Topic: Rigidarmnodes in 2D analysis

Hello,
I am currently trying to simulate a 2D beam subjected to an eccentric load (at node 3). I have a roller support(node 5) and a pin support (node 1). On top of the beam there is a steel plate which is not in the model. However due to the beam all the nodes along the top surface of the beam (nodes 6,4,3)  should have the same rotation and the same x-translation. As I am using elements with 2DOFS i have to enrich them first with the rotation DOF and make some of them rigidarmnodes. The OOFEM script is the following

Although the script seems to be running I have convergence problems:

Error: (/Users/demexen/oofemsvn/trunk/src/oofemlib/calmls.C:1190)
Class: CylindricalALM, number: 1
computeDeltaLambda: discriminant is negative, solution failed

Could you please help me with my problem?Is it due to a bug in the algorithm or due to the boundary conditions set? Could I use an alternative formulation?

Re: Rigidarmnodes in 2D analysis

It seems the file I attached was not attached so I am opting the input file. I am sorry about that:


square.out
Square
NonLinearStatic nmsteps 1 nsteps 1 nmodules 1 contextoutputstep 1
nsteps 1 rtolv 1.e-4 stiffMode 1 maxiter 5000 controllmode 0 psi 0.0 hpcmode 1 hpc 2 3 2 hpcw 1 1 steplength 0.0002 minsteplength 0.0002 donotfixload
vtk vars 7 1 4 13 12 64 65 66 tstep_all domain_all
domain 2dPlaneStress
OutputManager tstep_all dofman_all element_all
ndofman 6 nelem 4 ncrosssect 1 nmat 1 nbc 3 nic 0 nltf 1
node 1 coords 3 0.000000e+00 0.000000e+00 0.000000e+00 bc 2 1 1
node 2 coords 3 0.500000e+00 0.000000e+00 0.000000e+00
node 3 coords 3 1.000000e+00 1.000000e+00 0.000000e+00 load 1 2 ndofs 3 dofidmask 3 1 2 6 
rigidarmnode 4 coords 3 0.500000e+00 1.000000e+00 0.000000e+00 ndofs 3 dofidmask 3 1 2 6 master 3 masterMask 3 1 0 1 dofType 3 2 0 2
node 5 coords 3 1.000000e+00 0.000000e+00 0.000000e+00 bc 2 0 1
rigidarmnode 6 coords 3 0.000000e+00 1.000000e+00 0.000000e+00 ndofs 3 dofidmask 3 1 2 6 master 3 masterMask 3 1 0 1 dofType 3 2 0 2
TrPlaneStress2d 1 nodes 3 1 4 6 crossSect 1 mat 1
TrPlaneStress2d 2 nodes 3 4 2 1 crossSect 1 mat 1
TrPlaneStress2d 3 nodes 3 2 3 4 crossSect 1 mat 1
TrPlaneStress2d 4 nodes 3 3 5 2 crossSect 1 mat 1
SimpleCS 1 thick 1.0 width 1.0
isole 1 d 0.  e 29.28e9 n 0.2 tAlpha 0.
BoundaryCondition  1 loadTimeFunction 1 prescribedvalue 0.0
Nodalload 2 loadTimeFunction 1 components 3 0.0 -1.0 0.0
BoundaryCondition  3 loadTimeFunction 1 prescribedvalue 0.0
ConstantFunction 1 f(t) 1.0

Post's attachments

square.in 1.4 kb, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

Re: Rigidarmnodes in 2D analysis

Hi.
I can't understand what you are trying to achieve

There are no actual beams/plates or gradient theory elements here, so there is no way to prescribe a moment in any single node (and then no reason to use rigid arm nodes either)
Did you forget to add the beam element?

Re: Rigidarmnodes in 2D analysis

I am sorry! I thought this was done automatically by OOFEM. I misunderstood the command.
Thanks for the information!

Re: Rigidarmnodes in 2D analysis

If I understand your problem correctly, then I dont think you need to have any rigid arms to achieve this. You can prescribe a moment on rotation dofs directly if necessary.

Note that even if you apply a beam along the edge of a TrPlaneStress2d element, it will be as if the beam is connected in momentless joints on each side;
If you want the plane-stress solids to be connected you need to have compatible formulations, i.e. some type of gradient theory elements, which I'm not sure if its implemented.


I understood what you described as something like this (in a crude ASCII-drawing);

         F (load)
         |
/|       v
/|o------o  <--- beam
/||______|  <--- solid brick
/|

in which case rigidarmnodes would not help.