Learning Forum

[Rony Das]

what are the boundary conditions when BFAST is used in LUMERICAL FDTD for 3D structures?

[Amrita Pati]

Hi Rony,

BFAST has an in-built algorithm for enforcing periodic boundary conditions in the transverse direction. The
dependence of the incidence angle on the wavelength (which is seen with Bloch boundaries) is removed by transforming the E and H fields using a new set of variables. The transformed fields are then split into additional terms, known as split field method to reformulate the FDTD update equations. So, the BFAST has its own boundary conditions independent of the other BCs used in FDTD. For more details on the formulation, please refer to the article linked below:

https://ieeexplore.ieee.org/document/6651692

 Regards,

Amrita

[Rony Das]

I don't understand when I used BFAST, the value of reflection was more than one, and X and Y BCs were warned by default BCs. Please let what will be the condition in this case.

[Amrita Pati]

Hi Rony,

The values of reflection greater than 1 could be due to many reasons. First, it is possible that the simulation did not have sufficient time to reach the auto shutoff min value. I would suggest you to increase the simulation time ( by 100 times) and check the simulation status after the run. A simulation status of 2 indicates that the simulation has reached the auto shutoff min. If this doesn't work I will suggest you to reduce the auto shutoff min value by 100 times (to 10e-07) and see if that helps. May I know what values of incident angles are you using?

There is no need to worry about the warning. Since BFAST has its own boundary conditions, the regular boundary conditions will not apply. The BFAST boundary conditions are automatically applied to the simulation which is indicated by the warning message. In other words, it emphasizes that regardless of what BCs you choose in the transverse direction, BFAST's own BCs will be applied. Also, would you also be able to share the reflection graph? Thank you!

 

Regards,
Amrita

[Rony Das]

But If I use 100 times more than the simulation time the time required is almost 340 hours like that. Is it possible? The incident angle were 10 and 20  used in separate simulations.

 

[Rony Das]

 

 

[Amrita Pati]

Hi Rony,

As you can see on the left side, the simulation has diverged so your results are not accurate. BFAST simulations are numerically more unstable than FDTD simulations. The following link provides more information on the possible causes of divergence and the steps to address it. Please go through the article and let me know if you have any questions.

https://optics.ansys.com/hc/en-us/articles/360034902273-Broadband-Fixed-Angle-Source-Technique-BFAST-

Regards,
Amrita