Portail HAL IMT Mines Albi
Not a member yet
5440 research outputs found
Sort by
Optimisation des opérations et groupes d'opérations dans le cadre de la planification dynamique
No full textMaintenance plays a significant role in the running of an aircraft operator. Its primary aim is to ensure passenger safety and a well-maintained aircraft. Aircraft maintenance takes several forms: corrective (repairing the system after a failure), systematic preventive (following a pre-established maintenance program), and increasingly conditional preventive (consisting of observing the state of health of a system and intervening when a condition is triggered). Maintenance accounts for between 10% and 20% of an operator's budget, and mismanaged maintenance can lead to flight delays or cancellations and heavy financial losses. In this context, accurate and reliable planning is essential to prevent failures as far as possible and limit the costs of safe maintenance. This thesis focuses on line maintenance, managing short and medium-term task deadlines (from one week to several months). This type of maintenance management consists of spreading periodic inspection tasks over the slots between two flights while the aircraft is on the ground, thus avoiding the need to ground the aircraft for several hours to perform these tasks all at once. As a result, the aircraft's airworthiness is enhanced. However, the feasibility of planning is paramount to avoid disrupting flight schedules by ensuring that the necessary time and resources are available to complete the tasks. Scheduling can be complex and time-consuming for a human, and exploring all scheduling options makes optimizing difficult. Therefore, a scheduling assistant can help operators quickly simulate optimized maintenance scenarios. This thesis formalizes the line maintenance optimization problem as an extended Resource Constrained Project Scheduling Problem (RCPSP) with additional maintenance-related constraints and considers multi-skilled resources. The multi-objective optimization of this problem is managed by a lexicographic ordering of the industrial priorities. The study of exact solution methods is proposed through a Mixed Integer Programming (MIP) model and a Constraint Programming (CP) model. A comparative study justifies the choice of CP to solve this industrial problem. A search strategy inspired by business knowledge is used to return a first reproducible solution within a solving time acceptable to an expert. A practical use case on real maintenance data highlights the advantages of the proposed approach by comparing it with a heuristic industrial approach. Finally, decomposition methods are under investigation in the hope of solving large instances. Column generation (CG) models are proposed for application to different types of RCPSP to study their resolution quality. These preliminary studies outline the future work needed to model the industrial problem using these advanced solving methods.La maintenance joue un rôle majeur dans le fonctionnement d'un exploitant d'aéronefs. Elle a pour but de garantir avant tout la sécurité des passagers et le bon entretien de l'appareil. La maintenance aéronautique se décline sous plusieurs formes : corrective (restaurant le système après une défaillance), préventive systématique (suivant un programme de maintenance pré-établi) et de plus en plus préventive conditionnelle (consistant à observer l'état de santé d'un système et à intervenir lors du déclenchement d'une condition). La maintenance représente entre 10% et 20% du budget d'un exploitant et une mauvaise gestion de la maintenance peut mener à des retards ou des annulations de vols pouvant causer de lourdes pertes financières. Dans ce contexte, une planification précise et fiable est donc primordiale pour prévenir au maximum les défaillances et limiter les coûts de sûr-maintenance. Ces travaux de thèse se concentrent en particulier sur la maintenance en ligne, gérant les échéances de tâches à court et moyen terme (de l'horizon d'une semaine à quelques mois). Cette gestion de la maintenance consiste à répartir les tâches d'inspections régulières dans les créneaux entre deux vols lorsque l'avion est au sol et ainsi d'éviter de l'immobiliser plusieurs heures pour effectuer ces tâches en une seule fois. Par conséquent, cette pratique permet d'augmenter la navigabilité de l'appareil. Toutefois, la faisabilité du planning est primordiale pour ne pas causer de perturbations dans le planning de vols, en s'assurant d'avoir le temps et les ressources nécessaires à l'exécution des tâches. L'exercice de planification peut être complexe et chronophage pour un humain et l'étude de toutes les possibilités de planification rend l'optimisation difficile. C'est ainsi qu'un assistant à la planification peut présenter un réel bénéfice pour les exploitants d'aéronefs pour simuler rapidement des scénarios de maintenance optimisés. Cette thèse formalise ce problème d'optimisation de la maintenance en ligne comme un problème de gestion de projet à contraintes de ressources (RCPSP) étendu avec des contraintes additionnelles liées à la maintenance et considérant des ressources multi-compétences. L'optimisation multi-objectifs de ce problème est gérée par un ordre lexicographique permettant d'ordonner les priorités industrielles. L'étude de méthodes de résolution exacte est proposée au travers d'un modèle de Programmation Linéaire En Nombres Entiers (PLNE) et d'un modèle de Programmation Par Contraintes (PPC). La comparaison des modèles permet de justifier le choix de la PPC pour résoudre ce problème industriel. Une stratégie de recherche, inspirée de connaissances métier, permet de retourner une première solution reproductible dans un temps acceptable pour un expert. La mise en œuvre sur un cas pratique de maintenance met en évidence les avantages de l'approche proposée en la comparant à une approche industrielle heuristique. Enfin, des méthodes de décomposition sont explorées dans l'espoir d'arriver à résoudre de grandes instances. Des modèles en Génération de Colonnes (GC), appliqués à différents types de RCPSP, sont proposés afin d'étudier leur qualité de résolution. Ces études préliminaires permettent de dresser un état des lieux des prochains travaux nécessaires pour modéliser le problème industriel à l'aide de ces méthodes avancées de résolution
