Peridynamics theory, shown to be consistent with molecular dynamics, is enabling predictive simulation of failure in highly heterogeneous materials and shows promise for coupling atomistics to continuum scales. Peridynamics approaches are being used to investigate damage and delaminations in composites for future Boeing aircraft structures. A recently developed peridynamics-based stress tensor that is consistent with classical continuum mechanics enables coupling to finite-element methods. The image at left shows a peridynamics-based simulation of a projectile impact, showing failure, crack initiation, and propagation in the target material(1400) NW, ST&E, DS&A
As part of the Advanced Simulation and Computing Algorithms Milestone, Sandia demonstrated vertical integration of Trilinos algorithms with the Charon semiconductor device modeling code. Algorithms ranging from preconditioned iterative solvers through nonlinear solvers, transient solvers, and optimization were integrated in a modular yet high-performance manner, yielding up to ten- fold reductions in solution time while providing more accurate results. The algorithms are available to the Qualification Alternatives to SPR Program to optimize modeling parameters, and they are being used to explore the sensitivity of transistor device models to input parameters. (1400) NW, ST&E, DS&A
A cross-discipline team of researchers in the data mining, visualization, data analysis, and information security areas designed and built the Data Analysis and Visualization Laboratory in Sandia’s Distributed Information Systems Laboratory (DISL). The lab includes three high-resolution projectors for group analysis and visualization of data related to national security problems. The room includes lab space for working on group projects, several high-end computer workstations for open use, and facilities suitable for meetings and presentations to VIPs. (8100, 8900) HS&D, DS&A
Life is multidimensional, but many of the nation’s calculations and simulations are done using two-dimensional mathematical data structures called matrices. The Tensor Toolbox for MATLAB is enabling the use of multidimensional data structures, called tensors, for solving problems in domains such as data analysis, image and signals processing, psychometrics, and chemometrics. On a laptop, the software can easily do numerical computations on problems of size 10,000 x 10,000 x 10,000 with more than a half million nonzero entries. More information at http://csmr.ca.sandia.gov/~tgkolda/TensorToolbox/. (1400, 8900) HS&D, DS&A
The Design Analysis Kit for Optimization and Terascale Applications (DAKOTA) suite of systems analysis tools works with simulations to address assessment and design questions. Developed with Lockheed Martin, DAKOTA has been deployed within DOE to facilitate risk-informed national security decisions. DAKOTA 4.0 enables predictive simulations and performs decision making for stockpile stewardship. While DAKOTA applications can be proprietary, the software itself is not. Free downloads are available, allowing the emergence of DAKOTA as a de facto standard for systems analysis. (1400, 330) ER&N, ITS
OVIS-2, a tool for intelligent, scalable real-time monitoring of large computational clusters, has been developed and deployed for prerelease testing on Sandia’s Thunderbird and Talon clusters. It is being used for post-processing analysis of Red Storm system data. OVIS uses a unique statistical approach to characterize behaviors of cluster components and detect abnormalities and outliers that could be indicative of impending failures. Advances in OVIS-2 include a three-dimensional visual display, enhanced statistical capabilities, and a distributed clustered database back end for scalability and fault tolerance. (8900, 8300) NW
Ed Thomas (12227) and Tim Draelos (5635) have developed a method for authenticating concealed and statistically varying multidimensional data. Their approach provides a way to measure the statistical similarity of data objects in their concealed states. The calculated Euclidean distance metric of the concealed data is identical to the Euclidean distance metric between the data objects prior to the transformation used for their concealment. This technology can be used for authentication when the information itself is private in nature, such as for biometric authentication applications. (12200, 5600) DS&A
Capabilities were added to the Advanced Simulation and Computing SIERRA mechanics code that greatly enhance the ability to predict mechanical response for nuclear weapon applications. These capabilities include new elements for extremely large deformations, explicit dynamics algorithms that provide up to two orders of magnitude speed-up in computation time, and improved reentry pressure and acoustics modeling capabilities. These capabilities represent the next step in computational prediction of margins and uncertainties. They were applied to analyses performed for the W76-1 qualification and component design. (1500) NW