You are on page 1of 9

12

Paper 2

THE VELOCITY OF WATER HAMMER WAVES

By I. S. Pearsall*

Sudden flow changes in a pipeline cause water hammer waves to be transmitted up the pipe. The magnitude
of these pressure waves is directly proportioned to the acoustic velocity.
The value of the acoustic velocity depends on the bulk modulus or compressibility of the liquid. It is thus
affected by pressure, temperature and gas content of the liquid, as well as by the elasticity of the pipe.
For water, considerable data are available on the variation of acoustic velocity with temperature and
pressure. These are summarized and it is shown that, whereas temperature causes changes of the order of
1 per cent per 5 degC, the variation due to pressure is negligible except at very high pressures.
The presence of free gas causes a considerable increase in compressibility, and it is shown that even as little
as 1 part of air in lo4 parts of water by volume causes a 50 per cent reduction in acoustic velocity. The
damping of the pressure waves, which has an overall beneficial effect, is also greatly increased by the
presence of free gas, and data are given on these effects. Solids in liquid have a similar but less drastic
influence.
Experimental results are given of some tests on two sewage pumping stations in which good agreement was
obtained between theory and experiment.
The elasticity of the pipe also affects the acoustic velocity and a summary is given of the data available for
steel, concrete, and rock-lined tunnels ,with different types of pipe fixing.

INTRODUCTION a Mean pipe diameter.


IN MANY HYDRAULIC SYSTEMS sudden changes or complete E Young’s modulus of material of pipe wall.
stoppages of flow occur. The kinetic energy destroyed F Function representing direct wave.
during these changes is converted into potential energy, f Function representing reflected wave.
which is transmitted through the fluid as pressure waves. g Gravitational constant.
These pressure or water hammer waves are propagated H Static head on pipeline.
through the liquid at the acoustic velocity (or velocity of K Bulk modulus of liquid.
sound). The value of the acoustic velocity depends on the I Distance apart-reinforcement rods.
compressibility of the liquid and its container. Its value P Pressure.
for water and other pure liquids is accurately known, and Q Flow in pipeline.
this value, together with allowance for elasticity of pipe R Reflection coefficient.
walls, is normally used in water hammer calculations. r Radius of pipe.
The acoustic velocity is however affected by temperature, S Salinity in parts per thousand.
pressure and the presence of air or solid matter in the T Temperature.
liquid. It is shown how allowance can be made for these t Time.
factors. V Volume.
V Velocity in pipeline.
Notation X Distance from valve.
A Cross-section area of pipe. Y Proportion of gas (or solid) in liquid by volume.
a Acoustic velocity. a Acoustic velocity ratio
b Pipe wall thickness. Velocity in elastic media
d Pipe diameter. Velocity in rigid pipe at 20°C & atmosphericpressure
B Wave height.
The M S . of this paper was received at the Institution on 16th May S Diameter reinforcement rods.
1965.
* National Engineering Laboratory, East Kilbride, Glasgow. Y Ratio of specific heats (Cp/Cv).
Proc Instn Mech Engrs 1965-66 Vo1180 Pt 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


