Ivar KJELBERG
COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
15 giu 2012, 04:51 GMT-4
Hi
I would say a few (<10) wavelength of air, then 2-3 for the PML distance. but I havent modelled optics and fibre for some time, others out here might have better rule of thumbs
--
Good luck
Ivar
Hi
I would say a few (
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
15 giu 2012, 12:40 GMT-4
First, make PML very large. Then gradually decrease its thickness until your solution starts changing significantly. This way you can find an optimum that's right for you.
If you plan to solve for range of different wavelength, optimize PML for the largest wavelength
First, make PML very large. Then gradually decrease its thickness until your solution starts changing significantly. This way you can find an optimum that's right for you.
If you plan to solve for range of different wavelength, optimize PML for the largest wavelength
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
16 giu 2012, 10:52 GMT-4
Thanx Ivar.
Thanx Ivar.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
4 feb 2013, 09:27 GMT-5
how about number of layer ?
how about number of layer ?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
4 feb 2013, 19:52 GMT-5
I always use sweep mesh in PML with 3-5 layers. Again, you can increase the number of layers until you get the desired accuracy.
I always use sweep mesh in PML with 3-5 layers. Again, you can increase the number of layers until you get the desired accuracy.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
5 feb 2013, 04:59 GMT-5
According to Comsol Support answer I got, it does not matter. The PML will always be "one-wavelength" thick per default, you should just take care how you mesh it. They suggested using mapped-quadratic elements and make sure you have between 1-10 elements across the thickness of the PML. In my experience 4-6 usually is more than enough and I have never had any real need of 10 elements, the marginal computational accuracy vs. the time it takes to compute is not worth it.
Hope this helps
--
/RVN
According to Comsol Support answer I got, it does not matter. The PML will always be "one-wavelength" thick per default, you should just take care how you mesh it. They suggested using mapped-quadratic elements and make sure you have between 1-10 elements across the thickness of the PML. In my experience 4-6 usually is more than enough and I have never had any real need of 10 elements, the marginal computational accuracy vs. the time it takes to compute is not worth it.
Hope this helps
--
/RVN
