HAL - Université de Franche-Comté
Not a member yet
51444 research outputs found
Sort by
Becoming Emperor: gestures and political ideology in imperial investiture ceremonies
No full textInternational audienc
Robust Thermal Management in PEM Electrolysis: A Two-Loop Control Strategy
No full textInternational audience<div style=""><font face="arial, helvetica"><span style="font-size: 13px;">Proton Exchange Membrane (PEM) electrolysis is </span></font><span style="font-size: 13px; font-family: arial, helvetica;">a highly effective method for generating green hydrogen from </span><span style="font-size: 13px; font-family: arial, helvetica;">renewable energy sources. However, effective thermal control </span><span style="font-size: 13px; font-family: arial, helvetica;">is essential for maintaining high efficiency and prolonging </span><span style="font-size: 13px; font-family: arial, helvetica;">the stack lifespan, especially under variable power conditions. </span><span style="font-size: 13px; font-family: arial, helvetica;">This paper develops a four-state thermal model of the entire </span><span style="font-size: 13px; font-family: arial, helvetica;">PEM electrolysis system, including key Balance of Plant (BoP)</span></div><div style=""><font face="arial, helvetica"><span style="font-size: 13px;">components such as heat exchangers and separators, to capture </span></font><span style="font-size: 13px; font-family: arial, helvetica;">the system’s primary temperature dynamics. Building on this </span><span style="font-size: 13px; font-family: arial, helvetica;">model, we propose a two-loop control architecture: a slower PID </span><span style="font-size: 13px; font-family: arial, helvetica;">loop regulates the inlet temperature, while a faster sliding-mode </span><span style="font-size: 13px; font-family: arial, helvetica;">loop maintains a target temperature difference between the inlet </span><span style="font-size: 13px; font-family: arial, helvetica;">and outlet temperatures. We use Particle Swarm Optimization </span><span style="font-size: 13px; font-family: arial, helvetica;">(PSO) to determine the controller parameters, enabling robust </span><span style="font-size: 13px; font-family: arial, helvetica;">operation under changing load conditions. Simulation results </span><span style="font-size: 13px; font-family: arial, helvetica;">demonstrate that the proposed approach not only stabilizes </span><span style="font-size: 13px; font-family: arial, helvetica;">the overall temperature but also mitigates abrupt disturbances </span><span style="font-size: 13px; font-family: arial, helvetica;">from renewable energy sources, thereby improving both the </span><span style="font-size: 13px; font-family: arial, helvetica;">efficiency and durability of the PEM electrolyzer.</span></div><div style=""><span style="font-size: 13px; font-family: arial, helvetica;"><div>PEM electrolysis, thermal management, nonlinear control, sliding mode, particle swarm optimization</div></span></div&g
Développer les capacités d’anticipation des communautés locales en situation d’exclusion. Perspectives pour la résolution des conflits liés à l’exploitation des ressources naturelles et minières en Afrique
No full textInternational audienc
Intracellular Monitoring of Luminescent Ag<sub>2</sub>S Nanoparticles in Fibroblasts
No full textInternational audienceInteraction between nanoparticles (NPs) and cells is a key concern in nanotechnology-based biomedical applications. This study investigates cytocompatible silver sulfide (Ag2S) quantum dots (QDs), which emit strongly in the second near-infrared (NIR II) region, for their potential in image-guided therapies. Since cellular microenvironments contain acidic pH, hydrogen peroxide (H2O2), and reactive oxygen species (ROS), which can affect luminescence, Ag2S NPs (AS NPs) coated with three ligands: DTDTPA (dithiolated diethylene triamine pentaacetic acid), 11MUA (11 mercaptoundecanoic acid), and PEG (polyethylene glycol) were studied in fibroblast cells over 9 days. Long term photoluminescence (PL) tracking showed distinct behavior. AS DTDTPA and AS 11MUA exhibited increasing uptake but decreasing PL, while AS PEG showed both enhanced PL and uptake over time. No significant changes in NP size or structure explained the PL differences. Exposure to H2O2 revealed that AS DTDTPA and AS 11MUA lost PL due to oxidation or shell detachment, reducing surface passivation. In contrast, AS PEG showed intensified PL, possibly from removal of metallic silver and increased PL lifetime. These results highlight time dependent PL changes in cellular and oxidative environments. DTDTPA and PEG coatings may act as ROS scavenger and promoter, respectively, while enabling NIR II imaging and offering insights into NP design for stable optical tracking and therapeutic use.</p
Sensitivity Analysis for Sizing Standalone Green Datacenters
