Mercredi 21 Décembre 2011
14 h 00 salle de réunion LEA/SP2MI
OCE Post Doctoral Fellow
CSIRO Marine and Atmospheric Research, Australia
Invité par Jean-Paul Bonnet
Subgrid-scale (SGS) parameterisations of turbulence with self similar scaling laws are developed for the large eddy simulations (LESs) of atmospheric and oceanic flows. The SGS model coefficients are determined self-consistently from higher resolution reference simulations. The resulting LES then replicates the statistics of the reference data within the resolved scales. The flow fields are simulated using a spectral two-level quasi-geostrophic code that incorporates the processes of baroclinic instability and the interaction of synoptic-scale structures and inhomogeneous Rossby wave turbulence. Two specific basic flows are analysed: an atmospheric flow with large scale jets in the mid-latitudes; and an oceanic flow representative of the Antarctic Circumpolar Current. In both cases the turbulent energy in the system is injected at the Rossby wavenumber (kR). In the present simulations, for the atmosphere kR=14, whilst for the ocean kR=140. This makes the ocean a more computationally challenging case, as a finer grid is required to resolve the energy injection. It is found that for both the atmosphere and ocean, as the resolution increases the SGS model coefficients decrease in magnitude, and become steeper in wavenumber space. A unified scaling law is proposed that represents both the atmosphere and ocean SGS processes.
Mercredi 14 Décembre 2011
14 h 00 salle de réunion LEA/SP2MI
Associate Professor
Syracuse University, USA
Invité par Joël Delville
The abstract can be downloaded here.
Jeudi 8 Décembre 2011
11 h 00 salle de réunion CEAT
Doctorant
ENSTA ParisTech - Unité de Mecanique (UME)
Invité par Bernd Noack
The sensitivity of the global properties of a turbulent wake behind a D-shaped bluff cylinder is investigated experimentally by introducing a much smaller control cylinder of various diameters as a local disturbance. Hot-wire anemometry and particle image velocimetry are used to obtain local and global measurements of the turbulent wake. Aerodynamic forces acting on the main cylinder are derived from pressure measurements around its perimeter. The results are presented in the form of sensitivity maps of the Strouhal number and base pressure. The sensitivity of global properties is interpreted on the basis of the ability of the control cylinder to change the size of the formation region of the Karman vortex street mainly through the turbulent properties modification of the perturbed detached shear layers. The corresponding physical mechanisms are discussed with regard to the origins of drag reduction and global frequency modification.
Mercredi 30 Novembre 2011
11 h 00 salle de réunion CEAT
Post Doctoral Research Associate
Technische Universität Ilmenau, Germany
Invité par B. Noack
Two low-dimensional models (LDM) for turbulent convection in rectangular cells, based on the Galerkin projection of the Boussinesq equations onto a finite set of empirical eigenfunctions, are presented. The empirical eigenfunctions are obtained from a Proper Orthogonal Decomposition (POD) of the fields using the Snapshot Method. In the first step we determine the coherent structures, and how much they contribute to the global momentum and heat transfer. The first case is a three-dimensional cell in which a classical turbulent Rayleigh-Bénard flow evolves. The second case is based on two-dimensional DNS data of mixed convection in a cell with heated obstacles at the bottom as well as in- and outlets of air. In both cases, a quadratic inhomogeneous coupled ODE system is obtained for the evolution of the modal amplitudes. The truncation to a finite number (a few hundred) of degrees of freedom, requires the additional implementation of an eddy viscosity-diffusivity, η, to capture the missing dissipation of the small-scale modes. The magnitude of this additional dissipation mechanism is determined by requiring statistical stationarity and a total dissipation that corresponds with the original DNS data. We introduce a mode-dependent η, which turns out to reproduce the largescale properties of the turbulent convection qualitatively well. We also compare this with the Heisenberg-constant η (Aubry et al. 1988). In this way, we obtain more insights into the long term evolution of the flow.
