Population Based Ant Colony Optimization for Inventory Routing Problem

The inventory routing problem presented in this study is a one-to-many distribution network consisting of a manufacturer that produces multi products to be transported to many geographically dispersed customers. We consider a finite horizon where a fleet of capacitated homogeneous vehicles, housed at a depot/warehouse, transports products from the warehouse to meet the demand specified by the customers in each period. The demand for each product is deterministic and time varying and each customer request a distinct product. The inventory holding cost is product specific and is incurred at the customer sites. The objective is to determine the amount on inventory and to construct a delivery schedule that minimizes both the total transportation and inventory holding cost while ensuring each customer's demand is met over the planning horizon. The problem is formulated as a mixed integer programming problem and is solved using CPLEX to get the lower bound and upper bound (the best integer solution) for each instance considered. We proposed a population based ant colony optimization (ACO) where the ants are subdivided into subpopulations and each subpopulation represents one inventory level to construct the routes. In addition, we modify the standard ACO by including the inventory cost in the global pheromones updating and the selection of inventory updating mechanism is based on the pheromone value. ACO performs better on large instances compared to the upper bound and performs equally well for small and medium instances.