Fully developed turbulent flow in channels with mild to strong longitudinal curvature is studied by direct numerical simulations. The Reynolds based on the bulk mean velocity and channel half-width delta is fixed at 20 000, resulting in a friction Reynolds number of approximately 1000. Four cases are considered with curvature varying from gamma = 2 delta/r

Heat and mass transport are generally closely correlated to momentum transport in shear flows. This so-called Reynolds analogy between advective heat or mass transport and momentum transport hinders efficiency improvements in engineering heat and mass transfer applications. I show through direct numerical simulations that in plane Couette and Taylor-Couette flow, rotation can strongly influence wall-to-wall passive tracer transport and make it much faster than momentum transport, clearly in violation of the Reynolds analogy. This difference between passive tracer transport, representative of heat/mass transport, and momentum transport is observed in steady flows with large counter-rotating vortices at low Reynolds numbers as well as in fully turbulent flows at higher Reynolds numbers. It is especially large near the neutral (Rayleigh's) stability limit. The rotation-induced Coriolis force strongly damps the streamwise/azimuthal velocity fluctuations when this limit is approached, while tracer fluctuations are much less affected. Accordingly, momentum transport is much more reduced than tracer transport, showing that the Coriolis force breaks the Reynolds analogy. At higher Reynolds numbers, this strong advective transport dissimilarity is accompanied by approximate limit cycle dynamics with intense low-frequency bursts of turbulence when approaching the neutral stability limit. The study demonstrates that simple body forces can cause clear dissimilarities between heat/mass and momentum transport in shear flows.

In the present work the evolution of the boundary layer over a low-pressureturbine blade is studied by means of direct numerical simulations. The set-upof the simulations follows the experiments by Lengani et al. (2017), aimingto investigate the unsteady flow field induced by the rotor-stator interaction.The free-stream flow is characterized by high level of free-stream turbulenceand periodically impinging wakes. As in the experiments, the wakes are shedby moving bars modeling the rotor blades and placed upstream of the turbineblades. To include the presence of the wake without employing an ad-hoc model,we simulate both the moving bars and the stationary blades in their respectiveframes of reference and the coupling of the two domains is done throughappropriate boundary conditions. The presence of the wake mainly affects thedevelopment of the boundary layer on the suction side of the blade. In particular,the flow separation in the rear part of the blade is suppressed. Moreover, thepresence of the wake introduces alternating regions in the streamwise direction ofhigh- and low-velocity fluctuations inside the boundary layer. These fluctuationsare responsible for significant variations of the shear stress. The analysis of thevelocity fields allows the characterization of the streaky structures forced inthe boundary layer by turbulence carried by upstream wakes. The breakdownevents are observed once positive streamwise velocity fluctuations reach theend of the blade. Both the fluctuations induced by the migration of the wakein the blade passage and the presence of the streaks contribute to high valuesof the disturbance velocity inside the boundary layer with respect to a steadyinflow case. The amplification of the boundary layer disturbances associatedwith different spanwise wavenumbers has been computed. It was found thatthe migration of the wake in the blade passage stands for the most part of theperturbations with zero spanwise wavenumber. The non-zero wavenumbers arefound to be amplified in the rear part of the blade at the boundary betweenthe low and high speed regions associated with the wakes.

In this work, the free-stream turbulence (FST) induced transition of a flat plate boundary layer is studied using particle image velocimetry (PIV) under variable Reynolds number (Re), FST intensity, and adverse pressure gradient (APG). Overall, 10 different flow conditions were tested concerning the variation of these parameters. The streak spacing and the probability density function (PDF) of turbulent spot nucleation are computed for all cases. The streak spacing is shown to be constant in the transition region once scaled with the turbulent displacement and momentum thickness, with resulting values of around 3 and 5, respectively. Nucleation events are shown to occur near the position where the dimensionless streak spacing reaches such constant values. The streamwise position where most turbulent spots are formed is strongly influenced by the FST intensity level. Additionally, the PDF of spot nucleation becomes narrower with increase in the APG, while FST has the opposite effect. A common distribution of all the PDFs is provided as a function of a similarity variable accounting for the streak spacing, the shape factor of the boundary layer, and the FST intensity.

