No replies to this topic

greetings,

i hv been wandering in this forum since few months ago. i find it very beneficial for chemical eng students like me. thanks to all the experts and professionals who share their precious knowledge.

to complete my degree in chem eng, now i am doing my plant design project. my plant produces 1,3-butadiene. i found in literature (Ullmann's handbook), to separate butadiene from C4s, industry uses extractive distillation. it seemed so nice and easy.

however, when i do the simulation on hysys, n-butane has lower boiling point than butadiene even after the addition of solvent (extractive distillation approach) which is the opposite from what the book tells me (butadiene should has lower boiling point)...

what should i do now??

i have tried all the fluid packages in HYSYS, but nothing happened.

i'm looking forward to the advices. thank you..

[joerd]

You should do more research. Try to find a good description of the process, including solvent composition, flow rate, column pressure and temperature. If I read http://www.chemical-...20industry.html it seems that the solvent in the first column goes to the bottoms stream with the butadiene, then is stripped in the next column and gets recycled.
You will probably need NRTL as the fluid package, the usual PR EOS doesn't have the necessary binary interaction parameters for butadiene and solvent (check all this before you start simulating the column, if the thermodynamic basis is wrong, your simulation will be worthless).
But, to reiterate my point, do more reading first. Google is your friend - http://books.google....nG=Search Books

thanks for the response.

yes, butadiene and solvent are the bottom product of the first column while butanes and butenes are the overhead product. that is what the literatures tell me.

all the components follow this rule, except n-butane. that is where the peculiarity is: n-butane also comes as bottom product along with solvent and butadiene. so i assume the problem is in hysys.

i hv tried using NRTL (and the rest of liquid packages), but that doesnt help.

[joerd]

Did you check the interaction parameters for the NRTL package? Hysys doesn't have them so you need to estimate them.

[GGuerra]

Hi joerd and dariusity,

me and my colleagues are watching this forum closely. We're also "building up" a plant to produce 100000 t/y of 1,3-bd using DMF (dmiethylformamide) as the extractive solvent. We're graduating as chem engineers an that's our final job.

We're certainly going to face the same problems. Is NTRL a package tat has to be installed in Hysys? Our 1st problem is to install hysys...

We're using as guide the handbook of a brazilian enterprise who build up its plant here and some SRI / Chemsystems stuff. The former's capacity is of 100000 t/y and the later around 50000.

Hope we can exchange some information. To get in touch, just send an email to guilherme.b.guerra@gmail.com

See ya.

hello,

I got the same problem as above. my simulation did converged but the major problem is the component recovery for above is 0.6 (iso-butane, iso-butene and 1-butene) and for the bottom is 0.5 (1,3-Butadiene, n-butane, cis-2-butene, trans-2-butene, ...). So, can anyone help me on how to get the component recovery for both overhead and bottom at least 0.7 or 0.8?

p/s: iso-butane, iso-butene and 1-butene goes to top and i already tried in hysys also in pro II.


thanks.

[abhi_agrawa]

Hello,

I will reiterate what Joerd has said: Check the interaction parameters. Go to your library, and search in Dechema or Journal Of Chemical and Engineering Data. If you are not clear about this, I'd strongly suggest you to go through the following papers:

1) Carlson, Don’t Gamble with Physical Properties for Simulations, CEP, October 1996
2) Agarwal, et al, Uncovering the Realities of simulation Part I, CEP May 2001
3) Agarwal, et al, Uncovering the Realities of simulation Part II, CEP June 2001

Hope this helps,
Abhishek

[quelareine]

cool post! very interesting!
simulation rachat de credit

That's a very good topic and interesting work. Thank u so much for sharing.

simulation rachat credit

[shinichikudo]

simulation credit auto
Thanks to abhi_agrawa's papers I got it over.
Appriciate

[Johny4]

That's pretty good.

simulation
credit

[Padmakar Katre]

Dear,
I tried with Wilson (Activity Coefficient Model) and I had simulated this before. I don't remember exactly but the same thread was discussed in cheresources before.
http://www.cheresour...amp;hl=monolina I hope it's similar. Just give me your id I will send you the Hysys File which I sent it in reply to the above post.

