Ask Greg McMillan - What role do you see dynamic simulation playing in the future of best distillation column temperature controller tuning?

Greg's Response:

Dynamic simulations running faster than real time are key to ensuring the best column temperature controller tuning. Except for high purity columns, changes in the split (material balance) have a ten to one hundred times greater effect on the control plate temperature than changes in the separation (energy balance). The dynamics for manipulation of split are rather slow due to a time constant associated with each tray’s liquid holdup composition response. The time constants are interactive because the adjacent trays mutually affect each other through downcomer flow and vapor flow. This large number of interactive time constants in series establishes, for a first order approximation, a relatively constant dead time to equivalent time constant ratio of about 0.2. A column usually takes several hours to line out after an upset. For large columns, it may take several shifts. Packed columns tend to have one third the liquid holdup and, thus, are about three times faster than tray columns. The net result, in any case, is that the tuning of a temperature loop that manipulates split requires extreme patience. The chance of a disturbance occurring during the long period while the controller is in manual for an open-loop test is considerable. The closed-loop method is less disrupted by upsets. The tuning test time can be reduced to about 4 dead times by using the “near integrator” tuning method where the dead time and maximum excursion rate is estimated like the Ziegler Nichols Reaction Curve method.

The tuning can be checked by the identification of the dead time from the delayed start of the response to a setpoint change. The equivalent time constant is then estimated to be about five times the dead time for the manipulation of split. The open-loop gain is the dimensionless product of the valve (manipulated variable) gain, process gain, and measurement gain. The process gain is the change in temperature for a change in split (distillate to feed or bottoms to feed flow ratio) multiplied by the inverse of the feed flow. Composition controller responses can have much greater dead time due to the analyzer sample and cycle time best addressed by an enhanced PID.

For much more knowledge, see the ISA book Advanced Temperature Measurement and Control, Second Edition (LIST PRICE REDUCED by 50%. Use promo code ISAGM10 for an additional 10% discount on Greg’s ISA books).