Application of extended Kalman filtering to chemical reactor fault detection

In this paper, we present a method for detecting faults that can appear in some parts of a chemical plant. This method is based on statistical information generated by the Extended Kalman Filter (EKF) and is designed to reveal any drift from the normal behavior of the process. Although this method was originally developed for linear systems, our contribution consists in extending it to unsteady state nonlinear systems such as stirred reactors in the presence of exothermic chemical reactions. We examined the abnormal behavior of a chemical reactor due to three different faults in its control parameters. Each fault is caused by an abrupt variation in the value of one of the following parameters: the cooling temperature, the stirring rate of the reaction mass, and the feeding flow rate of the reactant. The objective of this experimental study is to detect the presence of such faults on-line and to pinpoint the moment each occurs, as well as to determine the delay in detection with respect to the decision threshold.