Search

International audience; Solid oxide electrolysis cells (SOEC) are a promising energy conversion technology for the production of green hydrogen via steam electrolysis. However, performance loss remains a bottleneck for large-scale commercialization. Here, we investigate the different types of degradation occurring in a state-of-the-art cell composed of La0.6Sr0.4Co0.2Fe0.8O3-Ce0.8Gd0.2O2 (LSCF-GDC) composite for the oxygen electrode, GDC for the barrier layer, Y0.16Zr0.84O1.92 (8YSZ) for the electrolyte and Ni-YSZ for the fuel electrode. Electrochemical impedance spectra measured before and after operation at 750 °C for 2000 hours at -1 A cm-2 in SOEC mode revealed no significant evolution for the oxygen electrode contribution. To investigate potential degradation of this electrode, we performed multi-modal chemical imaging based on synchrotron X-ray diffraction (µ-XRD) with a resolution at the micrometer-scale. In addition, we monitored the evolution of the chemical composition and chemical environment before and after ageing using laboratory X-ray photoelectron spectroscopy (XPS). All these analyses revealed a slight evolution of the crystal lattice parameter in the oxygen electrode and the inter-diffusion layer. It was also found that Sr segregation accelerates with ageing. However, more severe ageing conditions and/or longer ageing durations are required in order to observe an effect on the performance.

In the next few years, numerous satellites with high-resolution instruments dedicated to the imaging of atmospheric gaseous compounds will be launched, to finely monitor emissions of greenhouse gases and pollutants. Processing the resulting images of plumes from cities and industrial plants to infer the emissions of these sources can be challenging. In particular traditional atmospheric inversion techniques, relying on objective comparisons to simulations with atmospheric chemistry transport models, may poorly fit the observed plume due to modelling errors rather than due to uncertainties in the emissions. The present article discusses how these images can be adequately compared to simulated concentrations to limit the weight of modelling errors due to the meteorology used to analyse the images. For such comparisons, the usual pixel-wise ℒ2 norm may not be suitable, since it does not linearly penalise a displacement between two identical plumes. By definition, such a metric considers a displacement as an accumulation of significant local amplitude discrepancies. This is the so-called double penalty issue. To avoid this issue, we propose three solutions: (i) compensate for position error, due to a displacement, before the local comparison; (ii) use non-local metrics of density distribution comparison; and (iii) use a combination of the first two solutions. All the metrics are evaluated using first a catalogue of analytical plumes and then more realistic plumes simulated with a mesoscale Eulerian atmospheric transport model, with an emphasis on the sensitivity of the metrics to position error and the concentration values within the plumes. As expected, the metrics with the upstream correction are found to be less sensitive to position error in both analytical and realistic conditions. Furthermore, in realistic cases, we evaluate the weight of changes in the norm and the direction of the four-dimensional wind fields in our metric values. This comparison highlights the link between differences in the synoptic-scale winds direction and position error. Hence the contribution of the latter to our new metrics is reduced, thus limiting misinterpretation. Furthermore, the new metrics also avoid the double penalty issue.

International audience; The purpose of this paper is to propose a time-step-robust cell-to-cell integration of particle trajectories in 3-D unstructured meshes in particle/mesh Lagrangian stochastic methods. The main idea is to dynamically update the mean fields used in the time integration by splitting, for each particle, the time step into sub-steps such that each of these sub-steps corresponds to particle cell residence times. This reduces the spatial discretization error. Given the stochastic nature of the models, a key aspect is to derive estimations of the residence times that do not anticipate the future of the Wiener process. To that effect, the new algorithm relies on a virtual particle, attached to each stochastic one, whose mean conditional behavior provides free-of-statistical-bias predictions of residence times. After consistency checks, this new algorithm is validated on two representative test cases: particle dispersion in a statistically uniform flow and particle dynamics in a non-uniform flow.; L’objectif de ce papier est de proposer un algorithme robuste en pas de temps permettant l’intégration face à face de la trajectoire de particules dans des maillage 3-D non-structurés pour des méthodes lagrangiennes stochastiques. L’idée principale est de mettre à jour de façon dynamique les champs moyens utilisés dans l’intégration temporelle. Pour ce faire, pour chaque particule, on subdivise le pas de temps en sous pas de temps, de façon à ce que chacun de ceux-ci correspondent au temps de résidence de la particule dans une cellule. Ce faisant on réduit l’erreur de discrétisation spatiale. Étant donnée la nature stochastique du modèle, un aspect prépondérant est de pouvoir dériver des estimations du temps de résidence des particules dans les cellules qui n’anticipent pas le futur des processus de Wiener. Dans ce but, le nouvel algorithme est basé sur une particule virtuelle attachée à chaque particule stochastique, dont le comportement conditionné moyen permet d’obtenir des temps de résidence qui ne sont entachés d’aucun biais statistique. Après une vérification de consistance, ce nouvel algorithme est validé sur deux cas de tests représentatifs : une dispersion de polluants dans un écoulement statistiquement uniforme et un écoulement non-uniforme dans lequel on traque la dynamique des particules.

This paper deals with the numerical modeling of immiscible three-phase flows. The main focus here is on the numerical treatment of the source terms of the model. A new scheme based on a more coupled approach than the preexisting fractional step strategy is presented. Properties of this scheme are given. Numerical applications highlight the benefits of this scheme in terms of both accuracy and stability.

International audience

Maxwell models for viscoelastic flows are famous for their potential to unify elastic motions of solids with viscous motions of liquids in the continuum mechanics perspective. But the usual Maxwell models allow one to define well motions mostly for one-dimensional flows only. To define unequivocal multi-dimensional viscoelastic flows (as solutions to well-posed initial-value problems) we advocated in [ESAIM:M2AN 55 (2021) 807-831] an upper-convected Maxwell model for compressible flows with a symmetrichyperbolic formulation. Here, that model is derived again, with new details.

