[LUM#22] Solid predictions
Reactor containment, cooling towers, spent fuel storage pools... concrete is everywhere in a nuclear power plant. And it is subject to numerous stresses: radiological, thermal, chemical, hydraulic, and mechanical. To predict the watertightness of these facilities, Yann Monerie, a researcher at the Mechanics and Civil Engineering Laboratory, uses artificial intelligence.
More than 20 million tons of concrete are produced in France each year to build buildings, dams, bridges, and even nuclear power plants. These are high-risk facilities where this mixture of sand, cement, gravel, and water acts as a containment barrier. But how can we know if this enclosure will remain watertight over time? "The impermeability of concrete is determined by the way it cracks. The more tortuous the path of these cracks, the less easily fluids flow through them, and therefore the more impermeable it is, " explains Yann Monerie, a researcher at LMGC.1.
Tortuosity
To determine this tortuosity, it is first necessary to know the properties of the various components of the mixture, but also to understand how the aggregates are distributed in the mortar. "Concrete is not homogeneous, and the spatial distribution of its different phases determines its cracking, " explains the specialist in micromechanics of materials, who is collaborating on this issue with the Institute for Radiation Protection and Nuclear Safety.
This is because cracks do not spread randomly, but evolve gradually from one gravel particle to another . "The interfaces between the aggregates and the matrix are weak areas that will determine the spread of cracks." It is this trajectory that researchers want to predict in order to better understand the mechanical properties of this material.
From 4 months to 3 seconds
To make these predictions, Yann Monerie and his colleagues use artificial intelligence. "We have developed a model that allows the statistical reconstruction of the microstructure in three dimensions from two-dimensional sections, " explains the researcher.
After studying thousands of photos of concrete cross-sections and their 3D counterparts, AI enables them to deduce the size and shape of the aggregates and how they are distributed in the cement from simple images. This modeling allows researchers to create a digital simulation of crack propagation. "Without machine learning, it takes about four months of calculations to perform these simulations, but with AI, it takes two to three seconds!"
This predictive artificial intelligence makes it possible to predict the permeability of a power plant based on simple cross-sectional plans of concrete samples. And these applications go even beyond nuclear power: "With this method, we can also determine the watertightness of houses built on polluted soil, for example, " adds Yann Monerie.
Find UM podcasts now available on your favorite platform (Spotify, Deezer, Apple Podcasts, Amazon Music, etc.).
- LMGC (CNRS, UM)
↩︎