The Linear Algebra Team of Ansys LST has added two new eigen solver technologies for the standard vibration analysis problem of structural mechanics. The first, LOBPCG, is based on iterative solution technology to reduce the cost of the direct solution used by the default eigen solver Lanczos. The second, Fast Lanczos, is a new innovative implementation of the Block Shift and Invert Lanczos algorithm targeting the computation of thousands of eigenmodes with less accuracy for the Noise-Vibration-Harshness (NVH) application.
Many structures, machines, or devices are operated partially in water, like surface ships, vessels, semi-submersible platforms. Some others may work completely in water, like submarines. For any of them, water has an important influence on their dynamic response. For most cases, a strong coupling between structures and surrounding water is required to get a good simulation of vibration response of structures subjected to shock or wave loadings. Unfortunately, a fully coupled simulation involving both structures and water explicitly can be expensive. Besides, with the traditional finite element method, meshing a large volume water body and defining non-reflection boundary conditions (e.g., Perfectly Matched Layer [1]) on the truncated boundary can be challenging and needs some experience.
The multifrontal method of Duff and Reid [1] dominates the runtime of most LS-DYNA im-plicit analyses. Its complexity will range from O(N1.5) to O(N2), depending on the model. This paper will give an overview of attempts to reduce the run time of solving large systems of linear equations, both on the host processor as well as with accelerators. Most of what is discussed herein is available today in the development version of LS-DYNA and should be released with R15. Everything discussed herein only applies to our symmetric indefinite solver.
New Eigen Solver Technology in LS-DYNA