Solid solubility

[pw0908]

Some preliminary work on a solid solubility solver. Right now we can calculate both the eutectic point and the solubility of a solid at a given temperature, pressure and solvent composition. This makes use of the CompositeModel API. A Simple example of this would be:

julia> model = CompositeModel(["water","ethanol","paracetamol"]; liquid=PCSAFT,gas=nothing,saturation=nothing,melting=nothing,solid=SolidSolubility,solid_userlocations=(;Hfus=[6000.,4969.00,27000], Tm=[273.15,158.70,443.6]))
Composite Model:
 Liquid Model: PCSAFT{BasicIdeal}("water", "ethanol", "paracetamol")
 Solid Model: SolidSolubility("water", "ethanol", "paracetamol")


julia> solid_solubility(model,1e5,298.15,[0.5,0.5,0.];solute=["paracetamol"])
3-element Vector{Float64}:
 0.4905731673763385
 0.4905731673763385
 0.01885366524732288

[codecov]

Codecov Report

Merging #194 (a10c660) into master (6daaf22) will decrease coverage by 0.34%.
The diff coverage is 1.51%.

@@            Coverage Diff             @@
##           master     #194      +/-   ##
==========================================
- Coverage   83.42%   83.08%   -0.34%     
==========================================
  Files         245      249       +4     
  Lines       15935    16000      +65     
==========================================
  Hits        13293    13293              
- Misses       2642     2707      +65     

Files Changed	Coverage Δ
.../property_solvers/multicomponent/multicomponent.jl	76.81% <ø> (ø)
...ty_solvers/multicomponent/solids/eutectic_point.jl	0.00% <0.00%> (ø)
..._solvers/multicomponent/solids/solid_solubility.jl	0.00% <0.00%> (ø)
src/models/CompositeModel/CompositeModel.jl	64.15% <ø> (-0.34%)	⬇️
...odels/CompositeModel/SolidModel/SolidSolubility.jl	0.00% <0.00%> (ø)
src/models/types.jl	100.00% <100.00%> (ø)

[pw0908]

Things that need to be done before merging:

• Add tests
• Add API docs
• Add example notebooks?

Some feedback:

• The API is really flexibility because of CompositeModel. However, it is really intricate and not straight-forward. Could we create a sub-model of CompositeModel which handles solid solubility only where we'd only need to specify the liquid model?

Some additional stuff I noticed that could be improved:

• Propagating group-contribution structure to the liquid model but not the solid model (e.g. when using SAFT-gamma Mie, we'd need to specify the groups of the molecules, but only when initialising the liquid model, not the solid)
• Merging X_userlocations to just one universal userlocations (we've talked about this before but I'm debating when the right time to do this change is?)
• We should make it optional to load-in activity coefficient models without a pure model. In the case of SLE, there's no reason to have a saturation model.

[longemen3000]

Could we create a sub-model of CompositeModel which handles solid solubility only where we'd only need to specify the liquid model?

You could create a function that does that, something ĺike this?

function Solid(components, liquidmodel = PC-SAFT, solidmodel =SolidSolubility, liquid_userlocations = String[], solid_userlocations = String[], verbose) 
  li= init_model(liquid, components, liquid_userlocations, verbose)
  so = init_model(solidmodel, components, solid_userlocations, verbose) 
  return CompositeModel(components, liquid=li, gas=nothing,saturation=nothing,melting=nothing,solid=so)
end