{"id":10403,"date":"2026-04-10T16:12:50","date_gmt":"2026-04-10T14:12:50","guid":{"rendered":"https:\/\/ingenius.ecoledesponts.fr\/?p=10403"},"modified":"2026-04-10T16:12:51","modified_gmt":"2026-04-10T14:12:51","slug":"predicting-the-behavior-of-materials-using-artificial-intelligence","status":"publish","type":"post","link":"https:\/\/ingenius.ecoledesponts.fr\/en\/articles\/predicting-the-behavior-of-materials-using-artificial-intelligence\/","title":{"rendered":"Predicting the behavior of materials using artificial intelligence"},"content":{"rendered":"\n\n\n
\n
\"\"
Credit : Jean-Claude Moschetti \/ ENPC<\/figcaption><\/figure>\n<\/figure>\n\n\n\n

AUTOMATIX aims to transform the way in which the behavior of materials is simulated and modeled. What are the conventional approaches used in solid mechanics today?<\/strong><\/p>\n\n\n\n

Describing the behavior of a material generally involves conducting laboratory experiments. The way in which a material deforms when subjected to stress in a given direction is unique to each material and cannot be described by broad, general equations used in physics. This means that materials must be subjected to stress in relatively simple situations \u2013 tension, compression, etc. \u2013 and their deformation measured in relation to the forces applied.<\/p>\n\n\n\n

Tests of this kind have been carried out for a very long time, but remain difficult to perform. Numerous laboratory tests need to be conducted to capture their complexity. The behavior of the material varies depending on environmental factors (temperature, humidity, etc.) and the direction of the applied stress. Mathematical models therefore aim to represent as accurately as possible the behavior of materials observed in laboratory experiments.<\/p>\n\n\n\n

For example, wood, an anisotropic material, behaves differently depending on whether it is stressed along the grain (longitudinal) or across the growth rings (radial). Other materials, such as metals under tension or concrete under compression, undergo changes in their microstructure, notably the formation of voids or microcracks. In such cases, the behavior of the material is not a simple relationship between deformation and stress: it also depends on its history. A material that has already undergone numerous deformations will no longer react in the same way as it did in its initial state.<\/p>\n\n\n\n

\"\"
Automation of material behavior modeling: application to 3D printing. Credit : Provided by the author<\/figcaption><\/figure>\n\n\n\n

How does the AUTOMATIX project differ from these conventional approaches in practical terms?<\/strong><\/strong><\/p>\n\n\n\n

If we take rubber as an example, just two or three parameters may be sufficient in the mathematical model, whereas more complex materials such as soils may require the identification of several dozen different parameters.  Depending on the material, it can therefore take up to a year to develop, calibrate and implement in a software program the behavior that we want to study. With the AUTOMATIX project, we hope to be able to describe a material simply using just a few lines of code. To achieve this, we will use approaches based on artificial intelligence. <\/p>\n\n\n\n

In recent years, the rise of machine learning technologies has gradually replaced mathematical models with learning models. They work by training a neural network using experimental data or data derived from numerical simulations. In the case of the behavior of materials, the first attempts were made using models from other fields, such as image recognition, without fully understanding how they worked. These neural networks, known as \u201cblack-box\u201d models due to their complexity, present a significant challenge. If they are evaluated for mechanical conditions that differ greatly from those used during their training, they can behave in a highly erratic manner and produce completely unrealistic predictions. This lack of robustness is unacceptable if they are to be used for safety-critical applications, such as the design of a dam or a nuclear power station, for example.<\/p>\n\n\n\n

The AUTOMATIX project aims to propose a middle ground between the traditional approach and the more recent \u201cblack-box\u201d approach. Within the very architecture of the neural network, we incorporate controllable elements based on established mathematical models. Any aspects that remain unclear are entrusted to machine learning. In principle, this approach will require less data and computing time and will yield more reliable results.<\/p>\n\n\n\n

<\/a><\/p>\n\n\n\n

The project will be officially launched in June 2026. Although it marks a certain departure from your previous research, you already have preliminary results. Can you tell us about them? <\/strong><\/strong><\/p>\n\n\n\n

