CAPE-OPEN

I have a problem with using Refprop in COFE. I have set up a water-ammonia mixture through the Refprop Cape-open ppm, which is working fine for direct calculations, but apparently not in iteration loops. I.e. for units which cannot be calculated sequentially I get the error message:

"error: calculate failed for unit HeaterCooler_8: CalcEquilibrium failed: CalcEquilibrium failed: CalcEquilibrium failed: [TPFLSH error 219] outer iteration loop for composition in 2-phase state did not converge"

According to Refprop documentation, [TPFLSH error 219] is simply a failed-to-converge error message. Could this be caused by the data (or lack of) that is sent to the refprop routine?

Kristian

This is a flash convergence failure and is probably due to a bad iteration guess. Can you send me the fsd file in my private mail?

Sent example file to jasper@amsterchem.com. Thanks for quick reply.

I have not made any effort to provide good first guesses for the iteration. However, when using the built-in property packages(TEA) with Peng-Robinson, this and alot more complicated flowsheets easily converge with the same inputs. Unfortunately there are too large deviations between this and refprop calculations to use rely on them as is.

Your initial specification is fine (at least in what you sent me; there is no recycle, so no initial guesses are used).

The problem is simply that the flash at X = {water = 0.3, ammonia = 0.7}, T = 30 C, p = 2 bar does not converge. This gives the heat exchanger no starting point.

If you enter this in the NIST front-end you will find the same.

Let us forward the problem to NIST. I will get back to you on this.

Thanks Jasper, you are of course right, the initial state in question here (30 C, 2 bar) is by chance the "dead" state of the process which is why I confused it for a iteration/initial guess problem.

I am impressed by the level of support you provide, and not only because this problem is caused by a 3rd part.

Kristian

I thought I had posted the follow-up already, but I do not see it. So I am posting it again:

Please see http://www.boulder.nist.gov/div838/theo ... EFPROP.HTM

Quoting Eric Lemmon of NIST:

But ignore the stuff at the top and the blank space right below it, and scroll down to the very bottom. There’s a new ammoniaL.fld file that can help fix these problems, but it comes with a little bit higher uncertainty in the mixture calculations. One of these days I’ll get the fit finished for this system and then all of these problems will go away!

I checked and these ammonia data indeed address this particular convergence failure. A more permanent fix should be available in the future in the form of a new data fit.

Thanks again, problem solved.