Data-driven approaches offer novel opportunities for improving the performance of turbulent flow simulations, which are critical to wide-ranging applications from wind farms and aerodynamic designs to weather and climate forecasting. While conventional continuum Navier-Stokes solvers have been the subject of a significant amount of work in this domain, there has hitherto been very limited effort in the same direction for the more scalable and highly performant lattice Boltzmann method (LBM), even though it has been successfully applied to a variety of turbulent flow simulations using large-eddy simulation (LES) techniques. In this work, we establish a data-driven framework for the LES-based lattice Boltzmann simulation of near-wall turbulent flow fields. We do this by training neural networks using improved delayed detached eddy simulation data. Crucially, this is done in combination with physics-based information that substantially constrains the data-driven predictions. Using data from turbulent channel flow at a friction Reynolds number at $5200$, our simulations accurately predict the behaviour of the wall model at arbitrary friction Reynolds numbers up to $1.0 \times 10^6$. In contradistinction with other models that use direct numerical simulation datasets, our physics-informed model requires data from very limited regions within the wall-bounded plane, reducing by three orders of magnitude the quantity of data needed for training. We also demonstrate that our model can handle data configurations when the near-wall grid is sparse. Our physics-informed neural network approach opens up the possibility of employing LBM in combination with highly specific and therefore much more limited quantities of macroscopic data, substantially facilitating the investigation of a wide-range of turbulent flow applications at very large scale.
Lattice Boltzmann Method (LBM) simulations for turbulent flows over a fractal and non-fractal obstacles are presented. The wake hydrodynamics are compared and discussed in terms of flow relaxation, Strouhal numbers and wake length for different Reynolds numbers. Three obstacle topologies are studied, Solid (SS), Porous Regular (PR) and Porous Fractal (FR). In particular we observe that the oscillation present in the case of the solid square can be annihilated or only pushed downstream depending on the topology ot the porous obstacle. The Lattice Boltzmann Method (LBM) is implemented over a ran…
We develop an improved lattice-Boltzmann numerical scheme to solve magnetohydrodynamic (MHD) equations in the regime of low magnetic Reynolds numbers, grounded on a manifestly Galilean covariant modeling of the Navier-Stokes equations. The simulation of the magnetic induction equation within the lattice-Boltzmann approach to MHD has been usually devised along the lines of the simplest phenomenological description, the single relaxation time (SRT) model. In order to deal with well-known stability difficulties of the SRT framework, we introduce, alternatively, a multi-relaxation-time technique f…
Background and objective: Contrary to flows in small intracranial vessels, many blood flow configurations such as those found in aortic vessels and aneurysms involve larger Reynolds numbers and, therefore, transitional or turbulent conditions. Dealing with such systems require both robust and efficient numerical methods. Methods: We assess here the performance of a lattice Boltzmann solver with full Hermite expansion of the equilibrium and central Hermite moments collision operator at higher Reynolds numbers, especially for under-resolved simulations. To that end the food and drug administrati…
Here we introduce new coarse-grained variables for a turbulent flow in the form of moments of its Reynolds-averaged Boltzmann equation. With the exception of the collision moments, the transport equations for the new variables are identical to the usual moment equations, and thus naturally lend themselves to the variety of already existing closure methods. Under the anelastic turbulence approximation, we derive equations for the Reynolds-averaged turbulent fluctuations around the coarse-grained state. We show that the global relative entropy of the coarse-grained state is bounded from above by…
Turbulence modeling is a critical component in numerical simulations of industrial flows based on Reynolds-averaged Navier-Stokes (RANS) equations. However, after decades of efforts in the turbulence modeling community, universally applicable RANS models with predictive capabilities are still lacking. Recently, data-driven methods have been proposed as a promising alternative to the traditional approaches of turbulence model development. In this work we propose a data-driven, physics-informed machine learning approach for predicting discrepancies in RANS modeled Reynolds stresses. The discrepa…
This paper reports the investigations done to adapt the Characteristic Boundary Conditions (CBC) to the Lattice-Boltzmann formalism for high Reynolds number applications. Three CBC formalisms are implemented and tested in an open source LBM code: the baseline local one-dimension inviscid (BL-LODI) approach, its extension including the effects of the transverse terms (CBC-2D) and a local streamline approach in which the problem is reformulated in the incident wave framework (LS-LODI). Then all implementations of the CBC methods are tested for a variety of test cases, ranging from canonical prob…
The lattice Boltzmann method, now widely used for a variety of applications, has also been extended to model multi-phase flows through different formulations. While already applied to many different configurations in the low Weber and Reynolds number regimes, applications to higher Weber/Reynolds numbers or larger density/viscosity ratios are still the topic of active research. In this study, through a combination of the decoupled phase-field formulation -- conservative Allen-Cahn equation -- and a cumulants-based collision operator for a low-Mach pressure-based flow solver, we present an algo…
We present a regularized version of the color gradient lattice Boltzmann (LB) scheme for the simulation of droplet formation in microfluidic devices of experimental relevance. The regularized version is shown to provide computationally efficient access to Capillary number regimes relevant to droplet generation via microfluidic devices, such as flow-focusers and the more recent microfluidic step emulsifier devices.
