ETSAP Project Head
EFDA/ CSU, IPP, Max-Plank-Institut
Boltzmannstr.2, D-85748 Garching Bei Muenchen
Phone: +(4989)3299-4194 / cell.+39(335)537-7675
e-mail: email@example.com / firstname.lastname@example.org
The Energy Technology Systems Analysis Programme (ETSAP) of the International Energy Agency continues to provide a multinational capability to determine the most cost-effective national choices to limit future emissions of greenhouse gases, using consistent methodology that offers a basis for international agreement on abatement measures. The basic MARKAL model continues to serve national interests, as illustrated by its use for a major national RDD&D appraisal in the UK, its use to help develop the national least-cost energy strategy in the USA, and its acceptance by a wider international community. Outside ETSAP, MARKAL was used in Taiwan and (in the form of MENSA) in Australia to inform the debate on response strategies under the UN Framework Convention on Climate Change.
Primarily, however, Annex V produced and demonstrated a number of innovations and improvements that extend the usefulness of MARKAL, especially for climate policy analysis.
Contrary to earlier arguments that the two types of models are conceptually incompatible, MARKAL has now been hard-linked with MACRO, a long-term neoclassical economic growth model, and soft-linked with other macroeconomic models such as MACROEM to combine the insights of "bottom-up" technological and "top-down" economic models. MARKAL-MACRO has evolved into a new standard used by most ETSAP participants and an increasing number of users around the world.
The combined energy-economic models can evaluate energy technologies by their ability to reconcile economic growth targets with environmental constraints. This new generation of models makes it possible to evaluate the energy sector in a more comprehensive and satisfactory way. For example, the effect of emission restrictions on the "rebound effect" from energy conservation can be examined.
National results with the combined models indicate that severe emission reductions lead to reduced energy demands, thus relieving the need for drastic technological change. Nevertheless, long-term emission reductions depend more upon new technology than economic policy. The reduction in economic growth due to emission restrictions in itself contributes only a few percent of the emission reductions, however. Therefore, it may be sufficient to use partial equilibrium models, such as MARKAL-ED and MARKAL-MICRO, which introduce price-elastic energy demands but do not otherwise represent the economic system.
When material flows are added to the energy flows in MARKAL, the need for changes in energy technology to meet CO2 emission restrictions is reduced because many changes in manufacturing materials and recycling can be made at lower cost. When emission restrictions are applied to the complete set of greenhouse gases measured as "CO2 equivalents" through their Global Warming Potential only methane (other than the CFCs already bound by the Montreal Protocol) is found to be important in industrialized economies.
When probabilities are assigned to various future emission reduction scenarios in stochastic programming, the best mix of near-term technologies to hedge against the uncertainty may prove to be different from that most suitable for any one of the assumed scenarios.
Perhaps most significantly, the use of a common methodology permits both side-by-side comparisons of national capabilities to reduce future emissions, and integrated multinational analysis to calculate the benefits of national activities implemented jointly. The widespread use of MARKAL provides a basis for assessment of options in an international context, in particular for activities implemented jointly. The value of a consistent methodology among nations may be becoming more apparent.
During Annex V, the ETSAP group contributed to a number of international programs and studies, including the IEA study Electricity and the Environment, the OECD study Environmental Implications of Supports to the Energy Sector, and Climate Change 1995, the Second Assessment Report of the Intergovernmental Panel on Climate Change.
In Annex V (1993-1995), the participants have continued to cooperate in extending MARKAL, a mathematical model used to represent energy systems at the national, regional or community level, to better evaluate measures to mitigate the future emission of greenhouse gases. The countries participating in Annex V were Belgium, Canada, Germany, Italy, Japan, Korea, the Netherlands, Norway, Sweden, Switzerland, United Kingdom, and USA.
The United Nations Framework Convention for Climate Change calls for "stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system." For this stabilization to take place at today's levels, for example, carbon dioxide emissions from human activities would have to be reduced to less than half the current rate. The nations of the world must develop means to cooperate in making major emission reductions, taking into account their individual differences.
In the long run, agreement must be reached on both the implementation of abatement measures and the manner in which costs are to be shared. Not all countries now have the ability to evaluate the abatement measures most suitable to their situation. By common view, however, neither the ultimate climate stabilization target nor the strategy for achieving that objective needs to be fixed immediately. An important part of the initial response to the threat of global climate change, therefore, is to build capability where it is needed.
Annex V of ETSAP addressed the issues of greenhouse gases and national energy options by continuing to develop better models to evaluate technological solutions, economic impacts, and emission reduction strategies. Its main objectives were to:
At its simplest, technology assessment can take the form of a side-by-side comparison that takes the "rest of the world" as given and constant. Up to a certain level, such simple comparisons can be satisfactory if they concern relatively isolated problems in the short term. However, longer term assessments of a more structural than incidental nature can only be examined in more comprehensive frameworks. Mutual interdependencies between energy supplying and consuming sectors, or within subsectors thereof are then implicitly accounted for. A typical example clearly illustrating the importance of considering the entire system is, for example, mentioned in an analysis of Switzerland. With its electricity sector dominated by nonfossil sources (nuclear and hydro), it would hardly be expected that electricity-saving devices would be selected in strategies to reduce greenhouse gas emissions. Nonetheless, they were found to be, as this was the only way to make available and affordable "low-CO2" electricity to replace fossil fuels in traditionally fuel-based applications like space and water heating.
