PIPENET VISION

TRAINING MANUAL
TRANSIENT MODULE

CHAPTER 11

ADVANCED TOPICS

Revision 2.3, February 2016

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 1 OF 180 REVISION 2.3, FEB 2016

1. MISCELLANEOUS TOPICS: .................................................................................................................... 2

1.1. TIME STEP: .......................................................................................................................................... 2
1.2. INITIAL STATE OF THE SYSTEM: ........................................................................................................... 14
1.3. TABULAR OUTPUT: ............................................................................................................................. 15
2. PUMPS: ................................................................................................................................................... 18
2.1. SIMPLE PUMP .................................................................................................................................... 19
2.2. INERTIAL PUMP .................................................................................................................................. 25
2.3. TURBO PUMPS ................................................................................................................................... 33
3. TANKS: ................................................................................................................................................... 42
3.1. SIMPLE TANK: .................................................................................................................................... 43
3.2. ACCUMULATOR .................................................................................................................................. 47
3.3. SURGE TANK: .................................................................................................................................... 53
3.4. RECEIVING VESSEL: ........................................................................................................................... 67
4. FORCE CALCULATION: ........................................................................................................................ 76
4.1. SIMPLE FORCE:.................................................................................................................................. 76
4.2. COMPLEX FORCES: ............................................................................................................................ 86
4.3. TOTAL FORCE AND DYNAMIC FORCE ................................................................................................... 92
4.4. FORCES OUTPUT FILE ........................................................................................................................ 94
5. CONTROL SYSTEMS ............................................................................................................................. 98
5.1. THE IMPORTANCE OF RUN-IN TIME: ................................................................................................... 100
5.2. PRESSURE CONTROL SYSTEM. ......................................................................................................... 102
5.3. LIQUID LEVEL CONTROL SYSTEM ...................................................................................................... 117
6. CHECK VALVES: ................................................................................................................................. 130
6.1. NON-RETURN VALVE (SIMPLE CHECK VALVE) ..................................................................................... 132
6.2. CHECK VALVE (DYNAMIC CHECK VALVE)............................................................................................ 133
6.3. INERTIAL CHECK VALVE .................................................................................................................... 136
7. CAVITATION MODELLING: ................................................................................................................. 140
7.1. VAPOUR CAVITATION MODEL ............................................................................................................ 140
7.2. CHANNEL CAVITATION MODEL: ......................................................................................................... 149
8. DRY PIPE MODELLING: ...................................................................................................................... 151
8.1. BASE NETWORK: .............................................................................................................................. 151
8.2. OPTIMIZE NETWORK: ........................................................................................................................ 157
8.3. LOOPED NETWORK: .......................................................................................................................... 161
8.4. IMPROVING CALCULATION ACCURACY: .............................................................................................. 162
9. RELIEF/SAFETY VALVE MODELS: .................................................................................................... 168
9.1. USING A BURSTING DISC: ................................................................................................................. 168
9.2. USING A REGULATOR: ...................................................................................................................... 170
9.3. USING A LIQUID SURGE RELIEF VALVE .............................................................................................. 173
10. MILE POST DATA ............................................................................................................................ 176

This chapter covers some of the advanced topics for using PIPENET. The networks in this
chapter are very simple because the objective is to show a large number of examples
which can be quickly set up. This chapter can be used for training. It can also be used as
a reference document even without an associated training course.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 2 OF 180 REVISION 2.3, FEB 2016

1. MISCELLANEOUS TOPICS:

1.1. Time Step:

What are Numerical Stability and Maximum Time Step for Numerical Stability?
A numerical solution procedure often has a maximum limit on the time step. If we used a
time step bigger than this, the numerical procedure may become unstable. Essentially the
term numerical instability means any errors in the solution at one time step will increase at
the next time step. After a few time steps the real solution could be completely swamped
so that the calculated solution bears no resemblance to the real solution.
In the case of the method of characteristics which is used in PIPENET Transient module,
the maximum time step for numerical stability is easy to understand intuitively. A
knowledge of numerical methods is not necessary. Let us consider a single pipe filled with
a liquid. If we create a pressure disturbance at one end of the pipe, it will travel at the
wave speed along the pipe, reflect at the other end and return. In other words, the
disturbance will travel back and forth along the pipe at the wave speed. It is clear that it is
essential to model the reflection of the disturbance because without that we will not have
oscillations in the pressure.
Let us now consider what the maximum time step for the calculation should be. It is
obvious that the maximum time step has to be the time it takes for the wave to reach from
one end of the pipe to the other. This is because a pressure disturbance created at one
end of the pipe will go past the other end of the pipe if we took a bigger time step. It will
not be possible to predict the reflection of the wave if that happened. In the case of a
single pipe we can define the following.
Time step for numerical stability = length of pipe/wave speed
In the case of a system with multiple pipes, each pipe would have its own time step for
numerical stability. The smallest of such time steps would be the time step for numerical
stability for the complete system. In this context it is important to note that the wave speed
is usually different for different pipes.
How can We Look at the Time Step Determined by PIPENET?
It would be of interest to be able to see what time step PIPENET will use and how it
arrived at that time step. This can be done by following the steps below.

Choose the tab ‘Output’

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 3 OF 180 REVISION 2.3, FEB 2016

Click on tab
Output

Click on the Timestep button

Click on
Timestep
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 4 OF 180 REVISION 2.3, FEB 2016

How does PIPENET Determine the Time Step for Calculation?