THE VELOCITY OF WATER HAMMER WAVES 13

v Poisson's ratio. the pressure of the gas. In a liquid, y can usually be ignored
p Mass density. as it is nearly unity for an incompressible liquid. If great
I) Wall fixing constant (see Appendix 2.11). accuracy is required, or if the liquid is compressible, y must
be considered, thus y=1.001, for water at 8"C, so the
Subscr+ts change in a is 1 part in 2000. For sea water y = 1.0207 at
a Air. 30°C, which is sigdicant, but for ethyl ether at 18"C,
B Other fluids or solids. y = 1.32 and the effect is important.
c Concrete. Table 2.1 shows values of K for common liquids.
I Inner.
m Mixture. Table 2.1. Values of bulk modulus at atmospheric pressure
0 outer. for common liquids
R Reflection coefficient.
s Steel. T K
("C) (lb/in*)
T Total.
w Water. Water . 20
BASIC EQUATIONS oil (viscosity 20 &)
Oil (viscosity 500 c/s)
:. 22
22
It can be shown (I) (2)" that the basic equations for the Kerosene . 20 1.91 x 10-5'
propagation of pressure waves through a compressible fluid Paraffinoil . 34 1.76 x 10-6
Mobil A 40 2.56 x 105
in an elastic pipe, if friction is neglected, are Methyl alcohol
Ethyl alcohol
: 20
20
1.76 x 105
1.75 x 106
aH 1 av Carbon tetrachloiide
-=_ -
a x g a t -
* *
(2.1) - 20 1.63 x 105
Considerable differences in values of bulk moduli are
and quoted by different sources. Except where tolerances are
quoted the values used in these tables are therefore approxi-
mate. More comprehensive examination of data for water
is given by Dorsey (19).
where H i s head, t is time, v is velocity, with sign opposite
to that of the distance x, and a is the acoustic velocity. TEMPERATURE EFFECTS
These equations can be solved in the form The acoustic velocity has been measured at a number of
temperatures by a number of workers. Representative
values, as given by Greenspan and Tschiegg (4), are given
in Table 2.2 at atmospheric pressure for distilled water.
Linear interpolation will give errors less than 0.2 ft/s. It
a can be expressed by the polynomial (4), if T is in degrees
where F and f are functions representing the direct wave Celsius,t
and the reflected wave respectively. On a graph of head H, u = 4602.1566 + 16.5144T- 0.190 127T2+ 10.8804 10-4T3
against velocity v, these become straight lines, leading to - 4.765 82 10-'T4 + 9.989 84 10-'T5,
the graphical method of solution (Schnyder-Bergeron giving values accurate to 0.08 ft/s.
Method) (2) (3). The slope of these lines, or the 'pipeline For sea water (between 6" and 17°C) Wood ( 5 ) gives the
a
characteristic', is - ,or if transposed on to an H-Q diagram, equation
g u =4626 + 13.8T - 0.12T2+ 3.73s ft/~,
U
is -, where A is the cross-sectional area of the pipe. where s is the salinity in parts per thousand. Alternative
gA values are given by Wilson (6).
Thus the magnitude of the excess pressure generated
during a flow change is directly proportional to the acoustic It can be seen that the variation of acoustic velocity due
velocity. to temperature is some 1 per cent per 5 degC, which is a
The acoustic velocity can be shown to be, for a homo- significant factor. Table 2.2 and Fig. 2.1 also express the
geneous fluid, as differences as ratios with respect to the speed at 20°C.
P R E S S U R E EFFECTS
Numerous measurements have been made of the acoustic
where y is the ratio of the specific heats of the fluid, p is the velocity, but only a few have considered the effects of both
density and K is the bulk modulus. The bulk modulus is temperature and pressure. Wilson (7) has reviewed the
the reciprocal of the compressibility, defined by the existing information on these effects in distilled water and
relationship supplemented these with experimental results at up to
compressibility = -
K
1= -1 --dv
v dp
. 100°C and 14000 lb/in2. From these results he has
In a gas, K is measured isothermally and is equal to p, t The values are quoted verbatim from the onginal papers. It seems
unlikely that the number of significant figures is justified, in view
of the differences shown by dirfetent workers (Tables 2.2'2.3' and
References are given in Appendix 2.111 2.4).
Proc Instn Mech Engrs 1965-66 Vol180 P t 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


14 I. S. PEARSALL

Table 2.2. Values of acoustic velocity and acoustic velocity a =ao +a,T + a2T2+ a 3 P +a4T4,
ratio for distilled water where + 3.4460 10-2p + 5.3602 10-'p2
a0 = 4602.56
Atmospheric pressure - 12.7600 1 0 - 1 2 ~ ~
a, = 16.4825 +20.1381 10-5p - 3.5439 10-*p2
Temperature Velocity, a Velocity ratio +8.1289 1 0 - 1 3 ~ ~
"C (ft/s) to 20°C
a2 = - 0.1867 - 3.5143 10Vp
0 4602.17 0.9461 + 7.2696 1 0 - 1 0 ~-~1.6696 1 0 - 1 4 ~ ~
5
10
4680.12
4749.32
0.9621
0.9763
a3= +9.4650 + 5.1916 10Vp
15 4810.53 0.9889 - 7.9428 1 0 - 1 2 ~+~1.6687 1 0 - 1 6 ~ ~
20 4864.37 1 .oooo a4= - 2.0703 10-6 - 2.2440 10-lop
25
30
4911.42
4952.23
1.0130
1.0181
+ 3.1862 1 0 - 1 4 ~-~6.0538 10-19~~.
35 4987.27 1.0253 Table 2.3 shows the values of velocity and its ratio to
40 5017.00 1.0314
45 5041.74 1.0365 that at 20°C and at atmospheric pressure for various
50 5061.91 1.0406 pressures and temperatures (see footnote t page 13).
55 5077.76 1.0439
60 5089.57 1.0463
65 5097.64 1.0479
70 5102.10 1.0489
75 5103.19 1.0471
80 5101.08 1.0487
85 5095.97 1.0476
90 5087.90 1.0459
95 5077.10 1.0437
100 5063.68 1.0410

Greenspan and Tschiegg (4) report that the difference in


velocity of sound between distilled and tap water is less
than 0.15ft/s.

deduced the following expression (the standard deviation


0.94! lb io io do o;
TEMPERATURE-'C
60 o; 40 io b
1 0

being less than 19 parts in 10000 from experimental


values). Fig. 2.1. Acoustic velocity ratio as a function of temperature
Table 2.3. Values of acoustic velocity and acoustic velocity ratio as functions of temperature and pressure for distilled water
I
Temperature
"C

