Experimental analysis of the flow between stay and guide vanes of a pump-turbine in pumping mode

Experimental Analysis of the Flow Between Stay and Guide Vanes of a Pump-Turbine in Pumping Mode

André L. Amarante Mesquita

Departamento de Engenharia Mecânica

Centro de Tecnologia

Universidade Federal do Pará

66075-900 Belém, PA Brazil

[e-mail]

G. D. Ciocan

J. L. Kueny

LEGI ^_ Inpg

France

Abstract

An experimental study was conducted in a pump-turbine model in pumping mode, in order to characterize the flow field structure in the region between stay and guide vanes, using mainly the laser-Doppler anemometry in a two-color and back-scattered light-based system. The structure of the steady and unsteady flow was analyzed. The measurements were performed at three operation points. The obtained data provide appropriate boundary conditions and a good base of validation for numerical codes, and for the understanding of main loss mechanisms of this complex flow.

Keywords: Pump-Turbine, Stay and Guide Vanes, Laser-Doppler Anemometry, Spiral Casing, Volute.

Introduction

The hydraulic losses in the spiral casing of a pumped storage power plant are more important in pumping mode, and several studies have been conducted in order to provide prediction methods for improvement of this component (Song et al., 1996; Soares Gomes et al., 1994). The flow field inside the pump volute is strongly three-dimensional with an important vortex pair created by the curvature of the flow passage and by secondary flows, which are mainly generated by the flow through stay and guide vanes. This flow structure creates a very difficult problem for the theoretical modeling and the corresponding numerical solution, which needs to be validated by detailed experimental data.

Despite the diversity of turbines and pumps and the great effort to understand the flow to improve their performances, measurements in the stay and guide vanes regions are still poorly represented. Most of the experimental exploration is related to the runner (Wang and Hellman, 1996). Only a few experimental studies of the flow around the stay and guide vanes are found in the literature. Suzuki et al. (1996) have measured the pressure distribution on the vane surface in the turbine mode. The pressure taps were placed in a single flow inter-vane channel, and therefore the flow behavior around turbine axis was not characterized.

This work is related to an experimental study on the flow in the region between stay and guide vanes of a pump-turbine in pumping mode. The experiments were conducted out in an industrial model of a Nq 75 pump-turbine, whose volute was specially built to allow measurements with both laser-Doppler anemometer and 4-hole pressure probe. The instrumentation and method employed were well tested by Amarante Mesquita and Kueny (1993), and the three-dimensional flow structure in this volute pump model was obtained by Ciocan et al (1996) for both nominal and off-design flow conditions.

In this study, the flow in the inter-vane region was measured at three operating points. The unsteady turbulent velocity profile was obtained by using a miniature 2D laser probe, and total pressure evolution was given by a small 4-hole pressure probe. The measurements were obtained synchronized with the runner rotation, allowing the identification of the periodic component of the instantaneous velocity, and therefore, the computation of the true turbulence intensity of the flow.

The model built and the technique developed allowed the measurement of the total velocity field at the boundary of the volute. Such information is very important for establishing boundary conditions for numerical computation of the flow in the pump volute. As far as the authors know, this is the first time that these measurements were obtained.

Experimental Setup

The experimental study was conducted in the test rig for turbines and pumps at the CREMHyG - Centre de Recherche et d^'Essais de Machines Hydrauliques de Grenoble. The accuracy that can be achieved in the calculation of the efficiency and speed of the machines was 0.2 % and 0.01 %, respectively. The calibration of the instruments was performed on site. The pump-turbine model is a Nq 75 pump-turbine with a 7-blade runner. A spiral casing of 23 stay vanes and 24 guide vanes was built considering special access for instrumentation. Fig. 1 shows the spiral casing geometry and the location of the different measurement points. The spiral casing is equipped with 37 flat windows for LDA (Laser-Doppler Anemometry) measurements and 30 static pressure taps. A special microcomputer-controlled scanning system consisting of 32 electrically driven gates was used for the connection between the static pressure taps and a pressure transducer. Details of this configuration and of the complete machine geometry can be found in Amarante Mesquita (1992).

