Category UNSTEADY AERODYNAMICS, AEROACOUSTICS AND AEROELASTICITY OF TURBOMACHINES

Experimental results

A lot of the unsteady 3D ft>w measurements using three-sensor hot-wire probes were presented in earlier papers [Krysinski et al. 1999 – 2002].

Based on earlier tests one could observe that there were no significant differ­ences after the first stator and the first rotor while indexing the airfoils. It was clearly visible how the first stator wakes pass through the first rotor and change their circumferential position downstream. The small fbw-blocking effects (according to the vane clocking) observed during earlier tests were neutralised by keeping the inlet/outlet pressure ratio at the same level with an especially prepared control system (Fig. 1, pos. 13). Significant changes were observed inside the ft>w passing the second stage, especially after the second vane. New test results showed again visible differences due to the clocking effect when compared to the situation when two stators were at the design position (x/T = 0.0), especially it is showed for highly loaded machine (Fig. 7).

Figure 7. Measured relative torque and efficiency versus clocking position and rotational speed

Periodic and random components of unsteady pressure above the first rotor for rotational speed n = 42 Hz are presented in Fig. 8. The presence of the po­sition of the preceding stator is clearly visible for the periodic component. The minimum value of the periodic component is related to the situation when the stator wake (secondary flow structures) flows inside the testing zone (nearby transducer position) and the maximum value is obtained when the inside vane channel ft>w is passing by. Due to the low level of the measured pressure and the low frequency resolution of the pressure transducer the tests above the sec­ond rotor were not performed. In this case the clocking effect phenomena are hardly to measure using standard instrumentation.

Fig. 9 shows a comparison of the relative random (er /E) and periodic (ep/E) components of thermoanemometric signals measured on the suction sides of the first stator and the second stator vanes with a help of glue-on hot-film sen­sors. The results are similar for both vanes. The rapid grow of the signal is

0 Figure 8 RMS of periodic and random components of the unsteady pressure above rotor

clearly visible due to the laminar-turbulent transition zone. The beginning of this phenomenon on the surface of the second stator vane starts earlier and is characterised by higher RMS values. The end of the zone is almost at the same place and its position is related to the inlet conditions. A detailed ft>w measurement data comparison is beyond the scope of this paper and will be discussed in another paper.

Fig. 10 shows the results of the glue-on hot-film signals analysis from two stators compared to the variation of the torque of the turbine for the same clock­ing positions. The position of the sensor was at the leading edge of the vane and in the middle of the ft>w channel height. Additionally acoustic measure­ments with the microphone connected pneumatically to the pressure tap nearby the hot-film sensor were performed. The results are shown in Fig. 11. Small variations of the measured values are visible but they are not clearly correlated to the overall performance changes. This due to the fact that the most impor­tant performance changes are present nearby the hub and the casing. In this case more detailed tests along the flow channel height are necessary.

The next figure (Fig. 12a) shows differences of the thermodynamic tur­bine efficiency versus height of the ft>w channel position at different rotational

speeds. The differences are visible for all speeds and the shape of the curves is similar showing the visible drop in the near the walls regions, especially nearby the hub. From earlier studies it is known that the turbulence level is very high in these regions [Krysinski et al. 2001]. Fig. 12 b and c show the efficiency changes due to the clocking effect at different fbw channel heights. The variations in the region near the hub are about 1 %. Along the height of the flow channel the variations of the turbine efficiency are phase shifted. This shift makes the overall efficiency variation to be small and not clearly visible. If the shift does not exist (that means across the fbw channel height the effi­ciency obtains maximum value at the same clocking position) for the nominal rotating speed we will obtain the efficiency benefit about 0.6 %.

Figure 13. a) Stator and rotor leading edge (LE) and trailing edge (TE) compared to the radial direction (RL) b) Results of the numerical calculation downstream the first stator and the first rotor

2. Conclusions

Two-stage low-pressure turbine at the Institute of Turbomachinery (Tech­nical University of Lodz, Poland), was prepared for experimental investiga­tions of the stator-to-stator clocking effect. Special attention was directed into machine work conditions and strictly constant inlet conditions, especially in – let/outlet pressure ratio. The changes due to the fbw blocking effect were re­
duced with the help of special inlet condition control system. Continuation of earlier experimental investigations of the stator-to-stator interaction has been performed in the axial gaps, on the vane profile and outer casing.

The results of precise measurements of the power and torque output differed slightly according to the stator-to-stator clocking position. Only for the lowest rotational speed the effect of the clocking position was very clearly visible on overall performance. That gave the conclusion that it was very hard to find out the infhence of the clocking position measuring and considering only the external characteristics, especially for the case of not highly loaded turbine.

The ft)w parameters downstream the first turbine stage showed similarities for the different clocking positions. The stator vane wake could be clearly observed also after the rotor. With the improved resolution of the measuring system even the thermal wake after the first stator was observed (Krysinski et al. 1999).

The exit fbw of the second stator was strongly inflienced by the clock­ing position of the first stator. The differences of the fbw parameters became smaller downstream the second stage due to the strong flow mixing phenomena inside the second rotor but the time-averaged values downstream the turbine showed significant differences relative to the first stage and when compared with different clocking positions. The results showed that the presented phe­nomena are very complicated. The first stator wakes do not pass directly inside the second stator flow channel or directly impinge the noses of the second sta­tor vanes as other authors presented it. The wakes start immediately to rotate inside the first stator-rotor axial gap and next downstream the turbine. As a result they interact strongly with the channel boundary layers not only at hub and casing regions but also on both surfaces (suction and pressure side) of the following stator vanes.

Parallel to presented results the numerical calculations of the ft>w were per­formed to understand unsteady flow behaviour in multistage machines and to determine the clocking effect more deeply. Fig.13 shows the geometrical po­sition of the leading and trailing edges of the turbine blading as well some nu­merical results. The numerical models showed the good accordance with the experimental results. They show also the wake variations, especially nearby the hub region where the wake is strongly shifted. The numerical codes need more and more sophisticated data to be improved. The data presented in this paper provide the community with an understanding of the effects that indexing airfoils can have on the overall turbine efficiency giving the vision of a future turbomachinary performance improvement. The new design with less rotor-to – stator blading number ratio is prepared to find out the influence of the clocking effect for other geometries of the machines working at similar conditions.

