Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
MT surface field problem
Posted May 3, 2011, 5:54 a.m. EDT 2 Replies
Please login with a confirmed email address before reporting spam
Dear all,
I'm using Comsol 4.1 to simulate the MT (magnetotelluric) EM induction response, in frequency domain, of a 2D-two layer model, given an H field at its surface. The final goal is to optimize the model resistivities to obtain a response in terms of surface apparent resistivity ( =f(surface E/surface H) ) matching at best with data in a point at the surface of the model. I calculated the rhoapp in the following ways: once I set the geometry and material properties, I've choosen the Magnetic Field Interface, for which the Ampere's Law is available in mag potential form; in frequency domain, it's:
(j*omega*sigma-omega^2*eps0*epsr)*A+rotH-sigma*vxB=Je (where 'x' is the cross product and A is
the vector mag.potential)
In MT, Je (external current density) and v (velocity of the conductor) are = 0, so I simply didn't set neither the external current nor the velocity node in the Comsol model tree (I think that it's correct, but I didn't find anything about the default values on AC/DC manual; I hope they're zero).
The boundary conditions change according to EM field polarization:
TE (surface H has only x component):
at surface: Hx=1A/m
at bottom: magnetic insulation rotA=0
outer boundaries: perfect magnetic conductor rotH=0
inner boundary(between first and 2nd layer): surface current=0 ( nx(H2-H1)=0 )
TM (surface H has only z component):
at surface: Hz=1A/m
at bottom: magnetic insulation (rotA=0)
outer boundaries: perfect magnetic conductor (rotH=0)
inner boundary(between first and 2nd layer): surface current=0 ( nx(H2-H1)=0 )
My problem is that the surface H field is not held constant for all the frequencies, but decrease with frequency, and this lead me to an error in the rhoapp calculus. So, accordingly to Comsol 3.5. AC/DC user's guide, I reduced the electrical size in order to keep it smaller than 1/10 for most of the frequencies I need. The parameters I used are the following:
f=0.1-150 Hz
epsr=10
mur=8
max dimension=2500m
Accordingly with these parameters, the minimum wavelength is 2.2351e+005 m, so the electrical size is about 1/100, hence very smaller than 1/10. I slightly mitigated this drawback choosing a very fine (triangular) meshing shrinking the model (from 20000 to 2500m), but this lead, respectively, to a much heavier computational effort and to inadequate model dimensions for my purposes.
Does anybody know if the problem maybe the electrical size or anything else?
Thank you in advance
I'm using Comsol 4.1 to simulate the MT (magnetotelluric) EM induction response, in frequency domain, of a 2D-two layer model, given an H field at its surface. The final goal is to optimize the model resistivities to obtain a response in terms of surface apparent resistivity ( =f(surface E/surface H) ) matching at best with data in a point at the surface of the model. I calculated the rhoapp in the following ways: once I set the geometry and material properties, I've choosen the Magnetic Field Interface, for which the Ampere's Law is available in mag potential form; in frequency domain, it's:
(j*omega*sigma-omega^2*eps0*epsr)*A+rotH-sigma*vxB=Je (where 'x' is the cross product and A is
the vector mag.potential)
In MT, Je (external current density) and v (velocity of the conductor) are = 0, so I simply didn't set neither the external current nor the velocity node in the Comsol model tree (I think that it's correct, but I didn't find anything about the default values on AC/DC manual; I hope they're zero).
The boundary conditions change according to EM field polarization:
TE (surface H has only x component):
at surface: Hx=1A/m
at bottom: magnetic insulation rotA=0
outer boundaries: perfect magnetic conductor rotH=0
inner boundary(between first and 2nd layer): surface current=0 ( nx(H2-H1)=0 )
TM (surface H has only z component):
at surface: Hz=1A/m
at bottom: magnetic insulation (rotA=0)
outer boundaries: perfect magnetic conductor (rotH=0)
inner boundary(between first and 2nd layer): surface current=0 ( nx(H2-H1)=0 )
My problem is that the surface H field is not held constant for all the frequencies, but decrease with frequency, and this lead me to an error in the rhoapp calculus. So, accordingly to Comsol 3.5. AC/DC user's guide, I reduced the electrical size in order to keep it smaller than 1/10 for most of the frequencies I need. The parameters I used are the following:
f=0.1-150 Hz
epsr=10
mur=8
max dimension=2500m
Accordingly with these parameters, the minimum wavelength is 2.2351e+005 m, so the electrical size is about 1/100, hence very smaller than 1/10. I slightly mitigated this drawback choosing a very fine (triangular) meshing shrinking the model (from 20000 to 2500m), but this lead, respectively, to a much heavier computational effort and to inadequate model dimensions for my purposes.
Does anybody know if the problem maybe the electrical size or anything else?
Thank you in advance
2 Replies Last Post May 5, 2011, 5:29 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Daniele,
in the upcoming version 4.2 we have added a magnetotellurics model to the AC/DC model library. I suggest you contact your Italian COMSOL representative in Brescia and I am sure they will share that model with you even if the upgrade has not been shipped just yet.
Niklas
--
Niklas
in the upcoming version 4.2 we have added a magnetotellurics model to the AC/DC model library. I suggest you contact your Italian COMSOL representative in Brescia and I am sure they will share that model with you even if the upgrade has not been shipped just yet.
Niklas
Dear all,
I'm using Comsol 4.1 to simulate the MT (magnetotelluric) EM induction response, in frequency domain, of a 2D-two layer model, given an H field at its surface. The final goal is to optimize the model resistivities to obtain a response in terms of surface apparent resistivity ( =f(surface E/surface H) ) matching at best with data in a point at the surface of the model. I calculated the rhoapp in the following ways: once I set the geometry and material properties, I've choosen the Magnetic Field Interface, for which the Ampere's Law is available in mag potential form; in frequency domain, it's:
(...)
--
Niklas
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Niklas,
thank you for the quick answer.
I should succeed to set and resolve my physical problem by magnetic interface of the ac/dc module, I guess,
since it has already the right equations and boundary conditions I need. My only doubt is about the electrical
size I may use in my modeling, and I'd like to know why it doesn't work properly with a such small value (1/100),
despite the manual user's guide instuctions; if reduce it furtherly, changing the epsr and mur to very small
(and unrealistic for natural materials) values, I can keep the surface mag field constant through my frequency range
(0.1-150Hz). Maybe I'm wrong, but this seems to be the cause...
So, I'd like to know if this is really the cause, and if there is anyone in Comsol community who had similar problems.
Anyway,I'll contact the Comsol italian representative soon.
Thank you,regards
Daniele
thank you for the quick answer.
I should succeed to set and resolve my physical problem by magnetic interface of the ac/dc module, I guess,
since it has already the right equations and boundary conditions I need. My only doubt is about the electrical
size I may use in my modeling, and I'd like to know why it doesn't work properly with a such small value (1/100),
despite the manual user's guide instuctions; if reduce it furtherly, changing the epsr and mur to very small
(and unrealistic for natural materials) values, I can keep the surface mag field constant through my frequency range
(0.1-150Hz). Maybe I'm wrong, but this seems to be the cause...
So, I'd like to know if this is really the cause, and if there is anyone in Comsol community who had similar problems.
Anyway,I'll contact the Comsol italian representative soon.
Thank you,regards
Daniele