A Distributed Algorithm based 2-level Optimization for Multi-area Optimal Power Flow with Discrete Variables

Abstract

This paper presents a fast computation algorithm for a multi-area optimal power flow (OPF) with discrete variables. The multi-area OPF with discrete variables is formulated as a Mixed Integer Non-Linear optimization Problem (MINLP) that includes both continuous and discrete variables as decision variables. In this paper, in order to solve the multi-area OPF with discrete variables efficiently, we propose a resource allocation-type decomposition algorithm considering continuous relaxation of discrete variables. By using the resource allocation type decomposition, it is possible to guarantee the feasibility in the optimizing process, and it is possible to suppress the amount of error by absorbing and buffering the relaxation error by the resource allocation variables. In addition, since the variables that interfere between adjacent Areas is utilized as resource allocation variables, the sub-problems obtained by the decomposition become independent problems, and the original MINLP can be solved in completely distributed approach. Numerical examples against several multi-area test systems show that the computational costs can be significantly reduced with less computational errors compared to a centralized approach.