See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305318409

Effects of load variation on a Kaplan turbine runner

Article  in  International Journal of Fluid Machinery and Systems · May 2016

DOI: 10.5293/IJFMS.2016.9.2.182

CITATIONS READS

5 1,151

4 authors:

Kaveh Amiri Berhanu Mulu

Luleå University of Technology Vattenfall GmbH
14 PUBLICATIONS   76 CITATIONS    15 PUBLICATIONS   154 CITATIONS   

SEE PROFILE SEE PROFILE

Michel Cervantes Mehrdad Raisee Dehkordi

Luleå University of Technology University of Tehran
143 PUBLICATIONS   1,112 CITATIONS    126 PUBLICATIONS   1,014 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

An Empirical Analysis of Energy Demand in Tunisia Besma Talbi ([email protected]) and Duc Khuong Nguyen ([email protected]) Additional contact information
Economics Bulletin, 2014, vol. 34, issue 1, 452-458 View project

scleup View project

All content following this page was uploaded by Mehrdad Raisee Dehkordi on 17 July 2016.

The user has requested enhancement of the downloaded file.

International Journal of Fluid Machinery and Systems DOI: http://dx.doi.org/10.5293/IJFMS.2016.9.2.182
Vol. 9, No. 2, April-June 2016 ISSN (Online): 1882-9554

Original Paper

Effects of load variation on a Kaplan turbine runner

K. Amiri1, B. Mulu2, M.J. Cervantes1, 3, M. Raisee4

1
Department of Engineering Science and Mathematics, Division of Fluid and Experimental Mechanics,
Luleå University of Technology, Luleå, 971 87, Sweden, [email protected], [email protected]
2
Vattenfall Research and Development, Älvkarleby, 814 70, Sweden, [email protected]
3
Department of Energy and Process Engineering, Water Power Laboratory, Norwegian University of Science
and Technology, Trondheim, 7491, Norway
4
Mechanical Engineering Department, University of Tehran, 14155-6448, Tehran, Iran, [email protected]

Abstract

Introduction of intermittent electricity production systems like wind and solar power to electricity market together
with the deregulation of electricity markets resulted in numerous start/stops, load variations and off-design operation of
water turbines. Hydraulic turbines suffer from the varying loads exerted on their stationary and rotating parts during
load variations since they are not designed for such operating conditions. Investigations on part load operation of single
regulated turbines, i.e., Francis and propeller, proved the formation of a rotating vortex rope (RVR) in the draft tube. The
RVR induces pressure pulsations in the axial and rotating directions called plunging and rotating modes, respectively.
This results in oscillating forces with two different frequencies on the runner blades, bearings and other rotating parts of
the turbine. This study investigates the effect of transient operations on the pressure fluctuations exerted on the runner
and mechanism of the RVR formation/mitigation. Draft tube and runner blades of the Porjus U9 model, a Kaplan
turbine, were equipped with pressure sensors for this purpose. The model was run in off-cam mode during different load
variations. The results showed that the transients between the best efficiency point and the high load occurs in a smooth
way. However, during transitions to the part load a RVR forms in the draft tube which induces high level of fluctuations
with two frequencies on the runner; plunging and rotating mode. Formation of the RVR during the load rejections
coincides with sudden pressure change on the runner while its mitigation occurs in a smooth way.

Keywords: Kaplan turbine, Runner pressure measurement, Load variations, Rotating vortex rope formation, Rotating vortex
rope mitigation.

1. Introduction
Transient events are known as harmful conditions for hydraulic turbines in the literature. They may lead to high pressure
fluctuations on different parts of the turbine, including the rotating parts. Such fluctuations affect the turbine lifetime [1]. Hence,
transient events often account for most of the damages sustained by hydroelectric turbines during their operation [2]. They affect
the turbine lifetime both by accelerating crack propagation on the runner blades [3] and damaging the bearings. The exerted
unsteady fluctuations during the transients are a challenge for the turbine designers and power plant owners because they are
difficult to predict during the design stage. Hydraulic turbines are subject to cyclic stresses, asymmetric forces on the runner, wear
and tear during transient operations; each of them may affect the components lifetime [4]. From a controlling point of view, the
turbines behavior varies significantly with unpredictable loads, mainly because of the turbines complexity as non-linear and non-
stationary multivariable systems. Consequently, the transient operation of the turbines poses challenges to the control community
and the existing problems has not been completely solved yet [5]. At the same time, intermittent power generation has increased
the average number of transient events that a power plant may experience during its lifetime period [6].
Most of the studies are focused on the steady state operation of the turbines and more studies on transient operation of the
turbines are required as presented in the review paper prepared by Trivedi et al., [4]. Gagnon and Leonard [2] investigated runner
blades deformation of two hydropower plants during load rejection. The study showed that in both cases the fatigue damage
increases during load variation. Gagnon et al. [3] showed that the damage to a Francis runner during start-up event is significantly
dependent on the start-up scheme. Houde et al., [1] directly measured the pressure fluctuations exerted on the runner blades of a
propeller turbine during runaway and speed-no-load conditions. The results showed that in both cases the highest amplitudes
Received August 24 2015; accepted for publication January 16 2016: Review conducted by Prof. Yoshinobu Tsujimoto.
(Paper number O15079S)
Corresponding author: K. Amiri, [email protected]
This paper was presented at the 27th IAHR Symposium on Hydraulic Machinery and Systems, September 4, Montreal, Canada.

