International Journal of Fluid Machinery and Systems DOI: [Link]

304
Vol. 11, No. 3, July-September 2018 ISSN (Online): 1882-9554

Original Paper

Fast Transition from Pump to Turbine Mode of Operation

Magni F. Svarstad1 and Torbjørn K. Nielsen1

1
Waterpower Laboratory, Norwegian University of Science and Technology
Alfred Getz’ v. 4, 7034 Trondheim, Norway, [Link]@[Link], [Link]@[Link]

Abstract

The reversible pump turbine (RPT) is a suitable machine to control fluctuations in the energy market. The usage of
RPTs for this purpose will increase the number of operational mode changes of the machine. In order to reduce the response
time of the machine, fast transitions between the modes of operation are necessary. Therefore, increased knowledge of
how the machine operates during these fast transitions is needed. This includes the investigation of the transient
characteristics for the whole operating range of the machine. This paper presents experimental results from a fast transition
from pump to turbine mode. The flow rate is measured by the use of a modified pressure time method. The resulting
transient characteristics are compared with steady state characteristics. Experiments have been preformed on a model
scale reversible pump turbine in the Waterpower Laboratory at the Norwegian University of Science and Technology
(NTNU). The results show that the pressure pulsations are highest at low discharge in both pump and turbine mode of
operation and at runaway speed in turbine. Oscillations at runaway speed is reduced with lower opening degree on the
guide vanes. The results also show a difference between the steady state and transient characteristics in the pump mode
due to the inertia of the water masses.

Keywords: Reversible pump turbine, Tranient operation, Mode change

1. Introduction
When discussing the future of the energy market, the reversible pump turbine (RPT) are often given the role of balancing the
intermittent energy production. The introduction of intermittent renewable energy sources calls for a faster regulating force in the
energy market. Balancing the energy production can be done by RPTs and require the machines to change the mode of operation
more frequently. In such a scenario, the need is not only for an increase in operational mode changes, but also for a decrease in the
transition time between the pump and turbine mode of operation.
In reversible pump turbines, as in Francis turbines, it is well documented that the machines experience high fatigue loads in off-
design and start and stop operations [1]–[3].
For reversible pump turbines the instabilities that can occur during start up, both in pump mode [4] and turbine mode [5] are also
areas where a lot of research is available.
There is, however, less information about the transition between the different operational modes. Early, and most well-known
research on this topic is the four quadrant characteristics of a pump presented in [6] by Stepanoff. The experimental data describing
the four quadrant characteristics in [6] was done by Knapp [7].
Ruchonnet and Braun [8] have presented transient characteristics from experimental research done from pump to turbine mode
and from turbine to pump mode.
They refer to high-pressure pulsations during the transient period, and especially high amplitudes in energy dissipation mode.
Stens and Riedelbauch [9] have investigated the same transient operation using Computational Fluid Dynamics (CFD). They
conclude that the fast transitions lead to stall conditions between the guide vanes and vortices in the runner and draft tube, and that
this is the cause of the off-design conditions in the runner. Liu et al. [10] have done CFD on a transient process caused by power
failure from pump mode to turbine mode. The authors show that the minimum value for the transient head occurs when the flow
rate is approximately zero, which means between energy dissipation mode and turbine mode of operation.
When connecting a hydraulic machine to the grid, the rotational speed of the machine have to be equal the synchronous speed.
The most common way to achieve the correct rotational speed before connecting the machine to the grid is to let the machine rotate
at runaway speed with a given opening of the guide vanes. It is therefore of interest to look closer at the systems behaviour at
runaway speed after a fast transition from pump to turbine mode of operation.
This article investigates the transient characteristics of a reversible pump turbine, from pump to turbine mode of operation. An
important part of the work has been the development of a reliable method to measure the flow rate in this scenario. The method of
measuring the flow rate, together with the resulting transient characteristics, are presented in the article. The transient characteristics
are also compared to the steady state characteristics for the same machine.
Received November 12 2017; accepted for publication May 8 2018: Review conducted by Shogo Nakamura. (Paper number O17073J)
Corresponding author: Magni F. Svarstad, PhD candidate, [Link]@[Link]

