Abstract: Metamodels, or models of models, map defined model inputs to defined model outputs. When metamodels are constructed to be computationally cheap, they are an invaluable tool for applications ranging from topology optimization, to uncertainty quantification, to real-time prediction, to multi-scale simulation. In particular, for heterogeneous materials, metamodels are useful for exploring the influence of the (potentially massive) heterogeneous material property parameter space. By nature, a given metamodel will be tailored to a specific dataset. However, the most pragmatic metamodel type and structure will often be general to larger classes of problems. At present, the most pragmatic metamodel selection for dealing with mechanical data — specifically simulations of heterogenous materials — has not been thoroughly explored. In this work, we draw inspiration from the benchmark datasets available to the computer vision research community. These benchmark datasets have both made it feasible to compare different methods for solving the same problem, and inspired new directions for method development. In response, we introduce benchmark datasets for engineering mechanics problems (for example, the Mechanical MNIST Collection https://open.bu.edu/handle/2144/39371 [1,2,3, 4]). Then, we show some example problems that we are exploring with these datasets such as our methodology for constructing metamodels for predicting full field quantities of interest (e.g., full field displacements, stress, strain, or damage variable), for leveraging information from multiple simulation fidelities, and for creating well calibrated models. Looking forward, we anticipate that disseminating both these benchmark datasets and our computational methods will enable the broader community of researchers to develop improved techniques for understanding the behavior of spatially heterogeneous materials. We also hope to inspire others to use our datasets for educational and research purposes, and to disseminate datasets and metamodels specific to their own areas of interest (https://elejeune11.github.io/).

Biographical Sketch: Emma Lejeune is an Assistant Professor in the Mechanical Engineering Department at Boston University. She received her PhD from Stanford University in September 2018, and was a Peter O’Donnell, Jr. postdoctoral research fellow at the Oden Institute at the University of Texas at Austin until 2020 when she joined the faculty at BU. At BU, Emma has received the David R. Dalton Career Development Professorship, a Computational Science and Engineering Junior Faculty Fellowship, the Haythornthwaite Research Initiation Grant from the ASME Applied Mechanics Division, and the American Heart Association Career Development Award. Current areas of research involve integrating data-driven and physics based computational models, and characterizing and predicting the mechanical behavior of heterogeneous materials and biological systems.

Abstract: Sickle cells are known for their significantly shortened lifespan (10-20 days), which is much shorter than the lifespan (~120 days) of the normal red blood cells (RBCs). Similar to normal RBCs, sickle cells are also challenged by repeated hypoxia cycles as well as mechanical fatigue. To examine the impact of these repeated challenges toward the progressive degradation process of RBCs, we have developed in vitro microfluidic assays for testing RBCs in health and disease under cyclic hypoxia loading or cyclic mechanical loading. Both types of fatigue loading are found to cause significant RBC degradation in a cumulative manner. More importantly, our results show that sickle cells on average degrade much faster than normal healthy RBCs. These results provide new insights into the possible mechanisms underlying the significantly shortened lifespan of sickle cells. The developed assays can be used for drug efficacy screening and potentially disease severity testing in a patient-specific manner.

Biographical Sketch: Ming Dao is the Principal Investigator and Director of MIT’s Nanomechanics Laboratory, and a Principal Research Scientist in the Department of Materials Science and Engineering at MIT. His research interests include nanomechanics of advanced materials, cell biomechanics/biophysics of human diseases, and machine learning for engineering and biomedical applications. He has published over 160 papers in peer-reviewed journals, including Science, Nature Materials, Science Advances, Nature Communications, PNAS, etc. He was ranked within the Top 2% Scientists list established by Ioannidis/Stanford University in all four updates published in June 2019 (single year), October 2020 (single year & career), October 2021 (single year & career), and November 2022 (single year & career). He is also ranked as a top 0.5% researcher in both citation and h-index by Exaly.com (March 2023).

He is a Fellow of the American Society of Mechanical Engineers (ASME) and named the 2012 Singapore Research Chair / Professor in Bioengineering and Infectious Disease by MIT. He was a visiting professor with the National Institute of Blood Transfusion, Paris, France (INTS, 2016-2017) and an adjunct professor with Xi’an Jiaotong University, Xi’an, China (2011-2020). Since 2018, he has been a visiting professor at Nanyang Technological University, Singapore. He has also chaired or co-chaired 18 international symposiums/workshops/webinar series.

