arXiv subjects

Elie Hachem

Publications and source records attributed to Elie Hachem.

33 records · Page 2Linked to original sources

arXiv2022

Aggregation and disaggregation processes in clusters of particles: simple numerical and theoretical insights of the competition in 2D geometries

Aggregation and disaggregation of clusters of attractive particles under flow are studied from numerical and theoretical points of view. Two-dimensional molecular dynamics simulations of both Couette and Poiseuille flows highlight the growth of the average steady-state cluster size as a power law of the adhesion number, a dimensionless number that quantifies the ratio of attractive forces to shear stress. Such a power-law scaling results from the competition between aggregation and disaggregation processes, as already reported in the literature. Here, we rationalize this behavior through a model based on an energy function, which minimization yields the power-law exponent in terms of the cluster fractal dimension, in good agreement with the present simulations and with previous works.

cond-mat.soft↗

arXiv2022

Proposition of extension of models relating rheological quantities and microscopic structure through the use of a double fractal structure

Colloidal suspensions and the relation between their rheology and their microstructure is investigated. The literature showed great evidence of the relation between rheological quantities and particle volume fraction, ignoring the influence of the cluster. We propose to extend previous models using a new double fractal structure which allows, first, to recover the well-known models on the case of percolated system and, second, to capture the influence of the cluster size. This new model emphasises the necessity of such structure to account for recent experimental results. Then, the model is compared with data coming from the literature and shows close agreement.

cond-mat.soft↗

arXiv2021

Policy-based optimization: single-step policy gradient method seen as an evolution strategy

This research reports on the recent development of a black-box optimization method based on single-step deep reinforcement learning (DRL), and on its conceptual proximity to evolution strategy (ES) techniques. In the fashion of policy gradient (PG) methods, the policy-based optimization (PBO) algorithm relies on the update of a policy network to describe the density function of its next generation of individuals. The method is described in details, and its similarities to both ES and PG methods are pointed out. The relevance of the approach is then evaluated on the minimization of standard analytic functions, with comparison to classic ES techniques (ES, CMA-ES). It is then applied to the optimization of parametric control laws designed for the Lorenz attractor. Given the scarce existing literature on the method, this contribution definitely establishes the PBO method as a valid, versatile black-box optimization technique, and opens the way to multiple future improvements allowed by the flexibility of the neural networks approach.

math.OC↗

arXiv2021

A twin-decoder structure for incompressible laminar flow reconstruction with uncertainty estimation around 2D obstacles

Over the past few years, deep learning methods have proved to be of great interest for the computational fluid dynamics community, especially when used as surrogate models, either for flow reconstruction, turbulence modeling, or for the prediction of aerodynamic coefficients. Overall, exceptional levels of accuracy have been obtained, but the robustness and reliability of the proposed approaches remain to be explored, particularly outside of the confidence region defined by the training dataset. In this contribution, we present a twin auto-encoder network for incompressible laminar flow reconstruction with uncertainty estimation around 2D obstacles. The proposed architecture is trained over a dataset composed of numerically-computed laminar flows around 12,000 random shapes, and naturally enforces a quasi-linear relation between a geometric reconstruction branch and the flow prediction decoder. Based on this feature, two uncertainty estimation processes are proposed, allowing either a binary decision (accept or reject prediction), or proposing a confidence interval along with the flow quantities prediction (u,v,p). Results over dataset samples as well as unseen shapes show a strong positive correlation of this reconstruction score to the mean-squared error of the flow prediction. Such approaches offer the possibility to warn the user of trained models when the provided input shows too large deviation from the training data, making the produced surrogate model conservative for fast and reliable flow prediction.

physics.flu-dyn↗

arXiv2021

Graph neural networks for laminar flow prediction around random 2D shapes

In the recent years, the domain of fast flow field prediction has been vastly dominated by pixel-based convolutional neural networks. Yet, the recent advent of graph convolutional neural networks (GCNNs) have attracted a considerable attention in the computational fluid dynamics (CFD) community. In this contribution, we proposed a GCNN structure as a surrogate model for laminar flow prediction around 2D obstacles. Unlike traditional convolution on image pixels, the graph convolution can be directly applied on body-fitted triangular meshes, hence yielding an easy coupling with CFD solvers. The proposed GCNN model is trained over a data set composed of CFD-computed laminar flows around 2,000 random 2D shapes. Accuracy levels are assessed on reconstructed velocity and pressure fields around out-of-training obstacles, and are compared with that of standard U-net architectures, especially in the boundary layer area.

