CAPE-OPEN

Hi Jasper,

I'm about to implement a workaround in DWSIM for those cases where the flash algorithm asks the EOS for properties of a specific phase and there's not a root for it. Currently DWSIM is throwing an exception, breaking the flash calculation, but there's obviously a better, more elegant way to handle this issue.

I was taking a look at TEA's help file and read that you return pseudo properties for phases that fit in the above problem. Can you share your thoughts about this problem, what's the best way to handle it and how you did it in TEA?

Thanks,
Daniel

TEAs uses an implementation based on

Mathias, Boston and Watanasiri, "Effective utilisation of Equations of State for Thermodynamic Properties in Process Simulations", AIChEJ, vol 30, no.2, pp 182-186 (1984)

It is not ideal though - the best solution is to avoid having to evaluate properties in regions where the phase does not exist. Michelsen describes in his papers how to get a good initial guess for a phase split by stability analysis. This works a lot better in practice.

I agree. In fact Michelsen tells us in his book "Thermodynamic Models: Fundamentals & Computational Aspects" to always use the root that gives the lowest gibbs energy, even if the phase is currently being identified as 'liquid' or 'vapor'.

I did code these methods in DWSIM before but ended up not using them because they're very slow. Maybe it is time to take a look at them again.

Thanks!