5.1 lntroduction
\Vater distribution network analysisprovides the basisfor the designof new systemsand
the extension of existing systems.Design criteria are that specifiedminimum flow rates
and pressureheadsmust be attained at the outflow points of the network. The flow and
pressuredistributions acrossa network are affectedby the arrangementand sizesof the
pipes and the distribution of the outflows. Sincea changeof diameter in one pipe length
will affect the flow and pressuredistribution everywhere,network designis not an explicit
process.Optimal design methods almost invariably incorporate the hydraulic analysis
of the system in which the pipe diameters are systematicallyaltered (see,for example,
Featherstoneand El Jumailly, 1.983).
Pipe network analysisinvolves the determination of the pipe flow rates and pressure
headswhich satisfythe continuity and energyconservationequations.Thesemay be stated
as follows:
(i)
Continuity: The algebraic sum of the flow rates in the pipes meeting at a junction,
together with any external flows, is zero:
I:Np(/
Q'tFl:o'
/1'NJ
ts.1l
I:1
(ii)
in which Ql7 is the flow rate in pipe IJ at junction /, NP(/) the number of pipes
meeting at junction I, FI the external flow rate (outflow) at / and NJ the total
number of junctions in the network.
EnergyconseruAtion:Thealgebraicsum of the headlossesin the pipes,togetherwith
any heads generatedby inline booster pumps, around any closed loop formed by
pipesis zero.
/:NP
(1)
h r g  I d o 1 1: 0 ,
/ : 1,NL
15.21
I :1,
in which hy11 rs the head loss in pipe / of loop I and I{o.g is the manometric head
generatedby a pump in line {.
PipeNetwork Analysis
117
'When
the equationrelating energylossesto pipe flow rate is introducedinto Equations
5.1 or 5.2, systemsof nonlinearequationsare produced.No method existsfor the direct
solution of such setsof equationsand all methodsof pipe network analysisare iterative.
Pipe network analysisis therefore ideally suited for computer application but simple
networks can be analysedwith the aid of a calculator.
The earliest systematicmethod of network analysis,due to ProfessorHardyCross,
known as the head balanceor 'loop' method is applicableto systemsin which the pipes
form closedloops. Assumedpipe flow rates,complying with the continuity requirement,
Equation 5.1.,are successively
adjusted,loop by loop, until in every loop Equation 5.2
is satisfiedwithin a specifiedsmall tolerance.In a similarlater method, due to Cornish,
assumedjunction headelevationsare systematicallyadjusteduntil Equation 5.1 is satisfied
at everyjunction within a small tolerance;it is applicableto both open and closedloop
networks. Thesemethodsare amenableto desk calculationbut can also be programmed
for computer analysis.However convergenceis slow since the hydraulic parameter is
adjustedat one element(either loop or junction) at a time. In later methods systemsof
simultaneouslinear equations,derived from Equations 5.1 and 5.2 and the head lossflow rate relationship,are formed, enablingcorrectionsto the hydraulicparameters(flows
or heads)to be made over the whole network simultaneously.Convergenceis much more
rapid but sincea number of simultaneouslinear equations,dependingon the sizeof the
network, have to be solved,thesemethods are only realisticallyapplicableto computer
evaluation.
The majority of the worked examplesin this chapterillustratethe useof Equations5.1
and 5.2 in systemswhich can be analysedby deskcalculationusingeitherthe headbalance
or quantity balancemethods.In addition to friction losses,the effect of local lossesand
booster pumps is shown. The networks illustratedhave been analysedby computer but
the intermediatestepsin the computationshave beenreproduced,enablingthe readerto
follow the processas though it were by deskcalculation;the numbershave beenrounded
to an appropriatenumber of decimalplaces.An exampleshowing the gradientmethod is
also given.
!a
L
e)
*,
CL
G
7
lr,
b: KA
118
Hydraulics
CivilEngineering
Inflow
h : K ( Q + A Q ) 2: K l e + z Q L g + A d l
Neglecting LQ2 and assumingAQ to be small,
b  K(C +zQ^Q)
Now round a closed loop I h  0 and AQ is the same for each pipe to maintain
continuity.
l n  I K C + z ^ Q D K Q :o
+. 'A\rN
which may be written as AQ
estimatedflow Q.
D'KE  D,KE
zyxglg
zDxg:
EL,
rl
Figure 5.2
PipeNetworkAnalysis 1 1 9
If ZI is the true elevation of the pressurehead at I , the head loss along each pipe can
be expressedin terms of the differencebetween Z7 and the pressurehead elevation at the
other end.
For example: hy,61  Zn  21.
Expressing the head loss in the form h : Kg?, N such equations can be written as
(where N is the number of pipes)
$
l)^'l
f zo Ztl : r(srcN)KnrI a_ni
K)s r ( l S r l )I'
 4  Z t r _
 (SIGN
r
I
Ir : l l : l
15.31
Ktteu t),)
lz,  zt): LtsrcNl
and in general,(SIGN) is * or  accordingto the sign of (Zr  4). Thus flows towards
the junction are positive and flows away from the junction are negative.
