<p>Experimental findings show that transition from laminar toturbulent ow may occur also if the exponentially growingperturbations, eigensolutions to the linearised disturbanceequations, are damped. An alternative non-modal growthmechanism has been recently identi fied, also based on thelinear approximation. This consists of the transient growth ofstreamwise elongated disturbances, with regions of positive andnegative streamwise velocity alternating in the spanwisedirection, called streaks. These perturbation are seen toappear in boundary layers exposed to signi ficant levels offree-stream turbulence. The effect of the streaks on thestability and transition of the Blasius boundary layer isinvestigated in this thesis. The analysis considers the steadyspanwise-periodic streaks arising from the nonlinear evolutionof the initial disturbances leading to the maximum transientenergy growth. In the absence of streaks, the Blasius pro filesupports the viscous exponential growth of theTollmien-Schlichting waves. It is found that increasing thestreak amplitude these two-dimensional unstable waves evolveinto three-dimensional spanwiseperiodic waves which are lessunstable. The latter can be completely stabilised above athreshold amplitude. Further increasing the streak amplitude,the boundary layer is again unstable. The new instability is ofdifferent character, being driven by the inectional pro filesassociated with the spanwise modulated ow. In particular, it isshown that, for the particular class of steady streaksconsidered, the most ampli fied modes are antisymmetric andlead to spanwise oscillations of the low-speed streak (sinuousscenario). The transition of the streak is then characterisedby the appearance of quasi-streamwise vorticesfollowing themeandering of the streak.</p><p>Simulations of a boundary layer subjected to high levels offree-stream turbulence have been performed. The receptivity ofthe boundary layer to the external perturbation is studied indetail. It is shown that two mechanisms are active, a linearand a nonlinear one, and their relative importance isdiscussed. The breakdown of the unsteady asymmetric streaksforming in the boundary layer under free-stream turbulence isshown to be characterised by structures similar to thoseobserved both in the sinuous breakdown of steady streaks and inthe varicose scenario, with the former being the mostfrequently observed.</p><p><b>Keywords:</b>Fluid mechanics, laminar-turbulent transition,boundary layer ow, transient growth, streamwise streaks,lift-up effect, receptivity, free-stream turbulence, secondaryinstability, Direct Numerical Simulation.</p>

The later stages of separated shear layer transition within separation bubbles developing over a NACA0018 airfoil operating at a chord Reynolds number of 105 and at angles of attack of 0, 5, 8, and 10 degrees were investigated experimentally in a wind tunnel. Several experimental tools, including a rake of six boundary-layer hot-wire anemometers, were used to perform measurements over the model. Novel high-speed flow visualization performed with a smoke-wire placed within the separated shear layer showed that roll-up vortices are shed within separation bubbles forming on the suction side of the airfoil. The structures were found to convect downstream and eventually break down during laminar-to-turbulent transition. Top view visualizations revealed that, at angles of attack of 0, 5, and 8 degrees, roll-up vortices form coherently across the span and undergo significant spanwise deformations prior to breaking down. At angles of attack of 5 and 8 degrees, rows of streamwise-oriented structures were observed to form during vortex breakdown. Statistics regarding the formation and development of shear layer roll-up vortices were extracted from high-speed flow visualization sequences and compared to the results of boundary layer measurements. It was found that, on the average, roll-up vortices form following the initial exponential growth of unstable disturbances within the separated shear layer and initiate the later stages of transition. The onset of these nonlinear stages was found to occur when the amplitude of velocity disturbances reached approximately 10% of the free-stream velocity. The rate of vortex shedding was found to fall within the frequency band of the unstable disturbances and lie near the central frequency of this band. The formation of vortices has been linked to the generation of harmonics of these unstable disturbances in velocity signals acquired ahead of mean transition. Once shed, vortices were found to drift at speeds between 33% and 44% of the edge velocity. Vortex merging at an angle of attack of 5?? was investigated. It was found that the majority of roll-up vortices proceed to merge with either one or two other vortices. Vortex merging between two and three vortices was found to occur periodically in a process similar to vortex merging in plane mixing layers undergoing subharmonic forcing of the most amplified disturbance. The flapping motion of the separated shear layer was investigated by performing a cross-correlation analysis on the high-speed flow visualization sequences to extract vertical displacement signals of the smoke within the shear layer. The frequency of flapping was found to correspond to the unstable disturbance band. At an angle of attack of 5??, it was found that the separated shear layer has a low-frequency component of flapping that matches a strong peak in velocity and surface pressure spectra that lies outside the unstable disturbance frequency band. The spanwise development of disturbances was assessed in the aft portion of the separation bubbles by performing a cross-correlation analysis on signals acquired simultaneously across the span with the rake of hot-wires. The spanwise correlations between signals was found to be well-correlated ahead of shear layer roll-up, after which disturbances became rapidly uncorrelated ahead of mean reattachment. These results were found to be linked to the coherent roll-up and subsequent breakdown of roll-up vortices.