By choosing suitable system boundaries, MARKAL has without modification been extended geographically beyond national boundaries to take into account "upstream" energy-consuming and polluting activities in so-called full fuel cycle analyses, and extended temporally for "cradle-to-grave" life cycle analysis. In Annex V, MARKAL, pure and simple, has been used to evaluate the cost savings from "activities implemented jointly" by three countries cooperating in reducing carbon dioxide emissions, and to evaluate the relative importance of reducing emissions of the many different greenhouse gases.
It is a commonplace, however, that no single model can provide all the answers. There is no doubt value in the different perspectives provided by using more than one type of model in a complementary fashion. This view has characteristically justified the use of both bottom-up energy systems and top-down macroeconomic models in projecting energy futures. Although the two types of model were coupled as early as 1981 in ETA-MACRO, it has also quite recently been argued that the two approaches are conceptually incompatible.
For greenhouse gas abatement costing studies, the need for a hybrid approach that combines the essential elements of bottom-up technological models and top-down econometric models has come to be acknowledged. Individual ETSAP partners have investigated linkages of MARKAL with economic models of various types. For example, Italy has used an input-output model. Belgium has developed MARKAL-MICRO, a partial equilibrium model with price elasticities for each demand sector. Australia is developing a linkage between its version of MARKAL and a dynamic general equilibrium model. Japan has developed an interindustrial economic approach in its model MACROEM.
The inclusion of general economic aspects has expanded the analytical
capabilities available to the participants and the growing number of other
users of ETSAP methods and tools, not only in the sense that more
realistic results can be obtained but also because communication with
experts engaged in top-down modeling and other economic model
concepts is facilitated.
The ETSAP research partnership is pioneering the application of MARKAL-MACRO, a new model that combines the technological detail of MARKAL with the general economics of MACRO, a long-term neoclassical growth model. MARKAL-MACRO is the only detailed energy-environment model linked to a macroeconomic growth model generally available worldwide.
MARKAL-MACRO resulted from the collaboration of the Swedish and US participants in ETSAP with Professor Alan Manne of Stanford University. MARKAL-MACRO was designed specifically to estimate the costs and analyze alternative policies proposed for reducing environmental risks such as global climate change. If sizeable reductions in carbon dioxide emissions are required in the future, energy demands and energy prices are likely to be affected. The cost to a national economy is likely to be large enough to be measured as a percentage of Gross Domestic Product. In MARKAL, projections of the energy demands and resource costs up to its 40 to 50 year time horizon are input to the model, unchanged by other circumstances. In MARKAL-MACRO, on the other hand, energy demands and prices are calculated within the model through the interaction of the energy system with the rest of the economic system. The cost to the national economy is calculated directly as the change from the GDP that would otherwise be projected.
The linkage between MARKAL and MACRO is based upon the concept of an economy-wide production function. Inputs to the MACRO production function consist of capital, labor, and energy (K,L and E), the latter in the form of useful services delivered by MARKAL. Total economic output is used to invest in the capital stock required for an expanding economy, to cover the costs of energy and the remainder for consumption. The latter is maximized through the objective function of MARKAL-MACRO representing the utility. In the production function each of the inputs (K,L and E) may substitute for the others but with diminishing returns in the substitution process. In this way, MACRO incorporates price-induced energy conservation. In addition, there is the possibility for autonomous energy efficiency improvements: nonprice factors that reduce energy demands per unit of gross output with the passage of time. The integrated model simultaneously solves for energy and economic components using nonlinear optimization. For carbon dioxide emission studies, MARKAL-MACRO provides as primary results a ranking of the mitigation options, the cost of reducing emissions, and the implications for the economy.
The typical difference between results of MARKAL and MARKAL-MACRO is illustrated in Figure 1 for a 50 percent reduction in carbon dioxide emissions in 2030 in a MARKAL model of the Netherlands. Assuming in the cases considered that carbon dioxide removal and disposal is not a viable option, the main source of the drastic reduction is a big drop in the fossil fuel share. MARKAL-MACRO, on the other hand, introduces the possibility of emission reductions due to a reduction in GDP growth under these circumstances and to a reduction in the useful demand per unit of GDP. Together these reduce the change due to the fossil fuel share, resulting in a diminished call upon relatively expensive nonfossil energy supplies from abroad. Nevertheless, replacing fossil fuel remains the largest source of CO2 emission reductions. According to the MARKAL- MACRO model of Norway, this is the case even there where electricity production is almost totally dominated by hydropower.
Figure 1.With CO2 emission restrictions, lower useful energy demand per unit of GDP reduces the need for fossil fuel curtailment in models of the Netherlands.
