High-order Discontinuous Galerkin(DG) methods have been receiving more and more attentions in the area of Computational Fluid Dynamics(CFD) because of their high-accuracy property. However, it is still a challenge to ...High-order Discontinuous Galerkin(DG) methods have been receiving more and more attentions in the area of Computational Fluid Dynamics(CFD) because of their high-accuracy property. However, it is still a challenge to obtain converged solution rapidly when solving the Reynolds Averaged Navier–Stokes(RANS) equations since the turbulence models significantly increase the nonlinearity of discretization system. The overall goal of this research is to develop an Artificial Neural Networks(ANNs) model with low complexity acting as an algebraic turbulence model to estimate the turbulence eddy viscosity for RANS. The ANN turbulence model is off-line trained using the training data generated by the widely used Spalart–Allmaras(SA) turbulence model before the Optimal Brain Surgeon(OBS) is employed to determine the relevancy of input features.Using the selected relevant features, a fully connected ANN model is constructed. The performance of the developed ANN model is numerically tested in the framework of DG for RANS, where the‘‘DG+ANN' method provides robust and steady convergence compared to the ‘‘DG+SA' method. The results demonstrate the promising potential to develop a general turbulence model based on artificial intelligence in the future given the training data covering a large rang of flow conditions.展开更多
Data-driven turbulence modeling studies have reached such a stage that the basic framework is settled,but several essential issues remain that strongly affect the performance.Two problems are studied in the current re...Data-driven turbulence modeling studies have reached such a stage that the basic framework is settled,but several essential issues remain that strongly affect the performance.Two problems are studied in the current research:(1)the processing of the Reynolds stress tensor and(2)the coupling method between the machine learning model and flow solver.For the Reynolds stress processing issue,we perform the theoretical derivation to extend the relevant tensor arguments of Reynolds stress.Then,the tensor representation theorem is employed to give the complete irreducible invariants and integrity basis.An adaptive regularization term is employed to enhance the representation performance.For the coupling issue,an iterative coupling framework with consistent convergence is proposed and then applied to a canonical separated flow.The results have high consistency with the direct numerical simulation true values,which proves the validity of the current approach.展开更多
We present the approaches to implementing the k-√k L turbulence model within the framework of the high-order discontinuous Galerkin(DG)method.We use the DG discretization to solve the full Reynolds-averaged Navier-St...We present the approaches to implementing the k-√k L turbulence model within the framework of the high-order discontinuous Galerkin(DG)method.We use the DG discretization to solve the full Reynolds-averaged Navier-Stokes equations.In order to enhance the robustness of approaches,some effective techniques are designed.The HWENO(Hermite weighted essentially non-oscillatory)limiting strategy is adopted for stabilizing the turbulence model variable k.Modifications have been made to the model equation itself by using the auxiliary variable that is always positive.The 2nd-order derivatives of velocities required in computing the von Karman length scale are evaluated in a way to maintain the compactness of DG methods.Numerical results demonstrate that the approaches have achieved the desirable accuracy for both steady and unsteady turbulent simulations.展开更多
We introduce a framework for statistical inference of the closure coefficients using machine learning methods.The objective of this framework is to quantify the epistemic uncertainty associated with the closure model ...We introduce a framework for statistical inference of the closure coefficients using machine learning methods.The objective of this framework is to quantify the epistemic uncertainty associated with the closure model by using experimental data via Bayesian statistics.The framework is tailored towards cases for which a limited amount of experimental data is available.It consists of two components.First,by treating all latent variables(non-observed variables)in the model as stochastic variables,all sources of uncertainty of the probabilistic closure model are quantified by a fully Bayesian approach.The probabilistic model is defined to consist of the closure coefficients as parameters and other parameters incorporating noise.Then,the uncertainty associated with the closure coefficients is extracted from the overall uncertainty by considering the noise being zero.The overall uncertainty is rigorously evaluated by using Markov-Chain Monte Carlo sampling assisted by surrogate models.We apply the framework to the Spalart-Allmars one-equation turbulence model.Two test cases are considered,including an industrially relevant full aircraft model at transonic flow conditions,the Airbus XRF1.Eventually,we demonstrate that epistemic uncertainties in the closure coefficients result into uncertainties in flow quantities of interest which are prominent around,and downstream,of the shock occurring over the XRF1 wing.This data-driven approach could help to enhance the predictive capabilities of CFD in terms of reliable turbulence modeling at extremes of the flight envelope if measured data is available,which is important in the context of robust design and towards virtual aircraft certification.The plentiful amount of information about the uncertainties could also assist when it comes to estimating the influence of the measured data on the inferred model coefficients.Finally,the developed framework is flexible and can be applied to different test cases and to various turbulence models.展开更多
The numerical simulation based on Reynolds time-averaged equation is one of the approved methods to evaluate the aerodynamic performance of trains in crosswind.However,there are several turbulence models,trains may pr...The numerical simulation based on Reynolds time-averaged equation is one of the approved methods to evaluate the aerodynamic performance of trains in crosswind.However,there are several turbulence models,trains may present different aerodynamic performances in crosswind using different turbulence models.In order to select the most suitable turbulence model,the inter-city express 2(ICE2)model is chosen as a research object,6 different turbulence models are used to simulate the flow characteristics,surface pressure and aerodynamic forces of the train in crosswind,respectively.6 turbulence models are the standard k-ε,Renormalization Group(RNG)k-ε,Realizable k-ε,Shear Stress Transport(SST)k-ω,standard k-ωand Spalart-Allmaras(SPA),respectively.The numerical results and the wind tunnel experimental data are compared.The results show that the most accurate model for predicting the surface pressure of the train is SST k-ω,followed by Realizable k-ε.Compared with the experimental result,the error of the side force coefficient obtained by SST k-ωand Realizable k-εturbulence model is less than 1%.The most accurate prediction for the lift force coefficient is achieved by SST k-ω,followed by RNG k-ε.By comparing 6 different turbulence models,the SST k-ωmodel is most suitable for the numerical simulation of the aerodynamic behavior of trains in crosswind.展开更多