Abstract: Topology optimization has emerged as a promising and powerful approach to design engineered materials and components. Initially restricted to two-phase, solid-void design problems in linear elasticity, topology optimization approaches for multi-physics and multi-material problems have emerged. These problems are often dominated by interface phenomena, such as contact and delamination at material interfaces and boundary layer effects at fluid-solid interfaces. Accurately modeling these phenomena and, at the same time, allowing for topological changes in the optimization process pose interesting challenges on the formulation of the design optimization problem, the physics model, and the discretization method.

This talk will provide an overview of topology optimization approaches for problems, reviewing both density and level set topology optimization methods. This overview will show that level set methods combined with immersed finite element approaches provide a promising framework, especially for coupled multi-physics and multi-material topology optimization problems. The accuracy, robustness, and accuracy of the finite element analysis play a crucial role for such problems. This talk will present an isogeometric formulation of the eXtended Finite Element Method where the level set and state variables fields are discretized on adaptively refined meshes, using truncated hierarchical B-splines. Using approximate Lagrange extraction, this formulation can be integrated in standard finite element solvers.

The characteristics of this XFEM analysis and level set topology optimization framework will be illustrated with 2D and 3D problems in solid and fluid mechanics, including elastic, flow, and conjugate heat transfer problems.

Biographical Sketch: Dr. Maute is the Palmer Endowed Chair and a professor in the Ann and H.J. Smead Aerospace Engineering Sciences Department at the University of Colorado Boulder. Dr. Maute received a Bs/Ms. in Aerospace Engineering in 1992 and Ph.D. in Civil Engineering in 1998, both from the University of Stuttgart, Germany. After working as a postdoctoral research associate at the Center for Aerospace Structures, he started his faculty position at CUB in 2000. His research is concerned with computational mechanics and design optimization methods. He focuses on fundamental problems in solid and fluid mechanics and heat transfer with applications to aerospace, civil, mechanical engineering problems. For the past 30 years, Dr. Maute has worked on topology and shape optimization methods for a broad range of problems focusing on coupled multi-physics and multi-scale problems, such as fluid-structure interaction and chemo-mechanically coupling. Dr. Maute has published his work in over 200 journal articles, book chapters, and conference proceedings.

Abstract:

The emerging economic and societal needs such as advanced manufacturing, environmental treatment, and space exploration call for machines that can operate in harsh and complex environments. An attractive approach is to utilize a new paradigm of physical intelligence in material development: a rational material design will enable its on-board actuation, sensing, and analysis, without a need for central computing or complex control. This talk will present our recent progress in the fundamental research of embedding physical intelligence in soft active materials. A stretchable polymer network responds to an external stimulus such as light or heat, dramatically changes its shape or material property, and enables special functionality in its bulk or surface. The first part of the talk presents photoactive liquid crystal elastomers that can change their shape and generate work output under light illumination or temperature change. Emphasis is placed on the fundamental photo-thermo-mechanical coupling across many length scales, especially at the mesoscale where the polymer network and liquid crystal mesogens behave collectively, leading to multiple interesting phenomena and their consequences in the macroscopic actuation. The second part of the talk presents temperature-switchable adhesives with high adhesion strength, large switching ratio, fast switching speed, and good reversibility. A polymer network containing many free-end dangling chains is a strong adhesive at ambient environment due to the long chains, dense physical bonds, and large dissipation from the polymer matrix, and is completely non-adhering at an elevated temperature due to its thermo-responsive phase transition. This talk is hoped to help advance the fundamental knowledge of soft active materials, bring together communities of relevant research fields, and expand the potential large-scale applications.

Bio:

Ruobing Bai is an assistant professor in the Department of Mechanical and Industrial Engineering at Northeastern University. He received his BS in Theoretical and Applied Mechanics at Peking University in 2012, and PhD in Engineering Sciences at Harvard University in 2018. He was a postdoctoral fellow in the Department of Mechanical and Civil Engineering at California Institute of Technology from 2018 to 2020. He is the recipient of the Chun-Tsung Scholar in Peking University, the Haythornthwaite Research Initiation Award from the Applied Mechanics Division of American Society of Mechanical Engineers (ASME), and the Extreme Mechanics Letters (EML) Young Investigator Award. Research in the Bai group aims to combine theory and experiment in areas including solid mechanics, soft active materials, fracture and toughening of materials, adhesion, and sustainable materials, for applications such as soft robotics, advanced manufacturing, human-machine interfaces, and human health.

Abstract: The properties of materials are highly dependent on their structures, which include morphologies, grain boundaries, phases, atomic structures, Etc. In particular, the atomic structures determine the limit of their properties, which are the unique characteristics of each material. However, predicting physical properties and understanding their origins demand accurate electronic structure and atomic structure calculations, which require a lot of computational resources. Over the past few decades, computational power has been tremendously improved, enabling atomistic simulations on a reasonable length and time scale to answer such questions. Accordingly, high-throughput calculations, data mining, and machine-learning(ML) solutions have been becoming mainstream in materials research.

