GT Proceedings of ASME Turbo Expo 2013: Turbine Technical Conference and Exposition GT2013 June 3-7, 2013, San Antonio, Texas, USA.

Proceedings of ASME Turbo Expo 2013: Turbine Technical Conference and Exposition GT2013 June 3-7, 2013, San Antonio, Texas, USA GT2013-95687 A review of the Oxford Turbine Research Facility Kam Chana, Dave Cardwell and Terry Jones Department of Engineering Science University of Oxford Parks Road Oxford OX1 3PJ. UK ABSTRACT Gas turbine engine efficiency and reliability is generally improved through better understanding and improvements to the design of individual components. The life limiting component of the modern gas turbine is the high pressure (HP) turbine stage due to the arduous environment. Over the last 50 years significant research effort has been focused on advancing blade cooling designs and materials. Due to practical limitations little fundamental research on the turbine system is performed in the operating gas turbine engine. Consequently different types of experimental approaches have been developed over the last 4 decades to study the flow and in particular the heat transfer and cooling in turbines. In general the facilities can be divided into continuous running or short duration and cascade or rotating. Over the last 30 years short duration facilities have dominated the research in the study of turbine heat transfer and cooling. The Oxford Turbine Research Facility (formerly known as the QinetiQ Turbine Test Facility, The Isentropic Light Piston Facility and The Isentropic Light Piston Cascade) is a short duration facility developed and built in the late 1970s and early 1980s for turbine heat transfer and cooling studies. This paper presents the developments and measurements taken on the facility over the last 35 years, including the type of research that has been conducted and, the current capability of the facility. NOMENCLATURE c p Specific heat capacity at constant pressure H Total enthalpy flux I Polar moment of inertia m Mass-flow rate q Specific heat Q s t T γ η ω T coolant Tmean Tradial Tlocal Heat transfer rate Specific entropy Time Temperature, Torque Ratio of specific heat capacities Efficiency Angular velocity Coolant temperature injected through casing, hub endwalls and turbulence rods Mean temperature of the flow Radial temperature of the flow circumferentially averaged Local temperature of the flow Subscripts 0 Total conditions, Output value, Initial conditions 1 Stage inlet conditions 2 NGV exit conditions 3 Stage exit near plane conditions 4 Stage exit far plane conditions Averaged value Time differential INTRODUCTION Turbine design engineers continually strive to improve gas turbine efficiency and power by developing technologies to increase overall pressure ratios (beyond 50) and turbine inlet temperatures (in excess of 2000K). Heat transfer to the turbine blades, particularly in the first stage high pressure (HP) turbine, is extremely important as these components are generally the life limiting parts of the engine. Competing demands from the aerodynamicist, blade cooling specialist, metallurgist, stress engineer and the production engineer are uniquely challenging 1 Copyright 2013 by ASME

and make the evolution of efficient, reliable designs the ultimate test of the industry. Advanced cooling schemes are deployed that are crucial for the hot gas path components. Designs have to accommodate continuously increasing hot gas temperatures and pressures and consequently increasing coolant temperatures while at the same time satisfying allowable material temperatures. Over the years, gas path temperatures have increased significantly, with material temperature capability only marginally increasing over the same period. This entails the designer requiring accurate prediction methods to design an efficient cooling system. In turn this requires accurate heat transfer data from engine representative environments to develop and validate predictive design methods. Experimental research facilities for external turbine heat transfer investigations have over the last 30 years tended to be mainly full-scale engine demonstrators, short duration high speed facilities or large scale low speed rigs. The use of high speed continuous running warm flow facilities for heat transfer measurements has been rare for many decades. Measurements from gas turbine engine demonstrators are excellent to generate realistic turbine heat transfer data that includes all aspects such as leakage flows, combustor outlet flows and cooling. However, instrumenting engines with extremely high temperatures and pressures makes the measurements difficult, unreliable and in particular high-resolution measurements are virtually impossible. In addition, the cost of performing engine measurements is significant in comparison to test facilities. The use of short duration facilities for turbine heat transfer research was pioneered at the Department of Engineering Science, University of Oxford in the late 1960 s. Jones and Schultz [1] studied turbine blade cooling using shock tunnels and developed specialist transient instrumentation techniques based on those previously used in hypersonic studies, Shultz and Jones [2]. The Oxford researchers developed and built a new novel type of short duration tunnel in the early 1970 s called the Isentropic Light Piston Tunnel [3] at the Osney laboratory, Oxford. The Oxford tunnel went through a progression of developments, starting as a linear cascade followed by an annular cascade and subsequently rotation of an HP turbine was introduced. HP and HP/IP working sections were tested in this facility. For 4 decades the Oxford facility produced high quality aerodynamic and heat transfer data. Results from this tunnel have contributed significantly to the understanding of turbine flows and the design methods for turbines. This paper does not describe the Oxford tunnel further but gives the history of the version of the tunnel built at The Royal Aircraft Establishment (RAE). From the outset the tunnel was annular and had a volume five times that of the Oxford tunnel. The Isentropic Light Piston type of tunnel has been extremely popular for the measurement of heat transfer to turbines, predominantly because of their relatively low cost of operation and excellent scaling for Mach number, Reynolds number and gas-to-wall temperature ratio. The tunnels do not operate at a full engine temperature but provide significant heat transfer to the turbine components which is readily and accurately measured. Heat transfer coefficients resulting from short duration facilities are considered to be of high quality in terms of experimental uncertainties and of very high value to the turbine designer. Heat transfer instrumentation is generally advanced heat flux sensors with signal transmission via slip rings for the rotating frame and signal processing through software. Short duration tunnels for turbine heat transfer investigations have been constructed and operated by The University of Oxford, The Von Karman Institute for Fluid Dynamics, Ohio State University (previously at Calspan), Massachusetts Institute of Technology Gas Turbine Lab, Wright Patterson Air Force Base and QinetiQ UK [4,5,6,7]. The OTRF Facility was originally built at the Royal Aircraft Establishment (Part of the Ministry of Defence) and has now moved to the Department of Engineering Science, University of Oxford, Osney Themo-Fluids Laboratory and replaced the Original Oxford Rotor Facility. The facility has been renamed the Oxford Turbine Research facility and is the subject of this paper. This paper provides an overview of the Oxford Turbine Research Facility which has been operational at the Osney Thermo-Fluids Laboratory since 2011. Details of the facility, its development and work programmes over the last 3 decades are given including developments to incorporate a rotor, nonuniform inlet temperature generator, inlet swirl variation and the addition of film cooling. Operation of the isentropic light piston tunnel and the measurement of heat transfer The Isentropic Light Piston tunnel was originally conceived at the Department of Engineering Science, University of Oxford [8]. This novel short duration approach was required to produce flow Mach numbers, Reynolds numbers and gas-towall temperature ratios to match those in gas turbine stages. The temperature ratio required to match engine conditions is modest, with typical gas temperatures even for a modern engine of order 2000K and metal temperatures of order 1100K entailing that a temperature ratio of 2 is adequate. Thus, if the metal temperature in the facility is at ambient, the gas temperature requirement is about 600K, much lower than in the engine case. 2 Copyright 2013 by ASME

