12 replies to this topic

[Dacs]

Hi!

I'm trying to come up with a design (this is more for my self-benefit since my workload is kind of slow lately) for an absorber using MDEA as the amine solvent, using hysys for the simulation. Unfortunately, my company doesn't have that much experience handling FEED in amine systems, so the available literature (and experience) is limited.

So for starters, I'm trying to reproduce a result of an existing simulation (apparently not produced by hysys) just to get the gist of doing things. Also I tried reading up the manual for the Amine package as well. It says, the package uses real trays since apparently the tray specs affect the performance of the column.

So here are my questions:
1. How can I come up with a design suited for packed column? If I were to design a tray column, it's pretty quite straightforward since you can pretty much input all the relevant data. However, in my review of the projects that I've been involved with, I have yet to see an amine absorber that uses trays. All of them (so far) used structured packing. Normally, I need to establish the number of theoretical stages to satisfy a certain separation, then multiplying it to HETP for a specific packing to determine the height of the bed. If hysys uses real trays, will it be enough to just set the efficiency of both CO2 and H2S to unity, run the column, determine the # of trays and multiply it to HETP to get the bed height? My concern here is that the efficiency of H2S doesn't match of CO2. Will it be reasonable just to go after H2S (if that's the component that I'm after) efficiency and ignoring the CO2 efficiency? Or is this a sound approach in the first place? Something tells me that I cannot just simply set the efficiency to unity since amine treatment processes also involve chemical reactions and the concept of equilibrium doesn't necessarily hold.
2. Will it be just enough to use the packing data (void frac, HETP, etc) and just input it straight to hysys and adjust the # of stages instead?

Thank you guys

[Dacs]

I tried tinkering with hysys and changing packing internals, but the only thing that seems to change the tray efficiency is the diameter (regardless of packing type/settings).

When I changed the internals to tray instead, I changed the weir height/length and the efficiencies change. Changing the tray type (or its settings) doesn't do anything. It seems that the efficiency is a function of tray liquid holdup.

So for added questions: How can I even compare the performances of different packing materials when the only thing that matters in the column efficiency would be the diameter (for packed columns)?

[aliraza]

Dear Dacs,

I have also come across same questions which you asked while running the simulation and doing preliminary sizing of Amine Sweetening Package. I came to know that Pro Max is better than Hysys for Amine Sweetening & Sulfur Recovery Packages (especially). And I found a technical paper on Amine Sweetening Package of BR&E Research Institute which owns Pro Max Simulation Software. That technical paper shared alot of important parameters on which efficiency of Amine Tower depends. It also discusses some equations which are helpful in sizing the unit. I want to share this paper with you and hope that you will inform me after you find it fruitful.

I must appreciate if you start knowing the basic equation involved in sizing the Amine Tower thats how you will find "HOW THE THINGS HAPPEN IN AMINE TOWER WHILE USING MDEA", and after that please make a spread sheet for basic equations and then you run the simulation. I am sure that all the queries will be resolved and you will enjoy while running the simulation then.

Regards

Ali Raza

Attached Files

•  Amine_Sweetening_Process_Design.pdf   112.78KB   1661 downloads

[Andrei]

Gentlemen,

I've read all of your concerns regarding real vs. theoretical trays and pressure correlations and so on.
Nobody raised, or I didn't noticed anybody to raise the issue of the validity of the Amine thermodynamic package.

I think this should be your starting point. And this is because the Amine package gives very poor results and absolutely off any practical data. The main reason is an inaccurate prediction of the heat of reaction at CO2/H2S absorption, and from here will result higher amine circulation rates than should be in reality and much higher heat required to regenerate the rich amine. In MDEA's case you can get double or even triple circulating rates from Hysys than you will get from a licensor or from operation data and of course the same kind of proportion for regeneration heat required.

I suggest just to take a book like GPSA Databook, and compare your results with what you can find there.

The same amine thermodynamic system can be found also in PRO/II and I can tell, yields the same poor results.

In Hysys there is a system called DBRAmine that it is claimed to be based on real plant operation data and supposedly give better results. Unfortunately this is not working properly even in last Hysys version.

Somebody mentioned ProMax here as something that works. Yes it is true, it seems to give the best results among all existing simulators on the market. ProMax was innitially concieved for Sulphur simulations and it seems to give good results also for CO2.

Now about the number of stages required in the absorber. It is difficult to calculate a number without having good thermodynamics, and without being able to accurate predict the pressure drop in the tower. Usually this number comes from plant experience or licensor data and you have to play with it to fine tune your results.

