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 The information provided may be out of date.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.
Problem on 3d complex geometry CFD (problems are in the posts on April 27th&28th)
Posted Apr 22, 2015, 11:38 a.m. EDT Low-Frequency Electromagnetics, Computational Fluid Dynamics (CFD), Geometry Version 4.4 9 Replies
Please login with a confirmed email address before reporting spam
Hi!
I have using Comsol to draw a complex geometry as shown in the attachment, but it always has the error message: fail to insert point. And each time I refine the mesh, the insert point location might change... I am really frustrated since after reading many threads, I still can not figure out why.
Any help is really appreciated!
Thanks!!!
Leo
I have using Comsol to draw a complex geometry as shown in the attachment, but it always has the error message: fail to insert point. And each time I refine the mesh, the insert point location might change... I am really frustrated since after reading many threads, I still can not figure out why.
Any help is really appreciated!
Thanks!!!
Leo
Attachments:
9 Replies Last Post Apr 28, 2015, 3:13 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Leo,
The geometry contains a lot of very short edges, small and narrow faces, and faces with sharp angles that cause problems for the mesh generator. I recommend you to build this geometry by adding a work plane, defining a 2D geometry without any interior edges, and extruding it to 3D. You can then choose to mesh the 3D geometry using a Swept node or a Free Tetrahedral node.
Sincerely,
Bjorn Bretz
COMSOL Development
The geometry contains a lot of very short edges, small and narrow faces, and faces with sharp angles that cause problems for the mesh generator. I recommend you to build this geometry by adding a work plane, defining a 2D geometry without any interior edges, and extruding it to 3D. You can then choose to mesh the 3D geometry using a Swept node or a Free Tetrahedral node.
Sincerely,
Bjorn Bretz
COMSOL Development
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Bjorn Bretz,
Wow! I understand! Thank you so much! You really helped me a lot!
Best,
Leo
Wow! I understand! Thank you so much! You really helped me a lot!
Best,
Leo
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Bjorn Bretz,
I am sorry to trouble you again. By extruding the 2D symmetry to 3D is much more effective compared to my previous method. But it seems that the previous problem still exists. I wonder whether there is anyway to smooth the boundary?
I also found a strange thing, when changing the repair tolerance to a lower value(<10^(-6)), the extrusion of 2d plane to 3D will fail. Could you please explain the repair tolerance?
Many Thanks!
Best Regards,
Leo
I am sorry to trouble you again. By extruding the 2D symmetry to 3D is much more effective compared to my previous method. But it seems that the previous problem still exists. I wonder whether there is anyway to smooth the boundary?
I also found a strange thing, when changing the repair tolerance to a lower value(<10^(-6)), the extrusion of 2d plane to 3D will fail. Could you please explain the repair tolerance?
Many Thanks!
Best Regards,
Leo
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Leo,
When extruding from a work plane, COMSOL by default first unites all 2D objects in the work plane. This operation uses a repair tolerance to determine overlapping and intersecting vertices and edges. If the objects to unite are slightly misalign or contains very small entities, lowering this tolerance can lead to numerical problems.
I suggest that you manually unite all 2D objects and remove all interior boundaries before extruding by adding a Union node with the "Keep interior boundaries" option deselected. If the boundary of the 2D geometry contains very short edges the resulting extruded 3D geometry will typically contain corresponding narrow faces. You can then use the Ignore Edges or Form Composite Faces operation in 3D (available in the Virtual Operations menu) to join the narrow faces with adjacent faces.
I hope this helps.
Sincerely,
Bjorn Bretz
When extruding from a work plane, COMSOL by default first unites all 2D objects in the work plane. This operation uses a repair tolerance to determine overlapping and intersecting vertices and edges. If the objects to unite are slightly misalign or contains very small entities, lowering this tolerance can lead to numerical problems.
I suggest that you manually unite all 2D objects and remove all interior boundaries before extruding by adding a Union node with the "Keep interior boundaries" option deselected. If the boundary of the 2D geometry contains very short edges the resulting extruded 3D geometry will typically contain corresponding narrow faces. You can then use the Ignore Edges or Form Composite Faces operation in 3D (available in the Virtual Operations menu) to join the narrow faces with adjacent faces.
