Portail des publications scientifiques IMT Mines Alès
Not a member yet
5198 research outputs found
Sort by
Insights into the Effect of Recycled Glass Fiber Reinforced Polymer on the Mechanical Strengths of Cement Mortar
No full textInternational audienceThe incorporation of recycled glass fiber reinforced polymer (rGFRP) in cementitious materials is an interesting recycling and valorization method. However, this incorporation generally results in a significant loss of workability, often compensated by an adjustment of the water to cement ratio, which can affect mechanical performance, particularly compressive strength. The aim of this paper is to examine the effect of different size fractions of rGFRP (0.063 mm, 0.16 mm, 0.63 mm, 1.25 mm and 2 mm) on the mechanical strengths of cement mortars with a mixing method that is likely to maintain the workability of the mix without adjusting the water to cement ratio. For this, a substitution rate of 10% (in volume of sand by rGFRP), supposed to induce workability loss, is chosen. A pre-mixing of rGFRP with water before adding cement and sand is performed and allows for the workability to be maintained without increasing the water content. The results show that compressive and flexural strengths are almost maintained compared with reference mortar for two rGFRP size fractions (2 mm and 0/2 mm). For the 2 mm fraction, a slight improvement (3%) in flexural strength after 7 months of curing and a 5% reduction in compressive strength are observed. After 7 months of curing, fibers or clusters of rGFRP are still observed, although they are not alkali-resistant
Integrating Biological Treatment and Electro-Oxidation in an Anaerobic Biomass Membrane Bioreactor for Innovative and Integrating Urban Wastewater Treatment
No full textInternational audienceIn response to evolving environmental regulations, particularly the European Water Framework Directive, this study focuses on implementing an innovative Anaerobic Membrane BioReactor (AnMBR) configuration targeting the removal of organic recalcitrant micropollutants and energy recovery from urban wastewater. This novel configuration integrates biological treatment and electro-oxidation within a specialized AnMBR featuring granular sludge (e-GAnMBR). Notably, the engineered reactor intensifies biological reactions and addresses AnMBR challenges such as membrane fouling, upgrading methane production, and energy consumption mitigation. Tailored for the removal of organic micropollutants, with a particular emphasis on diuron, carbamazepine, and perfluorooctanesulfonic acid in this study, and an examination of their degradation pathways, transformation products, and toxicity, this innovative reactor represents a significant stride towards sustainable wastewater treatment
Can Large Language Models Learn Conceptual Modeling by Looking at Slide Decks and Pass Graduate Examinations? An Empirical Study
No full textInternational audienceLarge Language Models (LLMs) have been used in numerous studies to pass written examinations, with mixed performances. Prior studies have relied on the text capability of LLMs and close ended questions (e.g., multiple choices), often examining performances without comparison to real-world test-takers. Recent studies have emphasized the need to test multiple LLMs and situate performances with respect to students’ scores. In addition, recent LLMs can handle both text and images. In our empirical study, we leverage this technical improvement to examine whether LLMs can learn from the same material (i.e., readings and slide decks) that was given to students in a graduate class on conceptual modeling. Our experiments on GPT-4o and Claude 3.5 Sonnet demonstrate that both LLMs improve performances by learning from the material across five tests (covering different aspects of conceptual modeling) and across levels of mastery (e.g., understand, create). Both LLMs achieved passing grades. GPT-4o was often aligned with, or above, a median student. Our results pave the way for future studies in which LLMs may take a course in conceptual modeling, by decomposing video recordings of lectures into images
Design, development and maintenance of a digital twin of vitrification process: a methodological contribution
No full textInternational audienceThis paper explores the concept of the digital twin introduced by Industry 4.0. Classically considered, a digital twin serves as a virtual replica or a physical asset, enabling various usages such as real-time monitoring, predictive maintenance management, or optimization. The application case is a continuous vitrification process developed in France for the treatment of final nuclear waste here supported by different physical prototypes. Using both normative references and concepts and principles from Model Based Systems and Software Engineering (MBSSE), the contribution introduces a methodological approach for the design, implementation and maintenance of a so-called digital twin system. Through case studies and analysis, the paper intents to demonstrate the potential of this approach in improving vitrification process for nuclear waste
Human exposure related to nano or microplastics during recycling process
No full textInternational audienceThe aim of the PROTEGE-MIPLAS project is to provide new knowledge about theemission of micro and/or nanoplastics that may be aerosolised during the processing and re-generation of plastic waste. Global losses in the form of microplastics (plastics less than 5mmin length) across the plastics value chain are considerable and can be estimated at 3 milliontonnes per year. Although the primary production of plastics generates very few microplas-tics, very little data is available on their production for integrated plastic waste processingsites and regeneration in the form of recycled plastic granules or finished products. Variousunitary operations are capable of creating indoor air pollution in plastic waste processingfacilities, but can also lead to external pollution, depending on the configuration of the in-dustrial premises. The PROTEGE-MIPLAS project, which is being carried out jointly byIMT Mines Al`es (Materials Centre) and Mines St Etienne (Engineering and Health Centre),in partnership with Polyvia and with the support of the Laboratoire National de M´etrologieet d’Essais (LNE), involves manufacturers from the recycling sector and is structured in sev-eral phases. Initially, the aim is to identify the process stages at these companies’ integratedsites that are likely to generate aerosolisable microplastics, depending on the nature of theincoming waste and the types of process under consideration (packaging, end-of-life vehicleparts, WEEE, etc.)