Reynolds-averaged Navier-Stokes (RANS) equations are widely used in engineering turbulent flow simulations. However, RANS predictions may have large discrepancies due to the uncertainties in modeled Reynolds stresses. Recently, Wang et al. demonstrated that machine learning can be used to improve the RANS modeled Reynolds stresses by leveraging data from high fidelity simulations (Physics informed machine learning approach for reconstructing Reynolds stress modeling discrepancies based on DNS data. Physical Review Fluids. 2, 034603, 2017). However, solving for mean flows from the improved Reyn…
Although an increased availability of computational resources has enabled high-fidelity simulations of turbulent flows, the RANS models are still the dominant tools for industrial applications. However, the predictive capabilities of RANS models are limited by potential inaccuracy driven by hypotheses in the Reynolds stress closure. Recently, a Physics-Informed Machine Learning (PIML) approach has been proposed to learn the functional form of Reynolds stress discrepancy in RANS simulations based on available data. It has been demonstrated that the learned discrepancy function can be used to im…
We simulate the two-dimensional fluid flow around National Advisory Committee for Aeronautics (NACA) 0012 airfoil using a hybrid lattice Boltzmann method (HLBM), which combines the standard lattice Boltzmann method with an unstructured finite-volume formulation. The aim of the study is to assess the numerical performances and the robustness of the computational method. To this purpose, after providing a convergence study to estimate the overall accuracy of the method, we analyze the numerical solution for different values of the angle of attack at a Reynolds number equal to $10^3$. Subsequentl…
Despite their well-known limitations, RANS models remain the most commonly employed tool for modeling turbulent flows in engineering practice. RANS models are predicated on the solution of the RANS equations, but these equations involve an unclosed term, the Reynolds stress tensor, which must be modeled. The Reynolds stress tensor is often modeled as an algebraic function of mean flow field variables and turbulence variables. This introduces a discrepancy between the Reynolds stress tensor predicted by the model and the exact Reynolds stress tensor. This discrepancy can result in inaccurate me…
The lattice Boltzmann method has been successfully applied for the simulation of flow through porous media in the creeping regime. Its technical properties, namely discretization, straightforward implementation and parallelization, are responsible for its popularity. However, flow through porous media is not restricted to near zero Reynolds numbers since inertial effects play a role in numerous natural and industrial processes. In this paper we investigate the capability of the lattice Boltzmann method to correctly describe flow in porous media at moderate Reynolds numbers. The selection of th…
We revisit force evaluation methodologies on rigid solid particles suspended in a viscous fluid and simulated via lattice Boltzmann method (LBM). We point out the non-commutativity of streaming and collision operators in the force evaluation procedure and provide a theoretical explanation for this observation. Based on this analysis, we propose a discrete force calculation scheme with enhanced accuracy. The proposed scheme is essentially a lattice version of the Reynolds Transport Theorem (RTT) in the context of the lattice Boltzmann formulation. Besides maintaining satisfactory levels of reli…
The present work investigates the application of Artificial Neural Networks (ANNs) to estimate the Reynolds ($Re$) number for flows around a cylinder. The data required to train the ANN was generated with our own implementation of a Lattice Boltzmann Method (LBM) code performing simulations of a 2-dimensional flow around a cylinder. As results of the simulations, we obtain the velocity field ($\vec{v}$) and the vorticity ($\vec{\nabla}\times\vec{v}$) of the fluid for 120 different values of $Re$ measured at different distances from the obstacle and use them to teach the ANN to predict the $Re$…
Lattice Boltzmann method models offer a novel framework for the simulation of high Reynolds number dilute gravity currents. The numerical algorithm is well suited to acceleration via implementation on massively parallel computer architectures. Here we present two lattice Boltzmann method models of lock-exchange dilute gravity currents, in which the largest turbulent length scales are directly resolved. The three-dimensional simulations are accelerated by exporting computations to a graphics processing unit and are validated against experiments and high-resolution simulations for Reynolds numbe…