PIPENET determines the calculation time step taking into account many parameters.
The tabular output time step specified by the user
Graphical output time step specified by the user
Forces output time step specified by the user
The maximum time step for numerical stability determined by PIPENET
The ‘Calculation time step’ must be less than all the above. For convenience it is also
chosen to be an integral fraction of the tabular output time step.
In this case the time step for tabular output is 1 sec and that is divided into 15 steps. So
the time step for calculations is 1/15 = 0.066666… PIPENET Chose 15 steps because if it
had chosen 14 steps then the time step would have been 1/14 = 0.071429. This is bigger
than the time step for numerical stability which is 0.07009724.
We can see that if we changed the tabular output time step to 0.5 then the time step for
calculations changes to 0.0625 which is 0.5/8. This means 8 time steps are taken for each
tabular output time step of 0.5 sec.
Sub-division of a Single Pipe and the Grid Size:
Let us still stay with a single pipe. The above time step for numerical stability will not result
in the pipe being sub-divided because it is based on the total length of pipe. In order to
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 5 OF 180 REVISION 2.3, FEB 2016

make more accurate calculations and in order to follow the pressure wave along the pipe,
we need to subdivide the pipe.
If we specify a time step for calculation, which is considerably smaller than the time step
for numerical stability, then the pipe will be sub-divided. This is because the pressure
wave will only move a fraction of the length of pipe in one time step. Each such sub-
division of the length is called a grid and its length is called the grid size. We can say,
Grid size = the length travelled by the wave during the time step
= wave speed x time step specified by the user
In this way the pipe is sub-divided into sections or grids. This is important for accuracy. It
also enables us to follow the movement of the pressure wave along the pipe.
There are two ways this can be done.
1. Specify the time step for graphical output which is much smaller than the time step
for numerical stability.

This can be set to

a small value

The time step for calculation cannot be bigger than the graphical output time step.
For that reason the calculation time step will be taken as the graphical output time
step.
The drawback of this method is the fact that it is not possible to perform the
calculations with a short time step, with the graphical output using a bigger time
step. They both have to be the same.
2. The second option is simply to select user defined time step.

The advantage of this approach is to different values can be specified for the time
step for calculation and the time step for output.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 6 OF 180 REVISION 2.3, FEB 2016

Select user
defined time step

Specify the
time step

How to Select a Reasonable Time Step (Hints):

Generally small time step can accurately describe the dynamic response of valves and
devices; decrease the interpolation error and achieve more accurate results. However,
small time step increases the grid number and total calculation amount so that slows down
the calculation speed significantly. For example, the time step is cut to 10%, the number of
grids and the amount of calculation will both increase about 10 times. As the results the
calculation time may increase up to 10x10 = 100 times. Therefore, it is very important to
find a compromise time step which gives consideration of both calculation accuracy and
calculation speed. The following steps may be helpful to find a reasonable time step.
1. Select an initial time step with reasonable calculation speed.
2. Cut the time step to half manually and redo the calculation.
3. Compare the calculation results from the steps 1 and 2, e.g. global maximum and
minimum pressures of the system. If similar, it indicates the reasonable time step is
obtained. Otherwise, repeat the steps 2 and 3 until the calculation results are not
affected by the time step significantly.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 7 OF 180 REVISION 2.3, FEB 2016

Multiple Pipe Systems:

The above paragraphs give an introduction to how PIPENET calculates the time step and
some hints on how to over-ride the time step calculated by PIPENET. A network of pipes
will typically contain pipes of different lengths and different diameters. Usually pipes of
different diameters will have different wave speeds because the internal pipe diameter and
the wall thickness will be different. It is also possible that the elasticity is different because
the pipe material is different in different parts of the system. For these reasons the wave
speed of different pipes in the network typically will be different.
The formula used for the wave speed is shown below.

where:

is the wavespeed,
is the fluid bulk modulus,
is the internal pipe diameter,
is the Young's Modulus for the pipe material,
is the pipe wall thickness,
is the pipe distensibility (restraint factor),
the fluid density.

In general the wave speed and hence the time step will be different for different pipes. For
this reason, it will often be unnecessary for the user to sub-divide the pipes in the system
with multiple pipes. The time step for calculation will be determined by the pipe with the
shortest. This in turn will result in the longer pipes in the system being automatically sub-
divided. Nevertheless, it is a good idea to look at how pipes are sub-divided as shown in
the above section ‘How can we look at the time step determined by PIPENET?’
An example is shown below:
Data for pipes:

Calculation Timestep window:

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 8 OF 180 REVISION 2.3, FEB 2016

What is User Defined Time Step?

PIPENET automatically creates the wave speed and the time step for every pipe. Using
the time step for every pipe and the output time steps it will calculate the time step for
calculation. The user can over-ride this time step calculated by PIPENET by specifying a
user defined time step.
What Happens if a Higher Value is Specified for the User Defined Time Step than the
Time Step for Numerical Stability?
It is clear that if the user defined time step is smaller than the time step for numerical
stability, then the user defined time step will be used for the calculation. On the other
hand, if the user defined time step in bigger what will PIPENET do.
As discussed under the topic of Multiple Pipe Systems, PIPENET will calculate the time
step for every pipe in the system. As we know the wave speed could be different for
different pipes. For this reason the pipe with the smallest value for the length/wave speed
will have the shortest time step.
PIPENET will automatically use the ‘short pipe’ model for every pipe which has a time step
shorter than the user defined time step. This is a simplified model for the pipe. Essentially
the short pipe model ignores the elasticity of the pipe and the compressibility of the fluid.
This means any disturbance at one end of the pipe will be instantly reflected at the other
end of the pipe. The concept of the wave speed is not used in the model.
For every pipe which has a time step bigger than the user defined time step PIPENET will
use the elastic pipe model.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 9 OF 180 REVISION 2.3, FEB 2016

An example of the Calculation Timestep window is shown below. Please note that the
user defined time step is bigger than the time step for numerical stability.

What Does ‘Interpolation error (%)’ Mean?

