Portail HAL IMT Mines Albi
Not a member yet
5440 research outputs found
Sort by
Etude et définition d'une approche de modélisation et d'orchestration simultanée, agile, et sociale des processus métier, basé sur l'incomplétude, la flexibilité et l'adaptabilité
No full textBusiness Process Management (BPM) plays a vital role in organizations, overseeing the entire lifecycle of business processes to deliver valuable services and products to end-users. However, the rapidly changing business landscape (market volatility, extreme product/experience personalization) presents many diverse business processes alongside uncertainties, necessitating responding to changes reactively and efficiently. In a word, BPM needs to be agile. Agile BPM will enable organizations to effectively manage, execute, and adapt to various process variations while addressing unexpected changes. Indeed, despite the rigour and formalization they provide, business processes are generally seen as restricting creativity, curbing initiative, and constraining autonomy. Business processes are challenging to design, requiring business experts' interviews and data analysis to identify and document the targeted business processes. Their orchestration is also challenging, especially to face instability in the business process execution. These observations lead to the following research questions: (i) How to bring agility to BPM? (ii) How might tools and technology bring agility to business processes and BPM orchestrators? An agile and social business process management approach is proposed to address these questions. One of the critical aspects of agile BPM is the active involvement of stakeholders throughout the business process (BP) lifecycle. Achieving a standardized approach for stakeholders to access and participate in the BP lifecycle is essential to avoid issues like cultural resistance. In this context, Social BPM emerges as a valuable concept, facilitating stakeholders' seamless and efficient engagement in the BP lifecycle. This thesis begins by offering a comprehensive and precise definition of agility to embrace an agile and social BPM approach. It further introduces an agility framework, guiding experts and researchers seeking to implement and contribute to agility within BPM. Drawing inspiration from Social BPM, this study introduces a novel BPM platform that leverages social media capabilities to manage diverse BPs and uncertainties while integrating stakeholders into the BPM lifecycle. The proposed platform enhances communication, information sharing, and decision-making among stakeholders, removing the boundaries between design time and run-time when changes occur by merging these phases. This platform is meticulously aligned with the proposed agility framework, enabling it to adapt to changes across various timeframes. As a future perspective and extension of the proposed social and agile BPM platform, this study explores the potential role of AI agents (such as conversational agents) in enhancing the platform's overall functionality and its implementation of the agility framework. This analysis highlights the significance of artificial intelligence in further improving the platform's capabilities and effectiveness on various.La gestion des processus métier (BPM) joue un rôle essentiel dans les organisations, en supervisant l'ensemble du cycle de vie des processus métier afin de fournir des services et des produits de qualité aux utilisateurs finaux. Cependant, l'évolution rapide du paysage commercial (volatilité du marché, personnalisation extrême des produits/expériences) présente de nombreux processus métier différents ainsi que des incertitudes, ce qui nécessite de répondre aux changements de manière réactive et efficace. En un mot, le BPM doit être agile. Le BPM agile permettra aux organisations de gérer, d'exécuter et de s'adapter efficacement à diverses variations de processus tout en faisant face à des changements inattendus. En effet, malgré la rigueur et la formalisation qu'ils apportent, les processus métier sont généralement perçus comme limitant la créativité, l'initiative et l'autonomie. Les processus métier sont difficiles à concevoir, nécessitant des entretiens avec des experts métier et une analyse des données pour identifier et documenter les processus métier ciblés. Leur orchestration est également difficile, en particulier pour faire face à l'instabilité dans l'exécution des processus d'entreprise. Ces observations conduisent aux questions de recherche suivantes : (i) Comment apporter de l'agilité au BPM ? (ii) Comment les outils et la technologie peuvent-ils apporter de l'agilité aux processus métier et aux orchestrateurs BPM ? Une approche agile et sociale de la gestion des processus métier est proposée pour répondre à ces questions. L'un des aspects essentiels de la gestion agile des processus métier est l'implication active des parties prenantes tout au long du cycle de vie du processus métier. Il est essentiel de mettre en place une approche normalisée permettant aux parties prenantes d'accéder et de participer au cycle de vie du processus métier afin d'éviter des problèmes tels que la résistance culturelle. Dans ce