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 The information provided may be out of date.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.

1D wall in CFD

Please login with a confirmed email address before reporting spam

Hi everyone
Is it possible to have an 1D wall in 2D fluid?
Please some one guide me on this.
I could not subtract 'line' from 'rectangle'. Any idea please...

11 Replies Last Post 2011/11/09 4:25 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 2011/10/30 3:13 GMT-4
Hi Siva

you cannot substract a 2D from 3D nor a 1D from 2D. But you can do an union of a 1D line crossing or adjacent to a 2D structure (same for 2D crossing/cutting a 3D). This union is mostly done by the defulat "finis - Union node of COMSOL, where COMSOL analyses the geoemtry and transforms the different geoemtrric objects (volumes, surfaces, lines, points into corresponding, numbered FEM Entities (domains, Boundaries Edges and points)

Check the doc, as the naming changes slightly for 3D, 2D, 1D. Domains are always the highest, Boundary next level down, then edges and points if applicable

--
Good luck
Ivar
Hi Siva you cannot substract a 2D from 3D nor a 1D from 2D. But you can do an union of a 1D line crossing or adjacent to a 2D structure (same for 2D crossing/cutting a 3D). This union is mostly done by the defulat "finis - Union node of COMSOL, where COMSOL analyses the geoemtry and transforms the different geoemtrric objects (volumes, surfaces, lines, points into corresponding, numbered FEM Entities (domains, Boundaries Edges and points) Check the doc, as the naming changes slightly for 3D, 2D, 1D. Domains are always the highest, Boundary next level down, then edges and points if applicable -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/10/30 3:59 GMT-4
Thanks dr.Ivar.

I will look at those documents...

Actually I wanted to rotate a 2D box (rigid wall containing solid domain - subtracted from fluid domain) through velocity boundary condition. But after rotating few angles it deforms the rigid wall. I thought the deformation will be less if it is an 1D line.

Would you please advice me how can I rotate rigid wall, for cfd (2phase moving mesh module) analysis without much mesh deformation?

I could not use rotating machinery module because rotating direction change with respect to time.

Any help please...
Thanks dr.Ivar. I will look at those documents... Actually I wanted to rotate a 2D box (rigid wall containing solid domain - subtracted from fluid domain) through velocity boundary condition. But after rotating few angles it deforms the rigid wall. I thought the deformation will be less if it is an 1D line. Would you please advice me how can I rotate rigid wall, for cfd (2phase moving mesh module) analysis without much mesh deformation? I could not use rotating machinery module because rotating direction change with respect to time. Any help please...

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 2011/10/30 8:53 GMT-4
Hi

are you saying you want to study the transport and rotation of a solid rectangle pushed by a fluid ? If so your mesh will be easily twisted and ou nee to remesh often.
Or you try with a rotatonary circle, centred on the rectangle and you use the assembly/rotationary machiney "trick" on thies moving circle boundary. Some equation writing, but I believe it would work

--
Good luck
Ivar
Hi are you saying you want to study the transport and rotation of a solid rectangle pushed by a fluid ? If so your mesh will be easily twisted and ou nee to remesh often. Or you try with a rotatonary circle, centred on the rectangle and you use the assembly/rotationary machiney "trick" on thies moving circle boundary. Some equation writing, but I believe it would work -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/10/30 21:00 GMT-4
Dear Dr. Ivar,
Thanks for your thoughtful reply.
Yes my problem is about a body with two parts rotating in opposite direction and changes direction with respect to time. I want to study the net force experienced by body over a time period of one cycle (or in particular I am interested in finding net displacement of the body after a cycle of flapping).
I could not give rotating fluid domain as the parts are overlapping (please see the attached figure).
My problem is moving these two parts (2d boxes) in semi-circular paths back and forth with respect to time.
Would you please advice how can I achieve this?

PS: In my language (Tamil), Ivar stands for water, land, air, fire and atmosphere which present-day CFD handleswith. I am following all your posts. your detailed answers showing your expertise and letting us know and wonder how physics problems can be approached in comsol.

With regards,
Siva
Dear Dr. Ivar, Thanks for your thoughtful reply. Yes my problem is about a body with two parts rotating in opposite direction and changes direction with respect to time. I want to study the net force experienced by body over a time period of one cycle (or in particular I am interested in finding net displacement of the body after a cycle of flapping). I could not give rotating fluid domain as the parts are overlapping (please see the attached figure). My problem is moving these two parts (2d boxes) in semi-circular paths back and forth with respect to time. Would you please advice how can I achieve this? PS: In my language (Tamil), Ivar stands for water, land, air, fire and atmosphere which present-day CFD handleswith. I am following all your posts. your detailed answers showing your expertise and letting us know and wonder how physics problems can be approached in comsol. 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 2011/11/01 6:16 GMT-4
Hi

OK I get a better grasp of it like that ;)

It's true that you could do as well with only a line instead of the surface for the flap. And I suppose you have a "hinge" it the convergence point of the two flaps there.

