Portail HAL IMT Mines Albi
Not a member yet
5440 research outputs found
Sort by
Modélisation de la diffusion optique pour la mesure de température par thermoréflectométrie polychromatique
No full textTemperature measurement without contact, based on the observed radiance scene, is a valuable tool to optimize and control a process, and to anticipate material damages. The main problematic of these passive methods is the requirement of a previous knowledge of the emissivity of the targeted surface, which can evolve, especially at high temperature. Thermoreflectometry is an active method of true temperature measurement that does not require knowing the material emissivity. This last is measured on line, in an indirect way, by measuring the bidirectional reflectivity completed by a diffusion function model, which parameters are identified. The contributions of this thesis are the proposed models of diffusion function based on a physical description of the light/matter interaction on metallic surfaces. The approach of this work is to identify the models from geometrical optics and physical optics, with their validity domain and apply them to the case of rough surfaces, described statically by roughness parameter. The reflectivity distribution function (BRDF) is then simulated and used for the calculation of the spectral diffusion function, used in thermoreflectometry. These models are validated from roughness and spectrometric measurements in the near and mid infrared domain on metallic samples, covering a wide range of optical behaviors. Afterwards these models are applied to thermoreflectometry and their contributions (true temperature and model parameter identification) are evaluated. This validation is performed by simulation, using data from various materials characterizations, and experimentally with the thermoreflectometer on heated samples. The results demonstrated the well-founded of the approach and the developed device. Finally, it opens perspectives of the in-situ analysis of the surface state by the identification of the parameters of the physical model of the diffusion function.La mesure de champs de température sans contact, basée sur l'analyse de la luminance de la scène observée avec une caméra infrarouge, s'avère précieuse pour optimiser, contrôler des procédés et anticiper l'endommagement des matériaux observés. La principale problématique de ces méthodes passives est qu'elles impliquent une connaissance a priori de l'émissivité des surfaces en présence, qui sont sujettes à des évolutions, en particulier à haute température. La thermoréflectométrie est une méthode active de mesure de champs de température vraie sans connaissance a priori de l'émissivité de la surface du matériau. Cette dernière est mesurée en ligne de manière indirecte par une mesure de la réflectivité bidirectionnelle complétée par un modèle de fonction de diffusion dont les paramètres sont identifiés. La contribution de cette thèse est la proposition de modèles physiques de fonction de diffusion basés sur la description de l'interaction rayonnement/matière à la surface d'un matériau métallique. La démarche consiste à identifier les modèles issus de l'optique géométrique et/ou ondulatoire, ainsi que leur domaine d'application, et d'extrapoler leur application à des surfaces décrites statistiquement selon leurs paramètres de rugosité pour simuler la fonction de distribution des réflectivités (BRDF). A partir de cette grandeur est calculée la fonction de diffusion spectrale, intervenant en thermoréflectométrie. Ensuite, ces modèles sont validés à partir de mesures de rugosité et de mesures spectrométriques, sur un domaine spectral du proche au moyen infrarouge, ceci sur des matériaux métalliques couvrant une large gamme de comportements optiques. Dans une seconde étape, ces modèles sont introduits dans le système de thermoréflectométrie polychromatique, et l'évaluation de l'apport pour la mesure de température et du suivi potentiel de l'état de surface est menée. Cette validation est conduite à la fois sur des données de simulations issues de multiples caractérisations des matériaux et sur des données expérimentales issues de mesures acquises sur un banc dédié. Les résultats révèlent le bien fondé et la pertinence de la démarche et du système développé. Enfin ils ouvrent des perspectives pour le suivi "in-situ" de l'évolution de l'état de surface des matériaux via l'identification des paramètres des modèles physiques de la fonction de diffusion
On learning deep domain-invariant features from 2D synthetic images for industrial visual inspection