No full textInternational audienceSince the widespread use of social media and IT services (cloud, video streaming, etc.), the ecological impact of IT infrastructures such as datacenters has become a significant concern. This is due to the large amount of electricity these infrastructures consume, and the reliance on fossil fuels to meet this demand. Datacenters powered by renewable energy have been proposed as a solution. However, due to the intermittency of renewable energy sources, these platforms remain connected to the conventional power grid. The DATAZERO2 project [1] has developed an approach with a fully autonomous electrical infrastructure including wind turbines, solar panels, short- and long-term energy storage devices (batteries and hydrogen systems respectively). The work presented here focuses on the sizing stage of such an infrastructure. Although this stage encompasses both the data center and the electrical infrastructure needed to power it, the focushere is on IT sizing. The challenge is to determine the optimal size of the infrastructure: large enough to meet customer demand without service interruption, but small enough to avoid overprovisioning, which would lead to unnecessary costs, pollution from excess equipment and wasted energy on idle servers. This raises the question of balancing system robustness against frugality. The sizing process must account for the potential variability in client usage (such as the increase in demand observed during lockdown due to remote working and streaming) while simultaneously minimizing the infrastructure to avoid purchasing unnecessary hardware. The IT (datacenter) sizing takes as inputs a workload (requests submitted by clients) and the characteristics of the servers used. The workload is structured as a time series over one year, where requests (jobs) are presented hourly across 8760 hours. Each job Ji is defined by two values: its submission time si (the date the datacenter receives the request), and the job size wi (the amount of work servers must do to complete the task). The sizing process then returns the minimum number of servers mOPT required to satisfy this demand. IT sizing can be monitored by various metrics as the cost of the infrastructure (in US dollar), its carbon footprint (in kgeqCO2) resulting from the construction of the datacenter. The objective is to meet the expected QoS while minimizing both the ecological and economic impact of the infrastructure. This is where the first question of sensitivity analysis comes in: how does the workload, and therefore the number of servers, contribute to variations in quality of service? Besides, to ensure robustness, the datacenter must be able to meet the expected QoS even when the workload fluctuates over the years. To deal with, we propose the following framework: considering m servers, the sizing ischallenged across a set setW of workloads (tests) and for each test, the resulting QoS obtained. A guarantee is then established, for m servers, the datacenter commits to meeting an expected QoS of Q%, with a success rate of R%. This success rate represents the ratio of tests that meet the expected QoS, relative to the total number of tests. The robustness of the datacenter is evaluated on a set of 100 workloads, providing the associated guarantee. These workloads are based on a generator using Google traces [2], with distributions determined the number of jobs submitted per hour (which follows a Pareto distribution with an index of α = 5) and the size of the jobs in Million Instructions [MI] (which follows a log-normal distribution with parameters σ = 1.42 and μ = 6.25). The second question in the sensitivity analysis is how the number of servers is governed by the variation in the number of tasks and their size. Figure 1a shows the distribution of the number of servers obtained by considering |setW| = 100 workloads tested. Further analysis suggests that the number of servers mOPT (the minimum number of servers required for a QoS of 100%) follows a log-normal distribution. This result is confirmed using the normality test of D’Agostino with a p-value around 0.2. Figure 1b shows the distribution of the number of servers for a given expectedQoS. Notably, a 0.2% reduction in the number of servers results in a significant decrease in the value of m, which subsequently converges to a constant value. However, further sensitivity analysis indices will be explored to provide more accurate insights about the QoS on the number of servers m. Based on these two observations, the objective is to generalize the guarantee presented above for any given set setW, such that the guarantee no longer depends on the chosen set. This study is conducted in two phases: (1) Evaluating the impact of the workload on the sizing when the expected quality of service is at its maximum (100%). The goal is to precisely determine whichfactor, whether the number of submitted jobs or their size, has the greatest impact on the resulting sizing; (2) Evaluating the impact of the fixed number of servers on the final