
Filter by Topic
Release
Popular Topics
 List all discussions
Local Edge System
Thread index  Previous thread  Next thread  Start a new discussion 
November 26, 2012 9:41am UTC
Local Edge System
Hello,
I'm trying to implement a "Predescribed Displacement" Condition on an Edge using the local Edge Coordinate system. But i have the Problem that i couldn't find a reference to the "Local Edge System" and therefore I'm not sure which coordinate direction lies in the direction of the edge. And additionally i don't really know how the directions are called.
I really appreciate your help.
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
November 26, 2012 7:33pm UTC in response to Jan Kaul
Re: Local Edge System
Hi
the local variable for edges is "s" going from 0 to 1 along the arrow direction (COMSOL decides how) and you need to turn on the arrow plot in the view options to see the direction

Good luck
Ivar
Reply  Reply with Quote  Send private message  Report Abuse
November 27, 2012 10:03am UTC in response to Ivar Kjelberg
Re: Local Edge System
Hello,
Thank you very much Ivar. Is there a way to get for example u(s=0)?
kind regards
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
November 27, 2012 10:49am UTC in response to Jan Kaul
Re: Local Edge System
Hi
in COMSOL Solid physics u,v,w are the displacement field (implicitely u(x,y,z),v(x,y,z) ...) but you do not address u,v,w like that, you can define in the post processing a line graph on a solved Data set, then select an Edge entity and type u in the expression, this will give you the displacement u(x,y,z) along the edge for all (x,y,z) belonging to the selected edge.
by default COMSOL show this w.r.t. the arc length (s)*total_arc_length, but you might change the horizontal scale and define your own expression (scalar) based on any value, including x,y,z ...
in COMSOL if you want to know a dependent value at "point" location you need to define a "point integration" operator and select your point and "integrate" (as sum_over_i=1to1 => value at i=1) so the integration of "u" over a point at location (x0,y0,z0) is u(x0,y0,z0).
but now I have a doubt how you could implement this for "s" and get u(s=0) for a given edge over to a variable

Good luck
Ivar
Reply  Reply with Quote  Send private message  Report Abuse
November 27, 2012 12:05pm UTC in response to Ivar Kjelberg
Re: Local Edge System
Thank you very much Ivar!
I solved my problem using the intop Operator.
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
November 27, 2012 2:11pm UTC in response to Jan Kaul
Re: Local Edge System
Hi,
Concerning the original question about directions in Local Edge System:
I'm trying to implement a "Predescribed Displacement" Condition on an Edge using the local Edge Coordinate system. But i have the Problem that i couldn't find a reference to the "Local Edge System" and therefore I'm not sure which coordinate direction lies in the direction of the edge. And additionally i don't really know how the directions are called.
Local edge systems are available for some loads and boundary conditions in the Beam and Shell interfaces. Your question does not tell which, and the definitions differ:
Beam:
The local edge system is the same as the beam coordinate system as specified by the Section Orientation subfeature under Cross Section Data.
Shell:
The first direction (xl) is along the edge.
The third direction (zl) is the same as the shell normal direction.
The second direction (yl) is in the plane of the shell, and formed by the cross product zl X xl.
In both cases you can examine the directions by adding Arrow Line plots under a 3D Plot group. If you do not have any results yet (which you would not have during input data preparation) you need to add a Study, then Show Default Solver, and run until Compile Equations. Now you will have a data set which contains the geometrical information required.
Regards,
Henrik
December 3, 2012 9:32am UTC in response to Henrik Sönnerlind
Re: Local Edge System
Thank you very much for your reply.
Is there a way to compute the derivative of along the edge(beam), because although my beam is in the direction of z, uz doesn't seem to work.
kind regards
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
December 3, 2012 3:16pm UTC in response to Jan Kaul
Re: Local Edge System
Hi
isnt it what was just said above, in the beam theory the beam "edge" is "x", you have a coordinate change with the beam physics

Good luck
Ivar
Reply  Reply with Quote  Send private message  Report Abuse
December 4, 2012 8:41am UTC in response to Jan Kaul
Re: Local Edge System
Hi,
Is there a way to compute the derivative of along the edge(beam), because although my beam is in the direction of z, uz doesn't seem to work.
Derivatives with respect to spatial directions cannot be computed for the beam elements, nor can the tangential derivative operator dtang() be used. The beams use a special ('Hernitian') element formulation. A derivative like uz for a beam element in the z direction is however the same as the degree of freedom thy. In the beam element the derivatives of the transverse displacements are actually degrees of freedom.
Regards,
Henrik
December 4, 2012 8:42am UTC in response to Ivar Kjelberg
Re: Local Edge System
Hello Ivar,
I'm really sorry to bother you so often, but i can't seem to be able to compute the derivatives, no matter how I tried it. The reason for my question is that I want to apply an Edge Load depending on the derivative of u. I attached my Model, so that you might have a look at it.
The critical part is the Edge Load in the Solid Mechanics Physics, which seems to be zero although I multiplied it with a very big number.
I thank you very much for your time and hope that I won't be bothering you for a while after that.
kind regards
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
December 4, 2012 8:59am UTC in response to Jan Kaul
Re: Local Edge System
Thank you Hendrik,
that changes a lot. But actually for my case it would suffice to access the derivatives of the displacement field of the solid, which should be usual Lagrange tetrahedrons. I thought that if I want to compute the derivatives using the u variable of the solid mechanics physics instead of the u2 variable of the beam physics comsol would automaticly use the shapefunctions of the solid.
kind regards
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
December 4, 2012 7:58pm UTC in response to Jan Kaul
Re: Local Edge System
Hi
I have some problems understanding what you are doing
1) define materials on the domains and the edge, now you have a mix of undefined material variables and local user defined for part of it
2) the edge load of the solid is probably not d(u,x), first d(u,x) is ux in COMSOL (check the equations view), then u is the dependent variable of the "solid", I suspect you want the dependent variables of your "beam" that is u2x
3) why use a rigid connector and not just a "fixed" BC for the solid (and the beam) ?
I would suggest that you solve the two independently and look at the variables in specific plots, then you can combine them, probably it will be clearer.
As validation you should draw your beam as solid and solve the "solid" as a 2 material case and then compare with the use of your beam

Good luck
Ivar
Reply  Reply with Quote  Send private message  Report Abuse
January 8, 2013 12:38pm UTC in response to Ivar Kjelberg
Re: Local Edge System
Hello,
I still couldn't figure out how to compute the derivative along a certain edge. I looked at the beam equation view and there the derivative is accessed by the expression "mod1.umod1.vmod1.wts". The s at the end seems to compute the derivative of the local w variable in the edge direction. Sadly i wasn't able to apply this to an own weak formulation on an edge. Do you have an idea, how this might be done?
Thank you very much.
Jan Kaul
Reply  Reply with Quote  Send private message  Report Abuse
December 9, 2014 3:09pm UTC in response to Ivar Kjelberg
Re: Local Edge System
Dear Ivar,
How can I do this (turn on so that the arrows for each edge are shown) when using Comsol from Matlab. It seems that neither mphgeom nor model.view has the necessary switch, or am I wrong?
Sincerely,
Oskar Wallmark
Reply  Reply with Quote  Send private message  Report Abuse
Rules and guidelines