As general introduction I would recommend the (already mentioned) Properties of Gases and Liquids (Poling, Prausnitz, O'Connel) or similar books,
most documents mentioned in previous posts discuss the features available in a specific software (Aspen) and do not consider recent models which includes association or some complex mixing rules...
also the models may be quite different,
for example the extended Peng Robinson available in Aspen is very different from the extended Peng Robinson available in Prode (which has 5 parameters per chemical allowing a more accurate reproduction of vapor pressure and liquid phase...).
Of course the availability of some specific model or feature depends from your software, however you can try (within some limits) to adapt a general purpose model providing you have access to a utility allowing to regress data.
In my experience, recent EOS with temperature dependent parameters allow to manage even polar fluids (say water)
as example the extended Peng Robinson available in Prode allows to predict water dew point in hydrocarbons (up to 100 Bar) with an accuracy of about 1-1.5 K ,
I have adopted the same model for water-methanol and similar mixtures with good results.
For very non-ideal, polar, mixtures at high pressures you can adopt a EOS such as Extended Peng Robinson + NRTL with complex mixing rules (say, for example WS or MHV2)
Other models such as CPA or SAFT can accurately reproduce vapor pressure allowing good results for phase equilibria of many (including associating) fluids...
Finally, many models such as Pitzer, E-NRTL, E-UNIQUAC, CPA-Electrolyte etc. are available for electrolytes, the oldest (Pitzer) is probably the most flexible but requires many parameters, the others have different features depending from the specific variant avauilable in your software....
As you see there are no simple rules of thumb and the selection of the thermo package may require evaluation of predicted results vs. experimental data (when available)
Edited by serra, 20 October 2015 - 03:39 AM.