Interaction parameters within TEA property package

[Stephan]

Hallo,

I tried

NRTL and Uniquac as activity coefficient models. However, error message showed up within Tea:
All interaction parameters must be set.
However, there are interaction parameters visible in the interaction parameter tab.

Same thing for Uniquac and Hayden O'Connel as gas-EOS.

Is this a bug or have additional IPs to be provided that are not specified within the COCO ipd-Files?

[jasper]

Is this in COCO/COFE? If so, can you attached the fsd file for me to look at?

[Stephan]

I attached an example. Changing vapour EOS into Hayden O'Connel failes
and the mentioned error message pops up.

[jasper]

... but not all interaction parameters for Hayden O'Connel are set:

Screenshot 2020-09-22 193929.png (12.15 KiB) Viewed 17019 times

[Stephan]

... but not all interaction parameters for Hayden O'Connel are set:

In Chemsep the same (incomplete) IP-Set is used for NRTl.
There it works.

In the case of TEA: Does it need a complete interaction parameter set
to work with the IP-based models? The ipd-files delivered with TEA/Coco do not provide a complete set.

[jasper]

I am not clear on that. NRTL binary interaction coefficients are unrelated to Hayden O'Connel ones. And I am not sure how NRTL would work without specifying all coefficients. Perhaps it is assuming zero.

[jasper]

Quoting Harry Kooijman:

That it runs doesn't mean that zero IP's will give a proper phase equilibria prediction. I believe for Hayden O'Connell BIPs zero is indeed assumed as default value. With respect to NRTL the user can opt for prediction using UNIFAC. However, for these light gases that is unlikely to give good predictions. The user should always validate his thermo before starting to use it.

[jasper]

Quoting prof. Ross Taylor:

The absence of BIPs for the HOC method means they are not listed in this book. I know of no other convenient sources for these parameters.

https://books.google.com/books/about/Co ... SeAAAAIAAJ

As already stated, if they are missing they are assumed to be zero and calculations will proceed accordingly. Also as stated, whether this provides an adequate model is another question that is impossible to answer without appropriate experimental data. If you have data available you can try tuning the parmeaters to get a better fit.

Note that HOC is considered essential for mixtures like acetic acid and water. An alternative for this mixture might be PSRK. But, again, any model should be validated. If that is not possible (because data does not exist) then one is always in the position of taking the model predictions on trust. There is no guarantee in thermodynamics.

[Stephan]

Sorry for the confusion with NRTL and Hayden O'Connel.

The above mentioned error showes up whenever I try one of the IP-based models within TEA.
When I try those models within Chemsep using the same IP-datatables, it "works" without error.

Is that a bug within TEA?

As I understood I can manually set all missing IP's to zero to make the models running within TEA.
Is that correct?

Would that be a workaround for all IP-based models within TEA (NRTL, Hayden O'Connel etc.) ?

[jasper]

It depends on the model. For example, PR will work find and will assume zeros. Models for which zero is not necessarily a useful default will not accept a default. This is not a bug.

If you *do* want them to run with zeroes, put in zeros. But again - this is not necessarily a good 'default' value.