State struct

[longemen3000]

with #268 , the need for State struct seems to arise. this discussion is to discuss the details of a State struct for Clapeyron.

• What is a State struct? : it is a struct that contains a set of thermodynamic properties that constitute a complete state. For example: T = 300K, p = 1 atm, z = [0.5,0.5] is a complete state for a 2-component model. there are also states that depend on the phase (liquid, gas, solid) or equilibria conditions (at saturation, critical point)
• what interface do we want for the State constructor?
• how do we calculate properties with the State constructor?
• compatibility

[longemen3000]

Coolprop supports the following:

Input Pair	Description
QT_INPUTS	Molar quality, Temperature in K
QSmolar_INPUTS	Molar quality, Entropy in J/mol/K
QSmass_INPUTS	Molar quality, Entropy in J/kg/K
HmolarQ_INPUTS	Enthalpy in J/mol, Molar quality
HmassQ_INPUTS	Enthalpy in J/kg, Molar quality
DmassQ_INPUTS	Molar density kg/m^3, Molar quality
DmolarQ_INPUTS	Molar density in mol/m^3, Molar quality
PQ_INPUTS	Pressure in Pa, Molar quality
PT_INPUTS	Pressure in Pa, Temperature in K
DmassT_INPUTS	Mass density in kg/m^3, Temperature in K
DmolarT_INPUTS	Molar density in mol/m^3, Temperature in K
HmassT_INPUTS	Enthalpy in J/kg, Temperature in K
HmolarT_INPUTS	Enthalpy in J/mol, Temperature in K
SmassT_INPUTS	Entropy in J/kg/K, Temperature in K
SmolarT_INPUTS	Entropy in J/mol/K, Temperature in K
TUmass_INPUTS	Temperature in K, Internal energy in J/kg
TUmolar_INPUTS	Temperature in K, Internal energy in J/mol
DmassP_INPUTS	Mass density in kg/m^3, Pressure in Pa
DmolarP_INPUTS	Molar density in mol/m^3, Pressure in Pa
HmassP_INPUTS	Enthalpy in J/kg, Pressure in Pa
HmolarP_INPUTS	Enthalpy in J/mol, Pressure in Pa
PSmass_INPUTS	Pressure in Pa, Entropy in J/kg/K
PSmolar_INPUTS	Pressure in Pa, Entropy in J/mol/K
PUmass_INPUTS	Pressure in Pa, Internal energy in J/kg
PUmolar_INPUTS	Pressure in Pa, Internal energy in J/mol
DmassHmass_INPUTS	Mass density in kg/m^3, Enthalpy in J/kg
DmolarHmolar_INPUTS	Molar density in mol/m^3, Enthalpy in J/mol
DmassSmass_INPUTS	Mass density in kg/m^3, Entropy in J/kg/K
DmolarSmolar_INPUTS	Molar density in mol/m^3, Entropy in J/mol/K
DmassUmass_INPUTS	Mass density in kg/m^3, Internal energy in J/kg
DmolarUmolar_INPUTS	Molar density in mol/m^3, Internal energy in J/mol
HmassSmass_INPUTS	Enthalpy in J/kg, Entropy in J/kg/K
HmolarSmolar_INPUTS	Enthalpy in J/mol, Entropy in J/mol/K
SmassUmass_INPUTS	Entropy in J/kg/K, Internal energy in J/kg
SmolarUmolar_INPUTS	Entropy in J/mol/K, Internal energy in J/mol

[pw0908]

I see. That should work then?

Apart from that, would we now need a VT-solver?

[pw0908]

Oh! We could also include caches at this point!

[longemen3000]

Yeah, that is an option too

[pw0908]

So do we want to figure out this State framework before the next release?

[longemen3000]

I'm interested. At the moment I'm finishing the COSMO updates, after that I can work up on that

[Sush1090]

Hello,
I came across this discussion recently.
There is something similar I do locally. It is same as the State described here.

It has the following vars

mutable struct State
    model ::EoSModel
    z:: AbstractVector
    p::Any 
    T::Any 
    h::Any 
    s::Any
    v::Any 
    q::Any 
    method::Symbol
end

I store the method as a symbol to choose which flash computation it should follow

For example I can do the following:

model = cPR(["pentane","isobutane"],idealmodel = ReidIdeal)
z = [1.0,1.0]; p = 101325*3; T = 350; h = Clapeyron.enthalpy(model,p,T,z)

S = State(model=model,z= z,p = p,h = h)

Here I define the State to be a ph flash based solution. Then I use that to find the property I need for example :

julia> entropy(S)
-49.84418964661152

I would be glad to contribute to this if it is still of interest.

Thank you,
Sushrut

[longemen3000]

Hello,

This was originally a discussion on how to support multiple inputs, apart from PT, when i wrote this, i just didn't have any idea on how to implement those.

Now, we have a sort of structure on which we can work with:

• all (supported) solvers have a XY_property function
• all flashes return a FlashResult.
• all (supported) properties allow calling f(model, state)

FlashResult assumed the role i wanted for State to be, at the moment.

With that said, what do you have in mind? a way to call a flash with the keyword arguments defining the type of flash?

[Sush1090]

I had made this as a user friendly (just for myself haha) way to go through all the various Flash computations (pt,ph etc ...)

I was looking for something like PropsSI("output","var1"val1,"var2"val2,fluid) where I do not specify explcilty on which flash needs to be used. It will be automatic. As of now we still need to import the relevant xy_flash to do our computation.

With that said, what do you have in mind? a way to call a flash with the keyword arguments defining the type of flash?

Yes exactly as we define the State only once.

[Sush1090]

In other words its just a wrapper over all the flash computations for ease of usage

[longemen3000]

Yeah, that would be great to have, something like Clapeyron.flash(model;kwargs...)?

[Sush1090]

yes indeed. As of now my code is very "dirty". I can clean it up and make a pull request

[longemen3000]

Seems good to have, feel free to open up a PR