A Two-stage Approach To Solving Large-scale Optimal Power Flows

A two stage optimization algorithm is developed for the solution of the large-scale optimal power flow (OPF) problem. The solution scheme is arrived at by exploiting the structural properties of the OPF problem formulation, including the physical characteristics of its variables. The algorithm incorporates certain features of the generalized reduced gradient and the penalty function methods. At each iteration a modified reduced gradient is used to provide a direction of movement. The use of penalty functions in the optimization is kept at a minimum. The generalized reduced gradient ideas are adapted in such a way that a routine power flow is used in the optimization. Numerical results of test cases including a 1410 bus system are given.