Schematic of four major types of HP rotor tip designs
Abstract
A rotating high-pressure (HP) turbine, in modern aircraft engines, is subjected to high cyclic thermal and mechanical loads. Shroudless turbines experience high heat loads on the rotor tip and casing, which makes them one of the life-limiting components of an engine. If the rotor tip starts to erode, then the tip clearance increases, which increases the over-tip leakage flow, resulting in reduced stage efficiency and ultimately, the engine lifetime. Since the 1970s, extensive research has been undertaken to understand the over-tip leakage flow and develop mitigation strategies such as novel tip designs, to minimize the losses and heat load. However, the majority of the present findings are based on linear cascade or low-speed rotational studies, due to the high cost of the experiments and difficulty in instrumentation at engine conditions. Recent studies (mostly numerical) have shown the importance of testing at engine-representative conditions, specifically targeting the transonic tip flow in a high-speed rotational environment, which is essential for an accurate understanding. The Oxford Turbine Research Facility (OTRF) is a high-speed rotating transient test facility that allows unsteady aerodynamics and heat transfer measurements, at engine representative conditions. This article presents an experimental investigation of the HP turbine rotor casing aerodynamics, performed in the OTRF. Steady and unsteady casing static pressure measurements were acquired, using pneumatic lines and high-frequency response (∼100 kHz) Kulite pressure transducers, respectively. The single-stage HP turbine consisted of cooled vanes (featuring film and trailing-edge slot cooling) and uncooled rotor blades. Three different tip designs (two squealers and one flat) were tested, at two tip gaps, and with two inlet temperature profiles that included a spatially uniform total temperature profile and an engine-representative HP turbine inlet total temperature profile. In parallel, a numerical investigation of the experimental test cases is presented using unsteady computational fluid dynamics (CFD) predictions. The experimental results from this study provide a unique dataset to validate numerical models.
1 Introduction
A shroudless turbine rotor necessitates a gap between the rotor tip and the casing to prevent rubbing during rotation. The gap, however, enables the flow at the pressure surface (PS) of the rotor to migrate toward the suction surface because of the pressure differential between the two surfaces. This flow is termed an “over-tip leakage flow.” The turbine is unable to turn and extract work from the over-tip leakage flow, which lowers the stage efficiency. Further, the mixing of the over-tip leakage flow in the tip gap and with the passage flow as it flows onto the suction side generates losses. The over-tip leakage flow is a highly complex secondary flow that is influenced by numerous factors including, but not limited to, blade loading, rotor width, PS tip corner radius, flow speed, tip gap, and tip design.
In an engine, tip gap of the shroudless high-pressure (HP) rotor may vary over the course of operation because of continual wear, variable centrifugal forces, thermal stress, and rearward axial movement of the shaft. An increase in tip gap enhances the over-tip leakage mass flowrate that results in higher mixing losses [1]. Hourmouziadis and Albercht [2] studied the effect of changes in the tip gap for several rotating rigs and engine tests [3]. The tip gap exchange rate, defined as the change in turbine efficiency to change in tip gap-to-span ratio , was found to be ∼2 for most of the turbines tested despite differences in stage loading and reaction. Recently, Viera et al. [4] performed pressure-sensitive paint measurements at two tip gaps in the high-speed linear cascade (HSLC) test facility at Oxford. The smaller tip gap was reported to have an overall higher tip static pressure because of a reduction in the over-tip leakage flow momentum caused by a reduction in the flow area.
Since the 1970s, various mitigation strategies have been suggested in the literature to minimize the over-tip leakage flow and tip heat load, which include novel tip designs, casing treatment, and innovative cooling techniques. Shrouded rotor blades offer significant improvements in stage efficiency of the turbine through a reduction in the over-tip leakage flow. However, the shroud area requires extensive cooling, and the weight of the shroud increases the centrifugal forces resulting in enhanced mechanical stress at the blade root and disk, which limits the maximum rotational speed. Off-loading the rotor tip has also been proven to reduce the over-tip leakage flow [5]. Staubach et al. [6] achieved a 40% reduction in tip gap exchange rate by tangentially bowing the rotor, which redistributed the loading away from the tip region toward the mid-span, resulting in a lower driving pressure gradient across the tip. However, such mechanical design changes are accompanied by increased stresses in the rotor hub, which limits the design space in an engine environment.
Finally, a better aerodynamic seal from the over-tip leakage flow can be achieved by modifying the rotor tip design. Depending on the working principle, the different tip designs can be broadly categorized into four types: flat, rib, winglet, and squealer as illustrated in Fig. 1. Winglets are an extension of the tip on either pressure side or suction side, or sometimes both sides [7]. A suction-side winglet is found to lower the pressure gradient across the pressure and suction surface of the tip, which reduces the driving force responsible for the over-tip leakage flow. Suction-side winglets are often implemented on aircraft wings to reduce the lift-induced drag caused by the wingtip vortex. A pressure-side winglet acts as an obstacle to the over-tip leakage flow and reduces its discharge coefficient. The major disadvantage with the use of a winglet in an engine is the enormous increase in the tip heat transfer surface area that makes the rotor tip extremely difficult to survive at engine conditions. A rib on the flat tip surface acts as a blockage to the over-tip leakage flow and is typically incorporated with a squealer or winglet design.
A squealer tip consists of two rims extended radially outward on the pressure and suction sides, which provide a double seal against the over-tip leakage flow. In addition, the squealer cavity (groove) acts like a labyrinth seal that increases the flow resistance and thus reduces the over-tip leakage flow for a given pressure gradient across the tip [8]. Consequently, the heat load is significantly reduced in the cavity region in comparison to a flat tip. In the event of abrasion, squealer tips feature small contact area with the casing in comparison to a flat tip, which reduces the risk of a catastrophic blade failure and hence enables operation at a tighter tip gap. The smaller tip gap in turn reduces the aerodynamic losses. Hence, the squealer is one of most prevalent tip designs for the HP turbine in an engine.
In addition, advanced tip designs have been proposed in the literature, which mostly combine features from a squealer, winglet, and rib to improve the overall aerodynamic performance. Harvey and Ramsden [9] performed a computational study of a novel “gutter winglet” tip design that featured a winglet design with a cavity following the camber line with openings toward the rotor leading edge and trailing edge. The novel design showed 1.2–1.8% improvement in stage efficiency and 31% reduction in tip gap exchange rate. However, such designs pose difficulty in cooling and feature local regions of significantly high heat transfer, which makes survival in an engine environment extremely difficult. O'Dowd et al. [10] tested a similar gutter winglet as proposed in Harvey's computational study, but with an added recessed cavity near the leading edge of the tip. The experimental study was performed in the HSLC test facility at Oxford and found large regions of extremely high Nusselt number and low film cooling effectiveness.
The majority of the current understanding of over-tip leakage flow and its mitigation strategies is based on findings from low-speed flow studies and/or linear cascade experiments. This is because of the high cost of the experiments and difficulty in instrumentation at engine conditions. Wheeler et al. [11] captured the effect of compressibility on the over-tip leakage flow by developing a quasi-three-dimensional tip gap model in which the tip gap exit Mach number (Mexit) was varied from 0.1 to 1. As the Mexit increased from 0.1 to 0.6, a gradual reduction in the size of the separation bubble was observed with a little change in the nature of the flow. Beyond Mexit of 0.8, shock waves were observed in the tip gap with drastic reduction in the size of the separation bubble. The separation bubble at the pressure-side tip corner was found to act as a convergent-divergent nozzle. Beyond the separation bubble, the increase in area in the reattachment region led to supersonic acceleration, resulting in oblique shock waves within the tip gap. Consequently, the flow behavior in the tip region was dominated by the shock-boundary layer interaction. Similar observations have been reported in other transonic tip studies [12–15] that highlight the importance of modeling and testing at engine representative Reynolds number and Mach number.
Further, the complex flow field in the tip gap is greatly influenced by the rotation of the blade. The relative motion between the casing and rotor tip results in the scraping flow that has been reported to restrict the over-tip leakage mass flow as illustrated in Fig. 2. Additionally, the interaction between the scraping and over-tip leakage flow results in a vortex, termed “scraping vortex”, near the suction-side tip of the rotor [17]. Yaras and Sjolander [18] performed an experimental investigation using relative casing motion simulated by a moving belt above the stationary rotor tip in a linear cascade. A 50% reduction in the over-tip leakage mass flowrate, along with reduction in the driving pressure gradient across the tip, was reported because of the relative casing motion. Zhang et al. [19] performed computational investigations into the effect of a moving casing for a transonic tip in a linear cascade. The relative casing motion was again reported to decelerate the over-tip leakage flow, which increased the subsonic region on the transonic tip surface.
![Illustration of scraping flow caused by the relative motion between the rotor and the casing during rotation of the blade [16]](https://asmedc.silverchair-cdn.com/asmedc/content_public/journal/turbomachinery/147/6/10.1115_1.4066965/1/m_turbo_147_6_061017_f002.png?Expires=1742297453&Signature=mKnq52jc1yxf4FT-luCXxSvhM04flzB2X4oeGUATDaG~xpO23dD-wYpBUJLYmWxS1dGL83FySlgyBs7pacgKJ1zAaRKe1KzKHv0Tpp5DB7Q4rp1r7RdtmoHnKjWTww05F5rbpCMGIIwUyqIncWPJOSn7cvWZbxbtjHyjMnNfoR9wVm1VrYd8QpXkK5qAQpynGWBVXOPU7nWgjZWZ5N2PMWG3aa2xoBH5HNVikMHo1YubQ82P2UVBPAzt17nkStM3OYkgdKwTVN3Z-PXc2G89EivVeOkLE1dRzHxoqfyxTHdX6FeYvwPyD1KrOr8aWubXZkpwjBH4crVLfybwh3yr0g__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Illustration of scraping flow caused by the relative motion between the rotor and the casing during rotation of the blade [16]
![Illustration of scraping flow caused by the relative motion between the rotor and the casing during rotation of the blade [16]](https://asmedc.silverchair-cdn.com/asmedc/content_public/journal/turbomachinery/147/6/10.1115_1.4066965/1/m_turbo_147_6_061017_f002.png?Expires=1742297453&Signature=mKnq52jc1yxf4FT-luCXxSvhM04flzB2X4oeGUATDaG~xpO23dD-wYpBUJLYmWxS1dGL83FySlgyBs7pacgKJ1zAaRKe1KzKHv0Tpp5DB7Q4rp1r7RdtmoHnKjWTww05F5rbpCMGIIwUyqIncWPJOSn7cvWZbxbtjHyjMnNfoR9wVm1VrYd8QpXkK5qAQpynGWBVXOPU7nWgjZWZ5N2PMWG3aa2xoBH5HNVikMHo1YubQ82P2UVBPAzt17nkStM3OYkgdKwTVN3Z-PXc2G89EivVeOkLE1dRzHxoqfyxTHdX6FeYvwPyD1KrOr8aWubXZkpwjBH4crVLfybwh3yr0g__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Illustration of scraping flow caused by the relative motion between the rotor and the casing during rotation of the blade [16]
In addition to the scraping flow, the effect of rotation imparts rotational forces such as centrifugal force and Coriolis force upon the fluid domain, which are not simulated in a cascade study. Yang et al. [20] performed a comprehensive computation study on the effects of rotation for a flat and squealer tip by modeling three cases: (1) “rotating domain,” i.e., rotating fluid domain with stationary casing (capturing all rotational effects), (2) “moving casing,” i.e., stationary fluid domain with counterrotating casing (capturing only scraping flow), and (3) “stationary casing,” i.e., stationary fluid domain with stationary casing (no rotational effects captured). The three cases showed significant differences in aerodynamics in the tip region. At 20% axial chord, the scraping flow was found to occupy 50% of the tip gap toward the suction side for the “moving casing,” while the “stationary casing” featured 0% scraping flow providing least resistance to the over-tip leakage flow. As shown in Fig. 2, the directions of over-tip leakage flow and scraping flow are opposite. Hence, the Coriolis force acts radially outward for the over-tip leakage flow and radially inward for the scraping flow, while the centrifugal force acts radially outward for both the flows. Since the scraping flow velocity was greater than the over-tip leakage flow velocity, the effect of rotational forces was larger for the scraping flow. As a result, the scraping flow bent inward and occupied 70% of the tip gap toward the suction side at 20% axial chord. Hence, the sealing effectiveness from the over-tip leakage flow was reported to be the highest for the “rotating domain” case. Further, Yang et al. [20] performed the same analysis for a squealer tip and found that the three flow setups showed a completely different vortex structure in the squealer cavity. In the case of the “stationary casing,” the over-tip leakage flow was found to produce a jet from the pressure-side rim and rotate in the direction from the suction side to the pressure side at the cavity floor. For the “moving casing” and “rotating domain” cases, the over-tip leakage flow was found to move inward toward the cavity floor after passing the pressure-side rim and subsequently, form a counterrotating vortex with the scraping flow above.
Few experimental studies [21–25] in the literature have been able to model test conditions such as high-speed rotation or combustor exit temperature profiles that are necessary to correctly simulate the engine environment. Thorpe et al. [21,26] performed both steady and unsteady casing static pressure measurements in the Oxford Rotor Facility using a uniform temperature profile for a flat tip rotor. The time mean measurements showed highest pressure in the gap between nozzle guide vane (NGV) and rotor, with declining pressure thereafter as the flow was accelerated in the rotating frame of reference. Time-resolved measurements showed changes in over-tip leakage flow characteristics caused by variation in rotor lift profile resulting from periodic NGV fluctuations. NGV trailing-edge shock system and potential field interaction between moving and stationary components were reported to cause significant static pressure variation along rotor suction surface. Cernat et al. [22] measured time-resolved casing static pressure resulting from three different HP rotor tip designs in a rotating test facility with a uniform stage inlet temperature profile. A rainbow rotor configuration was employed for this study, which allows testing of multiple tip designs on the same rotor disk. This strategy minimizes the run-to-run measurement discrepancies, enabling direct comparison between flow features of different tip designs. In addition, the cost and time of testing reduces significantly. The time-resolved data captured distinct flow features caused by different tip geometries. Qureshi et al. [25] measured steady casing static pressure in the Oxford Turbine Research Facility (OTRF) using MT1 turbine featuring flat tip with both uniform and nonuniform stage inlet temperature profile. The casing static pressure measurements for both inlet profiles were reported to be very similar with a mean difference of less than 2%. To the author's knowledge, there has been no previous experimental study at engine representative conditions, which has tested intricate tip designs such as squealer tips with inlet temperature distortions and at multiple tip gaps as presented in this article.
2 Experimental Setup
The experiments for this study were conducted in the OTRF, a transient test facility used for steady and unsteady aerodynamic and heat transfer investigations at engine representative conditions. Hilditch et al. [27] first described the design and working of the test facility. The OTRF is capable of matching aerodynamics and heat transfer relevant nondimensional parameters such as Mach number, Reynolds number, gas-to-wall temperature ratio, turbulence intensity, and nondimensional speed. Singh et al. [28,29] presented the commissioning of the Strategic Investment in Low-Carbon Engine Technology (SILOET) turbine used for these experiments along with a detailed description of the operating conditions. The turbine features a subsonic NGV exit and transonic rotor with 50% reaction. The complete build of the stage consists of 40 cooled NGVs (featuring both film cooling and trailing-edge slot cooling) and 60 uncooled rotor blades. For this study, three tip designs were tested on the SILOET turbine rotor at two tip gaps with two stage inlet temperature profiles.
Figure 3 illustrates the three tip designs, namely, “flat”, “baseline,” and “novel” tested in the OTRF. The flat tip has a constant radius throughout the rotor tip surface and was the original design of the SILOET turbine, described by Shahpar and Caloni [30]. The baseline tip features a double-sided squealer tip with an open trailing edge. In the near-tip region, the rotor width of the baseline blade increases in the radial direction, from around mid-chord to trailing edge, on both pressure and suction sides. The novel tip is an optimized version of the baseline tip, designed based on a series of experimental and computational fluid dynamics (CFD) studies performed in the HSLC at Oxford [31,32]. Although the baseline tip provided a good aerodynamic seal from the over-tip leakage flow, it increased the rotor tip heat load. The novel tip was designed as a solution, featuring a suction-side cut-out and downward sloping squealer cavity that reduces the heat load and enhances the film cooling effectiveness near trailing edge, with some loss in the aerodynamic performance.
The three different tip designs were distributed in groups on the rotor disk to form a “rainbow” build, for the experiments in the OTRF, as illustrated in Fig. 4. The rainbow rotor build immensely reduces both time and cost of testing by allowing multiple tip designs to be tested in the same run. Within the rainbow rotor, the blades were distributed such that similar tip designs could be grouped together, and overall mass of the rotor disk could be balanced during the high-speed rotation.
A variety of stage inlet temperature profiles including radial distortion and swirl can be produced in the OTRF. For the present investigation, the rig was operated at two different inlet temperature profiles, namely, (1) uniform, i.e., a spatially constant total temperature profile at stage inlet, and (2) radial temperature distortion factor (RTDF), i.e., a radially varying total temperature profile at the stage inlet [28,29]. The RTDF profile is representative of radial temperature distortions, at the HP turbine inlet of a modern engine, caused by combustor exit profile and extensive platform cooling upstream of the NGVs. In addition, the experiments were conducted at two rotor tip gaps, namely, (a) nominal tip gap (0.6% span), i.e., representative of design tip gap in an engine and (b) large tip gap (1.3% span), i.e., representative of the increased tip gap over time, during in-service operation of an engine. The flat tip design was not tested at the large tip gap condition.
2.1 Steady Pressure Instrumentation.
The steady static pressure measurements at the rotor casing were obtained using pneumatic lines instrumented with Sensor Technics CTE8000 pressure transducers with a quoted nominal accuracy of ±0.1%. An over-tip casing block was manufactured from Perspex with 90 pneumatic tappings that were flush fitted to the casing surface of the turbine above the rotor. The block was assembled into the top casing cassette of the OTRF. A frequency analysis of the data from pneumatic lines, used for casing static pressure measurements, showed the presence of three main frequencies: (1) the shape of the run in OTRF at ∼3 Hz, (2) piston oscillations at ∼16 Hz, and (3) pressure fluctuations due to reflection of pressure waves travelling through the facility at ∼70 Hz. The uncertainty associated with the calibration and the data acquisition system for the pneumatic lines was calculated as ±0.019%. Figure 5 illustrates the schematic of time-averaged casing static pressure measurement instrumentation. The 90 pneumatic tappings on the over-tip block were distributed in an array of 6 axial rows and 15 circumferential positions. The measurement region covered two NGV pitches circumferentially and ∼84% rotor tip chord in the axial direction. The NGV wake lines have been plotted in the schematic based on the design flow whirl angle at NGV exit.