We define and analyze the notion of variational Wardrop equilibrium for nonatomic aggregative games with an infinity of players types. These equilibria are characterized through an infinite-dimensional variational inequality. We show, under monotonicity conditions, a convergence theorem enables to approximate such an equilibrium with arbitrary precision. To this end, we introduce a sequence of nonatomic games with a finite number of players types, which approximates the initial game. We show the existence of a symmetric Wardrop equilibrium in each of these games. We prove that those symmetric equilibria converge to an equilibrium of the infinite game, and that they can be computed as solutions of finite-dimensional variational inequalities. The model is illustrated through an example from smart grids: the description of a large population of electricity consumers by a parametric distribution gives a nonatomic game with an infinity of different players types, with actions subject to coupling constraints., arXiv admin note: substantial text overlap with arXiv:1806.06230

To simulate the interaction of ocean waves with marine structures, coupling approaches between a potential flow model and a viscous model are investigated. The first model is a fully nonlinear potential flow (FNPF) model based on the Harmonic Polynomial Cell (HPC) method, which is highly accurate and best suited for representing long distance wave propagation. The second model is a CFD code, solving the Reynolds-Averaged Navier-Stokes (RANS) equations within the \openfoam toolkit, more suited to represent viscous and turbulent effects at local scale in the body vicinity. Two one-way coupling strategies are developed and compared in two dimensions, considering fully submerged and fixed structures. A domain decomposition (DD) strategy is first considered, introducing a refined mesh in the body vicinity on which the RANS equations are solved. Boundary conditions and interpolation operators from the FNPF results are developed in order to enforce values at its outer boundary. The second coupling strategy considers a decomposition of variables (functional decomposition, FD) on the local grid. As the FNPF simulation provides fields of variables satisfying the irrotational Euler equations, complementary velocity and pressure components are introduced as the difference between the total flow variables and the potential ones. Those complementary variables are solutions of modified RANS equations. Comparisons are presented for nonlinear waves interacting with a horizontal cylinder of rectangular cross-section. The loads exerted on the body computed from the four simulation methods (standalone FNPF, standalone CFD, DD and FD coupling schemes) are compared with experimental data. It is shown that both coupling approaches produce an accurate representation of the loads and associated hydrodynamic coefficients over a large range of incident wave steepness and Keulegan-Carpenter number., Comment: 39 Pages, 12 figures, 2 tables

International audience; Introduction: Characterizing the thermal habitat of fish is key to understanding their ecological requirements in order to make appropriate management plans for the preservation of fish populations in the context of climate change. Little is known about the thermal habitat of the early life stages of fish (larvae and juveniles), as most methods have been designed for large, easily handled individuals. In particular, the study of otoliths holds great promise for unveiling the early thermal history of fish, although it is challenging to implement due to the very small size of the biological material. Methods: The stable oxygen isotope content (δ 18 O) of biogenic carbonates (such as otoliths) can be used to reconstruct the life temperature of fish individuals. However, relationships between δ 18 O of otoliths and ambient temperature are scarce and mainly developed for commercial species. In this study, we assessed the δ 18 O of juvenile European chub (Squalius cephalus, Linnaeus 1758) living in a section of the lower Rhône River in France. Results: The δ 18 O of otoliths showed significant relationship with ambient temperature [δ 18 O oto(VPDB)-δ 18 O w(VPDB) = 33.391-0.2641 * T (°C)], and the relationship between temperature and the fractionation factor (α) was 1,000 ln α = −43.472 + 21.205 [1,000/T(K)]. Discussion: As expected, back-calculated temperatures from these equations were in much better agreement with the measured in situ temperatures than previously published equations. This equation is the first to be developed for leuciscid fish in Europe and the second in the world.

International audience; A stable scheme is proposed in this paper in order to obtain approximate solutions of second-moment turbulent models for incompressible flows with or without thermal transport equation. The analysis of the convective terms, which includes the solution of the associated Riemann problem, enables to propose a standard projection scheme, and to get rid of spurious oscillations.

International audience; A Layered Double Hydroxide (LDH) compound LDH ([Mg 2 Al(OH) 6 ] + x 2 H 2 O) intercalated with a redox active organic anion, Anthraquinone-2-sulfonate (AQS), has been envisioned as an electrode material for high power aqueous based battery. The purpose is to use this interlayer redox active molecule for the enhancement of the specific capacity at the LDH composite electrode, which should allow fast charge transfer at the negative electrode for high power storage applications. This is achieved by the reduction of AQS in charge and oxidation in discharge within a redox inactive LDH matrix. The first charge of this new material [Mg 2 Al(OH) 6 ] + [AQSO 3 ] − x 2 H 2 O leads to a capacity of 100 mAh g −1 at − 0.78 V vs Ag/AgCl (based on the weight of the active material) when operated in aqueous 1 M sodium acetate electrolyte. However, low cycling stability was observed, since a drastic loss in specific capacity occurs after the first charge. This study focuses at elucidating the mechanism behind this phenomenon via in situ UV/vis experiments. Subsequently, the dissolution of charged AQS anions into the electrolyte during the first charge of the anode has been identified and quantified. Such understanding of fading mechanism might lead to the design of improved LDH-based electrodes, which utilize redox active anions working in the positive potential range with enhanced cycling ability.

The spatiotemporal resolution of Partial Differential Equations (PDEs) plays important roles in the mathematical description of the world's physical phenomena. In general, scientists and engineers solve PDEs numerically by the use of computationally demanding solvers. Recently, deep learning algorithms have emerged as a viable alternative for obtaining fast solutions for PDEs. Models are usually trained on synthetic data generated by solvers, stored on disk and read back for training. This paper advocates that relying on a traditional static dataset to train these models does not allow the full benefit of the solver to be used as a data generator. It proposes an open source online training framework for deep surrogate models. The framework implements several levels of parallelism focused on simultaneously generating numerical simulations and training deep neural networks. This approach suppresses the I/O and storage bottleneck associated with disk-loaded datasets, and opens the way to training on significantly larger datasets. Experiments compare the offline and online training of four surrogate models, including state-of-the-art architectures. Results indicate that exposing deep surrogate models to more dataset diversity, up to hundreds of GB, can increase model generalization capabilities. Fully connected neural networks, Fourier Neural Operator (FNO), and Message Passing PDE Solver prediction accuracy is improved by 68%, 16% and 7%, respectively.

International audience; The direct parametrisation method for invariant manifolds (DPIM) is applied to rotating structures. Reduced- order models of arbitrary order expansion can be derived for non-autonomous systems of nonlinear differential equations stemming from finite element models of continuous structures. The method is applied to a rotating simplified fan blade with comparisons to full order model simulations.

