1,721,176 research outputs found
Inferring turbulent environments via machine learning
No full textThe problem of classifying turbulent environments from partial observation is key for some theoretical and applied fields, from engineering to earth observation and astrophysics, e.g., to precondition searching of optimal control policies in different turbulent backgrounds, to predict the probability of rare events and/or to infer physical parameters labeling different turbulent setups. To achieve such goal one can use different tools depending on the system’s knowledge and on the quality and quantity of the accessible data. In this context, we assume to work in a model-free setup completely blind to all dynamical laws, but with a large quantity of (good quality) data for training. As a prototype of complex flows with different attractors, and different multi-scale statistical properties we selected 10 turbulent ‘ensembles’ by changing the rotation frequency of the frame of reference of the 3d domain and we suppose to have access to a set of partial observations limited to the instantaneous kinetic energy distribution in a 2d plane, as it is often the case in geophysics and astrophysics. We compare results obtained by a machine learning (ML) approach consisting of a state-of-the-art deep convolutional neural network (DCNN) against Bayesian inference which exploits the information on velocity and entropy moments. First, we discuss the supremacy of the ML approach, presenting also results at changing the number of training data and of the hyper-parameters. Second, we present an ablation study on the input data aimed to perform a ranking on the importance of the flow features used by the DCNN, helping to identify the main physical contents used by the classifier. Finally, we discuss the main limitations of such data-driven methods and potential interesting applications
Role of helicity for large- and small-scale turbulent fluctuations
No full textThe effects of the helicity on the dynamics of turbulent flows are investigated. The aim is to disentangle the role of helicity in fixing the direction, the intensity, and the fluctuations of the energy transfer across the inertial range of scales. We introduce an external parameter α that controls the mismatch between the number of positive and negative helically polarized Fourier modes. We present direct numerical simulations of Navier-Stokes equations from the fully symmetrical case, α=0, to the fully asymmetrical case, α=1, when only helical modes of one sign survive. We found a singular dependency of the direction of the energy cascade on α, measuring a positive forward flux as soon as only a few modes with different helical polarities are present. Small-scale fluctuations are also strongly sensitive to the degree of mode reduction, leading to a vanishing intermittency already for values of α∼0.1. If the analysis is restricted to sets of modes with the same helicity sign, intermittency is vanishing for the modes belonging to the minority set, and it is close to that measured on the original Navier-Stokes equations for the other set
Wet to dry self-transitions in dense emulsions: From order to disorder and back
No full textOne of the most distinctive hallmarks of many-body systems far from equilibrium is the spontaneous emergence of order under conditions where disorder would be plausibly expected. Here, we report on the self-transition between ordered and disordered emulsions in divergent microfluidic channels, i.e., from monodisperse assemblies to heterogeneous polydisperse foamlike structures, and back again to ordered ones. The transition is driven by the nonlinear competition between viscous dissipation and surface tension forces as controlled by the device geometry, particularly the aperture angle of the divergent microfluidic channel. An unexpected route back to order is observed in the regime of large opening angles, where a trend towards increasing disorder would be intuitively expected
Multiparticle collision dynamics for fluid interfaces with near-contact interactions
No full textWe present an extension of the multiparticle collision dynamics method for flows with complex interfaces, including supramolecular near-contact interactions mimicking the effect of surfactants. The new method is demonstrated for the case of (i) short range repulsion of droplets in close contact, (ii) arrested phase separation, and (iii) different pattern formation during spinodal decomposition of binary mixtures
Concentrated phase emulsion with multicore morphology under shear: A numerical study
No full textWe numerically study the dynamic behavior under a symmetric shear flow of selected examples of concentrated phase emulsions with multicore morphology confined within a microfluidic channel. A variety of nonequilibrium steady states is reported. Under low shear rates, the emulsion is found to exhibit a solidlike behavior, in which cores display a periodic planetarylike motion with approximately equal angular velocity. At higher shear rates, two steady states emerge, one in which all inner cores align along the flow and become essentially motionless and a further one in which some cores accumulate near the outer interface and produce a dynamical elliptical-shaped ring chain, reminiscent of a treadmillinglike structure, while others occupy the center of the emulsion. A quantitative description in terms of the (i) motion of the cores, (ii) rate of deformation of the emulsion, and (iii) structure of the fluid flow within the channel is also provided
Mesoscale modelling of droplets’ self-assembly in microfluidic channels