Mardi 18 Octobre 2011
14 h 00 salle de réunion CEAT
Group leader
Comparative Visualization Group - Zuse Institute Berlin
Invité par B. Noack
Automatic extraction of characteristic flow structures from numerical and experimental data of complex fluid flows plays an essential part in flow analysis. In visual data analysis such flow structures are often called "features". The basis for the development of any feature extraction method is a mathematically precise definition of "features of interest". Here, one can distinguish two different scenarios. First there is already a clear understanding of the feature. Typically, this results in feature identifiers defined as Galilean invariant flow quantities, e.g., based on pressure, acceleration, vorticity. Another scenario arises if the mathematical description is not a priori given, and a more intuitive idea of the feature exists. In this talk I will present two different data analysis approaches to extract and follow features related to these scenarios.
The first method focuses on acceleration as fundamental quantity of flow fields that captures the Galilean invariant properties of particle motion. Considering its magnitude features can be defined as extremal structures of the resulting scalar fields. In this context a particular challenge is a robust extraction of the related structures. This concerns dealing with noise as well as with high feature densities. Utilizing scalar field topology, vortex cores and regions can be robustly extracted and tracked using one algorithm framework.
The second method is based on the definition of a flow description using invariant moments. The resulting descriptors support an interactive selection of interesting flow structures in streamline representation, which are used as feature template for the extraction of similar structures and tracking over time.
Short Bio. Ingrid Hotz received the M.S. degree in theoretical Physics from the Ludwig Maximilian University in Munich Germany and the PhD degree from the Computer Science Department at the University of Kaiserslautern, Germany. During 2003 - 2006 she worked as a postdoctoral researcher at the Institute for Data Analysis and Visualization (IDAV) at the University of California. Currently she is the leader of a junior research group at the Zuse Institute in Berlin Germany. Her research interests are in the area of data analysis and scientific visualization with focus on tensor and vector fields.
Mercredi 21 Septembre 2011
14 h 00 salle de réunion LEA/SP2MI
Professor
Head Dept. Visualization and Data Analysis - Zuse Institute Berlin
Invité par B. Noack
In science information is deduced from more or less established theories or gained from experiments. In order to extend knowledge and to proceed scientifically hypotheses are built based on observations and they are verified or falsified. Since almost all information is tainted with uncertainty, is is part of the scientific paradigm to establish measures of reliability and to take uncertainties into account in all conclusions.
Today data analysis is often supported by visualization - both, in the phase of hypothesis creation and of hypothesis verification/falsification. However, only few attempts have been made to develop visualization techniques that depict uncertainties in the data. Furthermore, most of these techniques are rather heuristic and just qualitative.
I will discuss how uncertainty visualization can be put on a firm ground by building on established methods for representation, quantification and propagation of uncertainties in statistics. This will be illustrated using one of the simplest examples, namely iso-contours of a scalar field with uncertainties. In fluid dynamics these are often used, e.g. for vortex identification (utilizing some indicator quantity like lambda2 or Q). A probabilistic formulation leads to fuzzy analogs of crisp iso-contours that can be computed by Monte-Carlo methods. At the end of the talk challenges will be sketched for extending such methods to other spatio-temporal features of scalar, vector and tensor fields.
A short bio is available from here.
The slides can be downloaded here.
Mercredi 21 Septembre 2011
11 h 00 salle de réunion LEA/SP2MI
CNRS Research Director
PPRIME
Le résumé peut être téléchargé ici.
Mardi 20 Septembre 2011
14 h 00 salle de réunion LEA/SP2MI
Professor
Institut de Mathématiques de Bordeaux
Invité par B. Noack
We will briefly review some results in reduced order modeling of classical PDEs relevant to fluid mechanics. Compressible/incompressible and steady/unsteady flow models will be considered and achievements and open problems will be sketched. The discussion will focus on the representation of the solution by a small number of global basis functions and on possible alternatives to this approach. In this sense, one perspective that will be detailed is the use of optimal mass transfer theory to define transport modes that can more accurately model convection, as compared to usual Galerkin projection. Examples relative to model flows like vortex shedding or Korteweg-de Vries equation will be presented.
