Machine learning for multi-scale modeling of dispersed two-phase flows

IFP Energies nouvelles (IFPEN) est un organisme public de recherche, d’innovation et de formation dont la mission est de développer des technologies performantes, économiques, propres et durables dans les domaines de l’énergie, du transport et de l’environnement. 

IFPEN met à disposition de ses chercheurs un environnement de recherche stimulant, avec des équipements de laboratoire et des moyens de calcul très performants.

Dans le cadre de la mission d’intérêt général confiée par les pouvoirs publics, IFPEN concentre ses efforts sur :

• l’apport de solutions aux défis sociétaux de l’énergie et du climat, en favorisant la transition vers une mobilité durable et l’émergence d’un mix énergétique plus diversifié ;
• la création de richesse et d’emplois, en soutenant l’activité économique française et européenne et la compétitivité des filières industrielles associées.

Machine learning for multi-scale modeling of dispersed two-phase flows

This internship is part of the effort carried out at IFPEN on the multi-scale modelling of dispersed phase flows, a crucial issue for a large number of IFP Energies Nouvelles applications related to the energy transition: microplastics separation, biofuel production, ...

The fluid flows involved in these applications can be modelled with Eulerian formalism for the continuous phase and a Lagrangian one for the dispersed phase. This formalism requires the transport equations to be averaged. This averaging process, which results in more unknowns than equations, requires the introduction of closure laws, especially for the hydrodynamic loads (force and torque) acting on the inclusions. These laws play a key role because they allow the particle trajectories to be accurately predicted (for microplastics for instance).

It appears mandatory to predict as accurately as possible the forces acting on a swarm of inclusions. Although many models predicting the average stresses experienced by the swarm are available, very few models are capable of predicting the stresses acting on each particle according to the local arrangement of the neighbouring particles.

To answer this question, we will use machine learning techniques applied to direct numerical simulation results. Supervised learning methods (ANN, GNN) preserving or not the symmetries of the problem will be used to predict the forces exerted on the particles and their trajectories.

In parallel, we will use the theoretical results available in the Stokes regime (Faxen's laws, the method of images,...), to relate the predictions of the learning model to analytical laws based on hydrodynamic equations.

Your profile

The applicant should have a strong background in fluid mechanics and applied mathematics.

• Knowledge of machine learning techniques would be a plus.

Supervisors IFPEN : Jean-Lou PIERSON, Thibault FANEY
Supervisor IMFT : Jacques MAGNAUDET
Affiliation : IFP Energies Nouvelles (IFPEN)
Where ? IFP Energies Nouvelles - etablissement de Lyon Rond-point de l'echangeur de Solaize
Date: Mars 2023
Salary : 1000 euros brut par mois