Bogdan I. Epureanu

By |

• Computational dynamics focused on nonlinear dynamics and finite elements (e.g., a new approach for forecasting bifurcations/tipping points in aeroelastic and ecological systems, new finite element methods for thin walled beams that leads to novel reduced order models).
• Modeling nonlinear phenomena and mechano-chemical processes in molecular motor dynamics, such as motor proteins, toward early detection of neurodegenerative diseases.
• Computational methods for robotics, manufacturing, modeling multi-body dynamics, developed methods for identifying limit cycle oscillations in large-dimensional (fluid) systems.
• Turbomachinery and aeroelasticity providing a better understanding of fundamental complex fluid dynamics and cutting-edge models for predicting, identifying and characterizing the response of blisks and flade systems through integrated experimental & computational approaches.
• Structural health monitoring & sensing providing increased sensibility / capabilities by the discovery, characterization and exploitation of sensitivity vector fields, smart system interrogation through nonlinear feedback excitation, nonlinear minimal rank perturbation and system augmentation, pattern recognition for attractors, damage detection using bifurcation morphing.

Mihaela (Miki) Banu

By |

In the area of multi-scale modeling of manufacturing processes: (a) Models for understanding the mechanisms of forming and joining of lightweight materials. This new understanding enables the development of advanced processes which remove limitations of current state-of-the-art capabilities that exhibit limited formability of high strength lightweight alloys, and limited reproducibility of joining quality; (b) Innovative multi-scale finite element models for ultrasonic welding of battery tabs (resulting in models adopted by GM for designing and manufacturing batteries for the Chevy Volt), and multi-scale models for ultrasonic welding of short carbon fiber composites (resulting in models adopted by GM for designing and manufacturing assemblies made of carbon fiber composites with metallic parts); (c) Data-driven algorithms of prediction geometrical and microstructural integrity of the incremental formed parts. Machine learning is used for developing fast and robust methods to be integrated into the designing process and replace finite element simulations.

Tamas Gombosi

By |

Tamas Gombosi is the Konstantin Gringauz Distinguished University Professor of Space Science and the Gerstacker Professor of Engineering at the University of Michigan.
Over his four-decade-long career at Michigan he participated in a number of space missions (Cassini, Rosetta, Stereo, MMS and others). In the last two decades he has led a highly interdisciplinary team that developed the first solution adaptive (AMR) global magnetohydrodynamic (MHD) simulation code of space plasmas. His most recent research focus is to bring advanced machine learning to space weather modeling.
He is Fellow of the AGU (1996), Member of the International Academy of Astronautics (1997), recipient of AGU’s inaugural Space Weather Prize (2013), Van Allen Lecturer of AGU’s SPA section (2017), recipient of the Kristian Birkeland Medal (2018), and recipient of AGU’s John Adam Fleming Medal (2020).

Alex Gorodetsky

By |

Alex Gorodetsky’s research is at the intersection of applied mathematics, data science, and computational science, and is focused on enabling autonomous decision making under uncertainty. He is especially interested in controlling, designing, and analyzing autonomous systems that must act in complex environments where observational data and expensive computational simulations must work together to ensure objectives are achieved. Toward this goal, he pursues research in wide-ranging areas including uncertainty quantification, statistical inference, machine learning, control, and numerical analysis. His methodology is to increase scalability of probabilistic modeling and analysis techniques such as Bayesian inference and uncertainty quantification. His current strategies to achieving scalability revolve around leveraging computational optimal transport, developing tensor network learning algorithms, and creating new multi-fidelity information fusion approaches.

Sample workflow for enabling autonomous decision making under uncertainty for a drone operating in a complex environment. We develop algorithms to compress simulation data by exploiting problem structure. We then embed the compressed representations onto onboard computational resources. Finally, we develop approaches to enable the drone to adapt, learn, and refine knowledge by interacting with, and collecting data from, the environment.

Veera Sundararaghavan

By |

Veera Sundararaghavan is a Professor of Aerospace Engineering at the University of Michigan – Ann Arbor and the director of Multiscale Structural Simulations Laboratory. His research is on multi-length scale computational techniques for modelling and design of aerospace materials with a focus on microstructural mechanics (crystal plasticity, homogenization) and molecular simulation. He is particularly interested in new computational techniques that can revolutionize the way we compute in materials science: machine learning and quantum computing algorithms. He has made important contributions in the area of integrated computational materials engineering (ICME) including reduced order representations for microstructure-process-property relationships, Markov random fields approach for microstructure reconstruction, and parallel, multiscale algorithms for optimizing deformation, fatigue, failure and oxidation response in polycrystalline alloys, high temperature ceramic matrix composites and energetic composites. Methods of choice for data science include deep Boltzmann machines, undirected graph models (Markov random fields) and Support vector machines.

