HAL Portal de Univ. Gustave Eiffel
Not a member yet
72847 research outputs found
Sort by
Full-scale monitoring of cold mix asphalt during curing
No full textInternational audienceCold mix asphalts (CMA) could be an answer to the need to use road materials with a lower carbon footprint. They are implemented at room temperature without drying aggregates, allowing significant energy savings for the pavement industry. However, their behavior in road pavements is still little known. This paper investigates the structural evolution and damaging of monitored test sections containing CMAs and their link to mix and binder variations with time. The behavior of the pavement was assessed in regards to the mix composition, surface layer, number of loading cycles using accelerated pavement testing and weather conditions. Full-scale experimentations showed that structures containing CMAs as base course have a very distinctive behavior in response to loading. Firstly, they undergo high strain amplitudes, compared to structures containing hot mix asphalts (HMAs), and these amplitudes increase with the number of loading cycles without apparent structural damage. It was also demonstrated that a HMA surface layer shields the CMA course. Full-scale and in-lab tests demonstrated that CMA curing (after 36 months) is not influenced by a surface layer and that the performances of CMAs can be reduced by a lower binder content. Moreover an overloaded section displayed only transverse cracking. Permanent deformation was visible on this sacrificial section, from the CMA surface to the unbound granular material course. Generally, this work also gives more insight on structural behavior of pavements containing CMAs under loading. The testing procedures and equipment suitable for structures with HMAs are not always adapted to structures with CMAs
Performance and leaching behavior of concrete incorporating electric arc furnace slag aggregates
No full textInternational audienceThis study provides a balanced evaluation of the technical performance and environmental behavior of concretes incorporating EAF slag, offering key insights for the sustainable and safe reuse of this industrial by-product in construction. Fresh concrete properties, compressive strength, and durability indicators were evaluated for various replacement rates of natural aggregates with EAF slag. Five mixtures were produced: a control with 100% natural aggregates and four mixtures with varying sand and gravel substitution levels were tested. EAF slag increased compressive strength and elastic modulus by up to 38% and 16% respectively, mainly due to its high strength and angular shape, which enhance aggregate-matrix bonding. Despite a rise in porosity (14.7% to 16.8%) caused by additional water demand, chloride diffusion resistance improves by up to 26%, indicating a denser interfacial transition zone. However, visual signs of corrosion under chloride exposure raise concerns for long-term durability in aggressive environments. From an environmental perspective, leaching tests were conducted on both the raw slag and the resulting concretes in crushed and monolithic forms. Although EAF slag contains elevated levels of certain elements (Cr, Mo, and Fluorides), their leachability is significantly reduced in the cementitious matrix. Crushed concrete samples remain within regulatory thresholds for inert waste, and monolithic leaching confirms further pollutant immobilization. Nevertheless, some mobile species (Zn, sulfates, and chlorides) are less effectively retained in monolithic form
Lifetime occupational and para-occupational exposure to organic solvents and testicular germ cell tumor risk: a French case–control study—TESTIS
No full textInternational audienceAbstract Background Despite an incidence increase in recent decades, the etiology of testicular germ cell tumors (TGCT) remains poorly understood. The hypothesis of a two-stage development, combining initial alteration in utero followed by malignant transformation later in life, has been suggested. This study examined the association between cumulative lifetime occupational and para-occupational solvent exposure and TGCT risk. Methods The French multicenter case–control study TESTIS included 454 cases and 670 controls. Participants provided information on their occupational history; participants’ mothers (N = 547) provided information on their own and the father’s occupational history. Solvent exposure was assessed by using the Matgéné job–exposure matrices. The influence of the parental and subject’s occupational exposures over the lifetime and at different periods (i.e. fetal life/infancy; childhood; adolescence; subject’s exposure) on TGCT was examined. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by using conditional logistic regression models. Results An OR for TGCT of 1.03 (95% CI 0.59–1.79) was found for the lifetime solvent exposure. When each period was examined individually, the results showed an increased TGCT risk in adult males who were occupationally exposed to trichloroethylene (OR = 3.09; 95% CI 1.25–7.65); fuels and petroleum-based solvents (OR = 1.91; 95% CI 1.21–3.02); diesel, kerosene, and fuel oil (OR = 2.26; 95% CI 1.16–4.41); and ketones and esters (OR = 1.66; 95% CI 1.02–2.71), and suggested a positive association with solvent exposure during adolescence (OR = 1.77; 95% CI 0.95–3.31). Conclusion Overall, this study did not suggest a substantial role of cumulative lifetime solvent exposure and TGCT risk. The results showed an increased TGCT risk associated with solvent exposure during adulthood. Indirect exposure to certain solvents during adolescence might also promote TGCT development
Study and Design of a Dual-Cavity Differential Resonant Sensor for Small Liquid Samples
