Portail HAL IMT Mines Albi
Not a member yet
5440 research outputs found
Sort by
Etude prospective d’un environnement d’aide à la décision rendu configurable par des modèles
No full textInternational audienceDecision-making paths, frequency of decision-making, as much as their critical weight, have new prominences in today organizations. We are interested here in the decision-maker's capacity to include the characteristics specific to his/her ecosystem, its context and its often-changing issues in the way he/she practices. Decision-makers are often business experts to whom we want to help the design and adaption of their decision-making tools. The creation of a decision support framework using a Model Driven Engineering (MDE) approach is proposed. It seeks to link models of decision- making needs to logical models required by a combinatorial optimization solver. To do this, the functional and logical architectures of this framework are described on a case study, the traveling salesman problem. The assumptions and knowledge conveyed by the models all along the engineering chain are proposed. The design of the environment follows the principles of the MDA architecture. Several transformation rules between models are then explained. A prototype of this environment was developed using a meta modeler software. It provides a meaningful proof of concept, we mean the capacity to change problem formulations on the fly.Les circuits de décision, le rythme de prise de décision, autant que le poids de celles-ci, prennent de nouvelles dimensions dans les organisations. Nous nous intéressons à la capacité du décideur à inclure les caractéristiques propres à son écosystème, son contexte et ses enjeux, souvent changeants, dans ses pratiques. Les décideurs sont souvent des experts métiers à qui nous essayons de procurer la capacité de concevoir et adapter à moindre effort leurs outils de prise de décision. A cette fin, la création d’un environnement d’aide à la prise de décision utilisant une approche d’Ingénierie Dirigé par les Modèle (IDM) est proposée. Ce travail exploratoire cherche à lier des modèles de besoin de prise de décision à des modèles logiques d’appel à un solveur en optimisation combinatoire. Les architectures fonctionnelles et logiques de cet environnement sont décrites sur un cas d’études, le problème du voyageur de commerce. Nous expliquons les hypothèses et les connaissances véhiculées par les modèles peuplant la chaîne d’ingénierie en suivant les principes de l’architecture MDA. Plusieurs règles de transformation entre modèles y sont définies. Un prototype de cet environnement a été développé à l’aide d’un outil de méta-modélisation. La preuve de concept est sa capacité à reformuler des problèmes en ligne
Extraction générique de connaissances à partir de données textuelles et mesure de la performance des systèmes d’extraction de relations dans un contexte non supervisé.
No full textInternational audienceAmong the incoming challenges in the industrial domain and in the monitoring of industrial systems, the aggregation, synthesis and management of knowledge through ontological structures occupy an essential place. Existing knowledge extraction systems often use a supervised approach which rely on labelled data for which the annotation process is fastidious. This paper presents an unsupervised self-feeding rule-based approach for domain-independent ontology population from textual data. Moreover, the evaluation of such systems, performing knowledge extraction using natural language processing methods requires the use of performance indicators. The indicators usually used in such evaluations have limitations in the specific context of knowledge extraction for unsupervised ontology population. Thus, the definition of new evaluation methods becomes a need arising from the singularity of the harvested data, especially when these are unlabelled. Hence, this article also proposes a method for measuring performance in a context where reference data and extracted data do not overlap optimally. The proposed evaluation method is based on the exploitation of data that serve as a reference but are not specifically linked to the data used for extraction, which makes it an original evaluation method.Parmi les défis à venir dans le monde de l’industrie et dans le pilotage des systèmes industriels, l’agrégation, la synthèse et la gestion des connaissances au travers de structures ontologiques occupent une place primordiale. Beaucoup des systèmes d’extraction de connaissances actuels adoptent une approche supervisée, qui se base sur des données labellisées dont le processus d’annotation est long et fastidieux. Cet article présente une approche par les règles non supervisée, auto-alimentée et indépendante du domaine pour la population d’ontologie à partir de données textuelles. Par ailleurs, l’évaluation de tels systèmes, réalisant de l’extraction de connaissances par les méthodes de traitement automatique du langage, requiert l’utilisation d’indices de performance. Dans le cadre particulier de la population d’ontologie non supervisée, les indices habituellement utilisés pour réaliser ces évaluations présentent des limites dues notamment à l’absence de données annotées. Cet article propose donc également une méthode de mesure des performances dans un contexte où les données de référence et les données extraites ne se recouvrent intrinsèquement pas de manière optimale. Le mode d’évaluation proposé s’appuie sur l’exploitation de données faisant office de références mais qui ne sont pas spécifiquement liées aux données sur lesquelles est réalisée l’extraction, ce qui en fait sa particularité