I have begun working on methods to describe what is known as the yield surface of materials. When a sufficiently large force is applied to a material such as steel, it never fully returns to its original shape. This is known as plasticity.  This behavior can be mathematically represented by a surface defined in six-dimensional space. Our preliminary results show that it is possible, using a small number of experiments, to automatically reconstruct the shape of this surface without knowing it explicitly.<\/p>\n\n\n\n

This is a key point from an experimental perspective. If we were to characterize the surface using a very large number of points, this would require carrying out as many tests as there are points. This would be very difficult to achieve. The challenge therefore lies in striking a balance between simplifying the model and remaining faithful to the actual behavior of the material.<\/p>\n\n\n\n

The AUTOMATIX project has direct applications in <\/strong>3D concrete printing<\/strong><\/a>, a key area of research at the Navier Laboratory. What are the specific challenges of this emerging technology?<\/strong><\/strong><\/p>\n\n\n\n

3D concrete printing is a technology that uses high-performance concrete, applied in very thin layers, with optimized geometries designed to use as little material as possible. The challenge therefore lies in ensuring sufficient strength with minimal material, without resorting to conventional reinforcements such as steel rods.<\/p>\n\n\n\n

At the Navier laboratory, we have observed that beyond a certain threshold, the printed structures become brittle and begin to crack. To address this, we have chosen to incorporate carbon fibers into the concrete<\/a>. During the printing process, spools of carbon fiber are fed into the print head, and the filaments are deposited simultaneously with the concrete extrusion.<\/p>\n\n\n\n

We have now reached a stage where we wish to present to manufacturers the structures we are capable of producing, but to do so, we must demonstrate their strength. This requires calculations and, consequently, the development of behavioral models for both concrete and carbon fibers. As part of the AUTOMATIX project, data is generated directly on-site, within the laboratory, and involves materials that have not yet been extensively modeled. This makes it a particularly relevant application.<\/strong><\/p>\n\n\n\n

Finally, what does this ERC grant<\/strong><\/strong>1<\/a><\/sup> mean to you, and how will it transform your project and your research team?<\/strong><\/strong><\/p>\n\n\n\n

The main advantage of this grant lies in the scale of the projects it enables. It provides five years of funding for a \u201cConsolidator\u201d grant of up to \u20ac2 million. This provides the opportunity to conceive a large-scale project capable of funding numerous research positions (engineers, PhD students, postdoctoral researchers), as well as the necessary experiments and equipment.<\/p>\n\n\n\n

The ERC grant focuses on scientific excellence, and allows researchers to explore more risky avenues than usual, thereby pushing the boundaries of knowledge. For this reason, I believe it is essential to establish collaborations between colleagues working on similar topics, both in Europe and globally. In addition, we plan to make our tools open-source2<\/a><\/sup> in order to provide the entire community with accessible tools that can be applied regardless of the class of materials being studied. <\/p>\n\n\n\n

Interview conducted by Ad\u00e8le Mazurek, Head of Scientific Outreach at \u00c9cole nationale des ponts et chauss\u00e9es<\/p>\n\n\n

  1. This grant is designed to support early-career researchers (7 to 12 years’ expereince post-PhD) by enabling them to consolidate their research teams and further their career development. \u21a9\ufe0e<\/a><\/li>
  2. Freely accessible to the public \u21a9\ufe0e<\/a><\/li><\/ol>","protected":false},"excerpt":{"rendered":"

