You are on page 1of 32

chapter 5

Pipe Network Analysis

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't-Fl: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 non-linearequationsare 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 ProfessorHardy-Cross,
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 closed-loop
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,

5.2 The head balance method ('loop' method)


This methodis applicableto closed-looppipe networks.It is probably more widely applied
to this type of network than is the quantity balancemethod.The headbalancemethodwas
originally devisedby ProfessorHardy-Crossand is often referredto as the Hardy-Cross
method. Figure 5.1 representsthe main pipesin a water distribution network.
The outflows from the systemare generallyassumedto occur at the nodes (junctions);
this assumptionresultsin uniform flows in the pipelines,which simplifiesthe analysis.
For a given pipe systemwith known junction outflows, the head balancemethod is an
iterativeprocedurebasedon initially estimatedflows in the pipes.At eachjunction these
flows must satisfythe continuity criterion.
The headbalancecriterion is that the algebraicsum of the headlossesaround any closed
loop is zero;the sign conventionthat clockwiseflows (and the associatedhead losses)are
positiveis adopted.
The head loss along a singlepipe is

b: KA

118

Hydraulics
CivilEngineering

Inflow

Figure5.1 Closed-looppipe network.

If the flow is estimatedwith an error A Q

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-,

where E is the head loss in a pipe basedon the

5.3 The quantity balance method ('nodal' method)


Figure 5.2 shows a branched-type pipe system delivering water from the impounding
reservoir A to the servicereservoirs B. C and D. F is a known direct outflow from the
n o d e/.

rl
Figure 5.2

Branched-type pipe network.

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

Ktt|eu 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- Q-qi+Ql+Qc7*S.r-F: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.
R-earrangingEquation 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

analysisEquation 5.5 should be replacedby the Darcy-Colebrook-White combination:


th,
2.s1.v \
f sD;tos I k
Q: -2A\2
\:-'o*;@.)

15'61

For each pipe, h1,11(friction head loss) is initialised to'21 -.21, Qr1 calculatedfrom
Equation 5.6 and h1.77re-evaluatedfrom hr,rt : Vr - 4) - X^$t. This subroutine
follows the procedure of Example 4.2.

5.4 The gradient method


In addition to Equations 5.1-5.6, the gradient method needsthe following vector and
matrix definitions:
NT - number of pipelinesin the network
NN - number of unknown piezometrichead nodes
'connectivity matrix'
associatedwith each one of the nodes.Its dimension is
l[l2]:
NT x NN with only two non-zero elementsin the ith row:
-1 in the column correspondingto the initial node of pipe i
1 in the column correspondingto the final node of pipe i
:
NS number of fixed head nodes
tA10] - topologic matrix: pipe to node for the NS fixed head nodes.Its dimensionis
NT x NS with a -1 value in rows correspondingto pipelinesconnectedto
t1
t
Al
ttt+
o

fixed head nodes


Thus, the head loss in eachpipe betweentwo nodesis

l,

tAllllQl +lLt2l[H] : -[A10][Ho]

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

[Q] : dischargevector with NT x 1 dimension


[H] : unknown piezometrichead vector with NN x 1 dimension
[Ho] : fixed piezometrichead vector with NS x 1 dimension
Equation 5.7 is an energyconservationequation.The continuity equationfor all nodes
in the network is

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

tlA11ltA12l-l
[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+3r-l
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(nz-r)ool
a3Qlz-t)
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]

The objectiveof the gradientmethod is to solvethe systemdescribedby Equation 5. 11,


taking into account that in each iteration

l d Q l: [ Q i + r ] - t Q i l

[s.1s]

tdHl - [Hi+r]- [Hi]

[s.16]

and

Using matrix algebra,it is possibleto showthat the solutionto the systemrepresented


b y E q u a t i o n5 . 1 1 i s

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,

Figure5.3 Two-loop network.

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

Pipe Network Analysis

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 1.1229.1-40.42
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

Note: Flows in direction of pipe identifier, e.g. A -+ B.

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

of all pipesis 0.06 mm.


lVor"rRoughness

'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

Pipe network with valve losses.

