Piping network

Reference: Mechanics of Fluids, by MC. Potter and DC.

Wiggert (any Edition will be fine)
1. Applications:
➢City water distribution
➢Multiple “inlets” & “outlets”
2. Some definitions:
➢ Nodes: the joint points of the pipes
➢ Fixed-grade nodes: have known hydraulic-grade
line, such as the free surface of a tank,
➢ Piezometric head H = ( p / g + z )
➢ Hydraulic grade line drop: Wi = H A − H B
➢ Interior nodes: as B, C, D and E in next fig.
➢ Closed or interior loop – enclosed by pipes
➢ Pseudo-loop – imagined
➢ Sign convention for Q and hL in loops
Clockwise Q and head losses are positive,
otherwise negative
➢Sign convention for Q at the interior nodes:
1. Positive = flow into node
2. Negative = flow out of node
1
3. Main tasks:
➢Determine Q in each pipe
➢Determine the pressure head at the nodes
4. Mathematical equations:
➢Continuity at the nodes,  Qall = 0
➢Energy conservation: head losses, together
with pump head around any closed loop
formed by the pipes is zero.
5. In this chapter, we assume:
V 2 / 2 g  ( p /  + z ) , i.e. velocity head is much
smaller than the piezometric heads

6. Example
7 pipes , 2 reservoirs , 1 pump.
Piezometric heads H = ( p / g + z ) at A, B,
C, D, E, F: H A , H B , H C , H D , H E , H F

Given : H A , H F , QC , QD

12 Unknowns : H B , H C , H D , H E , H p

Q1, Q2 , Q3 , Q4 , Q5 , Q6 and Q7

2
(1) Governing equations:
(a) Energy equations (7 Eqs):

Pipe 1:  p + z  + H p =  p + z  + hL
 g A  g B

 H A − H B + H p = R1Q12 (3.1)

Pipe 2: H B − H D = R2Q22 (3.2)

Pipe 3: H C − H D = R3Q32 (3.3)

Pipe 4: H B − H C = R4Q42 (3.4)

Pipe 5: H C − H E = R5Q52 (3.5)

Pipe 6: H E − H D = R6Q62 (3.6)

Pipe 7: H F − H E = R7 Q72 (3.7)

3
(b) Continuity Eqs at 4 interior nodes:

Sign convention for Q at interior nodes:

• positive = flow into node
• negative = flow out of node
Node B: Q1 − Q2 − Q4 = 0 (3.8)
Node C: Q4 − Q3 − Q5 = QC (3.9)
Node D: Q2 + Q3 + Q6 = QD (3.10)

Node E: Q5 − Q6 + Q7 = 0 (3.11)

Pump head :
H p (Q1 ) = a0 + a1Q1 + a2Q12 (3.12)

where a0 , a1, a2 are known constants

12 equations = 12 unknown quantities

4
(c) Simplification of the equations
using concept of loops:
For “closed” loop I. Flow direction

[6]
[7] [2]

[3]
[1]
[5]
[4]

“+” : clockwise ↷
Energy Loss:
“−” : anti- clockwise ↶

Energy: R6Q62 − R3Q32 + R5Q52 = 0 (3.13)

For “closed” loop II,

Energy: R3Q32 − R2Q22 + R4Q42 = 0 (3.14)
For pseudo-loop connecting reservoirs
H A + H p − R1Q12 − R2 Q22 + R6 Q62 + R7 Q72 = H F
(3.15)

5
Total governing equations:
Continuity eqs:
Node B: Q1 − Q2 − Q4 = 0 (3.8)
Node C: Q4 − Q3 − Q5 = QC (3.9)
Node D: Q2 + Q3 + Q6 = QD (3.10)
Node E: Q5 − Q6 + Q7 = 0 (3.11)
Energy eqs:
R6Q62 − R3Q32 + R5Q52 = 0 (3.13)
R3Q32 − R2Q22 + R4Q42 = 0 (3.14)
𝐻𝐴 + 𝐻𝑃 − 𝑅̅1 𝑄12 − 𝑅̅2 𝑄22 + 𝑅̅6 𝑄62 + 𝑅̅7 𝑄72 = 𝐻𝐹

Or: 𝑅̅1 𝑄12 + 𝑅̅2 𝑄22 − 𝑅̅6 𝑄62 − 𝑅̅7 𝑄72 − 𝐻𝑃 + 𝐻𝐹 −𝐻𝐴 = 0
(3.15a)

