Convergence criteria.

[FrankMalone]

I'm using NX Nastran and SOL106. I have made a non linear analysis, and I have problems with the convergence of the solution.
In the convergence criteria menu there are three possibilities, displacement, load and work, by default load and work are checked by the software.
If I check displacement and work and uncheck load, the solution converge. But if I try the default load and work the solution does not converge.

Could be the solution accurate enough? What is the difference between the convergence options? Which must I choose?

Regards
Frank.

 

[MechanicalEngUCL]

I am not familiar with the software you are using but most FEA programs have similar non-linear solver options.
The first option provides a DISPLACEMENT CONTROLLED solution, i.e. the solver stops when a pre-defined (by user, which is you) displacement is reached. This may be in UX, UY, UZ directions. As long as you have an idea of the deformation (or displacement) the geometry is undergoing, this is a reliable convergence option.
The load option provides you a solution in which load is applied and nodal displacement calculated... depending on the material model you have used, the solution does not converge because nodal displacement tends to infinite.

Displacement control is fine to use, but if you are trying to obtain the ultimate load carrying capacity of your geometry, by defining a maximum displacement, you might interrupt the solution prior to obtain yielding.

Load controlled is usually employed however can be unstable. Try to use a different material model (such as Bilinear isotropic model defining yield stress and tangent modulus). Try to increase the load steps. Or try to use the arc-length method.