Discussion Forum

Ok, I try to describe the problem again. I want to aplly constant forces on a system in a certain direction. I do this by boundary loads. It is a cube. All other surfaces are free. Comsol doesn't find a soution. I do not know what I am doing wrong.

Hi

first of all in your model, your Material 2 has probably a wrong Young modulus, I suspect you are missing 1E9 as it looks like aluminium spheres (but I might be wrong ;)

If you apply a pressure it's always perpendicular to the surface. but you can change from pressure, to Total force (in Newton, COMSOl looks after Area calculations, and Surface loads in N/m^2 (basically Pascal) but independently for all 3 directions

Then its better to fix one boundary and appply the load on the other part. if you use symmetry, take care to block all free DoF otherwise your solver will have problems finding a solution

--
Good luck
Ivar

Thanks for the hint with the E-modulus. If I fix one side I will get fixed displacement on one side, but not on the other side, not? That worries me.
Will the solver be able to solve it if I use explicit loads instead of pressure (seems to be the same to me in the end as I want the force to act perpendicular anyway).
What are degrees of freedom that I could additionally restrict? I just want a force to stretch the body and to have no forces acting on the body from the other sides.

Hi

it depends what you are looking for, a constant pressure with a material mix like you have make deformed shapes, you can also apply a prescribed displacement but that implies variable pressure load.

in y<our case you have a nice symmetry so by applying a 1/8 symmetry you can fix all 6 dsof and use only one pressure or Force (equivalent up to the surface value) or a prescribed displacement

--
Good luck
Ivar

To the first part. I want to do both, prescribed displacement and prescribed force.

I didn't work with symetries so far. I can try to figure that out, but when I try to compute the study currently it doesn't work. so is there a chance that applying this symetry will fix the problem?

Edit: I am applying symetries now, trying to figure out when to use symetry and when antisymetry, still computing the study gives an error message that the residuum is too big.

Hi
If you look at modes you must check all combinations of symmetry and anti symmetry boundaries, but if its only symmetric load use symmetry.

Then ensure the mesh is fine enough, or even perhaps apply a boundary mesh on the " stiff side" to better resolve the stress gradients at high values

You also have "roller load" this does not block all DoF so you must add something

And do not forget that pressure is applied onto the surface, note the sign

And check your strain, you are outside normal "linear range"

--
Good luck
Ivar

ok, I am still trying to make it work by the measures you suggested, just to your last edit: I am aware that normally the material would not behave linear, but I would like to ignore all nonlinear effects, I was assuming that this would happen if I use the linear elastic model. Should I use the "force linear strains" option?

When I fix the displacement on one side it works, but I would like to avoid this. May it be that theoretically there is no reason for the body to stay in place otherwise and therefore I cannot solve it? (While the body recieves no acceleration (all forces cancel out), any position of the body in the space can be a solution.

Hi

It's more the validity of the equations for large deformation, I would rather use the Solver incliude geoemtry nonlinearity. Or better lower the load so one remains within resonable limits.

If you apply a compression on two sides of your cube, you have a symmetry plane in the middle, and it's as if you model just 1/2, still if you leave your volume free in the 2 other directions the solver will eventually diverge, so you need at least to apply some "soft" load to keep it in place, or add more symmetry, or some "soft" spring load

The issue comes from any nuemrical instabilities and assymetry in the mesh FEM is an approximation, no "real" continuity model

--
Good luck
Ivar

Ok, I understand now, what you mean, why it diverges. I will try to find a way to fix it. Thanks for all your help.

Ok, comsol can solve it now. I used prescribed displacement 0 on the ring around the sphere in the center in the xy plane.

Hello,

Is there a tutorial or instructions to do the simulation that yo did? I am trying to make a similar simulation. There will be a deformable soft object on a flat surface. It will not be fixed from any place. I will apply forces from different directions to observe the deformation. I found some tutorials but all of them are fixing the mesh from one side and I don't want to do. When I don't fix one side, the residuals are too big. So if there is a tutorial that explains things that should be done step by step, it would be great. Because I am new in FEM simulations and COMSOL, so most of the terminologies are new to me.

Best regards,
Püren