With MARKAL-MACRO, one can estimate the economic value of energy technologies in a more comprehensive and satisfactory way than was possible with MARKAL alone. MARKAL-MACRO proved useful in Swedish and Dutch studies of the effect of carbon dioxide emission restrictions on the much disputed concept of "conservation rebound." Conservation rebound is the tendency to demand more energy services as a consequence of using energy more efficiently, thus reducing the net effect expected from the technical measures taken. This was measured by running MARKAL-MACRO with and without conservation technologies in two situations: with restrictions on carbon dioxide emissions and without. In each case, introducing the possibility of conservation up to the level that is cost-effective increased the total useful energy demand, as shown in Figure 2. The increase was noticeably greater with restricted emissions. This accords with the hypothesis that rebound can be significant when energy cost or availability is a constraint on activity, in this case through the imposition of emission restrictions.
Figure 2.Rebound measured by the difference in total useful energy demand is greater when carbon dioxide emissions are restricted, according to this Swedish model.
Users of the MARKAL User's Support System (MUSS), which supports both MARKAL and MARKAL-MACRO, can make a seamless transition to the new technological-economic model. MARKAL has also been linked with other economic models, for example in Italy and Japan. To evaluate carbon dioxide emission control measures in the Italian energy system, MARKAL has been linked with an input-output model of the national economy. The economic production of each sector is not the outcome of a unique average producer, but of many producers each producing sectorial output with different processes. Therefore, each economic producer is modeled through a set of column vectors of technical coefficients instead of a single column vector. The objective function of the expanded model has additional terms due to the input-output matrix that is now incorporated.
In Japan, MARKAL has been soft-linked with a macroeconomic model of the economy, MACROEM, using an interindustrial econometric approach. In this approach, the effect of carbon dioxide reduction by energy conservation is divided between the technical part handled by MARKAL through conservation technologies, and the economic part to be accounted for through changes in useful demand in response to prices. The information that MARKAL transfers to MACROEM includes energy imports, investments in energy industries, investment in energy conservation by industries, production by energy industries, and carbon dioxide emissions by sectors. MARKAL can also provide such information as heating and lighting expenditures by the household sector, or expenditures on fuels by the transportation sector. MARKAL results were compared with those of MARKAL-MACROEM to distinguish carbon dioxide reductions due to technology changes from those due to changes in the economy. Notwithstanding the sizable contribution of economic adjustments to reach environmental targets, the larger part of CO2 emission reductions appears to be due to technical changes within the energy system. As shown in Figure 3, this is especially true in the longer run, when more advanced versions of existing technologies and entirely new concepts are expected to become available, and old equipment is retired. Very similar trends are observed in MARKAL-MACRO runs.
Figure 3.A Japanese model illustrates that technology changes outweigh reductions in energy demand in reducing CO2 emissions in the long term (two price elasticities -- ß = 0.2,0.3 -- are assumed).
3.2 Partial Equilibrium MARKAL
In MARKAL, emission restrictions are met by modifications of the energy system, such as changes in the fuel mix, new energy technology, and energy conservation. In MARKAL-MACRO, as we have seen, the necessary changes in the energy system to meet a specific emission cap are lessened by adjustments made with the rest of the economic system, both by reduced GDP and reduced energy needs. The effect of GDP on emissions is rather small, however, typically contributing just a few percent of the emission reduction. Therefore, a partial equilibrium model not representing the rest of the economic system, but allowing demands to be reduced in response to higher energy prices may suffice.
The partial equilibrium approach has been taken in separate
developments by Belgium, Canada, and Switzerland. Canada uses the
traditional MARKAL linear programming model with the OMNI modeling
language. The others take advantage of the more compact GAMS
language and its link with MINOS, a nonlinear optimizer.
3.2.1 Linear Partial Equilibrium Model
The partial equilibrium approach makes use of the Equivalence Theorem drawn from economics:
Figure 4.Illustration of a supply-demand equilibrium.
Point E is also the point at which the area between the two curves is maximized. (Think of the size of both hatched areas also bounded by a vertical line measuring quantity moving to the right. Obviously, it reaches its maximum when the vertical line reaches Point E.) The two hatched areas represent the sum of the producers' and the consumers' surpluses, sometimes called the net social surplus, which is a proxy for welfare.
The Equivalence Theorem is valid subject to the supply and demand curves meeting certain economic and mathematical conditions. One condition is that the area under the inverse supply curve is well defined, but that area is simply the value of the MARKAL objective function.
The objective of the MARKAL model is to minimize total energy system cost, a linear function. The objective of the partial equilibrium model is to maximize net social surplus, a nonlinear function. Since the latter objective function is separable, however, it is easily linearized by piecewise linear functions. By so doing, the resulting optimization problem becomes linear again, and it may now be formulated entirely within MARKAL. This is achieved simply by defining additional "dummy" technologies, each representing a portion of the energy demand that is reduced due to its elasticity.
MARKAL models with elastic demands are no longer driven by exogenous demands for energy services. Instead, each demand is also influenced endogenously in response to its shadow price computed by the model. Instead of fixing demands, the user specifies demand functions. The elasticities may be different for different demand categories.
MARKAL-ED, the "elastic" MARKAL model applied to the Canadian province of Ontario, was used for demonstration runs. The marginal CO2 reduction costs were smaller with elastic demands than those with fixed demands in all periods by amounts varying from 5 to 40 percent. The main difference occurs when CO2 emission restrictions begin to be imposed. The elastic model allows a smoother transition, whereas the fixed demand model has to absorb the full shock of the CO2 emission reduction by technological means only.
