An experimental investigation: uncertainty in MRP systems
Material Requirements Planning (MRP) has evolved as a technique for the planning and controlling of inventories and production of complex manufactured products. The problem addressed in this research is that of uncertainty in MRP systems. Demand uncertainties are those which involve variations in the gross requirements for a component. Supply uncertainties deal with variations in the scheduled receipts for a component.
The purpose of this research is to examine the impact of specific operating policies on the performance of an MRP system under conditions of supply/timing uncertainty. Supply uncertainties include either the timing or quantity type. Supply uncertainty that results because of timing involves the receipt of an order after its scheduled delivery date. The specific supply/timing uncertainty examined in this study is that which is caused by variability in the lead time of purchased parts. Experiments are conducted in order to assess the impact of lead time variability, the amount of safety stock buffering, the amount of safety lead time buffering, and the lot-size rule on the average total cost of an MRP system. In addition, the question of whether the results obtained are sensitive to changes in the system's cost parameters is examined. Based on the results, guidelines are developed for practitioners to use when making decisions involving uncertainty in MRP systems.
The simulation model used in this research is a versatile MRP/Production simulator designed to provide a framework for the investigation of a wide variety of MRP related problems. The hypothetical manufacturing system is simulated and its average total cost is recorded for varying levels of lead time variability, safety stock, safety lead time, lot-size rule and the holding cost and lateness penalty values. A 3x5x3x3 balanced factorial experiment is performed and multifactor analysis of variance is used to assess significant differences in the average total cost of the MRP system. On those means judged significantly different by Duncan's multiple range test, further analysis is provided in the form of general linear contrasts and confidence intervals.