9 replies to this topic

[danago]

Im currently working on a project and have come across something that i am not fully able to explain. Using the envelope utility in HYSYS, i have come up with the following phase diagram for an 80% methane/20% n-decane mixture:



I am still quite early in my engineering degree and have only very recently started learning how to use HYSYS, so i am unsure about the shape of the curve. All of the other phase envelopes i have looked at have had a similar "dome" shape; however this one is clearly in a class of its own. What could possible cause the shape of the curve above? Is it likely to be a problem with HYSYS, or is there some actual physical phenomenon behind it?

Thanks in advance, all help is greatly appreciated,
Thanks,

Daniel.

PS. I am using the Peng-Robinson EOS

[Zauberberg]

It's definitely a HYSYS issue and is not related to the real phase behavior. I have encountered similar issues in the past, but mostly when Hypo-components or Crude Assays were used in fluid characterization. Your graph apparently shows only the existence of dew-point curve while the bubble-point curve does not exist. That is not possible.

The strange fact is that you have only two components in the system, and both are true compounds. Does the same thing happen when you replace n-Decane with some other, lighter compound?

[danago]

It's definitely a HYSYS issue and is not related to the real phase behavior. I have encountered similar issues in the past, but mostly when Hypo-components or Crude Assays were used in fluid characterization. Your graph apparently shows only the existence of dew-point curve while the bubble-point curve does not exist. That is not possible.

The strange fact is that you have only two components in the system, and both are true compounds. Does the same thing happen when you replace n-Decane with some other, lighter compound?

Thanks Zauberberg for the reply.

I currently dont have access to HYSYS to try different things (i am at home), but i do remember that when i made an 85% methane/15% ethane mixture the phase envelope was normal with both a dew point and bubble point curve.

What could possibly make HYSYS mess up a calculation like this? Would it be related to the fluid package (PR) i have selected?

[Zauberberg]

I have just checked your composition and I'm getting the same results. Phase envelope is "normal" for Mixtures of Methane with Ethane, Propane etc. all the way up to n-Octane, and then it becomes similar to yours.

Interestingly, if I remove Methane from the gas stream and use any other Alkane compound coupled with n-Decane, I'm getting quite reasonable phase envelopes. I remember from one of my J.Campbell's training lessons that Methane-based mixtures can show quite specific behavior (e.g. appearance of two liquid hydrocarbon phases) under certain conditions, but I forgot what was all about.

Anyway, an interesting topic. We have a simulation expert here (Joerd) but I haven't seen him on the forums for quite some time. Perhaps someone else from forum members can help us with this issue.

[danago]

Ok thanks, glad to see that i haven't put something into HYSYS incorrectly

Anyway, thanks again for your input, it is greatly appreciated. I will keep on looking for a specific answer and hope that somebody else knows exactly what is going on.

Cheers,
Daniel.

PS. I just registered on this forum; it looks like a great resource!

Edited by danago, 02 May 2010 - 07:32 PM.

[joerd]

Anyway, an interesting topic. We have a simulation expert here (Joerd) but I haven't seen him on the forums for quite some time. Perhaps someone else from forum members can help us with this issue.

Thanks, pretty busy times here, which is good, I suppose.

Hysys' phase envelope routine is not always very robust, so it sometimes gives strange pictures, especially with non-condensibles (hydrogen) or with very different components (which you have here). I suspect that the utility uses a shortcut flash method which sometimes simply fails. Also, here, Hysys thinks you'll have two liquid phases at low temperatures, which may or may not be true, I don't know, but it does give trouble in the calculations.
One thing you can do, is to generate property tables for the stream. Set up one independent variable as Vapor Fraction, state = 0 and 1, and the second independent as pressure or temperature (try which one gives you the best curve, sometimes you have to do them both and paste everything into Excel to get a nice curve. Outside the phase envelope the flash will fail (there is no bubble / dew point to be found there), but that is not a problem.

Hope this helps

Attached Files

•  80methane20decane.bmp   2.08MB   130 downloads

[danago]

Anyway, an interesting topic. We have a simulation expert here (Joerd) but I haven't seen him on the forums for quite some time. Perhaps someone else from forum members can help us with this issue.

Thanks, pretty busy times here, which is good, I suppose.

Hysys' phase envelope routine is not always very robust, so it sometimes gives strange pictures, especially with non-condensibles (hydrogen) or with very different components (which you have here). I suspect that the utility uses a shortcut flash method which sometimes simply fails. Also, here, Hysys thinks you'll have two liquid phases at low temperatures, which may or may not be true, I don't know, but it does give trouble in the calculations.
One thing you can do, is to generate property tables for the stream. Set up one independent variable as Vapor Fraction, state = 0 and 1, and the second independent as pressure or temperature (try which one gives you the best curve, sometimes you have to do them both and paste everything into Excel to get a nice curve. Outside the phase envelope the flash will fail (there is no bubble / dew point to be found there), but that is not a problem.

Hope this helps

Thank you very much for your time. That helps a lot, gives me something to work from

Just one question though; how did you determine that HYSYS thinks there will be two liquid phases at low temperatures? Is that just an observation made from the shape of the curve, or is there somewhere in HYSYS that specifically implies the presence of two phases?

Thanks again.

[joerd]

Just one question though; how did you determine that HYSYS thinks there will be two liquid phases at low temperatures? Is that just an observation made from the shape of the curve, or is there somewhere in HYSYS that specifically implies the presence of two phases?

Thanks again.

If you go to the Phase Envelope utility, and look at the results Table (not the graph), one of the columns is "phases", which most of the time is L-V, but sometimes shows L-H-V (light liq. - heavy liq. - vapor)

[amitkatyal.asim]

Im currently working on a project and have come across something that i am not fully able to explain. Using the envelope utility in HYSYS, i have come up with the following phase diagram for an 80% methane/20% n-decane mixture:



I am still quite early in my engineering degree and have only very recently started learning how to use HYSYS, so i am unsure about the shape of the curve. All of the other phase envelopes i have looked at have had a similar "dome" shape; however this one is clearly in a class of its own. What could possible cause the shape of the curve above? Is it likely to be a problem with HYSYS, or is there some actual physical phenomenon behind it?

Thanks in advance, all help is greatly appreciated,
Thanks,

Daniel.

PS. I am using the Peng-Robinson EOS

Such a curve is not derived for 80 % CH4 and 20 % n-decane with binary interaction parameter =0 but such curves are pretty commom with high concentrations of non-condensable gases like N2. I have given one such curve in the home page of my website, www.eq-comp.com. EQ-COMP is a VLE software that I have developed which is described in detail on the website www.eq-comp.com. Such curves indicate the presence of LLV equilibrium for the mixture.

Regards

Amit

[PaoloPemi]

I do not think it's a numerical problem, I get similar results from Prode Properties (see attached file)
CH4 0.8
C10H22 0.2
K12 0.043609 (as for HYSYS standard settings)
model Peng Robinson with standard correlation
this phase envelope probably has 2 critical points (I calculated the VLE, not LLE), it is not uncommon to get a unusual phase enevelope, see for example
http://www.prode.com...aseenvelope.htm
possibly if you change something in your specifications you'll get a "standard " phase envelope (with one critical point)

Attached Files

•  PR_K12_0.043609_0.8CH4_0.2C10H22.jpg   211.83KB   36 downloads

Edited by PaoloPemi, 31 July 2011 - 11:26 AM.