Abstract: Deep learning approaches have become crucial tools across numerous engineering domains. However, they face various challenges, as they typically depend on representative data and large training datasets. Conversely, condition monitoring data for complex systems often lacks labels and representativeness, posing significant challenges for purely data-driven approaches. Additionally, deep learning models generally struggle in extrapolation regimes, which are common for assets characterized by long service lives and the frequent emergence of new operating regimes. 

In response to these challenges, the integration of physical laws and principles with deep learning methodologies has shown tremendous promise. This presentation will explore a variety of approaches that combine physics-based concepts with deep learning techniques. One focus will be on how incorporating structural inductive biases into learning architectures, such as through physics-enhanced graph neural networks, can address the aforementioned challenges.

To close the loop and bridge machine learning with physics, the talk will delve into novel approaches of symbolic regression through reinforcement learning to uncover symbolic equations.

Biographical Sketch: Olga Fink has been assistant professor of Intelligent Maintenance and Operations Systems at EPFL since March 2022.  Olga’s research focuses on Hybrid Algorithms Fusing Physics-Based Models and Deep Learning Algorithms, Hybrid Operational Digital Twins, Transfer Learning, Self-Supervised Learning, Deep Reinforcement Learning and Multi-Agent Systems for Intelligent Maintenance and Operations of Infrastructure and Complex Assets.

Before joining EPFL faculty, Olga was assistant professor of intelligent maintenance systems at ETH Zurich from 2018 to 2022, being awarded the prestigious professorship grant of the Swiss National Science Foundation (SNSF). Between 2014 and 2018 she was heading the research group “Smart Maintenance” at the Zurich University of Applied Sciences (ZHAW).

Olga received her Ph.D. degree from ETH Zurich with the thesis on “Failure and Degradation Prediction by Artificial Neural Networks: Applications to Railway Systems”, and Diploma degree in industrial engineering from Hamburg University of Technology. She has gained valuable industrial experience as reliability engineer with Stadler Bussnang AG and as reliability and maintenance expert with Pöyry Switzerland Ltd.

Olga has been a member of the BRIDGE Proof of Concept evaluation panel since 2023. Moreover, Olga is serving as an editorial board member of several prestigious journals, including Mechanical Systems and Signal Processing, Engineering Applications of Artificial Intelligence, Reliability Engineering and System Safety and IEEE Sensors Journal.

In 2018, Olga was honored as one of the “Top 100 Women in Business, Switzerland”. Additionally, in 2019, earned the distinction of being recognized as a young scientist of the World Economic Forum. In 2020 and 2021, she was honored r as a young scientist of the World Laureate Forum. In 2023, she was distinguished as a fellow by the Prognostics and Health Management Society.

Abstract: Water management is one of the most critical issues in proton exchange membrane fuel cells (PEMFCs). The water generated in catalyst layer as a product of the electrochemical reaction is mainly transported through porous media by diffusion if it’s vapor, or by capillarity in case of liquid. In flow channels, the liquid water is removed primarily by inertial force of the gas flows. In my research group (Multiscale Transport Process Laboratory) at Michigan Technological University, one of our focused research areas is the gas-liquid two-phase transport processes in PEMFCs.

In various aspects of the two-phase transport phenomena, this presentation is focused on the impact of land-channel geometry. If we look at the cross-section of PEMFC, land-channel geometry causes the difference in transport distance between the flow channel to the catalyst layer, and results in the uneven distribution of various factors, such as transport resistance, species concentration, and current generation. In order to investigate the distributions of various parameters in the land-channel direction, we developed a small-scale segmented cell with about 350-micron resolution, and successfully measured the current and high-frequency resistance distribution in the land-channel direction for two different flow fields.

Biographical Sketch: Dr. Kazuya Tajiri is an associate professor of Department of Mechanical Engineering-Engineering Mechanics at Michigan Technological University. He has obtained his Bachelor degree in Aeronautics and Astronautics from University of Tokyo, Master degree in Aerospace Engineering from Georgia Institute of Technology, and Ph. D in Mechanical Engineering from The Pennsylvania State University. After obtaining a Ph.D degree, he worked at Argonne National Laboratory as a postdoctoral researcher, and then in 2010 he joined Michigan Technological University as an assistant professor. He also has work experience at Nissan Research Center in Yokosuka, Japan. In 2013, he was selected as one of the finalists for the Distinguished Teaching Award at Michigan Technological University.

