MINLP formulation of optimal reactive power flow

A method for the solution of optimal reactive power dispatch which treats var sources and transformer tap ratios as discrete variables is presented. The optimal reactive power flow problem is inherently a mixed-integer nonlinear programming (MINLP) problem. For finding the global optimal solution of this problem, an MINLP formulation is proposed and executed. In this formulation, discrete variables, var sources and tap ratios, are modeled as binary variables. The MINLP problem with only continuous and binary variables is solved by an outer-approximation/equality-relaxation algorithm. In this algorithm, the MINLP problem is decomposed into a mixed-integer linear programming master problem, and an nonlinear programming (NLP) subproblem. These two subproblems are solved successively until convergence criteria are met. A sample network is used for testing the proposed method. The results verify that the MINLP approach can find the global optimum, while NLP algorithms give a suboptimal solution