Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

moving ALE body due to reaction force, should I have to subtract reaction force again for drag

Please login with a confirmed email address before reporting spam

Hi,
I made a time-dependent coupled (ALE Moving mesh) cfd study. I calculated reaction force and coupled back as volume force, and velocity through moving wall calculated through a PDE(like falling sand tutorial). I want to know the body can move or not due to reaction force? should I have to subtract the reaction force again from the calculated force because of drag experienced?
Please some one advice..

10 Replies Last Post 23 nov 2015, 02:10 GMT-5
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 19 gen 2012, 12:54 GMT-5
Hi

how do you define or distinguish lift and drag forces ?

for me its a definition issue w.r.t the initial velocity of my "wing" but I'm not a specialist, and there are many ways too look at this.

the reaction force is a total force so if you decompose it into the corresponding drag and lift direction you have everything I would say (again it depends if my definition holds)

Often one simulates the displacement of the wing in a air flow, as a fixed wing with a air velocity being the sum of air initial motion and object motion, no ?

But there are better specialists that out there that should know more precisely

--
Good luck
Ivar
Hi how do you define or distinguish lift and drag forces ? for me its a definition issue w.r.t the initial velocity of my "wing" but I'm not a specialist, and there are many ways too look at this. the reaction force is a total force so if you decompose it into the corresponding drag and lift direction you have everything I would say (again it depends if my definition holds) Often one simulates the displacement of the wing in a air flow, as a fixed wing with a air velocity being the sum of air initial motion and object motion, no ? But there are better specialists that out there that should know more precisely -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 19 gen 2012, 21:20 GMT-5
Thank you dr.Ivar.
I have used twophase cfd module (ale moving mesh). Prescribed velocity to the wall and same to mesh in the y-axis direction. I calculated reacf(u) in x axis direction. I have coupled this force (like falling sand tutorial)through volume force and velocity. so oscillatory lift is prescribed, drag is induced. Upon doing this I saw the wing moving horizontally very easily at all the situation without any critical value. then I understood that the induced force and induced velocity may not move the wing horizontally and experience another drag to reduce the horizontal motion. Here I became confused. How can I include the possible drag resistance force which has to be overcome by the induced horizontal reaction force. If I do so, i will be able to see a critical value after which motion may start.

Any idea Dr. Ivar or any other experts...Please share with us...

Thanks again Dr. Ivar for your elaborate and thoughtful reply.
Siva
Thank you dr.Ivar. I have used twophase cfd module (ale moving mesh). Prescribed velocity to the wall and same to mesh in the y-axis direction. I calculated reacf(u) in x axis direction. I have coupled this force (like falling sand tutorial)through volume force and velocity. so oscillatory lift is prescribed, drag is induced. Upon doing this I saw the wing moving horizontally very easily at all the situation without any critical value. then I understood that the induced force and induced velocity may not move the wing horizontally and experience another drag to reduce the horizontal motion. Here I became confused. How can I include the possible drag resistance force which has to be overcome by the induced horizontal reaction force. If I do so, i will be able to see a critical value after which motion may start. Any idea Dr. Ivar or any other experts...Please share with us... Thanks again Dr. Ivar for your elaborate and thoughtful reply. Siva

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 gen 2012, 01:34 GMT-5
Hi

I believe one should distinguish the physics: "pressure or force density exchange" in your case, and the math required to get the mesh to remain "connected" over a common boundary

forcing continuity of the mesh displacement between the fluid and the solid does not bring forces to the domains, even if we often talk about the mesh as a series of "coupled springs" to explain how the mesh squeeze compresses or expands with the deformation it's only an analogy no "physical" action is exchanged.

So on one side you need to tell the ALE what is fixed and what is moving and how, and on the other side you need to resolve the fluid equations to get the pressure applied onto the walls and apply them onto any small flex part or just calculate them on any fixed part.

What I'm not sure about is how you consider your "wing" is it fixed in your reference and the fluid is moving across, and you observe and model a fixed rectangular region around your wing ? or are you moving your wing up and down in your rectangle based on lift and drag forces ?

