Removal of CO2 with MDEA

[Zoobie]

I have been struggling with simulating (HYSYS) the sweetening of a 6mol% CO2, 2000 ppm H2S gas using 45wt% MDEA (operating at about 800 psig and 100F). Sweet gas has to be 2% CO2. The results that I am getting appear to be garbage. I am chalking this up to the fact that this is a tertiary amine used for bulk removal of CO2 (as per pg 21-10 in the GPSA data book, 10th ed.). I do not have a workaround for this yet.

After inputting my tower data I get CO2 stage efficiencies below 3%. This seems odd. Of course manually changing the efficiency makes a big improvement...duh. The H2S efficiency doesn't seem to far off...about 50%. This is for some ballpark estimates for max throughput through the unit but right now I don't think I am even in the parking lot.

Here are the questions that I have:
1) What would be a reasonable efficiency for CO2 per stage? 30%, more, less?
2) In the event the simulator does not satisfy my needs, what would be a reasonable mol/mol loading for CO2/MDEA in this sort of system?
3) What would be a reasonable BTU/gal for the still reboiler? GPSA book states that 'requires only a modest heat input'. Is this substantially below 800 BTU/gal?

Thanks for any help you can provide.

 

[Montemayor]

Zoobie:

As with all simulation programs, the accuracy and credibility of the results rests on:

1) the experience and expertise of the individuals writing the algorithms and the simulation program;
2) the experience and expertise of the individual running the simulation program.

I'm going to assume you surpass the qualities of #2; that leaves the HYSIM program in question. This is not surprising if one contemplates what other simulator companies are trying to do (like Aspen, SimSci, ChemCad, etc.). These are all in competition and they strive to simulate "everything" that can be simulated. However, successful simulation does not depend on being able to write an algorithm or program an algorithm. It depends on KNOWING how to do it with successful accuracy. To achieve that standard, one needs experience - not in academics or program writing, but in operations and actual hard design. Simulator are good for some things - they can't be expected to do all simulations correctly.

For example, have you any experience with TSweet? Go to

http://www.bre.com/technicalpapers/...layAbstract=Y&FilterCategory=AMINE SWEETENING

and you'll get a truckload of technical, first-hand information and data on MDEA and CO2 removal systems and their design. I'm not pushing BRE no more than I'm pushing HYSIM. I've used both. Each has a niche. I know of the people who put TSWeet together and who have worked on it. I don't know who put HYSIM's CO2 removal algorithm together or who wrote it. Like any other engineering product, one has to always take into account the degree of engineering experience and know-how that goes into it - not the quality or amount of "C+++++" code that went into it.

Are you designing a CO2 system for your company? Or are you designing it for practice or study? If you are doing it for your company, your company is in trouble. I say this not to demean your engineering competance, but rather to point out that a serious design enterprise like this should always be done by experienced individuals who have done this before - successfully. Your questions indicate that you haven't, so I can only surmise that you are doing this as an exercise for the sake of getting experience and background. If that's the case, then I would get in touch with BRE and ask them the specifics. I'm sure they will be glad to furnish you with a trial TSWEET program you can try out and also answer your questions.

 

[Zoobie]

Thank you Montemayor. When I saw that you had responded to my post I was sure that I would get something valuable from it. I am not designing a system. We have purchased a skid mounted amine unit and are moving it to an area with higher CO2 content. I was asked to provide a ballpark estimate as to the amount of gas that could be processed with the existing still reboiler (as this is believed to be the limiting factor). DMEA is the most likely solvent for this situation. Unfortunately, I have found out the CO2 removal mechanism using DMEA does not seem to fall in line with the HYSYS amine package. After talking with some other people internally, the historical approach has been to fudge the stage efficiencies to match vendor amine data.

I have since contacted and amine vendor and will hopefully get some guidance as to typical amine loading and reboiler duty requirements...in exchange for my veiled promise to buy their amine.

As an aside, DB Robinson was a professor at my Alma Mater and his AMSIM (which HYSYS Amines is based on) is great...it just misses the mark with MDEA.

Again, if anyone has some rule-of-thumb amine loadings for CO2/MDEA I would appreciate it.

 

[Montemayor]

Zoobie:

If you go to the recommended BRE website, you’ll find your specific answer in a variety of flavors and conditions – specifically in a detailed abstract that was published and given at the GPA in 1988 over the almost-exact scope that you define. The actual amine loadings for CO2/MDEA are given in detail.

There, you’ll find that Dome's North Caroline amine plant – which was Canada's first plant conversion from DEA to MDEA - was a reported a success, performing smoothly after being debottlenecked. Performance data taken for the MDEA absorber at several amine flow rates compare with safe accuracy to values calculated by the TSWEET program that was used as the basis for the conversion.

In defense of D.B. Robinson and his AMSIM program, I’m sure his scope of work involved writing a simulation program, not an Amine plant design program. There is a big, big difference between simulation programs and design programs – unbeknownst to may college professors and Chemical Engineering students. We find out the difference in our first real-world process design assignment when the words “liability” and “credibility” come up during engineering management discussions. Many academic types within the AIChE still persist in generating simulation results and labeling them “design” results. This is easy and practical for them, since their product would never be taken seriously and built anyway. Many of them have never had to confront the infamous question: “We have to make a decision on building this project and can’t waste anymore time; do we use your computer print-out answers or don’t we?”

Now that I know the background of your query, I’m more convinced you should be getting in contact with people like those at BRE who have done what you are proposing and have the credentials to prove it and discuss the results with you. As you can see by their publications and references, the Amines loading data is real and empirical – not simulated or divined. Amine vendors are good sources for physical properties and other values. However, they often have little or no experience in actual, operating Amine plants and the results of process changes or designs.

I wish you a lot of luck in carrying out your assignment. I think you’re on the right path and asking the right questions. In many ways I envy you because it is a challenging project where you can contribute a very useful and profitable work effort.

Good Luck.

 

[egber]

Zoobie,

Please note that MDEA is not the most appropriate ethanolamine to remove CO2. MEA, DEA and DGA are more suited for CO2 removal. MDEA is best for H2S removal while leaving the CO2 in the gas (selective H2S removal). For CO2 removal MDEA requires a promotor or activator which means a blend of amines.
As an alternative for Hysis or TSWEET which is now part of PROMAX, you can also look at Protreat. This is the only stand alone package I am aware of. Important is as well that it is a rate based program. So you do not have to bother about tray efficiencies. You can find some information about Protreat on

http://www.amineexperts.com

regards

 

[Zoobie]

Thank you for the reply egber. If you look at the papers that Montemayor pointed me to than you will see that MDEA has been applied to bulk CO2 removal. I actually found that it was being used on similar gases to what I was interested in. Since posting this last month, I have worked a bit closer with an amine vendor and they have recommended a solvent that is primarily MDEA (with a few snake oils thrown in). I won't be looking at other simulators...the expense does not warrant it for what I need it for. I have been satified with using vendor data to 'calibrate' my HYSYS models.

 

[AggieCHEN04]

Which thermodynamic model are you using for the amine? In 2004 (and probably earlier), there are two models: the Kent-Eisenburg, which is more correlational and the Li-Mather, which is more empirical. The default for HYSYS is Kent-Eisenburg which I've always found extremely difficult to use. I find that the Li-Mather model works much better for matching data from the amine vendors' simulations.

 

[Zoobie]

Thank you very much for your reply AggieCHEN04. Very useful. I have wondered what the difference between the two models is for some time. HYSYS documents it, but it is much more helpful to get some real-world opinions/experiences.