A new scalable and efficient implementation of the mesoscopic distinct element method for massively parallel numerical simulations of carbon nanotube systems is introduced. Carbon nanotubes are represented as chains of rigid bodies, linked by elastic bonds and dispersive van der Waals (vdW) forces. The enhanced vector model formalism of the elastic bond between rigid bodies, developed recently, is employed here to capture the elastic deformation of nanotubes. Dispersive interactions between the neighboring nanotubes are described with the coarse-grained vdW potential. Time integration is performed using a velocity Verlet integration scheme with tunable damping in order to describe the energy dissipation to the implicit degrees of freedom. Due to the scalable message passing interface (MPI) parallelization, enabled by rigid particle dynamics module (PE) of the waLBerla multiphysics framework, our method is capable of modeling large assemblies of carbon nanotubes. This advance enables us to move closer to the length and time scales required to extract representative mechanics of carbon nanotube materials. The promising scalability of the new implementation is probed in two examples of self-assembly of ultra-thin carbon nanotube films and carbon nanotube buckypapers, where formation of hierarchical networks of carbon nanotube bundles, storing both elastic and vdW adhesion energies is observed. The relaxation of one cubic micrometer of buckypaper illustrates the code scalability.
Bibliographical noteFunding Information:
Acknowledgments. I. A. Ostanin acknowledges the financial support from the Russian Science Foundation (RSF) under grant 17-73-10442 (the development of the system of CNT modeling based on waLBerla software). T. Dumitrica acknowledges support from NASA’s Space Technology Research Grant NNX16AE03G. V. A. Kuzkin acknowledges financial support from the Russian Science Foundation under grant № 17-71-10213. High-performance computations presented in the paper were carried out on Skoltech Pardus HPC cluster http://crei.skoltech.ru/cdise/hpc/the-skoltech-hpc-cluster-pardus-by-ibm-lenovo/.
© 2018, Institute for Metals Superplasticity Problems of Russian Academy of Sciences. All rights reserved.
- Carbon nanotubes
- Distinct element method
- Enhanced vector model
- Parallel computing