Three-dimensional corner separation is a common phenomenon that significantly affects compressor performance. Turbulence model is still a weakness for RANS method on predicting corner separation flow accurately. In th...Three-dimensional corner separation is a common phenomenon that significantly affects compressor performance. Turbulence model is still a weakness for RANS method on predicting corner separation flow accurately. In the present study, numerical study of corner separation in a linear highly loaded prescribed velocity distribution （PVD） compressor cascade has been investigated using seven frequently used turbulence models. The seven turbulence models include Spalart Allmaras model, standard k-e model, realizable k-e model, standard k-to model, shear stress transport k co model, v2-fmodel and Reynolds stress model. The results of these turbulence models have been compared and analyzed in detail with available experimental data. It is found the standard k-1： model, realizable k-e model, v2-f model and Reynolds stress model can provide reasonable results for predicting three dimensional corner separation in the compressor cascade. The Spalart-Allmaras model, standard k-to model and shear stress transport k-w model overesti- mate corner separation region at incidence of 0°. The turbulence characteristics are discussed and turbulence anisotropy is observed to be stronger in the corner separating region.展开更多
Despite dedicated effort for many decades,statistical description of highly technologically important wall turbulence remains a great challenge.Current models are unfortunately incomplete,or empirical,or qualitative.A...Despite dedicated effort for many decades,statistical description of highly technologically important wall turbulence remains a great challenge.Current models are unfortunately incomplete,or empirical,or qualitative.After a review of the existing theories of wall turbulence,we present a new framework,called the structure ensemble dynamics （SED）,which aims at integrating the turbulence dynamics into a quantitative description of the mean flow.The SED theory naturally evolves from a statistical physics understanding of non-equilibrium open systems,such as fluid turbulence, for which mean quantities are intimately coupled with the fluctuation dynamics.Starting from the ensemble-averaged Navier-Stokes（EANS） equations,the theory postulates the existence of a finite number of statistical states yielding a multi-layer picture for wall turbulence.Then,it uses order functions（ratios of terms in the mean momentum as well as energy equations） to characterize the states and transitions between states.Application of the SED analysis to an incompressible channel flow and a compressible turbulent boundary layer shows that the order functions successfully reveal the multi-layer structure for wall-bounded turbulence, which arises as a quantitative extension of the traditional view in terms of sub-layer,buffer layer,log layer and wake. Furthermore,an idea of using a set of hyperbolic functions for modeling transitions between layers is proposed for a quantitative model of order functions across the entire flow domain.We conclude that the SED provides a theoretical framework for expressing the yet-unknown effects of fluctuation structures on the mean quantities,and offers new methods to analyze experimental and simulation data.Combined with asymptotic analysis,it also offers a way to evaluate convergence of simulations.The SED approach successfully describes the dynamics at both momentum and energy levels, in contrast with all prevalent approaches describing the mean velocity profile only.Moreover,the SED theoretical framework is general,independent of the flow system to study, while the actual functional form of the order functions may vary from flow to flow.We assert that as the knowledge of order functions is accumulated and as more flows are analyzed, new principles（such as hierarchy,symmetry,group invariance,etc.） governing the role of turbulent structures in the mean flow properties will be clarified and a viable theory of turbulence might emerge.展开更多
A growing interest has been devoted to the contra-rotating propellers （CRPs） due to their high propulsive efficiency, torque balance, low fuel consumption, low cavitations, low noise performance and low hull vibrati...A growing interest has been devoted to the contra-rotating propellers （CRPs） due to their high propulsive efficiency, torque balance, low fuel consumption, low cavitations, low noise performance and low hull vibration. Compared with the single-screw system, it is more difficult for the open water performance prediction because forward and aft propellers interact with each other and generate a more complicated flow field around the CRPs system. The current work focuses on the open water performance prediction of contra-rotating propellers by RANS and sliding mesh method considering the effect of computational time step size and turbulence model. The validation study has been performed on two sets of contra-rotating propellers developed by David W Taylor Naval Ship R ＆ D center. Compared with the experimental data, it shows that RANS with sliding mesh method and SST k-ω turbulence model has a good precision in the open water performance prediction of contra-rotating propellers, and small time step size can improve the level of accuracy for CRPs with the same blade number of forward and aft propellers, while a relatively large time step size is a better choice for CRPs with different blade numbers.展开更多
A second-order moment two-phase turbulence model for simulating dense gas-particle flows （USM-Θ model）, combining the unified second-order moment twophase turbulence model for dilute gas-particle flows with the kin...A second-order moment two-phase turbulence model for simulating dense gas-particle flows （USM-Θ model）, combining the unified second-order moment twophase turbulence model for dilute gas-particle flows with the kinetic theory of particle collision, is proposed. The interaction between gas and particle turbulence is simulated using the transport equation of two-phase velocity correlation with a two-time-scale dissipation closure. The proposed model is applied to simulate dense gas-particle flows in a horizontal channel and a downer. Simulation results and their comparison with experimental results show that the model accounting for both anisotropic particle turbulence and particle-particle collision is obviously better than models accounting for only particle turbulence or only particle-particle collision. The USM-Θ model is also better than the k-ε-kp-Θ model and the k-ε-kp-εp-Θ model in that the first model can simulate the redistribution of anisotropic particle Reynolds stress components due to inter-particle collision, whereas the second and third models cannot.展开更多
A new and much simpler rotation and curvature effects factor, which takes the form of Richardson number suggested by Hellsten originally for SST k-ω model, is presented for Spalart and Shur＇s rotation and curvature ...A new and much simpler rotation and curvature effects factor, which takes the form of Richardson number suggested by Hellsten originally for SST k-ω model, is presented for Spalart and Shur＇s rotation and curvature correction in the context of Spalart-Allmaras （SA） turbulence model. The new factor excludes the Lagrangian derivative of the strain rate tensor that exists in the SARC model, resulting in a simple, efficient and easy-to-implement approach for SA turbulence model （denoted as SARCM） to account for the effects of system rotation and curvature, techniquely. And then the SARCM is tested through turbulent curved wall flows： one is the flow over a zeropressure-gradient curved wall and the other is the channel flow in a duct with a U-turn. Predictions of the SARCM model are compared with experimental data and with the results obtained using original SA and SARC models. The numerical results show that SARCM can predict the rotation-curvature effects as accurately as SARC, but considerably more efficiently. Additionally, the accuracy of SARCM might strongly depend on the rotation-curvature model constants. Suggesting values for those constants are given, after some trials and errors.展开更多
