Pump
p Division
Flowserve Pumps
IDP Pumps
Cavitation in Centrifugal Pumps
and Prediction Thereof
F k C.
Frank
C Vi
Visser
Flowserve Pump Division
Etten-Leur, The Netherlands
Tutorial
Presented at 2005 ASME Fluids Engineering Division Summer Conference,
June 19-23,, 2005,, Houston,, Texas,, USA
Outline
• Part 1: What is cavitation and what does it mean for
pumping machinery?
• Part 2: Prediction of cavitation in centrifugal pumps
– Scaling laws
– Thermodynamic effect (temperature depression)
– Effect
Eff t off dissolved
di
l d or entrained
t i d gases
– Calculating incipient cavitation (NPSH) from CFD
– Cavity length prediction
2
P t 1 – What
Part
Wh t iis cavitation
it ti
Cavitation is defined as the process of formation and disappearance
of the vapour phase of a liquid when it is subjected to reduced and
subsequently increased pressures.
The formation of cavities is a process analogous to boiling in a liquid,
although it is the result of pressure reduction rather than heat
addition.
Cavitation is a thermodynamic change of state with mass transfer
from liquid
q
to vapor
p p
phase and visa versa (( bubble formation &
collapse).
3
P t 1 – What
Part
Wh t iis cavitation
it ti
(cont.)
(
t)
Sheet cavity on pump
impeller vane leading
edge (suction side)
Speed = 2990 RPM
NPSHA = 70 m(230 ft)
Flow rate =1820 m3/h
(8015 gpm)
Vane marker stripes at
intervals of 10 mm (0.4 in)
Cavity length = 25-40 mm
(1.0 – 1.5 in)
(from Visser et al, 1998)
4
Part 1 – What is cavitation (cont
(cont.))
Cavitation causes or may cause:
• Performance
P f
loss
l
(head
(h d d
drop))
• Material damage (cavitation erosion)
• Vibrations
• Noise
• Vapor lock (if suction pressure drops
b l
below
b
break-off
k ff value)
l )
(Visser et al, 1998)
General Advice: TRY TO AVOID CAVITATION (under normal operation)
Unfortunately, economic or operational considerations often necessitate
operation with some cavitation
cavitation, and then it is particularly important to
understand the (negative) effects of cavitation.
 Design optimization to minimize cavitation
5
P t 1 – What
Part
Wh t iis cavitation
it ti
(cont.)
(
t)
Typical cavitation damages
Centrifugal pump impeller
cavitation pitting erosion @ inlet
(from Dijkers et al, 2000)
Francis turbine runner
cavitation damage @ discharge
(from Brennen, 1994)
6
Part 1 – What is cavitation (cont
(cont.))
Cavitation behavior is typically expressed in terms of cavitation
parameters.
t
• Cavitation number:
p1  pV
; (Centrifugal Pumps : U  Ueye  R1T )
 1
2
2 U
• Net Positive Suction Head:
p01  pV
NPSH 
g
• Thoma cavitation number:
 TH
NPSH

