sar computation

How to use openEMS. Discussion on examples, tutorials etc

Moderator: thorsten

thorsten
Posts: 1393
Joined: Mon 27 Jun 2011, 12:26

Re: sar computation

Post by thorsten » Sun 07 Apr 2013, 08:55

Hi,

but you don't have any change in conductivity?

Do you still look at the xy-plane? Maybe you only have tangential electric fields? They would be steady across the change in permittivity.
Try a step/change in conductivity and/or density. That should show up in the SAR values. The permittivity itself does not influence the SAR if the electric field is the same.

regards
Thorsten

sebastian
Posts: 114
Joined: Mon 27 Jun 2011, 12:36
Contact:

Re: sar computation

Post by sebastian » Sun 07 Apr 2013, 09:33

It looks like your geometry is correctly set up. Let's wait for Thorsten's opinion on your problem - I never did any sar computation myself (btw: is a difference in permittivity enough, or shouldn't there be a difference in conductivity?).

Sebastian

Edit: I was not really awake, Thorsten already answered this...

xrhstos
Posts: 9
Joined: Sat 30 Mar 2013, 21:57

Re: sar computation

Post by xrhstos » Mon 08 Apr 2013, 15:36

there is a difference between the computed sar but the far fields do not change at all :shock:
i have also tried changing density but there was no difference...there is also no change in Zin, reflection coefficient...

thorsten
Posts: 1393
Joined: Mon 27 Jun 2011, 12:26

Re: sar computation

Post by thorsten » Mon 08 Apr 2013, 15:44

Hi,

well I would not expect the antenna to be affected much by a change of loss etc. inside of the sphere...

Furthermore you should be a little but more specific, maybe attach your modified m-file.
What did you change/add ?

regards
Thorsten

xrhstos
Posts: 9
Joined: Sat 30 Mar 2013, 21:57

Re: sar computation

Post by xrhstos » Fri 12 Apr 2013, 23:27

the following 3 lines stand for the inner sphere

CSX=AddMaterial(CSX,'anomoiogeneia');
CSX=SetMaterialProperty(CSX,'anomoiogeneia','Epsilon',epsilon_anom,'Sigma',sigma_anom,'Density',puknothta_mazas_anom);
CSX=AddSphere(CSX,'anomoiogeneia',11,[0 0 0],1,'Transform',{'Scale',aktina_anom,'Translate',kentro_anom} );

where sigma_anom=2

one other aspect is that when i am using sigma as a parameter in

CSX=SetMaterialProperty(CSX,'anomoiogeneia','Epsilon',epsilon_anom,'Sigma',sigma_anom,'Density',puknothta_mazas_anom);

instead of kappa, it takes ages to finish the simulation... can i speed up the simulation?

thorsten
Posts: 1393
Joined: Mon 27 Jun 2011, 12:26

Re: sar computation

Post by thorsten » Sat 13 Apr 2013, 13:11

Hi,

don't use sigma!! I know that most people think of it as electric conductivity, but it is the magnetic conductivity, which is only there because of symmetry of the FDTD equations. It is a NON-PHYSICAL quantity!
Only use kappa for the conductivity!
Please read through the help to AddMaterial or here: http://openems.de/index.php/Material_Property

I don't know why a sigma is making the simulation take longer, just don't use it...

regards
Thorsten

thorsten
Posts: 1393
Joined: Mon 27 Jun 2011, 12:26

Re: sar computation

Post by thorsten » Sat 13 Apr 2013, 14:24

Hi,

I modified the script to use a stacked model of skin/headbone/brain.
The values are roughly taken from 300MHz, so they don't match to the 1GHz used. But the high contrast between skin and brain versus the dense and low conducting bone shows the effect nicely...

This figures shows the SAR in a logarithmic scale, to enhance the contrast even more...
SAR_xy_plane.png
log10 SAR on the xy-plane
SAR_xy_plane.png (8.57 KiB) Viewed 7542 times
Again, don't forget to apply a higher priority to the materials inside...

Simulation time on my old Core 2 Duo was less than 5 min...

regards
Thorsten
Attachments
Dipole_SAR.m
modified SAR calc
(6.28 KiB) Downloaded 289 times

xrhstos
Posts: 9
Joined: Sat 30 Mar 2013, 21:57

Re: sar computation

Post by xrhstos » Sat 13 Apr 2013, 14:34

that's great! thanks for the tip on sigma! it took like ages to finish the simulation :roll:

AMynster
Posts: 20
Joined: Wed 07 Nov 2012, 22:33

Re: sar computation

Post by AMynster » Wed 17 Apr 2013, 17:10

Looks great!

One thing though. If you set epsilon to 1 for the sphere and kappa to 0 S/m it should be like air. It gives SAR value 0 as expected due to no losses. When looking in paraview on the farfield radiation pattern it gives maximum gain of 3.859. Im not sure if it is linear or dB but in any case it doesnt match the gain of a half wave dipole which should be 1.64 linear scale or 2.15 dB. Any comments on how that happened?

thorsten
Posts: 1393
Joined: Mon 27 Jun 2011, 12:26

Re: sar computation

Post by thorsten » Wed 17 Apr 2013, 17:39

Hi,

I just simulated and removed all spheres...

nf2ff states:
nf2ff.Dmax

ans =

1.6538
Sounds (nearly) perfect to me?

The example does not dump a far-field pattern to vtk. How did you export it to Paraview??

If you do it like in the Simple_Patch_Antenna.m it looks like this:
far-field_pattern.png
far-field_pattern.png (130.01 KiB) Viewed 7525 times
I can't reproduce the problem.

regards
Thorsten

Post Reply