Generative AI for Business Model Generation (GAI4BM): from Textual Description to Business Process Model
No full textInternational audienceCurrently, companies are increasingly turning to process modeling as a tool to comprehend and enhance their internal operations, thereby boosting efficiency. However, executing these processes effectively poses a significant challenge for organizations. While process modeling serves as a pivotal step toward formal models that facilitate verification or simulation, the initial description of processes often relies on natural language. Transitioning from this textual description to a more formal BPMN (Business Process Model and Notation) representation remains a time-consuming and difficult task for experts. This article proposes exploring the use of Generative Artificial Intelligence and especially Large Model Language Models (LLM) such as GPT (Generative Pre-trained Transformer) to generate BPMN business process models from textual descriptions. This process includes various steps, including natural language understanding, extraction of pertinent information, and translation of this data into structured BPMN elements. By employing generative artificial intelligence, the objective is to automate and streamline this process, offering a more efficient approach to business process modeling
Nation-wide monitoring campaign of 49 biocides and surfactants in surface waters and wastewaters
No full textInternational audienceDespite their intensive use and their impact on ecosystems, biocides and surfactants are still poorly regulated and poorly monitored at large scale. In the frame of the revision of the national regulatory surveillance plan of surface waters, France planned in 2018 a monitoring campaign at national scale focused on these two types of substances of very emerging concern. Forty-nine contaminants (32 biocides and 17 surfactants) were investigated in surface water and sediment samples from 91 sampling sites, and in effluent and sludge samples of 7 wastewater treatment plants (WWTP), in mainland France and overseas regions. Between 33 and 52 % of the target contaminants were quantified at least once in water and sediment. High frequencies of quantification were observed for the surfactants (up to 91 % in water samples and up to 57 % in sediment samples for LAS C10-C13) and for the biocides (up to 64 % for fipronil in water samples and up to 90 % for methyl nonyl ketone in sediment samples). The median concentrations of surfactants were up to 2 μg/L in mainland surface water samples and up to 528 μg/kg in sediment samples, and for biocides, the median concentrations were up to 0.18 μg/L in mainland surface water samples and up to 104 μg/kg in sediment samples. PNEC exceedances in water and sediment were determined for both types of substances. The analysis of effluent and sludge suggested significant but not total removal of these substances in the WWTP. Temporal and spatial variations of the concentrations of both types of substances in surface water samples were also observed, suggesting both punctual and diffuse contamination sources of the surface water investigated
Conception et mise en œuvre d'une solution interopérable d'un hôpital de campagne mobile dédié à l'industrie pétrolière et gazière
No full textThis study examines the application of computational techniques and disaster response role-playing game to assess the feasibility of developing a reproducible and adaptable simulation model for disaster response and crisis decision-making in healthcare. A crisis management approach is used to deploy resources and evacuate victims from a hypothetical disaster site following a sudden onset of disaster. This approach requires diverse agents and their ability to coordinate and allocate resources effectively. While the model in no way replaces any real system, the simulations and exercise developed and implemented in this study can serve as an analytical tool to support process improvement, training and decision-making.The simulation model and game developed for this study analysed the results of both reactive and deliberative agents’ decisions, evaluated as mortality, and mean wait time outcomes from interacting emergency departments. The core elements of the developed agent-based model (NetLogo 6.3.0) were then transformed and implemented in a tabletop serious game where participants played through the model's decision-making rules, providing an answer to the study’s research questions. This research questions posed in this study are: (1) "How can the interaction between a regional hospital (RH) and a mobile field hospital (MFH) be modelled for a disaster response setting for the oil and gas industry?" and (2) "How can Disaster Response Tabletop Role-Playing Games (DRTRPGs) be used to verify/validate the representativeness of the ABMS model agents’ response process in the context of the disaster response scenario?" The aim of this study is to "improve the interoperability between a regional hospital and a mobile field hospital by detailing the process of interaction between the systems," describing a typical disaster response incident control system.In the event of an industrial explosion such as that hypothesized in this study, the parameters examined highlighted the uncertainties associated with the complexity of responding to a disaster scenario and the variability due to the crisis context as critical to patient outcomes. The examined indicators can serve two purposes: (1) enable an assessment of the overall effectiveness of disaster response strategies; and (2) draw attention to elements in the context of disaster response management that may have larger and complementary impacts. The need for informed decision-making in disaster situations to reduce patient mortality rates and mean waiting times between interoperable healthcare systems is demonstrated in this study through modelling, simulation and gamification. Therefore, the potential and effectiveness of using combined techniques to gain valuable insights into the interactions within complex systems is established.Cette