THIS MONTH’S PUZZLER
We’re having trouble controlling the reboiler in the regeneration column for diethylamine (DEA) — see Figure — at a refinery we recently bought. The DEA concentration is about 35% by weight. The tower pressure varies 2–3 psig from design. Originally, this DEA process fed our refinery sulfur recovery unit (SRU); as of the last turnaround, it only polishes the gas before the SRU. This system hasn’t been stable since startup.
This tower was built many years before modern simulations existed; plant records lack details about any models used for the column.
Should we move the feed tray, change the feed conditions, or reposition the controlling thermocouple further up the column? Would raising the condenser recycle ratio help? Can we alter the amine or the concentration to get better results? What else should we consider?
Make It Simpler
Your DEA regeneration seems complicated. Let’s see if we can’t make it simpler and address your questions first.
If the amount of acid gas absorbed in the DEA is lower, then the vapor/liquid ratio for the still trays may be below the minimum operating flow range (typically 10–15% for bubble cap trays). If amine flow was trimmed to compensate for reduced loading, then the lower rate could lead to solids settling out on the trays and flows could be channeling.
Generally, units built before good computer modeling (about 1985–1990) are overdesigned. Have an engineer with over 15 years of experience in troubleshooting amine units develop a model that incorporates detailed equipment evaluation to set your capacity limits. Then, you can know if you are operating at a lower or higher than design capacity. Alternatively, you can adjust operating conditions, e.g., amine flow or reboiler duty, to determine maximum capacity of the exchangers, trays, pumps and control valves.
Keeping the pressure steady with small, slow changes is the most critical control parameter for the column. When the pressure drops, so does the bubble point temperature (BPT) throughout the column and reboiler, thus causing increased vapor flow up through the trays. Vapor traffic in excess of tray capacity is the major cause of tray damage. I have also seen large pressure swings move the reboiler and break its foundations. Conversely, a pressure increase raises the BPT: until the reboiler catches up, the vapor traffic is reduced or stops in the tower.
Tower pressure is shown at the top of the column as 7 psig; the pressure varies by 2–4 psig. Typical design pressure for the still reflux drum is 7–10 psig, which depends on the pressure required for the downstream unit plus about 3–5 psig for the pressure control valve (PCV) to operate properly. Because this is a steam stripper and not a distillation column, the pressure sets the temperatures throughout the column. Higher pressure means higher temperatures, which makes condensing easier and reboiling harder: more energy is required.
The simplest solution may be that the pressure controller (PC) and PCV are causing the pressure swings. Normally, the PC is on the reflux drum and the PCV is on the acid gas outlet. The PC is shown on the tower overhead line, which does not account for changes in pressure drop through the condensers. Therefore, if the PC sees a higher pressure and opens the PCV, then the initial increased flow will cause more pressure drop and an even higher pressure, thus making the PC open even more. When it catches up, it will reverse the action and drop below set point, and repeat. Also, if the PCV is a full-line-sized butterfly valve, then its valve characteristics may be causing the oscillations. Consider adding a digital valve controller (DVC) to the PCV or changing the valve to a V-ball. Also, if there is another control valve (CV) downstream, then the two CVs may be fighting each other.
Another control problem may be two-phase flow. Do the lines from the still to the air heat exchanger (HE) and the trim HE to the reflux drum have liquid pockets? Remember, we are working with a condensing fluid and two-phase flow and flow should be continuously downhill. Pockets of liquids can cause slug flow, which would create pressure swings.
Do not move the feed tray. The still feed is shown going to tray 3 (simulation stage 2), which is fine. This tower uses trays 1 and 2 as wash trays. The reflux water washes out the last traces of amine so as to reduce amine loss in the overhead. Many amine stills do not even have wash trays.
Many gas plant still columns have 20 trays. Your 26 trays will provide a bit more efficiency.
Typical amine still tray efficiency is 50%. A proper model may indicate a reduced tray efficiency (less than 25% reliable), which could indicate some tray damage or fouling. During the last turnaround, was the still cleaned and its internals inspected? No other investigative technique is better than putting your own eyes on the internals.
Changing the feed conditions is very difficult. If the amine feed HE is designed and working properly, then the outlet will be about 200–230°F with about 2–5% vapor. Some of the acid gases are released and form steam. Trying to get any higher temperature than this is futile because, as more vapors are formed, the exchanger heat transfer coefficient drops dramatically. Nearly all of these exchangers are designed based on the rich amine liquid properties and, despite what the specification sheet may state, they won’t perform if there is any substantial amount of vaporization. Also, excessive heat input will have to be removed in the condenser, without much benefit. What good is the trim HE on the still feed? I suggest taking the feed trim HE out of service, at least until you get the rest of the unit under control.