physics.flu-dyn↗

arXiv2021

Robust deep learning for emulating turbulent viscosities

From the simplest models to complex deep neural networks, modeling turbulence with machine learning techniques still offers multiple challenges. In this context, the present contribution proposes a robust strategy using patch-based training to learn turbulent viscosity from flow velocities, and demonstrates its efficient use on the Spallart-Allmaras turbulence model. Training datasets are generated for flow past two-dimensional (2D) obstacles at high Reynolds numbers and used to train an auto-encoder type convolutional neural network with local patch inputs. Compared to a standard training technique, patch-based learning not only yields increased accuracy but also reduces the computational cost required for training.

physics.flu-dyn↗

arXiv2021

Deep reinforcement learning for the control of conjugate heat transfer with application to workpiece cooling

This research gauges the ability of deep reinforcement learning (DRL) techniques to assist the control of conjugate heat transfer systems governed by the coupled Navier--Stokes and heat equations. It uses a novel, "degenerate" version of the proximal policy optimization (PPO) algorithm, intended for situations where the optimal policy to be learnt by a neural network does not depend on state, as is notably the case in optimization and open-loop control problems. The numerical reward fed to the neural network is computed with an in-house stabilized finite elements environment combining variational multi-scale (VMS) modeling of the governing equations, immerse volume method, and multi-component anisotropic mesh adaptation. Several test cases of natural and forced convection in two and three dimensions are used as testbed for developing the methodology. The approach successfully alleviates the natural convection induced enhancement of heat transfer in a two-dimensional, differentially heated square cavity controlled by piece-wise constant fluctuations of the sidewall temperature. It also proves capable of improving the homogeneity of temperature across the surface of two and three-dimensional hot workpieces under impingement cooling. Various cases are tackled, in which the position of multiple cold air injectors is optimized relative to a fixed workpiece position. The flexibility of the numerical framework makes it tractable to solve also the inverse problem, i.e., to optimize the workpiece position relative to a fixed injector distribution. The obtained results showcase the potential of the method for black-box optimization of practically meaningful computational fluid dynamics (CFD) conjugate heat transfer systems.

physics.flu-dyn↗

arXiv2021

A review on Deep Reinforcement Learning for Fluid Mechanics

Deep reinforcement learning (DRL) has recently been adopted in a wide range of physics and engineering domains for its ability to solve decision-making problems that were previously out of reach due to a combination of non-linearity and high dimensionality. In the last few years, it has spread in the field of computational mechanics, and particularly in fluid dynamics, with recent applications in flow control and shape optimization. In this work, we conduct a detailed review of existing DRL applications to fluid mechanics problems. In addition, we present recent results that further illustrate the potential of DRL in Fluid Mechanics. The coupling methods used in each case are covered, detailing their advantages and limitations. Our review also focuses on the comparison with classical methods for optimal control and optimization. Finally, several test cases are described that illustrate recent progress made in this field. The goal of this publication is to provide an understanding of DRL capabilities along with state-of-the-art applications in fluid dynamics to researchers wishing to address new problems with these methods.

physics.comp-ph↗

arXiv2021

Folding instabilities in non-Newtonian viscous sheets: shear thinning and shear thickening effects

In this work, we extend the analyses devoted to Newtonian viscous fluids previously reported by Ribe [Physical Review E 68, 036305 (2003)], by investigating shear thickening (dilatant) and shear thinning (pseudoplastic) effects on the development of folding instabilities in non-Newtonian viscous sheets of which viscosity is given by a power-law constitutive equation. Such instabilities are trigged by compression stresses acting on viscous sheets that leave a channel at a very small initial velocity, fall, and then hit a solid surface or a fluid substrate. Our study is conducted through a mixed approach combining direct numerical simulations, energy budget analyses, scaling laws, and experiments. The numerical results are based on an adaptive variational multi-scale method for multiphase flows, while Carpobol gel sheets are considered for the conducted experiments. Two folding regimes are observed: (1) the viscous regime; and (2) the gravitational one. Interestingly, only the latter is affected by shear thinning/thickening manifestations within the material. In short, when gravity is balanced by viscous forces along the non-Newtonian viscous sheet, both the folding amplitude and the folding frequency are given by a power-law function of the sheet slenderness, the Galileo number (the ratio of the gravitational stress to the viscous one), and the flow behaviour index. Highly shear thickening materials develop large amplitude (and low frequency) instabilities, which, in contrast, tend to be suppressed by shear thinning effects, and eventually cease. Lastly, nonNewtonian effects on folding onset/cessation are also carefully explored. As a result, non-Newtonian folding onset and cessation criteria are presented.