Schematic of time-averaged casing steady pressure measurement instrumentation. Circles indicate the position of pneumatic tappings.
2.2 Unsteady Pressure Instrumentation.
Kulite XCQ-062 high-frequency response pressure transducers were employed for time-resolved casing static pressure measurements. With a frequency response of ∼100 kHz, these pressure transducers are able to resolve the periodic fluctuations in the casing static pressure field caused by rotor passing at ∼8.5 kHz. A passive temperature compensation unit is attached to the sensors, which accounts for changes in the resistive impedance of the piezo-resistive elements caused by temperature fluctuations. Ainsworth et al. [33] provided a detailed description of unsteady pressure measurements using Kulite sensors in turbomachinery and reported a measurement uncertainty of ±0.2% for the Kulite measurements with temperature compensation units.
An over-tip block with Kulite instrumentation was designed to fit in the same casing cassette arrangement as that used for the steady pressure measurements. The Perspex over-tip block was drilled with radial holes that were distributed in the axial direction and aligned along the rotor true chord. Figure 6 illustrates the schematic of the five Kulite pressure transducers, covering 86.2% rotor tip axial chord. The three tip designs have been drawn in different colors and one rotor pitch has been placed apart as shown in Fig. 6. As shown further in Sec. 5, time-resolved casing static pressure fields, unique to a tip design, could be extracted from the Kulite data acquired in a rainbow rotor build.

