Manufacturing modeling using RESQ

• Authors: Wayne J. Oates
• Venue: WSC '84 Proceedings of the 16th conference on Winter simulation
• Year: 1984

Queueing networks are often used to model complex systems when the performance of the system is affected by contention for resources. Complex queueing problems occur in areas such as communication networks, computer systems, and manufacturing systems. In manufacturing processes, contention for items such as automated tools, robots, conveyors, material handling resources, and human operators greatly affect the performance of the system. The system performance can be measured in terms of production capacity or throughput, buffer requirements, work-in-process levels, and system reliability. Modeling and simulation techniques can be used to analyze and understand the performance of manufacturing systems. They are often required to understand the effects of uncertain or random events such as tool failures and parts shortages, and to analyze the sensitivity of the system to key parameters such as buffer sizes and tool reliability parameters. This paper presents four examples of manufacturing applications and demonstrates the uses of modeling and simulation techniques to analyze the problems. The examples presented use IBM's Research Queueing Package (RESQ), developed at the IBM Watson Research Center in Yorktown Heights, New York. RESQ is a modeling system which allows the quick construction and solution of queueing models. RESQ provides either an analytic solution or a discrete simulation capability from the same model definition. With certain constraints the model will be solved analytically; however, if the constraints are not feasible, a discrete simulation approach can be used. A major advantage with RESQ is that both solution approaches can be applied to the same model definition. The manufacturing examples considered include: 1. A small sub-system consisting of a single tool with input and output buffers, a conveyor, and a set of transfer units. The model is solved analytically and demonstrates the effect of arrival rates upon the sub-system performance. 2. The first example is expanded to include tool down-time due to random failures. Discrete simulation is used to analyze the sensitivity of system performance to tool failures. 3. The concepts in the above examples are used to construct a model of a serial transfer line. The serial line includes multiple processing points and work-in-progress buffers between each station. The relationship between buffer size and system throughput is developed via simulation. Sensitivity of buffer size and throughput to tool failures is also analyzed. 4. An example of a supply and demand system in which the supply and demand rates follow random distributions. A model is constructed to analyze the relationship between the supply and demand points to determine the resulting buffer profile. The analysis also includes the expected maximum buffer size and the risk of starving the consumption point due to an empty buffer. The intent of this paper is to demonstrate some uses and benefits of modeling and simulation in a manufacturing environment. Although RESQ is used as the modeling language, the paper provides examples of manufacturing modeling applications and is not meant to be a comprehensive RESQ tutorial.