Rheophysical study of dispersed alumina suspensions
No full textInternational audienceThis paper studies the rheological behavior of aqueous suspensions of fine grained (d50 = 200 nm) alumina. Rheological measurements were performed on suspensions containing various amounts of solid and 0.26 wt% of ammonium polymethacrylate in order to ensure a good state of dispersion. Brownian motions, colloidal interactions and hydrodynamic interactions dictate the behavior. The viscosity is mainly influenced by the shear rate and by the solid volume fraction. A Maron and Pierce model was used to describe this last effect in the hydrodynamic regime. The maximum packing fraction was found to be 40 vol%. This value is related to the colloidal stability as described by the DLVO theory. In fact, the maximum packing fraction leads to a surface-to-surface separation distance of about 46 nm because of the repulsive potential. Finally, a dimensionless approach was achieved to quantitatively identify, on the rheogram, the different flow regimes associated to each dominant interaction type
Energy requirements to produce fine powders of raw and torrefied wood at pilot scale, and characterization of their flowability
No full textInternational audienceThis work aims at comparing different preparation chains to produce wood powders suitable for further gasification in an entrained flow reactor. Three wood powders, with particle size below 1 mm, have been produced at pilot scale from resinous wood chips: a powder of raw wood and a powder of torrefied wood both ground in a knife mill, and a powder of raw wood ground in a vibration mill. The respective requirements both in energy and in feedstock material resource have been determined for each chain of production. The production of the raw wood powder requires 0.83 MWh per ton of dry powder (tdp). The additional grinding step with the vibrating mill adds 0.63 MWhel.tdp−1. The production of the torrefied wood powder requires 3.56 MWh.tdp−1. Moreover, the wet wood chips requirements vary from 2.5 tons for the chains of production without torrefaction to 4.4 tons for the chain including a torrefaction step. In the latter, the step with the highest energy demand is the gas-cleaning step in an instrumented post-combustion reactor. Heat recovery from the combustion gases could supply energy to both the drying and torrefaction steps. It would reduce by half the total energy cost of the chain, down to 1.66 MWh.tdp−1. The resource requirements would be reduced down to 3.2 tons. The morphology and flowability of the powders have been investigated and compared. Torrefaction or vibration milling significantly improve the ability of the wood powder to flow both without any stress or when consolidated by a vertical stress
Supported catalysts for induction-heated steam reforming of methane
No full textInternational audienceNi60Co40 nanoparticles supported on γ-Al2O3 capable of simultaneously catalysing the steam reforming reaction of methane and supplying in-situ the heat necessary to activate the reaction by induction heating, have been synthesized and characterized. Energy is remotely and promptly supplied by an alternating radiofrequency magnetic field (induction heating system) to supported nanoparticles that act as dissipating agents by virtue of their ferromagnetic properties. The temperature reached on the Ni–Co based catalyst surface is high enough to obtain good catalytic performances for the steam methane reforming (SMR). By varying synthesis conditions, samples with two different metal loading (17 wt% and 30 wt%) and different particle size distribution were prepared and characterized. Experimental results evidence that the temperature reached on the catalyst surface is related to the metal loading and to the particles size distribution that strongly affect the ability of ferromagnetic nanoparticles to convert the externally applied radio frequency field into heat. Catalyst pellets proved their effectiveness reaching the temperature of 720 °C during SMR reaction and 80% methane conversion
Benchmarking et selection des technologies de pyrolyse et de gazéification adaptées à la valorisation des CSR et du Bois-B sous forme du gaz
