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 The information provided may be out of date.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.

Rigid Connector Rotation

Please login with a confirmed email address before reporting spam

Can anyone suggest or provide an example of Rigid Connector being used to introduce rotation into a 2D or 3D structural mechanics study?

There is no example in Model Exchange and the threads which mention it here do not go into detail.

Basically, I want to analyze deformation of a thin metal plate that is 1m long, 5 mm thick and 5 cm wide.

The 2D boundary conditions of the plate are rollers for the bottom 5 cm, and I would like to have rotation about the out-of-plane axis centered on the top short edge.

The exaggerated deformed shape would look similar to a question mark, ? Hope that is clear enough.

Thanks.

16 Replies Last Post 2011年9月29日 GMT-4 13:13
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年1月29日 GMT-5 10:29
Hi

I find the Rigid COnnector somewhat subtile and had some trouble in the beginning with the true implications when using it.
If you fix an Boundary,edge as a rigid connector, the full edge becomes fully RIGID, but you can then apply an averaged load, moment or a prescribed displacement or rotation, very nicely.

I wanted also to use it as an angle sensor by using it "unconstrained", but that does generally not work as the selected edge becomes rigid,

My first thoughts would be that with the moments and forces, as well as rotation/displacement constraints of the rigid connector, you should manage to deform a straight bar from an I to a ?
But I'm not shure what to propose to apply as combinations of loads. The thing though is that your section where you define the rigid connector becomes "rigid"

The best is to try out the RBC on a simple case, load by load
--
Good luck
Ivar
Hi I find the Rigid COnnector somewhat subtile and had some trouble in the beginning with the true implications when using it. If you fix an Boundary,edge as a rigid connector, the full edge becomes fully RIGID, but you can then apply an averaged load, moment or a prescribed displacement or rotation, very nicely. I wanted also to use it as an angle sensor by using it "unconstrained", but that does generally not work as the selected edge becomes rigid, My first thoughts would be that with the moments and forces, as well as rotation/displacement constraints of the rigid connector, you should manage to deform a straight bar from an I to a ? But I'm not shure what to propose to apply as combinations of loads. The thing though is that your section where you define the rigid connector becomes "rigid" The best is to try out the RBC on a simple case, load by load -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年1月29日 GMT-5 15:27
Ivar, thanks for the response.

The attached model is very simple, and is a working example of rotation via Rigid Connector. Key to obtaining a solution in this case was to:

1. enable geometric nonlinearity (Solid Mechanics > Linear Elastic Material Model > Geometric Nonlinearity)

No solution was found without enabling this feature.

2. add a fixed point on the axis of rotation

Without this feature, the axis of rotation was not held statically in space and a displacement of the body occurs.

The model currently sets Rigid Connector > Prescribed Rotation > Free so that deformation is achieved by applying a moment to the Rigid Connector boundary, but equally effective is setting Rigid Connector > Prescribed Rotation > Prescribed and selecting rotation axis and radians of rotation.

There is a limit to the amount of deformation that can occur. Solutions become increasingly difficult to achieve with larger deformations.

The attached model is from 4.0a where Rigid Connector is only available in 3D, but I will also try this in 2D with version 4.1 in a few days.
Ivar, thanks for the response. The attached model is very simple, and is a working example of rotation via Rigid Connector. Key to obtaining a solution in this case was to: 1. enable geometric nonlinearity (Solid Mechanics > Linear Elastic Material Model > Geometric Nonlinearity) No solution was found without enabling this feature. 2. add a fixed point on the axis of rotation Without this feature, the axis of rotation was not held statically in space and a displacement of the body occurs. The model currently sets Rigid Connector > Prescribed Rotation > Free so that deformation is achieved by applying a moment to the Rigid Connector boundary, but equally effective is setting Rigid Connector > Prescribed Rotation > Prescribed and selecting rotation axis and radians of rotation. There is a limit to the amount of deformation that can occur. Solutions become increasingly difficult to achieve with larger deformations. The attached model is from 4.0a where Rigid Connector is only available in 3D, but I will also try this in 2D with version 4.1 in a few days.


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年1月30日 GMT-5 04:42
Hi

nice model, but:

Why roller on "6" and not on "3" ? or just fixed on "3" ?