contexte, le BPM social apparaît comme un concept précieux, facilitant l'engagement transparent et efficace des parties prenantes dans le cycle de vie des processus métier. Cette thèse commence par offrir une définition complète et précise de l'agilité afin d'adopter une approche agile et sociale de la gestion des processus métier. Elle introduit ensuite un cadre d'agilité, guidant les experts et les chercheurs qui cherchent à mettre en œuvre et à contribuer à l'agilité au sein du BPM. En s'inspirant du BPM social, cette étude présente une nouvelle plateforme de BPM qui exploite les capacités des médias sociaux pour gérer divers processus métier et incertitudes tout en intégrant les parties prenantes dans le cycle de vie du BPM. La plateforme proposée améliore la communication, le partage d'informations et la prise de décision entre les parties prenantes, supprimant les frontières entre le moment de la conception et le moment de l'exécution lorsque des changements se produisent, en fusionnant ces phases. Cette plateforme est soigneusement alignée sur le cadre d'agilité proposé, ce qui lui permet de s'adapter aux changements dans différents horizons temporels. Dans une future perspective d'extension de la plateforme BPM sociale et agile proposée, cette étude explore le rôle potentiel des agents d'intelligence artificielle (tels que les agents conversationnels) dans l'amélioration de la fonctionnalité globale de la plateforme et de sa mise en œuvre du cadre d'agilité. Cette analyse met en évidence l'importance de l'intelligence artificielle dans l'amélioration des capacités et de l'efficacité de la plateforme
Microstructure and Mechanical Properties of Hybrid LPBF-DED Inconel 625
No full textInternational audienceThe hybridization of Laser Powder Bed Fusion and Direct Energy Deposition could increase the application range of additive manufacturing by benefiting from the resolution of LPBF and the flexibility of DED. However, the microstructures and mechanical properties obtained by these processes are very different. This causes the hybrid parts to be very heterogeneous. The mechanical behavior of hybrid LPBF-DED Inconel 625 parts are investigated under static and cyclic loading. Both as-built and heat-treated Inconel 625 are studied to investigate the role of the specific as-built microstructure. Tensile tests are performed using Digital Image Correlation, which allows the global behavior to be explained by the local behavior of the samples. Hybrid samples are also tested under high cycle fatigue, and fractographic observations permit to determine the initiation sites and mechanisms. The DED is found to be the limitation in both the static and the fatigue strengths of hybrid samples. Hybrid samples perform as well as or better than mono-process DED. In static loading, the local behavior of the hybrid samples and the yield tensile strength gap between LPBF and DED determine the global behavior. The heat-treatment is successful in reducing this gap, and improving the global behavior of the hybrid samples, especially the total elongation. In fatigue, the defects are preponderant compared to the mechanical heterogeneity of hybrid samples
Potentiel de la CIN-EF pour la caractérisation mécanique de renforts textiles
No full textFor composite materials with a thermoplastic matrix, the shaping of the textile reinforcement has a direct impact on the mechanical properties of the consolidated composites. However, measuring textile reinforcement materials using non-destructive techniques poses challenges due to their physical properties, structure, large deformation, and complex kinematics. To address this measurement error issue, the performance of "global" digital image correlation by finite elements (FE-DIC) is parametrically studied from various perspectives in relation to the mechanical properties of textile reinforcements. This research project is divided into three distinct parts. The first part involves conducting bias tests on several typical reinforced fabrics, such as glass fabric, carbon fabric, and carbon/PPS fabric. The measurement method entails determining the displacement field using the FE-DIC method to regularize the displacement field and exchange it with the facilitated simulation. The analysis of results involves calculating and comparing the ultimate error and residuals by varying the mesh size and mesh orientation to identify the most suitable parameters. The second part proposes two FE-DIC-based algorithms qualified by correlation residuals, which are used to reduce errors resulting from large deformations. The last part proposes an in situ test approach based on X-ray micro-tomography at the yarn scale, providing a feasible method to study inter-yarn kinematics and establish a link between experiments and numerical simulations. The results of these three research subjects significantly contribute to reducing measurement errors associated with tracking displacement fields and deformation fields.Pour les matériaux composites à matrice thermoplastique, tout comme pour d'autres types de matériaux composites, la mise en forme du renfort textile a un impact direct sur les propriétés mécaniques des matériaux