Decentralized Reactive Power Optimization Method for Transmission and Distribution Networks Accommodating Large-Scale DG Integration

With the integration of distributed generation systems, distribution networks and transmission networks have become more coupled, especially, with respect to voltage problems. This paper presents a decentralized reactive power optimization method for integrated transmission networks and distribution networks based on the generalized Benders decomposition (GBD). The second-order conic programming relaxation technique is used to make the subproblem (distribution network model) convex, which can guarantee that the GBD process converges. An improved GBD iteration procedure is proposed by exploiting the special structure of the problem, which does not need to generate optimal cut. Moreover, a revised feasible cut generating approach is also developed to improve the efficiency. Theoretical analysis and numerical tests show that the convergence of this decentralized method is guaranteed and the coordinated optimization scheme outperforms conventional separated methods by eliminating overvoltage and reducing network loss in the active distribution network.