QoS. The aim is to clarify the impact of the workload on the resulting QoS for a given fixed sizing. To do so, a global sensitivity analysis is performed. The focus is on the study of the impact of the input parameter variation on the variability of the output in the whole domain of interest [4].Among the existing GSA approaches we focus on: Variance based method where the idea is to quantify how the variability in model outputs can be attributed to the variations in input parameters, this can be performed using sensitivity indices such as Sobel indices [5]. Secondapproach is the screening methods where the variations in the output is measured for different variation of the input parameter. The third approach is based on Shapely coefficients and is more suitable for correlated model inputs [3]. In this work, it is attested that the size of job has greater impact on the servers numbers than the number of jobs. Further sensitivity analysis indices will be explored to accurately measure the impact of each variabl
Vers une meilleure compréhension de l’essai de déchaussement de microgouttes sur fibre élémentaire
No full textInternational audienceLes composites à matrice organique sont largement utilisés dans différents secteurs industriels du fait de leur bonne performance mécanique. En outre, leur recyclage en fin de vie reste actuellement limité en raison de plusieurs verrous scientifiques et technologiques. Dans ce contexte, le projet RECYCOMP (PEPR « Recyclabilité, recyclage et ré-utilisation des matières ») vise à développer des stratégies (solvolyse, fonctionnalisation des fibres) permettant de faciliter la séparation du renfort fibreux (carbone, verre ou lin) de la matrice. Afin d’évaluer l’impact des différentes approches sur la performance de nouveaux matériaux composites intégrant ces renforts recyclés, il est nécessaire de s’intéresser au comportement à l’interface fibre/matrice. En particulier, l’essai de déchaussement de microgouttes sur fibre élémentaire permet de déterminer la résistance mécanique à l’interface. Cette étude présente les premiers résultats obtenus sur des fibres de verre avec une matrice époxy. Ceux-ci sont comparables aux valeurs rapportées dans la littérature. En outre, l’observation au<br>tomographe à rayonsXd’un échantillon après déchaussement souligne la nécessité de développer un dispositif expérimental pour réaliser cet essai in situ afin d’acquérir une meilleure compréhension des mécanismes de déchaussement et, ainsi, de mieux appréhender le comportement à l’interface dans le composite
Cellular Automaton Simulation Model for Predicting the Microstructure Evolution of an Additively Manufactured X30Mn21 Austenitic Advanced High-Strength Steel
No full textInternational audienceAdditive manufacturing techniques, such as laser-based powder bed fusion of metals (PBF-LB/M), have now gained high industrial and academic interest. Despite its design flexibility and the ability to fabricate intricate components, LPBF has not yet reached its full potential, partly due to the challenges associated with microstructure control. The precise manipulation of the microstructure in LPBF is a formidable yet highly rewarding endeavor, offering the capability to engineer components at a local level. This work introduces an innovative parallelized Cellular Automaton (CA) framework for modeling the evolution of the microstructure during the LPBF process. LPBF involves remelting and subsequent nucleation followed by crystal growth during solidification, which complicates and burdens microstructure simulations. In this research, a novel approach to nucleation seeding and crystal growth is implemented, focusing exclusively on the final stages of melting and solidification, enhancing the computational efficiency by 30%. This approach streamlines the simulation process, making it more efficient and effective. The developed model was employed to simulate the microstructure of an austenitic advanced high-strength steel (AHSS). The model was validated by comparing the simulation results qualitatively and quantitatively with the experimental data obtained under the same process parameters. The predicted microstructure closely aligned with the experimental findings. Simulations were also conducted at varying resolutions of CA cells, enabling a comprehensive study of their impact on microstructure evolution. Furthermore, the computational efficiency was critically evaluated
Recension de La détresse existentielle (PUFC, 2025)
No full textDépouillement analytique complet de l’ouvrageRecension de : Benjamin Guérin, Sarah Carvallo, Régis Aubry (dir.), La détresse existentielle, Presses universitaires de Franche-Comté, 2025. 189 p
Recension de Liaisons pastorales (ediSens, 2023)
No full textDépouillement analytique complet de l’ouvrageRecension de : Aliènor Bertrand, Anne Blondeau Da Silva et Dominique Taurisson-Mouret (dir.), Liaisons pastorales : Coévolutions, ruptures, résistances, Paris, ediSens, 2023. 311 p