Optimization of Power Flow in Electric Power Systems via Parallel Processing

Abstract

An important nonlinear optimization problem arises in the management of large electric power systems. The great number of variables and constraints, typically characterizing this problem, makes very hard to pursue its solution via classical methods. At present, the only efficient and reliable way to compute the solution of a large dimensional optimal power flow problem, seems dependent on the use of parallel processing methods. In this paper, the economic dispatching of the thermoelectric power production, modelled as an optimal power flow problem, has been solved via the parallel implementation of a nonlinear optimization gradient algorithm. The numerical results, obtained by using a Vector Computer Cyber 205, show very sharply the high efficiency and reliability of the computations, in spite of the large-scale feature of the problem.