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

0.0ts7 664.2 9.s6


0.0208 4 526.s 0.4s
-4.34
0 . 0 1 7 52 7 1 , 1 , . 2
0.01.63t413.7 - 2 . 6 5

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

0.01.74 9 20s.2 23.01


3.39
0.0205 33 877.0
0.0190 8226.0 -3.29
0.0242 5 266.4 -0.10

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 head-dischargerelationship 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 head-dischargecurve 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

Figure5.5 Network of Example5.2 with 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,

0.01.57 664.2 9.56


0.0208 4 526.5 0.45
0.0175 271.1..2-4.34
0.01,63 413.7 -2.65

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

0.01,74 7 573.0 8.93


0.020s 33877.0 3.39
0 . 0 1 8 9 8 2 2 s . 9 6- 3 . 2 9
0.0242 5 266.4 -0.10

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: -6.34L|s.

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

Pipe Network Analysis

Figure 5.6

127

Network connecting multi-reservoirs.

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

sizeof all pipesis 0.06 mm.


Note:Roughness
The friction factor ), may be obtained from the Moody diagram, or using Barr's equation,
using an initially estimated velocity in each pipe. Subsequeniy, )" can be based on the
prerriouslycalculated discharges.However, unless there is a serious error in the initial
velocity estimates,much effort is savedby retaining the initial ). values until perhapsthe
penultimate or final correction.

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

2(-0'2862) -29.67; :1'20.33m.


:
: =
Zr
Offi

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

0 .0 145 649 79.67 0.3504


0 .0 150 472 -0.33 -0.0264
0 .0 155 1581 -20.33 -0.11,34
0 . 0 1 6 54 1 8 8 - 4 5 . 3 3 - 0 . 1 0 4 0

D
+ Lq

t#rl

+ 0.1066

(3ri
eth(x1o-3)
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 .

Comment: The velocity in BJ has changedsignificantly but it may oscillate;rt is therefore


estimatedat L.0 m/s for next correction.Note that ). (BJ)alteredaccordingly.
Third correction

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(^10-3)

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;

Ql. : 0.1'27^3 lt;

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

in which Ct, Cz and C3 are constantsfor a particular pipe.


The correspondingvelocitiesand l valueshavebeenevaluatedand tabulated;thesedata
may be useful for those who wish to work through the exampleusing the Moody diagram
as shown in Example5.4.
Note that Q is expressedin litres per second; in evaluating L, alh the flow is also
expressedin litres per secondso that the units in the correction t e r mA , Z : z ( D Q - F ) l
E Q/ h) areconsistent.

Example 5.5 calculation

Pipe
k /D

BJ
0.000171

AJ
0.000133

Note: EstimateZt :150.00

DJ
0.000240

a.o.a.

First correction

Pipe

JunctionJ

AJ
BJ
DJ

?-nti,t e(r/s) elh


50.00
-30.00
-7 5.00

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

+ LZt : 3.94m; ZI : I27 .05m.

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

4-66
340.2
-1'1,9.94 1'7.0t
-114.08 2.r9

2-t4
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

Head loss due to friction along JC:

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 1-0metreshead, determinethe flows in
to C and D uncontrolled
"nd
the pipes(seeFigure5.8).

lx
J
Figure 5.7

I
I

Network of Example 5.4 with valve losses.

PipeNetworkAnalysis 1 3 1

Figure5.8 Network of with pump.

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:Zl-Zs-Hp
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

Note: Roughnessof all pipes is 0.03 mm (:ft).

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

Network with multi-reservoirs.

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

Zr-Zt
(: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

+ LZB : 17 'tt m; Zs : 1L2.89m-

Proceedto Junction E noting that the amendedvalue of Zs is now used:

Junction E

Pipe

Zt - Zt

BE
FE
GE

1.7.89 1.1.2.81 6.31


-20.00 -7t.38 3.s7
-3s.00 -s3.38 1.53

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

+ LZB: -0.7m; Zs:1L2.L9 m.

)
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 2-3 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 L-2 and 1-6, 250 mm; pipe 6-5, 200 mm; pipes 2-3
and 4-5,1,50 mm; pipes 2-5 and 3-4, 100 mm. Analysethe network using the gradient
method.