This is NOT the error in the solution. This is a measure of the interpolation which arises
because for a given pipe, the length is not necessarily an integral multiple of the distance
by which the pressure wave moves in one time step. Let us take the example we saw
earlier:
Data for pipes:

Calculation Timestep window:

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 10 OF 180 REVISION 2.3, FEB 2016

Let us consider pipe label 4.

Length = 6.378 m
Number of divisions = 6
Length of each division = 6.378/6
= 1.063 m

On the other hand the pressure wave does not move exactly this distance during each
time step.
Time step (user defined) = 0.002 sec
Wave speed = 484.018 m/sec
Distance per time step = 0.002 x 484.018
= 0.968036 m

These are not exactly the same although they have the same order of magnitude.
Interpolation error (%) = 100 x (1.063 – 0.968036)/1.063
= 8.93358 %
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 11 OF 180 REVISION 2.3, FEB 2016

Why Are Short Pipes Undesirable?

In a word short pipes can potentially increase calculation time very significantly without
much benefit.
Case 1: Single pipe.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 12 OF 180 REVISION 2.3, FEB 2016

Case 2: Add a short pipe (0.01 m long)

This not only decreases the time step by a factor of 10,000, it also increases the number of
grids by a factor of 10,000. In other words, the amount of calculation which needs to be
done is 10000 x 10000 = 1.0e08 times more.
Why is this of little benefit?
In order to understand this, we must look at the cause of pressure surge intuitively. Please
note that the discussion below must not be used for calculations. It is merely for
explaining the concept intuitively.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 13 OF 180 REVISION 2.3, FEB 2016

Consider the system below:

The momentum of the liquid in the pipe is , where is the mass of fluid in the pipe
and is the velocity.

Suppose the valve closes in . Then the rate of change of momentum is . We

know that the rate of change of momentum is force.
If the area of cross-section of the valve is a, then the increase in the pressure will be equal
to .
Now consider the two-pipe system.

The mass of the liquid in the short pipe is negligible in comparison to the mass of liquid in
the longer pipe. Hence the momentum is also negligible. So, the contribution of the first
pipe to the pressure surge is negligible.
So essentially, PIPENET has to do is increased by a factor of 1.0e08 for little benefit.
For this reason, short pipes should be avoided whenever possible.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 14 OF 180 REVISION 2.3, FEB 2016

1.2. Initial State of the System:

This is defined using the following dialog box.

The initial steady state is sometimes difficult to determine. PIPENET has a number of
tools to assist the user in determining the initial steady state. The Initial State dialog box
shown above is for this purpose.
The principle of ‘Run-in time’ is one of the most important principles to understand. To put
it quite simply, it a very good way of generating sound initial guesses. We can start from
the meaning of the term steady state. The mathematical definition of steady state is that
the partial derivatives of all the variables with respect to time are 0. These equations are
sometimes difficult to solve because it is not easy to find good initial guesses for the
iterations. In contrast if the state of the system is known at one point in time, it is usually a
stable process to take the next step in time. So, rather than start with the mathematical
definition of steady state we can start with a more pragmatic definition. Basically, steady
state would be ultimately reached if we start with a random state of the system and run it
for an infinitely long time without changing the operating conditions. In fluid flow terms it
means if we start with a random state of the system and run it for sufficiently long time, the
system will reach its steady state. Run-in time is the time we allow the simulation to run in
PIPENET using the boundary conditions at time = 0. So, the run-in time can be used in
two different ways.

(i) Use a long run-in time which is long enough for the steady state to be reached.
The “Calculate initial steady state” button is usually unchecked under these
conditions. This is the preferred option if the system has control loops.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 15 OF 180 REVISION 2.3, FEB 2016

(ii) Use a run-in time which is shorter but which would nevertheless produce a good
initial guess starting from a random guess. The “Calculate initial steady state”
button must be checked under these conditions.

“Calculate initial steady state” solves the system of equations arrived at by

setting all the derivatives to 0. The run-in time can generate good initial guesses
for solving these equations.

There are many other ways the initial state can be specified or the initial guesses
generated. These tools are not often used but they are there if needed.

(i) Then initial guess file has two purposes.

(a) It can be the initial starting point for the simulations. This could for example be
used as a re-start file. If this is the case the run-in time should be set to 0 and
the “Calculate initial steady state” button is usually unchecked.

(b) The user can specify the initial starting guess for finding the steady state. This
could be used in conjunction with the run-in time and/or “Calculate initial steady
state”.

The initial guess file has a pre-defined structure and data can be
entered/modified by the user. The “Make initial guess file” and “Edit initial
guess file” buttons are for this purpose.

(ii) Default.

PIPENET uses its own default values using which the steady state calculation process will
start. It is generally use in conjunction with run-in time and Calculate initial steady state
button being activated.

(iii) Static (all flows zero)

Remember that static means all flows are 0 and is different from steady state. In some
cases it may be known that all the flows will be 0 at the start. In such cases this may be a
better to generate the initial guesses on the assumption that the flows are 0.
It is also important to bear the following point in mind. It is usually a good idea to run
the system in PIPENET without introducing events (changes to the boundary conditions)
for a certain length of the simulation. We can then assess if steady state has been
reached by looking at the graphs.

1.3. Tabular Output:

Tabular output has a useful purpose in that it gives exact numbers, rather than graphical
results.
By default it shows the following
Global maximum and minimum for the pressure thought the system.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 16 OF 180 REVISION 2.3, FEB 2016

The maximum and minimum pressures for every item in the network, at the input and
output nodes.
For pipes the maximum and minimum pressures which occur at the internal points
The maximum and minimum pressure for each pipe schedule which is used in the
system
The maximum/minimum pressure for each pipe schedule would be important under certain
circumstances. If the system has several pipe schedules, the worst point may not be at
the place where the maximum pressure occurs. There may a weaker pipe schedule in the
somewhere else where the pressure may be higher than the design pressure. Extracts
from the tabular output file are shown below.

