This month at SciTech, I participated in the AIAA special session on Hover Prediction, also known as the Hover Prediction Workshop. This particular workshop has special meaning to me because I started my career in CFD when I was a Research Scientist at the US Army Aeroflightdynamics Directorate at NASA Ames Research Center. At that time, I was part of a group of engineers tasked with producing the first full helicopter CFD simulation. That goal was achieved, but the community is still working to define the best practices for predicting helicopter hover. It has been well over 10 years since I've presented to this community, so it was like a homecoming having this opportunity to present to them again.
The workshop brought together seven participants submitting data on the same sample cases but using different meshes, different solver codes and different methods. The goal was to be able to compare all the data to maximize the knowledge extracted.
Intelligent Light supported this effort by contributing time and expertise to develop a standardized, automated, post-processing workflow that facilitated dataset comparison, report generation and knowledge extraction for a diverse set of CFD results.
Our solution allowed the users to upload data that was then run through the automated routines to normalize the data, produce XDB extracts and publish comparison images. Participants are able to explore the XDB data extracts with their own licensed version of FieldView or by downloading the free viewer: XDBview. Generating images and compact XDB files allowed all users to explore their data interactively on their local systems.
In addition to the workflow engineering, Intelligent Light ran two unsteady OVERFLOW simulations for the workshop. These were executed by Intelligent Light on a remote Cray supercomputer that provided the HPC capability to run multiple full transient solutions for the project. We'd like to thank our partners at Cray for their support of this Workshop.
Our team included Intelligent Light Application Engineers Atsushi Toyoda and Michael Burklund, and Intelligent Light Applied Research Group member Christopher P. Stone. We also brought in R-Systems, our on-demand HPC computing partner. R Systems provided an anonymous ftp server where the participants could upload their data and worked with our team to implement the workflow using PBS and parallel servers. It was easy to set up and run the post-processing tasks on the remote HPC systems. I'd like to thank R Systems for their help and support on this project.
Paper: "Direct Comparsion of Hover Prediction Workshop Results", 54th AIAA Aerospace Sciences Meeting, AIAA SciTech,(AIAA 2016-0035),
Earl P.N. Duque, Atsushi Toyoda and Michael D. Burklund, Intelligent Light; Nathan Hariharan, HPCMP CREATE-AV; Robert Narducci, Boeing; Christopher P. Stone, Computational Science & Engineering, LLC.
Participants included Intelligent Light, US Army Aeroflightdynamics Directorate, Boeing, Gerogia Tech, KAIST, ONERA, United Technologies Research Corporation.