Impact of Partitioning on the Performance of Decomposition Methods for AC Optimal Power Flow

The optimization problems in power systems become larger and larger due to the increased number of variables from distributed generation and flexible loads. Hence, there has been growing interest in decomposition methods that facilitate distributed decision making. However, limited effort has been spent on the actual implementation of decomposition methods including determining how to partition the problem and what information to exchange among subproblems, which may greatly impact the efficiency of decomposition methods. In this paper, we evaluate the effects of partitioning on the convergence speed of decomposition methods for solving the AC Optimal Power Flow problem. In addition, we propose a speed-up method for the Optimality Condition Decomposition by adding a correction term to refine the search direction. Simulations on the IEEE-30 system show that the convergence speed of the decomposition method can be significantly improved by using a proper partition of the system and the correction term.