Abstract

Due to the complexity of the supply chain, sourcing and distribution activities within the supply chain require a fair deal of orchestrating in order to eliminate delays and other inefficiencies. For this reason, researchers have worked effortlessly to incorporate a wide range of parameters in the modeling of the supply chain. The parameters integrated have touched many important issues. As important, issues pertaining to quality are of great importance in organizations. Some literature has discussed quality from the perspective of the supply chain and acknowledged the lack of a consistent vision pertaining to quality throughout the supply chain. With many industries today on the quest of improving their quality systems, finding ways to reduce nonconformities and failure of products is crucial. In industries such as the aerospace industry, the variable production cost is considerably high; hence producing extra parts to compensate for defectives would be a costly option. While Cost of Quality (COQ) is a very good indicator of how much poor quality is costing a company, the literature lacks a work that aims at integrating COQ into Supply Chain Network Design (SCND). This thesis aims at exploring the challenges in doing so and introduces a comprehensive supply chain model that minimizes a series of costs, in which COQ is integrated. The inclusion of COQ is done through the integration of quadratic quality function. The overall supply chain is mathematically modeled producing a nonlinearity in the objective function and in the constraints. Hence this thesis solves a constrained binary nonlinear programming problem. Further, this work integrates binary entities, to allow for assignable/set-up costs, into the model and introduces seven solution procedures to solve the model. A real life supply chain network is used to extract relevant results. The real life supply chain is in the domain of the aerospace industry and has an n level Bill of Material (BOM). Heuristics have been introduced to solve Binary Quadratic Programming (BQP) problems before. A majority of these heuristics are geared towards unconstrained problems where feasibility might not be a concern. Alternatively in the COQ model, constraints bind the objective function making feasibility a criterion for optimality. Therefore, the seven solution procedures entertain a feasibility check mechanism and one of the seven solution procedures is a hybrid solution procedure formulated to tailor for the special topography of the feasible solution region of the problem