COURSE NUMBER: ME 323

Fluid Mechanics II

Pipeline system analysis

Course teacher
Dr. M. Mahbubur Razzaque
Professor
Department of Mechanical Engineering
BUET

1
 Frictional losses
It is convenient to express the pipe element frictional loss in the exponential form

in which hL is the head loss over length L of pipe, K is the resistance coefficient, Q
is the discharge in the pipe, and n is an exponent. Comparing the Darcy - Weisbach
relation with Eq. 11.2.1, one gets n = 2, and the resulting expression for the
resistance coefficient denoted by R is

where f is the friction factor and may be determined from the Moody diagram or
various correlations.

Swamee and Jain

Combining Equations 11.2.2 and 11.2.3, one finds that,

2
 Frictional losses
Equations 11.2.3 and 11.2.4 are valid over the ranges 0.01 > e/D > 10-8, 108 > Re >
5000. The fully rough regime, where Re has a negligible effect on f, begins at a
Reynolds number given by,

For the values of Re greater than this, the friction factor is a function only of e/D,
and is given by,.

Series Piping

Consider the series system shown above. It consists of N pipe elements with a
specified number of minor-loss components K associated with each ith pipe

3
element. A single minor loss is equal to hL = KV2/2g = KQ2/2gA2.

For many flow situations, it is common practice to neglect the kinetic-energy terms
at the inlet and outlet; they would be significant only if the velocities were
relatively high.

So, the energy equation applied from location A to location B is

in which Ri is the resistance coefficient for pipe i. The statement of continuity for
the series system is that the discharge in every element is identical, or

4
Replacing Qi with Q, Eq. 11.3.1 becomes

200 kPa

5
6
7
8
Solution

Use 1 ft3/s = 449 US GPM 1 US Galon = 231 in3 = 0.1336 ft3

Colebrook- White equation

9
10
11
 PARALLEL PIPE FLOW, EQUIVALENT PIPES
In case of parallel pipes, it is advantageous to replace the set of parallel pipes by a single
"equivalent pipe." This equivalent pipe, which is devised so it has the same head loss as
the original set of parallel pipes and conveys the same total discharge, will in some
cases allow the analyst to avoid the use of iteration in seeking a solution.

Assume that pipes 1 and 2 are parallel pipes with frictional head losses described by
KQn, then the equivalent pipe (unsubscripted) must satisfy

12
13
 Parallel Piping

A parallel piping arrangement is essentially an arrangement of N pipe elements

joined at A and B with K minor loss components associated with each pipe
element i.

The continuity equation applied at either location A or B is given by

The algebraic sum of the energy grade line around any defined loop must be
zero.
14
As in the case of series piping, it is usual to assume that V2/2g << (p/g + z).
Hence, for any pipe element i, the energy equation from location A to B is

It is possible to convert the minor loss terms using an equivalent length. For
each pipe element i the equivalent length Le for K minor loss components is

Thus Eq. 11.3.5 simplifies to the form

in which the modified pipe resistance coefficient Ri is given by

The right-hand side is equivalent to the term Ri + K/(2gAi2) and Ri is given by

15
A solution employing the method of successive substitution is developed in the
following manner. Define the variable W to be the change in hydraulic grade
line between A and B; that is,

Then Eq. 11.3.7 can be solved for Qi in terms of W as

Equations 11.3.4 and 11.3.9 are combined to eliminate the unknown discharges
Qi, resulting in

The remaining unknown W is taken out of the summation sign since it is the
same in all pipes. Solving for W in Eq. 11.3.10 results in

16
An iterative procedure can be formulated to solve for W and the discharges Qi
as follows:

1. Assume flows in each line to be in the completely rough zone, and compute
an initial estimate of the friction factors in each line using Eq. 11.2.6.

