Article
Mechanics
Davide Perrone, J. G. M. Kuerten, Luca Ridolfi, Stefania Scarsoglio
Summary: Mixing of inertial point particles in turbulent channel flow at Re-z = 950 is studied using direct numerical simulations. The study considers the release of inertial particles, with varying Stokes number, from different positions inside the channel and analyzes their rate of coming close to each other. A Lagrangian framework is employed to analyze trajectories and study mixing and dispersion problems. The research provides a comprehensive understanding of mixing in an anisotropic turbulent flow by varying the release position of particles along the wall-normal direction. Moreover, the effects of particle inertia are analyzed and found to depend on the position and alignment of the sources due to the dependence of the flow timescales on the distance from the wall.
INTERNATIONAL JOURNAL OF MULTIPHASE FLOW
(2023)
Article
Mechanics
Paul Andrade, Yannis Hardalupas, Georgios Charalampous
Summary: This study uses a combination of graph theory and Voronoi analysis to investigate the structure and internal sub-structure of clusters. Experimental results show that large clusters have a significantly large number of neighboring particles for collisions, while small clusters have a random number of neighbors.
Article
Mechanics
Yixiang Wang, Kit Ming Lam, Kam Tim Tse
Summary: This study investigates the settling velocity and clustering behavior of bidisperse inertial particles in turbulent channel flow through direct numerical simulation. The results show a significant turbophoresis effect on smaller diameter particles in bidisperse cases, influencing clustering and turbophoresis. It is also found that terminal settling velocities in bidisperse cases are affected by final volume fractions at dynamic equilibrium state, with a strengthened preferential sweeping mechanism observed with decreasing Stokes number.
Article
Physics, Fluids & Plasmas
Jan Meibohm, Vikash Pandey, Akshay Bhatnagar, Kristian Gustavsson, Dhrubaditya Mitra, Prasad Perlekar, Bernhard Mehlig
Summary: In the dynamics of small, heavy identical particles in turbulence, the formation of singularities known as caustics leads to significant fluctuations in spatial particle-number density and collision velocities. While caustic formation for particles with large inertia is akin to Kramers escape, for particles with small inertia, caustics tend to form near trajectories with specific histories of fluid-velocity gradients characterized by low vorticity and violent strain surpassing a certain threshold. A theory is developed to explain these findings in terms of an optimal path to caustic formation in the small inertia limit.
PHYSICAL REVIEW FLUIDS
(2021)
Review
Thermodynamics
Eric Loth
Summary: Turbulence is crucial for spreading particles and drops in energy systems, and recent advances have focused on the complexities of particle motion in turbulent flows. This review discusses the fundamental features of turbulence and its influence on particle motion, including turbulent diffusivity, kinetic energy of particle velocity, and turbophoresis. It also examines turbulent biases such as non-linear drag bias and clustering bias, as well as recent progress in turbulence modulation and particle collision frequency. A generalized flow regime is presented to summarize the interactions based on particle size and concentration.
PROGRESS IN ENERGY AND COMBUSTION SCIENCE
(2023)
Article
Mechanics
J. G. Wang, J. L. Yu
Summary: The interactions between large and small scales in compressible turbulent mixing layers were investigated using numerical simulations. The study found that large-scale compressive velocity fluctuations modulate the small scales with different properties on each side of the mixing layer, and this modulation decreases with increasing convective Mach number.
Article
Physics, Fluids & Plasmas
Daiki Terakado, Taku Nonomura, Soshi Kawai, Hikaru Aono, Makoto Sato, Akira Oyama, Kozo Fujii
Summary: This study investigates the effects of shocklets on sound source characteristics through direct numerical simulations. It is found that shocklets not only affect the sound generation mechanism but also change the relationship between Reynolds stress and entropy terms. The study also discusses the applicability of predicting far-field acoustic wave characteristics and provides a possible explanation for the contributions of shocklets to crackle noise mechanisms.
