Geometric Process Control in the Petrochemical Industry

Huntsman Petrochemicals faced complications when analysing its 360,000 t/y paraxylene process at Wilton, UK. The reaction section of the plant was subject to rate swings, and there were questions about whether the recovery section was optimally operated. The liquid feedstocks are vaporised and superheated in two parallel heat trains before being fed into the reactor. The hot vapour from the reactor is passed back through the exchanger, where the heat is recovered. Two gas-fired heaters and two steam heat exchangers provide additional heat. The system is subject to rate swings, and the extensive heat integration made analysis of the root causes difficult. Conventional wisdom was that the level in the vaporiser shell was key and that the levels should be run as high as possible. As a result, the vaporiser levels were frequently being adjusted to try and maintain rates.

Downstream of the reactor, paraxylene is recovered by crystallising it out of solution and separating it from the residual liquid with centrifuges. The company had recently improved control in the section to increase the equipment reliability, but the complexity of the series/parallel centrifuge arrangement made it difficult to determine how much material should be recovered in each crystallisation process for optimum recovery.

PPCL’s award-winning Geometric Process Control (GPC) technology was able to shed some light on what was really happening. Using C Visual Explorer (CVE), Huntsman was able to see all of their several hundred variables simultaneously. The entire process could now be analysed in one step, from one screen, without seeing an equation. While traditional time series and XY plots would limit analyses to a few variables at a time, GPC enabled complete visualisation without the guesswork. By highlighting the areas of the data where rates to the reactor were highest and where the vaporiser levels were highest, CVE showed within a few minutes that pushing the vaporiser levels to the maximum was in fact, not the best strategy.

The same display was used to look for other potential disturbances. It was noted that although the furnace exit temperatures were closely controlled, the temperature at the reactor exit was varying and that these changes corresponded to the rate swings. A little more digging showed that a dominant factor driving the reactor temperature was - to everybody’s surprise - the wind speed! Within an hour and a half, the focus had shifted from vaporiser levels to the weather. With the problem understood, the heat input into the process could now be adjusted to account for changes in ambient conditions.

The data also indicated that one of the two parallel fired heaters was operating more efficiently than the other. The unit had been re-tubed and revamped less than a year beforehand and the second was awaiting this treatment. CVE had provided conclusive evidence that this maintenance procedure was worth every penny!

That done, PPCL started to analyse the plant’s centrifuges. The data showed that over the normal operating range, the recovery efficiency was quite flat. How the flow was biased between sets of centrifuges was less important than keeping the system steady. Again, this was a result from a mere few hours of analysis, which, if done using traditional methods, would have taken a considerable amount of time and effort. All in all, within six hours of use, CVE had solved two major operational issues - a tremendous return on investment. Unsurprisingly, this company continues to cooperate closely.