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.

Can a 3D "Mapped" mesh have elements of definite proportions (1:5:10)?

Please login with a confirmed email address before reporting spam

Hello,

I have meshed a 3D object using the Mapped mesh. My object is a sum of rectangular elements, and therefore this is most convenient solution (tetrahedron - too many elements, quad doesn't want to mesh 3D, this was easiest) . Since I have a high aspect ratio, I would like to have directional proportions of the element size (I manually adjusted the size of the minimum element but the ratio 1:1:1- ideally I would like to have 1:5:10) - is this option possible for mapped meshing?

Thanks


8 Replies Last Post 20 mai 2012, 08:26 UTC−4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 nov. 2011, 18:52 UTC−5
I kind of figured out this one on my own . Acctualy I was doing everything right it's just that the "distribution" in the mesh section was not physicaly above the mapped mesh icon.
I kind of figured out this one on my own . Acctualy I was doing everything right it's just that the "distribution" in the mesh section was not physicaly above the mapped mesh icon.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 nov. 2011, 19:04 UTC−5
Now that the meshing works I get the follwoing

error

Failed to find consistent initial values.
Singular matrix.

For mesh-case 0 there are 141198 void equations (empty rows in matrix) for the variable mod1.T
at coordinates: (0.0004,0.0381,0.0106667), (0.0004,0.0381,0.0105185), (0.0004,0.0381,0.0103704), (0.0005,0.0381,0.0106667), (0.0005,0.0381,0.0105185), ...
and similarly for the degrees of freedom (empty columns in matrix)
Last time step is not converged.

and a warning

Not all domains are meshed

Mesh - mapped - 40k elements, 3D structure, all materials are assigned. What am I doing wrong?
Now that the meshing works I get the follwoing error Failed to find consistent initial values. Singular matrix. For mesh-case 0 there are 141198 void equations (empty rows in matrix) for the variable mod1.T at coordinates: (0.0004,0.0381,0.0106667), (0.0004,0.0381,0.0105185), (0.0004,0.0381,0.0103704), (0.0005,0.0381,0.0106667), (0.0005,0.0381,0.0105185), ... and similarly for the degrees of freedom (empty columns in matrix) Last time step is not converged. and a warning Not all domains are meshed Mesh - mapped - 40k elements, 3D structure, all materials are assigned. What am I doing wrong?

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 26 nov. 2011, 03:45 UTC−5
Hi

singular matrix means essentially that your BC are not defined in sufficient numer or independently, or that you are missing some material data such that you model is not fully defined, there might be several solutions, and the maths, hence also COMSOL cannot decide for you which one to go for.
My advice, cgo through each physics, step by step and be sure you have given enough BC definitions for each to give an unique results. And check that all materials variables exist.

--
Good luck
Ivar
Hi singular matrix means essentially that your BC are not defined in sufficient numer or independently, or that you are missing some material data such that you model is not fully defined, there might be several solutions, and the maths, hence also COMSOL cannot decide for you which one to go for. My advice, cgo through each physics, step by step and be sure you have given enough BC definitions for each to give an unique results. And check that all materials variables exist. -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 nov. 2011, 15:56 UTC−5
Thank you for your insight. I am aware that a singular matrix has a 0 on the main diagonal - therefore not solvable.

All my materials are defined sufficiently for the heat-transfer in solids to be solved (cp - heat capacity, rho-density and k-thermal conductivity, where k is the only needed).

I have meshed the model using both mapped and triangular with and wihtout the asigned distribution. -- Same warrning

I have changed all meatrials to one -- Si --Same warrning

I have treated the geometry as both one body and as an assembly (by introducing all the contact pairs and heat continuity in the BC) -- Same warrning

I reduced my BC to only two - Same warrning

So what ever I did, the warrning "not al domains are meshed" was the same, but the error message changed

I believe it's a trivial mistake but I can't see to find it, and the model itself is simple in structure (though a high aspect ratio one).

I'll try redesigning the model from scratch. If the same problem occurs - you'll be hereing from me.


thanks for the reply
Thank you for your insight. I am aware that a singular matrix has a 0 on the main diagonal - therefore not solvable. All my materials are defined sufficiently for the heat-transfer in solids to be solved (cp - heat capacity, rho-density and k-thermal conductivity, where k is the only needed). I have meshed the model using both mapped and triangular with and wihtout the asigned distribution. -- Same warrning I have changed all meatrials to one -- Si --Same warrning I have treated the geometry as both one body and as an assembly (by introducing all the contact pairs and heat continuity in the BC) -- Same warrning I reduced my BC to only two - Same warrning So what ever I did, the warrning "not al domains are meshed" was the same, but the error message changed I believe it's a trivial mistake but I can't see to find it, and the model itself is simple in structure (though a high aspect ratio one). I'll try redesigning the model from scratch. If the same problem occurs - you'll be hereing from me. thanks for the reply

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 26 nov. 2011, 17:40 UTC−5
Hi

not all domains meshed means that you have a domain = solid in 3D or surface in 2D that is not meshed. You have probably meshed the boundaries (surfaces) but not the domains (volumes)

Why not use the default physics induced mesh once, then start to refine manually ?