It is of great significance to improve the accuracy of turbulence models in shock-wave/ boundary layer interaction flow. The relationship between the pressure gradient, as well as the shear layer, and the development ...It is of great significance to improve the accuracy of turbulence models in shock-wave/ boundary layer interaction flow. The relationship between the pressure gradient, as well as the shear layer, and the development of turbulent kinetic energy in impinging shock-wave/turbulent bound- ary layer interaction flow at Mach 2.25 is analyzed based on the data of direct numerical simulation （DNS）. It is found that the turbulent kinetic energy is amplified by strong shear in the separation zone and the adverse pressure gradient near the separation point. The pressure gradient was non-dimensionalised with local density, velocity, and viscosity. Spalart Allmaras （S A） model is modified by introducing the non-dimensional pressure gradient into the production term of the eddy viscosity transportation equation. Simulation results show that the production and dissipation of eddy viscosity are strongly enhanced by the modification of S-A model. Compared with DNS and experimental data, the wall pressure and the wall skin friction coefficient as well as the velocity profile of the modified S-A model are obviously improved. Thus it can be concluded that the mod- ification of S-A model with the pressure gradient can improve the predictive accuracy for simulat- ing the shock-wave/turbulent boundary laver interaction.展开更多
The Spalart-Allmaras （S-A） turbulence model, the shear-stress transport （SST） turbulence model and their compressibility corrections are revaluated for hypersonic compression comer flows by using high-order differ...The Spalart-Allmaras （S-A） turbulence model, the shear-stress transport （SST） turbulence model and their compressibility corrections are revaluated for hypersonic compression comer flows by using high-order difference schemes. The compressibility effect of density gradient, pressure dilatation and turbulent Mach number is accounted. In order to reduce confusions between model uncertainties and discretization errors, the formally fifth-order explicit weighted compact nonlinear scheme （WCNS-E-5） is adopted for convection terms, and a fourth-order staggered central difference scheme is applied for viscous terms. The 15° and 34° compression comers at Mach number 9.22 are investigated. Numerical results show that the original SST model is superior to the original S-A model in the resolution of separated regions and predictions of wall pressures and wall heat-flux rates. The capability of the S-A model can be largely improved by blending Catris＇ and Shur＇s compressibility corrections. Among the three corrections of the SST model listed in the present paper, Catris＇ modification brings the best results. However, the dissipation and pressure dilatation corrections result in much larger separated regions than that of the experiment, and are much worse than the original SST model as well as the other two corrections. The correction of turbulent Mach number makes the separated region slightly smaller than that of the original SST model. Some results of low-order schemes are also presented. When compared to the results of the high-order schemes, the separated regions are smaller, and the peak wall pressures and peak heat-flux rates are lower in the region of the reattachment points.展开更多
A variety of turbulence models were used to perform numerical simulations of heat transfer for hydrocarbon fuel flowing upward and downward through uniformly heated vertical pipes at supercritical pressure. Inlet temp...A variety of turbulence models were used to perform numerical simulations of heat transfer for hydrocarbon fuel flowing upward and downward through uniformly heated vertical pipes at supercritical pressure. Inlet temperatures varied from 373 K to 663 K, with heat flux rang- ing from 300 kW/m2 to 550 kW/m2. Comparative analyses between predicted and experimental results were used to evaluate the ability of turbulence models to respond to variable thermophysical properties of hydrocarbon fuel at supercritical pressure. It was found that the prediction performance of turbulence models is mainly determined by the damping function, which enables them to respond differently to local flow conditions. Although prediction accuracy for experimental results varied from condition to condition, the shear stress transport (SST) and launder and sharma models performed better than all other models used in the study. For very small buoyancy-influenced runs, the thermal-induced acceleration due to variations in density lead to the impairment of heat transfer occurring in the vicinity of pseudo-critical points, and heat transfer was enhanced at higher temperatures through the combined action of four thermophysical properties: density, viscosity, thermal conductivity and specific heat. For very large buoyancy- influenced runs, the thermal-induced acceleration effect was over predicted by the LS and AB models.展开更多
Presently developed two-phase turbulence models under-predict the gas turbulent fluctuation, because their turbulence modification models cannot fully reflect the effect of particles. In this paper, a two-time-scale d...Presently developed two-phase turbulence models under-predict the gas turbulent fluctuation, because their turbulence modification models cannot fully reflect the effect of particles. In this paper, a two-time-scale dis- sipation model of turbulence modification, developed for the two-phase velocity correlation and for the dissipation rate of gas turbulent kinetic energy, is proposed and used to simulate sudden-expansion and swirling gas-particle flows. The proposed two-time scale model gives better results than the single-time scale model. Besides, a gas tur- bulence augmentation model accounting for the finite-size particle wake effect in the gas Reynolds stress equation is proposed. The proposed turbulence modification models are used to simulate two-phase pipe flows. It can prop- erly predict both turbulence reduction and turbulence enhancement for a certain size of particles observed in ex- periments.展开更多
In previous attempts of rational subgrid-scale (SGS) modeling by employing the Kolmogorov equation of filtered (KEF) quantities, it was necessary to assume that the resolved-scale second-order structure function is st...In previous attempts of rational subgrid-scale (SGS) modeling by employing the Kolmogorov equation of filtered (KEF) quantities, it was necessary to assume that the resolved-scale second-order structure function is stationary. Forced isotropic turbulence is often used as a framework for establishing and validating such SGS models based on stationary restrictions, for it generates statistical stationary samples. However, traditional forcing method at low wavenumbers cannot provide an analytic form of forcing term for a complete KEF in physical space, which has been illustrated to be essential in the modeling of such SGS models. Thus, an alternative forcing method giving an analytic forcing term in physical space is needed for rational SGS modeling. Giving an analytic linear driving term in physical space, linearly forced isotropic turbulence should be considered an ideal theoretical framework for rational SGS modeling. In this paper, we demonstrate the feasibility of establishing a rational SGS model with stationary restriction based on linearly forced isotropic turbulence. The performance of this rational SGS model is validated. We, therefore, propose the use of linearly forced isotropic turbulence as a complement to free-decaying isotropic turbulence and low-wavenumber forced isotropic turbulence for SGS model validations.展开更多
The mechanism of local scour around submarine pipelines is studied numerically based on a renormalized group （RNG） turbulence model. To validate the numerical model, the equilibrium profiles of local scour for two c...The mechanism of local scour around submarine pipelines is studied numerically based on a renormalized group （RNG） turbulence model. To validate the numerical model, the equilibrium profiles of local scour for two cases are simulated and compared with the experimental data. It shows that the RNG turbulence model can give an appropriate prediction for the configuration of equilibrium scour hole, and it is applicable to this situation. The local scour mechanism around submarine pipelines including the flow structure, shear stress distribution and pressure field is then analyzed and compared with experiments. For further comparison and validation, especially for the flow structure, a numerical calculation employing the large eddy simulation （LES） is also conducted. The numerical results of RNG demonstrate that the critical factor governing the equilibrium profile is the seabed shear stress distribution in the case of bed load sediment transport, and the two-equation RNG turbulence model coupled with the law of wall is capable of giving a satisfying estimation for the bed shear stress. Moreover, the piping phenomena due to the great difference of pressure between the upstream and downstream parts of pipelines and the vortex structure around submarine pipelines are also simulated successfully, which are believed to be the important factor that lead to the onset of local scour.展开更多