I do not understand the fixed constraint on Point "9"

the RBC "makes boundary "10" fully rigid but then applies "an average" rotation, displacement, moment,and/or force as defined in the RBC sub nodes, I agree.
The cntre of rotation "automatic", from what I have heard by COMSOL calcualtes the CoG of the boundary selected (unfortunately I have found no way to know exactly what is uses, in numbers), but you might define it also manually

For me the "large deformation" is only required if you do really "large" (such that sin(a) >>a) displacement, and need the Green-Lagrange strain tensor

Often I use a parametrical (continuation) sweep to load it in "steps" to improve the convergence of the solver, directly a large deformation can be tricky to get in one solver step

--
Good luck
Ivar
Hi nice model, but: Why roller on "6" and not on "3" ? or just fixed on "3" ? I do not understand the fixed constraint on Point "9" the RBC "makes boundary "10" fully rigid but then applies "an average" rotation, displacement, moment,and/or force as defined in the RBC sub nodes, I agree. The cntre of rotation "automatic", from what I have heard by COMSOL calcualtes the CoG of the boundary selected (unfortunately I have found no way to know exactly what is uses, in numbers), but you might define it also manually For me the "large deformation" is only required if you do really "large" (such that sin(a) >>a) displacement, and need the Green-Lagrange strain tensor Often I use a parametrical (continuation) sweep to load it in "steps" to improve the convergence of the solver, directly a large deformation can be tricky to get in one solver step -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年2月16日 GMT-5 15:34
Hi Marcus and Ivar,

I'm currently trying to perform a frequency domain study on a cylindrical cantilever to determine the frequency response function at the tip of the beam. I am able to get a X/F solution by adding a probe to the tip of the beam and applying a unit force. However, I require an X/M solution as well; meaning I have to apply a unit bending moment (1 N.m) at the free end of the beam.

I followed Marcus's advice for the static solution of RBC.mph (posted) and I can get a stationary solution this way, but the solution fails when I try a frequency domain study. I keep getting a Singular Matrix error saying there are void eqns, etc.

I am not well enough acquainted with using Rigid Connectors to know what is wrong, but I think (in some way) this is giving me the error. Also, I feel uneasy about using the Roller BC at the end as was done in the posted model. Why shouldn't I use a fixed constraint at the end. I've included my model. If you could take a look at it, I would really appreciate it. Any information will be useful.

Thank you,
Chris
Hi Marcus and Ivar, I'm currently trying to perform a frequency domain study on a cylindrical cantilever to determine the frequency response function at the tip of the beam. I am able to get a X/F solution by adding a probe to the tip of the beam and applying a unit force. However, I require an X/M solution as well; meaning I have to apply a unit bending moment (1 N.m) at the free end of the beam. I followed Marcus's advice for the static solution of RBC.mph (posted) and I can get a stationary solution this way, but the solution fails when I try a frequency domain study. I keep getting a Singular Matrix error saying there are void eqns, etc. I am not well enough acquainted with using Rigid Connectors to know what is wrong, but I think (in some way) this is giving me the error. Also, I feel uneasy about using the Roller BC at the end as was done in the posted model. Why shouldn't I use a fixed constraint at the end. I've included my model. If you could take a look at it, I would really appreciate it. Any information will be useful. Thank you, Chris


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年2月17日 GMT-5 05:29
Hi

I would not use the RBC for eigenfrequency or modal analysis, as it "rigidifies" your edge too. You can create a moment load by applying a variable force distributed over a boundary.

i.e. in 2D you have a small vertical edge of coordinate y0[m] < y1[m] (length = (y1-y0)) then you can use a boundary load (Total Force) of

Fx[N] = M0[N*m]*3*2*(y0+y1-2*y)/(y1-y0)^2

Pls recheck the equation I might have missed something ;) This will result in a load of Mz = M0[N*m]

Always check your reaction moments, as its very easy to get the wrong load on your system with moments
--
Good luck
Ivar
Hi I would not use the RBC for eigenfrequency or modal analysis, as it "rigidifies" your edge too. You can create a moment load by applying a variable force distributed over a boundary. i.e. in 2D you have a small vertical edge of coordinate y0[m] < y1[m] (length = (y1-y0)) then you can use a boundary load (Total Force) of Fx[N] = M0[N*m]*3*2*(y0+y1-2*y)/(y1-y0)^2 Pls recheck the equation I might have missed something ;) This will result in a load of Mz = M0[N*m] Always check your reaction moments, as its very easy to get the wrong load on your system with moments -- Good luck Ivar

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年2月17日 GMT-5 07:25
Hi

concerning your model, it's not clear for me what you are doing, or want to do ?

I mostly use fixed constraints to block one end, then RBC's are good in static particular if you apply the load on a small area already very rigid.

