FDTD + CPML -> PDP - power decreasing with time

[roldao]

fdtd 2d cpml

Hi everybody.

I´m working with FDTD + CPML.
I´m simulating a typical scenario with 6 buildings (PEC structures) and in other situation (er = 3.0 e sigma = 0.005 S/m(electric conductivity).

For now I´m trying to plot the Power Delay Profile and I think, when comparing with the original document that something's wrong.

The Power Delay Profile from the original document for the PEC situation, the power is almost always at the same level (for example the multipath components are almost always in 12dB) , and in my simulation, the power is always decreasing with time.

What is the normal behaviour of a power delay profile? with the specific case of PEC buildings and with the other case explained?

What do you think that could be the problem?

The file attached is the plot of the original document and mine.

Please I need some help.

Cheers

Joao Roldao

 

[iyami]

Re: FDTD + CPML -> PDP

The Power Delay Profile from the original document for the PEC situation, the power is almost always at the same level (for example the multipath components are almost always in 12dB) , and in my simulation, the power is always decreasing with time.

What kind of excitation do you use? If you use a pulse this is to be expected. If you use a continuous
sine wave you should reach a stable state at some point. It may take a while for transients to die down,
however. (If you use a sine wave how do you switch it on?)

Sorry, I cannot read your document so I don't know if that information is there.

 

[roldao]

Re: FDTD + CPML -> PDP

Hi, thanks for answer.

I use a pulse.

Jmax =1000;
alpha = (4/(32*dt))^2;
source=zeros(1,nSteps);
for n=1:88%(9 e 30)
source= Jmax*(exp(-alpha*(((dt*n)-(32*dt))^2)));
end

then I apply

Hz(x,y)=source(x,y);

Thanks for attention

Added after 4 hours 14 minutes:

Hello.

Tell me please, if I use a continuous sine wave, how it is expected to switch on?

I made some simulations using a sine but only did:

Hz(is,js)= sin(omega*dt*time); for all time of the simulation.

There is any special manner to do it?

I saw in some examples something like:

Hz(is,js)=Hz(is,js)+ sin(omega*dt*time);

What is the better or what are the differences between this two excitation in practice?

Thanks

João Roldão

 

[iyami]

Re: FDTD + CPML -> PDP

* If you use a pulse you have to extract the behaviour for different frequencies using Fourier transforms.

* if you use H = sin() then you use a 'hard source' meaning no signal can travel through this point.
for a point source it ususally does not change much but if you have a line (in 2d) or a wall (in 3d)
of sources then you get perfect reflection of all signals hitting the source point
* switching on just means that you multiply the first few periods with half of a gaussian pulse.
This decreases transients.

Some comments (I do not have time to look closely at the paper (difficult to read)).
- you are adding the source to the magnetic field which is ok but usually you use the electric field.
(this can explain different scales of the results)
- you said that the energy level gets weaker over time (unlike in the reference paper
this is expected for a pulse. the absorbing boundary condition removes energy from your model.
if the reference paper uses a continuous sign wave then the result will be a steady state (after
transients died down)

I have the feeling that you have to take Fourier transforms of the input signal and the measured
energy levels to get the data you need if you are using a pulse