The vast majority of gas turbine combustor systems employ swirl injectors to produce a central toroidal recirculation zone (CTRZ) which entrains and recirculates a portion of the hot combustion gases to provide continuous ignition to the incoming air-fuel mix. In addition to these primary functions, swirl injectors often generate multiple aerodynamic instability modes which are helical in nature with characteristic frequencies that can differ by many orders of magnitude. If any of these frequencies are consistent with prevalent acoustic modes within the combustor there is a potential for flow-acoustic coupling which may reinforce acoustic oscillations and drive combustion instabilities via the Rayleigh criterion. The aerodynamic performance of the swirl injector is thus of great practical importance to the design and development of combustion systems and there is a strong desire within industry for reliable computational methods that can predict this highly unsteady behaviour. This assessment can be made under isothermal conditions which avoids the complex interactions that occur in reacting flow. The goal of the present work was to compare and contrast the performance of Unsteady Reynolds- Averaged Navier-Stokes (URANS) and Large-Eddy Simulation (LES) CFD methodologies for a combustion system equipped with a derivative of an industrial Turbomeca swirl injector as this exhibits similar unsteady aerodynamic behaviour under reacting and isothermal conditions. The influence of the level of swirl, SN = 0.51−0.8, was first investigated experimentally using Particle Image Velocimetry (PIV) by varying the inlet swirl vane angle. Based on a qualitative assessment of instantaneous velocity data, and a range of coherent structure eduction techniques, it was found that ®1 = 30± (SN ¼ 0.8) would be the most challenging test case for LES and URANS as this contained near and far-field instability modes that differ in frequency by around two orders of magnitude and the highest levels of normal Reynolds-stress anisotropy. Based on extensive simulations performed with both in-house (LULES and Delta) and commercial (Fluent) CFD codes it was found that, despite the relative modest computational cost of URANS which is between one-third (RST) to an order of magnitude (k−²) less than that demanded by LES, only LES captures the all-important frequency content in accordance with experimental evidence and, thus, only LES can be recommended for use in swirl injector flows. The increased cost is believed to be an absolutely worthwhile expense because of the high fidelity of the predicted results in the important area of flow instabilities.

This thesis examines the cooling performance and the flow on a gas turbine blade. Numerical and experimental methods are described and implemented to assess the influence of film cooling effectiveness. A modem gas turbine blade geometry has been used. The blade is considered as a solid body with the blade cross section from hub to shroud varying with a degree of skewness. Computational Fluid Dynamics (CFD) is employed to assess blade film cooling effectiveness via simulation of the effect of varying blowing ratios (BR=1, 1.5 and 2), varying coolant fluid temperature (Tc=153 K and Tc=287.5 K), various angles of injection (35°,45° and 60°), increasing the number of cooling holes (32 and 42) and increasing the cooling holes diameter (D= 0.5 mm and 1mm). A full three-dimensional finite-volume method has been utilized in this study via the FLUENT 6.3 code with a k-epsilon (RNG) turbulence model. Results of the CFD models were carefully validated by studying aerodynamic flow and heat transfer in turbine blade film cooling performance. A two-dimensional channel and NACA 0012 airfoil were selected to investigate turbulence effects. The solution accuracy is assessed by carrying out a sensitivity analysis of mesh type and quality effects with enhancement wall treatment and standard wall function effects also addressed for turbulent boundary layers. In this study, four different turbulence models were utilized (S-A, mu-epsilon, (RNG), and (SST) mu-o). The computations were compared with available Direct Numerical Simulation (DNS) and experimental data. Good correlation was observed when using the RNG turbulence model in comparison with other turbulence models. Film cooling effectiveness and heat transfer along a flat plate has been analyzed for four different plate materials, namely steel, carbon steel, copper and aluminum, with 30° angle of injection. The cooling holes arrangement was simulated for a hole diameter of D=1 mm and different sections of the blade showing cooling effectiveness and heat transfer characteristic variation with increasing (BR = 0.5, 1). Furthermore a symmetrical single hole at 35° angle of injection was studied both the solid and shell plate cases. Cooling effectiveness numerical results were compared with available experimental data and the effect of material thermal properties for the solid plate on cooling performance evaluated. Numerical modeling has clearly identified that there is no benefit in reducing the number of holes as this decreases film cooling effectiveness. The experimental investigation showed the effect of increasing volumetric flow rate V°=1000, 800 and 600 cm3/min, as a term of the blowing ratio (BR) and angle of injection (35°,45° and 60°) for a modem gas turbine blade specimen using Thermal Paint Technology (TPT) and a Thermal Wind Tunnel (TWT). Both methods confirmed that the blade specimen with angle of injection of 45°, blowing ratio of BR=2 (which corresponds to 1000cm[3]/min), cooling holes diameter D=lmm and 42 holes developed a better film cooling effectiveness compared with the 35° and 60° cases. In addition TPT is a sufficient and relatively easy method for evaluating temperature distributions in experimental studies.

