Modeling of Faults in Chemical Reactor with Continuous Agitation (CSTR)

Abstract

Fault detection is an important problem in process engineering. The early detection and diagnosis of process faults while the plant is operating in controlled areas can help avoid abnormal event branches and reduce productivity losses. Today there are a lot of literature on the detection and diagnosis of faults, from analytical methods to static methods approach. The methods of detection and diagnosis of faults can be classified into three categories (Venkatasubramanian et al., 2002), quantitative model-based methods, qualitative model-based methods and methods based on historical data. Each of these methods is subdivided in different detection and diagnosis of fault models. Model of chemical reactor with continuous stirring is selected, which is one of the most used in the chemical industry, to model and simulate failures in the application of fault detection models. Some methods that are studied in this work are the Kalman filter, Fault Trees and Wavelet transformed. It was concluded that no method is, individually, a complete study of fault detection and diagnosis. On the contrary, each of them can be complement each other, enabling the creation of a hybrid method.