Experimental Thermal and Fluid Science 32 (2008) 1620–1627

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science

journal homepage: www.elsevier.com/locate/etfs

Theoretical, numerical and experimental investigation of centrifugal pumps in

reverse operation
Shahram Derakhshan *, Ahmad Nourbakhsh
Department of Mechanical Engineering, Faculty of Engineering, University of Tehran, Tehran, P.O. Box 11365/4563, Iran

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

Article history: When a pump works as a turbine, its hydraulic behavior will be changed. Several methods have been
Received 1 December 2007 developed to predict the best efficiency of pumps running as turbines but their results are not in good
Received in revised form 13 May 2008 coincidence with experimental data for all pumps. Therefore, study and investigation of hydraulic behav-
Accepted 14 May 2008
ior of pumps in reverse operation can be useful. In this study, the best efficiency point of an industrial
centrifugal pump running as turbine was achieved using a theoretical analysis. This method tries to esti-
mate hydraulic components of reverse (turbine) mode using direct (pump) mode. In the next step, the
Keywords:
pump was simulated in direct and reverse modes by computational fluid dynamics. 3D full Navier–Stokes
Best efficiency point
Characteristic curves
equations were solved using FineTurbo V.7 flow solver. Using numerical results, complete characteristic
Computational fluid dynamics curves of the pump in direct and reverse modes were obtained. For experimental verification of theoret-
Experimental ical and numerical results, the pump was tested as a turbine in a test rig. All required parameters were
Pump as turbine measured to achieve complete characteristic curves of the reverse pump. The theoretical and numerical
results were compared with experimental data and some other methods.
Ó 2008 Elsevier Inc. All rights reserved.

1. Introduction Most recent attempts to predict performance of PAT, have been

made using computational fluid dynamics (CFD) [11,12]. However,
One of the cheap and attractive alternatives in small water- without verifying the CFD results by experimental data, they are
power resources is using pumps in reverse operation. Pumps are not reliable.
relatively simple machines, easy to maintain and readily available In this paper, the best efficiency point of an industrial centrifu-
in most developing countries. The research on using pumps-as- gal pump running as turbine was achieved using theoretical anal-
turbines was started around 1930. With increasing energy de- ysis based on ‘‘area ratio” method developed by Williams [5] and
mands, it will be more economical to exploit such energy sources Anderson [13]. This method estimates turbine mode hydraulic
in the future. Experiments have shown that in relatively low power components based on pump characteristics. Also, this pump was
outputs, pumps with high technological standards in reverse simulated in reverse and direct modes using computational fluid
operation can compete with conventional turbines in respect to dynamics. 3D full Navier–Stokes equations were solved using Fine-
maximum efficiency [1]. Pump manufacturers do not normally Turbo V.7 flow solver. Using numerical results, complete character-
provide characteristic curves of their pumps working as turbines. istic curves of the pump in direct and reverse modes were
This makes it difficult to select an appropriate pump to run as a obtained.
turbine for a specific operating condition. A complete mini hydropower test rig established in laboratory
Since behavior of a pump running as a turbine changes, predic- [9] was used for experimental verification of theoretical and
tion of characteristic curves of pump as turbine (PAT) is difficult. numerical results. The simulated centrifugal pump was tested as
Numerous papers have been published on this subject, but there a turbine in this test rig. All required parameters were measured
has been comparably limited use of this knowledge. Stepanoff to achieve complete characteristic curves of the reverse pump.
[2], Sharma [3], Wong [4], Gantar [5], Williams [6], Alatorre-Frenk The theoretical and numerical results were compared with exper-
[7], Ramos and Borga [8], Derakhshan and Nourbakhsh. [9], have imental data and some other methods.
presented some relations to predicate the best efficiency point
(BEP) of a PAT. The results obtained by these relations had almost
±20% deviation from experimental data [10]. 2. Theoretical analysis

When a centrifugal pump operates as a turbine, flow and rotat-

* Corresponding author. Tel.: +98 21 82 08 48 31; fax: +98 21 88 33 86 48. ing directions are reverse. There is one passage with two opposite
E-mail address: [email protected] (S. Derakhshan). flows and rotating directions. Using area ratio method [5,13], a

0894-1777/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.expthermflusci.2008.05.004
S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627 1621

