Renewable Energy 78 (2015) 1e6

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Technical note

Design and performance evaluation of a pump-as-turbine

micro-hydro test facility with incorporated inlet ow control
D.R. Giosio a, *, A.D. Henderson a, J.M. Walker b, P.A. Brandner b, J.E. Sargison c, P. Gautam d
a
School of Engineering, The University of Tasmania, Hobart, Tasmania, 7001, Australia
b
National Centre for Maritime Engineering & Hydrodynamics, Australian Maritime College, University of Tasmania, Launceston, Tasmania, 7248, Australia
c
JSA Consulting Engineers, Sandy Bay, Tasmania, 7005, Australia
d
Genesis Energy, Hamilton, Waikato, 3204, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: In the context of micro-hydro power schemes the initial cost of conventional Francis turbine units is
Received 5 February 2014 often prohibitive. As such there is growing interest in pump-as-turbine (PAT) technology offering a more
Accepted 10 December 2014 cost effective, yet still highly efcient, power generating alternative, nding uses in remote area power
Available online
supply and energy recovery systems. However, the implementation of a PAT is highly problematic in
terms of predicting the installed best operating point coupled with poor off-design performance due to
Keywords:
the xed geometry and absence of inlet ow control. In the current work a micro-hydro test facility and
Micro-hydro
turbine unit is developed utilising a commercially available pump impeller together with a customised
Hydro power
Pump-as-turbine
housing for incorporation of inlet ow control. Working initially from established PAT theory, this paper
Remote area power generation presents the design and performance testing of a hydraulic turbine unit suitable for use in rural micro-
Waste energy recovery systems hydro, and energy recovery installations. Maximum efciency of the unit was found to be 79%,
marginally higher than that of the parent pump, while the off-design efciency offered considerable
improvement over previously published data of traditional PAT systems. The design provides a cost
effective power generator in comparison to small scale Francis turbines, while providing a greater
operational range than traditional PAT units.
2014 Published by Elsevier Ltd.

1. Introduction However, a major drawback of PATs is that the performance away

The ability for pumps to operate efciently in reverse as turbines
was rst established by Thoma [1] in 1931 while mapping the full
operating characteristic of a centrifugal pump. In recent decades
there has been renewed interest in pump-as-turbine (PAT) tech-
nology which has found signicant use in remote area power
supply installations, both on- and off-grid, as well as industrial
application in energy recovery systems where a high pressure
water source exists that would otherwise require throttling.
In such instances, within the range designated micro-hydro
power (<100 kW), the initial capital cost of a conventional Francis
turbine generating set is often prohibitive, making PAT installations
an attractive alternative with signicantly shorter payback periods.
* Corresponding author.
E-mail addresses: Dean.Giosio@utas.edu.au (D.R. Giosio), Alan.Henderson@utas.
edu.au (A.D. Henderson), Jessica.Walker@utas.edu.au (J.M. Walker), P.Brandner@
utas.edu.au (P.A. Brandner), jane@jsaengineers.com.au (J.E. Sargison), prakash.
gautam@genesisenergy.co.nz (P. Gautam).
from best efciency point (BEP) is extremely poor due to the xed
internal geometry and absence of ow regulation. Various authors
have provided, with positive results, a number of relatively simple
modications such as impeller tip and hub/shroud rounding in
order to increase overall PAT performance [2e4]. However, the
rapid efciency drop-off at off-design conditions remains an
inherent and major limitation of PATs.
This is further compounded by the current lack of accurate and
reliable methods for predicting expected PAT performance (BEP)
from available pump manufacturer data. In a review of PAT per-
formance prediction methods Williams [5] denes a prediction
criterion, C, as a means to assess the accuracy of eight prediction
models suggested by various authors for predicting the turbine BEP
based on either pump performance data at BEP, or dimensional
pump specic speed, nq NQ0.5/H0.75 where N [rpm], Q [m3/s] and
H [m] at rated. Each method was compared to turbine test data of
35 pumps of various sizes and with nq ranging from 12.7 to 183.3.
The method proposed by Sharma [6] was found to be the most
accurate, however, 20% of the tested pumps were still outside the

