Professional Documents
Culture Documents
DOI 10.1007/s10010-006-0045-1
ORIGINALARBEITEN ORIGINALS
A. Nemdili (u)
Faculty of Mechanical Engineering,
Technical University of Oran,
Oran, Algeria
e-mail: alnemdili@yahoo.com
D. H. Hellmann
Department of Mechanical Engineering,
Institute of Fluid Flow and Positive Displacement Machinery,
University of Kaiserslautern,
Kaiserslautern, Germany
D. H. Hellmann
e-mail: hellmann@mv.uni-kl.de
Experimentelle Untersuchungen
der Leistungsermittlung durch Scheibenreibung
von Scheiben mit und ohne Umfangsausschnitten
in realen Kreiselpumpegehausen
Zusammenfassung Um den Zusammenhang zwischen der
Hohe der Scheibenreibungsverluste und der Geometrie der
Scheiben in realen Kreiselpumpengehause zu u berprufen,
werden Scheiben mit und ohne Umfangsausschnitten mit
verschiedener Zahl, Winkel und Breite nachgeforscht.
Scheiben mit Umfangsausschnitten wurden u berpruft, um
sich einem realen Laufrad im realen Kreiselpumpengehause
zu nahern. Die Austrittskanale dieses Laufrades werden
heraus gebildet, der Rest der Kanale wird durch festes Material ersetzt. Experimentelle Ergebnisse der Scheiben ohne
Umfangsausschnitten zeigen, dass der Scheibenreibungskoeffizient sinkt mit zunehmender Reynolds Zahl, aber wird
kleiner mit kleinen Werten der Scheibenbreite und hohen
Werten des Spaltes. Der Einsatz von Spiralen hat nur einen
geringen Einfluss auf den Scheibenreibungskoeffizient. Experimente auf den Scheibenreibungsverlusten, die mit und
ohne Radseitenraumabdeckung durchgefuhrt wurden, zeigen, dass die Radseitenraumabdeckung mehr Einfluss fur
hohe Werte des axialen Spaltes hat. Experimentelle Ergebnisse der Scheiben mit Umfangsausschnitten zeigen, dass
der Scheibenreibungskoeffizient sich nicht mit der Reynolds
Zahl a ndert, aber im Allgemeinen haben die Zahl und der
Winkel den Umfangsausschnitten der Scheiben einen wichtigen Einfluss auf den Scheibenreibungskoeffizienten.
List of symbols
b
D
g
width (m)
diameter of disk (m)
gravity (m/s2 )
13
60
k
ks
M
R
Re
s
t
n
1 Introduction
Friction losses, which occur during the rotation of disks and
cylinders in closed casings, filled with fluid play in many
cases a decisive role. Their possible exact knowledge constitutes an important step in the calculation of pumps.
Disk friction losses caused by the fluid flow in the clearance between the impeller and the housing are expressed as
a power loss.
Disk friction losses in impeller side rooms of turbomachines have been treated in numerous publications. A number of empirical equations to predict the disk friction losses
were suggested by many researchers, but generally there is
not an exact mathematical model to predict these losses.
Most of the published work on disk friction is concerned
with plain thin disks rotating in close-fitting cylindrical casings and has been generally presented in terms of the nondimensional torque coefficient C M .
The purpose of the investigations reported here is to examine disks with different axial clearances between the
disk and the end walls of real centrifugal pump casings of
different width. Experiments in water are conducted with
modified geometry of disks at outlet by variation of the
Fig. 1 Notation for the
description of disk friction loss
13
dFz
= u 2 = r 2 2 .
dA
2
2
(1)
61
2 2
r dA = r 2 2 2 r dr .
2
2
(2)
If this force is multiplied by the lever arm r and the angular speed of the fluid, then the energy dissipation for the
friction surface becomes after integration
D/2
PD f =
D/2
r dFz =
d/2
(3)
MD f = 2
2 r 2 dr .
(6)
d/2
PD f = u 3 D2 .
R
0
r 4 3 dr
3 D5 .
160
(5)
Pantell [1] in 1949 has stated that from the braking action
of the rotating parts around the flowing medium in turbomachines high losses result. Thus, kinetic energy is dissipated
in heat energy, whereby an energy loss for the process results with a negative influence on the hydraulic efficiency.
