Portail HAL edf
Not a member yet
11029 research outputs found
Sort by
HPC challenges and opportunities of industrial-scale reactive fluidized bed simulation using meshes of several billion cells on the route of Exascale
No full textInternational audienceInside fluidized bed reactors, gas-solid flows are very complex: multi-scale, coupled, reactive, turbulent and unsteady. Accounting for them in an Euler-nfluid framework induces significantly expensive numerical simulations at academic scales and even more at industrial scales. 3D numerical simulations of gas-particle fluidized beds at industrial scales are limited by the High Performances Computing (HPC) capabilities of Computational Fluid Dynamics (CFD) software and by available computational power. In recent years, preExascale supercomputers came into operation with better energy efficiency and continuously increasing computational resources. The present article is a direct continuation of previous work, Neau et al. (2020) which demonstrated the feasibility of a massively parallel simulation of an industrial-scale polydispersed fluidized-bed reactor with a mesh of 1 billion cells. Since then, we tried to push simulations of these systems to their limits by performing large-scale computations on even more recent and powerful supercomputers, once again using up to the entirety of these supercomputers (up to 286,000 cores). We used the same fluidized bed reactor but with more refined unstructured meshes: 8 and 64 billion cells. This article focuses on efficiency and performances of neptune_cfd code (based on Euler-nfluid approach) measured on several supercomputers with meshes of 1, 8 and 64 billion cells. It presents sensitivity studies conducted to improve HPC at these very large scales. On the basis of these highly-refined simulations of industrial scale systems using pre-Exascale supercomputers with neptune_cfd, we defined the upper limits of simulations we can manage efficiently in terms of mesh size, count of MPI processes and of simulation time. One billion cells computations are the most refined computation for production. Eight billion cells computations perform well up to 60,000 cores from a HPC point of view with an efficiency >85% but are still very expensive. The size of restart and mesh files is very large, post-processing is complicated and data management becomes near-impossible. 64 billion cells computations go beyond all limits: solver, supercomputer, MPI, file size, post-processing, data management. For these reasons, we barely managed to execute more than a few iterations. Over the last 30 years, neptune_cfd HPC capabilities improved exponentially by tracking hardware evolution and by implementing state-of-the-art techniques for parallel and distributed computing. However, our last findings show that currently implemented MPI/Multigrid approaches are not sufficient to fully benefit from pre-Exascale system. This work allows us to identify current bottlenecks in neptune_cfd and to formulate guidelines for an upcoming Exascale-ready version of this code that will hopefully be able to manage even the most complex industrial-scale gas-particle systems
Integrating Small Modular Reactors into hybrid energy systems: the tandem Modelica library
No full textInternational audienceThe rapid evolution of Small Modular Reactor (SMR) technology has triggered renewed interest in exploring innovative applications within the framework of nuclear hybrid energy systems. SMR, given their greater flexibility in terms of siting and power rates, are one of the most suitable candidates to be integrated in energy systems aimed at providing multiple energetic assets. In this context, feasibility studies as well as techno-economics analysis should be supported by proper modelling tools able to facilitate the integration of SMRs with other energy system components, such as renewable plants, energy storage systems, hydrogen production, etc. This paper presents the TANDEM Modelica library developed within the TANDEM Project – a Horizon Europe project aimed at developing methodologies and tools to facilitate the safe and efficient integration of SMRs into smart low-carbon hybrid energy systems. The TANDEM library is based on several staple libraries (e.g., ThermoPower, ThermoSysPro, Modelica Standard Library) from which the main components of a hybrid energy system are derived, e.g., NSSS, BoP and power conversion systems, electrical and thermal energy storage, electrical grid, conventional power plants, hydrogen production with low and high temperature electrolyzers, district heating…. The library is meant to be a versatile platform, offering a unified framework for the dynamic simulation and analysis of complex interactions within nuclear hybrid energy systems. Key features of the library include modularity, extensibility, and compatibility with existing library and simulation tools, including safety codes. Possible applications enabled by the library include the analysis of different operational strategies and the optimization of the hybrid energy system configuration and components’ design in terms of efficiency, reliability and/or CO2 emissions
Significant impact of urban tree biogenic emissions on air quality estimated by a bottom-up inventory and chemistry transport modeling
