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.
Using symmetry to create a heat sink
Posted 13.04.2017, 17:39 GMT-4 Heat Transfer & Phase Change, Geometry, Mesh, Studies & Solvers Version 5.2 2 Replies
Please login with a confirmed email address before reporting spam
Hi guys,
I am looking to design a heat sink with pin fins however I'm finding that if I make a full design it takes ages to compute. How can I use some form of symmetry so that the software only has to compute a small part of the array, rather than the whole thing? I have a small heat source in the centre of my sink (the sink sits on a chip that dissipates heat). Would that change anything?
Thank you very much
I am looking to design a heat sink with pin fins however I'm finding that if I make a full design it takes ages to compute. How can I use some form of symmetry so that the software only has to compute a small part of the array, rather than the whole thing? I have a small heat source in the centre of my sink (the sink sits on a chip that dissipates heat). Would that change anything?
Thank you very much
2 Replies Last Post 14.04.2017, 10:12 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
Hello George,
The software does offer capabilities to take advantage of symmetry and periodicity, if they exist in your physical phenomenon.
Here are some pointers:
www.comsol.com/support/knowledgebase/1038/
www.comsol.com/model/heat-sink-8574
www.comsol.com/model/forced-air-cooling-with-heat-sink-22041
Best,
Jeff
The software does offer capabilities to take advantage of symmetry and periodicity, if they exist in your physical phenomenon.
Here are some pointers:
www.comsol.com/support/knowledgebase/1038/
www.comsol.com/model/heat-sink-8574
www.comsol.com/model/forced-air-cooling-with-heat-sink-22041
Best,
Jeff
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
George-
Exploiting symmetry is a great way to reduce computational expense.
First, consider if perhaps 2D axisymmetry could work for you (revolution symmetry with no change in azimuthal direction).
If no, consider if 1 or more mirror planes would produce the same results (geometry + physics) on either side of the plane. Start slicing and dicing the full 3D geometry. Keep track of symmetry planes and impose zero flux conditions on these planes.
Important point: Always remember to reduce any TOTAL loading (heat source) amounts to be appropriate (e.g. a half symmetry total heat source should be reduced by a factor of 2.) If using a heat density however do NOT reduce the amount as the denominator in the term (W/m^3) will adjust for dimensionality reduction accordingly.
Also, one nice feature in COMSOL is that you can re-constitute the full 3D geometries using revolved, sector, or mirror data sets under Results>Data Sets. No one will ever know you only solved on reduced dimensionality! Of course, you know the results would be the same ;)
Good luck!
Josh
--
Best regards,
Josh Thomas
AltaSim Technologies
OOPS -- Just saw Jeff beat me to a response! Haha.
Exploiting symmetry is a great way to reduce computational expense.
First, consider if perhaps 2D axisymmetry could work for you (revolution symmetry with no change in azimuthal direction).
If no, consider if 1 or more mirror planes would produce the same results (geometry + physics) on either side of the plane. Start slicing and dicing the full 3D geometry. Keep track of symmetry planes and impose zero flux conditions on these planes.
Important point: Always remember to reduce any TOTAL loading (heat source) amounts to be appropriate (e.g. a half symmetry total heat source should be reduced by a factor of 2.) If using a heat density however do NOT reduce the amount as the denominator in the term (W/m^3) will adjust for dimensionality reduction accordingly.
Also, one nice feature in COMSOL is that you can re-constitute the full 3D geometries using revolved, sector, or mirror data sets under Results>Data Sets. No one will ever know you only solved on reduced dimensionality! Of course, you know the results would be the same ;)
Good luck!
Josh
--
Best regards,
Josh Thomas
AltaSim Technologies
OOPS -- Just saw Jeff beat me to a response! Haha.