Mechanics and Computation / View source

for Atomistic-to-Continuum Modeling for Multiscale and Multiphysics Simulation

=Jeremy Templeton, Ph.D.=
===February 9, 2012===

Atomistic-to-Continuum Modeling for Multiscale and Multiphysics Simulation 

As engineered nanosystems become increasingly common, simulation techniques are required capable of device discovery, characterization, and optimization at the scales on which these systems function.  Finite element (FE) analysis is the macrosale tool of choice for these tasks, but has difficulty accounting for effects which are strictly atomistic in nature, e.g. defects and interfaces.  In contrast, molecular dynamics (MD) can access the relevant  physics through explicit representation of atoms and their trajectories.  However, MD does not include many important processes, such as electron transport, and is computationally intractable for large systems.  In order to develop an appropriate computational methodology for engineering analysis at the nanoscale, multiscale and multiphyics coupling schemes for bridging FE and MD models have been developed.  The foundation of this approach is predicated on concurrent two-way coupling in which an embedded MD simulation informs the FE calculation in regions where high fidelity is required, while the FE represents dynamics which occur at larger scales beyond the MD region.  FE information is propagated to the MD via dynamic boundary conditions derived from Gauss' principle of least constraint.  A similar approach is used to enhance MD with models of electron transport, thereby adding physics not intrinsic to the atomistic representation.  This talk will describe the theoretical coupling framework and provide an overview of several applications of these methods.

Return to Atomistic-to-Continuum Modeling for Multiscale and Multiphysics Simulation.

Retrieved from "http://mc.stanford.edu/Atomistic-to-Continuum_Modeling_for_Multiscale_and_Multiphysics_Simulation"