A schematic of the Oxford Turbine Research Facility (OTRF) is shown in Figure 1. The main components are i) the pump-tube containing a light weight piston, ii) a fast acting plug valve, iii) the working section and iv) a dump tank. The operating principles and theory of this approach were first described by Jones et al. [8]. The test gas is contained in the pump-tube which is filled to a pre-determined level depending on the compression required. The test gas is compressed isentropically by a free piston which is propelled by allowing high pressure gas behind the piston. Once the test gas has been compressed to the required pressure level and heated to match the temperature ratio the fast acting valve is triggered to open. The heated gas flows into the working section. To maintain the pressure level, the gas flow into the pump tube behind the piston is matched to that leaving the tube through the working section. Figure 1: The Oxford Turbine Research facility. Measurement of heat transfer rates in short duration facilities is easier than in continuously operating warm flow rigs because, in continuous running facilities the model reaches an equilibrium temperature. In transient tunnels the heat transfer can be evaluated from the temperature-time history as the model does not reach temperature equilibrium. To measure heat transfer in continuously running facilities the model would require internal cooling to maintain the temperature difference between the gas and model, however, in transient facilities the models can be manufactured from relatively simple materials that are thermally insulating such as machinable glass (macor), quartz, plastic or aluminum alloy covered in a plastic material. The temperature-time history is measured using high frequency platinum thin film gauges. The theory of thin film heat transfer gauges for short duration facilities is given by Schultz and Jones [2]. THE OXFORD TURBINE RESEARCH FACILITY The Oxford Turbine Research Facility was originally built at the Royal Aircraft Establishment (RAE) Pyestock as an annular cascade in the early 1970 s (Called the Isentropic Light Piston Cascade). The University of Oxford, Department of Engineering Science were the design authority for the facility with the tunnel being operated by the Royal Aircraft Establishment (part of the Ministry of Defence). From the early 1980s to the early 1990s, once the tunnel had been commissioned, a number of HP turbine vanes were designed and studied for surface heat transfer. These included the T146 Turbine, High Rim Speed Turbine (HRST), MT1 turbine, High Rim Speed Turbine with axially profiled endwalls (HRST-profiled endwalls) and the High Temperature Demonstrator Unit 4X (HTDU4X). HRST and the HTDU 4X vanes were studied over 3 Mach Numbers and 3 Reynolds numbers. Exit area traversing was developed in the late 1980s to measure the loss at exit from the vane. Since the late 1980 s a number of major developments were carried out on the facility to bring the configurations studied closer to that found in engines. These major developments, turbines studies and measurements taken are summarised in Table 1 below. From 1990 to 2008 approximately 25M worth of developments and measurements were carried out whilst the facility was at RAE Pyestock and QinetiQ Farnborough Year Development Measurements taken 1982 T146 and HRST vanes, RAE Design Heat transfer & surface static pressures Aerofoil at 10,50,90% blade spans 1985 Hub and casing endwalls 1985 Installation of RR designed vane (MT1) with new plug Heat transfer & surface static pressures valve cassette system introduced for ease of Aerofoil 10,50,90% blade spans 1987 instrumentation Hub & casing fillet Hub and casing endwalls 1987 Film cooling of vane casing endwall MT1 vane Heat transfer on the casing endwall 1988 Varying blowing rate, density ratio and coolant temperature. 1988 1989 HRST with axially profiled hub endwall 3 Reynolds numbers 3 Mach numbers 1989 3D design HP vane HTDU 4X, RAE design 3 Reynolds numbers Heat transfer & surface static pressures on vane and endwalls. Exit area traversing for Po, Ma, Heat transfer & surface static pressure - Casing endwall 3 Copyright 2013 by ASME

3 Mach numbers - Hub endwall 1991 - Hub fillet - Casing fillet - Aerofoil 5, 10, 50, 90, 95% blade spans 1991 Optical measurements with MT1 vane Particle Image Velocimetry - Velocity within vane passage 1992 1994 Installation of new working section for full rotating stage with uniform inlet. MT1 turbine vane and rotor. Turbo-brake for constant speed In-shaft electronics for rotating measurements Heat transfer and surface static pressure - Vane aerofoil 10,50,90% blade spans - Vane hub and casing endwalls Heat transfer steady/unsteady - Rotor aerofoil 5,10,50,90, 95% blade spans Surface static pressure steady/unsteady - Rotor aerofoil 50% blade span Stage exit area traversing 1993 Optical access for particle image velocimetry Velocity within stator/rotor region 1995 Film cooling MT1 turbine stage Heat transfer Pressure & suction side aerofoil cooling Vane aerofoil 10,50,90% blade spans 1997 Varying blowing ratio, density ratio with Rotor aerofoil steady/unsteady 10,50,90%blade spans foreign gas injection. Surface static pressure Vane aerofoil 10,50,90% blade spans Rotor aerofoil steady/unsteady 50%blade span Steady and unsteady stage exit area traversing Po, To, Ma, 1998 Stage exit area traversing MT1 HP turbine Steady and unsteady Po, To, Ma,, 1999-2.2 vane pitches and 3 probes 1999 2000 OTDF1 & OTDF2 MT1 HP turbine stage Heat transfer on vane aerofoil, vane endwalls, rotor aerofoil, rotor tip, rotor casing Surface static pressure on vane aerofoil, vane endwalls, rotor aerofoil, rotor casing Inlet traverse Po, To Stage exit traverse for Po, To, Ma, 2000 2002 1.5 Stages with and without OTDF MT1 HP turbine stage - co-rotating design structural style, low number off IP vane 2003 Rotor casing film cooling (Static shroud) 2 designs of film cooling Varying blowing ratio 2004 Facility relocated to DERA Farnborough Independent from site services New IP compressor Additional storage tanks for air motor 10 bar air supply system 2004 Installation of a new industrial gas turbine HP turbine (Alstom Lincoln) 2005 HiREST shrouded turbine Highly loaded stage 2006 Efficiency measurements MT1 turbine stage Heat transfer and surface static pressures on IP vane aerofoil at 10, 50, 90% blade spans Static pressure on exit duct IP vane exit area traversing Heat transfer on the rotor casing Rotor exit area traversing Re-commissioning and comparison with pre-move data Heat transfer measurements with OTDF and film cooling of HP vane Heat transfer Vane aerofoil, 10,50 90% blade spans Rotor aerofoil Rotor blade shroud Measurements of Mass flow Torque 2007 Inlet traverse Stage exit near plane traverse Stage exit far plane traverse 2007 EOTDF & efficiency MT1 turbine stage Heat transfer & surface static pressure Vane aerofoil 10, 50, 90% blade spans Vane endwalls 2008 Rotor aerofoil 10, 50, 90% blade spans Rotor casing Stage exit area traversing near and far planes Po, To, Ma, Efficiency measurements (as above) & Mass flow EOTDF 2008 Swirl & efficiency MT1 turbine stage Heat transfer & surface static pressure Vane aerofoil 10, 50, 90% blade spans 4 Copyright 2013 by ASME

