How to decrease spurious currents in VOF ?

  • 78 Views
  • Last Post 26 November 2019
alucardz14 posted this 22 November 2019

 

My problem description:

I'm trying to solve the problem about the interface between water and air in a microchannel.

Bubble is stayed at the cavity and water is injected through the inlet.

Surface tension force is dominant rather than inertia and viscous forces.

 

Numerical results:

Inlet velocity is 0.01 m/s.

However, The numerical results show that the spurious velocity of 0.2 m/s at the interface near the wall.

 

Please see the attached *.cas and *.dat files.

 

 1) zoomed area (vectors of velocity components)

2) entire domain (contours of velocity magnitude)

Questions:

First, why a (non-physical) spurious current is observed when using the VOF method.

Second, how to reduce this spurious currents?

  

 

Attached Files

Order By: Standard | Newest | Votes
rwoolhou posted this 22 November 2019

You may find refining the mesh and reducing the time step to work. Also have a look in the Release notes and documentation: there are some numerics that can help but I can't remember which version they're in. 

alucardz14 posted this 23 November 2019

Thanks for your information. As far as I know, small time step with finer mesh can be the reason that spurious currents are getting worse. (Brackbill's aritcle, https://doi.org/10.1016/0021-9991(92)90240-Y)

I already did some simulations to check the effect of mesh and time step. It did not help with convergence and accuracy.

 

 

abenhadj posted this 25 November 2019

You are right. You probably need to look into beta feature after interracial dissipation. Other trick is to increase the number of smoothing steps for surface tension (docu)

Best regards, Amine

alucardz14 posted this 26 November 2019

Some tests were performed by adjusting smoothing steps  (solve>set>surface-tension options in the TUI).

The magnitude of the velocity was decreased by increasing smoothing step, however, the spurious currents (vortex) was maintained.

So, numerical results have still non-physical behaviors.

abenhadj posted this 26 November 2019

Try the other suggestion too. Spurious velocity is an evergreen topic for all multiphase codes. Make a sensitivity run without st force and with st stress instead of Brackbill force.

Best regards, Amine

Close