The rotating-disk boundary layer is not only a simpler model for the study of cross-flow instability than swept-wing boundary layers but also a useful simplification of many industrial-flow applications where rotating configurations are present. For the rotating disk, it has been suggested that a local absolute instability, leading to a global instability, is responsible for the small variation in the observed laminar-turbulent transition Reynolds number however the exact nature of the transition is still not fully understood. This thesis aims to clarify certain aspects of the transition process. Furthermore, the thesis considers the turbulent rotating-disk boundary layer, as an example of a class of three-dimensional turbulent boundary-layer flows. The rotating-disk boundary layer has been investigated in an experimental apparatus designed for low vibration levels and with a polished glass disk that gave a smooth surface. The apparatus provided a low-disturbance environment and velocity measurements of the azimuthal component were made with a single hot-wire probe. A new way to present data in the form of a probability density function (PDF) map of the azimuthal fluctuation velocity, which gives clear insights into the laminar-turbulent transition region, has been proposed. Measurements performed with various disk-edge conditions and edge Reynolds numbers showed that neither of these conditions a↵ect the transition process significantly, and the Reynolds number for the onset of transition was observed to be highly reproducible. Laminar-turbulent transition for a ‘clean’ disk was compared with that for a disk with roughness elements located upstream of the critical Reynolds number for absolute instability. This showed that, even with minute surface roughness elements, strong convectively unstable stationary disturbances were excited. In this case, breakdown of the flow occurred before reaching the absolutely unstable region, i.e. through a convectively unstable route. For the rough disk, the breakdown location was shown to depend on the amplitude of individual stationary vortices. In contrast, for the smooth (clean-disk) condition, the amplitude of the stationary vortices did not fix the breakdown location, which instead was fixed by a well-defined Reynolds number. Furthermore, for the clean-disk case, travelling disturbances have been observed at the onset of nonlinearity, and the associated disturbance profile is in good agreement with the eigenfunction of the critical absolute instability. Finally, the turbulent boundary layer on the rotating disk has been investigated. The azimuthal friction velocity was directly measured from the azimuthal velocity profile in the viscous sublayer and the velocity statistics, normalized by the inner scale, are presented. The characteristics of this three-dimensional turbulent boundary-layer flow have been compared with those for the two-dimensional flow over a flat plate and close to the wall they are found to be quite similar but with rather large differences in the outer region. / <p>QC 20150119</p>