Nomenclature

a area, m2 g efficiency
b impeller passage width p power number
c absolute velocity, m/s q density, kg/m3
D diameter, m x rotational speed, rad/s
e impeller blade thickness, m w head number
g gravitational acceleration, m/s2
H head, m Superscript
m diameter ratio, DD21 ‘‘ theoretical
N rotational speed, rpm
n rotational speed, rps Subscripts
Ns specific speed, (m, m3/s) b best efficiency point
P power, W, kW e blade exit
Q discharge, m3/s h hydraulic
u peripheral velocity, m/s i impeller
w relative velocity, m/s l leakage
z impeller’s blade number m meridian
me mechanical
Greek symbols n net
a absolute velocity angle (rad) u peripheral
b relative velocity angle (rad) p pump
b0 blade angle (rad) v volute
e utilization factor t turbine
u discharge number
l slip factor

correlation passage between direct and reverse modes of a centrif- Leakage value for pump operation is estimated using Thorne meth-
ugal pump may be found. To calculate head, discharge and effi- od [14]. Volumetric efficiency is
ciency of PAT at BEP, hydraulic losses in volute and impeller,
Q lp
mechanical losses related to power losses due to gland packing glp ¼ 1 ð5Þ
Q p þ Q lp
and bearing cases, disc friction losses in gaps between rotor and
stator, volumetric losses related to leakage from clearances be- The important parameter is the overall mechanical efficiency
tween rotor and stator shall be detected. including power losses due to disc friction, gland packing and bear-
Fig. 1 shows inlet and outlet velocity triangles in pump and tur- ing losses. Overall mechanical efficiency is determined using:
bine modes. In Fig. 1.a, the gray velocity triangle is related to slip gp
phenomenon (l is slip factor) on pump mode, since in impeller gmep ¼ :
outlet, real relative flow angle (b2) is smaller than blade angle
glp ghp
(b02 ). The absolute flow angle in position 2, a2, is approximately Hydraulic efficiencies in impeller and volute can be defined as
the same for both modes of operation in respect to the shape of
the volute. A good estimation of this angle is the volute angle, av. H0 H
gip ¼ ; gvp ¼ ð6Þ
In reverse mode, the flow angle in blade outlet (b3) can be assumed H00 H0
close to the blade inlet angle (b01 ) (no whirl at exit). Using momen- where H00 ; H0 , H are Euler, impeller and real heads, respectively. It
tum theorem and turbine inlet and outlet velocity triangles, Euler can be assumed friction losses in impeller and volute are the same,
head in turbine mode is [2] besides, the shock losses in impeller are approximately close to the
u2 cu2 u3 cu3 diffusion losses in volute [5]. So, total losses in volute and impeller
H00t ¼ ð1Þ are equal: H00 H0 ¼ H0 H.
g
Impeller and volute efficiencies are computed as follows:
Using velocity triangles Eq. (1) can be changed to
H0 H H
u3 Q 00t m cot av cot b01 u2 ¼ ) ghp ¼ gip gvp
H00t ¼ þ 3 ð2Þ H00 H0 H00
g a2 a1 g
8 H00 H0
Shockless discharge ðQ 00t Þ is obtained as < 1 gip ¼ H00
>
0
1 gvp ¼ HHH0 ) ð1 gip Þ ¼ ð1 gvp Þgip
u2 a2 >
: 00
Q 00t ¼ C m a2 ¼ ð3Þ 0
H H ¼H H 0
cot av þ cot b02
To avoid shock losses, the inlet relative velocity angle (b2) should be Then
as much as possible close to the blade inlet angle. 2ghp 1 þ ghp
Flow reaches the impeller after hydraulic losses in volute. These gvp ¼ ; gip ¼ ð7Þ
1 þ ghp 2
losses must be added to Euler head. Since computing of hydraulic
losses is difficult, these losses were estimated using pump mode. In general, contraction losses are less than expansion losses (in
The hydraulic efficiency of a pump at BEP can be computed using same shapes) [5]. Therefore, volute hydraulic losses in reverse mode
its overall efficiency [2]: are less than direct mode (in same conditions). In other words:
pffiffiffiffiffiffi
ghp gp ð4Þ 1 gvt ¼ kð1 gvp Þ; k<1 ð8Þ
1622 S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627

and pump modes are the same. Power losses due to leakage are
a (1-μ)u2 computed as
Plt ¼ c Q lt ðHt gvt Þ ð12Þ
w2
c m2 c2 Volute power losses of turbine mode are
β2
u2 c Pvt ¼ ð1 gvt Þ c Q t Ht ð13Þ
Using Eqs. (12) and (13), kinetic energy losses in turbine impeller
outlet is calculated as
Pet ¼ ð1 eÞ ðc Q t Ht Pvt P lt Þ ð14Þ
In same procedure of volute (Eq. (8)), hydraulic efficiencies of
impeller in turbine and pump modes are
1 git ¼ 0:9ð1 gip Þ ð15Þ

