Learning Forum

[Paulvh]

Hello everyone,

Currently I'm working on a static structural analysis with two models that describe the failure of individual material and the interface in a multi-material (PC) part. The individual material failure is modelled with the use of EKILL based on the ultimate plastic strain theory. The interface failure is modelled by a separation-distance based debonding (CZM) on contact elements.

The problem that I encounter in this simulation is the high occurence of element HDST violations, i.e. element distortions. To a degree this behavior is expected as when elements are killed via EKILL, their stiffness is reduced significantly, subjecting them to high degrees of deformation. I have however tried many approaches to minimize this effect, including:

• Model setup: decreased loading increments and increased amount of substeps (1mm displacement is loaded linearly ramped over 100 loadsteps, with each having minimum 5 and maximum 1*10^5 (!) substeps), varied CZM artificial damping factor, ensured contact pinball radius is at least the length of the contact gap at the completion of debonding.
• Mesh quality: varied between all mesh methods (multizone/quadrilateral/triangles), varied mesh size (ranging from 0.01mm to 1mm), introduced initial biased aspect ratios of elements that are to be stretched.

Unfortunately, none of the approaches minimizes the occurence of element distortions to an extend where the simulation is able to completely converge in a robust manner. Observing the force convergence solution output and the Newton-Raphson residual force plot (see figures), I can only conclude the simulation is prematurely halted because of the element distortion (highlighted in green in the NR plot).

[Ashish Khemka]

nnWhat is the number of equilibrium iterations you are using? If you have not changed the same then can you try increasing the number of equilibrium iterations using the command in the Setup branch:nneqit,100 ! This increases the equlibrium iterationsnnRegards,nAshish Khemkan

[Paulvh]

Hi Akhemka,nI have come across this suggestion on another forum as well and attempted with increased neqit,50. Unfortunately I noticed no changes. Thanks for thinking along though!nIn the meantime a promising change I have made was changing the discplacement boundary condition to a remote displacement boundary condition, allowing me to change its behaviour from rigid to deformable. Switching between these settings drastically affected the robustness of the model with limited effect on the results. nKind regards,nPauln