We present the construction of an original stochastic model for the instantaneous turbulent kinetic energy at a given point of a flow, and we validate estimator methods on this model with observational data examples. Motivated by the need for wind energy industry of acquiring relevant statistical information of air motion at a local place, we adopt the Lagrangian description of fluid flows to derive, from the $3$D+time equations of the physics, a $0$D+time-stochastic model for the time series of the instantaneous turbulent kinetic energy at a given position. Specifically, based on the Lagrangian stochastic description of a generic fluid-particles, we derive a family of mean-field dynamics featuring the square norm of the turbulent velocity. By approximating at equilibrium the characteristic nonlinear terms of the dynamics, we recover the so called Cox-Ingersoll-Ross process, which was previously suggested in the literature for modelling wind speed. We then propose a calibration procedure for the parameters employing both direct methods and Bayesian inference. In particular, we show the consistency of the estimators and validate the model through the quantification of uncertainty, with respect to the range of values given in the literature for some physical constants of turbulence modelling.
The high cost of high-resolution computational fluid/flame dynamics (CFD) has hindered its application in combustion related design, research and optimization. In this study, we propose a new framework for turbulent combustion simulation based on the deep learning approach. An optimized deep convolutional neural network (CNN) inspired from a U-Net architecture and inception module is designed for constructing the framework of the deep learning solver, named CFDNN. CFDNN is then trained on the simulation results of hydrogen combustion in a cavity with different inlet velocities. After training,…
In this paper, investigations are conducted using Reynolds-averaged Navier-Stokes (RANS) turbulence models to investigate the importance of turbulence modelling for nasal inspiration at a constant flow rate of 250 ml/s. Four different, standard turbulence models are tested in a model geometry based on pre-operative CT images of a selected obstructive sleep-apnea syndrome (OSAS) patient. The results show only minor differences between them. Furthermore, the turbulence models do not give significantly different results than a laminar flow model. Thus, the main conclusion is that effects of turbu…
In this work, we propose a model for the orientation of inertialess spheroidal particles suspended in turbulent flows. This model consists in a stochastic version of the Jeffery equation that can be included in a statistical Lagrangian description of particles suspended in a flow. It is compatible and coherent with turbulence models that are widely used in CFD codes for the simulation of the flow field in practical large-scale applications. In this context, we propose and analyze a numerical scheme based on a splitting scheme algorithm that decouples the orientation dynamics into its main cont…
With the growing size and complexity of turbulent flow models, data compression approaches are of the utmost importance to analyze, visualize, or restart the simulations. Recently, in-situ autoencoder-based compression approaches have been proposed and shown to be effective at producing reduced representations of turbulent flow data. However, these approaches focus solely on training the model using point-wise sample reconstruction losses that do not take advantage of the physical properties of turbulent flows. In this paper, we show that training autoencoders with additional physics-informed …
Dust explosions are among the most hazardous accidents affecting industrial facilities processing particulate solids. Describing the severity parameters of dust clouds is critical to the safety management and risk assessment of dust explosions. These parameters are determined experimentally in a 20L spherical vessel, following the ASTM E1226 or UNE 14034 standards. Since their reproducibility depends on the levels of turbulence associated with the dust cloud, a computational model of the multi-phase (gas-solid) flow is used to simulate the dispersion process with the open-source CFD code OpenF…
The growth of computational resources in the past decades has expanded the application of Computational Fluid Dynamics (CFD) from the traditional fields of aerodynamics and hydrodynamics to a number of new areas. Examples range from the heat and fluid flows in nuclear reactor vessels and in data centers to the turbulence flows through wind turbine farms and coastal vegetation plants. However, in these new applications complex structures are often exist (e.g., rod bundles in reactor vessels and turbines in wind farms), which makes fully resolved, first-principle based CFD modeling prohibitively…
In this paper, neural network (NN)-based models are generated to replace flamelet tables for sub-grid modeling in large-eddy simulations of a single-injector liquid-propellant rocket engine. In the most accurate case, separate NNs for each of the flame variables are designed and tested by comparing the NN output values with the corresponding values in the table. The gas constant, internal flame energy, and flame heat capacity ratio are estimated with 0.0506%, 0.0852%, and 0.0778% error, respectively. Flame temperature, thermal conductivity, and the coefficient of heat capacity ratio are estima…