Fig. 1 The spiral casing geometry and location of the measurements

In order to allow the flow measurement between stay vanes and guide vanes, the supporting ring on the bottom side of the guide vanes in the spiral casing has been manufactured with 4 inter-vane windows, 4 passages for 4-hole Pitot, probes and 12 static pressure taps. The supporting ring can be rotated, allowing the total velocity field at the boundary of the spiral to be measured. The traversing mechanism for the displacement of the laser probe was designed to be attached to the pump body to decrease the vibration effects on the spatial stability of the measurement point. The precision of the displacement is 0.01 mm. The measurement of the laser beams position relative to the spiral casing walls was determined by searching the maximum output voltage of the LDA-photomultiplier for the light scattered from the reference dashes (0.1-mm thick) engraved on the casing walls.

Nomenclature

D = volute cross-section diameter (m)

h = inter-vane canal height (m)

k = turbulence intensity (m²/s²)

P = static or total pressure (N/m²)

P* = dimensionless pressure (-)

Q = volumetric flow rate (m³/s)

= mean velocity component (m/s)

= period velocity component (m/s)

v' = fluctuating velocity component (m/s)

v_a = axial velocity component (m/s)

v_r = radial velocity component (m/s)

v_r* = dimensionless radial velocity component (-)

v_ref = reference velocity (m/s)

v_u = tangential velocity component (m/s)

v_u* = dimensionless tangential velocity component (-)

z = local axial coordinate (m)

z* = dimensionless local axial coordinate (-)

The velocimeter employed was a 5 W argon ion, two-color LDA, in the fringe mode, operated with back-scattered light and optical fiber transmission. A 14 mm- diameter miniature laser probe was used with a 30 mm beam expander. The focal length was 100 mm, the fringe spacing 3.86 mm, and the dimensions of the measurement volume were 0.15 x 0.15 x 2.1 mm. A 40 MHz Bragg cell was employed to solve the ambiguity in the velocity component direction. The signal processing for the Doppler frequency was accomplished by a spectrum analyzer. An optical encoder attached to the turbine shaft provided the clock signal necessary to analyze the runner rotation influence on the flow. This encoder delivers a reference impulse per one runner revolution and another with 2500 pulses by revolution. These signals are sent to the spectrum analyzer and used for the synchrony analysis. The reference impulse serves also to determine the runner angular position as compared to the fixed parts of the machine by means of marks with the help of a precision stroboscope. A computer program was developed for both data acquisition and data control. The spectrum analyzer setup was adjusted with the oscilloscope display of amplitude of the Doppler sine wave and of amplitude of the pedestal. The total pressure measurements were obtained with a miniature 4-hole probe (3 mm x 2.4 mm). After calibration, this probe allows the determination of the three components of the velocity vector.

The Phase-Averaging Technique

Due to the influence of the runner rotation, the turbomachinery flows are inherently unsteady. This aspect brings some difficulties in measuring the flow field, and it is necessary to apply some special techniques to take the periodic blade movement into account (Gostelow, 1977; Strazisar, 1985).

In a periodic turbulent flow, the instantaneous velocity at a fixed point, v(t), consists of three components: the mean velocity, , the periodic component, , with a period T, and the fluctuating component, v'(t). They are combined as

(1)

Thus, to measure the turbulence intensity it is necessary to separate the components. This can be accomplished by applying a phase-averaging technique - PAT (Lakshminarayama, 1981; Bendat and Piersol, 1986).

In this work, the synchronizing one-per-revolution pulse is employed as the phase reference and the pulse trains as the timing basis for the LDA measurements. These pulses are accumulated in a counter, and each velocity measurement is associate to a clock count which is used to determine the blade row rotational position. Therefore, experimental information consists of a data word pair (velocity, angular position). The PAT method is implemented by averaging the instantaneous velocity data in appropriate measuring windows (angular interval), small enough against the periodic flow variation, ^{see Fig. 2}. Due to the several tests performed in this work, the number of windows employed for the present work was 21 per canal which corresponds to an angular interval of 2.44 degrees. Further, tests were also conducted in order to obtain a good statistical calculation for the velocity measurement in each window. The number of measured points in each window can be increased if a typical canal formed by the average between the canals is employed. Thus, using the typical canal, the best compromise between the number of acquired points and the acquisition time was approximately 300 points per window.

