Bacterial Memetic Algorithms for Order Picking Routing Problem with Loading Constraints

Abstract

Order picking is the most labour and capital intensive warehousing operation whose primary development field is routing optimisation due to its time consuming nature. The Order Picking Routing Problem is a special case of the vehicle routing problem with loading constraints, when the operator visits picking positions and collects items to build transport unit.Where the stacking and stability challenges are relevant during the picking of ordered items and exact routing algorithms are not available, the order picking operators have huge challenges to sequence the order picking list. They should take into consideration several factors by themselves, such as product properties, order picking list characteristics, and order picking system properties. The goal of the proposed research is to support the order picking operators in order to make more objective decisions in decreasing the order picking lead time, building stable transport units, and avoiding product damages, when industrially relevant, but rarely discussed, order picking sequencing based on stacking property is necessary.The paper defines the Order Picking Routing Problem based on Pallet Loading Feature (OPRP-PLF) and presents Bacterial Memetic Algorithm (BMA) based solutions for it, which is compared to Simulated Annealing (SA) algorithms. BMA has already been applied for Travelling Salesman Problem (TSP) but never used for the defined OPRP-PLF. The paper describes several BMA operators, most of them have an alternative which can be completed with SA based decisions. Using the BMA operators with SA methodology is a novelty of the proposed algorithms, which might support a quicker approximation to the global optimum. The possible combination of BMA operators will be evaluated with shorter and longer order picking lists and compared to SA algorithms on the same basis.The simulation results highlight, that allowing unit load reconstruction could decrease the order picking lead time and the developed BMA algorithms are more effective for OPRP-PLF than the SA algorithms. The paper concludes that the SA combined BMA operators are more effective than the SA-less operators in the case of shorter (less than about 20 records) order picking lists. While the shorter lists are the most commonly occurring order picking lists of warehouses, the SA combined BMA operators can increase the effectiveness of the OPRP-PLF optimisation.