PART 7

Multiple-Pipe Systems
7.1. Head losses in a simple pipeline
The simple pipeline is a pipeline with constant diameter without
branches.
The portion of simple pipeline is described in Fig. 7.1.

Fig 7.1. Simple pipeline

Lts write the Bernoulli equation for cross-sections 1 and 2, assumed
that kinetic energy factors 1 2 1 and reducing the velocity heads in left
an right sides of it:
z1

p1
p
z2 2
g
g

or
p1 p2
z 2 z1
g
where

h H req

h summary head losses along stream from section 1 to section

2.
The difference between the pressure heads in left side of an equation is the
required head - H req . If this value is given, we will call it as available head H av . We can express the head loss as a function of discharge:
h f MQ n ,
(7.1)
where the coefficients M and n have different values depending on the flow
mode.

For laminar flow using the portions of pipe with equivalent lengths
instead of the local losses ( leq ) we will have:
128 ( l leq )
M
and n 1
.
(7.2)
gd 4
For turbulent flow the values for M and n change, depending on whether the
Darcy-Weisbach, Hazen-Williams equation is used.
Darcy-Weisbach equation.
Lets write the expression for head losses considering booth the friction
losses and local losses:
V2
l V2
hf
i
.
d 2g
2g
i
V2
Then factor out the common factor
and using the dependence
2g
d2
between velocity and discharge in round pipes : Q V
, we can rewrite
4
the exp. (7.2) as:
l
8Q 2
h f i
MQ 2 ,
(7.3)
2
4

gd
i

8
where M i
2 gd 4 is the pipeline characteristic factor. The
d
i

n
coefficient in this case is equal 2.
For the Hazen-Williams equation the exponent is n 1.852 and the coefficient M is:
C
M 1.852K 4.87 ,
(7.4)
C HW d
where

C K 10.67

and

C HW CHW is the Hazen-Williams roughness

coefficient. Table (7.1) gives values for C HW for some common pipe
materials.
Pipe Material
C HW Pipe Material
C HW
PVC
150 Wood, Concrete
120
Very Smooth
140 Clay, New Riveted Steel 110
Cement-lined Ductile Iron
140 Old Cast Iron, Brick
100
New Cast Iron, Welded Steel
130 Badly corroded Cast Iron 80
Table 7.1. Hazen-Williams Roughness
In summary, the best equation for computing the frictional head loss in a
given pipe for a given discharge, or the best equation for the discharge if the
head loss is known, regardless of the fluid, is the Darcy-Weisbach equation.
The range of applicability for the empirical equations is much more restricted.

Consequently, all engineers should consider using the Darcy-Weisbach

equation in professional practice even if it is sometimes more difficult to use
than the empirical equations.
7.2. Pipeline characteristic
Pipeline characteristic is the diagram of dependence of required head
H req from discharge Q . Where the value of H req is equal to head losses in
pipeline. According to expression (7.1) we can introduce the head losses in
n
pipeline as: h f MQ . The values of power index n and M coefficient are
different for the laminar and turbulent mode of flow and are defined according
Exp. (7.2) and (7.3) correspondingly.
The example of this dependence for laminar and turbulent flow is
described in Fig. 7.2.

Fig.7.2. Pipeline characteristic for laminar and turbulent flow

7.3. Three Types of Pipe-Flow Problems
The Moody chart or Murin chart can be used to solve almost any
problem involving friction losses in long pipe flows. However, many such
problems involve considerable iteration and repeated calculations using the
chart because the standard Moody chart is essentially a head-loss chart. One
is supposed to know all other variables, compute Re , enter the chart, find ,
and hence compute h f . This is one of three fundamental problems which are
commonly encountered in pipe-flow calculations:
1. Given d , L , and V or Q , , and g , compute the head loss h f (headloss problem).
2. Given d , L , h f , , , and g , compute the velocity V or flow rate Q (flowrate problem).
3. Given Q , L , h f , , , and g , compute the diameter d of the pipe (sizing
problem).

