Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Most manufacturing organizations may be viewed as multi-stage production-inventory systems in which production of goods proceeds from the acquisition of raw material to the fabrication of final product through a series of production stages. Material requirements planning (MRP) has been introduced as an inventory planning technique for such systems. Different lot sizing models have been used to improve the economic efficiency of MRP systems. The objective of this research was to examine the impact of lot size strategies, degree of demand variability, and degree of component commonality on the performance of MRP systems in a stochastic demand environment. A computer simulation model of a hypothetical multi-stage, multi-product, production-inventory system was used as a vehicle for this study. Three different lot size models (EOQ, WW, LTC) were used at the end-item level and four (EOQ, POQ, LFL, TLC) at the intermediate levels producing twelve lot size strategies. Two levels of component-commonality, a measure of the degree of commonality of component parts among all parent-items, were introduced. Fifteen finished goods were incorporated into the model for each product structure set. Three levels of demand variability were introduced with coefficient of variation ranging from .2 to 1.4. System performance was evaluated according to several performance measures. Some of the findings of the study are as follows. First, the system performance was affected by the choice of lot size strategy with respect to all performance measures employed. Second, the system performance was affected by demand uncertainty according to seven of the nine performance measures. Third, both demand variability and component commonality had significant effect on the performance of the system. Fourth, lot size strategy was affected by demand uncertainty only according to carrying cost, setup cost, and number of setups criteria. Determining a single best performing lot size strategy for each of the twelve factor combinations of this research was not possible. However, it was shown that for any of the environmental conditions (factor combinations) there exists a class of best performing strategies from anong which MRP users will be able to select one which they prefer.