Flowsheet fails to solve with COFE 2.6.0.31 update

[HenkF]

Hi Jasper,
thanks for your advice. You are right that a PT flash isn't really suitable for a single component phase calculation.
But should such an erroneous result not show an error state for the unit?

enjoy!
Henk

[jasper]

The flash is a bit of a special situation. Normally in a process simulator, the flash is performed only once. The phases that follow are in phase equilibrium with each other, but are also the phase equilibria for the product streams, just by changing the phase fractions.

CAPE-OPEN however requires that the phase equilibrium is re-calculated on each of the product ports. So in addition to the main flash calculation that produces the separation, there is an additional flash calculation for each of the product ports. Typically the P-VF flash is used for that, but the solution of the P-VF flash is not unique. Hence, the temperature of the P-VF flash is post-checked, and a TP backup flash is issued automatically in case the temperature deviates too much from what is expected. This check of course has a finite reliability and you can choose to override this immediately with TP flashes.

TP flashes are of course wrong in the sense that, for a mixture, the amount of the incipient phase will never be exactly zero. Either it will not be formed at all, or it will be formed with a finite, non-zero, phase fraction. The latter of course is a problem if the unit operation that follows does not like this phase. For example, pumps do not like a trace of vapor in the liquid.

So the TP flash is already an exception.

The only case in which the TP flash would be off by more than a little bit is in case of a discontinuity of H with T, such as is the case for a pure compound (or more generally, for n-1 compounds in an n-phase equilibrium). The check that the TP flash gives the right answer is getting fuzzier (requires a sensitivity on the enthalpy balance) and to perform this only for a pure compound is also not straight forward (requires definition of when the stream is pure enough). It seems a bit over the top to make such a check, especially since it requires an enthalpy calculation that is otherwise not required.

Diagnosing the problem was rather easy: if you have clicked the stream button with the flash selected, you would have found that both the products were in the liquid state.