Hi Frederique,

There are various reasons why this might be happening. 

Firstly, what are you trying to model? I cannot comment much without knowing what you are modelling.

Secondly, the convergence problem may be happening because you are having a source of non-linearity such as interactions which become exploited in the NLGEOM option. Nonlinear boundary conditions can become an issue in a non-linear analysis which makes the convergence problems more likely.

Once I get an answer, then I could offer more detailed advice.

Hi Mahmoud,

Thank you very much for helping me. The initial problem is complex but I simplified it in order to understand why I couldn't obtain results. So now it's a really simple problem of a homogeneous plate in 3 points-bending test. So it is simply supported at the two extremities and  a force is applied on nodes at the middle of the plate. The rigid body motions are suppressed and the material is elastic linear. No other non linearity are added except NLGEOM. Thank you forward

Hi Frederique,

It is good that you simplified the problem because it makes the problem more manageable. The model does not sound complicated. 

It may be because you are getting high-stress concentrations at the node where you apply the force.Bear in mind it is wrong to apply a force at a single node, a force must applied on multiple nodes to mimic real conditions.

Which solver do you use in Abaqus? Is it Riks, Standard, Explicit,etc?

 If you can attach a copy of the input file and the error messages you get, then this would help to solve the problem a lot quicker.

Hi Mahmoud,

I know it's not very good to put force directly on nodes but it's a very small force and it doesn't result in stress concentration with the linear calculations. Moreover, the linear results are fully the same than those I expected with the analytical solutions. I use standard resolution. It's a very simple problem that's why I don't understand why I can't obtain convergent results. It seems to me that something I don't know should be turn on or off ???

Thank you

HI Frederique,

I previously had issues with applying a force at a single node in Abaqus. I still think that it would be a good idea if you try applying it over a line of nodes. 

By the way, what is the material type? Also, Do you use Riks or Standard ?

Hi Mahmoud,

I'm not applying a force on only one node but on a line of nodes at the middle of the sample. The material is linear elactic and I use Standard calculations. I thought of trying explicit calculations with very slow loading to force calculations to be made with NLGEOM automatically on "on". I'm not having time enough now but I will try next week...

The second point I met with my problem is the following : As I will further use an own behaviour law with UMAT subroutine, I have to manage with "transverse shear stiffness" option with the shell section definition and it seems not too easy to deal with. Have you got some experience on it?