Portail HAL IMT Mines Albi
Not a member yet
5440 research outputs found
Sort by
Synthesis/separation of 5-hydroxymethylfurfural converted from fructose promoted by H2O–CO2 biphasic system with solid catalysts: Experimental and kinetic modeling approaches
No full textInternational audience5-Hydroxymethylfurfural (5-HMF) is an industrially valuable compound typically produced by dehydration of fructose converted from cellulosic biomass. However, the formation of 5-HMF is accompanied by its conversion into by-products, such as levulinic and formic acids, decreasing its yield. Therefore, biphasic processes are attracting much attention allowing the efficient simultaneous synthesis and separation of 5-HMF. In this study, a biphasic system using water with high-pressure carbon dioxide, the most environmentally friendly solvents, combining with a solid catalyst is used to achieve 5-HMF synthesis and separation at lower temperatures and shorter duration than previous studies. The main experiment was operated by a semibatch reactor with a continuous flow of CO2. Additionally, three experiments, i) measurement of the partition coefficient of 5-HMF between the two phases, ii) 5-HMF synthesis in H2O–CO2 biphasic batch system, and iii) 5-HMF extraction in H2O–CO2 biphasic semibatch system, were conducted to complement the main experiment. All results from complement experiments were used as parameters in the kinetic modeling. As the result, the maximum yield of 5-HMF (36.6 %) was achieved at a temperature of 403 K, a pressure of 25 MPa, and a reaction time of 180 min. Finally, the optimal conditions for the semibatch experiment for 5-HMF synthesis and separation could be estimated from the kinetic models with parameters determined in this work. The obtained results suggested that this yield can be further improved by increasing the reaction time, which should be confirmed by further experimental studies
Nickel and Iron‐Doped Biocarbon Catalysts for Reverse Water‐Gas Shift Reaction
No full textInternational audienceBiocarbon catalysts for reverse water‐gas shift reaction (RWGS) were produced from pyrolyzed fern and willow impregnated with iron and nickel nitrates. This reaction can partake during Fischer‐Tropsch synthesis (FTS) by consuming CO2 and lowering both the H2/CO ratio and the efficiency in the production of fuels. RWGS has attracted much attention to widespread utilization of CO2 through the production of syngas. The catalysts were therefore tested in a fixed‐bed reactor at 400°C as it is the maximal temperature for FTS and high RWGS. They showed high selectivity towards CO (>84%) and fair conversion (<17%) compared to rust (81%, 30%, respectively) and Fe‐impregnated alumina (100%, 8%). No loss in selectivity and conversion was observed for a longer residence time (288h). Biomass inherent metals could provide reactive gas adsorption sites that improve conversion by dispersing electrons which reduces adsorption and dissociation energy barriers. K, Mg and Ca in fern biocarbon catalysts may be related to the higher CO2 uptake compared to willow catalysts. Electron deficient sites produced by reduction of biocarbon oxygen functional groups may facilitate CO2 uptake and activation. Ni‐impregnated fern‐based biocarbon showed the highest activity, due to the synergetic effect of the inherent metals, O vacancies and strong metal‐carbon interactions
Kinetic modelling of biomass fast devolatilization using Py-MS: Model-free and model-based approaches
No full textInternational audienceFeasibility using quantitative data obtained from a micropyrolyzer coupled to an evolved gas analysis technique, mass spectrometry, for kinetics determination has been long demonstrated. This paper describes for the first time how to obtain intrinsic kinetics for fast devolatilization of biomass and its lignocellulosic fractions (e.i., holocellulose and lignins). The main challenge with online detection is assessing the time lag related to transport phenomena between the reactor and the detector. To do this, an experimental method was developed to derive the real-time biomass devolatilization profile; this provided 'corrected' datasets. Preliminary kinetic parameters were obtained from the differential isoconversional Friedman method combined with real-time sample temperature history. Not considering the effect of thermal lag and the delay in detecting pyrolysis products by the MS leads to a certain level of inaccuracies. Isoconversional activation energy (Eα) dependencies obtained in the absence of heat and mass transfer limitations for biomass and its components highly varied with conversion, confirming the multi-step nature of the fast pyrolysis process. After demonstrating the modeling limitations of a constant activation energy model (CAEM), both isoconversional functions were parametrized to propose a variable activation energy model (VAEM) and used as initial inputs for the distributed activation energy model
The physics of decision approach: a physics-based vision to manage supply chain resilience
No full textInternational audienceAs instability becomes the norm, supply chain management is becoming increasingly complex and critical. As a result, supply chain managers must adapt to complex situations. Managing instability is a key expectation for these managers. One way to help them to manage this instability is to study resilience. Resilience is related to in the literature as the ability of a system to resist, adapt and recover from disruptions. Measuring and controlling supply chain resilience has therefore become a key issue for managers, especially in a context of instability. In 2013, the World Economic Forum [2013. Global Risks 2013. Davos, Switzerland: World Economic Forum] highlighted in its study, this priority for the surveyed companies to master this concept of resilience. To address this need, this paper presents an innovative approach to disruption and resilience management based on physics principles. It considers disruptions as forces that impact supply chain performance. These forces are created as a result of changes in the internal or external attributes of the supply chain. In this approach, supply chain performance is represented and visualised as a physical trajectory modelled in the framework of its performance indicators. Thus, disturbances are considered as forces that displace and deviate the supply chain's performance trajectory in its performance framework
