Posted:
5 years ago
Apr 4, 2013, 1:22 AM EDT

Hi

for the wall conditions on the ball, I would answer that depends on what kind of surface treatment you have. Normally, even for a perfectly hydrophobic surface the molecules "at" the surface have a zero velocity (relative to the surface) and I understand the BC as setting the fluid velocity at the surface velocity as in CFD you solve for _U_ the velocity vector field and not the _u_ position vector field as for in SOLID)

For inlet and outlet conditions it all depends on what you desire, the main condition is that you ensure that both _U_ velocity and p pressure have enough BC to give an unique solution. The free p=0 outlet fixes the pressure somewhere (else you need to fix the gauge pressure at least on a point in the model). Now notice also that outlet and inlet are names, none of these imposes a strict direction of flow, our model may be made such that there is locally some outlet at the "inlet" boundary and vice versa. Check the model library, there was an interesting model showing the BC for an outlet, in presence of gravity, and how a little glip in the "correct" outlet BC set up changes drastically the outflow pattern

Mostly COMSOL is correct (at least when physics and BC are set up correctly) so often I take some time when a response is not ass expected to study it in more details, either to discover I had taken some hypothesis and defined my BC wrongly, or most often, that I had forgotten about a given case, the response is correct and the complexity of my model in fact tougher than I had imagined. With the detailed images and data you can extract from COMSOL you learn far more than before when we had to deduce the response from a few points ona graph, having taken us ages and pages to get to ;)

Another point, set the ALE physics node ABOVE the SOLID node, as solid has a "special" case of ALE inside (the spatial and material frame approaches) this is overridden by ALE the way you have set it, and that will give strange results, so if you move ALE first, then SOLID will override the ALE in the SOLID domain. Then use the BOUNDARY of domain 2 and impose the motion u,v from solid. Then be aware that as u,v,w is defined by solid for the position/deformation field, COMSOL is using u2,v2,w2 and p2 as variable names for the SPF velocity field.

Finally, is not really deforming you could exclude it and only use ALE and SPF + some equations for the rigid body motion. Ore use level set or phase set methods see the examples in teh model library

--

Good luck

Ivar

Hi
for the wall conditions on the ball, I would answer that depends on what kind of surface treatment you have. Normally, even for a perfectly hydrophobic surface the molecules "at" the surface have a zero velocity (relative to the surface) and I understand the BC as setting the fluid velocity at the surface velocity as in CFD you solve for _U_ the velocity vector field and not the _u_ position vector field as for in SOLID)
For inlet and outlet conditions it all depends on what you desire, the main condition is that you ensure that both _U_ velocity and p pressure have enough BC to give an unique solution. The free p=0 outlet fixes the pressure somewhere (else you need to fix the gauge pressure at least on a point in the model). Now notice also that outlet and inlet are names, none of these imposes a strict direction of flow, our model may be made such that there is locally some outlet at the "inlet" boundary and vice versa. Check the model library, there was an interesting model showing the BC for an outlet, in presence of gravity, and how a little glip in the "correct" outlet BC set up changes drastically the outflow pattern
Mostly COMSOL is correct (at least when physics and BC are set up correctly) so often I take some time when a response is not ass expected to study it in more details, either to discover I had taken some hypothesis and defined my BC wrongly, or most often, that I had forgotten about a given case, the response is correct and the complexity of my model in fact tougher than I had imagined. With the detailed images and data you can extract from COMSOL you learn far more than before when we had to deduce the response from a few points ona graph, having taken us ages and pages to get to ;)
Another point, set the ALE physics node ABOVE the SOLID node, as solid has a "special" case of ALE inside (the spatial and material frame approaches) this is overridden by ALE the way you have set it, and that will give strange results, so if you move ALE first, then SOLID will override the ALE in the SOLID domain. Then use the BOUNDARY of domain 2 and impose the motion u,v from solid. Then be aware that as u,v,w is defined by solid for the position/deformation field, COMSOL is using u2,v2,w2 and p2 as variable names for the SPF velocity field.
Finally, is not really deforming you could exclude it and only use ALE and SPF + some equations for the rigid body motion. Ore use level set or phase set methods see the examples in teh model library
--
Good luck
Ivar

Posted:
2 years ago
Apr 12, 2016, 1:24 AM EDT

Hi

Dear Ivar ! recently, i was plagued by how to setting "outlet pressure condition" of straight pipe,which is under the condition of gravity. Here, You talked about there have related cases in the model library. Do you yet remember which one is it？or, can you give me some guidance about this question?

Thank you so much !

Hi
Dear Ivar ! recently, i was plagued by how to setting "outlet pressure condition" of straight pipe,which is under the condition of gravity. Here, You talked about there have related cases in the model library. Do you yet remember which one is it？or, can you give me some guidance about this question?
Thank you so much !