304
2. Experimental setup
When doing transient experiments it is important to remove disturbances on the system caused by the pump, if run in closed loop.
The Francis test rig was therefore arranged to operate in an open loop configuration when testing the reversible pump turbine. As
seen in Fig. 1 the upper reservoir have a constant head regulated by an overflow valve (1) and fed by a centrifugal pump (6).
Upstream the turbine (3) there is a pressurized tank (2) with an air cushion. The draft tube ends up in an outlet tank (4), and the
tanks water level is held constant by another centrifugal pump (5). In order to disconnect the hydraulic inertia of the outlet system
a weir is installed in the outlet tank. The head is defined as the total energy difference between inlet of the turbine and outlet of the
draft tube according to the convention as seen in eq. (1).

Table 1 RPT runner parameters

D1 D2 B1 ned Qed H* Βhp Βlp α*
0.631 m 0.349 m 0.059 m 0.133 0.223 29.3 12° 12.8° 10°

The model runner is designed by Grunde Olimstad [11] during his PhD work. The laboratory set-up is, with some modifications,
the same as used by Eve Walseth as part of her PhD thesis [12]. The runner parameters are listed in Table 1.

Fig. 1 The open loop configuration of the Francis test rig with; upper reservoir (1), pressure tank (2), RPT (3), draft tube
tank (4), downstream feeding pump (5), upstream feeding pump (6), main water reservoir (7) and weir in draft tube tank (8)

2.1 Instrumentation
Flow rate, torque, rotational speed and pressure were measured during the experiment. Table 1 lists the different sensors, and
details for each instrument used indicate the placement of each instrument. The head calculated as shown in eq. (1). The sample rate
of 5000 Hz was used.

Table 2 Instrument details

Abbreviation Description Sensor Type Location
pi1 Inlet pressure Druck UNIK 5000 [Link]
pi2 Upstream inlet pressure Druck UNIK 5000 Inlet pipe
Pvl Vaneless space pressure Kulite Vaneless space
Pdt Draft tube pressure Kisler Draft tube upper
Po Outlet pressure PTX Draft tube lower
M Torquexxxxxxxxxxxxxxxxx Hottinger Main shaft
Qm Electromagnetic flowmeter Krohnexxxxxxxxx Inlet pipe
n Rotational speed Encoder Generator draft

𝑝𝑖1 − 𝑝𝑜 𝑄𝑔2 1 1
𝐻= + ∙ ( 2 − 2) (1)
𝜌𝑔 2𝑔 𝐴𝑖1 𝐴𝑜

305
 Fig. 2 Instrument locations in the test section

2.2 Transient flow rate measurement

To measure the discharge fast and with high accuracy, a modified pressure-time method was used. The pressure-time (Gibson)
method referred to in IEC 60193 [13] is used for measuring the initial flow using the transient pressure difference as the guide vanes
closes to zero opening. In this experiment however, a modified pressure-time method was used. By measuring the pressure
difference over a certain pipe length, the pressure difference can be used to calculate the transient flow during the mode change,
using the electromagnetic flow meter (EMF) as a verification of the flow at steady state, both at the start and the end of each transient
sequence. Nielsen [14] was the first to use the pressure-time method to find the flow at each increment of time in a transient sequence,
while doing tests in the turbine mode of operation.

(a) (b)
Fig. 3 The differential pressure ΔP (a) used to calculate the flow Q (b) where the steady state area at start and end are
given by the EMF

Figure 3(a) shows the measured differential pressure between two pressure sensors (pi1 and pi2) with a distance of 5.345 m, and
with a constant cross sectional area between them. The transient flow rate as seen in Fig. 3(b) is found by integrating the pressure
difference in accordance with the pressure-time method as stated in eq. (3) and eq. (4),

𝐴 𝑡𝑖
𝛥𝑄𝑖 = ∫ (𝛥𝑝 + 𝜁) ⅆ𝑡 (2)
𝜌𝐿 𝑡𝑖−1

𝜁 = 𝑘𝑄2 (3)

where 𝜁 is the friction loss in the pipe section, k is the friction factor, 𝛥Q is the change in flow in one time step. A is the cross
sectional area and L is the length of the pipe section. 𝜌 is the water density and 𝛥𝑝 is the differential pressure of the two pressure
sensors. Each measurement is divided into three parts. t0 ≤ Ts-s < t1 is the time period from the start of the steady state measurement
up to the start of the transient period. t1 ≤ TD < t2 is the transient period where the RPT changes from pump to turbine mode of
operation. t2 ≤ TS-E < t3 is the steady state period from the end of the transient measurement to the end of the measurement.

