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.

radiation boundary condition

Please login with a confirmed email address before reporting spam

Hi to all,
I'm trying to solve numerically a fluid-structure interaction problem involving an object which moves in a fluid which is both *compressible* and *viscous*.
I currently do not have the licence for the Acoustic Module.
Since the actual fluid domain is considered as unbounded, the computational domain should be truncated with an "artificial" boundary on which a radiation boundary condition should be implemented.
This condition should ideally reproduce a pure outgoing wave and avoid wave reflection towards the computational domain.
My equation is of Helmholtz type.
Any suggestion to implement such a condition?
Thanks in advance!

Alessandro Ricci

9 Replies Last Post Aug 16, 2010, 5:55 a.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 7:27 a.m. EDT
this subject have been investigated in great details by the applied mathematician community over the Years. there is a number of possibilities and no perfect solution.
So you need to learn what exist out there and decide what is best for your problem.
Google "transparent boundary condition" as a starter... but again this is a large subject. Typically you cannot find universally transparent BC but partially transparet that will remove some waves but reflect some others... so you need to decide what is best for your problem either from theory or trying different options and compare...

as an example you can choose to be perfectly transparent for normally incident wave but get large reflection for grazing waves.. or the opposite or a blend..

JF
this subject have been investigated in great details by the applied mathematician community over the Years. there is a number of possibilities and no perfect solution. So you need to learn what exist out there and decide what is best for your problem. Google "transparent boundary condition" as a starter... but again this is a large subject. Typically you cannot find universally transparent BC but partially transparet that will remove some waves but reflect some others... so you need to decide what is best for your problem either from theory or trying different options and compare... as an example you can choose to be perfectly transparent for normally incident wave but get large reflection for grazing waves.. or the opposite or a blend.. JF

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 9:00 a.m. EDT
Hi Jean Francois,
I have already read some papers about the topic.
I saw that that, usually, such a BC is implemented as a sort of generalization of the Sommerfeld radiation condition.
In my case I would like to simulate an object which vibrates in a free fluid.
My computational domain is a sphere and the analysis is of eigenfrequency type. I think it is better for my case to have a boundary which is transparent to normally incident waves.
I have already tried some different BCs (all of them are implemented like a mixed-type condition, i.e. Neumann + Dirichlet), but results are always unphysical.
I'm wondering if there exists a way in Comsol to write your own PML (or also an "infinite" element) to mimic a domain with an infinite extension.
Any further suggestion?
Thanks in advance.

Alessandro


Hi Jean Francois, I have already read some papers about the topic. I saw that that, usually, such a BC is implemented as a sort of generalization of the Sommerfeld radiation condition. In my case I would like to simulate an object which vibrates in a free fluid. My computational domain is a sphere and the analysis is of eigenfrequency type. I think it is better for my case to have a boundary which is transparent to normally incident waves. I have already tried some different BCs (all of them are implemented like a mixed-type condition, i.e. Neumann + Dirichlet), but results are always unphysical. I'm wondering if there exists a way in Comsol to write your own PML (or also an "infinite" element) to mimic a domain with an infinite extension. Any further suggestion? Thanks in advance. Alessandro

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 10:36 a.m. EDT
I have not done that on a long time.
but out of my memory,

If these BC are properly written and applied [ that is your modelling is correct for the problem you want to solve] the result should not be unphysical.
One way to test it is to get a larger domain [ possibly on symmetric case so that you can use symmmetry to control the number of DOF if memory or cpu time is an issue and compare the result.
after that make sure the bc you use is really the one you want to use [ these things can be tricky sometime]
JF
I have not done that on a long time. but out of my memory, If these BC are properly written and applied [ that is your modelling is correct for the problem you want to solve] the result should not be unphysical. One way to test it is to get a larger domain [ possibly on symmetric case so that you can use symmmetry to control the number of DOF if memory or cpu time is an issue and compare the result. after that make sure the bc you use is really the one you want to use [ these things can be tricky sometime] JF

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 11:31 a.m. EDT
a radiation boundary condition that is mathematically exact exists and works well; it derives from a boundary element method and has been applied to many problem ,see attached files :

_ application of bem on elastic wave(baw),can work on stokes infinite losses and infinite element for porous flow


the secret here is to use a green function ......
a radiation boundary condition that is mathematically exact exists and works well; it derives from a boundary element method and has been applied to many problem ,see attached files : _ application of bem on elastic wave(baw),can work on stokes infinite losses and infinite element for porous flow the secret here is to use a green function ......


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 11:46 a.m. EDT
Thanks for the update .. seems that science progress even when you dont monitor it :-) [ I worked on these problems in the late 80's early 90's...]
JF
Thanks for the update .. seems that science progress even when you dont monitor it :-) [ I worked on these problems in the late 80's early 90's...] JF

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 19, 2010, 11:53 a.m. EDT
I wont spent more time than that but reading the wejung_dong paper..

it works in spherically symmetry [ R is the only space variable] so that solutions have only "normal" components [ if you think of the solution as a pseudo differential integral in spherical coordinates " a la Hormander" [ a generalized green function] and this paper in fact is just an application of the well known result that you can write exact transparent condition in One direction of propagation but not in all of them.

My guess is that when you break the symmetry and angular variable teta and phi shows up you lose the "exacteness" of the BC..and then have to find the best compromise....
but this belong to my past so i really cannot say or do more than that o this subject ...

JF
I wont spent more time than that but reading the wejung_dong paper.. it works in spherically symmetry [ R is the only space variable] so that solutions have only "normal" components [ if you think of the solution as a pseudo differential integral in spherical coordinates " a la Hormander" [ a generalized green function] and this paper in fact is just an application of the well known result that you can write exact transparent condition in One direction of propagation but not in all of them. My guess is that when you break the symmetry and angular variable teta and phi shows up you lose the "exacteness" of the BC..and then have to find the best compromise.... but this belong to my past so i really cannot say or do more than that o this subject ... JF

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 20, 2010, 3:48 a.m. EDT
Thanks a lot to Jean Francois and Louvet.
I will read very carefully the pdf in attachment and if I will manage to solve my problem I will post the BC I have employed.
Thanks a lot again!

Alessandro
Thanks a lot to Jean Francois and Louvet. I will read very carefully the pdf in attachment and if I will manage to solve my problem I will post the BC I have employed. Thanks a lot again! Alessandro

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago May 30, 2010, 4:03 p.m. EDT
hi guys ,

i am doing close to that model a buoy loaded. the buoy motion created waves. i was wndring could I use the same mthod to abzorbe the waves?
Ricci : i am new in comsol i want to model this in frequancy domain is posible ? and wht kind of mode i should pick


thanks
hi guys , i am doing close to that model a buoy loaded. the buoy motion created waves. i was wndring could I use the same mthod to abzorbe the waves? Ricci : i am new in comsol i want to model this in frequancy domain is posible ? and wht kind of mode i should pick thanks

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Aug 16, 2010, 5:55 a.m. EDT
Hi,

I am a new user in COMSOL and trying to estimate the radiation force on a sphere using acoustic module. I wonder if the radiation boundary condition on the boundary of sphere makes it a compressible sphere as nothing gets reflected back !
Please help me out !

Kind regards
Puja
Southampton University
Hi, I am a new user in COMSOL and trying to estimate the radiation force on a sphere using acoustic module. I wonder if the radiation boundary condition on the boundary of sphere makes it a compressible sphere as nothing gets reflected back ! Please help me out ! Kind regards Puja Southampton University

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.