No full textFor a methanation reaction to occur, it is important that the synthesis gas has a minimum H2 / CO ratio of 3 or H2 / CO2 of 4. The higher these ratios, the greater the conversion of carbon oxides to methane. It is also important that the synthesis gas does not contain nitrogen for the reason that this reduces the partial pressure of the main reactive species in the produced syngas. However, due to the common practice of using air as an oxidant in gasification processes, the synthesis gas produced contains about 50% nitrogen. The purification of this gas has significant financial implications. Another important consideration is that the gas that will be in the methanation process must be tar-free. The low particle load of the syngas is also considered to be part of the criteria for measuring the quality of the syngas produced. However, a major constraint in this project is the issues linked to the thermochemical upgrading of SRFs and Bois-B.In this report, therefore, gasification and pyrolysis processes as well as pre-treatment techniques for CSR and Bois-B in order to be able to transform them into high quality syngas has been proposed. Knowing that this syngas must be transformed into biomethane by biological methanation, this report is divided into two parts: parite A - "Benchmarking of Existing Pyrolysis and Gasification Technologies"; and part B ‘Production of Biomethane from Syngas issued from SRFs and Bois-B’.Pour qu’une réaction de méthanation se produise, il est important que le gaz de synthèse ait un rapport minimum H2 / CO de 3 ou H2 / CO2 de 4. Plus ces rapports sont élevés, plus la conversion des oxydes de carbone en méthane est facile. Il est également important que le gaz de synthèse ne contienne pas d'azote pour la raison que ceci réduire la pression partielle des principales espèces réactives dans le syngaz produit. Mais, en raison de la pratique courante d'utilisation de l'air comme oxydant dans les procédés de gazéification, le gaz de synthèse produit contient environ 50% d'azote. La purification de ce gaz a des implications financières importantes. Une autre considération importante est que le gaz qui sera dans le processus de méthanation doit être exempt de goudron. La faible charge de particules du gaz de synthèse est également considérée comme faisant partie des critères de mesure de la qualité du gaz de synthèse produit. Cependant, une contrainte majeure dans ce projet est les enjeux liés à la valorisation thermochimique des CSR et du Bois-B.Dans ce rapport, donc, les technologies différent de gazéification et pyrolyse ainsi que les techniques de prétraitement de CSR et du Bois-B afin de les valoriser sous forme de gaz sont proposés. Sachant que ce syngaz doit être transformé en biométhane par méthanation biologique, ce rapport est devisé en deux partie: parite A - ‘Benchmarking des Technologies Existantes de Pyrolyse et de Gazéification’ ; et partie B ‘La Production de Biométhane à partir du Syngaz issu des CSR et du Bois-B
Experimental study and in-situ FBG monitoring of process-induced strains during autoclave co-curing, co-bonding and secondary bonding of composite laminates
No full textInternational audienceThe geometric stability of bonded structural parts, such as self-stiffened panels, is a critical issue in aeronautics. However, autoclave curing and bonding of large complex integrated primary structures often induce distortions compromising their geometric stability. Composite cure-induced strains and distortions have been extensively studied in the literature but the influence of the adhesive on these distortions has often been overlooked. More experimental data are therefore necessary to properly understand its influence on process-induced strains and its behaviour during cure. The present paper reports a wide range of new experimental investigations showing the non-negligible influence of the adhesive on cure-induced distortions of composite assemblies. The experimental data presented include post-cure distortion measurements, optical microscope imaging, micro-tomography imaging and FBG in-situ strain monitoring during an autoclave bonding process. These experiments demonstrate the impact of the adhesive film behaviour and of the manufacturing bonding process on final distortions and strains of composite assemblies
Chapter 2 - Nanocellulose-based composites: fundamentals and applications in electronics
No full textInternational audienceNanocellulose has been established to be one of the most promising green resources of modern times due to its inherent properties and renewability. In this chapter, we present the fundamentals of nanocellulose-based composite materials and their significance in electronics applications. The composites involving various types, including films, papers, and aerogels comprising nanocellulose incorporated with conducting polymers, carbon nanomaterials, metal nanoparticles and oxides, etc. have been discussed. The underlying mechanism for nanocellulose composites to be utilized for electronics device fabrication includes flexible and printed electronics, energy storage and conversion devices, and dielectric and microwave absorbing systems has been systematically presented. The ability of nanocellulose to provide mechanical supports for freestanding flexible and thermally stable electronic devices has been widely exploited. This chapter aims at providing an outline in the fabrication process and significance of nanocellulose composites in the entire electronics