The simulation of the turbulent and potentially separating flow around a rotating, twisted, and tapered airfoil is a challenging task for CFD simulations. This paper describes CFD simulations of the NREL Phase VI turbine that was experimentally characterized in the 24.4m x 36.6m NREL/NASA Ames wind tunnel (Hand et al., 2001). All computations in this article are performed on the experimental base configuration of 0o yaw angle, 3o tip pitch angle, and a rotation rate of 72 rpm. The significance of specific mesh resolution regions to the accuracy of the CFD prediction is discussed. The ability o…
We report measurements of the instantaneous viscous boundary layer (BL) thickness $δ_v(t)$ in turbulent Rayleigh-Bénard convection. It is found that $δ_v(t)$ obtained from the measured instantaneous two-dimensional velocity field exhibits intermittent fluctuations. For small values, $δ_v(t)$ obeys a lognormal distribution, whereas for large values the distribution of $δ_v(t)$ exhibits an exponential tail. The variation of $δ_v(t)$ with time is found to be driven by the fluctuations of the large-scale mean flow velocity and the local horizontal velocities close to the plate can be used as an in…
This paper proposes a simple new closure principle for turbulent shear flows. The turbulent flow field is divided into an outer and an inner region. The inner region is made up of a log-law region and a wall layer. The wall layer is viewed in terms of the well known inrush-sweep-burst sequence observed since 1967. It is modelled as a transient laminar sub-boundary layer, which obeys the Stokes solution for an impulsively started flat plate. The wall layer may also be modelled with a steady state solution by adding a damping function to the log-law. Closure is achieved by matching the unsteady …
A four component decomposition of the local instantaneous velocity is proposed. It brings out more readily the terms in the Navier-Stokes equations associated with different events and fluid structures of turbulent flows than the classic two component decomposition of Reynolds. In particular the new composition highlights the existence of two types of Reynolds stresses: fast and slow. The fast Reynolds stresses can be linked to a streaming process that describes the ejection of wall fluid in the bursting process. It also provides a simple method for modelling the wall layer. The four component…
This paper aims at extending the code verification methodology to high-order accurate scheme implementations on curved wall-bounded domains as well as for realistic turbulent flows such as the flat plate boundary layer modelled by the Reynolds-averaged Navier-Stokes (RANS) equations. Two new manufactured solutions (MSs) are introduced with demonstrated ability to verify the treatment of slip and no-slip boundary conditions in high-order frameworks on curved domains. These MSs serve as well to discuss the impact of the method of computation of boundary normals on the order of accuracy (OOA) of …
There are several approaches to describe flows with particles e.g. Lattice-Gas Automata (LGA), Lattice-Boltzmann method (LBM) or smoothed particle hydrodynamics (SPH). These approaches do not use fixed grids on which the Navier-Stokes equations are solved via e.g. finite volume method. The flow is simulated using a multitude of particles or particle density distributions, which interacts and due to statistical laws and an even more fundamental approach than the Navier-Stokes equation, the averaged flow variables can be derived. After a short summary of the most popular particle methods the new…
In the present work, an attempt is made to map the sensitivity of the existing zero pressure gradient (ZPG) turbulent boundary layer (TBL) wall-pressure spectrum models with different TBL parameters, and eventually, with different Reynolds Averaged Navier Stokes (RANS) turbulence models, simulated in OpenFOAM and ANSYS Fluent solvers. This study will help future researchers to choose a particular RANS turbulence model vis-à-vis a particular wall-spectrum model in order to obtain a reasonably accurate wind tunnel result predicting capability. First, the best-predicting pressure spectrum models …
Autoencoders and generative neural network models have recently gained popularity in fluid mechanics due to their spontaneity and low processing time instead of high fidelity CFD simulations. Auto encoders are used as model order reduction tools in applications of fluid mechanics by compressing input high-dimensional data using an encoder to map the input space into a lower-dimensional latent space. Whereas, generative models such as Variational Auto-encoders (VAEs) and Generative Adversarial Networks (GANs) are proving to be effective in generating solutions to chaotic models with high 'rando…
Despite the progress in high performance computing, Computational Fluid Dynamics (CFD) simulations are still computationally expensive for many practical engineering applications such as simulating large computational domains and highly turbulent flows. One of the major reasons of the high expense of CFD is the need for a fine grid to resolve phenomena at the relevant scale, and obtain a grid-independent solution. The fine grid requirements often drive the computational time step size down, which makes long transient problems prohibitively expensive. In the research presented, the feasibility …
A Kalman filter based sequential estimator is presented in the present work. The estimator is integrated in the structure of segregated solvers for the analysis of incompressible flows. This technique provides an augmented flow state integrating available observation in the CFD model, naturally preserving a zero-divergence condition for the velocity field. Because of the prohibitive costs associated with a complete Kalman Filter application, two model reduction strategies have been proposed and assessed. These strategies dramatically reduce the increase in computational costs of the model, whi…
