Search

The development of integrally cast turbine airfoils allows the production of narrow impingement channels in a double-wall configuration, where the coolant is practically injected within the wall of the airfoil providing increased heat transfer capabilities. This study examines the cooling performance of narrow impingement channels with varying jet diameters using a single exit design in an attempt to regulate the generated crossflow. The channel consist of a single row of five inline jets tested at two different channel heights and over a range of Reynolds numbers. Detailed heat transfer coefficient distributions are evaluated over the complete interior surfaces of the channel using the transient liquid crystal technique. Local jet discharge coefficients are determined by probe traversing measurements for each individual jet. A 10%-increasing and a 10%-decreasing jet diameter pattern is compared with a baseline geometry of uniform jet size distribution indicating a considerable effect of varying jet diameter on the heat transfer level and the development of the generated crossflow.

An experimental investigation was carried out to examine the influence of the axial spacing between rotor and stator. The test facility of the Swiss Federal Institute of Technology is made up of an inlet guide vane (IGV), a rotor and a stator. Temperature and aerodynamic measurements determined the characteristics of the compressor stage. Data were acquired over a range of flow coefficients for four different axial spacings. A refined measurement mesh was used for Pitot probe measurements in different sections of the compressor. As the compressor has a rather low aspect ratio of 1.21, the flow was governed by strong secondary flows. The interaction between the different secondary flows, especially the Influence of the IGV wake on the secondary flows in the stator, were detected by the detailed Pitot probe measurements. Shifting the stator axially changes the circumferential position of the IGV wake entering the stator. From this statement, the link between an axial variation of the distance between rotor and stator and the clocking between IGV and stator will be shown. Further Pitot measurement with different indexings between IGV and stator were performed. The Pitot measurements for the different axial spacings as for the different indexings brought to light strong interactions in the wall regions especially between the IGV wake and the clearances at the rotor tip and at the stator hub. The structure of the flow thus changes with the indexing and with the axial spacing.

This paper presents a new International Standard Configuration to be added to an already existing set of 10 configurations for unsteady flow through vibrating axial-flow turbomachine cascades. This 11th configuration represents a turbine blade geometry with transonic design flow conditions with a normal shock positioned at 75% real chord on the suction side. Out of a set of test cases covering all relevant flow regimes two cases were selected for publication: A subsonic, attached flow case and an off-design transonic case showing a separation bubble at 30% real chord on the suction side are published. The performed tests are shown to be repeatable and suitable for code validations of numerical models predicting flutter in viscous flows. The validity of the measured data of the two public cases was examined and comparisons with other tests were conducted. Sometimes a large difference in aerodynamic damping was observed on cases with similar flow conditions. This was investigated at three transonic cases with almost identical inlet flow conditions and only small variations in outlet Mach Number. It was found that the differences in the global damping are due to very local changes on the blade surface in the shock region, which obtain a large influence by the integration because of the discrete measuring points. Hence it is recommended not to look at the global damping for code validations but more precisely to the local values. These show a common tendency, which is reproducible with different numerical methods. This was demonstrated with a potential model, a linear Euler model, a nonlinear Euler model and a Navier-Stokes solver, all applied to predict flutter of each test case with a 2D/Q3D approach. The limitations of inviscid codes to predict flutter in viscous flow regimes is demonstrated, but also their cost advantage in attached flow calculations. The need of viscous code development and validation is pointed out. This should justify and encourage the publication of thoroughly measured test cases with viscous effects.

The present paper shows the comparison between steady-state and unsteady results from two versions of a flow solver and measurements in a linear cascade at transonic flow conditions for forced vibration of a single blade. Experiments have been conducted in a linear test facility with a cascade composed of five turbine blades. The center blade was forced to oscillate in a pure bending mode with a frequency of 160 Hz. Tailboards are fixed in this test facility at the trailing edges of the border blades to create a periodic flow at the outlet of the cascade. A 2D inviscid flow solver for unsteady flows through vibrating blade rows was adapted to simulate the flow conditions in the linear cascade. The initial adaptation of the flow solver was the limitation to only one vibrating blade. Comparisons with measurements showed discrepancies which were hypothesized to be due to the presence of the tailboards which can create reflections of pressure waves and shocks. Therefore, as a second modification of the program, the periodic boundaries downstream of the blades were replaced by solid walls. Based on these computations it can be concluded that for transonic flows the tailboards have a significant influence on both the steady-state and unsteady flow in the outlet region as well as on the blade in the shock region. In particular, the unsteady data measured in a linear cascade has to be analyzed very carefully to estimate the influence of the tailboards.