No full textInternational audienceDeep learning resulted in a huge advancement in computer vision. However, deep models require a large amount of manually annotated data, which is not easy to obtain, especially in a context of sensitive industries. Rendering of Computer Aided Design (CAD) models to generate synthetic training data could be an attractive workaround. This paper focuses on using Deep Convolutional Neural Networks (DCNN) for automatic industrial inspection of mechanical assemblies, where training images are limited and hard to collect. The ultimate goal of this work is to obtain a DCNN classification model trained on synthetic renders, and deploy it to verify the presence of target objects in never-seen-before real images collected by RGB cameras. Two approaches are adopted to close the domain gap between synthetic and real images. First, Domain Randomization technique is applied to generate synthetic data for training. Second, a novel approach is proposed to learn better features representations by means of self-supervision: we used an Augmented Auto-Encoder (AAE) and achieved results competitive to our baseline model trained on real images. In addition, this approach outperformed baseline results when the problem was simplified to binary classification for each object individually
A Heuristic-Based Genetic Algorithm for Scheduling of Multiple Projects Subjected to Resource Constraints and Environmental Responsibility Commitments
No full textInternational audienceDuring the last decades, resource-constrained project scheduling problems have been abundantly presented in extant literature. However, there are still some real-world challenges that have not been adequately considered. These challenges include environmental commitments and constraints related to the procurement of resources (as regards procurement commitment). This calls for the integration of the project planning and forward-reverse supply chain planning systems. To achieve this goal, this paper contributes to the existing literature by presenting a model that incorporates the two issues in the integrated planning system: (1) the procurement commitment objective is met through the just-in-time delivery of non-renewable resources to the project sites while considering the limited supply capacity of suppliers, and (2) the environmental commitment is satisfied by collecting and recycling the waste generated at project sites. A mixed-integer linear formulation of the problem is proposed. Since the model is NP-hard (non-deterministic polynomial time-hard), the paper develops a new heuristic-based genetic algorithm to solve the problem instances. The main parameters of the algorithm are tuned using the Taguchi method. The results show the efficiency of the algorithm in obtaining appropriate solutions in reasonable computational times. The integrated planning model that is proposed in this paper and its novel resolution method would help managers to make more responsive and efficient decisions
Ultrasound-assisted Extraction Process of Glycosides from Stevia Rebaudiana Bertoni Leaves
No full textInternational audienceStevia Rebaudiana Bertoni leaves contain glycosides that have a sweetening power of 100 to 300 times greater than sucrose. These glycosides are increasingly used as natural sweetener, mainly to replace aspartame, in order to reduce daily caloric intake. The two major glycosides are Stevioside and Rebaudioside A, which represent about 90 % [w/w] of total Steviol. The Rebaudioside A is the most interesting since it has the advantage of not having a bitter aftertaste, unlike stevioside. In recent years, the cultivation of organic stevia is growing in the Occitanie region in France. Thus, the valorization of this production chain is of economic interest for this region. It's known that conventional extractions to obtain these glycosides are time consuming and the extensive use of organic solvents may be harmful for human health and environment. Thus, the objective of this work was to develop a green extraction process suitable for industrial purposes, using water as solvent and ultrasound technique to obtain high yield of glycosides in the Stevia extract. The extractions are carried out in batch with recirculation and continuous assemblies at 45 °C. The effects of ultrasound power (at 20kHz), experience time/ residence time, and leaves concentration are studied. The soluble solids and stevia glycosides concentration are analyzed respectively by °Bx and HPLC. With the continuous assembly the experiments were carried out at 64 W and 144 W, which represents 50 % and 100 %, respectively, of the generator capacity. A low flowrate (or high residence time) is detrimental to the use of ultrasound, leaves sedimentation and their accumulation prevents probably the ultrasound diffusion in the medium, harming the steviol extraction. The extraction ultrasound-assisted allowed to increase in 9-10 % °Bx concentration for batch and continuous assembly, compared to extraction without ultrasound under the same conditions. Soluble species other than steviols are extracted with ultrasound. The HPLC results showed that the ultrasound favored lightly the extraction of stevioside
