21549 research outputs found
Sort by
SPECFEM2D-DG, an open-source software modelling mechanical waves in coupled solid–fluid systems: the linearized Navier–Stokes approach
No full textWe introduce SPECFEM2D-DG, an open-source, time-domain, hybrid Galerkin software
modelling the propagation of seismic and acoustic waves in coupled solid–fluid systems. For
the solid part, the visco-elastic system from the routinely used SPECFEM2D software is used
to simulate linear seismic waves subject to attenuation. For the fluid part, SPECFEM2D-DG
includes two extensions to the acoustic part of SPECFEM2D, both relying on the Navier–
Stokes equations to model high-frequency acoustics, infrasound and gravity waves in complex
atmospheres. The first fluid extension, SPECFEM2D-DG-FNS, was introduced in 2017 by
Brissaud, Martin, Garcia, and Komatitsch; it features a nonlinear Full Navier–Stokes (FNS)
approach discretized with a discontinuous Galerkin numerical scheme. In this contribution,
we focus only on introducing a second fluid extension, SPECFEM2D-DG-LNS, based on the
same numerical method but rather relying on the Linear Navier–Stokes (LNS) equations. The
three main modules of SPECFEM2D-DG all use the spectral element method (SEM). For
both fluid extensions (FNS and LNS), two-way mechanical coupling conditions preserve the
Riemann problem solution at the fluid–solid interface. Absorbing outer boundary conditions
(ABCs) derived from the perfectly matched layers’ approach is proposed for the LNS exten-
sion. The SEM approach supports complex topographies and unstructured meshes. The LNS
equations allow the use of range-dependent atmospheric models, known to be crucial for the
propagation of infrasound at regional scales. The LNS extension is verified using the method
of manufactured solutions, and convergence is numerically characterized. The mechanical
coupling conditions at the fluid–solid interface (between the LNS and elastodynamics systems
of equations) are verified against theoretical reflection-transmission coefficients. The ABCs in
the LNS extension are tested and prove to yield satisfactory energy dissipation. In an example
case study, we model infrasonic waves caused by quakes occurring under various topographies;
we characterize the acoustic scattering conditions as well as the apparent acoustic radiation
pattern. Finally, we discuss the example case and conclude by describing the capabilities of
this software. SPECFEM2D-DG is open-source and is freely available online on GitHub
Sunflower Hydrogenation in Taylor Flow Conditions: Experiments and Computational Fluid Dynamics Modeling Using a Moving Mesh Approach
Get PDFSunflower oil hydrogenation was carried out in a 2 mm diameter jacketed capillary reactor coated with a Pd/Al2O3 catalyst, mimicking a channel of a heat exchanger monolith reactor. The operating conditions were chosen to ensure Taylor flow conditions and to evaluate their impact on the reaction selectivity. CFD simulations of the experiments were performed using the unit cell approach. They accounted for the dependence of viscosity on the degree of oil saturation, kinetic laws describing the effects of pressure on cis/trans selectivity, and bubble shrinkage along the channel using a moving mesh strategy. The model captured the experimental trends, in which the fraction of monounsaturated cis fatty acids did not exceed 35% (vs 30% originally). Poor selectivity is mainly due to the strong mass transfer resistance at the catalyst wall, either from fatty acids (favoring their complete saturation) and/or from hydrogen (due to bubble shrinkage, favoring cis to trans isomerization
Network-driven anomalous transport is a fundamental component of brain microvascular dysfunction
Get PDFBlood microcirculation supplies neurons with oxygen and nutrients, and contributes to clearing their neurotoxic waste, through a dense capillary network connected to larger treelike vessels. This complex microvascular architecture results in highly heterogeneous blood flow and travel time distributions, whose origin and consequences on brain pathophysiology are poorly understood. Here, we analyze highly-resolved intracortical blood flow and transport simulations to establish the physical laws governing the macroscopic transport properties in the brain micro-circulation. We show that network-driven anomalous transport leads to the emergence of critical regions, whether hypoxic or with high concentrations of amyloid-β, a waste product centrally involved in Alzheimer’s Disease. We develop a Continuous-Time Random Walk theory capturing these dynamics and predicting that such critical regions appear much earlier than anticipated by current empirical models under mild hypoperfusion. These findings provide a framework for understanding and modelling the impact of microvascular dysfunction in brain diseases, including Alzheimer’s Disease
Clusters of Defects as a Possible Origin of Random Telegraph Signal in Imager Devices: a DFT based Study
Get PDFThe origin of the random telegraph signal (RTS) observed in semiconductors-based electronic devices is still subject to debates. In this work, by means of atomistic simulations, typical clusters of defects as could be obtained after irradiation or implantation are studied as a possible cause for RTS. It is shown that:(i)a cluster of defects is highly metastable,(ii)it introduces several electronic states in the band gap,(iii)it has an electronic cross section much higher than the one of point defects.These three points can simultaneously explain why an electron-hole generation rate can switch with time, while respecting the experimental measurement
Potentiel de l'efficacité digestive pour l'amélioration génétique de l'efficacité alimentaire du porc en croissance dans un contexte de diversification des ressources alimentaires, et rôle du microbiote intestinal