    AUTOMATIX aims to transform the way in which the behavior of materials is simulated and modeled. What are the conventional […]<\/p>\n","protected":false},"author":9,"featured_media":10410,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_related_content_post":[],"_related_content_subject":[],"_related_content_author":[],"_related_content_category":[],"_related_content_folder":[],"_excerpt":"Understanding how materials deform, crack and age is a key issue in modern engineering. J\u00e9r\u00e9my Bleyer, a researcher at the Navier Laboratory, is exploring new avenues combining solid mechanics and artificial intelligence. His project, AUTOMATIX, which was recently awarded a prestigious ERC grant, aims to provide the academic and industrial sectors with flexible and powerful digital tools to accelerate the development of safer, more efficient and better-controlled materials and structures.","_duration":5,"_manual_duration":false,"footnotes":"[{\"content\":\"This grant is designed to support early-career researchers (7 to 12 years' expereince post-PhD) by enabling them to consolidate their research teams and further their career development.\",\"id\":\"a2dfa255-e797-491e-a85d-16aee4518c3c\"},{\"content\":\"Freely accessible to the public\",\"id\":\"adb81c9d-decf-47f2-9dbb-71481733e0e0\"}]"},"article-types":[13],"class_list":["post-10403","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","article-types-article"],"has_blocks":true,"block_data":[{"blockName":"enpc\/excerpt","attrs":{"lock":[],"metadata":[],"className":"","style":""},"innerBlocks":[],"innerHTML":"","innerContent":[],"rendered":""},{"blockName":"core\/gallery","attrs":{"linkTo":"none","images":[],"ids":[],"shortCodeTransforms":[],"columns":0,"caption":null,"imageCrop":true,"randomOrder":false,"fixedHeight":true,"linkTarget":"","sizeSlug":"large","allowResize":false,"aspectRatio":"auto","lock":[],"metadata":[],"align":"","className":"wp-block-gallery has-nested-images columns-default is-cropped","style":"","backgroundColor":"","gradient":"","borderColor":"","layout":[],"anchor":""},"innerBlocks":[{"blockName":"core\/image","attrs":{"id":10410,"sizeSlug":"large","linkDestination":"none","align":"wide","blob":"","url":"https:\/\/ingenius.ecoledesponts.fr\/wp-content\/uploads\/2026\/03\/jcm-ENPC-33-1024x683.jpg","alt":"","caption":null,"lightbox":[],"title":"","href":"","rel":"","linkClass":"","width":"","height":"","aspectRatio":"","scale":"","linkTarget":"","lock":[],"metadata":[],"className":"wp-block-image alignwide size-large","style":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    \"\"
    Credit : Jean-Claude Moschetti \/ ENPC<\/figcaption><\/figure>\n","innerContent":["\n
    \"\"
    Credit : Jean-Claude Moschetti \/ ENPC<\/figcaption><\/figure>\n"],"rendered":"\n
    \"\"
    Credit : Jean-Claude Moschetti \/ ENPC<\/figcaption><\/figure>\n"}],"innerHTML":"\n
    <\/figure>\n","innerContent":["\n
    ",null,"<\/figure>\n"],"rendered":"\n
    \n
    \"\"
    Credit : Jean-Claude Moschetti \/ ENPC<\/figcaption><\/figure>\n<\/figure>\n"},{"blockName":"core\/paragraph","attrs":{"style":{"elements":{"link":{"color":{"text":"var:preset|color|red"}}}},"textColor":"red","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color has-link-color","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    AUTOMATIX aims to transform the way in which the behavior of materials is simulated and modeled. What are the conventional approaches used in solid mechanics today?<\/strong><\/p>\n","innerContent":["\n

    AUTOMATIX aims to transform the way in which the behavior of materials is simulated and modeled. What are the conventional approaches used in solid mechanics today?<\/strong><\/p>\n"],"rendered":"\n

    AUTOMATIX aims to transform the way in which the behavior of materials is simulated and modeled. What are the conventional approaches used in solid mechanics today?<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    Describing the behavior of a material generally involves conducting laboratory experiments. The way in which a material deforms when subjected to stress in a given direction is unique to each material and cannot be described by broad, general equations used in physics. This means that materials must be subjected to stress in relatively simple situations \u2013 tension, compression, etc. \u2013 and their deformation measured in relation to the forces applied.<\/p>\n","innerContent":["\n

    Describing the behavior of a material generally involves conducting laboratory experiments. The way in which a material deforms when subjected to stress in a given direction is unique to each material and cannot be described by broad, general equations used in physics. This means that materials must be subjected to stress in relatively simple situations \u2013 tension, compression, etc. \u2013 and their deformation measured in relation to the forces applied.<\/p>\n"],"rendered":"\n

