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.
3D mesh and mysterious triangular shape
Posted 2010年7月2日 GMT-4 06:02 2 Replies
Please login with a confirmed email address before reporting spam
Hello everybody,
I have some problems with my 3D-mesh of my geometry (just a folded tube).
I used 1600 tetrahedrons using the mesh automatic procedure of COMSOL and using a "normal" size.
The quality of the mesh is good (quality = 1) at the leg of my tube but bad at the body (quality = 0.4) (see the attachement)...
Due to this mesh (I think), the solution of my equations (the solution is the tension constraint on the surface of the tube) show some defaults...I mean some "triangular shapes" are visible looking at the solution (see the attachement)...I think that this default is due to my mesh (probably a little bad).
You will find in attachement two picturew showing the tube with mesh and the tube with the solution plotted (tension constraint at the surface of the tube). The defaults are visible looking at the solution (some "triangular shapes" unknown are visible on the tube in the solution)...
I do not know how I can remove these defaults...I tried to use a fine and a finer mesh but the problem is always there...Also, I tried to change the shape of the mesh but nothing is better...
Have you already see this problem in your model ? Have you a solution to solve it ?
Thank you very much for your help !
Best
Pierre
I have some problems with my 3D-mesh of my geometry (just a folded tube).
I used 1600 tetrahedrons using the mesh automatic procedure of COMSOL and using a "normal" size.
The quality of the mesh is good (quality = 1) at the leg of my tube but bad at the body (quality = 0.4) (see the attachement)...
Due to this mesh (I think), the solution of my equations (the solution is the tension constraint on the surface of the tube) show some defaults...I mean some "triangular shapes" are visible looking at the solution (see the attachement)...I think that this default is due to my mesh (probably a little bad).
You will find in attachement two picturew showing the tube with mesh and the tube with the solution plotted (tension constraint at the surface of the tube). The defaults are visible looking at the solution (some "triangular shapes" unknown are visible on the tube in the solution)...
I do not know how I can remove these defaults...I tried to use a fine and a finer mesh but the problem is always there...Also, I tried to change the shape of the mesh but nothing is better...
Have you already see this problem in your model ? Have you a solution to solve it ?
Thank you very much for your help !
Best
Pierre
Attachments:
2 Replies Last Post 2010年7月3日 GMT-4 05:02
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
For me these are natural "ripples" from the higher order polynomial fits, try a direct mesh wireframe plot witout interpolation and you will get the "coarse" true FEM view.
What I do not understand though are the rainbow waving variations but as I'm not sure about your oad case or which variable you are showing, it could well be perfectly correct.
Good luck
ivar
For me these are natural "ripples" from the higher order polynomial fits, try a direct mesh wireframe plot witout interpolation and you will get the "coarse" true FEM view.
What I do not understand though are the rainbow waving variations but as I'm not sure about your oad case or which variable you are showing, it could well be perfectly correct.
Good luck
ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I Ivar, I everybody,
Ivar, thank you for your answer ...
To be more clear, I tried another way to make my meshage to get better results. The tube (volume) embedded in a cube (which I call the environment and is also a volume). My calculation is simply to solve an integral on the surface of the tube. In the calcul of this integral I have to use the normal vector unit at the surface of my tube.
[I use :
Nx = (up(dom)>down(dom))*unx+(up(dom)<=down(dom))*dnx
Ny = (up(dom)>down(dom))*uny+(up(dom)<=down(dom))*dny
Nz = (up(dom)>down(dom))*unz+(up(dom)<=down(dom))*dnz]
The solution show in the attachement the constraint tension on the surface of the tube.
1) When I mesh my tube (a volume) (I do not mesh the environment), I get a really good solution and no horrible tasks appear in the solution -----> good (see the attachement)
2) If I both mesh tube (fine mesh) AND cube (a volume) (coarse mesh) into which my tube is embedded, I find the same solution to the previous BUT with some horrible tasks located in the solution ! (see the attachement)
Unfortunately I need this additional environment for the rest of my work.
I do not understand why the fact to mesh the external environment gives these tasks in the solution ! The presence of the mesh of the cube should not be a problem... More, I get some results completely incoherent when the tube is meshed only in surface ...and the normal vectors (Nx, Ny, Nz) are not good oriented.
Have you encountered this problem in your models ?
Unfortunately I cannot send you my complete modele due to the property of this work.
Thank you again for your help.
Thank you Ivar again !
Pierre
Ivar, thank you for your answer ...
To be more clear, I tried another way to make my meshage to get better results. The tube (volume) embedded in a cube (which I call the environment and is also a volume). My calculation is simply to solve an integral on the surface of the tube. In the calcul of this integral I have to use the normal vector unit at the surface of my tube.
[I use :
Nx = (up(dom)>down(dom))*unx+(up(dom)<=down(dom))*dnx
Ny = (up(dom)>down(dom))*uny+(up(dom)<=down(dom))*dny
Nz = (up(dom)>down(dom))*unz+(up(dom)<=down(dom))*dnz]
The solution show in the attachement the constraint tension on the surface of the tube.
1) When I mesh my tube (a volume) (I do not mesh the environment), I get a really good solution and no horrible tasks appear in the solution -----> good (see the attachement)
2) If I both mesh tube (fine mesh) AND cube (a volume) (coarse mesh) into which my tube is embedded, I find the same solution to the previous BUT with some horrible tasks located in the solution ! (see the attachement)
Unfortunately I need this additional environment for the rest of my work.
I do not understand why the fact to mesh the external environment gives these tasks in the solution ! The presence of the mesh of the cube should not be a problem... More, I get some results completely incoherent when the tube is meshed only in surface ...and the normal vectors (Nx, Ny, Nz) are not good oriented.
Have you encountered this problem in your models ?
Unfortunately I cannot send you my complete modele due to the property of this work.
Thank you again for your help.
Thank you Ivar again !
Pierre
Attachments: