Forum 360: High Performance Computing's Impact on Aerospace Prediction 9 January 2020 1400 - 1600
The Aerospace industry has a strong desire and need to extend development of their CFD capability as described in the CFD 2030 Vision set forth by NASA and Boeing with input from government, industry, and academia. However, limitations of Reynolds Averaged Navier-Stokes (RANS) methods are preventing us from analyzing vehicles at the edges of their envelopes where performance typically drives designs. Hybrid RANS/Large Eddy Simulation (LES), LES, and Direct Numerical Simulation (DNS) computations require emerging High Performance Computing (HPC) hardware to be useful to vehicle design. The trend we have been taking advantage of over the last several decades is called Moore’s Law, which has led to the performance of machines roughly doubling every 18 months. Many hardware specialists are predicting the end of this performance trend for current hardware and software architectures due to a variety of reasons. Continuation of Moore’s Law will likely require new computational strategies and computer architectures. Our current CFD software will not execute efficiently, if at all, without a conscious effort on our part to prepare it for the future because it is no longer as simple as just buying a bigger/faster computer. In order to accomplish the CFD2030 Vision, HPC issues will have to be discussed and strategies for overcoming them planned for.
Technical Fellow, Flight Sciences, Boeing Commercial Airplanes
Chief Scientist, High Performance Computing Modernization Program, Department of Defense
Director, Exascale Computing Project, Oak Ridge National Laboratory
Senior Research Scientist, Computational AeroSciences Branch, NASA Langley Research Center
DoD HPCMP CREATE-AV Project Manager, U.S. Army Engineering Research and Development Center