Constraint Oriented Neighbourhoods: an application to the hydrothermal coordination problem

Abstract

In hydro-thermal power production systems a proper coordination between the hydro and the thermal sub-systems is required to minimize the total production costs. This is generally achieved by decomposing the problem in two sub-problems, one handling the hydro component and the other handling the thermal one. Usually, the hydro scheduling subproblem is solved first and then, for the remaining demand (the total load demand minus the hydro production) the thermal sub-problem is solved. The process is repeated until convergence is achieved. Metaheuristic approaches dealing with the Hydro-Thermal Coordination problem (HTC) are also based on the same type of problem partitioning: the metaheuristic tackles the combinatorial thermal sub-problem, while other techniques are used to solve the hydro subproblem. Hierarchical approaches are then run in an iterative way and may lead to suboptimal results, due to lack of convergence. This paper proposes an integrated approach to solve the HTC using metaheuristic techniques with Constraint Oriented Neighbourhoods [1]. This is achieved by discretising the production levels of the hydro units and by defining appropriate neighbourhood movements that allow changes to be made in the schedules of both the thermal and the hydro units.