This study examines the local and averaged endwall heat transfer effects of a staggered array of porous pin fins within a rectangular channel. The porous pin fins were made from aluminum and had a pore density of 10 pores per inch (PPI). The pressure drop through the channel was also determined for several flow rates and presented in terms of the friction factor. Local heat transfer coefficients on the endwall were measured using Thermochromic Liquid Crystal (TLC) sheets recorded with a charge-coupled device (CCD) camera. Static and total pressure measurements were taken at the entrance and exit of the test section to determine the overall pressure drop through the channel and explain the heat transfer trends through the channel. Results are presented for Reynolds numbers between 25000 and 130000 and a blockage ratio (blocked channel area divided by open channel area) of 50%. All results were compared to the corresponding results obtained using solid pins. All experiments were carried out in a 150 mm by 500 mm channel with an X/D of 1.72, a Y/D of 2.0, and a Z/D of 1.72 for the porous pins. It was found that for the range of Reynold's numbers tested in this study, the porous pin array consistently resulted in a larger friction factor, and therefore greater losses than a geometrically similar array of solid pins. The friction factors for the solid pin array were between 9.5 and 10.5, similar to the results found in the literature. For the porous pins, however, the friction factors were significantly increased as the Reynold's number increased, reaching as high as 15.3 at the highest Reynold's number tested. The heat transfer enhancement for the porous pins was found to be between 150 and 170% while the solid pins resulted in a heat transfer enhancement between 190 and 230%.

Wirkungsgradsteigerungen moderner Turbomaschinen machen dort auch zukünftig verstellbare Leitschaufeln unverzichtbar. Infolge komplexer Ringraumgeometrie entstehen bei der Schaufelverstellung betriebspunktabhängige Radialspalte, welche hinsichtlich Effizienzoptimierungen eine zunehmende Rolle spielen. Die vorliegende Arbeit charakterisiert die aerodynamischen Mechanismen und das Potential von spezifischen radialen Teilspalt- und Seitenwandkonturgeometrien solcher Verstellstatoren. Anders als bei durchgehenden radialen Spalten variiert der Einfluss der Teilspalte abhängig von deren axialer Position. Ein Teilspalt im hinteren Schaufelsehnenbereich erreicht reduzierte Totaldruckverluste gegenüber einer spaltfreien Referenz- beschaufelung. Dieser Vorteil vergrößert sich mit steigender Gitterbelastung und erhöht sich nochmals im Vergleich mit Konfigurationen eines vorderen Teilspaltes. Die Ergebnisse zei- gen, dass die Position entlang der Sehne der wesentliche Ein- flussfaktor auf die Radialspaltcharakteristik ist, die Spaltlänge als auch deren radiale Variation spielen nur eine untergeordnete, quantitative Rolle. Auch die untersuchten modellhaften Seitenwandkonturen wirken sich aufgrund sekundärer geome- trischer Einflüsse positiv auf Verlust und Gitterumlenkung in der Abströmung aus. Reduzierte Abmessungen der Teilspalte schwächen auch charakteristische Periodizitäten in Verbindung mit Spaltströmung bzw. resultierendem Spaltwirbel ab. Diese Erkenntnisse erlauben aerodynamische Optimierungen der Geometrie eines spezifischen Statordesigns. Ebenso ergibt sich ein Anwendungspotential im Randbereich starrer Leitschaufeln.