Angelo Iollo is full professor of applied mathematics at the Institut de Mathématiques de Bordeaux and INRIA Bordeaux - Sud Ouest since 2004. He holds a doctorate in aerospace engineering earned at ICASE, NASA Langley and at the Politecnico di Torino in 1995. Junior lieutenant in the air force, staff scientist at ICASE, recipient of the Marie Curie Fellowship at INRIA Sophia Antipolis, he has been assistant professor of fluid mechanics at the Politecnico di Torino since 1998 and qualified professor in fluid mechanics at the University of Bologna in 2001. His main research interests are modeling and simulation with focus on applied sciences and medicine. He entertains established collaborations with researchers from the academic, clinic and industrial environment (University of Toulon, University of Poitiers, Imperial College, Institut Bergonié, CIRA, Politecnico di Torino, University of Pisa, Boeing, CEA, VALEOL) and is or has been responsible of national and international research projects. He is reviewer for the leading journals in applied mathematics, scientific computing and fluid mechanics and had 15 PhD students. He published over 40 peer reviewed journal papers and is recipient of the scientific excellence prize from the University of Bordeaux.
The slides can be downloaded here.
Mardi 13 Septembre 2011
14 h 00 salle de réunion LEA/SP2MI
ONERA DAFE
Invité par B. Noack
The objective of this talk is to present three different approaches to the stabilization or de-stabilization of unstable flows. First, we consider the case of an oscillator flow, the super-critical open cavity. After briefly recalling the case of a steady forcing targetting directly the modification of the base-flow (Marquet et al. 2008), we present the case of a harmonic forcing modifying the dynamics of an unstable global eigenmode thanks to non-linear interactions. For this, an amplitude equation is sought based on a weakly-non linear approach and a multiple time-scale analysis. The frequency ranges for maximal stabilization or destabilization are sought and validated against a forced Direct Numerical Simulation. Secondly, we consider an amplifier flow, a leading-edge boundary layer characterized by Re=600000 (based on the length of the plate and the upstream velocity). We will first characterize the dynamics of such a flow by computing the singular values/vectors of the resolvant operator: it will be shown that the TS instabilities and Lift-Up instabilities are nicely characterized by such an approach. The effect of a steady forcing on this flow will then be evaluated thanks to an adjoint approach: the gradient of a singular value (and not a global mode!) with respect to the introduction of a steady forcing will show where to introduce a small control cylinder to stabilize the TS and Lift-up instabilities.
Denis Sipp, born in 1972, is a researcher at ONERA since 2002. He is the head of the Fluid Mechanics unit at the Departement of Fundamental and Experimental Aerodynamics. He obtained a PhD degree from Ecole Polytechnique in 1999 on the stability of vortex pairs. He was an engineer at the "Délégation Générale pour l'Armement" from 2000 to 2002, in charge of funding in the aerodynamics field. His primary interest has now moved to control of flow instabilities encoutered in aeronautic applications, such as bluff bodies, cavities, boundary layers, jets, vortices or backward facing step flows. Both open-loop and closed-loop control strategies are considered with a special focus on linearized methods, adjoint techniques, model reduction, optimal control and numerical simulation. He obtained his Habilitation Degree in 2009 at the University Pierre et Marie Curie in Paris. He published 30 articles in the leading peer reviewed journals of fluid mechanics. He is associate Professor at the Department of Mechanics at Ecole Polytechnique since 2003.
Lundi 12 septembre 2011
14 h 00 salle de réunion LEA/SP2MI
Étudiant en post-doctorat
Université Fédérale de Santa Catarina, Brésil
Invité par E. Lamballais
An implicit large-eddy simulation (LES) method based on the approximate deconvolution model (ADM) was used to investigate the noise radiated by a high-subsonic round jet. Unlike eddy-viscosity models, the LES/ADM approach assumes that the subgrid scales may be determined by the resolved flow. This approach also preserves the Reynolds number of the jet, which might not be possible using eddy-viscosity models. The flow variables were solved by the compressible Navier-Stokes equation written in a non-conservative form. The major advantage of using a non-conservative formulation is that it avoids the inconsistent treatment of density weighting (or Favre avareging) of flow variables, commonly used in LES of compressible flows. The large difference of scales between the flow and acoustic field was solved using high-order schemes for spatial discretization and filtering, and a fourth-order Runge-Kutta method for time integration. Boundary conditions were prescribed by a characteristic-based formulation and a conceptual model based on the characteristic analysis. Implicit LES of the well-known test case of a Mach 0.9 round jet at Reynolds number 65000 were performed with a message passing interface (MPI) parallel solver with overlap communication. An extensive investigation of the round jet was performed for the analysis of jet flow dynamic characteristics, such as shear-layer thickness, mean velocity decaying and jet spreading. The present results were found to be in good agreement with previous numerical and experimental data at similar flow conditions.
