A hybrid stochastic search approach for unit commitment with hard reserve constraints

Reliability considerations are of paramount importance in the operation of modern power systems. In Unit Commitment (UC) these considerations are reflected in the form of hourly Spinning Reserve (SR) Capacity. Spinning reserve, more often than not, is treated as soft constraint by using penalty methods in Unit Commitment literature. This doesn't guarantee a strict compliance of minimum hourly reserve constraint throughout the schedule period which implies decreased reliability of the system. This paper presents a Hybrid Stochastic Search based approach for solving Unit Commitment (HSSUC) with hard reserve constraints. The method is a synergistic combination of Genetic Algorithms (GA) and Simulated Annealing (SA), thereby combining the advantages of both. Several other features are incorporated for faster convergence to better solution. For a given commitment, an analytical Economic Load Dispatch (ELD) is employed to get the minimum operating cost in each hour. This speeds up the algorithm considerably over the iterative ELD methods that are time consuming. The algorithm has been tested and results have been compared with those presented in the literature. The proposed algorithm gives solutions with lower cost with less computational effort, even with strict compliance of constraints.