Fig. 2 The Phase-Averaging Technique - PAT

The mean component, computed by means of an ensemble average, is extracted from the velocity data, and the PAT method is applied. As a consequence, the fluctuating component becomes null, and then the periodic component can be obtained. With the mean and periodic components, the fluctuating component is computed according to the following expression

(2)

where v_i is the instantaneous measured velocity and N is the number of measurements.

Since the order of magnitude of the two measured fluctuating velocity components (tangential, , and radial, ) was the same, the fluctuating component of the third component (axial, , not measured) was calculated by considering that its variance is the mean value of the variance of the random variables v_u and v_r. If these variables are considered as independent, the covariance is zero, and therefore, the turbulence intensity, k, was computed according to

(3)

Preliminary measurements showed that the difference in the computation of the turbulence velocity fluctuation using Eq. (2) and the conventional mean-square- root expression, overlooking , was approximately 10%. It should be mentioned here that any difference between the runner canals adds false information to the calculation of the turbulence intensity.

Results and Discussion

The measurements were performed for five different flow rates. For the 23-mm nominal guide vane opening, the flow rates were 0.8Q_n, 0.9Q_n, Q_n and 1.2Q_n, where Q_n is the nominal flow rate. For the 20-mm opening, the flow rate selected corresponds to the maximal efficiency point. Fig 3 shows the circumferential static pressure distribution for 0.8Q_n and Q_n. P* = (P_ref-P)/rY_n is the dimensionless pressure, where P_ref is the exhaust pressure and Y_n the nominal specific energy. In this figure, each set of three closed points is corresponding to a canal between two adjacent guide vanes and in the inter-vane region ( see the static pressure taps location in Fig. 1). This canal corresponds to an angular interval of 15 degrees. One can verify the influence of the guide vanes wake on the pressure profile for both off-design and nominal flow condition, in a similar way as was also observed by Sideris and Van der Braembussche (1987). Close to 180 and 270 degrees, the pressure variation is more intense due to the tongue region proximity. The tongue region is an important singularity in the turbine geometry, and where, following a spiral casing design procedure, one guide vane was removed. However, this variation is less than 15 % close to the tongue region and less than 2 % for other angular positions.

Fig. 3 Circumferential static pressure evolution in the inter-vane canal

The pressure probe survey has revealed, according to Fig. 4, a nonuniform total pressure profile in the inter-vane canal as function of the angular and axial canal position (z* = z/h, where z is the local axial coordinate, and h is the inter-vane canal height). Based on the static and total pressure measurements, a nonuniform velocity profile was also a expected in the inter-vane region. This is confirmed by the LDA measurements. Figure 5 shows the distribution of both radial velocity component (v_r*=v_r/v_ref, where v_ref is a reference velocity) and flow angle, averaged in time (the means have been performed on 7000 data). Following an interval of 15 degrees one can clearly observe the guide-vane wake zone (the radial velocity component decreases) and, also, the variation of the flow angle due to the nonuniformity of the flow from the runner.

Fig. 4 Total pressure evolution for Q

Fig. 5 Radial velocity component and flow angle distribution

The velocity field behavior in the inter-vane region produces asymmetry in the vortex pair existing in the volute cross-section. In this section the velocity field was measured by using a methodology developed by Amarante Mesquita and Kueny (1993), based on a two-color LDA system in non-orthogonal arrangements. Fig. 6 shows an example of the LDA measurements for the main flow velocity contours (in section B) and for the secondary flow vectors (in section C). In this figure, D is the cross-section diameter, and D_r a reference). The vortex pair with nonsymmetric centers is observed in all sections and for all flow conditions.

Fig. 6 Example of the velocity distribution in the volute cross-sections

The flow distribution, which fills the volute, has a great influence on the flow characteristics in the cross-sections. This is demonstrated by the results from a numerical simulation of the flow in the volute (Soares Gomes et al., 1994), which uses both uniform velocity profile and the measured velocity distribution as inlet boundary condition. The computed mean velocity component (v_u* = v_u/v_ref) through the volute cross-sections is shown in Fig. 7, where one can verify the difference between the results, which is more important close to the tongue region. The interaction between the complex flow issuing from the intervane channel and the narrow flow passage in the volute creates a strong secondary flow, which is responsible for this weaker main-flow velocity component.

