Heat 2D and 3D simulation results are very different, how to fix it?



Hello, Lumerical Team,

I’m using the Heat solver in Device to simulate a thermal tuner with air insulations on both sides.


However, the temperature results from 2D and 3D simulations in the waveguide region are very different, even though the configurations are the same (or I didn’t catch it).

For example:
The 2D simulation result gives a max temperature 323.295K in the waveguide region, whereas the 3D simulation result gives a max temperature 396.216K.

I also have some measured data from a real thermal tuner, and its performance lies between the 2D and 3D cases in terms of Pi phase power. But the experiment data still have a significant difference from the simulation results.

For example:

2D simulation -> 11.5137mW/Pi

Experiment Pi shift power is 8.3565mW/Pi,

3D simulation -> ~3.7mW/Pi

My problem is: what can be the reason that causes this big temperature difference in 2D and 3D simulation. How can I fix this problem to get both 2D and 3D simulation results close to my experiment data?

The Device model is attached in the following.
Simulation2D.ldev (5.9 MB)

Simulation3D.ldev (5.9 MB)

I really appreciate any suggestion and help from you!

W. Wang


Hi @wwang73

I modified the 3D simulation file and used a smaller simulation region with a coarser mesh. Results almost match the 2D simulations and I expect them to match even better with a refined mesh:

2D simulations:

3D simulations:

I have attached simulation file for your review:
Simulation3D_BK.ldev (5.9 MB)

Hope this solves the problem.


Thank you! I will try this and get back to you if I have more questions!



A post was split to a new topic: Heat source definition for phase shifter simulation