Laboratoire des Écoulements Géophysiques et Industriels




Nos tutelles

CNRS

Nos partenaires

Rechercher


Accueil > Actualités > Séminaires > Séminaires 2019

Mardi 12 mars 2019 à 11h00 en amphi K118

Thomas Engels, LMD UMR8539, École Normale Supérieure, Paris & Berlin Universität

Titre/Title : Bio-Inspired Flight in Turbulence : from Modeling to Data Analysis
Contact : Christophe Brun (équipe MEIGE)

Résumé/Abstract : The spectacular flight capabilities of flying insects have motivated researchers to consider bio-inspired flapping flight as an alternative to fixed wing or rotary flight. In this talk, we will discuss the fluid-dynamics aspects of flapping flight from the point of view of numerical modeling, high-performance computing and data analysis. The main questions that we have in mind are : How do flapping fliers control their flight in highly turbulent and obstacle-rich environments ? What can engineers learn from animals in their attempts to design bio-inspired flapping robots ? To answer those questions, the aim of the present work is to device models to predict and control the nonlinear dynamics of such fliers, even when challenged by turbulent perturbations.
The first step towards this goal is establishing a set of reliable data of flapping fliers in turbulent flow. Our approach is to use direct numerical simulation, solving the incompressible Navier-Stokes equations using high-order Fourier pseudospectral methods. The complex, moving and possibly deforming fluid-solid interface is taken into account by the volume penalization technique, a generic approach also applicable to other technical problems with moving boundaries, for example wind turbines. Our simulations typically require national-scale supercomputers and algorithms suitable for their exploitation, which is an important aspect of the present work. Outcome of these simulations are highly accurate (and typically large, of the order of 50 TB) flow data with all excited scales of fluid motion resolved. We present first results on model reduction of this data using proper orthogonal decomposition. Reducing the problem size by focusing on essential degrees of freedom motivates the new development of a wavelet-based adaptive multiresolution solver. The idea is based on the work of M. Farge and K. Schneider, who proposed to define coherent structures as what remains after denoising. The approach is called ‘coherent vortex simulation’ (CVS) and has so far only been applied to idealized setups like homogeneous isotropic turbulence. The computational grid is dynamically changing in time and tracks important features of the solution, which is challenging in parallel computing, but distinguishes it from the large-eddy simulation approach. Here, we provide such a parallel computational framework and extend CVS to complex geometries. It can then be used not only for simulations of flapping flight in turbulent flow, but for other technical applications as well, including reactive flows, wind turbines or medical flows.

Fig. 1 : Volume rendering of vorticity magnitude produced by a bumblebee model in forward flight with laminar inflow condition.

Portfolio