You are on page 1of 5

36

Feature
WORLD PUMPS

April 2011

Cavitation control

The NPSH performance


indicator, NPSH()
In this article, a new concept is proposed for the assessment
of cavitation risk in centrifugal pump operations. The method
based on NPSH() the net positive suction head performance
indicator is low cost and simple to apply and can assist both
manufacturers and users, as Dr Edward Grist explains.

entrifugal pump manufacturers


cannot be expected to enter into
guarantees against cavitation where
unacceptable consequences may be long
term or dependent upon the method
of pump operation. Equally, pump users
need reassurance that pumps are t for
purpose in that all reasonable measures
are demonstrated to have been taken so
that unacceptable performance is avoided.
The present methods of calculation
that attempt to bridge this gap are
complex and beyond the understanding
of new or infrequent pump users. A
low-cost, simple-to-apply method of
demonstrating the veracity of cavitation avoidance claims is needed. Such
a method based on the net positive
suction head performance indicator,
NPSH(), meets this need. It could form
part of a Pump Acceptance Test standard.

Changing the engineering approach


NPSH() and its corresponding best eciency ow rate Q() dene at a notional
reference speed the boundaries of unacceptable centrifugal pump performance
caused by cavitation. It is based on a pragmatic evaluation of known risk areas and
commercial cavitation test practices. It has
a very wide but not unlimited range of
application. It errs on the side of caution.
The reference speed is 3,600 rpm. The
areas of concern encompassed by
NPSH() are:
www.worldpumps.com

hydraulic performance loss


cavitation surging (hydrodynamically
induced) and
cavitation erosion.
The resulting composite boundary
of unacceptable performance is
described relative to test data where
cavitation causes a 3% fall in generated head. The base for this so-called
NPSH(3PC) data is a cold water
test in the temperature range 2C
to 40C. Its application is restricted
to flow rates in the range 50% to
120% of the best efficiency value.
For water, the reductions in NPSH(3PC)
resulting from the changes in the thermodynamic properties between 40C
and 150C are comparable with errors
commonly made in water temperature measurement. For all practical
purposes, an NPSH() value based on
cold water test data can be applied
to water, without correction, over the
temperature range 2C to 150C.
The NPSH() and Q() values together
characterize a particular pump.
The data from which the former is
derived allow cavitation performance assessments to be made reliably within the pump speed range
of 900 rpm to 5,400 rpm. The only
condition in using the NPSH() value
is that the flow rate for continuous
operation is within the 50120%
best efficiency flow rate range.

0262 1762/11 2011 Elsevier Ltd. All rights reserved

The method of applying NPSH()


starts with the pump manufacturer
carrying out a cold water cavitation test. This is at a convenient test
speed nT of between 900 rpm and
3,600 rpm. At this test speed the
value of best eciency ow rate QT
and NPSH(3PC)T are determined and
from them the values of NPSH() and
Q() are calculated. These are the
descriptors for the particular pump.
Choosing a pump for a particular
application necessitates evaluating
the value of the net positive suction
head required, NPSH(R), that minimizes the risks from unacceptable
cavitation. This is achieved by a simple
calculation using the pump duty ow
rate QG (sometimes referred to as
the guarantee ow), the associated
pump speed nG and the values of
NPSH() and Q() determined earlier.
The denition of NPSH is that given in
ISO 9906:2000 (Ref. 1). The notation used
is explained where it arises in the text.

The base for NPSH() assessments


NPSH(3PC)
Cavitation performance is usually dened
by a deviation from the curve for noncavitating generated head/ow rate. The
3% head-drop value, NPSH(3PC), is now
the internationally accepted basis for this
deviation in performance comparisons.

Feature
April 2011

Generated head,

NPSH(3PC) data is obtainable by either the


constant ow rate method shown in Figure
1 or by the constant NPSH method shown
in Figure 2. The former method can be quick
(less than 15 mins) but requires a test facility
where strict ow control can be maintained
as the value of NPSH is reduced; the latter
is time-consuming (because the extent of
cavitation within the pump is not readily
evident from external instrument readings) but requires only the simplest of test
facilities. The value of NPSH(3PC)T obtained
at QT is the same for the two methods.

