Approach to modifying binary interaction parameters

To be honest, I rarely see modifications to the binary interaction parameters. That’s how I understand it: Aspen’s physical property data also comes from literature data imported (AspenTech plans to collaborate with the U.S. **Standardization Bureau to use that bureau’s database in the future); if it’s experimental data, there isn’t much that needs to be modified. Besides, what is the basis for our modifications? Are our own experimental data necessarily accurate? If it is found that the simulation results do not match the actual data, the experimental data should not be modified directly; instead, the simulation methods should be checked or the simulation results corrected. It is possible that the binary parameters could be appropriately adjusted to make the simulation results closer to the actual values, but such adjustments must also be made on a small scale.

What was said upstairs is correct; I don’t know either what the basis for modifying the binary parameters is. However, I have seen literature stating that it was modified

An example from the day before yesterday: Before making the changes, no matter how many theoretical plates were added or how much reflux was used, the desired purity could not be achieved; however, it was possible to attain that purity in actual production. After making the adjustments, everything worked properly – though such situations are quite rare.

This post was last edited by Albertlu on 2010-1-6 at 17:45. The correction of binary interaction parameters includes substances for which some property interaction parameters are incomplete; vapor-liquid equilibrium data is difficult to obtain, and there are limited references available. UNIFAC-DMD can be used to estimate certain binary interaction parameters that are lacking, such as those in the NRTL model (UNIFAC-DMD first estimates the activity coefficients at infinite dilution).

I’ve heard that modifying the parameters of binary interactions can make things more in line with reality, but I’m not sure based on what those modifications are made. I’m still learning; I hope experts can give some guidance!

I think the approach to modifying the binary interaction parameters is as follows: first, conduct vapor-liquid equilibrium experiments and perform regression on them; once the regression is successful, a prompt will appear asking whether to replace the original parameters. Second, after obtaining the interaction parameters from the DECHEMA database in the library, they are directly added using the Dechema button available for binary interaction parameters in Aspen. For example, when simulating the liquid-liquid equilibrium at a certain temperature, this approach yields very accurate results. Because the existing parameters in Aspen are all derived from regression over a wide range.

How are UNIFAC-DMD estimates made when methods like NRTL are lacking?

After discussing this issue, it seems to be a database problem; however, it requires a lot of resources to address. I believe that’s why some large companies continue to improve in this area – they have invested a lot of effort in it.

Regression needs to be done using experimental data

I think modifying parameters is meaningful, especially in certain industrial processes. This issue can be illustrated from one perspective: as everyone can see, for certain specific processes, the software includes a set of interaction parameters suitable for those processes – such as those related to acid water systems or aromatics, etc. Aren’t these specific \"property packages\" designed precisely to make the calculation results more in line with actual industrial data? So of course, we can also modify the parameters to make our calculations more meaningful.