182
appeared in the signals during the transient events. The dominant frequencies were found to be in the sub-synchronous range,
showing that the source of the fluctuations is associated with draft tube instabilities. Trivedi et al. investigated the effect of load
variation [6] and also start-up and shutdown sequences [7] on the pressure fluctuations exerted on the runner of a high head
Francis turbine with the main focus on the rotor-stator interaction. The first paper results showed that the torque starts to oscillate
with the start of the guide vane movement. An unsteady vortical flow also developed in the vaneless space during load variations
resulting in large pressure difference between the guide vanes. Runner blades experience pressure fluctuations at their leading
edge as well. The second paper showed that the pressure fluctuations on the runner were higher during shut-down process
compared to start-up. The fluctuations in the vaneless space were also higher during shut-down. Simmons et al., [8] and Simmons
[9] investigated the effect of start-up on the loads exerted on the bearings of a Kaplan turbine prototype. The results showed that
the exerted loads on the journal bearings of the turbine during start-up are higher compared with the load during steady state
operation of the turbine. Jansson [10] used the same turbine to investigate the effect of start-up on the stresses exerted on the
rotating parts of the turbine. The results showed the presence of a wide band frequency in the torque and axial strain of the turbine
main shaft during start-up process. The abovementioned studies mostly focus either on the high frequency phenomena happening
during load variations such as rotor stator interaction or general effects of load variations on turbines life time. Focusing on low
frequencies during load variation processes is valuable for understanding physics of sub-synchronous phenomena in hydraulic
turbines such as RVR formation process and its effects on turbine performance.
The hydraulic turbine investigated for the study is a 1:3.1 scaled model of the prototype investigated during start-up by
Simmons et al, [8], Simmons [9] and Jansson [10]. The model has been the case study for different experimental and numerical
studies during on-design and off-design operating conditions [11-14]. This paper deals with the effects of load acceptance and
load rejection on the pressure variations on the runner blades of the above mentioned turbine. Different load variation schemes
were investigated to estimate the exerted pressures on the runner and find the most critical load variations for the turbine. The
turbine was investigated in off-cam mode to trigger the RVR formation which its formation and mitigation are investigated.

2. Experiments and experimental setup

2.1. Model specification and operating conditions

The model used during this study is a 1:3.1 geometrically similar scaled model of a Kaplan turbine, named Porjus U9, which is
located in northern Sweden. The turbine consists of a penstock, spiral casing, distributor, runner and an elbow type draft tube. The
runner has 6 blades and the spiral casing’s distributor consists of 20 guide vanes and 18 stay vanes. The model was investigated
under a head of 7.5 m and at a rotational speed of 696.3 rpm. The guide vanes angle corresponding to the best efficiency point of
the investigated propeller curve was found to be 26.5º. Operating points with guide vanes angles of 16º and 37.5º were selected as
part load and high load for this study, respectively. The turbine operating parameters at these three operating points are presented
in Table 1. The turbine was investigated during load variations between the three operating points. Similar operating conditions
were selected by Amiri et al. [15] to investigate transitions from one operating point to the other, but with different load variation
scenarios.
Measurements were performed at Vattenfall Research and Development, Sweden. More information about the test rig can be
found in [11].
Table 1 Operating condition parameters (PL: part load; BEP: best efficiency point; HL: high load)
Operating point PL BEP HL