--
Good luck
Ivar
Hi I believe one should distinguish the physics: "pressure or force density exchange" in your case, and the math required to get the mesh to remain "connected" over a common boundary forcing continuity of the mesh displacement between the fluid and the solid does not bring forces to the domains, even if we often talk about the mesh as a series of "coupled springs" to explain how the mesh squeeze compresses or expands with the deformation it's only an analogy no "physical" action is exchanged. So on one side you need to tell the ALE what is fixed and what is moving and how, and on the other side you need to resolve the fluid equations to get the pressure applied onto the walls and apply them onto any small flex part or just calculate them on any fixed part. What I'm not sure about is how you consider your "wing" is it fixed in your reference and the fluid is moving across, and you observe and model a fixed rectangular region around your wing ? or are you moving your wing up and down in your rectangle based on lift and drag forces ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 gen 2012, 07:22 GMT-5
thanks again dr. Ivar for the detailed reply.
I am giving a prescribed velocity to the wing (moving wall) in which the fluid is fixed. so I have given oscillating velocity to the wing through a sine function. Therefore the wing is going up and down in the fluid and experiencing reaction force. Now I want to couple the horizontal reaction force back to the wing so that it moves horizontally without 'prescribing' any motion horizontally. Coupling I have done through pde as in sand falling tutorial: X= Xt-Xdot and Xdott=Xdott-(react(u)/m). I am entering this horizontal velocity (Xdot) back to the moving wall in x component velocity. where as y component is prescribed, x component is calculated.

There should be some horizontal motion after certain critical frequency. But I am getting horizontal motion at all frequencies. So puzzled.

I hope I have explained the set up.

Please have a look on the set up and enlighten me on my mistakes.
thanks and regards
Siva
thanks again dr. Ivar for the detailed reply. I am giving a prescribed velocity to the wing (moving wall) in which the fluid is fixed. so I have given oscillating velocity to the wing through a sine function. Therefore the wing is going up and down in the fluid and experiencing reaction force. Now I want to couple the horizontal reaction force back to the wing so that it moves horizontally without 'prescribing' any motion horizontally. Coupling I have done through pde as in sand falling tutorial: X= Xt-Xdot and Xdott=Xdott-(react(u)/m). I am entering this horizontal velocity (Xdot) back to the moving wall in x component velocity. where as y component is prescribed, x component is calculated. There should be some horizontal motion after certain critical frequency. But I am getting horizontal motion at all frequencies. So puzzled. I hope I have explained the set up. Please have a look on the set up and enlighten me on my mistakes. thanks and regards Siva

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 gen 2012, 20:15 GMT-5
Also I am using quad mesh and now included boundary layers.
Also I am using quad mesh and now included boundary layers.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 gen 2012, 08:42 GMT-5
Hi

I would assume tets are more robust (and with boundary mesh layers, I would even start with COMSOL defaults) , in any case

If you are moving the object in static fluid, be aware of the different frame mappings (these are things at least I tend to get wrong, and would need a couple of days to be sure I have it right ;)


--
Good luck
Ivar
Hi I would assume tets are more robust (and with boundary mesh layers, I would even start with COMSOL defaults) , in any case If you are moving the object in static fluid, be aware of the different frame mappings (these are things at least I tend to get wrong, and would need a couple of days to be sure I have it right ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 gen 2012, 20:29 GMT-5
Hi Dr. Ivar,
Thanks for your post. Actually when there is ALE with moving mesh, i think that the default triangular mesh is not working properly. I am not sure why but if I change tri to tetra mesh, there is seem to be solution starting without any problem.
Actually I have been working on this problem almost a year but without knowing a suitable methodology to simulate my case.
I understood through some tutorials that reference frame x, y and X,Y should be properly used. So, I am slowly getting idea on which frame we have to see things. On comparing with earlier works on this problem, it is mentioned that they have solved in 'body-fitted' frame. But their plots and animation shows wing actually flapping and seen at different locations with respect to time.
when I made some simple simulation and solving, I could really see my wing flapping as I wanted also measure reac forces. But I could not really decipher what is going wrong.

Thanks a lot for your post. Every post has very thought full and illustrative insights on solving physics. In fact I am compiling and classifying your posts on different topics and started using as my 'pre-cooked' [ ;-) as you always call] reference.

with regards
Siva
Hi Dr. Ivar, Thanks for your post. Actually when there is ALE with moving mesh, i think that the default triangular mesh is not working properly. I am not sure why but if I change tri to tetra mesh, there is seem to be solution starting without any problem. Actually I have been working on this problem almost a year but without knowing a suitable methodology to simulate my case. I understood through some tutorials that reference frame x, y and X,Y should be properly used. So, I am slowly getting idea on which frame we have to see things. On comparing with earlier works on this problem, it is mentioned that they have solved in 'body-fitted' frame. But their plots and animation shows wing actually flapping and seen at different locations with respect to time. when I made some simple simulation and solving, I could really see my wing flapping as I wanted also measure reac forces. But I could not really decipher what is going wrong. Thanks a lot for your post. Every post has very thought full and illustrative insights on solving physics. In fact I am compiling and classifying your posts on different topics and started using as my 'pre-cooked' [ ;-) as you always call] reference. with regards Siva

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 gen 2012, 01:54 GMT-5
Hi

in principle tri or quad (2D), or tet, prism or hex (3D) should all work, it depeds a little on your geometry and on the physical phenomena you have.

