15th International Conference on Experimental Mechanics
PAPER REF: 2977
AXIAL THRUST IN CENTRIFUGAL PUMPS - EXPERIMENTAL
ANALYSIS
Vasant Godbole1(*), Rajashri Patil1, S.S. Gavade2
1
Kirloskar Brothers Ltd., India
2
RIT, Rajaramnagar, India
(*)
Email: vasant.godbole@kbl.co.in
ABSTRACT
Reliability of rotating machineries, especially centrifugal pumps is defined on account of wear
rate of components and bearing durability. One of the important performance parameter for
any centrifugal pump is its bearing life, which is dependent on the hydraulic thrusts, Radial[7,9]
and Axial Thrusts[7,9]. The radial thrust prediction can be done accurately and easily with the
help of prevailing methods available in literature [7,9]. Whereas to predict the axial thrust, the
prevailing methods are not found very much reliable to use as the number of parameters
influencing axial thrusts are more than those for radial thrust. Also to contribute to it are the
variations in the impeller geometries, which are governed by the hydraulic design and keeps
on changing based on the hydraulic design of the impeller. Therefore to predict the bearing
life and select the suitable bearing for getting optimum bearing life, it is really necessary to
formulate a basis to arrive at the reliable axial thrust values and formulation of such basis,
there is no better tool than experimental analysis.
This paper talks about the experimental analysis of the axial thrust of end suction long
coupled centrifugal process pump and the findings of the analysis.
1.0 INTRODUCTION
One of the important performance parameter for any centrifugal pump is its bearing life. To
have the minimum maintenance and repairs cost, the bearing life for any centrifugal pump
must be as prolonged as far as possible. Particularly in case of process pumps, which are used
at high temperatures, high pressure and more hazardous applications with maximum
reliability, bearing life is of more importance. Bearing life [12,13]. of centrifugal pump depends
upon two hydraulic forces acting on the impeller i. e. radial thrust and axial thrust. Amongst
the two forces, radial thrust is dependant solely on pressure generated by pump. There are
proven and reliable methods available to predict the radial thrust generated. Radial thrust can
be satisfactorily reduced to harmless extents by using double volute casings or diffuser type
casings.
Whereas axial thrust is dependent on the many aspects viz. shroud and casing clearances,
peripheral shroud speeds, head developed by the pump, impeller geometry etc. due to this one
cannot arrive at the conclusion that the theoretically calculated thrust value and the thrust
experienced by the pump in practice are same. This is mainly due to the manufacturing
limitations and assembly variations. The existing methods cannot be employed in practice to
estimate the correct value of axial thrust or even the permissible variation. The available
literature on measurements of axial thrust in pumps indicates the lack of adequate and
accurate experimental results data. Axial thrust data and improved analytical methods are
ICEM15
1
Porto/Portugal, 22-27 July 2012
critical to the proper selection of thrust bearings due to the consequences of rapid wear,
frequent maintenance and possible pump failure due to large axial forces, it is better to predict
and minimize axial thrust for a range of pump types in a variety of applications.
2.0 NEED FOR OPTIMIZATION OF AXIAL THRUST
The life and size of the bearings in centrifugal process pumps is greatly influenced by the
axial thrust developed.
High axial thrust loads can cause rapid thrust bearing wear and either subsequent pump failure
or frequent overhauls. The axial thrust is balanced by various methods like balancing chamber
and holes, pump out vanes, balancing ring and disc and by using balancing drum. By reducing
the pressure in the space between the back shroud and the casing wall, the axial thrust is also
balanced.
The axial thrust is dependent on the many aspects viz. shroud and casing clearances,
peripheral shroud speeds, head developed by the pump, impeller geometry etc. due to this one
cannot ensure that the theoretically calculated thrust value and the thrust experienced by the
pump in practice are same. This is mainly due to the manufacturing limitations and assembly
variations. None of the existing method can currently be employed in practice to estimate the
correct axial thrust value or even the permissible variance.