PRESSURE EXTREMA
----------------

Maximum pressure is 18.4501 bar G

on pipe 36 at the inlet
at time 10.10000 seconds

Minimum pressure is 0.00000 bar G

on pipe 24 at the inlet
at time 0.000000 second

MAXIMUM/MINIMUM PRESSURE
------------------------

Inlet max. Inlet min. Outlet max. Outlet min. Middle max. Middle min.
Component pressure Time pressure Time pressure Time pressure Time pressure Time pressure Time

bar G seconds bar G seconds bar G seconds bar G seconds bar G seconds bar G seconds

----------------------------------------------------------------------------------------------------------------------------------

PIPE
3 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
4 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.159E+02 0.10E+02 0.152E+02 0.95E+00
5 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.156E+02 0.10E+02 0.152E+02 0.95E+00
6 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
7 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
9 0.160E+02 0.10E+02 0.156E+02 0.95E+00 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.160E+02 0.10E+02 0.156E+02 0.10E+01
10 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.160E+02 0.10E+02 0.156E+02 0.10E+01
8 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.160E+02 0.10E+02 0.156E+02 0.95E+00 0.160E+02 0.10E+02 0.155E+02 0.95E+00
11 0.160E+02 0.10E+02 0.156E+02 0.95E+00 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.160E+02 0.10E+02 0.152E+02 0.95E+00
12 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.151E+02 0.10E+02 0.147E+02 0.95E+00 0.156E+02 0.10E+02 0.147E+02 0.95E+00
13 0.151E+02 0.10E+02 0.147E+02 0.95E+00 0.139E+02 0.10E+02 0.135E+02 0.95E+00 0.151E+02 0.10E+02 0.135E+02 0.95E+00
14 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.156E+02 0.10E+02 0.152E+02 0.10E+01 0.160E+02 0.10E+02 0.152E+02 0.10E+01
15 0.156E+02 0.10E+02 0.152E+02 0.10E+01 0.142E+02 0.10E+02 0.135E+02 0.10E+01 0.156E+02 0.10E+02 0.135E+02 0.10E+01
16 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.160E+02 0.10E+02 0.155E+02 0.95E+00
18 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.155E+02 0.10E+02 0.151E+02 0.95E+00 0.159E+02 0.10E+02 0.151E+02 0.95E+00
20 0.151E+02 0.10E+02 0.147E+02 0.95E+00 0.141E+02 0.10E+02 0.137E+02 0.95E+00 0.151E+02 0.10E+02 0.137E+02 0.95E+00
21 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.159E+02 0.10E+02 0.144E+02 0.90E+00
23 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00
26 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00
27 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.150E+02 0.10E+02 0.144E+02 0.90E+00
28 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.150E+02 0.10E+02 0.146E+02 0.80E+00
30 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.128E+02 0.10E+02 0.124E+02 0.75E+00 0.150E+02 0.10E+02 0.124E+02 0.75E+00
31 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.147E+02 0.10E+02 0.144E+02 0.80E+00 0.150E+02 0.10E+02 0.144E+02 0.80E+00
17 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
19 0.155E+02 0.10E+02 0.151E+02 0.95E+00 0.151E+02 0.10E+02 0.147E+02 0.95E+00 0.155E+02 0.10E+02 0.147E+02 0.95E+00
29 0.150E+02 0.10E+02 0.146E+02 0.80E+00 0.150E+02 0.10E+02 0.146E+02 0.90E+00 0.150E+02 0.10E+02 0.146E+02 0.80E+00
2 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
33 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.160E+02 0.10E+02 0.156E+02 0.10E+01 0.160E+02 0.10E+02 0.155E+02 0.95E+00
34 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
35 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
22 0.147E+02 0.10E+02 0.144E+02 0.80E+00 0.130E+02 0.10E+02 0.127E+02 0.80E+00 0.147E+02 0.10E+02 0.127E+02 0.80E+00
24 0.000E+00 0.00E+00 0.000E+00 0.00E+00 0.161E+01 0.59E+01 0.161E+01 0.10E+02 0.161E+01 0.59E+01 0.000E+00 0.00E+00
25 0.150E+02 0.10E+02 0.146E+02 0.90E+00 0.148E+02 0.10E+02 0.144E+02 0.90E+00 0.150E+02 0.10E+02 0.144E+02 0.90E+00
1 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.159E+02 0.10E+02 0.152E+02 0.95E+00
32 0.159E+02 0.10E+02 0.155E+02 0.95E+00 0.157E+02 0.10E+02 0.153E+02 0.95E+00 0.159E+02 0.10E+02 0.153E+02 0.95E+00
37 0.157E+02 0.10E+02 0.153E+02 0.95E+00 0.157E+02 0.10E+02 0.153E+02 0.95E+00 0.157E+02 0.10E+02 0.153E+02 0.95E+00
38 0.154E+02 0.10E+02 0.151E+02 0.95E+00 0.150E+02 0.10E+02 0.146E+02 0.90E+00 0.154E+02 0.10E+02 0.146E+02 0.90E+00
36 0.185E+02 0.10E+02 0.182E+02 0.12E+01 0.154E+02 0.10E+02 0.151E+02 0.95E+00 0.185E+02 0.10E+02 0.151E+02 0.95E+00
--------------------

