Radiating Boundary Condition

[houdali]

Hi,

After reading some patents and articles about the two method of simulation: Radiating Boundary Condition and PML( Perfectly Matched Layer).
I concluded that PML is better than RBC, because the PML have a property to absorb the incident wave from antenna, without reflecting energy into the antenna. While with the RBC we can have a small reflection into the interior.

So, Why we use the Radiating Boundary Condition if the PML is better?
What is the advantage of the RBC?

Thanks

 

[tallface65]

The ABC is computationally cheaper as it is merely a boundary condition that terminates the tangential E field while ignoring the normal component (hence the angle of incidence effect) while the PML is a 3D object that must be meshed and solved inside. Also, ABC boundaries are fine as long as you have a single radiator and the fields incident upon it are near normal (say lambda/4 away from radiator). PMLs are certainly preferred fro most antenna simulations and especially for multiple antennas in a single model simulations, but they can be computationally more expensive.

I like to think of ABC boundaries as 'good' and PMLs as 'great'. If I need absolute accuracy...PML if I need quick simulations to understand trends...ABC. It is important to have the option.

 

[houdali]

Thanks for your attention.

But the ABC boundaries is also a 3D object because you must to select a air box . and the air box must be also meshed and solved.

 

[tallface65]

You are incorrect. The air does indeed need to be meshed and solved, but that is simply due to the fact that HFSS solves explicitly for E and thus need to include fringing fields to correctly represent the electromagnetics of the problem. You can have an air box without a radiation boundary...however what do the fields do at the surfaces??? You need a boundary condition applied to the surface of your air box to terminate the fields in a matched condition if you want to prevent fields from reflecting back into your domain. If there is no boundary assigned...something is still needed from a computational perspective and a PEC is applied.

Again...All boundary conditions are assigned to faces of a 3D objects. However the Boundary condition itself is only operated on the surface and the Radiation boundary works as I described above.

 

[venu.gongal]

Hi
In addition to those discuused...

One more thing which I observed is that by using RBC simulations doesnot take long time for single radiators and simple antennas. But when u go for multiband antennas, more than one resonance PML seems to be much faster.

Otherwise usage of either PML or RBC is a just a matter of effient use of computational resource and the accuracy of ur result

 

[Azulykit]

I concur with the general comments above. PML's are more complex to create compared to an ABC. Another warning, If you are using a parametric model keep a close eye on any PML's. You may find that the parametric modeling does not adjust the PML as one would expect. It is sometimes necessary to re-create a PML after a model is parametrically adjusted. I have not seen similar failures when using ABC's.

 

[venu.gongal]

Hi

Do u mean when using ABC conditions and try parametric modeling the ABC box resizes itself ( or the size itself is lambda/4 doesn't effect hugely while performing parametric modeling)....... can u make it clearer??

I haven't encountered this problem while performing parametric variables using PML....

Thanks
venu

 

[Azulykit]

I have occasionally had PML boundaries not resize themselves as I expected and intended. The ABC boundaries have resized more easily. I have not seen, nor have I looked, for this in version 13. I bring it up as something to keep an eye on as a potential pitfall. It does not happen in all cases. It is not a big deal to manually create a PML with an otherwise parametric model. As I said, I like to keep an eye on PML's so I am not surprised.

HFSS is an excellent tool, just keep an eye on this aspect.

 

[houdali]

Thanks amigos