The available literature on measurements of axial thrust in pumps indicates the lack of
adequate and accurate experimental results data. Axial thrust data and improved analytical
methods are critical to the proper selection of thrust bearings due to the consequences of rapid
wear, frequent maintenance and possible pump failure due to large axial forces, it is better to
predict and minimize axial thrust for a range of pump types in a variety of applications.
3.0 HYDRAULIC THRUSTS IN CENTRIFUGAL PUMPS
During operation and working of centrifugal pumps, as mentioned earlier, the kinetic energy
of flowing liquid is converted into pressure energy. This high pressure liquid is continuously
flowing all over the circumference of the impeller and also gets entrapped inside the
clearances between impeller and casing / casing cover. This high pressure liquid exerts
pressure on the outlet passages and shrouds of the impeller resulting in generation of two
forces, one in lateral and another in longitudinal direction with respect o shaft axis.
The force generated in lateral direction is due to dissimilar pressure generation in
volute and called as Radial thrust[fig1], while another one generated in longitudinal direction is
on account of different areas of impeller exposed to trapped pressurized liquid called as axial
thrust.
3.1 RADIAL THRUST
The hydraulic radial load is due to the unequal velocity of the fluid flowing through the
casing. The unequal fluid velocity results in a non-uniform distribution of pressure acting on
the circumference of the impeller. The radial load is most influenced by the design of pump
casing. The pump casing is designed to direct the fluid flow from the impeller into the
discharge piping. In a theoretical situation at Best Efficiency point (BEP), the volute casing
has a uniform distribution of velocity and pressure around the impeller periphery. There are
proven formulas to predict the hydraulic radial thrust with fair amount of accuracy. Radial
2
15th International Conference on Experimental Mechanics
thrust can be minimized by making double volute casing [fig1] or by providing the diffuser type
casing.
Fig. 1 – Single volute and Double volute.
3.2 AXIAL THRUST
Axial thrust arises in the centrifugal pumps due to their asymmetry. The clearances between
casing cover and impeller back shroud, casing and impeller front shroud are filled with fluid
at delivery pressure. This pressure acts on the impeller shrouds. As the back shroud is having
larger surface area than front shroud, a net thrust acts on the impeller in the direction opposite
to that of the incoming flow [9]. The force contributed by the change in the momentum of
incoming flow is also to be considered.
F2
F1
Fm
UNBALANCED
THRUST
Figure 2 – Axial thrust components
As shown in Fig. 2 above the resultant unbalanced axial thrust is vector summation of the
following forces [1] –
Force acting on front shroud due to liquid of delivery pressure entrapped between pump
casing and front shroud. (F1)
Force acting on back shroud due to liquid of delivery pressure entrapped between casing
cover and back shroud. (F2)
ICEM15
3
Porto/Portugal, 22-27 July 2012
Force acting in the direction of the liquid flow due to its momentum change. (Fm)
The pressures generated by a centrifugal pump exert forces on both its stationary and rotating
parts. The design of these parts balances some of these forces, but separate means may be
required to counter-balance others. Axial hydraulic thrust is the summation of unbalanced
impeller forces acting in the axial direction. As reliable large-capacity thrust bearings are not
readily available, axial thrust in single-stage pumps remains a problem only in larger units.
Theoretically, a double-suction impeller is in hydraulic axial balance with the pressures on
one side equal to, and counter-balancing the pressures on, the other Fig 6. In practice, this
balance may not be achieved for the following reasons:
The suction passages to the two suction eyes may not provide equal or uniform flows to the
two sides.
DISCHARGE PRESSURE
DISCHARGE PRESSURE
SUCTION PRESSURE
DISCHARGE
PRESSURE
SUCTION
PRESSURE
SUCTION
PRESSURE
DISCHARGE
PRESSURE
1. External conditions such as an elbow being too close to the pump suction nozzle may
cause unequal flows to the suction eyes.
Figure 3 – Origin of pressure acting on impeller shrouds to produce axial thrust.
2. The two sides of the discharge casing may not be symmetrical, or the impeller may be
located off-centre. These conditions will alter the flow characteristics between the impeller
shrouds and casing, causing unequal pressures on the shrouds.
