Multiphase thermal and turbulent flow simulation with fast distributed visualization in GPU cluster environment

2020 Fiscal Year Final Research Report

• Project/Area Number: 18K11323
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 60090:High performance computing-related
• Research Institution: University of Yamanashi
• Principal Investigator: Ando Hidetoshi 山梨大学, 大学院総合研究部, 准教授 (50221742)
• Co-Investigator(Kenkyū-buntansha): 鳥山 孝司 山梨大学, 大学院総合研究部, 准教授 (50313789)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Keywords: GPU / 熱流体計算 / 深層学習 / 可視化

Outline of Final Research Achievements

This study constructed a cluster environment using GPU (Graphics Processing Unit), which is inexpensive and has high computational performance per unit of power. It introduced turbulence analysis for solid-gas-liquid multiphase thermo-fluid analysis. In addition, we have developed a method for high-speed distributed visualization of calculation results. For thermal radiation calculations, which have been challenging to handle in conventional thermo-fluid analysis, we have extended the photon mapping method on GPUs to achieve fast and accurate thermal radiation calculations. In addition, by incorporating deep learning technology, we have achieved higher accuracy in numerical calculations and higher resolution images in visualization.

Free Research Field

高性能計算，可視化，機械学習，画像処理

Academic Significance and Societal Importance of the Research Achievements

熱流体解析は気象予測や空調設計，原子炉の冷却設計そのほか社会的に重要な様々な分野で必須の技術であり，この研究成果によりGPUという安価なハードウェアを用いて高精度な固気液多相の熱流体解析と可視化を可能にした．GPUは熱効率も高く環境への負荷も小さい計算機資源であり，本研究の成果がエネルギー資源を節約する効果も期待できる．ここで用いられた深層学習技術や高速数値計算技術も個別に様々な分野に広く応用可能であり，生命科学や医学分野での共同研究も進めており，学術的な意義も大きい．さらに個別の分野での学術的な発展の成果を本研究の手法に組み込むことも可能であり，今後の大きな相乗効果が期待できる．