出版者・発行元：ACM International Conference Proceeding Series  

This paper presents a novel numerical method for the elastoplastic simulation of vehicle component structures under large deformation problems, such as crash-worthiness analysis. Elastoplastic simulation of vehicle structures is essential for designing safer and more efficient vehicles, but poses significant challenges for a conventional finite element method (FEM) due to the limitation of parallel performance and mesh distortion at large deformations. To overcome these challenges, we propose using the cell-centered finite-volume method (CCFVM) in the Eulerian description and the building-cube method (BCM). The Eulerian CCFVM is a robust scheme for solving large deformations of the continuum by spatially fixed collocated Cartesian mesh, which enables a simpler data structure than conventional FEM. The BCM is a hierarchical Cartesian mesh approach in which the computational domain is divided into cubic regions with adaptive refinement suitable for parallel computing. The combination of Eulerian CCFVM and BCM enables us to perform high-fidelity simulations of vehicle component structures. We demonstrate the validity and scalability of the present method by computing a three-point bending test of vehicle component structures using different geometries, such as impact beams. The numerical results show good agreement with the reference results obtained by LS-DYNA, a commercial FEM code. We also achieve a weak scaling of 73.5% in an elastoplastic simulation of about 540 million cell mesh using 131,072 cores of supercomputer Fugaku. The proposed method is expected to contribute to the development of geometrically complex crushable structures by giga-casting and 3D printing, which are emerging manufacturing techniques for vehicle structures.

DOI： 10.1145/3635035.3635038

Scopus

出版者・発行元：ACM International Conference Proceeding Series  

This paper presents a novel numerical method for the elastoplastic simulation of vehicle component structures under large deformation problems, such as crash-worthiness analysis. Elastoplastic simulation of vehicle structures is essential for designing safer and more efficient vehicles, but poses significant challenges for a conventional finite element method (FEM) due to the limitation of parallel performance and mesh distortion at large deformations. To overcome these challenges, we propose using the cell-centered finite-volume method (CCFVM) in the Eulerian description and the building-cube method (BCM). The Eulerian CCFVM is a robust scheme for solving large deformations of the continuum by spatially fixed collocated Cartesian mesh, which enables a simpler data structure than conventional FEM. The BCM is a hierarchical Cartesian mesh approach in which the computational domain is divided into cubic regions with adaptive refinement suitable for parallel computing. The combination of Eulerian CCFVM and BCM enables us to perform high-fidelity simulations of vehicle component structures. We demonstrate the validity and scalability of the present method by computing a three-point bending test of vehicle component structures using different geometries, such as impact beams. The numerical results show good agreement with the reference results obtained by LS-DYNA, a commercial FEM code. We also achieve a weak scaling of 73.5% in an elastoplastic simulation of about 540 million cell mesh using 131,072 cores of supercomputer Fugaku. The proposed method is expected to contribute to the development of geometrically complex crushable structures by giga-casting and 3D printing, which are emerging manufacturing techniques for vehicle structures.

DOI： 10.1145/3635035.3635038

Scopus

記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：Structural and Multidisciplinary Optimization  

Advances in 3D-printing technology have made it possible to manufacture structures containing periodic microstructures (lattice structures) and obtain superior mechanical performance that cannot be provided with uniform solids. To achieve the best performance, microstructures with different properties should be optimally designed and arranged in the right places within the macrostructure. However, in most of the basic approaches with homogenization theory, each microstructure is analyzed and optimized separately under periodic boundaries, so connectivity with adjacent microstructures is not considered (or guaranteed). Consequently, unrealistic discontinuous structures are often obtained, resulting in poor mechanical performance and manufacturability. To address this problem, we propose a method to introduce an "enlarged connective domain" around the boundary when dividing the macrostructure into multiple domains with different microstructures in advance. The microstructure of the enlarged connective domain is formed by combining the unit cells of adjacent microstructures, which allows micro-scale connectivity to be reflected in the macrostructure. Therefore, this method improves the mechanical connectivity between microstructures through the original optimization process of maximizing the performance of the entire macrostructure. This paper demonstrates that the proposed method improves connectivity and mechanical performance by performing optimization analyses for two- and three-dimensional problems. In addition, we focus on the width (volume ratio) of the connective domain as a design parameter that specifies the degree to which connectivity is considered and discuss how to set the parameter appropriately.

