than the 1995 stock of similar buildings. Over the forecast horizon, new building shells improve in efficiency by 6 percent relative to their efficiency in 1995.  For existing buildings, efficiency is assumed to increase by 4 percent over the 1995 stock average.  The shell efficiency index affects the space heating and cooling service demand intensities causing changes in fuel consumed for these services as the shell integrity improves.

Technology Choice Submodule

The technology choice submodule develops projections of the results of the capital purchase decisions for equipment fueled by the three major fuels (electricity, natural gas, and distillate fuel).  Capital purchase decisions are driven by assumptions concerning behavioral rule proportions and time preferences as well as projected fuel prices, average utilization of equipment (the “capacity factors”), relative technology capital costs, and operating and maintenance (O&M) costs.  

Decision Types

In each forecast year, equipment is potentially purchased for three “decision types”.  Equipment must be purchased for newly added floorspace and to replace a proportion of equipment in existing floorspace projected to wear out.23  Equipment is also potentially purchased for retrofitting equipment which has become economically obsolete.  The purchase of retrofit equipment occurs only if the annual operating costs of a current technology exceed the annualized capital and operating costs of a technology available as a retrofit candidate.

Behavioral Rules

The commercial module allows the use of three alternate assumptions about equipment choice behavior.  These assumptions constrain the equipment choice among three choice sets, which are progressively more restrictive. The choice sets vary by decision type and building type:

• Unrestricted Choice Behavior - This rule assumes that commercial consumers consider all types of equipment that meet a given service, across all fuels, when faced with a capital purchase decision.

• Same Fuel Behavior - This rule restricts the capital purchase decision to the set of technologies that consume the same fuel that currently meets the decision maker’s service demand.  

• Same Technology Behavior - Under this rule, commercial consumers consider only the available models of the same technology and fuel that currently meet service demand, when facing a capital stock decision.  

Under any of the above three behavior rules, equipment that meets the service at the lowest annualized lifecycle cost is chosen.  Table 10 illustrates the proportions of floorspace subject to the different behavior rules for space heating technology choices in large office buildings.

Table 10.   Assumed Behavior Rules for Choosing Space Heating Equipment in Large Office Buildings (Percent)

Time Preferences

The time preferences of owners of commercial buildings are assumed to be distributed among six alternate time preference premiums (Table 11).  Adding the time preference premiums to the 10-year Treasury Bill rate results in implicit discount rates, also known as hurdle rates, applicable to the assumed proportions of commercial floorspace.  The effect of the use of this distribution of discount rates is to prevent a single technology from dominating purchase decisions in the lifecycle cost comparisons.  The distribution used for AEO99 assigns some floorspace a very high discount or hurdle rate to simulate floorspace which will never retrofit existing equipment and which will only purchase equipment with the lowest capital cost.  Discount rates for the remaining five segments of the distribution get progressively lower, simulating increased sensitivity to the fuel costs of the equipment that is purchased.   

Table 11.   Assumed Distribution of Time Preference Premiums

Technology Characterization Database

The technology characterization database organizes all relevant technology data by end use, fuel, and Census Division.  Equipment is identified in the database by a technology index as well as a vintage index, the index of the fuel it consumes, the index of the service it provides, its initial market share, the Census Division index for which the entry under consideration applies, its efficiency (or coefficient of performance; efficacy in the case of lighting equipment), installed capital cost per unit of service demand satisfied, operating and maintenance cost per unit of service demand satisfied, average service life, year of initial availability, and last year available for purchase.  Equipment may only be selected to satisfy service demand if the year in which the decision is made falls within the window of availability.  Equipment acquired prior to the lapse of its availability continues to be treated as part of the existing stock and is subject to replacement or retrofitting. This flexibility in limiting equipment availability allows the direct modeling of equipment efficiency standards. Table 12 provides a sample of the technology data for space heating in the New England Census Division.

End-Use Consumption Submodule

The end-use consumption submodule calculates the consumption of each of the three major fuels for the ten end-use services plus fuel consumption for Cogeneration and district services.  For the ten end-use services, energy consumption is calculated as the end-use service demand met by a particular type of equipment divided by its efficiency and summed over all existing equipment types.  This calculation includes dimensions for Census Division, building type and fuel.  Consumption of the five minor fuels is forecast based on historical trends.

Equipment Efficiency

The average energy consumption of a particular appliance is based initially on estimates derived from CBECS 1995.  As the stock efficiency changes over the model simulation, energy consumption decreases nearly, but not quite proportionally to the efficiency increase.  The difference is due to the calculation of efficiency using the harmonic average and also the efficiency rebound effect discussed below.  For example, if on average, electric heat pumps are now 10 percent more efficient than in 1995, then all else constant (weather, real energy prices, shell efficiency, etc...), energy consumption per heat pump would now average about 9 percent less.  The Service Demand  and Technology Choice Submodules together determine the average efficiency of the stocks used in adjusting the initial average energy consumption.

Adjusting for Weather and Climate

Weather in any given year always includes short-term deviations from the expected longer-term average (or climate).   Recognition of the effect of weather on space heating and air conditioning is necessary to avoid projecting abnormal weather conditions into the future.  In the commercial module, proportionate adjustments are made to space heating and air conditioning demand by Census Division.  These adjustments are based on NOAA data for HDD and CDD.  A 10 percent increase in HDD would increase space heating consumption by 10 percent over what it would have otherwise been.  The commercial module makes weather adjustments for the years 1996 through 1998.   After 1998, long term weather patterns are assumed based on 30-year averages of HDD and CDD.

