Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
A clear conceptual framework in which one can adequately understand the process by which energt) is used in theeconomy is not yet available. Over the past 15 years,energt) input-output analysis and the end-use approach toenergy modelling have emerged as useful analytic tools.There is, however, a need to integrate these framer.uorksinto a coherent and comprehensive whole. This paperprovides an initial formulation of such an integratedframework, defines efficiency measures relating to it, andevaluates the usefulness of this approach for energydemand modelling and forecasting.
Nous ne disposons pas encore d'un cadre conceptuel clairpour apprehender de maniere adequate le processus parlequell'tnergie est utilisee dans l'economie. Au cours des15 dernieres annees, l'analyse input-output et l'approchede l'utilisation ultime qu'on a utilisees pour creer desmodeles energetiques se sont revelees des instrumentsutiles d'analyse. Ii faut cependant integrer ces cadresconceptuels dans un ensemble coherent. Cet article ofireune premiere fonnulation d'un tel cadre conceptuelintigre, definit les mesures d'efficacite qui lui sontrelatives et evalue l'utilite de cette approche pour ce quitouche iz la creation de modeles et aux previsionsconcernant la demande d'energie.
Douglas Gardner is in the Department of IndustrialEngineering, University of Toronto. John Robinsonis Director of the Sustainable Development ResearchInstitute at the University of British Columbia,Vancouver.
Energy 5htdies Review Vol. 5, No. 1, 1993 Printed in Canada
To What End?A ConceptualFramework for theAnalysis of EnergyUseDOUGLAST.GARDNERandJOHN B. ROBINSON
1. Introduction
The late 1970s and early 1980s saw anexplosion of interest in energy demand analysis due to energy price hikes and supply disruptions. Interest waned somewhat in the latter half of the 1980s with the collapse in oilprices. More recently, however, growing environmental awareness has brought energydemand analysis once more to the fore. This isdue largely to the critical role played by energyin the greenhouse effect. More generally,energy use is an important consideration inpolicies aimed at sustainable development. Akey issue in the current policy debate is boththe technical feasibility and the cost of changing present energy use patterns in order toreduce harmful environmental side effects.Such policies could be aimed at either moreefficient energy use, fuel substitution, or both.While aggregate energy-economic modelshave been used to examine some of theseissues (Manne and Richels, 1990), such analyses are clearly unable to estimate technicallimits nor follow in detail the process by whichsuch changes may be achieved. A more
1
detailed analysis is clearly necessary if policiesare to be developed and implemented whichbring about the desired changes at minimumcost. Understanding how changes in energyuse may be achieved requires both a clearconceptual framework of how energy is usedin the economy along with efficiency measuresto gauge progress toward the desired goal.Concepts from the fields of end-use modelling,energy input-output analysis and thermodynamics are useful in this respect.
Energy end-use models were developed inthe late 1970's in response to both theunreliability of conventional forecastingmodels and the need for more detailed analysis of the potential for increased energy efficiency (Robinson 1982a, 1982b, 1988). End-usemodels are based on the observation that thedemand for energy is a derived demand. Thatis, no demand exists for energy itself but onlyfor the end-use services that it helps to provide: space conditioning, lighting, and transportation, etc. Given the demand for theseenergy services and the efficiency with whichthey are provided, total energy demand maythen be calculated.
A related, but separate, research directionhas been the development of energy inputoutput analysis for determining the totalenergy requirements needed to produce various goods and services (Bullard and Herendeen, 1975; Casler and Wilbur, 1984). Directenergy is the energy used as an immediateinput in the production of a given good orservice. Indirect energy use is the total energyrequired to produce other inputs (e.g.,materials, equipment, etc.). Both are significantcomponents of energy demand. The sum ofthe direct and indirect energy required to produce a product or service is said to beembodied in the product or service. Energyinput-output analysis explicitly takes accountof all direct and embodied energy flows in theeconomy and thus takes a broader perspectivetoward energy use than end-use analysiS.
The importance of considering indirectenergy consumption in addition to directenergy consumption was underlined by a
2
recent report by the US Congress, Office ofTechnology Assessment (1990). In an energyinput-output analysis of the United States overthe 1972-1985 period, it concluded that energywas increasingly being consumed indirectlyand that the bulk of this increase was due tothe rapid growth in demand for services. Italso noted that the energy embodied inimported goods and services was equal to onehalf of direct US energy imports. Lastly, itobserved that one fifth of the reduction inenergy use over the 1972-1985 period was dueto indirect energy savings as energy-intensivematerials were used less. These findingsstrongly suggest the need to analyze energyuse in a broader framework than is commonlydone.
In order to evaluate the potential for efficiency improvementsf efficiency measures relating to a conceptual framework are also needed.As far as possible, these measures should bebased on thermodynamic principles. The relatively recent development of second law analysis and the concept of exergy are useful in thisrespect.
