Modeling of complex flows involving the combined effects of flow transition and streamline curvature using two advanced turbulence models, one in the Reynolds-averaged Navier-Stokes (RANS) category and the other in the hybrid RANS-Large eddy simulation (LES) category is considered in this research effort. In the first part of the research, a new scalar eddy-viscosity model (EVM) is proposed, designed to exhibit physically correct responses to flow transition, streamline curvature, and system rotation effects. The four equation model developed herein is a curvature-sensitized version of a commercially available three-equation transition-sensitive model. The physical effects of rotation and curvature (RC) enter the model through the added transport equation, analogous to a transverse turbulent velocity scale. The eddy-viscosity has been redefined such that the proposed model is constrained to reduce to the original transition-sensitive model definition in nonrotating flows or in regions with negligible RC effects. In the second part of the research, the developed four-equation model is combined with a LES technique using a new hybrid modeling framework, dynamic hybrid RANS-LES. The new framework is highly generalized, allowing coupling of any desired LES model with any given RANS model and addresses several deficiencies inherent in most current hybrid models. In the present research effort, the DHRL model comprises of the proposed four-equation model for RANS component and the MILES scheme for LES component. Both the models were implemented into a commercial computational fluid dynamics (CFD) solver and tested on a number of engineering and generic flow problems. Results from both the RANS and hybrid models show successful resolution of the combined effects of transition and curvature with reasonable engineering accuracy, and for only a small increase in computational cost. In addition, results from the hybrid model indicate significant levels of turbulent fluctuations in the flowfield, improved accuracy compared to RANS models predictions, and are obtained at a significant reduction of computational cost compared to full LES models. The results suggest that the advanced turbulence modeling techniques presented in this research effort have potential as practical tools for solving low/high Re flows over blunt/curved bodies for the prediction of transition and RC effects.

No description available.

In recent years, airports have experienced increasing airport congestion, partially due to the hub-and-spoke model on which airline operations are based. Current airline operations utilize large airports, focusing traffic to a small number of airports. One way to relieve such congestion is to transition to a more accessible and efficient point-to-point operation, which utilizes a large web of smaller airports. This expansion to regional airports propagates the need for next-generation low-noise aircraft with short take-off and landing capabilities. NASA has attacked this problem with a high-lift, low-noise concept dubbed the Cruise Efficient Short Take-Off and Landing (CESTOL) aircraft. The goal of the CESTOL project is to produce aircraft designs that can further expand the air travel industry to currently untapped regional airports. One method of obtaining a large lifting capability with low noise production is to utilize circulation control (CC) technology. CC is an active flow control approach that makes use of the Coanda effect. A high speed jet of air is blown over a wing flap and/or the leading edge of the wing, which entrains the freestream flow and effectively increases circulation around the wing. A promising tool for predicting CESTOL aircraft performance is computational fluid dynamics (CFD,) due to the relatively low cost and easy implementation in the design process. However, the unique flows that CC introduces are not well understood, and traditional turbulence modeling does not correctly resolve these complex flows (including high speed jet flow, complex shear flows and mixing phenomena, streamline curvature, and other challenging flow phenomena). The recent derivation of the v2-f turbulence model shows theoretical promise in increasing the accuracy of CFD predictions for CC flows, but this has not yet been assessed in great detail. This paper presents a methodical verification of several variations on the v2-f turbulence model. These models are verified using simple, well-understood flows. Results for CC flows are compared to those obtained with more traditional turbulence modeling techniques (including the Spalart-Allmaras, k-ε, and k-ω turbulence models). Wherever possible, computed results are compared to experimental data and more accurate numerical methods. Results indicate that the v2-f turbulence models predict some aspects of circulation control flow fields quite well, in particular the lift coefficient. The linear v2-f, nonlinear v2-f, and nonlinear v2-f-cc turbulence models have generated lift coefficients within 19%, 14%, and -26%, respectively of experimental values, whereas the Spalart-Allmaras, k-ε, and k-ω turbulence models produce errors as high as 85%, 36%, and 39%, respectively. The predicted stagnation points and pressure coefficient distributions match experimental data roughly as well as standard turbulence models do, though the modeling of these aspects of the flow do show some room for improvement. The nonlinear v2-f-cc turbulence model shows very non-physical skin friction coefficient profiles, pressure coefficient profiles, and stagnation points, indicating that the streamline curvature correction terms need attention. Regardless of the source of the discrepancies, the v2-f turbulence models show promise in the modeling of circulation control flow fields, but are not quite ready for application in the design of circulation control aircraft.