Nomenclature

b = chord length [ m ] m = mass few rate [ kg/s ] n = rotational (shaft) speed [ Hz ] p = pressure [ Pa ]

T = vane pitch [ m ], temperature [ K ] x = coordinate along circumferential direction [ m ] y, h = coordinate along radial direction [ m ] z = coordinate along axial direction [ m ]

a = circumferential (pitchwise) ft>w angle (absolute frame) [ deg ] в = circumferential (pitchwise) ft>w angle (relative frame) [ deg ]

Subscripts and Superscripts

= inlet conditions

= downstream the first stator

= downstream the first rotor

= downstream the second stator

= downstream the second rotor m = design (metal)

t = total (stagnation)

References

Arnone A., Marconcini M., Pacciani R. (2000): On the Use of Unsteady Methods in Predicting Stage Aerodynamic Performance. ISUAAAT’2000, Lyon, France, pp. 24- 46.

Arnone A., Marconcini M., Pacciani R., Schipani, Spano E. (2002): Numerical Investigation of Airfoil Clocking in a Three-Stage Low-Pressure Turbine. Trans. of the ASME, J. of Turbo­machinery, Jan. 2002, Vol. 124 pp. 61-68.

Dorney D. J., Sondak D. L., Cizmas P. G.A., Saren V. E., Savin N. M. (1999): Full-Annulus Simu­lations of Airfoil Clocking in a 1-1/2 Stage Axial Compressor. ASME 99-GT-023.

Eulitz F. (2000): Modeling and Simulation of Transition Phenomena in Unsteady Turbomachin­ery Flow. ISUAAAT’2000, Lyon, France, pp. 332-337.

Haldeman C. W., Krumanaker M. L., Dunn M. G. (2003): Infhence of Clocking and Vane/Blading Spacing on the Unsteady Surface Pressure Loading for a Modern Stage and One-Half Tran­sonic Turbine. ASME GT2003-38724.

He L., Chen T., Wells R. G., Li Y. S., Ning W. (2002): Analysis of Rotor-Rotor and Stator-Stator Interferences in Multi-Stage Turbomachines. Trans. of the ASME, J. of Turbomachinery, Oct. 2002, Vol. 124, pp. 564-571.

Howell R. J., Ramesh O. N., Hodson H. P., Harvey N. W., Schulte V. (2001): High Lift and Aft – Loaded Profiles for Low-Pressure Turbines. Trans. of the ASME, J. of Turbomachinery, Apr. 2001, Vol 123, pp. 181-188.

Huber F. W., Johnson P. D., Sharma O. P., Staubach J. B., Gaddis S. W. (1996): Performance Im­provement Through Indexing of Turbine Airfoils: Part 1 – Experimental Investigation. ASME J. of Turbomachinery, Vol. 118, pp. 630-635.

Hummel F. (2002): Wake-Wake Interaction and Its Potential for Clocking in a Transonic High – Pressure Turbine. ASME J. of Turbomachinery, Vol. 124, pp. 69-635.

Jouini D. B.M., Little D., Bancalari E., Dunn M., Haldeman C., Johnson P. D. (2003): Exper­imental Investigation of Airfoil Clocking Impacts on Aerodynamic Performance in a Two Stage Turbine Test Rig. ASME GT2003-38872.

Krysinski J., Smolny A., Blaszczak J. R., Gallus H. E., Walraevens R., Mertens B. (1995): Axial Two-Stage Turbine Test Rig For Unsteady Flow Measurements. SYMKOM’95, Int. Symp. CMP Pc 108, Lodz, p.99.

Krysinski J., Gallus H. E., Smolny A., Blaszczak J. R. (1999): Stator Wake Clocking Effects on 3D Unsteady Flows In a Two-Stage Turbine. 3rd EuroConf. On Turbomachinery, IMechE, C557/017/99. Vol. A, pp. 323-332, London, UK.

Krysinski J., Smolny A., Blaszczak J. R., Gallus H. E. (2000): 3D Unsteady Flow Experimental Investigations in a Two-Stage Low-Pressure Turbine. ISUAAAT’2000, Lyon, France, pp. 515-523.

Krysinski J., Blaszczak J. R., Smolny A. (2001): Stator Clocking Effect on Efficiency of a Two – Stage Low-Pressure Model Turbine. 4th European Conference on Turbomachinery: Fluid Dynamics and Thermodynamics, ATI-CST090/01, pp. 1045-1051, Firenze, Italy.

Krysinski J., Blaszczak J. R., Smolny A. (2002): Efficiency Improvement Through Indexing Of Stators Of A Two-Stage Turbine. HEFAT2002, KJ2, Kruger Park, South Africa.

Reinmoeller U., Stephan B., Schmidt S., Niehuis R. (2002): Clocking Effects ina a 1.5 Stage Axial Turbine – Steady and Unsteady Experimental Investigations Supported by Numerical Simulations. Trans. of the ASME, J. of Turbomachinery, Jan. 2002, Vol. 124, pp. 52-60.

Saren V. E., Savin N. M., Krupa V. G. (2000): Experimental and Computational Research of a Flow Structure in a Stator-Rotor-Stator System of an Axial Compressor. ISUAAAT’2000, Lyon, France, pp. 494-502.

Saren V. E., Savin N. M. (2000): Hydrodynamic Interaction of a Stator-Rotor-Stator System of an Axial Turbomachine. Fluid Mechanics 3/2000, UDK 533.6.011.34, Russian Academy of Sciences, Moscow, pp. 145-158 (in Russian).

Smolny A., Blaszczak J. R. (1996): Boundary Layer and Loss Studies on Highly Loaded Turbine Cascade. CP-571/4, AGARD.

Smolny A., Blaszczak J. R. (1997a): Experimental Investigations Of Unsteady Flow Fields In A Two-Stage Turbine. 2nd EuroConf. on Turbomachinery, Antwerp, Belgium.

Swirydczuk J., Gardzilewicz A. (2002): Analysis of the Stator-Rotor Interaction in the TM-3.00 Turbine. Institute of the Fluid Flow Machines of the Polish Academy of Sciences. Internal Report 2764/02 (in Polish), Gdansk.

[1] Corresponding Author: atassi@nd. edu

247

K. C. Hall et al. (eds.),

Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines, 247-259. © 2006 Springer. Printed in the Netherlands.