Vane endwalls 2009 Rotor aerofoil 10, 50, 90% blade spans Rotor casing Inlet traverse for Po, To, Ps,, Stage exit area traversing near and far planes Po, To, Ma, 2010 Facility moved and re-commissioned at University of Oxford Aerodynamic data compared to pre-move measurements May 2011 New HP compressor 3 new acquisition systems 2011 Effusion cooling Samulet/Siloet Programme 2012 2012 Entropy Noise MT1 vane High frequency pressure (21 Kulites) measurements upstream and downstream of HP vane 2012 Installation of 1.5 stage MT1 HP stage FACTOR EU 7 th Framework Programme 2013 IP vane for counter rotating stage 2013 Combined OTDF and swirl inlet generator Rolls-Royce UK 2013 Cooled HP turbine SAMULET/TSB/EPSRC Programme film cooled vane and rotor 2014 2014 2015 Installation of a 1.5 stage cooled HP turbine EU 7 th Framework Programme film cooled vane uncooled rotor Table 1: OTRF developments and measurements taken. The most significant development was commissioned by RAE from the University of Oxford during the late-1980 s when a full sized aero-engine high-pressure turbine stage working section fitted with the MT1 HP stage was developed [9]. A novel feature added to the facility was an aerodynamic turbobrake [10] on the same shaft as the turbine and driven by the turbine exit-flow. At the design speed the turbobrake power is matched with the turbine, and thus the turbobrake maintains constant speed during a run. The turbine shaft is supported on two sets of oil lubricated bearings. Additionally the shaft is hollow in order to house electronic circuit boards to condition signals from instrumented rotor blades. These signals are transmitted through a 24 channel slip-ring. blade counts for the HP stator and rotor are 32 and 60 respectively. The rotor module schematic and a photograph are shown in Figure 2. The pressure ratio is set by a second throat at exit from the stage. This is an annular variable area device that maintains axisymmetry in the exhaust flow, and since it is choked, it also isolates the turbine from disturbances originating downstream. The performance of the turbobrake is controlled using blockage rings and a bypass mechanism, which together reduce the flow through the turbobrake. Fine-tuning is achieved by adjusting the amount of flow that by-passes the turbobrake. Installation and commissioning of the rotating module with the tunnel took place at Pyestock during 1991/2. This major development allowed unshrouded HP turbine stage heat transfer and aerodynamics to be investigated on a high-pressure research turbine relevant to future civil aircraft applications. The turbine stage was a modern high pressure Rolls-Royce aero-engine design with stage pressure ratio of 3.2 and nozzle guide vane (ngv) Reynolds number of 2.54E6. The stage is unshrouded and Figure 2: The Oxford Turbine Research facility HP turbine stage working section photograph and schematic. 5 Copyright 2013 by ASME

The addition of film cooling to the ngv and rotor A three phase programme studying vane and blade cooling was commenced in November 1992. In the Phase I tests, neither the ngv nor the rotor were cooled; cooling was then added to the ngv only for Phase II, and to both rotor and ngv in Phase III. A coolant delivery system and new blade ring that allowed cooling to be delivered to the vane and blade internal cooling passages was designed and implemented. Figure 3 shows the mid-span cross section of the ngv cooling system. The cooled vane had two coolant cavities. Each cavity feeds cooling holes on both the pressure and suction surfaces. The cooling holes were arranged with two double rows of 0.6 mm diameter circular cooling holes on the pressure surface. The rows were staggered relative to each other and had an inclination angle of 50 to the local surface. The suction side had two single rows of cooling holes of the same diameter and inclination angle. For this test series the coolant supply to the vanes was performed using a combination of three blowing rates and two density ratios. In the real engine environment film cooling flows are cooler and hence denser than the mainstream flows. To simulate this in the facility the density ratio could be obtained by matching the mainstream/coolant temperature ratio, but the density range that can be achieved is limited unless the coolant is refrigerated. Another method is to use a foreign gas with a greater density, but with the same ratio of specific heats as air [11]. By using air alone at room temperature and a pre-mixed composition of sulphur hexafluoride and argon gas (SF6/Ar) density ratios of 1.6 and 2.9 were simulated. The composition of SF6/AR was chosen to give the same ratio of specific heats as air, i.e. γ = 1.4. Figure 3: Nozzle guide vane film cooling arrangement The addition of inlet temperature non-uniformity Time-averaged gas temperature at exit to a combustion system varies in both radial and circumferential directions because of incomplete mixing of hot and cold gas streams. The radial profile is a direct consequence of the combustor liner cooling, this may be tailored to minimise blade stresses at root and tip and the position of the peak temperature may be optimised. The circumferential peaks are a consequence of discrete fuel injectors and combustor dilution air. These peaks, known as hot spots are in various circumferential positions relative to the ngvs. Obvious alignments put the hot spot in line with the ngv passage or the ngv leading edges. In this study the number of hot spots was 32, the same as the number of ngvs. The system was designed to allow the hot spots to be rotated to any position relative to the ngvs. To relate radial and circumferential temperature variations from test rigs to engines, or visa versa, Overall Temperature Distortion Factor (OTDF) and Radial Temperature Distortion Factor (RTDF) distribution parameters are used. The definitions of these parameters for engines as proposed in the literature are given here: OTDF T = max ΔT T avg combustor However, in relating these to rig tests the ΔT combustor temperature is not valid and hence the parameters are redefined for comparison with engine data in the following manner: T OTDF = T Local mean T T mean coolant RTDF T RTDF = T radial = max, ΔT radial mean combustor T T T mean coolant A pilot study was commissioned by DERA at the Department of Engineering Science at Oxford University to study temperature generator implementation into the tunnel. Initially a computational study was carried out to investigate the means of generating radial temperature non-uniformity through injecting cold air from the hub and casing endwalls upstream of the turbine. Circumferential non-uniformity was proposed using cold air injection from radial bars fitted in the inlet section. However, the possibility of using existing bars fitted to the inlet for turbulence generation was proposed. The study involved the design and manufacture of a scale model of the chosen configuration. The model had the same cross-sectional dimensions as the facility to avoid side-wall effects, however the section was twodimensional. Detailed area traverses were conducted at the ngv leading edge locations with fast response thermocouples. Modified turbulence rods similar to those in the facility were used to inject cold air to produce the circumferential profile. Various hub and casing injection geometries were tested to optimise the radial temperature profiles. RTDF was implemented on the facility [12] by using part of the existing inlet boundary layer bleed system. The inlet section, which houses the plug valve, has radial ports open to an T avg 6 Copyright 2013 by ASME