There is increasing interest in design methods and performance prediction for turbine engines operating at low Reynolds numbers. In this regime, boundary layer separation may be more likely to occur in the turbine flow passages. For accurate CFD predictions of the flow, correct modeling of laminar-turbulent boundary layer transition is essential to capture the details of the flow. To investigate possible improvements in model fidelity, both two-dimensional and three-dimensional CFD models were created for the flow over several low pressure turbine blade designs. A new three-equation eddy-viscosity type turbulent transitional flow model originally developed by Walters and Leylek was employed for the current RANS CFD calculations. Flows over three low pressure turbine blade airfoils with different aerodynamic characteristics were simulated over a Reynolds number range of 15,000-100,000, and predictions were compared to experiments. The turbulent transitional flow model sensitivity to inlet turbulent flow parameters showed a dependence on free-stream turbulence intensity and turbulent length scale. Using the total pressure loss coefficient as a measurement of aerodynamic performance, the Walters and Leylek transitional flow model produced adequate prediction of the Reynolds number performance in the Lightly Loaded blade. Furthermore, the correct qualitative flow response to separated shear layers was observed for the Highly Loaded blade. The vortex shedding produced by the separated flow was largely two-dimensional with small spanwise variations in the separation region. The blade loading and separation location was sufficiently predicted for the Aft-Loaded L1A blade flowfield. Investigations of the unsteady flowfield of the Aft-Loaded L1A blade showed the shear layer produced a large separation region on the suction surface. This separation region was located more downstream and significantly reduced in size when impinged upon by the upstream wakes, thus improving the aerodynamic performance consistent with experiments. For all cases investigated, the Walters and Leylek transitional flow model was judged to be sufficient for understanding the separation and transition characteristics, and superior to other widely-used turbulence models in accuracy of describing the details of the transitional and separated flow. This research characterized and assessed a new model for low Reynolds number turbine aerodynamic flow prediction and design improvement. / Ph. D.

In this work, a three-dimensional turbulent boundary layer experiment was set up with alternating stream-wise and span-wise pressure gradients. The pressure gradients are generated as a result of the test section wavy side wall shape. Each side had six sine waves with a trough to peak magnitude to wavelength ratio of 0.25. Boundary layer control was used so that the flow over the side walls remains attached. The mean flow velocity components, static and total pressures were measured at six plane along the stream-wise direction. The alternating mean span-wise and stream-wise pressure gradients created alternating stream-wise and span-wise vorticity fluxes, respectively, along the test section. As the flow developed downstream the vorticity created at the tunnel floor and ceiling diffused away from the wall. The vorticity components in the stream-wise and span-wise directions are strengthened due to stretching and tilting terms in the vorticity transport equations. The positive-z half of the test section contains large areas that generate positive vorticity flux in the trough region and smaller areas generating negative vorticity around the wave peak. The opposite is true for the negative-z half of the test-section. This results in a large positive stream-wise vorticity in the positive-z half and negative stream-wise vorticity in the negative-z half of the test-section. The smaller regions of opposite sign vorticity in each half tend to mix the flow such that as they diffuse away from the wall, the turbulent stresses are more uniform. Turbulent fluctuating velocity components were measured using Laser Doppler Velocimetery. Mean velocities as well as Reynolds stresses and triple velocity component correlations were measured at thirty stations along the last wave in the test section. Profiles at the center of the test section showed three dimensionality, but exhibited high turbulence intensities in the outer layer. Profiles off the test section centerline are highly three dimensional with multiple peaks in the normal stress profiles. The flow also reaches a state where all the normal stresses have equal magnitudes while the shear stresses are non-zero. Flow angles, flow gradient angles and shear stress angles show very large differences between wall values and outer layer vlaues. The shear stress angle lagged the flow gradient angle indicating non-equilibrium. A turbulent kinetic energy transport budget is performed for all profiles and the turbulence kinetic energy dissipation rate is estimated. Spectral measurements were also made and an independent estimate of the kinetic energy dissipation rate is made. These estimates agree very well with those estimates made by balancing the turbulence kinetic energy transport equation. Multiple turbulent diffusion models are compared to measured quantities. The models varied in agreement with experimental data. However, fair agreement with turbulence kinetic energy turbulent diffusion is observed. A model for the dissipation rate tensor anisotropy is used to extract estimates of the pressure-strain tensor from the Reynolds stress transport equations. The pressure-strain estimates are compared with some of the models in the literature. The comparison showed poor agreement with estimated pressure-strain values extracted from experimental data. A tentative model for the turbulent Reynolds shear stress angle is developed that captures the shear stress angle near wall behavior to a very good extent. The model contains one constant that is related to mean flow variables. However, the developed expression needs modification so that the prediction is improved along the entire boundary layer thickness. / Ph. D.

