Electricity - GEM-E3
| Corresponding documentation | |
|---|---|
| Previous versions | |
| Model information | |
| Model link | |
| Institution | Institute of Communication And Computer Systems (ICCS), Greece, https://www.iccs.gr/en/. |
| Solution concept | General equilibrium (closed economy) |
| Solution method | Optimization |
| Anticipation | |
CGE models have been criticized for their simplified modelling approach of the energy system. The usual CGE representation of the energy production by means of aggregate production functions fails to capture crucial characteristics of the sector reducing the credibility of simulations related to energy policies and technology dynamics. The bottom up models employed instead, ignore the feedbacks from the interaction of the energy sector with the wider economy within which it operates.
The development of a modelling framework that encompasses the multi market equilibrium of top down models with an engineering consistent representation of power producing technologies constitutes a long-standing challenge in applied energy policy analysis. Many different approaches have been employed to link bottom up and top down models and can be classified in two main categories:
• Hard link approach, that is, integrating both bottom-up and top-down features in a consistent modelling framework. Such an integrated framework is provided by the specification of market equilibrium models as mixed complementarity problems.
• Soft-link or decomposition approach where bottom-up and top-down models are run independently of each other. In this case results from one model are fed into the other, and vice versa.
A characteristic example of the first category is found in Boehringer (1998) where the electricity generating technologies are modeled as specific activities within a mathematical-programming representation of the electricity sector, which is embedded directly in a computable general equilibrium model. In particular his approach is based on the complementarity formulation of the general equilibrium problem while the representation of the electricity producing sectors is based on Koopmans (1951) activity analysis framework. The standard aggregate production functions (C.E.S. or CD) used in the model are replaced by a set of discrete Leontief technologies (fixed input/output vector).
Towards the same direction lies McFarland et al. (2004), who suggest a more flexible format through a C.E.S. representation of energy technologies. Their approach consists of splitting the energy sector using engineering bottom up data and then calibrate the model’s smooth production functions on these data. In particular in their approach the cost estimates on capital, labour, and fuel inputs are used directly as the CES share parameters. The nesting scheme of the production function allows for the appropriate input substitution while the control of technology penetration rate is based on an endogenous quasi fixed factor coefficient introduced at the top level of the C.E.S. production function. Each technology produces electricity through a C.E.S. aggregation of its primary and secondary inputs (low elasticities of substitution chosen at this nesting level), while total electricity production results from a CES aggregation of all power technologies represented in the model (high elasticities of substitution at this nesting level).
A disadvantage of this approach lies in its treatment of investment decisions. That is, investment is either allocated to electricity technologies exogenously or decided at the level of the aggregate electricity sector and then allocated to each technology using a logit function. This investment formulation although it allows for multiple technologies with different costs to coexist is not sufficient to represent the investment behavior of the electricity sector (i.e. each sector should decide the level of investment as a function of its profit function and then this investment demand should be translated to demand for investment products produced by other sectors). In addition the non-smooth (kinked) representation of power supply results in sharp shifts in the technology mix of electricity production implying unrealistic swift switching between technologies.
The second category refers mainly to a decomposition method that links bottom up models with top down by combining different mathematical formats - mixed complementarity and mathematical programming. In Boehringer & Rutherford (2008) mixed complementarity methods (MCP) are used to solve the top-down economic equilibrium model and quadratic programming (QP) to solve the underlying bottom-up energy supply model. Then they reconcile equilibrium prices and quantities between both models through an iterative procedure portray this iterative solution process).
Hybrid Bottom Up Top Down (BUTD*)* CGE models are still rare in the policy modelling literature due to difficulties arising from the integration of macroeconomic and engineering data in a consistent way. E3M-Lab has designed and incorporated into the GEM-E3 model a bottom up top down module. The motivation for this development was the need for a better representation of the electricity sector investment decision. Toward this end, electricity producing technologies were treated as separate production sectors while their investment decision is discrete. The advantage of this approach is that it is fully consistent with the general equilibrium framework while it leads to a full identification of the technologies.
