Subject
Suspension Plasma Spraying (SPS) is an emerging industrial process, particularly for the creation of ceramic coatings resistant to thermomechanical stresses, used as long-life thermal barriers for aircraft engine. For the aeronautics industry, it is classified as a special process whose output elements can only be verified by monitoring or post-measurement, and whose deficiencies therefore only become apparent once the product is in use. In this process, the liquid suspension containing the submicron particles of the material to be deposited is injected into a thermal plasma jet to be fragmented and evaporated, releasing individual or agglomerated submicron particles that are then accelerated and melted and will impact and spread over the part to be coated to form a coating. The structure of the coating is a function of the operating conditions, from the plasma torch to the droplet impact conditions (shape, velocity, temperature, and substrate roughness). A dense or columnar structure may occur, which influences the final thermomechanical properties of the material. A full CFD simulation of the entire process is beyond reach due to limitations in the number of particles that can be simulated. Therefore, we propose a three-step approach, consisting of CFD simulations at the droplet scale combined with a stochastic approach [1] enriched by AI at the coating scale:
- The stochastic approach aims to represent realistic spray conditions (spatial, temporal, radius, velocity, and temperature distributions of the particles).
- Simulations of droplet impacts using the CFD code Notus [2] aim to populate a database representing the topology of various instantaneous representative sprayed surfaces.
- A neural network-based on CFD results aims to surpass CFD simulations capacity by representing large impact surfaces and amounts of particles. The AI tool's results can be verified and refined through additional CFD simulations.
The proposed postdoctoral position focuses on strengthening the project’s CFD component by using and improving the Notus code. It is a numerical framework dedicated to simulating incompressible and low-Mach flows on massively parallel architectures. Built on a finite-volume method on Cartesian grids, it integrates advanced interface-handling techniques (VOF-PLIC, surface tension, immersed boundaries) and is designed for easy extension and rapid testing of new numerical schemes.
In this project, complex geometries are treated using first-order volumetric penalization, which remains the preferred approach for handling arbitrary substrates and solidified particle layers. Notus already includes tools for computing solid volume fractions from OBJ geometries, and new developments will address moving obstacles, required for SPS processes.
A major objective is to port Notus to GPU architectures to increase domain size and particle count. This involves introducing loop-level parallelism with OpenACC (and potentially CUDA), and evaluating GPU-enabled sparse linear solvers. Notus is already interfaced with Hypre, which offers high-performance GPU support, and comparisons with PETSc, AMGx, and others will be performed. Development will start on a single-GPU workstation, then extend to multi-GPU systems in national computing centres (NVIDIA, AMD).
Research context
3-years PhD at I2M-Bordeaux; starting date: January 2026; French National Agency ANR funding; academic collaboration with IRCER laboratory & industrial collaboration with Safran Tech.
Expected skills
Computational Fluid Mechanics, Fortran, OpenACC, experience with GPU architectures and solvers
Contact
Please send a detailed CV, your PhD dissertation/review, and reference contacts. Please send these documents to all the following contacts:
Stéphane Glockner I2M-Bordeaux : glockner@bordeaux-inp.fr
Cédric Le Bot: cedric.lebot@bordeaux-inp.frReferences
[1] M. Xue et al 2008, A stochastic coating model to predict the microstructure of plasma sprayed zirconia coatings, Modelling Simul. Mater. Sci. Eng., 16 065006.
[2] Notus CFD code : https://notus-cfd.org
[3] Jingzu Yee, Daichi Igarashi, Shun Miyatake and Yoshiyuki Tagawa, Prediction of the morphological evolution of a splashing drop using an encoder–decoder, Machine Learning Science and Technology, 2023.
[4] M. Giselle Fernández‐Godino, Donald D. Lucas & Qingkai Kong Predicting wind‐driven spatial deposition through simulated color images using deep autoencoders. Scientific Reports, 2023
[5] Morgane Suhas, Modèles de comportement et loi de défaillance de systèmes enrichis par les données, thèse de l’École doctorale Sciences des métiers de l'ingénieur de l’ENSAM, 2025.
