MULTIVARIABLE CONTROL OF A CATALYTIC TUBULAR REACTOR USING BOTH WIENER-HOPF CONTROLLER DESIGN AND INTERNAL MODEL CONTROLLER DESIGN APPROACHES

A practical application of advanced model-based multivariate control to a pilot plant packed bed reactor carrying out highly exothermic butane hydrogenolysis reactions is presented. The system was first stabilized by using butane flow rate to control the reactor hotspot temperature. Propane production and butane conversion were then controlled using multivariate controllers which manipulated the hotspot temperature setpoint and hydrogen flow rate. An approximate linear dynamic-stochastic model was developed using multivariate time series and process identification methods.Two multivariate controllers were applied. The first was a simple version of Internal Model Control (IMC) utilizing a stabilized approximation of the model inverse and a tunable exponential filter. The second was a constrained minimum variance controller design using pulse transfer function models to characterize the process dynamics and autoregressive-moving average models to characterize the stochastic disturbances. The optimal control solution was arrived at via a spectral factorization solution to the Wiener-Hopf equations. The results indicated that both designs, if well-tuned, provide adequate performance and robustness over a wide region of operation.