    Describing the behavior of a material generally involves conducting laboratory experiments. The way in which a material deforms when subjected to stress in a given direction is unique to each material and cannot be described by broad, general equations used in physics. This means that materials must be subjected to stress in relatively simple situations \u2013 tension, compression, etc. \u2013 and their deformation measured in relation to the forces applied.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    Tests of this kind have been carried out for a very long time, but remain difficult to perform. Numerous laboratory tests need to be conducted to capture their complexity. The behavior of the material varies depending on environmental factors (temperature, humidity, etc.) and the direction of the applied stress. Mathematical models therefore aim to represent as accurately as possible the behavior of materials observed in laboratory experiments.<\/p>\n","innerContent":["\n

    Tests of this kind have been carried out for a very long time, but remain difficult to perform. Numerous laboratory tests need to be conducted to capture their complexity. The behavior of the material varies depending on environmental factors (temperature, humidity, etc.) and the direction of the applied stress. Mathematical models therefore aim to represent as accurately as possible the behavior of materials observed in laboratory experiments.<\/p>\n"],"rendered":"\n

    Tests of this kind have been carried out for a very long time, but remain difficult to perform. Numerous laboratory tests need to be conducted to capture their complexity. The behavior of the material varies depending on environmental factors (temperature, humidity, etc.) and the direction of the applied stress. Mathematical models therefore aim to represent as accurately as possible the behavior of materials observed in laboratory experiments.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    For example, wood, an anisotropic material, behaves differently depending on whether it is stressed along the grain (longitudinal) or across the growth rings (radial). Other materials, such as metals under tension or concrete under compression, undergo changes in their microstructure, notably the formation of voids or microcracks. In such cases, the behavior of the material is not a simple relationship between deformation and stress: it also depends on its history. A material that has already undergone numerous deformations will no longer react in the same way as it did in its initial state.<\/p>\n","innerContent":["\n

    For example, wood, an anisotropic material, behaves differently depending on whether it is stressed along the grain (longitudinal) or across the growth rings (radial). Other materials, such as metals under tension or concrete under compression, undergo changes in their microstructure, notably the formation of voids or microcracks. In such cases, the behavior of the material is not a simple relationship between deformation and stress: it also depends on its history. A material that has already undergone numerous deformations will no longer react in the same way as it did in its initial state.<\/p>\n"],"rendered":"\n

    For example, wood, an anisotropic material, behaves differently depending on whether it is stressed along the grain (longitudinal) or across the growth rings (radial). Other materials, such as metals under tension or concrete under compression, undergo changes in their microstructure, notably the formation of voids or microcracks. In such cases, the behavior of the material is not a simple relationship between deformation and stress: it also depends on its history. A material that has already undergone numerous deformations will no longer react in the same way as it did in its initial state.<\/p>\n"},{"blockName":"core\/image","attrs":{"id":10404,"sizeSlug":"large","linkDestination":"none","blob":"","url":"https:\/\/ingenius.ecoledesponts.fr\/wp-content\/uploads\/2026\/03\/automatix_EN-1024x366.png","alt":"","caption":null,"lightbox":[],"title":"","href":"","rel":"","linkClass":"","width":"","height":"","aspectRatio":"","scale":"","linkTarget":"","lock":[],"metadata":[],"align":"","className":"wp-block-image size-large","style":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    \"\"
    Automation of material behavior modeling: application to 3D printing. Credit : Provided by the author<\/figcaption><\/figure>\n","innerContent":["\n
    \"\"
    Automation of material behavior modeling: application to 3D printing. Credit : Provided by the author<\/figcaption><\/figure>\n"],"rendered":"\n
    \"\"
    Automation of material behavior modeling: application to 3D printing. Credit : Provided by the author<\/figcaption><\/figure>\n"},{"blockName":"core\/paragraph","attrs":{"style":{"elements":{"link":{"color":{"text":"var:preset|color|red"}}}},"textColor":"red","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color has-link-color","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    How does the AUTOMATIX project differ from these conventional approaches in practical terms?<\/strong><\/strong><\/p>\n","innerContent":["\n

    How does the AUTOMATIX project differ from these conventional approaches in practical terms?<\/strong><\/strong><\/p>\n"],"rendered":"\n

