Modeling micron size features

[mazmpv94]

I am trying to do static thermal modeling of a box with the following dimensions:

W: 0.400mm
L: 1.500mm
T: 0.0035mm (3.5um)

One side of this box generates heat and flows to a big heat sink below at a fixed temp. The big heat sink is much bigger (100x) than the heat source.

After many adjustments, COSMOS sucessfully created a mesh for the model. However, when i apply a heat load on one of the surfaces of the little box, the heat fails to conduct to its adjacent surfaces. I've tried setting bonded contacts, but the heat just wont flow.....

Any thoughts?

Thanks!
M.

My Specs: P4 2.4Ghz, 1GHz RAM, not very good video card (so sloooow......upgrading soon)

 

[mazmpv94]

Forgot to mention, the heat load is applied to a split surface on the box. The width of the heat load surface is only 0.0053 mm in width..... very narrow!

I suspect somehow that's too small for the mesh to catch?

 

[gfbotha]

You might also investigate your problem by using the following relationship to calculate the required time step size:

L^2/(4.alpha)
where
L = element dimension or conducting length in heat flow direction
alpha = diffusivity = k/(ro.C)

 

[corus]

A time step won't help in a static/steady state heat transfer problem.

If you're worried about the dimensions of the problem then choose some other units, such as micro mm so that your dimensions are 400 x 1500 x 3.5 and modify your conductivity and heat transfer coefficients accordingly. I would doubt that the size of the heat source has little to do with the problem though, but perhaps other sources of heat loss dominate the heat flow rather than the conductivity to other surfaces. Check your material properties too to make sure you have the right units.

corus

 

[gfbotha]

Yes, you're right about time step not being applicable to steady state heat flow, Corus. In fact I read over "static" without realising he meant steady state.

That said, sometimes a different approach, like having a look at transient heat flow, also might be informative.

Regards