Schematic of time-resolved casing pressure measurement instrumentation. Asterisks indicate the position of Kulite sensors.
The output voltage from each Kulite pressure transducer was amplified using a Fylde FE-579-TA bridge amplifier (a low noise differential amplifier that can be used to provide a constant gain value in full bridge configuration). Gain for each transducer was set to maximize the dynamic range of the data acquisition system. The Kulite pressure transducers were calibrated using a dead-weight tester, with each transducer connected to its respective amplifier channel. Multiple passes were used to check for hysteresis. A consistent linear response was observed, and the uncertainty from calibration was calculated as ±0.05% using 95% confidence interval on the linear fit.
3 Computational Fluid Dynamics Setup
The HP turbine under investigation had 40 NGVs and 60 rotor blades, which necessitated a 2:3 NGV-to-rotor blade ratio in the CFD domain. The NGV and rotor domains were meshed using BOXERmesh, which is an unstructured meshing tool, developed by Cambridge Flow Solutions Ltd. The tool creates an octree computational grid with body-fitted prismatic layers. The number of body-fitted layers, layer thickness, and expansion ratio were adapted, such that the y+ could be kept less than 1 on the rotor tip and the casing surface. The meshing strategy was informed through previous mesh accuracy and grid independence studies [34–38]. The overall computational domain comprised 18.5 million cells. Figure 7 illustrates the mesh on the baseline rotor surface along with a sectional cut showing extensive refinement near the walls, especially for the near-tip aerofoil surface and rotor casing surface.

