Supplier Selection under Product Breakdown Structure Constraints
Open Access
- Author:
- Lawit, Aitan Salamon
- Area of Honors:
- Industrial Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Dr. Soundar Rajan Tirupatikumara, Thesis Supervisor
Dr. Paul M Griffin, Thesis Honors Advisor - Keywords:
- Supplier Selection
Product Breakdown Structure
Non-preemptive Goal Programming - Abstract:
- The defense industry, like many manufacturing industries, suffers from lengthy product lifecycles due to inflexible manufacturing facilities and high design variability. Each time a new product is proposed or a product is changed, the firm completes a design-build-test-redesign iteration, significantly adding to the product lifecycle of the final product. With the goal of developing a “flexible, programmable, potentially distributed production capability capable of accommodating a wide range of systems and system variants with extremely rapid reconfiguration timescales,” DARPA has proposed a foundry-style model of manufacturing, which minimizes design-build-test-redesign iterations (DARPA). In such a model of manufacturing, a final product design is decomposed into a set of subsidiary products, which may represent components, subassemblies and processes required to manufacture the final product. As the final products often involve multiple components, subassemblies and processes, a product breakdown structure provides an effective means to organize the subsidiary products. The computation of certain performance criteria (e.g. makespan) may be influenced by the product breakdown structure, making it an important consideration. Supplier selection involves choosing a supplier, which in principle, represents a combination of resources (e.g. machines, personnel, tools, parts, etc.) to source each product. A foundry represents the full assignment of a supplier to each subsidiary product of the final product design. In essence, the foundry generation problem is a supplier selection problem under product breakdown structure constraints. No research has considered the supplier selection problem under product breakdown structure constraints. The primary objective of this thesis is to formulate the supplier selection problem under product breakdown structure constraints and to contribute a solution procedure to solve the problem. The solution procedure proposed integrates all steps involved in the supplier selection problem and is capable of identifying a solution in the solution space close to the performance criteria goals set by the decision maker. The main mechanism of the solution procedure is a non-preemptive goal program, which facilitates supplier selection under product breakdown structure constraints by minimizing the summation of weighted deviations from performance criteria goals set by the decision maker. To illustrate the solution procedure and non-preemptive goal program, a case study for sourcing the components, subassemblies and processes required to manufacture a ground vehicle is presented. The case study demonstrates the non-preemptive goal program’s effectiveness at returning solutions near multiple (criteria) goals set by the decision maker. The solve time of the non-preemptive goal program was benchmarked against a random generation technique, which rapidly returns feasible solutions by randomly contracting suppliers to produce the products. A full factorial experiment with five replications was performed with performance criteria as the factors and solve time as the response for both the non-preemptive goal program and a random generation technique. The results of the full factorial were analyzed using three techniques: comparison of means, comparison of variance and analysis of variance. Several key findings arise from employing these techniques. First, the non-preemptive goal program has a significantly lower mean solve time than the random generation technique for all trials with multiple criteria goals. Second, the non-preemptive goal program is more robust (i.e. contains less solve time variability) than the random generation technique at solving the supplier selection problem with multiple criteria goals. Third, all performance criteria, with the exception of PPC, PPC/M, M/NR, IC/PPC/NR and IC/PPC/M/NR, appear to be significant predictors of the non-preemptive goal program’s solve time. A regression model was presented, which can be used by the decision maker to approximate the solve time of the non-preemptive goal program. Finally, the random generation technique can return solutions that meet specified single criterion goals faster than the non-preemptive goal program, but is significantly slower when dealing with multiple criteria goals. The two techniques complement each other as the random generation technique is adept at rapidly returning feasible solutions while the non-preemptive goal program is adept at returning a solution nearest multiple criteria goals set by the decision maker (i.e. a solution with the minimum summation of weighted deviations from criteria goals set by the decision maker). Together, the techniques provide the decision maker with an effective approach to solving the supplier selection problem under product breakdown structure constraints. This thesis, for the first time in literature, shows that firms that effectively make supplier selection decisions while considering product breakdown structure constraints will have a better chance at surviving in the competitive, global marketplace.