Direct Numerical Simulation Results Research Articles

Investigations on the turbulentnon-turbulent interface in supersonic compressible plate turbulent boundary layer

The turbulent boundary layer (TBL) is a widely existing flow phenomenon in nature and engineering applications. Its strong mixing effect can achieve more sufficient material mixing, heat transport, etc. The understanding of the entrainment process and mechanism of irrotational fluids entering the turbulent region can be promoted by studying the geometric and dynamic characteristics of turbulent ${/}$ non-turbulent interfaces (TNTI). In compressible flow, it is unclear whether the properties of TNTI will change and whether the entrainment will show different features due to the influence of compressibility. Based on the direct numerical simulation results of supersonic compressible plate TBLs with Mach number of 2.9, the geometric and dynamic characteristics of TNTI are investigated in this paper. The interface is identified by the enstrophy method, and the height, thickness, fractal dimension, enstrophy transportation and entrainment characteristics of the interface are investigated. It is found that for the enstrophy transportation in a TBL, the contribution of compressibility-related terms accounts for approximately 13.4 % of the total enstrophy transportation, which tends to transfer the enstrophy of turbulence near the interface to both directions vertical to the interface. This promotes the expansion of the turbulent region towards the non-turbulent region, and the mean height, thickness and entrainment velocity are increased by approximately 3.7 %, 7.0 % and 8.5 %, respectively, while the fractal dimension is basically unaffected. Different from the incompressible flow, the contribution of the compressibility-related terms to the entrainment velocity is independent of the local curvature, and the intense entrainment process is more likely to occur on a highly curved concave surface.

Direct FE2 Based on Single-Integration Point for Modeling Damage Evolution of Materials with Micro-Cracks

The Direct FE2 method is a recently developed concurrent multi-scale simulation approach with a monolithic solution scheme, where the macroscopic and microscopic models are solved in a unit iteration process. The conventional Direct FE2 using quadrilateral element has some limitations in determining the failure state of a macro-element. In this work, we develop a new Direct FE2 method based on the Single-integration Point triangle element (Direct SP-FE2) and apply it to simulate the micro-crack-induced fracture problems for the first time. The cohesive element is inserted into a representative volume element (RVE), so that the failure process of a macro-element can be determined by the micro-crack propagation on RVE. Accordingly, the relationship between macro-deformations and microstructures can be established directly, thus avoiding complex derivation and utilization of a multi-scale damage constitutive model. In Direct SP-FE2, the failure state of a macro-element can be directly predicted by the RVE at the single-integration point inside, avoiding the uncertainty associated with predictions by multiple RVEs. As a result, compared to the conventional Direct FE2 with a quadrilateral element, the Direct SP-FE2 shows advantages in capturing complex geometric boundaries and predicting the damage evolution process and failure state more accurately. Various numerical examples are conducted to comprehensively validate the effectiveness of the present method in describing the influence of micro-cracks on macro-structure deformations. For the macro-damage behaviors, the simulation results by the Direct SP-FE2 show good agreement with the direct numerical simulation (DNS) results of the full FE model at a significantly lower computational time. With the developed Direct SP-FE2, a variety of complex micro-crack fracture problems in composite materials can be successfully modeled by manipulating the interior RVE structure.

Is a direct numerical simulation (DNS) of Navier-Stokes equations with small enough grid spacing and time-step definitely reliable/correct?

