Three parallel algorithms for classical molecular dynamics are presented. The first assigns each processor a fixed subset of atoms; the second assigns each a fixed subset of inter-atomic forces to compute; the third assigns each a fixed spatial region. The algorithms are suitable for molecular dynamics models which can be difficult to parallelize efficiently—those with short-range forces where the neighbors of each atom change rapidly. They can be implemented on any distributed-memory parallel machine which allows for message-passing of data between independently executing processors. The algorithms are tested on a standard Lennard-Jones benchmark problem for system sizes ranging from 500 to 100,000,000 atoms on several parallel supercomputers--the nCUBE 2, Intel iPSC/860 and Paragon, and Cray T3D. Comparing the results to the fastest reported vectorized Cray Y-MP and C90 algorithm shows that the current generation of parallel machines is competitive with conventional vector supercomputers even for small problems. For large problems, the spatial algorithm achieves parallel efficiencies of 90% and a 1840-node Intel Paragon performs up to 165 faster than a single Cray C9O processor. Trade-offs between the three algorithms and guidelines for adapting them to more complex molecular dynamics simulations are also discussed.

Fast parallel algorithms for short-range molecular dynamics

Experimental modelling of orogenic wedges: A review

The dynamic interaction that occurs in the soil tillage process includes a high rate of plastic deformation and soil failure, characterized by flow of the soil particles. The discrete element method (DEM) seems to be a promising approach for constructing a high-fidelity model to describe soil–implement interaction. Proper prediction of this interaction using DEM depends upon the model parameters. However, there is no robust method for determining the parameters for discrete element models. In this study, the determination of parameters was based on in situ field tests, which consisted of sinkage tests performed with different penetration tools. Based on each test, a plot of force versus displacement, or a so-called “real curve,” was drawn. Discrete element models were built in correspondence with the field tests. “Simulation curve” plots were obtained from the results of the simulation of force versus displacement. In order to minimize the area difference between the real and simulation curves, an inverse solution technique using the Nelder–Mead algorithm of optimization was employed. The optimization results of this particular problem are sensitive to the initial estimate of the parameters. In order to achieve a unique solution, the initial estimate must be close enough to the proper value of the parameters. An energy method and elastic–plastic rule were developed to determine the initial estimation for the optimization process. The described methodology was verified experimentally and numerically; good correlation was achieved between the soil mechanical behavior obtained by experiments and the discrete element simulations.

Determination of discrete element model parameters required for soil tillage

Modeling soil-tillage interaction is a complex process due to dynamic soil-implement interaction which includes a high rate of plastic deformation and soil failure, characterized by the flow of soil particles. The need for a sound modeling technique for soil-implement interaction is the motivation for the present work. The discrete element method (DEM) seems to be a promising approach for constructing a high-fidelity model to describe the soil-implement interaction and can serve as a predicting simulation tool in the process of designing the implement shape. The wide cutting-blade interaction was modeled using a 2D discrete element code-PFC2D and the soil particles by clumps of two disks with a cohesion force contact model between the particles. Four different blade shapes were analyzed by the discrete element model and experimentally by a soil box filled with sand. A very good correlation was obtained between the discrete element simulation and the experimental results. The simulations indicate an increasing horizontal force applied on the blades during motion, as a result of the piling effect of the soil in front of the blade. It was found that the soil flow beneath the blade tip can affect the vertical force applied on the blade. The simulation results were also compared with classical soil mechanics theories for straight blades (the McKyes approach). A good correlation was obtained between the simulation results and McKyes approach for the horizontal force applied on the blade. Weaker correlations were obtained in the vertical direction. This finding can be explained by the soil particle flow beneath the blade tip, which the McKyes approach, does not take into consideration. Observation of the simulation revealed that the failure curve could be reasonably described by a straight line, as assumed in the classical theories.

Interaction between soil and a wide cutting blade using the discrete element method

Algorithm and implementation of soil–tire contact analysis code based on dynamic FE–DE method

This paper describes the current work on a large deformation soil model to demonstrate the feasibility of particle models to simulate full-scale vehicle-soil interaction problems in which the soil undergoes large excavation-like deformation. To achieve this objective, boundary conditions that accurately represent the vehicle geometry had to be incorporated into a 3D discrete element model. The approach taken was to use a finite-element grid to model the vehicle component interacting with the soil and develop routines to model the particle-grid interactions. The particle-grid interactions were more complicated than the particle-particle interactions required for the soil simulations and pose the greatest challenge to the use of computational parallelism. Two examples are presented in which vehicle components are modeled by finite elements that interact with 10 million discrete soil elements. Important theoretical issues are briefly noted concerning mechanics of granular media that are critical to acceptance of the nascent discrete element modeling technology.

Large Scale Discrete Element Modeling of Vehicle-Soil Interaction

Video tracking for experimental validation of discrete element simulations of large discontinuous deformations

The Discrete Element Model (DEM) is an alternative to the classical continuum mechanics approach for problems with flow-like deformations of solids. DEM is especially well-suited for problems such as soil plowing where the media is subject to large discontinuous deformations. The goal of this project was to develop a large-scale three dimensional DEM soil modeling system capable of handling problems with potentially several million particles. This paper discusses the development of that model and the factors driving it toward a High Performance Computing (HPC) solution. It also discusses how work in progress is making use of heterogeneous HPC environments and high speed networks to produce a real-time user interaction and visualization capability.

Design of a Large Scale Discrete Element Soil Model for High Performance Computing Systems

It is widely recognized that simulation is pivotal to vehicle development, whether manned or unmanned. There are few dedicated choices, however, for those wishing to perform realistic, end-to-end s ...

/pdf/an-interactive-physics-based-unmanned-ground-vehicle-3ivf16fvg5.pdf

An Interactive, physics-based unmanned ground vehicle simulator leveraging open source gaming technology: Progress in the development and application of the virtual autonomous navigation environment (VANE) desktop

: A constitutive theory is developed for granular material undergoing arbitrarily large deformations. A three-dimensional discrete element model (DEM) was developed to simulate granular material. The computational efficiency of the DEM was improved to allow for modeling of large particle systems. The need for large particle simulations was to develop on ability to model laboratory experiments on a one-to-one basis so that the discrete elements model could be evaluated against real soils. A comparison was made between laboratory experiments involving very large discontinuous deformations in sand and numerical simulations using large-scale DEM computation. The magnitude of the simulation provided a unique opportunity to assess the validity of the DEM, based on experimental results. The agreement between the experimental and simulated particle motions in the plowing experiment indicates that details not captured by the simplistic particle interaction model may not be relevant in statistically large assemblies. Once it was established that the discrete element method provided a reasonable model for real granular material, an averaging scheme was developed to convert properties local to the particles (e.g., mass, momentum) into continuum attributes (e.g., density, velocity gradients). From this averaging scheme, a new constitutive law was developed to model large deformation of granular material.

/pdf/application-of-dem-to-micro-mechanical-theory-for-large-2blz00ve3o.pdf

David A. Horner

Papers

Large Scale Discrete Element Modeling of Vehicle-Soil Interaction

Video tracking for experimental validation of discrete element simulations of large discontinuous deformations

Design of a Large Scale Discrete Element Soil Model for High Performance Computing Systems

An Interactive, physics-based unmanned ground vehicle simulator leveraging open source gaming technology: Progress in the development and application of the virtual autonomous navigation environment (VANE) desktop

Application of DEM to micro-mechanical theory for large deformations of granular media