(216a) A Self-Tuning Regulator for Fluid Catalytic Cracking Units

The fluid catalytic cracking converter is considered one of the most important units in petroleum refinaries since it converts heavy residues of oil, without commercial value, into valuable products. Because of its large thoughtput and nonlinear behaviour, improvements or better control strategies and operation of the process can generate expressive improvements in profits. Because of its complex multivariable, nonlinear and nonstationary behaviour, the FCC process control is a challenging problem in chemical engineering. Several linear and nonlinear advanced control approachs were reported in literature for this case: DMC, GMC, Neural Networks, among others. But one common problem with these designs is the limited range of process operating conditions in which tunning update is not necessary. So, they require periodic online tuning to accompaing process operation changes in real plant, which is not desirable or sometimes possible in real time plant operation. Adaptative controllers are know to be more robust to process operating conditions changes because of their intrinsical self-regulating nature. In this work, an adaptive control of adjustable auto regulating type (Self-tuning Regulator) for a FCC unit is proposed. This controller uses a recursive estimator to achieve the process parameters and these are then used for calculation of the desired control action through the controller design module. The STR controller is composed by two parts: an external block, formed by a conventional controller directly connected to the real process, and an internal one, that takes into account the changes in the process and provides the new set of controller parameters to the conventional controller. This internal block is formed by two parts: an estimator of parameters/states and another one that calculates the best parameters for the internal controller, the controller designer. The developed adaptive controller performance is compared with the DMC controller one. The robustness of each approach is compared thought theirs responses behaviour under varying FCC unit operating conditions.