No full textA recently proposed mesoscale approach for the simulation of multicomponent flows with near-contact interactions is employed to investigate the early stage formation and clustering statistics of soft flowing crystals in microfluidic channels. Specifically, we first demonstrate the ability of the aforementioned mesoscale model to accurately reproduce main mechanisms leading to the formation of two basic droplet patterns (triangular and hexagonal), in close agreement with experimental evidence. Next, we quantitatively evaluate the device-scale clustering efficiency of the crystal formation process by introducing a new orientational order parameter, based on the Delaunay triangulation and Voronoi diagrams analysis of the droplet patterns. The mesoscale computational approach employed in this work proves to be an efficient tool to shed new light on the complex dynamics of dense emulsions, from short-scale thin-film hydrodynamics, all the way up to global structure formation and statistics of the resulting droplets ensembles
A Multiresolution Mesoscale Approach for Microscale Hydrodynamics
No full textA new class of multiscale scheme is presented for micro-hydrodynamic problems based on a dual representation of the fluid observables. The hybrid model is first tested against the classical flow between two parallel plates and then applied to a plug flow within a micrometer-sized striction and a shear flow within a microcavity. Both cases demonstrate the capability of the multiscale approach to reproduce the correct macroscopic hydrodynamics also in the presence of refined grids (one and two levels), while retaining the correct thermal fluctuations, embedded in the multiparticle collision method. This provides the first step toward a novel class of fully mesoscale hybrid approaches able to capture the physics of fluids at the micro- and nanoscales whenever a continuum representation of the fluid falls short of providing the complete physical information, due to a lack of resolution and thermal fluctuations
Reconstruction of turbulent data with deep generative models for semantic inpainting from TURB-Rot database
No full textWe study the applicability of tools developed by the computer vision community for feature learning and semantic image inpainting to perform data reconstruction of fluid turbulence configurations. The aim is twofold. First, we explore on a quantitative basis the capability of convolutional neural networks embedded in a deep generative adversarial model (deep-GAN) to generate missing data in turbulence, a paradigmatic high dimensional chaotic system. In particular, we investigate their use in reconstructing two-dimensional damaged snapshots extracted from a large database of numerical configurations of three-dimensional turbulence in the presence of rotation, a case with multiscale random features where both large-scale organized structures and small-scale highly intermittent and non-Gaussian fluctuations are present. Second, following a reverse engineering approach, we aim to rank the input flow properties (features) in terms of their qualitative and quantitative importance to obtain a better set of reconstructed fields. We present two approaches both based on context encoders. The first one infers the missing data via a minimization of the L z pixel-wise reconstruction loss, plus a small adversarial penalization. The second, searches for the closest encoding of the corrupted flow configuration from a previously trained generator. Finally, we present a comparison with a different data assimilation tool, based on Nudging, an equation-informed unbiased protocol, well known in the numerical weather prediction community. The TURB-Rot database of roughly 300 K two-dimensional turbulent images is released and details on how to download it are given
Lattice Boltzmann simulations capture the multiscale physics of soft flowing crystals: Multiscale physics of flowing crystals
No full textThe study of the underlying physics of soft flowing materials depends heavily on numerical simulations, due to the complex structure of the governing equations reflecting the competition of concurrent mechanisms acting at widely disparate scales in space and time. A full-scale computational modelling remains a formidable challenge since it amounts to simultaneously handling six or more spatial decades in space and twice as many in time. Coarse-grained methods often provide a viable strategy to significantly mitigate this issue, through the implementation of mesoscale supramolecular forces designed to capture the essential physics at a fraction of the computational cost of a full-detail description. Here, we review some recent advances in the design of a lattice Boltzmann mesoscale approach for soft flowing materials, inclusive of near-contact interactions (NCIs) between dynamic interfaces, as they occur in high packing-fraction soft flowing crystals. The method proves capable of capturing several aspects of the rheology of soft flowing crystals, namely, (i) a 3/2 power-law dependence of the dispersed phase flow rate on the applied pressure gradient, (ii) the structural transition between an ex-two and ex-one (bamboo) configurations with the associated drop of the flow rate, (iii) the onset of interfacial waves once NCI is sufficiently intense. This article is part of the theme issue 'Fluid dynamics, soft matter and complex systems: recent results and new methods'
Shear dynamics of polydisperse double emulsions
No full textWe numerically study the dynamics of a polydisperse double emulsion under a symmetric shear flow. We show that both dispersity and shear rate crucially affect the behavior of the innermost drops and of the surrounding shell. While at low/moderate values of shear rates, the inner drops rotate periodically around a common center of mass triggered by the fluid vortex formed within the emulsion generally regardless of their polydispersity; at higher values, such dynamics occurs only at increasing polydispersity, since monodisperse drops are found to align along the shear flow and become approximately motionless at late times. Our simulations also suggest that increasing polydispersity favors close-range contacts among cores and persistent collisions, while hindering shape deformations of the external droplet. A quantitative evaluation of these effects is also provided
- …