No full textInternational audienceThis paper presents the development andcharacterisation of a differential RF sensor specifically designedfor the dielectric analysis of small liquid samples. The proposedsensor architecture comprises a microstrip line integrated withtwo identically-dimensioned split-ring resonators, engineered toachieve a resonance at 1.89 GHz when unloaded. Experimentalmeasurements confirm the sensor’s high sensitivity to the liquidsunder investigation and demonstrate its versatility, enabledby the incorporation of two dedicated cavities per resonator.Furthermore, the design’s performance aligns with theoreticalpredictions, attesting its robust and reliable operation undervarious conditions
Sécurité, intelligence artificielle et confiance : impacts et solutions pour une meilleure implémentation au sein des organisations
No full textInternational audienceSociety is facing a security pervasiveness, leading to a predominant solutionist thinking within organisations. However, this reliance on AI-enhanced technology to solve problems without human intervention, reveals flaws and raises issues of legitimacy, perception, acceptability and recognition in the field of organisational security.La sociedad se enfrenta a una omnipresencia de la seguridad que conduce a un pensamiento solucionista predominante en las organizaciones. Sin embargo, esta confianza en la tecnología potenciada por IA para resolver problemas sin intervención humana, revela fallos y plantea cuestiones de legitimidad, percepción, aceptabilidad y reconocimiento en el ámbito de la seguridad organizativa.La société fait face à une pervasivité sécuritaire, conduisant à une pensée solutionniste prédominante au sein des organisations. Cependant, cette confiance dans une technologie augmentée par l’IA pour résoudre les problèmes sans intervention humaine, révèle des failles et pose des problèmes de légitimité, de perception, d’acceptabilité et de reconnaissance dans le domaine de la sécurité organisationnelle
PROSE+ project: offshore seismic measurements on the seabed to test the ability to assess the spatial variation of the small-strain shear modulus in the subsurface environment
No full textInternational audienceThe increase in the number of offshore wind farm siting projects, combined with the multiplicity of developments in anchoring techniques, means that foundations and anchors need to be optimized for the conditions of the offshore subsurface, requiring precise knowledge of the mechanical characteristics of the medium. According to the recommendations of the CFMS (French committee of Soils Mechanics), detailed geophysical reconnaissance is then necessary during the project phase (design and execution) to obtain the most accurate information possible at the locations of the structures. In this context, the PROSE+ project aims to increase knowledge and provide new methodological and technical elements, based on surface seismic and geoelectrical techniques. This will make possible to approach heterogeneous environments in a quantitative and non-destructive way, thereby reducing the number of costly and invasive geotechnical surveys. To this end, we developed numerically a 2D seismic inversion technique using Surface Seismic Waves based on Particle Swarm optimization methods. In order to validate it on experimental data, we carried out measurements off Concarneau using 70 4-components sensors (GPR -Sercel nodes) placed on the seabed, in an unprecedented manner. 241 seismic shots were fired over this sensor's network using an air gun at variable water depth. Finally, the sensors were left recording on the seafloor for 28 days. The recorded seismic data allow to test the capacity of both active and passive seismic imaging process to assess the shear modulus in a 2D medium under seabed
Débris plastiques sur les berges de Seine – Dynamiques de stockage, remobilisation et fragmentation
No full textInternational audienc
Perception adaptative et générique pour la détection et la compréhension de l'environnement routier dégradé. Application à la détection générique des conditions météorologiques
No full textThe development of driving automation systems requires the implementation of a robust and resilient perception architecture under degraded conditions. This thesis proposes a generic methodology for detecting the presence of environmental perturbations (such as fog and rain), applicable to a wide range of sensor technologies. Thus, this methodology enables a real-time evaluation of reliability on embedded sensors under degraded conditions. The combination of Machine Learning and data fusion enables the quantification of the impact of environmental conditions via a functioning score, while balancing robustness, performance, and explainability. Prior to constructing the generic methodology, Neural Networks (NN) were compared with non-neural Machine Learning methods. This initial study highlighted the advantages and limitations of these Machine Learning systems. The proposed methodology for degraded conditions detection is applicable to any sensor capable of producing data in either matrix or point cloud form. Firstly, we applied, tested, and validated on an RGB camera. Then, the methodology was tested on the LIDAR TOF on the point cloud form. The results obtained are highly promising. Nevertheless, to understand the few incorrect predictions, an explainability procedure was developed. This procedure has been built upon the differentiation between ante-hoc methods (incorporating interpretability during the design phase) and posthoc methods (analyzing the decisions after the training phase). Finally, to account for and manage ignorance, uncertainty, and conflicts, a multisources fusion based on Dempster-Shaffer's belief theory was proposed. The equations governing the fusion operators demonstrate that the optimization of fusion coefficients is possible through gradient descent and backpropagation within the fusion operators. In this context, a set of non-supervised criteria is defined intrinsically within the architecture. The collection of these studies greatly contributes to the increase of reliability of embedded sensors by integrating meteorological