Table 12.   Capital Cost and Efficiency Ratings of Selected Commercial Space Heating Equipment

Short-Term Price Effect and Efficiency Rebound

It is assumed that energy consumption for a given end-use service is affected by the marginal cost of providing that service.  That is, all else equal, a change in the price of a fuel will have an inverse, but less than proportional, effect on fuel consumption.  The current value for the short-term elasticity parameter is -0.25 for end uses affected by short-term price effects.  For example, for lighting this value implies that for a 1 percent increase in the price of a fuel, there will be a corresponding decrease in energy consumption of 0.25 percent.  Another way of affecting the marginal cost of providing a service is through equipment efficiency.   As equipment efficiency changes over time, so will the marginal cost of providing the end-use service.  For example, a 10 percent increase in efficiency will reduce the cost of providing the service by 10 percent.  The short-term elasticity parameter for efficiency rebound effects is -0.15 for affected end uses; therefore, the demand for the service will rise by 1.5 percent (-10 percent x -0.15).  Currently, all services except refrigeration are affected by the short-term price effect and services affected by efficiency rebound are space heating and cooling, water heating, ventilation and lighting.  

Cogeneration

Nonutility power production applications within the commercial sector are concentrated in education, health care, office, and warehouse buildings.  Historical data from Form EIA-867, Annual Nonutility Power Producer Report, are used to derive electricity cogeneration for 1996 by Census Division, building type, and fuel.  After 1996, a forecast of electricity cogeneration, as disaggregated above, is developed as follows:  first, relative prices of energy sources for generation are compared with the price of electricity; second, if the price of electricity increases relative to generation fuels, then cogeneration increases based on a sensitivity parameter.24 If the price of electricity falls relative to the prices of other fuels, then cogeneration decreases based on the same sensitivity parameter.  For each year of the forecast period, all cogenerated electricity is assumed to be sold to the grid and, subsequently, a portion is bought back to meet part of the consumption necessary to satisfy service demands.

Legislation and Other Federal Programs

Energy Policy Act of 1992 (EPACT)

A key assumption incorporated in the technology selection process is that the equipment efficiency standards described in the EPACT constrain minimum equipment efficiencies. The effects of standards are modeled by modifying the technology database to eliminate equipment that no longer meets minimum efficiency requirements.  For standards effective January 1, 1994, affected equipment includes electric heat pumps—minimum coefficient of performance of 1.64,  furnaces and boilers—minimum annual fuel utilization efficiency of 0.8,  fluorescent lighting—minimum efficacy of 75 lumens per watt, incandescent lighting— minimum efficacy of 16.9, air conditioners—minimum seasonal energy efficiency ratio of 10.5, electric water heaters—minimum energy factor of 0.85 and gas and oil water heaters—minimum energy factors of 0.78.

Climate Change Action Plan

The Climate Change Action Plan (CCAP) contains 5 Action Items which affect the commercial sector.   Action Items 1, 4 and 5 are designed to stimulate investment in more efficient building shells and equipment for heating, cooling and other end uses.  Action Item 2, EPA’s Green Lights Program targets the retrofitting of lighting equipment.  Action Item 3 was unfunded and therefore not modeled.  The commercial module includes several features that allow projected efficiency to increase in response to voluntary programs (e.g., the distribution of time preference premiums and shell efficiency parameters).  For Action Items 1, 2, 4 and 5, retrofits of equipment for space heating and air conditioning are incorporated in the distribution of premiums given in Table 11.  Also, based partly on these actions, the shell efficiency of new and existing buildings is assumed to increase from 1995 through 2020.  Shells for new buildings increase in efficiency by 6 percent over this period, while shells for existing buildings increase in efficiency by 4 percent.   In total, the action items result in energy savings which are estimated to reduce carbon emissions by the commercial sector by 13.0 million metric tons for the year 2010.

Commercial Technology Cases

In addition to the AEO99 reference case, three side cases were developed to examine the effect of equipment and building standards on commercial energy use—a 1999 technology case, a high technology case, and  a best available technology case.  These side cases were analyzed in stand-alone (not integrated with the NEMS demand and supply modules) commercial model runs and thus do not include supply-responses to the altered commercial consumption patterns of the three cases.

The 1999 technology case assumes that all future equipment purchases are made based only on equipment available in 1999.  This case further assumes building shell efficiency to be fixed at 1999 levels.  In the reference case, existing building shells are allowed to increase in efficiency by 4 percent over 1995 levels, new building shells improve by 6 percent by 2020 relative to new buildings in 1995.  

The high technology case assumes earlier availability, lower costs, and/or higher efficiencies for more advanced equipment than the reference case. Equipment assumptions were developed by engineering technology experts, considering the potential impact on technology given increased research and development into more advanced technologies.11 In the high technology case, building shell efficiencies are assumed to improve 50 percent faster than in the reference case from 2000 forward. Existing building shells, therfore, increase by 5.6 percent relative to 1995 levels and new building shells by 8.4 percent relative to their efficiency in 1995 by 2020.

The best available technology case assumes that all equipment purchases from 2000 forward are based on the highest available efficiency in the high technology case in a particular simulation year, disregarding the economic costs of such a case.  It is merely designed to show how much the choice of the highest-efficiency equipment could affect energy consumption. Shell effects in this case are assumed to be the same as for the high technology case above.

Fuel shares, where appropriate for a given end use, are allowed to change in the technology cases as the available technologies from each technology type compete to serve certain segments of the commercial floorspace market.  For example, in the best available technology case, the most efficient gas furnace technology competes with the most efficient electric heat pump technology.  This contrasts with the reference case, in which, a greater number of technologies for each fuel with varying efficiencies all  compete to serve the heating end use.  In general, the fuel choice will be affected as the available choices are constrained or expanded, and will thus differ across the cases.