K17 is composedof the friction loss and minor loss coefficients.
The continuity equation for flow rates at / is
IQtr
_ F Qqi+Ql+Qc7*S.rF:0
t5.41
Examination of Equations 5.3 and 5.4 shows that the correct value of 21 will result in
valuesof Qrl, calculatedfrom Equation 5.3, which will satisfyEquation 5.4.
RearrangingEquation 5.3 we have
t'1
zr
t O r l  l l s r c N(lY 5  r\tt r
ts.sl
L\tttl/J
The value of 21 can be found using an iterative method by making an initial estimateof
21, calculatingthe pipe dischargesfrom Equation 5.5 and testing the continuity condition
in Equation 5.4.
If (I Qtl  E) + 0 (with acceptablelimits), a correction LZ1 is made to 21 and the
procedure repeateduntil Equation 5.4 is reasonablysatisfied.A systematiccorrection for
L^21 can be developed:expressingthe head loss along a pipe as h:KQIL,for a small
error in the estimate ZT rthe correction L,Z1 can be derived as
aZ1 
z ( D Q r y F )
D Qr1lhlt
Example 5.7 shows the procedure for networks with multiple unknown junction head
elevations.
Evaluation of K77:
K1
,:ffi.#(:Kr+K)
where C is the sum of the minor losscoefficients.), can be obtained from the Moody chart
using an initially assumedvalue of velocity in the pipe (say 1 m/s). A closerapproximation
to the velocity is obtained when the discharge is calculated. For automatic computer
120
Civil EngineeringHydraulics
15'61
For each pipe, h1,11(friction head loss) is initialised to'21 .21, Qr1 calculatedfrom
Equation 5.6 and h1.77reevaluatedfrom hr,rt : Vr  4)  X^$t. This subroutine
follows the procedure of Example 4.2.
l,
ls.7l
where
[A11] : diagonal matrix of NT x NT dimension, defined as
tA11l
ut+ 6 ,
l"''n'"'";

Lo
0
otQY',t'+pr+&
:
0
0
0
o*t*i't'
+ pr, *
lI
ffil
ts.8l
lAzrl[Q]: [q]
t5.el
PipeNetworkAnalysis 121
where l ztl is the transposematrix of [A12] and [q] water consumption and water supply
vector in each node with NN x 1,dimension.
In matrix form, Equations5.7 and 5.9 are
tlA11ltA12ll
[ta]l  [lnt0]tnoll
LtA21ltol I LtHll L Lql I
[s.10]
The upper part is nonlinear, which implies that Equation 5.10 must use some iteratiue
algorithm for its solution. Gradient method consistsof a truncated Taylor expansion.
Operating simultaneouslyon ([Q], [H]) field and applying the gradientoperator,we can
write
r+3rl
tfq3ll: liqr]l
[LNlnnI
I
L IA21l LolI LIdHl.l L[aql
Ls11l
where [N] is the diagonalmatrix (nr, nz, ..., nNr) with NT x NT dimensionand [,{1L]'
: NT x NT matrix defined as
fo''Q'i'')
lou2q(nzr)ool
a3Qlzt)
o
o
tAlll':l
l::::::'l
L;
I
I
rs.1,2l
:::"*,ok'',1
 In any iteration i, [dE] is the energy imbalance in each pipe and [dq] is the discharge
imbalancein eachnode. Theseare given by
ldEl : [A11][Qi] + lLtzllHil + [A10][H0]
[s.13]
la
L
q,
t
and
ldql:tA2lllQtl tql
[s.14]
l d Q l: [ Q i + r ]  t Q i l
[s.1s]
[s.16]
and
1]')1[A12]]1{lA21l(tNl[A11]',)1
lHi+rl {[A21](tNl[A1
 tql]
 ([A21]tQrl)
(tA11l[Qi])
+ [A10][Ho]
ls.17l
[Qi+r]: {[r] (tNl[A11]') tA11l]tQ;l {(tNllA11l')1(tA12l
[s.18]
[Hi+r]+ [A10]lHol)]
The method has the advantageof fast convergenceand doesnot needcontinuity balancing
in eachnode to beginthe process.The method is not suitedfor hand calculation.Example
5.8 illustratesthe methodology.
E
G
a
t
122
C i v i lEn g i n e e ri n gH y d ra ul i cs
lg'
L20
220
10(
100
lx,
l3'
50
10
\7
t?,
ll,
Worked examples
Exomple
5.1
Neglecting minor lossesin the pipes, determine the flows in the pipes and the pressure
headsat the nodes(seeFigure5.3).
Data
t:
Length(m)
Diameter(mm)
600
250
;
600
150
200
100
600
150
200
200
600
150
200
100
Roughness
sizeof all pipes: 0.06 mm
Pressure
headelevation
at A:70 m o.d.
Eleuationof pipe nodes
XoA* ;
Elevation (m o.d.)