0
1
I
14.7
4603.05
2000
4673.53
Pressure lb/in2 abs.
4000
4748.17
6000
4825.90
8000
4906.04
I 10000
4988.06
10 4750.16 4822.94 4897.90 4974.25 5051.71 5 129.92
20 4865.22 4939.83 5015.36 5091.18 5167-23 5243.34
30 4952.96 5029.27 5105.68 5181.66 5257.25 5332.45
40 5017.39 5095.51 5173.20 5249.90 5325.79 5400.89
50 5061.95 5142.20 5221.49 5299.41 5376.08 5451.71
60 5089.44 5 172.15 5253.45 5332.98 5410.86 5487.34
70 5101.90 5187.47 5271.16 5352.59 5432.02 5509.72
80 5 100.85 5189.54 5275.92 5359.65 5440.95 5520.15
90 5087.08 5179.07 5268.38 5354.66 5438.22 5519.46
100 5060.73 5155.97 5248.36 5337.54 5423.89 5507.81

Temperature Pressure lb/in2 abs.


"C
14.7 I 2000 I 4000 I 6000 I 8000 I 10000
0
10
20 I 0.9461
0.9763
1.0000 1 0.9606
0.9913
1.0153 1 0.9759
1.0067
1.0309 1 0.9919
1.0224
1.0464
1.0084
1.0383
1.0621
1.0252
1.0544
1.0777
40
50
i
1.0313
1.0404
1.0461
1.0473
1.0569
1.0631 1 1.0633
1.0732
1.0798
1.0791
1.0892
1.0961
I
I
i.094-I
1.1050
1.1121
1.1101
1.1205
1.1279

1 1 1 1
70
60 1.0486 1.0662 1.0834 1.1002 1.1165 1.1325

1
80 1.0484 1.0667 1.0844 1.1016 1.1183 1.1346
90 1.0456 1.0645 1.0829 1.1006 1.1128 1.1345
100 1.0402 1.0597 ~ 1.0787 1.0971 1.1148 1.1321

Greenspan and Tschiegg (4) report that the difference in velocity of sound between distilled and tap water is less than 0.15ft/s.
Proc Instn Mech Engrs 1965-66 VoZ 180 Pt 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


THE VELOCITY OF WATER HAMMER WAVES 15
Wilson also estimated the systematic and problematic or
errors in measuring the velocity of sound and showed that
at pressures up to 14000 lb/in2 the overall accuracy is
1 part in 10 000. Table 2.4 compares results given by hence
1 l-Y+y.
Table 2.4. Values of velocity of sound (1s assessed by dijfment Ka KT=K
authors at 14-7lblin2 30°C At low pressures and ordinary temperatures the process is
isothermal rather than adiabatic, thus Ka becomes equal to
Distilled water Sea-water the absolute pressure of the gas pa,
35 per cent salinity
~ ~~~
thus
Mathews 1939 4935.7 5062.67
Kuwahara 1939 - 5063.00
Del Grosso 1952 4952.76 5073.17 and the acoustic velocity becomes
Greenspan 1958 4952.23 -
Wilson 1959 4952.96 5072.71
I I
Greenspan and Tschiegg (4) report that the difference in
velocity of sound between distilled and tap water is less
than 0.15ft/s.

different authors. From Table 2.3 it can be seen that over


the range 0 to 2000 lb/in2 at 20”C, the change in acoustic
For gas concentrations of less than 1 in by volume
velocity is some 1-3 per cent. Hence it can be considered
that, for normal pressures encountered in pumping and ypa<(l - Y ) P ~and ( 1 -y)’)pwEpw
turbine installations, the effect of pressure on the acoustic hence
velocity is negligible.
The actual velocities of sound derived by this equation
G A S CONTENTS EFFECTS
are given in Table 2.5 for various pressures at 20”C, whilst
If the liquid contains even a small quantity of undissolved the relative values are given in Fig. 2.2. Values measured
gas bubbles, the acoustic velocity is greatly reduced by Silberman (9) are given for comparison. A proportion of
(5) (8) (9). For a non-resonant system, i f y is the proportion the bubbles may be resonant, but this will be small and
of gas by volume, then the mean density of the liquid and there is no need to use the more complicated equations and
gas is temperature of Spitzer (10)for the resonant case.
pm=YPa+(l-y)pw. . . .
(2.3)
Considering the volumetric changes that occur if we con- .loo-
150 f t GAUGE
sider a pressure wave to compress a volume of water V by P -
I-
80- 100 f t GAUGE
an amount dV, 4
K
> -
then for the water
60-
9 -
dV,= - -
vwdp by definition,
KW ’4 0 -
0
and for the air In
0’ 20- o EXPERIMENTAL V A L U E S ( ~ * ~ )
u -
I I I I I
I , , , , I I I I ,,,,I , , , , ,,,,I
:0-e
1 0 -RATIO
~ OF AIR
10-4 TO WATER 1 0-3
BY VOLUME 10-2

