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.
Normal vectors for each certain domain (not whole geometry)
Posted Oct 31, 2011, 3:21 p.m. EDT Geometry, Modeling Tools & Definitions, Parameters, Variables, & Functions 4 Replies
Please login with a confirmed email address before reporting spam
Hi all,
I am working in PDE module and I have faced problem with the normal vectors (nx,ny) in my model. According to the definition of "n" in the manual, "n" is the outward unit normal pointing out from the domain. This is totally true for the outside boundaries of the geometries. However, the problem is that it only considers one unit normal vector at the internal boundaries which causes problem when i try to couple two adjacent domains by means of a relation based on "normal vector". I am familar with other definitions of "un and dn" but they are not what I want. For example when modelling the Acoustic-Structure Interaction (in other modules), in the boundary of Fluid to Solid we define the equality of traction by putting -ns.gamma=-ns.p. or -n_acpr.gamma=-n_acpr.p . In PDE module, the control on "n" is not like that as we can not refer to a special n like n_acpr or .... . If we had the normal vectors exactly like its definition that for example when we have 2 domains with one internal boundary and it could recognize the normal vector for each domain at the commen bounday, there was no problem, but it seems it cant. I am wondering if there is any way to make COMOL understand that get the normal vector for any arbitrary domain at boundaries.
For example when we have two general PDEs (g) and (g2). it would be great if it could understand n_g or n_g1. (like n_acpr or n_acpn or ...) but it seems it can not. Is there any way to solve this issue?
I appreciate your comments,
Masoud
I am working in PDE module and I have faced problem with the normal vectors (nx,ny) in my model. According to the definition of "n" in the manual, "n" is the outward unit normal pointing out from the domain. This is totally true for the outside boundaries of the geometries. However, the problem is that it only considers one unit normal vector at the internal boundaries which causes problem when i try to couple two adjacent domains by means of a relation based on "normal vector". I am familar with other definitions of "un and dn" but they are not what I want. For example when modelling the Acoustic-Structure Interaction (in other modules), in the boundary of Fluid to Solid we define the equality of traction by putting -ns.gamma=-ns.p. or -n_acpr.gamma=-n_acpr.p . In PDE module, the control on "n" is not like that as we can not refer to a special n like n_acpr or .... . If we had the normal vectors exactly like its definition that for example when we have 2 domains with one internal boundary and it could recognize the normal vector for each domain at the commen bounday, there was no problem, but it seems it cant. I am wondering if there is any way to make COMOL understand that get the normal vector for any arbitrary domain at boundaries.
For example when we have two general PDEs (g) and (g2). it would be great if it could understand n_g or n_g1. (like n_acpr or n_acpn or ...) but it seems it can not. Is there any way to solve this issue?
I appreciate your comments,
Masoud
4 Replies Last Post Nov 1, 2011, 11:08 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
there is no simple way to identify for internal boundaries "up and down" in a system build on "union"mode with different domains. For that you need to use the "assembly" mode, then all common boundaries are split into two (but its also up to you to define continuity on all these common boundaries whereit applies). Then w.r.t. a domain you can always define an up and down as each domain have "external" boundaries for the common "assembly boundaries".
If you have a complex model with only 1-2 assembly boundaries to split. Then make an union of all other ojects before , block by block such that you have only defined the boundaries to be assembled. Then "FInish assembly". In this way you have less boundaries to pair up
--
Good luck
Ivar
there is no simple way to identify for internal boundaries "up and down" in a system build on "union"mode with different domains. For that you need to use the "assembly" mode, then all common boundaries are split into two (but its also up to you to define continuity on all these common boundaries whereit applies). Then w.r.t. a domain you can always define an up and down as each domain have "external" boundaries for the common "assembly boundaries".
If you have a complex model with only 1-2 assembly boundaries to split. Then make an union of all other ojects before , block by block such that you have only defined the boundaries to be assembled. Then "FInish assembly". In this way you have less boundaries to pair up
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
You have access to "physics interface normals" for all physics interfaces (including PDE interfaces) in version 4.2a (and earlier versions, at least since 4.0). For exampe, if you have two General Form PDE interfaces, g1 and g2, you refer to them as g1.nx, g1.ny, and g1.nz for the normals "out" from the domains where g1 is active and g2.nx, g2.ny, and g2.nz for the normals "out" from the domains where g2 is active. For an Arrow Surface plot, for example, these normal vectors are predefined plot quantities. See the attached image for an example where g1 is active in the left block and g2 in the right block.
Best regards,
Magnus Ringh, COMSOL AB
You have access to "physics interface normals" for all physics interfaces (including PDE interfaces) in version 4.2a (and earlier versions, at least since 4.0). For exampe, if you have two General Form PDE interfaces, g1 and g2, you refer to them as g1.nx, g1.ny, and g1.nz for the normals "out" from the domains where g1 is active and g2.nx, g2.ny, and g2.nz for the normals "out" from the domains where g2 is active. For an Arrow Surface plot, for example, these normal vectors are predefined plot quantities. See the attached image for an example where g1 is active in the left block and g2 in the right block.
Best regards,
Magnus Ringh, COMSOL AB
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Ivar and Magnus,
Thanks to both because of your helpful comments. The way Ivar suggested seems to be interesting but the Magnus is easire. Actually I also had implemented the Magnus method. I had been looking for a way to use the application mode normal vectors, however, because of the different ways of implementing them in previous versions (n_g) and new version (g.n) I were not able to use them. Fortunately, I found the way and it works perfectly.
Thanks for the comments,
Masoud
Thanks to both because of your helpful comments. The way Ivar suggested seems to be interesting but the Magnus is easire. Actually I also had implemented the Magnus method. I had been looking for a way to use the application mode normal vectors, however, because of the different ways of implementing them in previous versions (n_g) and new version (g.n) I were not able to use them. Fortunately, I found the way and it works perfectly.
Thanks for the comments,
Masoud
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago