Discussion Forum

Hi

I'm looking rapidly into your model, and have a few comments

FEM and COMSOL is not really a multibody simulation tool, in the sens you need a mesh and cannot really change the topology, but you can indeed do many things close to multibody simulation.

For your body force, if you want to apply gravity use g_const*solid.rho (in V4) or G*rho_smsld (in 3.5) where you define G=1[lbf/lb] or just G=9.81[m/s^2]

When you have collision and bouncing you need to define this physics and I'm not sure this is done in COMSOL by default contact pairs, again you have here a multibody simulation demand

Now how to better explain the differences betwee standard FEM and a multibody simluation tool such as Modelica ? specially since COMSOL is somewhere in-between. I do not have the answer just now and here

--
Good luck
Ivar

Hi Ivar,
If I'm not wrong the majority of the models are multi-bodies models since, for instance, a cantilever electrically actuated by an electrode is a multi-body problem to, although without contact. But, anyway, I have done some simulations where a cantilever snaps-in the driving electrode and the "pair contact" worked (below I attached an exemple).
Do you really think that COMSOL is not able to solve this kind of problems? If you're right I have a very big problem!!
Thanks Ivar for your time!
Miquel

Hi

It's not that COMSOL is not able, (For me COMSOL is able to do about anything) but you might need to add some physics by hand yourself

--
Good luck
Ivar

Ok,
so my very big problem has been reduced to a big problem. That's a first step.
Thanks for all.
Miquel

Hi Miquel

I have take a look at your Capacitor model, and I have a few comments:

A general one, to really learn COSOl its worth to take some of the 1 day courses of COMSOL, but leave some 3-4 months inbeteen them and use the tool extensively in between. This allows you to ask the numerous questions one always have when attacking such a general physics tool as COMSOL :)

a) your geometry: you do not need to take all these differences etc, you can overlap the shapes, COMSOL will during it's "analysis" phase decompose the geoemtry into subitems. So its enought o draw the main rectanbgles and close/finish by a "union"

b) materials: the selection active in your model is probabl not the final one. Personally I mostly start with my own material data, because thenI add them one by one and I know what I use, as many of the library material data are complex, with changes, functions related to p, T or other external variables this tends to make the model non-linear or more complex than needed, not the easiest to debug when starting with a tool like COMSOL (start simple)

c) mesh: you do not have many elements through the beam thickness, try to have at least 3 in the thickness when you consider structural flexing items (more even better)

d) structural edge loads: you have different ways to apply edge loads, in your case nortmally one should consider a local edge pressure from the EM fields. You are calculating the total force and distriute this evenly over the area. This is not 100% the same thing, as if you have larger displaceemnts, the electric force in non-linear with the electrode distance hence yo ushould use a local pressure item to be more precise (but for a first approach your case is OK)

e) your physical model is forme somewhat strange, you have GND all around and you charge your electrodes symmetrically, but then you have forces applying on all surfaces, no ? not only on the one between the electrodes. I would have started with GND on the and a voltage (or charge) on theother electrode, and a PML / free air all around

f) your infinite element is rather thick, its enough to have 1/5 or 1/10 of the total size, but I would have left more "air" around the items to be less influences by the air to Infinite element transition.

g) I get some strange errors with your model so I suspect the file is somewhat corrupt, from too many changes. It's worth to restart from scratch in a newer version sucxh as the newly patched 4.0a that is less error prone

Finally some general comments:

think out the physics and how they couple before makign your model. Obviously here you should have two acses (as a first test, then later you might couple tighter): a) en ES model to get the forces, and then a structural case to see the deflection on the beam alone. I would have started with a two model (1 & 2) case each with one physics and then with a force from one to the next basd on a static case in the ES and on a deflection resulting solution for the structural.

Then later I would try out an ALE to have the mesh move and the resulting ElectroMagnetic force change with the deformation but this is a tw physics one model case with deformed mesh. There are a few examples in the doc (more still in the 3.5 doc worth still a study as mostly its portable, once you understand the rules).

--
Good luck
Ivar

Interesting. I've got a simple question about contact pairs in different types of analyses. I'm doing an ALE/plane stress/heat transfer application.

When I add a contact pair, it becomes available in the convection & conduction (cc) boundary settings but not in the plane stress (ps) boundary settings. Any thoughts on why this would occur?

I think I debugged it earlier by choosing the plane stress transient analysis from the Structural Mechanics Module rather than the COMSOL Multiphysics/Structural Mechanics Module. Would this make any difference?