F%. 2.2. Acoustic velocity as a function of gas content

Table 2.5. The effect of gas content on acoustic velocity and acoustic velocity ratio

Pressure
(ft
lb/in2
(abs.)
I Proportion of air by volume
head) I 10-6 I 10-6 I 10-4 I 10-3

0
30
50
100
150
14.70
27.68
36.34
57.98
79.61
II 4810.8
4837.6
4843.9
4851.2
4855.0
1I 0.9889
0.9944
0.9957
0.9972
0.9980
II 4414.8
4608.9
4665.3
4737.8
4772.4
1I 0.9075
0.9474
0,9590
0.9739
0.9810
II 2741.3
3324.6
3558.6
3914.2
4115.6
1I 0.5635
0.6834
0.7315
0.8046
0.8460
II :%::II
10265

1916.7
2184.3
0.2110
0.2839
0,3220
0.3940
0.4490
332.75
455.30
520.50
651.90
763.80
0.0684
0.0936
0.1070
0.1340
0.1570

a. =Acoustic velocity in gas-free liquid (taken at 14.7 lb/in2 and 20°C)


Proc Instn Mech Engrs 1965-66 Vo1180 Pt 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


16 I. S. PEARSALL

M I X T U R E S O F WATER WITH S O L I D S OR Table 2.6. Values of Young's modulus for various pipe
LIQUIDS materials
In sewage pumping installations or in chemical plant,
I
mixtures of liquids or liquids and solids may occur. In
these cases some reduction in acoustic velocity is caused;
the amount is calculable by a similar technique to that
Material
I E
(lb/in2)
V

given in the foregoing section.


Steel . I 29 t o 3 1 x106 0.30
Cast iron . 11.6 to 16.0 x 106 0.25
If the proportion of fluid (or solid) B in water (or second Transite 3.4 x 106
fluid) is y, then the ratio of the acoustic velocity in the Asbestos cement} . .
Aluminiumalloys . 10 x 106
fluid to that in water is Brass . 11.4 to 16 x 106 0.33
-=[ ((1-Y)KB Concrete
.
. 2.8 to 4.3 x 106 0.1 to 0.3
a"
a0
PwKB
+(I -Y>PW>
+YKW}(YPB I'
where subscript B refers to the fluid (or solid) B and sub-
Copper
Glass
Lead
.
Soft rubber
15.6 to 19 x 106
9.9 x 106
0.7 to 2.4 x 106
0.1 to 1.04 x 106
0.34
0.24
0.44
0.46 to 0.49
Rocks*
script w refers to water (or second fluid). It can be seen Sandstones . 0.285 to 0.70 x 106
that, unless KB and Kw are very different, this ratio will Quartzites . 3.55 to 6.5 x 106
Schists . 0.94 to 2.7 x 106t
be about 1. Plastics
ABS . . 2.5 x 105
P I P E WALL ELASTICITY PVC rigid . 3.5 to 4 x 106
Nylon . 2 to 4 x 105
The effect of the pipe walls is similar to that of compressible Phenalic glass reinforced 34 x 105
gas bubbles, in that the elasticity of the walls reduces the Polvethvlene . 0.21 to 0.27 x 105
wave velocity. Thus if we imagine a section of pipe of
radius r (diameter d and thickness b) and length 6x being
subjected to a pressure wave Sp, then the increase of
volume in the pipe is
*
Poljrpripylene
PTFE .
.
:I 1.4 to 1.7 x 105
0.5 x 106

Lower value corresponds to normal to strata, anc higher value


corresponds to parallel to strata.
t Grouting of the rock improved this to 3.45 x 106.
dVp =
2ddpSxT
(see Appendix 2.1)
by Linton (11). For thick-walled pipes or pipes of concrete
or, as the original volume V = m 2Sx, or other materials, some revision of this basic formula is
necessary. A summary of the information from different
d
dVp=Eb Vdp. authors has been collected in Appendix 2.11 for reference.

