Research

High-order contact detection between fibres

In this work we analyse the contact forces yielded by fibre models coupled to a segment-based contact detection algorithm. We note the occurrence of spurious jumps which, to the best of our knowledge, were never reported before. We demonstrate that these artifacts are actually directly linked to the low-order contact detection being used between the fibre and the obstacles, and that they worsen as curvature at contact increases. Low-order detection, based on segment proxys, is classically used due to its simplicity of treatment, even for fibre models possessing a higher-order geometry like super-helices. We show that spurious artifacts occur whatever the fibre model and contact response solver used, as soon as a segment-segment detection scheme is employed. To remove such numerical artifacts in the force profile, which can accumulate to yield large force errors, we introduce an efficient high-order detection algorithm, relying on an efficient adaptive pruning strategy.

This work has been published in the ACM Transactions on Graphics journal (ACM SIGGRAPH 2024) 10, and presented at the corresponding ACM SIGGRAPH 2024 conference. It was also presented to the French Nonlinear Physics community at the national RNL 2024 conference 19.

Numerical modeling of a feather’s vane, a highly anisotropic membrane

In the context of Jean Jouve’s doctoral thesis, in collaboration with Rahul Narain (Indian Institute of Technology Delhi, India) and Theodore Kim (Yale University, United States), we have developed a macroscopic shell simulator capable of dealing with highly anisotropic materials, such as feathers.

Feathers have a multi-scale structure with very interesting properties. From a central stiff rod (the rachis) hundreds of small fibers (barbs) come out to form a membrane-like structure, which is held together by yet another set of smaller fibers (barbules). This intricate structure drives the elastic response of the membrane and gives feathers the capability of reversible rupture. Our interest is to produce a simulator capable of realistically portray the deformation of a feather. This simulation is challenging because the membrane has two main directions with elastic coefficients differing by 4 orders of magnitude. We have compared and validated our results with experimental data. We have published and presented our findings in a conference paper at the ACM SIGGRAPH 2024 conference 11 .

Mixed elements for a curvature-based ribbon model

In collaboration with Sébastien Neukirch (Institut D’Alembert, Sorbonne Université) and our former Ph.D. student Raphaël Charrondière, we have improved our initial curvature-based numerical model for thin elastic ribbons 2 by developing a mixed strategy, where each ribbon element is treated independently, and glued to each other only at the final solving stage through well-chosen bilateral constraints.

Thanks to this mixed variational strategy, which yields a banded Hessian, our algorithm recovers the linear complexity of low-order models while preserving the quadratic convergence of curvature-based models. As a result, our approach is scalable to a large number of elements, and suitable for various boundary conditions and unilateral contact constraints, making it possible to handle challenging scenarios such as confined buckling experiments or Möbius bands with contact. Additionally, our numerical model can incorporate various ribbon energies, including the recent 20 model for quasi-developable ribbons recently introduced in Physics, which allows to transition smoothly between a rectangular Kirchhoff rod and a (developable) Sadowsky ribbon. Our numerical scheme is carefully validated against demanding experiments of the Physics literature, which demonstrates its accuracy, efficiency, robustness, and versatility.

This work has been published in the ACM Transactions on Graphics journal (ACM SIGGRAPH 2024) 9, and presented at the corresponding ACM SIGGRAPH 2024 conference. It was also presented to the French Mechanical Engineering community at the national CSMA 2024 conference 9.

Nonsmooth simulations of 3D Drucker-Prager granular flows and validation against experimental column collapses

In collaboration with Gauthier Rousseau (TU Wien, formerly post-doc in the team), Hugo Rousseau (INRAE) and Gilles Daviet (NVIDIA, formerly Ph.D. student in the team), we have performed thorough comparisons between the predictions of our numerical solver Sand6 for granular flows 7.1.2, and collapse experiments conducted in a narrow channel (in collaboration with EPFL). We have shown that our nonsmooth simulator, which relies on a constant friction coefficient corresponding to the yield angle of a granular heap, is able to reproduce with high fidelity various experimental granular collapses over inclined erodible beds. Our results, obtained for two different granular materials and for various bed inclinations, suggest that a simple constant friction rheology choice remains reasonable for capturing a large variety of unsteady granular flows at low inertial number. Using the versatility of our numerical approach, we have further analysed the possible biases pertaining to laboratory-scale experiments, and shown that, in the case of granular collapses, accurate predictions could be performed as long as care was taken in measuring yield angle of the granular material appropriately. Our numerical simulations demonstrate the specific nature of transient granular flows, which are mainly driven by successive layered avalanches, and we have explored the role of metastability and hysteresis in the onset of flow and inner dissipation within the material. This study has been published in the Journal of Fluid Mechanics 8 along with all the research data at doi.org/10.5281/zenodo.7288829. It has been presented at several international conferences 12, 15, 16.

Hydrodynamic model for fish locomotion

In collaboration with Bruno Ventéjou (co-supervised post-doc at LIPhy, UGA), Philippe Peyla (LIPhy, UGA) and Aurélie Dupont (LIPhy, CNRS), we study the respective roles of hydrodynamic and social interactions within schools of fish, in the context of the FISHSIF ANR project. As a first step toward the simulation of large assemblies of swimming fish, we have developed a simplified hydrodynamic model of a swimmer, able to account for individual fish swimming and stigmergy, in particular regarding the generation of vortices wakes, without the need to introduce deformation of the body of the fish. We have performed detailed hydrodynamic scaling analyses of the velocity of a moving immersed body, and shown that the motion of swimmers obeys a universal scaling law expressed in terms of only two dimensionless quantities describing the relative importance of inertia, viscosity and swimming forces. Using extensive numerical simulations, we have shown excellent agreement between our theoretical scaling laws and the swimming behaviour of our model fish. The validity of our scaling laws notably extend across a wide range of hydrodynamic regimes (from the Stokes to the turbulent regime), and demonstrates the ubiquitous decrease in swimming efficiency as the velocity increases. We have further compared our results to experimental data collected among many aquatic species, with very different body shapes, deformations, and swimming velocities. The overall collapse of swimmers’ data onto our single master curve supports the robustness and genericity of our analysis and model. The corresponding publication is available as a preprint 18 and is currently under review.