An illustration of the hybrid Quantum-Classical computation technique: Quantum Annealer is used as a Boltzmann sampler while the gradient optimization is carried out using classical computation

Lei Chen

By |

Lei Chen’s group focus on applying data science tools to advanced manufacturing. Chen’s research expertise and interests are to integrate the physics-based computational and experimental methods and data-driven approaches, to exploit the fundamental phenomena emerged in advanced manufacturing and to establish the design protocol for optimizing the materials and process parameters of as-fabricated parts for quality control. Current research can be summarized by:
1 One of significant challenges in additive manufacturing (AM) is the presence of heterogeneous sources of uncertainty involved in the complex layer-wise processes under non-equilibrium conditions that lead to variability in the microstructure and properties of as-built components. Consequently, it is extremely challenging to repeat the manufacturing of a high-quality product in mass production, and current practice usually reverts to trial-and-error techniques that are very time-consuming and costly. This research aims to develop an uncertainty quantification framework by bringing together physical modeling, machine-learning (ML), and experiments.
2 Computational microstructure optimization of piezocomposites involves iterative searches to achieve the desired combination of properties demanded by a selected application. Traditional analytical-based optimization methods suffer from the searching efficiency and result optimality due to high dimensionality of microstructure space, complicated electrical and mechanical coupling and non-uniqueness of solutions. Moreover, AM process inherently poses several manufacturing constraints e.g., the minimum feature size and the porosity in the piezoelectric ceramics as well as at the ceramics-polymer interface. It is challenging to include such manufacturing constraints since they are not explicitly available. This research aims to develop a novel data-driven framework for microstructure optimization of AM piezoelectric composites by leveraging extensive physics-based simulation data as well as limited amount of measurement data from AM process.
3 Lithium (Li) and other alkali metals (e.g., sodium and potassium) are very attractive electrode candidates for the next-generation rechargeable batteries that promise several times higher energy density at lower cost. However, Li-dendrite formation severely limits the commercialization of Li-metal batteries, either because dendrite pieces lose electrical contact with the rest of the Li-electrode or because growing dendrites can penetrate the separator and lead to short circuits. This research aims to develop a computational model to accelerate the design of dendrite-free Li-metal batteries.

9.9.2020 MIDAS Faculty Research Pitch Video.

Blueprint for the research: data-driven modelling of additive manufacturing. Stereolithography-based and laser melting-based additive manufacturing processes are used to fabricate the powder-based piezoelectric ceramics and metals respectively, with controllable complex microstructures and/or architectures to tune material properties. Physics-based numerical simulations are performed in an “in-house” multiscale computational framework, which includes macroscopic finite-element based manufacturing process modelling, mesoscopic phase-field modelling of microstructure evolution and design, and first principles/CALPHAD calculation of thermodynamics and kinetics. Data-driven approaches include machine learning and uncertainty quantification with surrogate models, such as polynomial chaos expansion, Gaussian process, radial basis functions, etc.

Zhen Hu

By |

I am an assistant professor in Department of Industrial and Manufacturing Systems Engineering (IMSE) at the University of Michigan-Dearborn. Prior to joining UM-Dearborn, I was a research assistant professor and postdoctoral research scholar at Vanderbilt University. My research areas of interest are uncertainty quantification, Bayesian data analytics, big data analytics, machine learning, optimization under uncertainty, and applications of data analytics and machine learning in aerospace, mechanical and manufacturing systems, and material science. The goal of my research is to develop novel computational methods to design sustainable and reliable engineering systems by leveraging the rich information contained in the high-fidelity computational simulation models, experimental data, and big operational data and historical data.

Xun Huan

By |

Prof. Huan’s research broadly revolves around uncertainty quantification, data-driven modeling, and numerical optimization. He focuses on methods to bridge together models and data: e.g., optimal experimental design, Bayesian statistical inference, uncertainty propagation in high-dimensional settings, and algorithms that are robust to model misspecification. He seeks to develop efficient numerical methods that integrate computationally-intensive models with big data, and combine uncertainty quantification with machine learning to enable robust and reliable prediction, design, and decision-making.

Optimal experimental design seeks to identify experiments that produce the most valuable data. For example, when designing a combustion experiment to learn chemical kinetic parameters, design condition A maximizes the expected information gain. When Bayesian inference is performed on data from this experiment, we indeed obtain “tighter” posteriors (with less uncertainty) compared to those obtained from suboptimal design conditions B and C.

Venkat Raman

By |

Prof. Raman’s work focuses on the simulation of large scale combustion systems – aircraft engines, stationary power turbines, hypersonic engines – with the goal of advancing computations-aided systems design. This involves large scale computations accounting for detailed behavior of the chaotic turbulent flow in these systems, combined with enabling science in computational chemistry and algorithms. One aspect of my research is the prediction of rare events that lead to catastrophic system failure (as in flight crash, engine failure etc.). This work also involves the understanding of uncertainty in models, and streamlining knowledge in the form of mathematical models.

raman_image_large-1024x765

Karthik Duraisamy

By |

Karthik Duraisamy, PhD, is Associate Professor of Aerospace Engineering in the College of Engineering at the University of Michigan, Ann Arbor.

Prof. Duraisamy’s group focuses on data-driven modeling of computational physics problems. Specifically, we use statistical inversion and physics-informed machine learning techniques to augment existing computational models. Another focus area is formal reduced order modeling using data-driven decompositions.

Our application areas are in turbulence, combustion and materials physics.

Turbulent flow in a trailing vortex.

Turbulent flow in a trailing vortex.