annular plenum chamber at the casing. This is connected via spokes to a hub plenum. Between these plenum chambers and the gas path perforated annular plates originally used for inlet boundary layer bleed were fitted. These were replaced with non-perforated plates to allow injection at the hub and casing. The annular plates were fitted such that they formed an injection slot between the plate and casing housing. A similar arrangement was fitted on the hub wall. Figure 4 shows how the new plates are fitted in relation to the working section to form the injection slots. For the OTDF profile generation the turbulence rods were initially modified as in the pilot study. The hot spots would then be generated by operating the OTRF close to the required peak temperature and simultaneously injecting cold air from the turbulence rods and the hub and casing injection slots. Thus a hot spot could be produced between the injection points. Figure 4: OTDF installation to the facility There are 32 turbulence rods and 32 ngvs and thus one hot spot per ngv is produced. The turbulence rods were designed such that they could be rotated in the circumferential direction to change the position of the hot spots relative to the ngvs. During early commissioning tests it became apparent that the cold air injected from the turbulence rods was mixing out so that almost no circumferential temperature profile was detectable at the ngv leading edge plane. Increasing the mass flow from the turbulence rods did not affect the mixing significantly. To overcome this, small tapering duct sections with internal vanes were added to the turbulence rods to reduce mixing and direct the flow better. A schematic and a photograph of these vanes are shown in Figures 4 and 5 (in blue). This reduced the distance over which the profile could mix with the freestream and gave the required circumferential profile. Changing the mass flow through the hub, casing and turbulence rod systems independently showed that the system could match the targeted profiles and enhance the profiles significantly if required. Figure 5 shows a photograph of the installation showing the turbulence rods with and without ducts. Turbulence rods with ducts Hub injection Turbulence rods without ducts Figure 5: Turbulence rods with and without ducts fitted The addition of a second vane for a 1.5 Stage configuration In the period from 1994 to 2000, the ILPF was used to investigate the aerodynamics and heat transfer of the HP nozzle guide vane and rotor row, in particular the interactions between these two rows [13]. Area averaged experimental traverse data at the HP rotor exit are: Mach number, M 0.45, whirl angle, α 15, and total pressure, p 0 1.65 10 5 Pa respectively. The rotor flow exits as a free jet into a larger annulus before passing through the Second Throat where the flow is choked. The Second Throat allows adjustment of the turbine pressure ratio and isolates it from downstream interactions. In 2001 the ILPF was extended to 1.5 stages, with the introduction of an IP nozzle guide vane designed by Chana [14], using the Rolls Royce design suite. The IP vane exit conditions, predicted by a Rolls Royce through-flow code (code Q263), are M = 0.792, α = 71.4º and p 0 = 1.529 10 5 Pa. The position of the IP vane relative to the HP stage is shown in Figure 6. Some of the design features for this turbine and a limited collection of unsteady test data are given by Chana et al., [15]. The test facility was extended to investigate the interaction of an IP vane with a HP turbine stage, for both uniform and non-uniform inlet-temperature distribution conditions [16,17]. Few experimenters have directly measured the second stator surface heat transfer rate in a 1.5-stage turbine. This study presented steady and unsteady surface heat transfer for an intermediate pressure vane operating in this configuration. The vane aerodynamics were measured in detail and blind predictions of heat transfer using an integral method and the TEXSTAN code were compared. These data are unique, as they were taken in a facility equipped with a temperature distribution generator that simulated both the radial and circumferential variations in heat transfer at the HP stage inlet as present in the combustor exit-flow of an engine. A summary of the operating point of the OTRF 1.5-stage turbine for both uniform and non-uniform inlet temperature is given in Table2. 7 Copyright 2013 by ASME