The flow rate in the transient section Qg was obtained as follows:

 The flow rate, Qm at TS-S and TS-E were measured by the flow meter.
Δp
 The average friction factor for the time period TS-S, k = 2Ts−s
QTs−s
 The friction loss was found at TS-S and TS-E using the relationship given in eq. (3).
 𝜁𝑇𝑆−𝑆 was used until the rotational speed changed direction, then 𝜁𝑇𝑆−𝐸 was used for the rest of the measurement.
 𝛥𝑄 was calculated by the pressure difference for each time step and 𝑄𝑖 = 𝑄𝑖−1 + Δ𝑄𝑔𝑖

306
2.2 Experiment procedure
The procedure consisted of first setting the required head in the upper and lower reservoir. Then the rotational speed in pump
mode was adjusted and the system was given time to stabilize. The generator torque was disconnected, causing the hydraulic pressure
to force the RPT from pump mode to turbine mode. The transient measurement ended at turbine runaway speed. Measurements
were preformed for constant guide vane angle (GV) 7º, 10º and 13º, for each GV the measurement was repeated seven times.

3. Results
3.1 Transient characteristics

(a) (b)
Fig. 4 Transient measurements for three different guide vane openings in non-dimensional (a) and physical (b) units. I is
turbine mode, II is dissipation mode, III is pump mode and IV are reverse pump mode.

The four quadrant, transient characteristics for three different guide vane openings are presented in Fig. 4, shown as machine
characteristics Fig. 4(a) and system characteristics Fig. 4(b). Machine characteristics have non-dimensional units QED and nED as
defined in eq. (4) and eq. (5). The non-dimensional units are independent of both the system around the machine and the
dimensions of the machine parameters.
Q
Q ED = (4)
𝐷2 √𝑔𝐻
nD
𝑛ED = (5)
√𝑔𝐻
In Fig. 4(b) the different guide vane openings give a similar flow rate per rotational speed in pump mode (III). When the flow
changes direction in quadrant II the effect of the guide vane positions becomes apparent. Finally, in turbine mode of operation the
result of an increased guide vane opening can be seen giving an increased spiral around the runaway position.

Fig. 5 The change in head (H) during the transient experiment. The numbering show start of T D (1), flow rate zero (2),
rotational speed zero (3), flow rate maximum (4), rotational speed maximum (5).

The pressure oscillations in the system can be seen in Fig. 5 and show the changes in head over time. Since the transient
measurements shown in Fig. 4(a) include H, the same oscillations can be seen in this figure. The highest oscillations are found in
the dissipation mode in quadrant II and at turbine runaway in quadrant I. All of the guide vane openings show similar magnitude
in the oscillations in dissipation mode. At runaway speed the oscillations increase with increasing opening.

307
 (a) (b)
Fig. 6 Transient and steady state characteristic in non-dimensional (a) and physical (b) units

Figure 6 compare the transient measurement with steady state measurements. Figure 6(b) has two areas where the transient and
steady state measurements differ significantly, when the rotational speed is negative and around runaway speed. In Fig. 6(a) the
difference are concentrated in quadrant II and III.
The difference between the steady state and transient measurements in turbine mode of operation has been addressed by Nielsen
[14]. By removing the inertia of the water masses as in eq. (6) when calculating the head, he showed that the transient and steady
state characteristics becomes equal around runaway speed.
ⅆQ
Ht = H − I (6)
ⅆt
A
I=∑
gL
In eq. (6) H is the measured head, I is the hydraulic inertia, dQ/dt is the change in flow rate over time and Ht is the transient
head. A and L is the cross-sectional area and the length of the pipe segment respectively.

Fig. 7 The difference between the measured transient characteristics with H, the adjusted transient characteristics with Ht and
the steady state characteristics.

