Solving methods for the Multi-item Inventory Routing Problem with Pickup and Delivery

As industry 4.0, IoT and clouding technologies take hold, real-time scheduling and resource allocation solutions by means of proven optimization techniques have a concrete interest and remarkable market opportunities. Concerning logistic and production sectors, some solid well-designed optimization system can be a precious ally to Just-In-Time philosophy at work. Speaking of logistics, it is a fast-growing sector responsible for a significant share of company costs, and its optimization has a major impact on economic rentability and competitiveness. The topic of this thesis is the research of dynamic and flexible scheduling methods for a generic indoor logistic support system. More specifically, the sought methods concern the optimal satisfaction of the expressed material needs of an indoor manufacturing facility, by means of a limited-capacity vehicle fleet. Part I is devoted to bibliographic research and the introduction of some optimization key-concepts. Part II introduces and discusses in detail the problem, its mathematical formulations, and the methods adopted to solve them. A series of MILP formulations are proposed, each with a specific set of hypotheses, the associated mathematical model, and some considerations about applicability and solvability. Then, an ad-hoc heuristic was designed to compare and evaluate the performance of the Discrete-Time 1-mainstock MILP (DT-1ms-MILP) formulation. The thesis ends with a joint evaluation of tested methods, with a special appreciation for DT-1ms-MILP by reason of its performances and flexibility. Last, some directions are given for a further development of this thesis work.