Two things that strike me:
1) you do not have a long inlet, not outlet, compared to the diameter, be sure you flow is well developed, or add some length, if you have an inlet velocity?

2) you have a symmetrical design, one could dare a hypothesis that its symmetrical with respect to the horizontal axis, hence start with a simpler case, or ?

To force the flaps to move/rotate, you could perhaps try to add a moment on the hinge side (where the two connect), or you impose the displacement via a coordinate system at the hinge point. you could probably also fix the hinge point and say that the motion forwards or backwards is the travel of the fluid along the horizontal axis, depends where you put your reference frame, on the shore or on the boat (Euler or Lagrange, or a mix ?) (check on Wiki for the definitions) you need then to define correctly the inlet/outlet boundaries, or just free-flow (not sure its in all physics)


And for my first name "Ivar" I have read somewhere that it originates from "ivharr" the "solder to foot with the if tree", which translates to the archer. But I do not know if it's really true, it supposely dates back to my forefathers the Vikings some thousands year ago. But I believe I heritated it from my father's uncle name ;) Indeed words and sounds have different meanings in different countries.
--
Good luck
Ivar
Hi OK I get a better grasp of it like that ;) It's true that you could do as well with only a line instead of the surface for the flap. And I suppose you have a "hinge" it the convergence point of the two flaps there. Two things that strike me: 1) you do not have a long inlet, not outlet, compared to the diameter, be sure you flow is well developed, or add some length, if you have an inlet velocity? 2) you have a symmetrical design, one could dare a hypothesis that its symmetrical with respect to the horizontal axis, hence start with a simpler case, or ? To force the flaps to move/rotate, you could perhaps try to add a moment on the hinge side (where the two connect), or you impose the displacement via a coordinate system at the hinge point. you could probably also fix the hinge point and say that the motion forwards or backwards is the travel of the fluid along the horizontal axis, depends where you put your reference frame, on the shore or on the boat (Euler or Lagrange, or a mix ?) (check on Wiki for the definitions) you need then to define correctly the inlet/outlet boundaries, or just free-flow (not sure its in all physics) And for my first name "Ivar" I have read somewhere that it originates from "ivharr" the "solder to foot with the if tree", which translates to the archer. But I do not know if it's really true, it supposely dates back to my forefathers the Vikings some thousands year ago. But I believe I heritated it from my father's uncle name ;) Indeed words and sounds have different meanings in different countries. -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/11/02 8:27 GMT-4
Thank you very much Dr. Ivar.
I am doing by using prescribing a point on those rectangles.
My case is open boundary (flapping makes fluid to flow -thereby flappers moving front).

Thanks a lot for all your inputs I am doing all the possible methods and post it outcomes.

PS: See how names are different in different languages.But one of my grandfather used to tell me people 'get' their name they deserve. or their life will be in the way of their name.


with lots of regards,
Siva
Thank you very much Dr. Ivar. I am doing by using prescribing a point on those rectangles. My case is open boundary (flapping makes fluid to flow -thereby flappers moving front). Thanks a lot for all your inputs I am doing all the possible methods and post it outcomes. PS: See how names are different in different languages.But one of my grandfather used to tell me people 'get' their name they deserve. or their life will be in the way of their name. with lots of 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 2011/11/02 11:01 GMT-4
Hi

well but life also changes, no need for archer today, much better to get the arrows to tickle the will to learn physics, and that is what COMSOL is for, at least in my opinion, so, so long I believe my arrows help to push other to physics I'll continue to be an archer, but I must admit I have never tried a bow ;)

--
Good luck
Ivar
Hi well but life also changes, no need for archer today, much better to get the arrows to tickle the will to learn physics, and that is what COMSOL is for, at least in my opinion, so, so long I believe my arrows help to push other to physics I'll continue to be an archer, but I must admit I have never tried a bow ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/11/03 23:11 GMT-4
Thank you very much for your reply Dr. Ivar.
As per your advice I made simple 2d geometry. With hinge at (0,0). I applied a time dependent point load expecting that it will turn back after a particular time. But it did not work.
I also tried for applied moment using rigid connector, body load etc. didnot work either.
Would you please kindly check the file and let me know what mistakes I am doing? It would be great if i can change the angle between flaps with respect to time in a simplest way without taking much memory.