Modern gas turbine designs' ever-increasing turbine inlet temperature raises challenges for the nozzle guide vane cooling. Two typical endwall cooling schemes, jump cooling and louver cooling, result in different interactions between the injected coolant and the mainstream, leading to different cooling effects. This study investigates these two cooling schemes on the endwall cooling experimentally and numerically. Wind tunnel tests and the CFD simulations are carried out with engine-representative conditions of an exit Mach number of 0.85, an exit Reynolds number of 1.5×10^6, and an inlet Turbulence intensity of 16%. The jump cooling scheme experiments investigate two blowing ratios, 2.5 and 3.5, two density ratios, 1.2 and 1.95, and three endwall profiles with different NGV-turbine alignments. Four coolant mass flow ratios from 1.0% to 4.0% are tested for the louver cooling. The results show that the cavity vortex, the horseshoe vortex, and the passage vortex are the main factors that prevent the upstream coolant from reaching the NGV passage. The jump cooling scheme generally provides high momentum to the cooling jets. As a result, the coolant at the design case density ratio of 1.95 and blowing ratio of 2.5 is sufficiently energized to penetrate the horseshoe vortex. It then forms a relatively uniform coolant film near the NGV passage inlet, leading to a minimum adiabatic cooling effectiveness of 0.4 throughout the passage. Reducing the coolant density or increasing the blowing ratio leads to higher coolant momentum, so the coolant jets can further suppress the horseshoe vortex. However, high momentum may cause coolant lift-off, mitigating the coolant reattachment. Therefore, the density ratio needs to be carefully balanced with the blowing ratio to optimize the cooling effect. This balance is also affected by the combustor-NGV misalignment, as a higher step height requires higher coolant momentum to overcome the step-induced vortices. On the contrary, the louver cooling scheme provides less momentum to the coolant. The results showed that only by exceeding a coolant mass flow rate of 1~2% can the coolant form a uniform film which provides good coverage upstream of the NGV passage inlet. As for the cooling of the NGV passage, the mass flow rate ratio of the range investigated is not sufficient for desirable cooling performance. The pressure side endwall proves most difficult for the coolant to reach. In addition, the fishmouth cavity at the combustor-NGV passage causes a three-dimensional cavity vortex that transports the coolant in the pitch-wise direction. Moreover, the coolant transport pattern is dependent on the coolant blow rate. Overall, the more-energized coolant film generated by the jump cooling tends to survive longer, but it is also more prone to lift-off. At the same time, the less-energized coolant film caused by the louver cooling is more susceptible to vortices and the discontinuity of the endwall geometry. However, it develops faster, especially in the lateral direction. The two schemes could be applied simultaneously for an ideal cooling system. The jump cooling can provide enough momentum for the coolant to persist in the NGV passage. Meanwhile, the louver cooling covers the upstream region before the jump cooling coolant reattaches to the endwall. / Doctor of Philosophy / Gas turbines, sometimes called combustion turbines, are widely used to generate power or propulsion for various applications. The three main components of a gas turbine are compressor, combustor, and turbine. Modern gas turbines run at a high turbine inlet temperature that exceeds the current metal limits to increase efficiency. However, this brings significant challenges to the cooling of the first stage of the turbine, the nozzle guide vane. In this research, two commonly used endwall cooling methods, jump cooling and louver cooling, are investigated under engine-representative conditions experimentally and numerically. In addition, flow physics is demonstrated to explain the endwall cooling performance, mainly the upstream boundary layer caused by the interaction between the mainstream and the coolant flow. The results show that the cavity vortex, the horseshoe vortex, and the passage vortex are the main factors that prevent the upstream coolant from reaching the NGV passage. The jump cooling scheme provides high momentum to the cooling jets. As a result, the coolant in the design case is sufficiently energized to penetrate the horseshoe vortex, providing a desirable cooling effect in the NGV passage. Reducing the density ratio or increasing the blowing ratio can help the coolant jets further suppress the horseshoe vortex but also causes more lift-off, which adversely affects the cooling performance. On the contrary, the louver cooling scheme provides less momentum to the coolant, forming a less energized coolant film. The lack of coolant causes the louver coolant film to provide good coverage immediately downstream of the louver scheme exit. However, due to unfavorable interaction with vortices and endwall discontinuity, the cooling effect decays quickly downstream. Overall, the more-energized coolant film generated by the jump cooling tends to survive longer, but it is also more prone to lift-off. At the same time, the less-energized coolant film caused by the louver cooling is more susceptible to vortices and the discontinuity of the endwall geometry. However, it develops faster, especially in the lateral direction. The two schemes could be applied simultaneously for an ideal cooling system to work mutually beneficially.