Get PDFLa durabilité de la filière porcine est fortement affectée par la volatilité du coût des matières premières, ce qui entraîne une hétérogénéité croissante des ressources alimentaires utilisées en élevage. Dans ce contexte, l’objectif principal de cette thèse était de caractériser la variabilité génétique de l’efficacité digestive chez le porc en croissance en fonction de son alimentation, nouveau levier de sélection possible pour consolider l’amélioration de l’efficacité alimentaire. Le projet s’appuie sur un dispositif dédié, constitué de 1663 porcs Large White, dont 880 nourris avec un régime conventionnel (CO) (blé, orge) et 783 de leurs pleins-frères nourris avec un régime alternatif, riche en fibres (F) (son de blé, coques de soja, pulpe de betterave). Des développements méthodologiques préalables ont permis de prédire l'efficacité digestive individuelle sur la base d'un échantillon de fèces analysé par spectrométrie dans le proche infrarouge. Ainsi, les coefficients d’utilisation digestive (CUD) individuels de l’énergie, de la matière organique et de l’azote ont été prédits. Les performances d’efficacité alimentaire, de composition du microbiote intestinal (issu de séquençage partiel du gène de l’ARNr 16S), partenaire majeur de la digestion, et les génotypes (puce 70K) étaient par ailleurs disponibles pour tous les animaux. Dans un premier temps, la variabilité génétique de l’efficacité digestive a été estimée pour la première fois chez le porc en croissance : les trois CUD étaient modérément héritables pour les porcs nourris avec le régime CO (~0,25) et fortement avec le régime F (~0,55). L’efficacité digestive peut donc alors être intéressante pour améliorer l’efficacité alimentaire, car héritable et favorablement corrélée à ce caractère du point de vue génétique (44 %), part qui était supérieure à celle expliquée par la génétique de l’hôte (<32%), contrairement aux autres caractères étudiés. Le microbiote intestinal semble donc être une source d’information pertinente pour prédire l’efficacité digestive. Pour confirmer cette hypothèse, une étude de validation croisée a été réalisée pour évaluer l’intérêt de combiner les informations génomiques de l’hôte et du microbiote dans des modèles pour prédire les phénotypes d’efficacité digestive. Les précisions de prédiction obtenues étaient élevées dans les deux régimes, confirmant que la composition du microbiote intestinal est une source d’information pertinente pour prédire l’efficacité digestive. Ainsi, l’efficacité digestive est un nouveau phénotype d’intérêt pour améliorer l’efficacité alimentaire chez le porc en croissance, notamment dans un contexte de diversification des ressources alimentaires. Le microbiote intestinal peut être envisagé comme une source d’information pertinente pour la prédire, et les interactions génétique x aliment semblent limitées. Pour utiliser ces informations en routine, des études simulant l’intégration de ces phénotypes dans les schémas de sélection devront être menées. Finalement, pour mieux comprendre les mécanismes biologiques de la variabilité génétique de l’efficacité digestive, l’étude d’autres facteurs, tels que la structure du tractus digestif, pourraient compléter les approches décrites dans cette thèse
Caractérisation du dérangement des cétacés en mer méditerranée française et méthodes d'évaluation de ce dérangement, de l'échelle individuelle à l'échelle des populations
Get PDFLes cétacés de la mer méditerranée sous soumis à de nombreuses activités anthropiques susceptibles de les déranger. Le dérangement anthropique est défini comme toute interaction humain-animal susceptible de modifier le comportement de ce dernier. Ces modifications peuvent avoir des impacts sur la physiologie des individus, par le biais d’un stress aigu à chronique. L’impact de nos activités sur les cétacés, à l’échelle populationnelle, est un sujet très discuté aujourd’hui dans la communauté scientifique. Il est en effet très difficile de monitorer ces animaux dans la nature. A l’échelle individuelle, il est possible de monitorer l’impact du dérangement grâce à des observations comportementales, ou en réalisant des échantillons afin de mesurer des facteurs de stress, ou des hormones. Pour le moment, les quelques outils proposés pour mesurer l’impact du dérangement à l’échelle populationnelle sont des modélisations. De nos jours, le manque de données concernant les espèces présentes en mer Méditerranée ne nous permettent pas de donner une estimation correcte des conséquences du dérangement à l’échelle des populations. Des indicateurs de bien-être de ces animaux et d’impact du dérangement sont en cours d’élaboration. En l’absence de résultats actuels à ce sujet, les sollicitations d’experts peuvent constituer une bonne méthode de prédiction pour orienter les choix politiques afin de protéger ces espèces au mieux. Une telle sollicitation a été effectuée au cours de cette thèse, qui ne nous permet pas de conclure à ce jour en raison d’un échantillon trop faible de réponses
Bayesian networks in risk informed decision-making