For the solution the first type of problem its necessary to calculate the
Reynolds number and define the flow mode in pipeline or in its sections. Then,
using the Moody chart or corresponding formula to define the value of .
After obtaining you should calculate the head loss according the follow
expression:
l V2
hf
(7.5).
d 2g
The second type of problem we solve with help of iteration method.
Lets consider that flow is turbulent and the value of in first
approximation : 0.015...0.04 . From exp. (7.5) we can obtain the value of V
and Reynolds number Re . Using the Moody chart taking into consideration
k
the given value of pipe relative roughness
, we obtain the next
d
approximation of . Then we repeat this algorithm of calculations while the
difference between the values of velocities in adjacent iterations will not be
less then given value .
The algorithm of solution of the third type of problem will be follows:
Lets write the expression for head losses (7.3) and express the d 4 :

l8Q 2
8Q 2
d
i 2
.
d 2 gH
gH
i
4

B i
i

8Q 2

2 gH

Then

designate

as:

l8Q 2

2 gH

and

, we can obtain the equation which gives the solution of

( d )
B.
d
Its convenient to solve this equation graphically.
For this purpose take into consideration some values of diameter d and
calculate the friction losses coefficient for each one using the
corresponding formulas depended on the flow mode. The required value of
d we can obtain as intersection point of two curves :
( d )
y1 A
B and y2 d 4 .
d
problem: d 4 A

7.4 Multiple-Pipe Systems

If you can solve the equations for one-pipe systems, you can solve
them all; but when systems contain two or more pipes, certain basic rules
make the calculations very smooth. Any resemblance between these rules
and the rules for handling electric circuits is not coincidental.

Fig 7.3. Pipes in series

Fig 7.4. Pipes in parallel;

Fig 7.5. The three-reservoir junction problem

Fig. (7.1)-(7.3) shows three examples of multiple-pipe systems. The first
is a set of three (or more) pipes in series. Rule 1 is that the flow rate is the
same in all pipes
Q1 Q2 Q3 const
or
(7.6)
V1d 12 V2 d 22 V3d 32 const
Rule 2 is that the total head loss through the system equals the sum of
the head loss in each pipe
ha b h1 h2 h3
(7.7)
In terms of the friction and minor losses in each pipe, we could rewrite
this as

V32 3 L3
V12 1 L1
V2 2 2 L2
ha b
K 1
K 2
K 3 (7.7)

2 g d1
2 g d 2
2 g d3

and so on for any number of pipes in the series. Since V2 and V3 are
proportional to V1 from Eq. (7.5), Eq. (7.7) is of the form
V12
ha b
(7.8)
0 1 1 2 2 3 3
2g
where the a, are dimensionless constants. If the flow rate is given, we can
evaluate the right-hand side and hence the total head loss. If the head loss is
given, a little iteration is needed, since 1 , 2 , and 3 all depend upon V1
through the Reynolds number. Begin by calculating 1 , 2 , and 3 , assuming
fully rough flow, and the solution for V1 will converge with one or two
iterations. EES is ideal for this purpose.

EXAMPLE 1
Given is a three-pipe series system, as in Fig. (7.3). The total pressure
drop is pa pb 150 000 Pa , and the elevation drop is za zb 5m . The pipe
data are
,mm
L,m
d ,sm
d
Pipe
1
100
8
0.24
0.00
2
150
6
0.12
0.00
3
80
4
0.20
0.00
5
The fluid is water, 1000 kg m 3 and 1.02 10 6 m 2 s .
Calculate the flow rate Q in m3/h through the system.
Solution
The total head loss across the system is
p pb
15000
ha b a
z a zb
5m 20.3m
g
1000 9.81
From the continuity relation (6.105) the velocities are
d 12
d 12
16
V2 2 V1 V1 , V3 2 V1 4V1
9
d2
d3
V2 d 2
4
Re1 Re1 , Re3 2 Re1
V1d 1
3
Neglecting minor losses and substituting into Eq. (7.7), we obtain
2

