Hinge rotated by Force field (e.g.: gravity)

Numerical methods and mathematical models of Elmer
zozitak
Posts: 7
Joined: 12 Aug 2022, 15:58
Antispam: Yes

Hinge rotated by Force field (e.g.: gravity)

Post by zozitak »

Dear Elmer Community,

I'm new to FEM, and with Elmer FEM. I've got a problem, that I couldn't overcome alone.
I need to rotate a body in a force field in a transient simulation around an axis.

What did I try so far :
- Extra mesh node in 1,1,0 and displacement 1,2,3 = 0 BC. Here i should have change the material, of the extra mesh to something non-physical... I couldn't solve this, so far.
- Extra body in the left hole, constrain it, and let the hinge slide. Here my extra body shoot out... Probably i did something wrong along setting the Mortar BC conditions, or the Shaft's mesh was too close to the Hinge's mesh.
- Extra holding Force to the Hinge's left circle. This was a partial success. I could calculate some functions, and implement them in MATC, but I couldn't finetune the solution. But the "strange mesh warping" didn't happen. So i think Normal-Tangential Displace = True plus Displacement 1 = 0 is the cause.

I appreciate any advice, comment or reference.

Sincerely,
Zoli
Attachments
Screenshot from 2022-08-14 15-47-16.png
(242.38 KiB) Not downloaded yet
rotatinghinge.zip
(81.95 KiB) Downloaded 14 times
raback
Site Admin
Posts: 4424
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Hinge rotated by Force field (e.g.: gravity)

Post by raback »

Hi,

Have you tried with shorter timesteps? I would guess there is a step short enough for convergence and you can continue from there.

Btw, there is are limiters, contact BCs or mortar BCs in your case so you have some extra keywords that probably are omitted.

-Peter
Rich_B
Posts: 183
Joined: 24 Aug 2009, 20:18

Re: Hinge rotated by Force field (e.g.: gravity)

Post by Rich_B »

Hello,

You describe at least three different scenarios. It would help if you were to post a working example of each scenario, with a specific issue for each one. In other words, break the post into separate, smaller posts, where each post has a specific question.

Also, in your posted solver.log, line 25:
CheckKeyword: Unlisted keyword: [initialize dirichlet conditions] in section: [solver 1]
This can be fixed by moving this statement:
Initialize Dirichlet Conditions = Logical False
from the Solver section to the Simulation section. Moving it doesn't seem to make a difference, but at least the warning goes away.

Rich.
raback
Site Admin
Posts: 4424
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Hinge rotated by Force field (e.g.: gravity)

Post by raback »

This is a broken version of test case "ElasticBeamHinged".
kevinarden
Posts: 1529
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Hinge rotated by Force field (e.g.: gravity)

Post by kevinarden »

I downloaded the zip file and after changing to quads and higher order elements, liner triangles and linear tets are not good for elastic solvers. The zip file appears to have a copy of the third attempt, normal-tangential BCs for the hinge.

The problem appears to be well behave for about a 1/3 of the times steps, but it then appears to hit a stop and bounces back up which also causes the rippling in the model.

Strange thing is there is nothing defined in the model or sif to create such a stop. Could be an anomaly of the N-T BC if it is updating every step perhaps it is creating an unexpected BC after so much rotation?
zozitak
Posts: 7
Joined: 12 Aug 2022, 15:58
Antispam: Yes

Re: Hinge rotated by Force field (e.g.: gravity)

Post by zozitak »

Hi Peter,

Peter thanks for the advise. I tried

Code: Select all

Boundary Condition 4
  Name = "LeftCircle" 
  Target Boundaries(4) = 5 6 7 8
  Normal-Tangential Displacement = True
  Displacement 1 Lower Limit = Real 0.0
End
and

Code: Select all

Boundary Condition 4
  Name = "LeftCircle" 
  Target Boundaries(4) = 5 6 7 8
  Normal-Tangential Displacement = True
  Displacement 1 Upper Limit = Real 0.0
End
with same results. I also tried to decrease the time step size to 0.005 and with 200 steps i've got the same results. I tried to tweak the elstic solver lin and non-lin parameters, but the anomaly didn't seem to solve itself. I tried to switch off and on Displacement 1 Conditions:

Code: Select all

Displacement 1 Condition = Variable Time
  Real MATC "switch(tx(0))"

other kind of deformation happend, which i think is corralated, with the original one.

