Detection of runaway reaction in a polyvinyl chloride batch process using artificial neural networks

In this paper, the method of detecting a runaway reaction in a batch process using artificial neural network (ANN) model is proposed. From the safety point of view, it is difficult to investigate what the causes of runaway reactions in the chemical plants are, besides as the consequences of the accident cannot be so well investigated. In this study, the parameters of temperature, pressure, and level of the reactor that influence a runaway reaction are studied. Based on historical data of the parameters (temperature, pressure, and level of the reactor) for normal and runaway reaction, the ANN model and simulation of Polyvinyl Chloride plant has been developed. Analysis of the learning and results of the ANN algorithm and simulation are presented and discussed. It is established that by using this approach, we can observe and compare what's the dominant factor of the runaway reaction. Finally, we also can visualize the characteristic of runaway reaction so we can anticipate and isolate this kind of fault.