Distillation column tray temperature control
Much has been written about distillation DCS control strategy, covering many configurations, constraints, considerations, and it’s not easy to see the forest from the trees. This blog proposes a “default” DCS configuration, good for most cases, to be served as a basis. Deviations from the basis are of course permitted, though reasons for these deviations should be discussed and understood.
Is that even feasible? Can one DCS configuration be valid for most cases? Distillation columns separate a mixture of components into a light boiling product (distillate) and a heavy boiling one. To be successful the control strategy must first set distillate yield (or “cut”) correctly, to draw all of the light components in the feed. That is important because failure to set the cut correctly would, by mass balance, leave one of the products contaminated. Once the cut is set correctly, a second requirement is setting column vapor and liquid traffic (or “fractionation”) correctly. Product separation is not absolute and cross product contamination is a function of liquid to vapor ratio. We need a method of defining reflux or reboil ratio to achieve the desired product purities. Thus, distillation control is about manipulation of two handles: “cut” and “fractionation”. That said, precise cut control is more important than fractionation control. A 1% cut error may cause significant contamination, whereas for fractionation, even a 5% shift in reflux ratio is of minor consequences.
The proposed default configuration is illustrated in figure 1, a common arrangement that would work for many cases. This DCS structure is called heat balance type configuration, augmented by a tray temperature controller (tray TC). Reflux flow is determined by the operator whereas distillate draw is on accumulator level control. Distillate yield is determined here by heat balance. Increase reboiler heat duty and then more vapor would reach the condenser, liquified and taken out as distillate. Hence the name – “heat balance”.
Figure 1. Heat balance control configuration with stripping section tray TC
Now for the tray TC contribution. The operator has no precise knowledge of feed composition and no clue as to what distillate yield is desirable. Beyond that, feed composition and flow may vary continuously. Tray TC is a rudimentary inferential controller of the cut. Should the feed become heavier, our tray would warm up. TC would cut down reboiler heat duty, reducing distillate yield. Is tray TC a perfect inference? No, but upon feed composition or enthalpy changes it responds in the right direction. Without this TC, distillation heat balance configurations are difficult to operate. A more elaborate inferential model is feasible, and such a model could be implemented as a part of APC, and it would reset tray TC setpoint.
What about reflux control? Shouldn’t the distillation control strategy include some form of fractionation control. Automatic reflux control is not as critical as cut control because the operator knows roughly what reflux ratio is needed for the separation. Slow DCS ratio controller of reflux to feed or to distillate is not out of the question, as shown in figure 3, although dynamic considerations suggest that fractionation control logic should be left for the APC. To illustrate this point, consider an event when feed flow is ramping up. Tray TC becomes colder and the TC is working hard, ramping the reboiler to correct the cut. The last thing we need at that moment is an increase in reflux flow that would further cool the column. APC on the other hand would have logic to permit increasing the reflux only when the TC is at or above setpoint.
Figure 3. Addition of reflux logic (not recommended)
What about pressure control? Per figure 1, this column has a total condenser where condensed liquid is subcooled, and we have chosen the hot vapor bypass pressure control method, employing a hot vapor stream, bypassing the condenser. This method works by partially filling the condenser with liquid, reducing condenser effectiveness. When column pressure is too low the hot bypass opens, sending superheated vapor to increase accumulator pressure. That reduces the column-accumulator pressure difference, slowing down condenser drainage to build up condenser liquid level, reducing condenser effectiveness. And vice-versa, when column pressure is too high hot vapor bypass is reduced, accumulator pressure comes down, encouraging condenser drainage, thus exposing more of the condenser area and increasing condensation to drop the column pressure.
Actually, a decision on whether or not employ the hot vapor bypass method is in the hands of the equipment designer, not the control engineers. If the column overhead is not designed with hot vapor bypass, other pressure control methods are feasible, though still, hot vapor bypass is popular and it makes sense to choose it as default for columns with total condensers.
On what tray should be the TC be located? Figure 1 does not specify whether the TC should be placed near to or far away from the reboiler. The object of distillation is to separate a light component, called light key (LK) from a heavier component, called heavy key (HK). Many of our columns deal with multiple components, some lighter than LK and others heavier than HK. The lighter components (LLK) would evaporate to become part of the distillate whereas the heavier components (HHK) would plop to the bottom. Each column has a composition profile, where vapor and liquid composition become heavier near the reboiler. While bottom product specification may call for 1% (or less) LK contamination, the tray TC vapor stream would have about much higher content of LK, ideally 15-20%. Tray temperature is a function of tray composition, and upon change of composition tray TC would manipulate the reboiler to correct the tray composition. Thus, the tray TC works by keeping tray composition relatively constant.
Having said that, tray TC location is often determined before control engineers can state their preference. We may be stuck with nonideal locations, which affect the tray TC reliability.
What about a tray TC locate in the rectifying section? For dynamic reasons we would want this TC to manipulate reflux flow rather than reboiler duty as shown in figure 2. From heat balance point of view reboiler manipulation is similar to reflux manipulation. Reducing the reflux would increase distillate yield, but dynamically manipulating the reboiler to control a rectifying section tray temperature is painful. Ideally this TC would be located on a tray where the HK content of tray liquid would be in the order of 15-20%.
Figure 2. Heat balance control configuration with rectifying section tray TC
In conclusion, we’ve come up with the simple configuration of figure 1 (or 2) as a basis default configuration for a simple distillation column DCS strategy.
