April 18, 2019

Fuel BU boosts technological innovation with its “Free to Innovate” initiative

What is the “Free to Innovate” initiative?

It’s all in the title: the founding principle of “Free to Innovate” is to give personnel the freedom to develop new ideas.

“Free to Innovate” was initiated by the Fuel BU in June 2018 and covers all of the unit’s activities—engineering, services, as well as fabrication—across the world scope.

The innovator is central to the initiative: at any time, a person with an idea about the technology can decide to spend time analyzing the feasibility of this idea and can call on his or her colleagues or other divisions of the BU.

The innovator can draw on local contacts and designated promoters within each technological domain. Once the feasibility study is finished, the promoter saves the idea in a database, and if it’s suitable for development, the launch of a project in the R&D program is validated.

By the end of March, 40 ideas had been collected as part of the BU’s development program. The idea of Thibaut Clouteau, a Mechanical Design Engineer, is an excellent illustration of the first results of Free to Innovate.

Thibaut, in 10 words or less, what is your idea?

To predict assembly deformations using artificial intelligence (AI).

Why did you use “Free to Innovate”?

Not a day goes by that AI isn’t mentioned by the media: face recognition, targeted advertising, self-driving vehicles, and so on. My hope was that these new technologies would help to address a complex mechanical issue: assembly deformations. But I didn’t have the time to develop this idea. Free to Innovate gave me the opportunity to devote 40 hours to developing the idea.

What is the advantage of your idea?

This idea involves developing a predictive AI model based on more than 30 years of core operating experience. What does this model do? It learns. We give it all available information on the assemblies: position, deformation at the start and end of the cycle, burnup, and design. The model learns; that is, it adjusts all of the coefficients of the regression functions that comprise it so that its error on the end-of-cycle deformations is minimal. Once the learning is over, we can use this model to predict the deformations in future cycles.

There are multiple benefits: analyses based on real measurements, instantaneous calculations once the AI models are trained, and complementarity with existing physical models.

Specifically, what has “Free to Innovate” done for you?

“Free to Innovate” is the “Here and Now” of innovation: the opportunity to work immediately in a well-defined context, with broad visibility for my colleagues specialized in the field, in particular in Germany and for R&D Leads.

Thanks to “Free to Innovate,” we now have a distinct project in the Fuel R&D program to study the robustness of the AI model and eventually apply it to predicting the deformation of our core assemblies. There is also the possibility of testing the efficiency of solutions already in place and continuing to improve our performance.