Hey -

Running a simulation in CFX-11 where the flow becomes super-saturated and particle nucleation is predicted by CNT (classical nucleation theory). Growth then follows. I have written various UDF's to attempt at capturing the physics - latent heat transfer (phase change), particle growth etc. However - fundamentally this may not be possible using the CFX-pre and solver:

- No particle variables to track the particle's exposure, i.e. nucleated or not, what mode of growth.

- Return variables are 'bulk' fluid only - effect of the Lagrangian particle tracking;

- ...hence, particles do not (in real-time) influence the flow properties

- Changes in particle diameter are not up-datable

- unable to differentiate particles in the same mesh-cell

Hoping I'm wrong on all these counts...thoughts?

Cheers!

Jon

Jon,

it is possible, but don't use the lagranian way. Use Euler-Euler instead: http://www.cfd.com.au/cfd_conf06/PDFs/027Pra.pdf

Succes, Bart Prast

Hi,

Also have a look at V12. There is a new tutorial example of non-equilibrium stuff which may be of use.

Glenn Horrocks

Thanks Bart and Glenn,

I've looked at the Euler-Euler form of solver, as well as the treatment of the 'particles'. Haven't been able to solve a scenario yet - but have a few questions:

- Particles are solved as a 'concept' (a group of N 'particles' subject to the vectorized PDE's) within the mesh-cells, where does their time/trajectory history come from? - Hence, individual particle drag, etc

- Particle size is broken up into discrete size ranges, with each range requiring it's own solution to the governing PDE's, solving for 9 size-ranges (1-2um, 2-3um....9-10um) that's 9x27 equation sets to be solved - expensive!

- Having Courant-number issues as the 'particles' have 'slip' and the time-marching action of the flow/particle movement progresses out of 'phase', lags.

So far I've learned a lot about CFX's particle capability. I've developed a temp. soln. by setting a new material with a high Cp and writing a UDF to increase particle mass incrementally w.r.t. time. This simulates the (latent) heat release of phase change, increases in particle dia. and interacts with the flow.

Not sure how much more can be interpreted by going the full-stretch and developing a E-E over E-L model. Thoughts?

Also, thank you both for the insightful replies thus far. I'm a recent graduate and have found the CFD community to be very friendly and dynamic. Hopefully I will be able to return the favour in the years to come.

Sincerely,

Jon

I think you will benefit considerably from the (inhomogeneous) DQMOM approach in CFX12. Check with your CFX representative.

Bart