3D composite coupon convergence issue

[IcarusAero223]

I'm doing an Abaqus/Standard model verification for a 3D composite coupon using C3D8R elements and LaRC05 failure criteria (Load-displacement curve).
Interlaminar properties are defined using cohesive interaction property and General contact interaction type.
When running my simulation the convergence seems to be very slow with an average simulation time being around 7-8 hours for 60ish increments.
Checking .msg file I get: "Displacement correction too large compared to displacement increment." and I'm not sure what this error is referring to.
Is there any way I could get my job to converge faster? I tried many things from adjusting load rate to fiddling with boundary conditions and nothing seemed to do the trick.

 

[FEA way]

Displacement correction too large compared to displacement increment.

Abaqus has 2 main convergence criteria - residual forces and displacement corrections. This message only means that the latter criterion is not met in a given iteration. You could try relaxing this criterion in solver controls but it can be risky in some cases (not as risky as messing with residuals though). Basically, the solver may skip some sudden change in the equilibrium path. Relaxing this criterion is a good idea in the cases where there’s no risk of buckling and only corrections are preventing iterations from converging.

I would consider different approaches first but those depend on the details of your model, potential reason of convergence difficulties and the actual progress of convergence during the whole analysis. It might be a matter of too coarse/distorted mesh, for instance. Some analyses always converge poorly in Abaqus/Standard due to their highly nonlinear/transient nature but many can be improved in this regard.

 

[SWComposites]

what are you attempting to do with the interlaminar cohesive elements / contact??

 

[IcarusAero223]

Abaqus has 2 main convergence criteria - residual forces and displacement corrections. This message only means that the latter criterion is not met in a given iteration. You could try relaxing this criterion in solver controls but it can be risky in some cases (not as risky as messing with residuals though). Basically, the solver may skip some sudden change in the equilibrium path. Relaxing this criterion is a good idea in the cases where there’s no risk of buckling and only corrections are preventing iterations from converging.

I would consider different approaches first but those depend on the details of your model, potential reason of convergence difficulties and the actual progress of convergence during the whole analysis. It might be a matter of too coarse/distorted mesh, for instance. Some analyses always converge poorly in Abaqus/Standard due to their highly nonlinear/transient nature but many can be improved in this regard.

Thanks for your answer. I'm validating LaRC05 criteria so I'm unfortunately stuck on Abaqus/Standard. Most definitely convergence issue comes from contact between 12 layers in my model and I'll have to check if the interlaminar properties affect the results in any way.

what are you attempting to do with the interlaminar cohesive elements / contact??

I'm inspecting delamination between certain layer orientation pairs. (interlaminar as opposed to intralaminar)