Therefore, it is essential to deal with this phenomenon of
disk friction for the optimization of operation conditions of
centrifugal pumps.
For this numerous flow theoretical investigations were
accomplished, which revealed however only fundamental
influencing parameters, but not the exact analysis of the disk
friction phenomenon including the arising losses, due to the
complex three-dimensional flow process.
Daily and Nece [2] conducted their experiments on test
stand with a DC motor of approx. 22 kW and a speed range
from 125 to 2000 RPM. Water was used at first as test fluid.
This was brought in the center, i.e. at the drive shaft of the
test stand, and in other case at the housing outlet diameter.
The radial clearance of disks, the Reynolds number as well
as the flow rate for the centric and/or the peripheral flowing
were constant for each test series. For the tested geometry of
the disk, the flow form of the induced flow between disk and
housing depends only on the Reynolds number and the axial
clearance.
In principle four flow forms are possible for a constant
axial clearance with variable Reynolds number. These will
be classified according to the flow between rotating disk
and housing, and to the geometrical expansion of the flow
boundary layers regarding to the axial gap. The four flow
modes are complete detailed in [2] for different Reynolds
MD f =
1
C M 2 R5 .
2
(7)
uR
R2
=
.
(9)
0.25 2 .
3.8 log10 kRS 2.4 Rs
(11)
150
.
R
(12)
The parameter k S is described as effective hydraulic roughness and must be interpreted as roughness characteristic
value of the flow-guiding surfaces of a centrifugal pump.
Geis [4] accomplished investigations on hydraulic pumps,
where the efficiency is affected by a multiplicity of parameters, for example the Reynolds number, the design mode of
the disk and/or the impeller, the fluid flow in the gap and the
size of the axial gap between rotating disk and housing.
13
62
13
with test data. The Author demonstrated that the main uncertainties in predicting disk friction losses are related to the
roughness and the inlet swirl.
3 Experimental procedure
The experimental installation for the identification of the
power demand of disk friction in water tests is shown in
Fig. 2. The torque is measured by means of a torque meter
mounted on (8) and the rotational speed by means of a stroboscope lamp in (9). The mechanical seal is cooled with
water, which is regulated by the valve (11). The ball valve
(14) is used to air escape of the trial equipment. The pump
(10) is filled with water by means of the shut-off slide (12).
The inlet and the outlet of the pump are direct connected
with socket (15 and 18), which at the same time support
the resistance thermometers (16 and 17). The temperature
difference between (16) and (17) is measured by means of
a precision temperature measurement device (21). The heat
63
Table 1 Geometrical
parameters of tested disks
Width
Axial gap Roughness
b2 (mm) s (mm) k S (m)
18
13.5
5.45
23
11
6.46
29
8
5.26
36
4.5
8.62
(14)
13
64
PS = M S
2n
60
(15)
PDF
2n 3
60
R5
(16)
Cn t n ;
(t) = e
110
110+t 3,152
106
(17)
n=0
with:
C0 = 999.84 ;
C2 = 9.106 10
C1 = 6.7983 2 ;
3
C4 = 1.12676 106 ;
C3 = 1.005273 104 ;
C5 = 6.5918 109 .
It is to note that the temperature is measured when a stationary temperature level for each rotational speed and each
disk width appears. It takes approximately 5 to 10 minutes.
A main value of the temperature t between the inlet and the
outlet of the pump is used.
Non-dimensional torque coefficients C M for disks without modified outlet sections and C MU A for disks with modified outlet sections are evaluated using Eq. 16 and drawn
versus the Reynolds number Re.
4 Experimental results
4.1 Analysis of results obtained for disks
without modified outlet sections
Influence of the axial clearance on C M . Figures 5 and 6
show the experimental results obtained for various clearances and the volute with width of 54 mm.
It can be seen from the curves C M = f(Re) that:
For each volute, the curve corresponding to the high
value of the axial clearance s = 13.5 mm (disk width of
b2 = 18 mm) lies below others curves (Fig. 5).