In this paper, we provide an initial formulation of a coherent and comprehensive energyuse framework. The main goal of this framework is to serve as a vehicle for the analysis ofboth current and future energy use patterns. Asecond goal is to enrich the commonly heldmental model of how energy is used, thuspossibly widening the range of future energyuse scenarios deemed feasible and perhapsdesirable. Efficiency measures relating to thisframework are then presented, followed by anevaluation of the usefulness of this approachfor energy-demand modelling and forecasting.
2. Conceptualization of EnergyDemand
The fundamental purpose of energy use is tohelp satisfy human needs and desires. Energymay be used either directly for this purpose(e.g., to provide space heating, lighting, cooking and transportation) or indirectly (e.g., toproduce goods and services which humans
In the direct use of energy, the energy servicesthat are provided are of immediate benefit tohwnans. Since the goal of the end-useframework is to describe the relationshipbetween energy and hwnan requirements, theend-use process described above is immediately applicable. In the example shown in
~
Natural Electricity Electricity
G"
Furnace Motor light Bulb
H~t Mechanical lightEnergy
House Stamping LightingMachine System
Space Shaping of BuildingH~t Sh<et Illumination
Mctal
1/ Henceforth, "direct use" and "indirect use" ofenergy will refer to these meanings (i.e., the process of meeting human needs). This terminology isslightly different than that commonly used inenergy input-output analysis which applies theseterms to any process in which energy is used (e.g.,the process of producing a given good or service).
2/ The energy field is bedeviled by complex andsometimes contradictory technical language. In thispaper, secondary energy use or input energy isequivalent to what is sometimes called final energyuse or even end-use energy: the energy actuallypurchased by consumers. Tertiary energy use, oroutput energy, refers to the useful energy outputof devices that "utilize" secondary energy, as discussed in more detail in the text.
Illumination(DirectUre)
AutomobileProduction(IndirectUre)
W"",th(DirectUse)
2.2 Direct and Indirect Use of Energy
Figure 1: The Energy End-Use Process
The end-use process, depicted graphically inFig. 1, begins where traditional descriptions ofenergy use, which trace the processes lyingbetween primary production and secondaryuse, leave off.' The first link in the end-useprocess chain involves the procurement ofsecondary or input energy by the conswner.An end-use technology next transforms theinput energy into tertiary or useful energy. Forexample, a residential furnace (an end-usetechnology) transforms natural gas or fuel oil(input energy) into heat (useful energy). Usefulenergy is then used by service technologies toprovide energy services such as space conditioning, illumination, and transportation. Forexample, an automobile requires mechanicalenergy from the engine to provide transportation services, while a house requires a supplyof heat from the furnace in order to remainwarm. Energy services represent quantifiablemeasures of human requirements or needs; forexample passenger-kilometers of travel, orkilograms of dried clothes.
The service technology, which uses usefulenergy as an input in the provision of anenergy service, defines the boundary betweenthe system that provides the energy serviceand the environment. In many cases the service technology is the physical system withinwhich the end-use technology operates. Thecharacteristics of the service technology determine the quantity of useful energy required toprovide the energy service. Thus, insulationlevels and infiltration rates determine thequantity of heat required to heat a home in agiven climate; an automobile's drag coefficientand rolling resistance determine the requiredmechanical energy to travel a certain distance.
2.1 The End-Use Process
conswne) (Ayres and Narkus-Kramer, 1976).1In this paper, each of these uses of energy willbe treated separately. First, however, a genericdescription of the energy end-use process, abasic component of both the direct and indirectuse of energy, will be presented (Robinson,1983; Gardner, 1987).
3
Fig. 1, space heat (an energy service) meetshuman requirements for warmth.
Energy that is used indirectly to meethuman needs (e.g., for freight transportation,metal refining, fabrication activities), however,is of a fundamentally different nature. It is thegoods and services produced by use of thisenergy that provide energy services directly.
Net energy analysis refers to the estimation of the energy embodied in a given commodity or service (Chapman and Roberts,1983). For a specific commodity, the grossenergy requirement (GER) includes not onlythe energy used directly in the productionprocess but also the energy used to producethe other inputs to the production process (e.g.,materials, buildings, machines, transportationservices). Energy input-output analysis hasbeen a major source of the data used in netenergy analysis. This methodology uses amodified input-output model to calculateenergy intensities for different products whichinclude both direct and indirect energy use.Input-output models are uniquely suited fortaking account of the complex network ofenergy and material flows in the economy.
Figure 2 illustrates a simplified model ofmaterial flows and indirect energy use. Primary production describes the processing ofraw materials (e.g., iron ore, limestone, wood)into primary materials (e.g., steel, glass, paper,cement). Primary materials are then used infabrication and assembly processes for theproduction of material goods. New scrap (i.e.,waste material that may be reused), producedin fabrication and assembly processes, isrecycled via secondary production. Materialgoods that are produced then deliver theirservices over their useful lives. After thisperiod, the materials contained within thegoods are either discarded, burned, or salvaged. In the latter case, the materials becomeold scrap to be reused after secondary production.
