Postdoctoral Position At Lheea (Ecole Centrale Nantes): Innovative Hybrid Numerical Method In Cfd

Context

Numerical simulation is an essential tool for the design of marine structures, both to improve the performance of maritime transport and in the field of marine renewable energies. These topics necessarily involve taking into account complex mobile geometries within flows. However, this operation remains today strongly limited by the technical difficulties posed by the meshes, as well as by the monolithic nature of the numerical methods used. This research work aims, in the long term, to create an innovative solver, a technological breakthrough with the digital tools currently available:

- Elimination of technical difficulties and engineering time related to the generation of meshes around moving geometries in interaction with the fluid

- Response to the problems posed by the large differences in temporal and spatial scales (e.g. taking into account both turbulent boundary layers and distant wakes; rapid and precise modeling of swells while capturing very local impact problems)

- Taking advantage of the qualities of the different hybridized methods while overcoming their respective weaknesses

- Unification of different solvers already developed within the LHEEA research dept.

Objectives – Scientific content

This work is part of a continuation of the expertise of the LHEEA around different numerical methods for hydrodynamics, and in particular CFD methods on Eulerian grids (Finite Differences, Finite Volumes with adaptive refinement), Lagrangian particle methods (Smoothed Particle Hydrodynamics), and potential methods (High Order Spectral method). The aim here is to take advantage of the complementarity of these methods, by intensive use of coupling and ALE formulation (Arbitrary Lagrangian-Eulerian) in order to achieve a unified numerical tool, while abolishing the inherent difficulties to the implementation of meshes around complex geometries in interaction with the fluid. To do this, the following points will be explored and combined:

- Automatic consideration of complex geometries with "simplified" adapted meshes

- Generalization of spatial interpolations: hybridization of resolutions on Eulerian grid / Lagrangian particle resolution, with ALE relaxation zones

- Extensive use of adaptive grid/particle refinement (AMR/APR)

- Intensive parallelization of the solver on distributed memories, deployable on several thousand CPU cores

The position is for 18 months starting December 2022 or as soon as possible thereafter.

Applicants are asked to send a short cover letter, CV and details of few relevant publications.

Contact: Guillaume Oger ([email protected]), David Le Touzé ([email protected])

- PhD in fluid mechanics

- Experience in numerical methods ; CFD Methods

- Algorithms and High Performance Computing (HPC)

- C/C++ or Fortran

- Good English level