Hello all, again I am trying to stretch my limits with Elmer, but this time I am not even sure if it is at all possible what I'd like to simulate. I need to model a fill study of a potting process, where a cavity around a detailed PCB has to be filled. So it would have to be observable, how the potting mass is gradually flowung and replacing the air, and where critical areas with remaining air would be... for a few "keywords" on what solvers to involve and which kinds of boundaries to set, I'd be grateful. So far my search has not shown any similar topics at all.
Thank you for your kind attention to my question.
Thomas
Trying to simulate a potting process
-
- Posts: 47
- Joined: 18 Jan 2019, 18:18
- Antispam: Yes
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Trying to simulate a potting process
Hi Thomas,
Haven't heard of potting before but looked some youtube video and I think I got the picture.
This is a challenging problem. You cannot use Lagrangian free surface techniques since the geometry changes. Hence you need to use Eulerian methods. Of these we have the Leveset method in Elmer, some codes use VOF (volume of fluid) method. Also I would think that also particle based methods are used for this kind of problem. Elmer does have particles but not in a way that would allow to model the flow itself with the particles. We could use the particles as trace particles (that might carry some property) but not for the flow equations. So we are really left with FlowSolver + Levelset.
Levelset uses smeared material properties where the density/viscosity changes based on the signed distance. That works better for fluids that are rather similar. Polymer & air is a more challenging combination. Also there are some challenges in the mold filling that I don't know how to address. For example, if we use no-slip BCs then basically there can be no velocity on the surface. How can then the surface get wetted? Probably on microlevel there are interesting things happening when the polymer folds on the surface having just normal velocity component. How would this be modeled with a coarse mesh? And how to deal with the encaptured air. Maybe one could end of with some kind of slip BCs not ease down the requirements on mesh density. I would guess that this interface physics has been studied somewhere. I'm just speculating here...
You will see the problems in the simple test case "levelset2". It is old, and probably many things could be made better now with little effort. For example, estimating the properties directly at integration points rather than nodes is currently supported but not used here. But these won't change the big picture, I think. There might be specialized codes for doing this kind of filling properly. Maybe OpenFOAM has something...
-Peter
Haven't heard of potting before but looked some youtube video and I think I got the picture.
This is a challenging problem. You cannot use Lagrangian free surface techniques since the geometry changes. Hence you need to use Eulerian methods. Of these we have the Leveset method in Elmer, some codes use VOF (volume of fluid) method. Also I would think that also particle based methods are used for this kind of problem. Elmer does have particles but not in a way that would allow to model the flow itself with the particles. We could use the particles as trace particles (that might carry some property) but not for the flow equations. So we are really left with FlowSolver + Levelset.
Levelset uses smeared material properties where the density/viscosity changes based on the signed distance. That works better for fluids that are rather similar. Polymer & air is a more challenging combination. Also there are some challenges in the mold filling that I don't know how to address. For example, if we use no-slip BCs then basically there can be no velocity on the surface. How can then the surface get wetted? Probably on microlevel there are interesting things happening when the polymer folds on the surface having just normal velocity component. How would this be modeled with a coarse mesh? And how to deal with the encaptured air. Maybe one could end of with some kind of slip BCs not ease down the requirements on mesh density. I would guess that this interface physics has been studied somewhere. I'm just speculating here...
You will see the problems in the simple test case "levelset2". It is old, and probably many things could be made better now with little effort. For example, estimating the properties directly at integration points rather than nodes is currently supported but not used here. But these won't change the big picture, I think. There might be specialized codes for doing this kind of filling properly. Maybe OpenFOAM has something...
-Peter
-
- Posts: 47
- Joined: 18 Jan 2019, 18:18
- Antispam: Yes
Re: Trying to simulate a potting process
Hello Peter,
thank you for checking on this and leaving your assessment. Honestly, as I am unfamiliar with many of the terms you used, I will probably not follow this further... the chunk to bite off is just too large. I know at least one package that contains solvers for this: Autodesk Moldflow. But its proprietary...
However I will check out the levelset example.
Thank you and keep up the good work!
Thomas
thank you for checking on this and leaving your assessment. Honestly, as I am unfamiliar with many of the terms you used, I will probably not follow this further... the chunk to bite off is just too large. I know at least one package that contains solvers for this: Autodesk Moldflow. But its proprietary...
However I will check out the levelset example.
Thank you and keep up the good work!
Thomas
Re: Trying to simulate a potting process
Hello Thomas,
Although not a finite element based method, this link might be interesting for you.
https://github.com/InteractiveComputerG ... lisHSPlasH
Best Regards, Rich.
Although not a finite element based method, this link might be interesting for you.
https://github.com/InteractiveComputerG ... lisHSPlasH
Best Regards, Rich.
-
- Posts: 47
- Joined: 18 Jan 2019, 18:18
- Antispam: Yes
Re: Trying to simulate a potting process
Rich,
truly thank you very much for this pointer! I managed to build SPlisHSPlasH on my machine and played with a few examples - I think that's just what I need. Appreciate your help on this!
Greetings, Thomas.
truly thank you very much for this pointer! I managed to build SPlisHSPlasH on my machine and played with a few examples - I think that's just what I need. Appreciate your help on this!
Greetings, Thomas.
Re: Trying to simulate a potting process
Hello Thomas,
Glad that link is helpful for you. Maybe post a picture or two of your results!
Thanks, Rich.
Glad that link is helpful for you. Maybe post a picture or two of your results!
Thanks, Rich.