Discussion Forum

If you use a prestressed eigenfrequency analysis, the linearization of the eigenvalue problem will include nonlinear effects from the prestress load case. This, for example, includes stress stiffening and geometrical changes caused be geometric nonlinearity.

Friction, however, is only taken into account to the extent that boundaries that are in contact in the prestress solution (and have friction) will be handled as 'glued'. Friction as such cannot be linearized. For a certain amplitude of vibration it could be sticking and at a larger amplitude sliding.

Thank you very much Henrik.

I found in this COMSOL blog "https://www.comsol.com/blogs/structural-analysis-with-thin-elastic-layers" that we can use Thin Elastic Layer for estimation of contact. Is there any linear equivalent to estimate sliding-slipping friction ?

Thanks.

Since slipping-sliding is inherently nonlinear, it cannot be linearized. However, since the underlying assumption for frequency domain analysis in general, and eigenfrequency analysis in particular, is that the perturbations are small, the most consistent approximation is 'sticking'. Thus, using a thin elastic layer with both normal and shear stiffness is the best you can do (if you choose that approach).

Note that the blog post you refer to is more than 10 years old. While it is possible to use a thin elastic layer for this purpose, a more general approach today would be to use Contact and Friction. However, if you beforehand know the contact regions, you can of course connect them either by thin layers, by identity pairs, or by using union operators in the geometry.

Thank you very much,

But neither contact nor friction can be applied to eigenfrequency analysis. Furthermore, there is a gap between the regions and therefore I cannot pair them. So, I guess the best I can do is thin elastic layer.