columns operate day in and day out at constant pressure, there What is the worst mistake? It is the one that is repeated in may be some seasonal changes, and the configuration in Fig. 2 practically every refinery, and yet, we havent managed to learn might save the operator one minute or so upon changing colfrom it. I have chosen to discuss the very simple distillation column operating pressure. In any case, this is a tiny DCS change, umn pressure control by partial flooding of the condenser. This so why not do it? method is used often when the column Because this tiny DCS modification overhead vapor is totally condensed into creates a not-so-tiny change of loop the accumulator. PC PC interaction pattern. In the previous conAs illustrated in Fig. 1, a valve is placed LC figuration, upon an increase of column on the condenser outlet to create a flow FC FC FC pressure the bypass valve closes, but with restriction and build up liquid level inside this new Fig. 2 configuration, when colthe condenser. As level rises, condensing umn pressure goes up, the bypass valve area shrinks and then the column presFC would open. That is, as the column sure starts trending up. And vice versa, LC pressure controller is trying to drain the upon draining the condenser, condensing FC condenser, the bypass control acts to area increases and pressure starts trending increase accumulator pressure, preventdown. This control method must allow ing condenser drainage. That interaction some hot vapor to bypass the condenser Fig. 1 Condenser hot vapor bypass control. makes the control loops next to imposto control accumulator pressure. Hot sible to tune. I must have seen at least vapor coming into the accumulator con100 Fig. 2 configurations and none of denses on the subcooled liquid surface, them worked. Invariably the hot vapor and hence, the hot vapor flow is always bypass was operating manually, and the unidirectional. Accumulator pressure PC DPC good intention of avoiding the nuisance should be set somewhat lower than colLC of changing two setpoints instead of one umn pressure to permit proper operation FC FC FC has caused a much bigger nuisance of of the condenser outlet valve. having to change the hot vapor bypass This system in Fig. 1 works well and position manually, and experiencing freis easy to tune. Accumulator pressure is FC quent pressure disturbances. tuned fairly tight, whereas the column LC Fig. 3 offers a way to have your cake pressure is tuned slowly, much like a level FC controller. That makes sense because in and eat it too. Use pressure difference response to a change in outlet valve posifor controlling the hot vapor bypass, but tion, uncondensed vapor keeps accumuinstead of the actual column pressure, use Fig. 2 Improved hot vapor bypass control. lating and the pressure acts like a pure the column pressure setpoint. Another integrator. While these two control loops tiny change of DCS structure will now interact, they interact in such a way as to restore the favorable dynamics of Fig. 1 Pressure setpoint help each other. When column pressure while maintaining the intended convegoes up the accumulator pressure connience of Fig. 2. Do please try that and PC DPC troller would shut the bypass and the colsee the magic of an impossible-to-tune umn pressure controller would open the loop becoming a cinch. HP LC condenser outlet, and both these actions FC FC FC go in the same direction of reducing the Y. Zak Friedman is a principal consultant in column pressure. advanced process control and online optimization In the enhanced system in Fig. 2, with Petrocontrol. He specializes in the use of firstFC principles models for inferential process control and the accumulator pressure controller is LC has developed a number of distillation and reactor replaced by a pressure difference controlFC models. Dr. Friedmans experience spans over 30 years ler. It is a minor DCS change with a very in the hydrocarbon industry, working with Exxon good intention. In the configuration of Research and Engineering, KBC Advanced Technology Fig. 3 Another tiny change of DCS structure and since 1992 with Petrocontrol. He holds a BS degree Fig. 1, when the column pressure is to be will now restore the favorable from the Israel Institute of Technology (Technion) and dynamics of Fig. 2 while maintaining changed the operator must change the a PhD degree from Purdue University. the intended convenience of Fig. 1. setpoints of both controllers. While these 150
Article copyright 2011 by Gulf Publishing Company. All rights reserved.
I June 2008 HYDROCARBONPROCESSING Not to be distributed in electronic or printed form, or posted on a Website, without express written permission of copyright holder.