In this talk, we introduce our effort to understand the atomic structure and materials property relationship to practical application in the next-generation battery and semiconductor materials using atomistic simulation tools such as; ab initio Density Functional Theory, Classical Molecular Dynamics, and ML algorithms. First, we present newly developed active materials for next-generation rechargeable battery applications, which include novel cathode and electrolyte materials. And then, we exhibit the structure-property relationship to describe dielectric properties using the ML algorithm and intuition of fundamental physics. These examples will sufficiently show atomistic simulation’s practical application to materials research.

Biographical Sketch: Dr. Shin is a Sr. Staff Engineer and Project Leader in the Advanced Materials Lab at Samsung Semiconductor Inc (SSI). He studies theoretical and computational materials science through computational modeling, simulation, and Artificial-intelligence driven materials discoveries for energy harvesting, conversion, and storage materials.

Dr. Shin received his Ph.D. in Materials Science and Engineering at Boston University in 2012 and held a Chemistry Postdoc Fellow position in the Energy Storage and Distributed Resources Division at Lawrence Berkeley National Laboratory. He worked as Research Engineer at Samsung Research America before joining SSI.

Abstract: The human brain undergoes a myriad of changes during its lifetime. From a mechanics perspective alone, it is mesmerizing how the brain develops during early life, transforms into this highly functional, albeit still very enigmatic, organ that makes us unique, and is subjected to injury, disease, and ultimately age-related degeneration like every other part of the body. Despite extensive efforts to mechanically characterize brain tissue for more than two decades, the relationship between microstructure, state of health, and mechanical behavior remains elusive. On the modeling side, the computational biomechanics community has had extensive interest in modeling traumatic brain injury, neurodegeneration, stroke, surgical guidance, and, the most intensely studied, brain folding during early development. Our group’s motivation to pursue multiphysics modeling of the brain is simple: while the biology of brain aging and many neurological diseases is very well established, its coupling to the brain’s mechanical response in the form of cerebral atrophy and tissue degeneration/damage remains understudied.

In the present talk, we will explore our work on inferring the growth field during brain development, modeling brain shape changes during Alzheimer’s disease, and the mechanical origin of white matter degeneration during brain aging.

 

Biographical Sketch: Johannes Weickenmeier is an assistant professor of mechanical engineering and the director of the Center for Neuromechanics at Stevens Institute of Technology. Dr. Weickenmeier leads the Soft Matter Biomechanics Laboratory that combines medical image analysis, mechanical testing, and numerical methods to understand and predict soft tissue behavior. His group’s current work focuses on understanding and developing physics-based models that describe brain changes during develop, healthy aging, in Alzheimer’s disease, and multiple sclerosis. For more information, go to www.weickenmeierlab.com or follow his group on Twitter @weickenmeierlab.

Abstract: The behavior of materials involve physics at multiple length and time scales: electronic, atomistic, domains, defects, etc.  The engineering properties that we observe and exploit in application are a sum total of all these interactions.  Multiscale modeling seeks to understand this complexity with a divide and conquer approach.  It introduces an ordered hierarchy of scales, and postulates that the interaction is pairwise within this hierarchy.  The coarser-scale controls the finer-scale and filters the details of the finer scale.   Still, the practical implementation of this approach is computationally challenging.  This talk introduces the notion of neural operators as controlled approximations of operators mapping one function space to another and explains how they can be used for multiscale modeling.  They lead to extremely high-fidelity models that capture all the details of the small scale but can be directly implemented at the coarse scale in a computationally efficient manner.  We demonstrate the ideas with examples drawn from first principles study of defects and crystal plasticity study of inelastic impact.

Biographical Sketch: Kaushik Bhattacharya is Howell N. Tyson, Sr., Professor of Mechanics and Professor of Materials Science as well as the Vice-Provost at the California Institute of Technology. He received his B.Tech degree from the Indian Institute of Technology, Madras, India in 1986, his Ph.D. from the University of Minnesota in 1991, and his post-doctoral training at the Courant Institute for Mathematical Sciences during 1991-1993. He joined Caltech in 1993. His research concerns the mechanical behavior of materials, and specifically uses theory to guide the development of new materials. He has received the von Kármán Medal of the Society of Industrial and Applied Mathematics (2020), Distinguished Alumni Award of the Indian Institute of Technology, Madras (2019), the Outstanding Achievement Award of the University of Minnesota (2018), the Warner T. Koiter Medal of the American Society of Mechanical Engineering (2015) and the Graduate Student Council Teaching and Mentoring Award at Caltech (2013).  He served as the editor of the Journal of Mechanics and Physics of Solids during 2004-2015.