[2]Work is executed at financial support of International scientific and technical centre (ISTC), the grant number 672.2.

261

K. C. Hall et al. (eds.),

Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines, 261-273. © 2006 Springer. Printed in the Netherlands.

[3](t) = (V1/c) Rn zn (t) + En q n (t).

The generalized aerodynamic forces are then written in the time domain as

(

c c

^nOfin 4" у -^nlfin 4" ^^2 J^n24n

(21)

Combining the reduced coupled system (5) with Eqs. (16) and (17), we obtain a linear system of second-order differential equations of dimension mn + np :

Instrumentation

The external characteristic experimental data (rotational speed, torque, power, e. g. Fig. 3) were obtained for 28 different clocking positions of the two stators (relative range about 2.5 based on the vane pitch) with high-accuracy eddy-

current brake controlling system. Additional pressure and temperature kiel – head probes (Fig. 2) fixed at the inlet and outlet section of the turbine were used. In this case two stator fixed together were moved every 1/12 of the pitch performing 1000 measurements in every position, next the clocking position was changed by 1/6 of the pitch, the stators again were fixed and the operation was repeated until all indexing range was passed.

The basic few parameters (e. g. pressures, temperatures, few vectors, tur­bulence levels) after the annular cascades in each main measuring plane (0, 1, 2, 3, 4 – Fig.2) were surveyed before the measuring session using classi­cal methods and the tests confirmed the results obtained from earlier sessions [Smolny, Blaszczak 1997; Krysinski et al. 1999-2002].

During tests presented in this paper supplementary measurements were per­formed. To find out if there is any influence of the indexing effect on leakage flow above the rotor blades, the pneumatic signals with the help of unsteady pressure transducers were received (Fig. 6b). The data from the pressure trans­ducer were simultaneously recorded by the digital multimeter (DC part of the signal) and the transient recorder (AC part of the signal) to obtain the maxi­mum resolution. The data acquisition was triggered by a photocell located at the hub of the rotor.

Unsteady pressure was sampled in a one-time window with a digital reso­lution of 256 points at a sampling frequency of 240 kHz. After one rotor rev-

Figure 5. Turbine stator vanes with glue-on probes for boundary layer measurements

olution, the next time-window was recorded, until 256 of these time-windows were stored.

The triggering and data acquisition systems for the thermoanemometric mea­surements were the same. For vane boundary layer phenomena identification glue-on hot-film probes were used. The results presented here were performed on the suction side of the vanes at the midspan of the ft>w channel. Addi­tionally, to find the correlation between hot-film signals and inside noise level another tests were performed including acoustic measurements with the help of 1/4” microphones perpendicularly connected to the vane surface tap tubing. The method was similar to the one used by [Sabah, Roger 2001]. The data from glue-on hot-film sensor and the microphone were sampled simultaneously at 50 kHz and next, recorded by a digital data acquisition system.

More details about data treatment methods can also be found in [Smolny, Blaszczak 1996, 1997].

Figure 6. Rotor geometry (a) and the position (b) of the unsteady pressure transducer (c)

Facility

The series of tests were conducted on the two-stage low-pressure model tur­bine with the eddy-current brake. The layout of the turbine test rig is presented in Fig. 1. A two-fan set with a specially equipped control system provided a continuous and strictly constant airflow to the test rig. The inlet air parameters during the tests presented here were as follows: total pressure pt0 = 15.668 ± 0.003 kPa, total inlet temperature Tt0 = 318 ± 2 K, mass fbw rate m = 3.565 ± 0.005 kg/s (n = 49.2 Hz) to m = 3.730 ± 0.005 kg/s (n = 30 Hz) – variations were due to the clocking effect. During the tests the rotational speed n was in the range 25.00 to 58.33 Hz (1500 rpm to 3500 rpm). Variation of the speed n at the working point was less then 0.02 Hz (1 rpm) during every measurement session. Fig. 2 shows the cross section of the test rig with the main measuring planes and Fig. 3 shows the performance curves of the machine.

The turbine geometry with some indexing positions of the first stator vane (pos. 5 ^ x/T = 0.0 ^ identical circumferential position of both stators) is presented in Fig. 4. For tests presented herein both identical stators have 16 constant-section cylindrical vanes with the trailing edge inclined to the radial direction at an angle of 22 deg. Both identical rotors have 96 twisted blades. This atypical machine with a very clear stator wake presence is an interesting case from the numerical point of view. A more detailed description of the test facility geometry and the unsteady flow measurement can be found in our earlier studies [Krysinski et al. 1995 – 2002, Smolny, Blaszczak 1997].

TWO-STAGE TURBINE EXPERIMENTAL INVESTIGATIONS OF UNSTEADY STATOR-TO-STATOR INTERACTION

Indexing Effect

Jan Krysinski

Institute of Turbomachinery, Technical University of Lodz krysinski@p. lodz. pl

Robert Blaszczak Jaroslaw

Institute of Turbomachinery, Technical University of Lodz blaszczk@p. lodz. pl

Antoni Smolny

Institute of Turbomachinery, Technical University of Lodz asmo1948@p. lodz. pl

Abstract The results of the stator-to-stator clocking effect in a two-stage low-pressure tur­bine are presented. The main goal is focused on a detailed investigation of the shape and its position of the stator wake after the first stage and the correlation between different parameters including acoustic signal levels. The numerical calculations confirmed the image of the flaw field obtained during the measure­ments.

Keywords: axial turbine, clocking effect, measurements, flow field, acoustics

1. Introduction

In order to improve performance and prediction methods accuracy for mul­tistage axial turbomachines, understanding of the unsteady fhw is essential. A number of experimental and numerical studies have been carried out in recent years to investigate these fhw phenomena [e. g. Arnone et al. 2000, 2002; Dorney et al. 1999, Eulitz 2000; Haldeman et al., 2003; He et al. 2002, Howel et al. 2001, Huber 1996, Hummel 2002; Jouini et al. 2003; Reinmoeller et al.

615

K. C. Hall et al. (eds.),

Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines, 615-626. © 2006 Springer. Printed in the Netherlands.

