Although numerical algorithms for 2D crack simulation have been studied in Modeling and Simulation (M&S) and
computer graphics for decades, realism and computational efficiency are still major challenges. In this paper, we
introduce a high-fidelity, scalable, adaptive and efficient/runtime 2D crack/fracture simulation system by applying the
mathematically elegant Peano-Cesaro triangular meshing/remeshing technique to model the generation of
shards/fragments. The recursive fractal sweep associated with the Peano-Cesaro triangulation provides efficient local
multi-resolution refinement to any level-of-detail. The generated binary decomposition tree also provides efficient
neighbor retrieval mechanism used for mesh element splitting and merging with minimal memory requirements essential
for realistic 2D fragment formation. Upon load impact/contact/penetration, a number of factors including impact angle,
impact energy, and material properties are all taken into account to produce the criteria of crack initialization,
propagation, and termination leading to realistic fractal-like rubble/fragments formation. The aforementioned parameters
are used as variables of probabilistic models of cracks/shards formation, making the proposed solution highly adaptive
by allowing machine learning mechanisms learn the optimal values for the variables/parameters based on prior
benchmark data generated by off-line physics based simulation solutions that produce accurate fractures/shards though at
highly non-real time paste. Crack/fracture simulation has been conducted on various load impacts with different initial
locations at various impulse scales. The simulation results demonstrate that the proposed system has the capability to
realistically and efficiently simulate 2D crack phenomena (such as window shattering and shards generation) with diverse potentials in military and civil M&S applications such as training and mission planning.
|