Abstract 

In numerical modeling of planetary and stellar convection, taking into account compressibility effects is crucial. However, using the exact equations may not be feasible due to the generation of fast acoustic waves, which distract from the slower convective motions caused by buoyancy. The Oberbeck-Boussinesq model simplifies the calculations by suppressing the acoustic waves making it easier for numerical simulations, but is so simple and pressure effects are relegated to a secondary role. Intermediate models, such as the anelastic and anelastic liquid models, have also been proposed to balance simplicity and accuracy. 

We investigated compressible convection under several different approximations for the thermodynamic state as well as using the exact equations. We tested two different classes of equations of state (EoS): one where entropy depends only on density, resulting in nearly constant density and minimizing non-Oberbeck-Boussinesq effects, and the Birch-Murnaghan equations of state, which are realistic models for condensed matter like the Earth’s mantle and core.  Our study showed that dissipation is closely linked to the fraction of heat flow carried by entropy flux. Additionally, we observe that small-scale convection is prevalent in the flow structure. Our results are mostly discussed in the framework of mantle convection, but the EoS is flexible enough to be applied in the inner core or in icy planets. 

 

Bio 

Jezabel Curbelo is a Ramon y Cajal Research Fellow at Barcelona School of Industrial Engineering at Universitat Politècnica de Catalunya, and currently visiting the Department of Earth and Planetary Sciences at Harvard University. She has previously held positions at various universities, including the Department of Atmospheric and Oceanic Sciences at UCLA and the Laboratoire de Géologie de Lyon. Her PhD thesis (Universidad Autónoma de Madrid, 2014) was awarded with the “2015 Donald L. Turcotte Award” (American Geophysical Union). She has received several awards for her research in geophysical fluid dynamics including the ”Leonardo Fellowships 2022” (BBVA Foundation) and the ”2021 L’Oréal-UNESCO For Women in Science” award (L’Oréal Spain). Her research focuses on the simulation and modeling of nonlinear fluid processes in the ocean and atmosphere and the analysis of convective motions in planetary mantles. Her webpage is web.mat.upc.edu/jezabel.curbelo/.

Abstract:

Turbulence is ubiquitous in natural and engineering systems. It can suppress energy loss in fusion reactors, affects stellar formation, has first order effects on processes critically important to society such as mixing of chemicals and pollutants in the atmosphere and oceans, climate dynamics and high-speed flight. It is therefore critically important to develop fundamental understanding of turbulent processes to improve predictive capabilities of turbulent fluid systems. An important hurdle in characterizing turbulent flows is the presence of extreme events, e.g. in dissipation, velocity gradients etc. These events are often very high-dimensional in nature and require large degrees of freedom/grid points to resolve accurately in simulations. The extreme events become stronger at high Reynolds number (Re, parameter characterizing the strength of turbulence) that are characteristic of realistic flows. Therefore, the focus of much of turbulence research has been to simulate very high-Re flows. This is a challenging computational task as the computational work load can grow as steeply as the fourth power of Reynolds number. Direct simulations of complex turbulent flows at realistic conditions currently remain elusive on the largest supercomputers. In this talk, we present a new theoretical perspective on understanding high-Re turbulence using well-resolved simulations at low to moderate-Re, that can be simulated on supercomputers available today. We will show that features of high-Re turbulence can be studied at finite and small values of Re and they are predictive of the infinite-Re limit. The simulations have the finest small-scale resolution in literature and long time-series. A primary focus is on the universality of small-scales and the scaling of extreme events. The consequences of these fundamental insights on modeling approaches, phenomenological and data-driven, in complex turbulent flows will also be discussed. The work also provides a new perspective on computational study of complex systems at very high values of dynamically relevant parameters. 

 Bio:

Sualeh Khurshid is a Computing Innovation Fellow and Postdoctoral Associate in Mechanical Engineering at Massachusetts Institute of Technology. His research is focused on understanding fundamental characteristics of complex turbulent flows in various regimes using direct numerical simulations and theory. His work includes developing high performance simulation codes and appropriate numerical methods to guide the development of reduced order models using phenomenological and data-driven methods. He completed undergraduate programs in Aerospace Engineering and Physics in 2016 and earned his Ph.D. in Aerospace Engineering in 2021 at Texas A&M University.