Regards

[smalawi]

Hi,

if you are only looking for a "feel" not detailed design then the equipment size should not change considerably and more important is the number of new equipment needed. Hysys should do,

Hysys / Pro II are as good as the thermo packages they have, even Tsweet/Promax have issues when the feed gas has significant concentration of CO2. Leave the detailed design to the experts, they know what they do and don't totally relay on the simulation software !

with Hysys, the tray internals should also matters, the weir height should play a role in the approach to equilibrium calculations that is in addition to the column dia (both changes the amine residence time in the tray), there is an amine tap in the column environment, check that (I have the latest unisim and its there). try to do a test run and match your unit results first.

be careful if your system has ammonia, Hysys has issues with it and give totally bad results, check aspen support and also check eng-tips.com for a more comprehensive explanation.

the column internals as service specific and more by experience, I've seen both trays and packing. in your case as i indicated, you need general dimensions so use actual trays and when using packing check with the packing vendor for the HETP for this service (I know the HETP should be independent of service but not in the case of amines, its more of a reaction than mass trasfer)

hope this helps,

cheers,

sm

[Dacs]

Thank you all guys!

I've been tinkering around Hysys and ProMax this past few days and I've also considered all your replies and I have to agree, this is still out of my league

Although I must admit after reading some literature, I've learned quite a lot especially with regards to rule of thumb when dealing with this system. Unfortunately, all of this pale in comparison with something that I do not have (yet): experience.

Another thing, if you have a working simulation for an amine absorber (with trays), how would you go on converting the column into a packed column, for comparison?

While I think I can handle designing a tray column, I'm still at a loss when dealing with packed columns.

I apologize if I ask too much questions, I'm just trying to gain something from here. And thanks again guys!

[smalawi]

Hi,

same principles, just download kgtower and sulterpack and start from there,

Ludwig also has good chapters on column internals and you can get access to Kister distillation design books on Google books, good sources.

again expriance in design is based on process knowledge not just column internals

regards,

sm

[Dacs]

I think I need to elucidate my queries further so that I'll solicit the answers more clearly.

I've had some experience working with columns and I think I've acquired some engineering judgment when dealing with these things. And I think I'm already in the stage where I've come to realize and understand the rationale behind some design principles that I've applied in line with my work as a process engineer.

That said, I sorely lack operating experience since I've worked in an industry very far from Oil and Gas (float glass production) and I acquired my Oil and Gas experience from my present (and past) companies (all are EPCM companies)

Going back to my present issue, I'm trying to reproduce the specs from one of my projects in which the front-end part was done by another company. Using whatever design software that I have in my disposal (Hysys and ProMax), I still cannot reproduce faithfully the result.

For instance, the datasheet for the amine absorber from FEED indicates 24 ideal stages using Mellapak 250.x (with an HETP = 0.5 m from available literature) with a diameter of 3.8 m). I tried reproducing this using ProMax and yet I cannot come up with a design that satisfies the requirements (3 ppm H2S) at the given feed conditions. I've used 24 stages and even increasing the diameter (since as per datasheet specs, it exceeds the flooding limit). What's strange however, even if I add stages, it doesn't do any good, decreasing the H2S that is.

Maybe I'm doing the simulation wrong (which in my honest opinion I've covered so far) so I'm thinking that I may not have enough engineering judgment to really analyze the system and in that part, I may have overlooked something that keeps me from getting it right. So it's either, the FEED people screwed their design (which is very very unlikely since they are a well established engineering company) or most likely, I'm screwed

Anyone can give me an insight into the right direction? I'll upload the file, if needed.

While I realize that I'm asking too much from you guys, I really do appreciate if you can give me pointers on this.

Thanks!

[Dacs]

I spent my whole day doing this simulation and here are my observations:

1. (Refer to attached clip): All feed flow/condition/composition and tower config fixed based on FEED docs, I cannot attain the proper H2S spec (6 ppm compared to 3 ppm from FEED) and CO2 slippage (0.48% compared to 0.94% from FEED). Increasing the # of stages won't help since it seems the H2S concentration stagnates in the middle of the column (reached an equilibrium wrt to kinetics?) and increasing the # of stages doesn't do much (I've tried that).
2. While lowering the lean amine temperature (36°C from 46°C, keeping the sour gas temperature still fixed at 30°C) helped (reduction to 3.55 ppm from 6 ppm), this is not inline with the results from the FEED docs.
3. I may be able to manipulate other parameters (amine loading, % amine in solution, etc), but as with #2, this will deviate with what specs that I have here, which at this point, should be in line with having a sound design that will satisfy the sweet gas specs.

At this point, I'm leaning towards
1. Lack of experience in my part: I may have overlooked something wrt simulation setup
2. Proprietary simulation used by FEED company: I'm quite sure that they used ProMax or Hysys for the simulation. If that's the case, why can't I arrive with a setup that is in agreement with theirs?

I realize that at this point that I may be asking some help that may involve some quite of analysis and effort. In that case, it's not my intention to burden you with this. But any insight will definitely be of help.