Various energy services are required ineach of these processing stages. The production of primary materials (steel, glass, cement,etc.) involves the embedding of a substantialportion of the energy used by these industries'
4
Raw MaterIals
SecondaryProduction
New Scrap
}-_.:O;;ld Scrap
Unrecovered Waste
Figuxe 2: Materials Flows and Indirect Energy Use
The transfer of primary materials or any goodcontaining primary materials thus involves atransfer of energy. Production processes thatuse recycled materials that contain embeddedenergy generally require less additional energythan processes that use mainly raw materials(e.g., aluminum). The traditional view ofenergy demand neglects to take account ofthese embedded energy flows.
Given this model of indirect energy use,the energy demand process may no longer beseen as a simple flow of energy from producers to consumers. Instead it must be regardedas a complex web of activities involving multiple transformations and transfers of bothdirect and embedded energy. The term processanalysis has been used for the study of suchcomplex material and energy transformationprocesses (Gault et aI., 1985).
3/ Energy which is embedded in a material is notlost to the environment after being used in theproduction process but is actually contained withinthe material. The term embodied energyencompasses both embedded energy and energywhich is released into the environment after beingused.
2.3 The Demandfor Energy Services
A focus on the direct and indirect energy flowsdescribed above directs attention to the processes and efficiencies by which energy servicesare provided. However, it takes as given theneed for these services or goods. At a morefundamental level, such demands are only areflection of more basic human needs anddesires, ranging from such basic human requirements as food, shelter and warmth tohigher needs such as security, communicationor self-esteem. Indeed all services, products(which are not desired in themselves but forthe services that they provide) and energydemands may be regarded as being derivedfrom human needs and desires. From thisperspective, for example, there is no demandfor space heat per se, but only a need to maintain the physical comfort of the persons in thehome.
Such an approach suggests that moreefficient ways of meeting these underlyingneeds may exist or may be envisioned. In principle, it would be desirable to evaluate theefficiency with which energy is used in meeting these more fundamental demands, that is,the relationship between energy services andhuman needs. In practice, of course, it may bedifficult to know exactly what human needsare being satisfied by a specific energy service,let alone how well they are being satisfied.
A second implication is that changes in thecomposition of these more fundamentaldemands may affect the demand for variousgoods and services and thus, by extension, thedemand for energy. The terms "structuralshift" and "sectoral shift" have been coined todescribe this effect. Its influence on energydemand has been documented in a number ofrecent studies (Huntington, 1989; Marbek,1989; US Congress, 1990; Howarth et aI., 1991;Gardner, 1993). The recent shift away frommaterial-intensive industries is attributed byWilliams et ai. (1987) to be at least partly theresult of the growing consumer preference forhigh value-added, low material-intensive products. Huntington, in a review of a number ofstudies that have examined this effect, con-
cluded that at least one third of the decline inenergy use per unit of output in manufacturing since 1973 may be attributed to this"sectoral shift" (Huntington, 1989). A recentstudy in Canada concluded that "27% of thetotal drop in the energy/GDP ratio (from 1972to 1984) was due to energy efficiency improvements in the business economy and 20% wasdue to indirect or "structural" change withinthe business economy" (Marbek 1989, p.A2S).It seems clear that a better understanding ofthe forces driving these changes is necessary.In tum this requires the use of appropriatemeasures of the efficiency of energy use.
3. Measures of Efficiency
3.1 Thermodynamic Measures of Efficiency
Thermodynamic measures of efficiency may beused whenever energy is converted from oneform to another. Thus a thermodynamicmeasure is called for when measuring theefficiency of a particular end-use technologythat converts input energy into useful energy.In addition, a thermodynamic measure is appropriate whenever a process involves achange in chemical or physical state. The production of primary materials from rawmaterials falls into this category.
When measuring the efficiency of anydevice that utilizes energy, engineers and otherenergy analysts have traditionally used the socalled first law efficiency (FLE), defined as theratio of energy output (in a desirable form) toenergy input. This may be written as
E,'1 =-,
E,
where '1 is the FLE, E, is the energy output (ina desirable form) of the device and E1 is theactual energy input. 4 In terms of the conceptual
4/ Since for any given end-use technology, theenergy output (in a desirable form) is equivalent tothe minimum energy requirement, FLE may alsobe thought of as the ratio of the theoretical minimum energy input of a device to the achtal energyinput.
S
framework outlined above, FLE is the efficiency of any end-use teclmology in convertinginput energy to useful energy (e.g., the efficiency of a furnace, electric motor, resistanceheater, automobile engine, etc.). This is theefficiency measure implicitly assumed inpopular discussions of energy efficiency.
The FLE is necessarily device-specific"rather than task-specific. That is, while the FLEof an electric resistance heater tells how efficient a particular heater is, it tells us nothingabout the heater's efficiency in the task ofusing electrical energy to heat a home compared to other means of heating the home.
In order to deal with this problem, there isa need for a method of evaluating energy efficiency independently of the device underquestion. Energy efficiency must be related tothe task being performed. Such an efficiencymeasure requires the incorporation of theSecond Law of thermodynamics, which, unlikethe First Law, sets limits on the efficiency ofany conversion process or task. The generallyaccepted definition is a quantity called secondlaw efficiency (SLE) which compares actualperformance to optimal performance permitted by both the First and Second Laws. Fora single-output device, SLE is the ratio of the"heat or work usefully transferred by a givendevice" to the "maximum possible heat orwork usefully transferable for the samefunction by any device" (American Institute ofPhysics, 1975).