Entropy represents the dissipation rate of energy. Through direct numerical simulation （DNS） of supersonic compression ramp flow, we find the value of entropy is monotonously decreasing along the wall-normal directi...Entropy represents the dissipation rate of energy. Through direct numerical simulation （DNS） of supersonic compression ramp flow, we find the value of entropy is monotonously decreasing along the wall-normal direction no matter in the attached or the separated region. Based on this feature, a new version of Baldwin-Lomax turbulence model （BL-entropy） is proposed in this paper. The supersonic compression ramp and cavity-ramp flows in which the original Baldwin-Lomax model fails to get convergent solutions are chosen to evaluate the performance of this model. Results from one-equation Spalart-Allmaras model （SA） and two-equation Wilcox k-x model are also included to compare with available experimental and DNS data. It is shown that BLentropy could conquer the essential deficiency of the original version by providing a more physically meaningful length scale in the complex flows. Moreover, this method is simple, computationally efficient and general, making it applicable to other models related with the supersonic boundary layer.展开更多
The internal turbulent flow in conical diffuser is a very complicated adverse pressure gradient flow.DLR k-ε turbulence model was adopted to study it.The every terms of the Laplace operator in DLR k-ε turbulence mod...The internal turbulent flow in conical diffuser is a very complicated adverse pressure gradient flow.DLR k-ε turbulence model was adopted to study it.The every terms of the Laplace operator in DLR k-ε turbulence model and pressure Poisson equation were discretized by upwind difference scheme.A new full implicit difference scheme of 5-point was constructed by using finite volume method and finite difference method.A large sparse matrix with five diagonals was formed and was stored by three arrays of one dimension in a compressed mode.General iterative methods do not work wel1 with large sparse matrix.With algebraic multigrid method(AMG),linear algebraic system of equations was solved and the precision was set at 10-6.The computation results were compared with the experimental results.The results show that the computation results have a good agreement with the experiment data.The precision of computational results and numerical simulation efficiency are greatly improved.展开更多
Cavitation typically occurs when the fluid pressure is lower than the vapor pressure at a local thermodynamic state, and the flow is frequently unsteady and turbulent. To assess the state-of-the-art of computational c...Cavitation typically occurs when the fluid pressure is lower than the vapor pressure at a local thermodynamic state, and the flow is frequently unsteady and turbulent. To assess the state-of-the-art of computational capabilities for unsteady cavitating flows, different cavitation and turbulence model combinations are conducted. The selected cavitation models include several widely-used models including one based on phenomenological argument and the other utilizing interface dynamics. The k-e turbulence model with additional implementation of the filter function and density correction function are considered to reduce the eddy viscosity according to the computed turbulence length scale and local fluid density respectively. We have also blended these alternative cavitation and lustrate that the eddy viscosity turbulence treatments, to ilnear the closure region can significantly influence the capture of detached cavity. From the experimental validations regarding the force analysis, frequency, and the cavity visualization, no single model combination performs best in all aspects. Furthermore, the implications of parameters contained in different cavitation models are investigated. The phase change process is more pronounced around the detached cavity, which is better illustrated by the interfacial dynamics model. Our study provides insight to aid further modeling development.展开更多
We propose an integrated method of data-driven and mechanism models for well logging formation evaluation,explicitly focusing on predicting reservoir parameters,such as porosity and water saturation.Accurately interpr...We propose an integrated method of data-driven and mechanism models for well logging formation evaluation,explicitly focusing on predicting reservoir parameters,such as porosity and water saturation.Accurately interpreting these parameters is crucial for effectively exploring and developing oil and gas.However,with the increasing complexity of geological conditions in this industry,there is a growing demand for improved accuracy in reservoir parameter prediction,leading to higher costs associated with manual interpretation.The conventional logging interpretation methods rely on empirical relationships between logging data and reservoir parameters,which suffer from low interpretation efficiency,intense subjectivity,and suitability for ideal conditions.The application of artificial intelligence in the interpretation of logging data provides a new solution to the problems existing in traditional methods.It is expected to improve the accuracy and efficiency of the interpretation.If large and high-quality datasets exist,data-driven models can reveal relationships of arbitrary complexity.Nevertheless,constructing sufficiently large logging datasets with reliable labels remains challenging,making it difficult to apply data-driven models effectively in logging data interpretation.Furthermore,data-driven models often act as“black boxes”without explaining their predictions or ensuring compliance with primary physical constraints.This paper proposes a machine learning method with strong physical constraints by integrating mechanism and data-driven models.Prior knowledge of logging data interpretation is embedded into machine learning regarding network structure,loss function,and optimization algorithm.We employ the Physically Informed Auto-Encoder(PIAE)to predict porosity and water saturation,which can be trained without labeled reservoir parameters using self-supervised learning techniques.This approach effectively achieves automated interpretation and facilitates generalization across diverse datasets.展开更多
基金co-supported by the Aeronautical Science Foundation of China (Nos. 20151452021and 20152752033)the National Natural Science Foundation of China (No. 61732006)
文摘High-order Discontinuous Galerkin(DG) methods have been receiving more and more attentions in the area of Computational Fluid Dynamics(CFD) because of their high-accuracy property. However, it is still a challenge to obtain converged solution rapidly when solving the Reynolds Averaged Navier–Stokes(RANS) equations since the turbulence models significantly increase the nonlinearity of discretization system. The overall goal of this research is to develop an Artificial Neural Networks(ANNs) model with low complexity acting as an algebraic turbulence model to estimate the turbulence eddy viscosity for RANS. The ANN turbulence model is off-line trained using the training data generated by the widely used Spalart–Allmaras(SA) turbulence model before the Optimal Brain Surgeon(OBS) is employed to determine the relevancy of input features.Using the selected relevant features, a fully connected ANN model is constructed. The performance of the developed ANN model is numerically tested in the framework of DG for RANS, where the‘‘DG+ANN' method provides robust and steady convergence compared to the ‘‘DG+SA' method. The results demonstrate the promising potential to develop a general turbulence model based on artificial intelligence in the future given the training data covering a large rang of flow conditions.
