Fault Tolerant Control of a Batch Reactor -A Case Study for Decomposition of Di-Cumyl-Peroxide

Chemical batch-reactors are used in process industries to produce agro-pesticides. Reactors can be operated in a stable or sometimes unstable mode for economic reasons. Some specific exothermic reactions exhibit runaway conditions, which need to be avoided (by proper control) as they lead to explosions. When the heat of the reaction exceeds or the coolant fails, the reaction tends to become unstable and may proceed to run away. The decomposition of di-cumyl peroxide is one such runaway reaction. The work aims to design a mechanism for removing exothermic heat by continuously monitoring reactor temperature and controlling coolant rate for the safe operation of the reactors.  Due to faults in the system, the temperature is estimated using an Unscented Kalman filter. The information is used in a closed loop to detect the fault in the sensor and to diagnose the fault using the fault tolerant control system to efficiently control the coolant flow rate through the jacket of the reactor. Out of neural–the proportional integral controller, gain scheduled model predictive control, and model predictive control - proportional integral controller, the latter is found to work well and satisfactorily to set-point and disturbance rejection. Under faulty sensor conditions, the model predictive control - proportional integral controller scheme showed performance with an Integral Absolute Error of ۴.۲۸ for set-point on temperature while ۲.۸۵ for load-disturbance (under ۱% change in feed concentration) rejection case. Authors opine that the accuracy of the model can be exploited in effectively designing and implementing fault tolerant control for the reactor to avoid runaway.

کلیدواژه ها:

Fault-Tolerant Control ، Gain scheduled model predictive control ، Model predictive control - proportional integral ، Neural- neural-proportional integral ، batch-reactor ، runaway reaction ، Decomposition