University of Notre Dame（圣母大学）；Missouri University of Science and Technology（密苏里科技大学）；Virginia Tech（弗吉尼亚理工大学）
Modeled Reynolds stress is a major source of model-form uncertainties in Reynolds-averaged Navier-Stokes (RANS) simulations. Recently, a physics-informed machine-learning (PIML) approach has been proposed for reconstructing the discrepancies in RANS-modeled Reynolds stresses. The merits of the PIML framework has been demonstrated in several canonical incompressible flows. However, its performance on high-Mach-number flows is still not clear. In this work we use the PIML approach to predict the discrepancies in RANS modeled Reynolds stresses in high-Mach-number flat-plate turbulent boundary lay…
Modeling blood flow in larger vessels using lattice-Boltzmann methods comes with a challenging set of constraints: a complex geometry with walls and inlet/outlets at arbitrary orientations with respect to the lattice, intermediate Reynolds number, and unsteady flow. Simple bounce-back is one of the most commonly used, simplest, and most computationally efficient boundary conditions, but many others have been proposed. We implement three other methods applicable to complex geometries (Guo, Zheng and Shi, Phys Fluids (2002); Bouzdi, Firdaouss and Lallemand, Phys. Fluids (2001); Junk and Yang Phy…
We numerically analyse the rotation of a neutrally buoyant spheroid in a shear flow at small shear Reynolds number. Using direct numerical stability analysis of the coupled nonlinear particle-flow problem we compute the linear stability of the log-rolling orbit at small shear Reynolds number, ${\rm Re}_a$. As ${\rm Re}_a \to 0$ and as the box size of the system tends to infinity we find good agreement between the numerical results and earlier analytical predictions valid to linear order in ${\rm Re}_a$ for the case of an unbounded shear. The numerical stability analysis indicates that there ar…
University of Tehran（德黑兰大学）；KTH Royal Institute of Technology（瑞典皇家理工学院）
Physics-informed neural networks (PINNs) are successful machine-learning methods for the solution and identification of partial differential equations (PDEs). We employ PINNs for solving the Reynolds-averaged Navier$\unicode{x2013}$Stokes (RANS) equations for incompressible turbulent flows without any specific model or assumption for turbulence, and by taking only the data on the domain boundaries. We first show the applicability of PINNs for solving the Navier$\unicode{x2013}$Stokes equations for laminar flows by solving the Falkner$\unicode{x2013}$Skan boundary layer. We then apply PINNs for…
We present three dimensional realizations of the model introduced recently by (Karlin, Bösch, Chikatamarla, Phys. Rev. E 2014) and review the role of the entropic stabilizer. The presented models achieve outstanding numerical stability in presence of turbulent high Reynolds number flows. We report accurate results for low order moments for homogeneous isotropic decaying turbulence and second order grid convergence for most assessed statistical quantities. The explicit and efficient nature of the scheme renders it a very promising candidate for both engineering and scientific purposes in the vi…
We develop a lattice Boltzmann (LB) model for immiscible two-phase flow simulations with central moments (CMs). This successfully combines a three-dimensional nonorthogonal CM-based LB scheme [A. De Rosis, Phys. Rev. E 95, 013310 (2017)] with our previous color-gradient LB model [S. Saito, Y. Abe, and K. Koyama, Phys. Rev. E 96, 013317 (2017)]. Hydrodynamic melt-jet breakup simulations show that the proposed model is significantly more stable, even for flow with extremely high Reynolds numbers, up to $O(10^6)$. This enables us to investigate the phenomena expected under actual reactor conditio…
Flows through medical devices as well as in anatomical vessels despite being at moderate Reynolds number may exhibit transitional or even turbulent character. In order to validate numerical methods and codes used for biomedical flow computations, the U.S. food and drug administration (FDA) established an experimental benchmark, which was a pipe with gradual contraction and sudden expansion representing a nozzle. The experimental results for various Reynolds numbers ranging from 500 to 6500 were publicly released. Previous and recent computational investigations of flow in the FDA nozzle found …
