Improving performance in flexible manufacturing systems

Abstract

The need for flexible manufacturing systems (FMSs) has become of utmost importance over the last years due to the fierce competition present in the manufacturing industry as well as the ever decreasing product life spans and the markets attempts to respond to customers’ changing needs. These systems allow us to react quickly to changes, however this flexibility costs both time and money. Given this fact, it would be desirable to reduce costs by testing potential implementations before using them. We will use a timed process algebra called BTC (for bounded true concurrency) to face up to this challenge. We have developed this algebra by extending CSP in order to consider the duration of actions and the context (resources) in which processes are executed. This new algebra is able to take into account that the resources in a system must be shared by all the processes. So, we can find two kinds of delays in the execution of a process: delays related to the synchronization of processes, and delays related to the allocation of resources. Once FMSs are specified by means of BTC, we will focus our attention on the performance of the system which will be variable depending on the number of available resources. For this, we have developed a performance algorithm and a tool which implements it. Three examples are used to show how an FMS can be specified and how the algorithm and the tool work, allowing us to obtain the best performance with the lowest cost.