The decrease in volume in the water is (by definition) A C O U S T I C VELOCITY-GENERAL EQUATION


The adoption of a general equation is to some degree
dVw= --VdP dependent on the form in which data are available. The
Kw
temperature (and pressure) effects on the water alone are
therefore the total change in volume is presented in Tables 2.2. and 2.3. It is then easiest to allow
dVp -dVw= ( t-b + - Vdp,iw) for these effects by utilizing the appropriate data. The other
effects, however (gas content and wall elasticity), have been
incorporated by revision of the bulk modulus, thus
and, if we imagine an effective K value for the pipe and
water,

and
1 dVT-dVl+dV2+dV3

e)
-

(A+&) =-Vdp Y1
-+-+-...
Y2
(Kl K 2 K3
Y3

where K,, K,, K3, . . . Kn are the bulk moduli of the


Hence the velocity of sound becomes
respective elements in the fluid, or the equivalent bulk
+-i b ) } ' . . . modulus of the walls, and yl, y2 . . .y n are the proportions
a"= p
{ -
(iw
(2.5)
of the elements. The equivalent bulk modulus of a wall is
or the ratio to water in a rigid pipe is -
d for a simple wall, or one of the functions given in
K, Eb
- = ( 1 + = )dKw
a" -* Appendix 2.11 for more complicated walls.
ao Thus
Values of E for various common pipe materials are given
inTable 2.6. Values for pipes to B.S. Specificationsare given
Proc Instn Mech Engrs 1965-66 Vo1180 P t 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


THE VELOCITY OF WATER HAMMER WAVES 17

and then the ratio of reflected to incident waves (reflection


Yi
a'= pm -
{ +-Y2 ...-+-
(Ki K2
Yn
Kn A)}-' . (2.6)
coefficient) is 0)
R2= pmam - p wa ( pmam + pwaw )
The ratio compared to the plain primary fluid is
hence the percentage of energy reflected can be related to
x, the proportion by volume of air in the water (Fig. 2.3).

where pm is the mean density of the mixture, ;100-

pm=YiPi + ~ 2 p 2 - - - Ynpn, - - (2.8) z


w0 80-
ATMOSPHERIC

and pw and Kw refer to the primary fluid, for instance, 5 0 f t GAUGE


water. 100 11 GAUGE
Alternatively, the acoustic velocity of the fluid can be z
2 40- 150 f t GAUGE
expressed as o
t- -
w
2 20-
w
L -

where y l , y 2 etc. are the proportions of the different ele- RATIO OF AIR TO WATER BY VOLUME
.
ments, such that y 1+ y 2 + y 3 . . y n = 1, and pm is the mean
Fig. 2.3. Reflectionfactor as a function of gas content
density of the fluid, equation (2.8).
The procedure is therefore, firstly to determine the The transmission is then 1 -R2, and if we assume the
relevant wave velocity in the pure liquid (Tables 2.2 and waves follow an exponential decay, the percentage trans-
2.3 for water). This value can then be amended by use mission is equal to the percentage decay of the pressure
of equation (2.7) to allow for gas bubbles, solid inclusions, wave. Referring to Fig. 2.4
mixtures of liquids, and pipe wall elasticity.
1 -RZ,&.
ATTENUATION O F P R E S S U R E WAVES PI
Any elasticity in the system will, in addition to reducing The above theory neglects frictional losses in the piping
the acoustic velocity, cause attenuation of the water system, which are small. These can, if necessary, be esti-
hammer waves owing to hysteresis in the energy cycle. In mated from conventional water hammer theory and be
most practical cases the first wave is the greatest in magni- included in the total attenuation.
tude and for this reason most attention is concentrated on
it. Attenuation caused by wall elasticity is negligible com-
pared to that caused by fiiction and other losses in the pipe
system. Attenuation caused by fluid elasticity is significant
in the case of a compressible fluid, or fluid containing gas
bubbles.

Attenuation caused by free gas bubbles


Most of the work done on attenuation of pressure waves in
liquids by gas bubbles concern resonant systems (9) (10). Fig. 2.4. Wave decay curve
Attenuation under these conditions is larger than in non-
resonant systems. Some of the bubbles in a pipeline will Attenuation caused by liquid and solid mixtures
correspond to resonant size, but the majority will be non- The further attenuation caused by a second component
resonant. will be small unless that component is very compressible.
From equation (2.3) the mean density is The theory given in the previous section is applicable to
pm =YPa + (1 - Y ~ W
this case if it is desired to calculate attenuation.
and the acoustic velocity EXPERIMENTAL RESULTS
[{ Acoustic velocity
a'= Kwpa
(1 - Y ) P ~+yKw 11 YPa + (1 -Y)PW
The acoustic resistance of the substance is pma' equal to
11'. The effects of pressure and temperature on the velocity in
pure water are based on experimental determinations and
confirmation can be found in Greenspan and Tschiegg (4)
and Wilson (7). The effects of gas content are also fairly
well established in the physical sense. Measured values are
plotted, where available, in Fig. 2.2. from Silberman (9).
If we assume that the layer of bubbly mixture is thick, Most water hammer tests reported have been conducted
Proc Instn Mech Engrs 1965-66 Vol180 Pt 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