Figure 6: Schematic of the 1.5 stage test-turbine; (R) The working section of the Oxford Turbine Research Facility. Parameter Predicted: Uniform Measured: Uniform Measured: OTDF inlet inlet inlet Inlet total pressure (bar) 4.6 4.6 ± 2% 4.6 ± 2% Nominal mean inlet total temperature (K) 444 444 ± 2% 444 ± 2% HP stator exit hub isentropic Mach number 1.054 1.034 ± 2% 1.063 ± 2% HP stator exit casing isentropic Mach number 0.912 0.925 ± 2% 0.926 ± 2% Speed (rpm) 9500 9500 ± 2% 9500 ± 2% Rotor exit hub static pressure (bar) 1.434 1.428 ± 2% 1.451 ± 2% Rotor exit casing static pressure (bar) 1.439 1.435 ± 2% 1.453 ± 2% Rotor relative inlet total pressure (bar) 2.697 2.707 ± 2% No data IP stator exit hub static pressure (bar) 0.883 0.869 ± 2% 0.886 ± 2% IP stator exit casing static pressure (bar) 1.158 1.097 ± 2% 1.121 ± 2% Table 2: Operating point of the ILPF 1.5-stage turbine for OTDF and Uniform inlet distributions The addition of shroud cooling Shroud cooling research was initiated in the ILPF in 2002, an initial two-dimensional design was tested and characterised with and without cooling. Results from this design were subsequently used to produce a more advanced threedimensional design with the aid of CFD. The first rotor casing film cooling design was based on a streamline two dimensional (2D) boundary layer approach and was designed by Chana [18] at Siemens Industrial-Turbines, Lincoln. This 2D approach was chosen over full three dimensional (3D) Navier Stokes predictions to give relatively quick and accurate estimates of the performance parameters like the heat transfer rates and cooling effectiveness. A 2D prediction scheme was particularly advantageous during the design as many conditions such as different blowing rates could be studied quickly. However, when tested in the ILPF this design did not perform well, as the nozzle guide vane (NGV) and rotor overtip secondary flows interacted with the cooling films stripping them from the casing local to the cooling holes. A new design was carried out utilising the lessons learnt and experience gained from the original design. In addition, extensive unsteady rotor casing heat transfer and pressure data (from the EU TATEF programme) were studied in order to understand and overcome the problems associated with film cooling the rotor casing and the interaction between the films and the secondary flows. An entirely new design philosophy was formulated which accepted that the cooling films would not survive the rotor passing and that the key to effective cooling was the re-establishment of the film between rotor passings. Validation of the design was carried out using 3D CFD in conjunction with Siemens Industrial-Turbines. With the design optimised, the cooling system was manufactured and implemented. The film-cooling design is representative of an engine configuration and was subjected to mechanical constrains similar to those for an engine component. The cooling configuration was tested at one blowing rate and one density ratio for a similar mass flow to the earlier design. The first rotor casing film cooling design is reported in reference GT2009-60242 part 1 [19] and the second design in reference GT2005-60246 [20]. The cooling holes were arranged in angled lines on the rotor casing. Before more detailed design work could be carried out, a list of requirements for the proposed design was produced. This specified requirements to maintain the integrity of the concept, for mechanical and manufacturing considerations and to comply with current standard practices in film cooling design. The main requirements are listed below. 8 Copyright 2013 by ASME

Approximate cooling hole row spacing (pitchwise) 5mm Minimum hole diameter 0.4mm Pitch/diameter (P/D) ratio 4 Length/diameter (L/D) ratio 5 Hole radial angle 15 Blowing ratio 1.0 Cooling fed from no more than 2 cavities The specification of an inlet radial angle of 15 is half the normally specified 30. However, it was considered important to maintain the film as close to the surface as possible. In order to avoid excessive complication in the design, cooling was to be fed from 2 cavities, front and rear. The result of this would be that for holes fed from the same cavity, the downstream holes would have a progressively higher blowing ratio, due to the drop in static pressure through the rotor row. The design was to be carried out so that the first hole in each row had a blowing ratio of 1.0, with the blowing ratio of the last at approximately 1.4. Ideally, each row of holes would be fed from a separate cavity so that a blowing ratio of 1.0 could be maintained at all rows. This was thought to be completely impractical to implement in the engine environment. Practical design for the MT1 annulus The initial design was tailored for the MT1 annulus with the following values: Cooling hole row spacing 6.19mm Hole diameter 0.42mm Pitch/diameter (P/D) ratio 4.05 Length/diameter (L/D) ratio 5.47 Hole radial angle 15 Nominal blowing ratio 1.0 Hole pitch 1.7mm Line angle 16 Holes per line 13 Number of lines (around entire annulus) 300 Cooling fed from 2 cavities Final design and manufacture Testing the design on the entire annulus was neither practical nor desirable. Instead, it was decided to carry out the test over a portion of the annulus by fitting a specially designed cooling block into one of the removable instrumentation cassettes. In addition, instrumentation in the form of platinum thin film gauges was fitted to the cooling block to measure the surface heat transfer rate from the leading edge to the trailing edge of the blade. During the design process, the first cooling hole fed from the after cavity had to be removed in order to allow a sufficiently thick dividing wall to be maintained between the feed cavities. Each cavity had an independent feed to control the mass flow via a calibrated orifice plate. Figure 7: Casing film cooling block fitted in a cassette Figure 7 shows the cooling block mounted in the instrumentation cassette. The vanes can be seen at the bottom of the picture and the rotor moves from right to left along the region marked by the abradeable strip. The cooling block is visible in the centre of the picture with the array of cooling holes over the track of the rotor. It is also possible to see where one hole from each row had to be removed in order to allow a sufficiently robust dividing wall between the 2 cavities. The installation of a Shrouded Turbine stage Shrouded HP turbines are still widely used since they give reduced tip leakage loss, less performance deterioration with time and improved vibration control compared to the unshrouded high rim speed alternative. They also avoid the requirement for complex Active Clearance Control systems in engines. However as engine cycle parameters (Turbine Entry Temperature) increase in the quest for improved efficiency, shroud reliability and life becomes an issue. Failures due to shroud curl caused by creep may occur and lead to high corrective costs. To investigate shrouded turbines a three-year programme called HiReST (High performance high Reliability Shrouded Turbine) was successfully carried out and reported in GT2007-27168 [21]. The main objectives were to study detailed aerodynamic, heat transfer and aero-mechanical test data over a range of test conditions for the high load HP stage. In particular shroud heat transfer levels were investigated to further improve shroud reliability and life and to explore innovative, efficient means to cool the shroud and reduce cooling flows. The high load, high lift, transonic shrouded Controlled Flow HP stage was previously tested in a Warm Air Turbine Rig. Measurements from the OTRF allowed steady and unsteady heat transfer and aerodynamic results to be obtained for unsteady methods validation. 9 Copyright 2013 by ASME