Fig. 7 Numerical simulation of the flow in the volute cross-sections

As far as the periodic velocity component is concerned, the propagation of the runner wake into the inter-vane region (1.33 outlet runner diameter) can be seen. This behavior is shown in Fig. 8, where the periodic velocity component is displayed as a function of both canal height and runner angular position (phase) in four different windows. The inclination of the synchrony wake is the same as the blade runner angle at the trailing edge. This is not the case of the configuration without the presence of stay and guide vanes, where the wake is crushed by the main flow in the volute. These data are very important for validation of unsteady numerical calculation.

Fig. 8 Periodic velocity component evolution as a function of the runner phase

Finally, the turbulence intensity measurements are reported. As already mentioned, the obtained error in the turbulence calculation due to overlooking the determining part of the synchrony velocity component is about 10%. By applying the PAT method, Fig. 9 shows one example of the turbulence intensity measurement (referenced to the v_ref²) in the inter-vane region for two different operation points. The turbulence intensity increases strongly with the removal of the nominal condition, according to the experiments performed by Casey et al. (1995). It is particularly important to note that due to the use of the PAT method, this result does not show the unsteadiness relative to the off-design flow condition but the turbulence level inherent to the physic of the flow.

Fig. 9 Turbulence level for Qn and 0.8Qn

Experimental Uncertainties

The experimental uncertainties are well analyzed by the methods described by Kline (1985), Abernathy et al. (1985) and Mofat (1985). For LDA measurements in non-orthogonal arrangements the error estimates are reported by Orloff and Snyder (1982). However, in this study the biasing errors and other error sources such as the electronic setup and the aberrations in the probe volume are not included. In fact, it is very difficult to have all these possible error sources taken into account. An analysis of the various factors which influence the overall accuracy in LDA measurements can be found in Boutier (1991). In the present work, the experimental uncertainty levels were estimated by using the methods already mentioned.

For the LDA measurements in the spiral casing sections, the uncertainty levels of the secondary velocity vectors are not uniform in the measuring sections, because the coupling angle of the non-orthogonal LDA channels are variable. The maximum values calculated are ± 0.05 m/s for the magnitude and ± 0.95 degree for the direction of the secondary velocity vectors, and ± 0.03 m/s for the through-flow velocity components. For the laser measurements in the region between stay and guide vanes, the uncertainties were estimated as 2%. The uncertainties for the 4-hole probe measurements were ± 0.75 deg for the angles, ± 0.85 % for the total pressure and ± 2.5% for the mean velocity. For the static pressure measurements (pressure tapping constructed according to the international code for model acceptance tests of hydraulic turbines) the global uncertainty is evaluated in ± 1 x 10² Pa.

Conclusion

An experimental system was developed to allow detailed flow analysis in the region between stay and guide vanes of a turbine-pump spiral casing in pumping mode. Despite the difficulty of the probe access and the complexity of the instrumentation adjustment, the flow field in the region between the stay and guide vanes of a turbine-pump model in pumping regime was characterized. The obtained experimental data allowed a detailed flow analysis for both steady and unsteady flow in this region. It was also possible to explain some secondary flow sources in the pump volute. It was demonstrated that the PAT method implemented is an adequate tool for the measurement of the unsteady and turbulent flows encountered in turbomachines. The obtained data provide appropriate boundary conditions and a good base for validation of numerical codes and for the understanding of main loss mechanisms of this rather complex flow.

Acknowledgments

EDF (Electricité de France) and NEYRPIC (GEC ALSTHOM) are gratefully acknowledged by the funding of this research. The authors would like to recognize the CAPES-COFECUB agreement (research project No. 001N/94) and CNPq (research project No. 523211/94-5) for the support in the preparation of this work.

Manuscript received: October 1997, revision received : August 1999. Technical Editor: Angela Ourívio Nieckele.

