General model structure

The proposed optimization model of the Russian electricity market is expected to establish functional dependences between various attributes of the actual market. These dependences can later be used to analyse and predict the behaviour of the market under different experimental conditions. However, in order to obtain the desired input-output characteristics, assumptions about the structure of the proposed model should be made.

The key structural elements of a typical optimization model of the electricity market or power system are: decision variables, uncontrollable model parameters, performance indicators, and objective function (Badinelli et al., 2010).

The factors that can affect the values of performance indicators are categorized into two sets or collections; uncontrollable factors and controllable factors. Uncontrollable factors are fixed parameters of the model, and they can have constant values in the course of modelling or be stochastic by nature. Typical examples of uncontrollable

parameters of electricity market models include pre-specified limits of available transmission capacity, installed capacity of plants, maximum fuel consumption limits, power system reliability criteria, outdoor temperature levels, or amounts of precipitation. Sometimes, certain uncontrollable model parameters and constraints, such as required operating reserve margins in the power systems or electric loads, may also be considered to be known in advance, and thus, they do not have to be modelled. Other input parameters, such as hydro reservoir inflow rates or unplanned outages of capacity, usually have to be estimated or modelled separately, for example, by using available historic data. Controllable model parameters or decision variables, on the other hand, represent alternatives or choices of resource usage that should be made to obtain the preferred and feasible results.

A model typically requires definition of how the controlled and uncontrolled variables are connected to the model output variables. Examples of these input–output relationships that have to be considered in the model can include:

• Production cost functions of generators

• Dependences of flows in the transmission network from power injections and withdrawals at nodes.

The proposed decision models can be characterized as a prescriptive model, that is, the model prescribes the actions that have to be done to achieve the best and most desired alternative. Generally, prescriptive solutions from the energy market models can be obtained using various optimization techniques including branch-and-bound or evolutionally heuristic algorithms (Badinelli et al., 2010).

For the modelling, a set of performance indications used as the model outputs should be defined. Examples of performance indicators can include:

• Hourly day-ahead energy market price estimates

• Energy production plans of the modelled generators

• Total net revenues obtained by the modelled plants from the day-ahead energy market over a single model run

The values of performance indicators determined by the model may be used directly in the decision making process or averaged over time and over the examined modelling scenarios to obtain generic measures of rationality and feasibility of each proposed alternative. Examples of the generic performance measures can include the number of start-ups and shutdowns of plants over a month, duration of operation under different loading levels, fuel economy, or average marker revenues.

Following the objectives of this doctoral dissertation, the total revenues of the modelled generators are chosen as the main generic performance measure. Evaluation of the generic measures requires estimation of two main decision variables: the prices and the production plans of generators in the market. In the actual wholesale energy market of Russia, market prices and generation schedules are obtained as a result of solution of two major optimization problems: unit commitment and economic dispatch. Therefore, the proposed optimization model of the short-term energy market is required to replicate in a sufficient detail the process of market clearing performed by the SO and the ATS in the actual market. Thereby, a mathematical problem of total market welfare maximization under given power sector constraints should be formulated. In the proposed optimization model, the group of controllable parameters should include binary commitment decisions of thermal generators and production output levels of hydro, thermal, and nuclear power plants in the short-term market. These variables are considered the most fundamental parameters for describing the physical state of the generation sector of Russia. To represent the values of state variables in a realistic manner, the constraints associated with actual electricity sector operation should be introduced.

In the context of the Russian electricity market, the list of basic constraints describing the boundary conditions of the actual electricity sectors operation should cover:

• Thermal generation sector constraints

o Operational constraints of condensing power plants (GRES).

o Operational constraints of co-generation plants (CHP)

• Nuclear generation sector constraints

• Hydro generation constraints

o Output and energy constraints of plants with water reservoirs o Run-of-river plant constraints

• Limitations in the national transmission system

• System-security constraints o Operating reserve margins

Figure 3.1 depicts the general structure of the proposed optimization model of the Russian day-ahead energy market.

Figure 3.1: General structure of the proposed optimization model

The functions of the SO in the model are narrowed down to performing the process of unit commitment in order to meet the forecasted customer’s demand at the lowest possible costs given by the production offers of generators. Activities of the commercial operator ATS are replicated by implementing an Economic Dispatch of the committed generation sources.