Experience with the Operation of an Energy Model Set

(1)

NOT FOR QUOTATION WITHOUT PERMISSION OF

THE

AUTHOR

EXPERIENCE WITH THE OPERATION

OF AN

ENERGY MODEL SET

L

S c h r a t t e n h o l z e r

October 1984 WP-84-76

Working Papers a r e i n t e r i m r e p o r t s on work of t h e I n t e r n a t i o n a l I n s t i t u t e for Applied Systems Analysis and have received only limited review. Views o r opinions expressed h e r e i n do not necessarily r e p r e s e n t t h o s e of t h e I n s t i t u t e o r of i t s National Member Organizations.

IlTI'ERNATIONAL INSTITUTE FOR APPLlED SYSTEMS ANALYSIS 2361 Lasenburg, Austria

(2)

PREFACE

The Energy Group of t h e International Institute for Applied Systems Analysis (IIASA) has developed a s e t of models to describe the global energy system. This was first used in t h e late seventies to formulate two scenarios for global energy demand and supply for t h e period 1980-2030. The original model s e t consists of individual models and blocks of assumptions t h a t are arranged in the form of a loop. The information flow between t h e elements of t h e set is deli- berately not fully automated, so as t o enhance h u m a n control of t h e modeling process. The model s e t operates iteratively, i.e., tentative assumptions made in one part of t h e s e t a r e subsequently modified in t h e light of results in other parts of t h e set until satisfactory consistency over t h e s e t as a whole is achieved. This paper reports on t h e conclusions arrived a t by t h e author during his participation in the work of formulating the IIASA scenarios and subse- quent applications of t h e model set. The main methodological conclusion is t h a t t h e r e is an important trade-off between model detail and model usability, which calls for great c a r e in t h e do not leave the terminal without writing, etc.

design of both t h e model and t h e analysis to be undertaken.

Hans-Holger Rogner Leader

Energy Development, Economy.

and Investments

(3)

EXPERENCE WITH THE OPERATION O F AN ENERGY MODEL SET*

L.

Schrattenholzer

INTRODUCTION AND sulUURY

Three y e a r s ago. IIASA's Energy Systems Program published t h e book Energy in a Enite World (Hafele e t al., 1981). A major p a r t of t h e book dealt with t h e description of a s e t of energy models and i t s application, i.e., t h e formulation of two scenarios for t h e development of t h e global energy s y s t e m over t h e period 1980-2030. The models themselves and t h e i r input d a t a h a v e been extensively d o c u m e n t e d (see, e.g., Schrattenholzer (ed.). 1982) a n d t h e global scenarios have been p r e s e n t e d on a n u m b e r of occasions. The emphasis in those presentations was usually on various aspects of t h e model results. In c o n t r a s t , this paper s e t s o u t t o r e p o r t on t h e experiences a n d conclusions arrived at by t h e a u t h o r during his participation in t h e work formulating t h e IIASA scenarios a n d l a t e r on. Although t h e global scenarios were t h e r e s u l t of a t e a m effort, responsibility for t h e conclusions presented h e r e r e s t s solely with t h e author.

The paper begins by describing s o m e of t h e deliberations t h a t took place a t t h e beginning of IIASA's global energy modeling activities a n d outlining a n u m b e r of perspectives in which t h e work was seen a s t h e results m a t u r e d The model s e t is t h e n described with an emphasis on methodological aspects, taking

*To be published in the Proceedings of the 9th World Congress of the International Federa- tion of Automatic Control (IFAC), Budapest. Hungary, July 2-6, 1984.

(5)

into account both t h e purpose of t h e modeling and conclusions drawn from t h e modeling process. This is followed by some remarks on t h e operation of t h e model s e t and experience of applying i t to other problems involving various geographical areas.

DESIGN ClUTERM FOR IIASA'S GLOBAL ENERGY MODEL

The main idea behind t h e development of t h e IIASA model s e t was to con- s t r u c t a tool with which plausible a n d consistent lines of development (scenarios) of the global energy system could be investigated and displayed.

The size of the model was t o be chosen in such a way a s to allow just enough room for the formulation of the plausibility arguments for scenarios described.

This meant, in practical t e r m s , t h a t t h e model s e t was to be comprehensive enough t o provide for the formulation of a wide variety of scenarios, whilst a t t h e same time being small enough t o be comprehensible to a single person.