PUMP
2 0.161E+01 0.59E+01 0.161E+01 0.10E+02 0.185E+02 0.10E+02 0.182E+02 0.12E+01
--------------------

VALV
1 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.000E+00 0.00E+00 0.000E+00 0.00E+00
3 0.550E+01 0.00E+00 0.550E+01 0.00E+00 0.159E+02 0.10E+02 0.155E+02 0.95E+00
2 0.156E+02 0.10E+02 0.152E+02 0.95E+00 0.000E+00 0.00E+00 0.000E+00 0.00E+00
4 0.157E+02 0.10E+02 0.153E+02 0.95E+00 0.000E+00 0.00E+00 0.000E+00 0.00E+00
5 0.154E+02 0.10E+02 0.151E+02 0.95E+00 0.000E+00 0.00E+00 0.000E+00 0.00E+00
6 0.142E+02 0.10E+02 0.135E+02 0.10E+01 0.000E+00 0.00E+00 0.000E+00 0.00E+00
--------------------
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 17 OF 180 REVISION 2.3, FEB 2016

NRTV
1 0.154E+02 0.10E+02 0.151E+02 0.95E+00 0.154E+02 0.10E+02 0.151E+02 0.95E+00
--------------------

VBRK
1 0.139E+02 0.10E+02 0.135E+02 0.95E+00 0.139E+02 0.10E+02 0.135E+02 0.95E+00
2 0.142E+02 0.10E+02 0.135E+02 0.10E+01 0.142E+02 0.10E+02 0.135E+02 0.10E+01
3 0.141E+02 0.10E+02 0.137E+02 0.95E+00 0.141E+02 0.10E+02 0.137E+02 0.95E+00
4 0.128E+02 0.10E+02 0.124E+02 0.75E+00 0.128E+02 0.10E+02 0.124E+02 0.75E+00
5 0.130E+02 0.10E+02 0.127E+02 0.80E+00 0.130E+02 0.10E+02 0.127E+02 0.80E+00

PRESSURE EXTREMA FOR PIPE TYPES

-------------------------------

Pipe type Max. pressure Time Pipe id Position Min. pressure Time Pipe id Position
bar G seconds metres bar G seconds metres
------------------------------------------------------------------------------------------------------------------------

PN20 GRE PIPE<+> 0.185E+02 0.101E+02 36 Inlet 0.000E+00 0.000E+00 24 Inlet

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 18 OF 180 REVISION 2.3, FEB 2016

2. PUMPS:

There are three types of pump in PIPENET. Typically the data would be input into a library
because pump curves are often used several times in one network or indeed in several
networks. Pump model is selected on the basis of three factors: available pump curves,
the exact network and the scenarios.
Simple pump: the performance curve H-Q (head – flow rate) is required in the pump
library. Simple pump is used to model variable speed pump within the working range,
i.e. pump head, flow rate, rotation speed and driving torque are positive. This pump
model only requires less data. However, the pump speed during run up and/or spin
down has to be input by the user. In addition, this pump cannot model reverse flow or
reverse rotation. The calculation accuracy also deteriorates beyond the working
range.
Initial pump: the pump curves H-Q (head – flow rate) and P-Q (power – flow rate)
are required in the library. Alternatively, the pump efficiency curve -Q (efficiency –
flow rate) could be input. This is used by PIPENET to calculate the power curve P-Q,
except the data at zero flow rate where the pump power has to be input by the user.
In addition the motor torque curve T-S (torque – speed) is required when the start-up
scenario is modelled. Similar to the simple pump, the initial pump cannot simulate
reverse flow or reverse rotation, and the calculation accuracy deteriorates beyond the
working range. The main advantage of the inertial pump is that it can calculate pump
speed automatically during pump start-up or trip, whilst the pump speed of simple
pump has to be set manually.
Turbo pump: the input Sutter curves WH and WB should cover all four quadrants,
i.e. pump head, flow rate, rotation speed and driving torque can be positive or
negative in any combination. Therefore, it can be used to model reverse flow or
reverse rotation. In addition, pump trip can be calculated automatically. This is the
most comprehensive pump model in PIPENET. However, Sutter curves may not be
readily available.
The following network is composed of a pump, a pipe and a high-position tank. Please
explain why the turbo pump model is recommended in the scenario of pump trip. If a fast-
acting check valve is installed at the pump output, can we use simple pump or initial pump
to model pump trip? Please explain the difference.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 19 OF 180 REVISION 2.3, FEB 2016

2.1. Simple Pump

This uses the normal pump performance curve which is usually supplied by the vendor.
This model can be used as a variable speed pump. The speed can be specified by the
user or set by a control loop. A simple pump can be used in cases where the head and
flow rate are always positive throughout the scenario.

In this example we use metric units

Points taken from the vendor’s pump performance curve are shown below.

Flowrate, lit/min Pressure, bar

0 15
4000 14
8000 12
10000 10

The input is shown by showing the dialog box sequentially below. The data in the dialog
boxes require little description.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 20 OF 180 REVISION 2.3, FEB 2016

The pump curve is input into the library as shown below.

We saw earlier the principles used for determining the time step. Usually if there is only
one pipe, the user has to provide the calculation time step which is smaller than the time
step determined by PIPENET. Remember that in general the maximum time for numerical
stability would be based on not sub-dividing the pipe if there was only one pipe. This
means the pipe would be regarded as a single grid. On the other hand, if the network
includes several pipes of different lengths the time step for numerical stability would be
usually based on the length of the shortest pipe. That is not the case here and so we have
to provide a calculation time step which is shorter than the time step for numerical stability.

The fluid used is water at 20 °C (default)

The pipe schedule used in ANSI 36.10 schedule 40
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 21 OF 180 REVISION 2.3, FEB 2016

The other data used in the system are shown below.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 22 OF 180 REVISION 2.3, FEB 2016

We learn how to model a valve closing, staying closed for a while and then opening again.
For this purpose we use the linear function for the boundary condition. PIPENET has
several options for boundary conditions. However, most commonly one of the following
three functions is used:
Constant
Power Ramp
Linear Profile

The output files are specified as shown below:

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 23 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 24 OF 180 REVISION 2.3, FEB 2016

The result is shown in the graph below.

The simple pump model has two significant limitations.