2002, Saren et al. 2002]. The experimental work presented herein is the con­tinuation of our earlier studies, starting from mid-nineties [e. g. Krysinski et al. 1995 – 2002; Smolny, Blaszczak 1997] and it is still in progress at the Insti­tute of Turbomachinery (Technical University of Lodz, Poland). The objective of the present test program is to experimentally investigate vane-indexing ef­fects (so called also the clocking effects) on the performance of the two-stage low-pressure turbine test rig. Special clocking mechanisms were designed to allow the first stage vanes to be moved circumferentially relative to the second stage vanes independently of the casing. This allows changing the clocking positions of the first and the second stator vanes during the tests without stop­ping the turbine and dismantling it. Herein, some unsteady fbw measurement results, wall pressure above the rotor blades and the external characteristics for different circumferential positions of the stator vanes are described.

Position of Unsteady Boundary Layer

Compressor at maximum effi ciency (m=6.4 kg/s). The next figures de­

scribe the development of the unsteady boundary layer along the rotor and stator blades at maximum compressor efficiency (nmax) using t-s-diagrams of the bridge output voltage e and the skewness ^3 (Figures 10, 11) as well as line plots of e, rms and ^(Figures 12 — 15). The boundary layer on compres­sor blades can be described by two typical paths, which occur periodically: the wake-induced path with the regions A, B, and C, and the path between two wakes that consists of the regions A, D, E and F. The laminar region (A) occurs at the leading edge of the blade and becomes transitional downstream (E, F).

The beginning of the transition is dominated by a calmed region (D) along the non-wake induced path. This region causes a delay of the transition process in the transitional region E and is generated by turbulent spots in the transi­tional region (B). Further downstream a turbulent region can be identified for x/c >63%. Line plots through the two paths are shown in the Figures 12 and 13 for the rotor, which compare, among other things, the beginning of by-pass transition. At the wake induced region the incoming wakes cause instabili­ties, which generate turbulent spots and, therefore, results in an earlier induced transition (Fig. 13: x/c=38%).

Line plots for a stator vane along the two typical paths are shown in the Figures 14 and 15. The distributions were measured using the surface hot wire technique. The transition points (^3=0) show different positions along the two paths due to the incoming wakes, induced by the IGV stage.

Compressor at reduced mass flow (m=5.85 kg/s). The consequences of a mass ft>w reduction on rotor and stator are shown in Figures 16 — 23. The rotor blade Campbell-diagrams of quasi wall-shear stress (e) and skewness in­dicate that transition above a separation bubble that begins at x/c=25% chord has occured, see Figures 16,17. A turbulent reattachment takes place between x/c=70-75% and is clearly modulated by the IGV-wakes. In the line plots (Fig­ures 18,19) the beginning of the separation area is indicated by a local rms – maximum. The global maximum at 60% chord, on the other hand, describes the region of laminar-turbulent transition. A typical zero-skewness (transition point) could not be detected due to the sensor positions that are directly placed on the wall. The mass flow reduction causes free-stream conditions which are similar to an increase of the angle of incidence. This leads to a destabiliza­tion of the boundary layer and results in a separation due to a local pressure increase.

Turbulent boundary layer reattachment can be identified by means of look­ing at cross-correlations of the last few array sensors, see Figure 20. The pre­sented correlations use the last sensor of the array for reference. The results show no correlation with the sensor at x/c= 67% but a clear amplitude-increase and a positive time shift at 89.1% chord, which indicates a reattached boundary layer. The correlation at x/c=74.1% shows an increased amplitude tendency but no significant amplitude peak due to the sensor position, which is near the reattachment line.

While the rotor boundary layer is separated above a blade region of approx­imately 40-50% of the chord, investigations on the stator vane (Figures 21 and 22) reveal that the boundary layer remains attached and turns turbulent between x/c=20-40%, modulated by the incoming wakes.

Compressor at Stall. A Rotating stall is characterized by one or more cells of reduced, and often reversed, fbw travelling with an angular velocity which is much smaller than that of the rotor stages. While a few investigations of stator vanes have been carried out in the past, simultaneous surface sensor measurements on rotor blades were performed for the first time. Figure 23 shows ensemble averaged power spectra at 6 different streamwise positions at the rotor blade with an excited blade frequency at f=1665Hz. It can be seen that the frequency does not dominate in each spectrum due to highly unsteady flow conditions.

A sequence of instantaneous boundary layer distributions at the stator is shown in Figure 24. Due to the stall cells moving with a reduced angular velocity compared to the rotating speed, it appears that the boundary layer phenomena results in a backwards movement in the t-s-diagrams. A common transition mode such as by-pass transition or the transition above a laminar separation bubble will not arise. The boundary layer development along the chord is controlled by turbulence effects generated near the leading edge.

2. Summary

The results presented in this paper were obtained in a highly-loaded, 1,5- stage, axial low-speed compressor. Typical wake and non-wake induced paths, as well as regions of laminar, transitional, and turbulent boundary layer flows were periodically detected on rotor and stator blades using surface sensor ar­ray techniques. In particular, the results obtained by the novel surface hot wire imply that the technique has the potential for shear stress measurements in unsteady flows. Investigations of the boundary layer development on a rotor blade were obtained by means of a telemetry system and miniature constant – temperature anemometry fixed to the rotor. Using such a data acquisition sys­tem, it can be shown that the boundary layer at the design point and at maxi­mum efficiency is based on a by-pass process. A mass flow reduction causes a laminar boundary layer separation in connection with a transition above a bubble, as well as a turbulent reattachment near the trailing edge.

Acknowledgments

The reported work was performed within the AdComB research project that is funded by the 5th European Framework. The project has been supported by Rolls-Royce Deutschland Ltd. and Co. KG, and carried out at the Pfleiderer – Institute of the Technical University of Braunschweig. The 16-channel minia­ture constant-temperature anemometry ring used for the measurements at the rotor blade was developed by the M. Baumann engineer-bureau.

Rotor-Stator Interaction in a Highly-Loaded Single-Stage Axial Compressor 609

References

Adamczyk, J. J.,"Wake Mixing in Axial Compressors" ASME 96-GT-29, 1996 Swoboda M., Teusch R., Guemmer V., Fottner L., Wenger U., "Experimental Investigation of Boundary Layer Transition in Compressor Cascades at Unsteady Flow Conditions", STAB Proceedings, 1998, pp. 474-482

Rohkamm, H., Wulff, D., Kosyna, G., Saathoff, H., Stark, U., Gummer, V., Swoboda, M., Goller, M., "The Impact of Rotor Tip Sweep on the Three – Dimensional Flow in a Highly – Loaded Single-Stage Low-Speed Axial Compressor: Part-II Test Facility and Experimental Results", 5th European Conference on Turbomachinery – Fluid Dynamics and Thermody­namics, Prague, Czech Republic, 18.-21. March 2003.

