DAEP 2020 seminars

Energy transfer and non-linear interactions between a canopy of cubes and the boundary layer above

• Tuesday, November 17, 2020 - 11:00 a.m. - room 38.250 and zoom - by Karin Blackman

Previous work in the smooth wall boundary layer has identified a nonlinear relationship between large-scale structures occurring in the boundary layer and small-scale structures near the wall (Mathis et al., 2011a). Blackman and Perret (2016) used experimental data from boundary layers growing over rough walls (consisting of 3D and 2D features) to study nonlinear interactions. They found a top-down mechanism similar to amplitude modulation in all roughness cases. Here we attempt to shed light on the link between the nonlinear amplitude modulation mechanism demonstrated for a boundary layer above a canopy of cubes, and the energy transfers between large and small scale structures. To this end, the triple decomposition method is applied to the turbulent kinetic energy balance equation in order to analyze the dominant terms in the energy transfers and to evaluate the interactions between scales.

Research on understanding and modeling the Bypass transition

• Thursday, November 5, 2020 - 11h00 - room 38.137 Hyflex - by Olivier Vermeersch (ONERA)

Aerodynamic and acoustic study of contra-rotating doublet propulsion configurations

• Monday, November 2, 2020 - 11h00 - room 38.250 and zoom - by Maxime Fiore (Cerfacs)

Turbulent modelling in wedge separation

• Tuesday, October 27, 2020 - 11:00 am - room 38.250 and zoom - by Jean-Francois Monier

Optimal perturbations in inhomogeneous round jets

• Friday, June 5, 2020 - 11:00 am - visio - by Gabriele Nastro

The round jet represents an ideal flow for studying the transitional dynamics of certain families of flows, such as the jet downstream of an engine injector. In this context, the understanding of the transitional mechanisms downstream of the jet is an essential initial step for the development of innovative injector concepts. In particular, when the ratio between the density of the jet and that of its environment is sufficiently low, the flow is characterized by the appearance of secondary lateral jets which cause a significant flare of the main jet (see figure). In this respect, Lopez-Zazueta, Fontane & Joly (2016) showed for the variable density mixing layer that the mechanism proposed by Monkewitz & Pfizenmaier (1991) for a homogeneous jet, based on the development of pairs of contra-rotating longitudinal vortices in the braid zone, is replaced, in an inhomogeneous situation, by the appearance of longitudinal velocity striations of alternating signs.

In this seminar, we will propose the description of the methodology and the analysis of the dynamics of optimal perturbations. In particular, this study aims at analyzing the two- and three-dimensional secondary instabilities of the variable density, incompressible, high Froude number round jet, in order to verify the robustness of this transition mode in the case of inhomogeneous round jets, and thus to definitively rule on the validity of the mechanism at the origin of the lateral jets.

Study of the operability of a high bypass ratio fan under unsteady distortion

• Friday 7 February 2020 - 11h00 - room 38.137 - by Amaury Awes-Cheynis

Inlet distortion can significantly impact turbofan operability. In the present contribution, the focus is on vortex ingestion, which is a type of distortion that has received less attention in the literature for high-bypass-ratio fans of civil engines than inlet separation or boundary layer ingestion. Due to the unsteady and non-axisymmetric nature of the inflow, full annulus computations are performed, using an unsteady RANS approach with a sliding mesh interface between the rotor and the stator. The goal is to provide a detailed analysis and understanding of the mechanisms responsible for operability loss when a vortex is ingested by the fan. To that end, a simplified model of vortex is derived from previous simulations in crosswind conditions including the presence of the ground plane. The study test case is a turbofan demonstrator representative of modern turbofans. Simulations are run with and without vortex at several operating points until rotating stall is observed. The global results show that the vortex ingestion induces a stall margin loss of up to 8%. Modal analysis indicates a stall pattern with 10 cells rotating at somewhere between 52 and 67% of the shaft speed. A comparison of the two configurations shows that the stall inception mechanism is similar with and without distortion. In particular, examining the incidence at the fan blade, a similar critical angle is observed for the two configurations, which is leveraged to establish a stall criterion valid for vortex ingestion cases.

Simulation of violent transients and pulsating flows in turbines

• Friday 24 January 2020 - 11:00 am - room 38.137 - by Florian Hermet(Presentation)

Pulsed flows in turbines have many industrial applications (turbocharger, Humphrey’s cycle, ...) However, the influence of pulsed flow on turbine performance is complex and imperfectly understood. There is no clear consensus in the literature regarding generic trends. Those are difficult to extract since a majority of studies are based on industrial geometries (mainly turbocharging radial turbines), which involve strong interactions between many physical phenomena. In order to isolate and understand the main expression of the physics, the problem needs further simplifications, and a clear assessment of the relevant time scales, and performance indicators of the problem. The seminar proposes a review of the research conducted on that purpose which led to consider the academic case of a severe transient regime inside a thin airfoil cascade, caused by a sudden change of inlet conditions. Simulations with increasing complexity (from 1D Euler to 3D LES) are deployed in order to isolate the driving physics.

An Introduction to Deep Reinforcement Learning for Fluid Mechanics

• Friday, December 10, 2019 - 11:00 am - room 38.051 - by Sandrine Berger

Recent publications have demonstrated the potential of Deep Reinforcement Learning (DRL) techniques to solve control and optimization problems in fluid mechanics. In this scope, our work aims at developing an innovative control strategy of transonic buffeting, relying on the coupling between unsteady Reynolds Averaged Navier-Stokes (URANS) simulations and DRL. The target application is a diamond airfoil designed for microsatellite-dedicated launchers and for which experimental and numerical investigations of the aerodynamic and aeroelastic behaviour were performed by Jeromine Dumon at ISAE-SUPAERO. Nevertheless, the CPU cost associated to this application is particularly high. A preliminary study is hence conducted on a flying flat plate for which low order models of increasing complexity can be considered to decrease the CPU cost of the simulation. The seminar talk will first introduce the key concepts of Deep Reinforcement Learning before proposing a quick review of its application for fluid mechanics. Finally, the first results of the flying flat plate preliminary study will be presented.