Discussion Forum

The default is that each physics interface has its own set of degrees of freedom. The simplest solution in your case, is to change the name of the Displacement field in the Dependent Variables section of one of the Truss interface, so that the same fields as in the Beam interface is used.

Hi, Thanks for your reply, that error is now gone. But I still have another error, please see the attached file. It seems as I have some error in my geometry which makes comsol unable to compute a specific variable used in the calculations, but I can't understand why.

This seems to be a bug that we will have to look into. Please try putting the Beam interface before the Truss interface in the model tree, and see what happens. Please report back here.

Putting the beam interface before the truss interface seemed to help, definitly seems like a bug(?). However now I have a third error, see the attached png file. I made a very simple truss and beam model, which I also have attached.

The model you have attached is not statically determinate (stable). It has two rigid body modes, something that you can see by computing the eigenfrequencies.

How is it unstable? Isn't the truss and the beam system connected through the geometry?

Hi Zeus,

As Henrik indicated, you can look at the modes to see how the structure is unstable. My guess is that you are assuming that there is rotational stiffness associated with the joints, which there is not.

Best,

Jeff

One problem is that since the beams only have translational constraints, they could simply be replaced by truss elements, and then one of the unstable modes is evident.

However, even if you completely remove the beams and put pinned conditions where they were connected, the truss structure in itself is not stable. It does fulfill the fundamental condition

(number of bars) + (number of constraints) = 2 * (number of joints).

That, however, is only a necessary condition for static determinacy. It is not sufficient.

Hi Zeus,

As Henrik indicated, you can look at the modes to see how the structure is unstable. My guess is that you are assuming that there is rotational stiffness associated with the joints, which there is not.

Best,

Jeff

Yes, that is what I was assuming. Could I move the beam elements, say e.g., 1 mm downwards and then use some type of coupling between the truss and beam?

I am afraid that changing how the truss structure is connected at its ends would not fully fix your problem: as Henrik mentions, your truss structure on its own (e.g. even if there were no beams and the truss structure were pinned to the ground at both ends) is not stable. That instability would be removed if the truss elements were replaced with beams (I am always a bit skeptical of the use of trusses); you would need to decide if it is a realistic representation of the real world situation you are modeling.