Reliability improvement of distribution system through distribution system planning: MILP vs. GA

Abstract

Distribution system planning (DSP) is very important because it can result in reliability enhancement and large cost savings for both utilities and consumers. DSP is a complex nonlinear problem, which can be solved with different optimization methods. This paper compares two such optimization methods, conventional (mixed-integer linear programming - MILP) and meta-heuristic (genetic algorithm - GA), applied to the DSP problem: construction of feeders in distribution power system from scratch. The main objective of DSP is to minimize the total cost, where both the investment and operational outage costs are considered, while the reliability of the whole system is maximized. DSP problem is applied to an actual distribution system. Solution methods are outlined, and computational results show that even though GA gives reasonably good results in faster computation time, MILP provides a better optimal solution with simpler implementation.