Le vendredi 21 avril à 11h00, salle de séminaires IRPHE
Abstract : During the last two decades, the understanding of subcritical transition in shear flows has significantly progressed thanks to the dynamical systems perspective. In this respect, the concept of edge states, which correspond to relative attractors on the basin boundary of the laminar fixed point, have proven to be particularly illuminating.Despite this success, little attention has been paid to these developments in the magnetohydrodynamic (MHD) community, where subcritical transition can be expected in both applications involving liquid metals and astrophysical contexts. For ducts and channels with electrically insulated walls subject to a magnetic field, the Lorentz force inhibits motion in the directions orthogonal to the field, whereas the velocity component parallel to the field is unaffected. For a doubly periodic channel with a spanwise magnetic field, this manifests itself as a preference in the system for structures that are elongated in span. Such anisotrophy naturally impacts both streamwise streaks and rolls, which implies that the traditional picture of edge states in hydrodynamic channel flows involving periodically bursting structures is modified. For a duct with four walls on the other hand, the presence of a transverse magnetic field leads to the formation of Hartmann and Shercliff layers along the walls that are orthogonal and parallel to the field, respectively. Traditionally, transition to turbulence in such flows has been characterized by a critical Reynolds number based on the Hartmann layer thickness. On the contrary, recent direct numerical simulations (DNS) suggest that turbulence first appears in the Shercliff, instead of in the Hartmann layers5, which implies a slight contradiction in literature.
In this seminar, the concept of edge states originally developed within the dynamical systems community will be introduced, along with the quasi-static MHD approximation valid for low magnetic Reynolds numbers. Subsequently, the above mentioned classical MHD geometries will be revisited and the outcome of non-linear high-order DNS be presented. Specifically, the impact of different Hartmann numbers on the edge states in MHD channel flow, as well as the interaction between the different boundary layers and possible transition routes for MHD ducts with square cross-section will be discussed.