Allow only needing to define the present phases and components in inlet Feed blocks and allow the Mixer unit model to handle it

[tannerpolley]

Having working on the HDA flowsheet tutorial for a while, I am seeing a potential benefit and clarity for defining the feed/inlet streams with only the present phases and components. For example, in the HDA flowsheet, one inlet only contains liquid Toluene and the other inlet only contains vapor Methane and Benzene. It would be nice if each Feed unit model could take a different property package that only has the necessary phases and components. This would also allow the fixing the inlet states to be more intuitive since you wouldnt have to include very small value in each component and phase when in reality they arent present. I am confident it is mathematically possible and maybe even better for the model to compute. The Mixer unit model would have to be able to handle the different property packages for each inlet stream but that is doable and then create the mixed outlet stream with all components and phases present. All in all this would allow all users a more intuitive approach to creating models.

Potential Preview

  m.fs.I101 = Feed(property_package=m.fs.thermo_params)
  m.fs.I102 = Feed(property_package=m.fs.thermo_params_vap)

  m.fs.M101 = Mixer(
      property_packages=[m.fs.thermo_params, m.fs.thermo_params_vap, m.fs.thermo_params],
      num_inlets=3,
      inlet_list=['inlet_1', 'inlet_2_vapor', 'inlet_3'],
  )

[dallan-keylogic]

A translating mixer model would be nice, but it comes with some pitfalls that users might run into. The biggest issue is that different property packages might have different zero points for each component's enthalpy. At the dev call, @jsiirola suggested maybe adding a method to property packages to return component enthalpy at STP. We still might end up with issues if one property package implements enthalpy of mixing and one does not.

I wanted to move to using phase component mole fractions as a way to avoid needing a translating mixer. Since the state definition contains all necessary information about vapor-liquid equilibrium, there's no need to perform flash calculations that may be ill-defined if some of the components are zero.

It would be nice if each Feed unit model could take a different property package that only has the necessary phases and components. This would also allow the fixing the inlet states to be more intuitive since you wouldnt have to include very small value in each component and phase when in reality they arent present.

I think most end users would think using multiple property packages is less intuitive than just fixing those mole fractions to some small number, but maybe I'm wrong.

[tannerpolley]

I think the main reason why it might be helpful is that is it unintuitive to have to specify the inlet conditions of species that are present in a certain stream to a really small number (1e-5) or 0 in a stream that is specifically a pure component stream just because the property package it is defined with expects these components. Maybe make the property package config more adaptable and not require each component and one can specify which species are actually present in the stream but only one property package is used?