In the face of growing competition, the Western manufacturing management is forced into becoming much more frugal in resource management. Just-in-time (JIT) is a set of principles which, by attacking all types of waste in manufacturing, can help management to achieve this competitive prerequisite. One of the main pillars of JIT manufacturing is JIT purchasing practices, which are predicated on selecting and maintaining close relationships with a few, albeit reliable and high-quality, vendors. Most traditional approaches to vendor selection have been based on the notion of maintaining a competitive supplier base to achieve the least invoice cost. However, since there are many other costs, some explicit some implicit, JIT adopts the more encompassing objective of ‘total material costs’, augmenting the invoice cost with such avoidable wastes as large lot sizes, paperwork, inspection, and with losses due to poor quality and delivery. Many authors have proposed that limiting the number of suppliers that a company deals with is a prerequisite in the pursuit of JIT purchasing practices. When a relatively small number of parts are externally procured, single sourcing on a part-by-part basis can be quite effective in achieving the required small vendor base. However, when the number of parts and components externally purchased is large, as in the case of most fabrication/assembly operations, decisions pertaining to which vendors to keep, what to buy from them and which ones to drop are not as straightforward as finding the best single source for each part. The difficulty stems from the need to consolidate varying numbers of parts to the same supplier and to resolve many trade-offs among important, conflicting and often non-quantifiable criteria. This paper proposes a decision support approach to selecting vendors under the conflicting criteria of minimizing the annual material costs, reducing the number of suppliers and maximizing supplier’s delivery and quality performances. The system operates with a database on the purchased items and all potential vendors. Scenarios or problem instances are defined and analyzed with respect to specific quality and delivery performance standards (non-structured elements of the problem) that the vendors must achieve. Several models are used to explore the trade-off between material costs and number of suppliers (structured elements) under a variety of such scenarios. One model finds the set of vendors that minimize the total invoice cost regardless of their numbers which may of course be unacceptably large. A second model is used to find the smallest set of vendors who can supply all materials within the desired minimal quality and delivery parameters. This, in turn, may result in unacceptably high invoice costs. The above two solutions are used as benchmarks and a third model is employed to explore the quantitative trade off-between these extreme solutions. The solution of the first model is trivial. The second and the third models respectively have the structure of the set covering and the K-median problems. These are solved using ‘greedy’ heuristics which have been shown elsewhere to give very good-quality solutions with reasonable computational effort. An example problem illustrates the proposed decision support tool.
