A multiscale approach: Developing a Molecular Dynamics-informed RDX chemistry model for characterizing hot spot criticality in continuum-level simulations

Reactive molecular dynamics (MD) simulations can describe the complex physical and chemical processes in high energy density materials and ultimately contribute to a predictive understanding of their shock initiation under dynamical loading. However, computational intensity limits MD to the nanoscale, making it difficult to simulate hot spot formation and eventual shock to detonation transition. Thus, we developed a multiscale model that uses MD simulations to inform a continuum model capable of reaching the microstructural scales. We use dimensionality reduction via unsupervised learning to establish a two-step reduced-order chemistry model for the decomposition of RDX. From both homogeneous isothermal and adiabatic simulations, we extract chemical kinetics and heat of reaction parameters. The continuum model, capable of capturing chemistry, thermal transport and mechanics, is verified using homogeneous cook-off simulations and the multiscale approach validated from hot spot calculations. We find good agreement between the predicted critical temperatures from explicit MD simulations and the continuum model for nanoscale hot spots. Finally, we predict critical hot spot temperature as a function of size and quantify the effect of uncertainties for various materials’ parameters.