PERANCANGAN SISTEM DISTRIBUSI PUPUK MENGGUNAKAN NEURAL NETWORK DAN MIXED LINEAR PROGRAMMING

Abstract

Business environment is pushing a fertilizer distribution company, which is currently concentrating in Java, to expand their market to another area in Indonesia. Kalimantan is one of the target areas, and several provinces in Kalimantan are then chosen to be potential areas to be reached.

To accommodate this need, it is necessary that an extensive distribution networks, involving ports, depots, warehouses, ships, and various land transportations, which is capable to fulfill the demand of each region with minimum cost, be considered. Optimization technique using Mixed Integer Programming is applied. The purpose of this optimization is to minimize transportation cost as well as other costs for building depots and warehouses, with a due consideration of technological and economical constraints.

One of the most important data required for this optimization is the prediction of demand for each region. Neural Network is used for this purpose. Using some parameters identified as factors affecting fertilizer consumption in several areas in Java, demand forecasting of fertilizer for some areas in Kalimantan could be well predicted.

By applying demand prediction excerpted from the output of Neural Network and optimization using Mixed Integer Programming, an optimal distribution system in Kalimantan could be procured. The result of the optimization will specify which ports, depots and warehouses, that should be built, as well as ships and land transportations that should be used, which accordingly will guarantee an optimal distribution system.