Hydraulic losses of impeller in turbine mode are evaluated as

ω Pit ¼ ð1 git Þ ðc Q t Ht P vt Plt Pet Þ ð16Þ
Turbine net power is

b c2
Pnt ¼ c Q t Ht Pvt Plt Pet Pit Pmt ð17Þ
w2
Turbine maximum efficiency is
c m2
β2 αv Pnt c Q t Ht Pvt Plt Pet Pit Pmt
u2 c u2 gt ¼ ¼ ð18Þ
c Q t Ht c Q t Ht
Using simple assumptions, the BEP of PAT was achieved.

3. Numerical simulation

A 3D model of the pump was generated where the model in-

cludes the whole impeller and the volute since a single blade chan-
w3 nel model of the impeller alone is not enough to study the
c m3 circumferential variation of the flow caused by volute (Fig. 2).
c3
β1 The model does not include the flow field in the space between
u3 c u3 impeller hub/shroud and casing nor the sealing gap. The grid size
was approximately 700,000 cells. The FineTurbo V.7 flow solver
ω developed by Numeca was used and grid was generated using
Fig. 1. (a) Pump impeller outlet velocity triangle, (b) turbine impeller inlet and IGG5.5 for the volute and Autogrid5 for the impeller.
outlet velocity triangles. FineTurbo is an integrated software based on finite volume dis-
cretization for multi-block structured grids. The multi-block struc-
tured grids on the blades were prepared by IGG5.5 and AutoGrid5
[15,16]. The physical model used in the solver is the Reynolds-
There are no empirical equations for the contraction losses and k
subsequently. However, Williams [5] assumed: k = 0.9.
In turbine impeller outlet, a little no-recovered energy exists in
form of kinetic energy. The utilization factor is defined as
utilized energy by impeller
e¼ ;
available energy for impeller
using Fig. 1.a, the utilization factor is:
H00t Q 00t
e¼ ; cu3 ¼ u3 ð9Þ
H00t þ
c2u3 a1 tan b01
2g

Real head at BEP of turbine operation is

Ht ¼ e gvt H00t ð10Þ
Turbine discharge at BEP can be evaluated as shockless discharge
and leakage value. Using pump leakage value, turbine leakage value
can be estimated by
sffiffiffiffiffiffi
Ht
Q lt ¼ Q lp ð11Þ
Hp

Discharge at BEP of turbine mode is Q t ¼ Q 00t þ Q lt .

To calculate maximum efficiency of PAT, total losses shall be
evaluated. In same rotational speeds, mechanical losses in turbine Fig. 2. 3D model of simulated centrifugal pump.
S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627 1623

Averaged Navier–Stokes equations in rotating frames of reference as turbine, a control system is needed to regulate the frequency.
coupled with various turbulence models and near-wall treatment The classical governor used for standard turbines are expensive
for low-Reynolds modeling. Without any restriction, the standard and not always recommended for small hydropower plants. Since
high Reynolds k–e turbulence model can be chosen with extended these types of plants are more used in remote areas, an electronic
wall functions [16–18]. The discrete schemes are physically based load controller with ballast loads was built and used for keeping
second order [18]. For fast convergence, mass flow rate, velocity the frequency of generator in these tests. A conventional synchro-
direction, turbulence kinetic energy k and turbulent dissipation e nous generator was installed to produce electricity. To measure the
were imposed at inlet boundary while at outlet boundary condi- turbine shaft torque, generator is changed to suspense state and
tion, static pressure was prescribed. Finally, periodic boundary using a scaled arm and several weights, turbine shaft torque can
condition was applied between two blades. be measured. The discharge was measured by the discharge law
To check if the grid used in the whole pump model is too coarse, and using various orifice plates for each test. Pressures were mea-
simulations were made with one impeller channel and two differ- sured by some barometers ranging from 0 to 5 bar. An industrial
ent grids. The first grid had the same grid coarseness with 70,000 low specific speed centrifugal pump with specific speed 23.5 (m,
cells for one impeller channel and the second one consisted of m3/s) was selected for testing as turbine. This pump has the max-
about 200,000 cells. The simulation results showed less than 1% imum input turbine shaft power, the maximum head and the max-
difference in calculated efficiencies and heads. imum discharge of 20 kW, 25 m and 120 l/s, respectively. For the
PAT testing, feed pump, several pipes, an orifice, a generator and
4. Experimental setup ballast loads were selected and installed in the test rig. In the
PAT applications, it should be considered: that if a generator is cou-
A complete laboratory model of mini hydropower plant was in- pled directly, a nominal speed corresponding to one of synchro-
stalled in University of Tehran [9] as shown in Fig. 3. The discharge nous speeds (e.g. 750, 1000, 1500 or 3000 rpm) should be
and head for pump working as turbine were generated in the chosen. For induction generators and also induction motors, slip
experimental setup by each tested pump. When a pump is working must be taken into account (the tested pump rotates with