http://dx.doi.org/10.1016/j.renene.2014.12.027
0960-1481/ 2014 Published by Elsevier Ltd.
2 D.R. Giosio et al. / Renewable Energy 78 (2015) 1e6
acceptable range of the prediction criterion. More recently, Ven-
trone et al. [7] presented a detailed method of turbine performance
prediction through the denition of a runner momentum coef-
cient, jR, which describes the specic momentum work performed
within the runner proper assuming zero incident losses. For the
four pumps tested the proposed method gives values within 4% of
experimental values, although a larger sample size is needed to
verify the reliability of the proposed method. Moreover, the
method requires specic knowledge of pump geometric parame-
ters, as well as performance curves, and is therefore limited in its
ability as a pump selection tool as these details are often not readily
available.
Consequently, due to the nature and constraints of micro-hydro
sites, if the installed BEP is found to differ to some degree from the
predicted operating point the PAT will operate at sub-optimal ef-
ciency for a high proportion of its operational life.
Recent work by Alexander and Giddens [8], and Alexander et al.
[9] has investigated an extensive range of modular, xed geometry
microhydro turbine units. The turbine units described are designed
for ease of manufacture and consistantly obtain hydraulic ef-
ciencies greater than 70%. Most importantly, perhaps, is the scale-
ability of the units, providing micro-hydro developers with units
whereby the performance at site conditions may be well predicted.
However, as with traditional PAT units, the lack of inlet ow control
may somewhat limit the efcient turbine operating range.
This paper presents the development of a turbine using an off-
the-shelf pump impeller together with a customised housing
incorporating inlet ow control. Working initially from established
PAT theory the design and performance testing of a hydraulic tur-
bine unit suitable for use in rural micro-hydro and energy recovery
installations is given. The resulting design provides a cost effective
power generator in comparison to small scale Francis turbines
while providing a greater operational range than traditional pump-
as-turbine units. In typical micro-hydro power applications, where
water storage catchments or industrial operating conditions may
vary, the improved off-design efciency is highly desirable.
2. Experimental facility design
A micro-hydro experimental facility was developed comprising
a supply pump tted with variable speed drive (VSD), elevated
Fig. 1. Pump-turbine unit installed at the University of Tasmania micro-hydro test facility.
water storage, adjustable turbine and generator support frame,
pump-turbine unit, clear acrylic circular section draft tube, and
open discharge tank with return channel to underoor storage pit.
Flow may be provided directly from the 22 kW supply pump,
delivering up to 10 m head at 150 L/s, or via a hydraulically isolated
storage tank for investigation of penstock pressure pulsations
associated with transient operation, see Figs. 1 and 2.
2.1. Pump as turbine unit design principles
In most site installations of PAT units an induction motor is used
in place of a synchronous generator [10]. A directly coupled in-
duction generator removes the need for any belts or gearing,
minimises any lateral force thereby prolonging bearing life, elimi-
nates the requirement for a turbine shaft bearing, and is low cost -
particularly compared to synchronous generators of sizes up to
30 kW [10]. However this requires that one of the synchronous
speeds, corresponding to the number of poles of the induction
generator, must be chosen as the turbine operating speed [4].
As a rst approximation the method of Sharma [6] was used to
determine the required pump characteristics. In order to allow for
operation at a reduced speed an additional correction was made
based on turbine afnity laws as Q fN and HfN2 such that
 2
Nt QpBEP Nt HpBEP
Qt ; Ht (1)
Np h0:8
pBEP Np h1:2
pBEP
Knowing the available turbine net head and ow rate, Ht and Qt,
and the required turbine speed, Nt, Eq. (1a) and Eq. (1b) can be used
to select a suitable pump impeller based on readily available
manufacturers data. The selected pump impeller was a KSB Ajax I.S.
series impeller with turbine runner outlet diameter, D 226 mm.
No blade geometry modications were performed. The rated values
(BEP) of the pump unit from which the impeller is taken, and the
predicted and actual turbine performance are given later in Table 1
(Section 4).
2.2. Incorporation of inlet ow control to the PAT unit
To provide optimum inlet incidence angle over a range of ow
conditions a guide vane assembly consisting of 13 hydrofoil shaped
vanes, with individual linkages to a common ring, was tted within
 D.R. Giosio et al. / Renewable Energy 78 (2015) 1e6 3

Fig. 2. Schematic of the micro-hydro test facility at the University of Tasmania.

a simple square section spiral case (Fig. 3). Position is controlled by

a programmable 5.3 kN SEW electro-mechanical actuator, however, p1  p2 v21  v22
E gz1  z2 J=kg (2)
this could also be achieved by manual operation of a hand wheel r 2
depending on the requirements and budget of the given
installation. where subscripts 1 and 2 refer to measurements at high and low
pressure side, respectively. The hydraulic power available to the
turbine is therefore a product of the test specic energy and the
2.3. Instrumentation mass ow rate calculated as

