DigITER: a program for the past, present and future
ITER is all about building a large tokamak that will demonstrate the feasibility of fusion energy. The project's true "end product," however, is not the machine. It is knowledge. Over the past 35 years, beginning with conceptual design and engineering design activities, then through construction, manufacturing, assembly and pre-commissioning, ITER has already generated an immense quantity of data. Millions of documents have been created on different media and organized using methods and technologies available at the time of their production. Turning these precious bits and pieces of accumulated data into accessible information now requires an in-depth digital transition, which every company or institution engaged in long-term projects needs to implement. In late 2021, a working group inside ITER established a roadmap for such a transition, and in autumn 2022 DigITER was officially established.
Like in all institutions and companies whose projects span several decades, ITER data is present and preserved ... but it is scattered, ramified, fragmented and not always easy to locate. In late 2021, the ITER working group compared notes with Airbus, Rosatom, the French utility provider EDF, and the Korean Institute of Fusion Energy; since, contacts have been established with CERN and Tesla. All have faced a similar challenge and all have devised a similar approach: data needs to be captured and stored in structured and centralized data repositories. "The digital storage capacity available today allows us to capture every single bit of available data, including the 'soft data' that might appear irrelevant today but might prove useful tomorrow."
Bécoulet explains the nature of "soft data" with the following example: "Let's take the ballpoint pen you're using to take notes. We can render its dimensions, volume, etc., with a 3D drawing, add information about its material and, since storage capacity is not an issue, mention historical elements that explain why the colour blue was chosen for your ballpoint cap. Is it important? Maybe not today. But one day, for a reason we cannot anticipate, it could be—that's what soft, or meta, data is about. It's like a tag that can be attached to any 3D object. And we can attach as many tags as we wish."
The ballpoint cap example can be extrapolated to any ITER component or system however large or complex. Combined with the capacity of digital tools to capture and render massive amounts of data, it opens the way to creating "digital twins" for any ITER environment.
"In ITER's Virtual Reality Room today, we are able to access 3D renderings of components and systems. But outside of a few exceptions, they are 'as-designed' and not 'as-built,' and they don't include meta data attributes. Using today's technologies it is becoming quite simple to scan a complete environment, virtually recreate it as it is in the real world, and connect it to 'soft' attributes.
Artificial intelligence will likely play a key part in the DigITER digital transition, both as the designated tool for "questioning" the databases and for mitigating the risk of human error. "A process that involves actors from different backgrounds and cultures, each operating within a certain mode and mindset, is prone to generating errors. In ITER, like in any big and complex project, the consequences can be dire."
Knowledge relies on access to organized data, and this is precisely what the DigITER program boils down to. "Data management is at the core of such activity," stresses Bécoulet. "By holding on to data, we're holding on to everything else." The "data continuity" project, aiming at consolidating the structure of ITER engineering data, is under active elaboration. A full year will be needed to define the new data architecture and another one to proceed to what Bécoulet calls a "re-digitalization" of all of ITER's accumulated data.
Building on ITER's current capabilities, three projects within DigITER have also been given priority status: the digitalization of as-built data and the construction of 3D models; an increase in the use of virtual and augmented reality for design, validation, control, simulation and training; and the introduction of digital tools for the advanced engineering design of the future Hot Cell Facility.
"DigITER represents a colossal amount of work," says Bécoulet, "but the tools now exist to make it possible." Following more than one year of detailed investigation, benchmarking and the production of a roadmap by the working group, the program is now ready to launch.