3. Unequal leakage through the two leakage joints will tend to upset the balance.
Combined, these factors create definite axial unbalance. To compensate for this, all
centrifugal pumps, even those with double-suction impellers, incorporate thrust bearings.
The ordinary single-suction radial-flow impeller with the shaft passing through the impeller
eye Fig 3 is subject to axial thrust because a portion of the front wall is exposed to suction
pressure, thus exposing relatively more back wall surface to discharge pressure. If the
discharge chamber pressure were uniform over the entire impeller surface, the axial force
acting towards the suction would be equal to the product of the net pressure generated by the
impeller and the unbalanced annular area.
4
15th International Conference on Experimental Mechanics
4.0 Axial thrust balancing
The thrust magnitude will be approximately equal to the product of the net pump pressure
and the annular unbalanced area. Actually, the axial thrust turns out to be about 70% to 80%
of this theoretical value. The axial thrust can be safely carried out by a thrust bearing; this is
the most efficient way to reduce the thrust on the bearing. This can be done only at the
expense of the pump efficiency. Following are the different methods used to eliminate the
axial thrust of a single-suction impeller,
4.1 Providing wear rings and balancing holes: [1], [7], [9]
One of the method which is usually employed to reduce or eliminate axial thrust in single
stage pumps in a chamber on the back of the impeller is provided with a closely fitted set of
wearing rings and suction pressure is admitted to this chamber either by drilling holes through
the impeller back shroud into the eye or by providing a special channel connecting the
balancing chamber to the suction nozzle [fig 4].
The use of drilled holes through the impeller shroud to the balancing chamber is
inferior to the arrangement using a special channel to connect the balancing chamber with the
suction nozzle because leakage through the holes is directed against the flow in the impeller
eye, causing disturbances. The balance by this method is never complete. From 10 to 25 per
cent of the axial thrust always remains depending on the size of the holes. For a complete
balance the diameter of the wearing rings of the balancing chamber should be greater than that
at the impeller eye.
IMPELLER
BALANCE HOLE
REDUCTION IN
AXIAL
THRUST DUE TO
BALANCE HOLES
Figure 4 – Axial thrust reduction with balance holes
4.2 Balancing with radial back vanes: [1], [7], [9]
In this method, radial ribs are used on the back shroud to reduce the pressure in the space
between the impeller, and the pump casing. Figure 5 shows the schematic of an impeller
housed inside pump casing the radial back vanes provided at the back shroud of the impellers
acts as auxiliary impeller which restricts the entry of liquid into the clearances between
impeller back shroud and casing cover.
It is evident that the balancing with wearing and holes method introduces the leakage loss of
the pump which, in turn, increases as wearing rings are worn. Balancing with ribs method
requires some additional power which, however does not change with time. In addition it is
ICEM15
5
Porto/Portugal, 22-27 July 2012
cheaper and more effective than the first method. In this method the power consumed is much
less than the power loss due to leakage through balancing holes under normal conditions.
BACK VANES
IMPELLER
REDUCTION IN
AXIAL
THRUST DUE TO
BACK VANES
Figure 5 – Axial thrust reduction with back vanes
There are some other thrust balancing methods like Balancing Disc and ring, Balancing Drum
used specifically to reduce the axial thrust in multistage centrifugal pumps.
5.0 EXPERIMENTAL ANALYSIS [14]. –
The actual thrust induced on the impeller with back vanes will be measured by virtue of an
experimental test rig. This will include direct measurement of the thrust exerted on the
impeller by employing suitable transducer. The methods of measuring the unbalanced
hydraulic axial thrust experimentally are –
•
Providing tapping in pump casing and casing cover for obtaining the pressure values
between the clearances of impeller and casing.
•
By making a provision for load cell at the end of the shaft
the axial end thrust in Kg.
[14]
, this will directly give
Finally the results obtained experimentally and those obtained by theoretical analysis were
compared to draw a conclusion.
