bug in SAFT-VR SW?

[TermeHansen]

Just started to look around in this very nice module - great work!

As a first I have tried to reproduce the Cp of CO2 plot from the frontpage.

using:

compounds = ["carbon dioxide"]

model1 = SRK(compounds; idealmodel=ReidIdeal)
model2 = GERG2008(compounds)
model3 = SAFTVRSW(compounds; idealmodel=ReidIdeal);
model4 = PCSAFT(compounds; idealmodel=ReidIdeal);
model5 = SAFTVRMie(compounds; idealmodel=ReidIdeal);

I get:

Which seems fine except for the SAFT-VR SW. Any pointers to why I see that difference to your plot?

[longemen3000]

hmm, no idea. SAFT-VR-SW is one of those "legacy" SAFT eos, that are superseeded by more modern approaches (like SAFT-VR-Mie). also, looking at the code, seems that there wasn't any (functional) changes since last year.
Maybe it was an error on labelling? the the original plot image seems really old (sep 2021, around v0.2.1)

[longemen3000]

on a preliminary testing, SAFTVRSW seems consistent:

#tests that the gibbs energy and chemical potential have the same values. the first number is the difference
julia> Clapeyron.gibbs_duhem(model3,0.03,300.) 
(0.0, 11561.822765804463, 11561.822765804463)

#test that the ideal model returns an ideal pressure,the result is the difference of pressures
julia> Clapeyron.ideal_consistency(model3,0.03,300.)
0.0

so, if any bugs present, seems to be a subtle one

[pw0908]

It's been a while since I made that figure but, based on the figure we used in our paper, it seems like we removed SAFT-VR SW from it. I believe that may have been because the results had changed to something like TermeHansen is showing. Ill quickly benchmark SAFT-VR SW to the original paper and see if we can still replicate their results.

[pw0908]

I've now done the benchmark and can confirm that the current implementation of SAFT-VR SW replicates the original papers. There must have been something wrong at the time with the code.

[TermeHansen]

Good to know, then I suggest to either remove or update the plot here on GitHub, to avoid that misunderstanding again :)

Maybe using a plot using plotly which is really nice in giving options to inspect data a bit closer with hovering and zooming.

Cp_CO2.html

Regards!

[longemen3000]

after a long time, i finally found the bug.

Clapeyron.jl/src/models/SAFT/SAFTVRSW/SAFTVRSW.jl

Lines 214 to 219 in af65700

	function gHS(model::SAFTVRSWModel,V, T, z, i, j)
	σ = model.params.sigma.values
	ζ3 = @f(ζn,3)
	D = σ[i]*σ[j]/(σ[i]+σ[j])*∑(@f(x_S,k)*σ[k]^2 for k ∈ @comps)/∑(@f(x_S,k)*σ[k]^3 for k ∈ @comps)
	return 1/(1-ζ3)+3*D*ζ3/(1-ζ3)^2+2*(D*ζ3)^2/(1-ζ3)^3
	end

here, σ[i] should be σ[i,i]. It is fixed on the master branch.