In this thesis the orientation of macro-sized fibres in turbulent flows is studied, as well as the effect of nano-sized fibrils on hydrodynamic stability. The focus lies on enabling processes for new materials where cellulose is the main constituent. When fibres (or any elongated particles) are added to a fluid, the complexity of the flow-problem increases. The fluid flow will influence the rotation of the fibres, and therefore also effect the overall fibre orientation. Exactly how the fibres rotate depends to a large extent on the mean velocity gradient in the flow. In addition, when fibres are added to a suspending fluid, the total stress in the suspension will increase, resulting in an increased apparent viscosity. The increase in stress is related to the direction of deformation in relation to the orientation of the particle, i.e. whether the deformation happens along the long or short axis of the fibre. The increase in stress, which in most cases is not constant neither in time nor space, will in turn influence the flow. This thesis starts off with the orientation and spatial distribution of fibres in the turbulent flow down an inclined plate. By varying fibre and flow parameters it is discovered that the main parameter controlling the orientation distribution is the aspect ratio of the fibres, with only minor influences from the other parameters. Moreover, the fibres are found to agglomerate into streamwise streaks. A new method to quantify this agglomeration is developed, taking care of the problems that arise due to the low concentration in the experiments. It is found that streakiness, i.e. the tendency to agglomerate in streaks, varies with Reynolds number. Going from fibre orientation to flow dynamics of fibre suspensions, the influence of cellulose nanofibrils (CNF) on laminar/turbulent transition is investigated in three different setups, namely plane channel flow, curved-rotating channel flow, and the flow in a flow focusing device. This last flow case is selected since it is can be used for assembly of CNF based materials. In the plane channel flow, the addition of CNF delays the transition more than predicted from measured viscosities while in the curved-rotating channel the opposite effect is discovered. This is qualitatively confirmed by linear stability analyses. Moreover, a transient growth analysis in the plane channel reveals an increase in streamwise wavenumber with increasing concentration of CNF. In the flow focusing device, i.e. at the intersection of three inlets and one outlet, the transition is found to mainly depend on the Reynolds number of the side flow. Recirculation zones forming downstream of two sharp corners are hypothesised to be the cause of the transition. With that in mind, the two corners are given a larger radius in an attempt to stabilise the flow. However, if anything, the flow seems to become unstable at a smaller Reynolds number, indicating that the separation bubble is not the sole cause of the transition. The choice of fluid in the core flow is found to have no effect on the stability, neither when using fluids with different viscosities nor when a non-Newtonian CNF dispersion was used. Thus, Newtonian model fluids can be used when studying the flow dynamics in this type of device. As a proof of concept, a flow focusing device is used to produce a continuous film from CNF. The fibrils are believed to be aligned due to the extensional flow created in the setup, resulting in a transparent film, with an estimated thickness of 1 um. / <p>QC 20141003</p>

In this study, numerical investigations of stability and transition problems were performed for 2D compressible boundary layers over a flat plate in adiabatic wall condition. Emphasis was placed on linear stability theory. The mathematical formulation for 3D boundary layers with oblique waves including detailed theoretical information was followed by use of the numerical techniques for the solution of resulting differential system of the instability problem, consequently an eigenvalue problem. First, two-dimensional sinusoidal disturbances were analyzed at various Mach numbers including the subsonic, transonic, supersonic and even hypersonic flow speeds. In this case, the second mode (acoustic mode), namely the Mack mode, and its behavior with the increasing Mach number were visualized. The results were then compared with the available data in literature concluding with good agreements. Secondly, similar analysis was carried out for oblique waves. Here, not only the effect of flow speed but also the effect of wave orientation was demonstrated. For this purpose, instability problem was solved for several wave angles at each Mach number in the range of M=0 and M=5. In this respect, the angle at which the waves were most unstable was also obtained at each investigated flow speed. The resultant stability diagrams corresponding to M=4 and higher Mach numbers for which both first and the second modes appear revealed that plane waves were more stable than oblique waves for the Tollmien-Schlichting mode, however, this was the opposite for the acoustic mode where oblique waves were observed to be more stable. As a final step, estimation of the transition location was handled for the most unstable wave condition. Smith-Van Ingen transition method was applied as the prediction device. The results representing the influence of Mach number on transition Reynolds number were then compared with the experimental data as well as the numerical ones in literature ending up with very good agreements.