International audience; Recent work has shown that evaluating functional trait distinctiveness, the average trait distance of a species to other species in a community offers promising insights into biodiversity dynamics and ecosystem functioning. However, the ecological mechanisms underlying the emergence and persistence of functionally distinct species are poorly understood. Here, we address the issue by considering a heterogeneous fitness landscape whereby functional dimensions encompass peaks representing trait combinations yielding positive population growth rates in a community. We identify four ecological cases contributing to the emergence and persistence of functionally distinct species. First, environmental heterogeneity or alternative phenotypic designs can drive positive population growth of functionally distinct species. Second, sink populations with negative population growth can deviate from local fitness peaks and be functionally distinct. Third, species found at the margin of the fitness landscape can persist but be functionally distinct. Fourth, biotic interactions (positive or negative) can dynamically alter the fitness landscape. We offer examples of these four cases and guidelines to distinguish between them. In addition to these deterministic processes, we explore how stochastic dispersal limitation can yield functional distinctiveness. Our framework offers a novel perspective on the relationship between fitness landscape heterogeneity and the functional composition of ecological assemblages.

International audience; We consider the problem of learning time-consuming functions defined over unordered sets of vectors. Such functions arise frequently, in particular in the context of networks of devices whose number is not fixed and that interact with each other. A working example is the modeling of a wind farm. Unordered sets of vectors are a mix of integer and continuous input variables suitable for functions that are permutation-invariant. The time-consuming aspect of the functions is, classically, treated by approximating them with a Gaussian process.This study addresses the problem of defining valid and efficient covariance kernels over cloudsof points in the context of Gaussian process surrogate modeling. We review methods for defining such kernels. These kernels are compared on a set of analyticalfunctions inspired from different engineering problems, such as the design of experiments andthe modeling of wind farms production. The extrapolation properties of the kernels are testedon geometrically transformed clouds.We show that modeling 2D clouds of points as supports of discrete uniform distributions should be preferred to a Gaussian representation of the clouds. A detailed investigation of the good performance of MMD-based kernels illustrates how they adapt their hyperparameters to the geometrical properties of the studied functions.

International audience; The influence of the chemical composition on the brittle fracture behaviour of pressure vessel steels was investigated. Two model materials with chemical compositions simulating zones of moderate and severe positive macrosegregation were characterized and compared with the non-segregated material. Fractographic analysis suggests that brittle fracture mechanisms depend on the chemical composition through the induced microstructure.

Conformal prediction is a theoretically grounded framework for constructing predictive intervals. We study conformal prediction with missing values in the covariates -- a setting that brings new challenges to uncertainty quantification. We first show that the marginal coverage guarantee of conformal prediction holds on imputed data for any missingness distribution and almost all imputation functions. However, we emphasize that the average coverage varies depending on the pattern of missing values: conformal methods tend to construct prediction intervals that under-cover the response conditionally to some missing patterns. This motivates our novel generalized conformalized quantile regression framework, missing data augmentation, which yields prediction intervals that are valid conditionally to the patterns of missing values, despite their exponential number. We then show that a universally consistent quantile regression algorithm trained on the imputed data is Bayes optimal for the pinball risk, thus achieving valid coverage conditionally to any given data point. Moreover, we examine the case of a linear model, which demonstrates the importance of our proposal in overcoming the heteroskedasticity induced by missing values. Using synthetic and data from critical care, we corroborate our theory and report improved performance of our methods., Code for our experiments can be found at https://github.com/mzaffran/ConformalPredictionMissingValues . To be published in the proceedings of the 40th International Conference on Machine Learning, Honolulu, Hawaii, USA

International audience; The direct parametrisation method for invariant manifolds is a nonlinear reduction technique which derives nonlinear mappings and reduced-order dynamics that describe the evolution of dynamical systems along a low-dimensional invariant-based span of the phase space. It can be directly applied to finite element problems. When the development is performed using an arbitrary order asymptotic expansion, it provides an efficient reduced-order modeling strategy for geometrically nonlinear structures. It is here applied to the case of rotating structures featuring centrifugal effect. A rotating cantilever beam with large amplitude vibrations is first selected in order to highlight the main features of the method. Numerical results show that the method provides accurate reduced-order models (ROMs) for any rotation speed and vibration amplitude of interest with a single master mode, thus offering remarkable reduction in the computational burden. The hardening/softening transition of the fundamental flexural mode with increasing rotation speed is then investigated in detail and a ROM parametrised with respect to rotation speed and forcing frequencies is detailed. The method is then applied to a twisted plate model representative of a fan blade, showing how the technique can handle more complex structures. Hardening/softening transition is also investigated as well as interpolation of ROMs, highlighting the efficacy of the method.

International audience; Large-scale ensemble runs typically consist of executing thousands of physical simulation instances according to a range of different input parameters. These ensemble runs enable sensitivity analyses, deep surrogate trainings, reinforcement learning, and data assimilation, but they rely on volumes of data that are too large to store. For example, a recent data assimilation ensemble study generated 1.3 PB of data [@yashiro2020]. These enormous volumes of data hinder scientific analyses in two ways: first, the I/O speeds are the slowest component in supercomputers; the incongruence between slow read/write speeds compared to the rapid generation of data leads to a degradation and plateau of performance. Second, the file systems on supercomputers are not designed to allocate such large volumes of data to singular studies. To avoid this I/O limitation, scientists reduce their study size by running low resolution simulations or down-sampling output data in space and time. However, the I/O problem only becomes more pronounced as the speed and size of supercomputers continues to advance faster than I/O speeds of storage disks.

We revisit the classical zero-thickness Generalized Sheet Transition Conditions (GSTCs) which are a key tool for efficiently designing metafilms able to control the flow of light in a desired way. It is shown that it is more convenient to use an enlarged formulation of the GSTC in which the original metafilm is replaced by GSTCs that exclude the layer from the physical or computational domain. These new "layer" transition conditions have the same form as their "sheet" analogues hence they do not necessitate additional complications in their use; their advantage is that they provide a well-posed problem hence guaranty the stability of numerical schemes in the timedomain. These assessments are demonstrated for an all-dielectric structure; the effective susceptibility tensors are derived thanks to asymptotic analysis combined with homogenization technique and bounds for the susceptibilities entering the balance of energy are provided. While negative constant susceptibilities appear in the classical zero-thickness GSTCs, their values in the enlarged formulation are always positive which ensure the stability of the effective problem. Validation of the effective model is provided by means of comparison with direct numerics in two and three dimensions.