Jeudi 8 septembre 2011
14 h 00 salle de réunion LEA/SP2MI
PhD student
ONERA Toulouse
Invité par P. Jordan
Dans l'étude du bruit de jet, l'importance des grosses structures turbulentes dans la caractérisation du rayonnement acoustique observé en champ lointain reste au coeur de récents travaux. Ces structures cohérentes, équivalentes à des ondes d'instabilité issues d'une couche de mélange, peuvent être étudiées par la résolution des équations de stabilité parabolisées (PSE). Se basant sur une telle approche, deux thèmes pourront être couverts lors de ce séminaire. La première thématique s'intéressera à l'étude hydrodynamique et acoustique d'un jet double-flux subsonique issu du projet CoJeN. Un deuxième axe de recherche qui pourra être abordé concerne l'utilisation des PSE non linéaires dans une optique de modélisation et de contrôle.
Jeudi 25 août 2011
14 h 00 salle de réunion LEA/SP2MI
Embraer, Brésil
Propulsions Systems Engineer
Invité par R. Manceau
Currently a large usage of computational fluid dynamics and heat transfer (CFD) is found in the Aeronautical Industry in different phases and areas of aircraft design process. Besides traditional external aerodynamic analyses, internal flows and heat transfer computations became also part of the daily activities of CFD engineers. This presentation focus on the use of CFD at Embraer, Empresa Brasileira de Aeronáutica / Brazilian Aircraft Company. After an initial overview of the company and its portfolio of products in the commercial, business and defense aviation, it is presented applications of CFD for both external and internal flows. The methodologies used are described and some of the gains obtained with the use of CFD are discussed. In the end, current challenges in terms of prediction accuracy and multi-disciplinary analyses are posed.
Jeudi 30 Juin 2011
14 h 00 salle de réunion LEA/SP2MI
Professeur
Université de Sherbrooke, Québec, Canada
Invité par Eric Foucault
Des recherches proposent une modification radicale du concept de la ventilation mécanique artificielle en considérant la ventilation liquidienne totale (VLT). Avec cette méthode, les poumons sont entièrement remplis avec un liquide perfluorocarbone (PFC) et un ventilateur liquidien assure un renouvellement cyclique d'un volume courant liquide.
Pour étudier tout le potentiel et les limites de la ventilation liquide, des cliniciens et des chercheurs en ingénierie de l'Université de Sherbrooke se sont associés en une équipe multidisciplinaire dont l'objectif final est de développer un ventilateur liquide pour les soins intensifs néonataux. Après 10 ans de recherche et développement, nous avons acquis une expertise unique en VLT avec notre propre appareil. Un nouveau prototype (Inolivent-5) en construction sera la maquette d'une version pour la clinique.Les activités de recherche du projet répondent aux exigences des cliniciens et à des problèmes existants, mais peu connus, qui sont propres à la ventilation liquide, comme le contrôle du collapsus en expiration forcée, ou la mesure de la dynamique pulmonaire en liquidien. Ainsi, les différents éléments mécaniques, fluidiques, et de contrôle du ventilateur sont spécifiquement optimisés pour assurer un certain niveau de robustesse de performance.
Des ventilations d'agneaux en détresse respiratoire permettent de valider mais aussi de mieux comprendre le potentiel et les limites de cette nouvelle thérapie qu'est la ventilation liquide totale. A court terme, l'application clinique visée consisterait à pratiquer un lavage des poumons moins agressif pour les nouveau-nés avec un syndrome d'aspiration méconial sévère. A long terme, nous considérons la ventilation liquidienne pour tous les âges et d'autres applications cliniques comme l'hypothermie thérapeutique pour la cardioprotection ou la neuroprotection.
Un résumé légèrement plus complet (avec références bibliographiques) peut être téléchargé ici.