This thesis considers the generation and influence of free-stream turbulence toboundary layer transition on both flat and curved bodies in the flow. Variousflow configurations such as flow around the flat plate with a sharp leading edgeand low-pressure turbine blades are considered. This study aims at contributingto a better understanding of stability characteristics and different transitionmechanisms in such flows, which are of great interest for fundamental andindustrial applications.In the first part of the thesis, we study the effects of the free-streamturbulence characteristic length scales and intensity on the transition in anincompressible flat-plate boundary layer through direct numerical simulations(DNS). Computations are performed using the spectral element code Nek5000.The numerical setup corresponds to the experimental investigations by Fransson & Shahinfar (2020). Numerically generated homogeneous isotropic turbulenceupstream of the leading edge is designed to reproduce the characteristics of thegrid-generated turbulence in the wind tunnel experiments. Various combinationsof integral length scales are simulated. To ensure the quality of the data, classicalturbulence statistics and integral quantities are carefully evaluated, showingclose agreement with the corresponding experimental data.In the second part, we study both the effect of the free-stream turbulencelevel and the effect of the wake on the low-pressure turbine blades. Thehomogeneous and isotropic free-stream turbulence is prescribed at the inlet asa superposition of Fourier modes with a random phase shift. In the secondstage of the study, cylinders moving in front of the leading edge of the turbineare included to model the effect of the wake coming from the upstream blade.That is done using the tool NekNek which simultaneously runs two differentsimulations that communicate with each other at each time-step through aspecific boundary condition.We also analysed laminar/turbulent regions in the boundary layer flow forboth cases mentioned earlier. To achieve this, we proposed a topology-basedmethod based on extracting the extrema of the flow data. The goal was topropose a method to reduce the subjective choices to a minimum and provideefficient results regardless of the chosen flow case.

The characteristics of the incoming turbulence are known to significantly affect the aerodynamic performance of immersed bodies and highly contribute to the overall losses. Due to the broad applicability of free-stream turbulence for defining initial and boundary conditions used in CFD, giving necessary attention to the generation of synthetic turbulent fields is essential for the more accurate development process. The present paper describes a synthetic homogeneous isotropic free-stream turbulence generation method. The method is validated using direct numerical simulations of a doubly periodic streamwise evolving channel, representing the free stream. It is demonstrated that input specification of turbulence intensity and turbulence integral length scale can reproduce realistic and self-consistent turbulence structures of the desired spectrum. By varying numerical and physical parameters, we show free-stream turbulence's spatial and temporal development. This method can be used as an inlet boundary for different types of flow found in industrial applications, like wings and turbine blades.

Free-stream turbulence (FST) and its effect on boundary-layer transition is a complex multiscale problem. Under action of FST, elongated streamwise streaky structures are generated inside the boundary layer, and their amplitude and wavelength are crucial for the transition onset. While turbulence intensity is strongly correlated with the transitional Reynolds number, characteristic length scales of the FST are often considered to have a slight impact on the transition location. Conversely, a recent experiment by Fransson & Shahinfar (2020) shows significant effects of FST scales. They found that, for low values of turbulence intensity, an increase in length scale advances transition, which agrees with literature. However, for high turbulence intensities, an increase in length scale postpones transition. Here, we aim at physically understanding and verifying the results of Fransson & Shahinfar (2020) by performing a series of high-fidelity simulations. These results provide understanding why the FST integral length scale affects the transition location differently depending on intensity. Knowing this relation is crucial for the development of transition models, which are commonly used in turbomachinery and aeronautics. A correct transition point is essential as all boundary layer properties, including friction and heat-transfer coefficients, drastically change from laminar to turbulent, and thus fundamentally affects the complete flow.

The transition from laminar to turbulent flow is a long-standing research subjectin the field of fluid mechanics. A basic necessity for such studies is a distinctionbetween laminar and turbulent flow. In particular, a binary segmentation isdesired where laminar and turbulent regions behave consistently over time.Previous work in this regard yield inconsistent results, or are restricted to 2Dmanifolds thereby neglecting the three-dimensional nature of the problem. Inthis paper, we present a novel use of feature-based methods to segmenting a3D time-dependent flow into regions of laminar and turbulent behavior. It isbased on the extraction of local extrema from a scalar field such as spanwisevelocity. It turns out that the existence of a large number of extrema in aregion is a good indicator for turbulence. We derive a density function fromthe extracted extrema using a Kernel Density Estimate (KDE) and thresholdit to achieve a binary segmentation into laminar and turbulent regions. Ourmethod shows consistent results and enables the domain scientists to study thethree-dimensional aspects of the laminar-turbulent transition that were difficultto assess before.

For turbulent flows that are influenced by an active scalar, the Reynolds stresses and scalar flux are coupled in a complicated way, which makes it difficult to model these flows. A framework has been derived for obtaining explicit algebraic Reynolds-stress and scalar-flux models for two-dimensional mean flows with stratification. For the specific case of stably stratified parallel shear flows, the derived model was shown to give good results. As an extension of these results, two more cases are considered: unstable stratification in a horizontal channel and natural convection in a vertical channel. 