5.1 Experimental analysis procedure[14].
The method of orthogonal array was used to conduct this experiment instead of making
simple iterations (conventional method) of predefined parameters. For conducting this
experiment conventionally it would have required 81 different trials based on three variations
of four parameters each, but by utilizing the orthogonal array method, the same results can be
obtained in three trials. With orthogonal array method, it is possible to analyze the effect of
influencing parameters on the axial thrust development quickly and with fewer trials.
In this technique it is proposed that an experiment organizes consecutive small series of trials
in each of which all the factors are simultaneously varied according to a definite rules. The
series are such arranged that after mathematical processing of a proceeding, it will be possible
6
15th International Conference on Experimental Mechanics
to analyze the effect of each influencing parameters on the response value. For this project
following four parameters are chosen –
1. The number of back vanes,
2. The back vane radius,
3. The back vane height,
4. Clearance between casing cover and back vanes.
All the above factors have three variations each as given in Table-1.
Experiment
no
No of back
vanes
(Z)
Back vane
diameter
(D) mm
Back vane
thickness
(T) mm
Clearance between back
vane and casing
(S-T) mm
1
8
107.00
6.00
5.00
2
8
87.00
3.50
3.00
3
8
67.00
1.00
1.00
4
6
107.00
3.50
1.00
5
6
87.00
1.00
5.00
6
6
67.00
6.00
3.00
7
4
107.00
1.00
3.00
8
4
87.00
6.00
1.00
9
4
67.00
3.50
5.00
Table-1- Orthogonal array experiment design
5.2 Experimental setup (Measurement technique) [14].
These tests are been carried out on an end suction pump of 66 specific speed (metric) with 214
mm diameter. The test rig (fig6) used consisted mostly of the test end suction pump long
coupled to electric motor running 2960 rpm, a modified side bearing arrangement consisting
of three strain gauge type load cells located at 1200 apart over the circumference and on the
both side of cartridge. The transfer mechanism for transmittal of axial thrust from rotating
members to the stationary load cells consists of two thrust ball bearings, which are mounted
on the impeller shaft. The shaft is supported by two radial bearings, with their outer races free
to slide inside the bearing housing. The load cells are pressed slightly against the modified
bearing bracket which comes in contact with the rotating assembly through the thrust ball
bearings. The load cells were used on either side of thrust bearings so as to respond to change
in the axial thrust direction.
The total axial thrust will be the algebraic sum of the forces experienced by all the three load
cells. To avoid any error, all the load cells were preloaded slightly and at the preloaded
condition, the null reading was set.
5.3 Calibration[14]
The test rig was statically calibrated before the start of experiment by pulling and pushing the
impeller shaft against known calibrated load values from UTM (Universal Testing Machine).
The load cell response (Eb mV) to the calibration null load was set based on the known static
ICEM15
7
Porto/Portugal, 22-27 July 2012
load application, the graph-1 shows the response of the load cells output voltage with respect
to the known load applied.
Graph 1 – Load cell output voltage (Eb) Vs Load (kg)
LOAD CELL
(Back)
LOAD CELL
(front)
DRIVER SIDE
PUMP SIDE
possible axial thrust direction
Fig-6 Experimental setup
5.4 Equations set used for theoretical analysis
The basic theory used to evaluate the axial thrust is taken from K J Zanker(1) paper
specifically for the impellers with back vanes. The theoretical analysis parameters were also
measured during the tests performed by mean of tapings on pump casing at various locations.
The theory presented is as given below (kg)
Force on front shroud
2 2
2
FR = π * ρ {g * Ho*(R -r )-W /4(R
2
-r12/2)2}
(kg)
Force on back shroud (with back vanes)
2
2
2
2
2
2
2
2
FB= π * ρ *{g HoR –w /4(R -b /2) -gHoS1-W /8(R -b )*(b
2
-S12)-B2 (b2-S12/2)2}
The momentum thrust due to radial out thrust
Tm = (ρ*Q2)/A1.
