Improving the B&B search for large-scale hydrothermal weekly scheduling problems

This paper presents an optimization based algorithm to solve the weekly scheduling problem of a large-scale hydrothermal power system, formulated as a mixed-integer linear programming model (MILP). The main drawback of the MILP approach is the high computational burden required to solve large-size problems. The proposed algorithm tackles this problem by providing an initial feasible and integer solution, which enhances the search of the Branch and Bound (B&B) over the space of feasible solutions, reducing the resolution time. A detailed representation of thermal, pumped storage, and hydroelectric units is considered, taking into account the net head dependence of hydro plants by means of an under-relaxed iterative process. The presented algorithm has been applied to real-scale study cases, obtaining satisfactory results in computational time and optimality.