基金This work was supported by the National Natural Science Foundation of China(91852108,11872230 and 92152301).
文摘Data-driven turbulence modeling studies have reached such a stage that the basic framework is settled,but several essential issues remain that strongly affect the performance.Two problems are studied in the current research:(1)the processing of the Reynolds stress tensor and(2)the coupling method between the machine learning model and flow solver.For the Reynolds stress processing issue,we perform the theoretical derivation to extend the relevant tensor arguments of Reynolds stress.Then,the tensor representation theorem is employed to give the complete irreducible invariants and integrity basis.An adaptive regularization term is employed to enhance the representation performance.For the coupling issue,an iterative coupling framework with consistent convergence is proposed and then applied to a canonical separated flow.The results have high consistency with the direct numerical simulation true values,which proves the validity of the current approach.
基金supported by the National Natural Science Foundation of China(Grant Nos.92252201 and 11721202)the Fundamental Research Funds for the Central Universities.
文摘We present the approaches to implementing the k-√k L turbulence model within the framework of the high-order discontinuous Galerkin(DG)method.We use the DG discretization to solve the full Reynolds-averaged Navier-Stokes equations.In order to enhance the robustness of approaches,some effective techniques are designed.The HWENO(Hermite weighted essentially non-oscillatory)limiting strategy is adopted for stabilizing the turbulence model variable k.Modifications have been made to the model equation itself by using the auxiliary variable that is always positive.The 2nd-order derivatives of velocities required in computing the von Karman length scale are evaluated in a way to maintain the compactness of DG methods.Numerical results demonstrate that the approaches have achieved the desirable accuracy for both steady and unsteady turbulent simulations.
文摘We introduce a framework for statistical inference of the closure coefficients using machine learning methods.The objective of this framework is to quantify the epistemic uncertainty associated with the closure model by using experimental data via Bayesian statistics.The framework is tailored towards cases for which a limited amount of experimental data is available.It consists of two components.First,by treating all latent variables(non-observed variables)in the model as stochastic variables,all sources of uncertainty of the probabilistic closure model are quantified by a fully Bayesian approach.The probabilistic model is defined to consist of the closure coefficients as parameters and other parameters incorporating noise.Then,the uncertainty associated with the closure coefficients is extracted from the overall uncertainty by considering the noise being zero.The overall uncertainty is rigorously evaluated by using Markov-Chain Monte Carlo sampling assisted by surrogate models.We apply the framework to the Spalart-Allmars one-equation turbulence model.Two test cases are considered,including an industrially relevant full aircraft model at transonic flow conditions,the Airbus XRF1.Eventually,we demonstrate that epistemic uncertainties in the closure coefficients result into uncertainties in flow quantities of interest which are prominent around,and downstream,of the shock occurring over the XRF1 wing.This data-driven approach could help to enhance the predictive capabilities of CFD in terms of reliable turbulence modeling at extremes of the flight envelope if measured data is available,which is important in the context of robust design and towards virtual aircraft certification.The plentiful amount of information about the uncertainties could also assist when it comes to estimating the influence of the measured data on the inferred model coefficients.Finally,the developed framework is flexible and can be applied to different test cases and to various turbulence models.
基金Supported by National Natural Science Foundation of China(Grant No.51605397)Sichuan Provincial Science and Technology Program of China(Grant No.2019YJ0227)Self-determined Project of State Key Laboratory of Traction Power(Grant No.2019TPL_T02)
文摘The numerical simulation based on Reynolds time-averaged equation is one of the approved methods to evaluate the aerodynamic performance of trains in crosswind.However,there are several turbulence models,trains may present different aerodynamic performances in crosswind using different turbulence models.In order to select the most suitable turbulence model,the inter-city express 2(ICE2)model is chosen as a research object,6 different turbulence models are used to simulate the flow characteristics,surface pressure and aerodynamic forces of the train in crosswind,respectively.6 turbulence models are the standard k-ε,Renormalization Group(RNG)k-ε,Realizable k-ε,Shear Stress Transport(SST)k-ω,standard k-ωand Spalart-Allmaras(SPA),respectively.The numerical results and the wind tunnel experimental data are compared.The results show that the most accurate model for predicting the surface pressure of the train is SST k-ω,followed by Realizable k-ε.Compared with the experimental result,the error of the side force coefficient obtained by SST k-ωand Realizable k-εturbulence model is less than 1%.The most accurate prediction for the lift force coefficient is achieved by SST k-ω,followed by RNG k-ε.By comparing 6 different turbulence models,the SST k-ωmodel is most suitable for the numerical simulation of the aerodynamic behavior of trains in crosswind.