The selection of elasticities has now been fully integrated in MARKAL-ED.
The user simply specifies the elasticities, which may be time-dependent,
and the range within which demands may vary, without the need to define
the dummy technologies.
3.2.2 Nonlinear Partial Equilibrium Models
Nonlinear programming techniques can also be used to define the demand response to price changes and avoid the piecewise approximation using dummy technologies. This is the approach that has been developed by the Belgian and Swiss participants.
MARKAL-MICRO, the partial equilibrium model of Belgium, was used for a study of cost-efficient CO2 reduction possibilities in the Belgian transport sector. The supply function was given by the traditional MARKAL model, whereas a demand function for each demand category was added. Supply and demand clear the markets through price fluctuations, taking into account any limit on the emission of pollutants. Technologies to reduce CO2 emissions in the transport sector are compared with technologies in other sectors on the same cost and emission basis, taking into account the repercussions in the entire energy system.
The contribution of the transport sector to the decrease in CO2 emissions was rather small and was nearly completely due to a reduction in mobility demand, induced by the price increase. Its contribution only became significant in the highest CO2 reduction case. These differences represent the decrease in the total of the consumers' and producers' surplus due to the high energy cost induced by the CO2 constraint. Mobility demand decreases more sharply with the higher CO2 reductions, because the price increase is much higher. Though a large number of technologies are available or under development which could contribute to a reduction of the CO2 emissions in the Belgian transport sector, none seems to be cost-efficient compared to the potential reduction in the other sectors.
MARKAL-MICRO now provides as an option substitution between selected demands through cross-price elasticities, so that, for example, a decrease in the relative price of passenger rail transport may reduce the demand for automobile traffic accompanied by a shift to rail transport.
The advantage of using a partial equilibrium, or elastic, MARKAL is that the model user will be able to define different price and income elasticities for different energy sectors and energy services. The disadvantage is that nobody knows how to estimate these elasticities exactly. The Swiss began to operationalize the concept for Switzerland through the specification of these elasticities.
An econometric model of Switzerland was linked with the bottom-up model SMEDE to define the demands for energy services. Demands and price elasticities were estimated for electricity and up to four types of fuel in four sectors: industry, households, services and transport.
Once the econometric relations are known, the short term elasticities are also specified. Short-term elasticity describes how consumer behavior affects energy demand using existing devices to provide energy services. Long-term effects are represented in MARKAL by the explicit treatment of conservation and technological change and substitution. The marginal cost of conservation is defined by a stepwise function.
MARKAL-MICRO is now operational under MUSS, the MARKAL User Support System, along with MARKAL and MARKAL-MACRO.
Policies and measures to deal with climate change should be cost-effective so as to ensure global benefits at the lowest possible cost. The most cost-effective measures should be implemented wherever they occur. The most cost-efficient way to allocate carbon dioxide emission reductions among the countries is to limit their emissions at a point where their marginal costs are equal. In total, it would be more costly for one country to pay a higher marginal cost to reduce carbon dioxide emissions when lower cost opportunities exist elsewhere.
The Framework Convention establishes several mechanisms for
international cooperation, including "burden sharing" among the developed
countries and "activities implemented jointly" that take into account
differences in countries' economic structure, resource bases, and
available technology. In effect, the question of how much each country
reduces emissions -- a matter of cost-effectiveness -- is separated from
the question of who pays for it -- a matter of equity.
4.1Burden sharing among Belgium, the Netherlands and Switzerland
The Swiss participants have worked out an example of joint implementation by linking the MARKAL models of three countries: Belgium, the Netherlands, and Switzerland. To contain all three models in an integrated international model, an advanced decomposition algorithm, ACCPM, developed at the University of Geneva, was used.
The individual national models indicate that Switzerland, which now uses little coal and generates electricity with hydroelectric and nuclear power, would find it most expensive to reduce carbon dioxide emissions. The Netherlands, on the other hand, has devised a number of alternatives to reduce its future carbon dioxide emissions at comparatively low cost, including in this example carbon dioxide removal and disposal. Belgium lies between the other two.
The integrated international model allocated future reductions among the three countries to reduce their combined emissions in 2030 by 20 percent from year 2000 levels. The marginal costs of emission reduction for each country alone and for the three in combination are shown in Figure 5. Individually, the marginal cost of equal national reductions is least for the Netherlands and by far the greatest for Switzerland. The joint marginal costs calculated for shared reduction is more than that of the Netherlands but less than that of each of the other two countries. The joint marginal cost corresponds to the tax on carbon dioxide emissions to be imposed in all three countries for them to a reach a combined 20 percent reduction by 2030.
Applying the calculated marginal cost to the three energy systems allows more emissions from Belgium and Switzerland at a cost savings in 1990 U.S. dollars of about $2 billion and $3.8 billion, respectively, but requires greater emission reductions in the Netherlands costing an additional $2.6 billion. There remains the question of how the net savings of $3.2 billion should be divided among the three countries. On whatever grounds an equitable distribution is decided, however, the amount of the savings virtually assures a win-win-win outcome.