In Denicol et al., Phys. Rev. D 85, 114047 (2012), the equations of motion of relativistic dissipative fluid dynamics were derived from the relativistic Boltzmann equation. These equations contain a multitude of terms of second order in Knudsen number, in inverse Reynolds number, or their product. Terms of second order in Knudsen number give rise to non-hyperbolic (and thus acausal) behavior and must be neglected in (numerical) solutions of relativistic dissipative fluid dynamics. The coefficients of the terms which are of the order of the product of Knudsen and inverse Reynolds numbers have b…
This paper deals with the investigation of flow field due to clap and fling mechanism using immersed boundary coupled with lattice Boltzmann method. The lattice Boltzmann method (LBM), an alternative to Navier Stokes solver, is used because of its simplicity and computational efficiency in solving complex moving boundary problems. Benchmark problems are simulated to validate the code, which is then used for simulating flow over two elliptic wing of aspect ratio 5 performing clap and fling flapping motion for different flow parameters such as Reynolds number (Re = 75,100,150), advance ratio (J …
This work presents a review and perspectives on recent developments in the use of machine learning (ML) to augment Reynolds-averaged Navier--Stokes (RANS) and Large Eddy Simulation (LES) models of turbulent flows. Different approaches of applying supervised learning to represent unclosed terms, model discrepancies and sub-filter scales are discussed in the context of RANS and LES modeling. Particular emphasis is placed on the impact of the training procedure on the consistency of ML augmentations with the underlying physical model. Techniques to promote model-consistent training, and to avoid …
A recently introduced family of lattice Boltzmann (LB) models (Karlin, Bösch, Chikatamarla, Phys. Rev. E, 2014) is studied in detail for incompressible two-dimensional flows. A framework for developing LB models based on entropy considerations is laid out extensively. Second order rate of convergence is numerically confirmed and it is demonstrated that these entropy based models recover the Navier-Stokes solution in the hydrodynamic limit. Comparison with the standard Bhatnagar-Gross-Krook (LBGK) and the entropic lattice Boltzmann method (ELBM) demonstrates the superior stability and accuracy …
In the lattice Boltzmann method (LBM), the widely utilized wall boundary is the bounce-back (BB) boundary, which corresponds to the no-slip boundary. The BB boundary prevents the LBM from capturing the accurate shear drag on the wall when addressing high Reynolds number flows using coarse-grid systems. In this study, we proposed a "wall-function bounce (WFB)" boundary that incorporates a wall function into the LBM's boundary condition and overcomes the limitation of the BB. The WFB boundary calculates the appropriate shear drag on the wall using a wall function model, and thereafter modifies d…
A double-distribution-function based lattice Boltzmann method (DDF-LBM) is proposed for the simulation of polyatomic gases in the supersonic regime. The model relies on an extended equilibrium state that is constructed to reproduce the first 13 moments of the Maxwell-Boltzmann distribution exactly. This extends the validity of the standard 5-constraint (mass, momentum and energy) approach and leads to the correct simulation of thermal, compressible flows with only 39 discrete velocities in 3D. The stability of this BGK-LBM is reinforced by relying on Knudsen-number-dependent relaxation times t…
In the present paper a lattice Boltzmann scheme is presented which exhibits an increased stability and accuracy with respect to standard single- or multi-relaxation-time (MRT) approaches. The scheme is based on a single-relaxation-time model where a special regularization procedure is applied. This regularization is based on the fact that, for a-thermal flows, there exists a recursive way to express the velocity distribution function at any order (in the Hermite series sense) in terms of the density, velocity, and stress tensor. A linear stability analysis is conducted which shows enhanced dis…
We study the deformation and motion of an erythrocyte in fluid flows via a lattice Boltzmann method. To this purpose, the bending rigidity and the elastic modulus of isotropic dilation are introduced and incorporated with the lattice Boltzmann simulation, and the membrane-flow interactions on both sides of the membrane are carefully examined. We find that the static biconcave shape of an erythrocyte is quite stable and can effectively resist the pathological changes on their membrane. Further, our simulation results show that in shear flow, the erythrocyte will be highly flattened and undergo …