Mardi 28 Juin 2011
14 h 00 salle de réunion LEA/SP2MI
Chercheur CNRS
Laboratoire de Mécanique des Fluides et d'Acoustique
École Centrale de Lyon
Invité par P. Comte
Les analyses de stabilité linéaire, appelées "Rapid Distortion Theory" (Cambridge) peuvent être prolongées par des simulations numériques pseudo-spectrales non-linéaires (Orszag & Patterson) qui suivent la déformation de l'écoulement moyen (Rogallo 1981). Ces techniques sont anciennes, mais se sont bien diversifiées et ont même été souvent redécouvertes par des communautés (maths appliquées, astrophysique) utilisant une autre terminologie ... mais les mêmes équations. Après un bref historique, je vais présenter quelques résultats récents dans mon équipe sur deux sujets :
- Ecoulement turbulent soumis à un cisaillement plan, avec effets de rotation, stratification, MHD. Ce qui peut partir d'un modèle simplifié de canal en rotation est maintenant l'objet d'une abondante littérature, en plein essor, sur la stabilité et la turbulence des disques d'accrétion (tournants bien sûr) en astrophysique ! Ce contexte sera évoqué rapidement, avec des résultats semi-analytiques. Je présenterai avec plus de détails, et en utilisant la simulation numérique, comment obtenir une turbulence soumise à l'instabilité barocline, sans effet explicite de confinement, par une combinaison de cisaillement plan, rotation et stratification à taux uniforme.
- Ecoulement tournant avec précession. Le cisaillement plan induit par le couple gyroscopique dû à la précession se combine à la rotation principale pour rendre les lignes de courant moyennes, rectilignes dans le cas précédent, fermées et elliptiques. Une forme particulière d'instabilité elliptique et de turbulence peut alors être traitée avec les outils d'analyse et de simulation, issus de la "Rapid Distortion Theory" et de la technique Orszag-Patterson-Rogallo (terminologie "old fashion").
Et pour finir, nous avons appris la triste nouvelle du décès de Steve A. Orszag, il y a quelques semaines; à ma connaissance, Robert Rogallo est bien vivant mais retraité: on peut donc parler d' héritage scientifique qui continue à fructifier.
Légende de la figure : Dévelopement de la turbulence avec flottaison dans une simulation numérique pseudo-spectrale reproduisant un forçage "naturel" par instabilité barocline.
Jeudi 23 Juin 2011
14 h 00 salle de réunion LEA/SP2MI
Professeur à l'Université de Nice Sophia-Antipolis
Invité par Eric Lamballais
For the simulation of fully nonlinear surface gravity waves, a fast, accurate and robust numerical scheme is presented. The method is based on a boundary integral formulation of the irrotational Euler equations, rewritten in a convenient form, together with a pseudo-spectral spatial scheme and a high-order temporal one. Various applications are presented.
Mercredi 22 Juin 2011
14 h 00 salle de réunion LEA/SP2MI
Associate Professor
Mechanical and Industrial Eng., University of Toronto
Toronto, Canada
Invité par A. Spohn
Le résumé peut être téléchargé ici.
Mardi 21 Juin 2011
14 h 00 salle de réunion CEAT
PhD Candidate
The Ohio State University, USA
Invité par B. Noack
Localized arc filament plasma actuators have demonstrated significant potential in controlling high-speed and high Reynolds number jets in open-loop. Feedback control is being developed for continued optimal performance in the presence of uncertainties in operating conditions. The development proceeds in several steps or modules, building on each other to achieve this objective. In an initial assay, model-free online optimization algorithms have been developed and implemented in experiments. However, the main focus of this work is the reduced-order model-based control strategy. A proper orthogonal decomposition/Galerkin projection strategy has been adopted to devise such a model for the unforced jet from experimental data. Subsequently, a compression wave-based paradigm is proposed to incorporate the effect of the particular actuator. As a parallel module, a dynamic model-based real-time state estimation strategy is developed. Apart from exploiting experimental data, validation was performed on two existing numerical simulation databases. The results show the promise of low-dimensional strategies for predicting and controlling highly turbulent flows of practical interest.