--
Good luck
Ivar
Hi not all domains meshed means that you have a domain = solid in 3D or surface in 2D that is not meshed. You have probably meshed the boundaries (surfaces) but not the domains (volumes) Why not use the default physics induced mesh once, then start to refine manually ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 nov. 2011, 13:06 UTC−5
Huh,huh,huh...although I love simulations - they sometimes are simply anoying (nothing a sixpack of beer and a few friends cannot resolve).

I have tried with general hysics seting and the simulation works. Thank you, I will be coninuing my work using that setup.

Why in my model I insist on mapped meshing is the following reason - one of my consisiting elements is a 100µm thick tape that covers a microscope cover slip (76.2x25.4x0.13mm). I reduced the calculation by introducing two axes of symetry and yet teh following is the issue (I wont say problem cause it will get done)

finest triangular mesh that does not meet basic convergence criteria (3 elements across smallest gap) - 780 000 elements

a mapped mesh with 7 elements across each of the thin layers -80 000 elements

Since this is my thesis work and research in progress, I'm reluctant to upload the file, but here is an image of the device just to have an idea. If you are interested in teh model and it's problems from a software troubleshouting point of view, I can send it to you.

Again, thank you for your help, the feedback is greatly appreciated
Huh,huh,huh...although I love simulations - they sometimes are simply anoying (nothing a sixpack of beer and a few friends cannot resolve). I have tried with general hysics seting and the simulation works. Thank you, I will be coninuing my work using that setup. Why in my model I insist on mapped meshing is the following reason - one of my consisiting elements is a 100µm thick tape that covers a microscope cover slip (76.2x25.4x0.13mm). I reduced the calculation by introducing two axes of symetry and yet teh following is the issue (I wont say problem cause it will get done) finest triangular mesh that does not meet basic convergence criteria (3 elements across smallest gap) - 780 000 elements a mapped mesh with 7 elements across each of the thin layers -80 000 elements Since this is my thesis work and research in progress, I'm reluctant to upload the file, but here is an image of the device just to have an idea. If you are interested in teh model and it's problems from a software troubleshouting point of view, I can send it to you. Again, thank you for your help, the feedback is greatly appreciated


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 27 nov. 2011, 14:44 UTC−5
Hi

Indeed one of the drawbacks of 3D modelling is the large numebr of elements needed, with a "brute force" approache. So I believe youre stuck to find a more clever way, i.e. 2D simplification, use of symmetry, and/or replacement of thin structures by surfaces with bulk behaviour via some additional physics

--
Good luck
Ivar
Hi Indeed one of the drawbacks of 3D modelling is the large numebr of elements needed, with a "brute force" approache. So I believe youre stuck to find a more clever way, i.e. 2D simplification, use of symmetry, and/or replacement of thin structures by surfaces with bulk behaviour via some additional physics -- Good luck Ivar

Victor Manuel Rodriguez Zermeno

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 mai 2012, 08:26 UTC−4

Now that the meshing works I get the follwoing

error

Failed to find consistent initial values.
Singular matrix.

For mesh-case 0 there are 141198 void equations (empty rows in matrix) for the variable mod1.T
at coordinates: (0.0004,0.0381,0.0106667), (0.0004,0.0381,0.0105185), (0.0004,0.0381,0.0103704), (0.0005,0.0381,0.0106667), (0.0005,0.0381,0.0105185), ...
and similarly for the degrees of freedom (empty columns in matrix)
Last time step is not converged.

and a warning

Not all domains are meshed

Mesh - mapped - 40k elements, 3D structure, all materials are assigned. What am I doing wrong?


I think you are not meshing the inside of the material, to do so, you can try a sweep mesh (after you meshed the outer faces (at least 1)).

Alternatively, you can use the free tetrahedral and the scale geometry feature to reach the 1:5:10 desired, however, you need to have your domain aligned in the x y z....
[QUOTE] Now that the meshing works I get the follwoing error Failed to find consistent initial values. Singular matrix. For mesh-case 0 there are 141198 void equations (empty rows in matrix) for the variable mod1.T at coordinates: (0.0004,0.0381,0.0106667), (0.0004,0.0381,0.0105185), (0.0004,0.0381,0.0103704), (0.0005,0.0381,0.0106667), (0.0005,0.0381,0.0105185), ... and similarly for the degrees of freedom (empty columns in matrix) Last time step is not converged. and a warning Not all domains are meshed Mesh - mapped - 40k elements, 3D structure, all materials are assigned. What am I doing wrong? [/QUOTE] I think you are not meshing the inside of the material, to do so, you can try a sweep mesh (after you meshed the outer faces (at least 1)). Alternatively, you can use the free tetrahedral and the scale geometry feature to reach the 1:5:10 desired, however, you need to have your domain aligned in the x y z....

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.