H
7
Part 1 – What is cavitation (cont
(cont.))
In g
general, cavitation performance is related to some “critical” value:
NPSHA (=available) > NPSHc or NPSHR (=critical or required)
Typical “critical”
critical characteristics identified for centrifugal pumps:
• Incipient cavitation (NPSHi)
• Developed
p cavitation causing
g 3% head drop
p ((NPSH3%))
• Developed cavitation causing complete head breakdown
( vapor lock).
Choice of NPSHR is rather arbitrary, but usually NPSHR=NPSH3%
Alternative choices:
• NPSHR=NPSH1% or NPSHR=NPSH5%
• NPSHR=NPSHi (cavitation free operation)
8
Part 1 – What is cavitation (cont
(cont.))
Cavitation Phenomena
9
Cavitation Visualization Test Pump
Pump Division
Begin Visual Cavitation
3% head drop
1% head drop
0% h
head
dd
drop
Begin visual cavitation
Head (m)
4.05
4.00
3.95
3.90
3.85
3.80
3 75
3.75
3.70
0
10
20
30
40
50
60
70
80
90
100
NPSH (m)
Pump Division
0% Head Drop
3% head drop
1% head drop
0% head drop
Begin visual cavitation
Head (m)
4.05
4.00
3.95
3.90
3.85
3.80
3.75
3.70
0
10
20
30
40
50
60
70
80
90
100
NPSH ((m))
Pump Division
1% Head drop
3% head drop
1% head drop
0% head drop
Begin visual cavitation
Head (m)
4.05
4.00
3.95
3.90
3.85
3.80
3.75
3.70
0
10
20
30
40
50
60
70
80
90
100
NPSH ((m))
Pump Division
3% Head drop
3% Head drop
1% head drop
0% head drop
Begin visual cavitation
Head (m)
4.05
4.00
3.95
3.90
3.85
3.80
3.75
3.70
0
10
20
30
40
50
60
70
80
90
100
NPSH ((m))
Pump Division
Recirculation
3% head drop
1% head drop
Recirculation
0% head drop
Begin visual cavitation
Head (m)
4.05
4.00
3.95
3.90
3.85
3.80
3.75
3.70
0
10
20
30
40
50
60
70
80
90
100
NPSH ((m))
Pump Division
Part 1 – What is cavitation (cont
(cont.))
10
Part 1 – What is cavitation (cont
(cont.))
Typically (in practice):
• NPSHA > NPSH3%
• NPSHi > NPSHA (especially for low capacity)
 Pumps
u ps run
u o
okay,
ay, BUT
U with
t some
so e de
developed
e oped cavitation.
ca tat o
General misconception:
NPSHA > NPSHR  No Cavitation
(This will only hold if NPSHR = NPSHi.)
11
Part 2 – Cavitation prediction
• Scaling laws
• Thermodynamic effect
• Effect of dissolved or entrained gases
• Ca
Calculating
cu at g incipient
c p e t ca
cavitation
tat o (N
(NPSHi)
S i) from
o C
CFD
• Cavity length prediction
12
Part 2 – Cavitation prediction (cont
(cont.))
Predicting NPSH at speeds other than reference or test speed
( scaling laws)
2
 N 
2





NPSH
N
NPSH
NPSH
NPSHi:
i
i
i , REF 
 N REF 
NPSH
( TH 
 constant)
H
2
NPSH3%:
 N 


NPSH 3%  f NPSH 3%,
% REF 
 N REF 
N  N REF , f  1 ; N  N REF , f  1
“Postulate”: Amount of developed cavitation depends on residence
time  f depends on size of the pump and ratio N/NREF
13
Part 2 – Cavitation prediction (cont
(cont.))
Alternative approach to account for deviation from affinity law:
NPSH 3%
 N

 NPSH 3%,
% REF 
 N REF
1  2




Choice of  is rather arbitrary and relies heavily on empiricism
Conservative choice:
N < NREF ,  = 1
N > NREF ,  = 2
14
Part 2 – Cavitation prediction (cont
(cont.))
Thermodynamic
y
effect
(temperature depression)
Cavitation performance
depends on:
• Temperature of liquid
• Type of liquid
 NPSHR reduction
(E.g. Stepanoff method, or
Hydraulic Institute correction
chart)
(from Brennen, 1994)
15
Part 2 – Cavitation prediction (cont
(cont.))
Predicting thermodynamic effect
NPSH 3%  NPSH 3%, REF  NPSH
Equilibrium theory:
h fg
V
L
NPSH  B 2
;B V
V v fg g C p T
VL
Stepanoff (1965
(1965, 1978):
2
B  B1 NPSH
2
 L  g C pT
1
1

B1  
;
[
m
]
or
[
ft
]
2
 V  h fg
29  4 3
64  4 3
NPSH 
B1 ; [m 1 ] or NPSH 
B1 ; [ ft 1 ]
HV
HV
Non-equilibrium theory  bubble dynamic (CFD) calculations, involving
time-dependent
time
dependent two-phase
two phase flow calculations
16
Part 2 – Cavitation prediction (cont
(cont.))
Influence of dissolved and/or entrained gases:
 “conceptual effective or artificial” vapor pressure:
PE = PV + 
PE = yP
P0
(Ch
(Chen,
1993)
Key characteristic:
Performance (breakdown) comes from gas evolution and gas
expansion, rather than classical vapor formation.
Dissolved and/or entrained gases result in reduction of (effective)
field NPSHA:
NPSHA* = (P01 – PE) / g
“Hidden danger”: NPSHA > NPSHR but NPSHA* < NPSHR
17
Part 2 – Cavitation prediction (cont
(cont.))
Predicting incipient cavitation (NPSHi)
(
) from
f
CFD
C
T i l approach:
Typical
h
Create 3D geometry model/grid of impeller passage