With the growing evidence that high particle number concentrations may impact health, modelling their emissions and understanding formation processes is necessary, especially in cities where many people are exposed. As emission inventories of particle numbers and size distribution over cities are usually not available, a methodology is defined to estimate them from PM2.5 emissions and ratios of PM1 / PM2.5 and PM0.1 / PM2.5 by activity sector. In this methodology, a fitting parameter αem is used to redistribute the number concentrations in the lowest emission diameter range. This parameter is chosen by comparing measured and simulated number concentrations during non-nucleation days. The emission size distribution is then finely discretised by conserving both mass and number in each of the size ranges where emissions are specified. The methodology is applied over Greater Paris during the MEGAPOLI campaign (July 2009). Three-dimensional simulations are performed using the chemistry transport model Polair3D/Polyphemus coupled to the aerosol module SSH-aerosol to represent the evolution of particles by condensation, evaporation, coagulation, and nucleation, with a sectional approach for the size distribution. The model is first compared to measurements during non-nucleation days, and the influence over the month of July 2009 of three different nucleation parameterisations is assessed, i.e. binary (sulfuric acid, water), ternary (sulfuric acid, ammonia, water), and heteromolecular (extremely low-volatility organic compounds (ELVOCs) from monoterpenes and sulfuric acid). The modelled number concentrations compare very well to measurements, with an average normalised mean error of 42 % for the daily number concentrations of particles larger than 10 nm and 37 % for the number concentrations of particles larger than 100 nm. The influence of the binary nucleation is low, and the ternary nucleation scheme leads to better simulated number concentrations (in terms of bias and error) at only one site out of three, but it systematically reduces the model to measurement correlation, suggesting that ternary nucleation may not be the dominant process in new particle formation. However, the relative bias and error, as well as the correlation at suburban sites, are systematically improved using the heteromolecular nucleation scheme involving sulfuric acid and ELVOCs from monoterpenes. This suggests that heteromolecular nucleation may be important in cities, especially at suburban sites in summer, and that a better characterisation of the emissions of ELVOC precursors from traffic is needed.

Performing (variance-based) global sensitivity analysis (GSA) with dependent inputs has recently benefited from cooperative game theory concepts.By using this theory, despite the potential correlation between the inputs, meaningful sensitivity indices can be defined via allocation shares of the model output's variance to each input. The ``Shapley effects'', i.e., the Shapley values transposed to variance-based GSA problems, allowed for this suitable solution. However, these indices exhibit a particular behavior that can be undesirable: an exogenous input (i.e., which is not explicitly included in the structural equations of the model) can be associated with a strictly positive index when it is correlated to endogenous inputs. In the present work, the use of a different allocation, called the ``proportional values'' is investigated. A first contribution is to propose an extension of this allocation, suitable for variance-based GSA. Novel GSA indices are then proposed, called the ``proportional marginal effects'' (PME). The notion of exogeneity is formally defined in the context of variance-based GSA, and it is shown that the PME allow the distinction of exogenous variables, even when they are correlated to endogenous inputs. Moreover, their behavior is compared to the Shapley effects on analytical toy-cases and more realistic use-cases.

In the context of regression analysis, importance measures are effective tools to perform feature selection or to interpret a model by ranking the most influential regressors. In particular, variance-based importance measures (VIMs) are prominent in the field of statistics, but also in the most recent field of global sensitivity analysis, due to their accessible interpretation as variance shares of the explained variable. By focusing on the linear regression model, this work aims at revisiting the overview of the most well-founded methods (some of them being rather old and sometimes, misunderstood), while clarifying their respective positioning, conditions of use, intrinsic capabilities, and interpretation. Some challenges are discussed, such as the case of dependent inputs and the case of a high input dimension. A particular set of VIMs, called the Johnson indices, has shown the potential to solve both previously mentioned challenges. Their equivalence with other known VIMs, the Lindeman-Merenda-Gold (LMG) indices, is proved in the case of two inputs. The practical relevancy of such tools is highlighted through their empirical study on simulated data, as well as public datasets. An application to classification tasks (logistic regression) is also presented in the supplementary material.

Solid oxide cells (SOCs) have attracted a growing attention thanks to their high efficiency and ability to operate in both electrolysis (SOEC) and fuel cell (SOFC) modes. Despite its great potential, the current SOC technology faces significant cell degradation during long-term operation. The degradation phenomena are still not well understood as they involve complex and intricate processes arising at different length scales. Additionally, it has been shown that the degradation rate under SOEC operation is generally higher than in SOFC mode. To address this issue, durability tests were performed in SOEC mode at 750, 800 and 850 °C for 2000 h at -1 A cm-2 in 10/90 vol.% H2/H2O and dry air. The electrochemical performance of the cells was assessed using electrochemical impedance spectroscopy. In addition, advanced characterization techniques with nanometer and atomic resolution have been used to study material degradation after long-term testing.

International audience; This article presents the engineering of disodium naphthalene dicarboxylate (Na2-NDC) based negative organic electrodes for sodium-ion batteries. Firstly, different synthesis strategies were used to modify the morphology and crystal structure of the active material particles in order to improve their electrochemical performance. The engineering of the electrodes was then carried out to maximize the composite performances, which were initially evaluated in half-cells against Na. After electrode optimization, a complete cell was assembled by coupling the Na2-NDC negative electrode with Na3V2(PO4)2F3 (NVPF) as an inorganic-based cathode, to obtain a complete so-called “hybrid” cell. A thorough electrochemical study of the complete cell was carried out to evaluate the cell stability during cycling using different electrolyte additives. Thanks to the electrode and electrolyte optimizations, the NVPF/Na2-NDC hybrid cell demonstrated remarkable electrochemical performance, in terms of reversible capacity, cycling stability and rate performance. The full cells in the present work showed an energy density between 260 and 300 Wh/kg in the first cycle at low C-rate, and an energy density of 120–155 Wh/kg after 400 cycles, superior to those previously reported in the literature for the hybrid cell.In addition, a series of post-mortem analyses were carried out, based on GC/MS analysis, UV–Vis spectroscopy and electron microscopy (SEM/EDX), to understand the evolution of the performance over time and thus guide future research in this field to design the next generation of more durable and easily recycled sodium-ion battery.