2. Compute Ri for each pipe and evaluate W with Eq. 11.3.11.

3. Compute Qi in each pipe with Eq. 11.3.9.

4. Update the estimates of the friction factors in each line using the current
values of Qi and Eq. 11.2.6.

5. Repeat steps 2 to 4 until the unknowns W and Qi do not vary within a desired
tolerance.
Note that if friction factors are in the completely rough zone so that they are
independent of the discharge and therefore constant, steps 4 and 5 are
unnecessary and a solution results on the first iteration. The technique is
illustrated in the following example.
17
18
19
20
 Branch Piping

The branching network shown is made up of three elements connected to a

single junction. In the analysis, one assumes the direction of flow in each
element; then the energy equation for each element is written using an
equivalent length to account for minor losses:

21
The piezometric heads at locations A, C, and D are considered known. The
unknowns are the piezometric head at B and the discharges Q1, Q2, and Q3. The
additional relation is the continuity balance at location B, which is

Thus there are four equations with four unknowns. One convenient ad hoc
method of solution is outlined below and illustrated in Example 11.4:

1. Assume a discharge Q1 in element 1 (with or without a pump). Establish

the piezometric head H at the junction by solving Eq. 11.3.12 (a category 1
problem).
2. Compute the discharge Qi in the remaining branches using Eqs. 11.3.13
and 11.3.14 (a category 2 problem).
3. Substitute the Qi into Eq. 11.3.15 to check for continuity balance.
Generally, the flow imbalance at the junction Q will be nonzero. In Eq.
11.3.15, Q = Q1 - Q2 - Q3.
4. Adjust the flow Q1 in element 1 and repeat steps 2 and 3 until Q is within
desired limits.

22
If a pump exists in pipe 1 as shown below, Eq. 11.3.12 is altered in the manner

An additional unknown, namely, the pump head HP, is introduced. The

additional necessary relationship is the head-discharge curve for the pump.

23
24
25
26
27
28
29
30
31
EXAMPLE
A pipeline consists of two pipe segments in series (Fig. P11.20). The specific
gravity of the fluid is 0.81. If pump A has a constant power input of 1 MW, find
the discharge, the pressure head in pumps A and B, and the required power for
pump B. The minimum allowable pressure on the suction side of pump B is 150
kPa, and both pumps have an efficiency of 0.76.

SOLUTION
Energy eqn. from junction at A to suction side of pump at B:

where HPA = pump head and pB = pressure. The only unknown in the relation
is Q.
32
Pipeline resistances

Putting these values in the energy equation,

Solving,

The energy eqn. from suction side of pump at B to the downstream reservoir:

Pipeline resistances

Energy equation

33
EXAMPLE
The solution of flow of water in a pipe network is shown below. Compute:
(a) The hydraulic grade line throughout the system.
(b) The pressure at each node.

SOLUTION
The continuity of flow gives the following

34
(a) The hydraulic grade lines at the nodes (i.e., the nodal piezometric heads) can
now be computed:

(b) The pressure at each node.

35
Example:
Consider the small 5-pipe network shown below. Since no supply sources are
shown for this network, only J - 1 junction continuity equations are available.
Writing these continuity equations for nodes 1, 2, and 3 leads to

The continuity equation at node 4 is - Q3 - Q2 + Q5 = 0. However, this equation

is not independent of the other three nodal equations since it is, except for sign,
the sum of these three equations.

Now let us use the Hazen-Williams equation to define the head loss in each
pipe. In expressing these head losses, the exponential equation will be used. The
coefficients are the following:

36
The energy equations around the two loops may be written as

These two work-energy equations are obtained by starting at nodes 1 and 2,

respectively, and traversing the respective loops I and II in the clockwise
direction. If the assumed direction of flow opposes this traverse, a minus sign
precedes the head loss term for that pipe.

37
Now, the elevations H of the HGL are determined by starting where the HGL is
known, in this case at node 1, and computing the head losses in the pipes that
join this node; then the frictional head losses hf are subtracted from the known
38
values of H, etc. until all of the nodal heads have been determined. The
pressures are then determined by subtracting the nodal elevations z from the
heads H and multiplying this by the specific weight, i.e., p = γ(H - z). The tables
which follow present the computed values for this network:

39