For a more complex system, SLE may bedefined using the concept of exergy (alsoknown as available work or available energy),defined as the "maximum work that may beprovided by a system (or fuel)" (AmericanInstitute of Physics, 1975). In any real worldprocess, exergy is consumed, in contrast toenergy which is conserved in accordance withthe First Law. Thus exergy corresponds to thelayman's concept of energy: the "something"which is consumed in any real (Le., non-ideal)process. While FLE measures the efficiencywith which energy is used, SLE measures theefficiency with which exergy is used. It may bedefined, in its most general form, as the ratio ofthe minimum exergy required to perform a
6
task to the actual exergy used. This may bewritten as
B,£ =-
B,
where < is the SLE, 8, is the minimum exergyneeded to perform the task and 8, is the actualexergy used.
The ratio of exergy to energy is known asenergy quality, a quantity that may varybetween 0 and 1 (Van Gool, 1987). High qualityforms of energy include electricity, mechanicalenergy and chemical energy. The quality ofheat energy depends on the ratio of its absolutetemperature to that of the general reservoir(e.g., the atmosphere). Using the concept ofenergy quality, the SLE equation may berewritten as
B2
C2E
2t =_= __8
1C
1E
1
where C, is the minimum required quality ofthe input energy and C, is the actual quality ofthe input energy (Gyftopoulos and Widmer,1980). Since E,/ E, is simply the FLE, this expression is equivalent to
C,< ~-11
C,
This expression is important because it demonstrates that SLE may be interpreted as the product of the FLE and the ratio of the output toinput energy qualities. That is, SLE focusesattention not just on the quantity of energyrequired or produced but also its quality. Improvements in SLE thus depend equally onimprovements in FLE and better matching ofsupplied and required energy qualities.
The difference between FLE and SLE ismost pronounced when the task at handinvolves heating or cooling. While the FLE ofan electric resistance heater used for spaceheating is 100%, for example, the SLE is only afew percent. The potential for efficiencyimprovement, rather than being nonexistent assuggested by the FLE, is thus enormous. Themismatching of supplied and required energyqualities (e.g., converting high quality electricity to heat) is a major factor in the low over-
all efficiency of the economy.Using the SLE concept, Gyftopoulos and
Widmer (1980) have estimated that the overallefficiency of the American economy is verylow (8.1%), especially in relation to those sectors where space or process heating comprisemajor end-uses. Such an analysis, however,still presents an incomplete picture of the theoretical potential for improvement since it considers neither the potential for reducing enduse energy requirements by improving theefficiency of service technologies nor the potential for reducing indirect energy requirements through improved materials practicess
In the industrial sector, for example, thepotential savings from increased reuse, recycling, thermal conversion and reduced materialsuse are not included. A broader and morecomprehensive overview of energy use isclearly required incorporating both servicetechnologies and material flows in the economy.
3.2 Efficiency of the End-Use Process
As discussed above, the end-use processinvolves two technologies, each with its owncorresponding efficiency. Since end-use andservice teclmologies act in series, the overallefficiency is equal to the product of their respective efficiencies. The efficiency of end-usetechnologies can be assessed in a straightforward fashion. It consists of the second lawefficiency with which those technologies convert input energy into useful energy. However,the situation with regard to service technologies is more complex.
Service technologies are of two types. Forcertain technologies, a minimum requiredquantity of useful energy is easily specifiedgiven a certain desired level of service. In thecase of lighting, for example, a certain minimum amount of light flux (useful energy) mustreach the target area in order to provide agiven level of illumination. Such servicesmight be referred to as energy-intrinsic services. The efficiency of an energy-intrinsic service technology may be measured using theSLE concept by comparing the minimum exer-
gy required to perform the service to the exergy produced by the end-use technology. Otherenergy-intrinsic services include cooking,drying and dehumidification.
For certain service tedmologies, however,it is difficult to specify a minimum theoreticalexergy requirement. The amount of heatrequired to keep a house warm when the external environment is cold, for example, could, intheory, always be reduced by increasing thehouse's insulation level and air tightness. Inthe limit, an ideal house would require nointernal source of heat. Any real house, whencompared to this ideal, would necessarily havean efficiency of zero. Such services might becalled energy-extrinsic in the sense that theactual service is not a form of energy. For suchservices, only a relative measure of efficiencymay be used. The Relative Service Efficiency(RSE) of Krause is the ratio of the useful exergyrequirements of the state-of-the-art servicetechnology to that of the technology used inproviding a given service (Krause, 1981). Thismeasure takes no account of the theoreticalpotential for improvement in efficiency, but isuseful for gauging the practical potential, atleast in the near term. Clearly, it is importantthat a definition of "state-of-the-art" be explicitly specified if the efficiencies obtained are tohave meaning. Other energy-extrinsic servicesinclude space cooling, refrigeration and transportation (assuming equal initial and finalelevations). Table 1 summarizes the end-useprocess efficiency measures.
