IDENTIFICATION AND ROBUST CONTROL OF AN INDUSTRIAL STEAM REFORMER

The dynamic behaviour of a methane fired steam reformer is studied in an industrial environment. A system identification approach is used to determine transfer functions linking coil outlet temperature and fuel rate. All models have in common an unstable zero and large gain fluctuation over different operating conditions. A statistical clustering technique was used to determine the reactor dynamic model, for different operating conditions. To counteract the impact of gain fluctuation, a minimax error process model selection was used to determine a nominal model for the design of an IMC controller. The tuning of the controller was shown to be equivalent to the tuning of the time constant of the robustness filter. Improved temperature regulation can be used to increase the operating temperature of the reactor, thus leading to reducing the steam to carbon ratio and in turn improving process ecconomics.