Department Seminar: Gemma Berenguer, University of California-Berkeley Monday, Feb 13, 2012 @ 8 AM, IOE 1680
"Supply Chain Design: A Novel Solution Approach and a Humanitarian Setting"
Abstract: The models presented in this talk belong to the family of integrated supply chain design problems. Models in this family simultaneously consider location, shipping, and inventory decisions in the same optimization problem. Usually, these models' goal is to minimize costs when demand is stochastic. To deal with demand uncertainty, organizations can join their inventories in specific storage locations that allow them to better respond to demand peaks from different retailers. This is known as risk pooling strategy. This practice and other complexities of the supply chain design translate into models that are hard to solve. In this talk, I present a new theoretical approach based on conic quadratic mixed-integer programming (CQMIP) to solve popular models of this family and novel versions that were not tractable before. The novel models assume correlated retailer demand, stochastic lead times, and multiple commodities with correlated commodity demand. The numerical experiments presented show computational efficiency that is comparable to that of other solution methods. The experiments also yield interesting managerial insights. The integration of different decisions in the design of a supply chain is pertinent for humanitarian initiatives, which often face especially challenging conditions, such as unreliable operations and limited resources. I will finish this talk by presenting a model that includes these challenges and involves a recurrent issue in humanitarian settings: the service delivery problem. This model is a novel formulation of the family of integrated supply chain design problems. I solve it via CQMIP and illustrate it with examples using data from Rwanda.
Audience-Based Site-Wide Navigation:back to top