Nonlinear and adaptive multivariable control of a distillation column

Multivariable nonlinear and adaptive control is experimentally applied to two-point control of a binary distillation column with a nonlinear behavior in the desired operating range. In this range, linear experimentally determined models describing the behavior of the distillation column at different operating points are available. A nonlinear model is constructed by letting the parameters of a model with the same structure as the linear ones depend linearly on the operating point. This results in a simple bilinear process model, which is fitted to the linear models at the respective operating points. The controller used in the study is a multi-input/multi-output PID controller containing model-dependent parameters and some tuning parameters by which the tradeoff between performance and robustness can be affected. When fixed parameters are used in this controller, adequate performance is not obtained over the whole operating range of interest. However, if the model-dependent parameters are updated at each sampling instant on the basis of a linearized form of the bilinear model, the performance is uniform over the whole operating range.