Figure 5.The undiscounted marginal costs of CO2 reduction in each of the three countries under joint implementation would be less than Belgium and Switzerland would require individually, but more than the Netherlands acting alone.
A more precise calculation of the marginal costs of emission reduction
would take into account more than the difference in direct energy system
costs, however. In each country, the cost is influenced by the adjustment
that would be made in the country's economy as the energy system is
changed. Moreover, the countries are linked not only by the shared
resource of the earth's atmosphere but by commercial trade including
energy carriers and energy-intensive commodities. Steps toward refining
these results are being taken with further extensions of MARKAL.
4.2Common Action and Electricity Trade in Northern Europe
Common action to reduce carbon dioxide emissions through electricity trading in northern Europe was investigated by the Norwegian and Swedish ETSAP participants in concert with Denmark. Extended MARKAL, which features multiple electricity grids, was used for the evaluation. Various scenarios were examined to compare common action with country-by-country reductions, limited trade with unlimited trade, and no carbon dioxide emission constraints with no increase in emissions and future reductions of 20 and 40 percent.
Preferred trading patterns differ among the scenarios, regarding both electricity and emission rights. Norway is a net exporter of electricity and a net importer of emission rights in all cases. Denmark is a supplier of emission rights for all levels of carbon dioxide emission reductions, but its contribution declines as the emission constraint becomes more stringent. For Sweden, assumed to maintain the present level of nuclear capacity, the picture is mixed. For moderate emission constraints, Sweden is a net importer of both electricity and emission rights. With more stringent emission constraints, Sweden becomes an exporter of both electricity and emission rights. These changes are due mainly to the increased cost-effectiveness of nuclear power under constrained conditions. The value of traded electricity is in all cases higher than the value of traded emission rights.
Changes in the use of industrial materials could ease the adjustment to carbon dioxide emission restrictions, particularly in the short term, according to a study in the Netherlands. Using a meta-MARKAL model originally developed in Canada that represents material as well as energy flows, the Dutch group found additional cost-effective options for reducing carbon dioxide, especially in materials production and waste management. With severe emission reductions, there could be major changes in the materials used for some products, such as trucks and automobiles.
The metamodel can represent the industrial material flow system from primary production to waste disposal. More than 250 types of technology were added to the Dutch MARKAL energy system model, representing 29 types of materials and 20 types of products. Using the expanded model, the group projected future carbon dioxide emissions from the combined energy and materials systems and investigated how they could best react to future restrictions on emissions.
Emissions from the energy system are about three times those attributable to the industrial material flow, and in the absence of emission restrictions they would grow much faster. With restrictions in the form of a penalty of 200 Dutch guilders per tonne CO2, however, calculated emissions would be about 20 million tonnes per year less when the material system is taken into account; see Figure 6. In the year 2010 this would increase the carbon dioxide reduction by about one-third.
In the longer term, there could be some unexpected changes in the dominant materials in some familiar products. It is perhaps not too surprising that concrete, the manufacture of which is a major source of carbon dioxide, is replaced with other materials. But one might not expect that much of the steel used in cars and trucks would be replaced by aluminum, a material with higher energy content. However, reductions in carbon dioxide emissions due to their lighter weight during the operating lives of the vehicles more than compensate for the higher emissions in the production of the aluminum.
The net change in the use of materials in the entire energy and industrial materials system of the Netherlands would not be great during the next few decades. Most options for these reductions are in the hands of Dutch policy makers. Looking at carbon dioxide reductions in 2025, for example, about half is due to domestic shifts from fuels and materials with high carbon content: substituting hydrogen for natural gas in fertilizer production, using carbon dioxide removal from emissions in the manufacture of steel and cement, and changing the fuel generating electricity. Only about one-fifth is due to the expansion of sustainable biomass feedstocks from abroad such as elastomers, paint, tropical wood, and compost.
Figure 6.With the same carbon tax, changes in the use of materials as well as energy can further reduce carbon dioxide emissions, according to a Dutch model.
Compared to carbon dioxide, relatively little is known about other greenhouse gas emissions that might influence preferred solutions to direct carbon dioxide abatement. In an evaluation of the effects of other greenhouse gases, the Netherlands participant estimated that by the year 2000, methane emissions will make up about 5 percent of the total CO2-equivalent emissions attributable to the Dutch energy system. Energy-related halocarbon, nitrous oxide, and carbon monoxide will constitute less than one percent of the total. These other emissions are expected to increase disproportionately after 2000 due to differences in the primary energy mix and the start of natural gas imports from Russia. To calculate these emissions and how they can be limited, the MARKAL model of the Netherlands was used with a database that accounted for emissions of methane, nitrous oxide, carbon monoxide, and several halocarbons, as well as abatement options for such emissions. The carbon dioxide equivalents of these emissions were calculated using the global warming potentials recommended by the IPCC for a 100-year time horizon. A range of penalties of up to 1,000 Dutch guilders per tonne of carbon dioxide equivalent was applied in the model to all greenhouse emissions. This was compared to model results when only direct carbon dioxide emissions are penalized, as shown in Figure 7.
