A Glimpse into Mechanics
SwissMech Seminars is a monthly webinar series taking place at 16:00pm on the second Thursday of each month of the academic year. The talks are organized jointly by ETH Zürich and EPFL Lausanne. The speakers cover theoretical, computational and experimental aspects of solid and fluid mechanics in the broadest sense.
Academic Year 2021/2022
Every second Thursday of each month at 16:00pm during the academic year.
Mechanics of spontaneously arrested laboratory earthquakes
Computational Mechanics of Building Materials, ETH Zurich
Date/Time: October 14, Thursday, 16:00-17:00
Abstract: Earthquakes, which we experience as ground shaking, consist of sudden relative motion between tectonic plates. The underlying mechanical process involves three phases – the initiation of local slip, its growth along the tectonic fault, and its arrest. This rupture-like process is governed by physics at multiple length scales, making it a highly complex phenomenon that remains only partially understood. This is particularly true for earthquake arrest, which directly affects its magnitude. The current understanding of earthquake arrest is almost exclusively based on remote measurements because most laboratory experiments are too small to allow rupture arrest to occur naturally. However, recently developed large-scale laboratory experiments on granite blocks provide the necessary fault length to generate laboratory earthquake ruptures that not only nucleate and propagate, but also arrest, spontaneously. These experiments provide an opportunity to reexamine and better understand the physics governing earthquake arrest conditions. In this talk, we will discuss various analytical and numerical models that enable in-depth analysis, interpretation and extrapolation of results from such large-scale laboratory earthquake experiments. The results suggest that rupture arrest (at least in the laboratory) is controlled by the driving force rather than by the resistance, as often assumed. Further, we will discuss fault fracture energy, which is a key parameter in the arrest of earthquake ruptures. We will present a minimal numerical model with scale-invariant fault fracture energy in accordance with laboratory observations. However, when applying seismological approaches to estimate the fault fracture energy, it appears to be scale-dependent, similar to field observations, despite being scale-invariant. Therefore, the model reconciles conflicting observations from the field and the laboratory, and provides a pathway for more realistic models of earthquake arrest mechanisms.
A virtual lab tour of the flexible structures laboratory at EPFL
Institute of Mechanical Engineering, EPFL Lausanne
Date/Time: November 11, Thursday, 16:00-17:00
Abstract: It used to be that academic seminars involved a researcher visiting a host institution to deliver a talk and meet with colleagues (do you remember those days?!). Times have changed, at least temporarily, but this situation is also opening opportunities. In this ‘talk’, we will be inviting you for a virtual tour of our Flexible Structures Laboratory (fleXLab) at EPFL in Switzerland. We will show you some of our experimental facilities and share some of our recent research activities, focusing on the mechanics of magneto-active structures. Multiple members of our team will be involved in this tour. Research at our fleXLab focuses is centered in the general area of the mechanics of slender structures, which leverage their post-buckling regime for novel modes of functionality. Methodologically, we recognize scaled high-precision model experiments as a powerful tool for discovery in mechanics, supported by theory and computation, in a vision of science-enable engineering and engineering-motivated science. Recently, we have become fascinated with active structures made out of magneto-rheological elastomers that can be actuated in the presence of an external magnetic field. After introducing some recent advances in experimentation, modeling, and computational for this class of systems, we will present a series of concrete examples. Specifically, we will discuss (i) (re)programmable mechanical metamaterials with programmable memory; (ii) magneto-active beams and Kirchhoff-like rods; and (iii) magnetic shells with tunable buckling properties.
This virtual lab will involve the participation of a few members of the fleXLab at EPFL.
Machine learning based plasticity modeling
Date/Time: December 9, Thursday, 16:00-17:00
Abstract: Machine learning offers a data-driven approach to the development of constitutive models as an alternative to classical physics-based modeling. Recent applications of machine learning in the context of metal plasticity are presented ranging from temperature and rate dependent hardening laws to 3D constitutive models for anisotropic solids. In addition to developing mechanics-specific neural network architectures, new robot-assisted experimental procedures are presented that generate “big data” for the identification of machine-learning based plasticity and failure models from experiments.