An automated remeshing algorithm for the numerical analysis of propagating delaminations
Topic(s) : Special Sessions
Co-authors :
Martulli LUCA MICHELE (ITALY), Salvi LEONARDO GUIDO , Andrea BERNASCONI (ITALY)Abstract :
Delaminations are a critical failure mode for composite laminates under fatigue loading. The prediction of such defects and their propagation is thus a of paramount importance. The Virtual Crack Closure Technique (VCCT) is one of the most adopted techniques for this purpose [1]. VCCT is a nodal technique, which means that the computation of the Strain Energy Release Rate (SERR) is performed at the node-level. This makes this technique particularly sensitive to the relative orientation between the delamination front and the mesh. When large and curved delamination fronts are present, the non-orthogonality between the mesh and the front causes non-physical SERR concentration [1].
In this work, we developed a Python algorithm that performs a sequence of simulations while maintaining the mesh front as orthogonal as possible to the mesh front. The algorithm discretises a single load case into multiple incremental simulations. It first automatically launches a simulation and extract the results on the propagating delamination front; this front is then smoothed and a new simulation is launched with a new mesh orthogonal to it. Figure 1a and Figure 1b show a representative delamination front modelled on a standard mesh and on an orthogonal mesh, after it was smoothed by the algorithm.
Figure 1c shows the SERR distribution on a crack front of a non-standard double cantilever beam specimen, [2]: as shown, the standard mesh presents numerical peaks of the SERR where the delamination front presents step-like discontinuities; these are absent in the remeshed delamination front, which is smoother. As a result, local values of the SERR are smaller, which prevents premature delamination propagation. The algorithm is further validated against the benchmark case of [2].
Figure 1a also shows how the standard meshes necessarily require step-like discontinuities to model the delamination fronts: in these locations, an orthogonal direction to the front is not defined, and thus it is not possible to decompose the SERR into its mode II and mode III. This is solved completely by the remeshing algorithm, which ensures a smooth and continuous description of the delamination front.
The algorithm thus allows a more reliable adoption of the VCCT to model delaminations. It has been validated against prevailing mode I experimental cases, while mode II and mixed-mode validation is currently ongoing. Finally, the remeshing algorithm will be implemented in other sequential algorithms targeting fatigue predictions [1].
In this work, we developed a Python algorithm that performs a sequence of simulations while maintaining the mesh front as orthogonal as possible to the mesh front. The algorithm discretises a single load case into multiple incremental simulations. It first automatically launches a simulation and extract the results on the propagating delamination front; this front is then smoothed and a new simulation is launched with a new mesh orthogonal to it. Figure 1a and Figure 1b show a representative delamination front modelled on a standard mesh and on an orthogonal mesh, after it was smoothed by the algorithm.
Figure 1c shows the SERR distribution on a crack front of a non-standard double cantilever beam specimen, [2]: as shown, the standard mesh presents numerical peaks of the SERR where the delamination front presents step-like discontinuities; these are absent in the remeshed delamination front, which is smoother. As a result, local values of the SERR are smaller, which prevents premature delamination propagation. The algorithm is further validated against the benchmark case of [2].
Figure 1a also shows how the standard meshes necessarily require step-like discontinuities to model the delamination fronts: in these locations, an orthogonal direction to the front is not defined, and thus it is not possible to decompose the SERR into its mode II and mode III. This is solved completely by the remeshing algorithm, which ensures a smooth and continuous description of the delamination front.
The algorithm thus allows a more reliable adoption of the VCCT to model delaminations. It has been validated against prevailing mode I experimental cases, while mode II and mixed-mode validation is currently ongoing. Finally, the remeshing algorithm will be implemented in other sequential algorithms targeting fatigue predictions [1].