Flow
Valve
Meter

Barometer
Torque
Controller
Meter

Consumer Generator Turbine Vessel Pump Motor

Fig. 3. The mini hydropower established in University of Tehran.

10 100
ψ
η
8 Exp 80

CFD

6 60
η (%)
ψ

4 40

2 20

0 0
0 0.02 0.04 0.06 0.08 0.1 0.12

φ
Fig. 4. Measured and numerical head number and efficiency curves of pump mode.
1624 S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627

1450 rpm in pump mode). Practically, synchronous generators are with experimental data, leakage value estimated using Thorne
usually used. The PAT was tested in Nt = 1500 rpm. method [14] was reduced from CFD discharges in pump mode. Also
After measuring all parameters, PAT head, discharge, output mechanical and disc friction losses evaluated using Stepanoff
power and efficiency were obtained. A first-order uncertainty anal- method [2] were added to the pump input power obtained by
ysis is performed using constant odds combination method, based CFD. The results resemble the experimental data as Figs. 4 and 5.
on a 95% confidence level as described by Moffat [19]. The uncer- Table 2 shows the BEP of the pump predicted by CFD and experi-
tainty of head, flow rate, power and efficiency are, ±5.5%, ±3.4%, ment. CFD predicted 4.1%, 0.3% and 4.2% and 1.3% higher discharge
±5.1% and ±5.5%, respectively. number, head number, power number and efficiency than experi-
mental data at BEP, respectively.
5. Results CFD results for turbine mode were not in good coincidence with
experimental data. In turbine mode, the measured discharge is
Table 1 shows BEP of the pump predicted by theory and more than real discharge passing from impeller. Using Eq. (11), tur-
experiment. bine leakage value was added to CFD discharges of turbine mode.
Here, w, u, p are defined as Mechanical and disc friction losses in reverse mode are close to di-
rect mode in same rotational speeds [5]. These losses were reduced
gH Q P
w¼ ; /¼ ; p¼ ð19Þ from turbine output power computed by CFD. Figs. 6 and 7 show
n2 D2 nD3 qn3 D5 results for PAT.
Theoretical method predicted 1.1%, 4.7%, 5.25% and 2.1% lower dis- Table 3 compares the values of BEPCFD and BEPExp of turbine
charge number, head number, power number and efficiency than mode. The deviations of CFD results from experimental data at
experimental data at BEP, respectively. The head and power pre- BEP are 1.1%, 22.9%, 16.4% and +5.5% for discharge number,
dicted by this method are lower than experimental data due to head number, power number and efficiency, respectively. These
the volute and the impeller losses evaluation. deviations maybe in respect to the loss mechanisms in various
The characteristic curves of head, discharge, power and effi- zones of the PAT control volume and rotational momentum theory
ciency in pump and turbine modes achieved by experimental and across the impeller.
CFD methods were shown in Figs. 4–7. CFD could not simulate Deviations of the CFD results for the PAT also suggest a lower
leakage, disc friction and mechanical losses since numerical model determination of the net rotational momentum compared to mis-
did not include flow field in the space between impeller hub/ calculation of hydraulic losses. However, the miscalculation of
shroud and casing nor the sealing gap. To compare CFD results losses cannot be ruled out. CFD results of pump mode resemble
experimental data. In turbine mode, numerical results and experi-
mental data are not in good coincidence. These deviations are con-
Table 1 firmed by other published numerical results [12]. Difference
PAT dimensionless BEP predicted by theoretical method and experiment between real geometry and CFD model is only in interaction be-
Theoretical Experimental. Theoretical error (%) tween volute and impeller according to Fig. 8, since the model does
/ 0.089 0.09 1.1
not include flow zone in the space between impeller hub/shroud
w 9.15 9.6 4.7 and casing. There are some miscalculation losses by CFD where
p 0.578 0.61 5.25 head and output power show lower rotational momentum for all
g 0.71 0.725 2.1 regions of discharges. Maybe there are two phenomena in Fig. 8,