Net Axial Thrust
8
(kg)
15th International Conference on Experimental Mechanics
T
= (Force on back shroud considering back vanes –pressure force on front shroud –
momentum thrust)
T
= FB - FR - Tm
(kg)
•
W
=Angular velocity of impeller
(Rad/s)
•
B
= Angular velocity of fluid
(Rad/s)
•
ρ
= Density
(Kg/m3)
•
H
= Pump head
(meters)
•
Ho
= Static head at the impeller outlet (meters)
•
R
= Impeller radius (meter)
•
r1
= Impeller wear ring radius (meter)
•
b
= Radius of back vanes (meter)
•
h
= Hydraulic efficiency of pump
•
Cd
= Coefficient of contraction of jet flow at the impeller back vane
•
u2
= Peripheral velocity at the impeller outlet
(Rad/s)
•
S
= Axial gap between casing and impeller back shroud
(mm)
•
T
= Vane width
(mm)
•
(s-t)
=Vane clearance
(mm)
•
Z
= Number of back vanes
•
Re
=Reynold’s number
•
γ
= Kinematic viscosity
(m2/sec)
•
S1
=Radius of shaft
(meter)
•
Q
= Discharge
(m3/hr)
•
ηh
= Hydraulic efficiency of pump.
•
N
= speed in
(rpm)
•
A1
= Eye opening area
(m2)
Where,
6.0 Observations [14]
1) The thrust data was obtained throughout the performance range (H-Q curve) of the
pump. It is observed that the parameter controlling axial thrust are as follows in order
of importance:
•
Back vane radius
•
Back vane height
•
No. of back vane
•
Clearance between casing cover and Impeller back Vane.
ICEM15
9
Porto/Portugal, 22-27 July 2012
(Kg)
2) From graph it is observed that as number of back vanes increases the back vane force
developed also increases since the increase in number of back vanes increases the
angular velocity of the fluid between the back shroud of impeller and pump casing.
But after some limit, if no. of back vanes increased that will not be served. In these
experiments the optimum number of back vanes was found to be 8 [graph 2].
Graph 2- Variation of back vane force with number of back vanes
3) From graph-3. it is observed that back vane force increases as the radius increases. The
increase of radius results in increasing the rotation of the fluid. (Over a radial length).
A small increase in back vane radius affect the back vane force for a great extent. If
full diameter back vanes are provided it will rotate the total water trapped in the axial
gap. There by increasing the angular velocity of the fluid over the whole region of
impeller. If the back vanes provided are less than the impeller diameter, the fluid
trapped in the region of back vanes and casing will have the angular velocity higher
than the angular velocity of fluid trapped in the remaining region of the impeller
diameter. In this case the back vane force developed will be due to combined effect of
the two angular velocities which will result in the lower back vanes force.
Graph 3 – Back vane force Vs Back vane radius
4) If the axial gap is very small, the angular velocity of the fluid in the gap will be high
and hence the back vane force as well. Back vane height also helps to increase the
rotation of the fluid in the axial gap. The optimum value has been found to be 5 mm
from these experiments.
10
(Kg)
15th International Conference on Experimental Mechanics
Graph 4 – variation of back vane force with back vane height
5) Graph 5 shows that the back vane force developed is higher for low clearance values
as well a with very high clearance values, when the clearance is very low the angular
velocity of the fluid is higher in this small gap. This will result in decrease in pressure
on the impeller shroud. As the clearance increases a very large value the eddy of
recirculatory flow will start through the axial gap. This will result in equalization of
pressure behind the impeller at the outer diameter and at the inner most diameters
naturally increasing the total back vane force.
Graph 5 – Axial thrust Vs Clearance between casing cover and Back vane face
6) The axial thrust variation with respect to flow for all 9 tests is as given below in graph
6-
ICEM15
11
Porto/Portugal, 22-27 July 2012
Axial thrust Vs Flow
800
700
Expt 4
600
Expt 1
Expt 3
Axial Thrust (kg)
500
Expt 8
Expt 2
Expt 7
400
Expt 9
Expt 6
300
Expt 5
200
100
0
0
20
40
60
80
100
Flow (lps)
Graph 6 – Axial thrust Vs Flow
7.0 Conclusion
1. Maximum influencing parameter
From the experimental data, it is observed that the back vane radius contributes to the
maximum extent to the variation of axial thrust.