A few words about t h e variety of scenarios a r e necessary here. Clearly, t h e two main scenarios (High and Low) described in Energy in a EZnite World did not fully exhaust t h e range of possible f u t u r e developments. Rather, t h e y were intended to span a range which covers what appeared to be a "maximum plausibility" path in the judgment of t h e IIASA group a n d a significant n u m b e r of outside advisors. There was n o t too much c o n c e r n about representing all or most possible f u t u r e developments, first because it seemed impossible to make

"everybody happy" and, second, because i t was believed t h a t t h e scenarios would also be useful for anyone in disagreement with some of t h e numbers con- tained therein, since t h e High and Low scenarios could be used a s reference cases against which disagreements could be formulated and quantified. As expected. criticisms were raised of having considered too narrow a range of scenarios, but it is worth noting t h a t no critique has gone s o far a s to present an alternative to the IIASA scenarios in any comparable form.

One important point t h a t is crucial for a n understanding of t h e IIASA study (and misconceptions here appear to be a continuing source of criticism) is the distinction between a scenario and a forecast. The difference between t h e two t h a t is important here is t h a t t h e global energy scenarios a r e very m u c h t h e result of expert judgment. Thus, t h e dynamics relevant in t h e global modeIing process a r e the changes of the outputs a s a function of changes of t h e input

(6)

assumptions. In c o n t r a s t , t h e relevant p a r t of a forecasting model mimics t h e dynamics of t h e real-world system. Therefore, t h e "art" of forecasting is t o cap- t u r e t h e system's dynamics as a c c u r a t e l y as possible ( t h i s may involve a huge m a t h e m a t i c a l apparatus) whereas t h e "art" of s c e n a r i o writing consists more of finding a good selection of variables in t e r m s of which t h e scenario c a n be defined. In addition t o be obvious qualitative c r i t e r i a ( s u c h as comprehensiveness in spite of having only a small n u m b e r of variables, logical consistency in t h e combination of variables, t r a n s p a r e n c y , e t c . ) , I want to s t r e s s t h e impor- t a n c e of one t h a t m i g h t be called uniform comprehensiveness. Meeting t h i s l a s t objective should prevent, figuratively speaking, t h e third decimal place being provided in o n e p a r t of t h e s c e n a r i o when a t t h e s a m e time t h e very position of t h e decimal point is a variable controlled by a different part. Expressed in this way, t h e a r g u m e n t should sound convincing even to t h e most casual reader, b u t in i t s application t o modeling it takes so m a n y forms t h a t i t is both difficult t o apply a n d easy t o overlook.

One p r a c t i c a l consideration in t h e design of t h e TlASA model s e t was t h a t time a n d manpower c o n s t r a i n t s favored t h e use of existing, albeit modified models. This t u r n e d o u t t o imply n o p a r t i c u l a r shortcomings - since t h e models s e l e c t e d were adaptable enough t o serve t h e purpose

-

but i t explains some of t h e differences in style of t h e various p a r t s of t h e model set.

THE

IIASA MODEL

SET

A s c h e m a t i c description of t h e model s e t a n d t h e logical connections between i t s p a r t s is given in Figure 1. The basic s t r u c t u r e of the model s e t is t h a t of a loop. Accordingly, t h e model s e t was o p e r a t e d iteratively, i.e., prelim- inary assumptions made in one place were l a t e r c h a n g e d as consequence of model r e s u l t s in o t h e r p a r t s of t h e loop. This process was continued until satisfactory consistency within t h e whole model s e t was achieved.

The model s e t was applied t o seven world regions (see Figure 2), the major part of t h e s e t (those e l e m e n t s above t h e d a s h e d line in Figure 1) being applied t o e a c h region. The seven regions were linked by a procedure t h a t established a global balance between primary energy flows ( r e p r e s e n t e d by t h e box below t h e dashed line in Figure 1).

(7)

Scenarios Delinition

r---"---'

(economic, popu.

I

r

I

¹

Intmrgional Energy Trade

i

^I ^L ¹

I I

I Econ. Slructure,

I k - -

-,

^{E n a w} Lifestyles,

I I Consumption

I MEDEE - 2 Technical Efficiencies

I I ^I

I ^I

I I _I

+

I Secondary Fuel Mix

md Substitutions

Maximum Buildup R m s . Cuts

MESSAGE R ~ s w r o r

Production Lhnitr

I for each

( )

Assumptions, judgments, manual ulculrtiorn F m l m a t h e n n i d modtb

--

Direct flow of infonnation (only major Horn shorn)

----

Feedback flow of information (onty major R o w rhownl

---_ _-_-__---- -- ---

I

Figure 1. Schematic Description of t h e Model Set.

world region

--a---

---

f l \

(8)

Region I (NA) North America

Region II (SUIEE) Soviet Union and Eastern Europe

Region Ill (WEIJANZ) Western Europe, Japan, Australia. New Zealand.