406
Figure 5.10

Pipe network with valve loss.

134

Civil EngineeringHydraulics

00

405
Figure5.11 Network solution.

The iterative processcan be summarisedin the following steps:


1. Assume initial dischargesin each of the network pipes. (They can be unbalanced at
eachnode.)
2. Solve the system representedby Equation 5.17 using a standard method for the
solution of simultaneouslinear equations.
3. \fith the calculated[H1a1](Step2), [Qr+r] is solvedby Equation 5.18.
4. With ttre new [Qi+r], Equation 5.L7 is solved(Step2) to find a new [Hi+t].
5. Processcontinuesuntil

'$

[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

[A2T): transposedmatrix of [A12]

-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.

[A10] - topologic matrix node to node; dimension(7 x I)


[Q] : dischargesvector; dimension(7 x 1')
[H] : unknown piezometrichead vector; dimension(5 x 1 )
[Ho] : fixed piezometrichead vector; dimension(1 x 1)
[q] : water demandvectorl dimension(5 x 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

[I] = identity matrix; dimension (7 x 7)

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

[A21] : transposedmatrix of [A12)

-1. 0 0 - r
11
00
l0Lr
10
0
1
l0
-1. r
1000
00
1000

001
001
001
,L 0l
1. f[

[A10] : topologic matrix node to node; dimension(7 x 1')


[Q] : dischargesvector; dimension(7 x 1')
[H] : unknown piezometrichead vectorl dimension(5 x 1 )
[Ho] : fixed piezometrichead vector; dimension(1 x 1)
[q] - water demandvector; dimension(5 x 1)

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

[I] = identity matrix; dimension(7 x 7)

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

tA1l] : diagonalmatrix; dimension(7 x 7);having the valuea;Q"i-t1 on the diagonal,


with coefficientsB and y zero as no pumps exist in the network
The following table shows the calculatedvalues for a:

Pipe

(m3ls)

1,-2
2-3
3-4
5-4
2-5
6-5
6-1

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

a-a
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

[A11]' : diagonal matrix; dimension (7 x 7); having the value ui Q"i-t\


For this network, [A1L'] - [A11].

62.23000000
0822.0800000
00271.0J.60000
000668.90000
0
0
0
00000230.880
0
0
0

0
0

2709.89
0

37.34

To find Ht+r by Equation 5.L7 following a step-by-stepanalysis,the following matrices


can be found:

[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

([N ][A 11]' ,)-1

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

tA21l (tNl [A11]' )-1 lAL2l


-0.00061
-0.00018
0
.
0
0
8
0
4
0
0
|
-0.00018
0.00079
0
0
l-0.00061
-0.00018
0.00093 -0.0007s
0
l0
-0.00075 0.00310 -0.00217
0
l-0.00018
-0.0021,7 0 . 0 1 5 5 6
0
"10
- (IA21l(tNl [A11]')-t 11.tz11-t
-100.256
-33.383 -1,6.879 -2.3s01
l - r 2 0 . s 4 1 -1423.476
-365.444 -1,04.31.0-14.s221
l-100.2s6
- 3 3 . 3 8 3 - 3 6 5 . 4 4 4 - 1 4 6 0 . 1 , 5 7- 3 9 2 . 5 4 8 -s4.6s1l
| -16.87e -104.310
- 3 9 2 5 4 8 - 4 6 3 . 6 8 9 -64.55s1
| _2.3s0 - 1 , 4 . 5 2 2
- 5 4 . 6 5 1 - 6 4 . 5 5 5 -73.2731
|
tAllltQl
6.223
82.208
271.01.6
66.890
270.989
23.088
3.734

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))

l-0.6e3s
l-0.06

| 0.03
l-0.02
| -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

To find Qi*r by Equation 5.L8 following a step-by-stepanalysis,the following matrices


can be found:

[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

([N] [A1 1',]-1X ([A12] [Hi+r] )

+(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

(tll - (lNltAl1l')-1,. tA11l)tQl

(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

Pipe Network Analysis

139

Thus
- ((tNllA11'l)-1([A12][Hi+r]
Qi*r : (trl- [N][A11',])-1tA11l)tQl
+[A10][Ho]))

Pipe
1.-2
2-3
3-4
5-4
2-5
6-5
6-1,

(-3lr)
0.1.1.4
0.006
0.034
0.064
0.049
0.046
0.086

After only five iterations the following are the results.


