I am a new user of DWSIM and just playing around with the program testing its features. I came across when mixing different liquids, for exampel Ethylene glycol/water and Ethanol/Water, that the mixture properties seem to be way of looking at Cp and density. I haven’t looked at the other properties so they might differ as well. I have been trying different thermo settings but the result is basically the same. Also tried “Use Ideal Mixing rule for liquid phase density” which gives better results on density but still not accurate enough (Cp still way of). Used ChemSep database components only.
You're right, I've been discussing this issue with Gregor Reichert (co-developer) and we haven't decided yet what to do here. Currently most of property packages use an equation of state to calculate Cp (Peng-Robinson, SRK or Lee-Kesler) and the Rackett correlation to calculate liquid density.
What happens is that the results of liquid Cp are bad for highly non-ideal liquids and even worse if we use the theoretical definition of Cp as calculated from the derivatives in the case of activity coefficient models. None of them (NRTL, UNIQUAC, UNIFAC) seems to give good results here.
The only remaining option is to do an ideal mixing rule with the experimental individual values from the ChemSep database, and it seems to give good results, but this hasn't been done yet.
My intention is to let the user choose the source of calculation from one of the three options described above.
Regards,
Daniel
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Jens,
in the meantime this promised new option was realised by Daniel now. You will find this within the next release.
There is a new option field in Property Package configuration parameter table where you can make your choice.
regards
Gregor
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Gregor, Daniel,
I've also been doing some testing with water glycol mixtures (Ethylene glycol, 50 / 50% with water, VOLUME fraction). I've found a strange thing, that when I fill in the mass flow it seems to work okay (and I did activate the ideal mixing rule in the propterty package conf table). But when I use volume flow instead of mass flow, something goes wrong. See the image. Allthough I use almost the exact same input data (mass flow on the left, volume flow on the right), something goes wrong in the calculation. The overall liquid properties go wrong when I use volume flow as input.
Regards,
Checked it and now it seems to work better Daniel. At first I thought it didn't work because after changing mass flow to volume flow input, the calculated outputs downstream did not update correctly. But after hitting F5 it calculated again and better. Thanks!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi
I am a new user of DWSIM and just playing around with the program testing its features. I came across when mixing different liquids, for exampel Ethylene glycol/water and Ethanol/Water, that the mixture properties seem to be way of looking at Cp and density. I haven’t looked at the other properties so they might differ as well. I have been trying different thermo settings but the result is basically the same. Also tried “Use Ideal Mixing rule for liquid phase density” which gives better results on density but still not accurate enough (Cp still way of). Used ChemSep database components only.
BR
Jens
Last edit: Jens 2014-09-29
Hi Jens,
You're right, I've been discussing this issue with Gregor Reichert (co-developer) and we haven't decided yet what to do here. Currently most of property packages use an equation of state to calculate Cp (Peng-Robinson, SRK or Lee-Kesler) and the Rackett correlation to calculate liquid density.
What happens is that the results of liquid Cp are bad for highly non-ideal liquids and even worse if we use the theoretical definition of Cp as calculated from the derivatives in the case of activity coefficient models. None of them (NRTL, UNIQUAC, UNIFAC) seems to give good results here.
The only remaining option is to do an ideal mixing rule with the experimental individual values from the ChemSep database, and it seems to give good results, but this hasn't been done yet.
My intention is to let the user choose the source of calculation from one of the three options described above.
Regards,
Daniel
Hi Jens,
in the meantime this promised new option was realised by Daniel now. You will find this within the next release.
There is a new option field in Property Package configuration parameter table where you can make your choice.
regards
Gregor
Hi Gregor, Daniel,
I've also been doing some testing with water glycol mixtures (Ethylene glycol, 50 / 50% with water, VOLUME fraction). I've found a strange thing, that when I fill in the mass flow it seems to work okay (and I did activate the ideal mixing rule in the propterty package conf table). But when I use volume flow instead of mass flow, something goes wrong. See the image. Allthough I use almost the exact same input data (mass flow on the left, volume flow on the right), something goes wrong in the calculation. The overall liquid properties go wrong when I use volume flow as input.
Regards,
Guillaume
Hi Guillaume,
Thanks for informing us about this bug. It was also fixed for the next release.
Regards
Daniel
Checked it and now it seems to work better Daniel. At first I thought it didn't work because after changing mass flow to volume flow input, the calculated outputs downstream did not update correctly. But after hitting F5 it calculated again and better. Thanks!
I think that volumetric flow input doesn't fire the flowsheet calculator, only the stream itself. I'll take a look.