3% H

NPSH(3PC)T=2.1 m

Cold water

Figure 1. NPSH(3PC)T determined by the constant ow rate method.

NPSH = 2.1 m constant


Pump speed nT : 2,000 rpm
Pumped liquid: water at 20C

QT

3% H
H

NPSH(3PC)T= 2.1 m

Pump eciency,

An error of 2C in temperature measurement for water at 20C produces an


error in NPSH(3PC) of less than 0.03 m.
Compared to typical NPSH(3PC) values
for pumps running at speeds of >900
rpm this is insignificant. Additionally,
there is much evidence to show that
the NPSH(3PC) requirements for water
change very little over the temperature range 2C and 40C; a possible
reason for this is described in the later
section on cavity growth in water.

Net positive suction head, NPSH

Generated head,

Industrial cavitation tests using cold water


yield NPSH(3PC) results that are reliable,
repeatable and obtained without difficulty.
This is not the case for very hot water and
most other liquids. In the context of evaluations relating to NPSH(), the limits of the
description cold water are 2C and 40C.

Pump ow rate QT : constant


Pump speed nT : 2,000 rpm
Pumped liquid: water at 20C

Best eciency ow rate

WORLD PUMPS

QT

Reference speed 3,600 rpm


Most of the pumps manufactured worldwide
run at a speed of less than 3,600 rpm. This
results from the vast majority being powered
by alternating current electric motors at
electrical supply frequencies of 50 Hz or
60 Hz. Adopting 3,600 rpm as the reference
speed has practical signicance. Pumps
below 3,600 rpm have low NPSH(3PC)
typically 35 m at most. The inaccuracy
incurred in scaling from 3,600 rpm to lower
speeds is less than 0.2 m which, in plant
geometry terms, is very small indeed.

Scaling to other pump speeds


Scaling non-cavitating generated head
and pump flow rate values to another
pump speed using the so-called affinity
rules is well established and of proven
accuracy. The extent to which these
same rules can validly be applied
to NPSH data is now reviewed.
If dimensional similarity between
both the pump internal passages and
the cavitation within them existed

Flow rate, Q
Figure 2. NPSH(3PC)T determined by the constant NPSH method. The value of NPSH(3PC)T obtained at QT is the same for the
two methods.

the affinity rules would apply, giving


x = 2 where: NPSH(3PC) varies with
[pump speed]X; and the flow rate
varies directly with pump speed.
Figure 3 shows test results based on
NPSH(B), the net positive suction head
when cavitation causes the complete
breakdown of the generated head.
Since non-cavitation and NPSH(B) are
the two possible operating extremes,
it is evident that a head-drop criterion for performance comparisons such
as NPSH(3PC) will have an intermediate value. Within the accuracy of the
test results, Figure 3 shows that x = 2
applies for the particular pump tested.
A more-detailed examination of the test
data in Figure 3 has shown that the index
might be as low as x = 1.98. Other pumps,
where a different internal geometry/

cavity volume relationship inevitably


exists, with index values of less than two
have been reported in the literature.
The conclusion is that scaling NPSH(3PC)
data up in speed using x = 2 slightly
overestimates the NPSH required to bring
about comparable conditions. So, assessing
risk, scaling from 3,600 rpm to a higher
speed using a pump speed squared
relationship errs on the side of safety.
Scaling from 3,600 rpm to a lower speed
introduces a small but insignificant error.
Pump speed scaling in the evaluation of cavitation risk using
NPSH(3PC) is valid, where:
(i) NPSH(3PC) varies with
pump speed squared
(ii) Best efficiency flow rate varies
directly with pump speed.
www.worldpumps.com