Guide vane angle αGV [deg] 16 26.5 37.5

Volume flow rate Qm [m3/s] 0.51 0.71 0.77

𝑄
Reduced flow rate 𝑄𝐸𝐷 = [-] 0.238 0.322 0.359
𝐷2 �𝑔𝐻

𝑛𝐷
Reduced speed 𝑛𝐸𝐷 = [-] 0.676 0.676 0.676
�𝑔𝐻

Turbine Reynolds number 𝜋𝑁𝐷2 9.11×106 9.11×106 9.11×106

𝑅𝑒 = [-]
60𝜈

Relative efficiency to BEP η-ηBEP [%] -15.2 0.0 -3.7

Output power (Pout- Pout,BEP)/Pout,BEP [%] -26 0 12

2.2. Instruments and measurement techniques

Twelve pieso-resistive pressure sensors manufactured by Kulite (LL-080) were installed on the pressure and suction sides of
two adjacent runner blades to measure the pressure distribution in a blade channel; six on each. The sensors were installed at the
vertices of a net formed by the imaginary circles of 1/3 and 2/3 of the blade span and 1/4, 1/2 and 3/4 of the blade chord; see Fig.
1. Calibration was performed in a specially manufactured tank [14]. The maximum uncertainty for all the sensors calibrations was
less than 50 Pa. The data acquisition is described in [14]. Five membrane-type pressure transducers from Druck (PDCR810)
separated by 127.5 mm were installed along the inner part of the draft tube cone and five others were installed along the outer part

183
(see Fig. 1b). The maximum uncertainty in the calibration of the cone pressure sensors was less than 100 Pa.
The signals from all the runner pressure sensors, a magnetic encoder and 10 pressure sensors installed on the inner and outer
part of the draft tube cone were recorded with a sampling frequency of 4 kHz. The encoder signal was used for phase resolving the
measurement results and runner angular velocity calculation. Its accuracy was 0.03º. The data were recorded 330 s in each case,
starting a couple of seconds before the start of the load variation process.

a) b)

Fig. 1 Pressure and LDA measurements locations. a: on the runner; b: in the draft tube cone

Laser Doppler anemometry (LDA) measurements were performed close to the middle and at the end of the draft tube cone.
These locations are shown and denominated as “section II” and “section III” in Fig. 1b. The measurement locations are 493 and
691 mm downstream the runner hub center, respectively. More details about the measurements, LDA system and data acquisition
can be found in [13].

3. Data Analysis
The main data analysis tools were developed in MATLAB. Analysis of the load variation data started with smoothing the
acquired signals of the pressure sensors installed on the stationary and the rotating parts of the turbine. Each signal can be
decomposed to a mean and fluctuating part:
𝑃(𝑡) = 𝑃�(𝑡) + 𝑃�(𝑡) (1)
where 𝑃(𝑡) is the acquired pressure signal, 𝑃�(𝑡) is the mean part and 𝑃� (𝑡) is the fluctuating part of the signal. A Matlab smooth
function using the Savitsky-Golay filter was applied on the measured signals. A smooth function with a polynomial order of 8 and
a frame size of 2 s were used to keep the frequencies ranging from the RVR frequency to the guide vanes passing frequency for
further analysis. Then the mean signal was subtracted from the original one to get the fluctuating part of each signal. Short Time
Fourier Transform (STFT) was applied on the fluctuating part of the signals to get the spectrograms. Goertzel algorithm was used
for the data analysis. Window size of 2.05 s with a 96% overlap was selected after sensitivity analysis. For detailed data analysis
procedure refer to [14,15].
Experimental investigation of the model performed by Amiri et al. [13] showed that RVR formation in the draft tube cone of
the water turbine results in pressure oscillations with rotating and plunging modes. The pressure signals acquired at the draft tube
cone were used to decompose the plunging and rotating modes of the RVR in the draft tube. Since the sensors are located on the
stationary frame, both plunging and rotating modes contribute in pressure oscillations at fRVR,st. The plunging mode results in a
synchronous oscillation in the pressure signals of the two sensors located at the same height on the inner and outer part of the draft
tube cone at fRVR,st. Whereas, the rotating mode results in oscillation of the pressures with fRVR,st but with 180º phase difference.
The pressure signals acquired at the draft tube cone were filtered with a narrow band pass filter around fRVR,st to isolate the effect
of the RVR and then subtracted from the average of the signals as proposed by Amiri et al. [15]. The resulting signals can be
written as the superposition of the plunging and rotating modes:
~
PI (t ) = A sin( 2 πf RVR ,st t + ϕ st ) + B sin( 2 πf RVR ,st t + ϕ rot ) (2)

