AN ADAPTIVE CONTROL ALGORITHM FOR A THERMAL CRACKING FURNACE

A nonlinear first order model is used to represent the dynamic behaviour of a pyrolysis furnace. In particular it is capable of being readily identified using either control or load change variables. The proposed nonlinear model enables the furnace dynamics to be fitted very satisfactorily. This is because the dynamic behaviour of hydrocarbon pyrolysis furnaces changes with time as coke is deposited on the internal tube surfaces, so that it is necessary to continuously adjust the model parameters to the current state, good performance can be expected from adaptive control. To do this, it is necessary to develop control algorithm based on the proposed nonlinear model. In particular, it is possible to adapt the classical PID controller by continuously updating the controller settings as new model parameters become available. Investigations covering typical operating conditions encountered in an industrial furnace show that significant product and energy savings are possible using the proposed adaptive strategies.