Fluid in 2D channel dambreak extension

 Register Blogs Members List Search Today's Posts Mark Forums Read

 August 9, 2019, 13:46 Fluid in 2D channel dambreak extension #1 Senior Member   Josh McCraney Join Date: Jun 2018 Posts: 164 Rep Power: 4 Hi all! I'm running the tutorial dambreak, and have changed the geometry to a rectangle (no bump on the bottom) and the initial condition to a circular arc (see video here https://imgur.com/a/nsTBIO8#32DCvNL). I've turned gravity and viscosity off for both fluids. I've also enforced a static contact angle of 71 degrees. After a very long time the surface starts to become static, getting closer and closer to equilibrium: why (perhaps not shown in this short video, but if I run for longer times this happens)? Seems odd to me since there is no gravity or viscosity acting to stabilize the interface.

 August 12, 2019, 21:45 Numerical Dissipation #2 Member   Peter Brady Join Date: Apr 2014 Location: Sydney, NSW, Australia Posts: 54 Rep Power: 8 Hi Josh, My first guess would be numerical dissipation. Even with a higher order schemes there will be some small truncation errors, which can be interpreted as "numerical turbulence". This is actually the basis of some implicit turbulence models and will act to extract energy from the flow field. Hence, if you are removing energy, the flow field will eventually damp towards an equilibrium solution. Could this be a mechanism to explain what you are seeing? Cheers, -pete

August 13, 2019, 07:55
#3
Senior Member

Josh McCraney
Join Date: Jun 2018
Posts: 164
Rep Power: 4
Quote:
 Originally Posted by pbrady2013 Hi Josh, My first guess would be numerical dissipation. Even with a higher order schemes there will be some small truncation errors, which can be interpreted as "numerical turbulence". This is actually the basis of some implicit turbulence models and will act to extract energy from the flow field. Hence, if you are removing energy, the flow field will eventually damp towards an equilibrium solution. Could this be a mechanism to explain what you are seeing? Cheers, -pete
Hi Pete

Yea, this is something I was worried about. I reduced the time step, and it looks like the issue is going away, which would imply numerical dissipation could be the issue.

However, when I reduce the time step sufficiently small I actually get some instabilities in the interface, and even topological breakup! I can post a video, but have you ever seen this? (CFL number is restricted less than 0.5)

 August 20, 2019, 22:54 Not Seen #4 Member   Peter Brady Join Date: Apr 2014 Location: Sydney, NSW, Australia Posts: 54 Rep Power: 8 Hi Josh, Sorry for the late reply, I've been on the road the last few days. Actually no, I've not seen what you describe at dambreak scales. When I've been working with free surface flows reducing the timestep usually stabilises the simulations. Mind you, I've never had a need to push the timesteps as low as you have. The only time I've seen topology breakdowns like you describe is with surface tension enabled and working a capiliary scales. But this was in CFD-ACE+ and not OpenFOAM. Further, capiliary waves are also a known issue at micro scale simulations so there is specific damping to account for this. From what you describe you should be way above that scale but its the only mechanism I can think of right now. Cheers, -pete

August 21, 2019, 03:11
#5
Senior Member

Josh McCraney
Join Date: Jun 2018
Posts: 164
Rep Power: 4
Quote:
 Originally Posted by pbrady2013 Hi Josh, Sorry for the late reply, I've been on the road the last few days. Actually no, I've not seen what you describe at dambreak scales. When I've been working with free surface flows reducing the timestep usually stabilises the simulations. Mind you, I've never had a need to push the timesteps as low as you have. The only time I've seen topology breakdowns like you describe is with surface tension enabled and working a capiliary scales. But this was in CFD-ACE+ and not OpenFOAM. Further, capiliary waves are also a known issue at micro scale simulations so there is specific damping to account for this. From what you describe you should be way above that scale but its the only mechanism I can think of right now. Cheers, -pete
Pete, let me elaborate: I began with the dambreak tutorial, but I turned gravity and viscosity off and surface tension on. Then the capillary length scale is infinite, and surface tension is a dominate force. Does this change your analysis of my issue?

 August 21, 2019, 17:40 Potentially Yes #6 Member   Peter Brady Join Date: Apr 2014 Location: Sydney, NSW, Australia Posts: 54 Rep Power: 8 For sure, that would change the analysis for sure. Its a mathematically valid case that you are playing with be has no physical analogy. For example, there are some studies of liquids in micro-gravity where droplets are quickly formed and surface waves damp out. So yeah, you've essentially removed all damping other that that inherent in the numerics so what you are observing is totally possible. -pete

August 22, 2019, 02:27
#7
Senior Member

Josh McCraney
Join Date: Jun 2018
Posts: 164
Rep Power: 4
Quote:
 Originally Posted by pbrady2013 For example, there are some studies of liquids in micro-gravity where droplets are quickly formed and surface waves damp out.
Experimental or theoretical? I'm very interested if you have the source.

Quote:
 Originally Posted by pbrady2013 So yeah, you've essentially removed all damping other that that inherent in the numerics so what you are observing is totally possible. -pete
So you're suggesting waves are damping out due to numerical dissipation? Is there a good way to know which scheme I could use? How would you quantitatively measure numerical dissipation?

Lastly, at long run times with more refined time steps and mesh, the free surface loses symmetry. Do you have any idea why?