I hope this helps.
Sincerely,
Bjorn Bretz
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Bjorn Bretz,
Thank you so much for your reply. I can start running model now. Your suggestions are very very very helpful for me.
Many thanks!
Best,
Leo
Thank you so much for your reply. I can start running model now. Your suggestions are very very very helpful for me.
Many thanks!
Best,
Leo
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Björn,
I am so sorry to trouble you again. I have encountered a problem that I could not make it out.
My model now always gives messaga: fail to find consistent initial values. Last time step is not converged.
The model consists of electrostatic, transport of diluted species and laminar flow, to ensure species number is conserved, I also add a global ODE and DAEs part. And I use the default time dependent solver suggested by comsol ,i.e, the segregated groups and iterative.
There is no error message if I only calculate es,chds and ODE part. However, once I started calculating the laminar flow altogether, the error message comes out. My settings in laminar flow are:
1. inlet -laminar inflow average velocity equals to the outlet average velocity
2. outlet- pressure 0pa
3. other boundaries -wall slip length given
4. volume force-it is here that unites with other physics(expressed with gradient of species density and potential, expression should be allright since no unit mismatch indication)
5.initial values -velocity field 0,pressure 0
My guess is something wrong with initial values? But my model calcutes the process from static(velocity 0) to some time, the initial value can only be 0.
Thank you so much for your patience!
Best Regards,
Leo
I am so sorry to trouble you again. I have encountered a problem that I could not make it out.
My model now always gives messaga: fail to find consistent initial values. Last time step is not converged.
The model consists of electrostatic, transport of diluted species and laminar flow, to ensure species number is conserved, I also add a global ODE and DAEs part. And I use the default time dependent solver suggested by comsol ,i.e, the segregated groups and iterative.
There is no error message if I only calculate es,chds and ODE part. However, once I started calculating the laminar flow altogether, the error message comes out. My settings in laminar flow are:
1. inlet -laminar inflow average velocity equals to the outlet average velocity
2. outlet- pressure 0pa
3. other boundaries -wall slip length given
4. volume force-it is here that unites with other physics(expressed with gradient of species density and potential, expression should be allright since no unit mismatch indication)
5.initial values -velocity field 0,pressure 0
My guess is something wrong with initial values? But my model calcutes the process from static(velocity 0) to some time, the initial value can only be 0.
Thank you so much for your patience!
Best Regards,
Leo
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
If the pressure at the outlet is zero, then if the temperature is finite, then the number density is zero, and since flux is average velocity times number density the flux must be zero for finite velocity. Correct? I don't think this is what you want.
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Dear Daniel,
Thank you first for your reply.
I guess the flux you mentioned is the flux in chds model? In that model, I use no flux at side walls and periodic boundary condition at the begining and end points. The flux in chds contains diffusive flux and electric migration flux and ion number density*velocity.
The chds and es altogher could give me results, however, when combined with laminar flow, the errors comes out.
Best Regards,
Leo
Thank you first for your reply.
I guess the flux you mentioned is the flux in chds model? In that model, I use no flux at side walls and periodic boundary condition at the begining and end points. The flux in chds contains diffusive flux and electric migration flux and ion number density*velocity.
The chds and es altogher could give me results, however, when combined with laminar flow, the errors comes out.
Best Regards,
Leo
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hi,
Update of today's tests: I have ONLY computed the laminar flow part (remove the body force) in the complex 3d geometry, time dependent: range(0,10^(-7),10^(-5))s, set the boundaries conditions the same as the previous post.
Again the error comes out: Failed to find consistent initial values. Last time step is not converged.
I am using the fludic controlled mesh coarse for the whole but finer at all the edges.
Any suggestions are appreciated!
Best,
Leo
Update of today's tests: I have ONLY computed the laminar flow part (remove the body force) in the complex 3d geometry, time dependent: range(0,10^(-7),10^(-5))s, set the boundaries conditions the same as the previous post.
Again the error comes out: Failed to find consistent initial values. Last time step is not converged.
I am using the fludic controlled mesh coarse for the whole but finer at all the edges.
Any suggestions are appreciated!
Best,
Leo