Divergence

[siggi84]

Hi, I am using Ansys to calculate temperature distributions in nanostructures. I am doing thermal-electric analysis in 3D and I am using the solid69 element.

The structure is made of silicon, silicon dioxide and platinum. On top of a thin silicon dioxide layer I have a nanowire which is heated resistively (by applying a voltage we get power dissipation P=I*V). Under the thin silicon dioxide layer I have thick layer of silicon.

My problem is that when I use the same thermal conductivity for both silicon dioxide and silicon I get convergence and everything works (k=1.36 W·m?1·K?1), but when I increase the thermal conductivity of silicon more the ~20 W·m?1·K?1 I always get divergence (the real value is around 150 W·m?1·K?1).

I have mostly been using the ICCG Solver and the PCG solver. I am using between 1 and 2 million nodes so I can not use the sparse solver.

What can be the reason for this? What can I do to get convergence?

 

[bagarbilla]

Maybe the difference in conductivity causing large enough temperature gradient which causing divergence. Try to increase nodes in the thin layer region of your model.
i hope i helped you.

 

[siggi84]

Thank you. I actually found one error in my code, but I also had to increase the number of nodes in the SiO2 layer.

Best regards,
Siggi84