étude examine l'application de techniques computationnelles et d'un jeu de rôle de réponse aux catastrophes pour évaluer la faisabilité de développer un modèle de simulation reproductible et adaptable pour la réponse aux catastrophes et la prise de décision en situation de crise dans le domaine de la santé. Une approche de gestion de crise est utilisée pour déployer des ressources et évacuer les victimes d'un site de catastrophe hypothétique suite à un début soudain de catastrophe. Cette approche nécessite la présence d'agents divers et leur capacité à coordonner et allouer efficacement les ressources. Bien que le modèle ne remplace en aucun cas un système réel, les simulations et exercices développés et mis en œuvre dans cette étude peuvent servir d'outil analytique pour soutenir l'amélioration des processus, la formation et la prise de décision.Le modèle de simulation et le jeu développés pour cette étude ont analysé les résultats des décisions des agents réactifs et délibératifs, évalués en termes de mortalité et de temps d'attente moyen dans les services d'urgence interagissant. Les éléments centraux du modèle basé sur les agents développé (NetLogo 6.3.0) ont ensuite été transformés et mis en œuvre dans un jeu sérieux de table où les participants ont traversé les règles de prise de décision du modèle, fournissant ainsi une réponse aux questions de recherche de l'étude. Les questions de recherche posées dans cette étude sont les suivantes : (1) "Comment l'interaction entre un hôpital régional (HR) et un hôpital de campagne mobile (HCM) peut-elle être modélisée pour une situation de réponse aux catastrophes dans l'industrie pétrolière et gazière ?" et (2) "Comment les jeux de rôle de table sur la réponse aux catastrophes peuvent-ils être utilisés pour vérifier/valider la représentativité du processus de réponse des agents du modèle ABMS dans le contexte du scénario de réponse aux catastrophes ?" L'objectif de cette étude est d'"améliorer l'interopérabilité entre un hôpital régional et un hôpital de campagne mobile en détaillant le processus d'interaction entre les systèmes," décrivant ainsi un système typique de contrôle des incidents de catastrophe.En cas d'explosion industrielle telle que celle envisagée dans cette étude, les paramètres examinés ont mis en évidence les incertitudes associées à la complexité de la réponse à un scénario de catastrophe et la variabilité due au contexte de crise comme étant critiques pour les résultats des patients. Les indicateurs examinés peuvent servir à deux fins : (1) permettre une évaluation de l'efficacité globale des stratégies de réponse aux catastrophes ; et (2) attirer l'attention sur des éléments dans le contexte de la gestion de la réponse aux catastrophes qui peuvent avoir des impacts plus importants et complémentaires. Le besoin de prise de décision éclairée dans les situations de catastrophe pour réduire les taux de mortalité des patients et les temps d'attente moyens entre les systèmes de santé interopérables est démontré dans cette étude par la modélisation, la simulation et la ludification. Par conséquent, le potentiel et l'efficacité de l'utilisation de techniques combinées pour obtenir des informations précieuses sur les interactions au sein de systèmes complexes sont établis
Combining XRD and nanoindentation to characterize mechanical properties evolution of flax cell walls during controlled heat treatment
No full textInternational audienceOver the past decades, a tremendous interest has been developed concerning plant fibre used to reinforce composite materials as these fibers show comparable mechanical properties and higher ecological benefits than synthetic fibres such as glass fibres. Environmental concerns are currently driving interest in the replacement of thermoset matrices by thermoplastic polymers and pave the way for recycling and composting as end-of-life routes for plant fibre composites. However, using thermoplastic polymers as a matrix involves long processing cycles at high temperature, which may negatively affect both structure and mechanical properties of plant fibres. In this work, structural evolution and mechanical behaviour of flax fibre cell walls were dynamically monitored in-situ by X-ray diffraction and nanoindentation from ambient to 230°C. A drop in local mechanical performance of flax cell walls, associated with structural evolution at different scales and changesin the biochemical composition, were noticed while increasing temperature. This work, proposing for the first time an in-situ investigation of the dynamic evolution in temperature of the flax cell wall properties, also highlighted the reversible behaviour of their crystalline structure and local mechanical properties, after cooling to room temperatur
Transformer with Controlled Attention for Synchronous Motion Captioning
No full textIn this paper, we address a challenging task, synchronous motion captioning, that aim to generate a language description synchronized with human motion sequences. This task pertains to numerous applications, such as aligned sign language transcription, unsupervised action segmentation and temporal grounding. Our method introduces mechanisms to control self-and cross-attention distributions of the Transformer, allowing interpretability and time-aligned text generation. We achieve this through masking strategies and structuring losses that push the model to maximize attention only on the most important frames contributing to the generation of a motion word. These constraints aim to prevent undesired mixing of information in attention maps and to provide a monotonic attention distribution across tokens. Thus, the cross attentions of tokens are used for progressive text generation in synchronization with human motion sequences. We demonstrate the superior performance of our approach through evaluation on the two available benchmark datasets, KIT-ML and Hu-manML3D. As visual evaluation is essential for this task, we provide a comprehensive set of animated visual illustrations in the code repository: https://github.com/ rd20karim/Synch-Transformer