Storage Location Optimization for a real-world stock warehouse: the non-food sector

Globally, online shopping has gained traction; a natural consequence of the constant advances on digital platforms, both in terms of increasing speed and reducing costs. The coronavirus pandemic predictably boosted online market for all retail sectors; bringing to the market older and less adventurous consumers that have started to embrace the security and convenience that technology offers (Nielsen, 2020). The scenarios created by coronavirus are important indicators for retailers and producers, who need now to answer efficiently to the higher number of online orders. Online orders are usually smaller and customized (Nielsen, 2020), bringing additional operational challenges. Thus, it is important to present efficient order-picking (OP) operations. There are several ways of improving OP, namely: improving the picking route, the zoning policy, the batching strategy and /or the storage policy of the products (De Santis et al., 2018). This doctoral project is inspired by a real-word Portuguese food retailer company, in Northeast Portugal and it aims to improve the OP process by improving the storage policy of the products. This subject is called, in the literature, the storage location assignment problem (SLAP). In a study performed by Reyes et al. (2019), it is possible to see that the number of the publications, on this subject is still increasing, proving the actuality and pertinence of the topic. This thesis is composed of four essays that deal with SLAP when there are real-world constraints. In the first paper, we developed a two-phase heuristic procedure of first clustering and then weigh-ordering, considering the precedence constraint of picking heavy products before light ones. In the second paper, we developed a weight parameter, to measure the similarity of two products, in terms of weight and, we included it in a well-known assignment model, designed by Liu (2004), to set the products with similar weight together (considering weight constraints). In the third paper, we developed the model presented in the second paper to integrate shape constraints. For this purpose, we designed a parameter to measure the similarity of two products in terms of shape (using the volume as a proxy). In the fourth paper, we put forward a new clustering similarity index that, besides the similarity of two products in terms of demand, considered the similarities in terms of weight and shape, in the grouping process and, consequently, in the allocation procedure. The third and fourth paper considering both weight and shape constraints. In the four papers, the solutions were designed for warehouses with a high number of non-uniform fast-moving products (in terms of weight and/or shape). The results were compared to the results achieved by applying the usual strategies presented in the literature and, we proved that the designed methods worked better than the traditional ones. We applied more than seventeen experimental designs to test possible variations within the solutions and we showed that, by applying our method, the savings, in terms of the distance travelled by the picker, can go up to 40%. A percentage of up to 30% higher than the one achieved with the traditional methods. The main contribution of this thesis is the development of four alternative solutions to deal with SLAP when there are real constraints, a subject that was defined by van Gils et al. (2018) as a gap in the literature and as a subject of interest for future research. SLAP was proved to be a non-deterministic polynomial-time hard problem (Frazele & Sharp, 1989). Thus, finding exact methods or heuristic procedures to deal with it, is relevant. The solutions that we present, in this thesis, may be used to improve the performance of many other warehouses. Since they allow managers to have some sort of flexibility in the products allocation procedure, within the aisle and, they were thought to be easy to understand and to implement in practice.