Computational grid, generated using boxer mesh, showing (a) rotor surface along with the hub endwall and (b) an axial sectional cut through the rotor, illustrating extensive refinement near the aerofoil and casing walls
The CFD solutions were obtained using HYDRA, a Rolls-Royce in-house solver, which is based on preconditioned time marching of Reynolds-averaged Navier–Stokes [34,37]. The code uses an edge-based finite volume method with a second-order upwind discretization scheme to spatially solve the governing flow equations. The steady-state convergence is improved by using multigrid and local time-stepping acceleration techniques [34]. The shear stress transport (SST) model was used for turbulence closure, which combines elements of both and turbulence models. The model retains the advantages of in near-wall regions, ensuring its applicability down to the wall through viscous sublayer and the advantages of in the freestream region, thereby avoiding the problem of sensitivity to inlet freestream turbulence properties. This balanced approach enables accurate predictions of both near-wall and far-field turbulence effects. Further, the SST model has shown to accurately predict secondary flows and heat transfer in turbomachinery and has also been extensively validated against previous experimental data [34]. An unsteady nonlinear HYDRA model was used to perform unsteady time-varying flow simulations. The unsteady solver employs the explicit Runge–Kutta time-marching scheme. In total, 120 time-steps were solved, and 20 time-steps were written for each unsteady computation. Convergence was first checked on the steady-state solutions that were used as the initial boundary condition for the unsteady solutions. The mixing plane technique was employed at the interface of the NGV and rotor for steady-state solutions, while the sliding plane technique was employed for the unsteady solution to transfer data at each time-step.
The boundary conditions in the CFD model were adapted to match the measured operating conditions in the OTRF. A comparison of key operating parameters between experiments and CFD, at both Uniform and RTDF inlet, is summarized in Table 1. The experimental parameters were calculated as average values over 16 consecutive runs in the OTRF. With uniform inlet, the stage inlet total temperature and pressure were set spatially constant at measured values of 469.4 K and 5.4 bar, respectively. With RTDF inlet, the NGV inlet total temperature was set as a radial profile that was calculated from the circumferential average of total temperature survey data acquired at the inlet of the NGV [28].
Comparison of key operating parameters between experiments and CFD, for both uniform and RTDF inlet conditions
Parameter | Uniform inlet | RTDF inlet | ||||
---|---|---|---|---|---|---|
CFD | Experiments | CFD | Experiments | |||
Mean (16 runs) | % Standard deviation (16 runs) | Mean (16 runs) | % Standard deviation (16 runs) | |||
469.4 | 469.4 | ±0.23 | 470.4 | 472.3 | ±0.81 | |
5.4 | 5.4 | ±0.32 | 5.4 | 5.4 | ±0.49 | |
2.58 | 2.58 | ±0.45 | 2.58 | 2.58 | ±0.34 | |
392.33 | 392.33 | ±0.50 | 391.93 | 391.97 | ±0.27 | |
0.75 | 0.75 | ±1.06 | 0.75 | 0.75 | ±0.99 | |
1.84 × 106 | 1.85 × 106 | ±0.72 | 1.84 × 106 | 1.85 × 106 | ±0.48 | |
67.27 | 67.38 | ±0.37 | 67.32 | 67.81 | ±0.43 | |
1.50 | 1.51 | ±0.36 | 1.49 | 1.51 | ±0.29 |
Parameter | Uniform inlet | RTDF inlet | ||||
---|---|---|---|---|---|---|
CFD | Experiments | CFD | Experiments | |||
Mean (16 runs) | % Standard deviation (16 runs) | Mean (16 runs) | % Standard deviation (16 runs) | |||
469.4 | 469.4 | ±0.23 | 470.4 | 472.3 | ±0.81 | |
5.4 | 5.4 | ±0.32 | 5.4 | 5.4 | ±0.49 | |
2.58 | 2.58 | ±0.45 | 2.58 | 2.58 | ±0.34 | |
392.33 | 392.33 | ±0.50 | 391.93 | 391.97 | ±0.27 | |
0.75 | 0.75 | ±1.06 | 0.75 | 0.75 | ±0.99 | |
1.84 × 106 | 1.85 × 106 | ±0.72 | 1.84 × 106 | 1.85 × 106 | ±0.48 | |
67.27 | 67.38 | ±0.37 | 67.32 | 67.81 | ±0.43 | |
1.50 | 1.51 | ±0.36 | 1.49 | 1.51 | ±0.29 |
Note: Experimental values are averaged over 16 runs for each inlet condition.
Figure 8 shows the radial profiles of NGV inlet total temperature implemented in CFD, for uniform and RTDF conditions, along with the circumferential average of the measured RTDF profile. Turbulence intensity at stage inlet was set at 6.5% based on previous hot wire anemometer measurements in the OTRF [39]. The rotor exit static pressure measurements at hub and casing were linearly interpolated to produce a radial profile of static pressure that was used as an outflow boundary condition in the CFD. The measured disk speed was used to specify the rotational speed in CFD. The film cooling on the NGV aerofoil surface was modeled as source terms, while the trailing-edge slot cooling was modeled as an inflow boundary condition. The walls were set at the isothermal condition, with a viscous wall boundary condition.
4 Time-Averaged Casing Static Pressure Results
Low-frequency static pressure data, acquired using pneumatic tappings, were time averaged in the stable portion of a run in the OTRF to obtain a spatial distribution of casing static pressure. Given the distribution of three tip designs on the rotor disk in a rainbow build, the measured static pressure on the casing would be a weighted average of the different tip designs at any measurement location. The weights being the blade count for each tip design in the rainbow rotor build. The blade counts for the nominal tip gap experiments were as follows: 24 flat, 18 baseline, and 18 novel, while those for the large tip gap experiment were 30 baseline and 30 novel. Separate CFD models were built for each tip design and tip gap. Hence, CFD solutions from multiple tip designs, at a given tip gap, were weighted averaged to allow direct comparison with the steady experimental data.
4.1 Map of Casing Static Pressure.
Figure 9(a) illustrates the contour of measured casing static pressure, normalized by NGV inlet total pressure (P01), with uniform inlet at the nominal tip gap. For the purpose of plotting, the data have been linearly interpolated between the measurement locations, indicated by circles, in the contours. Figure 9(b) shows the corresponding CFD contour that was generated by first time averaging the unsteady CFD solutions in the stationary frame of reference for each tip design and then averaging the time-averaged solutions, from multiple tip designs, weighted by the blade count in the experimental rainbow rotor build. The process of time averaging in the stationary frame of reference of CFD averages out the pressure fluctuations at the casing caused by rotor passing. Hence, the CFD results can be directly compared with the experimental results.

Contour plots of casing static pressure with uniform inlet at the nominal tip gap: (a) experimental measurements from the rainbow rotor build and (b) time-averaged unsteady CFD predictions in stationary frame of reference
The static pressure at the casing is the highest at the rotor inlet and reduces by ∼1.7 bar (31.5% of P01), in the axial direction at rotor exit, due to the rotor work extraction. The largest reduction in the static pressure is observed between 35% and 65% axial chord, which is the region experiencing the largest flow acceleration. The circumferential variation is shown in Fig. 10 on a different scale to highlight the effect of the NGVs. Experimental data show a maximum circumferential difference of ∼0.15 bar (2.8% of P01) at the rotor inlet. The circumferential variation in the time-averaged measurements reduces in the axial direction and is only prominent till ∼10% rotor axial chord. Observing the experimental data in isolation gives an illusion of casing static pressure being higher in line with the wake of the NGVs. Figure 10(b) produced by extending the axial extent of the CFD results on the same scale gives a more comprehensive picture. First, CFD predictions show low casing static pressure in the direction of the NGV wake, which was expected because of the pressure loss due to flow separation in the wake region. Second, the increase in casing static pressure observed at the rotor inlet is a result of the interaction between the NGV and rotor passing as described further.

Contour plot of casing static pressure for uniform inlet at the nominal tip gap highlighting the NGV wake effect: (a) experimentally measured from the rainbow rotor build and (b) time-averaged unsteady CFD predictions in stationary frame of reference
Thorpe et al. [21] studied the complex unsteady interaction between rotor and NGV associated with the high Mach number flow at NGV exit and periodic blade passing. Three main physical mechanisms explained the interaction: (1) potential field interaction that modulates the casing static pressure field for a given rotor position, (2) NGV trailing-edge shock interaction, and (3) the total pressure and velocity deficits in the wake of the NGV. A detailed study of the unsteady interaction between NGV and rotor is out of the scope of the measurements presented in this article. The time-resolved casing static pressure measurements (presented later in Sec. 5) were acquired using an axial array of Kulites, which could only capture the unsteady pressure field due to rotor passing. With the lack of circumferential distribution of the unsteady pressure data, the rotor phase-specific interaction with the NGV could not be studied. However, the interaction between the rotor and NGVs could be inferred from the time-averaged measurements shown in Fig. 10. An increase in static pressure is observed as the rotor passes the NGV trailing edge. Thorpe et al. [21] explained such an increase with the presence of a shock or succession of small compression waves that form due to reduction in effective throat area between NGV trailing edge and rotor leading edge.
4.2 Comparison of Measured Circumferentially Averaged Profiles.
The axial variation in time-averaged casing static pressure measurements was significantly higher than the circumferential variation caused by the NGVs. In addition, the circumferential variation was found to be very similar for both uniform and RTDF inlet conditions at both nominal and large tip gaps. Figure 11 compares circumferentially averaged experimental data for different inlet conditions and tip gaps. The experimental uncertainty of ±0.1% in the casing static pressure measurement is indicated on the plot using vertical error bars. However, the uncertainty values are too small and concealed in the scale of the plot shown.