International audience; Programme d'augmentation de la part des énergies renouvelables dans le mix réseau(Plan Pluriannuel de l'Energie : PPE)• Installations massives d'EnR : de 48,6 GWc fin 2017 à 100 -110 GWc en 2028• PV : 35 à 44 GWc, Eolien : ~ 33 - 35 GWc, dont 5 à 6 GWc en mer, hydro. ~26,5 GWp• connectés au réseau (36 % de l'électricité fournie par ENR).• Développer les infrastructures de recharge des véhicules électriques, l'écrêtement des pointes etl'autoconsommation, la flexibilité et la résilience des réseaux• En 2050: besoin en énergie finale ~1 060 TWh (1 628 TWh en 2018), dont ~580 TWh électriquesOpérer une transition responsable• Qui mette l’humain au centre , qui soit bien appropriée socialement;• Qui soit conforme au développement durable et minimise les impacts environnementaux;Plan de réindustrialisation et de souveraineté (France Relance)• Moderniser l’appareil productif et décarboner l'industrie ;• (Re)localiser les activités stratégiques critiques ;• Innover, en soutenant les investissements en R&D tout en développant les compétences.

International audience; Chlorine 36 (36Cl, T1/2 = 301,000 years) is a radionuclide with natural and anthropogenic origin that can be rejected during decommissioning of nuclear power plants or chronically during recycling of used nuclear fuels. Once emitted into the atmosphere, chlorine 36 (gas and particles) can be transferred to the soil and vegetation cover by dry and wet deposition. However, quantitative constraints of these deposits are very scarce. Because of its relatively high mobility in the geosphere and its high bioavailability, chlorine 36 fate in the environment should be studied for environmental and human impact assessments. The aim of this study is therefore to develop dry depositional models for the gaseous and particulate fractions of chlorine 36. The model used for the parameterization of gaseous chlorine 36 dry deposition on grass is the << Big-leaf >> model based on the electrical analogy (Seinfield 1985). It was adapted from the Bah (2020) model developed for iodine. For particulate chlorine 36, the Damay Pellerin (2017) model was used. This model requiring the diameter of particulate chlorine 36 to be known, a sample of aerosol was taken in the chlorine 36 plume to determine on which particle size it is fixed. The sampling was performed on a low pressure impinger (LPI, DEKATI) of 13 levels for particles with diameters between 30nm and 10μm. In order to obtain model input data, meteorological and micrometeorological parameters were measured continuously at the IRSN La Hague technical platform (PTILH, France). The PTILH is located 2 km north of Orano La-Hague plant which emits small quantities of chlorine 36. An ultrasonic anemometer (Young 81000V) installed at a height of 4,5 m from the ground measured the direction (°), the velocity u (m/s) and the air friction velocity u* (m/s), the sensible heat flux H, the Monin-Obukhov length (L) and the atmospheric stability (1/L). A weather station (Spectrum Watchdog, series 2000) measured temperature Ts (°C), relative humidity RH (%), dew point (°C), global radiation SR (Wat/m^2) and photosynthetic active radiation PAR (μM/m^2.s). Sampling campaigns of 2 weeks were also conducted at the PTILH site to determine experimental depositional rates of chlorine 36. Chlorine 36 measurements were carried out by acceleration mass spectrometry at CEREGE-LN2C (France).The particle size distribution of the aerosol sample from chlorine 36 plume shows two peaks, a main one at 2μm and a second one at 10μm, typical distribution of marine aerosol. The modelsyield chlorine 36 depositional velocities between 6,7.10-3 and 10-2 m/s for the particulate fraction,and between 5.10-3 and 1,3.10-2 m/s for the gaseous one. The total dry depositional velocities(particles and gas) calculated from the model are less than one order of magnitude than the onesobtained experimentally.

Homogeneous and heterogeneous U1-xCexO2 (with 0≤ x≤ 0.25) materials were prepared via wet and dry chemistry routes, respectively before being submitted to dynamic leaching experiments. The feeding solution containing 0.20 mmol.L−1 H2O2 was kept under air and renewed to guarantee the stability of H2O2 during the experiment. Normalized alteration rates were determined from U concentration in the leachates. For homogeneous (U,Ce)O2 materials, the dissolution rate was divided by a factor of 3 when increasing the Ce content from 0.08 to 0.25. Surface characterizations revealed that studtite precipitated all over UO2 pellet surface and only on the UO2 grains of heterogeneous U0.92Ce0.08O2 samples. The behaviour of this heterogeneous material was similar to that observed for (U,Pu)O2 in the same conditions, which revealed the reliability of cerium as a plutonium analogue.

We consider the framework of convex high dimensional stochastic control problems, in which the controls are aggregated in the cost function. As first contribution, we introduce a modified problem, whose optimal control is under some reasonable assumptions an $\varepsilon$-optimal solution of the original problem. As second contribution, we present a decentralized algorithm whose convergence to the solution of the modified problem is established. Finally, we study the application of the developed tools in an engineering context, studying a coordination problem for large populations of domestic thermostatically controlled loads (TCLs), 24 pages