The meridional view of the turbine is shown on Figure 8 with details of the instrumentation. Further details of the rig, hardware, instrumentation and tests can be found in [22]. To achieve the objective, measurements of the turbine power and stage mass-flow were developed in the facility. Unlike other facilities of its type, the OTRF includes an aerodynamic braking system that keeps the rotor torque constant during a test run. This provides a unique advantage for the transient facility as the aerodynamics and power remain constant during a test run. The technique of measuring the adiabatic efficiency on a high pressure turbine in the transient Turbine Test Facility is given by Beard [23] and is summarised below. The measurements were made on the Rolls-Royce MT1 high pressure turbine under engine representative conditions. An emphasis was placed on achieving a precision accuracy of ±0.25% for the efficiency measurements with an absolute accuracy of order ±2%. QQ new statics WATR statics Thin film gauges Kulites Total pressure/temperature Figure 8: High load shrouded HP turbine stage installed in the OTRF Three conditions were tested - design, 80% speed and high pressure ratio as shown on Table 3. Measurements included static pressure tappings through the annulus and the vane surface, thin film gauges on the vane and rotor aerofoil and shroud, high frequency Kulite pressure transducers on the rotor surface at mid-span and total pressure/temperature at inlet and exit to the stage. Condition P0 in (Bar ) T0 in (K) Total/total pressure ratio Pstatic exit hub (Bar) Speed (rpm) N/(T) Design 2.9 393 3.43 0.765 9300 469 80% speed 2.9 393 3.24 0.755 7500 379 High pressure ratio 2.9 393 4.35 0.479 9300 469 Table 3: The HiReST Turbine stage operating conditions The addition of efficiency measurements Collection of high quality experimental data is essential for studying the steady performance and flow phenomena associated with an HP turbine. For many years, short duration facilities have been used to obtain time-averaged and timeresolved heat transfer and pressure data. The next major step with the use of these facilities was to capitalise on experience with high-speed data acquisition and high frequency response instrumentation to advance the capability by measuring the turbine stage performance both accurately and inexpensively. The adiabatic efficiency of a turbine is defined as the ratio of the actual power extracted by the rotor from the fluid to the ideal power obtained from an isentropic expansion between the same inlet and exit pressures. It is important to measure adiabatic efficiency as any heat transfer to or from the stage will affect the stage exit enthalpy and hence the measured efficiency. Efficiency of a turbine under non-adiabatic conditions is termed the indicated efficiency, but this is dependent on the quantity of heat transfer during the test. Hence, it is of more interest to determine the adiabatic efficiency which simply considers the irreversibility as a result of aerodynamic losses and allows meaningful, direct comparisons between different turbine configurations. The gas-to-wall temperature ratio in the test facility is engine representative, and hence the testing is not adiabatic. As a result of this heat loss, the exit entropy of the fluid is lower (ds=dq/t) than that in the adiabatic case. Therefore, to determine the adiabatic efficiency of a turbine in a transient facility, the heat transferred to the surrounding surfaces and rotor blades from the fluid must be accounted for. The correction term for the heat loss through the stage has been estimated from previous heat transfer measurements throughout the stage [24], and was found to be small at approximately 3%. As a result the main terms in this definition are the actual and isentropic power terms. In summation, the adiabatic efficiency of a turbine stage tested in the test facility can be evaluated by considering η adiabatic T = gauge + I rotor mc p T dω ω + Q dt p 011 p 04 12 01 + Q γ 1 γ 23 ΔH 03 10 Copyright 2013 by ASME

Hence to evaluate the adiabatic efficiency of a turbine stage in the OTRF, measurements of mass flow rate, torque and massaveraged inlet total pressure, inlet total temperature and exit total pressure are required. Measurement of rotational velocity is a standard measurement on the facility and the heat transfer throughout the stage has been evaluated in previous programmes. An area survey of inlet total pressure and inlet total temperature was performed 21mm upstream of the ngv leading edge. Turbine exit conditions were taken at a semi-mixed out plane positioned 4.5 mid-height axial rotor chords downstream of the rotor trailing edge at mid-height. A mixed out plane was implemented in an attempt to gain a more accurate measurement of the turbine efficiency, as the disturbances in all of the flow fields will be weaker than at a close plane. In addition an area-survey of the flow fields was taken at a rotor exit near plane at 0.5 mid-height axial rotor chords downstream of the rotor trailing edge at mid-height. Mass flow Evaluation of mass flow rate involves deriving the measurement from a list of variables, thus it has a large potential for error. The flow total temperature is measured using stage inlet thermocouples located at the turbine inlet plane. Instrumentation consists of 10 ngv leading edge thermocouples (25.4m bare) and 27 thermocouples placed in an inlet traverse cassette. The venturi upstream total pressure is measured from tappings on the end-wall of the piston tube. The throat is instrumented with four evenly distributed total-to-throat differential pressure measurements. The throat Mach number is found by measuring the total-to-static pressure ratio at the venturi throat. Location of the throat pressure tappings was optimised using three-dimensional steady CFD. Figure 9 shows the instrumented pump tube contraction, or exit, which acts as a converging-diverging venturi once the plug valve has opened. Δp = p 1 p 01-2 2 p 0 Plug Va Figure 9: Pump tube contraction with mass flow rate instrumentation A blowdown technique was used to calibrate the effective area of the pump tube contraction over a range of Reynolds number and Mach number. During an isentropic blowdown from a vessel, the stored mass in the vessel can be evaluated by considering a single pressure and temperature measurement. The mass-flow rate leaving the vessel is given by the change in stored mass. However, as the gas expands isentropically it cools and heat transfer occurs between the cold gas and the vessel walls. As a result isentropic conditions only exist at the beginning of the blowdown process. The difference in the exiting mass flow rates between an isentropic and nonisentropic blowdown was quantified, allowing a correction factor to be evaluated. The mass flow calibration in the OTRF involved two separate blowdown experiments with the piston removed from the pump tube. Firstly, a new technique which involves performing a series of blowdown experiments through an ISO calibrated sonic venturi was performed to evaluate the correction factor in mass-flow from isentropic conditions. Secondly, the correction factor is applied to a number of blowdown runs through the pump tube contraction to evaluate its effective area. Measurement of gas temperature in the pump tube during a blowdown with thermocouple instrumentation is not possible to the degree of accuracy that is required. The fluid velocities are typically of order 1ms -1. Instead, for both blowdown experiments a pseudo-isentropic mass-flow rate from the pump tube was calculated from initial pressure and temperature in the pump tube and the pressure decay during the blowdown. Inlet total pressure and total temperature were measured at the turbine stage to monitor the condition of each test run. Each rake has five pneumatic pitot tubes with Kiel heads (bevelled inlet) to allow accurate measurements up to relative flow angles of approximately ±20. Two ngvs were instrumented with 5 leading edge thermocouples to monitor the inlet total temperature to the turbine stage. The thermocouples are k-type 25.4m bare bead thermocouples. With a typical recovery factor of 0.75, the recovery factor error is only 0.07%. The radiation error is low enough to be neglected. Torque The accurate measurement of shaft torque is essential for the evaluation of the overall efficiency of the turbine stage. The OTRF includes an aerodynamic braking system to keep the turbine rotating at constant speed. The turbobrake provides a unique advantage for a transient test facility as it allows the turbine torque to be constant during a run, unlike other facilities of this type. Numerous measurement techniques were considered, but ultimately a device using strain gauges was implemented into the OTRF. Figure 10 shows the strain gauge based torque measurement system mounted on the rotor disc in the OTRF. Strain gauges are commonly used in industrial and engineering situations to measure strain as they are so robust and simple. With careful design, these systems can provide highly accurate torque measurements (typical accuracy of order 0.1%) and can 11 Copyright 2013 by ASME

