Modeling of Failures in Continuous Stirring Chemical Reactor (CSTR)

Abstract

The fault detection is a relevance problem in engineering processes. The early fault detection and diagnosis in processes while the plant is operating in controlled areas can help avoid the ramifications of abnormal events and reduces productivity losses. Today there are a lot of literature about fault detections and diagnosis, from analytical methods to static approach methods. The fault detection and diagnosis methods can be classified in three categories: methods based on quantitative models, qualitative models and historical data. Each of these methods can be subdivided in different models of fault detection and diagnosis. The model of Continuous Stirring Chemical Reactor, one of the most used in chemical industry, is selected to model and simulate process failures of applying the methods. Some method were studied in this work are Kalman filters, fault trees and Wavelet Transform. It was concluded that no methods are, individually, a complete study of fault detection and diagnosis. Rather, each can complement each other, enabling of creation of a hybrid method