25
20
20
22
25
Procedure:
'Sfhen
1. Identify loops.
using hand calculation the simplest way is to employ adjacent
loops, e.g. Loop 1: ABEFA; Loop 2: BCDEB.
2. Allocate estimatedflows in the pipes.Only one estimatedflow in eachloop is required;
the remaining flows follow automatically from the continuity condition at the nodes;
e.g. since the total required inflow is 220 Lls, if Qns is estimated at'1,20 L/s, then
Qar  100 Lis. The initial flows are shown in Figure 5.3.
3. The head loss coefficientK  ),Ll2gDA2 is evaluatedfor eachpipe, I being obtained
from the )" vs. Re diagram (Figure 4.2) correspondingto the flow in the pipe. Alternatively,Barr's equation (Equation4.1,2)may be used.
123
If the Reynolds numbers arc fatly high (+10s), it may be possibleto proceedwith
the iterations using the initial ,1.values, making better estimatesas the solution nears
convergence.
The calculations proceed in tabular form. Note that Q is written in litres per second
simply for convenience;all computations are based on Q in cubic metres per second.
However, hlQcould have been expressedin m/(L/s) yielding AQ directly in litres per
second.
D,
Pipe,
AB
BE
fLooP 4
t
EF
FA
Re!
0.00024 120.00
10.00
0 .0 0 0 6 0
o.ooo4o 60.00
0.00030  1 0 0 . 0 0
0u),
5.4L
1,.3L
4.51
s.63
hl',
li
1.1..48
0.01.57 797.0
3.39
0.0205 33877.0
0.0172 1.1229.140.42
0.01.62 336.6 8.36
t
+ AQ:
rLooPz^
Dh
2LhIQ
BC
cD
DE
EB
(33.91)
:0.0t423 :
2 x 1.1,91,.82
5 0 .0
10.0
_zo.o
24.23
3.76
1,.13
1.50
2.73
9s.64
338.77
673.75
83.66
33.91. 1r9L.82
t 4 . 2 3L t s .
0.01.74
0.0205
0.0189
0.0189
28.40
l l 359.7
3.39
33 877.0
1.2338.9 4.94
31,232.9 18.34
8.51
s67.98
338.77
246.78
756.77
1910.30
+ AQ: 2.23L1s.
(Note that the previouslycorrectedvalue of flow in the 'common'pipe EB has beenused
in Loop 2.)
Loop t
Pipe
Q (L/s
AB
BE
EF
FA
L 3 4 .23
2 6 .46
4 5 . 77
8 s .77
Ri x tOs;
6.05
2.98
3.44
4.83
h trril
7 9 1 , . 9 1,4.27
0.0156
0.0188 31.067.7 21..75
L1,424.9 23.93
5
0.01.7
6.23
846.9
0.01.64
5.86
t,
106.30
822.0s
522.92
72.64
L523.91,
= +A Q :  1 . 9 2 L 1 s .
Proceedto loop 2 agarn,and continuing in this way the solution is obtained within the
required specifiedlimit on lh in any loop after severalfurther iterations. The solution
given is obtainedfor Dh .0.01 m but an acceptableresultmay be achievedwith a larger
tolerance.
12 4
C i v i l E n g i n e e ri n gH y d ra ul i cs
heads
Presswre
Final ualues
Pipe
Q (L/s)
h (il
AB
BE
FE
AF
BC
CD
ED
131.s5
25.02
48.4s
8 8 .4 5
4 6 .5 3
6.55
2 3 .4 7
1,3.70
19.55
26.67
6.s9
24.74
1,.52
6.69
Node
Pressurehead (m)
40.00
31,.29
1.t.57
10.05
1,4.74
38.41,
A
B
c!
D
E
F
Example5.2
In the network shown in Figure 5.4 a valve in BC is partially closedto produce a local
head loss of 1,0v!r12g. Analysethe flows in the network.
AB
500
250
Pipe
Length(m)
Diameter(mm)
BC
400
150
CD
200
100
DE
400
150
BE
200
150
EF
600
200
AF
300
2s0
'n
.5
E
tn
Solution:
The procedureis identical with that of the previous problem. Kss is now composedof the
valve loss coefficientand the friction loss coefficient.
\7ith the initial assumedflows shown in the table below, Sc : 50 L/s; Re : 3.7 x l}s
k1D  0.0004;)"  0.0174 (from the Moody chart).Hence, KI:7573, K : 1.632and
Ksc : 9205.
Valve
10e.
;F
+40
Figure 5.4
PipeNetwork Analysis
Pipe
AB
BE
fLooP4
t
EF
FA
klD,
Q ( L / s ) R e( x 1 0 s )
0.00024 1,20.00
0 .0 0 0 4 0 10.00
o.ooo3o 40.00
0.00024 80.00
h (m) blQ
IK
5.41
0.75
2.25
3.61,
3.03
125
'A
79.70
4s.26
108.4s
33.10
266.51
= + A Q : 5 .6 9 L l s .