The Reynolds-averaged Navier-Stokes (RANS) equations for steady-state assessment of incompressible turbulent flows remain the workhorse for practical computational fluid dynamics (CFD) applications. Consequently, improvements in speed or accuracy have the potential to affect a diverse range of applications. We introduce a machine learning framework for the {surrogate modeling of steady-state turbulent eddy viscosities for RANS simulations, given the initial conditions. This modeling strategy} is assessed for parametric interpolation, while numerically solving for the pressure and velocity equa…
Hybrid Reynolds-averaged Navier Stokes large eddy simulation (RANS LES) methods have become popular for simulation of massively separated flows at high Reynolds numbers due to their reduced computational cost and good accuracy. The current study aims to examine the performance of LES and hybrid RANS-LES model for a given grid resolution. For better assessment and contrast of model performance, both mean and instantaneous flow fields have been investigated. For studying instantaneous flow, proper orthogonal decomposition has been used. Current analysis shows that hybrid RANS-LES is capable of a…
This study has numerically investigated the effect of surface wettability on two-phase immiscible flow and dynamics of droplet pinch-off in a T-junction microchannel using finite element method. A conservative level set method (CLSM) has been adopted to capture the interface topology in squeezing regime ($Ca_c <10^{-2}$) for wide flow rate ratio ($1/10 \leq Q_r \leq 10$) and contact angle ($120^{\circ} \leq θ\leq 180^{\circ}$). Based on the instantaneous phase profiles, droplet formation stages are classified as initial, filling, squeezing, pinch-off and stable droplet. Wettability effects are…
The main objective of this paper is to present a generic modelling framework, for the diffusive mass transport through the turbulent, reactive boundary layer of multi-component fluid mixtures that precipitate on the wall. The modelling is based on Maxwell-Stefan diffusion in multi-component mixtures, the relaxation to chemical equilibrium model, and the single-phase Navier-Stokes equations. Final-ly, turbulence is introduced by Reynolds-averaging. The governing equations are simplified in accordance with common assumptions of computational fluid dynamics (CFD), and based on the assumption that…
In this paper, we present the results of a numerical study of air-water turbulent bubbly flow in a periodic vertical square duct. The study is conducted using a novel numerical technique which leverages Volume of Fluid method for interface capturing and Sharp Surface Force method for accurate representation of the surface tension forces. A three-dimensional geometry construction method is employed during solution of interface equation which gives absolute conservation of mass and sharp interface between gas and liquid phases. The entire algorithm has been implemented on a data parallel mode on…
Airflow through the nasal cavity exhibits a wide variety of fluid dynamicsbehaviour due to the intricacy of the nasal geometry. The flow is naturallyunsteady and perhaps turbulent, despite CFD in the literature that assumesa steady laminar flow. Time-dependent simulations can be used to generatedetailed data with the potential to uncover new flow behaviour, although theyare more computationally intensive compared with steady-state simulations. Furthermore, verification of CFD results has relied on reported pressure drop(e.g. nasal resistance) across the nasal airway although the geometries use…
This paper reports on the investigation of combustion instabilities in a Methane air nonpremixed half dump combustor for different flow Reynolds number using Computational aero acoustic simulation. In order to simulate the flow physics under turbulent flow conditions, Detached Eddy Simulation in conjunction with generalized eddy dissipation model is adopted, while a CAA formulation based on Lighthills acoustic analogy is used for computing the acoustic field. It is observed that the unsteady pressure signals either predominantly arise from the natural acoustic modes of the duct or the local fl…
Turbulent characteristics within shear layer have been studied, recently, within vegetative canopy, buildings, dunes. Kevin-Helmholtz instability triggered hairpin vortex shedding has been widely concluded as the "signature" of mixing layer analogy. However, convoluted roughness types complicate the observing practice of turbulent evolving progress. To simplify that, spanwise homogeneous roughness has been adopted in this work to capture the streamwise flow alternations under different streamwise roughness element distances. Streamwise successive object distance $λ_x$ is used to show the rough…
Neural Operator-Based Surrogate Model for CFD:Helical Coil Steam Generator in Small Modular Reactor
Department of Quantum System Engineering, Jeonbuk National University（全子系统工程系，全州国立大学）；Department of Nuclear Engineering, Hanyang University（核工程系，翰阳大学）；Graduate School of Integrated Energy-AI, Jeonbuk National University（整合能源-人工智能研究生院，全州国立大学）
Idaho National Laboratory（爱达荷国家实验室）；Massachusetts Institute of Technology（麻省理工学院）；North Carolina State University（北卡罗来纳州立大学）
Multiphase flow phenomena have been widely observed in the industrial applications, yet it remains a challenging unsolved problem. Three-dimensional computational fluid dynamics (CFD) approaches resolve of the flow fields on finer spatial and temporal scales, which can complement dedicated experimental study. However, closures must be introduced to reflect the underlying physics in multiphase flow. Among them, the interfacial forces, including drag, lift, turbulent-dispersion and wall-lubrication forces, play an important role in bubble distribution and migration in liquid-vapor two-phase flow…
