Development of a Tool for Profit Based Unit Commitment in Deregulated Electricity Markets Using a Hybrid Lagrangian Relaxation – Evolutionary Particle Swarm Optimization Approach

Abstract

As electricity markets undergo deregulation all over the world, the approach to generation scheduling or unit commitment (UC) changes significantly. In tra ditional electricity markets with electricity utilities which act as system opera tors and also own generation units, UC is done based on a cost minimization objective. However, in deregulated markets, individual generation companies (GENCOs) have to carry out UC independently based on forecasts of energy and reserve prices for the scheduling period. The Generation Company (GENCO)’s UC strategies are developed with the aim of maximizing expected profit in what is known as Profit Based Unit Commitment (PBUC). Such profits are not only dependent on revenues from sale of energy and ancillary services such as reserve, but also on the cost characteristics of the generating units owned by the GENCO. This research develops a tool for carrying out PBUC for GENCOs in deregulated electricity markets. The tool is presented as a collection of MATLAB m-files that can be easily applied to any test system with the data stored in a specified format in an excel file. The MATLAB code is an implementation of solution algorithms that are developed and tested using simulations carried out for typ ical test systems. First, a solution methodology for the PBUC problem using a hybrid of the Lagrangian Relaxation (LR) and Particle Swarm Optimization (PSO) algorithms is implemented in MATLAB software. The PSO algorithm is used to update the Lagrange multipliers resulting in an optimal solution. It is found that the final solution is dependent on the values of the PSO algorithm parameters that have to be specified before running the algorithm. An analysis of the solution quality for various PSO algorithm parameters is carried out to determine the parameters that give the best solution. The algorithm is tested for a GENCO with 54 thermal units adapted from the standard IEEE 118-bus test xiii system. To tackle the challenge of the solution quality being dependent on the algorithm parameters, the Evolutionary Particle Swarm Optimization (EPSO) algorithm is explored. EPSO is chosen based on previous research which showed that it generally results in better solutions than PSO because of a “self-tuning” characteristic of the parameters. Simulation results for a test GENCO show that the EPSO algorithm provides better solutions and has better convergence charac teristics than the classic PSO algorithm. A second important consideration in the solution of the PBUC problem is the GENCO’s market power i.e. it’s influence on the market prices and/or demand. While a GENCO’s bilateral demand is previously agreed on and therefore well known, allocations from the spot energy market depend largely on the GENCO’s bidding strategy which is dependent on the GENCO’s market power. A GENCO thus requires an optimal bidding strategy (OBS) which when combined with a PBUC approach would maximize its profits. A solution of the combined OBS-PBUC problem is therefore devel oped. Simulation results carried out for a test power system with GENCOs of differing market strengths show that the OBS depends largely on a GENCO’s market power. Larger GENCOs with significant market power would typically bid higher to raise prices, while smaller GENCOs would typically bid lower to capture a larger portion of the spot market demand.