3.3 Direct and Indirect Energy Efficiency
Since the direct use of energy produces energyservices of direct benefit to humans, the enduse process efficiency measures describedabove may be applied directly. The manner inwhich energy is used in material processes
5/ This would include, for example, upgrading theinsulation level of buildings in order to reduce thequantity of heat (end-use energy) required (see Fig.1). Using the terminology of the American Instituteof Physics, such measures fall lUlder the rubric of"task redefinition," an issue which was explicitlyexcluded from consideration in this same study.
7
Table 1: End-use Process Efficiency Measures
(i.e., indirect energy use), however. is morecomplex, involving four processes in whichenergy is either an input or output: primaryproduction, secondary production, fabricationand assembly and scrap processes. Using theframework shown in Fig. 2, methods by whichthe efficiency of energy used in the productionof goods and services may be measured will bepresented. As in the analysis of direct energyefficiency, the goal is to enable the assessmentof the efficiency with which exergy is used toprovide services of benefit to humans.
As an initial step, the energy efficiency ofindividual operations in materials processesmay be evaluated using the end-use processefficiency framework. For example, the efficiency of a metal stamping operation performedby an electrically driven stamping machinemay be broken down into the SLE of the electrical motor in converting electricity to mechanical energy and the SLE of the machine inusing the mechanical energy to stamp themetal (see Fig. 1). In general, operations withlarge theoretical minimum exergy requirements are mainly in the primary productionstage and involve chemical changes of state.Many operations in fabrication and assembly,secondary production and scrap processinginvolve only physical rearrangements at themacroscopic level with very small theoreticalminimum exergy requirements (Ross, 1985).Large efficiency improvements are thus, inprinciple, possible.
The fabrication and assembly process alsoinvolves engineering technologies that determine the quantities of the various materials inthe good being produced and the expectedproductive lifetime of the good. It seems clear
End-UseTechnology
ServiceTechnology
EnergyIntrinsicServices
Second LawEffIciency
Second LawEfficiency
EnergyExtrinsicServices
Second LawEfficiency
Relative ServiceEfficiency
that the potential for improvements in this areais also large. As engineers have learnedimproved design techniques and gained newknowledge about the nature of materials, thetraditional over-designing of all manner ofgoods has given way to lighter, leaner and lessmaterial-intensive designs. This has led to areduction in the quantity of energy embeddedin finished products. Longer-lived productsare of obvious benefit, as well, reducing thedemand for replacement goods.
Engineering technologies also determinethe ease with which scrap materials can berecovered from depleted goods at the end oftheir useful lives. Exergy embedded in scrapmaterials may be recovered either throughthermal conversion or simply by recycling thematerial (and its embedded energy) in a newproduct. Of course, collection, sorting, burningand other ancillary operations associated withscrap processes reduce the net exergy that isrecovered. A relative material recovery efficiency may be defined which compares the netexergy actually recovered to that which couldbe practically recovered with state-of-the-arttechnology. The preceding efficiency measuresor factors are summarized in Table 2.
3.4 Production Efficiency
Developing methods by which the flow andconsumption of exergy may be incorporatedinto an exergy input-output model is beyondthe scope of this paper but is an importantissue that needs to be addressed in futurework. One way in which the efficiencymeasures described above could be used is toevaluate the overall efficiency of producing agiven product. Considering a single goodincorporating n primary materials, totalindirect exergy expended per unit good (Bi"d)may be written as
where qi is the quantity of material i requiredper unit good, f,. is the fraction of material iproduced in primary production, Bpp'i is theexergy required in primary production per
8
Table 2: Indirect Energy Efficiency Measures ment per unit service (8 ).101 •
Fabrication and Assembly End-Use Process Measures• Manufacturing Product Life
Technology Material Use Efficiency• Engineering
Teclmology
unit of material i, Bsp.; is the exergy required insecondary production per unit of material i, Bf"
is the exergy required in fabrication and assembly per unit good and B~"p is the exergythat may be recovered from the depleted good.These terms represent the exergy expended (orgained) per unit good in primary and secondary production, fabrication and assembly andscrap processes, respectively. The exergyexpended indirectly per unit of service (E i,d ) is
. BB _ i"di"d-T
where L is the expected lifetime in units ofservice of the good. It is clear from the form ofthis expression that the expected lifetime of thegood is of primary importance. The fraction fof a material produced in primary productionas opposed to secondary production can be animportant factor if the exergy requirements ofsecondary production are substantially lower.For goods that contain a large quantity ofembedded exergy, the recovery of the scrapmaterials can also lead to significant savings.