Prévoir la transition laminaire-turbulent de la couche limite sur l'avant-corps d'un véhicule hypersonique est importantpour optimiser l'entrée d'air du superstatoréacteur qui lui est associé, mais reste très difficile après un demi-siècle derecherches intensives sur le sujet. Dans ce travail, les approches numériques et expérimentales sont mises en oeuvre etcomparées. Expérimentalement, la transition naturelle est détectée à Mach 4 et Mach 6 dans la soufflerie continue T-313de l'ITAM à Novossibirsk à l'aide de mesures de pression Pitot. Dans une autre soufflerie de l'ITAM, la AT-303 à rafale,on a détecté la transition naturelle à Mach 6 et la transition déclenchée par rugosités à Mach 8 à l'aide d'un procédéoptique basé sur l'emploi de peintures thermosensibles. Ces essais ont été réalisés sur maquette à échelle 1/3. Toutesles rugosités testées se sont montrées efficaces. La prévision théorique de la transition naturelle a été réalisée au moyende la théorie de la stabilité linéaire locale modale couplée à la méthode du eN. En vol, sur avant-corps à échelle 1, lesfacteurs N atteignent difficilement 8 à 9, ce qui est insuffisant pour assurer la transition avec certitude. Pour appliquer laméthode aux essais au sol, on a besoin de connaître les facteurs N de transition des souffleries, ce qui est réalisé à partird'essais de calibration sur plaque plane dans T-313. Un excellent accord théorie/expérience est obtenu à Mach 4. AMach 6, on doit prendre en compte la présence d'instabilité ‘’crossflow’’ inflexionnelle au nez de l'engin, moyennant quoil'accord est aussi très bon. Les calculs de stabilité ont été réalisés sur des solutions de base obtenues par simulationnumérique (CFD) des conditions de vol ou des essais au sol. Ces simulations ont également permis de bien comprendrela structure de l'écoulement autour de l'avant-corps et de concevoir en grande partie les moyens d'essai. / The prediction of the laminar-turbulent transition in the boundary layer on a hypersonic vehicle forebody is important tooptimize the air inlet of the associated scramjet engine, but is still very difficult after half a century of intensive research onthe subject. In this work, numerical and experimental approaches are applied and compared. Experimentally, the naturaltransition is detected at Mach 4 and Mach 6 in the blow down wind tunnel T-313 in ITAM Novosibirsk using Pitot pressuremeasurements. In the impulse AT-303 wind tunnel in ITAM, the natural transition at Mach 6 and the roughness inducedtransition at Mach 8 are detected using an optical method based on thermosensitive paints. These tests have beenperformed on a 1/3 scale model. All the trips tested have shown their effectiveness. The theoretical prediction of thenatural transition has been performed using the local modal linear stability theory coupled with the eN method. In flight, onthe full scale forebody, N factors hardly reach 8 to 9, which is insufficient for the transition. To apply the method to groundtests, the wind tunnels transition N factors are needed. They are obtained from calibration tests on a flat plate in T-313. Avery good agreement with experiments is found at Mach 4. At Mach 6, the presence of inflexional crossflow instabilitynear the nose of the body must be taken into account, which gives also a good agreement. Stability calculations havebeen done for mean flow solutions obtained by numerical simulations (CFD) of flight or ground tests conditions. Thesesimulations have also helped to understand the structure of the flow around the forebody and to design efficiently theexperimental setup.

This thesis presents the use of an unsteady total pressure measurement to detect laminar-to-turbulent transition. A miniature dynamic pressure transducer, Kulite model XCS-062-5D, was utilized to measure the total pressure fluctuations, and was integrated with an autonomous boundary layer measurement device that can withstand flight test conditions. Various sensor-probe configurations of the Kulite pressure transducer were first examined in a wind tunnel with a 0.610 m (2.0 ft) square test section with a maximum operational velocity of 49.2 m/s (110 mph), corresponding dynamic pressure of 1.44 kPa (30 psf). The Kulite sensor was placed on an elliptical nose flat plate where the flow was known to be turbulent. The Kulite sensor was then evaluated to measure total pressure fluctuations in laminar, turbulent, and transition of boundary layers developed on the flat plate in the same wind tunnel. The root-mean-square value of total pressure fluctuations was less than 1 % of the local free-stream dynamic pressure in the laminar boundary layer, but was about 2 % in the turbulent boundary layer. The value increased to 4 % in transition, indicating that the total pressure fluctuation measurements can be used not only to distinguish the laminar boundary layer from the turbulent boundary layer, but also to identify the transition region. The unsteady total pressure measurement was also conducted in a with a 2.13 m (7.0 ft) by 3.05 m (10.0 ft) section with similar operational velocity range as the previous wind tunnel. The Kulite sensor was placed on a wing model under laminar and transition conditions. The testing yielded similar results, demonstrating the usefulness of total pressure measurement for identifying the laminar-to-turbulent transition.