~
( ) (
PO (t ) = A sin 2 πf RVR , st t + φst + B sin 2 πf RVR , st t + φrot + π )
~ ~
where PI (t ) and PO (t ) are the oscillating parts of the pressure signal from a sensors located on the inner and outer part of draft
tube cone, respectively. A and B are the plunging and rotating mode amplitudes, respectively. φst and φrot are the phase of the
plunging and rotating modes of the filtered signals at the inner part of the draft tube, respectively. The amplitude of the rotating
and plunging modes can be calculated as:
𝑃�𝐼 (𝑡)+𝑃�𝑂 (𝑡) 𝑃�𝐼 (𝑡)−𝑃�𝑂 (𝑡)
𝐴 = 𝑎𝑚𝑝 � 2
�, 𝐵 = 𝑎𝑚𝑝 � 2
� (3)

where amp (x) is the amplitude of the oscillating part of signal x. The method is applied both on the steady state and the load

184
variation results to decompose the two modes of the RVR.
The LDA measurement results were used to find the location of the RVR center. For that reason, the velocity measurement
results have been phase resolved with respect to the RVR frequency. Since no reference pressure measurements was recorded
simultaneous to the LDA measurements, the results were phase averaged using the tangential velocity component. Similar
procedure to the one used by Amiri et al. [16] and Jonsson et al. [12] was employed for phase resolving purpose.

4. Results and Discussion

The model was investigated during transitions between the three different operating points presented in Table 1. It was run
during off-cam mode to investigate the formation and mitigation process of the RVR. The results showed that the main features of
the flow is similar in each of the following pairs of load variations; high load to BEP and BEP to high load, high load to part load
and BEP to part load, part load to BEP and part load to high load. In the following sections, the acquired results from one of each
pairs are presented and discussed.

4.1. Load acceptance from BEP to high load

The load acceptance from the BEP to the high load starts with steady state operation of the turbine while the guide vanes angle
is set to 26.5º. After system stabilization, the guide vanes angle starts to increase linearly to 37.5º. The load variation process takes
about 13 s as presented in Fig. 2. The pressure signals are normalized with respect to their initial values, P0, in the figure. Since
the test rig is a closed loop one, there is a delay in the test rig head adjustments due to the head control system. Opening of the
guide vanes in this case results in a sudden decrease of the turbine head in the beginning of the guide vanes movement. This is due
to the fluid acceleration in the test rig circuit and pressure drop in the upper pressure tank. It takes around 130 s after the start of
the load variation process before the turbine head is completely recovered.
Figure 2 illustrates the pressure variation on two pressure sensors located at the runner blades suction side and pressure side,
close to the blade hub and trailing edge. The results show that during the load acceptance process, the pressure on the pressure
side increases whereas it decreases on the suction side. This results in higher pressure difference on the blade, higher torque and
higher output power. The results show that the variation happens in a smooth way and no specific phenomenon was captured
during the transition. Similar results were obtained by the other blade pressure sensors as well as the sensors installed on the draft
tube cone. The standard deviation of the signals increased on both blade sides during the guide vanes opening process due to the
higher flow rate, higher Reynolds number of the flow and thus increased turbulence in high load compared to the BEP.
a) pressure side sensor, PS6 b) suction side sensor, SS6

Fig. 2 Pressure variation on the runner blades during load acceptance from BEP to high load. Blue dot: instantaneous pressure,
black line: smoothed pressure, green line: guide vanes angle.

As discussed in the data analysis section, the fluctuating part of the signals are used to obtain the spectrograms. The
spectrograms of four different sensors located on the pressure and suction sides of the runner blades close to the hub and tip are
presented in Fig. 3. For all cases, the runner frequency has the largest amplitude before the start of the load variation process. The
other distinct frequencies in the spectrograms are the harmonics of the runner frequency. The amplitude of the runner frequency
increases with the guide vanes opening because of the higher energy level at the high load compared to the energy content at the
BEP. The noise in the signal increases because of the increased turbulence level as mentioned before. The transition between the
two operating points happens without occurrence of any special phenomenon; i.e., there is not any frequency appearing or
mitigating during the load variation. The acquired results during the load variation from the high load to the BEP were similar; see
Amiri et al, [15].

185
 a) suction side-hub sensor (SS4) b) suction side-tip sensor (SS1)

c) pressure side-hub sensor (PS6) d) pressure side-tip sensor (PS2)

Fig. 3 Spectrograms of the pressure sensors during load acceptance from the BEP to the high load operating point. The black
curve represents the variation of the guide vanes angle. The scale of the power spectral density is logarithmic and slightly different
for the figures.