Mardi 7 Juin 2011
14 h 00 salle de réunion LEA/SP2MI
Researcher
Konrad-Zuse Institut (ZIB)
Berlin, Allemagne
Invité par B. R. Noack
The analysis of time-dependent flow fields is a complicated task due to the amount and complexity of the generated data. A promising approach to handle such data is the extraction of structures as well-defined entities, i.e., extremal structures of derived quantities of the flow field. An interesting quantity that helps to identify different kinds of structures is the acceleration. In my talk, I will present a pipeline that enables the robust extraction of flow features based on the acceleration combined with a spatial and temporal filtering.
Lundi 16 Mai 2011
14 h 00 salle de réunion LEA/SP2MI
SRC Simulation Technology, Institute of Hydraulic Engineering (LH2)
University of Stuttgart, Germany
Invité par B. R. Noack
CO2 storage in geological formations is currently being discussed intensively as a technology for mitigating CO2 emissions. The construction of accurate models of underground reservoirs and the reproduction of real flow behaviours using simulation is a very challenging task for the applied research. Once a virtual model has been established, the history matching to the real field becomes an extremely important issue in order to improve the confidence of prediction. First of all, the accuracy of the history matching depends on the quality of the established physical model including all aspects, such as seismic, geological and hydrodynamic characteristics, fluid properties etc. Secondly, the history matching procedure itself is very time consuming from the computational point of view. Even one single forward deterministic simulation may require parallel high-performance computing. This fact makes a brute-force optimization approach not feasible, especially for large-scale simulations. We propose an advanced framework based on a massive model reduction which consists of two main steps. Step one - the original full complex model is projected onto an integrative response surface via data-driven polynomial chaos expansion. This projection is totally non-intrusive and optimally constructed for available reservoir data, using a data-driven orthonormal polynomial basis. Thus, the integrative response surface keeps all nonlinearity of the initial model and incorporates all suitable parameters, such as uncertain parameters (porosity, permeability etc.) and design or control parameters (injection rate, depth etc.). Technically, the computational power for the construction of such a response surface depends on the number of modeling parameters and the expansion degree. However, according to the probabilistic collocation approach, only the minimum number of forward simulations is required, which makes the proposed framework very powerful in comparison to classical sequential approaches. Step two consists of the Bayesian update in order to match the simulation model to available measurements of state variables or other the real-time observations of system behavior, such as pressure at monitoring wells during a certain time period, etc. Practically, we apply particle filtering on the integrative response surface constructed at step one. First of all, particle filtering is a strong technique for the reduction of predicting uncertainty which takes into consideration the nonlinearity of the model. At second, Bayesian updating becomes an interactive task and can incorporate real time measurements due to the fact that even sequential simulation on the constructed polynomials is very cheap. To summarize, we propose a novel framework for history matching of reservoir behavior which takes into consideration the nonlinearity of the model and provides a cheap but highly accurate tool for the reduction of prediction uncertainty.
Keywords: history matching, Bayesian update, particle filters, CO2 storage; multiphase flow; porous media; uncertainty; integrative response surfaces
Lundi 9 Mai 2011
14 h 00 -- 17 h 00 salle de réunion CEAT
Berlin Institute of Technology, Allemagne
Invité par B. R. Noack
Vendredi 6 Mai 2011
14 h 00 -- 17 h 00 salle de réunion CEAT
Berlin Institute of Technology, Allemagne
Invité par B. R. Noack
Mercredi 4 Mai 2011
14 h 00 -- 17 h 00 salle de réunion CEAT
Berlin Institute of Technology, Allemagne
Invité par B. R. Noack
Motivated from recent sucess of model-based cumulant closures for free shear flows, a turbulence closure from analytical turbulence theory is revisited, which has been proposed by Kraichnan in the 50's of the last century: the Direct Interaction Approxmation (DIA). In the talk, the basic ideas, results and extensions are presented and discussed. The talk is separated into three parts. Several classes of turbulence closures are provided via Renormalized Pertubation Theory in the first part of the talk. Subsequently, the basic ideas, results and consequences of the DIA are presented. In the last part of the talk, consistency of extensions of the DIA with turbulence theories like Kolmogorov inertial scaling law and similarities to thermodynamic based turbulence closures are demonstrated.
