Marching toward the 2030 vision of CFD

By MUJEEB R. MALIK|December 2018

In 2018, AIAA approved formulation of the CFD Vision 2030 Integration Committee to advocate for, inspire and enable community activities recommended by the vision study for revolutionary advances in the state-of-the-art of computational technologies needed for analysis, design and certification of future aerospace systems.

The NASA-sponsored CFD Vision 2030 Study presented a bold vision for physics-based computational capabilities when it was released in 2014. This road map guides research toward accomplishing this vision, and important work was carried out in 2018.

The vision study highlighted the role of leadership-class computing in achieving the simulation goals of the aerospace community. In a joint collaboration with NVIDIA Corp.; Old Dominion University; ParaTools Inc.; the U.S. Department of Defense High Performance Computing Modernization Program’s Productivity Enhancement, Technology Transfer, and Training organization; and the Oak Ridge Leadership Computing Facility, researchers at NASA’s Langley Research Center in Virginia in May demonstrated impressive next-generation computational performance for the NASA Navier-Stokes solver FUN3D, an application for analyzing complex fluid flows. The team explored a broad range of programming models appropriate for the many core architectures driving the current landscape of leadership-class high-performance computing facilities and used an approach based on the CUDA model to port and optimize the complete suite of FUN3D kernels. To demonstrate the efficiency of the implementation at scale, the team leveraged early access to Summit, the new IBM AC922 leadership-class system with NVIDIA Tesla V100 graphics processing units, or GPUs, installed at Oak Ridge National Laboratory in Tennessee. This system in June earned the rank of the world’s most powerful supercomputer. Experiments were performed in April and May using as many as 1,024 hardware nodes, comprising a total of 6,144 Tesla V100 GPUs. The performance of the GPU approach was compared to that of a conventional hybrid MPI/OpenMP formulation running exclusively on the dual-socket IBM Power9 central processing units with 44 cores each. Tests showed a nominal 30x speed advantage for the GPU approach, or a total throughput equivalent to that of approximately 1 million central processing cores.

The study also emphasized the role of CFD validation experiments. Turbulence modeling research for high temperature flows has been rather limited, largely due to the difficulty of making detailed measurements in such flows. A series of tests, referred to as the Turbulent Heat Flux experiments, have been underway at NASA’s Glenn Research Center in Ohio to collect data that could lead to advances in computational modeling of such flows. A square nozzle delivering a hot jet exhaust over a rectangular plate with 135 cooling holes was tested in May and June using particle image velocimetry and the Raman spectroscopy technique for velocity and temperature measurements, respectively. Extensive data was obtained for a range of nozzle exhaust temperatures and Mach numbers along with cooling hole blowing ratios. The configuration with a large number of cooling holes is representative of a high temperature nozzle exhaust or turbine blade.

In August, NASA initiated negotiation for 12 research awards to various universities and industry to make advances in turbulence simulations, numerical algorithms, benchmark experiments, multidisciplinary analysis and optimization, and to establish requirements for aircraft certification by analysis.

Also this year, researchers attending the Future CFD Technologies Workshop agreed to recommend better leveraging of advances made in fundamental disciplines within the aerospace CFD community, and fostering better collaboration between relevant stakeholder government agencies including NASA, the Energy and the Defense Departments.

Contributors: Eric J. Nielsen, Nicholas J. Georgiadis and Dimitri J. Mavriplis

Image: Experiments to learn about cooling requirements for hot nozzle surfaces and turbine blades were run on this Turbulent Heat Flux test article at the Aeroacoustic Propulsion Laboratory at NASA’s Glenn Research Center in Ohio. Hot air exits from the nozzle on to the plate (the test article), which has 135 tiny holes through which cold air is provided from the plenum. Air velocity and temperature measurements were made above the test plate. Credit: NASA