The disk friction coefficient C M declines approximately
steady in the same form with increased values of axial
clearance s and increased values of Reynolds-number Re
(Fig. 6).
By increasing of the s/R ratio the disk friction coefficient
C M decreases.
13
Fig. 7 Influence of the width of the volute on the disk friction coefficient
65
13
66
The comparison of the results shows that the best approximate equation, which can predict with minimum error
the corrected experimental results by taking into account
the factor indicated in Eq. 5 and Fig. 6 above, seems to
be Eq. 13 given by Poullikkas [5]. However certain deviations are found, because the experiments are done using
a real centrifugal pump casing.
The correlation 8 given by Daily and Nece [2] appears to
be not accurate, because there was an error in the original
work about the factor 0.0102, which must be equal to 0.102
according to Poullikkas [5].
13
5 Conclusions
Friction losses, which occur during the rotation of disks and
cylinders in closed casings filled with fluid, play in many
cases a decisive role. Their possible exact knowledge constitutes an important step in the calculation of pumps.
Disk friction losses caused by the fluid flow in the clearance between the impeller and the housing are expressed as
a power loss. This power absorbed by disks rotating in a volute pump casing is measured.
Investigations with disks with and without modified outlet sections and different geometrical parameters in real centrifugal pump housing of different width are conducted.
The obtained results of disks without modified outlet sections show that:
The disk friction coefficient C M declines approximately
steady in the same form with increased values of axial
clearance s and increased values of Reynolds-number
Re.
By increasing the s/R ratio the disk friction coefficient
C M decreases.
67
References
1. Pantell K (1949/1950) Versuche u ber Scheibenreibung. Forsch
Geb Ingenieurwes 16(4):97108
2. Daily JW, Nece RE (1960) Chamber Dimension Effects on Induced Flow and Frictional Resistance of Enclosed Rotating Discs.
ASME J Basic Eng 82(1):217232
3. Nixon RA, Cairney WD (1972) Scale effects in centrifugal cooling water pumps for thermal power stations. NEL Report 505,
April 1972
4. Geis H (1988) Pumpen optimieren. Berechnen und minimieren
der Radseitenverluste von hydraulischen Machinen. Maschinenmarkt / Forsch Konstruktion 54(20):7882
5. Poullikkas A (1995) Surface Roughness Effects on Induced Flow
and Frictional Resistance of Enclosed Rotating Discs. Trans
ASME J Fluids Eng 117:526528
6. Gulich JF (2003) Disk friction losses of closed turbomachine
impellers. Forsch Ingenieurwes 68:8795, DOI: 10.1007/s10010003-0111-x
7. Lunzmann H (1995) Einflu des Spaltstroms bei Spiralgehausepumpen mit glattem Kegelspalt und mit Bauchschaufeln. In: Mitteilungen des Pfeilderer-Instituts fur Stromungsmaschinen, Verlag
und Bildarchiv W. H. Faragallah, Heft 2, Januar 1995
8. Nemdili A (2000) Einzelverluste von Kreiselpumpen mit spezifischen Drehzahlen n q = 1535 min1 . SAM Forschungsberichte,
Band 1, Dissertation, Marz 2000, Verlag Universitat Kaiserslautern, ISBN 3-925178-45-7, ISSN 16156587
9. Nemdili A, Hellmann DH (2004) Experimental study of the influence of geometrical parameters on disc friction loss of a centrifugal pump. In: Proceedings of 22nd IAHR Symposium on
Hydraulic Machinery and Systems, June 29July 2 2004, Stockholm, Sweden, Volume B, B10-2.doc, 1(10)10(10)
10. Nemdili A, Hellmann DH (2004) Development of an empirical
equation to predict the disc friction loss of a centrifugal pump.
Scientific Bulletin of the Politehnica University of Timisoara,
Trans Mech 49(63):235240, Special issue, ISSN 1224-6077
11. Nemdili A (2005) Elaboration of calculation procedure of stresses
acting on turbine impeller disc due to its non uniform heating.
12`emes Journees Internationales de Thermique, Tanger, Maroc du
15 au 17 Novembre 2005, pp 311314
13