compensate for temperature variations, axial, centrifugal and bending forces. However, obtaining good performance from such a system requires on shaft calibration, careful alignment and bonding of the strain gauges to the shaft and in-shaft electronics. Figure 10: OTRF strain gauge torque system, strain gauges mounted on the MT1 rotor disc Design of the OTRF strain gauge system primarily involved two factors: the positioning of strain gauges on the rotor disc in an area of high local shear strain, and their mounting orientation and electrical connection to amplify the individual strain gauge signals and reject unwanted strain components. The torque applied to the rotor disc is constant throughout a typical OTRF test run, and therefore the thinnest sections of the disc at large radius will experience the largest local shear strain. The end-cap of the turbine, where the rotor disc is connected to the rotating shaft, was chosen as the location of the strain gauge instrumentation. This section is relatively thin and at a large enough radius to obtain a detectable signal from the strain gauges. This region experiences relatively low centrifugal forces at this radius. Modelling the end-cap as a thin walled cylinder, and application of standard torsion theory, results in a calculated local shear strain of 79.5strain in the chosen location. As the changes in strain gauge resistance are so small, 16 strain gauges (8 half- bridges) were placed in a standard Wheatstone bridge circuit. The bridge amplifies the change in resistance by combining a number of strain gauges. Variation of strain gauge sensitivity with temperature originates from two sources. Firstly, changes in gauge resistance as a result of changes in the temperature of the gauge itself, and secondly, strains experienced by the gauge due to a differential temperature between the gauge and the specimen. The entire strain gauge bridge was covered in a thin weave of glass reinforced polymer (GRP) to protect the gauges from the harsh environment in the OTRF. The signal from the strain gauge bridge is processed by an in-shaft electronics processing unit (EPU) which is mounted along the center axis of the shaft. In-shaft electronics provides a 12V nominally smooth DC bridge excitation voltage and its output is connected to PC software through the slip ring. The EPU contains a high-gain strain gauge amplifier, analogue-todigital converter, microprocessor and volatile memory. A torque calibration facility was developed and manufactured (University of Oxford) to allow quick recalibration of the MT1 turbine disc (blades removed) strain gauge torque system. A highly accurate HBM TB2 reference inline torque transducer and the rotor disc are mounted on the same vertical shaft, with a direct path for the applied torque between them. The TB2 reference transducer has been calibrated by the German Calibration Service (DKD) Laboratory based at HBM, which is accredited by the German Metrology Institute (PTB) and traceable to International Standards. The strain gauge torque system on the rotor disc is calibrated using the TB2 reference transducer. The calibration transducer which has a nominal accuracy of 0.03%, also utilises a strain gauge measuring system, and is highly insensitive to axial, lateral and bending forces. A pulley system and Kevlar webbing wrapped around the disc was used to apply a static torque of up to 2500Nm to the rotor disc. To calibrate the strain gauge torque system for any temperature dependency, the turbine disc was placed in an insulated box. Two heater pads were placed on the upper and lower surfaces of the disc to control its temperature. Hot air from a mesh heater was blown through the insulated box by a centrifugal fan, so that the heat loss from the surface of the rotor disc is minimised. In combination the two heating systems allow the rotor disc to be slowly heated and cooled in the range of 20-120 C. The shaft between the rotor disc and reference transducer is cooled with a water jacket as it operates to its specifications over a range of 10-60 C. Enhanced inlet temperature non-uniformity and efficiency The second generation temperature distortion generator developed for the OTRF was named Enhanced Overall Temperature Distortion Factor (the EOTDF) generator, which simulates an aggressive combustor temperature profile of a modern engine. The design and commissioning of this generator is reported by Povey et al. [24]. The target profile was based on a combustor profile of a modern engine, measured at the most extreme point in the engine cycle. Low temperature regions at the inner and outer walls arise as a result of the combustor liner cooling flows, while the central hotter region is formed downstream of a burner. The hot-streak simulator is a second-generation design, in which cold gas is introduced into a hot mainstream though 12 Copyright 2013 by ASME

