| This research applies simulation modelling to the study of dispatching rules and lot sizing rules in a material requirements planning (MRP) environment. The study was inspired by an article by H. F. Mather, "Dynamic lot-sizing for MRP: help or hindrance?" (Production and Inventory Management, 1985, vol. 26, no. 2), which contended that fixed order quantities were preferable to discrete lot sizing techniques as discrete lot sizing techniques create excessive nervousness. The shop modelled included 18 end items, 158 part numbers, passing through 15 work stations. The shop was simulated over a 52 week horizon, with slight changes to the master production schedule (MPS) made monthly. Some significant findings were made.; The paper provides empirical substantiation to Mather's contention. A fixed order quantity, EOQ, did in fact perform better than all other single level lot sizing procedures.; The Silver-Meal heuristic was found to perform poorly in a multi-level production/inventory environment with MPS instability. A multiple level version of the Silver-Meal heuristic was developed by the application of a simple cost modification. The cost modification attempts to account for setups of component items which would be necessary given any lot sizing decision at the end item level. This multi-level version of the Silver-Meal heuristic outperformed EOQ.; As an attempt to reduce nervousness among discrete lot sizing techniques, safety stock was applied at the component level. The performance of the discrete lot sizing techniques improved and the performance of the EOQ declined when component safety stock was applied.; The performance of dispatching techniques varied with the choice of lot sizing technique. Earliest Job Due Date (EDD) and EDD with Truncated Shortest Imminent Operation Processing Time applied at bottleneck work centers performed well with each lot sizing technique. |