18 I. S. PEARSALL

on pure water systems, where very often the head is high It will be noted that although the total gas content of the
and it is unlikely that free gas is present, and no authors pipeline will remain substantially constant for any one test,
have reported unusual values of wave velocity. The origin the percentage value will increase as the lower pressure
of the present investigation was two sets of tests under- wave is transmitted. Hence one would expect a difference
taken on sewage pumping stations at Kingston (12) and between the positive and negative waves which is evident
Derby, in both of which considerably reduced values of in Figs. 2.5 and 2.6.
Table 2.7. Expm'mental wave velocity and attenuation results
- -
(11 Volume
Calculated air to Calculated Experimental
wave water Measured teflection ransmissiol decay
Test Nos and type of test
velocity
in water
wave -I
I
velocity (Fig. 2.4) 1 from a2
(Fig. 2.2)
gas
content
:oefficieni
R2
1-R2
(per cent)
82/81
(per cent)
Static
head
Kingston tests
K2, K10, K11. No 2 pump
stopped 3770 1844 34 26 10.49 4.5 x l O - 4 - 16 84 76.5 45
K4, K8. Nos 1 and 2 pump!
stopped 1844 45.2 35 0.49 4.5 x - 16 84 77.5 45
K13. Pumps 4,5 and 6
stopped 2340 59 48 0.62 2 x ~ O - ~ - 8 92 81.5 45
K17. Pumps 5 and 6 stoppec 2510 43.5 35.8 0.66 1.06 x - 6 94 82.3 45
K20. Pump 5 stopped 2560 34 26 0.68 1.05 x10-* - 5 95 76.5 45
----
Derby tests
Al. Pump 1 stopped from
750 revlmin 3090 697 28 19.5 0.226 1.5 x - 37 63 69.6 29.5
A3. Pump 1 stopped from
738 rev/min 888 36 23.5 0.287 1 x 10-3 - 31 69 65 29.5
A4. Pump 1 stopped from
625 revlmin 661 24 12.5 0.214 1.8 x 10-3 ~ . x2 10-3 40 60 52 29.5
A6. Pump 1 stopped from
550 rev/min 692 28.5 18.5 0.224 1.6 x 10-3 1.2 x 10-3 38 62 65 29.5
A7. Pumps 1 and 5 stopped
from 580 revlmin 545 58.7 27.5 0.176 2.7 x 10-3 1.2 x 10-3 43 39 47 29.5
A10. Pumps 1 and 5 stopped
from 700 revlmin 870 47.5 303 0.282 1 x 10-3 1.7 x 10-3 31 69 64 29.5
C1. Pump 5 stopped from
755 revlmin 418 26.5 11.5 0.135 3.8 x 10-3 saturated 54 46 43 29.5
C3. Pump 5 stopped from
750 revlmin 503 31.5 14.5 0.163' 3 x 10-3 saturated 48 52 46 29.5
C6. Pump 5 stopped from I
750 revlmin 550 30.5 14.8 0.178 2.7 x 10-3 1.5 x 10-3 43 57 48 293
-
wave velocity were observed (1850 ft/s in the former and
600 fils in the latter).
No measurements of gas content were made at Kingston
but some measurements were made at Derby although, . STATIC .
because of the high proportion of gas, great accuracy was w LEVEL

unattainable. Table 2.7 shows the calculated values of wave


velocity for these stations (assuming pure water and allow- ~ o " " " " "5 " " " " " 10 15 20
TIME-s
ing only for the elasticity of the pipes) compared to the
measured values. Also included are the estimated gas Fig. 2.6. Derby test A4, pump 1 stopped from 42.6 cusecs
content from these values, as well as attenuation data as
described in the next section. The results from these tests support the theory although,
as the exact gas content was not known, it is not possible
to know how accurate is the agreement. However, it is seen
- 80 -
that variation in gas content may cause reduction in wave
velocity of as much as 75 per cent.
As far as the elasticity of the pipe walls affects the wave
velocity, the main difficulty is in estimating the type of
fixing and the particulars of the pipe wall. Comparison
between theory and practice has shown good agreement
I I (13) (14) (15) even where the elastic properties are not
4'0 20 30 40 50 60
TIME-s known exactly. Any errors from this source are likely to
Fig. 2.5. Kingston test K2, pump 2 stopped from 6 cusecs be less than 2 per cent, which is acceptable.
Froc Instn Mech Engrs 1965-66 V01180 Pt 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