physics.flu-dyn↗

arXiv2020

Direct shape optimization through deep reinforcement learning

Deep Reinforcement Learning (DRL) has recently spread into a range of domains within physics and engineering, with multiple remarkable achievements. Still, much remains to be explored before the capabilities of these methods are well understood. In this paper, we present the first application of DRL to direct shape optimization. We show that, given adequate reward, an artificial neural network trained through DRL is able to generate optimal shapes on its own, without any prior knowledge and in a constrained time. While we choose here to apply this methodology to aerodynamics, the optimization process itself is agnostic to details of the use case, and thus our work paves the way to new generic shape optimization strategies both in fluid mechanics, and more generally in any domain where a relevant reward function can be defined.

cs.CE↗

arXiv2020

Analysis and comparisons of various models in cold spray simulations : towards high fidelity simulations

Cold spray technology is a quickly growing manufacturing technology which impacts lots of industries. Despite many years of studies about the comprehension of the phenomena and the improvements of the performance of the system, ensuring high fidelity simulations remains a challenge. We propose in this work a detailed high fidelity modeling and simulations giving more insight of the phenomena appearing in cold spray such as turbulence, oblique shocks, bow shocks, fluctuations, particles motion and particles impacts. It is mainly based on a richer model known as the Detached Eddy Simulation (DES) model. Moreover, we present several analysis of various existing models for both validations and comparisons purposes. Finally, this high fidelity framework will allow us to deal with a new configuration showing an improved performance assessed with the previous models.

physics.flu-dyn↗

arXiv2020

A supervised neural network for drag prediction of arbitrary 2D shapes in low Reynolds number flows

Despite the significant breakthrough of neural networks in the last few years, their spreading in the field of computational fluid dynamics is very recent, and many applications remain to explore. In this paper, we explore the drag prediction capabilities of convolutional neural networks for laminar, low-Reynolds number flows past arbitrary 2D shapes. A set of random shapes exhibiting a rich variety of geometrical features is built using Bézier curves. The efficient labelling of the shapes is provided using an immersed method to solve a unified Eulerian formulation of the Navier-Stokes equation. The network is then trained and optimized on the obtained dataset, and its predictive efficiency assessed on several real-life shapes, including NACA airfoils.

physics.comp-ph↗

arXiv2019

U-net architectures for fast prediction of incompressible laminar flows

Machine learning is a popular tool that is being applied to many domains, from computer vision to natural language processing. It is not long ago that its use was extended to physics, but its capabilities remain to be accurately contoured. In this paper, we are interested in the prediction of 2D velocity and pressure fields around arbitrary shapes in laminar flows using supervised neural networks. To this end, a dataset composed of random shapes is built using Bezier curves, each shape being labeled with its pressure and velocity fields by solving Navier-Stokes equations using a CFD solver. Then, several U-net architectures are trained on the latter dataset, and their predictive efficiency is assessed on unseen shapes, using ad hoc error functions.

physics.comp-ph↗

arXiv2019

Exploiting locality and physical invariants to design effective Deep Reinforcement Learning control of the unstable falling liquid film

Instabilities arise in a number of flow configurations. One such manifestation is the development of interfacial waves in multiphase flows, such as those observed in the falling liquid film problem. Controlling the development of such instabilities is a problem of both academic and industrial interest. However, this has proven challenging in most cases due to the strong nonlinearity and high dimensionality of the underlying equations. In the present work, we successfully apply Deep Reinforcement Learning (DRL) for the control of the one-dimensional (1D) depth-integrated falling liquid film. In addition, we introduce for the first time translational invariance in the architecture of the DRL agent, and we exploit locality of the control problem to define a dense reward function. This allows to both speed up learning considerably, and to easily control an arbitrary large number of jets and overcome the curse of dimensionality on the control output size that would take place using a naive approach. This illustrates the importance of the architecture of the agent for successful DRL control, and we believe this will be an important element in the effective application of DRL to large two-dimensional (2D) or three-dimensional (3D) systems featuring translational, axisymmetric or other invariants.

physics.flu-dyn↗

arXiv2017

Inelastic accretion of inertial particles by a towed sphere

The problem of accretion of small particles by a sphere embedded in a mean flow is studied in the case where the particles undergo inelastic collisions with the solid object. The collision efficiency, which gives the flux of particles experiencing at least one bounce on the sphere, is found to depend upon the sphere Reynolds number only through the value of the critical Stokes number below which no collision occurs. In the absence of molecular diffusion, it is demonstrated that multiple bounces do not provide enough energy dissipation for the particles to stick to the surface within a finite time. This excludes the possibility of any kind of inelastic collapse, so that determining an accretion efficiency requires modelling more precisely particle-surface microphysical interactions. A straightforward choice is to assume that the particles stick when their kinetic energy at impact is below a threshold. In this view, numerical simulations are performed in order to describe the statistics of impact velocities at various values of the Reynolds number. Successive bounces are shown to enhance accretion. These results are put together in order to provide a general qualitative picture on how the accretion efficiency depends upon the non dimensional parameters of the problem.

physics.flu-dyn↗