Multivariable model reference adaptive control of a pilot scale packed bed tubular reactor

Abstract

A pilot scale nonadiabatic packed bed tubular reactor carrying out the hydrogenolysis of butane over a nickel on silica gel catalyst is used to evaluate the performance of a multivariable model reference adaptive control algorithm. The multiple reactions are highly exothermic causing steep radial gradients within the reactor. Measurements of temperature along the center of the reactor bed and of exit concentrations are used. The model reference algorithm is based upon Liapunov's direct method. The reference model is based upon a low order state space model derived from a three dimensional partial differential equation description of the process. Model parameters were fitted to reactor data to form the basic model. The model reference algorithm is implemented on a minicomputer which is interfaced to the reactor. The performance of the control algorithm is compared with results from low order conventional controllers and with an empirical low order controller. Problems related to derivation for, implementation on, and performance of the model reference technique on the highly nonlinear reactor system are discussed.