| dc.description.abstract | This dissertation addresses two topics. The first topic is strategic dynamic supply chain reconfiguration (DSCR) problem, in which the proposed capacity configuration network is employed in the second topic: healthcare facility location and capacity configuration under stochastic demand. The second topic investigates two problems: the stochastic, single healthcare facility location and capacity configuration problem (SSHFCP) in a competitive environment and the stochastic, multiple healthcare facility location and capacity configuration problem (SMHFCP) based on a location-allocation model.
The DSCR problem is to prescribe the location and capacity of each facility, select links used for transportation, and plan material flows through the supply chain, including production, inventory, backorder, and outsourcing levels. The objective is to minimize total cost. The network must be dynamically reconfigured (i.e., by opening facilities, expanding and/or contracting their capacities, and closing facilities) over time to accommodate changing trends in demand and/or costs. This research proposes a network-based model of DSCR and compares it with a traditional mixed integer programming (MIP) formulation via extensive, large-scale computational tests and sensitivity analyses, showing that the network-based model offers superior solvability.
The SSHFCP is to prescribe the location and multi-service, multi-period capacity configuration of facility facing competition from existing facilities under uncertain patient demand, so that the expected excess revenue (i.e., the amount by which revenue exceeds cost) of the new facility is maximized. This dissertation describes a solution methodology that relates practical features relative to healthcare, including a multiplicative competitive interaction (MCI) model to reflect competition among providers and a method to model the stochastic problem as a deterministic resource constrained shortest path problem (RCSPP) on a specially constructed network, which can be solved in pseudo-polynomial time. This dissertation proposes two solution methods to SMHFCP. The dissertation shows that first method, a column-generation heuristic, solves test instances to near optimality; and the second one, an approximation method, provides a fast runtime with a bounding procedure to assess the quality of a solution. The application of SSHFCP and SMHFCP in locating and configuring new primary care centers in mid-Texas rural area validates the real business decision of industrial collaborators. | en |