No description available.

No description available.

No description available.

Computational fluid dynamics (CFD) has been widely used to simulate turbulent flows. Although an increased availability of computational resources has enabled high-fidelity simulations (e.g. large eddy simulation and direct numerical simulation) of turbulent flows, the Reynolds-Averaged Navier-Stokes (RANS) equations based models are still the dominant tools for industrial applications. However, the predictive capability of RANS models is limited by potential inaccuracies driven by hypotheses in the Reynolds stress closure. With the ever-increasing use of RANS simulations in mission-critical applications, the estimation and reduction of model-form uncertainties in RANS models have attracted attention in the turbulence modeling community. In this work, I focus on estimating uncertainties stemming from the RANS turbulence closure and calibrating discrepancies in the modeled Reynolds stresses to improve the predictive capability of RANS models. Both on-line and off-line data are utilized to achieve this goal. The main contributions of this dissertation can be summarized as follows: First, a physics-based, data-driven Bayesian framework is developed for estimating and reducing model-form uncertainties in RANS simulations. An iterative ensemble Kalman method is employed to assimilate sparse on-line measurement data and empirical prior knowledge for a full-field inversion. The merits of incorporating prior knowledge and physical constraints in calibrating RANS model discrepancies are demonstrated and discussed. Second, a random matrix theoretic framework is proposed for estimating model-form uncertainties in RANS simulations. Maximum entropy principle is employed to identify the probability distribution that satisfies given constraints but without introducing artificial information. Objective prior perturbations of RANS-predicted Reynolds stresses in physical projections are provided based on comparisons between physics-based and random matrix theoretic approaches. Finally, a physics-informed, machine learning framework towards predictive RANS turbulence modeling is proposed. The functional forms of model discrepancies with respect to mean flow features are extracted from the off-line database of closely related flows based on machine learning algorithms. The RANS-modeled Reynolds stresses of prediction flows can be significantly improved by the trained discrepancy function, which is an important step towards the predictive turbulence modeling. / Ph. D.

Reynolds-Averaged Navier-Stokes (RANS) simulations are widely used for engineering design and analysis involving turbulent flows. In RANS simulations, the Reynolds stress needs closure models and the existing models have large model-form uncertainties. Therefore, the RANS simulations are known to be unreliable in many flows of engineering relevance, including flows with three-dimensional structures, swirl, pressure gradients, or curvature. This lack of accuracy in complex flows has diminished the utility of RANS simulations as a predictive tool for engineering design, analysis, optimization, and reliability assessments. Recently, data-driven methods have emerged as a promising alternative to develop the model of Reynolds stress for RANS simulations. In this dissertation I explore two physics-informed, data-driven frameworks to improve RANS modeled Reynolds stresses. First, a Bayesian inference framework is proposed to quantify and reduce the model-form uncertainty of RANS modeled Reynolds stress by leveraging online sparse measurement data with empirical prior knowledge. Second, a machine-learning-assisted framework is proposed to utilize offline high-fidelity simulation databases. Numerical results show that the data-driven RANS models have better prediction of Reynolds stress and other quantities of interest for several canonical flows. Two metrics are also presented for an a priori assessment of the prediction confidence for the machine-learning-assisted RANS model. The proposed data-driven methods are also applicable to the computational study of other physical systems whose governing equations have some unresolved physics to be modeled. / Ph. D. / Reynolds-Averaged Navier–Stokes (RANS) simulations are widely used for engineering design and analysis involving turbulent flows. In RANS simulations, the Reynolds stress needs closure models and the existing models have large model-form uncertainties. Therefore, the RANS simulations are known to be unreliable in many flows of engineering relevance, including flows with three-dimensional structures, swirl, pressure gradients, or curvature. This lack of accuracy in complex flows has diminished the utility of RANS simulations as a predictive tool for engineering design, analysis, optimization, and reliability assessments. Recently, data-driven methods have emerged as a promising alternative to develop the model of Reynolds stress for RANS simulations. In this dissertation I explore two physics-informed, data-driven frameworks to improve RANS modeled Reynolds stresses. First, a Bayesian inference framework is proposed to quantify and reduce the model-form uncertainty of RANS modeled Reynolds stress by leveraging online sparse measurement data with empirical prior knowledge. Second, a machine-learning-assisted framework is proposed to utilize offline high fidelity simulation databases. Numerical results show that the data-driven RANS models have better prediction of Reynolds stress and other quantities of interest for several canonical flows. Two metrics are also presented for an a priori assessment of the prediction confidence for the machine-learning-assisted RANS model. The proposed data-driven methods are also applicable to the computational study of other physical systems whose governing equations have some unresolved physics to be modeled.