Turbulence is strongly associated with the vast majority of fluid flows in nature and industry. Traditionally, results given by the direct numerical simulation (DNS) of Navier-Stokes (NS) equations that relate to a famous millennium problem are widely regarded as ‘reliable’ benchmark solutions of turbulence, as long as grid spacing is fine enough (i.e. less than the minimum Kolmogorov scale) and time-step is small enough, say, satisfying the Courant-Friedrichs-Lewy condition (Courant number < 1). Is this really true? In this paper a two-dimensional sustained turbulent Kolmogorov flow driven by an external body force governed by the NS equations under an initial condition with a spatial symmetry is investigated numerically by the two numerical methods with detailed comparisons: one is the traditional DNS, the other is the ‘clean numerical simulation’ (CNS). In theory, the exact solution must have a kind of spatial symmetry since its initial condition is spatially symmetric. However, it is found that numerical noises of the DNS are quickly enlarged to the same level as the ‘true’ physical solution, which finally destroy the spatial symmetry of the flow field. In other words, the DNS results of the turbulent Kolmogorov flow governed by the NS equations are badly polluted mostly. On the contrary, the numerical noise of the CNS is much smaller than the ‘true’ physical solution of turbulence in a long enough interval of time so that the CNS result is very close to the ‘true’ physical solution and thus can remain symmetric, which can be used as a benchmark solution for comparison. Besides, it is found that numerical noise as a kind of artificial tiny disturbances can lead to huge deviations at large scale on the two-dimensional Kolmogorov turbulence governed by the NS equations, not only quantitatively (even in statistics) but also qualitatively (such as spatial symmetry of flow). This highly suggests that fine enough spatial grid spacing with small enough time-step alone could not guarantee the validity of the DNS of the NS equations: it is only a necessary condition but not sufficient. All of these findings might challenge some of our general beliefs in turbulence: for example, it might be wrong in physics to neglect the influences of small disturbances to NS equations. Our results suggest that, from physical point of view, it should be better to use the Landau-Lifshitz-Navier-Stokes (LLNS) equations, which consider the influence of unavoidable thermal fluctuations, instead of the NS equations, to model turbulent flows.

Spectral analysis of dispersed multiphase flows in the presence of fluid interfaces

Spectral analysis of dispersed two-phase flows is highly desirable to reveal the interplay of the various flow scales, much larger or much smaller than the size of the dispersed bodies. This is a challenging task as the matching conditions at the body interfaces generate singularities in the fields describing the two-phase mixture. The nature of these singularities and their consequences on the spectra are theoretically analyzed for bubble or droplet flows. Results of direct numerical simulations are reported and spatial spectra of the mixture velocity, the flow forces and their power are examined. The regular part of the spectral densities of energy production, dissipation and transfers between scales are separated from their singular part. The resulting spectral energy balance, free of the footprint of the singularities, is found in agreement with coarse-grained simulations where the interfaces are filtered out before solving the Navier–Stokes equations. These results pave the way for the spectral analysis of more complex turbulent dispersed flows.

Large eddy simulation of a turbulent channel flow using dynamic Smagorinsky subgrid scale model and differential equation wall model

A large eddy simulation (LES) is conducted in a plane channel flow which is turbulent. For understanding the mechanism of the turbulent motion more efficiently, the present study is an attempt to conduct the LES using a Dynamic Smagorinsky Model (DSM) and a Differential Equation Wall Model (DEWM). DSM is a subgrid-scale model and DEWM is for approximating the near wall layer. The simulation is performed in a domain of 2 π δ × 2 δ × π δwhich is discretized by 32 × 20 × 32grid points where δ is the channel half width. The discretization techniques have been used in the simulation, are the 3rd order low storage R-K method in time and 2nd order finite difference formulation in space. To assess the performance of the LES with DSM and DEWM, the simulation results are compared with direct numerical simulation (DNS) data and LES data using Standard Smagorinsky Model (SSM). Comparing the results in the essential turbulence statistical field it has been found that DSM performs better than SSM in most of the cases and shows better agreement with the DNS results. Flow structures in the turbulent flow field are compared in isosurfaces and contour plots, and found that tube-like vortical structures are randomly distributed over the entire flow field, and the number of the vortical structures are more for the DSM case than that of SSM.

The Basic k-ϵ Model and a New Model Based on General Statistical Descriptions of Anisotropic Inhomogeneous Turbulence Compared with DNS of Channel Flow at High Reynolds Number

Predictions are presented of mean values of statistical variables of large-scale turbulent flow of the widely used basic k-ϵ model, and of a new model, which is based on general statistical descriptions of turbulence. The predictions are verified against published results of direct numerical simulations (DNSs) of Navier–Stokes equations. The verification concerns turbulent channel flow at shear Reynolds numbers of 950, 2000, and 104. The basic k-ϵ model is largely based on empirical formulations accompanied by calibration constants. This contrasts with the new model, where descriptions of leading statistical quantities are based on the general principles of statistical turbulence at a large Reynolds number and stochastic theory. Predicted values of major output variables such as turbulent viscosity, diffusivity of passive admixture, temperature, and fluid velocities compare well with DNS for the new model. Significant differences are seen for the basic k-ϵ model.