(i) The simple pump model should not generally be used to model pump trip. This is
because when a pump trips the pressure at the outlet of the pump usually drops
below the pressure at the inlet of the pump. The pressure difference across the
pump is negative even though the flowrate itself may be positive. Such an
operating point cannot lie on a simple pump performance curve.

(ii) If it is to be used in pump start up or shut down conditions the user has to define the
ramp up or spin down profile. Vendors sometimes provide these times. However,
typically the network characteristics will have an influence on the times and for this
reason the data provided by the vendor could be inaccurate.
For these reasons two other pump models have been made available with PIPENET,
namely, inertial pump and turbo pump.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 25 OF 180 REVISION 2.3, FEB 2016

2.2. Inertial Pump

The inertial pump model is intended to predict the behaviour of the pump during pump spin
down or run up. With a simple pump the run up or spin down time has to be input by the
user. Usually this cannot be calculated accurately by the pump vendor because these
times would depend on the network system. They are not characteristics of the pump
working in isolation. For example, if the back pressure of the system builds up quickly it
will take longer for a pump to achieve its full speed. Typically the vendor would have no
knowledge of the network system. The inertial pump can predict the run up and spin down
times. This model can be used if it is necessary to estimate the pump run up or spin down
time.
Additional data is necessary for the inertial pump model for this purpose.
The moment of inertia (typically pump impeller + motor rotor + retained liquid) and the
rated rpm of the pump must be given for both ramp up and spin down.
For pump ramp up and spin down, the Head vs Flowrate (H-Q) and Power vs Flowrate (P-
Q) curves need to be entered. Instead of the power, it is possible to input the efficiency for
all flowrates except 0. For the flowrate 0 point the power must be entered by the user
because it cannot be calculated using the efficiency.
In addition, for pump run up the motor synchronous speed and the Torque vs Flowrate (T-
Q) curve must also be specified.
In the example we will only consider pump trip. We will also use the efficiency curve as
the input rather than the power curve. For modeling pump spin down the actual time at
which power is lost is specified by using the step function boundary condition.

Use metric units for all cases

Flowrate, lit/min Pressure, bar fluid Efficiency Power, KW

0 15 100
4000 14 90
8000 12 80
10000 10 60

As mentioned above, it is not possible to calculate the power at 0 lit/min given the
efficiency. So at 0 lit/min flowrate it is a mandatory requirement to specify the power.
Alternatively, it is possible to input points from the power against flowrate curve.
Base case – Pump Trips at 8 seconds.
In addition to the above data, the following information is also necessary.
Moment of inertia (Impeller + Motor rotor + retained liquid) = 1000 kg m2
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 26 OF 180 REVISION 2.3, FEB 2016

Rated speed = 360 rpm

The input dialog box for the Inertial Pump for input into the library is shown below.

The rest of the data, including the relevant dialog boxes is shown below.

Use water at 20 °C for the fluid (default)

Use ANSI 36.10 schedule 40 for pipe type
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 27 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 28 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 29 OF 180 REVISION 2.3, FEB 2016

Case 2: Effect of Low Moment of Inertia of the Pump and Bigger Pipe with +ve
Elevation
Make the following changes and see the effect.
Change the Moment of Inertia of the pump to 10 kg m 2, pipe diameter to 300 mm and the
pipe length to 4000 m with an elevation change of 30 m.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 30 OF 180 REVISION 2.3, FEB 2016

Note elevation
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 31 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 32 OF 180 REVISION 2.3, FEB 2016

Case 3: Effect of Low Moment of Inertia of the Pump and Bigger Pipe with -ve
Elevation

Note elevation
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 33 OF 180 REVISION 2.3, FEB 2016

2.3. Turbo Pumps

The turbo pump is the most comprehensive pump model in PIPENET transient module. It
can model the behaviour of a pump in all 4 quadrants. Positive and negative flowrates as
well as positive and negative pressure differences. However, this model depends on the
use of Suter Curves. Only three well established data sets for Suter Curves are available
and are published in the book by Wylie and Streeter. The data sets published in Wylie and
Streeter cover the following cases.
Axial pump
Radial pump
Mixed flow pump
These are standard curves and are empirically derived. So, they must not be applied in all
cases. Sunrise Systems has developed an Excel Spread sheet which can estimate the
Suter curve from the normal pump performance curve which is readily available. In the
training example below we will assume that an axial pump is being used.
Base case – Pump Trip
Turbo pump data used in all cases is shown below.
Rated flowrate = 4000 lit/min
Rated pressure = 14.8 bar G
Rated speed = 6 rev/sec
Torque = 600 N.m
Moment of inertia= 1000 kg m2
Suter Curve type = axial flow
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 34 OF 180 REVISION 2.3, FEB 2016

Trip time

Note that the trip time is given here and not through the information node. This is because
the information node may be connected to a control loop which sets its speed. In general
this speed will not be known to the user when a control system is in operation. The user
can specify the trip time through the properties window. This is the time which the trip
starts.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 35 OF 180 REVISION 2.3, FEB 2016

In this case the pump is assumed to run at 100% rpm when it trips. The trip is assumed to
occur at 8 sec into the simulation as specified through the above dialog box.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 36 OF 180 REVISION 2.3, FEB 2016

Case 2: Effect of Low Moment of Inertia and Bigger Pipe, with +ve elevation
Make the following changes and see the effect.
Change the Moment of Inertia of the pump to 10 kg m 2, the torque to 50 Nm, pipe diameter
to 300 mm and the pipe length to 4000 m with an elevation of 30 m.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 37 OF 180 REVISION 2.3, FEB 2016

Note Values
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 38 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 39 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 40 OF 180 REVISION 2.3, FEB 2016

Case 3: Effect of Low Moment of Inertia and Bigger Pipe, with -ve elevation
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 41 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 42 OF 180 REVISION 2.3, FEB 2016