[monkeyking]

Hello,

im having same troubles here.. any inputs would be great!

Thanks!
simulation assurance vie

[abhi_agrawa]

Padmakar,
Wilson activity coefficient model is no doubt a good choice for modeling liquid phase non-ideality as it is a simple but quite robust model. However, it suffers from a very serious drawback: it can not predict two liquid phases in equilibrium. Wilson's model is not equipped to handle two liquid phases. So in the process if there is a chance of formation of two liquid phases, then Wilson model will not predict the same. I am not sure of this process, so I do not know if there is any chance of two liquid phases anywhere in this process.

For anybody doing this simulation I'll also insist that the person should get the solubility data (it should not be too difficult to obtain) and validate the coefficients before starting the simulation. This is important to understand the accuracy of the method you are going to use. My choice would be to go with NRTL.

Hope this helps,
Abhishek

[Lovella]

Well said Abhishek. Great job.

Thanks a lot.


Simulation pret

[batman009]

wonderful! thanks for the info..

simulation assurance vie

[Padmakar Katre]

Dear Abhishek,
Yes thats true that Wilson Activity Coefficient model doesn't support the VLLE(Vap-Liq-Liq Equil.) But as far as this simulation is concerned there is no issue to use Wilson as even I tried with all the activity coefficient models but the results are not satisfactory so I tried with the wilson and I compare the results with the Test-Run which I found match. So I recommended Wilson. Just waiting for your comments.

[abhi_agrawa]

Dear Padmakar,

I have no idea of this flowsheet, so really I can not comment if there will be formation of two liquid phases in the system. But what surprises me is that you did not get satisfactory results using other Activity Coefficient Model. My approach for validating any Activity Coefficient Model would be:

1) Try to get the VLE/VLLE of major components in the system, if possible then at the temperature and pressure similar to in the process. While gathering this data, try to get as many sets of points as possible. This step may be cumbersome, but is essential to the entire exercise.

2) Using the Activity Coefficient model chosen, predict the VLE/VLLE and if the prediction is not good, then estimate the coefficients. If you have data at multiple pressure/temperature points, use the regressed coefficients to predict at least one set.

3) Once the above two steps are set, you can use the coefficients thus obtained in your simulation.

If you are using a commercial simulator, then I'd strongly suggest that you go through the above step. I'll give you a small example that I faced with ASPEN Plus. I was trying to reproduce some result of gas separation simulation which was done quite some time back in my organization. The original simulation was also done in Aspen, however, somehow we lost the simulation file. We still had the simulation results. I put in a lot of effort but could not get to match the data. Then I simulated the flowsheet in HYSYS and we got a very good match. Then people in my organization started complaining that Apen in no good for gas processing (they conveniently forgot that the original simulation was in Aspen). I took this problem and then analyzed, only to realize that the Binary Interaction Parameters for Methane and H2S were missing. And that was a big problem, because we were guaranteeing H2S slippage. When I inserted Interaction parameters, which were regressed from data in DECHEMA, we had a very good match with the ole simulation as well as actual data.

Moral of the story is that it is easy to say that some thermodynamic models are just not good enough, without understanding the issues. So always when you do your simulation understand the thermodynamic method you are selecting. Look and ensure you have all the Interaction Parameters/ Activity Coefficients beforehand. Be sure that you know the limitations of the method you are using/going to use. Also beware, Aspen had the BIP's (originally) and at some point of time they removed the BIP's for CH4/H2S from the database (and was made available in Aspen Properties, which you had to buy!!!).

Hope this clarifies few things.
Let me know of your comments/similar experiences.

abhishek

[jkchiche]

Thank u so much for ur sharing.

simulation credit

[maliguo]

what "abhi_agrawa" saying points to the fact of extracitive distillation. As the solvent is added to the C4 fraction,relative volatility enlarged, so robust calculation need exact equilibrium data, otherwise it will results in error results. I have been doing the process simulation of extractive distillation, as the simulator lacks of phase equilibriumdata, so I have to go to libraries many times to search for vapor-liquid equilibrium data. It's very necessary!