1 100
π
η
0.8 Exp 80

CFD

0.6 60
η (%)
π

0.4 40

0.2 20

0 0
0 0.02 0.04 0.06 0.08 0.1 0.12

φ
Fig. 5. Measured and numerical power number and efficiency curves of pump mode.
S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627 1625

30 100
28
ψ
26
η
24 80
22 Exp
20 CFD
18 60

η (%)
16
ψ

14
12 40
10
8
6 20
4
2
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

φ
Fig. 6. Measured and numerical head number and efficiency curves of turbine mode.

2.4 100
2.2 π
2 η
80
1.8 Exp
1.6 CFD
1.4 60

η (%)
1.2
π

1 40
0.8

0.6
20
0.4
0.2
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

φ
Fig. 7. Measured and numerical power number and efficiency curves of turbine mode.

Table 2 Table 3
Pump dimensionless BEP predicted by CFD and experiment PAT dimensionless BEP predicted by CFD and experiment

CFD Experimental CFD error (%) CFD Experimental CFD error (%)
/ 0.067 0.064 +4.1 / 0.089 0.09 1.1
w 5.66 5.64 +0.3 w 7.4 9.6 22.9
p 0.49 0.47 +4.2 p 0.51 0.61 16.4
g 0.78 0.77 +1.3 g 0.765 0.725 +5.5

changing the flow angle in impeller inlet (zone A) and additional

In pump mode, flow enters from impeller outlet zone into great-
losses related to zone B.
er zone as volute inlet. Therefore there are no miscalculated losses
1626 S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627

Table 5 compares the results of theoretical and numerical

methods with experimental data and some other methods. Numer-
ical and theoretical methods estimated the discharge adequately.
Predicting the head by theoretical methods is nearly close to
experimental data. Numerical method cannot estimate the
head accurately. The validity of any prediction method is
based on accurate predictions of both head and the discharge
accurately.
B B Stepanoff [2], Sharma [3] and Alatorre-Frenk [7] did not show
A A acceptable results in comparing with the experimental data for
both of the head and the discharge.

6. Conclusions

In this paper, a theoretical method was presented to calculate

best efficiency point of reverse pump based on pumping mode
characteristics (geometric and hydraulic). This method’s predicted
values are slightly lower than experimental data. In fact, the losses
in volute and impeller must be evaluated exactly. In the next step,
a centrifugal pump (NS = 23.5 (m, m3/s)) was simulated using CFD
a b in direct and reverse modes. To verify numerical results, simulated
pump was tested as a turbine in the test rig. CFD results were in
Fig. 8. Volute and impeller interaction: (a) real- and (b) CFD-model. good coincidence with experimental data for pump mode not only
at best efficiency point but also in part-load and over-load zones.
But numerical results were not in acceptable coincidence with
Table 4
experimental data for turbine mode. Turbine head and power val-
PAT dimensionless part-load point at same discharge predicted by CFD and
experiment ues achieved by CFD were lower than experimental data at same
discharges. The only difference between direct and reverse modes
CFD Experimental CFD error (%)
was in flow and in rotating directions. The difference between real
/ 0.06 0.06 – geometry and CFD model was in interaction between volute and
w 4.6 6.0 23.3
impeller since flow zone in the space between impeller hub/shroud
p 0.18 0.2 10
g 0.633 0.55 +15.1 and casing was not included in the model. The effect of geometric
simplification is likely to be greater in turbine mode, because its ef-
fect on downstream flow is more than upstream flow, i.e. flow
through the impeller.
Characteristics of best efficiency point predicted by numerical
methods were not in good coincidence with experimental data. Fu-
w2real c 2CFD ture works on CFD application can be improved, since the applica-
c m2 tion of CFD fails in the turbine boundary as proved with this study.
c 2real
w2CFD αCFD References
αreal