Head at each node:

Node
2
.,

(m)

4
5
6

92.960
81.358
81.780
89.81,2
96.727

Pipe
L-2
2-3
3-4
5-4
2-5
6-5
6-l

(-3lr)
0.10667
0.03658
0.00342
0.03342
0.01009
0.05333
0.09333

Pipe discharges:

References and recommended reading


Featherstone,
for pipenetworks.
R. E. andElJumailly,K.K. (1983)Optimaldiameterselection
Journal of the HydraulicsDiuision,AmericanSocietyof Ciuil Engineers,1,09,221,-234.
SantafedeBogota,
Saldarriaga,J.(998) HidraulicadeTuberia.s,
McGraw-Hill Interamericana,
Colombia.
'Water
Twort, A. C., Ratnayaka,D. D. and Brandt, M. J. (2000)
Supply, Sth edn, Arnold,
2001,.
London; reprintedby Butterworth-Heinemann,
DistributionModeling,Haestad
S7alski,T. M., Chase,D. V. and Savic,D. A. (2001,)'Water
Press,'Waterbury,
CT.

140

Hydraulics
CivilEngineering

Figure5.12 Pipesin parallel.

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 head-discharge curve in the mid-discharge 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

Pipe Network Analysis

Figure 5.13

141

Network with reservoirs.

In the system illustrated in Figure. 5.t4, a pump is installed in pipe BC to provide a


flow of 40 Lls to Reservoir C. Neglecting minor losses calculate the total head to be
generated by the pump and the power consumption assuming an overall efficiency of
60%. Determine also the flow rates in the other pipes.

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

Network with reservoirs.

Dialneter t-mt

600
250
450
300
250

Roughness (mm)

0.06
0.06
0.06
0.06
0.06

142

Civil EngineeringHydraulics

Figure 5.15

Network with reservoirs.

Determinethe flows in a pipe systemsimilar in configurationto that in Problem 5. A


valve is installedin BC producinga minor loss of 20Vzl2g; otherwise consider only
friction losses.
Pipe

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
closed-loop 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

Pressurehead elevation at A : 60 m a.o.d.

+
Inlet

Figure 5.16

Two-loop 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.)'

Determinethe flow distributionin the pipe systemillustratedin Figure5.1,7andthetotal


head loss betweenA and F. Neglectminor losses.A total dischJrgeof 200 L/s passes
through the system.

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 Three-loopnetwork.

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.

Solve the network in Problem 10 using the gradient method.

13.

Analyse the network of Example 5.1 by the gradient method.

,,/uo

o\,0

Figure 5.19

Network with reservoirs.

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.

A pipeline L0 km long is to be designed to deliver water from a river through a pumping


station to the inlet tank of a treatment works. Elevation of delivery pressure head at
pumping station is 50 m o.d.; elevation of water in tank is 30 m o.d. Neglecting minor
losses,compare the discharges obtainable using
(a)
300 mm diameter plastic pipeline which may be considered to be smooth
"
(i)
using the Colebrook-\(/hite equation
(ii) using the Blasius equation
(b)
300 mm diameter pipeline with an effective roughness of 0.6 mm
"
(i)
using the K6rmin-Prandtl rough law
(ii) using the Colebrook-\7hite equarion.

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.

Flow of lncompressibleFluidsin Pipelines

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.

A pipeline 10 km long is to be designed to deliver water from ariver through a pumping


station to the inlet tank of a treatment works. Elevation of delivery pressure head at
pumping station is 50 m o.d.; elevation of water in tank is 30 m o.d. Neglecting minor
losses,compare the discharges obtainable using
(a) a 300 mm diameter plastic pipeline which may be considered to be smooth
(i) using the Colebrook-lfhite equation
(ii) using the Blasius equation
(b) a 300 mm diameter pipeline with an effective roughness of 0.6 mm
(i) using the K6rmin-Prandtl rough law
(ii) using the Colebrook-White equation.

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.