The direct exergy requirement per unitservice (Edi,) is simply the direct exergy intensity of a service. For example, an automobilerequires a certain number of Joules per kilometre travelled. Adding the direct exergyrequirement (Edi,) to the indirect exergy requirement (Ei'd) gives the total exergy require-
Process
Primary Production
Secondary Production
Scrap Processing
Efficiency Measureor Factor
End-Use Process Measures
End-Use Process Measures
End-Use Process MeasuresMaterial Recovery
Efficiency
8 =8 +8tol i"d dir
This expression allows the total system energyrequired to provide a given energy service ormaterial good to be evaluated. Such an exerciseallows the tradeoffs made in any given designto be evaluated, at least within the boundariespreviously mentioned' For example, longerlived automobiles may require heavier parts ormore energy-intensive materials; reusablebottles may require more glass thandisposables. Similarly, substituting lessenergy-intensive materials may increase theenergy requirements in fabrication andassembly or make the recovery of scrapmaterials from the depleted good more difficult. For goods that are direct energy users,tradeoffs must also be made between indirectenergy use and direct energy use. Substitutingaluminum or plastics for steel, for example,may improve the gasoline mileage of automobiles at the expense of increasing their GER aswell.
4. Utility of the ConceptualFramework
Despite the developments in end-usemodelling methodology, energy input-outputanalysis and the growing literature on secondlaw efficiency analysis, a common conceptualframework of the end-use process that can beused for modelling and demand analysis islacking. The need for such a comprehensiveframework is particularly apparent in a number of areas of energy policy analysis. Thegrowing complexity of the economy and therapid growth of the service sector means thatthe indirect use of energy in the form of vari-
6/ It should be kept in mind that the relationshipbetvveen human needs and the services providedby energy and material goods has been excludedfrom this analysis. The analysis of that relationshipraises issues that go far beyond questions of physical efficiency.
9
ous goods and services is increasing relative tothe direct use of energy. This trend is likely toaccelerate as the economies of both developedand developing nations become more technologically advanced and information-based.
The strength of energy input/outputmodels lies in their ability to systematicallymeasure the total impact on energy use of suchchanges in demand by linking the inputs andoutputs of all sectors of the economy. Theirweaknesses lie in the fact that they are basedon a static"snapshot" of how energy is used inthe economy, one that is typically a number ofyears out of date. In addition, they are alsobased on a number of unrealistic assumptions.For example, the demand for different forms ofenergy (along with all other inputs) is assumedto be strictly proportional to output.' Possibilities for fuel substitution, process change orefficiency improvements are excluded. PinpOinting where and how much energyefficiency may be improved is thus difficult.
End-use models, on the other hand, are atechnolOgically based description of howenergy is used in the economy. They are thuswell suited to the task of exploring the potential for fuel substitution and efficiency improvements through the introduction of newtechnologies and process changes. Since theyare based on a physical description of howenergy is used, efficiency measures may beused to evaluate exactly where and how muchenergy may be saved. In addition, by focusingattention on the link between the tasks whichenergy performs (e.g., heating) and the services which it helps to provide (e.g., humancomfort), end-use models facilitate the processof task redefinition, going beyond strict secondlaw analyses.
They lack, however, the necessary structure to methodically analyze the implicationsof changing patterns of materials use anddemand for goods and services. End-usemodels (as well as energy input-outputmodels) also neglect, in large part, the exergyembedded in scrap material flows. Thus enduse oriented analyses must be regarded asincomplete measures of the total potential forenergy efficiency and substitution, since the
10
potential for many systems savings is ignored.Failure to understand these complexinterconnections and the derived nature ofenergy demand overestimates the presentefficiency of energy use, thus necessarily diminishing the range of future scenarios deemedfeasible.
Energy input-output models and end-usemodels thus complement each other in manyrespects, the weaknesses of one being strengthsin the other. Combined with SLE measures andan appreciation of the intimate relationshipbetween energy use and material flows, theycould provide a powerful tool for the analysisof both how energy is used and might possiblybe used in the future.
A related concern is the problem of energydemand data. A framework for the organization of the available data is needed to facilitateits use in policy design and programevaluation. Moreover, a framework makesclear where new data are needed, allowingpriorities for the collection of such data to beset. The need for such a data framework isparticularly acute given the paucity and inadequacy of energy end-use data. In the absence ofcomprehensive end-use data bases, scarce datacollection and management resources need tobe allocated on the basis of some understanding of which data are likely to be most usefulfor analysis. In addition, a widely acceptedframework would promote the exchange ofdata among researchers. For example, while itis beyond the scope of this paper to trace thefull implications of this framework for datacollection efforts, even a cursory overview
7/ The method by which the energy required tomake different commodities is estimated is alsosuspect. Many industries produce a number ofcommodities, some of which are also produced inother industries. The "commodity technologyassumption" implies that the energy required tomake a given commodity is the same no matterwhat industry produces it (Casler and Wilbur,1984). Alternatively, the "industry technologyassumption" may be made, implying that theallocation of inputs to the commodities producedby a particular industry should be based on theindustry output proportion of each commodity.
suggests the need for data on the age structureand performance characteristics of the energyusing stock, particularly in the industrial sector.
