a newbie - how to set w to zero
 

Dear colleagues,

I was running a simple 2D steady state plate flow problem using cfx, but out of curiosity I modeled it as a 3D problem. I used a 1m width on the z direction and put 5 grids and periodicity in this (z) direction. While the problem gets solved quite as expected, I somehow also got a non-zero (but very small) z velocity component (w) - may be due to numerical diffusion.

So my aim is to enforce w=0 at every iteration.

Can someone guide me through how I can do this.

Cheers

You can pull it towards w=0 using a momentum source term. Or even a volumetric resistance in the z direction.

Not knowing the nature of your problem, but why would you put periodic BCs on the sides if you are trying to enforce w=0. Use symmetry, and if symmetry isnt correct, perhaps you shouldnt try to force w=0?

Thanks for the responses so far.

To ghorrocks: I have weighted your suggestions and another one (article) I found somewhere else. I thought to figure out the exact amount of momentum source to knock of 3D effects may be difficult, including the volume resistance - evenif I will have them in mind.

The article is found here http://www.padtinc.com/blog/post/2011/01/12/ANSYS-CFX-Going-2D-in-CFX.aspx#

- it suggest about adding a thin span-wise surface. and worked fine for my problem.

To singer1812, I think I can use periodic bc, or symmetry boundary condition, they are both applicable. I use this approach on the 3D fortran code I developed to model 2D problems and I have never faced any problem.

Thanks, you guys are terrific

Those 2 BCs are very different. Symmetric BCs indicate no mass flow (velocity) across the BC, while periodic has no such restriction (what goes out one side comes in the other).

Periodic may approach symmetric if your model is built right, but will not enforce a symmetric BC at the interface.

Hi Singer1812

I am running a laminar flow problem, and there is no forcing in the span wise direction, like an infinite length span wise plate. In so far as there is no disturbance, other than numerical disturbances, for the plate (stream-wise) length I used the flow remains laminar. In other words no three dimensional effects will be expected, but I am observing some non-zero w values. You are right I may better enforce symmetry condition, based on the definition you gave me.

I will yet have to try that.

Thanks any way.