A nonlinear programming approach to a very large hydroelectric system optimization

AbstractA large scale hydroelectric system optimization is considered and solved by using a non-linear programming method. The largest numerical case involves approximately 6 000 variables, 4 000 linear equations, 11 000 linear and nonlinear inequality constraints and a nonlinear objective function. The solution method is based on(i)partial elimination of independent variables by solving linear equations,(ii)essentially unconstrained optimization of a compound function that consists of the objective function, nonlinear inequality constraints and part of the linear inequality constraints. The compound function is obtained via penalty formulation. The algorithm takes full advantage of the problem's structure and provides useful solutions for real life problems that, in general, are defined over empty feasible regions.