Abstract: The 9-month journey home from Mars could begin with a 9 ns laser pulse.  Ignition in rocket combustors is typically accomplished using a spark plug, a pyrotechnic charge, an injection of hypergolic fluid, or a hot gas torch. These methods involve significant mechanical complexity, increase the inert mass with ancillary subsystems, limit the potential for engine re-ignitions throughout a mission, and require additional (often toxic) propellants. Non-resonant breakdown ignition is an alternative method in which the ignition energy is provided through a focused pulse of laser light. If the local flow conditions in the vicinity of the spark are suitable, a flame will develop and stabilize within the combustor. Laser-induced spark ignition holds significant promise for rocket combustion systems because the point of energy deposition can be precisely placed at an optimum location that minimizes the ignition energy requirement.  This talk will focus on an experimental characterization ignition probability in a gaseous oxygen and gaseous methane combustor. The oxygen-centered shear co-axial injector generated a widely varying mixture field and velocity field to create significant variability in both the ignition process and outcome.  Results from time-resolved imaging diagnostics will be discussed to explain the mechanisms that manifest the final ignition probability.

Biographical Sketch: Dr. Carson Slabaugh is the Paula Feuer Associate Professor in the School of Aeronautics and Astronautics at Purdue University.  Since joining Purdue in 2015, he has developed an education and research program focused on propulsion.  Dr. Slabaugh’s laboratory is housed within the Purdue Zucrow Laboratory complex, with high pressure, high flow-rate system capabilities to enable experimental replication of the flow and flame conditions (pressure, turbulence level, thermal power density) found in the most advanced propulsion and combustion systems.  Ongoing research projects cover a wide range of topics: from the fundamental exploration of detonations and turbulent flames to the development of advanced combustion technologies for liquid rocket engines and rotating detonation engines. His group also maintains a continuous effort in the advancement of high-bandwidth (typically, laser-based) measurement techniques to non-intrusively probe the physics of these complex, reacting flows.  Support for these research projects has been provided by AFOSR, AFRL, DARPA, DOE, NASA, ONR, and numerous industrial partners.  Prof. Slabaugh has published extensively in the field and is involved with multiple national efforts to transition advanced concepts into aerospace propulsion technologies.

Abstract: Chalcogenide phase change materials (PCMs) have promising properties for photonic applications thanks to their nonvolatile and large refractive index modulation [1]. The last decade has seen a growing interest in such a combination of properties for a variety of nonvolatile programmable devices, such as metasurfaces, tunable filters, phase/amplitude modulators, color pixels, thermal camouflage, photonic memories/computing, plasmonics, etc.—giving rise to the so-called Phase-change Photonics field.[1] PCM-based devices rely on the precise switching between the amorphous and the crystalline states, which can be achieved through optical or electrical pulses via optical absorption and Joule heating, respectively. Optical pulse switching is the fastest and most precise method; however, it lacks scalability given the difficulty of on-chip pulse routing when considering many PCM cells. It is also limited to absorptive PCMs, such as Ge2Sb2Te5. As a scalable alternative, on-chip microheaters using multiple material platforms have been proposed, e.g. doped-silicon, graphene, ITO, metals, etc. Doped-silicon microheaters are particularly interesting since they are CMOS compatible and can be fabricated onto silicon-on-insulator (SOI) wafers—the same platform used for silicon photonic integrated circuits. However, this electro-thermal switching also has shortcomings. It lacks stable multi-level response due to the stochastic nature of both amorphization and crystallization processes in the typical bow-tie-like devices where the microheater heats the entire cell to an almost flat temperature [2,3]. Because of this, the focus is shifting towards the microheater’s geometry in addition to the choice of conductive material and its integration. One common goal is to control the hotspot size within the PCM cell to deterministically switch specific areas (i.e., spatially resolved amorphous domains in a crystalline cell) and, thus, achieve controllable and reproducible optical modulation [4,5]. In this talk, I will review the fundamentals of PCM for photonics and the proposed electro-thermal mechanisms. I will then focus on our current efforts to re-engineer doped-silicon microheaters for optimum performance.

[1] M. Wuttig, et, al. “Phase-change materials for non-volatile photonic applications,” Nat. Photon.11(8), 465–476 (2017).

[2] J. Feldmann, et, al. “Calculating with light using a chip-scale all-optical abacus,” Nat Commun 8(1), 1256 (2017)

[3] T. Tuma, et, al. “Stochastic phase-change neurons,” Nat. Nanotechnol.11(8), 693–699 (2016).

[4] C. Ríos, et, al “Ultra-compact nonvolatile phase shifter based on electrically reprogrammable transparent phase change materials,” PhotoniX 3(1), 26 (2022).

[5] X.  Li, et, al, “Fast and reliable storage using a 5 bit, nonvolatile photonic memory cell,” Optica 6, 1-6 (2019)

[6] Y. Zhang, et, al, “Myths and truths about optical phase change materials: A perspective.” Appl. Phys. Lett. 118 (21): 210501.

Biographical Sketch: Carlos A. Ríos Ocampo is an Assistant Professor at the University of Maryland, College Park, where he has led the Photonic Materials & Devices groups since 2021. Before joining UMD, Carlos was a Postdoctoral Associate at MIT, received a DPhil (PhD) degree in 2017 from the University of Oxford (UK), an MSc degree in Optics and Photonics in 2013 from the KIT (Germany), and a BSc in Physics in 2010 from the University of Antioquia (Colombia). Carlos’s scientific interests focus on studying and developing new on-chip technologies driven by the synergy between nanomaterials and photonics.