Circumferential average of measured casing static pressure for two tip gaps and two inlet temperature profiles
Casing static pressure was found to be 1.3% higher on average with the nominal tip gap in comparison to the large tip gap, at both uniform and RTDF inlet conditions. One of the consequences of decreasing the tip gap is the increase in static pressure caused by the reduced over-tip leakage flow momentum. De Maesschalck et al. [40] performed a numerical study to understand the effect of tip gap on the aerothermodynamics of a high-speed rotating unshrouded turbine. The study also reported an increased static pressure at the rotor tip and casing surface with a reduction in the tip gap. Another explanation for the observed increase in casing static pressure with the smaller tip gap could be derived from the findings of Thorpe et al. [21]. The study found that the turbine rotor could perform isentropic work on the tip gap flow. Additionally, the authors argued that such work processes in the tip gap could only be identified through high-speed rotating facilities. A smaller tip gap would result in a smaller compression volume for the rotor, which would lead to an increase in pressure according to the fundamentals of the isentropic work process.
A smaller tip gap provides a better aerodynamic seal by restricting the over-tip leakage mass flowrate. In the OTRF, a reduction in the tip gap reduces the effective flow area through the rotor domain, which decreases the rotor mass flow capacity. In addition, the OTRF was operated to maintain the same stage pressure ratio across all test conditions. As a result, the pressure ratio (P/P01) in Fig. 11, can be seen to converge to a common value toward the rotor exit. The reduction in the rotor mass flow capacity through reduction in tip gap could be further confirmed by comparing the time duration of runs between the two tip gaps. Upon inspection, the nominal tip gap experiments were found to have a slightly longer stable run period in comparison to the large tip gap experiments. Given that the same quantity (mass) of test gas is compressed by the piston tube for experiments at both the tip gaps, the mass flowrate at the nominal tip gap must decrease to justify the observed increase in the run duration.
The casing static pressure was found to be 0.4% and 0.2% higher on average with the uniform inlet compared to the RTDF inlet at the nominal and large tip gap, respectively. These differences, between uniform and RTDF inlet, are likely to be a result of changes in the rotor secondary flows, caused by radial temperature distortions in the stage inlet profile. A detailed discussion of the aerothermodynamics in the near-tip region along with casing and rotor tip heat transfer measurements will be presented in a subsequent paper.
4.3 Comparison With Computational Fluid Dynamics Results.
The HP turbine casing at NGV exit (−35.5% rotor tip axial chord) was instrumented with static pressure tappings to monitor the OTRF run conditions. Data from these tappings were compiled with the rotor casing pressure measurements to extend the axial coverage of the experimental results for comparison with CFD predictions. Figure 12 compares the experimental results with the circumferential average of the time-averaged unsteady CFD predictions in the stationary frame of reference. The CFD solutions from different tip designs were weighted averaged, by the blade count in the experimental rainbow build, to allow a direct comparison with the time-averaged measurements. Similar differences between experiments and CFD were observed at both inlet conditions. Hence, the comparison is only shown for the nominal tip gap case with uniform inlet for the reason of brevity. Like the trend observed in the experimental results, CFD predictions also showed higher casing static pressure at the nominal tip gap in comparison to the large tip gap. Overall, a good match between experimental results and CFD predictions is found, except for some discrepancy at ∼17.8% axial chord where CFD is found to underpredict the casing static pressure, by 2.1% at the nominal tip gap and 2.8% at the large tip gap.

Comparison of circumferentially averaged casing static pressure between experimental results and time-averaged unsteady CFD predictions at two tip gaps for uniform inlet
This discrepancy was further investigated by comparing the CFD predictions between different tip designs as shown in Fig. 13. CFD predictions of both types of squealer tips show an accelerated drop in static pressure relative to the flat tip, from 0% to 35% axial chord. The experimental results match better with the flat tip CFD predictions in this region. These observations indicate that CFD is underpredicting the casing static pressure for the squealer tip geometries in the region of discrepancy. A similar discrepancy was observed when unsteady CFD predictions were compared with the time-resolved casing static pressure measurements as discussed further in Sec. 5.

Comparison of casing static pressure predictions between different rotor tip designs, with uniform inlet at the nominal tip gap
Beyond the region of discrepancy, it is interesting to note the differences in casing static pressure due to different tip designs. The design intent of the baseline tip was to provide a good aerodynamic seal by minimizing the over-tip leakage flow. The aerodynamics of this double-sided squealer reduces the flow acceleration in the tip gap causing an increase in casing static pressure, described further in the time-resolved results. The novel tip, on the other hand, enhances the flow acceleration with no suction-side squealer rim and a downward sloping squealer cavity that further increases the effective tip gap. The flat tip CFD predictions remain close to the baseline tip CFD predictions between 30% and 60% axial chord but show a rapid reduction in casing static pressure beyond 60% axial chord.
5 Time-Resolved Casing Static Pressure Results
This section presents the unsteady static pressure measurements at the rotor casing obtained using Kulite XCQ-062 pressure transducers. The unsteady pressure measurements were also conducted in the rainbow rotor build to save both the cost and the time of testing. The blade counts for the nominal tip gap experiments were as follows: 28 flat, 16 baseline, and 16 novel, whereas those for the large tip gap experiments were 30 baseline and 30 novel. Figure 14(a) shows a typical voltage signal from a Kulite transducer instrumented on the rotor casing at 53% axial chord. Figure 14(b) illustrates a magnified view showing three rotor passes. There are approximately 35 rotor revolutions during the stable portion of the run in the OTRF, and each revolution consisted of 60 rotor passes with different tip designs.

Typical pressure data, sampled at 1 MHz, from high-frequency response Kulite pressure transducer: (a) during a run in the OTRF with stable portion indicated by vertical lines and (b) magnified view to show three rotor passes
5.1 Ensemble and Pitch-Averaging Process.
Using the calibration data, measured voltage from the Kulite transducer was reduced to time-varying static pressure. The pressure data and rotor rotational speed data were correlated in time to extract individual rotor revolutions and blade passes. The rotational speed was recorded using optical encoders that measured the angular position of a toothed speed disk mounted on the main shaft. 1-line and 60-line toothed speed disks were used, which allowed speed measurement once per revolution and 60 per revolution, respectively. Both the Kulite data and rotor speed data were triggered and acquired using the same data acquisition system. By using the one-line speed disk data, Kulite pressure data could be separated for each revolution of the rotor. The pressure could then be averaged over all the revolutions of the rotor, within the stable portion of a run, to obtain an ensemble average as shown in Fig. 15 for the nominal tip gap rainbow rotor build. The presence of three tip designs can be distinctly identified as each tip design produces a unique casing static pressure field. The ensemble-averaged data show small variations, for different rotor blades of the same tip design, which could be caused by manufacturing tolerances between different rotor blades.

