FpcTP with Vapor Liquid Equilibrium at specified unit model

[lucbrinkman]

I'm stuck with a modeling problem in IDAES. I was using FpcTP state variables. This defines the flow of each component for each phase independently, so there is no need for vapor liquid equilibrium (VLE) calculations in the model. Now, I want to do VLE calculations in one specific heater unit model. I can't get that to work with FpcTP.

If I add
"phase_equilibrium_form": {("Vap", "Liq"): fugacity}, for the H2O component,
and

"phases_in_equilibrium": [("Vap", "Liq")],
"phase_equilibrium_state": {("Vap", "Liq"): SmoothVLE},

for the whole thermo_props config,
then VLE is calculated at every unit model.

If I leave these three terms out completely, or just leave out "phases_in_equilibrium": [("Vap", "Liq")],, then VLE is calculated nowhere.

I also tried both has_phase_equilibrium=True and False for the heater block, but that didn't seem to change anything. In fact, I couldn't find any difference in the m.fs.heater.pprint() between the True and False settings.

As I couldn't get calculating VLE in one specific model to work, I tried switching from FpcTP to FcTP. While this does technically work, it increases the calculation times and also seems to give me convergence problems for most cases. This is probably just because my model is poorly scaled. I tried the autoscaler but that didn't help, and would prefer not to spend days setting up my own scaling routines. Plus, from a first principles perspective, I would prefer to simplify the calculations at the source, which would mean using FpcTP and not calculation VLE at each unit block.

What is the supposed workflow for using FpcTP with VLE only at specified unit models? I tried everything that came to mind but I probably missed something. I'll try your suggestion in a minimal example and see if I can get it to work that way, and then get back to you.

[dallan-keylogic]

It appears that the has_phase_equilibrium flag does nothing in the modular property framework at present. I'll open an issue, but it may be a while before it gets fixed. Feel free to open a PR if you'd like to handle this sooner.

Which unit models do you want to exclude phase equilibrium in? You would also need to pass the option material_balance_type = MaterialBalanceType.componentPhase to whatever unit model you wish to avoid phase equilibrium in.

As a workaround, you could have two separate ParameterBlocks, one defined with "phases_in_equilibrium": [("Vap", "Liq")] and one without it. Use the one with phases_in_equilibrium defined for the unit models you want flash calculations in and the one without it for the unit models you don't.

For scaling, the new tools aren't fully rolled out yet, but you can probably get pretty far with the old scaling tools. See this test for how to set default scaling parameters for a property package (if you make one with VLE and one without, you'll need to scale each of them). There are some additional variables, like heater heat duties, that also need scaling factors. Note that the fugacity form of VLE is somewhat poorly scaled by default. log_fugacity is generally better.

[lucbrinkman]

Thanks a lot, I think these are some very useful suggestions. I'll try them out