Greenfield distribution network expansion strategy with hierarchical GA and MCDEA under uncertainty

Distribution network expansion planning (DNEP) is becoming more complex in nature. Addition of new load centers, due to increasing conversion of greenfield areas into habitats, have generated need of more intense and highly structured planning strategies. Micro level work on distribution expansion planning has been ignored by most of the researchers mainly in Indian scenario. Since practical distribution networks are quite large, number of candidates (load centers) will be more and, hence, number of variables (electrical parameters and new load center feasible connections with the existing system) are remarkable. Optimizing a large system may result in significant decrease of accuracy and increase of computation time. For deciphering this issue, segmentation procedure has been applicable. For this purpose, a sensitivity analysis has been applied to find dependent variables. It is obvious that a correct segmentation can decrease computation time (as a single task is operated in segments simultaneously) while accuracy decreases negligibly. In present work, a scheme has been introduced to connect three greenfield load centers with existing primary distribution system by using hierarchical genetic algorithm (HGA). HGA is an integrated approach of analytical hierarchical process and genetic algorithm. The paper reports best selection of investment with finest voltage profile and least losses while maintaining radiality of the system. DNEP has been done at micro level and proposed methodology has been tested on a small dimension practical distribution system. The novelty of this paper is to optimize the best possible selection of connection of new load centers with existing system with the help of AHP and GA and results have been verified with advance optimal tool multiple criteria data envelopment analysis (MCDEA). HGA and MCDEA are applied to practical nine bus distribution system and the results are presented and compared. (C) 2016 Elsevier Ltd. All rights reserved.