A FNRS research fellowship for Louis Denis : Numerical models of large scale superconducting magnets for the newest generation of particle accelerators
Cursus
In September 2018, Louis Denis began studying civil engineering at the Faculty of Applied Sciences at ULiège. His interest in physics developed fairly quickly and it was quite natural that he turned to the Engineering Physics Master's degree, which in his eyes offers a perfect balance between theoretical physics, numerical development and experimental methods. During his studies, Louis had the opportunity to work as a student instructor on a number of courses, including “Finite Element Method”, “Mesures électriques : fondements et applications” or “Mathématiques Appliquées”, among others.
In June 2023, Louis obtained his degree in Civil Engineering Physics with Summa Cum Laude and the Congratulations of the Jury after completing his Master’s thesis under the supervision of Prof. Benoît Vanderheyden and Prof. Christophe Geuzaine. His thesis focused on the magnetothermal study of a superconducting coil in a carbon therapy cyclotron magnet, in collaboration with Ion Beam Applications (IBA) in Louvain-la-Neuve. As part of this project, Louis had the opportunity to present some of his results at the 14th International Particle Accelerator Conference (IPAC), held in Venice in May 2023.
In October 2023, he joins the Applied and Computational Electromagnetics (ACE) department, where his FNRS fellowship will allow him to devote himself fully to the study of numerical models for superconducting magnets, under the supervision of Prof. Christophe Geuzaine and Prof. Benoît Vanderheyden. Louis' research is carried out in close collaboration with Dr. Mariusz Wozniak's team at CERN, where several research visits are planned during his doctoral thesis.
Research
The accelerator physics community is currently designing the next generation of high-field magnets with high-temperature superconductor (HTS) technology, using them for instance as coils inserted in the bore of conventional low-temperature superconducting (LTS) magnets. A critical issue is to predict the effects of the magnetic forces induced in the inserted HTS coil by a sudden quench of the surrounding LTS magnet (called outsert). Due to the multiscale structure of the HTS magnets and their intricate three-dimensional geometry, a standard finite-element model based on a brute-force approach is currently not an option, which calls for the development of a specific modelling approach based on recent model reduction and acceleration techniques.
The goal of Louis’ research is to exploit advances in multiscale analysis to develop numerical tools able to predict the response of large-scale HTS inserts, under various magnetic field excitations and in particular in case of a sudden quench of the LTS outsert. To this end, an accurate homogenized model is constructed, with a coarse-grained macroscopic model of the magnet being coupled with mesoscale models containing only a few conductors, which are based on recently developed thin-shell approximations to reduce the number of degrees of freedom, while exchanging information with the macroscopic model in a bidirectional way. Neural networks are also investigated to construct efficient lightweight surrogate models, after having learned the input-output relationships of the mesoscale models. Finally, domain decomposition methods are studied to scale up simulations to realistic problem sizes.
The expected outcome of Louis’ research is a tool able to accurately predict the behaviour of large scale hybrid HTS-LTS magnets, and to efficiently investigate protection and security measures to be implemented against sudden quenches of the LTS magnet.