All your answers are nothing but arrows enlightening our physics knowledge and put it into real field test. I dont think that there is a better way to learn physics than using COMSOL. Once upon a time lab experiments determined the fate of theories and postulates, but your comsol put everything into scrutiny that too even a novice with little knowledge can play with any hypothesis. You are done!


Thank you very much for your reply Dr. Ivar. As per your advice I made simple 2d geometry. With hinge at (0,0). I applied a time dependent point load expecting that it will turn back after a particular time. But it did not work. I also tried for applied moment using rigid connector, body load etc. didnot work either. Would you please kindly check the file and let me know what mistakes I am doing? It would be great if i can change the angle between flaps with respect to time in a simplest way without taking much memory. All your answers are nothing but arrows enlightening our physics knowledge and put it into real field test. I dont think that there is a better way to learn physics than using COMSOL. Once upon a time lab experiments determined the fate of theories and postulates, but your comsol put everything into scrutiny that too even a novice with little knowledge can play with any hypothesis. You are done!


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 2011/11/04 5:57 GMT-4
Hi

I havent had time to goo fully through your model, but I believe you are "overdoing it slightly". You have large deformation engaged on the structural, but your "cast" iron parts will not be severely bend by water I beleive (it might be a COMSOL default but in your case its not requried, it adds alo non-linearity to your model, hence its tougher to simulate).

Then formally, do you need the structure ? if you say only fluid and assume that the "steel" part is infinite rigid, you can remove the structural part and remain with fluid and ALE / deformed geometry physics to simulate the deformation. But this means using different physics. It would make a lighter model that should solve better.

Your point loads, as bolean functions are to close to a Dirac step, you should rather use the built in step() or pulse() functions to smoothen the transitions

You need boundary layers for the mesh on the no-slip flap fluid boundaries.

And I beleive the mesh at the junction point of the two flaps will be very squeezed, it might be an idea to open up the flaps, with a small gap to help the mesh to "topologically survive"

then with all open boundaries (symmetry for the top bottom could probably also do, I suspect that you do not define a pressure value, I would add a p=0 point load on one of the external corners of your fluid volume

Nice problem, but there are many physics to sort out

--
Good luck
Ivar
Hi I havent had time to goo fully through your model, but I believe you are "overdoing it slightly". You have large deformation engaged on the structural, but your "cast" iron parts will not be severely bend by water I beleive (it might be a COMSOL default but in your case its not requried, it adds alo non-linearity to your model, hence its tougher to simulate). Then formally, do you need the structure ? if you say only fluid and assume that the "steel" part is infinite rigid, you can remove the structural part and remain with fluid and ALE / deformed geometry physics to simulate the deformation. But this means using different physics. It would make a lighter model that should solve better. Your point loads, as bolean functions are to close to a Dirac step, you should rather use the built in step() or pulse() functions to smoothen the transitions You need boundary layers for the mesh on the no-slip flap fluid boundaries. And I beleive the mesh at the junction point of the two flaps will be very squeezed, it might be an idea to open up the flaps, with a small gap to help the mesh to "topologically survive" then with all open boundaries (symmetry for the top bottom could probably also do, I suspect that you do not define a pressure value, I would add a p=0 point load on one of the external corners of your fluid volume Nice problem, but there are many physics to sort out -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/11/04 15:31 GMT-4
Thank you very much for your answers. i understand that i have to start with simple one.
i will do only with fluid without solid.
could you please tell ne which physics should i choose for ALE deformed geometry?
if could see the related tutorial, i hope i can learn.
please advice me.
thanks a lot for all your support.
Thank you very much for your answers. i understand that i have to start with simple one. i will do only with fluid without solid. could you please tell ne which physics should i choose for ALE deformed geometry? if could see the related tutorial, i hope i can learn. please advice me. thanks a lot for all your support.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011/11/09 4:25 GMT-5
Hello Siva,

I'm facing a similar problem and I'm using the Moving Mesh ALE (using a Prescribed Mesh Displacement). Unfortunately, I'm still trying to make the fluid move in response to the flap movement. Using the Moving Wall boundary condition seems to be the answer although I still didn't manage to make it work. I'm using a Parametric Curve (1D Wall) to represent the flap in a 2D simulation.

Cheers,
Marcelo
Hello Siva, I'm facing a similar problem and I'm using the Moving Mesh ALE (using a Prescribed Mesh Displacement). Unfortunately, I'm still trying to make the fluid move in response to the flap movement. Using the Moving Wall boundary condition seems to be the answer although I still didn't manage to make it work. I'm using a Parametric Curve (1D Wall) to represent the flap in a 2D simulation. Cheers, Marcelo

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.