Thanks

Attached Files

•  column.JPG   63.35KB   263 downloads

[Andrei]

Dacs,

I just want to answer at your second question.
Most of the amine unit similations I've seen were "fixed", in other words the people "tricked" Hysys to make the results to look like something they know from their experience that works. You will never obtain the same results from a simulation if you don't do the same "tricks". Most probably this is the reason why you get different results.

[Aggie03]

Dacs,

I'm probably a little late on this response, but I thought I would try to point you in the right direction.

First, if your column is lean-end pinched, then changing the diameter and/or the number of trays will not help you at all. If you have ProMax, the software will calculate an approach to equilibrium for you. If it comes back 100%, then the only recourse you have is to increase the reboiler duty or add something such as a strong acid to aid in stripping. Either way, you need a cleaner lean amine.

Secondly, most amine absorbers are assumed to have a 3 to 1 real/ideal stage ratio or an efficiency of around 33%. If the design did indeed contain 23 ideal stages, this would equate to a column with almost 75 trays and over 150ft in height. I have never heard of a column of this proportion in industry.

It's more likely that the column has 24 actual trays and should be modeled as 8 ideal stages. This is much more inline with what is typical in industry. Too many trays could cause you to pick up too much CO2. While H2S is favored by the kinetics, CO2 is actually favored by equilbirum. Picking up too much could reduce your capacity to remove H2S.

The last thing I should mention is if the FEED study was done with a "proprietary" amine, such as Sulfinol, Ucarsol, etc., these solvents are not pure MDEA and contain additives such as piperazine, phosphoric acid, DEA, etc. These will all affect the kinetics as well as the equilbrium (and subsequently your results).

[Dacs]

Dacs,

I'm probably a little late on this response, but I thought I would try to point you in the right direction.

First, if your column is lean-end pinched, then changing the diameter and/or the number of trays will not help you at all. If you have ProMax, the software will calculate an approach to equilibrium for you. If it comes back 100%, then the only recourse you have is to increase the reboiler duty or add something such as a strong acid to aid in stripping. Either way, you need a cleaner lean amine.

Secondly, most amine absorbers are assumed to have a 3 to 1 real/ideal stage ratio or an efficiency of around 33%. If the design did indeed contain 23 ideal stages, this would equate to a column with almost 75 trays and over 150ft in height. I have never heard of a column of this proportion in industry.

It's more likely that the column has 24 actual trays and should be modeled as 8 ideal stages. This is much more inline with what is typical in industry. Too many trays could cause you to pick up too much CO2. While H2S is favored by the kinetics, CO2 is actually favored by equilbirum. Picking up too much could reduce your capacity to remove H2S.

The last thing I should mention is if the FEED study was done with a "proprietary" amine, such as Sulfinol, Ucarsol, etc., these solvents are not pure MDEA and contain additives such as piperazine, phosphoric acid, DEA, etc. These will all affect the kinetics as well as the equilbrium (and subsequently your results).

It's better late than never! Actually I shelved this one since I've been loaded with work lately, thank you for your reply however

The tower uses structured packing (Mellapak 250.X with total bed height of 12 m. Assuming that I got the HETP right of 0.5 m in which I believe so, this will equate to 24 theoretical plates. Or is it?

I used Real/Ideal Stage ratio of 1 and Murphree Efficiency of 33% in the simulation. To add, I haven't attempted to simulate the whole circuit yet, only the absorber part.

BTW, this tower employs split-flow configuration, presumably to regulate the CO2 slip.

I reviewed the feed docs and there's no mention of any "proprietary" amine or any additives (aside from the antifoaming agent). It's just plain 40 wt% MDEA.

Here's the basic flow diagram with the simulation values that I got. The tower is operating at 75 barG. As a reference, the CO2 = 666.3 kmoh/hr and H2S = 0.09 kmol/hr as per FEED HMB. This is quite far from what I got in the simulation (CO2 = 151 kmol/hr and H2S = 0.20 kmol/hr).

All MDEA goes to Stage 1 Feed. As far as I know, I've reproduced faithfully all the specs from the FEED documents. And still I'm at lost here

Attached Files

•  absorber.JPG   114.24KB   262 downloads

[nomoro]

Hi Dacs

Hope late than never, to contribute to yuor post:

If you are using ProMax do not forget that Packed columns are always modeled as ideal stages with no efficiencies, especially in design cases. You said that you used Murphree efficiencies of 33%!!

Also, residence time for packing must be estimated by the user and entered. Thus packed column results always have a degree of uncertainty due to the residence time parameter. For this reason it is always a good idea to run several cases using a range of residence times to determine the sensitivity to this parameter. Residence time cannot be determined precisely for any type of packing.

The residence time is usually between 5-7 sec for absorbers. For this service take care about the approach to equilibrium and do not forget to use the TSWEET Kinectic model in your absorber

regards

noralba