T H E VELOCITY OF WATER HAMMER WAVES 19

Attenuation have been found to substantiate the theory. In addition a


In normal water hammer calculations, attenuation is based section has been included giving results of tests on two
on frictional losses in the system and elasticity of the walls sewage pumping stations, to substantiate the theory on the
is ignored. There is no experimental or theoretical basis for presence of free gas. This appears to be a very important
supposing this leads to serious errors. However, where free factor in this type of installation, and has not previously
gas is present this is not the case and some eight tests on been noted.
sewage pumping stations are used to substantiate the Little mention is made of fluids other than water but all
theory in the previous section. the experimental and theoretical knowledge indicates that
In both sets of tests at Kingston and Derby the exact these follow the same laws, although uncertainty as to the
gas content was unknown, although at Derby some measure- bulk modulus may limit the accuracy of such calculations.
ments were made. However, knowledge of the wave
velocity, found from the periodic time, allows an estima- APPENDIX 2.1
tion of the gas content to be made, using Fig. 2.2, which Considering an element of water in a pipe, subject to a pressure
can then be used to calculate the attenuation. A comparison wave of magnitude 6p over a length ax, then the water will
between calculated values by this method and measured compress by a certain amount, by definition, equivalent to
values gives, on the whole, good agreement, Table 2.7. In dVw= --.SP V . . . . (2.9)
addition, the estimated free-gas content is substantiated to K
some extent by measurements on site. No allowance has The change in hoop stress will be
been made for solids content as this was less in both
magnitude and elasticity than the gas. (2.10)
Clearly the methods of calculation are approximate for, 6r
as well as experimental errors, the wave velocity and so that the strain - (neglecting axial stress) is
attenuation will vary as the pressure in the liquid varies. 6r -rap
- _- (2.11)
Thus one would expect a decrease of free gas in the r Eb‘ * . ’ ’

upsurges and an increase in the downsurges, hence the The increase in volume is
wave velocity would be higher and the attenuation less in 2rrr6r6x.
the upsurges than in the downsurges. Also one would Substituting for 6r from equation (2.3),
expect the free-gas content to decrease as the mean head 2Tr3
increased and the surges damped out. There is some increase in volume = --
Eb
6p6x. . (2.12)
evidence to support these observations but it is not con-
Axial strain will alter this value in a ratio varying from approxi-
clusive. mately 0.92 to unity according to the conditions.
This indirect experimental support for the theory seems
to confirm its usefulness, and show the major effect of free APPENDIX 2.11
gas in the liquid.
THICK-WALLED, CONCRETE AND ROCK-LINED T U N N E L S

CONCLUSIONS Perhaps the best summary of the effect of pipe fixture, and thick-
walled and rock tunnels is given by Halliwell (16),and is sum-
A summary has been given of the factors affecting the rate marized below. Concrete pipes are considered by Kennison (14).
of propagation of pressure waves in a piping system. It is Tests on a transite, or cement asbestos, pipe are reported by
dependent on the bulk modulus of the fluid and the Kessler (IS).
elasticity of the walls, and its prediction depends on the Table 2.6 contains a list of E values that can be used in equation
(2.5) for thin-walled pipes. If, however, the pipe has thick walls,
accuracy with which these are known. Even for water there or is made of a composite material (concrete), or is in rock, this
is considerable disagreement on exact values for bulk equation can no longer be used.
modulus but for engineering purposes the values are The general equation for such a case is (16)

- +-,:I}-*
sufficiently accurate.
Three factors can be computed separately. These are the a”= { (A
p . . (2.13)
variation of liquid bulk modulus with temperature and
pressure presented here as the change of velocity of sound
in which
* =E(2cr + c,)
in water with temperature and pressure, the effect of pipe where Zr represents the circumferential strain caused by unit
wall elasticity allowable by Young’s modulus of the material pressure inside the pipe, the longitudinal strain caused by unit
pressure rise inside the pipe. E’ is the Young’s modulus of the
and by a factor representing the method of fixture of the material in the pipeline immediately adjacent to the water.
pipes, and finally, the presence of free gas, allowance for
which can be made by modifying of the bulk modulus of Thick-walled pipes
the liquid. Strictly one should use a combined equation Halliwell’s results can be best expressed by the generalized
of the form of equation (2.6). However, in most cases a equation
satisfactory approximation can be made by using equa- A d b
+ = A -do2 +d12 + B = - - + -
tions relating the velocity to that in pure water for each do2 -d12 2( b a)+B
case, and then to combine the total effect. where do, d~and 2are respectively the outer, inner and mean pipe
Reference is made to sources where experimental data diameters, and b is the wall thickness.
Proc Instn Mech Engrs 1965-66 Vol 180 Pt 3 E