The development of a conceptual framework for energy demand analysis is alsobound up with the development of newapproaches to energy-demand forecasting.Until the mid-1970s, simple econometricmodels, trend analysis and expert judgementwere virtually the sole sources of analyses offuture energy demand. The oil price shocks ofthe 1970s and the growing concern over environmental problems led to the developmentof new methodologies in order to illustrate thefeasibility of alternative evolution paths of thesocioeconomic system. lnstead of trying topredict the most likely future, such studies(Solar Energy Research lnstitute, 1981; Lovinset aI., 1982; Brooks et aI., 1983; Goldemberg etal., 1988) sought to provide "existence proofs"(Keepin, 1986) of more preferable futures froman economic, social and envirorunentalstandpoint.
In order to analyze the potential for improving energy efficiency explicitly, suchstudies necessarily require an end-useapproach, since traditional top-down methodologies are inherently predictive and onlyimplicitly consider the teclmical processes bywhich energy is used to provide energy services. However, to date, most analyses ofenergy efficiency have focused on directenergy use only and on analyzing energy useprocesses individually and partially, ratherthan in an integrated and comprehensivefashion. Given the changes in indirect energyuse that have occurred in the recent past andthe growing complexity of the economy, itseems clear that broader analyses using a comprehensive conceptual framework of energyuse are needed if the full potential for changeis to be measured.
To date the most extensive application ofan end-use approach has occurred in the fieldof energy demand modelling, especially withregard to electricity modelling and load forecasting. By the mid-1980s, end-use forecastingdominated residential load forecasting in both
Canadian and large American utilities, andaccounted for one-quarter (US) to one-third(Canada) of commercial sector forecasts, andone-eighth (US) to one-quarter (Canada) offorecasts in the industrial sector (Huss, 1985;Goudie, 1987).
The emergence of end-use modelling andload forecasting is clearly related to changes inthe role such forecasts play in utility planningand analysiS. The increasing recognition of thepossibility of managing electrical demand,rather than simply responding to it, has led toa growing interest in the development ofdemand-side management (DSM). A key requirement of DSM analysis and planning is theneed to understand and model the underlyingphysical determinants of energy demand. Enduse models which in their pure form aresimple physical accounting frameworks,devoid of theory or statistically derived behavioural relationships, are ideally suited for thesimulation of alternative patterns of electricityuse, based for example upon different assumptions about the effects of strategic conservationor strategic load growth programs.
Looking beyond energy modelling per se, afocus on both the direct and indirect uses ofenergy may also be incorporated into thedesign approach to socioeconomic modelling(Gault et al., 1987). This modelling paradigm,borrowing from control theory, separates thecontrol space, where all information flowsoccur and control variables are set, from themachine space where all phYSical flows andtransformation processes occur. The designapproach emphasizes the use of engineeringdesign information in the description ofphYSical transformation processes, includingthe requirements for energy and materials. Theobject of an analysis is not to predict the future,but rather to explore possible alternative scenarios. This goal is facilitated by leavingoptimization out of the model and by the accessibility of the control variables to the user.
The design approach to modelling hasbeen implemented in a set of long-term simulation models at Statistics Canada and theUniversity of Waterloo. The Socio-EconomicResource Framework (SERF) consists of a large
11
set of simulation models of the Canadian economy and society (Hoffman and Mcinnis,1988). The SERF system incorporates a set ofdetailed energy end-use models as well as aninput-output model. The SERF system is thuswell suited for analyses that are concernedwith both direct and indirect energy consumption. Recently the SERF system was used tointegrate detailed sector-specific analyses ofenergy efficiency potential into an aggregate"efficiency" scenario of energy use in Canadato the year 2030 (Peat Marwick Ste"enson &Kellogg, 1991). While no explicit attempt wasmade to track indirect and direct energy use,the scenario analysis accounted for the indirectenergy used in achieving the energy efficiencypotential in each sector.
If the urgent need for better informationregarding both the potential and the costs ofchanging future energy use patterns is to bemet, energy models will need to be based upona consistent and comprehensive representationof the energy use process (Stem, 1984). Only inthis way will it be possible to obtain an accurate understanding of the potential for, andimplications of, changes in future energy use.There is, of course, a trade-off betweenincreased analytical capability, on the onehand, and the increased cost of gatheringenergy data and building improved energy usemodels, on the other (Robinson, 1982a). However, making an informed judgement aboutthat trade-off itself depends on developing acoherent picture of the energy use process thatthese models and their supporting data areintended to represent. This paper is intendedto provide a preliminary outline of one suchpicture. The next step is to attempt to implement the measures discussed above within anenergy modelling framework.
References
American Institute of Physics (1975) EfficientUse of Energy. Part I - A Physics Perspective,K W. Ford, G.!. Rochlin, Marc Ross, RH.Socolow (eds.) (New York).
Ayres, RU. and M. Narkus-Kramer (1976) 'An
12
Assessment of Methodologies for EstimatingNational Energy Efficiency,' AmericanSociety of Mechanical Engineers, 76WAjT5-4.
Braithwait, S. (1987) End-Use Models: An EPRIPerspective, paper presented at the CanadianElectrical Association Annual Meeting,Energy Forecasting Section, Calgary, May 68.