Toward a generic information system for projects portfolio management using Physics of Decision (POD)
No full textInternational audienceProject portfolios are complex and unique systems. Projects within the portfolio share the same resources to achieve the set objectives. Monitoring and controlling the progress of these projects requires implementing and understanding techniques such as risk analysis or decision support. In a complex and uncertain world, decision-making is a critical process that can cause uncontrollable and unanticipated disruptions. It is necessary to predict the consequences of and identify the effects of changes within the internal and external environment. However, the uniqueness of project portfolios poses a challenge for prediction because it cannot be based solely on historical data. This research presents a management tool based on decision support techniques and defined rules of behavior that a project portfolio could have. These rules are generic and adaptable and the goal is that managers can use and adapt these. These rules are defined using the Physics of Decision (PoD) vision and the adaptation of this vision using first-order logic. example of its use. approach allows us to present a solid basis for creating a decision support system by being generic to the field of project management but leaving the possibility of adapting the functioning of this tool
Study of Thermomechanical Behavior of Refractory Materials Under Thermal Gradient. Part I – Presentation of ATHORNA Device and Experimental Protocol
No full textInternational audienceBackground: Improving the understanding of how a refractory material responds to thermal shocks and allowing the validation of finite element models require a valuable tool for experimental data collection.Objective: This paper introduces an innovative, sophisticated, and highly reliable experimental device designed to apply a controlled cyclic thermal gradient in a disk-shaped ceramic refractory sample and to simultaneously monitor thermome-chanical response and potential damage.Methods: This device, named Advanced measurements for in-situ Thermomechanical monitORing of large sample uNder thermal grAdient, is based on a CO2 laser beam to generate a calibrated thermal flux sequence at the top face while accu-rately measuring temperature field at the bottom face by an infrared camera. The displacement field of the bottom face is also continuously monitored by a stereo-vision system, enabling a precise measurement of 3D displacements and, thus, of the local strains. An accurate monitoring of the crack extension is performed thanks to the Two-Part Digital Image Correla-tion technique.Results: Throughout the thermal cycling sequence applied to an exemplar sample, the device has proved to be a robust and reliable system able to provide very accurate experiment data in terms of displacement, strain, temperature fields and crack length/opening.Conclusions: This device represents a significant advancement in in-situ monitoring of a refractory sample and contributes to the comprehensive characterization of materials under thermal gradients. More investigations and comparison with ther-momechanical Finite Element modelling are shown in a second part of this paper
Experimental and numerical study on the bending behavior of glass-fabric rubber composites
No full textInternational audienceThe study mainly focuses on the instantaneous mechanical behavior of glass-fabric rubbercomposites. Aside from principal-invariant-driven isotropic hyperelasticity, the strain energycontribution of fabric layer to the overall constitutive behavior of the composite can be defined bypseudo-invariant-driven anisotropic hyperelasticity. The approach accounts for the anisotropy of thesilicone rubber matrix reinforced by plain-wave glass fabric using pseudo-invariants defined by vectors aligned with the warp and weft directions. Subsequently, the composite specimens with 0- and 45-degree fabric orientations were considered for the uniaxial tensile tests on top of the tests on pure rubber under different modes. The parameters of the analytical models with the above-mentioned approach were then identified from the test results. Moreover, different anisotropic hyperelastic material models were scripted and subsequently implemented into Abaqus using the UANISOHYPER_INV material subroutine. After checking the developed material subroutines with the coupon tests, this paper constructively paves a path to delve deeper into bending behavior of fabric rubber composites with the developed methodology accompanied with 3-point bending tests
Donald Katz Award 2025 presentation
