Mathematical models for the deterministic, capacitated, single kanban system.
Mousa George Mitwasi · UA Campus Repository (The University of Arizona) · 1991
The kanban system is the most popular technique for implementing the Just-In-Time philosophy. In this dissertation we develop mathematical models for the deterministic, capacitated, single kanban system. Three different production structures are studied. The models are used to analyze the system, understand the need and behavior of kanbans, and compute good solutions for the number of kanbans to allocate for each part. The first model applies to the single-stage, single-item system. Optimal solutions for the number of kanbans for this system are developed. The second and third models are built for the multi-stage, single-item system and the single-stage, multi-item system respectively. Necessary and sufficient conditions for the feasibility of a set of kanbans are developed for the last two models. The conditions are used to develop heuristic and optimal solution procedures. The heuristic procedures are tested over randomly generated problems and are shown to perform very well compared to the optimal solution procedures.