    How does the AUTOMATIX project differ from these conventional approaches in practical terms?<\/strong><\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    If we take rubber as an example, just two or three parameters may be sufficient in the mathematical model, whereas more complex materials such as soils may require the identification of several dozen different parameters.  Depending on the material, it can therefore take up to a year to develop, calibrate and implement in a software program the behavior that we want to study. With the AUTOMATIX project, we hope to be able to describe a material simply using just a few lines of code. To achieve this, we will use approaches based on artificial intelligence. <\/p>\n","innerContent":["\n

    If we take rubber as an example, just two or three parameters may be sufficient in the mathematical model, whereas more complex materials such as soils may require the identification of several dozen different parameters.  Depending on the material, it can therefore take up to a year to develop, calibrate and implement in a software program the behavior that we want to study. With the AUTOMATIX project, we hope to be able to describe a material simply using just a few lines of code. To achieve this, we will use approaches based on artificial intelligence. <\/p>\n"],"rendered":"\n

    If we take rubber as an example, just two or three parameters may be sufficient in the mathematical model, whereas more complex materials such as soils may require the identification of several dozen different parameters.  Depending on the material, it can therefore take up to a year to develop, calibrate and implement in a software program the behavior that we want to study. With the AUTOMATIX project, we hope to be able to describe a material simply using just a few lines of code. To achieve this, we will use approaches based on artificial intelligence. <\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    In recent years, the rise of machine learning technologies has gradually replaced mathematical models with learning models. They work by training a neural network using experimental data or data derived from numerical simulations. In the case of the behavior of materials, the first attempts were made using models from other fields, such as image recognition, without fully understanding how they worked. These neural networks, known as \u201cblack-box\u201d models due to their complexity, present a significant challenge. If they are evaluated for mechanical conditions that differ greatly from those used during their training, they can behave in a highly erratic manner and produce completely unrealistic predictions. This lack of robustness is unacceptable if they are to be used for safety-critical applications, such as the design of a dam or a nuclear power station, for example.<\/p>\n","innerContent":["\n

    In recent years, the rise of machine learning technologies has gradually replaced mathematical models with learning models. They work by training a neural network using experimental data or data derived from numerical simulations. In the case of the behavior of materials, the first attempts were made using models from other fields, such as image recognition, without fully understanding how they worked. These neural networks, known as \u201cblack-box\u201d models due to their complexity, present a significant challenge. If they are evaluated for mechanical conditions that differ greatly from those used during their training, they can behave in a highly erratic manner and produce completely unrealistic predictions. This lack of robustness is unacceptable if they are to be used for safety-critical applications, such as the design of a dam or a nuclear power station, for example.<\/p>\n"],"rendered":"\n

    In recent years, the rise of machine learning technologies has gradually replaced mathematical models with learning models. They work by training a neural network using experimental data or data derived from numerical simulations. In the case of the behavior of materials, the first attempts were made using models from other fields, such as image recognition, without fully understanding how they worked. These neural networks, known as \u201cblack-box\u201d models due to their complexity, present a significant challenge. If they are evaluated for mechanical conditions that differ greatly from those used during their training, they can behave in a highly erratic manner and produce completely unrealistic predictions. This lack of robustness is unacceptable if they are to be used for safety-critical applications, such as the design of a dam or a nuclear power station, for example.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    The AUTOMATIX project aims to propose a middle ground between the traditional approach and the more recent \u201cblack-box\u201d approach. Within the very architecture of the neural network, we incorporate controllable elements based on established mathematical models. Any aspects that remain unclear are entrusted to machine learning. In principle, this approach will require less data and computing time and will yield more reliable results.<\/p>\n","innerContent":["\n

    The AUTOMATIX project aims to propose a middle ground between the traditional approach and the more recent \u201cblack-box\u201d approach. Within the very architecture of the neural network, we incorporate controllable elements based on established mathematical models. Any aspects that remain unclear are entrusted to machine learning. In principle, this approach will require less data and computing time and will yield more reliable results.<\/p>\n"],"rendered":"\n

    The AUTOMATIX project aims to propose a middle ground between the traditional approach and the more recent \u201cblack-box\u201d approach. Within the very architecture of the neural network, we incorporate controllable elements based on established mathematical models. Any aspects that remain unclear are entrusted to machine learning. In principle, this approach will require less data and computing time and will yield more reliable results.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    <\/a><\/p>\n","innerContent":["\n

