Simheuristics for Designing Resilient Supply Chains under Uncertainty Scenarios

Supply Chain Network Design (SCND) is a topic broadly studied since several decades ago. A SCND problem "comprises the decisions regarding the number and location of production facilities, the amount of capacity at each facility, the assignment of each market region to one or more locations, and supplier selection for sub-assemblies, components and materials" (Carvalho et al., 2012). Many works addressing this topic consider that design parameters, such as demand, transportation times or costs are deterministic, i.e., their values are perfectly known in advance. However, in real-world cases design parameters are not known exactly, i.e., they are uncertain. Addressing uncertainty is one of the more challenging responsibilities in SCND. Anticipating the future is crucial on planning and design processes, although future is intrinsically uncertain.In recent years, a trend in the literature has been the consideration of resilience for designing supply chains in order to face uncertainty. Supply Chain Resilience (SCRES) can be defined as the “ability of a supply chain to return to its original state or move to a new, more desirable state after being disturbed” (Mohammed et al., 2018). Since SCRES and Resilient Supply Chain Network Design (RSCND) are a recent tendency, it is not still usual to find in the literature quantitative approaches, therefore, most works are still conceptual. Given the growth in computational power, the use of hybrid simulation-optimization (Sim-Opt) methods has increased in recent years, mainly because of its suitability to address uncertainty. Nevertheless, in the more specific topic of SCND, applications of hybrid Sim-Opt methods are still scarce and, to the best of our knowledge, it is almost null in SCRES.Besides, SCND problems can easily become NP-hard, i.e., solving this kind of problems may increase dramatically the computational time and approximate methods or algorithms such as heuristics or metaheuristics are required to deal with real and large instances. However, both heuristics and metaheuristics methods are barely used in RSCND. If simulation is combined with metaheuristics, a novel approach to tackle problems related to, but not limited to logistics and supply chain management is called Simheuristics (Juan et al., 2015). These authors state that it allows to model uncertainty in a natural way, which will be very useful in this PhD thesis to achieve the proposed objectives. Therefore, this PhD thesis will contribute to the field by exploring the use of simheuristics in RSCND, which will be useful as a tool for researchers and decision makers who want to obtain supply chain configurations as resilient as possible through considering uncertainty scenarios.