4 replies to this topic

[g3wtter]

hello,

 

I am trying to do this area test but the values of ln(gamma1/gamma2) are all positive so when I plot a graph of ln(gamma1/gamma2) vs x1 I get only one area above the x-axis and no are below.

 

Does this mean the data set is thermodynamically inconsistent or am I missing something?

 

 

I have also done the van ness point to point test which said that the data is thermodynamically consistent.

 

any help would be appreciated

 

kind regards

[g3wtter]

the picture got lost when posting this for some reason:

[MrShorty]

From what I can gather from the derivations of the standard area test (Herington test), the criteria that the total area should be 0 applies to data that covers the entire composition range from 0 to 1. Your data extend from 0.05 to 0.3.

[g3wtter]

From what I can gather from the derivations of the standard area test (Herington test), the criteria that the total area should be 0 applies to data that covers the entire composition range from 0 to 1. Your data extend from 0.05 to 0.3.

Thank you very much for your input Mrshorty

 

does this mean that the area test is only applicable for experimental data over the entire mixture range or can I use a correlation model to generate data points over the entire range and then apply the test?

[MrShorty]

I see no reason why you could not use a correlation model to generate data points over the entire range. However, a good correlation model should be thermodynamically consistent, so it should pass the area test. One might question whether such a test truly tests the measured data for thermodynamic consistiency, or if it tests the correlation model for thermondynamic consistency.

 

In their book Molecular Thermodynamics of Fluid-Phase Equilibria Prausnitz et al. suggest that the most reliable test for thermodynamic consistency is to use the Gibbs-Duhem equation to compute the 4th variable (y, for example) from the other three variables (P, T, x) and see how closely the computed values match the measured values.