Bayesian optimization (BO) based on Gaussian process regression (GPR) is applied to different CFD (computational fluid dynamics) problems which can be of practical relevance. The problems are i) shape optimization in a lid-driven cavity to minimize or maximize the energy dissipation, ii) shape optimization of the wall of a channel flow in order to obtain a desired pressure-gradient distribution along the edge of the turbulent boundary layer formed on the other wall, and finally, iii) optimization of the controlling parameters of a spoiler-ice model to attain the aerodynamic characteristics of …
The volume of fluid (VOF) method is a popular approach for multiphase flow modeling. The open-source computational fluid dynamics (CFD) software, OpenFOAM, implements a variety of VOF-based solvers and provides users a wide range of turbulence models. Since isothermal multiphase flows under the VOF framework belong to the variable-density incompressible flow category, the isothermal VOF-based solvers in OpenFOAM fail to use the correct turbulence models. varRhoTurbVOF is designed to solve this issue and with the hope to replace all the corresponding existing solvers in the future. With the obj…
This review focuses on Direct numerical simulations (DNS) of turbulent flows laden with droplets or bubbles. DNS of these flows are more challenging than those of flows laden with solid particles due to the surface deformation in the former. The classification of the discussed numerical methods is based on whether the initial diameter of the bubble/droplet is smaller or larger than the Kolmogorov length scale, and whether the instantaneous surface deformation is fully resolved or obtained via a phenomenological model. Numerical methods that account for the breakup of single droplet/bubble as w…
This work determines the inaccuracy of using Reynolds averaged Navier Stokes (RANS) turbulence models in transition to turbulent flow regimes by predicting the model-based discrepancies between RANS and large eddy simulation (LES) models and then incorporates the capabilities of machine learning algorithms to characterize the discrepancies which are defined as a function of mean flow properties of RANS simulations. First, three-dimensional CFD simulations using k-omega Shear Stress Transport (SST) and dynamic one-equation subgrid-scale models are conducted in a wall-bounded channel containing …
A molecular dynamics (MD) simulation of planar Couette flow is presented for the minimal channel in which turbulence structures can be sustained. Evolution over a single breakdown and regeneration cycle is compared to computational fluid dynamics (CFD) simulations. Qualitative similar structures are observed and turbulent statistics show excellent quantitative agreement. The molecular scale law of the wall is presented in which stick-slip molecular wall-fluid interactions replace the no-slip conditions. The impact of grid resolution is explored and the observed structures are seen to be depend…
JAX-Fluids 2.0: Towards HPC for Differentiable CFD of Compressible Two-phase Flows
Technical University of Munich, School of Engineering（慕尼黑工业大学工程学院）；Technical University of Munich, Munich Institute of Integrated Materials, Energy and Process Engineering（慕尼黑工业大学慕尼黑集成材料、能源与过程工程研究所）
Vortex induced vibration (VIV) occurs when vortex shedding frequency falls close to the natural frequency of a structure. Investigation on VIV is of great value in disaster mitigation, energy extraction and other applications. Following recent development in machine learning on VIV in laminar flow, this study extends it to the turbulent region by employing the state-of-the-art parameterised Navier-Stokes equations based physics informed neural network (PNS-PINN). Turbulent flow with Reynolds number $Re$ = $10^4$, passing through a cylinder undergoing VIV motion, was considered as an example. W…
The progress made in accelerating simulations of fluid flow using GPUs, and the challenges that remain, are surveyed. The review first provides an introduction to GPU computing and programming, and discusses various considerations for improved performance. Case studies comparing the performance of CPU- and GPU- based solvers for the Laplace and incompressible Navier-Stokes equations are performed in order to demonstrate the potential improvement even with simple codes. Recent efforts to accelerate CFD simulations using GPUs are reviewed for laminar, turbulent, and reactive flow solvers. Also, …
A computational fluid dynamics (CFD) simulation framework for fluid-flow prediction is developed on the Tensor Processing Unit (TPU) platform. The TPU architecture is featured with accelerated dense matrix multiplication, large high bandwidth memory, and a fast inter-chip interconnect, making it attractive for high-performance scientific computing. The CFD framework solves the variable-density Navier-Stokes equation using a low-Mach approximation, and the governing equations are discretized by a finite-difference method on a collocated structured mesh. It uses the graph-based TensorFlow as the…
Reduced Order Modeling (ROM) for engineering applications has been a major research focus in the past few decades due to the unprecedented physical insight into turbulence offered by high-fidelity CFD. The primary goal of a ROM is to model the key physics/features of a flow-field without computing the full Navier-Stokes (NS) equations. This is accomplished by projecting the high-dimensional dynamics to a low-dimensional subspace, typically utilizing dimensionality reduction techniques like Proper Orthogonal Decomposition (POD), coupled with Galerkin projection. In this work, we demonstrate a d…
