CAE Large-scale CAE Simulation Supporting New Value-creation あらまし CAE Computer Aided Engineering LS-DYNA CAE CAE afjrls-dyna CAE Abstract In the manufacturing industry, numerical simulation assisted by computer-aided engineering (CAE) has become a widely available technology, being applied to various tasks in product development to enhance product quality for example. In this way, it is reducing time to market and costs. Previously, the computational environment and software quality did not meet the advanced requirements of high-precision, high-performance computation, and thus the application was limited to certain physical phenomena. However, the scope of application has significantly improved in recent years due to enhancements made in parallel computation, data-processing capacity, and software performance. LS-DYNA is widely employed in product development, and together with other CAE-based software, Fujitsu has been promoting their application to highly parallel computation technology, and assessing their performance. This paper presents a case of applying this large-scale CAE simulation through the afjr Project being implemented by the Japan Aerospace exploration Agency (JAXA). In this case, LS-DYNA was employed in simulating bird strikes on aircraft engine components. In this project, we gained know-how on high-parallel computation through evaluations on test models. Leveraging the know-how, we realized a large-scale CAE simulation for a practical application model, and succeeded in obtaining more accurate results than before. 94 FUJITSU. 67, 6, p. 94-100 11, 2016
まえがき CAE Computer Aided Engineering CAE CAE CAE CAE CAE CAE JAXA JAXA afjr Advanced Fan Jet CAE ソフトウェア LS-DYNA Research 1 CAE LS- DYNA 2 CAE CAE LS-DYNA FEM Finite Element Method 3 LSTC Livermore Software Technology Corporation One Code Strategy LS- DYNA LSTC MPP Massively Parallel Processing Hybrid LS- DYNA MPP Hybrid CAE MPP FUJITSU. 67, 6 11, 2016 95
Hybrid MPP MPP Hybrid LS-DYNA FUJITSU Supercomputer PRIMEHPC FX10 FX100 afjr プロジェクト afjr JAXA 図 -1 afjr CO 2 CAE 1,500 4 JAXA -1 JAXA afjr 96 FUJITSU. 67, 6 11, 2016
ファンブレード大規模 CAE シミュレーション afjr LS-DYNA CAE CAE LS-DYNA 大規模 CAEシミュレーションにおけるLS- DYNAの高速化施策図 -2 FEM SPH Smoothed Particle Hydrodynamics 5 FEM FE SPH SPH -2 LS-DYNA Hybrid 1 Hybrid LS-DYNA Hybrid MPP 表 -1 Hybrid 2 MPP 0.44 Hybrid 1.55 LS-DYNA FE MPP Hybrid 0ミリ秒 3ミリ秒人工鳥 薄型平板 5 ミリ秒 8 ミリ秒 -2 FUJITSU. 67, 6 11, 2016 97
SPH 4 FE 2,000 x 1,024 MPP 1,024 Hybrid 128 8 y 2,048 MPP 2,048 Hybrid 256 8 x / y 2-1 Hybrid MPP a MPP b Hybrid a / b 1,470.46 312.12 4.7 612.00 1.99 307.5 858.46 310.13 2.8 3,309.81 201.24 16.4 2,559.00 2.40 1,066.3 750.81 198.84 3.8 0.44 1.55 0.24 0.83 1.14 1.56 MPP 1,0241,024 Hybrid 128 1,024 /8 MPP Hybrid 8 1,024 2,048 2 Hybrid MPP Hybrid MPP Hybrid 2 Hybrid LS-DYNA 図 -3 Hybrid -3 a b 512 4,096 / 8 a b 512-3 a b 1,024 128 8 4,096 512 8-3 c a b SPH LS-DYNA FEMSPH SPH SPH SPH a a SPH b SPH 実用モデルの大規模 CAEシミュレーション結果 10 2 図 -4 a 98 FUJITSU. 67, 6 11, 2016
(a) デフォルト分割 (512 分割 ) (b) 人工鳥と平板をそれぞれ 512 分割 計算時間 ( 秒 ) 50,000 40,000 30,000 20,000 10,000 そのほか 接触計算 要素計算 0 分割 ( a) 128 8 並列 分割 ( b) 128 8 並列 分割 ( a) 512 8 並列 分割 ( b) 512 8 並列 (c) 計算時間の比較 -3 1.75 ミリ秒 3.25 ミリ秒 材料強軸方向のひずみ時刻歴 引張 小規模モデル大規模モデル実験値 0.0 圧縮 0.0 2.0 4.0 6.0 8.0 10.0 時刻 ( ミリ秒 ) 5.95 ミリ秒 8.75 ミリ秒 材料弱軸方向のひずみ時刻歴 引張 小規模モデル大規模モデル実験値 0.0 圧縮 0.0 2.0 4.0 6.0 8.0 10.0 時刻 ( ミリ秒 ) (a) 変形図 (b) ひずみ ( 物体の伸縮の比率 ) の時刻歴 データ提供 :( 株 )IHI -4 FUJITSU. 67, 6 11, 2016 99
b CAE -4 b CAE CAE afjr むすび CAE JAXA afjrls-dyna CAE JAXA afjr IHI JAXA 参考文献 1 JAXA afjr http://www.aero.jaxa.jp/research/ecat/afjr/ 2 Livermore Software Technology Corporation http://www.lstc.com/ 3 T. Belytschko et al Nonlinear Finite Elements for Continua and Structures Wiley 2000 4 2015 http://www.mlit.go.jp/koku/koku_tk15_000001.html 5 J. J. Monaghan Smoothed particle hydrodynamics Annual review of astronomy and astrophysics Vol.30 p.543-574 1992 著者紹介貞本将太 ( さだもとしょうた ) TC LS-DYNA 金堂剣史郎 ( こんどうけんしろう ) TC LS-DYNA 永濱由徳 ( ながはまよしのり ) TC LS-DYNA 100 FUJITSU. 67, 6 11, 2016