Optimizing the Warehouse Location and Distributor Allocation: A Case Study of the LPG Distribution in Sri Lanka

Abstract

In this paper, we deal with a real world warehouse location and distributor allocation problem of Liquefied Petroleum Gas (LPG) distribution in Sri Lanka. The existing supply chain has a single storage plant and 33 different distributors throughout the country. With this system, approximately 65 km (6.49466+004 m) should be travelled to satisfy a unit demand. We choose to proceed with the P-median model, which locates “p” facilities among “n” demand points and allocates each demand point to one of the facilities by assuming every demand point can be elected as a median. Then the problem was solved by computational Myopic algorithm and the computational Lagrangian algorithm. As the first median, both Myopic and Lagrangian algorithms selected the same distributor node “no. 16 ” as the warehouse with the average distance of 5.2926e+004m to satisfy a unit demand. In the case of selecting the two medians, while the Myopic algorithm proposed the node “no.16” and node “no.18” as best locations with 3.79180+00401 average travelling distance, the Lagrangian algorithm selected node “no.18” and node “no.06” as best locations with 3.66730+00401 average distance. In later case, optimum demand point allocation could be done by assigning nodes 1,2,3,4,7,14,18,19,20,21,23 and 28 to the warehouse which will be located at node “no.18” and nodes 5,6,8,9,10,11,12,13,15,16,17,22,24,25,26,27, 29,30,31,32 and 33 to the warehouse which will be located at node “no.16” .The resulted computerized user interface provides drop down menu to select the number of warehouses to be located and then outputs the best nodes to be elected as warehouses and displays the best possible demand point allocation method.