The research carried out as part of this EngD is aimed at understanding the high temperature materials used in modern gas turbine applications and providing QinetiQ with the information required to assess component performance in new propulsion systems. Performance gains are achieved through increased turbine gas temperatures which lead to hotter turbine disc rims and blades. The work has focussed on two key areas: (1) Disc Alloy Assessment of High Temperature Properties; and (2) Thermal Barrier Coating Life Assessment; which are drawn together by the overarching theme of the EngD: Lifing of Critical Components in Gas Turbine Engines. Performance of sub-solvus heat treated N18 alloy in the temperature range of 650°C to 725°C has been examined via monotonic and cyclically stabilised tensile, creep and strain controlled low cycle fatigue (LCF) tests including LCF behaviour in the presence of a stress concentration under load-control. Crack propagation studies have been undertaken on N18 and a particular super-solvus heat treatment variant of the alloy LSHR at the same temperatures, in air and vacuum with 1s and 20s dwell times. Comparisons between the results of this testing and microstructural characterisation with RR1000, UDIMET® 720 Low Interstitial (U720Li) and a large grain variant of U720Li have been carried out. In all alloys, strength is linked to a combination of γ' content and grain size as well as slow diffusing atoms in solid solution. High temperature strength improves creep performance which is also dependent on grain size and grain boundary character. Fatigue testing revealed that N18 had the most transgranular crack propagation with a good resistance to intergranular failure modes, with U720Li the most intergranular. Under vacuum conditions transgranular failure modes are evident to higher temperature and ΔK, with LSHR failing almost completely by intergranular crack propagation in air. For N18 significant cyclic softening occurs at 725°C with LCF initiation occurring at pores and oxidised particles. An apparent activation energy technique was used to provide further insights into the failure modes of these alloys, this indicating that, for N18 with 1s dwell, changes in fatigue crack growth rates were attributed to static properties and for LSHR, with 20s dwell in air, that changes were attributed to the detrimental synergistic combination of creep and oxidation at 725°C. Microchemistry at grain boundaries, especially M23C6 carbides, plays an important role in these alloys. Failure mechanisms within a thermal barrier coating (TBC) system consisting of a CMSX4 substrate, PtAl bond coat, thermally grown oxide (TGO) layer and a top coat applied using electron beam physical vapour deposition have been considered. TGO growth has been quantified under isothermal, two stage temperature and thermal cyclic exposures. An Arrhenius relation was used to describe the TGO growth and produce an isothermal TGO growth model. The output from this was used in the QinetiQ TBC Lifing Model. Thermo-mechanical fatigue test methods were also developed including a novel thermocouple placement permitting substrate temperature to be monitored without disturbing the top coat such that the QinetiQ TBC Lifing Model could be validated. The importance of material, system specific knowledge and performance data with respect to a particular design space for critical components in gas turbine engines has been highlighted. Data and knowledge regarding N18, LSHR and TBC systems has been added to the QinetiQ’s databank enhancing their capability for providing independent advice regarding high temperature materials particularly in new gas turbine engines.