Solve flow field with CFD code (non-cavitating)

Calculate incipient NPSH from CFD pressure field (next slide)
18
Part 2 – Cavitation prediction (cont
(cont.))
Streamline through point of minimum pressure
NPSH i 
p01,i  pV
g
p01,i  p1,i  12 U 2
p1,i  p1  ( pmin  pV )

NPSH i 
p01  pmin
g
So: NPSHi follows from pmin and p01 of calculated pressure field, and
does not require pV to be known!
19
Part 2 – Cavitation prediction (cont
(cont.))
Running simulations for several flow rates produces NPSHi curve:
(from Visser
Visser, 2001)
20
Part 2 – Cavitation prediction (cont
(cont.))
Note: CFD calculated characteristic is for impeller flow!
To project it on pump throughput one needs to account
for volumetric efficiency
y (
( eye
y wear ring
g leakage
g flow):
)
Qimpeller = Qpump + Qleakage  Qpump = Qimpeller - Qleakage
 Computed curve shifts left by amount Q = Qleakage
Qleakage
p
 f(p, D, L,  , ,  ) ~  D u ; u 
 L
1
2
laminar  24 / Re ; turbulent  0.2373 / Re
0.25
u
; Re 
2
It becomes particularly important to take Qleakage into account for low
NS (specific speed) impellers. For high NS the relative influence is less.
21
Part 2 – Cavitation prediction (cont
(cont.))
What if NPSHA < NPSHi ?
 Find region on impeller blade surface where p < pV
• physically unrealistic
unrealistic, but it gives
• first “indication” of cavitation area, and
• first approximation of cavity bubble length
Note: The actual cavity will be bigger
 bubble length will be underestimated
22
Part 2 – Cavitation prediction (cont
(cont.))
To visualize p < pV region from non-cavitating flow simulation:
 Plot isotimic surface for threshold value pV*
pV*  pV  ( p1  p1, A )
 p1  12 U 2 
( p1, A  12 U 2  pV )
 p01   g NPSHA
 p01  NPSPA
23
Part 2 – Cavitation prediction (cont
(cont.))
Example:
Plot of p < pV region
NPSHA = 15.5 m (51 ft)
NPSHi = 28 m ((92 ft))
N = 2980 RPM
Q = 400 m3/h
(1760 USGPM)
Cavitation on blade
suction side
24
Part 2 – Cavitation prediction (cont
(cont.))
Putting
g LCAV = m L(p<p
(p pV), m=O(3),
( ), one can get
g some impression
p
of
expected cavitation erosion rate
n
 LCAV 
2 3
6 3
2



