Principal Investigators:
Michael Sprague is a senior scientist at the National Renewable Energy Laboratory. Mike’s research interests include high-performance computing and computational mechanics. He is leading several projects in wind energy, including a U.S. Department of Energy Exascale Computing Project called ExaWind. He was an assistant professor of applied mathematics at the University of California, Merced (2005-2010), and he was a postdoctoral fellow in applied mathematics at the University of Colorado at Boulder (2003-2005). His degrees are in mechanical engineering, with a B.S. from the University of Wisconsin-Madison (1997) and a Ph.D. from the University of Colorado at Boulder (2002).
Description:
OpenFAST is an open-source software package for wind turbine simulation and analysis. It is supported by the National Renewable Energy Laboratory under the U.S. Department of Energy (DOE) Wind Energy Technologies Office. OpenFAST encompasses models and associated simulation modules for aerodynamics, substructure hydrodynamics for offshore systems, control and electrical systems, and structural dynamics. OpenFAST modules are coupled to allow for nonlinear analysis of aero-hydro-servo-elastic interactions in the time domain.
OpenFAST serves as the high-fidelity turbine model (structures and control system) in DOE-supported efforts to enable predictive high-performance-computing simulations of whole wind farms, for which the complex flow dynamics are simulated with computational fluid dynamics. The DOE-supported efforts include the Exascale Computing Project, ExaWind, which is focused on creating an exascale-ready simulation capability for wind farms, and the Atmosphere-to-Electrons High-Fidelity Modeling project. OpenFAST also serves as a computer-aided-engineering tool. The wind energy industry relies heavily on computer-aided-engineering tools for analyzing wind turbine performance, loading, and stability. For example, under its use in design and optimization, OpenFAST is run thousands or even tens of thousands of times under various conditions, in which each simulation can take several hours.
Parallelizing OpenFAST is the objective of this Intel Parallel Computing Center. At the start of this project, FAST was a serial-computation code, but its modularity presents obvious pathways for parallelization (e.g., each of the three blade solvers can be run in parallel). Beyond modular parallelism, individual modules present additional opportunities for parallelism, e.g., through the nonlinear solvers. In this project, we are parallelizing OpenFAST with a focus on Intel Xeon and Xeon Phi processors, under a variety of use cases. Success will be measured in time-to-solution-improvement comparisons against the baseline serial-computation cases, and strong scaling tests.
Related Websites:
http://openfast.readthedocs.io/
https://github.com/openfast/openfast
https://www.exawind.org/