Mercredi 4 Mai 2011
9 h 00 -- 12 h 00 salle de réunion CEAT
Potsdam University
Senior Lecturer
Group "Statistical Physics / Theory of Chaos"
Invité par B. R. Noack
I will discuss the various formulas for entropies and dimensions and their relation to phase transitions and moments of increments in turbulence. The meaning and interpretation will be given, hopefully as unbiased as possible. The calculation from data is discussed and the advantage/disadvantage of different estimation techniques, as exit-times will be explained.
Mardi 3 Mai 2011
9 h 00 -- 12 h 00 salle de réunion CEAT
Senior Lecturer
Group "Statistical Physics / Theory of Chaos"
Potsdam University
Invité par B. R. Noack
Thermodynamics involves typically infinite-size, infinitely long living systems and the state considered is a stationary one (equilibrium or nonequilibrium). However, the question where a system evolves to in finite time is highly relevant. Some important step has been done by Jarzinski, who developed the finite-time fluctuation-dissipation formalism, further developement has been done by Chetrite and Gawedzki. I will follow their ideas and hint to possible applications. This is a tutorial using stochastic calculus.
Lundi 2 Mai 2011
9 h 00 -- 12 h 00 salle de réunion CEAT
Senior Lecturer
Group "Statistical Physics / Theory of Chaos"
Potsdam University
Invité par B. R. Noack
A typical task for dimension reduction is: given an input from a high-dimensional system, extract the minimal set of necessary variables needed to model the essential features of the system. In recent years such tools have been developed in mathematical statistics, physics and engineering. I will give an overview on the techniques I know about and show some applications.
Mardi 26 Mai 2011
14 h 00 salle de réunion CEAT
Researcher
Konrad-Zuse Institut (ZIB)
Berlin, Allemagne
Invité par B. R. Noack
The analysis of time-dependent flow fields is a complicated task due to the amount of generated data. Thereby, feature-based analysis is one approach to handle this data. The extraction of flow features and their evolution aids the understanding of the many flow phenomena. While the amount of structures can be large for complex datasets, a filtering of the extracted structures helps to focus on specific important structures. In the talk, I will present some approaches that deal with the extraction and filtering of flow features and I will show where the problems of this task are.
Jeudi 31 Mars 2011
14 h 00 salle de réunion CEAT
Responsable Recherche Développement MEMS
FLOWDIT
Invité par J. Borée
Flowdit SAS est une TPE spécialisée dans le prototypage de micro systèmes dédiés au contrôle d'écoulement fluidique. Nous développons des MEMS à finalité industrielle (en série => bas cout de production, robustesse, faible consommation énergétique), à destination des acteurs du transport. L'objectif de la présentation est d'exposer notre savoir faire à travers des exemples concrets de micro systèmes et les prochaines structures que nous imaginons développer prochainement, dans le but d'initier un échange pouvant conduire à des partenariats.
Jeudi 24 Mars 2011
14 h 00 salle de réunion LEA/SP2MI
Post-doc
ISVR Southampton
Invité par P. Jordan
Our inability to identify the physical sources of sound in jets hampers our understanding of the sound generation process. This in turns makes it difficult to reduce jet noise. This work presents a novel post-processing technique that allows us to compute the physical sources of sound in an axisymmetric laminar jet. This technique is compared to more traditional methods used in aeroacoustics to highlight its advantages. It is then used to explain how sound is generated in the flow. This technique has the potential to provide new insights and a better understanding of the noise generation mechanism in jets, which can translate into efficient noise reduction strategies.
Lundi 14 Mars 2011
14 h 00 salle de réunion LEA/SP2MI
Em.O.Univ.Prof. Dipl.-Ing. Dr.techn. Dr.h.c.