Robust determination of cubic elastic constants via nanoindentation and Bayesian inference
No full textInternational audienceNanoindentation is a promising tool for advancing the estimation of single crystal elastic constants in multiphase materials. In this study, a novel protocol is presented that couples high-speed nanoindentation mapping with the Vlassak and Nix’s model and Bayesian inference simulations to statistically estimate the elastic constants of cubic materials. The originality lies in considering ratios of indentation modulus as input data. For cubic elasticity, these ratios depend solely on two dimensionless parameters, which can be chosen as the Zener ratio A and the directional Poisson’s ratio ν<100>. Using ratios mitigates the influence of experimental calibration parameters. Only two constants are varied in the Bayesian simulations, and the computation time is further reduced by employing an optimized Vlassak and Nix’s model. This approach has also the great advantage to bound the search domain of ν<100> and A directly from elastic stability conditions. Furthermore, the method efficiency allows for continuous variation of the uncertainty considered in the experimental moduli, leading to stabilized Bayesian inference results. The choice of the finally retained values is thus simplified, converging to the uniqueness of the single crystal elastic constants. This method is successfully applied to high-purity Ni and Inconel 718, with the predicted elastic constants aligning well with literature data
Utilisation de l'apprentissage automatique pour prédire la qualité de service d'un centre d'appels-relais pour sourds et malentendants selon sa dotation en agents
No full textInternational audiencePara evaluar si el número de agentes asignados a un centro de llamadas puede alcanzar la calidad de servicio deseada, la investigación ha propuesto hasta ahora modelos analíticos o modelos de simulación.Gracias a la abundancia de datos registrados en los centros de llamadas, ahora es posible entrenar modelos de aprendizaje automático capaces de predecir el rendimiento de los centros de llamadas. En este artículo,utilizamos los datos de un centro de llamadas que atiende a la comunidad de personas sordas y con dificultades auditivas para entrenar modelos de aprendizaje automático que permitan predecir la calidad del servicio enfunción del número de agentes asignados. A continuación, los comparamos con el modelo clásico de colas Erlang C, ampliamente utilizado en este sector de actividad. Los primeros resultados demuestran la superioridad de losmodelos de aprendizaje automático.Pour évaluer si le nombre d'agents affectés à un centre d'appels permet d'atteindre la qualité de service visée, les travaux de recherche proposaient jusqu'à présent soit des modèles analytiques, soit des modèles de simulation. Grâce à l'abondance des données enregistrées dans les centres d'appels, il est aujourd'hui possible d'entrainer des modèles d'apprentissage automatique capables de prédire les performances du centre d'appel.Dans cet article, nous utilisons les données issues d'un centre d'appels-relais au service de la communauté des sourds et malentendants afin d'entrainer des modèles d'apprentissage automatique pour prédire la qualité deservice selon le nombre d'agents alloués. Nous les comparons ensuite au modèle de file d'attente classique Erlang C, largement utilisé dans ce secteur d'activité. Les premiers résultats mettent en évidence la supérioritédes modèles d'apprentissage automatique
Amélioration de la compressibilité des non-tissés en fibre de carbone recyclées pour mise en oeuvre par le procédé d'infusion
No full textInternational audienc
Plastic deformation delocalization at cryogenic temperatures in a nickel-based superalloy
No full textInternational audienceA nickel-based superalloy is examined during monotonic deformation from ambient to cryogenic temperatures, reaching as low as liquid helium temperature. A detailed multimodal analysis of the microstructure and plasticity is conducted to discern changes in deformation mechanisms and plastic deformation localization under cryogenic conditions. This study employs high-resolution digital image correlation and transmission electron microscopy to identify the deformation mechanisms and understand their influence on plastic deformation localization as the temperature varies. At cryogenic temperatures, unusual plastic deformation localization processes are observed, attributed to the competing activation of a range of deformation processes. Furthermore, a mechanism of slip delocalization, i.e., local plastic deformation homogenization through closely spaced slip, is noted at these extreme temperatures. Ultimately, the impact of the microstructure is identified across the temperature range, from room to cryogenic temperatures
Defect Characterization on Complex Shape Aeronautical Parts via 3D Point Cloud Processing
No full textCopyright IEEEInternational audienceThis paper presents an approach to characterizingdefects - a process that consists of accurate measurement of theirgeometric properties such as depth and surface area, assumingthat the defect detection process has been successfully performedpreviously. Our methodology for addressing this problem involvesthree key steps. The first is to reconstruct an ideal or defectfreesurface using scattered points obtained from a point cloudscan of the inspected part. Subsequently, the distance from eachcloud point to this surface is computed, and points at a distancehigher than a specified threshold are identified as defect points.The maximum of these distances corresponds to the depth of thedefect. Finally, the minimal 3D bounding box encapsulating thedefect points is determined, where the two largest dimensions ofthis box represent the length and width of the identified defect