Fault Detection in the Green Chemical Process: Application to an Exothermic Reaction

The chemical industry’s activities are often controversial due to the high risks that they represent. Over the past decades, serious industrial events affecting lives, facilities and environment have heightened society's awareness of the negative effects of technology. Among the most hateful events in the chemical industries are the phenomena of thermal runaway that often result in operator errors. Predicting and controlling them is essential to the processes design and safe operation. The objective of this work is to develop a method for early detection of malfunctions in a chemical reactor based on a reference model, before resulting in a critical event. For this, the reaction of perhydrolysis of formic acid to peroxyformic acid by hydrogen peroxide is used as a test case to simulate the reaction in abnormal operating mode. This exothermic reaction is composed of several secondary decomposition steps. The kinetic model of the reaction was determined in order to simulate the reaction in an abnormal mode with defects related to operator error. The detection method has been validated by simulation data. A performance analysis of the proposed detection method was carried out showing the robustness and the efficiency of this method, in presence of various errors due to the operator. The proposed method can contribute to the safety of chemical reactors in the chemical industry.