Ensemble-averaged Kulite pressure data, at 10.2% rotor tip axial chord, for nominal tip gap experiments with uniform inlet, distinctly showing three tip designs distributed over 60 rotor blades
Further, averaging the pressure data over all rotor blades of the same tip design over multiple revolutions of the rotor gives the pitch-averaged value. The pressure data, in the transition region between rotor blades of different tip designs in the rainbow build, were removed during the pitch-averaging process. Figure 16 shows the pitch-averaged casing static pressure, at 10.2% axial chord location, for the flat tip blade obtained as an average of 875 rotor passes (i.e., 35 rotor revolutions multiplied by 25 flat tip blades). The differences in pitch-wise casing static pressure, observed between different rotor passes of the same tip design, was calculated as 1.36% peak-to-peak or ±0.22% standard deviation (pitch-wise averaged). This difference is likely to be associated with manufacturing and assembly tolerances between different rotor blades.

Pitch-averaged Kulite pressure data for the flat tip blade, at nominal tip gap with uniform inlet. Different colors show multiple flat tip rotor passes over multiple rotor revolutions during the stable portion of the run in OTRF.
The rotor speed disk was referenced manually to one rotor position, during the rainbow build of the turbine disk, when the optical encoder square wave data went from high to low. Although this method was able to identify and label different rotor tip designs, it was not accurate enough to identify the correct phase of the blade, to fix its position in the pitch or ensemble-averaged data. Hence, unsteady CFD predictions of casing static pressure were cross correlated to the experimental data at several Kulite measurement locations to infer the exact phase of the rotor.
5.2 Measured Contours and Comparison With Computational Fluid Dynamics.
Figure 17 illustrates the contour plots of experimental data and CFD predictions of casing static pressure, normalized by NGV inlet total pressure (P01), for the three tip designs at the nominal tip gap with uniform inlet. The experimental contours are generated by linearly interpolating the pitch-averaged data between measurement locations. The CFD contours, shown in Figs. 17(d)–17(f), are time-averaged unsteady CFD predictions in the rotating frame of reference. The x-axis for experimental contours has been labeled as “rotor passing pitch,” while for CFD contours, the x-axis has been labeled as “rotor frozen pitch.” In the absolute frame of reference, the rotor is rotating in the experiments, while the rotor is stationary and the fluid domain is rotating in the CFD, hence, the nomenclature. In general, CFD predictions show good qualitative agreement with the experimental results for all three tip designs. The experimental contours show a gradual change in static pressure around the aerofoil boundaries, while the CFD predictions show a distinct and sharp pressure field. The difference could be associated with having perfectly sharp rim edges in the CFD predictions compared to experimental rotor blades that on inspection were found to have small radii around the corners of the tip rims due to manufacturing tolerances. The pressure-side corner radius of the rotor tip controls the over-tip leakage discharge coefficient, separation bubble size, and the subsequent reattachment region on the aerofoil. Bindon and Morphis [41] found that using the tip PS corner makes the separation bubble smaller, resulting in a smooth entry of the flow in the tip gap and a consequent reduction in mixing losses. However, this reduction in tip gap mixing loss came at the cost of increased over-tip leakage mass flowrate, which was reported to enhance the downstream mixing losses resulting in an overall lower aerodynamic performance. Hence, some discrepancies between experiments and CFD predictions could be associated with the difference in the geometry tested versus the geometry modeled. A sensitivity of the present CFD models around tip corner radii of the aerofoil is suggested for the future work.

Contour plots of casing static pressure for three tip designs at the nominal tip gap with uniform inlet from experimental data (a)–(c) and time-averaged unsteady CFD predictions in rotating frame of reference (d)–(f). Horizontal dashed lines in CFD contours mark the axial extent of experimental measurements. (a) Flat experiment, (b) baseline experiment, (c) novel experiment, (d) flat CFD, (e) baseline CFD, and (f) novel CFD.

Contour plots of casing static pressure for three tip designs at the nominal tip gap with uniform inlet from experimental data (a)–(c) and time-averaged unsteady CFD predictions in rotating frame of reference (d)–(f). Horizontal dashed lines in CFD contours mark the axial extent of experimental measurements. (a) Flat experiment, (b) baseline experiment, (c) novel experiment, (d) flat CFD, (e) baseline CFD, and (f) novel CFD.
In addition to the differences around aerofoil boundaries, there are subtle differences in the casing static pressure field between experiments and CFD predictions. For instance, the experimental results show a localized region of high casing static pressure near the aerofoil pressure side, at around 25% axial chord for all three tip designs, which is not exhibited by the CFD predictions. These differences are likely to be associated with differences in the NGV–rotor interaction between experimental results and CFD predictions. Time averaging the unsteady CFD solutions in the rotating frame of reference averages the circumferential pressure variations at casing caused by the NGV–rotor interaction. However, the experimental data were acquired from a single row of five Kulites distributed axially in the chord wise direction. Unlike time-averaged CFD results, the experimental results will be affected by the presence of NGVs since each Kulite experiences flow from a unique location on the casing relative to the NGV. Measurements by Thorpe et al. [21] showed that the casing aerodynamics is significantly influenced by the NGV–rotor interaction. The casing static pressure contours could be pulled forward or pushed backward in the axial direction depending on the rotor position relative to the NGV. Thorpe's findings could explain why the experimental contours, in Fig. 17 (especially in the passage near suction side of the aerofoil), look pulled forward relative to the CFD contours.
The following discussion, divided into three sections, presents a detailed analysis of the time-resolved casing static pressure results. The first part focusses on the aerodynamics characteristic of an unshrouded tip that causes the static pressure field at the casing as observed in Fig. 17. The second part focusses on comparison of near-tip aerodynamics across the three tip designs. Finally, the third part presents the experimental data and unsteady CFD predictions as pressure variations over a rotor pitch, allowing a detailed comparison across the three tip designs, the two tip gaps, and the two inlet temperature profiles.
5.3 Aerodynamics Characteristic of an Unshrouded Transonic Tip.
The static pressure field at the casing of unshrouded transonic tips, as illustrated in Fig. 17, is a result of three major factors: (1) rotor blade loading (lift) distribution, (2) NGV–rotor interaction, and (3) over-tip leakage flow. These factors are interdependent and govern the aerodynamics at the casing and the near-tip region. Thorpe et al. [21] measured time-resolved casing static pressure in the Oxford Rotor Facility using an axial and circumferential array of Kulites. The results from their study showed that the pressure distribution around the aerofoil changes significantly with a change in the rotor position relative to the NGV. The resultant change in pressure distribution across the pressure and suction side consequently changes the amount and the direction of the over-tip leakage flow.
All three tip designs feature a steep pressure gradient along the pressure surface of the aerofoil (marker A), which is responsible for flow acceleration into the tip gap. Along the suction surface of the aerofoil, large pressure gradients (that are almost parallel to the tip profiles) are present up to 50% axial chord (marker B). The geometrical throat of the blade row intercepts the suction surface of the aerofoil at this location, and the pressure gradient reduces significantly beyond 50% axial chord (marker C). In the rotor passage, the flow accelerates and expands as the rotor extracts work. The migration of passage boundary layer from the pressure surface to the suction surface of the aerofoil can be seen as the passage pressure contours stretch toward the aerofoil early suction side (marker D). This is caused by the relative endwall motion and cross-passage pressure gradient [42].
5.4 Comparison Across Different Tip Designs.
The differences in the casing static pressure for the three tip designs, as seen in the contours of Fig. 17, are discussed further in this section by studying the near-tip aerodynamics of the flat, baseline, and novel tips through CFD predictions, as shown in Fig. 18. From left to right, each figure shows tip surface static pressure with flow streamlines, cuts along the rotor showing relative total pressure, and cuts along the rotor showing the relative Mach number. The flat tip features a distinct pressure gradient in the axial direction on both the casing (Fig. 17) and tip surface (Fig. 18(a)), which is caused by the blade loading around the aerofoil and is typical of an unshrouded flat tip, as also reported by Thorpe et al. [21]. The baseline tip, on the other hand, features an almost constant static pressure in most of the squealer tip cavity and the pressure-side rim. This observation is common to the aerodynamics of double-sided squealer tips and was also reported by Cernat et al. [22]. The novel tip features the lowest casing static pressure in the over-tip region. A lower casing static pressure implies higher acceleration of the flow from the pressure side into the tip gap. Conversely, a higher pressure in the tip gap region is indicative of a good aerodynamic seal against the over-tip leakage flow. The baseline tip would, thus, rank better than the flat tip, which would rank better than the novel tip in the terms of minimizing flow acceleration in the tip gap. However, additional efficiency measurements are required for a detailed aerodynamic ranking of the three tip designs.