基金supported by the National Natural Science Foundation of China(No.51376001,No.51420105008,No.51306013,No.51136003)the National Basic Research Program of China(2012CB720205,2014CB046405)+2 种基金the Beijing Higher Education Young Elite Teacher Projectthe Fundamental Research Funds for the Central Universitiessupported by the Innovation Foundation of BUAA for Ph.D.Graduates
文摘Three-dimensional corner separation is a common phenomenon that significantly affects compressor performance. Turbulence model is still a weakness for RANS method on predicting corner separation flow accurately. In the present study, numerical study of corner separation in a linear highly loaded prescribed velocity distribution （PVD） compressor cascade has been investigated using seven frequently used turbulence models. The seven turbulence models include Spalart Allmaras model, standard k-e model, realizable k-e model, standard k-to model, shear stress transport k co model, v2-fmodel and Reynolds stress model. The results of these turbulence models have been compared and analyzed in detail with available experimental data. It is found the standard k-1： model, realizable k-e model, v2-f model and Reynolds stress model can provide reasonable results for predicting three dimensional corner separation in the compressor cascade. The Spalart-Allmaras model, standard k-to model and shear stress transport k-w model overesti- mate corner separation region at incidence of 0°. The turbulence characteristics are discussed and turbulence anisotropy is observed to be stronger in the corner separating region.
基金supported by the National Natural Science Foundation of China(90716008)the National Basic Research Program of China(2009CB724100).
文摘Despite dedicated effort for many decades,statistical description of highly technologically important wall turbulence remains a great challenge.Current models are unfortunately incomplete,or empirical,or qualitative.After a review of the existing theories of wall turbulence,we present a new framework,called the structure ensemble dynamics （SED）,which aims at integrating the turbulence dynamics into a quantitative description of the mean flow.The SED theory naturally evolves from a statistical physics understanding of non-equilibrium open systems,such as fluid turbulence, for which mean quantities are intimately coupled with the fluctuation dynamics.Starting from the ensemble-averaged Navier-Stokes（EANS） equations,the theory postulates the existence of a finite number of statistical states yielding a multi-layer picture for wall turbulence.Then,it uses order functions（ratios of terms in the mean momentum as well as energy equations） to characterize the states and transitions between states.Application of the SED analysis to an incompressible channel flow and a compressible turbulent boundary layer shows that the order functions successfully reveal the multi-layer structure for wall-bounded turbulence, which arises as a quantitative extension of the traditional view in terms of sub-layer,buffer layer,log layer and wake. Furthermore,an idea of using a set of hyperbolic functions for modeling transitions between layers is proposed for a quantitative model of order functions across the entire flow domain.We conclude that the SED provides a theoretical framework for expressing the yet-unknown effects of fluctuation structures on the mean quantities,and offers new methods to analyze experimental and simulation data.Combined with asymptotic analysis,it also offers a way to evaluate convergence of simulations.The SED approach successfully describes the dynamics at both momentum and energy levels, in contrast with all prevalent approaches describing the mean velocity profile only.Moreover,the SED theoretical framework is general,independent of the flow system to study, while the actual functional form of the order functions may vary from flow to flow.We assert that as the knowledge of order functions is accumulated and as more flows are analyzed, new principles（such as hierarchy,symmetry,group invariance,etc.） governing the role of turbulent structures in the mean flow properties will be clarified and a viable theory of turbulence might emerge.
基金supported by the National Natural Science Foundation of China(Grant No.51079157)
文摘A growing interest has been devoted to the contra-rotating propellers （CRPs） due to their high propulsive efficiency, torque balance, low fuel consumption, low cavitations, low noise performance and low hull vibration. Compared with the single-screw system, it is more difficult for the open water performance prediction because forward and aft propellers interact with each other and generate a more complicated flow field around the CRPs system. The current work focuses on the open water performance prediction of contra-rotating propellers by RANS and sliding mesh method considering the effect of computational time step size and turbulence model. The validation study has been performed on two sets of contra-rotating propellers developed by David W Taylor Naval Ship R ＆ D center. Compared with the experimental data, it shows that RANS with sliding mesh method and SST k-ω turbulence model has a good precision in the open water performance prediction of contra-rotating propellers, and small time step size can improve the level of accuracy for CRPs with the same blade number of forward and aft propellers, while a relatively large time step size is a better choice for CRPs with different blade numbers.
基金the Special Funds for Major State Basic Research of China(G-1999-0222-08)the National Natural Science Foundation of China(50376004)Ph.D.Program Foundation,Ministry of Education of China(20030007028)
文摘A second-order moment two-phase turbulence model for simulating dense gas-particle flows （USM-Θ model）, combining the unified second-order moment twophase turbulence model for dilute gas-particle flows with the kinetic theory of particle collision, is proposed. The interaction between gas and particle turbulence is simulated using the transport equation of two-phase velocity correlation with a two-time-scale dissipation closure. The proposed model is applied to simulate dense gas-particle flows in a horizontal channel and a downer. Simulation results and their comparison with experimental results show that the model accounting for both anisotropic particle turbulence and particle-particle collision is obviously better than models accounting for only particle turbulence or only particle-particle collision. The USM-Θ model is also better than the k-ε-kp-Θ model and the k-ε-kp-εp-Θ model in that the first model can simulate the redistribution of anisotropic particle Reynolds stress components due to inter-particle collision, whereas the second and third models cannot.
基金supported by National Natural Science Foundation of China (No. 11002117)
文摘A new and much simpler rotation and curvature effects factor, which takes the form of Richardson number suggested by Hellsten originally for SST k-ω model, is presented for Spalart and Shur＇s rotation and curvature correction in the context of Spalart-Allmaras （SA） turbulence model. The new factor excludes the Lagrangian derivative of the strain rate tensor that exists in the SARC model, resulting in a simple, efficient and easy-to-implement approach for SA turbulence model （denoted as SARCM） to account for the effects of system rotation and curvature, techniquely. And then the SARCM is tested through turbulent curved wall flows： one is the flow over a zeropressure-gradient curved wall and the other is the channel flow in a duct with a U-turn. Predictions of the SARCM model are compared with experimental data and with the results obtained using original SA and SARC models. The numerical results show that SARCM can predict the rotation-curvature effects as accurately as SARC, but considerably more efficiently. Additionally, the accuracy of SARCM might strongly depend on the rotation-curvature model constants. Suggesting values for those constants are given, after some trials and errors.