Now 7 unknown: Q1 ~ Q7 = 7 equations

6
7. Generalized pipe network Eqs
(1) Continuity at jth interior nodes
(such as “B”)
( Q ) j − Qe = 0 (3.16)
j : pipes connected to this node
Qe : external demand
(2) Energy balance around an
“interior” loop (assuming no pump)
( W )i = 0 (3.17)
ith : pipes making up the loop
“+” : clockwise ↷;
“−” : counter clockwise ↶
Wi : loss of ith pipe Ri Qi2
(3) Energy balance along pseudo-loop
connecting 2 fixed-grade nodes
( W )i − ( H p )i + H = 0 (3.18)
7
ith : pipes making up the loop
“+”: clockwise; “−”: counter-clockwise
Wi : loss of ith pipe Ri Qi2
(H p )i : pump head in ith pipe.
Sign for Hp: + for clockwise flow direction
in the loop, Otherwise, use negative.
H : Head difference of 2 fixed-grade nodes
in the path ordered in the clockwise direction
across the imaginary pipe in the pseudo-loop.
Therefore, for the above figure,
H = H F – H A.
(4) Pump head-discharge curve
(a) H p (Q1 ) = a0 + a1Q1 + a2Q12 (3.19)
a0 , a1, a2 are known constants
(b) or 𝐻𝑝 (𝑄1 ) = 𝑊/(𝜌𝑔𝑄) (3.20)
W = pump power (known)

8
(5) Linearization of the following Eq:
( W )i − ( H p )i + H = 0 (3.18)
• Eq. (18) is valid for any loops

• If H p = 0, H = 0  ( W )i = 0 (3.17)
• H is a known constant (= z ).
• Wi = (R Q 2 )i
• H p = a0 + a1Q + a2Q 2 or 𝐻𝑝 = 𝑊/
(𝜌𝑔𝑄)
 Eq(3.18) is a non-linear function of Q .
( W )i − ( H p )i + H = 0
Qi2 Qi2

It needs to be linearized for finding

the solution by an iterative method.
Let  (Q) = W (Q) − H p (Q) (3.21)
9
Taylor’s series
𝑓 ′ (𝑥0 ) 𝑓 ′′ (𝑥0 )
𝑓(𝑥 ) = 𝑓(𝑥0 ) + (𝑥 − 𝑥0 ) + (𝑥 − 𝑥0 )2 +..
1! 2!
Expanding  (Q) at Q (closing to Q0)
d
→  (Q) =  (Q0 ) + (Q − Q0 ) + ...
dQ Q 0

  (Q) = (W − H p ) Q +
0

 dW dH p 
 −  (Q − Q0 ) (3.22)
 dQ dQ  Q 0

 Eq. (3.18) is replaced by

( Wi − ( H p ) )Q +
0

 dWi dH p 
 −  (Q − Q0 )i + H = 0 (3.23)
 dQ dQ Q 0

Or
− ( Wi − ( H p ) )Q − H
Q = (Q − Q0 )i = (3.23a)
(dWi / dQ − dH p / dQ )Q
0

 We have now linear Q →can use iteration for Qi

10
If H p = a0 + a1Q + a2Q 2 ,
 (dH p / dQ )Q = a1 + 2a2Q0 (3.24)
0

Or if 𝐻𝑝 (𝑄1 ) = 𝑊/(𝜌𝑔𝑄)
 (𝑑𝐻𝑝 /𝑑𝑄) = −𝑊/𝜌g𝑄02 (3.25)
𝑄0

For W = R Q 2 →
dWi / dQi = 2 ( RiQi ) = 2 (Wi / Qi ) (3.26)
As Wi and Qi are always in the same sign,
dWi / dQi should always be positive values.

For a closed loop without pump, the above

equation is simplified as:
Q = − (Wi ) / 2 (Wi / Qi ) (3.27)
or Q = − (Wi ) / 2 ( R i Qi ) (3.27a)

Eqs. (3.23a), (3.27) and (3.27a) can be further

simplified if we considering sign convention for Q
& HL (or Wi):

11
• As W = R Q 2 , from this relation, it means that W
will always be positive. But actually, we have
known that it can be positive or negative,
depending on the sign convention. Therefore, to
keep Wi and Qi to have the same sign, Wi can be
written as:  Wi = Ri Qi Qi .
• From Eq. (3.26), it can be seen that as Wi and Qi
have the same sign, dWi/dQi will always be
positive, i.e. the denominator in (3.27) should
always be positive. But in (3.27a), the sign of the
denominator is the same as for Q (which can be
positive or negative, depending on its direction in
the loop). So we can use the absolute value of Q
as: ∑(2𝑅 ̅𝑖 |𝑄𝑖 |). In this way, the denominator
will always be positive.

Therefore, Eq. (3.23a) can be re-written as:

− ∑(𝑅̅𝑖 𝑄𝑖 |𝑄𝑖 |−(±𝐻𝑃 ))@𝑄 −∆𝐻
0
∆𝑄 = ∑((2𝑅̅𝑖 |𝑄𝑖 |)−𝑑𝐻𝑃 /𝑑𝑄)
(3.27b)
@𝑄 0

Eq. (3.27b) is valid for both closed and pseudo

loops.