This dissertation presents a detailed experimental and numerical analysis of the aerothermal characteristics of the turbine extremity regions i.e. the blade tip and endwall regions. The heat transfer and secondary flow characteristics were analyzed for different engine relevant configurations and exit Mach/Reynolds number conditions. The experiments were conducted in a linear blowdown cascade at transonic high turbulence conditions of Mexit ~ 0.85, 0.60 and 1.0, with an inlet turbulence intensity of 16% and 12% for the vane and blade cascade respectively. Transient infrared (IR) thermography technique and surface pressure measurement were used to map out the surface heat transfer coefficient and aerodynamic characteristics. The experiments were complemented with computational modeling using the commercial RANS equation solver ANSYS Fluent. The CFD results provided further insight into the local flow characteristics in order to elucidate the flow physics which govern the measured heat transfer characteristics. The results reveal that the highest heat transfer exists in regions with local flow reattachment and new-boundary layer formation. Conversely, the lowest heat transfer occurs in regions with boundary layer thickening and separation/lift-off flow. However, boundary layer separation results in additional secondary flow vortices, such as the squealer cavity vortices and endwall auxiliary vortex system, which significantly increase the stage aerodynamic losses. Furthermore, these vortices result in a low film-cooling effectiveness as was observed on a squealer tip cavity with purge flow. Finally, the importance of transonic experiments in analyzing the turbine section heat transfer and flow characteristics was underlined by the significant shock-boundary layer interactions that occur at high exit Mach number conditions. / Ph. D.

Thesis (PhD) -- Stellenbosch University, 2014. / ENGLISH ABSTRACT: With increasing environmental concerns and the drive for a greener economy comes an increased desire to improve turbine engine fuel efficiency and reduce emissions. Unfortunately weight reduction techniques used increase the blade loading, which in turn increases the losses. Non-axisymmetric endwall contouring is one of several techniques being investigated to reduce loss in a turbine. An investigation at Durham University produced a non-axisymmetric endwall design for a linear cascade. An adaption of the most promising endwall was investigated in an annular rotating test rig at the CSIR using steady state instrumentation. The current investigation extends those investigations into the unsteady time domain. Previous investigations found that a generic rotor endwall contour improved efficiency by controlling the endwall secondary flow vortex system in both a linear cascade and an annular 1½ stage rotating test turbine. The current research was aimed at determining if there were any unsteady effects introduced by the contoured endwall. The approach was unique in that it investigated the unsteady effects of an endwall contour originally designed for a linear cascade both experimentally and numerically at three incidence angles (positive, zero and negative to represent increased load, design load and decreased load respectively), the results of which are openly available. Unsteady experimental hotfilm results showed that the endwall contour made the velocity profile more radially uniform by reducing the strength of the endwall secondary flow vortex system. The fluctuations in the velocity were also reduced producing a more temporally uniform velocity profile. The FFT magnitude of the velocity at the blade passing frequency below midspan was also reduced. It was found that the reduction in the endwall secondary flow vortex system due to the contour increased with increasing loading. Numerical results showed that the oscillations in the flow were small and did not penetrate the boundary layer. The contoured rotor was forward and aft loaded when compared to the annular rotor, resulting in a weaker cross passage pressure gradient which allowed the endwall secondary flow vortex system to be less tightly wrapped. Numerical results did not show a significant difference in the oscillations observed in the annular and contoured rotor. A new objective function for use in the endwall optimisation process was proposed that acts as a proxy for efficiency, but is less prone to uncertainty in the results. When used on the current results it shows the same trend as efficiency. It remains to be used to design an endwall for full validation. / AFRIKAANSE OPSOMMING: Met ’n toenemende omgewingsbesorgdheid en die strewe na ’n groener ekonomie kom ’n toenemende behoefte om turbine enjin brandstofdoeltreffendheid te verbeter en vrystellings te verlaag. Ongelukkig het gewigsbesparingstegnieke wat gebruik is die lemlading verhoog, wat op sy beurt die verliese verhoog. Nie-assimmetriese endwandprofilering is een van verskeie tegnieke wat ondersoek word om verliese in ’n turbine te verminder. ’n Ondersoek by die Universiteit van Durham het ’n nie-assimmetriese endwandontwerp vir ’n lineêre kaskade gelewer. ’n Aanpassing van die mees belowende endwand is in ’n annulêre roterende toetsopstelling by die WNNR getoets, deur gebruik te maak van bestendige toestand instrumentasie. Die huidige ondersoek brei daardie ondersoeke uit na die nie-bestendige verwysingsraamwerk . Vorige ondersoeke het bevind dat die generiese rotor endwandprofiel doeltreffendheid verbeter as gevolg van die beheer van die endwand sekondêre vloei draaikolkstelsel in beide ’n lineêre kaskade sowel as ’n annulêre 1½ stadium roterende toetsturbine. Die huidige navorsing was daarop gemik om vas te stel of die endwandprofiel enige onbestendige effekte tot gevolg gehad het. Die benadering was uniek in die sin dat dit die onbestendige effekte ondersoek het van ’n endwandprofiel wat oorspronklik ontwerp is vir ’n lineêre kaskade beide eksperimenteel en numeries op drie invalsshoeke (positief, nul en negatief om onderskeidelik verhoogde lading, ontwerplading en verlaagde lading te verteenwoordig), waarvan die resultate algemeen beskikbaar is. Onbestendige eksperimentele warmfilm resultate het getoon dat die endwandprofiel die snelheidsprofiel meer radiaal uniform gemaak het deur die vermindering van die sterkte van die endwand sekondêre vloei werwelstelsel. Die skommelinge in die snelheid is ook verminder wat ’n meer tyduniforme snelheidsprofiel gelewer het. Die FFT (Fast Fourier Transform) grootte van die snelheid van die lem verbygaan frekwensie onder lem midbestek het ook verminder. Daar was bevind dat die vermindering in die endwand sekondêre vloei draaikolkstelsel as gevolg van die endwandprofiel toeneem met toenemende lading. Numeriese resultate het getoon dat die ossilasie in die vloei klein was en nie die grenslaag binnegedring het nie. Die rotor met gevormde wand het ’n voor- en agterlading gehad in vergelyking met die rotor met annulêre wand, wat tot ’n laer drukgradient dwarsop die vloeirigting gelei het, die endwand sekondêre vloei draaikolkstelsel minder beperk het. Numeriese resultate het nie ’n beduidende verskil in die ossilasies tussen die annulêre en gevormde rotorwand getoon nie. ’n Nuwe doelwitfunksie vir gebruik in die endwand optimersproses is voorgestel wat dien as ’n plaasvervanger vir doeltreffendheid, maar minder geneig is tot onsekerheid in die resultate. Wanneer dit gebruik word op die huidige resultate toon dit dieselfde tendens as doeltreffendheid. Dit moet nog gebruik word in die ontwerp van ’n endwand vir volledige bevestiging.

