| A significant role of intermediate storage in noncontinuous manufacturing systems is to decouple stages of operations. It eliminates delays and interruptions caused by the uncertainty in process operations, and increases the availability of the process. The thrust of this thesis is the comprehensive analysis and design of intermediate storage for the performance of these roles in noncontinuous processes.The storage sizing models are used to derive analytical expressions for estimating the duration of continuous operation in the process for a given storage size, initial inventory and confidence on continuity. These results in conjunction with the storage sizing models are useful for assessing storage needs in noncontinuous processes.The focus of analysis is a 2-stage process with multiple, nonidentical parallel batch/semicontinuous units in each stage and a storage facility between the stages. For this system, a stochastic model is first derived for the inventory in the storage. In contrast to the existing models in the literature, the new model permits simultaneous treatment of arbitrary types of random process disturbances such as operation time variations, unit failures and repairs, batch failures, batch size variations, etc. The analysis of the model using well-known results from the probability theory, yields general analytical expressions for sizing intermediate storage to maintain the continuity of operation in the process with a given degree of confidence. To evaluate the sizing results, they were first applied to the 2-stage system with only process parameter fluctuations, and obtained specific results. For each example of this process, the model performance was compared with the existing models in the literature and Monte-Carlo simulation. The new model predicts smaller storage sizes than these models and compares very well (within 15% overdesign on average) with the simulation results for several examples. The model was also specialized to obtain specific storage sizing results for the 2-stage process undergoing parameter variations, batch failures, and equipment failures and repairs. For several examples of this case, the theoretical size estimates compare closely (15.8% overdesign average) with the simulation results. |