3. TANKS:

PIPENET offers several types of tank model. They are listed below.
Simple tank: this is a two-node tank with constant pressure above the liquid level
which is specified by the user. The inlet node locates at the top gas side whilst the
outlet node connects to the bottom liquid side. The calculation will abort if a simple
tank drains or overflows or flow reverses at the inlet node.
Accumulator: this is sealed, pressurized one-node vessel. Some air is trapped
inside to provide cushioning effect. The air pressure is variable depending on the
liquid level in the vessel. The calculation will abort if it completely drains.
Surge tank: this is an open, atmospheric one-node vessel. The vessel top is open
to the ambient conditions and its pressure is constant 0 bar G. The calculation will
abort if it overflows or completely drains (dry pipe model is disabled).
Receiving vessel: this model is same as the surge tank except (1) The calculation
will not abort when it overflows; (2) the calculation can calculate the overflow rate
and total overflow; (3) it can consider the effect of weir crest.
The user can select regular tank from the predefined list (vertical cylinder, horizontal
cylinder, sphere, conical frustum and uniform section) or define an irregular tank (e.g.
reservoir) in the tanks library, see an example below. This would be invaluable in
modeling hydro-electric power stations where the ‘surge tank’ can actually be a lake, a
cave or whatever is available. In such a case the height vs area may have to be input in
the form of a table.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 43 OF 180 REVISION 2.3, FEB 2016

3.1. Simple Tank:

A simple tank has a flow in and a flow out. These flowrates may be different because the
level in the tank is allowed to change. Tanks can be pressurised or atmospheric.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 44 OF 180 REVISION 2.3, FEB 2016

Base Case
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 45 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 46 OF 180 REVISION 2.3, FEB 2016

Case 2: The inlet pressure is given the following specification.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 47 OF 180 REVISION 2.3, FEB 2016

Give an explanation for why the simulation stops prematurely. The simulation time was
input as 1200 secs. But the simulation stopped at about 110 secs.

3.2. Accumulator
In all the cases below we consider valve closure scenarios.
Case 1: System without an Accumulator:

Use the following data

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 48 OF 180 REVISION 2.3, FEB 2016

Use metric units

Use water at 20 °C
Use ANSI 36.10 Schedule 40 pipe
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 49 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 50 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 51 OF 180 REVISION 2.3, FEB 2016

Case 2: With a large accumulator

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 52 OF 180 REVISION 2.3, FEB 2016

Case 3: With small accumulator

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 53 OF 180 REVISION 2.3, FEB 2016

The calculation results show the pressure surge can’t absorbed by a small accumulator,
which indicates that PIPENET can be used to optimize the tank size.

3.3. Surge Tank:

A surge tank is an open tank. It is predominantly used for reducing pressure surges in
systems using water. The first case we consider is valve closure without a surge tank.
Base Case without a Surge Tank
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 54 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 55 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 56 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 57 OF 180 REVISION 2.3, FEB 2016

Case 2: With a Surge Tank Installed

Data for surge tank.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 58 OF 180 REVISION 2.3, FEB 2016

Case 3: Change the height of the surge tank to 20 m

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 59 OF 180 REVISION 2.3, FEB 2016

How do you interpret the results?

Case 4: Surge tank and pump trip

We use a turbo pump because we need to model the vacuum which the pump trip would
cause. We start with the same network as the turbo pump modelling case which we
considered earlier. The data however is different and is shown below.
NOTE THAT VAPOUR CAVITY MODELLING HAS BEEN CHOSEN BECAUSE PUMP
FAILURE IS LIABLE TO CAUSE CAVITATION.
Case 4 a: No Surge Tank Installed.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 60 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 61 OF 180 REVISION 2.3, FEB 2016

Trip time
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 62 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 63 OF 180 REVISION 2.3, FEB 2016

Case 4b: Surge Tank Installed

Surge tanks are sometimes used to prevent vacuum pressure downstream of the pump
when there is pump trip so that cavitation is avoided. Surge tanks which are used for the
purpose can be very big indeed. When there is an item in the network which is large it will
take a long time for the steady state to be reached. This is because a small change in the
height will produce a big change in the volume of the liquid. So it is difficult to determine
the steady state height. So, for this simulation we can expect an exceptionally large run-in
time.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 64 OF 180 REVISION 2.3, FEB 2016

The trip time is set to 100 sec so that we can clearly see the initial steady state solution.
As discussed above the run-in time will be very long.

We also have to increase the simulation time and this is shown below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 65 OF 180 REVISION 2.3, FEB 2016

Why does the simulation stop before the 5000 sec simulation time?
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 66 OF 180 REVISION 2.3, FEB 2016

Case 4c: Effect of adding a non-return valve and elevation change in the pipe

Note elevation
change
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 67 OF 180 REVISION 2.3, FEB 2016

Explain why this makes such a big difference.

3.4. Receiving Vessel:

The main header in a series of projects used 3.5 m and 4.5 m GRP pipes. The receiving
vessels were a fundamental aspect of their design. Their function is to prevent air coming
into the system if all the pumps tripped simultaneously because of power failure. As GRP
is a fairly elastic material as the pressure fluctuates during water hammer, the volume
inside the pipes will also fluctuate. The receiving vessel will prevent air being sucked back
into the system when the pipes expand. The water in the receiving vessel will flow back
into the system instead of air. It is essential to make sure that the receiving vessel is big
enough not to get drained.
This is similar to the network with the turbo pump model, with the small moment of inertia
and 300 mm diameter pipe. However a receiving vessel has been inserted and the pipe
data has been changed.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 68 OF 180 REVISION 2.3, FEB 2016

Select metric for units

Select water at 20 °C for fluid
Create the following user defined data in the library
Pipe Schedule Data
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 69 OF 180 REVISION 2.3, FEB 2016

Turbo Pump Data

Valve Data (This has the role of a check valve)

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 70 OF 180 REVISION 2.3, FEB 2016

Click on the ‘Linear’ button to develop the derivatives. Then click on ‘apply’.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 71 OF 180 REVISION 2.3, FEB 2016

The Network Data is shown below

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 72 OF 180 REVISION 2.3, FEB 2016

25 sec valve closure time

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 73 OF 180 REVISION 2.3, FEB 2016

50 sec valve closure time

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 74 OF 180 REVISION 2.3, FEB 2016

Smaller Receiving Vessel (5.223 m diameter)

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 75 OF 180 REVISION 2.3, FEB 2016

What do the above graphs indicate and what action do you think we should take?
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 76 OF 180 REVISION 2.3, FEB 2016

4. FORCE CALCULATION:

The overall philosophy which is used in PIPENET is shown below.