But if you have doubts it's always useful to test in another way. To add a moment you can use the method explained in the v3.5a doc adding an ode and having COMSOL updating it to get the desired reaction forces back, basing your results that action = reaction (holds also for COMSOL ;)

Main issue I have with frequency sweeps, in V4 the solver settings must be different from 3.5a, as I did not have to many problems to scan through structural resonances in 3.5a, while v4 is not very happy and I'm fighting with the damping factors to get the solver to be able even to pass the first resonance, not sure why, any advices out here ?

--
Good luck
Ivar
Hi concerning your model, it's not clear for me what you are doing, or want to do ? I mostly use fixed constraints to block one end, then RBC's are good in static particular if you apply the load on a small area already very rigid. But if you have doubts it's always useful to test in another way. To add a moment you can use the method explained in the v3.5a doc adding an ode and having COMSOL updating it to get the desired reaction forces back, basing your results that action = reaction (holds also for COMSOL ;) Main issue I have with frequency sweeps, in V4 the solver settings must be different from 3.5a, as I did not have to many problems to scan through structural resonances in 3.5a, while v4 is not very happy and I'm fighting with the damping factors to get the solver to be able even to pass the first resonance, not sure why, any advices out here ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月7日 GMT-4 19:25

Dear Ivar,

I was reading the thread on the rigid connector which you have been discussing. I have a clarification inquiry: so the rigid connector is a way of implementing rigid body rotations in a model?

that is to say, I am trying to simulate a rotating cylinder which has a plate attached to one side. Therefore, I need to apply the rotation on the cylinder and look at the strain produced on the plate as a result of rotating the cylinder. However, I would like to know if applying the comsol rigid connector will allow me to rotate the full 3D cylinder and not just its surface, in other words, to treat it as a rigid body.

I would definitely appreciate your comments/ feedback about this,

Best regards,

Lina

Dear Ivar, I was reading the thread on the rigid connector which you have been discussing. I have a clarification inquiry: so the rigid connector is a way of implementing rigid body rotations in a model? that is to say, I am trying to simulate a rotating cylinder which has a plate attached to one side. Therefore, I need to apply the rotation on the cylinder and look at the strain produced on the plate as a result of rotating the cylinder. However, I would like to know if applying the comsol rigid connector will allow me to rotate the full 3D cylinder and not just its surface, in other words, to treat it as a rigid body. I would definitely appreciate your comments/ feedback about this, Best regards, Lina

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月7日 GMT-4 19:39
Dear Ivar,

I was reading the thread on the rigid connector which you have been discussing. I have a clarification inquiry: so the rigid connector is a way of implementing rigid body rotations in a model?

that is to say, I am trying to simulate a rotating cylinder which has a plate attached to one side. Therefore, I need to apply the rotation on the cylinder and look at the strain produced on the plate as a result of rotating the cylinder. However, I would like to know if applying the comsol rigid connector will allow me to rotate the full 3D cylinder and not just its surface, in other words, to treat it as a rigid body.

I would definitely appreciate your comments/ feedback about this,

Best regards,

Lina

Dear Ivar, I was reading the thread on the rigid connector which you have been discussing. I have a clarification inquiry: so the rigid connector is a way of implementing rigid body rotations in a model? that is to say, I am trying to simulate a rotating cylinder which has a plate attached to one side. Therefore, I need to apply the rotation on the cylinder and look at the strain produced on the plate as a result of rotating the cylinder. However, I would like to know if applying the comsol rigid connector will allow me to rotate the full 3D cylinder and not just its surface, in other words, to treat it as a rigid body. I would definitely appreciate your comments/ feedback about this, Best regards, Lina

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月8日 GMT-4 11:14
Hi

the rigid boundary connector (as I understands it) makes your boundary fully rigid/stiff and allows you to apply distributed moments, or average forces, as well as giving you the opportunity to measure the angles

In 3D there are only u,v,w as dependent variables, the (small angle) rotations are obtained from the curl of the displacements solid.curlUX, solid.curlUY, ....UZ ... (theswe are the double of the small displacement angles, and can be integrated/averaged over a boundary to give you the rotations (just as you might use the boundary average normal to identify boundary orientation changes.

