Sampled-Data Control of Distillation Columns

Most sampled-data control systems employ discrete versions of PI and PID control algorithms although computers are certainly not limited to only these types. Special-purpose algorithms can be construaed in the software to deal with the multivariable, nonlinear nature of a distillation column. Adaptive control, for example, updates the parameters in control algorithms and sampling rates to compensate for nonlinearities in the process. Optimal control is a 

---

Having discussed some general points about sampled-data control techniques and algorithms, we now look specifically at how sampled-data control is applied to a distillation column. Fundamentally we are interested in achieving good control in the face of set-point changes and load disturbances—traditionally called servo and regulator control. 

Shunta compares the results with and without sampled-data interaction compensators for controlling both ends of a 20-tray distillation column. Figures 21.7a and 21.7b show the closed-loop responses of composition on trays 2 and 18 for set-point changes and Figures 21.7c and 21.7d for a 10 percent change in feed composition. Tlie improvement in control with interaction compensators is sign cant. 

Flow rates of isoamylene fraction, methanol and TAME fraction are measured using tensiometer balances, and reflux and distillate flow rates using a Coriolis-type flowmeter. The samples are analysed off line. Process data, such as temperatures, flows, levels and pressures are stored and visualised by a monitoring system based on microprocessor controller MSC68. This gives a way to identify steady states in the column. 

---