基金supported by the National Natural Science Foundation of China (No.11302012,51376001,51136003)the National Basic Research Program of China (No.2012CB720205)+3 种基金the National Magnetic Confinement Fusion Research Program of China (No.2012GB102006)the Aeronautical Science Foundation of China (No.2012ZB51014)the ‘‘111’’ Project(No.B08009)the Astronautical Technology Innovation Foundation of China
文摘It is of great significance to improve the accuracy of turbulence models in shock-wave/ boundary layer interaction flow. The relationship between the pressure gradient, as well as the shear layer, and the development of turbulent kinetic energy in impinging shock-wave/turbulent bound- ary layer interaction flow at Mach 2.25 is analyzed based on the data of direct numerical simulation （DNS）. It is found that the turbulent kinetic energy is amplified by strong shear in the separation zone and the adverse pressure gradient near the separation point. The pressure gradient was non-dimensionalised with local density, velocity, and viscosity. Spalart Allmaras （S A） model is modified by introducing the non-dimensional pressure gradient into the production term of the eddy viscosity transportation equation. Simulation results show that the production and dissipation of eddy viscosity are strongly enhanced by the modification of S-A model. Compared with DNS and experimental data, the wall pressure and the wall skin friction coefficient as well as the velocity profile of the modified S-A model are obviously improved. Thus it can be concluded that the mod- ification of S-A model with the pressure gradient can improve the predictive accuracy for simulat- ing the shock-wave/turbulent boundary laver interaction.
基金Foundation items： National Basic Research Program of China （2009CB723801） National Natural Science Foundation of China （11072259）
文摘The Spalart-Allmaras （S-A） turbulence model, the shear-stress transport （SST） turbulence model and their compressibility corrections are revaluated for hypersonic compression comer flows by using high-order difference schemes. The compressibility effect of density gradient, pressure dilatation and turbulent Mach number is accounted. In order to reduce confusions between model uncertainties and discretization errors, the formally fifth-order explicit weighted compact nonlinear scheme （WCNS-E-5） is adopted for convection terms, and a fourth-order staggered central difference scheme is applied for viscous terms. The 15° and 34° compression comers at Mach number 9.22 are investigated. Numerical results show that the original SST model is superior to the original S-A model in the resolution of separated regions and predictions of wall pressures and wall heat-flux rates. The capability of the S-A model can be largely improved by blending Catris＇ and Shur＇s compressibility corrections. Among the three corrections of the SST model listed in the present paper, Catris＇ modification brings the best results. However, the dissipation and pressure dilatation corrections result in much larger separated regions than that of the experiment, and are much worse than the original SST model as well as the other two corrections. The correction of turbulent Mach number makes the separated region slightly smaller than that of the original SST model. Some results of low-order schemes are also presented. When compared to the results of the high-order schemes, the separated regions are smaller, and the peak wall pressures and peak heat-flux rates are lower in the region of the reattachment points.
基金funding support from National Natural Science Foundation of China(No.51406005)Defense Industrial Technology Development Program of China(No.B2120132006)
文摘A variety of turbulence models were used to perform numerical simulations of heat transfer for hydrocarbon fuel flowing upward and downward through uniformly heated vertical pipes at supercritical pressure. Inlet temperatures varied from 373 K to 663 K, with heat flux rang- ing from 300 kW/m2 to 550 kW/m2. Comparative analyses between predicted and experimental results were used to evaluate the ability of turbulence models to respond to variable thermophysical properties of hydrocarbon fuel at supercritical pressure. It was found that the prediction performance of turbulence models is mainly determined by the damping function, which enables them to respond differently to local flow conditions. Although prediction accuracy for experimental results varied from condition to condition, the shear stress transport (SST) and launder and sharma models performed better than all other models used in the study. For very small buoyancy-influenced runs, the thermal-induced acceleration due to variations in density lead to the impairment of heat transfer occurring in the vicinity of pseudo-critical points, and heat transfer was enhanced at higher temperatures through the combined action of four thermophysical properties: density, viscosity, thermal conductivity and specific heat. For very large buoyancy- influenced runs, the thermal-induced acceleration effect was over predicted by the LS and AB models.
基金State Key Development Program for Basic Research of China (No.2006CB200305), the National Natural Sci-ence Foundation of China (No.50376004), and Ph.D. Program Foundation of Ministry of Education of China (No.20030007028).
文摘Presently developed two-phase turbulence models under-predict the gas turbulent fluctuation, because their turbulence modification models cannot fully reflect the effect of particles. In this paper, a two-time-scale dis- sipation model of turbulence modification, developed for the two-phase velocity correlation and for the dissipation rate of gas turbulent kinetic energy, is proposed and used to simulate sudden-expansion and swirling gas-particle flows. The proposed two-time scale model gives better results than the single-time scale model. Besides, a gas tur- bulence augmentation model accounting for the finite-size particle wake effect in the gas Reynolds stress equation is proposed. The proposed turbulence modification models are used to simulate two-phase pipe flows. It can prop- erly predict both turbulence reduction and turbulence enhancement for a certain size of particles observed in ex- periments.
基金the National Natural Science Foundation of China (Grant 11772128)the Fundamental Research Funds for the Central Universities (Grants 2017MS022 and 2018ZD09).
文摘In previous attempts of rational subgrid-scale (SGS) modeling by employing the Kolmogorov equation of filtered (KEF) quantities, it was necessary to assume that the resolved-scale second-order structure function is stationary. Forced isotropic turbulence is often used as a framework for establishing and validating such SGS models based on stationary restrictions, for it generates statistical stationary samples. However, traditional forcing method at low wavenumbers cannot provide an analytic form of forcing term for a complete KEF in physical space, which has been illustrated to be essential in the modeling of such SGS models. Thus, an alternative forcing method giving an analytic forcing term in physical space is needed for rational SGS modeling. Giving an analytic linear driving term in physical space, linearly forced isotropic turbulence should be considered an ideal theoretical framework for rational SGS modeling. In this paper, we demonstrate the feasibility of establishing a rational SGS model with stationary restriction based on linearly forced isotropic turbulence. The performance of this rational SGS model is validated. We, therefore, propose the use of linearly forced isotropic turbulence as a complement to free-decaying isotropic turbulence and low-wavenumber forced isotropic turbulence for SGS model validations.