MARKAL identifies the sources of emission reductions. In these projections, carbon dioxide removal and disposal is assumed to be an option in the Dutch energy system. As emission penalties increase, this becomes the dominant source of carbon dioxide emission reductions with the main fuel changing from coal to natural gas. There is little change from applying the emission penalty only to carbon dioxide (left bars) to all greenhouse gases (right bars). However, methane emissions would be reduced by more than half with the higher penalties applied to all greenhouse gases.
Figure 7.According to a model of the Netherlands, methane emissions would be substantially reduced with higher penalties on all greenhouse gas emissions rather than only CO2, mostly due to reduced imports of coal and natural gas, as shown here for the year 2030.
With all greenhouse gases penalized, more renewables -- principally offshore wind turbines, photovoltaics, and energy crops -- are found to be cost-effective, even though renewable energy technologies are penalized for the emissions resulting from their manufacture, and biogenic nitrous oxide from energy crops is counted as a greenhouse gas. Slightly more end-use savings are also found to be cost-effective.
The results of the MARKAL model of an energy system are usually evaluated through scenario analysis. The optimal mix of technologies found to comprise the energy system varies with the scenario, that is, with a set of assumptions as to projected energy demands and prices, restrictions on emissions of pollutants, and so on. Certain technologies, described as robust, may prove to be preferred in every scenario. Generally, however, major differences in the technology mix result from different assumptions. Thus, the decision maker is left to make choices according to his judgement of the likelihood of the various scenarios.
Stochastic programming applies an explicit probability of occurrence to each scenario, applicable until the uncertainty is resolved at an assumed future date when corrective action may be taken. Taking these additional assumptions into account, the model is then solved once and for all, selecting a single course of action that is optimal in the face of uncertainty. The model thus determines the hedging strategy: the singular optimal mix of technologies for the near term until the uncertainty is resolved. Thus, in one optimization run, a contingent policy that adapts to the possible evolution of the scenarios is determined. This adaptive policy is much closer, in spirit, to the way decision makers have to deal in real life with uncertain futures.
Results show that the technologies and fuel use may be different in this combined case from what they are in any of the individual scenarios. In the stochastic model of the Quebec energy system, for example, the decisions recommended by stochastic programming usually differ from those recommended for individual deterministic scenarios.
For the canton of Geneva, the Swiss participants used a five-stage stochastic program, with two major uncertainties: CO2 constraints and demand-side management (DSM) potential. Each stage marks a decision or a chance occurrence. The stochastic model determined that an early investment should be made in a combined cycle power plant as a hedge against the risk of the DSM programmes being abandoned or otherwise not bringing the long term electricity savings anticipated.
Sensitivity analysis may then be used to test how the stochastic model results may vary according to the assumed probabilities, resolution date, technologies available, and the nature of the emission constraint. Figure 8 from a Dutch study shows the carbon dioxide emission trajectories for the hedging path resulting from stochastic programming compared with the three determinate alternatives. In their work, the Dutch also took into account a risk aversion factor that penalizes more risky choices (i.e., those sets of alternatives with greater variance).
Even considering the recent improvements in hard- and software performance, dealing with uncertainties places a heavy burden on computational capabilities. For example, the two major uncertainties (CO2 constraints and DSM potential) and five decision stages, made the medium sized Geneva MARKAL model grow from 2,400 constraints and 3,000 variables to 21,000 constraints and 26,500 variables. It could still be solved in 10 to 26 minutes, but only by using a decomposition algorithm running on an IBM work station. To shorten computer time using desktop computers, the MARKAL database was reduced in size by eliminating some technologies and using fewer time periods in the initial Dutch and Canadian tests.
The full-scale, two-stage stochastic model formulation developed in the Netherlands is also implemented under MUSS, together with MARKAL, MARKAL-MACRO and MARKAL-MICRO.
Figure 8.Example of CO2 emission trajectories using stochastic programming which assigns probabilities to each of three deterministic cases to calculate a hedging path for the near term, using a Netherlands model.
The application of stochastic programming to the canton of Geneva is a recent example of the evolution of local energy planning (LEP). Since the first energy crisis, LEP has come to be understood as an element in strategic planning of whole community systems, primarily in countries with decentralized decision structures for town planning and energy supply. LEP is an interdisciplinary subject comprising energy supply and demand-side management, town planning, and■an important driving force in its development■environmental protection.
Beginning in 1980, MARKAL was used for developing energy-environmental strategies in seven Swedish communities. The model has also been used for local energy planning in Italy, in 1993, for example, to develop an urban energy plan for Turin that considers the need to reduce carbon dioxide emissions. A key concept in this planning has been the "reference energy system," a flow chart that includes both energy supply and demand options, or "integrated resource planning." MARKAL proved to an excellent tool for larger communities and energy utilities with fairly complex systems and the need to evaluate strategies for investments and emissions control, in part by facilitating communication with "reference groups" comprised of representatives of the affected community.
With the development and proliferation of personal computers, a wide variety of software has become available to planners. Often, there is a tendency toward sophisticated optimization of single components of community energy systems rather than simultaneous strategic consideration of the system as a whole. In all countries where local energy planning is commonly practised, however, there has been a trend toward to more integration and complexity.
Further development of the set of models for local planning in Geneva will include a geographical information system database for traffic planning and mapping pollution dispersion. In a few years, most major cities in Europe, North America and Japan are expected to have digitized city maps that will provide a huge information base for such fully integrated community planning.