I use tri and tet when I have no particular symmetry, and quad, prism or hex when I have "long" objects with slow gradients along the axis. Typically square or rectangular (box) type geometry fits nicely with square mesh. But you also have some limitations with square mesh w.r.t. cut planes and interpolation in the result section.

Example laminar flow in a long pipe there I use quad or prism/hex mesh, quite longer in the direction of the axis (provided I have no turbulence, that would require a more symmetric mesh of higher "quality")

Then "motion" can be virtual in FEM. I.e. a rotationg object in "solid", you might apply a rotating frame with equations and radial body load applied, and leave the object fixed in your frame. Just as with CFD, you might push your boat/wing in the fluid water/air, or leave the water/air flow around a fixed "boat/wing". But when you mix both, you easily get some headache. Then it's important to validate your case with vey simple models to be sure you understand what is happening. train yourself to fully get all the subtilities of eulerian and lagrangian reference frames, and how tat is implemented in COMSOL (check the doc and good textbooks, there are many in your library, I'm sure ;)

--
Good luck
Ivar
Hi in principle tri or quad (2D), or tet, prism or hex (3D) should all work, it depeds a little on your geometry and on the physical phenomena you have. I use tri and tet when I have no particular symmetry, and quad, prism or hex when I have "long" objects with slow gradients along the axis. Typically square or rectangular (box) type geometry fits nicely with square mesh. But you also have some limitations with square mesh w.r.t. cut planes and interpolation in the result section. Example laminar flow in a long pipe there I use quad or prism/hex mesh, quite longer in the direction of the axis (provided I have no turbulence, that would require a more symmetric mesh of higher "quality") Then "motion" can be virtual in FEM. I.e. a rotationg object in "solid", you might apply a rotating frame with equations and radial body load applied, and leave the object fixed in your frame. Just as with CFD, you might push your boat/wing in the fluid water/air, or leave the water/air flow around a fixed "boat/wing". But when you mix both, you easily get some headache. Then it's important to validate your case with vey simple models to be sure you understand what is happening. train yourself to fully get all the subtilities of eulerian and lagrangian reference frames, and how tat is implemented in COMSOL (check the doc and good textbooks, there are many in your library, I'm sure ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 9 feb 2012, 00:28 GMT-5
Thanks Dr. Ivar.
Now I successfully implemented using 2D spf. I learnt falling sand tutorial and understood the moving coordinates system. I think basically now I am able to simulate what I thought in my mind.
Thanks a lot.
With regards
Siva
Thanks Dr. Ivar. Now I successfully implemented using 2D spf. I learnt falling sand tutorial and understood the moving coordinates system. I think basically now I am able to simulate what I thought in my mind. Thanks a lot. With regards Siva

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 23 nov 2015, 02:10 GMT-5

Thanks Dr. Ivar.
Now I successfully implemented using 2D spf. I learnt falling sand tutorial and understood the moving coordinates system. I think basically now I am able to simulate what I thought in my mind.
Thanks a lot.
With regards
Siva


Hello Siva, Ivar

I know this post is quite old. I am curious. In the falling sand tutorial, the physics is prescribed in axisymmetric coordinate, which means the material frame would have been R, PHI and Z. Why is it that the ODE is written in terms of Xdot and X instead of R, PHI or Z?

Thanks.

EH
[QUOTE] Thanks Dr. Ivar. Now I successfully implemented using 2D spf. I learnt falling sand tutorial and understood the moving coordinates system. I think basically now I am able to simulate what I thought in my mind. Thanks a lot. With regards Siva [/QUOTE] Hello Siva, Ivar I know this post is quite old. I am curious. In the falling sand tutorial, the physics is prescribed in axisymmetric coordinate, which means the material frame would have been R, PHI and Z. Why is it that the ODE is written in terms of Xdot and X instead of R, PHI or Z? Thanks. EH

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.