Transvalor, a French specialist in materials forming simulation, and the engineering school CentraleSupélec officially launched an industrial research chair dedicated to AI on June 1st. Spanning over 4 years, its goal is to revolutionize the simulation tool in decision-making for manufacturing companies and make Industry 4.0 and digital twins an operational reality in their daily lives.
For companies, numerical simulation has become an essential means of assisting in the design of industrial parts to boost innovation and remain increasingly competitive. It allows for reducing the number of prototypes, detecting manufacturing defects, and significantly reducing development cycles and costs.
Founded in 1984, Transvalor, a French company based in Biot, in the Sophia Antipolis technopole, provides a unique platform of solutions capable of simulating the entire manufacturing process, from raw materials to product properties during use.
Their suite of highly performant simulation software covers a wide range of solid or liquid metal forming processes, as well as polymers, and serves various industrial sectors such as energy, automotive, aerospace, construction, medical, watchmaking, sports, and leisure.
Reducing calculation time to approach real-time
In March 2022, Transvalor and CentraleSupélec, through its French center of excellence in Artificial Intelligence (Hub AI), signed a contract for the preliminary stage of industrial research dedicated to AI. The objective was to explore all the possibilities offered by AI applied to numerical simulation, specifically Transvalor's solutions.
Researchers, doctoral students, post-doctoral fellows, and students from CentraleSupélec, along with Transvalor engineers, have been working on various scientific axes specific to intensive computing, and there have already been numerous advances in terms of proof of concept.
While Transvalor's simulations are known for their accuracy today, they will also become faster, easier to execute, and understand in the future thanks to machine learning (ML) and deep learning.
José Alves, the AI referent at Transvalor, explains:
Machine learning will help reduce calculation time to approach real-time computing, a key factor in digital twins. One of the key axes will be research on hybrid methods that combine raw data with physical knowledge. This will ultimately allow for the creation of more robust, reliable, and, above all, generalizable models.
The designs received from the design office will enable the design engineers of Transvalor's client companies to understand optimal industrial processes better and then explain the necessary process adaptations to the manufacturing technicians.
Previously, determining the lifespan and behavior of a new material required numerous physical tests, and the results were analyzed by an engineer for months or even longer.
Thanks to machine learning, data, or digital twins, it is now possible to predict and generalize the best industrial practices. In addition to accelerating calculation speeds, AI helps reduce waste and improve energy efficiency in manufacturing processes. This allows industries to participate in the circular economy and reduce their carbon footprint.
Shared interests
The collaboration between the two partners last year aimed to optimize Transvalor's solutions.
Stéphane Heitz, CEO of Transvalor, states:
All our employees have benefited from machine learning education thanks to CentraleSupélec. In addition, around thirty of them have received more advanced technical training to participate, on the one hand, in the development of algorithms with the school's researchers and, on the other hand, to introduce these new technological solutions into all our software eventually.
CentraleSupélec, one of the founding members of the DataIA Institute, has structured its AI activities through the creation of the Hub IA in 2020. With 80 researchers and 150 doctoral students, the Hub IA coordinates all the school's initiatives in this field and facilitates innovative projects.
CentraleSupélec's expertise in AI, at the intersection of mathematics and computer science, promotes the acceleration of applied research in the industrial sector. The creation of this chair will allow them, among other things, to confront innovative methods in industrial applications of the French SME.
This synergy of respective skills thus represents a major opportunity for the future industry and fully aligns with the France 2030 plan.
Frédéric Magoulès, the chair holder, concludes:
A majority of Transvalor's members come from research, enabling our research teams to interact in this partnership on complex industrial applications effectively. This chair also offers our students the opportunity to work on projects for major industrial players, which is a great way to showcase their work.