Simulated annealing approach for optimal batching in a warehouse

Abstract

Order-picking in a warehouse is normally the most expensive operation as it takes the longest time. In conjunction with this, order batching is introduced to expedite the orders for the picking process. An optimized batching order is one of the methods to minimize the traveling path, which eventually reduces operation costs in a warehouse. This problem is known as the NP-hard problem (Nondeterministic Polynomial time), typically solved using metaheuristic approaches. One of the metaheuristic approaches is called Simulated Annealing (SA). The main purpose of this study is to propose a SA algorithm for solving this problem to optimize order batching so that it can minimize the total travel distance in a warehouse. Under the precedence of a randomly defined list of orders and a fixed warehouse layout, a simulation was done to investigate the performance of SA in batching solutions. This is done by using different sample sizes and evaluating the SA algorithm's performance to batch orders compared to other existing batching methods. This research work found that SA method performs better than other existing batching methods such as Sort batching and First-Come First-Served (FCFS). From the results obtained, SA found the shortest distance for all different sample sizes compared to other existing batching methods, that is, by reducing the total distance by 64.27% and 15.46%, respectively, for FCFS method and Sort batching method. This proves that SA is efficient in reducing the total distance, and future implications of this method may be applied in advanced warehouse technology that uses drones or other robots.