Hysys dew point and bubble point results differ, databook vs. envelope 1

[JackWhite]

Good day, gentlemen and ladies,

I'm dealing with gas mixtures, involving light hidrocarbons, water and sulphur oxides. As an experiment i created a stream with a composition of methane 0.4%mol, ethane 0.3, water 0.2, and SO3 0.1. P=1 bar, Flow=1 kmole.
The property package PR seems gives very close results compared with PR-sour, PR-TWU, SRK and so on.

The envelope for this stream gives a dew point (no bubble point) that looks a bit weird. Reading about this i took Mr. Joerds' suggestion and calculated in a case study, setting the vapour phase=1 and pressure 1 bar as independent vars, and temp as a dependent var.

The results are puzzling, to say the least. Temp keeps rising with pressure indefinetly, and the graph looks nothing like the envelope (values differ a lot).Same pb for a bubble point calculation ((20 bar,60 dC)on envelope; (20 bar, -100 dC)on table).

Wich do i believe?
Thanks in advance, hope this issue isn't too silly to deal with.

 

[PaoloPemi]

possibly there is a numerical problem with your simulator (since calculated dew point should match the dew point line in phase envelope) however those packages (PR or SRK with conventional one-binary-parameter mixing rule) do not appear suitable to model that VLE.
With polar fluids it would be better to select for fugacities a model based on complex mixing rules, however these models are difficult to solve at high p, t and plotting the entire phase envelope may become difficult for very non ideal mixtures...
If you attempt to model with PR or SRK and convenctional mixing rules a simple binary as Water 0.9 and Methane 0.1 you'll find a unusual behaviour and possibly convergence problems,
there are many examples of unusual behaviour for binaries as CH4-H2S etc. see

http://www.prode.com/en/phaseenvelope.htm

 

[JackWhite]

Thanks for the quick reply, Paolo. The thing i found most puzzling was the huge difference between the two ways of calculating the dewpoint and bubblepoint curves. Here is a printscreen of the bubblepoint curves. What do you make of it? I think every value over 10 bar could be ignored, but the differences are still there. I'll try some more property packages and return.

Regards,
JW

 

[PaoloPemi]

I have not access to your simulator (I do my calc's in Excel with Prode Properties which apparently don't show problems in your test case with PR or SRK and Kij 0.5 for water-hydrocarbons).
As written in previous post I think this is not a honest test case since selected thermodynamics are not suitable for that mixture.
However you could verify if the problem appears at very low pressures (the beginning of the phase envelope), if that is TRUE possibly the difference depends from different VLE initializations, if not (it works at very low pressures) it could depend from the different ways the procedure adopts to calculate a dew point value (possibly initializing at specified p or t and then attempting to converge) and the phase envelope (starting from very low pressure and then proceeding from converged values) , the second case being certainly more stable.

 

[JackWhite]

You are right, Paolo. I hoped that Hysys would be able to predict the correct buble points curves and dew point curves for flashing a mixture while using a property package like sour-PR (deals with water, SOx, and HC mixtures). But the truth is it wasn't designed for that. Apparently AspenTech recommends using another of their products for such cases (Aspen Distill).

There is, however, a conclusion to be drawn from this. After testing a lot of property packages and comparing the results, i found that PR, PR-sour give the same results for bubble points of such a mixture between a 0-5 bar range, and Kobodi-Danner gives simmilar results with these two. Another important thing to note is to not trust the envelope Hysys prints for the same mixture. Calculating it yourself seems to give more accurate result, since the envelope looks as if it ignores the presence of water and SO(therefore the results differed with +50 dC).

In conclusion: envelope bad, var calculation much better but not best. In need of accurate dewpoints and bubblepoints for a complex mixture, Hysys is not your tool.

Thanks again Paolo for your posts, they guided me in the right direction.