Figure 7 shows that the difference between the transient and steady state characteristics in quadrant II and III are also removed
by Nielsens method. Calculating QED and nED from eq. (4) and eq. (5) for the transient characteristics use the measured H, the
adjusted QED-nED characteristics use Ht. The transient characteristics with Ht and its correspondence to the steady state
measurements and can thus be explained by the inertia of the water masses.
3.2 Uncertainty
Figure 8 show the repeatability for QED vs nED for four repetitions at 10º guide vane opening.

308
 Fig. 8 The repeatability at 10º guide vane opening for measurement series 2 to 5

The absolute uncertainty for N and Q in the transient period TD for one measurement series can be seen in Fig. 9. Where the
uncertainty in the flow rate is shown in Fig. 9(b) and the uncertainty in the rotational speed is shown in Fig. 9(c). It is prudent to
note that the uncertainty in flow measured during a transient phase makes it difficult to decide the standard deviation. In this
experiment it was decided to use the standard deviation from TSS when calculating the uncertainty in TD. Due to the chosen standard
deviation the uncertainty is probably marginally higher than calculated. For the rotational speed, it is noticeable that the uncertainty
increases rapidly around zero. This is due to the encoders dependence on the pulses to determine the rpm. There are 1024 pulses for
one rotation, but with close to zero rotation the result is a high increase in the uncertainty around that point. Excluding the uncertainty
at zero, the mean absolute uncertainty, eN for N is 0.0034 rpm. The mean absolute uncertainty, e Q for the flow rate is 0.000156.

(a)

(b) (c)
Fig. 9 (a) is the time dependent change in Q and N, (b) and (c) show the absolute uncertainty and the real value in the
transient period TD for respectively Q and N. The uncertainty band is multiplied with 100 to enhance visibility

309
4. Discussion
The use of a modified pressure-time method for measuring the flow rate in a transient phenomena with big changes in the flow
is previously unpublished. Nielsen [14] and Walseth [12] have used this method in the turbine mode of operation, but to the authors
knowledge this is the first time the method is used to measure the transient flow from pump to turbine mode of operation. The
uncertainty shown in Fig. 9(b) shows this method to be a reliable way to measure the flow.
There is a significant difference between the steady state and transient measurements in pump and pump break mode. Where, for
the same flow, the rotational speed is smaller for the transient measurements than the steady state measurements. This phenomenon
was also observed by Liu et al. [10] and Stens [9], but dismissed as an inaccuracy of the CFD in pump mode. The inertia of the
water masses explains the difference. Nielsen [14] showed that by taking into account the inertia of the water masses removes the
difference between the transient and steady state characteristics. The transient measurements are influenced by the system dynamics.
By accounting for this and subtracting the hydraulic inertia in the system from the head, we get the same results as in a steady
state situation without system dynamics.
The pressure fluctuations that can be seen in Fig. 4(a) and Fig. 5 are highest where the discharge is low and at runaway speed in
turbine mode of operation. This is in accordance with previous research by Zuo [1] and Tanaka [3], where it was found that these
areas give the highest vibrations due to the off-design conditions.
In Fig. 4(b) the flow and rotational speed oscillations at runaway speed is reduced with lower opening degree on the guide vanes.
These flow and speed oscillations are caused by the slower head oscillations seen in Fig. 5. The decrease in these oscillations
dependent on the decrease in guide vane opening degree is because of the reduced flow through the turbine, and a smaller amplitude
of the head oscillations after reaching the turbine quadrant (I). When doing a fast mode change ending at runaway speed it can
therefore be recommended to do this in a manner that ends up at a low guide vane angle when the RPT reaches runaway speed. This
can be done with a constant low GV angle from pump to turbine mode of operation, or with a gradual reduction in the GV angle
during the transition.

5. Conclusion
This article presents an investigation of the transient characteristics of a reversible pump turbine, from pump to turbine mode of
operation measured experimentally. It has been shown that the modified pressure time method is a reliable method to measure the
discharge in this transient phase. Pressure fluctuations are highest where they can be expected to be, i.e at low discharge in both
pump and turbine mode of operation and at runaway speed in turbine. The comparison of the steady state and transient characteristics
show a difference in the pump mode due to the inertia of the water masses.