For determination of test specic energy Druck UNIK 5000 Ph ErQ W (3)
gauge pressure transducers, 0e100 kPa range, were tted to ring
The power transferred to the turbine shaft is given by
manifolds with individually valved tappings located at spiral inlet,
and draft tube outlet. Each unit has a frequency response of 3.5 kHz
Ps 2pn$TW (4)
and was calibrated over the full range with a maximum uncertainty
of 0.06% full scale. Flow rate measurement is achieved using a where the torque, T, is calculated from the measured shaft strain as
Yokogawa US300PM ultrasonic ow meter tted upstream the
turbine inlet with an accuracy of 0.01 m/s over the sonic path.
pG D4s Fc
An optical sensor and 500 hole disc was mounted on the non- T Nm=v (5)
8D 106
drive end of the generator for test speed measurement. For
steady state tests a rolling average was used, updated every pulse. where G is the shear modulus of the solid stainless steel shaft (of
An independent induction pick-up was mounted on the drive end diameter Ds) and Fc is the strain gauge calibration coefcient.
for grid synchronisation and overspeed protection. Shaft torque Turbine performance is presented in terms of three key
measurement was achieved with a full bridge 350 U strain gauge dimensionless parameters: energy coefcient, EnD, ow coefcient,
installed on an exposed section of turbine shaft with data acqui- QnD, and the power coefcient, PnD, dened as
sition via a KMT digital telemetry system. The encoder scanning
rate was 6.94 kHz with a system accuracy of 0.2%. Turbine unit E Q Pm
efciency is therefore stated with an uncertainty of 1.3% based on EnD ; QnD ; PnD 3 5 (6)
n2 D 2 nD3 rn D
the propagation of errors method [11].
The hill chart of turbine efciency contours shown in Fig. 4 is
given in terms of overall efciency, ho, which is dened as the po-
3. Experimental methodology wer transferred to the turbine shaft relative to the available hy-
draulic power.
During steady state performance testing data was recorded at
each operational point with a sample frequency of 100 Hz, acquired Ps 2pn$T
ho (7)
by an NI M Series Multifunction DAQ SCB-68 card. For the con- Ph E$rQ
struction of the turbine efciency hill chart tests were conducted
As such the overall efciency incorporates mechanical ef-
for guide vane opening angles ranging between 20 and 35 ,
ciency, taking into account mechanical losses in shaft seals and
equivalent to a range of between 53% and 92% full stroke. Tests were
bearings; volumetric efciency, taking into account leakage and
conducted at net specic energy values of 34.3 J/kg up to 83.3 J/kg
cooling water ow; and hydraulic efciency, dening the amount of
in 5 J/kg increments to ensure a thorough exploration of the entire
power extracted by the runner from the power available from the
operating range.
water. Owing to the tight internal tolerances of the casing the
In accordance with IEC Standards [12] the net specic energy
volumetric efciency can be approximated to hv z 1, with only
across the machine was calculated as
small losses due to cooling water ow, so that

Table 1 ho hm hv hh zhm hh (8)

Best efciency point values of the selected PAT operating as pump, turbine predicted
& turbine actual.
Mechanical efciency due to losses in bearings, shaft seals, and
wear rings was estimated by running the turbine up to rated speed
Values at BEP nq [e] H [m] Q [m3/s] N [rpm] P [kW] h [%] and connecting to the grid. Guide vanes were then closed and air
Pump as rated 104.3 12.25 0.222 1450 34.0 78.5 injected into the draft tube in order to ll the runner chamber. With
Turbine predicted 92.4 4.38 0.139 750 4.68 78.5 cooling lines open the torque required to maintain constant speed,
Turbine actual 71.9 5.98 0.133 754 6.20 79.0
Tloss, was recorded.
4 D.R. Giosio et al. / Renewable Energy 78 (2015) 1e6

Fig. 3. Square section spiral case incorporating the guide vane mechanism prior to nal assembly.