CFD predictions, at nominal tip gap and uniform inlet, for (a) flat, (b) novel, and (c) baseline tip designs, showing tip surface static pressure with surface streamlines, cuts with contours of rotor relative total pressure, and cuts with contours of relative Mach number
From around 50% axial chord till the trailing edge, both flat and novel tips show regions of low casing static pressure on the suction side and over the rotor tip (blue region in Fig. 17). The reduction in static pressure is caused by an overexpansion of the flow [42] to supersonic Mach numbers in the tip gap as illustrated through the contour cuts shown in Fig. 18. A comparison of the tip surface streamlines shows the complexity of the aerodynamics for the two squealer tips in comparison with the flat tip. The streamlines follow a smooth trajectory from the pressure side to the suction side of the flat tip blade, while the baseline and novel tip feature an impingement of high-pressure flow in the squealer cavity at ∼15% axial chord.
Rotor relative pressure and relative Mach number contours at various locations along the rotor show how the tip leakage flow separates at the pressure side of the aerofoil and enters the tip gap. The flow reattachment location in the tip gap is dependent on the blade width-to-tip gap ratio, rotor tip pressure-side corner radius, and the local Reynolds number in the tip gap [1,41,43]. For the flat tip, the flow reattaches in the tip gap till ∼65% chord length beyond which the flow does not reattach and directly entrains into the over-tip leakage vortex. For the baseline and novel squealer tips, the flow structures are complex and include the formation of many vortices as indicated through localized regions of low rotor relative total pressure in the tip gap. The different vortices have been labeled as cavity vortex, scraping vortex, and corner vortex in the literature. Numerous CFD studies have been published that show how the vortices in the squealer tip cavity increase the sealing effectiveness in the tip gap [44–47]. However, there seem to be no consistent conclusion on the physical mechanism controlling the over-tip leakage flow.
The evolution of the over-tip leakage vortex in the axial direction can be seen through the contours of rotor relative total pressure on the suction side. The vortex core is characterized by the low rotor relative total pressure. The extent of the low-pressure region on the suction side is indicative of the size of the over-tip leakage vortex. For all three tip designs, the size of the over-tip leakage vortex increases in the axial direction due to the loss core separation in the positive suction corner pressure gradient. The baseline tip has the smallest over-tip leakage vortex at most axial locations, in comparison with flat and novel, which is indicative of minimal mixing loss and good aerodynamic efficiency of the baseline tip. As can be seen through the tip surface streamlines, the squealer cavity for the baseline tip channels the part of the over-tip leakage flow through the cavity opening at the trailing edge. Both novel and flat tip show low casing static pressure (blue region) toward the trailing edge of the blade, while the baseline tip shows a pressure gradient as the cavity flow accelerates and exits through the opening at the trailing edge of the double-sided squealer.
5.5 Comparison in Pitch-Wise Line Plots.
The last section showed that the aerodynamics at the casing and near-tip region was significantly different for the three tip designs. However, the differences in time-resolved casing static pressure measurements with a change in the tip gap or a change in the inlet condition were subtle and difficult to capture through contours, such as the ones shown in Fig. 17. Such differences were better interpreted in the pitch-wise line plots, as shown in Fig. 19. The plots show the casing static pressure variation over one rotor pitch at a given Kulite axial measurement location. Time-averaged (in rotating frame of reference) unsteady CFD predictions were extracted at the same axial locations and compared with experimental results. No experimental data were acquired for the large tip gap with RTDF inlet because of the shortage of testing time available on the OTRF. In addition, the Kulite sensor instrumented at 75% axial chord got damaged during the large tip gap testing, and hence, no large tip gap experimental data were available at this axial location. Following information is given to help the reader in interpreting the pitch-wise line plots as shown in Fig. 19:
Each of Figs. 19(a)–19(e) corresponds to one of the five axial locations on the casing that were instrumented with Kulite pressure transducers.
Each subfigure of Fig. 19 has three plots for the three tip designs: Fig. 19(a): flat, Fig. 19(b): baseline, and Fig. 19(c): novel. The left y-axis label has been displayed in the leftmost plot, while the right y-axis label has been displayed in the rightmost plot.
Aerofoil sections from the rotor tip have been drawn using the right y-axis, and the x-axis covering one rotor pitch.
The axial location of the comparison has been drawn using the right y-axis as a horizontal dashed-asterisk line.
The static pressure ratio has been plotted over one rotor pitch using the left y-axis and x-axis. The scale of the left y-axis has been adapted for each axial location to focus on the range of pressure variation. The scale is, however, kept the same for all three tip designs at a given axial location for ease of comparison.
The legend has been put on the top left of Fig. 19 with experimental results shown in dashed lines and CFD predictions shown in solid lines.
For correlating pitch-wise variation of static pressure with the phase of the rotor blade, the axial location of the measurement (indicated by horizontal dashed-asterisk line) should be traced along the rotor pitch (x-axis). For example, the intersection of dashed-asterisk line with aerofoil gives the pitch range for which the casing static pressure is plotted for the rotor tip region. The pitch range outside the intersection of aerofoil and dashed-asterisk line corresponds to the passage region.


Experimental and CFD plots of casing static pressure at two inlet conditions and two tip gaps for flat, baseline, and novel tip designs at various axial locations. CFD predictions have been extracted at the same axial location as Kulite pressure transducers. (a) 10.2% rotor tip axial chord, (b) 31.8% rotor tip axial chord, (c) 53.3% rotor tip axial chord, (d) 74.9% rotor tip axial chord, and (e) 96.4% rotor tip axial chord.