Simultaneous hydrogen sulfide removal and wastewater purification in a novel alum sludge-based odor-gas aerated biofilter
No full textInternational audienceA novel alum sludge-based odor-gas aerated vertical flow biofilter (Al-OAF) was developed, which aims to removal the pollutants in wastewater and simultaneously to eliminate H2S generated from wastewater treatment facility. Three lab-scale parallel columns were operated in batch model while intermittently aerated with 200 ppm H2S (Al-OAF), air (Al-AF) and unaerated (Al-F, as blank), respectively. The pollutants in wastewater and the effluent H2S concentration from Al-OAF were monitored regularly. Results showed that three columns presented a high removal efficiency (>98%) of total phosphorus (TP) and a completed removal of H2S (100%) in Al-OAF. Al-OAF and Al-AF could enhance the removal efficiency of chemical oxygen demand (COD) of 94.3 ± 3.0, 94.8 ± 1.9%, and total nitrogen (TN) of 86.2 ± 14.2, 91.6 ± 5.4%, respectively. In particular, there was no significant difference regarding the COD, TN removal performances between the “H2S driven” Al-OAF and the “air driven” Al-AF. The H2S removal mechanism lies in the alum sludge ability of H2S adsorption and the reaction by the biofilm in the biofilter. This demonstrates that the novel Al-OAF (aerated with waste gas) would be a promising “wise choice” for intensified biofilter with dual-goals of simultaneous wastewater purification and H2S elimination
Influence of composite preform fabrication on the integrity of poly‐aryl‐ether‐ketone matrices
No full textInternational audiencePoly-Aryl-Ether-Ketones (PAEK) matrices of different composite preforms were evaluated to assess the influence of the preform manufacturing on the matrix integrity. Yarns, films, and impregnation suspensions were character- ized by gel permeation chromatography, Fourier transform infrared spectros- copy, thermogravimetric analysis, differential scanning calorimetry, and rheometry. As described in literature, changes in the matrix properties at mol- ten state are observed for all materials. The degradation process is related to a crosslinking process that results in an increase of viscosity and a decrease of crystallization temperature. Differences in degradation behavior were observed depending on the type of preform. Film and yarn extrusion has a limited effect on PAEK behavior, apart from a slight shift in molecular weight, without sig- nificant effect on degradation kinetics. However, the organic products that are used for the preparation of semifinished composite products, that is, yarn sizing or suspension additives, have a detrimental effect on stability. The deg- radation of additives considerably increases the degradation kinetics of the PAEK matrices, and by modifying the thermo-oxidation mechanisms, the advantages of using complex preforms for consolidation can be annihilated
Influence of the cooling rate below Ms on the martensitic transformation in a low alloy medium-carbon steel
No full textInternational audienceThe influence of cooling rate below the martensite start temperature, Ms, on the kinetics of martensitic transformation in a medium carbon low alloy steel was determined using high resolution dilatometry, optical microscopy and X-ray diffraction techniques. A two-stage transformation was observed for slow cooling rates while at higher cooling rates, martensitic transformation occured through a single stage process. It is shown that the Koistinen–Marburger equation cannot adequately describe the observed two-stage transformation. A new equation is proposed in order to model the evolution of martensitic transformation by considering the influence of post Ms cooling rate. The method considers contributions from both, the initial austenite and the carbon enriched austenite. The underlying mechanisms are discussed and validated with experimental findings
Performance of lightweight mortar reinforced with doum palm fiber
No full textInternational audienceNatural fibers are increasingly used in composites because of their low cost and good mechanical properties. Cement reinforced with natural fibersis contemplates as a new generation of construction materials with superior mechanical and thermal performance. This study of three sizes'effect of Doum palm fiber explores the mortar's behavior reinforced with different fiber ratio. The aim is to determine the optimal addition to improve mechanical and thermal properties of natural fiber reinforced cements. Physical, mechanical and thermal properties of composite are examined. Tensile properties of Doum fibers are verified to determine their potential as reinforced material. Findings prove that the use of alkali-treated Doum fiber as reinforcement in cement mortar composite leads to the upgrading of the mechanical properties including thermophysical properties against composites reinforced with raw fibers and control cement mortars. While, the compression and flexural strength of the cement mortar reinforced with alkali-treated Doum fiber with diameter 0.3 mm (CT3) are metered to be 11.11 MPa, 5.22 MPa, respectively for fiber content 0.5%. Additionally, based on thermo-physical tests, it is assessed that the thermal conductivity and diffusivity decrease for cement mortar reinforced with Doum fiber with diameter 0.2 mm (CT2)