agreement with the predicted best efciency of a pump impeller

Pt
hm
2pnT Tloss
The mechanical efciency as determined by Eq. (9), was
assumed constant over the entire operating range, however in
practice this will vary to a relatively small degree.
4. Results
The efciency of the designed micro-hydro pump-turbine unit
over the entire operating range is illustrated in the hill chart shown
in Fig. 4. Overall efciency contours are given, while guide vane
opening angles are represented by the dotted lines running diag-
onally left to right. The best efciency point was found to be at
EnD 7.27, QnD 0.925, giving a maximum overall turbine ef-
ciency of 79.0%, measured at turbine shaft. This result is in good
acting as turbine which is generally reported to be within the range
(9)
of 2% of the best efciency in pump operation [5].
Performance of the micro pump-turbine unit in comparison to
the parent pump and predicted turbine values is given in Table 1.
While the maximum efciency shows a slight increase of 0.5% from
that in pumping mode, the net head and output power at rated
were considerably higher than predicted by the method of Sharma.
Furthermore the ow rate in turbine operation was slightly less
than that expected.
It is widely accepted that PATs require a higher head and ow at
BEP point than in pump mode, with the head ratio generally the
larger of the two. Ventrone et al. [7] suggests that this is due to a
combination of a signicant increase in head coefcient with
increasing ow and the simultaneous decrease in mechanical los-
ses relative to power output as well as a residual negative
tangential velocity at operating condition just greater than design.

Fig. 4. Performance Hill Chart diagram of the micro-hydro turbine unit.

 D.R. Giosio et al. / Renewable Energy 78 (2015) 1e6 5

Fig. 5. Turbine efciency (B) and power coefcient (,) at constant specic energies corresponding to tank min. level, BEP and tank max. level.

This is in contrast with the belief that slip, due to a relative rota-
tional velocity within the blade passages, is only present in pump
mode such that ow should be increased in order to minimise
incidence losses. However, a degree of slip is also present in turbine
operation, and an increase in incidence loss due to higher ow is
balanced in some degree by the mechanical efciency. This
decreasing inuence of mechanical losses can be seen in Fig. 6
whereby the overall efciency approaches the hydraulic efciency
at normalised ow Q/QBEP > 1.
Of practical interest is the general atness of the efciency map,
particularly around the BEP. This is also shown in Fig. 5, presented
for three cases of constant specic energy corresponding to tank
minimum supply level, BEP condition and tank maximum supply
level. The variation in head represents a range of static head be-
tween approximately 0.85HBEP and 1.2HBEP. The efciency in each
case remains considerably high even with signicant decreases in
ow. Indeed, in all cases a 30% ow reduction is required to see a
10% reduction in overall efciency, and a further 20% decrease in
ow for an additional 10% efciency reduction at rated EnD. This
compares favourably with previously presented data on PAT per-
formance by Yang et al. [13], Williams et al. [10], Derakhshan and
Nourbakhsh [14] and Singh [2] that all report signicant reduction
in turbine operation efciency due to the inherent lack of inlet ow
control and associated large incidence losses. In the sphere of
micro-hydro remote area power installations the ability to operate
efciently at off-design conditions is extremely important as water
storage reservoirs are often limited in capacity and catchments are
subject to highly uctuating, often seasonal inows.
4.1. Weighted-average-efciency (WAE)
The weighted average efciency provides developers of micro-
hydro installations a means to quantitatively compare generation
alternatives, taking into consideration the likely percentage of time
the unit will be operating under off-design conditions. As dened
by Eq. (10), the WAE is determined at 1.0Pr, 0.8Pr and 0.6.Pr

Fig. 6. Turbine efciency (B) and power () at constant guide vane opening of 30 . Power and ow rate normalised by values at B.E.P.
6 D.R. Giosio et al. / Renewable Energy 78 (2015) 1e6

acknowledgements to Prakash Gautam and Paul Coull for their

contributions to the turbine design.

Nomenclature

D turbine outlet diameter [m]

Ds shaft diameter [m]
E specic energy [J/kg]
Fc strain gauge coefcient [e]
G shear modulus [Pa]
H head [m]
N rotational speed [rpm]
P power [W]
Q ow rate [m3/s]
Fig. 7. Normalised efciency of the new micro-hydro turbine unit (UTAS-MH) and that
g acceleration due to gravity [m/s2]
of an optimised PAT [2] at three load points 1.0Pr, 0.8Pr, and 0.6Pr.
n rotational speed [s1]
nq specic speed, dimensional [rpm]



