Security versus Computation Time in IV-ACOPF with SOCP Initialization

Achieving the optimal solution to the security-constrained AC optimal power flow (SCACOPF) problem in real-time is a computationally challenging task. Successive linear programming (SLP) with a near optimal initialization has shown promising potential for solving the problem. It is, however, essential to improve the computational performance of the algorithm by only choosing the most critical contingencies, instead of modeling the extensive list of all the possible contingencies. This paper examines the trade-off between different levels of reliability, i.e., post-contingency network violations, and computation time in the SCACOPF problem. It is shown that a similar level of reliability can be obtained by only modeling the most critical contingencies, while substantially improving the computation time. The paper uses a combination of second order conic programming (SOCP) relaxation of the ACOPF and fast decoupled power flow to initialize the SLP algorithm for solving the SCACOPF. The method is tested on a range of test systems, using off-the-shelf solvers, i.e., Gurobi and CPLEX, and the results are presented and compared.