Downloaded from pcp.sagepub.com by guest on January 19, 2015


20 I. S. PEARSALL

The coefficients A and B are defined as follows. Equation (2.5) is then used with this value of b and E for steel
pipe.
For higher pressure pipes, a thin steel membrane is often used.
Under these circumstances the material has two portions to its
Pipeline anchored against longitudinal
movement throughout its length I 2(1 -v*) 1 2v(l +v)
stresslstrain curve, initially, where the concrete is taking the load
and, secondly, where the majority of the load is taken by the steel
reinforcement cylinder. Kennison (14)shows that by these means
Pipeline anchored against longitudinal decrease in effective surge velocity of 30 per cent can be effected.
movement only at end remote from H e suggests using a mean value based on a mean value of the two
valve curves.
His data refer solely to pipes made to U.S.A. Waterworks
Pipeline with frequent expansion joints
1 2 1 2 v standards, for reinforced and prestressed pipes. The reinforce-
ment in all cases being of a cylinder type.
Thin-walled pipes A P P E N D I X 2.111
In this case a Ail
= 0 and 4 = - -
- 91, and radial stress is 0 so that B REFERENCES
b 2 b’ JAEGER, C. Engineeringfluid mechanics 1956 (English transl.,
values at A being taken from the previous table according to the Blackie, London).
end fixture. BERGERON, L. Water hammer in hydraulics and wave surges
Pipeline through solid rock in electricity 1961 (Wiley, New York and London).
d
This case can be covered by putting d o = to or - = 1 hence
SCHNYDER,0. ‘Druckstosse in pumpensteigleitungen’
b (Water hammer and discharge mains of pumps), Schweiz,
* = A + B = 2 ( 1 +v). Bauztg 1929 94, 271, 283.
GREENSPAN, M. and TSCHIEGG, C. E. ‘Tables of the speed
Lined tunnels of sound in water’, J. Acoust. SOC.Amer. 1959, 31, 75.
(i) Steel and concrete lining WOOD,A. B. A text-book of sound. Revised edition 1955
If the steel lining has a mean diameter isand thickness b, the (Bell & Sons, London).
concrete lining between it and the rock has mean diameter d and WILSON,W. D. ‘Speed of gas in sea water as a function of
thickness b. Halliwell’s expression simplifies to temperature, pressure and salinity’, J. Acoust. SOC.
Amer. 1960 32.
(1 - ”2) WILSON,W. D. ‘Speed of sound in distilled water as a
4=
function of temperature and pressure’, J. Acoust. SOC.
lbs 1-v E, Amer. 1959 31, 1067.
RIPKEN,J. F. and OLSEN,R. M. ‘A study of the gas nuclei
1 -2v on cavitation scale effects in water tunnel tests’, St.
Anrhony Falls Hyd. Lab. Proj. Report No. 58. Minnea-
In this equation Poisson’s ratio v is assumed to be the same for polis : University of Minnesota, 1958.
steel, concrete and rock. SILBERMAN, E. ‘Some velocity attenuation in bubbly mix-
tures measured in standing wave tubes’, 3. Acoust. SOC.
(ii) Concrete lining Amer. 1957 29, 925.
For a concrete lining alone the above equation can be used with SPITZER,L. ‘Acoustic properties of gas bubbles in a
bs = 0 , hence liquid’, Columbia Univ. O S R D Report No. 1705, Sect.
6.1/Sr20-911. New York 27, N.Y.: Columbia Univer-
sity, July 1943.
LINTON, P. ‘Notes on pressure surge calculations by the
graphical method’, B H R A Publ. No. TN477.
WHITEMAN,K. J. and PEARSALL, I. S. ‘Reflux-valve and
Substantially Halliwell’s equations agree with other workers, surge tests at a station’, Fluid Handling, 1962 (152), 248
the cases where they do not are cited in his paper, being generally (153),
. .~282.
due to over-simplification of equations by these workers. PARMAKIAN, J. Water hammer analysis, 1955 (Prentice Hall,
New York: Bailey and Swinfer, London).
Reinforced concrete pipes KENNISON, H. F. ‘Surge-wave velocity-concrete pressure
For concrete pipes used with reinforcement bars, then Kennison pipe’, Trans. Am. SOC.mech. Engrs 1956 78, 1323.
(14)states the strength is limited by the tensile strength of the KESSLER, L. H. ‘Speed of water hammer wave in transite
concrete and the modulus of elasticity is essentially that of the pipe’, Trans. Am. SOC.mech. Engrs 1939 61, 11.
concrete. Parmakian (13),however, makes an allowance for the HALLIWELL, A. R. ‘Velocity of a water hammer wave in an
reinforcement bars, by quoting an equivalent pipe thickness elastic pipe’, Proc. Amer. SOC.civil Engrs (J. Hyd. Div.)
1963 89 (Hy4), 1.
GIBSON,A. H. Hydraulics and its applications 5th edition
1952 (Constable, Londoii).
DUNCAN,W. J., THOM, A. and YOUNG,A. The mechanics
where 6 is the diameter of the reinforcing rods and 1 the distance of fluids 1960 (Edward Arnold, London).
apart (the factor J6 represents the ratio of the modulus of elasticity DORSEY,N. E. Properties of ordinary water substance
of concrete to that of steel (& to $6) with a margin of safety. 1940 (Reinhold Publ. Corp.).

Proc Instn Mech Engrs 1965-66 Vol180 P t 3E

Downloaded from pcp.sagepub.com by guest on January 19, 2015

You might also like