PHYSICAL REVIEW FLUIDS
(2022)
Article
Mechanics
Kyle Pietrzyk, Jeremy A. K. Horwitz, Fady M. Najjar, Roger W. Minich
Summary: This study analyzes particle-laden, isotropic turbulence in three dimensions to understand the dynamics of inertial particles from a kinetic energy perspective. By identifying data trends, it is found that particles tend to accumulate in regions of low flow kinetic energy over time, as they lose kinetic energy and slow down in such regions. A particle kinetic energy equation is derived and hypotheses regarding the temporal change of particle kinetic energy and particle behavior are evaluated using simulation data. The steady-state probability density function of particle kinetic energy is derived using a Fokker-Planck equation. The model fits the simulation data well and provides a tool for investigating preferential concentration and predicting particle kinetic energy in turbulent flows.
Article
Mechanics
Du-Chang Xu, Xiao-Ying Tang, Ao Li, Jing-Tao Ma, Yuan-Qing Xu
Summary: The particle focus in channel flow refers to the equilibrium position reached by an initialized particle. The binding focus is a phenomenon where two adjacent particles can form a new equilibrium position. This study explores the external force attached binding focus and presents a three-dimensional model using the immersed boundary-lattice Boltzmann method. The migration conditions of the soft particle and its application in single-cell separation are discussed and validated numerically.
Article
Mechanics
Bo Yang, Cheng Peng, Guichao Wang, Lian-Ping Wang
Summary: In this study, direct numerical simulations of turbulent downward channel flow laden with finite-size spherical particles were conducted using the lattice Boltzmann method. It was found that the settling particles in the downward channel flow have an overall positive slip velocity at the center, causing the lateral hydrodynamic force to drive particles away from the center region. Additionally, an increase in particle terminal velocity leads to higher levels of particle accumulation near the wall.
Article
Physics, Fluids & Plasmas
A. D. Bragg, D. H. Richter, G. Wang
Summary: Even when the settling parameter is small, gravitational settling can still have a leading order contribution to concentration profiles. In the boundary layer, there is always a region where settling cannot be neglected, regardless of how small the settling parameter is.
PHYSICAL REVIEW FLUIDS
(2021)
Article
Engineering, Chemical
Mohammad Mahtab Alam, Zafar M. Malikov, Bekhzod Z. Malikov, Akermi Mehdi, M. R. Gorji, Walid Belhadj
Summary: The paper presents a two-fluid approach to modeling turbulent compressible flow. A mathematical model of turbulent compressible heat is constructed and used to study various flow scenarios. The results show that the developed two-fluid model accurately describes turbulent compressible flows.
JOURNAL OF THE TAIWAN INSTITUTE OF CHEMICAL ENGINEERS
(2023)
Article
Environmental Sciences
Hasan Zobeyer, Abul B. M. Baki, Saika Nowshin Nowrin
Summary: This study presents experimental results on the mean and turbulence characteristics of flow generated by a pair of cylinders placed in tandem in an open-channel flume, using an acoustic Doppler velocimeter (ADV) to measure the instantaneous three-dimensional velocity components. The downstream cylinder significantly influences the flow development between cylinders, with turbulent peak generally occurring near the end of the recirculation zone in all scenarios.
Article
Engineering, Aerospace
Jiseop Lim, Minwoo Kim, Seungtae Kim, Solkeun Jee, Donghun Park
Summary: An efficient high-fidelity simulation approach combining large-eddy simulation (LES) with the parabolized stability equations (PSE) analysis is investigated for laminar-to-turbulent transition in compressible boundary layer flow. The PSE analysis is used for an efficient treatment of instability modes in the flow, while LES is chosen for high-fidelity simulation of the transitional flow. The current study demonstrates a PSE+LES framework for turbulent transition in supersonic boundary layer flow, showing a significant reduction in computational cost compared to relevant DNS cost.
AEROSPACE SCIENCE AND TECHNOLOGY
(2021)
Article
Mechanics
Ming Yu, Dong Sun, QingQing Zhou, PengXin Liu, XianXu Yuan
Summary: In this study, we used direct numerical simulation databases to investigate the evolution of turbulent statistics and coherent structures in hypersonic turbulent boundary layers subjected to oblique shock waves. We found that large-scale structures are amplified within the interaction zone and gradually decay downstream, leading to discrepancies between the predictions of Reynolds-Averaged Navier-Stokes simulations and the actual flow behavior. To address this issue, we proposed refining the model parameters as functions of wall pressure and other flow quantities, leading to improved accuracy in predicting various flow properties downstream of the interaction zone.