In this paper, an efficient three-dimensional lattice Boltzmann (LB) model with multiple-relaxation-time (MRT) collision operator is developed for the simulation of multiphase flows. This model is an extension of our previous two-dimensional model (H. Liang, B. C. Shi, Z. L. Guo, and Z. H. Chai, Phys. Rev. E. 89, 053320 (2014)) to the three dimensions using the D3Q7 (seven discrete velocities in three dimensions) lattice for the Chan-Hilliard equation (CHE) and the D3Q15 lattice for the Navier-Stokes equations (NSEs). Due to the smaller lattice-velocity numbers used, the computional efficiency…
Harvard University（哈佛大学）；Imperial College London（帝国理工学院）；Dassault Systemes, Simulia Corp.（达索系统Simulia公司）
The Reynolds-averaged Navier-Stokes (RANS) equations require accurate modeling of the anisotropic Reynolds stress tensor. Traditional closure models, while sophisticated, often only apply to restricted flow configurations. Researchers have started using machine learning approaches to tackle this problem by developing more general closure models informed by data. In this work we build upon recent convolutional neural network architectures used for turbulence modeling and propose a multi-task learning-based fully convolutional neural network that is able to accurately predict the normalized anis…
Most Computational Fluid Dynamics (CFD) studies of hemodynamics in intracranial aneurysms are based on the assumption of laminar flow due to a relatively low (below 500) parent artery Reynolds number. A few studies have recently demonstrated the occurrence of transitional flow in aneurysms, but these studies employed special finite element schemes tailored to capture transitional nature of flow. In this study we investigate the occurrence of transition using a standard Lattice Boltzmann method (LBM). The LBM is used because of its computational efficiency, which in the present study allowed us…
Low order Carleman simulation of steady state flows at mild Reynolds numbers
Fondazione Istituto Italiano di Tecnologia（意大利理工大学基金会）；Center for Life Nano-Neuroscience at la Sapienza（罗马一大生命纳米神经科学中心）；Istituto per le Applicazioni del Calcolo “Mauro Picone（毛罗·皮科内计算应用研究所）；Italian National Research Council（意大利国家研究委员会）
We present a new lattice kinetic method to simulate fluid dynamics in curvilinear geometries. A suitable discrete Boltzmann equation is solved in contravariant coordinates, and the equilibrium distribution function is obtained by a Hermite polynomials expansion of the Maxwell-Boltzmann distribution, expressed in terms of the contravariant coordinates and the metric tensor. To validate the model, we calculate the critical Reynolds number for the onset of the Taylor-Couette instability between two concentric cylinders, obtaining excellent agreement with the theory. In order to extend this study …
Entropic lattice Boltzmann methods have been developed to alleviate intrinsic stability issues of lattice Boltzmann models for under-resolved simulations. Its reliability in combination with moving objects was established for various laminar benchmark flows in two dimensions in our previous work Dorschner et al. [11] as well as for three dimensional one-way coupled simulations of engine-type geometries in Dorschner et al. [12] for flat moving walls. The present contribution aims to fully exploit the advantages of entropic lattice Boltzmann models in terms of stability and accuracy and extends …
We propose a single-step simplified lattice Boltzmann algorithm capable of performing magnetohydrodynamic (MHD) flow simulations in pipes for very small values of magnetic Reynolds numbers $R_m$. In some previous works, most lattice Boltzmann simulations are performed with values of $R_m$ close to the Reynolds numbers for flows in simplified rectangular geometries. One of the reasons is the limitation of some traditional lattice Boltzmann algorithms in dealing with situations involving very small magnetic diffusion time scales associated with most industrial applications in MHD, which require …
Vortex-induced vibration (VIV) exists widely in natural and industrial fields. The main approaches for solving VIV problems are numerical simulations and experimental methods. However, experiment methods are difficult to obtain the whole flow field information and also high-cost while numerical simulation is extraordinary time-consuming and limited in low Reynolds number and simple geometric configuration. In addition, numerical simulations are difficult to handle the moving mesh technique. In this paper, physics informed neural network (PINN) is proposed to solve the VIV and wake-induced vibr…