Lakshminarayana B., "A Method of Measuring Three-Dimensional Rotating Wakes Behind Tur­bomachinery Rotors", Journal of Fluids Engineering, June 1974, Vol. 96, pp. 87-91 Sturzebecher, D., Anders, S., Nitsche, W., "The Surface Hot-Wire as a Means of Measuring Mean and Fluctuating Wall Shear Stress", Experiments in Fluids, Vol. 31, 2001, pp. 294-301

Figure 1. Measurement set up

Figure 4. Instrumented stator vanes Figure 5. Rotor with anemometry-

circuits and telemetry transmitting – component

Figure 17. Rotor: Campbell-diagram of the skewness (m = 5.85kg/s)

Figure 23. Rotor at stall: power spectra at different streamwise positions

Figure 24. Stator at stall: t-s-diagram sequence

Measurement Techniques

The results shown in this paper include a comparison of the thermal-resistive techniques used, and shows the inflience of mass ft>w and rotating speed on the boundary layer transition position on the compressor blades. These tech­niques also give detailed information of the unsteady boundary layer develop­ment on rotor and stator blades at m=6.4 and 5.85 kg/s, as well as at near stall conditions.

Surface Hot Wire Application

Figures 6 and 7 compare the skewness distribution on stator vanes at the compressor design point (m=6.82 kg/s) using hot film and hot wire techniques. The figures are plotted as Campbell-diagrams (t-s-diagram) over a time span of t=2 ms, corresponding to four wake periods. The results show a good concor­dance of both transition lines, which indicates the point of ^3=0 at streamwise positions between ж/с=30-40%. The transition lines are clearly modulated by the periodically incoming wakes generated by the up-stream rotor. The results from individual measurements at two stator vanes show slight changes in posi­tion, as well as expansion of the lines. These results imply a good performance of the novel hot wire technique used for unsteady measurements on turbo ma­chine components.

Boundary Layer Transition shifting

A general overview of the mass ft>w and speed of rotation infhence on the axial-compressor rows are given in the Figures 8 and 9. The results show the streamwise transition development at 3 different rotating speed frequencies over a range of mass fbws (m). The results indicate clear up-stream shifting of the transition due to a throttling of the machine and a corresponding mass flow decrease. Additionally, Figure 8 shows a good correlation of the data of both thermal-resistive sensors used for unsteady measurements on the stator vanes.

Data Acquisition. For measurements of periodically unsteady conditions in the cascade, a phase-locked ensemble average technique was used. This technique was proposed by Lakshminarayana et al. [4]. Data acquisition was carried out using a shaft encoder coupled with the rotor shaft to get a trigger-impulse. In the case of the present measurements sensor-signals were recorded 200 times using sample rates up to n=1024 and a scanning frequency of f=20 kHz. The ensemble-averaged sensor signals were analyzed using the AC-output voltages representing quasi wall-shear stress fhctuations. Statisti­cal values of root-mean-square (rms) and skewness (^3), as well as analysis functions such as power spectra and cross-correlation, were used to analyze the data.

Surface Hot Wire. As an alternative to a surface hot film, a new flish mounted hot wire with a better dynamic response was used for measurements on a stator vane for the experimental project part. The surface hot wire (SHW) is a thermal-resistive sensor, which is directly welded over a tiny cavity, flush – mounted to the surface [5]. The device offers a promising technique for skin friction measurements on compressor blades. The sensor design considerably reduces the heat conduction into the substrate and, therefore, yields an im­proved signal-to-noise ratio and a higher frequency response compared to a common surface hot film. Figure 3 shows a sketch of the sensor. A platinum – coated, 5 ^m diameter tungsten wire is used as sensor element while a 30 ^m copper-layer serves as substrate. The bonded sensor element has a resistance of approximately R=5.5 ^ and is positioned several wire diameters above the wall. The array is very flexible and allows for easy attachment onto curved surfaces.

Blade Instrumentation. The measurements on rotor and stator were con­ducted at blade mid-span using Senfhx hot film arrays. 16 and 25 single sen­sors were applied along the streamwise direction of a rotor and stator blade, respectively. In addition, a stator vane was equipped with a 16 single sensor surface hot wire array during the first experiment (see Figure 4). All sensors were operated in the constant-temperature mode with typical overheat ratios. The sensors on the rotor blade were adjusted using miniature anemometry – circuits. 16 electric circuits were designed on a ring-shaped printed circuit board (PCB) and implemented on the rotor disk (Figure 5). The integrated telemetry-system was used for a cable-less data transfer as well as a power supply for the 16-channel miniature constant-temperature anemometry ring.

Low-Speed Axial Compressor

All measurements were performed on the highly-loaded, single-stage, low – speed axial compressor shown in Figure 1. The test stage is designed with custom tailored CDA profiles and uses an inlet guide vane (IGV) to ensure that the inlet swirl to the following stage is truly representative of middle stage of a high-pressure compressor [3]. The axial compressor has a maximum capac­ity of m=9 kg/s at a speed of n=2800 rpm. The rotor consists of 43 blades, while the stator and the inlet guide vane use 45 vanes. The rotor as well as the stator blade chord is 1=75 mm and the aspect ratio is h/1=1.0. The axial dis­tance between the blade rows is about 50% chord at mid-span. The test section of the compressor has a casing diameter of DC=600 mm and a hub diameter of DH =450 mm with parallel annulus lines at constant inner and outer radii. The maximum motor power is Pmax=58 kW and allows a free-stream velocity of U=60 m/s. The measured compressor performance characteristics are pre­sented in Figure 2. The diagram shows three different trend lines of the static pressure ratio and the efficiency n due to a mass fbw variation. In addition, the operating points of interest are marked in the diagram.