Your drawing doesn’t indicate if there is a feed control valve (FCV) and if it is located upstream or downstream of the feed temperature element (TE). If there is not a CV or the TE is located downstream of the CV, then the fluid is essentially at tower pressure and the temperature will be nearly constant. The TE should be upstream of the CV to monitor amine exchanger performance, with a temperature approach of about 50°F.
Reboiler control looks like your biggest issue. Many have tried using a lower tray (25), an upper tray (8) or overhead temperature to control the reboiler. Amine stills are not distillation columns. Unless the tower is at least 75% loaded, it doesn’t work very well because of the large time lag between the boilup and the reflux. What does work is to set the reboiler duty proportional to the amine flow set point. The required reboiler duty, in btu/h, = 60 × amine flow (gpm) × F, where F = 1,000–1,600 btu/gpm. Check your simulation to get a rough btu/gpm starting value, or use 1,200 as a rough guess. Use the amine flow set point and not the actual amine flow, which may swing a bit. This is the set point for the reboiler duty controller (also called the QIC).
Knowing the flow of steam (mS) and the pressure in the reboiler, you can calculate how much duty (QS) the steam is providing (QS = mS × ΔH). This becomes the process variable for the QIC. If you do not have a steam flow element, then you can use the control valve characteristics if it has a DVC. For those with a hot oil system, you use the hot oil flow rate and hot oil temperatures in and out of the reboiler to calculate the duty.
Once the system is steady, check the lean amine loading and adjust the boilup factor (F) to get the lean load to target; I suggest 0.035 mol/mol for DEA. F is mostly dependent on the amine circulation rate and, to some degree, on the performance of the amine/amine exchanger. To a lesser degree, the lean load depends on other factors such as acid gas loading. You can use your simulation to see how F changes with loading. The factor also can be tracked to indicate changes in the unit performance such as exchanger fouling, reduced amine properties, tower performance, etc. However, if the amine regen is lightly loaded, the trays will require a minimum amine flow rate and reboiler duty to be properly loaded. In this case, you should reduce your amine concentration and increase the amine flow.
The reboiler temperature is also important. Monitor the amine inlet and vapor return as well as the temperature at tray 25. Stripping should occur in the tower, not the reboiler. Because the column is a steam stripper, the column temperatures should be nearly constant (just like the simulation), ranging from about 235°F just below the feed to 245°F in the reboiler.
Reboiler hydraulics also are important. The drawing indicates that tray 28 is a draw tray and the reboiler is located below the tower. Kettle reboilers have a weir to keep the tubes covered. How does liquid over the weir flow back to the bottom of the still? It should be flowing downhill all the way to the bottom of the still, which serves as the unit surge volume. How do vapors below tray 28 equalize with vapors above the tray? For other units that have vertical thermosiphon reboilers, proper hydraulics and exchanger design is much more complicated.
One improvement is to put a high and low limit on the reflux rate and use the reflux drum as a true surge tank. The condenser recycle ratio is determined by the steam boilup. Most of the acid gas exists as vapor and the condensed water is returned as reflux. At reasonable loads, a simple level control generally works well.
As a last resort, alter the amine type or the concentration to get better results. DEA has been around for a long time. Unless your rich amine loading is over 0.35 mol/mol, and your amine flow exceeds capacity, I would not change. If your lean loading is low, then you can reduce your amine flow down to 50% of design to save energy.
Here’s some other thoughts to consider:
1. The still overhead temperature and condensing temperature are very low. Set the air HE to 220°F. Condensing the reflux below 110–130°F has minimal return. If the overhead is 90–130°F and the reflux is 90–100°F, heavier hydrocarbons (C5+) are being condensed and recycled, building up on trays 1 and 2. These heavies then slug down to the hotter portion of the still, where they meet the hot steam and rapidly vaporize, contributing to the swings.
2. This may not show up in your model because it is using an amine property equation of state (EOS). Take the vapor stream properties from the feed stage and create a new model of the overhead condensing system using Ping-Robinson or other suitable EOS. Use both EOS methods to model and troubleshoot the condensers. This will show you how changing reflux temperature affects the amount of heavies in the reflux and the acid gas stream. If heavy hydrocarbons are not an issue with the downstream unit, then simply raise the condensing temperature. Raising the reflux temperature from 100°F to 130°F may reduce the condensate in the reflux by as much as 50%.