Article
Mechanics
Qingjia Meng, Zhou Jiang, Jianchun Wang
Summary: Fully connected neural networks (FCNNs) have been developed for closing subgrid-scale stress and heat flux in large-eddy simulations of compressible turbulent channel flow. The FCNN-based SGS model shows better results than the dynamic Smagorinsky model (DSM) in reconstructing SGS unclosed terms. It also performs well in predicting various variables such as mean velocity profiles, turbulent intensities, and total Reynolds stress.
THEORETICAL AND APPLIED MECHANICS LETTERS
(2023)
Article
Mechanics
Yunpeng Wang, Zelong Yuan, Jianchun Wang
Summary: Current research shows that turbulence small-scale errors can be eliminated through data assimilation, but only if large-scale Fourier modes below a critical wavenumber are continuously enforced. However, when large-scale data is insufficient, an artificial jump in the energy spectrum appears. Several approaches have been attempted to address this issue, including ensemble averaging, temporally sparse data assimilation, and filtering the penalty term. Each approach has its limitations and drawbacks, but a re-scaled ensemble average method shows improved accuracy in energy spectrum and small-scale reconstruction.
Article
Mechanics
Wenhui Peng, Zelong Yuan, Zhijie Li, Jianchun Wang
Summary: Modeling three-dimensional turbulence with neural networks is challenging due to its highly nonlinear nature and memory-intensive simulation. An attention mechanism, specifically the linear attention, is proposed to overcome the computational bottleneck of traditional self-attention mechanism. The linear attention coupled Fourier neural operator shows significant improvement in accuracy and efficiency for 3D turbulence simulation, making it applicable to other high-dimensional data problems.
Editorial Material
Mechanics
Xiang Yang, Jianchun Wang
THEORETICAL AND APPLIED MECHANICS LETTERS
(2023)
Article
Computer Science, Interdisciplinary Applications
Wenjin Zhao, Guiyu Cao, Jianchun Wang, Kun Xu
Summary: This study validates the effectiveness of high-order non-compact and compact reconstruction in turbulence simulation using the high-order gas-kinetic scheme (HGKS). The accuracy of HGKS is confirmed through numerical simulations of three-dimensional density perturbation advection. Both non-compact 7th-order and compact 5th-order reconstruction schemes are shown to provide accurate solutions for turbulent flows.
COMPUTERS & FLUIDS
(2023)
Article
Biology
Yu Wang, Yu Zhang, Jianchun Wang, Fang Xie, Dequan Zheng, Xiang Zou, Mian Guo, Yijie Ding, Jie Wan, Ke Han
Summary: Drug discovery is a complex process that requires significant investment and resources, as well as professional knowledge and skills. Machine learning methods, such as the neighborhood regularized logistic matrix factorization, can be used to predict drug-target interactions and reduce the cost and time of drug development.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Mechanics
Dehao Xu, Jianchun Wang, Changping Yu, Shiyi Chen
Summary: In this paper, an artificial-neural-network-based (ANN-based) nonlinear algebraic model, called MANA model, is proposed for the large-eddy simulation (LES) of compressible wall-bounded turbulence. The model incorporates innovative modifications to the invariants and tensor bases, and utilizes local grid widths to normalize the flow variable gradients. The MANA model outperforms traditional eddy-viscosity models in terms of correlation coefficients, relative errors, and accuracy of predicting flow statistics and mean subgrid-scale fluxes in both a priori and a posteriori tests.
JOURNAL OF FLUID MECHANICS
(2023)
Article
Mechanics
Chenglong Hu, Ke Xu, Yantao Yang
Summary: Convective dissolution plays a significant role in long-term CO2 sequestration in deep saline aquifers, and the presence of an unstable geothermal gradient affects the dissolution process. In this study, direct numerical simulations were conducted in a three-dimensional porous medium under different concentration and thermal Rayleigh numbers. The results show that the flow structures change with increasing thermal Rayleigh number, leading to alterations in the distribution and motions of concentration fingers. A theoretical model is developed to describe the evolution of concentration flux and volume averaged concentration. The dissolved CO2 into the interior exhibits non-monotonic variations as the thermal Rayleigh number increases, with the maximum increment occurring when the density ratio is around unity.