12
The is the so-called Hardy-Cross Method.
For a closed loop without pump (and also ∆𝑯 =
𝟎), Eq. (3.27b) is simplified as:
− ∑(𝑅̅𝑖 𝑄𝑖 |𝑄𝑖 |)@𝑄0
∆𝑄 = ∑(2𝑅̅𝑖 |𝑄𝑖 |)@𝑄
(3.28)
0

• (Qi ) new for each pipe in a loop:

(Qi )new = (Qi )old + Q (3.29)
If a pipe belongs to two loops, (3.29) should be
modified. For loop I, we get QI and for loop II,
we get QII . Within the calculation of loop I:
(Q)3,new = (Q)3,old + QI − QII (3.30)
Within the calculation of loop II,:
(Q)3,new = (Q)3,old + QII − QI (3.31)

4
Q1 Q4
1

3 Loop 2
Loop I
Q3
Q5
Q2
2 5

13
9. Steps in Hardy-Cross Method
1) Assume initial values Q0 for all pipes
with continuity satisfied at all nodes,

2) For each loop, calculate ∆𝑄 using Eq.

(3.27b) or (3.28), depending on the type of
loops and if there are pumps or not. E.g.,
for an interior (or called closed-) loop
without a pump, we use Eq. (3.28):
− ( Ri Qi Qi )
Q = (3.28)
(2 Ri Qi )

For a pseudo-loop, Eq. (3.23a) should be

used:
− (( Ri Qi Qi ) − ( H p ) )Q − H
Q = (Q − Q0 )i = 0
((2 Ri Qi ) − dH p / dQ )Q
0

(3.27b)
3) Update Q in each pipe and in each loop by
Qi = (Q0 )i + Qi (3.29)

14
 Note: If a pipe belongs to two loops, Eq.
(3.30) and (3.31) should be used.

4) Checking ( 0.01    0.05)

 Qi − (Q0 )i
? (3.32)
 Qi
Yes: Stop iteration
No: Repeat 2 & 3 until (3.32) is satisfied.

Example 1
Piping system shown below.
Find flow rate distribution Q1, Q2 and Q3.

Assuming flow direction:

15
From Eq. (3.28),
− (R1 Q1 Q1 + R 2 Q2 Q2 + R 3 Q3 Q3 ) (1)
Q =
2(R1 Q1 + R 2 Q2 + R 3 Q3 )

Iterations steps:
1. Assuming Q1 = 25 L / s , from
continuity at the node → Q2 = 10 ,
Q3 = 25 L/s.
2. Find Q. (Note Q3 is negative in
the loop)
− (3  252 + 5  102 − 2  252 )
Q = = −3.21
2(3  25 + 5  10 + 2  25)

16
3. Correction to Q: Qi = (Q0 )i + Q
Q1 = 25 − 3.21 = 21.79, Q2 = 10 − 3.21 = 6.59,
Q3 = −25 − 3.21 = −28.21,
4 2nd iteration: Q = −0.2
Q1 = 21.5, Q2 = 6.59, Q3 = −28.41,

5 3rd iteration: Q = −0.0041

Q1 = 21.6, Q2 = 6.6, Q3 = −28.4

Example 2. Find Q1 , Q2 and H J

B
A

Water:  = 9800 N / m3
H = HB–HA = 25 – 5 = 20 m
17
 𝑊 200000 20.4
𝐻𝑝 = = = (1)
𝛾 (𝑄)1 9800(𝑄)1 (𝑄)1
𝑑𝐻𝑝 −𝑊 −20.4
 = = (2)
𝑑 (𝑄 )1 𝜌g(𝑄)21 (𝑄)21

W1 = R1 (Q)12 = 30(Q)12 (3)

dW1
 = 60(Q)1 (4)
d (Q )1

W2 = R2 (Q) 22 = 20(Q) 22 (5)

dW2
 = 40(Q ) 2 (6)
d (Q ) 2

Expression for Q is (from Eq3.23a)

− (W1 − H p + W2 ) − H
Q = (7)
dW1 dW2 dH p
+ −
d (Q)1 d (Q) 2 d (Q)1
Iterations:
Assume initial Q0,

Pipe 1: (Q0 )1 = 1 m3/s

18
Pipe 2: (Q0 )2 = (Q0 )1 − 0.5 = 0.5 m3/s
Substitute Q0 to (1)-(6) and from (7)
 Q = –0.345
Iteration (Q)1 + Q (Q) 2 + Q Q
1 –0.345
2 0.655 0.155 –0.024
3 0.631 0.131 +0.001
4 0.632 0.132 810-9
−9
Q = 8  10 is small enough.
 (Q )1 = 0.632 m3/s

 (Q )2 = 0.132 m3/s
From (1) → H p = 32.3 m

Hydraulic grade line at J:

Energy Eq between J and A:
p 
H =  + z  = 5 + H p − R1Q12 = 25.3m
 J
19