Inventory, Storage and Routing Optimization with Homogenous Fleet in the Secondary Distribution Network Using a Hybrid VRP, Clustering and MIP Approach

Abstract

AbstractThe paper considers the inventory routing and storage problem and suggests a satisfactory solution by finding dispatch quantities and vehicle route allocations when the objective is to minimize transit cost, vehicle cost and storage cost. To be specific, the problem can be categorized as a cyclic inventory routing problem (CIRP) with homogenous fleets. The approach mentioned here is a hybrid of vehicle routing problem (VRP), graph-based clustering (GC) and mixed integer programming model (MIP) to find a solution when the scale is large enough which makes it difficult to solve it using exact methods. The VRP module is used to find the feasible routes of the customers from the depot using metaheuristic approach. The GC module is further used to decompose the route network into clusters using connected graph networks, and eventually, a MIP model is used to select the routes, find the daily dispatch of gases and thus also find the optimal storage required both at the depot and the customers. The MIP formulation is designed in a way to reduce the solving time complexity by converting the binary variables which are used in a traditional formulation of the inventory routing problem to integer variables by decomposing the constraint. The approach has been tested on a simulated business case that spans two hundred customer locations, demands fulfilment for a week and homogenous fleet with a truck carrying capacity of four cylinders. The scaling studies have been done on the GC module by analysing the time complexity and the optimization feasibility with respect to the cluster size. The MIP approach is designed to solve the problem to less than 1% MIP gap considering the number of customer locations.KeywordsInventory optimizationRouting optimizationStorage optimizationSupply chain optimization