ROTOR-STATOR INTERACTION IN A HIGHLY-LOADED, SINGLE-STAGE, LOW-SPEED AXIAL COMPRESSOR: UNSTEADY MEASUREMENTS IN THE ROTOR RELATIVE FRAME

Unsteady Measurements in the Rotor Relative Frame

O. Burkhardt, W. Nitsche,

Technical University of Berlin
Institute of Aeronautics and Astronautics
Marchstr. 12, Sekr. F2, 10587 Berlin, Germany

M. Goller, M. Swoboda, V. Guemmer,

Rolls-Royce Deutschland Ltd. and Co. KG
Eschenweg 11, 15827 Dahlewitz, Germany

H. Rohkamm, and G. Kosyna

Technical University of Braunschweig, Pfkiderer-Institute
Langer Kamp 6, 38106 Braunschweig, Germany

Abstract Unsteady measurements were performed on rotor and stator blades of a highly – loaded 1.5-stage, low-speed axial compressor cascade using dynamic skin-friction measurement techniques. The project focusses on advanced 3D compressor blade design and has the potential to improve tip-clearance fbw control at high loading by introducing different advanced unsteady measurement techniques. As an alternative to typical hot films, a novel surface hot wire with a better dy­namic response was used. This device is capable of measuring the effects of unsteady aerodynamics in turbo machinery components. The results show the inflience at different operating points (mass flaw) and rotor-stator interactions on the boundary layer transition development using statical values as well as analysis functions such as power spectrum and correlation. In addition, prelim­inary results of the unsteady boundary layer development at rotating stall are being presented.

Keywords: Rotor-stator interaction; Unsteady aerodynamics; Boundary layer transition; Mea­

surements in rotating frame.

603

K. C. Hall et al. (eds),

Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines, 603-614. © 2006 Springer. Printed in the Netherlands.

0. Introduction

The boundary-layer transition process on turbo machine blades is mainly in – fbenced by periodically incoming wakes, generated by an upstream blade row. Investigation by Adamczyk [1] has shown that the total pressure loss of a com­pressor blade row can be reduced if the incoming wakes mix out after rather than before the following blade row. Preliminary measurements, e. g. [2], were carried out using simple and suitable set ups to investigate interactions be­tween rotating and stationary blade rows. Traversable bars with diameters of 3 mm placed in front of a linear stator cascade were used to simulate and in­vestigate wake-induced transition. In the last few years measurements were directly conducted on real turbo machines with single or multi-stage blade rows. Detailed boundary layer measurements on stator vanes were done us­ing surface hot film sensors. However, until now simultaneous data acquisition using a high number of sensors has never been carried out, especially on rotor blades due to the complexity of the setup. Such measurements were performed in this project using a telemetry-system and a 16-channel miniature constant – temperature anemometry ring.

1. Setup

On influence of stators clocking on losses of total pressure in a compressor

The results of the presented above analysis of a fbw in system of rows IGV – R-S allow to make the conclusion, concerning observably in experiments [3-8] dependence of total pressure losses in a compressor stage from stators clocking

V.

As follows from the experimental data received here, the parameter v does not influence on averaged on time aerodynamic characteristics of the R and S. At the same time the unsteady (periodic) part of velocity circulation on the R blades rather considerably may depend on a mutual circumferential position of stators. In this connection in works [5-8] the assumption was stated, that inflience v on losses of total pressure in a stage is caused by dissipation of the free vortexes, descending with R blades at change on them of velocity circulation. The results received in the given work directly allow to compare dependences from v intensity of free vortexes and losses of total pressure.

According to Thomson theorem running intensity of free vortexes 7 behind an airfoil at the moment of time t in a point of a wake with arc coordinate s, counted from a leading edge of an airfoil, is equal

where T is the period of a ft>w pulsation on the R blades.

The value є^) may be calculated according to experiment, which illustra­tion is given on Fig.7. Function r(t) thus is determined as a result of the de­cision of the Cauchy-Lagrange equation concerning of velocity discontinuity on an airfoil pressure and suction sides at value of pressure difference known from experiment [16].

In turn the value of total pressure losses behind R is determined according to measurements by probes of the stagnation pressure, located behind the R and S, by calculation of value

APt = ((Pt2)t)y – ((Pt3)t)y

As shows experiment, as against time-averaged aerodynamic loadings on the R and S, values AP appear essentially dependent from v in particular for the assembly 1.

For the specified assembly on Fig. 14 dependences are presented

The received data show that in spite of the fact that r(t) is determined for peripheral section of the R blades, and value APt was measured on mean ra­dius, character of dependences fi = fi(v) and A = A(v) are close. Areas v, appropriate to the lowered (increased) values of free vortexes intensity coincide with areas of the lowered (increased) values of total pressure losses behind the R.

It is necessary to notice that this result specifies also a source of the addi­tional losses, caused a rotor-stator interaction in axial turbomachines.

2. Conclusions

The received results allow to formulate the following conclusions.

The presented experimental installation and the technique of data pro­cessing developed by authors allow to carry out complex researches of unsteady fbws, in particular, effects of a rotor-stator interaction in the subsonic axial compressors.

Researches of fbw properties show the following: – the rotor-stator in­teraction in a subsonic axial turbomachine at mean level of axial gaps (~ 15-25% of rotor blade pitch) does not infhence on time-averaged aerodynamic loadings on blades and structure of vortical wakes behind them;

– total pressure losses in a flow, caused a rotor-stator interaction, are bound up with additional losses owing to dissipation of free (periodic) vortexes in wakes behind blades.

Stator clocking effects in system of rows stator-rotor-stator of the axial compressor are bound up with influence of a mutual circumferential po­sition of stators on unsteady aerodynamic characteristics of a rotor. The effects are most significant at equal number of stators vanes and essen­tially depend on axial gaps between rows. Stators clocking may serve as an effective practical control facility a level of the rotor-stator interaction in axial turbomachines.

References

Saren, V. E. (1994) Some Ways of Reducing Unsteady Blade Loads Due to Blade Row Hydrodynamic Interaction in Axial Flow Turbomachines, Second International Confer­ence EAHE, Pilsen, Czech Republic, pp.160 – 165.

Saren, V. E. (1995) Relative Position of Two Rows of an Axial Turbomachine and Ef­fects on the Aerodynamics in a Row Placed Between Them, Unsteady Aerodynamics and Aeroelasticity of Turbomachines, Elsevier, pp.421 – 425.

Huber F. W., Johnson P. D., Sharma O. P. et ol. (1995) Performance Improvement Through Indexing of Turbine Airfols. Part 1. Experimental Investigation, ASME Paper No. GT-27.

