A recent version of FEA software is said to give engineers a single solver to perform multidisciplinary analyses including linear statics, dynamics, and implicit and explicit nonlinear features to simulate applications that span everyday common to complex. High performance computing productivity and nonlinear improvements in MSC Nastran are the focus of its latest release.
Improve fidelity of nonlinear simulations
New methods for contact detection through the new segment-to-segment approach improves the fidelity of contact analysis. Major enhancements also include implementation of non-symmetric stiffness matrix for sliding friction and finite sliding capability for deformable bodies. These significant improvements provide accuracy coupled with easy set up of contact analysis, helping users achieve higher productivity and accuracy.
Improved simulation performance
Through collaboration with Intel, developer MSC Software has enhanced MSC Nastran to take advantage of the Advanced Vector Extension (AVX) on Intel’s Sandy Bridge processor on the Windows operating systems in addition to its support for Linux systems completed earlier in 2012. Users with this processor can benefit with improved performance as the AVX is designed to help with numerically intensive applications such as MSC Nastran.
Accelerate nonlinear analysis
The developer says MSC Nastran’s nonlinear analysis is capable of running analyses in distributed and shared-memory parallel modes (DMP and SMP, respectively). This productivity enhancing capability executes nonlinear element stiffness, stress and force calculations, thereby providing improved parallel scalability and applicability for a wider range of nonlinear applications. Improved I/O caching is also now implemented to reduce wall time for large engineering models.
Enhancements to MSC Nastran’s explicit solver’s DMP technology also let users achieve significant performance gains for complex fluid structure interaction problems.
“Tire safety is of great importance and is a top design priority for Hankook Tire,” says Mr. Joonyong Heo, Manager, Structure Analysis Research Team of Hankook Tire. “To predict the tire performance on wet surface conditions, a complex and CPU intensive fluid structure interaction application, we used MSC Nastran 2012.2 to run a tire model with approximately 2 million elements in a Distributed Memory Parallel (DMP) environment with multiple cores. The performance improvements of the explicit solver in MSC Nastran 2012.2 were impressive with more than 3X performance improvement.”
Filed Under: News