TFSF formulation and Far field
I am not sure what you mean by "analytical Far Field formulations". But I often use TFSF and NTFF together.
TFSF = total field scattered field => used to insert planewaves near the scatterer
NTFF = near to far field (?) => used to extrapolate fields near the scatterer to fields far away
So both address completely different issues.
This is precisely the point. I am under the impression that with TF/SF I can model any far field scenario that could possibly arise. i.e the TF simulation is a far field simulation in itself. Is this not correct?
The NTFF would only give more accurate magnitude values I assume.
Could I ask you to give me an example of an NTFF that I can not do with a TF simulation. I understand the theory of what is happening from Taflove, but do not have the experience to translate this into practical scenarios.
I am not clear what your 'This' refers to.
No, it is not. Both methods are often used together but are not connected.
- TFSF is a method to introduce plane waves it is not a simulation method,
that is still FDTD.
- NTFF is a method to avoid computing field values for large homogeneous regions, a short cut.
Simple example.
1um x 1um x 1um scatterer. You use 100nm cells to model,
you want to know the field values at 1m distance (2mx2mx2m cube).
If you use standard FDTD you would need 8e21 cells which is a bit
difficult to handle with most computers.
if you use a 10um x 10um x 10um grid insert your plane wave using TFSF
and later use NTFF you need 1e6 cells for the FDTD simulation, possible with
any laptop nowadays.
Usually the opposite, since you have to approximate the field values at the
surface you use as input for NTFF. This is difficult to do well it seems unless
you use quite fine grids.
I did not mean to imply that I would model the whole field between source and target with FDTD using TF/SF. I am interested only in the reflection bfrom the wing edges of an aircraft at 2km(for example). This would entail an FDTD modelling space around the wing using a planar incident wave. i.e. the wing is in the "far field" and my modelling takes place only around the wing edges.
Could you advise what the NTFF aproach(using FDTD) would be for the situation described above? Are there any practical NTFF examples I could look at? I have looked at several academic articles and Taflove on the NTFF and they mostly describe the solution to the NTFF with FDTD but do not give clear examples of how they would be deployed.
First of all what wavelength are we talking about?
Secondly, the standard way to apply TFSF and NTFF would be
1) make cube containing the whole plane
2) the surface of that cube would be your TS-SF boundaries
3) simulate by inserting a plane wave and measure the reflected fields at some surface
surrounding this cube (i.e. in the scattered field)
4) this measured data can then be used to computed field values much further (e.g.2km) away
using the NTFF method.
5) the reason why the TF-SF cube surface or the NTFF surface should be all in air is that both methods
get very difficult (impossible) to apply if you have material boundaries cutting those surfaces.
I am not sure what you are asking for here. Try to reread Taflove with my description above in mind.
I will take the advice and am going to re-examine the NTFF in Taflove and Schneider. I find the TF/SF implementation(using FDTD) and modelling very intuitive but am still struggling with why anyone would want to use the much more complex(to me at least) NTFF. Thanks for the feedback on this.
Simply because NTFF needs less resources (the example I mention earlier would be 384,000,000,000TB with TFSF and 48MB with TFSF and NTFF.