Begell House Inc.
International Journal for Multiscale Computational Engineering
JMC
1543-1649
18
4
2020
ATOMISTICALLY INFORMED TEMPERATURE AND RATE-DEPENDENT MECHANICAL RESPONSE OF FUSED SILICA
409-420
10.1615/IntJMultCompEng.2020035707
Yang
Yu
Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, NY 10027, USA
Yang
Jiao
GE Global Research, Niskayuna, New York, 12309, USA
Jacob
Fish
Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, New York 10027, USA
amorphous glass
thermal effect
strain rate
yield criteria
critical state
machine learning
This manuscript studies the mechanical response of amorphous silica based on data mining from molecular dynamics simulations. The temperature- and densification-dependent yield criteria have been established. The proposed modified Drucker-Prager-cap yield criterion adequately captures the temperature effect on the initial yield surface. The main focus of this study has been on understanding the critical state, which defines the onset of an inhomogeneous plastic deformation. The machine learning-based nonparametric regression method has been employed to describe complex shear critical state dependence on temperature, pressure, and strain rate.
MULTI-LEVEL K-d TREE-BASED DATA-DRIVEN COMPUTATIONAL METHOD FOR THE DYNAMIC ANALYSIS OF MULTI-MATERIAL STRUCTURES
421-438
10.1615/IntJMultCompEng.2020035167
Zhangcheng
Zheng
International Research Center for Computational Mechanics, State Key Laboratory of
Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, Faculty
of Vehicle Engineering and Mechanics, Dalian University of Technology, Dalian 116024,
People's Republic of China
Hongfei
Ye
International Research Center for Computational Mechanics, State Key Laboratory of
Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, Faculty
of Vehicle Engineering and Mechanics, Dalian University of Technology, Dalian 116024,
People's Republic of China
Hongwu
Zhang
International Research Center for Computational Mechanics, State Key Laboratory of
Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, Faculty
of Vehicle Engineering and Mechanics, Dalian University of Technology, Dalian 116024,
People's Republic of China
Yonggang
Zheng
International Research Center for Computational Mechanics, State Key Laboratory of
Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, Faculty
of Vehicle Engineering and Mechanics, Dalian University of Technology, Dalian 116024,
People's Republic of China
Zhen
Chen
International Research Center for Computational Mechanics, State Key Laboratory of
Structural Analysis for Industrial Equipment, Department of Engineering Mechanics, Faculty
of Vehicle Engineering and Mechanics, Dalian University of Technology, Dalian 116024, China; Department of Civil and Environmental Engineering, University of Missouri, Columbia, MO 65211, USA
data-driven solver
multi-material database
data addition method
multi-level K-d tree
dynamic response
The model-free distance-minimizing data-driven computational method has recently become a novel paradigm for solving various mechanics problems. However, the paradigm may suffer from low efficiency since tremendous iterative searches of key data points in the material dataset are needed during the solution process. A fast data-driven solver is therefore proposed here for the accurate and efficient analysis of multi-material structural responses to dynamic loading. In the proposed approach, a multi-material database (MMD) with different kinds of constituents is constructed, and a multi-level K-d tree (MKT) is developed for effective data addition and fast data search in the MMD. An efficient data-driven dynamics solver (DDDS) is then designed based on the MMD/MKT, which can deal with the complicated dynamic analysis of different structures containing multiple material datasets. Representative types of dynamic problems are considered to verify and demonstrate the capability of the proposed approach. Numerical results demonstrate that the MMD/MKT and the corresponding DDDS possess high accuracy and efficiency, which might be further developed for the dynamic analysis of composite structures containing constituents at different scales.
A COUPLING OF MULTISCALE FINITE ELEMENT METHOD AND ISOGEOMETRIC ANALYSIS
439-454
10.1615/IntJMultCompEng.2020034287
Mateusz
Dryzek
Chair for Computational Engineering, Cracow University of Technology, ul. Warszawska 24,
31-155 Krakow, Poland
Witold
Cecot
Institute for Computational Civil Engineering, Cracow University of Technology, ul. Warszawska 24, 31-155 Krakow, Poland
multiscale finite element method
higher-order shape functions
B-splines
In this paper, we propose to use modified B-splines spanned on several macroelements as a basis for building the multiscale finite element method (MsFEM) trail functions. The main benefit of our approach is that the calculations of a multiscale function are done in one step on the whole support, in contrast to standard MsFEM shape functions that are evaluated coarse element by element and require a cumbersome gluing. Selected numerical experiments for flow in porous media with periodic and random material properties distributions were performed to test our modified MsFEM with the new basis functions. We found that the method indeed improves standard MsFEM for fast oscillating material properties. We observed that the resonance effect, when the ratio of inclusion size and coarse mesh size approaches 1 (ε/H → 1) can be reduced by increasing the order of B-splines.