If you want to imply a rigid rotation on a domain you must impose a displacement, but that means that your mesh is not really needed either as the volume is "fully rigid" (if you add a subnode rigid domain v4.2, I understand it as you will replace your domain by a virtual "point mass and inertia)

For correct RB BCs pls use latest patched version, as these features were very new in 4.0 I do not know how well they were debugged ;)

--
Good luck
Ivar

--
Good luck
Ivar
Hi the rigid boundary connector (as I understands it) makes your boundary fully rigid/stiff and allows you to apply distributed moments, or average forces, as well as giving you the opportunity to measure the angles In 3D there are only u,v,w as dependent variables, the (small angle) rotations are obtained from the curl of the displacements solid.curlUX, solid.curlUY, ....UZ ... (theswe are the double of the small displacement angles, and can be integrated/averaged over a boundary to give you the rotations (just as you might use the boundary average normal to identify boundary orientation changes. If you want to imply a rigid rotation on a domain you must impose a displacement, but that means that your mesh is not really needed either as the volume is "fully rigid" (if you add a subnode rigid domain v4.2, I understand it as you will replace your domain by a virtual "point mass and inertia) For correct RB BCs pls use latest patched version, as these features were very new in 4.0 I do not know how well they were debugged ;) -- Good luck Ivar -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月26日 GMT-4 16:07

Thanks very much Ivar, I have been thinking a lot about your answer

I still have a couple of questions about this,

- Does this mean that I can use Comsol for a rigid body dynamics simulation? i.e. do a time dependent study of say a cilinder that is rotating in 3D space?

-Does the model take into account inertial forces? or do I need to define them?

-Adding up the previous two: can i have the rotation of the rigid body (cilinder) affect the stress on other structural components attached to it?( a cantilever beam for instance)

Kind regards,

Lina
Thanks very much Ivar, I have been thinking a lot about your answer I still have a couple of questions about this, - Does this mean that I can use Comsol for a rigid body dynamics simulation? i.e. do a time dependent study of say a cilinder that is rotating in 3D space? -Does the model take into account inertial forces? or do I need to define them? -Adding up the previous two: can i have the rotation of the rigid body (cilinder) affect the stress on other structural components attached to it?( a cantilever beam for instance) Kind regards, Lina

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月27日 GMT-4 01:52
Hi

a rigid connector says that the boundary (surface in 3D, or edge in 2D) is becoming ridgid, as sub node you can also reduce a domain to it's CoG I believe COMSOl will improve this in v4.2a coming out next week at the conference.

But if you want to do true multi-rigid-body simulations you are mostly better off using the pde module in 1D I would believe

--
Good luck
Ivar
Hi a rigid connector says that the boundary (surface in 3D, or edge in 2D) is becoming ridgid, as sub node you can also reduce a domain to it's CoG I believe COMSOl will improve this in v4.2a coming out next week at the conference. But if you want to do true multi-rigid-body simulations you are mostly better off using the pde module in 1D I would believe -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月27日 GMT-4 13:51

Sounds good, I will look into the 1D pde module,

Many thanks again,

Lina
Sounds good, I will look into the 1D pde module, Many thanks again, Lina

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月28日 GMT-4 01:41
Hi
You can also work in 3D PDE, it's just slightly tougher, easier to start in 1D ;)


PS try to get hand on the latest version, it's far more stable amd 4.2a will come, I have heard, at the conference in October in US, good reason to go there (or to the one in EU 2 weeks later, or Asia a few weeks thereafter again)

--
Good luck
Ivar
Hi You can also work in 3D PDE, it's just slightly tougher, easier to start in 1D ;) PS try to get hand on the latest version, it's far more stable amd 4.2a will come, I have heard, at the conference in October in US, good reason to go there (or to the one in EU 2 weeks later, or Asia a few weeks thereafter again) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月29日 GMT-4 00:20

Thanks a lot Ivar,

yes, I am sure that even the pde 1D is probably not that easy :)

I would like to attend the Comsol conference in EU, I'm sure it's worth it,

Best,

Lina
Thanks a lot Ivar, yes, I am sure that even the pde 1D is probably not that easy :) I would like to attend the Comsol conference in EU, I'm sure it's worth it, Best, Lina

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月29日 GMT-4 00:58
Hi

I can only confirm that the COMSOL conferences are the best places to learn more and to meet other users, for the cost you gain a lot, so you should convince your boss or supervisor, that the conference is a very good investment ;)

That's how I got most out for my learning curve of the tool, then I followed a few 1d courses and one quickly becomes a specialist, provided that you have not forgotten your "physics courses" ;)

--
Good luck
Ivar
Hi I can only confirm that the COMSOL conferences are the best places to learn more and to meet other users, for the cost you gain a lot, so you should convince your boss or supervisor, that the conference is a very good investment ;) That's how I got most out for my learning curve of the tool, then I followed a few 1d courses and one quickly becomes a specialist, provided that you have not forgotten your "physics courses" ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年9月29日 GMT-4 13:13

Thanks Ivar,

Definitely will try :) I am sure that it is well worth it, hopefully I have not forgotten all my Physics!

Best,

Lina





Thanks Ivar, Definitely will try :) I am sure that it is well worth it, hopefully I have not forgotten all my Physics! Best, Lina

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.