Pipe
BC
CD
Loop 2
DE
EB
Q (L/s) R e ( x 1 0 s )
klD
0.0004 50.00
0.0006 10.00
0.0004 =20.00
0.0004 4.31,
3 . 75
1.r3
1.50
0.32
h (il
23.0L
htQ
m
3^
460.26
338.77
L64,52
22.70
986.25
+ AQ  tL.67 Lls.
Proceedingin this way the solution is obtainedwithin a small limit on lh
in any loop:
Final ualues
Pipe
Q (L/s)
hr @)
FE
AB
BE
t1.1..s2 1.6.48 48.48
1.1s
6.26
8.31
ED
CD
FA
BC
88.48 35.05 4.95 34.95
3.20 1r.57 0.91. 9.52
Example5.3
If in the network shown in Example 5.2 a pump is installed in line BC boosting the flow
towards C and the valve removed, analysethe network. Assumethat the pump deliversa
head of 10 m. (Note: In reality, it would not be possibleto predict the head generatedby
the pump sincethis will dependupon the discharge.The headdischargerelationship for
the pump, e.g. H : Ae + BQ * C, must thereforebe solvedfor the dischargein the pipe
at eachiteration. However, for the purposeof illustration of the basic effectof a pump the
head in this caseis assumedto be known.) An example of a network analysisin which the
pump headdischargecurve is usedis given in Chapter 6 (Example6.8). Considerlength
BC (seeFigure5.5).
The net lossof headalong BC(Zg  kl is (fu  Hp), where tlo is the total headdelivered
by pump. The value of K for BC is now due to friction only; the head loss for BC in
the table now becomesEp,s6: (K A"r 10) m. Otherwisethe iterative procedureis as
before.
EI
Li
*l
g
fitl
E
126
Civil EngineeringHydraulics
Pump
Solution:
Pipe
AB
BE
,LooP4
t
EF
FA
Q (L/s)
klD
0.00024 1,20.00
0.00040 10.00
o.ooo3o 40.00
0.00024  8 0 . 0 0
5.41.
0.75
2.25
3.61,
79.70
45.26
108.45
33.10
266.51
3.03
+ A Q :  5 . 6 9L t s .
klD',
BC
rLooPL
^ cD
DE
EB
tt til i,l'
R*, x tOs
3.76
t.13
1.50
0.32
0.00040 50.00
0.00060 10.00
o.ooo4o 20.00
0.00040  4 . 3 t
8.93
1,78.66
333.77
764.s2
22.70
704.6s
+ AQ: 6.34Ls.
10
Loop L
AB
BE
EF
FA
11,4.31,
1 0 .6 5
4 5 .6 9
8 5 .6 9
5.15
0.80
2.57
3.66
668.4
0.0158
0.0206 4 482.9
0.01.73 2 680.2
4Lr.2
0.01,62
8.73
0.51
5.59
3.02
76.4L
47.74
1,22.46
35.24
0.63
281.85
+ A8 : l .1 L L /s .
After similarfurtheriterotions:
Final ualwes
Pipe
Q (L /s )
h r (m)
AB
BE
1 .1 3 .2 1 8.90
8 .5 7 0.37
FE
46.79
5.83
FA
86.79
3.10
BC
44.30
4.95
CD
4.30
0.71
ED
25.70
5.29
Figure 5.6
127
Exomple5.4
Determinethe dischargesin the pipes of the network shown in Figure 5.6 neglectingminor
losses.
AJ
BJ
CJ
DJ
450
350
300
250
10000
2 000
3 000
3 000
Solution:
Estimate27 (pressurehead elevationat J)  150.0 m a.o.d. (Notezthe elevationof the
pipe junction itself doesnot affectthe solution.)Seetablesbelow and on p.128.
I
=* correct ionto Zt
0.2862
0.0193
128
Civil EngineeringHydraulics
Second correction
o"(xlos)
n,n.'J'i$#'
AJ
BJ
CJ
DJ
As
initial
estimate
7 .9 6
6 .2 0
5.31
4.42
K zr  zt
D
+ Lq
t#rl
+ 0.1066
(3ri
eth(x1o3)
4.39
80.1,2
5.58
2.29
2.20
0.27
r.60
2.20
+ 0.092
: a 2 . 3 0^ ; Z t : 1 2 2 . 6 3 m .
Pipe
Velocity
(estimate)
(m/s)
AJ
BJ
CJ
DJ
2.0
1.0
1.8
2.3
K
0 .0 1 4 5 649
503
0.01.6
0.0155 1581
4 061
0 .0 1 6
!r
It
.l
tr
77.37
2.63
22.63
47.63
?o
zr  zt
t#rrl
flA
Qth(^103)
0.3452
0.0723
0.1,196
'0.1083
4.46
27.50
+0.0450
0.0395
s.29
2.27
2.t7
0 . 75
1.69
2.21,
+ L Z t : 2 . 2 7 ^ ; Z t : 1 . 2 4 . 9 0m .
