O***p
发帖数: 1333 | 1 What's the best way to calculate the spreading resistance through multi-layer?
My backgroud is in EE but I need to calculate the thermal performace of our
high power module and it involves heat spreading through several layers of
substrates before reaching the heatsink. I have spoken with a few ME
professors in our school and they all said that there is no easy analytical
way of doing it. One numerical way is to use ANSYS but I am not sure if our
school has it or not. Is there any other software |
s*i
发帖数: 5025 | 2 It depends on the shape of the layers. For example, if they are relatively
thin layers, the one-dimensional heat transfer model could be enough. The
analytical solution should exist.
====================================== |
O***p
发帖数: 1333 | 3 Well, one lay is 100microns but it's copper, the other layer is 1mm but has a
thermal conductivity of between 30-45W/mk.
our
analytical
our
【在 s*i 的大作中提到】
: It depends on the shape of the layers. For example, if they are relatively
: thin layers, the one-dimensional heat transfer model could be enough. The
: analytical solution should exist.
: ======================================
|
s*i
发帖数: 5025 | 4 what is the XY dimension?
a
of
don't
【在 O***p 的大作中提到】
: Well, one lay is 100microns but it's copper, the other layer is 1mm but has a
: thermal conductivity of between 30-45W/mk.
:
: our
: analytical
: our
|
O***p
发帖数: 1333 | 5 from 9mmx9mm to 55mmx55mm substrate then to 80mmx80mm heatsink.
【在 s*i 的大作中提到】
: what is the XY dimension?
:
: a
: of
: don't
|
s*i
发帖数: 5025 | 6 Calculate the Biot Number for each bi-layer:
Bi = alpha*(V/A)/lamda, where alpha is heat transfer coefficient, V is volume,
A is area, and lamda is thermal conductivity.
I believe the number will be <0.1, which means one dimensional solution should
be fairly good.
【在 O***p 的大作中提到】
: from 9mmx9mm to 55mmx55mm substrate then to 80mmx80mm heatsink.
|
c*******e
发帖数: 8624 | 7 这个问题也许要用Green's function,因为对substrate和heatsink
来说boundary condition不是uniform的,不过我想如果用数值解法的话
选择余地应该比较大,比如用finite difference也许就不会很困难
呵呵,不如什么时候我帮你搞定,到时候加我个第二作者?
【在 O***p 的大作中提到】
: from 9mmx9mm to 55mmx55mm substrate then to 80mmx80mm heatsink.
|
O***p
发帖数: 1333 | 8 我的背景是EE，而且这是我的一个小task,
所以我想用最简单又不离谱的方法来approximate. |
c*******e
发帖数: 8624 | 9 我以前算过一个2-D的，就是长度方向是无限的
用的是finite difference,你这个是3-D的，
要困难不少。如果你就想用软件的话，简单
的，我估计像这种问题就FEMLAB这样的软件就
可以了。
【在 O***p 的大作中提到】
: 我的背景是EE，而且这是我的一个小task,
: 所以我想用最简单又不离谱的方法来approximate.
|
O***p
发帖数: 1333 | 10 Is there such software that's free or trial-version on the Internet? |
|
|
s*i
发帖数: 5025 | 11 如果你再仔细描述一下几何形体及边界条件，我给你来个解析的近似解吧。
【在 O***p 的大作中提到】
: Is there such software that's free or trial-version on the Internet?
|
t*n
发帖数: 14458 | 12 u may try 1-D heat conduction
it should give you reasonable results
there are solutions on textbooks
if you want more accurate results, write a finite difference code. pretty
simple for any ME guy in heat transfer area. you may wanna ask them.
【在 O***p 的大作中提到】
: from 9mmx9mm to 55mmx55mm substrate then to 80mmx80mm heatsink.
|
s*i
发帖数: 5025 | 13 Nod。热阻串联，和和
【在 t*n 的大作中提到】
: u may try 1-D heat conduction
: it should give you reasonable results
: there are solutions on textbooks
: if you want more accurate results, write a finite difference code. pretty
: simple for any ME guy in heat transfer area. you may wanna ask them.
|
O***p
发帖数: 1333 | 14 I think I got enough information for a good approximation, so thanks y'all. |
z***a
发帖数: 7 | 15 Albers, J., 1995, "An Exact Recursion Relation Solution for the
Steady-State Surface-Temperature of a General Multilayer Structure,"
IEEE Transactions on Components Packing and Manufacturing Technology
Part A, 18(1), pp.31-38.
Look at the references. The author published his Fortran codes in
a NIST special report. Good luck.
【在 O***p 的大作中提到】
: What's the best way to calculate the spreading resistance through multi-layer?
: My backgroud is in EE but I need to calculate the thermal performace of our
: high power module and it involves heat spreading through several layers of
: substrates before reaching the heatsink. I have spoken with a few ME
: professors in our school and they all said that there is no easy analytical
: way of doing it. One numerical way is to use ANSYS but I am not sure if our
: school has it or not. Is there any other software
|
