Research project title

Artificial intelligence to accelerate the numerical simulation of large-scale superconducting tapes

Education level

Master or doctorate

Director/co-director

Director: Frédéric Sirois

End of display

December 31, 2026

Areas of expertise

Modelling and simulation studies

Superconductors

Primary sphere of excellence in research

Innovative Materials

Secondary sphere(s) of excellence in research

Modeling and Artificial Intelligence

Energy, Water and, Resources

Unit(s) and department(s)

Department of Electrical Engineering

Conditions

The project is aimed at people who have completed a bachelor's degree in computer engineering, materials engineering, electrical engineering, engineering physics or a related discipline, and who are interested in the development of models based on artificial intelligence to simulate the electrothermal behavior of superconducting tapes.

Detailed description

Project's summary

High critical temperature superconducting tapes are now manufactured industrially over lengths of several kilometers. This allows considering building large-scale applications such as superconductive magnets for medical applications such as proton therapy, energy applications such as nuclear fusion, transport applications such as electric motors for aeronautics, etc. One of the problems with superconducting tapes is that their physical properties can vary along their length over distances of less than one millimeter, which can lead to the presence of potentially destructive hot spots. One strategy for finding solutions to this problem is to simulate the complete electrothermal response of these long tapes with a fine level of detail for their physical properties. However, carrying out these simulations requires considerable computation time.

Recent results have made it possible to achieve enormous speed-ups in calculation times (of the order of several thousand times), thanks to a judicious application of machine learning techniques. The calculations carried out until now have mainly concerned simple superconducting tapes, and we are now seeking to extend these calculations to the scale of complete superconducting coils, in order to take advantage of the approach in real designs of high-field electromagnets. Our team is therefore looking for a student at the master’s-research or doctoral level to help us achieve this objective. It is very likely that the project will be done with a partner such as the Atomic Energy Commission (CEA), based in Saclay, France.

Main responsibilities

The selected candidate will have to be perform his duties with creativity and dynamism. In particular, he/she will:

• Learn how to create and optimize finite element datasets to train machine learning models;
• Generalize the simulation model by machine learning to make it applicable to real applications;
• Participate in meetings (with supervisors, group meetings, teleconferences with the partner, etc.), as well as other project-related activities;
• Prepare technical reports and write scientific articles.

Interests and skills sought

• Knowledge (or at least a marked interest) in modeling and numerical calculation;
• Programming skills and interest in scientific computing;
• Interest in artificial intelligence and machine learning;
• Interest in laboratory experiments, which will be used to validate the models developed (it is possible to get help from other students if this aspect is unknown to the candidate).

Financing possibility

This project can only be carried out if the candidate has an academic record competitive enough to obtain a scholarship in a competition of excellence, for example at NSERC (Canadian citizens only), at FRQNT, or in a program such as IVADO. It is also possible that funding could be provided by a partner. Contact the professor to discuss it further.