Brooks, D.B., J.B. Robinson and RD. Torrie(1983) 2025: Soft Energy Futures for Canada,Prepared by Friends of the Earth for theDepartment of Energy, Mines and Resources(Ottawa).
Bullard, C and R Herendeen (1975) 'TheEnergy Costs of Goods and Services,' EnergyPolicy 3.
Casler, S. and S. Wilbur (1984) 'Energy InputOutput Analysis: A Simple Guide,' Resourcesand Energy 6:187-20l.
Chapman, P.P. and P. Roberts (1983) MetalResources and Energy (London: Butterworths).
Gardner, D.T. (1987) Evaluating the Potential forImproved Energy Efficiency with Application tothe Socio-Economic Resource Framework(SERF), M.ASc. dissertation, Department ofSystems Design Engineering, University ofWaterloo.
-(1993) 'Industrial energy use in Ontario from1962 to 1984,' Energy Economics 15(1):25-32.
Gault, P.O., RB. Hoffman and B.C Mcinnis(1985) 'The Path to Process Data,' Futures17(October): 509-27.
Gault, P.O., KE. Hamilton, RB. Hoffman andB.C Mcinnis (1987) 'The Design Approachto Socia-Economic Modelling,' Futures 19(February): 3-25.
Goldemberg, J., T.B. Johansson, AKN. Reddyand RH. Williams (1988) Energy for aSustainable World (London: Wiley-Eastern).
Goudie, S.R (1987) 'How Canadian ElectricalUtilities Forecast,' Proceedings from CanadianElectrical Association Annual Meeting - EnergyForecasting Section, v.I.
Gyftopoulos, E.P. and T.P. Widmer (1980)'Benefit-Cost of Energy Conservation,' Thermodynamics: Second Law Analysis, RA
Gaggioli (ed.) (Washington: AmericanChemical Society).
Hoffman, RB. and B.C. McIrmis (1988) 'TheEvolution of Socio-Economic Modelling inCanada,' Technological Forecasting and SocialChange 33:311-23.
Howarth, RB., L. Schipper, P.A Duerr andS. Strom (1991) 'Manufacturing Energy Usein Eight OECD Countries,' Energy Economics13(4):135-42.
Huntington, H.G. (1989) 'The Impact ofSectoral Shifts in Industry on U.s. EnergyDemands,' Energy 14(6):363-72.
Huss, W.R (1985) 'Can Electric UtilitiesImprove Their Forecasts?, The HistoricalPerspective,' Public Utilities Fortnightly116(11):27.
Keepin, B. (1986) 'Review of Global Energyand Carbon Dioxide Projections,' AnnualReview of Energy 11: 357-82.
Krause, F. (1981) 'The Industrial Economy-AnEnergy Barrel Without a Bottom?,' SecondInternational Conference on Soft Energy Paths.Rome.
Lovins, A.B., L.H. Lovins, F. Krause and W.Bach (1982) Least-Cost Energy, Solving the CO,Problem (Andover, Massachusetts: BrickHouse Publishing).
Manne, AS. and RG. Richels (1990) 'C02
Emission Limits: An Economic CostAnalysis for the U.S.A,' The Energy Journal11(2):51-74.
Marbek Resource Consultants (1989) EnergyDemand in Canada, 1973-1987 - A RetrospectiveAnalysis, Volume 1: Report, report preparedfor Energy Mines and Resources Canada(Ottawa) revised version, August.
Peat, Marwick, Stevenson and Kellog (1991)
The Economicaily Attractive Potential forEnergy Efficiency Gains in Canada (Ottawa:Energy, Mines and Resources Canada).
Robinson, J.B. (1982a) 'Bottom-Up Methodsand Low-Down Results: Changes in theEstimation of Future Energy Demands,'Energy 7(7): 627-35.
-(1982b) 'Energy Backcasting: A proposedmethod of policy analysis,' Energy Policy10(4): pp. 337-44.
-(1983) An Energy End-Use Data Framework,report prepared for Energy Conservationand Oil Substitution Branch, Department ofEnergy, Mines and Resources, Ottawa.
-(1988) 'Loaded Questions: New Approachesto Utility Load Forecasting,' Energy Policy16(1): 58-68.
Ross, M.H. (1985) Industrial Energy Conservation. Energy Sources: Conservation and Renewables, AIP Conference Proceedings No. 135,(New York).
Solar Energy Research Institute (1981) A Neu)Prosperity - Building a Sustainable EnergyFuture (Andover Massachusetts: Brick HousePublishing).
Stem, P. (ed.) (1984) Improving Energy DemandAnalysis (Washington: National AcademyPress).
US Congress, Office of Technology Assessment(1990) Energy Use and the U.S. Economy(Washington: US Government PrintingOffice).
Van Gool, W. (1987) 'The Value of Energy Carriers,' Energy 12(6):509-18.
Williams, RH., E.D. Larson and M.B. Ross(1987) 'Materials, Affluence and IndustrialEnergy Use,' Annual Review of Energy 12: 99144.
13