Calculated results for tip flow around two different blade configurations are presented and compared with experimental data. The first configuration (case number 1) is a flat-plate profile tested in a linear transonic tunnel — the profile is an idealized representation of the aft-section of some highly curved turbine blades. The second configuration (case number 2) originates from the outer profile on the last-stage-blade of a steam turbine, however it is also reminiscient of a section from a turbine blade with supersonic exit flow. This configuration was tested in an annular cascade at Mach numbers representative of engine operating conditions. The computed results were obtained using a parallel 3D unstructured Navier-Stokes code. The code runs on a work-station cluster, as well as being optimized for the 256 processor Cray T3D at EPFL: the code is capable of gigaflop performance using more than 3 million cells — adaptive mesh refinement thus allows enhanced resolution within the tip gap region. For each configuration we have calculated two Runs. In both cases, Run-1 is similar to the experimental conditions, so that direct comparison between measured and calculated results is possible. With case number 1/Run-2 we re-calculated the flow without imposing a prescribed inflow boundary-layer along the sidewall. Comparison between the two runs helped reveal how free-stream total pressure can establish itself within the tip gap region. For the second configuration — in the annular cascade — we were interested in observing the influence of relative movement between the blade tip and adjacent sidewall. Hence for case number 2/Run-2 we imposed a circumferential velocity on the adjacent sidewall. This modified the effective sidewall boundary-layer and had a noticeable influence on the development of the tip-leakage flow.

This paper presents a simple method for predicting tip leakage losses in transonic axial-flow turbines. The method is based upon experimental work conducted on a flat plate at 5° incidence and with isentropic exit Mach number of 1.26. The tip gap height was varied from zero up to 15% of chord. Measurements were made (using Laser-2-Focus) of velocity vectors around the tip gap region. These revealed a strong shear layer emerging from the gap onto the suction side of the plate. The relative angle between the leakage flow and the freestream was identified as a key parameter determining the subsequent mixing and overall loss generation. The proposed model applies two-dimensional potential flow analysis to estimate the flow angle as a function of tip gap height and the angle of incidence. Subsequently, comparisons were made with experimental results obtained in an annular cascade on the outer profile of the last-stage blade of a steam turbine. The predicted tip leakage losses compare favourably with the measured values.

This paper presents heat transfer measurements on a turbine airfoil in a linear cascade at various exit Reynolds and Mach numbers ranging from 3.2e5 to 1.6e6 and 0.2 to 0.8, respectively, which have been conducted with the transient liquid crystal technique. Two series were performed at turbulence intensities of 5.5% and 10%, the latter being created by a squared-bar mesh placed 10 meshsizes upstream of the turbine airfoils. While normally polished liquid crystals were used additional experiments were done at the high turbulence intensity with naturally rough liquid crystals. All measurements indicate a gradual increase in heat transfer and an upstream shift of the laminar-to-turbulent transition with increasing Reynolds number and turbulence intensity. The leading edge heat transfer agrees well with correlations if the turbulence length scale is taken into account. The measurements conducted with rough liquid crystals show an earlier transition on the suction side. Calculations with a two-dimensional boundary layer code agree well with the measurements.

Integrally cast turbine airfoils with wall-integrated cooling cavities are greatly applicable in modern turbines providing enhanced heat exchange capabilities compared to conventional cooling passages. In such arrangements, narrow impingement channels can be formed where the generated crossflow is an important design parameter for the achievement of the desired cooling efficiency. In this study, a regulation of the generated crossflow for a narrow impingement channel consisting of a single row of five inline jets is obtained by varying the width of the channel in the streamwise direction. A divergent impingement channel is therefore investigated and compared to a uniform channel of the same open area ratio. Flow field and wall heat transfer experiments are carried out at engine representative Reynolds numbers using particle image velocimetry (PIV) and liquid crystal thermography (LCT). The PIV measurements are taken at planes normal to the target wall along the centerline for each individual jet, providing quantitative flow visualization of jet and crossflow interactions. The heat transfer distributions on the target plate of the channels are evaluated with transient techniques and a multilayer of liquid crystals (LCs). Effects of channel divergence are investigated combining both the heat transfer and flow field measurements. The applicability of existing heat transfer correlations for uniform jet arrays to divergent geometries is also discussed.

The current capabilities of the foundry industry allow the production of integrally cast turbine airfoils. Impingement cooling effectiveness can be then further increased due to the manufacturing feasibility of narrow impingement cavities in a double-wall configuration. This study examines experimentally, using the transient liquid crystal technique, the cooling performance of narrow cavities consisting of a single row of five impingement holes. Heat transfer coefficient distributions are obtained for all channel interior surfaces over a range of engine realistic Reynolds numbers varying between 10,900 and 85,900. Effects of streamwise jet-to-jet spacing (X/D), channel width (Y/D), jet-to-target plate distance (Z/D), and jet offset position (Δy∕D) from the channel centerline are investigated composing a test matrix of 22 different geometries. Additionally, the target plate and sidewalls heat transfer rates are successfully correlated within the experimental uncertainties providing an empirical heat transfer model for narrow impingement channels. The results indicate similarities with multijet impingement configurations; however, the achievable heat transfer level is about 20% lower compared to periodic multijet impingement correlations found in open literature.