Direct numerical simulation and mode analysis of turbulent transition flow in a compressor blade channel

The separation and turbulent transition of the flow in a compressor blade channel are investigated through direct numerical simulations (DNS) at a Reynolds number of 1.367 × 105. Based on the original DNS data, both time-averaged statistics and instantaneous vortex structures of the flow field are extensively analyzed. The vortices are visualized and studied by the Liutex method, and the streaming dynamic mode decomposition (SDMD), a low-storage variant of conventional DMD, is applied to the large datasets obtained on both pressure and suction sides. The physical quantity analyzed with SDMD is the Liutex magnitude R. The DNS results indicate that flow separation occurs on both sides of the blade. On the pressure surface, the separation is weak and the flow remains in a natural transition dominated by viscous Tollmien–Schlichting instabilities. In contrast, owing to the presence of a large laminar separation bubble, the flow experiences a separation transition governed by inviscid Kelvin–Helmholtz instabilities on the suction surface. The SDMD results suggest that a broad range of vortex frequencies exist in the transition flow, and the scale of the spatial structures is negatively correlated with the frequency of the mode. On the pressure surface, the extracted SDMD modes are primarily related to Kelvin–Helmholtz rolls, whereas on the suction side, influenced by the separated boundary layer, the modal structures exhibit greater diversity.

The role of nonlinear interactions in the onset of drag increase in flow over riblets

Characterizing the mechanisms that contribute to the onset of drag increase over micro-grooves (riblets) as the spacing increases is critical to design strategies for riblet-based drag reduction. This study decomposes the roughness function to investigate different mechanisms associated with the breakdown of drag reduction as riblet spacing is increased. We obtain the roughness function through direct numerical simulations (DNS) in a minimal channel and restricted nonlinear (RNL) models. Both the traditional RNL decomposition and an augmented RNL (ARNL) model that includes additional nonlinear interactions are employed as computationally tractable, reduced order representations of the flow field. RNL and ARNL results are compared to those of DNS in minimal channels to investigate the role of the different scale-dependent nonlinear interactions contributing to the roughness function. A comparison of the co-spectra arising from the minimal channel DNS with that from RNL and ARNL simulations indicates that general trends are captured by both reduced order models. However, the additional nonlinearity introduced in the ARNL model produces closer correspondence in the observed structural features of the DNS results. In particular, the ARNL better captures the signatures of the dispersive flow and the texture-coherent fluctuations. There is also a noticeable improvement observed in the profiles of the added stress contributions obtained with the ARNL model versus the RNL model.

A Combined Model of Diffusion Model and Enhanced Residual Network for Super-Resolution Reconstruction of Turbulent Flows

In this study, we introduce a novel model, the Combined Model, composed of a conditional denoising diffusion model (SR3) and an enhanced residual network (EResNet), for reconstructing high-resolution turbulent flow fields from low-resolution flow data. The SR3 model is adept at learning the distribution of flow fields. The EResNet architecture incorporates a long skip connection extending from the input directly to the output. This modification ensures the preservation of essential features learned by the SR3, while simultaneously enhancing the accuracy of the flow field. Additionally, we incorporated physical gradient constraints into the loss function of EResNet to ensure that the flow fields reconstructed by the Combined Model are consistent with the direct numerical simulation (DNS) data. Consequently, the high-resolution flow fields reconstructed by the Combined Model exhibit high conformity with the DNS results in terms of flow distribution, details, and accuracy. To validate the effectiveness of the model, experiments were conducted on two-dimensional flow around a square cylinder at a Reynolds number (Re) of 100 and turbulent channel flow at Re = 4000. The results demonstrate that the Combined Model can reconstruct both high-resolution laminar and turbulent flow fields from low-resolution data. Comparisons with a super-resolution convolutional neural network (SRCNN) and an enhanced super-resolution generative adversarial network (ESRGAN) demonstrate that while all three models perform admirably in reconstructing laminar flows, the Combined Model excels in capturing more details in turbulent flows, aligning the statistical outcomes more closely with the DNS results. Furthermore, in terms of L2 norm error, the Combined Model achieves an order of magnitude lower error compared to SRCNN and ESRGAN. Experimentation also revealed that SR3 possesses the capability to learn the distribution of flow fields. This work opens new avenues for high-fidelity flow field reconstruction using deep learning methods.

