It was noted in Chapter 9 that there are usually at least two major process objectives in reactive distillation: to control the purity of the desired product; and to maximise the conversion of reactants. Only one of these objectives can be satisfied automatically with a one-point control structure although some implicit control of the secondary objective might be realisable if the control system is formulated well. If more direct and exact control of the secondary objective is required, two-point control must be implemented. A third process objective (e.g. the degree of separation of key components, the ETBE or ethanol purity in the distillate or sustained operation at an equipment constraint) cannot be achieved due to insufficient degrees of freedom.
Clearly, there are several combinations of process objectives that could be given priority. One-point control was shown to be effective when the product purity and throughput were considered to be most important. A conventional two-point distillation controller could be implemented when the ETBE or ethanol concentration in the distillate was the preferred secondary point of control. This is equivalent to controlling the separation or fractionation provided by the column and is incompatible with effective constraint management. However, the desire to simultaneously control a product purity and a reactant conversion is specific to reactive distillation and a standard approach is not available. This combination is considered here and implemented (via dynamic simulation) on the 10 stage ETBE column.
The use of a stripping section temperature to infer the bottoms composition was demonstrated in Chapter 9. The control configuration that was established with this measurement (using the reboiler duty) was also demonstrated and requires no modification for two-point control. However, a second control loop must now be configured to provide control of the reactant conversion. This requires the selection of appropriate controlled and manipulated variables. It was noted in Chapter 9 that the direct measurement of the isobutene conversion was unrealistic because it required synchronous results from at least three process analysers. Therefore, some form of inferential measurement is required.
Was this article helpful?