A New Method for Large Scale Molecular Dynamics Simulations of Shock-Induced Ejecta Production on Petaflopic Computer

We propose a new method for modelling large scale shock-induced ejecta production using petaflopic molecular dynamics (MD) simulations. A copper crystal with a sinusoidal surface finish representative of the roughness arising from a machine polishing is divided into a bulk and a surface part. The bulk part is simulated using the Hugoniostat technique, which allows a very large number of particles to reach a Hugoniot equilibrium state in a short physical time by the mean of a quasi-equilibrium MD simulation. The surface part is simulated with the NVE ensemble in order to account for the non-equilibrium character of the ejection process. With this method, the particle size distribution generated by a system with 5$\times $10$^{8}$ to 10$^{9}$ atoms and different depths of defect is studied. Our MD results show that the particle size distributions exhibit a power law scaling in good agreement with the percolation theory. This theory suggests that the production of ejecta clusters is due to large-scale fluctuations present near a critical point, where there is no dominant characteristic scale.