Structure and role of the pressure Hessian in regions of strong vorticity in turbulence

Amplification of velocity gradients, a key feature of turbulent flows, is affected by the non-local character of the incompressible fluid equations expressed by the second derivative (Hessian) of the pressure field. By analysing the structure of the flow in regions where the vorticity is the highest, we propose an approximate expression for the pressure Hessian in terms of the local vorticity, consistent with the existence of intense vortex tubes. Contrary to the often used simplification of an isotropic form for the pressure Hessian, which in effect inhibits vortex stretching, the proposed approximate form of the pressure Hessian enables much stronger vortex stretching. The prediction of the approximation proposed here is validated with results of direct numerical simulations of turbulent flows.

Vortex model of plane turbulent air flows in channels

We present a theoretical model of plane turbulent flows based on the previously proposed equations, which take into account both the longitudinal motion and the vortex tube rotation. Using the simple model of eddy viscosity, we obtain the analytical expressions for the mean velocity profiles of stationary turbulent flows. In particular, we consider the near-wall flow over a flat plate in a wind tunnel as well as Couette and Poiseuille flows in rectangular channels. In all these cases, the calculated velocity profiles are in good agreement with experimental data and results of direct numerical simulations.

Modeling of high-speed, methane–air, turbulent combustion, Part I: One-dimensional turbulence modeling with comparison to DNS

The ability of the one-dimensional turbulence (ODT) model to serve as a surrogate direct numerical simulation (DNS) is assessed for highly turbulent flames. The ODT model is applied to freely propagating premixed methane–air flames at Karlovitz numbers 10, 102, 103, and 104, and results are compared with DNS. The ODT model solves the conservation equations for momentum, energy, and species on a one-dimensional domain, which corresponds to a streamwise line of sight spanning the DNS domain. The effects of turbulent advection are modeled via a stochastic process, in which the Kolmogorov and reactive length and time scales are explicitly resolved. Molecular transport and chemical kinetics are concurrently advanced in time. Both the ODT and DNS simulations use a 21-species skeletal chemical model for methane combustion. The accuracy of the ODT model is assessed by comparing its predictions of several key characteristics of the flames for each Karlovitz number tested, including the turbulent flame speed and width and the joint probability density functions (jPDFs) of major and selected minor species as well as the heat release rate conditioned on temperature with the results of DNS under comparable conditions. The ODT model is shown to yield qualitative and quantitative agreement with the DNS data for most of the above flame characteristics. Discrepancies are observed primarily for the jPDFs of several minor species examined. Overall, the ODT approach is shown to be an effective surrogate of DNS, potentially useful for guiding chemical reaction model reduction and for assessing the sensitivities of the flame structure and the burning rate to chemistry under highly turbulent conditions.Novelty and Significance:The direct numerical simulations (DNS) of premixed turbulent methane–air flames presented in this work span a uniquely wide range of turbulent intensities, from relatively modest corresponding to Karlovitz number Ka=10 to ultra-high intensities at Ka=104. This represents virtually the entire range of turbulent intensities that could be encountered in any realistic situation. This is also the first time that such a wide range of conditions is probed for methane in high-fidelity, fully resolved simulations, which use a fully compressible set of flow equations. The one-dimensional turbulence (ODT) model utilizes the same forcing that is present in the DNS enabling a direct comparison between the ODT and DNS. The results show that ODT captures the key features of the DNS results. ODT is shown to be an effective surrogate for DNS and may be useful in guiding chemical reaction model reduction, where many simulations are required.

