An optimization method for a multi-day distribution problem with shortage supplies

Abstract

We investigated a multi-day distribution problem while supplies were limited. This scenario can be found in post-natural disasters or economic crises such as floods, earthquakes, palm oil shortage crises, etc. The objective function of this problem was to minimize total travelling distance, unsatisfied cost, and variance of supply delivery proportion. In order to solve this multi-day problem optimally, it requires a large computing memory and takes a long computational time. Therefore, we divided these large problems into multiple daily sub-problems and solved the sub-problems using an exact method. The sub-problems were solved sequentially for which the prior daily subproblem is to be tackled first and the following daily sub-problems were defined based on the prior daily sub-problem solution. Changes were applied to update demands and to adjust delivery priority. There were three delivery priority setups proposed in this thesis. Also, we presented an experiment using three proposed setups to solve a modified Solomon's vehicle routing problem datasets which extended a single period vehicle routing problem with time windows to be seven-day routing problems.