4.2. Load rejection from BEP to part load

Investigation of the load rejection process from the BEP to the part load starts with steady state operation of the turbine at the
BEP followed by a linear guide vanes angle change from 26.5º to 16º. Pressure signals from the sensors located on the blades
suction and pressure sides close to the blade hub and the trailing edge are presented in Fig. 4. Figure 4a and 4b present the raw and
smoothed signals from the two sensors. The dashed red and green lines indicate the time of appearance of the RVR plunging and
rotating modes, respectively. The two modes and their appearance in the signals will be discussed in the following paragraphs. In
the beginning of the load rejection process, the pressure on the pressure side starts to decrease whereas it increases on the suction
side. The swirl at the runner inlet increases with the guide vanes closure; however, the runner blades are unable to extract all the
energy from the flow. The combination of the increased swirl and decreased flow rate results in the formation of a RVR in the
draft tube during the load variation [12]. At the same time as the RVR rotating component forms, the pressure on the sensors close
to the hub and trailing edge drops suddenly and then increases again. This phenomenon is marked with a red circle in Fig. 4b. This
variation was just found on the suction side for the sensors close to the hub. This results in a sudden pressure difference between
the two sides of the runner blade as presented in Fig. 4c. This may result in a sudden change in the turbine’s torque and an axial
lift affecting the forces on the thrust bearing. Following the RVR formation, the pressure signals standard deviation increases
suddenly, especially on the suction side sensors which are located in the vicinity of the RVR; see Fig. 4d. This is because of the
pressure fluctuations with respect to the RVR frequency. Formation of RVR within the draft tube induces fluctuation with respect
to the RVR frequency to the runner which is the main source of increased pressure fluctuations.
Spectrograms of the signals from different pressure sensors during the load variation process are presented in Fig. 5. Before
the start of the guide vanes movement, the only significant frequencies in the spectrograms are the runner frequency and its
harmonics. Decreasing the guide vanes angle results in a vortex breakdown in the draft tube with the formation of a RVR. The
RVR induces pressure fluctuations in the circumferential and axial directions; referred as rotating and plunging mode, respectively
[13]. The plunging mode induces pressure fluctuations with dimensionless frequency of 0.2f* and the rotating mode induces
pressure fluctuations with dimensionless frequency of 0.8f* on the runner. Figure 5 shows that the plunging mode appears in the
signal 1.3 seconds before the rotating mode appearance in the signal independent of the sensors location (this may not be clearly
seen in Fig. 5 due to the small time difference). After the RVR formation, frequency of the plunging mode increases with the
guide vanes closure; see Amiri et al, [15]. The RVR amplitude in both the rotating and plunging modes increases as well. The
rotating mode of the RVR is the dominant frequency after the load variation process. Further closure of the guide vanes induces a
wide band noise ranging from zero to around 500 Hz on the sensors located on the blades suction side close to the hub. This is
explained by the increase in the dead zone diameter below the runner with the guide vanes closure. At a guide vane angle, the
RVR radius is equal to the sensors located close to the hub and the RVR covers the sensors in this location. The wide band noise
indicates a high level of fluctuations exerted to the runner by the RVR. The fluctuations may result in the resonance and eventual
failure of some components if their natural frequencies lie within the frequency band.

186
 a) pressure side, sensor PS6 b) suction side, sensor SS6

c) cp d) standard deviation

Fig. 4 Pressure development on the runner blade surfaces during load rejection from BEP to part load together with the
variation in the standard deviation of the results. Blue dot: instantaneous pressure, black line: smoothed pressure and red line:
guide vanes angle; dashed red: start of the formation of the RVR plunging mode; dashed green: start of the formation of the RVR
rotating mode

