JAXA Supercomputer System Annual Report February 2025-January 2026

Report Number: R25EDA201G24

Subject Category: Aeronautical Technology

PDF (to be added)

• Responsible Representative: Kazuyuki Nakakita, Aviation Technology Directorate, Fundamental Technology Research Unit
• Contact Information: Andrea Sansica(sansica.andrea@jaxa.jp)
• Members: Kenji Hayashi, David Lusher, Yuya Ohmichi, Andrea Sansica

Abstract

Acceleration and testing of FaSTAR-GPU code for large scale simulations of full aircraft configuration at flight Reynolds number on GPUs using RANS & URANS, for off-design aerodynamics workshop contributions. Fundamental research of transonic airfoil buffet using the OpenSBLI DNS solver on GPUs is also performed for high-fidelity analysis.

Reference URL

N/A

Reasons and benefits of using JAXA Supercomputer System

Development and acceleration of GPU-enabled OpenACC version of FaSTAR in preparation for JSS4. GPU nodes are used to accelerate the runtime of large scale simulations for aircraft stall and buffet research applications.

Achievements of the Year

FaSTAR-GPU: The capability and application of the OpenACC accelerated version of FaSTAR were significantly upgraded in FY25. GPU support was added for all turbulence models and flux schemes, plus unsteady RANS capability was added. Excellent agreement was found in validation against the existing CPU code. The GPU code was used to perform CRM-HL stall calculations on very large meshes on over 1.5 billion cells (Figure 1) for JAXA off-design workshop contributions, and a wide parameter study was completed.

OpenSBLI (DNS): Large-scale high-fidelity simulations were performed on NASA-CRM wings (Figure 2), considering the effect of sweep angle on the 2D and 3D buffet modes at AR=3 for the first time. The relationship between sweep angle and buffet cell frequency was determined by modal analysis, with good agreement to previous low-fidelity simulations and experiment.

Fig.1: Full aircraft CRM-HL stall RANS simulations on very large meshes (up to 1.7 billion cells) using FaSTAR-GPU on TOKI-RURI JSS3 GPU nodes.

Fig.2: High-fidelity scale resolving wide-span transonic buffet simulations of NASA-CRM wings up to aspect ratio 3 with sweep effects, using the TOKI-RURI JSS3 GPU nodes.

Publications

- Non peer-reviewed papers

[1] D.J. Lusher, A. Sansica. Scale-resolving simulations and data-driven modal analysis of turbulent transonic buffet cells on infinite swept wings. arXiv preprint arXiv:2601.11137.

[2] D.J. Lusher, A. Sansica. The effect of sweep angle on shock-induced cellular separation bubbles in wide-span 3D turbulent transonic aerofoil buffet. The 35th International Symposium on Shock Waves (ISSW35) Brisbane, Australia, (July 2025).

[3] D.J. Lusher, A. Sansica, T. Matsuzaki. Performance and scaling of FaSTAR-GPU for CRM high-lift stall configurations on JSS3.第57回流体力学講演会／第43回航空宇宙数値シミュレーション技術シンポジウム (July 2025).

- Invited Presentations

[1] D.J. Lusher. OpenSBLI automatic code-generation framework for scale-resolving compressible fluid dynamics simulations. OPS/OpenSBLI/SENGA2 Training Workshop, Oxford, United Kingdom (September 2025).

- Oral Presentations

[1] D.J. Lusher, A. Sansica. The effect of sweep angle on shock-induced cellular separation bubbles in wide-span 3D turbulent transonic aerofoil buffet. The 35th International Symposium on Shock Waves (ISSW35) Brisbane, Australia, (July 2025).

[2] D.J. Lusher, A. Sansica, T. Matsuzaki. Performance and scaling of FaSTAR-GPU for CRM high-lift stall configurations on JSS3. 第57回流体力学講演会／第43回航空宇宙数値シミュレーション技術シンポジウム (July 2025).

[3] A. Sansica, D.J. Lusher, T. Matsuzaki. Contribution of JAXA to APC-10 using FaSTAR on the CRM high-lift configuration. 第57回流体力学講演会／第43回航空宇宙数値シミュレーション技術シンポジウム (July 2025).

- Poster Presentations

D.J. Lusher, A. Sansica. FaSTAR-GPU: OpenACC+MPI GPU acceleration of a Computational Fluid Dynamics (CFD) solver applied to full aircraft high-lift aerodynamics simulations on the JSS3 supercomputer. Supercomputing Conference (SC25)

Usage of JSS

Computational Information

• Process Parallelization Methods: GPU
• Thread Parallelization Methods: N/A
• Number of Processes: 120
• Elapsed Time per Case: 72 Hour(s)

JSS3 Resources Used

Fraction of Usage in Total Resources*1(%): 2.04

Details

Please refer to System Configuration of JSS3 for the system configuration and major specifications of JSS3.

*1: Fraction of Usage in Total Resources: Weighted average of three resource types (Computing, File System, and Archiver).

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.

ISV Software Licenses Used

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.

JAXA Supercomputer System Annual Report February 2025-January 2026