Optimising the Storage Location Assignment Problem Under Dynamic Conditions

The assignment of products to storage locations has a major impact on the performance of a warehouse, especially if the warehouse is not automated, but serviced by human pickers. Although the static storage location assignment problem has been studied for more than fifty years, the interrelations with up- and downstream processes and the effects of dynamic fluctuations in demand are still not well under- stood. In this thesis, we model and optimise dynamic and integrated storage assignment problems based on real-world data from the auto- motive and steel industry. Order picking is the main bottleneck in both scenarios, therefore the quality of a warehouse assignment is evaluated via picker travel distance required to supply products to downstream processes. Affinity based slotting strategies place products that are frequently ordered together closer to each other. Part i of this thesis focuses on the formalisation of the novel Pick Frequency / Part Affinity score, which combines popularity and affinity measures. The new score is coupled with various metaheuristics and compared to standard assignment strategies in two empirical case studies. The downside of the algorithms studied in the first part of this work is that implementing the generated assignments in a fully oper- ational warehouse requires extensive movements of products. Part ii therefore focuses on the development of a generic multi-period model of the storage location assignment problem. By considering storage, re-location, and picking efforts, the costs and benefits of extensive re- locations versus iteratively moving a small number of products per period are analysed. Greedily selecting re-locations has a couple of disadvantages, which were mitigated by switching to a "robust" selec- tion strategy.