Accurate Lagrangian stochastic model on homogeneous turbulence under two shear orientations

The paper reports a numerical modelisation of homogeneous and stratified turbulence under vertical and horizontal shear by way of Lagrangian stochastic model (LSM) versus Froude number. Vertical and horizontal shear (θ = 0 and θ = π/2) are adopted. First, a comparison of predictions by LSM with results of direct numerical simulation of Jacobitz (DNSJ) is carried out. Current results obtained by LSM demonstrate the existence of asymptotic equilibrium states for various parameters of the problem particularly for the horizontal shear. Compared to the case of a vertical shear, an excellent agreement between predictions of LSM related to the case of horizontal shear and results of DNSJ is detected for growth rate, ratio kinetic energy to total energy and potential energy to total energy beyond τ = 40. For two shear orientations, results of normalized production and turbulence dissipation are comparable with different approaches for τ ≥ 70. Second, a comparative study of various thermal and dynamic parameters governing the considered problem for horizontal and vertical shear is performed versus Froude number. This study indicates that the LSM can present a significant contribution for numerical modelisation of stratified and homogeneous turbulence especially in the presence of a horizontal shear.

Asymptotic scaling relations for rotating spherical convection with strong zonal flows

We analyse the results of direct numerical simulations of rotating convection in spherical shell geometries with stress-free boundary conditions, which develop strong zonal flows. Both the Ekman number and the Rayleigh number are varied. We find that the asymptotic theory for rapidly rotating convection can be used to predict the Ekman number dependence of each term in the governing equations, along with the convective flow speeds and the dominant length scales. Using a balance between the Reynolds stress and the viscous stress, together with the asymptotic scaling for the convective velocity, we derive an asymptotic prediction for the scaling behaviour of the zonal flow with respect to the Ekman number, which is supported by the numerical simulations. We do not find evidence of distinct asymptotic scalings for the buoyancy and viscous forces and, in agreement with previous results from asymptotic plane layer models, we find that the ratio of the viscous force to the buoyancy force increases with Rayleigh number. Thus, viscosity remains non-negligible and we do not observe a trend towards a diffusion-free scaling behaviour within the rapidly rotating regime.

Machine learning-based vorticity evolution and super-resolution of homogeneous isotropic turbulence using wavelet projection

A wavelet-based machine learning method is proposed for predicting the time evolution of homogeneous isotropic turbulence where vortex tubes are preserved. Three-dimensional convolutional neural networks and long short-term memory are trained with a time series of direct numerical simulation (DNS) data of homogeneous isotropic turbulence at the Taylor microscale Reynolds number 92. The predicted results are assessed by using the flow visualization of vorticity and statistics, e.g., probability density functions of vorticity and enstrophy spectra. It is found that the predicted results are in good agreement with DNS results. The small-scale flow topology considering the second and the third invariants of the velocity gradient tensor likewise shows an approximate match. Furthermore, we apply the pre-trained neural networks to coarse-grained vorticity data using super-resolution. It is shown that the super-resolved flow field well agrees with the reference DNS field, and thus small-scale information and vortex tubes are well regenerated.

Upscaling of Reactive Mass Transport through Porous Electrodes in Aqueous Flow Batteries

Porous electrodes (PEs) are an important component of modern energy storage devices, such as lithium-ion batteries, flow batteries or fuel cells. Their complicated multiphase structure presents a considerable challenge to modeling and simulation. In this paper, we apply the volume-averaging method (VAM) as an efficient approach for the evaluation of effective macroscopic transport parameters in PEs. We consider the transport of electro-active species coupled to heterogeneous Butler-Volmer type reactions at the electrode surface. We identify the characteristic scales and dimensionless groups for the application to aqueous flow batteries. We validate the VAM-based model with direct numerical simulation results and literature data showing excellent agreement. Subsequently, we characterize several simplified periodic PE structures in 2D and 3D in terms of hydraulic permeability, effective dispersion and the effective kinetic number. We apply the up-scaled transport parameters to a simple macroscopic porous electrode to compare the overall efficiency of different pore-scale structures and material porosity values over a wide range of energy dissipation values. This study also reveals that the Bruggeman correction, commonly used in macroscopic porous electrode models, becomes inaccurate for realistic kinetic numbers in flow battery applications and should be used with care.