    <\/a><\/p>\n"],"rendered":"\n

    <\/a><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"style":{"elements":{"link":{"color":{"text":"var:preset|color|red"}}}},"textColor":"red","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color has-link-color","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    The project will be officially launched in June 2026. Although it marks a certain departure from your previous research, you already have preliminary results. Can you tell us about them? <\/strong><\/strong><\/p>\n","innerContent":["\n

    The project will be officially launched in June 2026. Although it marks a certain departure from your previous research, you already have preliminary results. Can you tell us about them? <\/strong><\/strong><\/p>\n"],"rendered":"\n

    The project will be officially launched in June 2026. Although it marks a certain departure from your previous research, you already have preliminary results. Can you tell us about them? <\/strong><\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    I have begun working on methods to describe what is known as the yield surface of materials. When a sufficiently large force is applied to a material such as steel, it never fully returns to its original shape. This is known as plasticity.  This behavior can be mathematically represented by a surface defined in six-dimensional space. Our preliminary results show that it is possible, using a small number of experiments, to automatically reconstruct the shape of this surface without knowing it explicitly.<\/p>\n","innerContent":["\n

    I have begun working on methods to describe what is known as the yield surface of materials. When a sufficiently large force is applied to a material such as steel, it never fully returns to its original shape. This is known as plasticity.  This behavior can be mathematically represented by a surface defined in six-dimensional space. Our preliminary results show that it is possible, using a small number of experiments, to automatically reconstruct the shape of this surface without knowing it explicitly.<\/p>\n"],"rendered":"\n

    I have begun working on methods to describe what is known as the yield surface of materials. When a sufficiently large force is applied to a material such as steel, it never fully returns to its original shape. This is known as plasticity.  This behavior can be mathematically represented by a surface defined in six-dimensional space. Our preliminary results show that it is possible, using a small number of experiments, to automatically reconstruct the shape of this surface without knowing it explicitly.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    This is a key point from an experimental perspective. If we were to characterize the surface using a very large number of points, this would require carrying out as many tests as there are points. This would be very difficult to achieve. The challenge therefore lies in striking a balance between simplifying the model and remaining faithful to the actual behavior of the material.<\/p>\n","innerContent":["\n

    This is a key point from an experimental perspective. If we were to characterize the surface using a very large number of points, this would require carrying out as many tests as there are points. This would be very difficult to achieve. The challenge therefore lies in striking a balance between simplifying the model and remaining faithful to the actual behavior of the material.<\/p>\n"],"rendered":"\n

    This is a key point from an experimental perspective. If we were to characterize the surface using a very large number of points, this would require carrying out as many tests as there are points. This would be very difficult to achieve. The challenge therefore lies in striking a balance between simplifying the model and remaining faithful to the actual behavior of the material.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"style":{"elements":{"link":{"color":{"text":"var:preset|color|red"}}}},"textColor":"red","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color has-link-color","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    The AUTOMATIX project has direct applications in <\/strong>3D concrete printing<\/strong><\/a>, a key area of research at the Navier Laboratory. What are the specific challenges of this emerging technology?<\/strong><\/strong><\/p>\n","innerContent":["\n

    The AUTOMATIX project has direct applications in <\/strong>3D concrete printing<\/strong><\/a>, a key area of research at the Navier Laboratory. What are the specific challenges of this emerging technology?<\/strong><\/strong><\/p>\n"],"rendered":"\n

    The AUTOMATIX project has direct applications in <\/strong>3D concrete printing<\/strong><\/a>, a key area of research at the Navier Laboratory. What are the specific challenges of this emerging technology?<\/strong><\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    3D concrete printing is a technology that uses high-performance concrete, applied in very thin layers, with optimized geometries designed to use as little material as possible. The challenge therefore lies in ensuring sufficient strength with minimal material, without resorting to conventional reinforcements such as steel rods.<\/p>\n","innerContent":["\n

    3D concrete printing is a technology that uses high-performance concrete, applied in very thin layers, with optimized geometries designed to use as little material as possible. The challenge therefore lies in ensuring sufficient strength with minimal material, without resorting to conventional reinforcements such as steel rods.<\/p>\n"],"rendered":"\n