radial and circumferential slots upstream of the turbine stage. The simulator can be indexed, so the effect of clocking (relative circumferential position of hot streak and nozzle guide vane leading edge) can easily be investigated. An emphasis was placed on accurate measurement of turbine inlet enthalpy flux so that the impact of hot streaks on turbine efficiency could be investigated. The hot-streak simulator differs from all previous systems in that a pronounced radial and circumferential temperature profile has been generated, with a hot-streak to vane count of 1:1. Figure 11 gives a schematic of the second generation generator. The profile generated was very well matched (nondimensionally) to the target profile. The most accurate area survey of a simulated temperature profile conducted to date was taken, and this demonstrated that the simulator offers an exceptionally high degree of circumferential symmetry and runto-run repeatability. Figure 12: EOTDF inlet temperature profile generated Inlet swirl variation and efficiency In order to carry out detailed experimental investigations of the effects of swirl on the HP turbine stage, an inlet swirl simulator was designed and commissioned in the Oxford Turbine Research Facility. The target swirl profile at the turbine stage inlet was based upon combustor exit measurements performed by Rolls-Royce in a modern engine (Figure 13), with peak yaw and pitch angles of up to +/-40 degrees. Figure 11: Enhanced inlet temperature profile generator The circumferentially averaged profile generated is given in Figure 12, this is compared to the first generation inlet profile generator profiles (OTDF1, OTDF2) and profiles from other facilities [25,26,27]. The figure shows how significantly enhanced the profile is in comparison to other profiles generated to date. Vane and rotor heat transfer measurement with the OTDF1, OTDF2, and EOTDF generator are reported in references [28, 29, 30, 31]. Figure 13: Inlet swirl target profile at NGV leading edge The development, installation and commissioning of the swirl simulator in the OTRF is discussed in detail by Qureshi et al [33]. The design of an individual swirler to generate the desired swirl profile was conducted using a pilot study at Oxford University. The pilot study was performed in a subsonic tunnel fitted with a two-dimensional replica of the inlet contraction of MT1 HP turbine stage. During the pilot study, various swirler designs were investigated to find the best match to the target swirl profile. The swirl simulator module is designed to mount in the tunnel flow path upstream of the OTRF inlet contraction. The module allows rotation so that clocking of the vortex core with respect to the NGV leading edge could be achieved. To aid in CFD comparison, and to allow repeatability assessment around the annulus, an integer swirler to vane count ratio of 1:2 was 13 Copyright 2013 by ASME

chosen (16 swirlers and 32 NGVs). The manufactured swirl generation system, about 1m in diameter, is shown in Figure 14. Each swirler is composed of six flat-plate vanes inclined at 40º angle to the axial direction. cycle for uncooled and cooled tests. With the system frequency response around 90kHz, flow features within the rotor passing period are clearly evident. The cooled data showed areas of lower total temperature than the uncooled results, these corresponded to the ngv pitch. These areas can be identified with the wake of the ngv being convecting through the rotor passage. Figure 14: The assembled inlet swirl simulator module. The effect of inlet swirl variation on heat transfer and aerodynamics are reported by Qureshi [34]. Figure 15: Casing Nusselt number distributions. SAMPLE RESULTS A significant amount of data has been generated from the OTRF since the mid-1980s initially as an annular cascade and subsequently as a complete stage and 1.5 stage configuration. A significant portion of the measurements and results obtained have been published. Only a very small sample of the results will be presented in this paper however, references to papers have been given that cover the results in greater detail. As an annular cascade the HTDU4x highly threedimensional vane was studied for heat transfer and aerodynamics over three Mach numbers and three Reynolds numbers [35]. The casing measurements from this study are given in Figure 15. Notably the Nusselt number shows little variation with increasing Mach number whereas with Reynolds number there is a significant increase. Figure 16: Time mean total temperature (K). Total temperature measurements taken using a dual thin film probe (described by Buttsworth [36]) at exit to the stage for an uncooled and cooled vane and are reported in [37]. These measurements have been taken at 0.5 axial chord downstream of the rotor at mid-span. The contour map of total temperature for the uncooled tests presented in Figure 16 shows the timeaveraged temperature varying by up to 40K. As expected the highest temperatures were seen towards the casing where the tip leakage flows will have had less work extracted than the mainstream flow. The lower temperature regions near the hub correspond with the ngv pitch and are thus likely to be a result of the ngv wakes and secondary flow. The major advantage of using the dual thin film probe is its ability to measure time-resolved temperature data. Figure 17 shows time intervals of unsteady data during a rotor rotation Figure 17: Time resolved stage exit total temperature (K). 14 Copyright 2013 by ASME

HP vane surface heat transfer measurements have been taken with and without film cooling and with and without inlet temperature distortion and inlet swirl variation [37]. Nusselt number comparisons (Figure 18) for a cooled and uncooled case show good run-to-run repeatability. The influence of the cooling flows on the Nusselt number distribution on the late pressure surface shows a clear benefit; the Nusselt number decreases with increased blowing rate, and at all coolant conditions is significantly lower than the uncooled case. A similar reduction in Nusselt number, when cooling is present is seen on the late suction surface although the variation with blowing rate is not monotonic, suggesting that at the highest blowing rate the film cooling near the hole is not well attached to the surface. Immediately downstream of the first row of suction surface cooling holes, the Nusselt number has increased sharply in comparison with the un-cooled case. This could be due to the boundary layer undergoing transition at the site of injection. However, the enhancement increases with blowing rate suggesting that another mechanism is also present. The blowing rate from this row of holes is high, due to the simplified cooling geometry, which requires both the suction and pressure surface row to be fed from the same cavity. Consequently, the film lifts off the surface, except perhaps at the lowest blowing rate, and the heat transfer rate local to the hole is enhanced. This is due to the vane wake impinging on the rotor aerofoil. However near the suction surface crown region on the aerofoil the frequency doubles, this is due to the aerofoil experiencing both the wake and the ngv trailing edge shock. Figure 19: Rotor surface time resolved heat transfer rate. 2 x 105 gauge 1410 mean q = 6.7404*10 4 W/m 2 1.5 Heat transfer rate fluctuations (W/m 2 ) 1 0.5 0 0.5 1 1.5 2 0 0.5 1 1.5 2 2.5 Ngv index Figure 20, Instrumented rotor tip and measured heat transfer rate fluctuations for 2 vane passing periods. Figure 18, HP vane Nusselt number with and without film cooling. Rotor surface heat transfer has been taken on the aerofoil, tip and casing region both steady and unsteady and studied with and without film cooling, inlet temperature distortion and inlet swirl variation. Sample unsteady results are given here for the rotor tip and surface at mid-span. Figure 19, 20 gives unsteady heat transfer rate measurements at various locations around the rotor aerofoil and at the rotor tip. The results are presented as fluctuations with the mean level removed. Most of the thin film sensors show the heat transfer fluctuating at the passing frequency of the HP vane around the aerofoil and at the tip. CONCLUSIONS Transient testing has been recognized as a robust means of determining the aerodynamic and heat transfer characteristics of turbine vanes and blades for over three decades. The development of high frequency instrumentation and the availability of high frequency acquisition systems have allowed transient tunnels to routinely study and supply the gas turbine industry with results to help improve the designs methods. Tunnels such as the Oxford Turbine Research Facility have been in operation for nearly 4 decades and have been significantly upgraded to allow more of the engine features to be included. 15 Copyright 2013 by ASME