Can I solve this with Elmer

[kmeld]

Hi,
I am new to elmer (found it only few days ago) so I still have no experience. I havewer have some experience with FEM.
Programs that I had used till now do not alow me to solve problem I now have in the way I want, so I would to ask you if this will be possible with elmer.

The thing is (not exactly my case but wuite the same still easyer to explain):
1) I have two elements: cylinder and pipe, cylinder is inside the pipe touching iner surface, dimetre of the cylinder is smaler then iner dimeter of the pipe
2) this can be 2d or 3d (no diference since they are wery long)
3) Cylinder is steel and pipe is steel
4) Cylinder is pusched with volume force (8489 * gravity) against the pipe wall

What I need is to solve the problem (linear elasticity) but NON linear contact (adaptive mash to simulate real deformation of both elements during contact)

Can this be done with elmer (preferably in easy to understand way)?

Regards
Cris

[raback]

Hi, Unfortunately not. The little work on contact models has been experimental and limited in functionality. -Peter

[kmeld]

OK so straihgt awey it is not possible but....

I was trying to solve this (not exactly this but similar [easier to model] )problem in softwear I know beter and come up with an folowing idea.
For previously described case it would be as follows:
1) model pipe and cylinder but not touching (littlr distance)
2) all remining space in the pipe model as "soft" material (small/very small Yang Module, Poissson ratio =0) devided in small radial segments.
3) constrains: each "soft" segment would be "fixed in such a way that it could ony deform propendicoulary to pipe surface.

I think this should mor o less work becauce penetration shouln not have place and force from cylinder would be ewentualy transfeded to pipe cosing it to deform in some area because at list few "softis" would take the force.
Unfortunatly in my case for some reasins non linear solution did not work (I thin the solver have some bug because iterations did converge but it didnt stop iterating)

4) solve this problem using iterative metod preferably with mesch adaptation.

Dose this make sence? or I had forgot of something?
Dose anyone have any constructiv opinion on above?

Regards,
Cris

[raback]

Hi Cris

Well, I guess it might work. There is actually a nonlinear version (called ElasticSolver) of the StressSolver which would eliminate the need for mesh adaptation (unfortunately it doesn't have stress computation...). No matter which you choose there would probably be a limit on how soft the material could be. Too soft material would result to elements that have poor aspect ratios. Also overshooting in the nonlinear iteration might spoil the mesh completely, and the linear systems might be difficult if the Youngs modulus would vary a lot. But still its worth a try.

-Peter

PS. We've been coding particle tracker into Elmer the last few weeks. Previously the generality of the contact problem was diffucult since the contact is difficult to determine between arbirtrary shapes. Here actually the particles tracker machinery might work nicely since it would help finding nodes (~particles) that are within elements and help them trace back to the line/surface of contact, and all this at quite a low cost. Unfortunately this would still be quite a coding effort but might be an idea for us to test time permitting.

[kmeld]

Hi,
Thanks for the remark about elemet's aspect ratio. This shourly will cose problems.