Multidomain and Integrative Capabilities for Large-Scale Systems Simulations with High-Order Methods In this project we seek to develop an efficient, accurate and scalable computational tool for high fidelity simulations of realistic large-scale and multi-scale engineering and physical systems on latest petascale and future exascale platforms. Specifically, we will extend a high-order spectral-element computational code Nek5000 to include a new class of computational techniques to allow for integrative simulations on multiple domains and with multiple computational modules. Each module is treated independently with a separate message-passing interface (MPI) communicator, thus allowing for a flexibility of extension to different integrative scenarios: overlapping grids, moving grids, multiscale coupling, multiphysics coupling. The first three scenarios will be addressed in the current proposal. We seek to implement and systematically validate the proposed techniques versus known analytical solutions and existing experimental data.
|Effective start/end date||9/15/12 → 8/31/16|
- National Science Foundation (NSF): $264,720.00