Fi,nalualues:
Q A J 0 .3 4 4m3 l s ;
Q , : 0 .105m3l s;
QJo : 0.1'1,2m3/s
Exomple5.5
If in the network of Example 5.4 the flow to C is regulatedby a valve to 100 L/s, calculate
the effect on the flows to the other reservoirs;determine the head loss to be provided by
the valve.
The principle of the solution is identical with that of the previous example except that
the flow in JC is prescribedand simply treated as an external outflow at J. In this example
the flow ratesin the pipeshave beenevaluateddirectly from Equation 5.6.
Q_ 24
in which h : Zr  Zl,since thereare no minor losses.This approachis ideal for computer
analysis;if minor lossesare presentusethe iterativeproceduredescribedin Example4.2.
PipeNetworkAnalysis 129
The method is also suitable for desk analysis using an electronic calculator since for
each pipe the only variable is h and Equation 5.6 can be written as
g(rr*9)
Q :  c r r t b \
.rtl
Pipe
k /D
BJ
0.000171
AJ
0.000133
DJ
0.000240
a.o.a.
First correction
Pipe
JunctionJ
AJ
BJ
DJ
Correction to
5.59
8.53
1.94
1.76
2.66
2.81,
I
0.0143
0.0146
0.0155
114.s3 1.s.96
279.32
25 5.95
1,37.90
V (m/s)
P)
zl : 2 ( D Q D,Q/h
zl  1 2 3 . 1m1
2 (  1 4 4 . s 3 1 0 0 )  2 6 . 8 9
:
m
1.5.96
Second correction
JunctionJ
Pipe
Zt  Zt
AJ
BJ
DJ
76.89
3.11,
48.1,1
Q (L/s)
Ql h
V (m/s)
349.70
77.61,
109.50
4.55
24.96
2.28
2.20
0.81
2.23
0.0140
0.0164
0.0158
ta
L
e)
+.
a
(t
130
Hydraulics
CivilEngineering
[
J
Junction
AJ
BJ
DJ
72.9s
7 .05
s2.05
466
340.2
1'1,9.94 1'7.0t
114.08 2.r9
2t4
1'25
2.32
0.0141
0'0156
0.01s8
+ L Z I : 0 . 5 2 m ; Z I : 1 ' 2 7  5 7m .
Final ualues
Pipe
Q (L/s)
AJ
338.98
JB
r24.36
JD
1,14.65
mls
Q  0.100m3ls; V  1'.41'5
1,.415x 0.3
^
.. x 1 0'bt
: 3 '. / 6
0.0002
Re:
i:
113 1b=
D i a me te r 3 0 0 mm;
A :0 .0 707 ^' ;
" .
I
3 0 0 0x t.4I52
0 . 0 1 6x 3000
= 1.6.33m
w h e n c e X  0 . 0 1 . 6 ;h r : @
(seeFigure5.7).
* Headlossat valve Zl  k  ht
 1.6.33
 t 2 7 . 5 5 1 0 0 . 0 0
 1.1.22
m
1
!
!
I
Exomple5,6
In the network as before, a pump P is installed on JB to boost the flow to B. \fith the flows
ttr. pump delivering 10metreshead, determinethe flows in
to C and D uncontrolled
"nd
the pipes(seeFigure5.8).
lx
J
Figure 5.7
I
I
PipeNetworkAnalysis 1 3 1
Note: In the caseof rotodynamic pumps the manometric head delivered varies with the
discharge(seeChapter 6). Thus it is not strictly possibleto specify the head and it is
necessaryto solvethe pump equation I+: AA + BQ+ C togetherwith the resistance
equation for JB. However to illustrate the effectof a pump in this example let us assume
that the head does not varv with flow.
Solution:
The analysisis straightforward, and follows the procedure of Example 5.5.
The head giving flow along JB is
hL,B:ZlZsHp
The final solution is as follows:
Pipe
I(L )
AJ
357.7
JB
141,.6
JC
110.8
JD
105.3
N o te :Z t:119.66 m o.d.
Exomple5.7
{
Determine the flows in the network shown in Figure 5.9 neglectingminor losses.
Pipe
Length(m)
Diameter(mm)
AB
10000
450
BC
3 000
250
BD
4 000
250
BE
6 000
3s0
EF
3 000
250
EG
3 000
200
132
C i v i lEn g i n e e ri n gH y d raul i cs
Figure 5.9
Solution:
In this casethere are two unknown pressurehead elevationswhich must therefore be both
initially estimatedand corrected alternately.
EstimateZs  1.20.0m o.d.;
Zp  95.0 m o.d.
First correction
JuncuonI'
Pipe
ZrZt
(:h)
AB
CB
DB
EB
21,9.77
30.00
71..38
20.00
40.00
86.7s
2s.00 13s.00
Q (L/s)
73.35
Ql h
7.33
3.57
4.34
5.40
V (m/s)
1.38
1.45
t.77
t.40
I
0.01,39
0.0155
0.0151
0.0145
20.63
Junction E
Pipe
Zt  Zt
BE
FE
GE
T
+ LZE: 2.I m3Zn: 9 2 . 9 m .