S. Africa, and Israel

...

Region I V (LA) Latin America

p j

I I Region V (Af/SEA) Africa (except Northern Africa and S. Africa), South and Southeast Asia

Region VI (ME/NAf) Middle East and Northern Africa

:. . .. Region VII (C/CPA) China and Centrally Planned Asian Economies

F i g u r e 2. The IIASA World Regions.

(9)

The operational s t a r t i n g point of t h e model loop was t h e definition of economic scenarios (see t h e top box in Figure 1). The outputs of this procedure a r e economic activity (represented by GDP) a n d population development for all seven world regions for t h e year 1980-2030. These r e s u l t s a r e very m u c h judg- m e n t a l , b u t t h e degree of arbitrariness was constrained by qualitative r u l e s , e.g., r e s t r i c t i n g t h e differences between economic growth r a t e s in the different world regions, requiring judgmental consistency with historical developments a n d with model r e s u l t s for t h e whole set, e t c .

The economic growth projections a r e a major input t o t h e e n e r g y consumption model MEDEE-2. Other groups of i n p u t d a t a For t h i s model define.

e.g.. t h e s t r u c t u r e OF t h e economy ( t h e disaggregation of GDP into economic sectors), lifestyle ( a m o u n t of dwelling space, room t e m p e r a t u r e , average daily traveling distance, etc.), a n d energy intensities of production processes. Thus MEDEE-2 is a model OF physical e n e r g y flows - in c o n t r a s t t o an e c o n o m e t r i c demand model, which calculates d e m a n d as a Function of GDP, prices, a n d elasticities. The reason for choosing this particular approach was again t h e specific purpose of t h e modeling, which involved a t i m e horizon t h a t was considered too long for t h e m o r e "conventional" approach. A g e n e r a l a r g u m e n t m a y s e r v e t o support t h i s point. If one wants t o know t h e position of some particle p a t a given t i m e t in t h e future, t h e r e a r e two distinctly different ways OF proceeding.

The first is to take t h e initial position a n d velocity of p a n d make a straightfor- ward extrapolation. The second possible approach s t a r t s with t h e assumption t h a t t i m e t is too f a r in t h e f u t u r e for t h e initial position and velocity t o be of m u c h predictive help. This problem description is simple enough, a n d o n e c a n easily think of real-world problems t h a t clearly Fall i n t o one or o t h e r of t h e categories. A m u c h m o r e tricky problem is t o decide where t o s e p a r a t e t h e two situations if one is given an initial location of p a n d asked to e s t i m a t e t h e new positions a t several points OF t i m e in t h e future. For t h e specific problem of projecting f u t u r e energy demand, t h e m e t h o d chosen by t h e IIASA groups fol- lows essentially the second approach, i.e., physical factors o t h e r t h a n c u r r e n t energy d e m a n d a n d elasticities (corresponding to t h e particle's velocity in t h e illustration above) were used t o derive Future energy demand. This procedure certainly did n o t go unquestioned by those who favored t h e first approach. Con- sequently (and this was m o r e t h a n a m e r e cornpromise), elasticities were calculated a p o s t e r i o r i , t h u s serving a s a n o t h e r consistency check and maintaining a m e a n s of communication with t h e a d h e r e n t s of t h e "classical" approach.

(10)

The r e s u l t s of MEDEE-2, which a r e final e n e r g y projections for t h e period studied, a r e subsequently converted into t i m e series of secondary e n e r g y demands. Since s o m e of t h e final e n e r g y d e m a n d is calculated in t e r m s of sub- stitutable fuels, various assumptions on t h e fuel mix a n d on t h e dynamics of interfuel substitution were required in order t o perform t h e conversions.

Secondary e n e r g y projections, in t u r n , a r e m a j o r inputs t o t h e energy supply model MESSAGE-I, a dynamic linear programming model t h a t minimizes total costs of energy supply for e a c h world region over t h e given period.

Although f u t u r e costs a r e somewhat easier t o p r o j e c t t h a n f u t u r e prices, one may ask why t h e role of prices was neglected i n o n e place (viz., t h e d e m a n d model) when a t t h e same time costs were allowed t o play a c r u c i a l r o l e in a n o t h e r place ( t h e supply model). There a r e s e v e r a l answers t o t h i s question.

The m o s t i m p o r t a n t one is t h a t t h e determination of a feasible solution for t h e supply model was more i m p o r t a n t for t h e s c e n a r i o s t h a n t h e e x a c t location of t h e optimal solution within t h e s e t of feasible points. This is readily u n d e r - standable when one looks a t t h e setting in which t h e IIASA study was initiated.

