Modified GA and data envelopment analysis for multistage distribution network expansion planning under uncertainty

Summary form only given. An approach is proposed to solve multistage distribution network expansion planning problems considering future uncertainties, guiding the planner from production of expansion plans, evaluation of the plans under various future uncertain scenarios, to the selection of the best strategy. A new balanced genetic algorithm (BGA) is introduced that improves the intensification of the solution search procedure by trading-off diversification ability. This facilitates searching for the optimal solution, but also the efficient production of suboptimal solutions for the planner to take into consideration. The features of the BGA allow a multistage planning problem to be solved more efficiently; the BGA can consider a set of expansion plans in an early planning stage in a single run and produce planning strategies required to solve network problems in a later stage along the planning horizon. The overall performance of each plan under different uncertain scenarios is evaluated using a modified data envelopment analysis to assist decisions on which solution to adopt. The approach is applied to a multistage “greenfield” distribution network expansion problem considering scenarios for the location of future loads. The results clearly show the advantages of the approach over more conventional methods.