Experimental and CFD plots of casing static pressure at two inlet conditions and two tip gaps for flat, baseline, and novel tip designs at various axial locations. CFD predictions have been extracted at the same axial location as Kulite pressure transducers. (a) 10.2% rotor tip axial chord, (b) 31.8% rotor tip axial chord, (c) 53.3% rotor tip axial chord, (d) 74.9% rotor tip axial chord, and (e) 96.4% rotor tip axial chord.
5.5.1 Comparison Between Experiments and Computational Fluid Dynamics Predictions.
In general, a good match in trends is observed between time-resolved experimental measurements and unsteady CFD predictions with some localized discrepancies. The experiments show a gradual change of static pressure around the aerofoil in the pitch-wise direction, while the CFD predictions show distinct and sharp changes particularly at the aerofoil boundaries. As discussed earlier, the difference could be associated with having perfectly sharp tip in the CFD geometry, compared to the experimental rotor blades that were found to have a small radius at the corners of the tip surface due to manufacturing tolerance.
The maximum discrepancy between CFD predictions and experimental results is observed in Fig. 19(a) that plots the time-resolved casing static pressure at 10.2% axial chord. CFD predictions show a large deficit in casing static pressure in the rotor tip region compared to the experiments. In this region, the novel tip shows the largest deficit of ∼30%, followed by ∼23% for baseline and ∼11% for flat. It is important to note that the discrepancy is significantly higher for the squealer tips compared to the flat tip. Comparison of circumferentially averaged CFD predictions of different tip designs with the steady pressure measurements, in Sec. 4, also showed a similar deficit in casing static pressure predictions for baseline and novel tips. It was initially suspected that the mesh in the tip gap region of both the baseline and novel tip was not able to resolve the complex flow structures associated with squealer tips. To test the hypothesis, the mesh for the baseline model was extensively refined especially in the tip gap region. As a result, the overall mesh size increased from 18.5 million cells to 44.6 million cells. However, CFD predictions even with the significantly refined mesh showed no improvements in the pressure deficit observed around 10.2% axial chord. Further investigation is needed in the CFD models to better understand the pressure discrepancy observed with squealer tip geometries. The first recommendation would be to model rotor tips with nonsharp edges as found in the experiments, and the second recommendation would be to evaluate the CFD solutions with different turbulence models.
5.6 Comparison Between Flat, Baseline, and Novel Tips.
The effect of blade passing is distinctly visible in the pressure variation as shown through the pitch-wise line plots. In the passage region, casing static pressure reduces gradually from the aerofoil pressure surface to the suction surface. The three tip designs show significantly different casing static pressure over the tip region but show very similar flow characteristics in the passage region till 53.3% axial chord. Beyond 53.3% axial chord, casing static pressure in the passage region also starts showing differences for the three tip designs. Two reasons could explain the observed differences: (1) The three tip designs have significantly different rotor blade width beyond 50% axial chord, which would change the aerofoil loading around this region, and (2) the over-tip leakage vortex and its interaction with the passage flow is very different beyond 50% axial chord for each tip design, as shown earlier in Fig. 18.
At 10.2% axial chord shown in Fig. 19(a), the measured casing static pressure, for the two squealer tips, rises from the suction side to the pressure side over the rotor and then gradually falls from the pressure side to the suction side over the passage. The flat tip blade, on the other hand, shows the peak pressure in the rotor region ahead of the pressure side. The pressure fluctuations in the pitch-wise direction are higher for the squealer tips in comparison to the flat tip.
At 31.8% axial chord shown in Fig. 19(b), both squealer tip blades show a sharp rise in pressure across the pressure-side rim. The novel tip shows the lowest casing static pressure, over the rotor, followed by baseline and flat tips. A lower casing static pressure in the rotor region implies high over-tip leakage flow acceleration. The flat tip shows a continuous rise in casing static pressure from suction to the pressure side over the rotor, while baseline and novel tips show a mixed trend with a fall in casing static pressure across the suction-side rim.
At 53.3% axial chord shown in Fig. 19(c), all three tip designs show significantly different pressure variation in the rotor region. For the baseline tip, the casing static pressure first rises across the suction-side squealer rim, then remains almost constant in the cavity, followed by a rise again across the pressure-side squealer rim. The novel tip shows a similar trend to the baseline except in the cavity region where the casing static pressure falls from the suction side to the pressure-side rim for the novel tip. The flat tip shows a gradual rise from the suction side to the pressure side in the rotor region.
At 74.9% axial chord shown in Fig. 19(d), the baseline tip measurements show a steep rise in casing static pressure, from the suction side to the pressure side, in the squealer cavity. While the flat and novel tips do not show much variation in this region and exhibit around 30% lower static pressure compared to the baseline tip.
At 96.4% axial chord shown in Fig. 19(e), all three tip designs feature low measured casing static pressure with no distinct trace of blade passing, unlike the other axial locations. CFD predictions consistently show higher casing static pressure compared to experiments at this location for all test conditions.
5.6.1 Comparison Between Uniform and RTDF Inlet.
The time-resolved casing static pressure measurements between uniform inlet and RTDF inlet are very similar at most axial locations, except for some local differences. For instance, at 10.2% axial chord (Fig. 19(a)), the measured casing static pressure is higher with RTDF inlet toward the suction surface of all three tip designs, in comparison to the uniform inlet. The observation is more evident through CFD predictions that overestimate the pressure increase with RTDF inlet. The change in casing static pressure could be associated with change in the relative velocity vectors and secondary flow, near the casing due to the presence of cold air toward the endwall of the RTDF profile. At 53.3% axial chord (Fig. 19(c)), measured casing static pressure with RTDF inlet is up to 8.7% lower than the uniform inlet in the tip region for the flat tip blade. The squealer tips do not show such large differences between RTDF and uniform inlet in this region.
5.6.2 Comparison Between Nominal and Large Tip Gap.
The comparison of experimental data across multiple axial locations shows that differences in unsteady casing static pressure field are higher with the change in the tip gap, compared to the change in inlet condition. A similar observation was made for steady casing static pressure data acquired using pneumatic lines. The casing static pressure is higher with the nominal tip gap compared to the large tip gap at most axial and pitch-wise locations, for the reasons discussed earlier in Sec. 4. At 10.2% axial chord (Fig. 19(a)), the nominal tip gap shows up to 3% higher casing static pressure compared to the large tip gap. Interestingly, the increase in casing static pressure with reduction in the tip gap, in both the rotor and passage regions, is higher for the novel tip compared to the baseline tip. As discussed earlier, the baseline tip already offers significantly better aerodynamic sealing from the over-tip leakage flow in comparison to the novel tip. Hence, the improvement in sealing effectiveness with reduction in the tip gap is anticipated to be higher for the novel tip. On the contrary, casing static pressure is found to be higher with the large tip gap, near the trailing edge, as shown in Fig. 19(e) for 96.4% axial chord.
6 Conclusions
An investigation on the aerodynamics of the rotor casing, of an unshrouded HP turbine, is presented in this article. The findings were based on time-averaged and time-resolved casing static pressure measurements acquired in the OTRF. The experimental findings were complemented with results from unsteady CFD predictions. The effect of three different tip designs, two tip gaps and two inlet temperature profiles, on the casing aerodynamics was studied. A summary of the key conclusions from this article is as follows:
Casing static pressure was found to drop by ∼1.7 bar (31.5% of P01) in the axial direction, from rotor inlet to exit, as the rotor extracted work. The largest decline was observed, between 35% and 65% axial chord, which corresponded to the region of highest flow acceleration in the rotating frame of reference.
Contour plots of time-averaged measurements illustrated the interaction between the rotor and NGVs. An increase in casing static pressure was observed in the gap, between NGV exit and rotor inlet, as the rotor passed the NGV trailing edge.
A reduction in tip gap was found to decrease the swallowing capacity of the rotor. Further, the reduced over-tip leakage flow momentum resulted in higher casing static pressure with the nominal tip gap, in comparison to the large tip gap.
Time-resolved measurements showed that the casing static pressure, in the passage region, falls gradually from the pressure side to the suction side. In the rotor region, the three tip designs showed significantly different aerodynamics. The baseline tip showed the highest casing static pressure (and hence, the lowest over-tip leakage flow acceleration) in the tip region, followed by the flat tip and then the novel tip. CFD predictions showed that the baseline tip featured the smallest over-tip leakage vortex among the three tip designs.
Time-averaged measurements found the casing static pressure to be ∼0.3% lower with RTDF inlet in comparison to the uniform inlet. Time-resolved casing static pressure showed similar trends, in the pitch-wise direction, at both tip gaps.
Comparisons across different tip designs and tip gaps suggested that a higher casing static pressure is indicative of good aerodynamic seal effectiveness from the over-tip leakage flow.
CFD predictions showed good qualitative agreement with experimental results. Possible reasons for discrepancies between experiments and CFD were explored; these included differences in NGV–rotor interaction and differences in corner radii on the rotor tips. CFD predictions captured the differences in casing static pressure caused by different tip designs and different tip gaps, better than the differences caused by the change in inlet condition from uniform to RTDF.
Acknowledgment
The authors would like to express gratitude to T. Godfrey for his expertise on instrumentation and S. Chana for his assistance in running the OTRF.
Funding Data
Rolls-Royce plc.
Conflict of Interest
There are no conflicts of interest.
Data Availability Statement
The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.