[JohanSoder]

Hi fellow CFX users and experts (although I am not an expert

I have a problem that I would like to ask if anyone can help me with.

I am tracking sub micron particles in a laminar flow (low pressure) in a tube. In the tube a set of orifices are places which results in a focusing effect of the sub-micron particles. The pressure is rather low giving a low Knudsen number, so low that the flow is in the free slip regime (0.001 < Kn <0.1).

I have written an external FORTRAN routine to handle the particle drag force. In the simplest routine (I have several) I calculate CD according to Schiller-Naumann and correct this with the Cunningham correction. I should mention that all these routines will be public (as I am a researcher) once everything is working as expected.

This now works, to some extent at least. This means that the particles are indeed being focused (without the Cunningham correction they will just follow the streamlines).

I can compare my results to published results obtained using FLUENT, and although they are qualitatively similar they are quantitatively different. One thing might be the fact that CFX does not handle the free slip regime, while FLUENT does – see equations 14.2-14 – 14.2-21 at http://www.ipt.ntnu.no/manuals/Fluen...ug/node613.htm

As this must be a common problem for CFD analysis in low pressure regimes I wonder if anyone have publicly available FORTRAN code to handle the slip regime? If so you will off course receive full credit for the code in any publication I make where this is included.

I hope that someone can give me some valuable advice here, I got the paper http://prace.ippt.gov.pl/IFTR_Reports_5_2007.pdf from ANSYS CFX, however this is 56 pages of reading (and it is for FLUENT!).

All the best wished possible,
~Johan S

[ghorrocks]

Hi,

I am not quite sure what you mean by the free slip regime. If you mean free slip at the wall boundaries, then CFX can do this and there is no problems with it (that I am aware of anyway!). If you mean potential flow or inviscid flow or something like that then yes, CFX has a problem there. You have to either specify a low viscosity, or do a thermal simulation and map the gradient of the thermal field as the velocity.

Glenn Horrocks

[JohanSoder]

Hi Glenn

Thank you for your answer, let's see if we can clarify things here...

I asked the CFX support about the free slip regime and this is their reply (edited for length):

The short answer to your question is : CFX doesn't handle free slip regime. [...] The range in between these two extremes is called the slip regime (0.001< Kn<0.1). CFX can be applied in the continuum regime but will give inaccurate results in the free molecular flow regime. In the slip regime the continuum assumption can still be applied but the wall boundary functions would need to be changed, to account for some free slip at the wall and modified heat transfer : the wall boundary functions will be inaccurate as they do not account of some free slip at the wall. There is no built in capability in CFX to do this so you would have to add in your own wall function variation if you were to carry out calculations in the slip regime, using Fortran routine.

My problem is very simple. I am tracking sub-micron particles in a tube with a set of orifices in it. At the inlet the pressure is ~300 Pa and the outlet ~1 Pa. The pressure/velocity gradients at the orifices causes an effective focusing of the sub-micron particles into a particle beam. This has been shown using FLUENT.

I have written FORTRAN code to correctly handle the drag force on these small particles which gives qualitatively similar results as FLUENT, however they are quantitatively different. Right now the only thing I know of that is significantly different is the free slip regime, in FLUENT this seems well taken care of, and as I understood it in CFX it is not - but perhaps you are correct, then I am even more surprised at my results...

Best,
Johan

[ghorrocks]

Hi,

OK thanks for the info - you learn something every day. CFX does not have anything like this so it would have to be applied as user fortran or maybe CEL expressions.

Glenn Horrocks