37

Feature
WORLD PUMPS

April 2011

Scaling to higher water temperatures


2,750
2,700
2,600
2,500

A reliable datum

Pump speed, n

Generated head,

NPSH(B) for each speed obtained at the


corresponding best eciency ow rate

2,400
2,310
2,200
2,100
2,000
1,900
1,800
1,700
1,600
1,500
1,400
1,200
1,000

Net positive suction head, NPSH

4
Test liquid: Water at 20C
Test pump: 80 mm dia. inlet

**
*

* *

* *

*
* *
**

2,700
2,750

2,600

2,500

2,400

2,310

2,200

1,900
2,000
2,100

1,800

*
1,400
1,500
1,600
1,700

1,200

n (rpm)
8

Industrial-quality NPSH(3PC) tests using


water between 2C and 40C provide a
reliable datum for NPSH() values. At higher
temperatures the rapid rate of increase of
vapour pressure presents problems with
regard to the possible variation of temperature within the liquid ow and the accuracy
of the temperature measuring instruments.
The consequences of underestimating the
pumped liquid temperature can be serious.
Temperature measurement at the wall of
the hot inlet pipe may not be representative of that within the main body of liquid.
Many pumps have bypass balancing ows
that return liquid that is heated by losses
in the pump back from the pump outlet
branch to near the inlet branch. Incomplete
mixing leads to hot spots that cavitate rst.
Table 1 shows for water how a temperature
understimate of 2C above 100C rapidly
increases the uncertainty in NPSH values.
The general consensus is that there is a
reduced NPSH requirement when pumping
water at higher temperatures. For pumps
below 5,400 rpm the cold water NPSH(3PC)
values are often much less than 10 m.
The quality of data from hot water cavitation tests in the literature is variable but,
even if a 50% reduction from cold data
was possible, the potential inaccuracy in
temperature measurement at a pump inlet
will always compromise any potential benet.

1,000

NPSH at breakdown, NPSH(B) (m)

38

Pump speed squared, n (rpm x 10-6)


Figure 3. (above) Determining NPSH(B). (below) For the pump tested, NPSH(B) varies with pump speed squared.
d

The pragmatic choice is to use cold water


NPSH(3PC) data for all pump speeds up
to 5,400 rpm and allow benefits that
might accrue from the thermodynamic
properties of water at higher temperatures
to provide a small but unquantifiable
additional safety margin. The extrapolation
of NPSH(3PC) data to deduce acceptable protection measures does err on

Table 1. Additional NPSH to compensate for a water temperature underestimate of 2C


Temperature, T (C)

20

40

...

110

130

150

Vapour pressure (bar)

0.0071

0.0234

0.0738

...

1.433

2.701

4.760

Temperature at T+2 (C)

22

42

...

112

132

152

Vapour pressure at T+2 (bar) 0.0081

0.0264

0.0820

...

1.532

2.867

5.021

Vapour pressure error (bar)

0.0010

0.0030

0.0082

...

0.099

0.166

0.450

NPSH error at T+2 (m)

0.01

0.03

0.08

...

1.1

1.8

5.0

www.worldpumps.com

WORLD PUMPS

Pump speed (rpm)

3,000

Feature
April 2011

30.4 mm
dia.

Minimum NPSH to prevent performance loss = 1.3 NPSH(3PC)

Maximum diameters of notional spherical


cavities when generated head collapses as
inlet pressure falls to give NPSH(B).

2,750 rpm

2,500
28.4 mm
dia.

27.8 mm
dia.

25.6 mm
dia.

2,000

1.18 mm
dia.

2,000 rpm

0.38 mm
dia.

27.0 mm
dia.

1,500

Cavitation surging

24.0 mm
dia.

1,000
Ice

This gure xes the permissible limit of


transient cavitation. Its use is conned to ow
rates where surging cannot occur and to short
periods less than ve minutes where the
much higher NPSH levels needed to protect
from cavitation erosion can be relaxed. Typically this is used to determine the acceptable
limits of inlet vessel pressure uctuations.

