Simulation take a long time to finish


I am trying to simulate a bowtie antenna on a Si substrate at 30 THz. The issue is the simulation is taking too long and I have not used mesh override yet. Also the E field is not concentrated in the gap instead on the outer edges of the antenna.I do not understand what is the problem. Can you plz have a look at my simulation file and point out the issues. Thanks.Bowtie_Antenna.fsp (349.0 KB)

Mesh override and minimum mesh step

Hi @fayyaz.kashif1

Simulation took only a few minutes in my machine to finish running. How long does it take for you to run?

You can add mesh override over the gap region and perform convergence testing to make sure that results are reliable:


Thank you for the answer.
It took around 15 min on my computer. Although I have set mesh accuracy to 2 and I have not used mesh override yet. After I applied the mesh override around the antenna structure only, the simulation data increased to GBs. Why it is so? I have simulated many examples from Lumerical which have mesh overrides and higher accuracy settings. Even then those required only few MBs.

  1. I wanted to use S parameters Analysis Group in my simulation. The default monitors used in this group use 1000 data points which is too high. How we can change that. Because it says “You are using constructed objects and NO changes can be saved”. How I can reduce the data points to 50 ?


Hi @fayyaz.kashif1

  1. You can reduce the number of frequency points if you want to reduce the size of the file.

  2. You can modify the number of frequency points from global properties of monitor:

Hope this answers your inquiry.


Thanks for the answer. It helped for analysis group monitor geometry settings.
I have tried simulating the structure at a single frequency only(30 THz) still it takes 15 mins to run. I have not used mesh override and mesh accuracy setting is at 2.

  1. I have PML boundary in Z direction from where the plane wave is coming and the distance of the PML boundary from the structure should be at least half of the maximum wavelength. Do we need to maintain same distance in x and y directions too ? Do we also need to maintain the same distance inside the substrate ? Because this can make the simulation region quite large.


Hi @fayyaz.kashif1

You cannot use plane wave source with PML BCs as it can cause edge effects:

instead you can use TFSF source:

And if your design is periodic, you can modify the BCs to periodic and use plane wave source.

If you are using PML, you need to have at least half wavelength between PML and edge of critical object (Bow_Tie antenna here) in every direction. This means that with the current setup, you need to increase FDTD span in the -z, x and y directions.

Hope this answers your inquiry.


Thank you.
If we have periodic BCs in x and y directions and PML BCs in z direction can we use plane wave source?


Hi @fayyaz.kashif1

Yes, you can if the plane wave source is on the xy plane (periodic boundary conditions) and injecting fields in the z-direction.



In this example, I am confused about the following line.

area = getdata(“source”,“area”); # get source area (it’s not exactly 2um^2 due to finite sized mesh)

I wonder how this line can pick the value of area from source? There no such term “area” in source window.


I have an antenna on a Si substrate. Is it OK to use TFSF source such that it spans from air above the antenna to inside the Si layer?


Hi @fayyaz.kashif1

This quantity can be found under source:

Please note that you need to sun the simulations to access this quantity.

That seems correct. For correct usage of TFSF please see the link below:

Also please make sure that your follow up questions are related to this topic otherwise feel free to create a new topic.