No full textInternational audienceIt is an incredible honor to stand here today as the recipient of the Donald L. Katz Award. When I first learned the news, my reaction was pure disbelief—I even double-checked the email to be sure it wasn’t a mistake!To receive an award named after a pioneer like Donald Katz, and to join a list of recipients who are giants in thermodynamics, is truly humbling. My journey with GPA began more than 20 years ago, with projects that were anything but simple—building apparatus, measuring data, and ensuring results mattered for industry. Those early collaborations shaped my career and taught me that molecules are mischievous: they escape, they mix when they shouldn’t, and they always find a way to ruin a perfect curve.Along the way, I’ve been fortunate to work with brilliant colleagues and friends — at IMT Mines Albi, Mines Paris, Heriot-Watt, DTU, and across the GPA community. I owe a special thanks to my mentor Prof Dominique Richon (D. Katz award 2001), to many wonderful collaborators, and to my students, who were the true lab heroes with pipettes and pressure regulators in hand.Some of our proudest achievements include modelling acid gas solubility, uniting kinetics and thermodynamics, developing new solvents, and building apparatus that—while sometimes temperamental—always came through in the end.This award is not just a personal honour; it reflects decades of teamwork, curiosity, and shared passion for understanding complex fluids. Looking ahead, I remain committed to advancing accurate models and reliable data for challenges like carbon capture, decarbonization, and new energy systems.I never set out to win awards—I set out to understand nature, one phase diagram at a time. That this passion has led me here is the greatest reward of all
Estimation de la demande de transport régional de passagers et de marchandises dans un environnement hyperconnecté (thèse sous embargo)
No full textDemand estimation in low-density territories presents a critical challenge for developing accessible, efficient, and sustainable transport services. This thesis proposes a unified framework to estimate and adapt demand by incorporating Physical Internet principles. First, a demand characterization method is developed, combining user profile segmentation with dynamic preference weighting across hours, routes, and transport modes. This module captures behavioral heterogeneity in regional contexts. Second, a utility-based demand estimation model is designed. It simulates mode choice across eight experimental scenarios by integrating different utility calculation approaches, time-dependent service availability, and variability in trip characteristics. Third, a decision-support system is developed to inform strategic, tactical, and operational planning. Implemented as an interactive interface, the system enables users to visualize estimated demand, compare scenarios, and adapt services in real time. The approach was validated through a case study on the French ECOTRAIN project, a modular, autonomous rail shuttle, demonstrating the framework's relevance for supporting multimodal transport planning in low-density networks.L'estimation de la demande dans les territoires à faible densité représente un défi critique pour le développement de services de transport efficaces, accessibles et durables. Cette thèse propose un cadre unifié pour estimer et adapter la demande en intégrant les principes de l'Internet Physique. Premièrement, une méthode de caractérisation de la demande est développée, combinant la segmentation des profils utilisateurs et la pondération dynamique des préférences selon l'heure, le type de trajet et le mode de transport. Ce module permet de capter l'hétérogénéité comportementale dans les contextes régionaux. Deuxièmement, un modèle d'estimation de la demande basé sur l'utilité est conçu. Il simule la sélection modale à travers huit scénarios expérimentaux, en intégrant différentes approches de calcul de l'utilité, la disponibilité horaire des services et la variabilité des caractéristiques de trajet. Troisièmement, un système d'aide à la décision est développé pour soutenir les décisions stratégiques, tactiques et opérationnelles. Ce système, implémenté dans une interface interactive, permet de visualiser la demande estimée, comparer les scénarios et adapter l'offre en temps réel. L'approche a été validée à travers une étude de cas sur le projet français ECOTRAIN, une navette ferroviaire autonome et modulaire, démontrant la pertinence du modèle pour accompagner la planification multimodale dans des réseaux à faible densité
Bad solutions are good to print? 3D printing of cellulose-ionic liquid solution and making aerogels
No full textInternational audiencePolysaccharide-based 3D printed objects are promising materials with customized shapes for life science applications. A classical approach is to use gelation to stabilize the shape of printed solution. To vary solution rheological properties, nanocellulose is often used. In this work, we developed a new approach allowing printing dissolved cellulose with no crosslinking and no additives, only by varying solution rheological properties and adjusting cellulose solubility. These printed cellulose objects were then transformed into aerogels, lightweight and nanostructured materials. In addition, we demonstrated that this approach can be applied not only to neat reference cellulose, but also to dissolved cellulose textile.Microcrystalline cellulose and viscose textile were dissolved in ionic liquid. By adjusting cellulose solubility, solution viscoelastic properties were varied with the goal to obtain yield-stress fluid. The printed objects were of various shapes, composed of many layers and self-standing. The solvent was then washed out by a non-solvent, and cellulose was dried with supercritical CO2. 3D printed cellulose aerogels obtained from MCC and viscose possess porosity higher than 90%, a specific surface area around 250 m²/g and a low density (< 0.2 g/cm3).The developed approach opens new easy ways of printing cellulose solutions and valorizing waste textile.AcknowledgementsWe are grateful to Institut Carnot MINES and PSL for the financial support of this work. We thank Julien Jaxel (PERSEE, MINES Paris) for supercritical drying