detection, explainability, and multisources fusion with backpropagation for the estimation of sensor reliability. The proposed multisources architecture and generic methodology is, undoubtedly, a significant innovation for improving perception and decision-making in autonomous driving.Le développement des systèmes d'automatisation de la conduite nécessite la mise en œuvre d'une architecture de perception robuste et résiliente aux conditions dégradées. Ce travail de thèse propose une méthodologie générique de détection de la présence de perturbations (pluie, brouillard) applicable à de très nombreuses technologies de capteurs. Cette méthodologie permet ainsi d'évaluer en temps réel la fiabilité des capteurs embarqués en conditions dégradées. En combinant le Machine Learning et la fusion de données, nous pouvons quantifier l'impact des conditions environnementales via un score de fonctionnement, tout en conciliant robustesse, performance et explicabilité.Avant de construire la méthodologie générique, les réseaux de neurones (NN) ont d'abord été comparés aux méthodes de Machine Learning non neuronales. Cette première étude a clairement mis en évidence les avantages et les limites des systèmes de Machine Learning.L'application de la méthodologie de détection des conditions dégradées est applicable à tous les capteurs pouvant fournir des données sous forme matricielle. Cette méthodologie a d'abord été appliquée, testée, et validée en utilisant une caméra RGB. Ensuite, la méthodologie a été étendue au LIDAR TOF en exploitant le nuage de points.Les résultats obtenus sont très prometteurs. Néanmoins, pour prendre en compte les quelques mauvaises prédictions, une procédure d'explicabilité a été construite. Cette procédure a été construite en distinguant les méthodes ante-hoc (prévoyant l'interprétabilité dès la conception) et post-hoc (permettant d'analyser les décisions après entraînement). Enfin, afin de prendre en compte et de gérer l'ignorance, l'incertitude, et les conflits, une fusion multi-sources fondée sur la théorie des croyances de Dempster-Shafer a été proposée. Les équations régissant les opérateurs de fusion montrent que l'optimisation des coefficients de fusion est possible par descente de gradient et rétro-propagation dans les opérateurs de fusion. Dans ce cadre, un ensemble de critères non supervisé est définie de manière intrinsèque à l'architecture.L'ensemble de ces travaux contribue fortement à la fiabilisation des capteurs embarqués en intégrant la détection météorologique, l'explicabilité, et fusion multi-sources avec rétro-propagation pour estimer la fiabilité des sources de données. Cette architecture multi-sources et cette méthodologie générique est, sans conteste, une innovation significative pour améliorer la perception et la prise de décision en conduite autonome
Tanzanie : la fermeture de l’espace numérique, élément clé de la répression
No full textThe Conversation https://theconversation.com/tanzanie-la-fermeture-de-lespace-numerique-element-cle-de-la-repression-269119This article analyzes the Tanzanian political crisis of October 2025 through the closure of digital space, which became a central instrument of repression. Following highly contested elections, the authorities resorted to a coercive triptych—curfew, military deployment, and Internet shutdown—transforming an electoral dispute into a systemic crisis of confidence.Drawing on contributions from information and communication sciences, the study shows that the digital blackout is not solely aimed at combating disinformation: it restructures the very conditions of public visibility by making the state the sole legitimate producer of discourse. Deprived of social networks, citizens, independent media, and NGOs find themselves without a means of expression or the ability to document events.The analysis also reveals the social, political, and economic effects of this strategy: fragmentation of public debate, information asymmetry, weakened trust, and major economic losses linked to the interruption of digital services. Through the Tanzanian case, the article highlights a growing trend in digital Africa: governance through the restriction of information space.It concludes that the stability produced by censorship is a fragile “performative stability” built on silence rather than democratic mediation—a dynamic that directly questions the future of digital sovereignty and communication security on the continent.Cet article analyse la crise politique tanzanienne d’octobre 2025 à travers la fermeture de l’espace numérique, devenue un instrument central de la répression. À la suite d’élections fortement contestées, les autorités ont recouru à un triptyque coercitif — couvre-feu, déploiement militaire et coupure d’Internet — transformant une contestation électorale en crise systémique de confiance.Mobilisant les apports des sciences de l’information et de la communication, l’étude montre que le blackout numérique ne vise pas uniquement la lutte contre la désinformation : il restructure les conditions mêmes de la visibilité publique en rendant l’État seul producteur légitime du discours. Privés de réseaux sociaux, les citoyens, médias indépendants et ONG se retrouvent sans relais d’expression ni capacité de documentation.L’analyse révèle également les effets sociaux, politiques et économiques de cette stratégie : fragmentation du débat public, asymétrie informationnelle, affaiblissement de la confiance et pertes économiques majeures liées à l’interruption des services numériques. À travers le cas tanzanien, l’article met en lumière une tendance croissante en Afrique numérique : la gouvernance par la restriction de l’espace informationnel.Il conclut que la stabilité produite par la censure est une “stabilité performative”, fragile, construite sur le silence plutôt que sur la médiation démocratique — une dynamique qui interroge directement l’avenir de la souveraineté numérique et de la sécurité communicationnelle sur le continent
