Learning Forum

[Patrick Kirby]

I was curious if the population balance model available, in particular QMOM, has incorporated means of moment correction since its original inclusion in the Fluent package? I came across a few publications describing the inclusion of QMOM in fluent from last decade and then additional follow-up publications describing correction algorithms. I'm curious if these tools to prevent the creation of corrupt moments has been included or if it should be through the use of UDF. I have also read that if one keeps the moment solver 1st order than moment corruption can also be avoided. Any thoughts?

Thanks and have a great week

Pat

 

 

[Rob]

I'm not sure how you'd corrupt a moment, and the 2023R1 Theory is here  https://ansyshelp.ansys.com/account/Secured?returnurl=/Views/Secured/corp/v231/en/flu_th/flu_popbal_popbal_theory.html  Given 10 years ago would be about R14-R15 there may well have been changes, you'd need to check all the release notes from then to now for changes. 

[Patrick Kirby]

Thanks Rob! In terms of moment corruption I am refering to that discussed in the citation below and others of the like:

Petitti, M., Nasuti, A., Marchisio, D.L., Vanni, M., Baldi, G., Mancini, N. and Podenzani, F. (2010), Bubble size distribution modeling in stirred gas–liquid reactors with QMOM augmented by a new correction algorithm. AIChE J., 56: 36-53. https://doi.org/10.1002/aic.12003

 

Thanks

 

Pat

[Rob]

Fluent v6.2 was about 2004-ish so was very likely about the first release of the population balance models, they may even have been UDF only at that point. Fluent v6.0 was 2001, and brought the multiphase models into the unstructured code. As I don't have the paper it's hard to comment, and having tried a couple of times, I'm straying very close to "engineering insight" and "proprietary knowledge". 

Are they adding a model correction or damping into the maths? Ie are they shifting a correlation by some factor or just damping a feedback loop?

[Patrick Kirby]

They are including multiple corrections in this particular work. One of which involves evaluating the moments and identifying, if any, moments that may require correction. Then the moment that would result in the best solution is chosen and corrected with the goal of preventing unphysical results. The correction appears to occur after the iteration solving for the moments. I hope my explanation makes sense. 

 

I take it from your previous answers that I may not need to consider this, but can track the moments throughout the solution as I go in case I have concerns.

[Rob]

Yes, I'd see what Fluent gives with time, and then review.  I'm always wary of models that look at the result and then modify it based on another correlation: if the correlation is that good why not use it as a UDF, and if not, why would I trust it over the initial model? Unless it happens to give exactly the result I want..... then I'd publish it! ;) 

[Patrick Kirby]

Thanks Rob! In a related but unrelated question. When making use of a sliding mesh, it would be incorrect to make use of the frozen flow field when solving the population balance equations correct? It is my understanding that the flow can be frozen when using the MRF technique, but not the transient sliding mesh.

[Rob]

The frozen flow is fine, but I can't remember if the rotor position will update regardless. If you freeze the flow I don't think it'll alter the flow if you turn the rotor speed off. Check first! 

[Patrick Kirby]

Thanks Rob, I was curious as I came across a source that stated that when making use of the sliding mesh you can't disable the flow field as it necessary for each time step when solving for mixing time. I was assuming that something similar would be true when solving for volume fraction and bubble size distribution as the flow field will be unsteady.

[Rob]

I suspect that's because you're updating the rotor position rather than using a reference frame (source term) to force the motion. Hence turning the rotor off too!  Any time you turn things on/off to speed up a calculation there are other factors you need to consider: a high proportion of the commercial support I handle is model or "short cut" related.