The method explained in the data analysis section was used to decompose the plunging and rotating modes of the RVR from
the pressure signals captured by the pressure sensors installed on the draft tube cone. Figure 6 shows the amplitude growth of the
plunging and rotating modes of the RVR within the cone during load rejection from the BEP to part load. Shown in Fig. 6a, the
plunging mode starts to appear simultaneously in all pressure signals captured on the draft tube cone at around 29 s. This is
compatible with the definition of the plunging mode which results in pressure fluctuations in axial direction. The phase of the
plunging mode oscillation along the draft tube cone has also been checked. FFT has been applied on the average of the pressure
signals at each DT location (Cone1 to Cone4) after test rig stabilization; i.e. starting from 75 s. Phase of plunging mode at each
location was found using the FFT results at the RVR frequency. Presented in Fig. 7, the phase is nearly 12.5° at locations Cone2
to Cone4 which is in agreement with the definition of plunging mode. The 7° deviation of the phase at Cone1 compared with the
other locations, can be attributed to the high level of disturbances and fluctuations induced to the flow by the runner resulting in
some inaccuracy in FFT results. Comparing Fig. 6a and b, amplitude of the rotating mode close to draft tube cone inlet dominates
the amplitude of the plunging mode which is in agreement with the results of the runner where the rotating mode has a larger
amplitude. Illustrated in Fig. 6b, the appearance of the rotating mode starts from the end of the draft tube cone and then appears
sequentially from downstream to upstream (location Cone 4 to location Cone 1). This shows that the RVR starts to form at the end
of the draft tube and moves upstream with the guide vane closure. There is a delay of about 1.22 s in the appearance of the
rotating mode in location Cone 1 compared with location Cone 4 which can be translated to a RVR travelling velocity of 0.31 m/s
in upstream-wise direction. The results also show that the amplitude of the plunging mode is almost equal all along the draft tube
cone, whereas, the rotating mode has higher amplitude at the cone entrance compared with its amplitude at the cone outlet. This
can be either related to the shorter distance between the RVR and the draft tube wall at the cone inlet compared to the one at the
cone outlet or stronger vortex induced to the flow by the RVR at the cone inlet compared to the cone outlet. Location of the RVR
center within the draft tube should be found accurately in order to find the distance between the RVR and the draft tube wall and
find the main source of higher amplitude at the cone entrance compared to the one at the outlet. The RVR radius at each location
can be found using the LDA measurement results performed in the draft tube cone. The RVR does not induce any tangential
velocity to the water at its center. Hence, after phase resolving the tangential velocity with respect to the RVR frequency, the RVR
center can be considered as the point where the tangential velocity fulfills the condition of V=2πr×fRVR,st.

187
 a) suction side-hub sensor (SS5) b) suction side-tip sensor (SS1)

c) pressure side-hub sensor (PS6) d) pressure side-tip sensor (PS3)

Fig. 5 Spectrograms of the pressure sensors during load rejection from BEP to part load. The black curve represents the
variation of the guide vanes angle. The scale of the pressure amplitude is logarithmic and slightly different for the figures

a) Plunging b) Rotating

Fig. 6 RVR amplitude growth during load variation from BEP to part load. a) Plunging mode, b) rotating mode.

188
 Fig. 7 Phase of plunging mode along the draft tube cone.

Figure 8a and b show contours of tangential velocities acquired by the LDA system at sections II and III shown in Fig. 1b and
phase resolved with respect to the RVR frequency, respectively. The black circles at each sub-figure show the center of the RVR
at the measurement locations. The radius of the RVR precession at section II is close to 0.6r* and its distance from the draft tube
cone is approximately 0.5r*. At section III, the RVR precession radius is close to 0.7r* and the distance from the RVR to the draft
tube cone is equal to the one at section II; i.e. 0.5r*. Since the distance from the RVR to the draft tube cone is equal at these two
sections, it can be concluded that the RVR strength decreases in the draft tube cone in stream-wise direction.
a) Section II b) Section III

Fig. 8 Tangential velocity contours resolved with respect to the RVR frequency at a) section II, and b) section III. The black
circles show the RVR center at each of the two positions.

4.3. Load acceptance from part load to the BEP

The mitigation process of the RVR was investigated during load acceptance from the part load operating point to the BEP. The
signals of the sensors located close to the hub and trailing edge of the blades are presented in Fig. 9. The results show that the
pressure on the pressure side increases whereas it decreases on the suction side during the load acceptance process. The larger
pressure difference between the pressure and suction side results in a higher torque and thus a higher output power at the BEP
compared to the part load operating point. The standard deviation of the pressure fluctuations decreased during the load
acceptance process because of the RVR mitigation. The dashed green and red lines in Fig. 9 represent the mitigation time of the
rotating and plunging mode of the RVR, respectively. As shown in the figure, the RVR mitigates at a guide vane angle of
approximately 26º while this angle correspond to the BEP of the turbine. It shows that the transition process cannot be considered
as a quasi-steady phenomenon. The RVR mitigation happens in a smooth way and the sudden change in the pressure of the
sensors located on the suction side and close to the hub cannot be seen as during RVR formation. This is in agreement with the
results captured by Amiri et al. [16].

189
 a) Pressure side, sensor PS6 b) Suction side, sensor SS6

Fig. 9 Pressure development on the runner blade surfaces during load acceptance from the part load operating point to the
BEP. Blue dot: instantaneous pressure, black line: smoothed pressure, red line: guide vane angle, dashed green line: rotating mode
mitigation time, dashed red line: plunging mode mitigation time