Comprendre la cristallisation dans les composites à matrice biosourcée
No full textNational audienceFace à l'urgence environnementale, comment repenser nos modèles industriels pour construire une économie véritablement circulaire ?A l'heure où l'exploitation intensive des ressources naturelles et l'accumulation des déchets atteignent des niveaux critiques, les cadres réglementaires européens et français fixent des objectifs ambitieux : recycler 55 % des emballages plastiques, incorporer 30 % de plastique recyclé dans les bouteilles d'ici 2030, ou encore interdire la destruction des invendus non-alimentaires. Autant de défis technologiques et organisationnels que les industriels doivent aujourd'hui relever. C'est dans ce contexte exigeant que les laboratoires du Carnot M.I.N.E.S ont conduit un programme de recherche structurant sur trois ans, mobilisant un large collectif scientifique. L'objectif était de développer des solutions concrètes pour améliorer la recyclabilité des polymères dans trois secteurs clés : bouteilles, textiles et composites. Au-delà de ce projet central, l'ouvrage présente également d'autres initiatives portés par le Carnot M.I.N.E.S, témoignant de son engagement constant pour des filières plus durables, plus responsables et plus innovantes.Destiné aux industriels, chercheurs, ingénieurs et décideurs, ce livre s'adresse également à toute personne curieuse de comprendre les enjeux techniques, environnementaux et sociétaux de la transition vers une économie circulaire. Il offre une synthèse rigoureuse et accessible des avancées scientifiques et technologiques au service du recyclage de demain : tri optimisé, intégration de matière recyclées, conception d'éco-matériaux et développement de composites recyclables. Un livre essentiel pour penser l'industrie de demain et accompagner les mutations vers un avenir plus soutenable
Comprendre l'ampleur des catastrophes naturelles par l'écoute des réseaux sociaux
No full textInternational audienc
Coupling a DEM model and a homogeneous Markov Chain to accelerate powder mixing simulation
No full textInternational audiencePowder mixing is a key step in many industrial processes. However, it remains difficult to control and predict due to the complexity of granular flows. Several numerical methods, notably DEM, have been developed to simulate particulate mixing. This approach provides access to individual particles positions and velocities, at each time step, at grain scale. Subsequently, DEM is computationally expensive[1] and non-applicable to process scale. Markov chain models, a stochastical approach, offer a macroscopic view of the mixing describing the overall dynamics of particulate systems, with significantly lower computational cost. Markov chain operators constructed based on DEM data were used to study particulate systems in horizontal rotating drums [2], [3], [4]. Mixing behaviors predicted by Markov chain models closely align with DEM results.From a process engineer point of view, investigating mixing at full scale own to limit the computational cost, even if this means losing some of the precision of the final result. This work proposes an approach to predict the dynamics of particle-based systems in a rotating drum using a homogeneous Markov Chain model determined from DEM data. The dense region of the mixture is identified and divided into cells with cylindrical coordinates. At each time step, particle transition probabilities between cells are computed. Several tests were conducted to determine the relevant parameters which are required to compute the average homogeneous transition matrix. The objective was to find a balance between mixing time simulated by DEM required to calculate the transition matrix and results accuracy with DEM reference simulation results.The mixture's homogeneity is assessed over time using well-chosen mixing indexes. The methodology was applied to various initial configurations and three observation scales. The impact of these parameters was investigated, revealing a significant influence on the mixing homogeneity. As a perspective, this methodology could be extended to segregating particulate.REFERENCES[1] F. Bertrand, L.-A. Leclaire, et G. Levecque, « DEM-based models for the mixing of granular materials », Chemical Engineering Science, vol. 60, no 8, p. 2517‑2531, avr. 2005, doi: 10.1016/j.ces.2004.11.048.[2] J. Doucet, N. Hudon, F. Bertrand, et J. Chaouki, « Modeling of the mixing of monodisperse particles using a stationary DEM-based Markov process », Computers & Chemical Engineering, vol. 32, no 6, p. 1334‑1341, juin 2008, doi: 10.1016/j.compchemeng.2007.06.017.[3] J. D. Tjakra, J. Bao, N. Hudon, et R. Yang, « Analysis of collective dynamics of particulate systems modeled by Markov chains », Powder Technology, vol. 235, p. 228‑237, févr. 2013, doi: 10.1016/j.powtec.2012.10.012.[4] Z. Hu et X. Liu, « A novel Markov chain method for predicting granular mixing process in rotary drums under different rotation speeds », Powder Technology, vol. 386, p. 40‑50, juill. 2021, doi: 10.1016/j.powtec.2021.03.041