The Input-Output tables represent the electricity sector as an aggregate of two activities, the power generation and the transmission and distribution of electricity. This detail is not sufficient for the development of the bottom up model, so it has been necessary to split the Input-Output column and row in different activities, some corresponding to power generation by technology and the rest corresponding to transmission and distribution of electricity. The split was performed by combining data from energy balances and company- related economic data on generation and transmission and distribution activities by country. The aggregate data were based on Eurostat, IEA and USA DOE statistics. For example, the disaggregation shows that the generation cost accounts for over half of total cost and in most E.U. countries they account for over 60% while transmission costs range between 5% and 10%.
In order to disaggregate the power sector appropriate mapping has been specified between the entries of the Input-Output table and the engineering information retrieved from the technical databases. For this purpose data on capital cost, fixed operating and maintenance cost, fuel cost and other variable operating and maintenance costs, related to the energy producing technologies to be incorporated in the model following cost elements have been extracted from the engineering database.
The unit costs have been associated with the corresponding cost elements of the Input-Output statistics, according to the following principles: i) annualised capital costs correspond broadly to operating surpluses, ii) fuel costs correspond to the fuel input, iii) fixed operating and maintenance cost correspond to non-energy inputs (materials), iv) variable operating and maintenance costs are associated with wages and salaries paid to employees in power generation. Since the entire GEM-E3 model is calibrated on the social accounting matrices the macroeconomic data have been kept constant and the market and cost shares of the technologies have been appropriately adjusted. The purpose of the calibration has been to depart as little as possible from the flows suggested by the engineering information while respecting exactly the totals appearing in the original input output table. For this purpose a cross entropy method has been applied.
The model represents separately 10 conventional and RES power generation technologies. The technologies incorporated in the GEM-E3 model are presented in the following table.
Table 4: Electricity producing technologies represented in GEM-E3 model
| No | Description | No | Description |
| 1 | Coal fired | 6 | Hydro electric |
| 2 | Oil fired | 7 | Wind |
| 3 | Gas fired | 8 | CSP and Photovoltaics |
| 4 | Nuclear | 9 | Coal CCS |
| 5 | Biomass | 10 | Gas CCS |
Generation costs are conceived in three categories: i) investment costs, ii) operating and maintenance costs and iii) fuel costs. Unit cost data and projections to the future for the first two categories are extracted from the TECHPOL and PRIMES database. The fuel costs depend on other variables of the GEM-E3. The data for each technology as introduced in the model are presented in the following table.
Table 5: Electricity production cost shares
| Coal fired | Oil fired | Gas fired | Nuclear | Biomass | Hydro electric | Wind | PV | |
| Agriculture | 25.0 | |||||||
| Coal | 24.3 | |||||||
| Oil | 70.6 | |||||||
| Gas | 73.2 | |||||||
| Chemicals | 6.7 | |||||||
| Other equipment goods | 5.0 | 0.5 | 0.5 | 0.5 | 1.5 | 1.0 | 9.8 | 0.8 |
| Construction | 3.0 | 2.0 | 4.7 | 1.0 | 1.5 | 3.0 | 5.8 | 6.7 |
| Capital | 56.6 | 22.3 | 19.3 | 87.6 | 67.4 | 80.3 | 8.0 | 83.2 |
| Labour | 11.1 | 4.7 | 2.2 | 4.2 | 4.6 | 15.7 | 4.4 | 9.2 |
| Total | 100 | 100 | 100 | 100 | 100 | 100 | 100 | 100 |
Source: Calculations based on TECHPOL and PRIMES databases
The shares of each technology in power generation in the base year are introduced from energy balance statistics. Some of the potential technologies that may develop in the future are not used in the base year. Since the production function for power generation is calibrated to the base year, it is necessary to introduce artificially small shares even for the non existing technologies in order to allow for the possibility of their penetration in the future under market conditions.
The development of the database on technology market shares and share of transmission and distribution cost to total cost of electricity production has been based on the TECHPOL database, the ENERDATA database and the PRIMES model database.