composites consolidés. Dans le but de réduire les erreurs de mesure, les performances de la corrélation "globale" d'images numériques par éléments finis (CIN-EF) sont étudiées de manière paramétrique, en se focalisant sur les propriétés mécaniques des renforts textiles sous différentes perspectives. Ce projet de recherche est divisé en trois parties distinctes. La première partie concerne les essais de traction réalisés sur plusieurs types de tissus renforcés courants, tels que le tissu de verre, le tissu de carbone et le tissu de carbone/PPS. La méthode de mesure consiste à mesurer le champ de déplacement à l'aide de la méthode CIN-EF afin de régulariser le champ de déplacement et d'échanger avec les simulations assistées. Dans l'analyse des résultats, l'erreur ultime et les résidus sont calculés et comparés en faisant varier respectivement la taille du maillage et l'orientation du maillage, afin de trouver les paramètres les plus appropriés. La deuxième partie consiste à proposer deux algorithmes basés sur CIN-EF qualifiés par les résidus de corrélation, qui sont utilisés pour réduire les erreurs générées par les grandes déformations. Enfin, la dernière partie propose une approche de test in situ basée sur la micro-tomographie aux rayons X à l'échelle de la chaîne, qui fournit une méthode réalisable pour étudier la cinématique inter-chaînes et établir un lien entre les expérimentations et les simulations numériques. Les résultats de ces trois sujets de recherche permettent de réduire de manière significative les erreurs de mesure issues du suivi des champs de déplacement et des champs de déformation
Impact des systèmes cyber-physiques sur le système de santé
No full textInternational audienceComme dans le domaine industriel, mais sans doute avec beaucoup moins de maturité, les Systèmes Cyber-Physiques de Santé (HCPS) constituent un élément clé de l’évolution et de la modernisation du système de santé. Cependant, il est délicat et risqué d’imaginer une transposition directe des systèmes industriels cyber-physiques (ICPS) sans prendre en compte les spécificités du système de santé, présentées dans ce chapitre. Le terme ” système de santé ” désigne l’ensemble des organisations, institutions, ressources et personnes dont l’objectif principal est d’identifier et de satisfaire les besoins de santé de la population
Physics of Decision: managing and preparing critical supply chains to supply disruptions
No full textInternational audienceToday, supply chains face many uncertainties and making well-informed decisions requires performant decision support systems and methods. The purpose of this study is to apply a new perspective of decision support: the Physics of Decision (PoD). This approach considers risks or opportunities (potentialities) as physical forces and which are assessed regarding their intensity and contribution towards or as deviations of the system’s performance trajectory compared to a target. Such an evaluation permits studying the effect of different mitigation actions to support the decision-making process and proritize corrective measures. The approach is applied to an aerospace manufacturing case study facing a supply shortage, a high stake in this secto
Capacity Planning for Ambulatory Surgeries in Collaborative Network of Hospitals
No full text© 2023 IFIP International Federation for Information ProcessingPart 13: Collaborative Networks in Personalized HealthcareInternational audienceIn this study, we investigate the required capacity of the ambulatory surgery units in a network of hospitals considering different outpatient surgeries within various specialties. The objective is to find the minimum cost capacity decisions that satisfy the increasing outpatient demand and keep the ‘waiting times to surgery’ below the specified thresholds. We also explore the potential improvement of a collaborative network to increase the resilience of day surgery services and prepare the territories for the expected increase in outpatient demand. For this, we propose integer programming optimization models for non-collaborative and collaborative networks. We demonstrate the benefits of potential collaboration between hospitals through a numerical study
Contribution of high-temperature 3D-Digital Image Correlation to investigate the creep behaviour of refractory materials
No full textInternational audiencePotential contributions of 3D-Digital Image Correlation are investigated to understand the dilatant behaviour of a refractory concrete during high-temperature creep experiments. An experimental analysis of high-temperature image quality is performed. The measurement uncertainty is assessed and compared to the sensitivity of computed kinematic fields to the parameters of the Drucker-Prager model that control the dilatancy. Promising results are found i) concerning deflection and neutral axis location measurement in bending creep tests and ii) contactless measurement of axial strain in compression creep tests, iii) to support the assumption of dilatant behaviour from bending and diametral compression tests and iv) to provide data for a rough parameter identification from compression creep tests
