CAPE-OPEN

I use a Chemsep Prop Pack within COFE.
and userdefined physical properties within Chemsep.

When setting liquid component density to temperature correlation
and mixture liquid density to Costald, it gives correct pure component densities for water and methanol
and slightly incorrect mixture densities within COFE.

When changing mixture liquid density to Racket (DIPPR) it gives quite incorrect pure component density for water (7% above "true" value),
despite pure component density is still set to temperature correlation and temperature correlation
for water is correctly implemented within Chemsep.

How can that be? Is that a bug?

Quoting prof. Ross Taylor:

When calculating mixture density ChemSep does not check to see if the mixture really is a pure component. It uses the selected mixture model. Many liquid density models do not use pure component densities in the calculation of mixture properties. That is true for COSTALD and Rackett and several others.

COSTALD is a corresponding states model, see here for more information:

https://www.cocosimulator.org/index_hel ... ensity.htm