Nomenclature
A Area [m] Q Flow rate [m3/s]
B1 Inlet runner height [m] p Pressure [kPa]
D Diameter [m] α Guide vane angle [º]
g Gravitational acceleration [m/s2] β Mean velocity component [º]
GV Guide vane angle [º] ρ Density [kg/m3]
H Head [m] ϛ Friction loss [-]
I Hydraulic inertia [1/s2] dt Position indication [-]
k Friction factor [-] i1 Position indication [-]
L Length [m] o Position indication [-]
M Torque [Nm] vl Position indication [-]
N,n Rotational speed [rad/s] 1 Inlet position turbine direction [-]
nED Non-dimensional speed [-] 2 Outlet position turbine direction [-]
QED Non-dimensional flow [-] * Best operation point [-]

References
[1] Z. Zuo, S. Liu, Y. Sun, and Y. Wu, “Pressure fluctuations in the vaneless space of High-head pump-turbines—A review,” Renew.
Sustain. Energy Rev., vol. 41, pp. 965–974, Jan. 2015.
[2] E. Egusquiza, C. Valero, D. Valentin, A. Presas, and C. G. Rodriguez, “Condition monitoring of pump-turbines. New challenges,”
Meas. J. Int. Meas. Confed., vol. 67, pp. 151–163, 2015.
[3] H. Tanaka, “Vibration Behavior and Dynamic Stress of Runners of Very High Head Reversible Pump-turbines,” Int. J. Fluid
Mach. Syst., vol. 4, no. 2, pp. 289–306, Jun. 2011.
[4] L. Deyou, W. Hongjie, X. Gaoming, G. Ruzhi, W. Xianzhu, and L. Zhansheng, “Unsteady simulation and analysis for hump
characteristics of a pump turbine model,” Renew. Energy, vol. 77, pp. 32–42, May 2015.
[5] G. Olimstad, T. Nielsen, and B. Borresen, “Stability limits of reversible-pump turbines in turbine mode of operation and
measurements of unstable characteristics,” J. Fluids Eng., vol. 134, no. 11, p. 111202, Nov. 2012.
[6] A. J. Stepanoff, Centrifugal and axial flow pumps: theory, design, and application, 2nd ed. New York: Wiley, 1957.
[7] R. T. Knapp, “Complete characteristics of centrifugal pumps and their use in the prediction of transient behavior,” Trans. ASME,
pp. 683–689, 1937.
[8] N. RUCHONNET and O. BRAUN, “Reduced scale model test of pump-turbine transition,” in IAHR WG Meeting on Cavitation

310
and Dynamic Problems in Hydraulic Machinery and Systems.
[9] C. Stens and S. Riedelbauch, “Investigation of a fast transition from pump mode to generating mode in a model scale reversible
pump turbine,” IOP Conf. Ser. Earth Environ. Sci., vol. 49, p. 112001, Nov. 2016.
[10] J. Liu, S. Liu, Y. Sun, L. Jiao, Y. Wu, and L. Wang, “Three-dimensional flow simulation of transient power interruption process
of a prototype pump-turbine at pump mode,” J. Mech. Sci. Technol., vol. 27, no. 5, pp. 1305–1312, May 2013.
[11] G. Olimstad, “Design of a Reversible Pump-Turbine -with Purpose to Investigate Stability,” presented at the 4th International
Meeting on Cavitation and Dynamic Problems in Hydraulic Machinery and Systems, 2011.
[12] E. C. Walseth, T. K. Nielsen, and B. Svingen, “Measuring the Dynamic Characteristics of a Low Specific Speed Pump—
Turbine Model,” Energies, vol. 9, no. 3, p. 199, Mar. 2016.
[13] IEC, “Hydraulic turbines, storage pumps and pump-turbines - Model acceptance tests,” Technical Standard NEK EN
60193:1999, 1999.
[14] Torbjørn K. Nielsen, “Transient characteristics of high head Francis turbines,” Universitetet i Trondheim, Norges tekniske
høgskole, Institutt for hydro- og gassdynamikk, Trondheim, 1990.

311