In the present work, the laminar-turbulent transition of the flow evolving arounda low-pressure turbine blade has been investigated. Direct numerical simulationshave been carried out for two different free-stream turbulence intensity (FSTI)levels to investigate the role of free-stream oscillations on the evolution of theblade boundary layer. Emphasis is posed on identifying the mechanisms drivingthe formation and breakup of coherent structures in the high FSTI case andhow these processes are affected by the leading-edge receptivity and/or bythe continuous forcing in the blade passage. Proper orthogonal decomposition(POD) has been adopted to provide a clear statistical representation of theshape of the structures. Extended POD projections provided temporal andspanwise correlations that allowed us to identify dominant temporal structuresand spanwise wavelengths in the transition process.The extended POD analysis shows that the structures on the pressure sideare not related to what happens at the leading edge. The results on the suctionside show that the modes defining the leading edge and the passage basescorrelate with coherent structures responsible for the transition. The mostenergetic mode of the passage basis is strongly related to the most amplifiedwavelength in the boundary layer and breakup events leading to transition.Modes with a smaller spanwise wavelength belong to the band predicted byoptimal disturbance theory, they amplify with a smaller gain in the rear suctionside, and they show the highest degree of correlation between the passage regionand the rear suction side.

This paper consists of three parts. In the first part, we demonstrate the performance of the explicit algebraic (EA) subgrid-scale (SGS) stress model at Reτ = 934 and Reτ = 2003, based on friction velocity and channel half-width, for the case of large eddy simulation (LES) of turbulent channel flow. Performance of the EA model is compared to that of the dynamic Smagorinsky (DS) model for four different coarse resolutions and statistics are compared to the DNS of del Álamo & Jiménez (2003) and Hoyas & Jiménez (2008). Mean velocity profiles and Reynolds stresses are presented for the different cases. The EA model predictions are found to be reasonably close to the DNS profiles at all resolutions, while the DS model predictions are only in agreement at the finest resolution. The EA model predictions are found to be less resolution dependent than those with the DS model at both Reynolds numbers. In the second and third parts, we use Langevin stochastic differential equations to extend the EA model with stochastic contributions for SGS stresses and scalar fluxes. LES of turbulent channel flow at Reτ = 590, including a passive scalar, is carried out using the stochastic EA (SEA) models and the results are compared to the EA model predictions as well as DNS data. Investigations, show that the SEA model provides for a reasonable amount of backscatter of energy both for velocity and scalar, while the EA models do not provide for backscatter. The SEA model also improves the variance and length-scale of the SGS dissipation for velocity and scalar. However, the resolved statistics like the mean velocity, temperature, Reynolds stresses and scalar fluxes are hardly affected by the inclusion of the stochastic terms.

The atmospheric boundary layer (ABL) is in continuous turbulent motion. The heating and cooling of the Earth’s surface drives mechanic and thermodynamic processes in the ABL through enhancing and damping of atmospheric turbulence. The surface forcing has a profound effect on the diurnal cycle of temperature,wind and related variables in the ABL. Efforts have been made to model atmospheric turbulence with linear algebraic relations such as the eddy-viscosity hypothesis. Modelling of atmospheric turbulence, however, still remains a great challenge and forms an important problem in the context of numerical weather prediction and climate models. In this thesis a recently developed non-linear turbulence model, the so-called explicit algebraic Reynolds-stress (EARS) model, implemented in the context of a single-column model is used to simulate dry, stratified ABLs.

We propose a new boundary-condition treatment in the EARS model. The boundary conditions correspond to the relations for vanishing buoyancy effects that are valid close to the ground. In the simulation of an idealized diurnal cycle the solutions for the stratified surface layer is in agreement with the surface scaling physics and the Monin–Obukhov functions.

We have carried out simulations of the ABL with varying levels of stratification using the EARS model implemented in the context of a single-column model. We use the same model formulation and coefficients in these simulations with different thermal stratifications of the ABL. Even in the SCM formulation the EARS model solution produces a full Reynolds-stress tensor and heat flux vector. The set-up of the numerical experiments are taken from previously published large-eddy simulation (LES) studies of ABL.

Simulations of stably-stratified ABL show that the EARS model is able to accurately predict the development of a low-level jet and wind turning for different levels of stratification. In addition to first-order statistics, the model also predicts more intricate features of the turbulent ABL such as the relation between vertical and horizontal fluctuations for different stratifications and horizontal heat fluxes caused by wind shear. In the simulations of convective ABL the EARS model correctly predicts the horizontal wind speed and potential temperature profiles. The study also shows that the non-gradient term in the vertical heat flux equation, that naturally appears in the model formulations, gives a large contribution to the heat flux and has a significant influence on the predicted potential temperature profile of the convective ABL. Finally, we study the effects of transitional turbulence in the simulation of diurnal cycle extended to several days. The comparison with the LES shows that the EARS model correctly predicts the mean profiles and surface fluxes at different times of the day, including the low-level jet close to the surface. The model also predicts residual turbulence near the top of the ABL at night. The study demonstrates that the EARS model is able to capture key features of stably-stratified and convective ABLs as well as transitional processes that drive the ABL from one stratification to another.