Combined hydro-wind generation bids in a pool-based electricity market

Abstract

Present regulatory trends are promoting the direct participation of wind energy in electricity markets. The final result of these markets sets the production scheduling for the operation time, including a power commitment from the wind generators. However, wind resources are uncertain, and the final power delivered usually differs from the initial power committed. This imbalance produces an overcost in the system, which must be paid by those who produce it, e.g., wind generators among others. As a result, wind farm revenue decreases, but it could increase by allowing wind farms to submit their bids to the markets together with a hydro generating unit, which may easily modify its production according to the expected imbalance. This paper presents a stochastic optimization technique that maximizes the joint profit of hydro and wind generators in a pool-based electricity market, taking into account the uncertainty of wind power prediction.

Introduction

The generation of wind power has increased around the world in recent years. The growth in Europe has been encouraged by the EU Directive 2001/77/EC [1]. The goal of the directive is that 12% of gross energy consumption in the European Union must be provided by renewable energy sources (RES). In some countries, like Spain, wind power supplies about 10% of the power demand, although some days in March 2007 reached up to 23% of the electric demand. Nowadays the wind power installed means 13,606 MW, which accounts for 15% of the total installed power. This high level of wind energy entails some new technical and financial challenges.

Wind farms have significant difficulties predicting their power output accurately [1], [2], [3]. This uncertainty involves energy imbalances with regard to the power committed in pool-based electricity markets, and these imbalances usually result in financial penalties [4], [5], [6]. One way to reduce the expected imbalance cost is to use a stochastic optimization tool. Taking into account the penalties due to imbalances, this kind of tool allows optimal wind energy to be traded on the market [7]. Another method for reducing the expected imbalance cost is to work together with other types of generating units, like a Hydro Generation Company (HGENCO). Some references use a pure hydro system or a hydro generation/pumping system combined with a Wind Generation Company (WGENCO) to provide the committed power in the electric power system [8], [9], [10], [11], [12], [13]. In these cases, the aim is to find the optimal hydro or hydro/pumping operation in order to provide a reliable energy supply. This strategy is useful in small or isolated systems. In [14], [15], a wind and pumping ensemble system is considered in a deregulated market. In both research papers, starting from a deterministic and well-known wind generation forecasting, hydro generation and pumped water are found. This algorithm is enhanced in [16], where the expected wind generation is considered as a stochastic parameter. Several simulations using the Monte Carlo method are carried out for different values of wind generation.

In this paper, the imbalances are treated differently. In a deregulated market, the WGENCO–HGENCO ensemble (WH-GENCO) tries to maximize its own profit. Thus, the company reduces imbalances only if it increases this profit. In that case, the market incentive to avoid imbalances will be defined by the penalty price of imbalances.

The problem is set for Spanish market rules, although they are not followed exactly. The Spanish electricity market is a pool-based market where all market players must present their bids for a whole day at 10:00 AM of the previous day. The accepted bids program entails a power commitment for the whole day (24 h). This market is called daily market. This commitment may be modified six times a day, every 4 h, in the intraday markets, buying or selling energy by means of bids submitted 3 h before the operation time. Finally, if there are differences between the last accepted bid of a market player and its actual production, the system overcost due to this mismatch between generation and demand must be paid by those agents who incur it, in accordance with the amount of this deviation. It involves a decrease in the revenue of these agents (wind generators are usually found among them). Under this regulatory framework, two alternative strategies could be followed:

• Combined operation strategy: Once the WGENCO and HGENCO energy is traded on the electricity market, both companies have a committed power. Nevertheless, in real time (about 1 h ahead) and starting from the current generated wind power, the WGENCO can calculate with small uncertainty the imbalances for the energy committed in the next hour. With that information, at this time, the HGENCO may change its own generation in order to obtain the maximum profit for the WH-GENCO ensemble. Thus, the HGENCO will move its generation towards the optimal values to reduce the wind power imbalances whenever this action increases the combined profit. Finally, notice that if the HGENCO traded all its rated power in the pool market, it could not support the WGENCO position.

• Combined bidding strategy: In this case, depending on the energy to be traded on the electricity market, the HGENCO plans its power reserve and, consequently, its availability to reduce imbalances caused by wind power. In accordance with the plan, the decision is taken considering the expected wind power distribution probability, which is a stochastic variable. It should be noted that this bid design strategy sets the optimal energy to be submitted to the market with the aim of obtaining the maximum profit for the WH-GENCO joint optimal operation (previous item), but with the bid defined hours before instead of in real time. That is, the combined bids design strategy has a higher hierarchical decision level than the operation planning.

In this paper, the authors follow this last novel strategy. The WH-GENCO ensemble will be considered a price-taker and a detailed hydro model is used. The proposed algorithm is a mixed-integer (0/1) linear problem which has been solved under GAMS mathematical modelling language using the solver CPLEX 9.0 [17].

This paper is organized as follows. Considerations about the wind power forecast and hydro system are shown in Sections 2 Uncertainty of wind prediction, 3 Hydro model, respectively. Section 4 formulates the mathematical model for defining the combined bid. Section 5 provides an application example, comparing the results with other possible ways to decide how energy is traded. Finally, conclusions are stated in Section 6.