TU Vienna
Invité par B. Noack
At heated (or cooled) horizontal plates buoyancy forces act in the direction normal to the plate, giving rise to a hydrostatic pressure perturbation in the boundary layer. The additional term in the boundary-layer equations, though appearing rather simple, leads to remarkable results. The laminar flow over a semi-infinite plate is considered first. Of particular interest is the lower side of a heated plate (or, equivalently, the upper side of a cooled plate), i.e. the case of an adverse pressure gradient. If the plate is adiabatic except in a small region near the leading edge, where the total heat transfer takes place, there are dual self-similar solutions of the boundary-layer equations [1]. If, however, the plate temperature is kept constant, conventional methods for solving the boundary-layer equations fail to give a solution. The failure seems to be unique in boundary-layer theory. In particular, well-established numerical methods break down before the point of vanishing wall shear stress, indicating separation, is attained [2, 3]. Secondly, the horizontal plate of finite length is considered. In this case, both laminar and turbulent flows show peculiarities that are associated with a hydrostatic pressure jump at the trailing edge, which must be compensated by an outer potential flow with circulation in order to satisfy the Kutta condition at the trailing edge. It turns out that the induced lift force not only points in the direction opposite to the buoyancy force, but it is also larger in magnitude. Furthermore, due to the potential-flow singularity at the leading edge there is a buoyancy-induced tangential force in the direction opposite to the free stream, i.e. a thrust force [5, 6]. Numerical solutions of the Navier-Stokes equations, as far as available at present [4], lend support to the results of the boundary-layer analysis, but important points remain open, in particular concerning the stability of the steady-state solutions.
[1] W. Schneider: A similarity solution for combined forced and free convection flow over a horizontal plate. Int. J. Heat Mass Transfer 22 (1979), 1401-1406.
[2] W. Schneider and M. G. Wasel: Breakdown of the boundary-layer approximation for mixed convection above a horizontal plate. Int. J. Heat Mass Transfer 28 (1985), 2307-2313.
[3] H. Steinrück: Mixed convection over a cooled plate: non-uniqueness and numerical instabilities of the boundary-layer equations. J. Fluid Mech. 278 (1994), 251-265.
[4] V. Noshadi and W. Schneider: A numerical investigation of mixed convection on a horizontal semi-infinite plate. Invited paper in: Advances in Fluid Mechanics and Turbomachinery (Eds. H. J. Rath and C. Egbers), pp. 87-97. Springer-Berlin 1998.
[5] W. Schneider: Lift, thrust and heat transfer due to mixed convection flow past a horizontal plate of finite length. J. Fluid Mech. 529 (2005), 51-69.
[6] M. Müllner and W. Schneider: Laminar mixed convection on a horizontal plate of finite length in a channel of finite width. Heat and Mass Transfer 46 (2010), 1097-2010.
Lundi 7 Février 2011
14 h 00 salle de réunion LEA/SP2MI
Assoc. Professor
National Technical University of Athens, School of Mechanical Engineering, Fluids Section
Invité par S. Huberson
Wind turbines have evolved very rapidly over the last 25 years. Starting from 150kW nominal power in mid 80's they increased up to 5MW two years ago. Current design trends indicate that wind turbines could get even bigger in near future. In many aspects also the design tools followed a similar evolution. Starting from simple engineering tools and comprehensive type of design procedures, gradually the tools became rather sophisticated. There is an ever increasing need for accuracy and reliability in design and multi- physics optimization. The aim of the talk is to follow certain aspects of this evolution of the knowledge basis that gave wind energy its leading position amongst all renewable energy sources. We will focus on aerodynamic and aeroelastic problems; outline the basic theories that have been applied and the issues that are still open to research.
Jeudi 6 Janvier 2011
14 h 00 salle de réunion LEA/SP2MI
Chargé de Recherche CNRS
Laboratoire Jean-Alexandre Dieudonné
Invité par L. David
Dans cet exposé, j'expliquerai l'analogie entre la propagation d'une onde hydrodynamique (son, vagues,...) dans un milieu en mouvement et la propagation de la lumière dans un espace-temps courbé par la présence d'un trou noir gravitationnel. J'introduirai la notion de trou blanc ou fontaine gravitationelle. Je décrirai la condition de blocage d'une vague par un contre-courant (apparition d'un horizon blanc hydrodynamique) en tenant compte des effets de profondeur finie, de tension de surface et éventuellement de vorticité. L'apparition d'ondes à fréquence Doppler négative par conversion de mode sera discutée en lien avec le rayonnement quantique de Hawking issu d'un trou noir. Si le temps le permet, le ressaut circulaire sera présenté comme un modèle de trou blanc hydrodynamique.
Laurent Cordier Dernière modification le 31/01/2011