International audience; We are interested in how issues relating to privacy and confidentiality are incorporated when using interactive services. Two smart home services based on the use of data from "smart" electricity meters have been evaluated to establish the guarantees and forms of interaction necessary for their appropriation. We show that, given the contextual, dynamic and social nature of privacy, normative responses such as those promoted by the regulator (predefinition of sensitive data, prior to use consent, privacy as the default setting) are insufficient. As mentioned in previous works, the design choices must promote as components of the systems the user control of data throughout the interaction, as well as transparency of the data, processing and actors. However, our results highlight an important dimension not previously addressed: the need to adopt a temporal and diachronic vision of privacy and to broaden the question of interaction beyond use by considering the life cycle of the service and of the relationship with its suppliers.; Nous nous intéressons à la façon dont les questions de vie privée et de sa préservation s'instancient dans l'utilisation de services interactifs. Deux services relatifs à l'habitat et basés sur l'exploitation de données issues de compteurs électriques « intelligents » ont été évalués pour établir les garanties et formes d'interactions nécessaires à leur appropriation. Nous montrons que, compte tenu de la nature contextuelle, dynamique et sociale de la privacy, les réponses normatives telles que celles promues par le législateur (prédéfinition des données sensibles, consentement a priori, privacy intégrée seulement au stade de la conception du système) sont insuffisantes. Comme cela a pu être évoqué dans des travaux antérieurs, les choix de conception doivent favoriser la maîtrise des données dans l'interaction en proposant des moyens d'actions et une transparence des données, traitements et acteurs, en tant que composantes des systèmes. Toutefois, nos résultats mettent en exergue une dimension importante non traitée jusqu'alors : la nécessité d'adopter une vision temporelle et diachronique de la privacy, pour élargir la question de l'interaction au-delà de l'utilisation en considérant la durée de vie du service et de la relation avec ses fournisseurs.

International audience; Misspecifying external forcings (such as rain) on a hydrological model can directly affect subsequent parameter calibrations. Indeed, by using classical calibration and problem inversion methods, the error in the external forcings is propagated to the model output, and then, if not treated correctly, this error is compensated by overcalibrating the model parameters. As a consequence, parameter values that were found optimal for one value of the external forcings, are not granted to be optimal for another one. Ideally however, estimated parameter values (that describe time-invariant soil properties) should be the same no matter the value of the external forcing.

International audience; We consider the task of estimating functions from a restricted number of observations where the inputs are in the form of varying-size sets of vectors. A classical method in this expensive functions context is to approximate the true expensive function with a Gaussian process that relies on semi-definite positive kernels. Varying-size sets of vectors have some peculiar characteristics : the number of vectors is not fixed and the vectors are not ordered. These properties prevent the use of classical, fixed dimension, kernels. Sets of vectors can be represented as probability distributions or as vectors composed of wellchosen characteristic features. The distribution associated with each set of vectors can be either a (non-informative) discrete uniform distribution supported by the points of the set, or an empirical Gaussian distribution characterized by a mean and a covariance matrix estimated from the vectors of the set as in [5]. With this representation, we can define semi-definite positive kernels between distributions or vectors. The kernels can be classified into explicit ones, with known feature map and scalar product, and implicit kernels where the feature maps and the scalar product come mixed together. For implicit kernels, we restrict ourselves to distance substitution kernels with an exponential form. It is known that this approach for constructing kernels requires the substitute to be conditionally semi-definite negative, see [1]. In this context, it is necessary and sufficient to have an Hilbertian metric (isometric to a L 2 norm) [4]. The distances between distributions verifying the latter condition which we consider are the sliced Wasserstein distance [3], the maximum mean discrepancy, the approximated Gaussian Wasserstein (see [2]), and the Euclidean distance for vectors of features. For explicit kernels, we investigate the probability product kernel between Gaussian distributions [5] and the scalar product of embeddings of discrete uniform laws. The prediction performance of theses kernels are compared on three types of analytical functions. The first ones mimic wind-farms productions for different winds, the second function is the inertia of a set of vectors and the third is the maximin function, well-known in the design of experiments. We first analyze the predictions over sets of vectors drawn uniformly over the same rectangular search space. Afterwards, we investigate the predictions over sets of vectors modified through geometrical transformations such as dilatation, rotation and translation. We show that some geometrical properties of the wind-farm functions can be better learned by MMD-based kernels than others. As an example, the MMD-based kernel prediction together with the true function is represented in Figure 1.

International audience; The Paris region (France) is increasingly the focus of urban atmospheric research. Numerous national and international research projects have chosen Europe’s largest metropolitan region as their study area to better understand and predict critical hazards (incl. heat, air pollution, thunderstorms) in the context of a changing climate. Located on rather flat terrain in continental, mid-latitude climates, the densely populated Paris region is very suitable for the evaluation of urban processes in numerical simulations at different scales. The European research infrastructures ACTRIS and ICOS are developing strategies for the improved operational monitoring of air pollution and greenhouse gas budgets, respectively. Various research projects are conducting fundamental process studies and model developments to investigate the dynamics and chemistry of the urban atmosphere and its interactions with the rural surroundings and regional-scale flow to better quantify associated health risks and inform sustainable planning.In addition to numerous modelling activities (chemistry-transport, numerical weather prediction, climate projections), diverse atmospheric observations are collected. These include dense surface station networks, turbulent flux towers, and ground-based atmospheric remote sensing to monitor the atmospheric boundary layer. This multi-project context motivates the pooling of resources.To facilitate efficient project synergy and to optimise the coordination of the individual experimental campaigns, the PANAME initiative (https://paname.aeris-data.fr/) was established. PANAME provides a framework to optimise the design of the Paris region measurement network and helps to standardise the operations. A professional, multi-disciplinary data portal is developed at the French AERIS atmospheric data centre to host the PANAME observations and model results. Here, data are collected and formatted, standardised advanced products are derived from the diverse sensor networks and high-quality visualisations are generated in near real-time. The presentation will provide an overview on the scientific objectives of the on-going projects, the deployment of measurements and simulation tools, and the data portal design.

We consider a revenue maximization model, in which a company aims at designing a menu of contracts, given a population of customers. A standard approach consists in constructing an incentive-compatible continuum of contracts, i.e., a menu composed of an infinite number of contracts, where each contract is especially adapted to an infinitesimal customer, taking his type into account. Nonetheless, in many applications, the company is constrained to offering a limited number of contracts. We show that this question reduces to an optimal quantization problem, similar to the pruning problem that appeared in the max-plus based numerical methods in optimal control. We develop a new quantization algorithm, which, given an initial menu of contracts, iteratively prunes the less important contracts, to construct an implementable menu of the desired cardinality, while minimizing the revenue loss. We apply this algorithm to solve a pricing problem with price-elastic demand, originating from the electricity retail market. Numerical results show an improved performance by comparison with earlier pruning algorithms.