ETSAP originated as an International Energy Agency program to help establish energy technology R&D priorities on the basis of the needs of all the IEA countries. A common methodology and comparable databases have been the touchstone of the program since its very beginnings. The standard MARKAL model has continued to be the focus of the group's analyses, and recurring efforts have been made to assure reasonable consistency in the national databases.
Responsibility for technology data rests at the national level, in many cases subject to guidance or approval from reference groups or steering committees. To the extent that studies have been primarily aimed at national (and subnational) levels, the need for "harmonized" data declined. Locally available and "approved" data sources became increasingly important. In addition, since the early Eighties enormous amounts of studies have been carried out in all countries, often making available the data collected and used.
Nonetheless, most MARKAL users regularly exchange their databases, so that newly developed insights in one country are made available to others (subject to restrictions on proprietary information), and national data can be compared with those adopted in other countries. Comments received in the process can help to further improve and expand the individual databases. In this respect, the MUSS system has proved its value in facilitating the exchange of data, offering space for comments and documentation for individual technologies and printing standardized technology data sheets.
Expecting a series of revised model calculations in the early part of Annex
V, and anticipating that more countries would be included, comparative
overviews of input data for various subsectors in the MARKAL model
specifications were made by the Operating Agent. Together with
observations and comments, these were circulated and discussed at one
of the meetings. However, because of the focus on exploring and
developing new approaches and tools for establishing links with economic
models, the revised and expanded common study using MARKAL was
taken off the Annex V program of work. Nevertheless, the annotated
overviews prepared for the central fossil and nuclear power plants, for
renewable electricity generation options, and for automobile technologies
may well have served to upgrade individual databases.
9.1 The CHALLENGE Project
Although a common set of runs among the ETSAP participants was delayed, four countries participated in CHALLENGE, a cooperative international project on energy and environment systems analysis. CHALLENGE consists of a network of scientists from East and West European countries. The project is intended to facilitate international negotiations and cooperation by providing a scientific basis for decisions on response strategies to reduce environmental stresses and climate risks due to energy use.
The first phase of the CHALLENGE project was to produce a coherent set of national case studies of emission reduction strategies with a generally consistent framework of emission control strategies and common assumptions. The Netherlands, Norway, Sweden and Switzerland used the MARKAL or MARKAL-MACRO models to calculate the consequences of successively higher carbon taxes or emission reduction requirements on the national energy and economic systems. For each scenario in successive decades out to 2030, the effect on GDP and CO2 emissions was estimated, together with production and consumption of primary energy, oil, natural gas, coal, nuclear energy and renewables.
CHALLENGE plans to aggregate the individual national results to evaluate the consequences of joint action on international trade and energy markets.
During Annex V, the Operating Agent and some participating countries
provided inputs to major international studies by the International Energy
Agency, Organization for Economic Cooperation and Development, and
the Annex I Expert Group on the UN Framework Convention on Climate
10.1 IEA Study "Electricity and the Environment"
With the cooperation of the participants from Italy, Japan, UK and USA,
the Operating Agent contributed to the International Energy Agency study,
"Electricity and the Environment." Detailed descriptions were provided of
technologies available for electricity supply and demand in the short and
medium term. The information included technical performance and
engineering costs. Specific data were drawn from the MARKAL databases
of the four cooperating countries.
10.2 OECD Study of Environmental Implications of Energy Subsidies
As part of a study by the OECD Secretariat of the environmental implications of energy and transport subsidies, the Italian participant used an "elastic" version of MARKAL to evaluate the impact of removing financial subsidies from the electric sector in Italy. The many ways in which financial interventions affect the electric supply industry were searched out, and MARKAL was used to assess their effect on electric and energy system costs and CO2 emissions.
Interest in energy subsidies has been spurred by the possibility of imposing carbon taxes to reduce emissions of carbon dioxide. Should we not first remove any subsidies that encourage the use of fuels that emit carbon dioxide?
Broadly speaking, a subsidy is any intervention, or failure to intervene, that results in the prices of goods or services to producers or consumers, or the quantities produced or consumed, differing from the those that would occur in a fully competitive market with all social and environmental costs internalized. In this study, financial subsidies were analyzed in depth; other economic subsidies were not analyzed because of the difficulty in quantifying their amounts and impacts. A "net subsidy" is defined as a financial transfer from the economy as a whole to electric producers and consumers. A "cross subsidy" is a financial transfer from one subsector of the electric market to another. Both kinds of subsidy perturb the market and make actual prices different from long-term marginal prices.
The main finding was that, in the early 1990s, about 40 percent of the electric market was affected by financial subsidies. Net public support of the electric sector by the economy was in the range of 15-20 percent of the value of electric production. Cross subsidies among different consumers and producers amounted to 20-25 percent of the market.
If subsidies are removed from the electric sector, the following effects are likely, among others:
The interaction of carbon dioxide emission control policies with the removal of electricity subsidies is synergistic. When applied independently each of them improves the welfare of the system; when applied together the welfare improvement is higher.