A TWO-SCALE APPROACH FOR THE DROP SHOCK SIMULATION OF A PRINTED CIRCUIT BOARD PACKAGE CONSIDERING REFLOWED SOLDER BALL GEOMETRIES
455-476
10.1615/IntJMultCompEng.2020035631
Wei
Hu
Computational and Multiscale Mechanics Group, Livermore Software Technology, An Ansys
Company, Livermore, CA, USA
Xiaofei
Pan
Computational and Multiscale Mechanics Group, Livermore Software Technology, An Ansys
Company, Livermore, CA, USA
Dandan
Lyu
Computational and Multiscale Mechanics Group, Livermore Software Technology, An Ansys
Company, Livermore, CA, USA
Ashutosh
Srivastava
Ansys Inc., Canonsburg, PA, USA
Siddharth
Shah
Ansys Inc., Canonsburg, PA, USA
Cheng-Tang
Wu
Computational and Multiscale Mechanics Group, Livermore Software Technology, An Ansys
Company, Livermore, CA, USA
particle method
solder ball
Navier-Stokes
co-simulation
multiscale model
In this paper, we introduce a new computational approach for linking the information of mesoscale solder ball shapes to the macroscale drop test of a printed circuit board. The approach starts with a numerical prediction of mesoscale solder bump profiles using a novel full-implicit Lagrangian particle method to approximate the Navier-Stokes equations and efficiently simulate the incompressible free surface reflow soldering process. The surface tension of the molten solder and the wall adhesion between the solder and the substrate are considered in the simulation. Subsequently, the predicted solder ball shapes from the reflow analysis are used in a chip package model for the drop shock analysis. The mesoscale solder joint model is coupled concurrently with the macroscale chip package model using the co-simulation to achieve the nonintrusive scale-bridging effect. To attend practical explicit-explicit two-scale co-simulations, an algorithm that handles properly the load-balancing, heterogeneity of processors, and memory also has been developed. We present four numerical examples that showcase the effectiveness of the new approach.
COMPUTATIONAL ANALYSES OF FLEXURAL BEHAVIOR FOR ULTRAHIGH PERFORMANCE FIBER REINFORCED CONCRETE BRIDGE DECKS
477-491
10.1615/IntJMultCompEng.2020035500
Yan
Wang
Key Laboratory for Wind and Bridge Engineering of Hunan Province, College of Civil
Engineering, Hunan University, Changsha 410082, China; Department of Civil Engineering and Engineering Mechanics, Columbia University, 610
Seeley W. Mudd Building, 500 West 120th Street, Mail Code 4709, New York, 10027, New
York, USA
Timothy
Artz
Department of Civil Engineering and Engineering Mechanics, Columbia University, NewYork, New York 10027, USA
Andrew
Beel
Department of Civil Engineering and Engineering Mechanics, Columbia University, 610
Seeley W. Mudd Building, 500 West 120th Street, Mail Code 4709, New York, 10027, New
York, USA
Xudong
Shao
Key Laboratory for Wind and Bridge Engineering of Hunan Province, College of Civil
Engineering, Hunan University, Changsha 410082, China
Jacob
Fish
Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, New York 10027, USA
multiscale
ultrahigh performance fiber-reinforced concrete (UHPFRC)
post-cracking flexural performance
overload damage assessment
The manuscript describes a multiscale paradigm for predicting post-cracking flexural behavior in ultrahigh performance fiber reinforced concrete (UHPFRC) structures. A comparative study was conducted to simulate the flexural behavior of reinforced UHPFRC beams used in the Malukou Bridge, China. In this study, UHPFRC was modeled as a heterogeneous composite medium made of a matrix and steel fibers using a reduced-order multiscale approach, and for comparison purposes, using a conventional single-scale homogeneous isotropic medium. The multiscale approach has been shown to be considerably more accurate in predicting the post-cracking flexural behavior of the reinforced UHPFRC deck than the conventional single-scale approach.
RANDOM WALK-BASED STOCHASTIC MODELING OF DIFFUSION IN SPHERICAL AND ELLIPSOIDAL COMPOSITES
493-505
10.1615/IntJMultCompEng.2020033217
Jian
Qiu
Department of Mechanical & Materials Engineering, University of Denver, Denver, Colorado
80208, USA
Jide
Williams
Department of Mechanical & Materials Engineering, University of Denver, Denver, Colorado
80208, USA
Yun-Bo
Yi
Department of Mechanical and Materials Engineering, University of Denver, Denver, Colorado 80208, USA
random walk
heterogeneous materials
diffusion
ellipsoids
Monte Carlo method
finite element analysis
Diffusion in randomly dispersed, spherical, and ellipsoidal composite systems is studied using the random walk simulations. The outcome of the computational analysis is validated by finite element analyses. A Monte Carlo scheme is applied to generate the particulate system. The composite is assumed to have a lower diffusivity in the inclusions and a higher diffusivity in the matrix. The effective diffusion coefficient is found to agree with the theory relating volume fractions of permeable and impermeable inclusions to the diffusion coefficient. The effect of the particle aspect ratio is investigated numerically and compared with the closed-form, effective medium solutions. In the case of ellipsoidal inclusions, it is found that the effective diffusion coefficient is strongly dependent on the particle aspect ratio and that it rapidly decreases with the volume fraction of inclusions. The interfacial effect in the setting of anomalous diffusion for permeable systems is also tentatively investigated.