    3D concrete printing is a technology that uses high-performance concrete, applied in very thin layers, with optimized geometries designed to use as little material as possible. The challenge therefore lies in ensuring sufficient strength with minimal material, without resorting to conventional reinforcements such as steel rods.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    At the Navier laboratory, we have observed that beyond a certain threshold, the printed structures become brittle and begin to crack. To address this, we have chosen to incorporate carbon fibers into the concrete<\/a>. During the printing process, spools of carbon fiber are fed into the print head, and the filaments are deposited simultaneously with the concrete extrusion.<\/p>\n","innerContent":["\n

    At the Navier laboratory, we have observed that beyond a certain threshold, the printed structures become brittle and begin to crack. To address this, we have chosen to incorporate carbon fibers into the concrete<\/a>. During the printing process, spools of carbon fiber are fed into the print head, and the filaments are deposited simultaneously with the concrete extrusion.<\/p>\n"],"rendered":"\n

    At the Navier laboratory, we have observed that beyond a certain threshold, the printed structures become brittle and begin to crack. To address this, we have chosen to incorporate carbon fibers into the concrete<\/a>. During the printing process, spools of carbon fiber are fed into the print head, and the filaments are deposited simultaneously with the concrete extrusion.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    We have now reached a stage where we wish to present to manufacturers the structures we are capable of producing, but to do so, we must demonstrate their strength. This requires calculations and, consequently, the development of behavioral models for both concrete and carbon fibers. As part of the AUTOMATIX project, data is generated directly on-site, within the laboratory, and involves materials that have not yet been extensively modeled. This makes it a particularly relevant application.<\/strong><\/p>\n","innerContent":["\n

    We have now reached a stage where we wish to present to manufacturers the structures we are capable of producing, but to do so, we must demonstrate their strength. This requires calculations and, consequently, the development of behavioral models for both concrete and carbon fibers. As part of the AUTOMATIX project, data is generated directly on-site, within the laboratory, and involves materials that have not yet been extensively modeled. This makes it a particularly relevant application.<\/strong><\/p>\n"],"rendered":"\n

    We have now reached a stage where we wish to present to manufacturers the structures we are capable of producing, but to do so, we must demonstrate their strength. This requires calculations and, consequently, the development of behavioral models for both concrete and carbon fibers. As part of the AUTOMATIX project, data is generated directly on-site, within the laboratory, and involves materials that have not yet been extensively modeled. This makes it a particularly relevant application.<\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"style":{"elements":{"link":{"color":{"text":"var:preset|color|red"}}}},"textColor":"red","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-red-color has-text-color has-link-color","backgroundColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    Finally, what does this ERC grant<\/strong><\/strong>1<\/a><\/sup> mean to you, and how will it transform your project and your research team?<\/strong><\/strong><\/p>\n","innerContent":["\n

    Finally, what does this ERC grant<\/strong><\/strong>1<\/a><\/sup> mean to you, and how will it transform your project and your research team?<\/strong><\/strong><\/p>\n"],"rendered":"\n

    Finally, what does this ERC grant<\/strong><\/strong>1<\/a><\/sup> mean to you, and how will it transform your project and your research team?<\/strong><\/strong><\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    The main advantage of this grant lies in the scale of the projects it enables. It provides five years of funding for a \u201cConsolidator\u201d grant of up to \u20ac2 million. This provides the opportunity to conceive a large-scale project capable of funding numerous research positions (engineers, PhD students, postdoctoral researchers), as well as the necessary experiments and equipment.<\/p>\n","innerContent":["\n

    The main advantage of this grant lies in the scale of the projects it enables. It provides five years of funding for a \u201cConsolidator\u201d grant of up to \u20ac2 million. This provides the opportunity to conceive a large-scale project capable of funding numerous research positions (engineers, PhD students, postdoctoral researchers), as well as the necessary experiments and equipment.<\/p>\n"],"rendered":"\n

