Effectiveness of Inferential Conversion Control

One-point control of the ETBE reactive distillation column using the reboiler duty (or the bottoms draw rate) to control the bottoms composition inferred via a stripping section temperature (e.g. stage 7) was found to be closed-loop stable, unlike similar control schemes that used other temperatures (e.g. the reboiler temperature) to infer the ether purity. This control scheme has good disturbance rejection properties and yields fast responses but provides only implicit control of the reactant conversion.

Isobutene conversion in this column can be predicted to within 1.0% using a statistically based inferential model and measurements of only two column temperatures (e.g. the temperature at the top of the reactive section and a temperature near the reboiler). The predicted conversion is closed-loop stable with the reflux rate, provided the bottoms composition is also controlled in closed-loop. The combination of bottoms composition control (using a stripping section temperature, with or without set-point cascade from a process analyser) using the reboiler duty and inferential conversion control using the reflux rate produces an effective two-point control system which is appropriate for reactive distillation and, specifically, for ethenfication columns.

The primary advantages of combined composition and conversion control, compared with composition-only control, are:

• automated adjustment of the operating conditions to optimise isobutene conversion, and therefore to maximise profitability;

• highly effective rejection of feed rate disturbances from both the composition and inferential conversion control loops;

• tighter control of the ether composition where an inferential composition model is used (e.g. via a stripping section temperature);

• increased sensitivity in the composition control loop;

• no loss in dynamic responsiveness.

The control scheme relies on a process model to predict the conversion and the effective implementation of a composition controller. The composition controller might also rely on a secondary process measurement (e.g. a stripping section temperature) to infer the process objective (i.e. ether purity). Consequently, the control scheme has some intrinsic limitations, including a reduced effectiveness for feed composition disturbances, which is a requirement for robust and accurate temperature sensors and transmitters (especially at the top of the reactive section), and the inability to control the actual reactant conversion to a set-point without incorporating laboratory updates into the control scheme to compensate for model uncertainty. These limitations are essentially minor for most practical applications. Set-point control can be achieved using process analysers to measure various compositions directly, but the presence of measurement errors (which may be as high as the inferential model) and dead-time are detrimental, and the overall control system effectiveness is not guaranteed to be enhanced.

Was this article helpful?

0 0

Post a comment