In decommissioning projects of nuclear facilities, the radiological characterisation step aims to estimate the quantity and spatial distribution of different radionuclides. To carry out the estimation, measurements are performed on site to obtain preliminary information. The usual industrial practice consists in applying spatial interpolation tools (as the ordinary kriging method) on these data to predict the value of interest for the contamination (radionuclide concentration, radioactivity, etc.) at unobserved positions. This paper questions the ordinary kriging tool on the well-known problem of the overoptimistic prediction variances due to not taking into account uncertainties on the estimation of the kriging parameters (variance and range). To overcome this issue, the practical use of the Bayesian kriging method, where the model parameters are considered as random variables, is deepened. The usefulness of Bayesian kriging, whilst comparing its performance to that of ordinary kriging, is demonstrated in the small data context (which is often the case in decommissioning projects). This result is obtained via several numerical tests on different toy models, and using complementary validation criteria: the predictivity coefficient (Q${}^2$), the Predictive Variance Adequacy (PVA), the $\alpha$-Confidence Interval plot (and its associated Mean Squared Error alpha (MSEalpha)), and the Predictive Interval Adequacy (PIA). The latter is a new criterion adapted to the Bayesian kriging results. Finally, the same comparison is performed on a real dataset coming from the decommissioning project of the CEA Marcoule G3 reactor. It illustrates the practical interest of Bayesian kriging in industrial radiological characterisation.

International audience; Recent studies showed both experimental and numerical evidence that the occurrence probability of freak waves could be significantly enhanced as results of non-equilibrium dynamics induced by strong depth variations. The sea-state is characterized by strong non-Gaussian behavior in a short spatial extent after the depth transition, covering a few wavelengths. In this work, we investigate the complete equilibration process of an out-of-equilibrium sea-state via high-fidelity numerical simulations. In the simulations, the region after the depth transition is set as long as around one hundred wavelengths, such that the spectral adaptation develops and terminates eventually. The results are analyzed with spectral, cross-spectral and statistical approaches. It is shown that there are two stages with different spatial scales in the equilibration process. In the short scale, the sea-state is characterized by significant changes in wave statistics, freak wave occurrence probability is intensified. In the long scale, the wave spectrum undergoes strong modulation, the spectral peak disintegrate into a relative broad band, and low-frequency waves are enhanced as well. We show evidence that the spectral changes in the long scale are due to interactions of free components. The wave nonlinearity is shown to be positively correlated to the magnitude of the dynamical responses, but irrelevant to the length of the spatial scales in the equilibration process. In the established shallow-water equilibrium, the freak wave occurrence probability becomes less than Gaussian expectation and the waves are asymmetric in the vertical direction and symmetric in the horizontal.

International audience; We consider an economy made of competing firms which are heterogeneous in their capital and use several inputs for producing goods. Their consumption policy is fixed rationally by maximizing a utility and their capital cannot fall below a given threshold (state constraint). We aim at modeling the interactions between firms on the markets of the different inputs on the long term. The stationary equlibria are described by a system of coupled non-linear differential equations: a Hamilton-Jacobi equation describing the optimal control problem of a single atomistic firm; a continuity equation describing the distribution of the individual state variable (the capital) in the population of firms; the equilibria on the markets of the production factors. We prove the existence of equilibria under suitable assumptions.

International audience; Nanostructure control is an important issue when using electroactive materials in energy conversion and storage devices. In this study, we report various methods of synthesis of nanostructured copper (II) hydroxide nitrate (Cu-2(OH)(3)NO3) with a layered hydroxide salt (LHS) structure using various synthesis methods and investigate the correlation between nanostructure, morphology, and their pseudocapacitive electrochemical behavior. The variations in nanostructure size and morphology were comprehensively explored by combining X-ray diffraction (XRD) and scanning electron microscopy (SEM), while the electrochemical activity was characterized using cyclic voltammetry. We demonstrate that Cu-2(OH)(3)NO3-LHS nanostructured submicron particles produced by alkaline precipitation with 88% of the copper cations can cycle with a two-electron redox process. Unfortunately, the electroactivity decreases rapidly from the first cycle due to the occurrence of structural transformations and subsequent electrochemical grinding. However, samples obtained by ultrasonication and microwave synthesis, two original synthesis methods for LHS materials, formed of nanosized crystalline domains agglomerated in micron-sized particles, represent a good compromise between capacity and cyclability. Moreover, by using pair distribution function analysis on electrode materials after repeated cycling, we were able to follow the chemical and structural changes occurring in Cu-2(OH)(3)NO3 materials during electrochemical cycling with first a quick transformation to Cu2O and then the appearance of Cu metal and copper acetate Cu(II)(2)(O2CCH3)(4)center dot 2H(2)O.

The use of serial robots for industrial and research purposes is often limited by a flawed positioning accuracy, caused by the differences between the robot nominal model, and the real one. Such an issue can be solved by means of kinematic calibration, which are usually tedious and intricate tasks. In this paper, we propose a complete kinematic calibration procedure relying on established geometric modeling, measurements design and parameters identification methods, as well as multiple integration tools, to provide a high adaptability as well as a simplified handling. The overall process was bundled up in a ROS-based modular and user-friendly kinematic calibration package for serial robots, whose main objective is to offer a smooth and fully integrated framework. Our solution was successfully tested using a motion tracking device, and allowed to increase the overall positioning accuracy of two different serial robots by 75% in a matter of hours.

International audience; Optimizing nuclear unit outages is of significant economic importance for the French electricity company EDF, as these outages induce a substitute production by other more expensive means to fulfill electricity demand. This problem is quite challenging given the specific operating constraints of nuclear units, the stochasticity of both the demand and non-nuclear units availability, and the scale of the instances. To tackle these difficulties we use a combined decomposition approach. The operating constraints of the nuclear units are built into a Dantzig-Wolfe pricing subproblem whose solutions define the columns of a demand covering formulation. The scenarios of demand and non-nuclear units availability are handled in a Benders decomposition. Our approach is shown to scale up to the real-life instances of the French nuclear fleet.