Two types of shear flows are investigated in this thesis; numerical simulations are performed for the analysis and control of the perturbation arising in a boundary layer over a flat plate, whereas PIV measurements are analysed for the investigation of a confined turbulent jet. Modal structures of the flows are identified: the aim is to understand the flow phenomena and to identify reduced-order models for the feedback control design. The attenuation of three-dimensional wavepackets of streaks and Tollmien-Schlichting (TS) waves in the boundary layer is obtained using feedback control based on arrays of spatially localized sensors and actuators distributed near the rigid wall. In order to tackle the difficulties arising due to the dimension of the discretized Navier-Stokes operator, a reduced-order model is identified, preserving the dynamics between the inputs and the outputs; to this end, approximate balanced truncation is used. Thus, control theory tools can be easily handled using the low-order model. We demonstrate that the energy growth of both TS wavepackets and streak-packets is substantially and efficiently mitigated, using relatively few sensors and actuators. The robustness of the controller is investigated by varying the number of actuators and ensors, the Reynolds number and the pressure gradient. The configuration can be possibly reproduced in experiments, due to the localization of sensing and actuation devices. A complete analysis of a confined turbulent jet is carried out using timeresolved PIV measurements. Proper orthogonal decomposition (POD) modes and Koopman modes are computed and analysed for understanding the main features of the flow. The frequencies related to the dominating mechanisms are identified; the most energetic structures show temporal periodicity. / QC 20110214

Laminar-turbulent transition has a strong impact on aerodynamic performance in many practical applications. Hence, there is a practical need for developing reliable and efficient transition prediction models, which form a critical element of the CFD process for aerospace vehicles across multiple flow regimes. This dissertation explores machine learning approaches to develop transition models using data from computations based on linear stability theory. Such data provide strong correlation with the underlying physics governed by linearized disturbance equations. In the proposed transition model, a convolutional neural network-based model encodes information from boundary layer profiles into integral quantities. Such automated feature extraction capability enables generalization of the proposed model to multiple instability mechanisms, even for those where physically defined shape factor parameters cannot be defined/determined in a consistent manner. Furthermore, sequence-to-sequence mapping is used to predict the transition location based on the mean boundary layer profiles. Such an end-to-end transition model provides a significantly simplified workflow. Although the proposed model has been analyzed for two-dimensional boundary layer flows, the embedded feature extraction capability enables their generalization to other flows as well. Neural network-based nonlinear functional approximation has also been presented in the context of transport equation-based closure models. Such models have been examined for their computational complexity and invariance properties based on the transport equation of a general scalar quantity. The data-driven approaches explored here demonstrate the potential for improved transition prediction models. / Doctor of Philosophy / Surface skin friction and aerodynamic heating caused by the flow over a body significantly increases due to the transition from laminar to turbulent flow. Hence, efficient and reliable prediction of transition onset location is a critical component of simulating fluid flows in engineering applications. Currently available transition prediction tools do not provide a good balance between computational efficiency and accuracy. This dissertation explores machine learning approach to develop efficient and reliable models for predicting transition in a significantly simplified manner. Convolutional neural network is used to extract features from the state of boundary layer flow at each location along the body. These extracted features are then processed sequentially using recurrent neural network to predict the amplification of instabilities in the flow, which is directly correlated to the onset of transition. Such an automated nature of feature extraction enables the generalization of this model to multiple transition mechanisms associated with different flow conditions and geometries. Furthermore, an end-to-end mapping from flow data to transition prediction requires no user expertise in stability theory and provides a significantly simplified workflow as compared to traditional stability-based computations. Another category of neural network-based models (known as neural operators) is also examined which can learn functional mapping from input variable field to output quantities. Such models can learn directly from data for complex set of problems, without the knowledge of underlying governing equations. Such attribute can be leveraged to develop a transition prediction model which can be integrated seamlessly in flow solvers. While further development is needed, such data-driven models demonstrate the potential for improved transition prediction models.

No description available.