20

43

Ice Water

50

103.1
100

143.3
150

Water temperature (C )

Figure 4. Theoretical support for using 240C as a base for NPSH performance projections.

the side of safety and always produces


commercially practical solutions at a
pump speed of less than 5,400 rpm. At
the higher pump speeds, this cautious
approach leads to the NPSH(R) having
a value exceeding tens of metres. There
are ways to provide this: raising either
the height or the operating pressure of
the inlet vessel is one; a second booster
pump operating in tandem is another2.
Note that for pumps running at speeds
of >5,400 rpm, several possibilities may
have to be evaluated to get the best
commercial outcome. These should include
looking beyond the NPSH head-drop
methods described here and evaluating
whether the lower NPSH(R) requirement resulting from the very expensive
and technologically demanding visual
cavitation tests3 are more appropriate.

Cavity growth in water


A theoretical prediction4 of the interplay
between cavity size and pump internal
geometry is shown in Figure 4. The
underlying nding is that cavity growth is
controlled by heat transfer limitations at the
vapour/liquid boundary. This has been shown
to have validity for comparative performance purposes when it is assumed that
the radius of curvature of all the cavitating
surfaces can be represented by the singlevalued radius of a number of spherical
cavities. The results of this work show that:
(i) In the temperature range 2C to 43C the
notional cavity size is large and changes little.
(ii) At 20C cavity size grows slightly
with increasing pump speed.

(iii) At >100C cavity size is much reduced.


(iv) For the same blockage at 143.3C
thousands of tiny cavities are forecast.
The ndings rearm that choosing 2C to
40C as a base for assessments has validity.
It also supports the conclusion that speed
scaling using x = 2 is a safe assumption.
Interestingly, the theoretical results might
also explain why cold water cavitation is often evidenced by a crackling
noise that is not heard at higher water
temperatures. Also, it throws light on
the observation that the most damaging
cavitation erosion occurs with cold water.

Scaling to other liquids


Liquids that behave like water when
cavitating can be considered on a case
by case basis. Denitely outside the valid
usage of NPSH() are liquid metals such
as mercury, which have exceptional heat
transfer properties, and volatile hydrocarbons. Slurries and highly viscous liquids
also invalidate comparisons of cavitation
performance using a head-drop concept.

Cavitation performance boundaries


Hydraulic performance loss
The NPSH(3PC) value at each ow rate
denes the point at which the occurrence
of cavitation reduces the generated head.
The addition of one third of the NPSH(3PC)
value at these ow rates ensures that
pumping performance always remains as
described by the non-cavitating curve:

The potential for damage from hydrodynamic surging is very great when cavitation takes hold. It is prudent not to make a
concession to pump usage where this risk
exists. In general, such surging is found at
ow rates below 30% of the best eciency ow rate. In high-energy pumps
a low-ow protection system is required
to prevent operation below this gure2:
Minimum ow rate to prevent cavitation
surging = 30% of best eciency ow rate
It should be noted that hydrodynamically
induced cavitation surging occurs where the
pumped liquid is free to expand or contract
with temperature. The possibility of therr
modynamicallyy induced cavitation surging
exists in pump pipework systems that can
become closed; for example, where there are
upstream and downstream non-return valves.
This is a safety issue for high-energy pumps2.

Cavitation erosion
Cavitation erosion is a vicious form of
attack that rapidly degrades the materials
used to construct impellers. It is known
to be especially severe when pumping
cold water. The cavities producing damage
usually result from the mismatch of
the impeller blade inlet angle with the
incoming flow at off-design conditions.
Cavitation erosion damage to a pump
impeller can render it unusable within
hours. The worst excesses are frequently
observed at flow rates that are either
much higher than the best efficiency flow
rate or well below it. Commercially viable
boundaries to avoid cavitation erosion
based upon NPSH(3PC) data are definable if continuous operation of a pump
is limited to flow rates within the 50%
to 120% best efficiency flow rate band.
NPSH(3PC) data related to a specified
operational flow rate can be used to
calculate an adequate protection level for
a particular pump. The problem, seen time
and time again, is that pumps very often
operate far from the intended flow rate.
www.worldpumps.com

39

40

Feature
WORLD PUMPS

Systems designed by competent system


designers often get close; those designed
by infrequent pump users seldom do.
The minimum NPSH for continuous
operation within the 50% to 120% best
efficiency flow rate band for pump
speeds less than 5,400 rpm should be
2.5 NPSH(3PC). [NB: Operating experience obtained since the recommendations in Ref. 3 were produced in 1998
has shown that use of the multiplier
2.5 has validity up to 5,400 rpm.]