 

Abstract: Aerosol particles are ever-present in both natural environments, and in engineered systems.  In large part, control over aerosol particle transport hinges upon control of particle inertia, i.e. the propensity of particles to maintain a particular trajectory whilst the surrounding fluid moves in a distinct direction.  Increasing the inertia of increasingly small particles typically involves increasing particle initial velocities, such that when fluid velocities slow down, particle-fluid velocity differences are maximized.  While inertial principles have long been exploited in a wide variety of aerosol instruments, including impactors, virtual impactors, and aerodynamic particle spectrometers, application of inertially-governed aerosol systems wherein the particle Mach number (based on its velocity difference with the fluid) approaches or even exceeds unit value are much more limited.  This presentation will overview current understanding of aerosol particle behavior in high-speed systems, and subsequently discuss ongoing studies of high speed particle behavior.  Specifically, the drag force on particles as a function of both the Knudsen number and the Mach number, derived from direct simulation Monte Carlo, will first be discussed.  Subsequently, results from ongoing studies of crater formation due to microparticle impacts at up to 1 km s-1 speeds with high-speed flight-relevant materials will be presented.  Also discussed will be insights from atomistic simulations into the changes in crystallinity and defect density experienced by particles during high-speed impacts, relevant to aerosol deposition of coatings, and the optimization of virtual impactor aerosol concentrators operating near the sonic limit, capable of submicrometer particle concentration enhancement.

Biographical Sketch: Dr. Chris Hogan is the Carl and Janet Kuhrmeyer Chair Professor and the Associate Department Head in the Department of Mechanical Engineering at the University of Minnesota, Twin Cities, where has been a faculty member since 2009.  His research group focuses largely on fundamental and applied research in aerosols, including the development of theories to describe transport and reactions in aerosols, the design of new measurement principles and instruments for aerosols, and the evaluation of HVAC control technologies.  He has published more than 150 peer reviewed papers focusing in these areas, and is the editor-in-chief of the Journal of Aerosol Science.

Abstract: In this talk, we’ll discuss how engineering students can utilize KeyShot to elevate the aesthetics and presentation of their concepts, increasing the likelihood their ideas will garner attention and be valued. Through an in-person demonstration, the Education Program Manager from KeyShot will illustrate just how quick and painless learning the industry-leading 3D-visualization software can be.

Biographical Sketch: Nick Abbott is a seasoned Industrial Designer with a rich five-year journey in the field. He has contributed to numerous projects for prominent clients, constantly expanding his creative horizons and refining his product design skills. Nick’s versatility led him to co-teach a design course at ASU, which paved the way for his latest endeavor at KeyShot. His bold decision to leave Purdue for startup adventures in the music industry reflects his risk-taking spirit and innovative mindset. Outside of work, Nick is an avid photographer, often found capturing the stunning landscapes of the mountains.

Abstract: Multiscale modeling methods are typically envisioned as precise and predictive simulation tools to solve complex science and engineering problems. However, even conventional atomistic models often lack computational efficiency and accuracy, making them inadequate for providing reliable information for large-scale continuum models. In this seminar, I will discuss the developments aimed at overcoming these critical limitations. 

At the beginning of the talk, I will introduce how atomistic models can enhance our understanding of the experimental observations of crystal growth in 2D materials using empirical reactive forcefield (FF). Although the models offer valuable insights at the atomic scale, the development of reliable FFs is severely limited due to the fixed potential expressions. Recently, neural network potentials (NNPs) have emerged to surpass the longstanding limitations of empirical potentials.

While NNPs can provide higher accuracy than other empirical FFs and lower computational costs compared to quantum calculations, efficient sampling or data generation for training becomes increasingly critical. I will present recent advancements in an automated active learning (AL) framework for NNPs, focusing on accurately simulating bond-breaking in hexane chains through steered molecular dynamics sampling with trained NNPs and assessing model transferability to other alkane chains.

In the end, I will introduce one of the sampling approaches, the multiorder-multithermal (MOMT) ensemble, to capture a broad range of liquid- and solid-phase configurations in a metallic system, nickel. Data distillation through active learning can significantly reduce sampled data without losing much accuracy. The NNP, trained from the distilled data, can predict different energy-minimized closed-pack crystal structures even though those structures were not explicitly part of the initial data. Moreover, this data can be applied to other metallic systems (aluminum and niobium) without repeating the sampling and distillation processes.

The capabilities developed through the research will provide valuable tools for a fundamental understanding of the chemical process and mechanistic insights into the predictive design and interpretive simulations of materials processes and properties.

Biographical Sketch: GS Jung is a Research Staff at Oak Ridge National Laboratory. His research interests are in the multiscale modeling of materials to understand their fundamental properties from synthesis and growth to performance and failures. Before joining ORNL, he earned his Ph.D. in multiscale modeling for 2D materials from the Massachusetts Institute of Technology.