Multi-item picking methodology in warehouses

Abstract

Order picking is the retrieval of items from warehouses to satisfy in-house orders. Recent research shows that order picking counts for more than half of the warehouse operational costs. In this study a storage assignment clustering methodology and computer implementation is developed to reduce the walking required for an order picking operation. The proposed methodology does not consider routing policy, and does not alter the material handling and storage equipment. The basic principle behind the proposed method is known: items frequently ordered together are clustered in the warehouse. The difference from earlier solutions is in that it avoids full enumeration of the occurrences; a potentially time consuming task. The methodology proposed is based on the transitivity property of implication, allowing for quick computerized processing. The implementation part is potentially the most crucial aspect, as the number of possible clustering combinations is high for construction production systems with thousands of items in the warehouse. Computerised implementation of the population-based enumeration of the occurrences of possible order-combinations is resource-demanding according to computer processing time. Given the waste elimination through reduced order retrieval time, the method is a potentially valuable tool in a company's 'lean toolbox'. The method is primarily suitable for large construction production systems (for example vessel construction) with variable picking orders, and multiple warehouse distribution systems. That said, the method is equally useful for single-warehouse systems. The developed demonstrator is expected to be implemented and tested at a leading Norwegian offshore shipbuilder.