Integer Linear Programming per Ottimizzare le Logiche di Inizio degli Ordini in un Magazzino Automatizzato

Abstract This thesis explores the use of Integer Linear Programming (ILP) to design the order start logic in an actual warehouse. The project presents two algorithms that effectively initiate real outbound orders by maximizing the utilization of different areas within the warehouse while adhering to each order's specific constraint. In order to analyze the effectiveness of this optimization method, the author conducted a case study based on his professional experience in designing a highly automated warehouse for one of the biggest footwear and apparel companies from the sports sector (called “Client” in the following). Knowing the layout of the structure under consideration (called “Warehouse” in the following) and making the necessary simplifications, a model of the Warehouse is identified to simulate the behavior of the real system. Then, having a high-level knowledge of the logic used to start the orders in the Warehouse, a more methodical approach is followed to obtain a similar behavior, with the final aim of potentially improving it. The first algorithm implemented (called “Algo 1” in the following) aims to replicate the logic in place in the Warehouse using the ILP, while the second algorithm (called “Algo 2” in the following) aims to improve the existing logic. In both cases, critical case studies are examined to gain a deeper understanding of the proposed algorithms and to validate the effectiveness of the optimization method. Comparing the results obtained from the two algorithms, it is possible to evaluate the efficiency enhancement achieved through the implementation of Algo 2. The insights gained from these critical case studies have proven to be invaluable in this evaluation process. Eventually, in order to enhance the practicality and relevance of the study for potential future implementation, a volume test is conducted to investigate how the proposed solutions would perform when subjected to realistic order quantity as a typical day of production in the Warehouse. Although the study made certain assumptions and simplifications, the conclusions drawn from this project confirm that Algo 1 is a reliable representation of the real system, as the results obtained are in line with the expectations from the real system, while Algo 2 has the potential to improve order management efficiency within the storage facility under consideration.