This Month’s Puzzler
Our azeotropic distillation column serves as the first step in separating water from benzyl alcohol (see Figure). Unfortunately, this trayed tower suffers periodic hiccups; flooding occurs, causing an off-specification bottom stream.
Initially, production tried boosting the reboiler temperature to reduce the periodic flooding. This led to condenser loading problems and high pressure in the condenser decanter that raises concerns about popping our pressure relief valve.
We looked at the feedstock for water contaminations. One operator suggested that the decanter is too small.
Another brainstorming session produced the idea of adding a feed preheater to vaporize the feed. The column designer says this is a bad idea. He warns it would increase the load on downstream columns because of more alcohol in the condensate. He suggests running a gamma back-scanner to determine where liquid is accumulating in the tower. His boss recommends a neutron back-scanner because of insulation.
What do you think is causing the problem? What tools should we use to determine the flooding tray(s)? What ideas do you have on how to address the periodic flooding?
Consider Three Cures
The periodic hiccup symptoms described are typical of component trapping. I discussed in detail this phenomenon, causes, troubleshooting and cures in “Component Trapping in Distillation Towers: Causes, Symptoms and Cures,” Chem. Eng. Progr., p. 22, August 2004. In addition, Case Studies 2.10 through 2.23 in my book “Distillation Troubleshooting” describe various hiccup experiences. Of these, Case Study 2.22, which focuses on three hiccup incidents in azeotropic distillation towers, is particularly relevant here.
Hiccups occur when the tower feed contains components with boiling points intermediate between the light and heavy key components. These components tend to concentrate toward the middle of the tower. In some cases, the tower top is too cold and the bottom is too hot to allow these components to leave the tower at a rate sufficiently high to match their feed rate. Having nowhere to go, these components accumulate in the tower, causing periodic flooding and hiccups. As observed, heating up the tower (by adding reboil or more preheat) permits these components to leave in the overhead — but at the price of loading up the condenser and raising pressure (as the plant observed) or losing product in the overhead (about which the designer expressed concerns).
Testing can help confirm the diagnosis. If practical, drawing internal samples can identify the trapped component. Judicious sampling of feeds and products during a hiccup also may determine it. Observing trays’ temperatures is valuable. In most hiccup experiences, the tower temperatures hold steady for some time, then one or more temperatures deviate, sometimes slightly. If the one that changes is near the control tray, it may interact with the controls. The temperature deviation often intensifies until the hiccup occurs, returning to normal after the hiccup. Gamma scans during a hiccup typically show flooding midway up the tower when hydraulic calculations indicate the tower operates a healthy margin away from the flood points.
There are three cures to a hiccup problem. One is to heat the top or cool the bottom to allow the component out. Per the problem statement, in this case, heating the top has unacceptable side effects. Cooling the bottom probably would run into product specification issues or may not go far enough because the bottom temperature is quite hot. The second solution is to remove the trapped component using a properly engineered side draw. Gamma scans and simulations can find the point where the component concentrates, and the side draw can be added there. The third solution is to remove the trapped component from the feed or reflux.
Removing the trapped component from the feed or reflux generally is the preferred solution in azeotropic distillation; it was implemented in two of the three incidents in Case Study 2.22 as well as in other cases we’ve had. A common way of achieving this is by adding stripped or fresh water to the decanter inlet. (Yes, this is correct!) This will work when the trapped compound is both water- and benzene-soluble, such as phenol. The atmospheric boiling point of phenol is 181°C, less than that of benzyl alcohol (205°C) and much higher than those of benzene and water. In the decanter, the benzene will extract it into the reflux and back into the tower, preventing most of it from coming out in the water product. Adding water will extract it out of the refluxed benzene into the water draw. Another solution may be to replace the commercial-grade benzene makeup with analytical-grade (i.e., high-purity) benzene. However, this only will work if the trapped component comes from the benzene makeup.
Henry Kister, Senior Fellow &
Director of Fractionation Technology
Fluor, Aliso Viejo, Calif.
Remove A Contaminant
There’s a contaminant in the benzene. The reflux probably is carrying it back to the tower where it is accumulating. Increasing the tower bottom temperature likely won’t do much to eliminate the impurity but might improve the bottom product purity.
Where do you get your benzene? Talk to the supplier after taking some samples from the decanter and from your fresh stock. Also, consider the possibility of something forming in the tower as a result of recirculation through the reboiler. Compare benzene from different times. Have your supplier do some of the lab work.
Don’t stop there. Have the operators take fresh samples from the top, feed point and bottom — and everywhere else available. Samples should be fresh, so either teach the operators to do the tests or have the laboratory do them within the hour. Also, set aside all the samples for a week to allow settling. Simple settling tests can provide valuable insights.
A preheater might work in boosting the vapor content in the tower but this will increase the number of trays needed and perhaps require re-location of the tower feed tray. This seems like a lot of trouble.
A neutron back-scanner makes sense only if insulation is thick, the tower diameter is large, foaming or sediment is suspected, or the densities of the condensate and bottoms liquids are similar as is the case here with water and benzyl alcohol. So, the boss is right.
As for the decanter, if you’re popping relief valves for a change as minor as increasing the reboiler steam rate, you should consider a larger decanter. This will allow you to change the internal dimensions to improve separator efficiency.
Include the decanter in sampling and testing. Also, if possible, look inside the decanter for a rag layer, i.e., a film or even thick stratum that forms between the top and bottom layers when a contaminant is created. Such stratum can be a site for corrosion; so, if you can, check for corrosion. Don’t forget to talk to operators and maintenance staff about past inspections of the column and the decanter.
Dirk Willard, consultant
We experienced a massive quality control failure in our batch process when a problem occurred in a high pH (12–12.5) control loop. We’re down and our sister plant is handling our customers until this problem is fixed. In our process, we neutralize a weak acid feed solution with a caustic solution. The automatic control for caustic dosing worked fine in the past but doesn’t anymore. So, operators have been doing lab tests and manually manipulating the caustic addition loop. Titration has changed recently because a particular indicator was out of stock. Because the company guideline doesn’t call for a specific indicator, the operators have used several but can’t seem to get the titration right. We also have tried pH probes from various vendors but all suffer fouling problems. Now we’re regularly facing situations where the pH probe slows down and then the control valve bangs up and down on its limits. In addition, I’m concerned about our use of a ball valve for dosing the caustic instead of a smaller globe valve. A corporate electrical engineer chose it — and asserts it works as well as a globe valve for caustic addition and, besides, it’s cheaper. That valve handles the addition of the large quantity of caustic that sometimes is necessary during startup as well as the small amount usually needed during normal operations.
I also question the accuracy of the pH probe because calibration relies on only two points, at 7 pH and 10 pH, at 25°C when the process operates at about 55°C.
The plant instrument engineer installed three pH meters a few years ago with the idea of a voting logic to average the pH signal for valve control. Currently, none of the probes and meters are in working order. He’s convinced we should opt for installing a pair of valves, one large and the other small, to control caustic flow.
What do you think we need to do to get the process running smoothly?
Send us your comments, suggestions or solutions for this question by September 14, 2018. We’ll include as many of them as possible in the October 2018 issue and all on ChemicalProcessing.com. Send visuals — a sketch is fine. E-mail us at ProcessPuzzler@putman.net or mail to Process Puzzler, Chemical Processing, 1501 E. Woodfield Rd., Suite 400N, Schaumburg, IL 60173. Fax: (630) 467-1120. Please include your name, title, location and company affiliation in the response.
And, of course, if you have a process problem you’d like to pose to our readers, send it along and we’ll be pleased to consider it for publication.