Q(L/s)
11.95
Qlh
11,.40
V (m/s)
1..1.7 0.01.49
r.45
0.0155
1.70 0.0159
PipeNetworkAnalysis 133
Second correction
JunctlonI,
Pife
i,Zt
AB
CB
DB
EB
37.1.t
1,2.89
32.89
1,9.99
246.21,
56.38
78.16
 1 1 9 . 75
6.63
4.37
6.06
5.99
8.07
23.06
1.55
1.15
1.59
1,.2s
0.0t37
0.0160
0.0153
0.0148
t.22
r.37
1..64
0.0148
0.0156
0.0159
)
JunctionE
Pipir Zr  Zt
Q (L/s)
Ql h
BE
FE
GE
92.9
t7.9
32.9
117.48
67.26
51..64
6.09
3.76
1.57
1.43
Lr.42
+ L Z E :  0 . 2 5m ; Z n : 9 2 . 6 5 m .
Exomple5.8
In the network shown in Figure 5.L0, a valve in pipe 23 is partially closed,producing a
local head lossof 10vt3l2g. The head at node 1 is 100 m of water. The roughnessof all
pipes is 0.06 mm. The pipe lengths are in metres and the demand dischargesare in litres
per second.
The pipe diametersare pipes L2 and 16, 250 mm; pipe 65, 200 mm; pipes 23
and 45,1,50 mm; pipes 25 and 34, 100 mm. Analysethe network using the gradient
method.
406
Figure 5.10
134
Civil EngineeringHydraulics
00
405
Figure5.11 Network solution.
'$
[Hi+r] r [Hi]
For all pipes initial dischargesof 100 L/s have been assumedwith the directions as shown
in Figure5.11.
Solution:
All the matrices and vectors neededfor the gradient method are as follows:
NT_7
NN_5
NS_1
lLLzl  corinectivity matrix; dimension (7 xX 5)
10
1, L
01.00
1 0
00
00
00
0
00
0
t
00
0
1. 1.
L0
0
1. L
0
0t
0
PipeNetworkAnalysis 135
1. 0 0 1,
11
0
1011.0
1
0
0
1
_
l0
1,
0
0
1
l0
0
loo0o
00
00
00
10
*11 1.
lA10l
tHl
tQl
(3l r)
1,
0
0
0
0
0
1,
LIz
EI3
0.10
0.10
0.10
0.10
0.10
0.10
0.10
lHol
Iql
(m)
(tn3lr)
Irool
H4
Hs
LI5
l 0 . 0 6
1 0 . 0 4
1 0 . 0 3
1 0 . 0 3
1 0 . 0I 4
[N] : diagonal matrix; dimension (7 x 7l; having 2 in the diagonal (from the DarcyWeisbachhead lossequation)
20
02
00
00
00
00
00
00
00
20
02
00
00
00
000
000
000
000
2 0 0
0 2 0
002
1.
0
0
0
0
0
0
00
10
OT
00
00
00
00
00
00
00
10
01,
00
00
00
00
00
00
00
10
01,
First iteration:
The previous matricesand vectorsare valid for all the iterations. The following matrices
changein each iteration:
PipeNetworkAnalysis 135
1. 0 0  r
11
00
l0Lr
10
0
1
l0
1. r
1000
00
1000
001
001
001
,L 0l
1. f[
lA10l
tHl
tQl
lHol
(m)
m3l s)
1
0
0
0
0
0
I
Flz
rI3
0.10
0.10
0.10
0.10
0.10
0.10
0.10
Irool
H4
IIs
Fte
tql
m3ls)
0.06
0.04
0 . 0 3
0 . 0 3
0.04
[N] : diagonal matrix; dimension (7 x 7); having 2 in the diagonal (from the Darcy$Teisbachhead loss equation)
2 0 0 0
0 2 0 0
0 0 2 0
0 0 0 2
0 0 0 0
0 0 0 0
0000002
0
0
0
0
2
0
0
0
0
0
0
2
0
0
0
0
0
0
10
01.
00
00
00
00
00
00000
00000
1,0000
01000
00100
000L0
00001,
First iteration:
The previous matricesand vectorsare valid for all the iterations.The following matrices
changein each iteration:
136
\
C i v i l En g i n e e ri n gH y d ra ul i cs
Pipe
(m3ls)
1,2
23
34
54
25
65
61
0.10
0.10
0.10
0.10
0.10
0.10
0.10
(m/s)
0.0159
0.0166
0.0178
0.0166
0.01,78
0.0161
0.0159
1,.974
5.482
1,2.335
5.482
1,2.335
3.084
1,.974
6.22
66.89
271..02
66.89
270.99
23.09
3.73
6.22
82.21,
271,.02
66.89
270.09
23.09
622.28
8220.77
27 101,.65
6 688.98
27 098.90
2308.78
373.42
aa
J./J
Matrix [A11]:
62.23000000
0822.0800000
00271.0.1.60000
000668.9000.0
0
0
0
00000230.880
0
0
0
0
0
2709.89
0
37.34
on the diagonal
62.23000000
0822.0800000
00271.0J.60000
000668.90000
0
0
0
00000230.880
0
0
0
0
0
2709.89
0
37.34
[N][A11]',
000
46000
1.24.
