Date of Award


Degree Type


Degree Name



Industrial and Manufacturing Systems Engineering

First Advisor

Caron, Richard

Second Advisor

Selvarajah, Esaignani


Capacitated facility location, Mathematical Modeling, Optimization, Supply chain network design




This dissertation is concerned with mathematical modeling and optimization of three-echelon capacitated supply chain network design (SCND) with suppliers, distribution centers (DCs) and customer zones. An introduction to SCND is provided followed by a literature review. Being inspired by a real world SCND problem, a Mixed Integer Linear Program (MILP) is developed for a three-echelon SCND. The model takes into account the operational costs of a built DC based on its actual activity level rather than the assumption that a built DC operates at the maximum capacity. The suppliers and customer zones are at fixed known locations and the DCs are picked from a set of potential DC locations. Then, a well-known model on three-echelon multi-capacitated SCND [1] is studied where both suppliers and DCs and their corresponding capacity levels are picked from a set of potential supplier and DC locations and a set of predetermined discrete capacity levels, respectively. We characterize the complete set of alternate optimal solutions in [1]. We extend the model through the addition of a constraint set that eliminates certain undesirable optimal solutions and we show that the extended model requires, essentially, the same computational effort as the original. We then deploy a new set of variables and present a new formulation for three-echelon multi-capacitated SCND. We show that the new formulation is more efficient as it offers lower computational times. We then present two approaches which allow an exponential increase in available capacity levels to facilities. We demonstrate the merits of the exponentially increased flexibility. Inspired by the merits of increased flexibility in capacity assignment, we present a technology-based, variable-capacitated, supply chain design model that is unique in that it allows complete variability in the choice of capacity level. This avoids the need to determine, a priori, a set of potential capacity levels. Another merit of the model is that the built facilities do not have unused capacity. Last, the conclusions and future work are provided.