Bent waveguide simulation in FDTD



Can i simulate bent waveguide on FDTD to study the modal losses?


Bending loss decrease with larger bending?

Hi @nouman.zia

The best approach for this problem would be to use FDE solver, an example of which can be found here for the bent waveguide and then to run some converging testing until your loss is stable.

You can tackle the problem down with FDTD, however, it is not timely efficient. My idea would be to have the identical design in FDTD, inject the light through one side of the fibre and analyse the tranmission on the fibre output after the bend. Generally there are two sources of loss here, radiation loss as a result on bend and mode overlap between straight waveguide and the bend section, which FDTD will count for both effects automatically.

Bending loss and Propagation loss of Si Ring Modulator


Thank you for suggestion.

I would like to explain more my problem: I want to simulate relatively small bent ( 7 degree bent angle) of a ridge waveguide with one facet. My aim is to optimize my design to reduce radiation losses and increase reflection or modal reflection losses.

Right now i have only access to FDTD, but if I use small bent angle then is FDTD fine for my above mentioned problem?
I have attached a schematic to give idea of my geometry.
Thanks once again


Actually using FDTD for estimating the total loss of a bent waveguide is even more accurate than MODE. You can use a setup like the one below. It is the similar to the suggestion of @bkhanaliloo . Here I added input and output waveguide are extended to allow the excited mode to propagate before entering the bend. But the field monitors are placed at the input and output of the 180 bend. To estimate the loss, you can divide the transmission at the output over the transmission at the input port. The radius can be as small as 0.6 um. Do you have any estimation for the radius you are using?


hI @aya_zaki,

Thank you for this information.

I have shown the schematic of my bent in last reply. I have bent angle of just 7 degree but not more than 10 degree. I don’t know exactly the bent radius but it would not be too small.

I would like to sweep the bending end of the waveguide. Is there any built-in command for bent waveguide or I can use polygons to make a sweep?



Hi @nouman.zia
Lumerical gives you the option to draw a bend as you wish. No matter how large the radius of the bend is it can be calculated. The confusing thing about your design is that we usually define a curve by its radius and corresponding angle.

To create bends, you can use the waveguide component from the Integrated elements library. You can find more details, here:


Hi @aya_zaki, I want to calculate the propagation loss using FDTD soln,in a straight wave guide whose length is approx. 2cm. Can you suggest me that how can I calculate? my second question is that, In last message you told about the bent loss that “To estimate the loss, you can divide the transmission at the output over the transmission at the input port”.I just want to confirm that how can I convert this loss in dB/cm .Is this formula is ok? Loss=10*Log[{Tout/Tin}/Total length (um) 10000]. In this formula,first I find the loss per micron just by divide the loss (Tout/Tin) by total length of semi circle and then to get loss per cm ,I multiply it by 10,000. and finally to convert in dB I took 10 Log of the answer.


Yes, the formula is okay. I am sorry for the delayed response.

As for a straight waveguide, as long as you are interested in a specific mode, you can trust the losses calculated by the eigenvalue solver (Mode analysis).

For bent waveguides, I recommend the FDTD simulations though. To get the bend losses in dB/cm, the formula shoud be:
Loss = -10log10(T)/Total length
where the total length is pi
R for a180 degree bend.
and T is the transmission which is the (S21^2).
It can be calculated as in the script in the following reply:


Hi @bkhanaliloo,

I am getting back to this my old post regarding the bending losses. I have a bit different case now in which a bend angle is fixed at a certian value and the bend radius is quite large, so we can assume the bending losses are negligible. But I still see about 80% drop in the power level at the facet of bent waveguide. Secondly the expansion monitor in the middle of bend waveguide gives really small back reflection than the monitor after the mode source, which is not true.
My final goal is the calculate the reflection which should be very low due to the bend angle at facet; on the other hand bendibent_sld_400um.lms (1.1 MB)
ng loss should be very low.
p.s. my case is similar to the tilt waveguides I discussed earlier. I am not sure if the angle is causing a problem here again!


Hi @nouman.zia

Sorry for a late reply. We were quite busy.

It looks like the waveguide is not completely extended into the PML region in +x direction. Is this intentional? Can you please make sure that it is extended into PML region and repeat the simulations? I am concerned that this might be a similar limitation that you were experiencing before:
Simulation tool and geometry