Cross over theory for SAFT-VR Mie

[branch171]

Hi

I have Kiselev critical point renormalisation working. However I face a problem that for parameter estimation I need to calculate the component critical point on the fly based on any change in the base SAFT model parameters. Is there a way to do that? I'm using SAFTVRMie. I have a base model SAFTVRMieKiselevNoCrit which need to get the Tc and Vc from. Then use SAFTVRMieKisekev in the estimation.

So far I have a very hand made method. It works but I would like a more systematic method.

[pw0908]

Hi @branch171,

Do you have the original paper so that we could take a look at the method? I have a feeling you might be able to include all of this in one model. At worst, you can probably leverage the CompositeModel structure to get what you need.

Best,

Pierre

[pw0908]

From a brief read through, this doesn't sound like just an EoS, but an actual correction applied on top of the EoS... this might need a new model structure...

[branch171]

Here is a base paper there are many all slightly different but the concept is to scale a base eos

IJT_21(6)_1373.pdf

[pw0908]

Could you also open a pull request with the changes you've already made? It'll make it easier for us to work out what best to do.

@longemen3000 looks like we could create a new model type (CrossoverModel) to do something like this:

model = SAFTVRMie(["water"])

model_c = CrossOver(model)

where, during model construction, the critical point of the original EoS is computed and cached.

I'm guessing, for parameter estimation to work, in the recombining process, it'll need to re-compute the critical point. It also means that, when updating the parameters, it needs to do the base model first.

[longemen3000]

Uff, i assume it will be funny to try to apply AD through parameter estimation on crossover models 😅, as pierre said, on recombining, the critical point will need to be recalculated.

[pw0908]

Yea... this would have to be one of those models which can only use Metaheuristics.jl

[longemen3000]

related issue: #109