To obtain these results, the "elastic MARKAL" approach was used in
which fixed energy demand projections are replaced by price-dependent
demand profiles. The standard MARKAL can be seen as a shadow price
generator: once the demand levels for private commodities (like useful
energy) and public commodities (like environmental emission restrictions)
are defined, the model calculates the shadow prices in each time period
for each commodity, making use of the very detailed technological supply
curve determined by many hundreds of technologies that comprise the
database. With this version of MARKAL, however, it is possible to
increase or decrease the shadow prices of resources and commodities by
amounts equal to taxes or subsidies. In other words, it is possible to build
separate scenarios with shadow prices similar to present and projected
internal prices, which depart from long-term marginal productions costs
and include taxes and subsidies.
10.3 UN-FCCC Working Paper on Full Cost Pricing
For a project on "Policies and Measures for Common Action" that was conducted by the Annex I Expert Group on the UN Framework Convention on Climate Change, the Operating Agent prepared an analysis of the effect on national carbon dioxide emissions of full cost pricing of electricity that takes into account emissions of particulates, SO2, and NOx. In conjunction with the International Energy Agency Secretariat, case studies were made of the electric sectors in Italy and the Netherlands using, respectively, MARKAL and MARKAL-MACRO. Full cost pricing for local or regional pollutants ■ to the extent that it may also reduce CO2 ■ would provide an added greenhouse benefit at no additional cost.
Full cost pricing is also called internalizing external costs, or externalities. Externalities are unintended by-products of commercial activities such as power generation. Examples are impacts on health, the loss of environmental quality or recreational facilities. While externalities are easily felt as influences on well-being, their measurement in exact terms is difficult. Huge uncertainties exist in the valuation of externalities due to particulates, SO2, NOx, and CO2 ■ sometimes of the size of several orders of magnitude ■ depending primarily upon location and population density.
The purpose of the study was to determine whether full cost pricing that takes into account particulates, SO2 and NOx would also reduce emissions of carbon dioxide. Results of the modeling in the two countries indicate that CO2 reductions would range from less than 1 percent (through 2010 in Italy) to about 10 percent (through 2020 or 2030 in the Netherlands). In the short term, even the possibility of higher CO2 emissions cannot be precluded because of the introduction of some measures, such as switching from high-sulphur fuel oil to low-sulfur coal. Judging by these two case studies, the incidental impact on CO2 emission reduction is likely to be too small to influence any decisions regarding the introduction of full cost pricing that considers only particulates, SO2, and NOx.
Full cost pricing as a common action among countries could help to level the international playing field for the trade of electric power to the extent that it would equally reflect the full cost of production in each country. The amount of the full cost adders seems small in comparison with the electricity losses due to transmission between countries. While full cost pricing is technically feasible, however, numerous large hurdles remain, including the consensus about the acceptability and validity of damage cost estimates. The damage cost values in one country cannot be transferred to another.
Expansion of environmental cost adders explicitly to include carbon dioxide emissions would be an effective way to assure reduction of CO2 emissions as well as those of the local and regional pollutants.
The primary focus of Annex V was the linking of the technology-rich MARKAL model with econometric models. This was accomplished in part by hard-linking MARKAL with economic drivers to create MARKAL-MACRO and MARKAL-MICRO. This has brought MARKAL into the realm of nonlinear optimization, a much more complex and difficult world.
Nonlinear solvers have not evolved at anywhere near the rate of improvement of linear optimizers. There was initially concern that models containing more than 3,000-4,000 rows might prove too difficult to solve, but full production models of more than 12,000 rows are now being handled, but sometimes with difficulty. In order to achieve this, it was necessary to remove all nonlinear constraints in the model, moving them directly into the objective function. This was accomplished with the assistance of the GAMS Development Corporation.
In addition to these major developments, a number of additional features were added to the basic model formulation to allow the MARKAL modeler to better represent the combined energy-economic systems. Notably:
The initial achievement was to ensure the availability of an integrated modeling environment where all the model formulations shared the same date orientation and operated from within the user environment. Another key feature is that a mechanism was developed to map the GAMS internal representation of the model into its OMNI-MARKAL equivalent. In addition to steps to foster the transition to the GAMS-based model formulation, a number of enhancements were made to the capabilities of MUSS to make it even more user-friendly.
A completely new, updated and expanded "PC-MARKAL/MUSS Manual" was published. For new users to familiarize themselves with the system, a "Getting Started" tutorial was produced, supplied with hands-on training exercises.
The metamorphosis of MARKAL and its extensions into highly automated, user-friendly software for desktop computers has made them accessible to a new class of users. Like MARKAL and unlike most other models being used to assess possible future costs of greenhouse gas abatement, MARKAL- MACRO is a generic model that can be applied to wide range of countries, including those for which detailed macroeconomic models have not otherwise been developed.
As a result of bilateral programs, there are already more countries outside than inside the OECD using MARKAL. Transferring know-how and building capacity in developing countries is a requirement of the Framework Convention, adding impetus to the spread of MARKAL to additional non-OECD countries. With more support from national and international organizations, the transfer of the soft technology developed by the ETSAP research partnership could be greatly enhanced.
The long-term benefit will be contribution to building an international
consensus, founded on a strong scientific understanding of the feasibility
and costs of technological options for reducing greenhouse gas