p static pressure [Pa]
WAE a1 $ht;1:0 a2 $ht;0:8 a3 $ht;0:6 (10) z elevation [m]
v ow velocity [m/s]
where a1 0.6, a2 0.2, a3 0.2 are weight factors representing h efciency [%]
the expected percentage of operational time at each off-design p mathematical constant [e]
loading. Values given were choosen such that a direct compar- r density [kg/m3]
ision may be made with data presented by Singh [2], although these
may be changed to reect conditions at a given site. Abbreviations
The performance of the designed micro-hydro turbine unit with BEP best efciency point
incorporated inlet ow control is compared to performance results PAT pump-as-turbine
of a similar traditional PAT unit presented by Singh [2]. The tradi-
tional PAT tested by Singh [2] had undergone numerous stages of Subscripts
optimisation including impeller blade tip rounding and tapered h hydraulic
ring inserts, resulting in an increased peak efciency of 83.4%, while m mechanical
the designed micro-hydro turbine runner had not undergone any o overall
modications. Even so, the improved performance at off-design p pump
conditions has essentially made up for a 4.4% disparity in peak ef- r rated
ciency with WAEs of 76.8% and 76.9% for the new micro-hydro s shaft
turbine and the optimised PAT, respectively. This is further illus- t turbine
trated in Fig. 7. v volumetric

5. Conclusions
References

The development of a 6.2 kW micro-hydro turbine unit and test
facility was presented. The turbine unit was designed to incorpo-
rate a commercially available pump impeller operated in turbine
mode, together with custom spiral casing and guide vane mecha-
nism for inlet ow control. Steady state experimental testing over
the full operating range indicated a maximum overall efciency of
79%, in good agreement with PAT theory. The net head and output
power at rated conditions were somewhat higher than predicted by
the method of Sharma [6], while the ow rate at rated was slightly
less than expected. Importantly, in the context of micro-hydro and
energy recovery systems, while slightly more complex than a
traditional PAT system the turbine demonstrated near peak ef-
ciency operation over a wide range of ow conditions, as indicated
by the weighted-average-efciency. The designed micro-hydro
turbine unit thereby addresses the main drawbacks of pumps
operating in turbine mode, providing a low cost alternative
generating solution for application in remote area power supply
and industrial energy recovery systems.
Acknowledgements
This project was funded through ARC Linkage Grant (LP
110200244) and industry partner Hydro Tasmania. Turbine design
in collaboration with Pentair Flow Technologies, Tasmania. Special
[1] Thoma D, Kittredge CP. Centrifugal pumps operated under abnormal condi-
tions. Power 1931;73:881e4.
[2] Singh P. Optimization of the internal hydraulics and of system design for
pumps as turbines with eld implementation and evaluation. Ph.D. thesis.
Germany: University of Karlsruhe; 2005.
[3] Singh P, Nestmann F. Internal hydraulic analysis of impeller rounding in
centrifugal pumps as turbines. Exp Therm Fluid Sci 2011;35(No. 1):121e34.
[4] Derakhshan S, Mohmmadi B, Nourbakhsh A. Efciency improvement of cen-
trifugal reverse pumps. J Fluids Eng 2009;131(No. 2). Article ID 021103.
[5] Williams AA. The turbine performance of centrifugal pumps: a comparison of
prediction methods. Proc Inst Mech Eng 1994;208:59e66.
[6] Sharma KR. Small hydroelectric projects e use of centrifugal pumps as tur-
bines. Bangalore, India: Kirloskar Electric Co; 1985.
[7] Ventrone G, Ardizzon G, Pavesi G. Direct and reverse ow conditions in radial
ow hydraulic turbomachines. Proc Inst Mech Eng Part A: J Power Energy
2000;214(6):635e44.
[8] Alexander K, Giddens E. Microhydro: cost-effective, modular systems for low
heads. Renew Energy 2008;33(6):1379e91.
[9] Alexander K, Giddens E, Fuller A. Radial- and mixed-ow turbines for low
head microhydro systems. Renew Energy 2009;34(7):1885e94.
[10] Williams AA, Smith NPA, Bird C, Howard M. Pumps as turbines and induction
motors as generators for energy recovery in water supply systems. J CIWEM
1998;12:175e8.
[11] Figliola RS, Beasley DE. Theory and design for mechanical measurements. 3rd
ed. John Wiley & Sons, Inc; 2000.
[12] International Standards. IEC 60193 second edition 1999-11.
[13] Yang SS, Derakhshan S, Kong FY. Theoretical, numerical and experimental
prediction of pump as turbine performance. Renew Energy 2012;48:507e13.
[14] Derakhshan S, Nourbakhsh A. Experimental study of characteristic curves of
centrifugal pumps working as turbines in different specic speeds. Exp Therm
Fluid Sci 2008;32:800e7.