3. The reflux should be clear and contain less than 1% amine and very little hydrocarbons. There might be condensate in the reflux drum. Check the sight glass for condensate or, better yet, drop the level to the lower sight glass tap and take a sample for analysis. If your reflux is not clear, then dump the reflux drum and refill it. With a minimum reflux level, start at a minimum reflux rate and increase to normal. Don’t worry, the system will operate without reflux. You may need to repeat for several days to clear it up. Also check the hydrocarbon skimmer in the rich amine flash tank.
4. Add a chilled-water overhead trim HE. In hot climates, trim HE is added to assist the air cooler during summer months. Once the air temperature drops below 80°F, the water cooler is bypassed. Cooling water cost money.
5. The reflux TE is shown on the reflux drum. Poor circulation in the drum will cause lag. Move the TE to the pipe before (or after) the drum.
6. Amine quality was not mentioned. If there is a considerable amount of solids (FeS, rust, etc.) in your amine, then improve your filtration. Also, maintain high flow rates to keep solids suspended.
You should be able to immediately implement many of these suggestions, while others will have to wait for the next turnaround. After making the changes to the reboiler controls, bypass the feed trim HE and the overhead trim condenser. Startup is easier and quicker with fewer control loops in service. An amine unit is pretty simple, don’t try to make it complicated.
Larry Tarkington, consultant
San Antonio, Texas
Consider Your Option
Amine regeneration columns work best when operated with feedstock that is steady in concentration and flow rate; this is generally true of most towers. Although the gas absorber won’t be affected much by a reduction in H2S concentration, the regenerator performance will suffer. Let’s consider your options.
Even a McCabe-Thiele analysis of the column shows that moving the feed tray further down should improve the tower performance slightly with reduced feed concentration of H2S. Note that CO2 also will be present in the tower distillate.
Changing the state of the feed from a saturated vapor to a liquid won’t work because this has limited effect on tower efficiency. Besides, in this tower, the absorber section, above the feed tray, is currently less than 5% of the total tower tray area. In addition, going to a liquid feed adds to the reboiler load.
Raising the condenser recycle ratio will reduce the number of tower theoretical stages (and real stages) but put an additional load on the condenser and the reboiler. Increasing reflux cools the absorber section of the tower (above the feed tray). Liquid cascades down the tower and the reboiler runs harder to evaporate the liquid. If the tower feed tray were further down, the feed heaters could take up some of this heat requirement.
Changing the flow of the amine in the gas absorber probably won’t have much effect on the performance of the regeneration column. It’s worth a try on a simulator though.
I find it curious that the tower efficiency is 40%. Normally, 40–50% is used for amine regenerators. This seems highly conservative.
Now, the tough question: the location of the thermocouple for the reboiler steam valve control. Ideally, you want the reboiler thermocouple located on a lower tray where it will see the greatest differential temperature change. (Maximum differential is a control axiom.) For this column the new location could be further up the column. At least that’s what a McCabe-Thiele plot shows; you will get a more precise definition from a full simulation. Temperature control may not be the answer.
Kister, in his book “Distillation Troubleshooting,” discusses a similar situation in case study 27.1, p. 374. He suggests controlling steam flow to the regenerator by a flow meter instead of a thermocouple. Perhaps this is the answer for a stable running regenerator column. Only a full simulation involving the entire amine system will validate this solution; a gamma scan, while eventually useful, disrupts normal operation so see what the simulation shows first.
Dirk Willard, consultant
Our corporate engineering department can’t agree on how to deal with the wastewater at our new beet sugar plant. The flow averages about 1,600 gpm and peaks at about 2,400 gpm. We also have to consider rainwater, namely, a once-in-10-yr flood event of 16-in./h. Our current plan is to use a flocculant in a clarifier to eliminate the dissolved solids that discolor the water. However, both the flocculent vendor and the clarifier manufacturer say we must ensure the clarifier feed won’t contain any suspended solids.
The production manager insists we need a centrifuge but dismisses bench tests as useless for predicting the operation of full-scale equipment. One centrifuge maker wants to rent several units to us on a trial basis.
However, the project manager and process manager both want to test the samples on a bench scale.
Our problem is becoming urgent. The city water-treatment works where the wastewater goes is tired of our brown tea and sludge. It claims that ultraviolet light can’t sanitize our water because of its cloudiness. They’ve already cited us once. Maybe, we have three months before we are cut off.
Our supplier of boiler chemicals agreed to send several samples to its lab in Houston to evaluate the flocculent. That’s 1,000 miles away! Unfortunately, I discovered the pH had to be neutralized to ship the samples; I think this ruined the tests.
What do you think we should do?
Send us your comments, suggestions or solutions for this question by October 13, 2017. We’ll include as many of them as possible in the November 2017 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.