An Optimization and Heuristic Based Algorithm for the Short Term Scheduling of Large-Scale Hydrothermal Systems

Abstract

This paper presents an optimization and heuristic based algorithm to solve the weekly scheduling of large-scale hydrothermal systems. The aim of the presented algorithm is to find the optimal hydrothermal scheduling with a detailed representation of the hydroelectric system (cascaded reservoirs, head dependent units, input-output surfaces, water rights and minimum-maximum operating outflows). Besides hydro and thermal generators, pumped storage units are considered using composite energy representation. The proposed algorithm can be divided in three steps. Firstly, the global problem is solved but relaxing the commitment decisions and grouping consecutive hours within periods up to 4 hours. This phase allows to quickly reach a good initial solution thanks to the particular structure of the problem when applied to real input data. The second step uses hourly periods but thermal unit-commitment decisions are still relaxed. After that, electrical power produced at each hydro plant is computed using the current water discharged and the input-output curve corresponding to the net head obtained in the last execution. Finally, after fixing the hydro generation obtained in the previous step, commitment and output power of thermal units are obtained by mixed integer linear programming. The proposed method has been applied to real scale study cases with excellent results in computational time as well as in mismatches from the global optimal solution.