Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
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.
parametric sweep in a contact problem
Posted 10 avr. 2013, 13:49 UTC−4 Studies & Solvers, Structural Mechanics Version 4.3a 3 Replies
Please login with a confirmed email address before reporting spam
The attached file shows a simple axisymmetric problem in 4.3a where a screw-shaped part is being pushed into some uniform bone by an axial force.
In the attached model, there are identity pairs between the screw and bone, and the model runs fine with a parametric sweep involving an increasing force on the screw. However, my question is this: Is there any reason this problem would not work properly when I have contact pairs replacing the identity pairs in this same geometry? I cannot get that contact version to run in a parametric sweep mode although it runs fine as a single contact problem. When I try a sweep, the error reads "error in multiphysics compilation". (While the contact problem runs OK without the parametric sweep, I am curious whether I can use the parametric sweep mode in a contact problem.)
Second question: In one of Ivar's notes to a user asking about parametric sweeps, he wrote the note below, see 3rd paragraph. That makes me double-check one point: Does 4.3a by default use a "continuation mode" type of sweep? My model attached did seem to run this way based on the results, but I did not see any place in the solver node where I could select a "continuation" sweep vs. the other type of sweep. (I'm fine with a continuation sweep, but I was just wondering.)
October 18, 2012 11:09am UTC in response to kishore naik
Re: parametric sweep
Hi
add the parameters in the Parameter list, replace the circle centre position by these parameters, call up a specific Parametric node in the solver sequence, add in all relevant parameters, and adjusts ways to combine them. and off you go
Note: when changing the geometry, try to avoid changing topology, as this renumbers the entities (domain, boundary edge, points IDs) and often COMSOL get confused (note this is getting less and less true, as COMSOL's algorithms seem to drastic improve with the latest versions, and it's working even better if you use systematically the Definitions Entity Selections nodes)
Another point in v4.3a, there is a new feature, COMSOL checks if the parametric sweep needs to be "outside" the solver sequence or can be included into the solver as a continuation mode (automatic), sometimes now you need to turn this feature "off" to impose an external (looping around the full model, not only the solver sequence) parametric sweep. Check the latest doc
--
Good luck
Ivar
In the attached model, there are identity pairs between the screw and bone, and the model runs fine with a parametric sweep involving an increasing force on the screw. However, my question is this: Is there any reason this problem would not work properly when I have contact pairs replacing the identity pairs in this same geometry? I cannot get that contact version to run in a parametric sweep mode although it runs fine as a single contact problem. When I try a sweep, the error reads "error in multiphysics compilation". (While the contact problem runs OK without the parametric sweep, I am curious whether I can use the parametric sweep mode in a contact problem.)
Second question: In one of Ivar's notes to a user asking about parametric sweeps, he wrote the note below, see 3rd paragraph. That makes me double-check one point: Does 4.3a by default use a "continuation mode" type of sweep? My model attached did seem to run this way based on the results, but I did not see any place in the solver node where I could select a "continuation" sweep vs. the other type of sweep. (I'm fine with a continuation sweep, but I was just wondering.)
October 18, 2012 11:09am UTC in response to kishore naik
Re: parametric sweep
Hi
add the parameters in the Parameter list, replace the circle centre position by these parameters, call up a specific Parametric node in the solver sequence, add in all relevant parameters, and adjusts ways to combine them. and off you go
Note: when changing the geometry, try to avoid changing topology, as this renumbers the entities (domain, boundary edge, points IDs) and often COMSOL get confused (note this is getting less and less true, as COMSOL's algorithms seem to drastic improve with the latest versions, and it's working even better if you use systematically the Definitions Entity Selections nodes)
Another point in v4.3a, there is a new feature, COMSOL checks if the parametric sweep needs to be "outside" the solver sequence or can be included into the solver as a continuation mode (automatic), sometimes now you need to turn this feature "off" to impose an external (looping around the full model, not only the solver sequence) parametric sweep. Check the latest doc
--
Good luck
Ivar
3 Replies Last Post 11 avr. 2013, 08:16 UTC−4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
a few comments for your model:
1) you could use "geometry" union and leave COMSOL define the default continuity, then no need to define identity pairs
2) normally in 2D-axi one should keep to the positive "r" quadrant, but perhaps you get correct results so long youre geometry does not cross the axis. But look up for sign issues with a model on that side !
3) be aware that in 2D axi you can define a vertical load (along the axis, but if you define a load along R its not pending by a true centrifugal type load going along "R" BUT FOR ALL PHI !
4) to be more representatives, I would fix the boundary to the left on your case and not the bottom, or the one with R extreme. As you will get stress concentration at the interface with the two parts on the bottom line like taht, this would change your results
--
Good luck
Ivar
a few comments for your model:
1) you could use "geometry" union and leave COMSOL define the default continuity, then no need to define identity pairs
2) normally in 2D-axi one should keep to the positive "r" quadrant, but perhaps you get correct results so long youre geometry does not cross the axis. But look up for sign issues with a model on that side !
3) be aware that in 2D axi you can define a vertical load (along the axis, but if you define a load along R its not pending by a true centrifugal type load going along "R" BUT FOR ALL PHI !
4) to be more representatives, I would fix the boundary to the left on your case and not the bottom, or the one with R extreme. As you will get stress concentration at the interface with the two parts on the bottom line like taht, this would change your results
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
2) normally in 2D-axi one should keep to the positive "r" quadrant, but perhaps you get correct results so long youre geometry does not cross the axis. But look up for sign issues with a model on that side !
Actually, the results are unpredictable if a model is to the left to the symmetry axis. Anything with geometric nonlinearity (as contact is) will definitely be wrong.
An axisymmetric geometry should always be modeled at positive R values.
Regards,
Henrik
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Henrik
I find that you should issue a warning for that, COMSOL should easily find out if something has been set to the left
I believe I sent a suggestion along those lines quite some time ago, probably its far down on the priority list, but it would make life easier for new beginners ;)
And a "frustrated client, destroys for ten new purchases" I have been told ;)
--
Good luck
Ivar
I find that you should issue a warning for that, COMSOL should easily find out if something has been set to the left
I believe I sent a suggestion along those lines quite some time ago, probably its far down on the priority list, but it would make life easier for new beginners ;)
And a "frustrated client, destroys for ten new purchases" I have been told ;)
--
Good luck
Ivar