Production of fuel additives by direct conversion of softwood bark using a cheap metal salt
No full textInternational audienceLevulinates are used as oxygenated fuel additives or as blending components in biodiesel. A metal salt was used for the direct conversion of real biomass, ie. softwood bark, to produce methyl levulinate (ML) and levulinic acid (LA). The experimental data were analyzed through response surface methodology (RSM), using a central composite design (CCD). Three dependent responses (ML yield, LA yield, and residue production) were studied to determine the optimum combination of the four factors. The total yield of levulinates was 62% at the optimum process parameters, including acid concentration (0.067 mol/L), reaction time (5.67 h), and softwood bark concentration (2.5% wt.) at 200 °C. Finally, the results showed that Al2(SO4)3 allowed the production of levulinates for its good BrØnsted/Lewis acidity, but also permitted a decrease in corrosion inside the reactor (as compared to homogeneous acids such as H2SO4). This shows that the use of these metal salts for this specific application could positively affect production costs (either CAPEX or OPEX) on a larger scale
Analyse des facteurs favorisant l’innovation et la réussite des projets complexes afin de proposer un cadre de sélection des partenaires
No full textCompetitive pressures and the need for innovation are shaping strategic partnerships. These partnerships share knowledge, collaborate in project activities, and make joint decisions to achieve complex project objectives. However, in alliances one of the difficulties is achieving effective collaboration: miscommunication, missing skills, missing resources, and lack of trust create a high risk of failure. To decrease this risk, and to overcome potential collaboration inefficiency, partner selection must take place among firms that are able to trust each other and communicate well while at the same time having the required knowledge to achieve their objectives. This thesis proposes a new conceptual model based on hypotheses extracted from the literature review, to investigate first the effect of different knowledge criteria on trust between partners and second the effect of trust on project innovation. We also considered the role of project complexity in project innovation. We analyzed our suggested model using a survey filled by managers of European complex projects conducted with partnerships. Partial Least Squares Structural Equation Modeling (PLS-SEM) is used to analyze the proposed structural model. Then, the correlation between project managers’ work experience and their opinion about the success of a complex project is analyzed. Finally considering the result of our analyses and through a deeper look at the partner selection research, a novel framework is proposed that could help decision-makers of partner selection in alliance formation. The framework also presents potentials for future studies.Les pressions concurrentielles et le besoin d'innovation façonnent les partenariats stratégiques. Ces partenariats partagent des connaissances, collaborent à des activités de projet et prennent des décisions conjointes pour atteindre des objectifs de projet complexes. Cependant, dans les alliances, l'une des difficultés est de parvenir à une collaboration efficace : une mauvaise communication, des ressources et des compétences manquantes et un manque de confiance génèrent un risque. Pour diminuer ce risque, et pour surmonter une inefficacité de la collaboration, la sélection des partenaires doit se faire parmi des entreprises capables de se faire confiance et de bien communiquer tout en ayant les connaissances requises pour atteindre leurs objectifs. Cette thèse propose un nouveau modèle conceptuel basé sur des hypothèses extraites de la revue de la littérature. Elle étudie d'abord l'effet de différents critères de connaissance sur la confiance entre partenaires puis l'effet de la confiance sur l'innovation du projet. Nous avons également considéré le rôle de la complexité du projet dans l'innovation du projet. Une enquête réalisée auprès de gestionnaires de projets européens complexes, réalisés en partenariats, a permis d'analyser le modèle que nous proposons. La modélisation des équations structurelles par les moindres carrés partiels (PLS-SEM) est utilisée pour analyser le modèle structurel proposé. Ensuite, la corrélation entre l'expérience professionnelle des chefs de projet et leur opinion sur le succès d'un projet complexe est analysée. Enfin, compte tenu des résultats de nos analyses et d'un examen plus approfondi de la recherche sur la sélection des partenaires, nous proposons un nouveau cadre permettant d'aider les décideurs à sélectionner les partenaires dans la formation d'alliances. Ce cadre présente également des possibilités d'études futures