The NextSim interview series aims to introduce readers to the people behind our research. Marc Montagnac, a CFD expert at CERFACS, is involved in NextSim´s work on the improvement of algorithms for efficient solution process and for data management, visualization and modelling. Learn more about his thoughts and motivations regarding his work and the NextSim project.
What is your position at CERFACS and what is your research area?
I am currently a research engineer in the Computational Fluid Dynamics (CFD) team at CERFACS in France. CERFACS is a research centre specialized in modelling and numerical simulation in the aeronautics, space, climate, energy and environmental fields. In this respect, the CFD group develops numerical methods and tools to solve major scientific and technical research problems of public and industrial interest.
My overall ambition is to foster high-performance computing in the aeronautical industry and to improve the CFD capabilities towards more efficient and accurate design tools in aerodynamics.
My main activities focus on the research, development, and optimisation of advanced numerical methods in performance portable software to leverage cutting-edge high-performance computing (HPC) facilities to perform challenging CFD simulations.
In this context, I am involved in national and European research and industrial projects, in collaboration with the main players in the field of aerodynamics, namely Airbus, ONERA, and DLR.
You are involved in the improvement of algorithms for efficient solution process and for data management, visualization and modelling. Can you explain why your work in these areas is important for NextSim?
There is a general consensus that simulation tools in general, and industrial simulation tools in particular, do not exploit the full capacity of HPC systems. This is especially true as HPC architectures evolve at a faster rate than is necessary to adapt software to a new hardware architecture. EuroHPC’s NextSim project is an important step towards solving this problem.
All this helps to understand why improving the algorithms and the overall CFD workflow is of the utmost importance to cope with exascale computers. To this end, CERFACS develops state-of-the-art high-fidelity numerical features in the next-generation CFD solver CODA developed by ONERA, DLR, and Airbus. This kind of methods exhibits good properties in terms of accuracy but also in terms of computing performance compared to older techniques.
On top of that, massive CFD simulations generate a huge amount of data that has to be analysed during the simulation itself and at the end of the simulation. CERFACS also studies different methodologies that can be applied efficiently when using exascale computers.
What would you have done if you were not a researcher?
Perhaps I could have worked in the field of Information & Technology since I was interested in computer science at an early stage. But I also found applied mathematics and fluid dynamics to be very attractive areas of science. All this added to the desire to build practical tools made me choose a career as a research engineer. It is still striking to me that with mathematical modelling of physics and computers, we can get close to phenomena that occur in real life. I had the opportunity to join CERFACS where all the people share this common passion to take on ambitious challenges in these scientific fields and that is why I have not asked myself this question for 20 years now.
Have you encountered any challenges in pursuing your research career?
After my PhD, the first challenge was to fully enter this career at a time when the job market was difficult and competitive. Secondly, the challenges I faced are the classic ones in the research field. The concerns are to get financial support for research activities and otherwise to adapt to new strategic orientations of my research centre.
How is your experience with NextSim so far?
The experience has been very positive so far. Even with virtual exchanges, CERFACS has established very fruitful collaborations with all partners, DLR, ONERA, and Airbus for the CODA software implementation and the simulation of the nozzle jet noise flow, BSC for the high-fidelity methods and the Large Eddy Simulations, and UPM for the data post-processing and visualisation aspects. There is no doubt that these collaborations will continue until the end of the project. I am confident that this joint research effort will pay off and have a significant impact on the design and development of numerical methods relevant to the application of CFD in future exascale machines.
What do you hope NextSim can achieve beyond the life of the project?
In a short-term period, thanks to all the innovative work coming out of the EuroHPC NextSim project, I hope that the next-generation CFD solver CODA will become a major software for performing large-scale CFD simulations on European HPC infrastructures.
In the longer term, I am confident that the NextSim project will have strengthened the relationship between all the partners, BSC, CIMNE, UPS, ONERA, DLR and that this relationship will serve to address future challenges of the aeronautical industry.
What advice would you give young researchers who would like to follow in your footsteps?
Be open-minded and ready for a long trip. The research process is made up of small steps towards a goal and the path to get there is full of pitfalls. Even a small step may require significant effort to achieve convincing results. Thus, it is also necessary to be lucid enough to abandon a presumed false direction and open up other directions at certain times. But hard work pays off in the end and you have to value it.