A New Cost-Optimal and Fault-Tolerant Instrumentation Sensor Network Design Methodology for Chemical Plants

Abstract

In this paper, an optimal redundant instrumentation sensor network design methodology is presented for complex chemical process plants using a combinatorial particle swarm optimization search (CPSO) engine. The approach, which is more flexible and general in comparison with previous works, aims to minimize cost as a main design factor, similar to the usual trend in the literature. Besides, it caters for fault-tolerance issue as a crucially important feature in the design procedure which has not been addressed sufficiently in the reported research works. For this purpose, weak redundancy degree (WRD) and sensor network reliability (R) are incorporated in the proposed design scheme as three evaluating measures. This enables the designer to maintain a desired fault-tolerant redundancy in the proposed sensor network to cope with a possible set of sensor failures. Thus, the developed CPSO engine searches in a diverse variety of sensor networks to adopt the most fitted one based on the imposed fault-tolerant design constraints. This facilitates the network realization of the fault-tolerance as the most attractive feature which is practically very demanding. Implementation of the proposed design methodology is illustrated in a simulated continuous stirred tank reactor (CSTR) as a benchmark process plant used in a large-scale design to show its effective capabilities.