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 The information provided may be out of date.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.

Defining wires in 3d models

Please login with a confirmed email address before reporting spam

I am trying to simulate a process that includes flow, electrostatics and mass conservation.

The model is simply a rectangular. One edge is grounded and the opposite includes 59 wires.

The grounded edge also includes a slit where particles are supposed to enter.

I simulate the wires by adding lines very close to each other (0.127 mm distance). Each "wire" is separated from the other (ie two lines form a single wire, no common edges). The space between the wires is supposed to be the insulation.

My problem is that I run out of memory every time I try to mesh, because the geometry is rather big compared to the "wire" size.

I have also tried to use a coarser (global) mesh without luck. I have also tried to divide my domain in half (one where the wires are located and one for the rest), without luck also.

Is there a better way to simulate wires in a 3-d model?
I have tried to use a single line as a wire but it will only allow me to set it as a line charge density. I want to input it as Volts.

I am using a very old version of FEMLAB (3.1) and I am also wondering if the current version works better in handling memory.

Thank you in advance

3 Replies Last Post 8 dic 2012, 03:50 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 5 dic 2012, 01:29 GMT-5
Hi

certainly the newer v4 can do much better than the old version, but it also depends on the PC hardware. But even, the most important I believe is to check your model to see if you cannot work in 2D rather than 3D, and to check if you cannot use surface boundaries, or edges/points as you suggested to mimic your thin wires. There are far more pre-cooked BC conditions in the newer v ersion

--
Good luck
Ivar
Hi certainly the newer v4 can do much better than the old version, but it also depends on the PC hardware. But even, the most important I believe is to check your model to see if you cannot work in 2D rather than 3D, and to check if you cannot use surface boundaries, or edges/points as you suggested to mimic your thin wires. There are far more pre-cooked BC conditions in the newer v ersion -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 7 dic 2012, 12:02 GMT-5
Thank you for your reply.

I am trying the new version and indeed works a lot better.

I am worried though because I know I get wrong results. I am trying to solve 3 PDE's. Two of which (electrostatics and laminar flow) are inputs to the 3rd on (advection - diffusion). The solution of the first two should not be very different if solved alone or in the system.

Is it possible to solve for the first two and use them as inputs in the third.
Can segregate solving do that? I have tried to use that but with poor results (system not converged). Any hints how that should be implemented.

My 3rd model (advection and diffusion) is non-linear as it uses differentation of the parameters solved for the other two as inputs.

Michael
Thank you for your reply. I am trying the new version and indeed works a lot better. I am worried though because I know I get wrong results. I am trying to solve 3 PDE's. Two of which (electrostatics and laminar flow) are inputs to the 3rd on (advection - diffusion). The solution of the first two should not be very different if solved alone or in the system. Is it possible to solve for the first two and use them as inputs in the third. Can segregate solving do that? I have tried to use that but with poor results (system not converged). Any hints how that should be implemented. My 3rd model (advection and diffusion) is non-linear as it uses differentation of the parameters solved for the other two as inputs. Michael

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 8 dic 2012, 03:50 GMT-5
Hi

Under the same study, a fully coupled solves all equations in one big matrix, segregated, in a loop with a given order, both cases the coupling si normally bidirectional. But you can also solve only for a few dependent variables in study1 "store the results into a first Data Set (=DB) and then solve only for the remaining dependent variables, you have than a truely one wy coupling from the first to the second study (check the tab where initial conditions are taken, for both solved and unsolved variables.

This is also sometims interestong then to change some BC (enable/disable per study) between the two study steps

--
Good luck
Ivar
Hi Under the same study, a fully coupled solves all equations in one big matrix, segregated, in a loop with a given order, both cases the coupling si normally bidirectional. But you can also solve only for a few dependent variables in study1 "store the results into a first Data Set (=DB) and then solve only for the remaining dependent variables, you have than a truely one wy coupling from the first to the second study (check the tab where initial conditions are taken, for both solved and unsolved variables. This is also sometims interestong then to change some BC (enable/disable per study) between the two study steps -- Good luck Ivar

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.