DOI： 10.1007/s00158-023-03613-w

Web of Science

Scopus

記述言語：日本語   出版者・発行元：一般社団法人 日本計算工学会  

<p>本研究では，固体材料と材料界面の両方の強度を考慮したマルチマテリアルトポロジー最適化手法を提案する．ヤング率と強度の複数の材料物性を同時に考慮するために，係数が等価に与えられるDMO材料内挿関数を適用する．応力制約付きトポロジー最適化の特異点問題を緩和するために，DMO材料内挿関数に適応した応力定義を提案する．さらに，材料界面を密度法ベースの手法で表現するために，2段階の密度フィルタリングを用いたgraded interfaceと呼ばれる手法を応用的に導入する．界面に引張/圧縮非対称応力規準を適用することにより，材料界面の引張の負担が軽減されるような最適化計算結果を得ることが可能となる．本研究で提案した手法の妥当性について，3つの構造モデルで検証を行う．</p>

DOI： 10.11421/jsces.2023.20230001

CiNii Research

記述言語：日本語   出版者・発行元：公益社団法人 土木学会  

<p>近年，積層造形技術の発展により，構造物の形状（マクロ構造）だけでなく材料の微視構造（ミクロ構造）を制御し，所望の特性を持つ材料を製造することが可能になりつつある．そのため，最適なミクロ構造を設計するマルチスケールトポロジー最適化が注目を集めているが，これを実設計に適用するには課題が多い．特に，ミクロ構造解析の高い計算コストは，高解像度での解析や非線形問題への拡張の妨げとなっている．本研究は，これを解決するため，高速フーリエ変換を用いた均質化手法に着目し，より高速で省メモリな新しい最適化手法を開発する．線形弾性材料を想定した剛性最大化問題を対象に，従来手法との比較に基づく性能検証を行い，提案手法の妥当性を確認するとともに，計算コスト削減への効果について検証する．</p>

DOI： 10.2208/jscejj.22-15026

CiNii Research

記述言語：日本語  

<p>Infill structure - which consists of porous and solid phases over the structures, is getting a lot of attention due to its multi-functional performance, such as light weight, high-thermal convention and the potential redundancy. Nowadays, the manufacturing for this kind of geometrically complicated structure becomes possible by the recent development of Additive Manufacturing technology and furthermore topology optimization is often applied for the design to improve those structural performances. In infill topology optimization, too slender members often remain as a result of the severe density local constraints: this leads to undesirable local failure in the members even under small deformation. In this research, we propose a filtering method to avoid those thin members and local failure by introducing a variable length radius.</p>

DOI： 10.1299/jsmecmd.2018.31.312

記述言語：日本語  

DOI： 10.1299/jsmeoptis.2018.13.113

記述言語：日本語  

<p>The present paper proposes a topology optimization method considering finite deformation for loading uncertainty. The loading angle is assumed to be uncertain as a condition. The objective is to minimize mean and standard deviation of structural compliance. In case of finite deformation theory, an analytical estimation of the mean and the standard deviation is not allowed. In order to solve this problem, we approximate the objective function by a Taylor series expansion and derive the mathematical formulation. In this approach, the second derivative of the objective function is necessary to keep the accuracy in sensitivity. This phenomenon is investigated in terms of numerical validations. Finally, some numerical examples demonstrate the usefulness of the proposed method.</p>

DOI： 10.1299/jsmecmd.2018.31.277

記述言語：日本語  

<p>This study deals with the ideal topological design of vibrating structures. For static structures, well-known compliance minimization is defined as the inner product of the external force and static displacement. However, for dynamic structures with time-varying forces and displacements, defining a measure for dynamic stiffness are required. Therefore, various objective functions have been proposed to suppress the response of vibrating structures. In this study, these various objective functions are compared from the viewpoint of vibration suppression. The examples of numerical calculation when adopting mean strain energy and mean squared dynamic compliance as objective functions are presented.</p>

DOI： 10.1299/jsmecmd.2018.31.291

開催年月日： 2023年12月

記述言語：英語   会議種別：口頭発表（招待・特別）  

開催地：Aachen   国名：ドイツ連邦共和国  

開催年月日： 2023年9月

記述言語：英語   会議種別：口頭発表（一般）  

開催地：Barcelona   国名：スペイン  

開催年月日： 2023年7月

記述言語：英語   会議種別：口頭発表（一般）  

開催地：Albuquerque, New Mexico   国名：アメリカ合衆国  

開催年月日： 2023年5月 - 2023年6月

記述言語：日本語   会議種別：口頭発表（一般）  

開催地：つくば   国名：日本国  

開催年月日： 2021年7月

記述言語：英語   会議種別：口頭発表（一般）  

国名：アメリカ合衆国