Wake dynamics in buoyancy-driven flows: Steady-state-Hopf-mode interaction with O(2) symmetry revisited.

We present a detailed mathematical study of a truncated normal form relevant to the bifurcations observed in wake flow past axisymmetric bodies, with and without thermal stratification. We employ abstract normal form analysis to identify possible bifurcations and the corresponding bifurcation diagrams in parameter space. The bifurcations and the bifurcation diagrams are interpreted in terms of symmetry considerations. Particular emphasis is placed on the presence of attracting robust heteroclinic cycles in certain parameter regimes. The normal form coefficients are computed for several examples of wake flows behind buoyant disks and spheres, and the resulting predictions compared with the results of direct numerical flow simulations. In general, satisfactory agreement is obtained.

Advanced Modeling of Flow and Heat Transfer in Rotating Disk Cavities Using Open-Source Computational Fluid Dynamics

Abstract Code_Saturne, an open-source computational fluid dynamics (CFD) code, has been applied to a range of problems related to turbomachinery internal air systems. These include a closed rotor–stator disk cavity, a co-rotating disk cavity with radial outflow and a co-rotating disk cavity with axial throughflow. Unsteady Reynolds-averaged Navier–Stokes (RANS) simulations and large eddy simulations (LES) are compared with experimental data and previous direct numerical simulation and LES results. The results demonstrate Code_Saturne's capabilities for predicting flow and heat transfer inside rotating disk cavities. The Boussinesq approximation was implemented for modeling centrifugally buoyant flow and heat transfer in the rotating cavity with axial throughflow. This is validated using recent experimental data and CFD results. Good agreement is found between LES and RANS modeling in some cases, but for the axial throughflow cases, advantages of LES compared to URANS are significant for a high Reynolds number condition. The wall-modeled large eddy simulation (WMLES) method is recommended for balancing computational accuracy and cost in engineering applications.

Direct numerical simulation of contaminant removal in presence of underfloor air distribution system

Indoor contaminant removal over 0.5 ≤ FrT ≤ 5.0, 0.5 ≤ N ≤ 5.0, and 50 ≤ Re ≤ 500 was investigated numerically, wherein FrT refers to the Froude number, N refers to the buoyancy ratio, and Re refers to the Reynolds number. As demonstrated, the ventilation effectiveness increased with increasing contaminant source intensity and air supply intensity at a constant air temperature, indicating that increase the fresh air can effectively eliminate contaminants in this case. At high air supply temperatures, the heat retention time and contaminant transport was extremely short, and the fresh air induced by strong natural convection floating lift was rapidly discharged. Additioanlly, the air supply intensity had significant effects on contaminant removal. Quantification of the ventilation effectiveness under the combined effects of air supply intensity, air supply temperature and contaminant source intensity was determined based on the results of direct numerical simulations.

Bifurcation analysis of double cavity flows

The first few bifurcations in a two-dimensional incompressible double cavity flow are investigated using the linear stability analysis, the Floquet analysis, and the nonlinear direct numerical simulations (DNS). The prediction of the critical Reynolds number and the type of bifurcation (Hopf, pitchfork, inverse pitchfork, and Neimark–Sacker), which depend on cavity configuration, by the linear stability analysis and the Floquet analysis is consistent with nonlinear DNS. The nonlinear DNS results show that the state of the system passes through multiple intermediate (unstable) states before it reaches the stable attractor (heteroclinic chain), and the type of intermediate states depends on initial conditions. The intermediate states are reported as the asymptotic state in the literature for some flow conditions because it is not known a priori how long it will take to reach the asymptotic state in nonlinear simulations. The present study reports the actual asymptotic state for those flow conditions.