2. The order of importance of other back vane parameters is as follows:
Back vane height
12
15th International Conference on Experimental Mechanics
Number of vanes
Clearance between back vanes and casing cover
3. Close control of back vane parameters is essential to maintain optimum thrust values
for mass produced pumps. The variation in thrust values due to change in the
dimensions of back vane height and clearance has been tabulated[14]
Axial Thrust at BEP
condition (kg)
Experiment
no. (Refer
table 1)
Without
back vane
With back
vane
Axial Thrust at 20% Left
of BEP condition (kg)
Pump efficiency at BEP
condition
Without
back vane
Without
back vane
With back
vane
With back
vane
Test
1
Test
2
Test
1
Test
2
Test
1
Test
2
Test
1
Test
2
Test
1
Test
2
Test
1
Test
2
1
640
642.5
100
101.5
815
812
295
298
82.3
82
78
78.5
2
630
628.5
378
377
795
794
594
592
82.3
82
78
78.5
3
590
588
553
552
673
674
707
580
82.3
82
78
78.5
4
590
588
282
283.5
673
674
497
501
82.3
82
78
78.5
5
640
642.5
556
551
815
812
749
748
82.3
82
78
78.5
6
630
628.5
510
513
795
794
690
691.5
82.3
82
78
78.5
7
630
628.5
545
547.5
795
794
730
733
82.3
82
78
78.5
8
590
588
388.5
385
673
674
547
546
82.3
82
78
78.5
9
640
642.5
482.5
480
815
812
688
688
82.3
82
78
78.5
References
1) K J Zanker, “Experiments with back vanes used for balancing axial thrust on centrifugal
pump impellers”, British Hydromechanics Research Association, Vol. 3, 5th Annual meet.
(Apr. 1962), pp.7-14.
2) George Schaefer and Eric Olson, ”Experimental Evaluation of Axial Thrust”, Journal of
World Pumps, Issue 393, June 1999, pp. 34-37.
3) Majan Gantar, Dr. Dussn Florjancic, Dr. Brane Sirok, “Hydraulic Axial Thrust in
Multistage Pumps - Origins and Solutions”, proceedings of ASME FEDSM’01, May 29 June1.2001, pp. 1-8.
4) Marc P. Mignolet Byeong-Keun Choi, “Robust Optimal Positioning of Strain Gages on
Blades”, journal of turbo machinery- transactions of ASME Vol. 125, October 2002. pp.
155-164.
5) J. Szwedowicz, S. M. Senn and R. S. Abhari, “Optimum Strain Gage Application to
Bladed Assemblies”, journal of turbo machinery- transactions of ASME Vol. 125, January
2003. pp. 606-613.
6) Igor J Karassik & Ray Cartor, “Centrifugal pumps”, Tata McGraw Hill Book company2nd Edition, 1960. pp. 3.129-3.339.
7) A.J. Stepanoff, “Centrifugal and Axial flow pumps - Theory, Design & Application” ,
John Wiley and Sons – 2nd edition – 1953. pp. 182-223.
ICEM15
13
Porto/Portugal, 22-27 July 2012
8) Stephen Lazarkiewicz & Adam T. Troskolanski, “Impeller pumps”, Pergramon Press- 1st
edition 1965. pp. 346-365.
9) Val S. Lobanoff & Robert R. Ross, “Centrifugal Pumps design and application”, Jaico
publishing house – 2nd edition 1995. pp. 333-353.
10) “Centrifugal pumps for petroleum, heavy duty chemical and gas industry services”,
American Petroleum institute – 610 10th edition-October 2004
11) “General catalogue” SKF bearings.
12) “Bearings in Centrifugal Pumps – Application handbook – Part I &II”, SKF100955_1,
SKF bearings.
13) “ISO 281: Rolling Bearings - Dynamic Load Ratings and Rating Life.”, Second edition
1990-03-25.
14) “Experimentation on suction pump for axial thrust evaluation”, R&D, Kirloskar Brothers
limited, Kirloskarvadi.
14