Griffin L. W. , Huber F. W. and Sharma O. P. (1995) Performance Improvement Through Indexing of Turbine Airfols. Part 2. Numerical Simulation, ASME Paper No. GT-28.

5 Dorney D. J. and Sharma O. P. (1996) A Study of Turbine Performance Increases Through Airfoil Clocking, AIAA Paper No.2816.

6 Hohn W. (2001) Numerical and Experimental Investigation of Unsteady Flow Interac­tion in a Low Pressure Multistage Turbine, XV Intern. Sump. On Airbreathing Engines, Sept. 2-7, Bangalore, India.

7 Saren V. E., Savin N. M., Dorney D. J., Zacharias R. M. (1997) Experimental and Numer­ical Investigatin of Unsteady Rotor-Stator Interaction on Axial Compressor Stage (with IGV) Performance, Unsteady Aerodynamics and Aeroelasticity of Turbomachines: 8 th Intern. Symp., Stockholm, Dordrecht et al.: Kluwer, P. 407-416.

8 Saren V. E., Savin N. M., Dorney D. J., Sondak D. L. (1998) Experimental and Numerical Investigatin of Airfoil Clocking and Inter-Blade-Row Gap Effects on Axial Compressor Performance, International Journal of Turbo and Jet Engines, 15, P. 235-252.

9 Savin N. M. and Saren V. E (2000) Hydrodynamic Interaction of the Blade Rows in the Stator-Rotor-Stator System of an Axial Turbomachine, Fluid Dynamics. Vol.35, _3, pp. 145-158.

10 Saren V. E. (1971) On Hydrodinamical Interaction of Cascades of Profiles in Potential Flow, Izv. gS of USSR, > GG, _4, pp.75-84.

11 Judin V. A. 1981 Calculation of Hydrodinamical Interaction of Cascades of Profiles with account of wakes, Aeroelasticity of Turbomachines Blades. Tras. of the СІАМ, _953, pp.52-66.

12 Saren V. E., Savin N. M., Krupa V. G. (2000) Experimental and Computional Research of a Flow Structure in a Stator-Rotor-Stator System of an Axial Compressor, The 9th Internaional Symposium on Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines (ISUAAAT), Lyon, France, September 4-8, pp.494 – 502.

13 Ivanov M.. Ja, Krupa V. G., Nigmatullin R. Z. (1989) Implicit High Accuracy S. K. Go­dunov Scheme for Integration of Navier-Stokes Equations, Journal of Calculating Math­ematic and Mathematical Physic, vol.29, No.6.

14 Saren V. E., Savin N. M., Krupa V. G., Petrovitchev A. M. (2001) Inflience of a Rotor – Stator Interaction on the Steady and Unsteady Characteristics of the Axial Compressor, Paper from the XV ISABE, Bangalore, India.

15 Saren V. E and Smirnov S. A. (2003) Unsteady Vortical Wakes behind Mutually Mov­ing Rows of Axial Turbomachine, Thermophysics and Aerodynamics, Vol.10, No.2.

16 Saren V. E, Smirnov S. A. (2003) Structure of Unsteady Vortical Wakes behind Blades of Mutually Moving Blade Rows of an Axial Turbomachine, The 10th Internaional Sym­posium on Unsteady Aerodynamics, Aeroacoustics and Aeroelasticity of Turbomachines (ISUAAAT), Durham, North Carolina, USA, Sept. 7-11.

Figure 1. Compressor flow path and measurements

Fig.2 Radial distributions of stagnation pressure Pt and temperature Tt along flow path.

0 05 У* 1

0 0 5 Yigv 1 0 05 Yt

Fig. 4 Pitch-wise distributions of total pressure before R (left) and static pressure after R (right).

0 02 04 06 08 1 1 о 02 04 06 08 1 1

аамтЫу 3

Fig. 7. Time variations of velocity circulation T.

Fia.8. Structure of vortical wakes behind R blade

0 0.2 0.4 0.6 0.8 1

Fig.9.Pitch-wise distributions of flow velocity behind R. Assembly 1.

Assembly 1

Assembly 2

Assembly 3

a) Measure of circumferential distortion of time-average flow

Assembly 1

Assembly 2

Assembly 3

b) Pitch-average measure of flow fluctuation

Assembly 1 Assembly 2 — Assembly 3

c) Measure of circumferential distortion of flow fluctuation

Fiq 10. Influence of mutual clocrinq positions of stators on relative flow velocity behind rotor.

Design operating mode: n=2000 rpm. Gcorr=30.8kg’s

0. 2 0.4 0.6

Pressure surface

0. 2 0.4 0.6

Suction surface

Fig. 12.Static pressure pulsation on the S vanes, assembly 1

Structure of unsteady flow in system of rows stator-rotor-stator of the axial compressor

Rotor blade fbw

The R flow analysis is based on measurements of static pressure pulsations on the R case with the help of 6 high-response sensors, uniformly distributed along an axis of the compressor within of an axial projection of the R blade from a leading (the sensor No.1) up to trailing (the sensor No.6) edges of R (see Fig.1).

On Fig.6 as illustration distributions on a R blade pitch (the sensor No.4, X = 0,6) of values AP(y, t) = P(y, t) — ((P)t )y, where P(y, t) is static pressure in a point of measurement are presented. Static pressure is expressed in the R system of reference and thus, corresponds to its blade to blade distri­bution at the various moments of time at fixed mutual circumferential position of the IGV and S (v = 0). Axis Oy is directed opposite to R rotation. The difference of the maximal and minimal values of AP reflects difference of static pressure on pressure and suction surfaces of the R blade, a little bit dis­tinguished (to reduction) from true owing to leakages in a radial clearance.

The analysis of results of measurements has shown, that time-averaged val­ues of static pressure difference A(P)t = (P)t — ((P)t)y practically do not depend on v for all investigated compressor assemblies 1, 2 and 3. Thus, av­eraged on time parameters of relative ft>w in the R do not depend on unsteady interaction of rows. The unsteady part of static pressure thus rather essentially depends on axial gaps, numbers of vanes in stators and their mutual circum­ferential position. Fluctuations of AP (y, t) on time for various sensors make 10 — 20% from it time-averaged value A(P)t. These fhctuations characterize unsteady interaction in system of rows stator-rotor-stator.