Origin of Stereoselectivity in a Mechanochemical Reaction of Diphenylfulvene and Maleimide
No full textInternational audienceMechanochemical reactions sometimes yield unexpected products or product ratios in comparison to conventional reaction conditions. In the present study, we theoretically reveal the origin of the mechanochemical selectivity by considering the Diels–Alder reaction of diphenylfulvene and maleimide as an example. The application of an external force is equivalent to the production of a structural deformation. Here, we show that a mechanical force applied in a direction orthogonal to the reaction mode can lower the activation barrier by varying the potential energy curvature in the transition state. In the case of the Diels–Alder reaction, the endo-type pathway was found to be more mechanochemically favorable than the exo-type pathway, which is consistent with the experimental observations
Reduced mechanical models of trunk–lumbar belt interaction for design-oriented in-silico clinical trials
No full textInternational audienceLumbar belts are one of the possible indication among therapeutic strategies for low back-pain management, but few studies and insufficient data have been reported to evaluate their clinical effects.Recently, a semi-analytical model has been developed for in-silico trials of lumbar belts on patients' morphologies (Molimard et al., 2019a), incorporating for the first time the trunk deformation and providing an estimation of the pressure applied on the trunk. A unique indicator of the belt mechanical efficiency was proposed: Pressure is integrated into a bending moment, calculated at the center of curvature in the lumbar region, characterizing the action of reducing lumbar spine lordosis. One important finding of this study indicates that patient BMI impacts belt efficacy, suggesting that overweight is predictive of decreased belt effectiveness. However, more morphologies in the high Body Mass Index (BMI) range were needed to confirm this trend, showing the importance of controlling the cohort characteristics.To investigate this association between belt efficacy and morphological indicators, 24 different morphologies were artificially generated using a Proper Orthogonal Decomposition (POD) from a base of 13 experimental trunk geometries, with most of them being overweight. They were used for a new in-silico clinical trial using one lumbar belt model and the waist-to-hip (WHR) ratio as a morphological indicator. The results confirmed our hypothesis showing that an increase in WHR is associated with a decrease in bending moment. In addition, no difference in these trends could be found between experimental and in-silico patients, nor between men and women
Paramétrisation du DDMRP avec l' apprentissage par renforcement
No full textInternational audienceConsidering more demanding customers and the diversity of the conventional products, the industrials face new stakes of production and lead times. Nevertheless, the production methods cannot reach those new goals anymore. The Demand Driven Material Requirements Planning (or DDMRP) is a demand-driven production method, which is included in a new era of industrial innovation. However, little attention has been given to the parametrization of DDMRP. This study aims to dynamically adjust order spike thresholds and horizons to improve the parametrization of the method. The proposed methodology is to integrate a reinforcement learning method to the simulation model of an hybrid DDMRP-run flowshop, subject to a peak demand distribution. We study the performance of the learning process and the industrial indicators of the flowshop. We manage to show that it is possible to adjust the parameters of the flowshop while improving its performance regarding customer satisfaction and inventory levels. The results of the study point out the possibility to drive DDMRP parameters with an automatic method using reinforcement learning.Avec l’augmentation des exigences clients et de la diversité des produits conventionnels, les industriels font face à de nouveaux enjeux de production et de délais. Néanmoins, les méthodes de production ne peuvent plus répondre à ces nouveaux enjeux. Le Demand Driven Material Requirements Planning (ou DDMRP) est une méthode de production pilotée par la demande qui s’inscrit dans une nouvelle aire d’innovation industrielle. Cependant, peu d’attention a été accordée à la paramétrisation du DDMRP. Cette étude vise à apporter des ajustements dynamiques à des seuils et horizons de détection de pics afin d’améliorer la paramétrisation de la méthode. La méthodologie proposée est d’intégrer un algorithme d’apprentissage par renforcement au modèle de simulation d’un atelier hybride piloté en DDMRP soumis à des pics de demande. Nous étudions la performance de l’apprentissage, ainsi que l’évolution des indicateurs industriels de l’atelier. Nous parvenons à montrer qu’il est possible de piloter les paramètres de l’atelier tout en améliorant ses performances en termes de satisfaction client et de niveaux d’inventaire. Les résultats de l’étude démontrent la possibilité de piloter les paramètres d’un DDMRP avec une méthode automatique d'apprentissage par renforcement