At t h a t t i m e ( a r o u n d 1977) serious questions were raised (see, e.g., t h e Workshop on Alternative Energy Stragegies' r e p o r t on global e n e r g y prospects, 1985-2000) a s t o whether t h e global energy d e m a n d could be m e t a t all in t h e decades to come. Accordingly, m a n y c o n t e m p o r a r y expectations - about both supply a n d d e m a n d - h a d t o be s t r e t c h e d in o r d e r t o arrive a t a solution a t all.

Thus, c o n s t r a i n t s on t h e buildup of new technologies a n d on the r a t e s of utiliza- tion of p r i m a r y energy sources were just p u s h e d u p high enough to overlap with pushed-down d e m a n d projections, leaving very little room for optimization.

Another a r g u m e n t for using a linear programming model was t h e relative fami- liarity of analysts with this m e t h o d worldwide, which made i t e a s i e r for t h e IlASA g r o u p t o disseminate t h e model results.

The box labeled "Interregional Energy Trade" in Figure 1 r e f e r s t o a procedure which, a t t h a t time (around 1979), had n o t y e t been developed i n t o a for- mal model. I t consisted of a s e t of r u l e s for standardized h a n d calculations, with t h e l a r g e s t p a r t dealing with a global balance in oil trade. Thus i t was con- n e c t e d with all seven world regional supply models. This connection was two- w a y information about a region's oil import r e q u i r e m e n t s (or export potential) went i n t o t h e balancing procedure, and inforrnation on t h e availability of (or demand for) c r u d e oil went into t h e seven regional supply models. The s c e n a r i o variables determining t h e outcome of the balancing procedure in t e r m s of

(11)

prices and quantities traded between regions a r e the short-term and long-term export strategies of the oil-exporting countries as well as assumptions about quantities and costs of oil substitution in t h e importing regions.

The IMPACT model of the economic impact of energy supply strategies is fed by the output of t h e supply model. It is a dynamic input-output model with particular emphasis on energy activities. It calculates direct investments in t h e energy sector and indirect investments in other industrial sectors t h a t deliver products to t h e energy sector, as well as the requirements for other principal resources like water, land, and manpower. One aim in applying the IMPACT model was t o use its results t o judge whether the increased overall costs of energy could be carried by t h e economies in question in t h e future. It t u r n e d out, however, t h a t the results were less conclusive than expected because t h e energy system is a significant but still relatively small part of the whole economy, so t h a t t h e feedback from t h e small energy sector is much harder to assess t h a n the direct impact t h a t goes t h e other way. Nonetheless, t h e IMPACT results on t h e magnitude and s t r u c t u r e of direct a n d indirect investments in t h e energy sector remain an important part of t h e global energy scenarios.

OPERATING EXPERZENCE WITH THE MODEL

SET

The most important feature of t h e operation of t h e model loop was t h e fre- quent h u m a n interaction with t h e model, which controlled each information flow between t h e elements of t h e model set. Although this lack of complete automation was considered a shortcoming t o begin with, i t was felt l a t e r t h a t such a setup was actually more appropriate for the desired goal, t h e generation of plausible scenarios. A once-through operation would certainly have led more smoothly t o results but often a t t h e expense of blurring t h e path between inputs and outputs, whereas the presence of a larger number of check points made it easier to assess t h e validity of t h e results and t o gain insights into the dynamics of t h e system.

Further experience with the IIASA model s e t was gained during its application t o o t h e r energy systems, each of which included a t least one of two different ways of using the model set. The first was t o use the model set or some of its parts as a tool to be applied to a new problem. The second approach

(12)

used t h e global s c e n a r i o s t o provide a consistency check on e n e r g y scenarios for geographical regions o r countries ( t h a t lie within a p a r t i c u l a r IIASA world region), by trying t o fit s u c h energy scenarios plausibly into t h e r e s u l t for a world region t h a t is, in t u r n , a p a r t of a globally consistent p i c t u r e . Both types of application were combined in a study for t h e European Community (Sassin e t al., 1983) and in t h e development of a n energy supply model for Austria (Schrat- tenholzer, 1979).

The g e n e r a l problem in applying a model t h a t has been developed for one specific purpose for a different one is t h a t t h e new problem usually h a s pecu- liarities t h a t need individual t r e a t m e n t . Building a n "all-purpose" model t o begin with is no solution, first, for fairly obvious economic reasons and, second, because a model c a n in f a c t be too big for a given purpose

-

a s I will argue now.