u2 c u2CFD [1] J.M. Chapallaz, P. Eichenberger, G. Fischer, Manual on Pumps used as Turbines,
Vieweg, Braunschweig, 1992.
c u2real [2] A.J. Stepanoff, Centrifugal and Axial Flow Pumps, Design and Applications, John
Wiley and Sons, Inc., New York, 1957.
[3] K. Sharma, Small hydroelectric project-use of centrifugal pumps as turbines,
Fig. 9. Turbine mode inlet velocity triangles in real (derived from theory and
Technical , Report, Kirloskan Electric Co., Bangalore, India, 1985..
experimental head and flow) and CFD.
[4] W. Wong, Application of centrifugal pumps for powers generation, World
Pump (1987) , pp. 348–381.
[5] M. Gantar, Propeller pumps running as turbines, in: Conference on Hydraulic
Machinery, September 13–15, 1988, Ljubljana, Slovenia, pp. 237–248.
in CFD simulation. On the other hand, the effect of geometric sim- [6] A. Williams, Pumps as turbines used with induction generations of stand-alone
plification is likely to be greater in turbine mode, because its effect micro-hydroelectric power plants, Ph.D.M.E. Thesis, Nottingham Polytechnic,
1992.
on downstream flow is more than upstream flow, i.e. flow through
[7] C. Alatorre-Frenk, Cost minimization in microhydro systems using pumps-as-
the impeller. Table 4 shows the deviations of numerical results turbines, Ph.D.M.E. Thesis, University of Warwick, 1994.
from experimental data at a part-load point. In simulation, flow en- [8] H. Ramos, A. Borga, Pump as turbine: an unconventional solution to energy
production, Urban Water 1 (1999) 261–263.
ters to the impeller by greater absolute flow angle especially at BEP
[9] S. Derakhshan, A. Nourbakhsh, Experimental study of characteristic curves of
(Fig. 9). Losses of zone B are miscalculated by CFD in all discharge centrifugal pumps working as turbines in different specific speeds,
regions. Experimental Thermal and Fluid Science 32 (2008) 800–807.

Table 5
PATs dimensionless BEP predicted by various methods

Pump Experimental CFD Theoretical Stepanoff [2] Sharma [3] Alatorre- Frenk [7]
Ns (m, m3/s) wt/wp /t//p wt/wp /t//p wt/wp /t//p wt/wp /t//p wt/wp /t//p wt/wp /t//p
23.5 1.7 1.41 1.31 1.40 1.62 1.39 1.30 1.14 1.37 1.23 1.65 1.63
S. Derakhshan, A. Nourbakhsh / Experimental Thermal and Fluid Science 32 (2008) 1620–1627 1627

[10] A. Williams, Pumps as Turbines Users Guide, International Technology [14] E.W. Thorne, Design by the area ratio method, in: Sixth Technical Conference
Publications, London, 1995. of the BPMA, 1979.
[11] A. Tomm, A. Braten, B. Stoffel, G. Ludwig, Analysis of a standard pump in [15] AutoGrid v.5 User Guide, Numeca, 2006.
reverse operation using CFD, in: 20th IAHR Symposium, 2000, Charlotte North [16] N. Hakimi, Preconditioning methods for time dependent Navier–Stokes
Carolina, USA, Paper No. 5. equations, PhD thesis, Vrije Universiteit Brussel, 1997.
[12] A. Rodrigues, P. Singh, A. Williams, F. Nestmann, E. Lai, Hydraulic analysis of a [17] B. Launder, D. Spalding, The numerical computation of turbulent flow,
pump as a turbine with CFD and experimental data, in: ImechE, Seminar Proc. Computer Methods in Applied Mechanics and Engineering 3 (1974) 269–289.
of Advances of CFD in Fluid Machinery Design, 2003, London, England. [18] FineTurbo v.7 User Guide, Numeca, 2006.
[13] H.H.Anderson, Anderson, Modern developments in the use of large single [19] R.J. Moffat, Contributions to the theory of single-sample uncertainty analysis,
entry centrifugal pumps, Proc. Inst. Mech. Eng. 177 (31) (1993). ASME J. Fluids Eng. 104 (1982) 250–260.