This thesis investigates, both theoretically and experimentally, the phenomenon of ingress through gas turbine rim seals. The work presented focuses on modelling and measuring the required sealing flow levels to purge the wheelspace against combined ingress and the effect of externally-induced ingress on the surface temperature and heat transfer to the rotor. Combined ingress is driven by a pressure difference between the mainstream annulus and wheelspace cavity resulting from the combination of the asymmetric external pressure profile in the annulus and the rotation of fluid in the rotor-stator wheelspace cavity. Ingress can be prevented by pressurising the wheelspace through the supply of sealant flow. The Owen (2011b) combined ingress orifice model was solved to predict the required levels of sealant flow to prevent ingress into the wheelspace. The model was validated using prepublished data and data collected experimentally over the course of this research. Gas concentration measurements were made on the stator of the Bath single-stage gas turbine test rig to determine the variation of sealing effectiveness with sealant flow rate for an axial clearance seal geometry at design and off-design operational conditions. The measured variation of the required sealant flow rate with the ratio of the external and rotational Reynolds numbers, ReW / Reϕ, was consistent with the findings of other workers: at low values of ReW / Reϕ, ingress levels were influenced by the combined effects of the disc rotation and the annulus pressure profile and were therefore considered to fall into the combined ingress region; the influence of rotation diminished as ReW / Reϕ increased and the ingress levels were dominated by the annulus pressure field (externally-induced ingress). The orifice model was in good agreement with the experimental measurements and the prepublished experimental data. Thermochromic liquid crystal (TLC) was used to determine effect of ingress on the heat transfer coefficient, h, and adiabatic wall temperature, Tad, on the rotor of the Bath gas turbine rig. Concurrent gas concentration measurements were made on the stator to compare the effects of ingress on the two discs. Data was collected at the design condition, where ReW / Reϕ = 0.538 and at an overspeed off-design condition, where ReW / Reϕ = 0.326. The comparison between a newly defined adiabatic effectiveness, εad, on the rotor and the concentration effectiveness, εc, on the stator, showed that the rotor was protected against the effects of ingress relative to the stator. The sealing air, which is drawn into the rotor boundary layer from the source region, thermally buffers the rotor against the ingested fluid in the core. A thermal buffer ratio, η, was defined as the ratio of the minimum sealant flow required to purge the stator against ingress to the minimum sealant flow required to purge the rotor against ingress. The thermal buffer is dependent upon the flow structure in the wheelspace, which itself is governed the turbulent flow parameter, λT. A hypothesis relating η to λT was developed and shown to be in good agreement with the experimental data. The local Nusselt numbers, Nur, on the rotor were shown to be fairly constant with radius and increased as λT was increased. The latter finding can be explained by the flow structure in the wheelspace: as λT is raised, the swirl in the fluid core reduces, which results in an increase in the moment coefficient and Nur on the rotor. Difficulties in measuring Tad during the experiments suggested a new technique from which to solve for h and Tad using TLC surface temperature measurements. The solution Fourier’s equation for a step-change in the temperature of a fluid flowing over a solid of semi-infinite thickness (the ‘semi-infinite solution’) is limited to relatively low Fourier numbers if Tad is to be calculated accurately. A two-layer composite substrate made from, for example, polycarbonate and Rohacell, could be used to achieve accurate estimates of h and Tad over a larger range of Biot numbers than for a single material substrate. TLC could be used to measure the surface temperature history of the composite substrate during an experiment; this would allow h and Tad to be solved from the numerical solution of Fourier’s equation or from a combination of the semi-infinite and steady-state solutions. The work presented in this thesis has uncovered some interesting findings in areas where research was limited. The measurements of the minimum sealant flow required to purge the wheelspace at off-design operation for a rotor-stator system with blades and vanes and the measurements of the adiabatic effectiveness on a rotating disc affected by ingress are unique and provide a platform for further experimental studies and validation of CFD models.