Swirling flows abound in nature and numerous engineering applications. Under conditions which are not completely understood, the swirling cores could undergo a sudden enlargement of their vortex core, leading to a ’vortex breakdown’. The physics of vortex breakdown and strategies to control it have been active areas of research for nearly half a century. There are many competing theories of vortex breakdown in the literature; broadly, these are surmised on similarities to flow separation, hydrodynamic instability or transition from a supercritical to a subcritical state. However, a rational criterion for vortex breakdown continues to be elusive. One of the most well known criteria in the literature is the one due to Brown and Lopez (1990) based on an inviscid vortex dynamics model which suggests that the helix angle of the velocity vector should enclose the helix angle of the vorticity vector. However it appears that this only suggests that the stream surface would diverge and not necessarily constitute a condition for breakdown. In this work, we propose a new criterion based on helicity (scalar product of velocity and vorticity vectors) for characterizing breakdown since it has fundamental topological interpretations relating to change in linkages of vortex lines. In particular, it is suggested that the breakdown location corresponds to the location where helicity becomes zero. We study the problem of vortex breakdown in a cylindrical container with a rotating top lid in order to clarify and elucidate our hypothesis. We present results from Direct Numerical Simulation of this problem for three different Reynolds numbers and evaluate the utility of our proposed helicity criterion. Our studies indicate that helicity is indeed a better choice for characterizing vortex breakdown.

No description available.