No full textInternational audienceBiogenic volatile organic compounds (BVOCs) are emitted by vegetation and react with other compounds to form ozone and secondary organic matter (OM). In regional air quality models, biogenic emissions are often calculated using a plant functional type approach, which depends on the land use category. However, over cities, the land use is urban, so trees and their emissions are not represented. Here, we develop a bottom-up inventory of urban tree biogenic emissions in which the location of trees and their characteristics are derived from the tree database of the Paris city combined with allometric equations. Biogenic emissions are then computed for each tree based on their leaf dry biomass, tree-species-dependent emission factors, and activity factors representing the effects of light and temperature. Emissions are integrated in WRF-CHIMERE air quality simulations performed over June–July 2022. Over Paris city, the urban tree emissions have a significant impact on OM, inducing an average increase in the OM of about 5 %, reaching 14 % locally during the heatwaves. Ozone concentrations increase by 1.0 % on average and by 2.4 % during heatwaves, with a local increase of up to 6 %. The concentration increase remains spatially localized over Paris, extending to the Paris suburbs in the case of ozone during heatwaves. The inclusion of urban tree emissions improves the estimation of OM concentrations compared to in situ measurements, but they are still underestimated as trees are still missing from the inventory. OM concentrations are sensitive to terpene emissions, highlighting the importance of favoring urban tree species with low-terpene emissions
Assessment of two non-invasive techniques for measuring turbulent benthic fluxes in a shallow lake
No full textInternational audienceBenthic fluxes refer to the exchange rates of nutrients and other compounds between the water column and the sediment bed in aquatic ecosystems. Their quantification contributes to our understanding of aquatic ecosystem functioning. Near-bed hydrodynamics plays an important role at the sediment-water interface, especially in shallow lakes, but it is poorly considered by traditional measuring techniques of flux quantification, such as sediment incubations. Thus, alternative sampling techniques are needed to characterize key benthic fluxes under in-situ hydrodynamic conditions. This study aimed to evaluate the performance of two promising methods: relaxed eddy accumulation (REA) and mass transfer coefficient (MTC). We applied them in a hyper-eutrophic shallow lake to measure the fluxes of ammonium, phosphate, iron, and manganese ions. For the first time, REA revealed hourly nutrient flux variations, indicating a strong lake biogeochemical dynamics at short time-scales. Daily average fluxes are of similar orders of magnitude for REA and MTC for ammonium (24 and 42 mmol m2 d-1), manganese (1.0 and 0.8), and iron (0.8 and 0.7) ions. They are one order of magnitude higher than fluxes estimated from sediment incubations, due to the difficulty in reproducing in-situ oxygen and hydrodynamic conditions in the laboratory. Although the accuracy of both techniques needs to be improved, the results revealed their potential: REA follows the short-term biogeochemical dynamics of sediments, while MTC could be widely used for lake monitoring because of its simpler implementation
Understanding copper sulfide formation from layered template and their use as power electrode materials in aqueous electrolyte
No full textInternational audienceCopper sulfide has received increasing attention as an electrode material in past decades. In this study, we report the synthesis of copper sulfide with layered copper hydroxide salt (LHS) (Cu2(OH)3NO3) precursors using different protocols. X-ray diffraction suggests the presence of numerous non-stoichiometric phases (Cu1-xS) and not a pure covellite phase and SEM images show particles with flower-like shape but different in size. The solidstate kinetic parameters of the reaction refined by the JMAK method indicate a pseudomorphic transformation controlled by 1D diffusion, different in term of precursors phase, reagents and protocol applied. The microwave method needs less energy to achieve the transformation than the amine digestion (AD) method and the morphology of particles is also different. Tested in sodium acetate electrolyte, CuS provides a maximum capacity of 67 mAh/g for AD, which is much higher than for the commercial grade CuS. This is explained by the difference in nanostructuration of the flower-like shape particles obtained from the layered template. Finally, CuS is used as both a positive and negative electrode material in a complete aqueous battery system but its redox process, which is strongly diffusion limited especially in the lower potential domain, prevents the whole system from operating at high power
: L'agroforesterie premier puits de CO2 atmosphérique
