Computing Performance Optimization And Application In Nesoscale Numerical Simulation Of Energetic Materials

The numerical simulation of energetic materials is an important method to study the mechanical behavior of energetic materials.The study of the mechanical behavior of energetic materials by material point method can not only save the manpower and material cost in the experiment process,but also avoid the risk of experiment.But also for the material to suppress the process design to provide theoretical guidance.The material point method separates a whole material area into a number of particles relative to the background grid movement.Each particle represents a corresponding material area and records all the material information in the region.All the discrete particles The collection represents the entire material area.The main calculation process is to map the material information carried by the particle to the background grid,with the background grid to solve the calculation,and finally the calculation results on the grid back to the particle.It can be seen that how to disperse the energetic material quickly into the distribution of the explosive particles in the mold is one of the key problems in the numerical simulation of the mechanical properties of the energetic materials.In addition,the inclusion of energetic materials into many particles,making the calculation of the material point method is very large,how to speed up the numerical simulation of the speed of energy to improve the numerical simulation of energy content of the key.Aiming at this problem,the computational performance optimization scheme of numerical simulation under the mesoscale scale of the energetic material is designed.The scheme mainly includes two parts: the randomized discrete digitization of energetic material particles and the numerical simulation of material point method.For the randomized discrete digitization of energetic materials,the digital model is quickly established by using the stochastic discrete algorithm in the high-level programming language in the case of satisfying the requirements of the properties of the energetic material and the boundary.For the parallel computing optimization scheme,a high performance parallel computing cluster system is built,and an openMPI parallel message delivery library is used to allocate a calculation task to multiple computing nodes.The mathematical model of mesoscale scale simulation is established by random discretization.The numerical simulation of nanostructures of energetic materials on MPI-based high performance parallel system is carried out.The results of simulationdata are analyzed.The experimental results show that under the mesoscopic scale of energetic materials The simulation results show that the filling rate of the mold is increased to 78%,and the random distribution characteristics of the energetic materials are closer to the experimental results in the whole simulation process,and the results are validated.The effectiveness of the random discrete scheme.At the same time,the numerical simulation is carried out on the parallel computing cluster system composed of 12 compute nodes.The speed of completing a calculation task is improved by 8-9times,and the required calculation time is greatly reduced.