The encompassing NPSH(3PC)


boundary

April 2011

NPSH() assessing the adequacy of NPSH provision


Stage 1: Pump manufacturer conduct a test to
quantify NPSH() and Q()
1 Conduct a cavitation performance test at a convenient pump speed
nT between 900 rpm and 3,600 rpm.
Determine the best efficiency flow rate QT and NPSH(3PC)T, the net
positive suction head value at that flow rate.
2 Calculate for pump speed 3,600 rpm the values of NPSH() and Q()
using:
NPSH() = 2.5 NPSH(3PC)T (3,600/nT )2 and Q() = QT 3,600/nT

It is evident from the preceding observations that adequate protection in


the form of minimum risk from unacceptable cavitation is achieved if the
following conditions are met:

3 Record the values of NPSH() and Q() for the particular pump.

1. Pump operation is limited to 50% to


120% of the best eciency ow rate value.

1 Data needed to start the assessment:


For the particular application:
(i) Flow rate QG (a continuous operating condition)
(ii) Pump speed nG

2. For pump speeds below 5,400


rpm, the minimum NPSH provided
is at least 2.5 NPSH(3PC).

The limitations of NPSH()


The use of NPSH() has validity in relation
to cavitation protection only. It does not
address the strength of the pump casing
pressure containment or the suitability of the
materials of construction. Validity extends to
water in the temperature range 2C to 150C
and to pump speeds within the range 900
to 5,400 rpm. Other liquids that behave like
water when cavitating can be considered on
a case by case basis. Liquids for which the
use of NPSH() is inappropriate are volatile hydrocarbon mixtures, liquid mercury,
slurries and other highly viscous liquids.

Applying NPSH()
The validity of NPSH() over a wide range
of pump operating conditions has been
demonstrated. The ease of obtaining test data
sucient to characterize cavitation performance has been shown. Finally, as listed in the
box, the way in which it is applied by pump
manufacturers and pump users is shown to
be simple. These three features make the use
of the NPSH performance indicator a potential
cornerstone for cavitation risk assessment in
the everyday usage of centrifugal pumps.

References
[1] ISO 9906:2000, Rotodynamic
pumps Hydraulic performance acceptance tests Grades 1 and 2, (2000).
www.worldpumps.com

Stage 2: Pump user evaluate the NPSH(R) that


minimizes the risk of unacceptable cavitation

For the particular pump


(iii) NPSH()
(iv) Q()
2 Check that the flow rate is within the acceptable best efficiency
band using:
Flow rate % = (3,600 QG)/[nG Q()] 100%
Note: A percentage figure outside the 50% to 120% band makes
using the NPSH() methodology invalid.
3 Calculate the minimum NPSH(R)
For a pump speed nG below 5,400 rpm but greater than 900 rpm:
NPSH(R) = NPSH() (nG/3,600)2

Stage 3: Evaluate the adequacy of the NPSH available


The NPSH available at the pump inlet is determined by the pump
system design. To minimize the risk of unacceptable cavitation its
value must exceed the NPSH(R) calculated in Stage 2.

[2] E. Grist, Power station feedwater


pump specifications; paper presented
at IMechE seminar Power Station
Pumps, Leeds, 2009. [Available from
the Institution of Mechanical Engineers,
London, UK; www.imeche.org/library]
[3] E. Grist, Cavitation and the centrifugal
pump a guide for pump users, Taylor
& Francis, Philadelphia, USA, (1998).
[4] E. Grist, The volumetric performance
of cavitating centrifugal pumps. Part 1:

theoretical analysis and method of prediction and Part 2: predicted and measured
performance, Proc. Instn Mech. Engrs, Vol.
200, pp. 159167 & 168172, (1986).

Contact
Dr Edward Grist
Congleton, UK
Email: pumps@gristgen.co.uk
[Dr Grist was formerly Head of Pumps & Water Turbines,
Central Electricity Generating Board, Gloucester, UK]

You might also like