Parametrized SEIR model for performance-based decision support: A case study of COVID-19 epidemic in the state of Georgia (USA)
No full textInternational audienceThe question of the impact of pandemics on human society is clearly of first priority. This paper aims at building on top of the classical SEIR pandemic model paradigm in order to bring a new vision of decision support. The gen-eral principle is to use System Dynamics modeling but the originality is that the impact of a pandemic is assessed through the deviation of the per-formance trajectory of the social system within the geometrical framework of its Key Performance Indicators (KPIs). Consequently, the model is based on two types of parameters: (i) the ones from the social system, and (ii) the ones from the pandemic. The simulation approach provides the values of the KPIs all along the observed period. The accuracy of the approach is fi-nally evaluated through the comparison with the actual values of the state of Georgia during one year of the COVID-19 crisis (03/2020-03/2021)
Etude et optimisation des conditions de refroidissement de pièces en alliage de titane obtenues par le procédé de formage superplastique
No full textThe new programs of aeronautical industry use titanium alloy parts for many applications. This is particularly the case for panels, air inlets and fairings parts of pylons. Some of these parts are obtained by hot forming, in particular by the superplastic forming process (SPF). The analysis of the different manufacturing operations has highlighted variability in the operating conditions during the unmolding process. One way of improvement is to a better control of the conditions of cooling and handling of the parts after the forming stage. The purpose of this thesis is to implement a finite element model to reproduce the pot-forming steps of sheet metal in Ti-6Al-4V titanium alloy formed with the SPF process. The cooling step combined with the mechanical loading encountered during the extraction generates distortions of the parts at the end of the manufacturing process. A first step consisted in characterizing the material behavior under the process conditions so as to identify a rheological law implemented in a finite element code. A second step aimed to reproduce, measure and characterize the heat transfers present during the different phases of unloading to define the thermal boundary conditions of the model. Finally, a global model of the forming, unmolding and trimming steps was developed and applied to an industrial part. This model has enabled to highlight the importance of the thermomechanical loading undergone by the part during the cooling and part removing operations.Les nouveaux programmes de l'industrie aéronautique font appel pour de nombreuses applications à des pièces en alliage de titane. C'est notamment le cas pour des pièces d'entrée d'air et carénages du mât réacteur. Certaines de ces pièces sont obtenues par mise en forme à chaud, notamment par le procédé de formage superplastique (SPF). L'analyse des différentes opérations de fabrication a mis en évidence des variabilités de conditions opératoires lors du défournement. Une des voies d'amélioration concerne une meilleure maîtrise des conditions de refroidissement et de manipulation des pièces après mise en forme à chaud. Cette thèse a pour but de développer un modèle éléments finis pour modéliser l’étape de défournement de tôles en alliage de titane Ti-6Al-4V formées avec le procédé SPF. Cette étape de refroidissement, combinée au chargement mécanique subi lors de l’extraction, génère des distorsions des pièces en fin de processus de fabrication. Une première étape a consisté à caractériser le comportement du matériau dans les conditions du procédé de façon à identifier un modèle de comportement qui est ensuite implémenté dans un code éléments finis. Une deuxième étape a visé à reproduire, mesurer et caractériser les échanges thermiques en présence lors des différentes phases du défournement pour définir les conditions limites thermiques du modèle. Pour finir un modèle global du procédé de formage, défournement et détourage a été développé et appliqué à une pièce industrielle. Ce modèle a permis de mettre en évidence l’importance du chargement thermomécanique subi par la pièce au cours du défournement