Yes i took the model from "ElasticBeamHinged" test for demonstration purposes. It is easier to speak about something, if everybody knows the geometry.
Sorry that i broke it. It wasn't my intention.

By the way, if you "drive" it by omega deg/s as Displacement 2 = Real $ omega, then it works like expected.
If i set Displacment 1 = Real 0.0 after Normal-Tangential Displacement = True , it should let only rotation around the axis, am i correct about it, or does this keyword means something else ?

I intend to use the Elastic Solver coupled with Electrokinetic Solver, but i cant post that project out, thats why i tried to make preparation.
zozitak
Posts: 7
Joined: 12 Aug 2022, 15:58
Antispam: Yes

Re: Hinge rotated by Force field (e.g.: gravity)

Post by zozitak »

Hi Rich,

I was maybe confusing so sorry.

The main problem is to rotate the Hinge around it's left circle. For this i tried to get workarounds.
What i posted in the attachment is the main problem, because if Normal-Tangential Displacment would work as i pictured, i wouldn't do the other 3.

Yes you are right Initialize Dirichlet Conditions doesnt really needed, but https://github.com/ElmerCSC/elmerfem/bl ... se.sif#L29
and i thougt it might help somehow.
zozitak
Posts: 7
Joined: 12 Aug 2022, 15:58
Antispam: Yes

Re: Hinge rotated by Force field (e.g.: gravity)

Post by zozitak »

Hello Kevin,

Thank you for your time and effort.
Its defenetly the N-T BC's D = 0 part. If you dont use N-T, and set some Force 2 = 400000.0, then it might flies up, but the strange stress doesnt happen in the hinge. So if i could replace, the N-T part with some Force 1,2 = Variable A,B; Real MATC "shaftforces(tx(0),tx(1))" than i would gadly do it. I will try it definitly.

If you have any idea please share it on the forum. This forum contains invaluable information for Elmer FEM.

And here i want to grab the chance, to thank Elmer's Developer Team, because its really an outstanding FEM Software.
I've seen elmerfem youtube channel, and thats how i decided to join in.
raback
Site Admin
Posts: 4424
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Hinge rotated by Force field (e.g.: gravity)

Post by raback »

Hi,
zozitak wrote: 15 Aug 2022, 19:24 And here i want to grab the chance, to thank Elmer's Developer Team, because its really an outstanding FEM Software.
I've seen elmerfem youtube channel, and thats how i decided to join in.
Thanx zozitak! We appreciate.

Now to get back to this little problem. I did this example (and the 3d variant of it) some day when this was previously discussed. Things are coming back to me now gradually... The challenge with normal-tangential condition is that it only defines a normal vector and 1 or 2 tangent directions. This is not a sufficient constraint for nonlinear hinge. You can follow the tangential direction a little but eventually you should turn to follow the circular path. So in reality any displacement (u_x,u_y) at node (x,y) should be such that (x+u_x)^2+(y+u_y)^2=R^2. This is a nonlinear constraint that can't be implemented by simply defining the n-t coordinate system. I guess it would not be too hard to add it but still might take some hours... Anybody willing to do rigorous testing if we did this?

The normals may be computed from the mesh or as commented out from the fact that we are an a circle. The latter is of course exact and fairs better results.

The alternative currently could be to mesh also the circle and use contact mechanics. There the mortar projector would be updated for each nonlinear iteration. It is an overkill but currently the only way to follor the circular path.

-Peter
raback
Site Admin
Posts: 4424
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Hinge rotated by Force field (e.g.: gravity)

Post by raback »

PS. Actually even simpler way would be to create a stiff circle and set its center to be fixed. The question is what kind of physics needs to be there in the end. Simply a frictionless rotation around a point has its limitations.
Post Reply