No full textGlobal agroforestry would absorb around 21 billion tonnes (Gt or Pg) of CO2 in 2022. This is the main global contribution to the reduction of atmospheric CO2, which is assessed here on the basis of human use of agroforestry products from photosynthesis. It represents two to three times the oceanic sink and offsets 60% of anthropogenic emissions through the combustion of fossil hydrocarbons. The share of the Brazilian forest, which includes Amazonia, is approximately 0.7% of this agroforestry CO2 sink. This value is much lower than those which attributed to Amazonia a preponderant role in the production of oxygen and global climate regulation.L'agroforesterie mondiale absorbait environ 21 milliards de tonnes (Gt ou Pg) de CO 2 en 2022. C'est la principale contribution planétaire à la réduction du CO 2 atmosphérique, laquelle est évaluée ici sur la base de l'utilisation humaine des produits de la photosynthèse. Elle représente deux à trois fois le puits océanique et compense 60% des émissions anthropiques par combustion d'hydrocarbures fossiles. La part de la forêt brésilienne, qui inclut l'Amazonie, est d'environ 0,7% de ce puits de CO 2 agroforestier. Cette valeur est très inférieure à celles qui ont attribué à l'Amazonie un rôle prépondérant dans la production d'oxygène et la régulation climatique mondiales
Evaluation de petits LLM quantifiés sur une tâche de classification de textes en français
No full textInternational audienceCet article présente une évaluation de plusieurs "Large Language Models", quantifiés (codés sur 4bits) de taille 7 milliards réalisée sur une tâche de classification de données textuelles (des avis AlloCiné). Les textes sont d'abord transformés en vecteurs, à l'aide d'un LLM, puis servent à entraîner un classifieur. Le meilleur résultat est obtenu avec le modèle NeuralHermès-2.5-Mistral
Response of biogenic secondary organic aerosol formation to anthropogenic NOx emission mitigation
No full textInternational audienceThis study investigates the effects of anthropogenic nitrogen oxide (NOx) mitigation reduction on secondary organic aerosol (SOA) formation from monoterpene and sesquiterpene precursors across Europe, using the three-dimensional (3-D) Chemical Transport Model (CTM) CHIMERE.Two SOA mechanisms of varying complexity are employed: the GENOA-generated Biogenic Mechanism (GBM) and the Hydrophobic/Hydrophilic Organic mechanism (H2O). GBM is a condensed SOA mechanism generated by automatic reduction from near-explicit chemical mechanisms (i.e., the Master Chemical Mechanism - MCM and the peroxy radical autoxidation mechanism - PRAM) using the GENerator of Reduced Organic Aerosol Mechanisms version 2.0 (GENOA v2.0). Conversely, the H2O mechanism is developed primarily based on experimental data, with simplified chemical pathways and SOA formation yields reflecting those from chamber experiments.In the 3-D simulations conducted for the summer of 2018 over Europe, the implementation of GBM significantly improved the model's performance in comparison to simulations using the H2O mechanism, yielding results more consistent with measured aerosol concentrations extracted from the EBAS database.In response to NOx emission mitigation, simulated SOA concentrations increase with GBM but decrease when using the H2O mechanism, unless a highly oxygenated molecules (HOMs) formation scheme is incorporated. The SOA composition becomes more oxidized and concentrations elevate after NOx reduction, particularly in simulations using GBM. These higher concentrations are likely due to enhanced reaction rates of organic peroxy radicals (RO2) with HO2, resulting in more oxidized products from monoterpene degradation that favors HOM formation. The results suggest that detailed SOA mechanisms including autoxidation are necessary for accurate predictions of SOA concentrations in 3-D modeling
Forecasting Electric Vehicle Charging Station Occupancy: Smarter Mobility Data Challenge
No full textInternational audienceThe transport sector is a major contributor to greenhouse gas emissions in Europe. Shifting to electric vehicles (EVs) powered by a low-carbon energy mix would reduce carbon emissions. However, to support the development of electric mobility, a better understanding of EV charging behaviours and more accurate forecasting models are needed. To fill that gap, the Smarter Mobility Data Challenge has focused on the development of forecasting models to predict EV charging station occupancy. This challenge involved analysing a dataset of 91 charging stations across four geographical areas over seven months in 2020-2021. The forecasts were evaluated at three levels of aggregation (individual stations, areas and global) to capture the inherent hierarchical structure of the data. The results highlight the potential of hierarchical forecasting approaches to accurately predict EV charging station occupancy, providing valuable insights for energy providers and EV users alike. This open dataset addresses many real-world challenges associated with time series, such as missing values, non-stationarity and spatio-temporal correlations. Access to the dataset, code and benchmarks are available at https://gitlab.com/smarter-mobility-data-challenge/tutorials to foster future research
HPC atomic simulations of defect formation and kinetic Monte Carlo simulation of material ageing
No full textInternational audienceKinetic Monte Carlo (KMC) simulations are developed to simulate microstructure evolution under irradiation of structural materials of nuclear power plants. Methods based on rigid lattice atomic KMC and object KMC, despite some approximations, present the advantage to reach significant time, hence irradiation doses. The physical inputs such as the primary damage and defect cluster properties are the results of intensive atomic simulation on high performance computers. Here, atomic KMC is applied to model microstructure evolution of reactor pressure vessel steels and object KMC is applied to zirconium cladding materials