first: Great software! I’ve used several commercial FEM suites in the past (for example Abaqus or Altairs HyperWorks) and I can say: the usability is great! Sure, other software had more functionalities, but the GUI and usability is just awesome.
I’d like to simulate the dynamic behaviour of an air bearing as a contact with stiffness. For this I use the Static State Dynamic step after a Frequency analysis with perturbation. In my case the perturbation is the static step just with gravity, to close the contact, respectively “to load” the surfaces. The simulation runs, but the results are nonsense. To eliminate other possible bugs, I’d created a simple model with just a plate and a cube.
It runs, but the results are nonsense as well. If I delete the contact pair (and surface interactions) the simulation results are acceptable. For this, I added a simple spring. In the final model I didn’t need it, but here it helps to find the issue.
My question: Are this kind of simulation (1: static step with contact, called “SA”; 2: frequency analysis, called “MA” and 3: steady state dynamics, called “SSD”) not possible? Or is it the contact definition? What could it be?
Instead of tied contact, I would just use tie constraint. It provides reasonable results.
Otherwise, I would use regular contact (not tied). But then the eigenfrequencies can change a lot and you should adjust the SSD analysis range to account for this. And tensile load doesn’t really make sense. In this case, compressive force with contact may not make sense as a harmonic load either since it implies contact changes which are not possible in SSD analyses.
It changes the modes but SSD still gives 0 displacements and stresses. I think that a different approach should be used, as explained in my previous post.
If I use a tie constraint, I disable the interaction between the plate and the cube. In this small example it doesn’t matter, but in my real application (air bearing) I need this interaction.
I’ll try it and take a look at the eigenfrequencies.
At the end the amplitude of the harmonic force is way less than the gravity load. So, I didn’t expect any contact opening. But I agree: the contact changing can’t solve with the SSD analysis. I hoped the changing harmonic force will not effect the contact to much.
The K value belongs to a measurement of the stiffness of the air bearing. So, if I change the value, I didn’t expect any comparable result with my measured system.
I’ll give it a try with another contact definition. Another plan is to add a thin layer, which represents the air gap. But this gap is just 11µm, so the elements will be very small …
Just as final feedback, may it help other users…
I use another approach for my case. I’ve modelled a small (1mm) layer between the two bodies and set a tied contact between each body. This layer represents my air gap and get an anisotropic material via Keywords editor (@Matej nice feature!)
Now I can easily set up and control the stiffness – in each direction separately. Works best for my approach, even in the “big model”.
Actually, it’s orthotropic to be precise. But yeah, thin orthotropic layers can be really helpful as workarounds in various cases where contact can’t be used properly or is too limited.
I just added a sample of numbers to test it with the cube and to upload some pictures here. These are probably different numbers then last time. In my real case it’s fits fine the experimental data.
I was just curious because your first Frequency mode is the same as mine. .
Mine cheked against a spring surface with K=1N/mm3 (425.4Hz matched perfectly).
That 0.005 is two orders of magnitude difference based on the fact that your initial " K value belongs to a measurement of the stiffness of the air bearing".
Same frequency, same stiffness, same geometry. Why is displacement so much different.?
.