000
0
0
1,644.1,5
0
000
0
5420.33
0
0
1.337.80 0 0 0
0
0
0
0
541,9.78 0
000
0
461..76 0
0
000
0
74.68
0
0
000
0
PipeNetworkAnalysis 137
0.00804
0
0
00
00
0
0.00061
0
00
00
00
0.00018
00
00
00
0
0.00075
0
00
00
0
0
0.0001s
00
00
0
00
0.0021.9
0
00
0
00
0
0.01339
lA21l[N] [A11]')',
0.00018
00
00
0.00061 0.00018
0
0
00
0.00018 0.00075
0
00
0.00075 0 . 0 0 0 1 8 0.00219
0
0
0.00219 0.01.339
00
0
o.00061
lo.ooso+

100
100
100
lA10l[Ho]
100
0
0
0
0
0
100
(tA21l(tNl[A11]')1
(tA11ltQllA10l[Ho]))
0.853
0.1
0
0.05
1.339
[A11]tQl) + ([A10][Ho])
93.777
82,2OB
27r.01.6
66.890
270.989
23.088
96.266
tA2lltQl
0.1
0.2
0
0.1
0
(tA21l(tNl [A11]')1([A11]
 (lA21l
tQl + [A10][Ho])
tQl tql))
l0.6e3s
l0.06
 0.03
l0.02
 1.2ee
138
Hydraulics
CivilEngineering
Thus
1
1
Hi+r :  (tA21l(tNl [A11]')1(tnrzl ) (lez r] ([N][A11]',) (tAl 1ltQl
 ([A2l]tQl Iql))
+[A10][Ho]
(m)
92.000
164.922
80.1"15
99.3r7
97.335
Node
2
3
4
5
6
[A12][Hi*r]+
lAl0llHol
8.00
72.92
84.81
r9.20
7.32
t.98
2.67
lAl2llHi+rl
92.00
72.92
84.81
1.9.20
 7.32
t.98
97.33
+(tAl0llHol))
0.0643
0.0444
0.0157
0.01.44
0.00L4
0.0043
0.0357
(tNltA11l';t
111lJ
0
0.5
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0.5
0
0
0
00
0.s
0
00
00
00
00
0
0.5
0
0
0
0.5
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0.5
(trl (tNl)tA1
1l/)1tAl1l)
0.5
0
0
0
0
0
0
0
0
0
0
0
0.5
0
0
0
0
0
0
0
0.5
0.05
0.05
0.05
0.0s
0.0s
0.05
0.05
139
Thus
 ((tNllA11'l)1([A12][Hi+r]
Qi*r : (trl [N][A11',])1tA11l)tQl
+[A10][Ho]))
Pipe
1.2
23
34
54
25
65
61,
(3lr)
0.1.1.4
0.006
0.034
0.064
0.049
0.046
0.086
Node
2
.,
(m)
4
5
6
92.960
81.358
81.780
89.81,2
96.727
Pipe
L2
23
34
54
25
65
6l
(3lr)
0.10667
0.03658
0.00342
0.03342
0.01009
0.05333
0.09333
Pipe discharges:
140
Hydraulics
CivilEngineering
Problems
1.
Calculate the flows in the pipes of the pipe systemillustrated in Figure 5.L2. Minor
are given by C^V212g.
AB
BCr
BCz
400
250
250
5000
7000
7000
10.0
15.0
10.0
0.15
0:15
0.06
(No/e: \7hile this problem could be solved by the method of Example 4.I, the method
of quantity balance facilitates a convenient method of solution. Note that the pressure
head elevations at the ends of Ct and C2 are identical.)
In the system shown in Problem 1.,an axialflow pump producing a totalhead of 5.0 m
is installed in pipe BCr to boost the flow in this branch. Determine the flows in the pipes.
(No/e: Although it is not strictly possible to predict the head generated by a rotodynamic
pump since this varies with the discharge (seeChapter 6), axial flow pumps often produce
a fairly flat headdischarge curve in the middischarge range.)
3.
Determine the flows in the network illustrated in Figure 5.13. Minor lossesare given by
C^Vz12g.
AB
BC
BDr
BDz
20 000
5 000
6 000
6 000
500
350
300
250
0.3
0.3
0.3
0.06
20
10
10
10
Figure 5.13
141
5.
Pipe
Length (m)
AB
BC
BD
10000
4 000
5 000
Diameter (mm)
400
250
250
Ro,rghn*rs(mm),
0.06
0.06
0.06
Determine the pressure head elevations at B and D and the discharges in the branches in
the systemillustrated in Figure 5.15. Neglect minor losses.