For an estimation of influence of v on an unsteady part of static pressure distribution and, hence, on aerodynamic loading on the R blades the value was used

m = Ry{t]v [Ry)t ■ ioo%,

(Ry )t

where Ry (t) is instant value of the linear circumferential aerodynamic load, acting on the R blade, and (Ry )t is it averaged on time value. Necessary for definition Ry (t) the difference of static pressure on the R blade was determined on measurements of static pressure on the case by sensors No.1-6. Examples of such measurements are submitted on Fig.6. Further, after project of the local loading determined thus on a circumferential direction, total loading was calculated as result of integration on 6 values of the local loading, received on each of sensors.

As illustrations on Fig.7 dependences of Г = r(t) for compressor assem­blies 1 and 3 are presented at values v = 0 and 0,6. The similar data, received for all assemblies and various values v show, that the circumferential aerody­namic loading, acting on the R blade, has on the period significant fluctuation from 5 up to 10% from its averaged on time loading. Thus the amplitude of fluctuations of loading is essentially various for various values v, in particular for assembly 1. Thus, overall aerodynamic interaction of the R with IGV and S at their mutual circumferential shift is changed significantly.

Velocity fi elds behind the rotor

Research of a velocity field behind R is based on laser flow anemometry in an axial gap between R and S and calculation of viscous flow by the numerical simulation of averaged on Reynolds 2-D unsteady Navier-Stokes equations.

Influence of mutual circumferential shift of the IGV and S on time-averaged and unsteady parameters of a vortical wake is illustrated on Fig.8, where for three compressor assemblies the measured distributions on the R pitch of time – averaged relative velocities behind R |(W)t | /Wav and their RMS deviations on time < W >t for two values of v = 0; 0,5 are presented. It is visible
that distributions of |(W)t | /Wav practically coincide at v = 0 and v = 0,5. Thus, averaged on time fl>w parameters of the R blades do not depend from unsteady aerodynamic interactions of R with IGV and S.

Apparently on Fig.8 that unsteadiness of a velocity vector in a vortical wake behind R, estimated in the value < W >t, first, it is rather significant especial in the field of a vortical wake where it achieves 25% and more, and, second, essentially changes at mutual circumferential shift of the IGV and S. As one would expect, for assembly 1 of the compressor stronger inflience of circum­ferential shift of stators on structure of unsteady vortical wakes behind R, than for assemblies 2 and 3, is observed. Besides assembly 3 differs more a high level of unsteadiness in a flow core and less expressed clocking effect, that is connected to strong potential inflience of the S upwards on a stream owing to small (Д23 = 0,05) an axial gap between R and S.

On Fig.9 the data on measurements (W)t, shown on Fig.8 for the assem­ble 1, are compared to results of direct numerical simulation (see item 2). Apparently, calculation well is coordinated to experiment in a flow core and considerably differs in the zone of vortical wake.

As a whole the received data testify that at a rotor-stator interaction flow velocity pulsations appear behind the R, especially significant in the zones of vortical wakes. From here it is necessary to expect that stators clocking most essentially may affect on value << w >t>y, which determines RMS deviation on the R pitch of RMS deviation on time of a velocity vector. Really, depen­dence from v value << w >t>y, presented on Fig.10 for various compressor assemblies, shows essential influence of stators clocking on velocity pulsa­tions in vortical wakes, in particular for assembly 1. The detailed description of structure of vortical wakes behind R blades is given in [15, 16].

Pressure pulsations behind R

Measurements of stagnation pressure pulsations were made by the insert probes, equipped with high-response sensors, on mean radius in an axial gap behind R and on an exit from the compressor. By results of static and dynamic calibrations receivers of probes are insensitive to a fl>w angles ±15° and have the uniform amplitude-freauency characteristic up to frequency 20kHz.

For an estimation of influence of the R and stators interaction on a level of stagnation pressure pulsations Pt2 (behind R) and Pt3 (behind S) values were used

[< (Pt2)t >y? + [«Pt2 >t>y]2 – 12 (< Pt2 >t )y

where Pt2 and Pt3 are expressed in the stator system of reference. It is easy to notice, that for isolated R and S values SPt2 and SPt3 are equal to zero and are nonzero for interactive rows. SPt2 and SPt3 were determined both for periodic (SPt2,3 )p and for random (SPt2,3)r pressure pulsations. Thus in expressions for SPt2 and SPt3 denominators are determined in both cases by a periodic part of pressure pulsations.

Dependences of values SPt2p, SPt2r and also SPt3p, SPt3c from parameter v are presented on Fig.11 for assembly 1. Apparently, behind R the periodic part of stagnation pressure pulsations, caused by vortical wakes, considerably surpasses random component and essentially depends from v. On an exit from the compressor, on the contrary, random component has essentially increased concerning periodic component, which poorly depends from v. The similar result is received also for assembly 2.

Undoubtedly the received data testify to evolution of vortical wakes behind R, which dissipate during mixture.

The increase of random pulsations concerning periodic pulsations on all compressor assemblies is shown also for static pressure p, which pulsations were measured on mean radius on the S vanes. Measurements were carried out by 12 high-response sensors, which were allocated in regular intervals along suction and pressure surfaces of the next S vanes (on 6 sensors on one vane) from leading up to trailing edges.

As an illustration on Fig.12 dependence from v value is presented

< P>t

A (p)t for the sensors No. 4, located on middle of a S vane chord from the suction and pressure sides. Here Ap = (pp. s.)t — (Ps. s.)t is difference of time-averaged static pressure. The value A is presented for assembly 1 both for periodic (Ap), and for random (Ar) components of pressure pulsations of p.

Apparently from Fig.12, both on suction and pressure sides of the S vanes random pressure pulsations practically do not depend from v and surpass peri­odic pulsations, for which dependence from v remains appreciable.

The important conclusion follows from results of measurement of value A(p)t for various mutual circumferential positions of stators. Dependences from v of A(p)t are presented on Fig.13 (assembly 1) for 5 pairs of sensors, where sensor No.1 and No.6, accordingly, concern to vicinities of leading and trailing edges of the S vanes. The received data show that as distributed and
overall averaged on time aerodynamic loading on the S vanes practically does not depend on a mutual circumferential positions of the IGV and S vanes.