In this paper, a new approach for short-term multiperiod planning of nonlinear systems is proposed. To find a more exact solution for the multiperiod planning problem of utility systems within an allowable computation time, a three-step approach has been introduced. In the first step, alternatives for optimum configuration in each period are generated, and a nonlinear programming problem is solved for generated configurations. In the second step, the optimal configuration sequence that minimizes the sum of the operating cost and the switching cost is determined using dynamic programming. In the third step, a fine search for tile optimum is performed using an iterative search to consider transition cost. With the decomposition of the original MINLP (mixed integer nonlinear programming) problem into NLP (nonlinear programming) subproblems and a dynamic programming problem, a more reliable and accurate solution that considers nonlinear characteristics is obtained and the computation time is greatly reduced. The case study has shown that the proposed approach shows good performance in finding the optimum solutions considering changeover costs between periods, and this approach can be applied to ether various MINLP-type problems by adopting appropriate heuristics.