• Abernathy, R. B., Benedict, R. P., and Dowdell, R. B., 1985, "ASME Measurement Uncertainty", J. of Fluids Engineering, Vol. 107, 161-164.
• Amarante Mesquita, A. L., 1992, "Contribution à l'Analyse Expérimentale des Ecoulements 3D en Turbomachines: Vélocimétrie Laser dans une Bâche de Turbine-Pompe", in French, Ph.D. Thesis, INPG, France.
• Amarante Mesquita, A. L. and Kueny, J. L., 1993, "A Methodology for 3D LDA Measurements Using Two-Color Probes", Proceeding of the 12th Brazilian Congress of Mechanical Engineering, Vol. 2, pp. 917-920, Brasília, Brazil.
• Bendat, J. S. and Piersol, A. G., 1986, "Random Data: Analysis and Measurement Procedures", 2nd ed., Wiley-Interscience, New York.
• Boutier, A., 1991, "Accuracy of Laser Velocimetry", VKI Lecture Series on Laser Velocimetry, Rhode St. Genèse, Belgium.
• Casey, M. V., Eisele, K., Zhang, Z., Gulich, J., and Schachenmann, A., 1995, "Flow Analysis in a Pump Diffuser: LDA and PTV Measurements of the Unsteady Flow, ASME FED-Vol. 229, Laser Velocimetry, pp. 89-99.
• Ciocan, G. D., Amarante Mesquita, A. L., and Kueny, J. L., 1996, "Three-Dimensional Flow Patterns in a Pump Volute at Nominal and Off-Design Flow Conditions by LDA", Proceedings of the 6th International Symposium on Transport Phenomena and Dynamics in Rotating Machinery - ISROMAC - 6, Vol. 2, pp. 229-237, Honolulu, USA.
• Gostelow, J. P., 1977, "A New Approach to the Experimental Study of Turbomachinery Flow Phenomena", Journal of Engineering for Power, pp. 97-105.
• Kline, S. J., 1985, "The Purpose of Uncertainty Analysis", ", J. of Fluid Engineering, Vol. 107, pp. 153-160.
• Lakshminarayana, B., 1981, "Techniques for Aerodynamic and Turbulence Measurements in Turbomachinery Rotors", Journal of Engineering for Power, Vol. 103, 374-392..
• Moffat, R. J., 1985, "Using Uncertainty Analysis in the Planning of an Experiment", J. of Fluid Engineering, Vol. 107, pp. 173-178.
• Orloff, K. L., and Snyder, P. K., 1982, "Laser Doppler Anemometer Measurements using Nonorthogonal Velocity Components: Error Estimates", Applied Optics, Vol. 21, n. 2, 339-344.
• Sideris, M. T., and Van der Braembussche, R., 1987, "Influence of a Circumferential Exit Pressure Distortion on the Flow in a Impeller and Diffuser", Journal of Turbomachinery, Vol. 109, pp. 48-54.
• Song, C. C. S., Chen, C., Ikohagi, T., Sato, J., Shinmei, K. and Tani, K., 1996, "Simulation of Flow Through Pump-Turbine", in Hydraulic Machinery and Cavitation, ed. Cabrera, E., Espert, V. and Martínez, F., Vol. I, pp. 277-283, Kluwer Academic Publishers.
• Soares Gomes, F., Amarante Mesquita, A. L., Ciocan, G., and Kueny, J. L., 1994, "Numerical and Experimental Analysis of the Flow in a Pump-Turbine Spiral Casing in Pump Operation", Proceedings of the 17th Symposium of the International Association of the Hydraulic Research - IAHR, pp. 249-258, Beijing, China.
• Strazisar, A. J., 1985, "Measurement Techniques in Turbomachines", Von Karman Lectures Series 1985-03, Belgium,
• Suzuki, T., Nogafuji,T., Komiya, H., Shimada, T., Kobayashi, T. and Tanigushi, N., 1996, "Flow Behavior Around Stayvanes and Guidevanes of a Francis Turbine", Journal of Fluids Engineering, Vol. 118, pp. 110-118.
• Wang, B. and Hellman, D. H., 1996, "Turbulent 3D Flows near the Impeller of a Mixed-Flow Pump", in Hydraulic Machinery and Cavitation, ed. Cabrera, E., Espert, V. and Martínez, F., vol. I, pp. 1044-1052, Kluwer Academic Publishers.

Publication Dates

• Publication in this collection
  11 Oct 2001
• Date of issue
  Dec 1999

History

• Reviewed
  Aug 1999
• Received
  Oct 1997