Application of Particle Swarm Optimization (PSO) on Determining Warehouse Locations: A Case Study on Rice Distribution Aunur Rofiq Mulyarto*, Hafizh Nizar Dani Nurrahman, Imam Santoso, Sucipto, Sri Suhartini
Dept. of Agroindustrial Tecnology, Faculty of Agricultural Technology, Universitas Brawijaya, Malang - Indonesia
*aunurrm[at]ub.ac.id
Abstract
The main obstacle faced in the distribution of The Rice for Poor Families Program (RASKIN) is the high distribution cost due to the large number of distribution points and the wide area covered. This distribution cost depends on the distance from the rice warehouse to the predetermined distribution points. Therefore, warehouse location analysis is very important in distribution efficiency. This study aims to determine the reallocation of distribution points and the potential for adding or relocating optimal rice distribution warehouses. The problem is formulated as integer programming. Particle Swarm Optimization (PSO) is used to obtain the optimal solution. To find the best performance of PSO, 9 parameter recommendations are tested. Based on these experiments, PSO^s parameters used are : swarm size 50, dimension 11, number of iterations 1000, acceleration coefficient c1 0.875 and c2 2.625. There are two optimization scenarios compared, namely scenario 1 - without adding a new warehouse and scenario 2 - with the addition of new warehouses. Using the actual data, the results of the analysis show that optimization with PSO (scenario 1 and scenario 2) provides better results than actual conditions for all criteria (total distance, average distance, allocation of distribution point and budget allocation). Scenario 2 produces the total distance of 18,954.426 km which is significantly shorter than in scenario 1 (23,339.597 km). However, in term of implementation, the scenario 1 will be a better option.