Güli h (1986
Gülich
(1986, 1988
1988, 1989)
1989): E  C



U

A
8
T
e
S
L
 A
 CAV ,10 
n
or
(*)
E  Ln  E  E L L 

CAV
with

2
1
 
2
1
n = 2.83 for blade suction side and
n = 2.6
2 6 for blade pressure side
Equation
q
((*)) is especially
p
yp
powerful when comparing
p
g designs
g and
evaluate susceptibility to cavitation erosion (in a relative sense).
 Design optimization studies
25
Part 2 – Cavitation prediction (cont
(cont.))
• Results and theory thus far do not require two-phase flow
calculations.
l l ti
• Still it p
provides important
p
information of an impeller
p
design
g
regarding cavitation performance.
• Next level of improvement has to come from CFD calculations
with cavitation model.
• Calculations with a cavitation model are time consuming and
tend to be “CPU-expensive”
• Several cavitation models exist to date, and development of
cavitation models is still ongoing
26
Part 2 – Cavitation prediction (cont
(cont.))
CFD Cavitation models
Typically two approaches:
• Equilibrium models
– Barotropic or pseudo density models;  =(p)
– Somewhat “simplistic”,
p
, yet
y
– Attractive since they can be used in single phase codes
• Bubble dynamic
y
models
–
–
–
–
–
Rayleigh-Plesset equation
Vapor-liquid interaction (time-dependent mass & heat transfer)
Closer to reality
More complicated and more “CPU-expensive”
E.g. Volume of Fluid (VOF) model
27
Part 2 – Cavitation prediction (cont
(cont.))
Example:
Plot of cavity bubble
Equilibrium
q
model
CFX-TASCflow
(CEV-model)
NPSHA = 15.5
15 5 m (51 ft)
NPSHi = 28 m (92 ft)
N = 2980 RPM
Q = 400 m3/h
(1760 USGPM)
m3
Cavitation on blade
suction side
28
Part 2 – Cavitation prediction (cont
(cont.))
Application:
With CFD cavitation
it ti models
d l one can predict
di t NPSH3% from
f
CFD
calculated head drop curves
((from Visser,, 2001;; CEV-model p
prediction))
29
Concluding Remarks
• Cavitation is a phenomenon which can seriously impact
performance and operation of pumps.
• Predicting cavitation performance is an important topic
topic,
not only for pumps, but for fluid machinery in general.
• Traditional (scaling) methods are still important and
useful.
• CFD methods provide further insight and are becoming
more and more common.
• Bubble dynamic (CFD) methods are emerging and hold a
promise for the future.
30
References
Brennen, C.E.
H d d
Hydrodynamics
i off P
Pumps.
Oxford University Press (1994)
Chen, C
Chen
C.C.
C
Cope with dissolved gases in pump calculations.
Chemical Engineering, vol. 100 (1993), pp. 106-112.
Dijkers, R.J.H., Visser, F.C. & Op De Woerd, J.G.H.
Redesign of a high-energy centrifugal pump first-stage impeller.
Proceedings of the 20th IAHR Symposium,
Symposium August 6-9
6 9, 2000
2000, Charlotte
Charlotte,
North Carolina, USA.
Gülich, JJ. F
Gülich
F. and Pace
Pace, S.
S
Quantitative Prediction of Cavitation Erosion in Centrifugal Pumps.
Proceedings of the 13th IAHR Symposium (1986), Montreal, Canada.
31
References (cont
(cont.))
Gülich, J. F. and Rösch, A.
Cavitation Erosion in Centrifugal Pumps.
World Pumps, July 1988, pp. 164-168.
Gülich, J. F.
Guidelines for Prevention of Cavitation in Centrifugal Feedpumps.
EPRI Final Report GS-6398, (1989).
Gülich, J. F.
Beitrag zur Bestimmung der Kavitationserosion in Kreiselpumpen auf Grund der
Blasenfeldlänge und des Kavitationsschalls
Kavitationsschalls.
Thesis, Technische Hochschule Darmstadt, Germany, 1989.
Stepanoff, A
Stepanoff
A.J.
J
Pumps and Blowers – Two-Phase Flow.
John Wiley & Sons (1965), Krieger Publishing (1978)
32
References (cont
(cont.))
Visser, F.C., Backx, J.J.M., Geerts, J., Cugal, M. & D. Miguel Medina Torres
Pump impeller lifetime improvement through visual study of leading-edge cavitation.
Proceedings of the 15th International Pump Users Symposium, Turbomachinery
Laboratory,Texas
y,
A&M University,
y, College
g Station,, Texas,, USA,, pp.
pp 109-117.
Also in: Pumping Technology, vol. 2 (1998), pp. 149-157.
Visser, F
Visser
F.C.
C
Some user experience demonstrating the use of CFX-TASCflow computational fluid
dynamics for cavitation inception (NPSH) analysis and head performance prediction
of centrifugal pump impellers. FEDSM2001-18087
Proceedings of the 4th ASME International Symposium on Pumping Machinery,
May 29 – June 1,
1 2001,
2001 New Orleans
Orleans, Louisiana
Louisiana, USA
USA.
33