JOURNAL OF FLUID MECHANICS
(2023)
Article
Mechanics
Dehao Xu, Jianchun Wang, Shiyi Chen
Summary: The effects of Reynolds number and wall cooling on correlations between thermodynamic variables in hypersonic turbulent boundary layers are investigated. The Kovasznay decomposition is used to analyze the fluctuating density and temperature. Results show the presence of alternating positive and negative structures in the fluctuating pressure and acoustic modes, as well as streaky entropic structures in the fluctuating entropy and entropic modes near the wall. Correlations involving density and temperature are primarily contributed by entropic modes. Fluctuating temperature is strongly positively correlated with streamwise velocity near the wall in strongly cooled wall cases due to the presence of alternating structures.
JOURNAL OF FLUID MECHANICS
(2023)
Article
Physics, Multidisciplinary
Leiqi Yuan, Shufan Zou, Yantao Yang, Shiyi Chen
Summary: In this study, a novel strategy is proposed to enhance heat transfer in convection turbulence. By introducing a standing-wave type boundary deformation, flow modulation can be achieved when the amplitude is comparable or larger than the boundary-layer thickness. The heat-flux enhancement primarily occurs in the near-wall regions affected by the boundary deformation at large wave numbers. The findings suggest that oscillating deformations of the boundary can effectively break the boundary layers and open new possibilities for modulating convection turbulence.
PHYSICAL REVIEW LETTERS
(2023)
Article
Mechanics
Ning Chang, Zelong Yuan, Yunpeng Wang, Jianchun Wang
Summary: In this study, the effects of filter anisotropy and sub-filter scale (SFS) dynamics on the accuracy of large eddy simulation (LES) of turbulence were investigated using various SFS models. The results showed that the direct deconvolution model (DDM) was capable of accurately predicting SFS stresses at highly anisotropic filters. The DDM outperformed the dynamic Smagorinsky model (DSM) and dynamic mixed model (DMM) in predicting various turbulence statistics.
Article
Mechanics
Yunpeng Wang, Zelong Yuan, Jianchun Wang
Summary: An ensemble Kalman filter-based mixed model is proposed for the subgrid-scale closure in large-eddy simulation of turbulence. The model coefficients are determined through data assimilation, and the performance of the model is compared to traditional subgrid-scale models. The results show that the proposed model consistently outperforms the traditional models, demonstrating its potential in improving the large-eddy simulation of turbulence.
Article
Mechanics
Yuhang Du, Yantao Yang
Summary: This work investigates convection turbulence driven by heat-releasing point particles that absorb energy from external sources. Both momentum and temperature fields are considered through two-way coupling, with particle dynamics including Stokes drag and gravity force. The main focus is on the gravity effect of particles on convection turbulence. Two regimes are identified at large and small Froude numbers, respectively. Within the large Froude number regime, transport properties show weak dependence on Froude number but strong dependence on Stokes number. In the regime with small Froude numbers, particles accumulate near the bottom plate in the boundary layer region. Scaling laws for the critical Froude number between the two regimes are derived and found to agree well with numerical results.
Article
Mechanics
Zhijie Li, Wenhui Peng, Zelong Yuan, Jianchun Wang
Summary: In this paper, an implicit U-Net enhanced Fourier neural operator (IU-FNO) model is proposed for stable and efficient predictions on the long-term large-scale dynamics of three-dimensional turbulence. The IU-FNO model incorporates implicit recurrent Fourier layers and the U-net network to improve accuracy and predict small-scale flow structures. Numerical simulations demonstrate that the IU-FNO model outperforms other FNO-based models in terms of accuracy and stable predictions on various statistics of turbulence.
Article
Physics, Fluids & Plasmas
Candi Zheng, Yang Wang, Shiyi Chen
Summary: Finding extended hydrodynamics equations that can be applied from the dense gas region to the rarefied gas region is a difficult task. Accurate constitutive relations for stress and heat flux are crucial for success. Data-driven models offer a new approach to learning these relations, but the choice of derivatives in these models is arbitrary and lacks a physical explanation.