The spectrograms of the fluctuating part of the pressure signals are presented in Fig. 10. The spectrograms are similar to the
previous case, but in a reverse order. The runner frequency and its harmonics are present in the spectrograms during the steady
state operation of the turbine at the part load operating point. However, the rotating and plunging frequencies of the RVR
dominate the spectrogram. A wide band noise is present in the signals of the sensors located on the suction side and close to the
hub. The RVR radius decreases with the guide vanes opening and the wide band noise disappears. The RVR frequency of the
plunging mode decreases, while it increases for the rotating mode with the guide vane opening; see Amiri [14]. Both RVR related
amplitudes decrease during the load acceptance process as well. The plunging mode of the RVR is more persistent than the
rotating mode like in the previous case. The plunging mode mitigates around 3 s after the rotating mode disappearance.
Variation in the plunging and rotating modes of the RVR extracted from the pressure signals of the draft tube cone sensors
during the load acceptance process is presented in Fig. 11. Opening the guide vanes is followed by a decrease in the amplitude of
both plunging and rotating modes of the RVR. This starts with mitigation of the rotating mode at the draft tube cone entrance, i.e.
location Cone1. This is followed by a sequential mitigation of the rotating mode from the cone entrance to its outlet. The rotating
mode of the RVR disappears from the signals of the sensors at location Cone4 at around 23 s and simultaneously, the plunging
mode disappears in all the signals. There is a delay of about 1.87 s in the disappearance of the rotating mode in Cone4 compared
with Cone1 which can be translated to a RVR travelling velocity of 0.2 m/s in downstream-wise direction.

a) suction side-hub sensor (SS5) b) suction side-tip sensor (SS1)

c) pressure side-hub sensor (PS6) d) pressure side-tip sensor (PS3)

Fig. 10 Spectrograms of the pressure sensors during load acceptance from part load to the BEP.

190
 a) Plunging b) Rotating

Fig. 11 Variation of the RVR amplitude during load acceptance from the part load operating point to the BEP.
Figure 12 shows the RVR travelling velocity along the draft tube cone during the load variations. The x-axis shows the
absolute value of the guide vane movement velocity. The results show that the RVR travelling velocity increases with increasing
the guide vane angular velocity in both cases of RVR formation and mitigation. Although, there is not any overlap between the
angular velocity of the guide vanes in two cases of RVR formation and mitigation, 50% higher velocity during RVR formation
can be an indication of higher RVR travelling velocity during RVR formation in comparison with RVR mitigation case.

Fig. 12 RVR travelling velocity vs. guide vane angular velocity.

5. Conclusion
The effect of the load variations on the pressure fluctuations exerted on the rotating parts and the draft tube of a Kaplan turbine
model was investigated. Different load acceptance and load rejections were investigated while the turbine operated under off-cam
mode. The results showed that the transitions between the high load operating point and the BEP occur in a smooth manner. The
distinctive phenomenon happening during a load rejection to the part load operating point was the RVR formation in the draft tube.
The RVR formation starts with a sudden change in the pressure on the suction side of the blade, which can result in a sudden
change of the thrust bearings load and output torque. The standard deviation of the pressure signals increases suddenly after the
RVR formation. The RVR results in induction of two dominant frequencies on the runner; the rotating and plunging mode
frequencies. The rotating component showed to be the dominant frequency during the part load operation. The plunging mode
frequency appeared on the signals a couple of seconds ahead of the rotating mode. This is explained by the formation of the RVR
at the end of the draft tube cone where the swirl number is higher. The RVR travels upstream with the guide vanes closure. The
amplitude of the RVR in the plunging and rotating mode increases with the guide vane closure. The RVR mitigation process is
showed to be similar but in a reverse order. The only difference was in the smooth disappearance of the RVR during the load
acceptance compared to the RVR formation. Generally, the load rejection to the part load was the the most harmful load variation
process for the turbine runner blades between the investigated load variations while the load variation in a RVR-free region was
comparably safe.

Acknowledgement
The authors’ gratitude goes to the Swedish Hydropower Center (SVC) for the financial support.

191
 Nomenclature

cp = ΔP/(ρgH) Pressure coefficient [-] r Radius [m]

Rotational frequency of the runner
f Re Reynolds number of the turbine [-]
[Hz]
Frequency of the rotating vortex rope
fRVR,st in the stationary frame (plunging t Time [s]
mode) [Hz]
g Gravitational acceleration [m/s2] tgen Transition time [s]
H Head during the model test [m] U Axial velocity [m/s]
nED Reduced speed of the turbine [-] V Tangential velocity [m/s]
P Static pressure [Pa] αGV Gguide vane angle [deg]
Fluctuating component of pressure
P̂ η Turbine efficiency [-]
[Pa]
~ Oscillating component of pressure
P ρ Fluid density [kg/m3]
[Pa]

