State Estimation of a Packed-Bed Reactor Using Ensemble Kalman Filter

This paper describes a method of estimation of concentration and temperature distribution in a packed-bed reactor with high exothermic reactions. One dimensional reactor model which was expressed as nonlinear partial differential equations was discretized to nonlinear lumped parameter system by use of orthogonal collocation. Since the derived lumped system had a large number of state variables, the ensemble Kalman filter was applied to estimate those state variables. Results of our simulation showed that in spite of model parameter uncertainty, the ensemble Kalman filter could estimate hot spot in the reactor well by using several temperature sensor data.