V12
16
2
1250 1 2500 2 2000 4 3
ha b

2 g
9

and Re2

or

V12
20.3m
(*)
1250 1 7900 2 32000 3
2g
This is the form which was hinted at in Eq. (7.8). It seems to be
dominated by the third pipe loss 32000 3 . Begin by estimating 1 , 2 , and 3
from the Moody-chart fully rough regime: 1 0.0262 , 2 0.0234 ,
3 0.0304 .
Substitution into Eq. (*) to find V12 2 g 20.3

33 85 973 . The first

estimate thus is V1 0.58m / s , from which

Re1 45400 , Re2 60500 Re3 90800
Hence, from the Moody chart,
1 0.0288 , 2 0.0260 , 3 0.0314
Substitution into Eq. (*) gives the better estimate

2
3 3
V1 0.565m / s , Q 41 d 1 V1 2.84 10 m / s or Q1 10.2m 3 / h
A second iteration gives Q 10.2m / h , a negligible change.

The second multiple-pipe system is the parallel-flow case shown in Fig.

(7.4). Here the loss is the same in each pipe, and the total flow is the sum of
the individual flows
ha b h1 h2 h3
(7.9a)
Q Q1 Q2 Q3
(7.9b)
If the total head loss is known, it is straightforward to solve for Qi in
each pipe and sum them, as will be seen in Example 2. The reverse problem,
of determining Qi when h is known, requires iteration. Each pipe is related
to h f by the Moody relation

h f L d V 2 2 g Q 2 C , where C 2 gd 5 8L .
Thus each pipe has nearly quadratic nonlinear parallel resistance, and
head loss is related to total flow rate by
hf

Q2

C i i

, where C i 2 gd i5 8Li

(7.9c)

Since the i vary with Reynolds number and roughness ratio, one
begins Eq. (7.9c) by guessing values of i (fully rough values are
recommended) and calculating a first estimate of h f . Then each pipe yields a

flow-rate estimate Q C h
i
i f
i

1
2

and hence a new Reynolds number and a

better estimate of i . Then repeat Eq. (7.9c) to convergence.

It should be noted that both of these parallel-pipe casesfinding either
Q or i are easily solved by EES if reasonable initial guesses are given.
EXAMPLE 2
Assume that the same three pipes in Example 1 are now in parallel
with the same total head loss of 20.3 m. (Fig. 7.4). Compute the total flow
rate Q , neglecting minor losses.
Solution
From Eq. (7.9a) we can solve for each V separately
V32
V12
V22
(**)
20.3m
1250 1
2500 2
2000 3
2g
2g
2g
Guess fully rough flow in pipe 1 0.026 , V1 3.49 m s hence
Re1 V1d 1 273000 . From the Moody chart read 1 0.0267 ; recompute
V1 3.46 m s , Q1 62.5m 3 / h .]

Next guess for pipe 2: 2 0.0234 , V2 2.61m s ; then Re2 153000 ,

and hence 2 0.0246 , V2 2.55 m s , Q2 25.9m 3 / h .
Finally guess for pipe 3: 3 0.0304 , V3 2.56 m s ; then Re2 100000 ,
and hence 3 0.0313 , V3 2.52 m s , Q3 11.4m 3 / h
This is satisfactory convergence. The total flow rate is
Q Q1 Q2 Q3 62.5 25.9 11.4 99.8 m 3 h Ans.
These three pipes carry 10 times more flow in parallel than they do in
series.
From the given equations and rules for calculation series and parallel
pipelines (7.6), (7.7), (7.9a), (7.9b) it is possible to make follows conclusion:
- at the series connection the discharge is constant in any cross-section
and head loss of the system is the sum of head losses of elements;
- at the parallel connection the resulting discharge is the sum elements
discharges and the head losses of all elements are equal to each other and in
turn equal to head loss of all system. (Fig.7.6).

