CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Surface tension problem in CFX (http://www.cfd-online.com/Forums/cfx/122754-surface-tension-problem-cfx.html)

lefuliang August 27, 2013 09:52

Surface tension problem in CFX
 
2 Attachment(s)
I am gonna simulate a viscous source flow that going across bumps. The wall of bump is no slip wall and adhesion force should be considered on these bump walls.
The boundaries is shown in fig.1, and the Pre-setting is listed as follow:
1. fluids: two continuous fluid (air, fluid 1:another is viscous fluid 0.8Pa.s), no bouyant;
2. fluid models: standard free surface/homogeneous model, laminar flow, no heat transfer, no combustion;
3. fluid pair models: air|fluid 1, surface tension coefficient:0.033 N/m, continuum surface force, primary fluid: fluid 1, no interphase transfer, no mass transfer;
4. inlet source: fluid 1: 2.6 mg/s;
5. bump wall: no slip wall, wall contact angle: 20 deg;
6. other walls: free slip wall, no wall adhesion;
7. open: static pressure: 0 Pa, normal to boundary;
8. Analysis type: transient, total time: 5 s, adapative timesteps, min timestep:0.0001 s, max timestep: 0.1s;
The solver finished very soon, but the result is totally wrong as fig.2 at 5 second simulation time. The fluid 1 almost stayed in same position as fig.2 during different timesteps and not yet passed one bump, whereas experiment will almost fill the domain.
I really spent plenty of time to find solution, but failed. Could anyone give me some suggestions? Thanks a lot.

ghorrocks August 27, 2013 18:33

Make sure you did not hit your max or min time step size. Widen the limits if necessary. Small time steps are essential for surface tension models.

Can you post a picture of your mesh?

Also, an inlet to produce the liquid might be better than a source term. Not sure of that, but worth a try.

lefuliang August 27, 2013 23:13

2 Attachment(s)
:) Thanks a lot!
I posted meshing grids in enclosed fig.3a and fig.3b. Fig.3a is overview whereas fig.3b is partial enlarged view.
The initial timestep is 0.0001 s (same as min timestep), then it keeps increase until it touches max timestep 0.1s. I also used double precision as CFX manual suggests.

Quote:

Originally Posted by ghorrocks (Post 448403)
Make sure you did not hit your max or min time step size. Widen the limits if necessary. Small time steps are essential for surface tension models.

Can you post a picture of your mesh?

Also, an inlet to produce the liquid might be better than a source term. Not sure of that, but worth a try.


lefuliang August 27, 2013 23:24

1 Attachment(s)
A meshing grid view in thickness direction is also attached in figure3c.
Attachment 24889

Quote:

Originally Posted by ghorrocks (Post 448403)
Make sure you did not hit your max or min time step size. Widen the limits if necessary. Small time steps are essential for surface tension models.

Can you post a picture of your mesh?

Also, an inlet to produce the liquid might be better than a source term. Not sure of that, but worth a try.


ghorrocks August 28, 2013 01:55

Some comments:
* Surface tension will not be very accurate with the sort of aspect ratioes you have here. I would check your mesh is OK before proceeding too far.
* If this simulation is 2D then only use a single element thickness. Even better, use a CFD model with a proper 2D model like Fluent. Fluent is much better than CFX for simulations with surface tension anyway.

lefuliang August 29, 2013 04:37

Thanks a lot. Fluent result of 2D case seems much better.:D

Quote:

Originally Posted by ghorrocks (Post 448436)
Some comments:
* Surface tension will not be very accurate with the sort of aspect ratioes you have here. I would check your mesh is OK before proceeding too far.
* If this simulation is 2D then only use a single element thickness. Even better, use a CFD model with a proper 2D model like Fluent. Fluent is much better than CFX for simulations with surface tension anyway.


ghorrocks August 29, 2013 05:49

Yes, if this can be done 2D you will do a lot better with Fluent.


All times are GMT -4. The time now is 15:24.