The eventual size of a n y model is necessarily a compromise between accu- r a c y on t h e one h a n d a n d c l a r i t y on t h e o t h e r , the first objective calling for larger models a n d t h e second for smaller ones. Unfortunately, t h e second objective s e e m s t o be frequently neglected by modelers. This is somewhat surprising in view of t h e s h e e r practicality of a t least starting small. In my view, t h e c o r r e c t answer t o questions about model size should be "as small a s possible", i.e., one should c o n s t r u c t the smallest possible version t h a t still gives a reliable a n d r e l e v a n t answer t o the problem modeled. My p r e f e r r e d way of proceeding is t o divide t h e original problem i n t o parts and t h e n t o s t a r t with a simpler subproblem a n d t h e simplest model form t h a t will yield r e s u l t s for the subproblems. Adding one subproblem a t a time, t h e model should t h e n be expanded s t e p by step. Clearly, this procedure presupposes t h a t t h e r e is a problem or a question t o begin with. (Frequently t h e r e is n o s u c h starting point - a t least none t h a t i s visible in t h e model description.)

The r e c o m m e n d e d stepwise procedure i s particularly useful if t h e problem t o be modeled is posed by a decision maker or i f policy r e c o m m e n d a t i o n s a r e t o be derived from model results. In this case i t is usually unrealistic t o expect final model r u n s t o be understood in isolation by the "customers" ( a n d t h u s t o have any impact). Rather, t h e decision makers should be p e r m a n e n t l y involved a n d clearly u n d e r s t a n d e a c h s t a g e of t h e modeling process in a n active sense;

in other words t h e y should know why t h e model yields p a r t i c u l a r results.

(preemptive answers t o t h e question "why?" a r e rarely found in published descriptions of model results.)

(13)

How does the IIASA modeling process look in t h e light of these recommen- dations?

I

have described t h e underlying purpose of the modeling in some detail, but how about t h e questions posed by t h e decision makers? There were (and still are) no decision makers a t t h e global level. On t h e one hand this is somewhat unfortunate because a variety of assumptions (such as a certain degree of international cooperation) had to replace decisions. On t h e other hand, i t was perhaps lucky because i t was impossible to make a mistake in this regard. Since I attribute so much importance to t h e interaction with decision makers, I conclude t h a t the absence of this deficiency contributed significantly t o t h e success of t h e IIASA scenarios.

CONCLUDING REMARKS

Although I have used t h e IIASA model set to illustrate some methodological and procedural ideas about modeling, m y conclusions have been influenced by a significantly larger number of encounters with modeling activities. My most general conclusion is t h a t i t is vital to clearly specify t h e modeling aim before starting t h e process.

I

a m convinced t h a t adherence to this principle can preempt several questions t h a t would otherwise arise during the modeling exer- cise. These questions primarily concern t h e eventual shape of t h e model, and most importantly its size. An important side effect of proceeding in this way could be b e t t e r communication between t h e modelers and the "outside world", including decision makers, peer groups, and t h e interested public. All of these audiences would have an easier t i m e trying to understand the modeler's con- cerns if t h e models and their results were described as answers t o t h e specific questions raised in t h e s t a t e m e n t of modeling aims.

(14)

References

Hafele, W. e t al. (1981) E n e r g y in a f i n i t e World, Vol. 2: A Global 3 y s t e m s A n a l y s i s . Report by t h e Energy Systems Program of t h e International I n s t i t u t e for Applied Systems Analysis. Ballinger Publishing Co., Cam- bridge, Mass.

Schrattenholzer, L. (1979) M o d e l l u n t e r s u c h u n g e n l a n g f r i s t i g e r S t r a t e g i e n d e r

~ e r g i e u e r s o r g u n g D s t e n e i c h s . Doctoral Thesis. Technical University of Vienna, Austria.

Schrattenholzer, L. (ed.)(1982) m e IIASA S t of E n e r g y Models: L b c m e n t a t i o n of Global h n s . International Institute for Applied Systems Analysis, Lax- enburg, Austria.

Sassin, W. e t al. (1983) f i e l i n g Europe in the F b t u r e . Long T e r m E n e r g y P r o b - l e m in the EC C o u n t r i e s : A l t e r n a t i v e

R&D

S t r a t e g i e s , RR-83-9. Interna- tional Institute for Applied Systems Analysis; and EUR 8421-EN, Cornmis- sion of t h e European Communities, Brussels.

Workshop on Alternative Energy Strategies (1977) E n e r g y : Global P r o s p e c t s , 19832000. McGraw-Hill, New York.