No full textWe live in an era where every human entity, from a simple citizen to the head of an entity as large as a country, is forced to make risky decisions. Indeed, risk is omnipresent in human activities or subject to humans today. We can take the example of the COVID 19 pandemic which surprised everyone; of course, in this case we are completely helpless given the fact that there is no feedback from former experiences. Fortunately, there are cases of risk where feedback can be used to formalize and model future events in order to control or manage the risk. This requires adequate modeling, analysis and decision support tools that can be used for decision support purposes for industrial systems managers, policy makers or others. The objective of this chapter is therefore to present the possibilities of risk assessment and management assistance built around Bayesian tools. The Bayesian networks have this faculty to consider the uncertainty linked as well to the characterization of the elements of the decision problem as to the relations linking these elements. In this chapter, we will therefore show how Bayesian technology can be effectively exploited to construct models for risk evaluation and management in almost all socio-economic and even environmental fields. The principle we adopt is one of presentation - illustration in the sense that each Bayesian technology will first be presented followed by an illustration of its use in solving concrete problems. We will try to diversify the problems of illustrations through the socio-economical fields as much as possible in order to demonstrate the possibilities around the Bayesian network technology, which go from prediction, diagnosis, prognosis, to the optimization of decision-making processes
Sunflower oil hydrogenation mechanisms and kinetics
Get PDFA kinetic model for sunflower oil hydrogenation on a palladium catalyst is proposed based on Horiuty–Polanyi type mechanism and considering either a dissociative or associative adsorption of hydrogen. The derived kinetic laws allow to explain the distinct dependency of saturation and isomerization reactions on hydrogen pressure. Kinetic parameters are identified based on batch slurry hydrogenations carried out at 60–160◦C and 2–31 bar with a powdered Pd/Al2O3 catalyst. A statistical analysis is used to select the most suitable adsorption mechanism for H2. Evaluation of the Weisz–Prater modulus reveals that limitations to intraparticle diffusion occur despite the particle diameter does not exceed 40 μm
Contribution to numerical optimization with applications to engineering problems
Get PDFThe rapid development of artificial intelligence and computational sciences has attracted much more attention from researchers and practitioners to numerical optimization. In this manuscript our goal is to propose competitive numerical optimization methods with demonstrable performances on a variety of application problems from data assimilation, aircraft design or data science. More specifically, our contribution to the field of numerical optimization addresses the following three challenges.
The first challenge is related to optimization problems where the derivative information is not available or hard to obtain in practice. We will show how ideas from deterministic derivative free optimization can improve the efficiency and the rigorousness of a class of evolution strategies. Our proposed framework achieves rigorously global convergence under reasonable assumptions. The obtained method is also extended to handle general constrained optimization problems.
The second challenge is related to the Levenberg-Marquardt algorithm (LM) which is one of the most popular algorithms for the solution of nonlinear least squares problems. Motivated by exploiting the problem structure in data assimilation, we consider solving general nonlinear least-squares problems where one may have a solution with a non zero residual. We will present and analyze the convergence properties of a novel LM method that carefully balances the opposing objectives of ensuring global convergence and stabilizing a fast local convergence regime. An extension of the proposed framework to solve large-scale constrained inverse problems is also proposed.
The third challenge is related to optimization problems arising in machine learning or in parameter identification, where it can be computationally challenging or even infeasible to evaluate the exact objective function or its derivatives. The presence of noise is particularly challenging for most of the classical optimization methods. In this context, we will discuss novel methods with both favorable convergence properties and satisfactory practical performance.
Finally, we will present algorithmic and computational aspects to solve engineering optimization problems related to structural and aircraft design efficiently
A Reconstruction Algorithm for Temporally Aliased Seismic Signals Recorded by the InSight Mars Lander
No full textIn December 2018, the NASA InSight lander successfully placed a seismometer on the
surface of Mars. Alongside, a hammering device was deployed at the landing site that penetrated into the
ground to attempt the first measurements of the planetary heat flow of Mars. The hammering of the heat
probe generated repeated seismic signals that were registered by the seismometer and can potentially
be used to image the shallow subsurface just below the lander. However, the broad frequency content of
the seismic signals generated by the hammering extends beyond the Nyquist frequency governed by the
seismometer's sampling rate of 100 samples per second. Here, we propose an algorithm to reconstruct the
seismic signals beyond the classical sampling limits. We exploit the structure in the data due to thousands
of repeated, only gradually varying hammering signals as the heat probe slowly penetrates into the
ground. In addition, we make use of the fact that repeated hammering signals are sub-sampled differently
due to the unsynchronized timing between the hammer strikes and the seismometer recordings. This
allows us to reconstruct signals beyond the classical Nyquist frequency limit by enforcing a sparsity
constraint on the signal in a modified Radon transform domain. In addition, the proposed method reduces
uncorrelated noise in the recorded data. Using both synthetic data and actual data recorded on Mars, we
show how the proposed algorithm can be used to reconstruct the high-frequency hammering signal at
very high resolution