∑ (P - P )
N 2 Standard deviation
P Time-averaged pressure [Pa] i
σ= i =1
N
Phase of the pressure fluctuations due to the
Pout Output power [kW] φrot
rotating mode of the RVR [rad]
Phase of the pressure fluctuations due to the
Qm Model flow rate [m3/s] φst
plunging mode of the RVR [rad]
QED Reduced flow rate of the turbine [-] ωGV Guide vane angular velocity [deg/s]
Super- and sub-scripts
Frequency is made dimensionless with respect to the
rotational frequency of the runner, dimensions with
* O Sensors is located on the outer part of draft tube
respect to the runner diameter and velocities with
respect to the reference velocity
I Sensors is located on the inner part of draft tube BEP best efficiency point

References
[1] Houde, S., Fraser, R., Ciocan, G., Deschênes, C., 2012, “Experimental study of the pressure fluctuations on propeller turbine
runner blades: part 2, transient conditions,” Earth and Environmental Science, Vol. 15, 062061.
[2] Gagnon, M., Leonard, F., 2013, “Transient response and life assessment: Case studies on the load rejection of two
hydroelectric turbines,” Proceedings of the International Conference Surveillance, Chartres, France, pp. 1-11.
[3] Gagnon, M., Tahan, S. A., Bocher, P., Thibault, D., 2010, “Impact of startup scheme on Francis runner life expectancy,” Earth
and Environmental Science, Vol. 12, 012107.
[4] Trivedi, C., Gandi, B. K., Cervantes, M. J., 2012, “Effect of transients on Francis turbine runner life: a review,” Journal of
Hydraulic Research, Vol. 51, No. 2, pp. 121-132.
[5] Kishor, N., Saini, R. P., Singh, S. P., 2007, “A review on hydropower plant models and control,” Renewable and Sustainable
Energy Reviews, Vol. 11, No. 5, pp. 776-796.
[6] Chirag, T., Cervantes, M. J., Bhupendrakumar, G., Dahlhaug, O. G., 2014, “Pressure measurements on a high-head Francis
turbine during load acceptance and rejection,” Journal of Hydraulic Research, Vol. 52, No.2, pp. 283-279
[7] Trivedi, C., Cervantes, M. J., Gandhi, B. K., Dahlhaug, O. G., 2014, “Experimental investigations of transient pressure
variations in a high head model Francis turbine during start-up and shutdown,” Journal of Hydrodynamics, Vol. 26, No., 2, pp.
277-290.
[8] Simmons, G. F., Aidanpää, J., Cervantes, M. J., Glavatskih, S., 2013, “Operational transients in the guide bearings of a 10 MW
Kaplan turbine,” International journal on hydropower and dams, Vol. 20, No.5, pp. 94-100.
[9] Simmons, G. F., 2013, “Journal Bearing Design, Lubrication and Operation for Enhanced Performance,” Ph. D. Thesis, Luleå
Univ. of Tech., Luleå, Sweden.
[10] Jansson, I., 2013, “Vibrant bodies of swirling flow: On the limits of mechanical power transformation,” Ph. D. Thesis, Luleå
Univ. of Tech., Luleå, Sweden.
[11] Mulu, B. G., Jonsson, P. P., Cervantes, M. J., “2012 Experimental investigation of a Kaplan draft tube – Part I: Best
efficiency point,” Applied Energy, Vol. 93, pp. 695-706.
[12] Jonsson, P. P., Mulu, B. G., Cervantes, M. J., 2012, “Experimental investigation of a Kaplan draft tube – Part II: Off-design
conditions,” Applied Energy, Vol. 94, pp. 71-83.
[13] Amiri, K., Cervantes, M. J., Mulu, B. G., 2015, “Experimental investigation of the hydraulic loads on the runner of a Kaplan
turbine model and the corresponding prototype,” Journal of Hydraulic Research, Vol. 53, No.4, pp. 452-465.
[14] Amiri, K., 2014, “An Experimental Investigation of flow in a Kaplan runner: steady-state and transient,” Licentiate Thesis,
Luleå Univ. of Tech, Luleå, Sweden.

192
[15] Amiri, K., Mulu, B. G., Raisee, M., Cervantes, M. J., 2015, “Unsteady pressure measurements on the runner of a Kaplan
turbine during load acceptance and load rejection,” Journal of Hydraulic Research, DOI:10.1080/00221686.2015.1110626.
[16] Amiri, K., Mulu, B. G., Cervantes, M. J., 2015, “Experimental Investigation of the Interblade Flow in a Kaplan Runner at
Several Operating Points Using Laser Doppler Anemometry,” Journal of Fluids Engineering, Vol. 138, 021106-2.

193
View publication stats