Scalable Knowledge Capture is Essential to Avoid CFD Bottlenecks
NASA’s CFD Vision 2030 Study details the many challenges that remain to routinely obtain accurate physics-based predictions of complex turbulent flows, including how to streamline and automate analysis to gain knowledge. Evolving HPC architectures will produce huge amounts of data, and future CFD technologies must be built to both realize the promise and avoid the pitfalls of this uncertain landscape. At Aviation 2015 this summer, Intelligent Light’s Dr. Earl Duque participated in an expert panel that discussed visions for post-processing and knowledge capture to meet the NASA 2030 CFD goals. Dr. Duque will be the lead author on the summary paper targeted for SciTech 2016.
Reduced Order Modeling Identified in the Study as an Enabling Technology
Reduced Order Modeling (ROM) can both compress and summarize, in a physics-oriented way, large unsteady CFD results and experimental data. Dr. Duque’s Applied Research Group at Intelligent Light has been successfully collaborating with BYU in an Air Force Research Laboratory-funded research effort to apply ROMs and Self-Organizing Maps (SOMs) to turbomachinery CFD. This is one example of how a partnership of government, industry and university researchers is working to make NASA’s 2030 CFD vision a reality.