Discussion Forum

If you only put equal forces on both sides your problem is not well defined. There is an infinite number of possible steady state solutions. The net force being zero only means that the accelleration of your object will be zero, but the velocity can have any value. So you have to restrict the motion of your object in some way.

Thanks for the reply; that does make sense. Do you have any recommendations as to how to do this? I would really just like to pin down the center of mass if I could and watch the rest of it displace, but I haven't gotten very comfortable with the features of COMSOL yet. Thanks again.

I think you can fix any point/line/plane/vollume that you want with a 'fixed constraint' node in your structural mechanics physics interface.

Bear in mind that fixing the center of mass is not enough to make the problem well-posed: you need to constrain the problem in such a way that the three rigid body rotations are prevented as well.
Jeff

Thank you for these responses! Unfortunately, I have not been able to constrain the center of mass, because it seems I can only constrain entire objects (i.e. the whole cylinder). I've tried defining a point at the middle and constraining that, but I cannot get the mesh to work; do you know any resources I can use for getting this to work?

Also, Jeff, could you elaborate a little bit on the rotation constraint? Can I use some sort of prescribed rotation or something like this?

Thanks again!

Like the linear velocity can have any value, so also the rotational velocity can have any value, so you need to constrain both linear and rotational motion. But why do you want to constrain specifically the center of mass and not for example one end face?

If the cylinder is uniform over the length the center of mass is in the middle. You can put a plane there perpendicular to the axis of the cylinder, and put a defined displacement constraint on that plane and constrain it only in the direction of the axis of the cylinder. Then you put another fixed constraint at the point of intersection of this plane with the axis and constrain it in the other two directions. Then your problem would be well posed I guess.

Thanks very much. I am sorry for this silly question, but I am very new.

I have been trying to build what you said, but I am using a "work plane" that is set up at the center. I cannot figure out how to put any constraints on its motion, because it is not shown in my geometry, but it is also messing up my mesh. Do you have any advice or anywhere I should look to figure this out?

Again, thanks!

Can you maybe upload your model? I do not completely understand what you mean.

Certainly, here it is. Thanks for looking!

I think this is about what I mean, but I don't know if you had any specific ideas for the mesh, I changed it.

Yes, yes, this is just right! Thanks so much for the help! I have learned a lot from this model that you posted. Out of curiosity, though, is there any reason that you used the first boundary load but then canceled it out with the second and switched the load face?

You will need opposite forces on both sides. The constraints halfway the cylinder are just to define the steady state linear and rotational velocity of the object as 0, as we discussed above. The forces on the opposite faces define the actual load on the cylinder. There should be no actual load on the constraints in the middle.

Hi

I'm not sure which of the files I got (as you use the same name for original and updated one).
I see 2-3 things:

1) the boundary load is selected for top surface 7 and the opposite on the bottom AND the TOP id's 3+7
I believe the entity selections should be 7 on top and 3 on bottom only, as these loads adds up
2) your model is fully symmetric, so if you cut a "Pie slice" and put a symmetry plane in the middle, and add symmetry conditions on the side of the "Pie slice" you will naturally constrain your volume and you can forget your plane and point (singular not a good numerical approach) prescribed displacement of "0"
3) be aware that if you use symmetry conditions on eigenfrequency modal analysis, you will restrain the number of modes to the symmetrical ones, so you need to run a combination of symmetry and anti-symmetry boundary conditions, to catch all modes
4) just a geometry "trick", you can get your 2 cylinders and the common boundary in the middle by using a "centred" cylinder and a "layer of half height"

--
Good luck
Ivar

Thanks for the input, Ivar! I was wondering about your first point as well. I will try these improvements. Thanks again, everyone.