Pipe
AB
BC
BD
DE
DF
Figure 5.14
Length (m)
20 000
2 000
2000
2 000
2 000
Dialneter tmt
600
250
450
300
250
Roughness (mm)
0.06
0.06
0.06
0.06
0.06
142
Civil EngineeringHydraulics
Figure 5.15
7.
0.06
0.06
0.06
0.06
0.06
450
300
400
250
300
20 000
2000
10000
3 000
4 000
AB
BC
BD
D E
DF
Determine the flow in the pipes and the pressure head elevations at the junctions of the
closedloop pipe network illustrated, neglecting minor losses. All pipes have the same
roughness size of 0.03 mm. The outflows at the junctions are shown in litres per second
(seeFigure 5.1'6).
Pipe
Length(m)
Diameter(mm)
AB
s00
200
BC
600
150
+
Inlet
Figure 5.16
Twoloop network.
CD
200
100
DE
600
150
EA
600
200
BE
200
100
PipeNetworkAnalysis
200Lls
Figure 5.17
143
200Lls
Pipes in parallel.
(No/e: A more rapid solution is obtained by using the head balance method. However
the
network can be analysed by the quantity balance method but in this case four unknown
pressure heads, at B, C, D and E, are to be corrected. If the quantity balance method
is
used, set a fixed arbitrary pressure head elevation to A, say roo m.)'
Pipe
Length (m)
Diameter (mm)
Roughness (mm)
BCE
3000
300
0.06
AB
1000
450
0.15
BE
2000
2s0
0.15
BDE
3000
350
0.06
EF
1000
450
0.15
9.
In the system shown in Problem 7 (Figure 5.16) a pump is installed in BC to boost the
flow to C. Neglecting minor losses determine the flow distribution and head elevations
at the junctions if the pump delivers a head of 15.0 m.
10.
Determine the flows in the pipes and the pressure head elevations at the junctions in the
network shown in Figure 5.18. Neglect minor lossesand take the pressurehe"d elevation
at A to be 100 m. The outflows are in litres per second. All pipes have a roughness of
0.06 mm.
+
/t'
c _/
,/30
D ,/
20
Inlet
1o
Figure5.18 Threeloopnetwork.
4o
,/*
r0
144
11.
Civil EngineeringHydraulics
150
FG
400
150
GA
300
200
DH
300
150
DE
150
150
EF
300
150
Pipe
Length(m)
Diameter(mm)
AB
400
200
BH
150
200
1s0
Pipe
Length (m)
Diameter (rnttt)
BC
300
1s0
CD
150
150
HF
Analyse the flows and pressureheads in the pipe system shown in Figure 5.19. Neglect
minor losses.
Pipe
Length (m)
Diameter (mm)
Roughness(mm)
AB
1000
250
0 .06
BC
400
200
0.15
CD
300
150
0.15
DE
400
150
0.15
'42.
13.
,,/uo
o\,0
Figure 5.19
EF
800
250
0.06
EF
300
200
0.1s
Flowof Incompressible
Fluidsin Pipelines 1 1 s
The pipeline terminates in a nozzle (G : 0.9S) which is 15 m below the bvel in the
reservoir. Determine thenozzle diameter such that the jet will have the maximum possible
power using the available head and determine the jet power.
Oil of absolute viscosity 0.07 N s/m2 and density 925 kglm3 is to be pumped by a
rotodynamic pump along a uniform pipeline 500 gr long to discharge to armosphere at
an elevation of +80 m o.d. The pressurehead elevation atthe pump delivery is 95 m o.d.
Neglecting minor losses, compare the discharges attained when the pipe of roughness
0.06 mm is (a) 100 mm and (b) 150 mm diameter, and state in each whether the flow is
laminar or turbulent.
:
I
{
i
9.
10.
Determine the hydraulic gradient in a rectangular concrete culvert 1 m wide and 0.6
m high of roughness size 0.06 mm when running full and conveying water at a rate of
z.) m"/s.
115
The pipeline terminates in a nozzle (C]/ : 0.98) which is 15 m below the bvel in the
reservoir. Determine thenozzlediameter such that the jet will have the maximrom possible
power using the available head and determine the jet power.
Oil of absolute viscosity 0.07 N s/m2 and denslty 925 kglm3 is to be pumped by a
rotodynamic pump along a uniform pipeline 500 m long to discharge to atmosphere at
an elevation of +80 m o.d. The pressurehead elevation at the pump delivery is 95 m o.d.
Neglecting minor losses, compare the discharges attained when the pipe of roughness
0.06 mm is (a) 100 mm and (b) 150 mm diameter, and state in each whether the flow is
laminar or turbulent.
9.
10.
Determine the hydraulic gradient in a rectangular concrete culvert 1 m wide and 0.6
m high of roughness size 0.06 mm when running fulf and conveying water at a tate of
2.5 m3/s.