This research work investigates the thermal performance of a film-cooled gas turbine endwall under two different mainstream flow conditions. In the first part of the research investigation, the effect of unsteady passing wakes on a film-cooled pitchwise-curved surface (representing an endwall without airfoils) was experimentally studied for heat transfer characteristics on a time-averaged basis. The temperature sensitive paint technique was used to obtain the local temperatures on the test surface. The required heat flux input was provided using foil heaters. Discrete film injection was implemented on the test surface using cylindrical holes with a streamwise inclination angle of 35? and no compound angle relative to the mean approach velocity vector. The passing wakes increased the heat transfer coefficients at both the wake passing frequencies that were experimented. Due to the increasing film cooling jet turbulence and strong jet-mainstream interaction at higher blowing ratios, the heat transfer coefficients were amplified. A combination of film injection and unsteady passing wakes resulted in a maximum pitch-averaged and centerline heat transfer augmentation of ? 28% and 31.7% relative to the no wake and no film injection case. The second part of the research study involves an experimental and numerical analysis of secondary flow and coolant film interaction in a high subsonic annular cascade with a maximum isentropic throat Mach number of ? 0.68. Endwall (platform) thermal protection is provided using discrete cylindrical holes with a streamwise inclination angle of 30? and no compound angle relative to the mean approach velocity vector. The surface flow visualization on the inner endwall provided the location of the saddle point and the three-dimensional separation lines. Computational predictions showed that the leading-edge horseshoe vortex was confined to approximately 1.5% of the airfoil span for the no film injection case and intensified with low momentum film injection. At the highest blowing ratio, the film cooling jet weakened the horseshoe vortex at the leading-edge plane. The passage vortex was intensified with coolant injection at all blowing ratios. It was seen that increasing average blowing ratio improved the film effectiveness on the endwall. The discharge coefficients calculated for each film cooling hole indicated significant non-uniformity in the coolant discharge at lower blowing ratios and the strong dependence of discharge coefficients on the mainstream static pressure and the location of three-dimensional separation lines. Near the airfoil suction side, a region of coalesced film cooling jets providing close to uniform film coverage was observed, indicative of the mainstream acceleration and the influence of three-dimensional separation lines.

Shock-tube experiments are typically performed at high temperatures (≥1200K) due to test-time constraints. These test times are usually ~1 ms in duration and the source of this short, test-time constraint is loss of temperature due to heat transfer. At short test times, there is very little appreciable heat transfer between the hot gas and the cold walls of the shock tube and a high test temperature can be maintained. However, some experiments are using lower temperatures (approx. 800K) to achieve ignition and require much longer test times (up to 15 ms) to fully study the chemical kinetics and combustion chemistry of a reaction in a shock-tube experiment. Using mathematical models, analysis was performed studying the effects of temperature, pressure, shock-tube inner diameter, and test-port location at various test times (from 1 - 20 ms) on temperature maintenance. Three models, each more complex than the previous, were used to simulate test conditions in the endwall region behind the reflected shock wave with Ar and N2 as bath gases. Temperature profile, thermal BL thickness, and other parametric results are presented herein. It was observed that higher temperatures and lower pressures contributed to a thicker thermal boundary layer, as did shrinking inner diameter. Thus it was found that a test case such as 800K and 50 atm in a 16.2-cm-diameter shock tube in Argon maintained thermal integrity much better than other cases - pronounced by a thermal boundary layer ≤ 1 mm thick and an average temperature ≥ 799.9 K from 1-20 ms.

<p>In rotating detonation engines the turbine inlet conditions may be transonic with unprecedented unsteady fluctuations, very different from those in conventional high-pressure turbines. To ensure an acceptable engine performance, the turbine passages must be unchoked at subsonic and started at supersonic conditions. Additionally, to maximize the aerodynamic performance potential, ad-hoc designs are required, suited for the oscillations in Mach number and flow angle. This manuscript focuses on designing and characterizing diffusive turbine vanes that can operate downstream of a transonic rotating detonation combustor.  </p> <p><br></p> <p>First, the phenomenon of unstarting is presented, concentrating on the effect of pressure loss on the accurate prediction of the starting limit. Afterward, a multi-objective optimization with steady Reynolds Averaged Navier Stokes simulations, including the endwall and 3D vane design, is performed. The results are discussed, highlighting the impact of the throat-to-inlet area ratio on the pressure loss and the geometric features of the top-performing designs. Compared to previous  research on stator passages with contoured endwalls, considerable reductions in pressure loss and stator-induced rotor forcing are obtained, with an extended operating range and preserving high turning.  </p> <p><br></p> <p>Subsequently, the influence of the inlet boundary layer thickness on the vane performance is evaluated, inducing remarkable increases in pressure loss and downstream pressure distortion. Employing an optimization with a thicker inlet boundary layer, specific endwall design recommendations are found, providing a notable improvement in both objective functions. The impact of the geometry variations on flow detachment is assessed as well.</p> <p><br></p> <p>Finally, the impact of the inlet flow angle on the vane design is studied through a multi-point, multi-objective optimization with different inlet angles. The effect of incidence on the flow field and vane performance is evaluated first. Then, by comparing the optimized geometries with those optimized for axial inflows, several design guidelines are identified </p>