Selection of the Final Control Scheme

Usually, catalytic cracking units are operated close to the constraint boundaries, which define an optimum for the oil feed throughput. The process is running at maximum throughput, hence some valves ate fully open. However, there are still some measured variables and correcting variables that have to be paired and the following section will discuss this in more detail. For a detailed process description and discussion of the operation of the process the reader is referred to McFarlane et al. (1993). 

There are three more process variables to control the reactor temperature 7) to ensure that cracking takes place but the catalyst is not damaged by excessive temperatures, the temperature in the regenerator to ensure that all the coke is burned from the catalyst and the oxygen concentration in the stack gas Xq to ensure that the regenerator runs with an excess of oxygen. 

This reduced control problem definition agrees with the solution that Khandalekar and Briggs (1995) presented. This problem is schematically shown in Fig. 35.8. 

The authors used the Amoco model and compared it with a model developed by Lee and Groves (1985). They parameterized the Lee model to match the more complicated Amoco model by adjusting the heat of reaction and coke formation rate constant. They also studied non-linear multivariable control of process variables that showed much inteiactioa 

The authors developed macroscopic non-linear approximate models to calculate the manipulated variables as shown in Fig. 35.8, and compared control performance to regulatory proportional-integral-derivative (PID) control. In this chapter the discussion will be limited to PI(D) control. 

---