We propose several algorithms to solve McKean-Vlasov Forward Backward Stochastic Differential Equations. Our schemes rely on the approximating power of neural networks to estimate the solution or its gradient through minimization problems. As a consequence, we obtain methods able to tackle both mean field games and mean field control problems in moderate dimension. We analyze the numerical behavior of our algorithms on several examples including non linear quadratic models., 29 pages, revised version, to appear in Methodology and Computing in Applied Probability

International audience; A novel approach to damage modeling for quasi-brittle solids is presented relying upon a differential inclusion that is closely related to the one of implicit gradient models. The proposed formulation naturally fits in the socalled non-local standard approach, whereby the framework of GeneralizedStandard Materials is extended to include gradients of internal variables to account for the physics of the fracture phenomenon in a regularized sense, i.e. via extended constitutive equations in which a length scale parameter brings to the macro level information about material microstructure. This concept is fully embodied into the present approach to quasi-brittle fracture, whereby progressive damage occurs in layers of finite thickness where the gradient of damage is bounded and a fully damaged region is understood as a fracture with no ambiguity. Key to the effective implementation of the model are the choice of two constitutive functions and the implicit tracking of regions in a state of progressive damage via Lagrange multipliers acting on internal constraints. The ideas are developed for a general Cauchy continuum and representative numerical simulations are included that demonstrate the model capabilities.

International audience; The main objective of this article is to investigate the capability of the flamelet progress variable (FPV) model to capture the extinction processes observed in under-ventilated fire scenarios. To this end, large eddy simulation (LES) of the methane line fire plumes in oxygen-reduced environments down to global extinction, investigated experimentally at the University of Maryland (UMD), is performed. Two experimental burner configurations, that differ by the presence (anchored) or not (non-anchored) of an oxygen anchor to stabilise the flame base, are considered leading to two different extinction modes. Both the FPV and the steady laminar flamelet (SLF) model coupled with a presumed filtered density function (FDF) are considered. The Rank Correlated Full Spectrum k-distribution (RCFSK) model is used as a gas radiative property model. In both non-anchored and anchored scenarios, the FPV model reproduces with fidelity the evolution of the fire plume structure, radiative loss, and combustion efficiency with decreasing XO2 down to global extinction, without introducing any adjustable constant. The extinction in the non-anchored scenario occurs owing to flame-based detachment coupled to the generation of a buoyancy-driven vortex and is found to be very sensitive to the grid resolution in the near burner region. The present results suggest that these processes can be adequately resolved with a spatial resolution of 2.5 mm in this region. The SLF model, for its part, provides reliable predictions comparable to the FPV as long as no local extinction/re-ignition process occurs.

Optimizing maintenance scheduling is a major issue to improve the performance of hydropower plants. We study a system of several physical components of the same family: either a set of turbines, a set of transformers or a set of generators. The components share a common stock of spare parts and experience random failures that occur according to known failure distributions. We seek a deterministic preventive maintenance strategy that minimizes an expected cost depending on maintenance and forced outages of the system. The Auxiliary Problem Principle is used to decompose the original large-scale optimization problem into a sequence of independent subproblems of smaller dimension while ensuring their coordination. Each subproblem consists in optimizing the maintenance on a single component. Decomposition-coordination techniques are based on variational techniques but the maintenance optimization problem is a mixed-integer problem. Therefore, we relax the dynamics and the cost functions of the system. The resulting algorithm iteratively solves the subproblems on the relaxed system with a blackbox method and coordinates the components. Relaxation parameters have an important influence on the optimization and must be appropriately chosen. An admissible maintenance strategy is then derived from the resolution of the relaxed problem. We apply the decomposition algorithm on a system with 80 components. It outperforms the reference blackbox method applied directly on the original problem.

We study the forecasting of the power consumptions of a population of households and of subpopulations thereof. These subpopulations are built according to location, to exogenous information and/or to profiles we determined from historical households consumption time series. Thus, we aim to forecast the electricity consumption time series at several levels of households aggregation. These time series are linked through some summation constraints which induce a hierarchy. Our approach consists in three steps: feature generation, aggregation and projection. Firstly (feature generation step), we build, for each considering group for households, a benchmark forecast (called features), using random forests or generalized additive models. Secondly (aggregation step), aggregation algorithms, run in parallel, aggregate these forecasts and provide new predictions. Finally (projection step), we use the summation constraints induced by the time series underlying hierarchy to re-conciliate the forecasts by projecting them in a well-chosen linear subspace. We provide some theoretical guaranties on the average prediction error of this methodology, through the minimization of a quantity called regret. We also test our approach on households power consumption data collected in Great Britain by multiple energy providers in the Energy Demand Research Project context. We build and compare various population segmentations for the evaluation of our approach performance.

In this paper, we propose an algorithmic framework to automatically generate efficient deep neural networks and optimize their associated hyperparameters. The framework is based on evolving directed acyclic graphs (DAGs), defining a more flexible search space than the existing ones in the literature. It allows mixtures of different classical operations: convolutions, recurrences and dense layers, but also more newfangled operations such as self-attention. Based on this search space we propose neighbourhood and evolution search operators to optimize both the architecture and hyper-parameters of our networks. These search operators can be used with any metaheuristic capable of handling mixed search spaces. We tested our algorithmic framework with an evolutionary algorithm on a time series prediction benchmark. The results demonstrate that our framework was able to find models outperforming the established baseline on numerous datasets.

International audience; Traditional computational approaches in simulating crack propagation in perfectly brittle materials rely on the estimate of stress intensity factors along the rupture front. This proves highly challenging in 3D when the crack geometry departs from very specific cases for which analytical solutions are available, like e.g. the penny-shaped crack geometry. Here, we extend the first-order theory of Gao and Rice (1987), and predict the distribution of the mode I stress intensity factor KI along the front of a tensile coplanar crack that is slightly perturbed from a reference penny-shaped configuration, up to second order in the perturbation amplitude. Our theory is validated against analytical solutions available for embedded elliptical cracks, and its range of validity is further assessed using numerical simulations performed on cosine front perturbations of varying mode and amplitude. It is then used to develop a homogenization framework for the toughness of weakly disordered media. The effective toughness and its fluctuations are bridged quantitatively to the intensity of the toughness fluctuations and their spatial structure. Our theoretical predictions are compared to the results of ~1 million simulations of crack propagation building on our second-order theory and Fast Fourier Transforms. We show that the impact of toughness heterogeneities is size-dependent, as they generally weaken the material when the crack size is lower or comparable to the typical heterogeneity size, but reinforces it otherwise. It results in an apparent R-curve behavior of the brittle composite at the macroscale.