    The main advantage of this grant lies in the scale of the projects it enables. It provides five years of funding for a \u201cConsolidator\u201d grant of up to \u20ac2 million. This provides the opportunity to conceive a large-scale project capable of funding numerous research positions (engineers, PhD students, postdoctoral researchers), as well as the necessary experiments and equipment.<\/p>\n"},{"blockName":"core\/paragraph","attrs":{"align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","gradient":"","fontSize":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    The ERC grant focuses on scientific excellence, and allows researchers to explore more risky avenues than usual, thereby pushing the boundaries of knowledge. For this reason, I believe it is essential to establish collaborations between colleagues working on similar topics, both in Europe and globally. In addition, we plan to make our tools open-source2<\/a><\/sup> in order to provide the entire community with accessible tools that can be applied regardless of the class of materials being studied. <\/p>\n","innerContent":["\n

    The ERC grant focuses on scientific excellence, and allows researchers to explore more risky avenues than usual, thereby pushing the boundaries of knowledge. For this reason, I believe it is essential to establish collaborations between colleagues working on similar topics, both in Europe and globally. In addition, we plan to make our tools open-source2<\/a><\/sup> in order to provide the entire community with accessible tools that can be applied regardless of the class of materials being studied. <\/p>\n"],"rendered":"\n

    The ERC grant focuses on scientific excellence, and allows researchers to explore more risky avenues than usual, thereby pushing the boundaries of knowledge. For this reason, I believe it is essential to establish collaborations between colleagues working on similar topics, both in Europe and globally. In addition, we plan to make our tools open-source2<\/a><\/sup> in order to provide the entire community with accessible tools that can be applied regardless of the class of materials being studied. <\/p>\n"},{"blockName":"core\/paragraph","attrs":{"fontSize":"small","align":"","content":null,"dropCap":false,"placeholder":"","direction":"","lock":[],"metadata":[],"className":"has-small-font-size","style":"","backgroundColor":"","textColor":"","gradient":"","fontFamily":"","borderColor":"","anchor":""},"innerBlocks":[],"innerHTML":"\n

    Interview conducted by Ad\u00e8le Mazurek, Head of Scientific Outreach at \u00c9cole nationale des ponts et chauss\u00e9es<\/p>\n","innerContent":["\n

    Interview conducted by Ad\u00e8le Mazurek, Head of Scientific Outreach at \u00c9cole nationale des ponts et chauss\u00e9es<\/p>\n"],"rendered":"\n

    Interview conducted by Ad\u00e8le Mazurek, Head of Scientific Outreach at \u00c9cole nationale des ponts et chauss\u00e9es<\/p>\n"},{"blockName":"core\/footnotes","attrs":{"lock":[],"metadata":[],"className":"","style":"","backgroundColor":"","textColor":"","fontSize":"","fontFamily":"","borderColor":""},"innerBlocks":[],"innerHTML":"","innerContent":[],"rendered":""}],"seo":{"title":"Predicting the behavior of materials using artificial intelligence"},"media":{"img":"\"Jean-Claude","src":"https:\/\/ingenius.ecoledesponts.fr\/wp-content\/uploads\/2026\/03\/jcm-ENPC-33-scaled.jpg"},"url":"\/en\/articles\/predicting-the-behavior-of-materials-using-artificial-intelligence\/","related":{"post":[],"author":[],"subject":[],"category":[],"folder":[]},"translated":"https:\/\/ingenius.ecoledesponts.fr\/articles\/predire-le-comportement-des-materiaux-grace-a-lintelligence-artificielle\/","icon":"icon-article","duration":"5","custom_excerpt":"Understanding how materials deform, crack and age is a key issue in modern engineering. J\u00e9r\u00e9my Bleyer, a researcher at the Navier Laboratory, is exploring new avenues combining solid mechanics and artificial intelligence. His project, AUTOMATIX, which was recently awarded a prestigious ERC grant, aims to provide the academic and industrial sectors with flexible and powerful digital tools to accelerate the development of safer, more efficient and better-controlled materials and structures.","duration_type":"","_links":{"self":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/10403","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/comments?post=10403"}],"version-history":[{"count":4,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/10403\/revisions"}],"predecessor-version":[{"id":10524,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/posts\/10403\/revisions\/10524"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/media\/10410"}],"wp:attachment":[{"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/media?parent=10403"}],"wp:term":[{"taxonomy":"article-types","embeddable":true,"href":"https:\/\/ingenius.ecoledesponts.fr\/en\/wp-json\/wp\/v2\/article-types?post=10403"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}