| Effective supply chain management is paramount to organizational success for public and private firms alike. This dissertation focuses on the location of facilities within the supply chain and the allocation of distribution resources for the organization. The research first develops a methodology for designing and/or improving an existing network within which goods and services must be delivered. The methodology is generalizable because it emphasizes the location of facilities and the assignment of delivery vehicles to opened facilities. The optimization approach is then complemented with an extensive evaluation of 5 heuristic algorithms. These algorithms present different rules for solving the facility location-vehicle assignment problem stage. For this problem, the specific objective is to minimize the total fixed and variable cost of opening, closing and combining facilities, and assigning a sufficient number of delivery vehicles to each opened facility within the service area. The approach implicitly considers customer location and demand at this decision point.; Each algorithm deploys a Simulated Annealing strategy for facility location and vehicle assignment. Three algorithms are adapted from the literature and two are modifications. Also, each implements a probabilistic switching strategy in which a subset of the possible facilities is opened and a sufficient number of vehicles are assigned to that facility in order to meet all customer demand. An assignment of vehicles to open facilities constitutes a configuration. At each iteration, a configuration is perturbed by re-assigning a vehicle. Then the algorithm solves a Linear programming problem for the customer-to-facility assignment. This ensures that only feasible assignment solutions are returned to the main heuristic.; Within all five heuristics, five cooling schedules are implemented. Algorithm performance is examined across five dimensions. There are 2400 problem instances analyzed. All algorithms are capable of reliably solving the facility-location vehicle assignment problem with up to 25 depots, 100 vehicles, and 80 customers. Further analysis is conducted to examine the variance within and across the heuristics under changing parameters. |