基金supported by the Program for Changjiang Scholars and Innovative Research Team in University of China under contract No,IRT0420the National Natural Science Foundation of China under contract No.50409015.
文摘The mechanism of local scour around submarine pipelines is studied numerically based on a renormalized group （RNG） turbulence model. To validate the numerical model, the equilibrium profiles of local scour for two cases are simulated and compared with the experimental data. It shows that the RNG turbulence model can give an appropriate prediction for the configuration of equilibrium scour hole, and it is applicable to this situation. The local scour mechanism around submarine pipelines including the flow structure, shear stress distribution and pressure field is then analyzed and compared with experiments. For further comparison and validation, especially for the flow structure, a numerical calculation employing the large eddy simulation （LES） is also conducted. The numerical results of RNG demonstrate that the critical factor governing the equilibrium profile is the seabed shear stress distribution in the case of bed load sediment transport, and the two-equation RNG turbulence model coupled with the law of wall is capable of giving a satisfying estimation for the bed shear stress. Moreover, the piping phenomena due to the great difference of pressure between the upstream and downstream parts of pipelines and the vortex structure around submarine pipelines are also simulated successfully, which are believed to be the important factor that lead to the onset of local scour.
基金National Basic Research Program of China(No.2009CB724104)the Innovation Foundation of BUAA for Ph.D.Graduates(No.300521)
文摘Entropy represents the dissipation rate of energy. Through direct numerical simulation （DNS） of supersonic compression ramp flow, we find the value of entropy is monotonously decreasing along the wall-normal direction no matter in the attached or the separated region. Based on this feature, a new version of Baldwin-Lomax turbulence model （BL-entropy） is proposed in this paper. The supersonic compression ramp and cavity-ramp flows in which the original Baldwin-Lomax model fails to get convergent solutions are chosen to evaluate the performance of this model. Results from one-equation Spalart-Allmaras model （SA） and two-equation Wilcox k-x model are also included to compare with available experimental and DNS data. It is shown that BLentropy could conquer the essential deficiency of the original version by providing a more physically meaningful length scale in the complex flows. Moreover, this method is simple, computationally efficient and general, making it applicable to other models related with the supersonic boundary layer.
基金Projects(59375211,10771178,10676031) supported by the National Natural Science Foundation of ChinaProject(07A068) supported by the Key Project of Hunan Education CommissionProject(2005CB321702) supported by the National Key Basic Research Program of China
文摘The internal turbulent flow in conical diffuser is a very complicated adverse pressure gradient flow.DLR k-ε turbulence model was adopted to study it.The every terms of the Laplace operator in DLR k-ε turbulence model and pressure Poisson equation were discretized by upwind difference scheme.A new full implicit difference scheme of 5-point was constructed by using finite volume method and finite difference method.A large sparse matrix with five diagonals was formed and was stored by three arrays of one dimension in a compressed mode.General iterative methods do not work wel1 with large sparse matrix.With algebraic multigrid method(AMG),linear algebraic system of equations was solved and the precision was set at 10-6.The computation results were compared with the experimental results.The results show that the computation results have a good agreement with the experiment data.The precision of computational results and numerical simulation efficiency are greatly improved.
基金supported by the National Natural Science Foundation of China (10802026)
文摘Cavitation typically occurs when the fluid pressure is lower than the vapor pressure at a local thermodynamic state, and the flow is frequently unsteady and turbulent. To assess the state-of-the-art of computational capabilities for unsteady cavitating flows, different cavitation and turbulence model combinations are conducted. The selected cavitation models include several widely-used models including one based on phenomenological argument and the other utilizing interface dynamics. The k-e turbulence model with additional implementation of the filter function and density correction function are considered to reduce the eddy viscosity according to the computed turbulence length scale and local fluid density respectively. We have also blended these alternative cavitation and lustrate that the eddy viscosity turbulence treatments, to ilnear the closure region can significantly influence the capture of detached cavity. From the experimental validations regarding the force analysis, frequency, and the cavity visualization, no single model combination performs best in all aspects. Furthermore, the implications of parameters contained in different cavitation models are investigated. The phase change process is more pronounced around the detached cavity, which is better illustrated by the interfacial dynamics model. Our study provides insight to aid further modeling development.
基金supported by National Key Research and Development Program (2019YFA0708301)National Natural Science Foundation of China (51974337)+2 种基金the Strategic Cooperation Projects of CNPC and CUPB (ZLZX2020-03)Science and Technology Innovation Fund of CNPC (2021DQ02-0403)Open Fund of Petroleum Exploration and Development Research Institute of CNPC (2022-KFKT-09)
文摘We propose an integrated method of data-driven and mechanism models for well logging formation evaluation,explicitly focusing on predicting reservoir parameters,such as porosity and water saturation.Accurately interpreting these parameters is crucial for effectively exploring and developing oil and gas.However,with the increasing complexity of geological conditions in this industry,there is a growing demand for improved accuracy in reservoir parameter prediction,leading to higher costs associated with manual interpretation.The conventional logging interpretation methods rely on empirical relationships between logging data and reservoir parameters,which suffer from low interpretation efficiency,intense subjectivity,and suitability for ideal conditions.The application of artificial intelligence in the interpretation of logging data provides a new solution to the problems existing in traditional methods.It is expected to improve the accuracy and efficiency of the interpretation.If large and high-quality datasets exist,data-driven models can reveal relationships of arbitrary complexity.Nevertheless,constructing sufficiently large logging datasets with reliable labels remains challenging,making it difficult to apply data-driven models effectively in logging data interpretation.Furthermore,data-driven models often act as“black boxes”without explaining their predictions or ensuring compliance with primary physical constraints.This paper proposes a machine learning method with strong physical constraints by integrating mechanism and data-driven models.Prior knowledge of logging data interpretation is embedded into machine learning regarding network structure,loss function,and optimization algorithm.We employ the Physically Informed Auto-Encoder(PIAE)to predict porosity and water saturation,which can be trained without labeled reservoir parameters using self-supervised learning techniques.This approach effectively achieves automated interpretation and facilitates generalization across diverse datasets.