Fig.7.6. Pipeline characteristic for series and parallel pipelines

Consider the third example of a three-reservoir pipe junction, as in Fig.
(7.5). If all flows are considered positive toward the junction, then
Q1 Q2 Q3 0
(7.10)
which obviously implies that one or two of the flows must be away from the
junction. The pressure must change through each pipe so as to give the
same static pressure p j at the junction. In other words, let the HGL at the
junction have the elevation
pj
hj z j
g
where p j is in gage pressure for simplicity. Then the head loss through each,
assuming p1 p2 p3 0 (gage) at each reservoir surface, must be such
that

V12 1 L1
h1
z1 h j
2 g d1
V22 2 L2
h2
z2 h j
2 g d2
h3

(7.11)

V32 3 L3
z3 h j
2 g d3

We guess the position h j and solve Eqs. (7.11) for V1 , V2 , and V3 and
hence Q1 , Q2 , and Q3 , iterating until the flow rates balance at the junction
according to Eq. (7.10). If we guess h j too high, the sum Q1 Q2 Q3 will be
negative and the remedy is to reduce h j , and vice versa.
EXAMPLE 3
Take the same three pipes as in Example 1, and assume that they
connect three reservoirs at these surface elevations
z1 20 m , z 2 100 m , z 3 40 m
Find the resulting flow rates in each pipe, neglecting minor losses.
Solution
As a first guess, take h j equal to the middle reservoir height,
z 3 h j 40 m . This saves one calculation ( Q3 0 ) and enables us to get the
lay of the land:
zi h j m
i Vi ,m s Q ,m 3 h
Li d i
Reservoi h j m
i
r 1
40
-20 0.0267 -3.43
-62.1
1250
2
40
60 0.0241 4.42
45.0
2500
3
40
0
0
0
2000

Q 17.1

Since the sum of the flow rates toward the junction is negative, we
guessed h j too high. Reduce h j to 30 m and repeat:
zi h j m
i Vi ,m s Q ,m 3 h
Reservoi h j m
i
r
1
30
-10 0.0269 -2.42
-43.7
2
30
70 0.0241 4.78
48.6
3
30
10 0.0317 1.76
8.0

Q 12.9

This is positive Q , and so we can linearly interpolate to get an

accurate guess: h j 34.3m . Make one final list:
Reservoi
r
1
2

hj m

34.3
34.3

zi h j m

Vi ,m s Q ,m 3 h
i
-14.3 0.0268 -2.90
-52.4
65.7 0.0241 4.63
47.1

34.3

5.7

0.0321

1.32

6.0

Q 0.7

This is close enough; hence we calculate that the flow rate is 52.4m 3 / h
toward reservoir 3, balanced by 47.1m 3 / h away from reservoir 1 and
6.0m 3 / h away from reservoir 3.
One further iteration with this problem would give h j 34.53m , resulting
in Q1 52.8 , Q2 47.0 , and Q3 5.8m 3 / h , so that Q 0 to three-place
accuracy. Pedagogically speaking, we would then be exhausted.
7.5. Multiple pipe system
The ultimate case of a multipipe system is the piping network illustrated
in Fig. 6.25. This might represent a water supply system for an apartment or
subdivision or even a city. This network is quit 1 1 e complex algebraically but
follows the same basic rules:

Fig.7.7. Multiple pipe system

1. The net flow into any junction must be zero.
2. The net head loss around any closed loop must be zero. In other words,
the HGL at each junction must have one and only one elevation.
3. 3.All head losses must satisfy the Moody and minor-loss friction
correlations.
By supplying these rules to each junction and independent loop in the
network, one obtains a set of simultaneous equations for the flow rates in
each pipe leg and the HGL (or pressure) at each junction.