Design Stage Analysis Stage

Structural Design Stage
Steady state calculations Dynamic calculations
using PIPENET using PIPENET Transient Pipe stress analysis
Standard and Module
Spray/Sprinkler Modules

`
Two very important conceptual differences between PIPENET and pipe stress analysis
programs are described below.
(i) If a disturbance is made to fluid in a pipe network system, it will travel throughout
the connected parts of the network at the wave speed. In comparison pipe
stress is essentially a local phenomenon. It can generally be assumed that
forces and stresses do not go past anchors.
(ii) When it comes to flow analysis the exact 3 D geometry of the network usually is
not important. It is the length of pipes and elevation changes which are
important. Three dimensional changes in direction are not important. Fittings
can be regarded as attributes of pipes. Only valves which operate are regarded
as models. Otherwise they can simply be added as fittings.
We also recognise that it would be extremely inconvenient for a professional engineer to
input the drawing as a piping isometric into PIPENET. It could take many times as long to
input the piping drawing as an isometric into PIPENET in comparison to a schematic. That
is why the network in PIPENET is represented as a schematic drawing. On the other
hand, it is essential in pipe stress analysis to work with isometrics data albeit in relatively
small parts of the network. The force calculation capability in PIPENET is intended to
bridge this dichotomy.
The forces calculation capability in PIPENET was developed with pipe stress analysis in
mind. For this reason it recognises the following;
(i) PIPENET analysis will have to be carried out by defining the complete network.
(ii) For each such case the pipe stress engineer would require the forces to be
calculated for a large number of load cases. Furthermore, care has to be taken
to take into account that the forces should lend themselves to be defined and
calculated quickly for each load case. PIPENET is designed in such way that
the forces for the load cases can be calculated easily without changing the basic
network.
4.1. Simple Force:

The principle of unbalanced force.

Even in steady state conditions the elbows will experience hydraulic forces. However the
unbalanced forces will be small because the components will cancel each other.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 77 OF 180 REVISION 2.3, FEB 2016

Unbalanced
Force = 0

In the following graphics the symbol represents the position of the pressure assuming
that it is a sharp spike. The symbol represents the steady state force.
The symbol represents the high force caused by the pressure surge.

Unbalanced
Force = down

Unbalanced
Force = up

Given the positions of the elbows PIPENET can calculate the unbalanced forces. It is
important to understand three key points.
(i) The first key point is that the elbows do not need to be on one single pipe. The
elbows can have items such as valves, T pieces and so on between them. The
important point is the elbows and the items in between need to be on a straight.
In fact the end points need not be elbows. If there is a point across which there
is no transfer of the force (such as an anchor) such a point could be defined as
an end point. The typical end points are summarized in the table below. We
should also remember that PIPENET uses schematic drawings. So an elbow
may not be shown in the schematic. There may not be a one-to-one
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 78 OF 180 REVISION 2.3, FEB 2016

correspondence between the pipes in PIPENET and pipe sections in a pipe

stress analysis program. The user has complete flexibility in how the forces
calculation capability is used in PIPENET.

Interruption End Point

Pipe section for stress analysis Elastic
Elbow, Angle valve, Reducer and Enlarger Rigid
Anchor, Axis restriction Elastic
Dead end, Entrance and Exit (e.g. nozzle, tank, reservoir) Rigid
Elastic joint Elastic

(ii) The second key point is the position is always defined with reference to the input
node of a pipe.
(iii) We also recognise that a simple or unbalanced force will only have one
component because it will always be in the direction of the pipe. In fact, that is
why it is called a simple force. Furthermore, it is not necessary to create nodes
at the positions where unbalanced forces need to be calculated. As we will see
later a complex force will have three dimensional components and must have
nodes where the pipeline changes directions.
PIPENET uses the concept of control volumes to calculate the forces. A control volume
need not be between two elbows at all. This may be because a pipe may be anchored at
a point. We assume that the force cannot go past an anchor. This is important from a
pipe stress analysis point of view. Typically each straight section of pipe will have a
control volume and each anchor will also divide a control volume. Let us consider the
control volumes as shown below:

Control volume

ANCHOR

ANCHOR ANCHOR

200 m

100 m

100 m
100 m
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 79 OF 180 REVISION 2.3, FEB 2016

Control volumes are defined by their intersection with pipes. By definition a control volume
for a simple force can only have two such intersections.

Network data.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 80 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 81 OF 180 REVISION 2.3, FEB 2016

Dialog boxes for forces

The dialog boxes for the forces are shown below. Forces are not physical entities in the
schematic and the user needs to have complete freedom in defining control volumes. For
this reason the forces tab in the data window must be selected for inputting forces

Forces tab

Click on … and complete the dialog boxes as shown below.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 82 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 83 OF 180 REVISION 2.3, FEB 2016

Forces are not components in the schematic and so it is not possible to point and select
them for graphical output.

Click on the double arrow to move to the next line.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 84 OF 180 REVISION 2.3, FEB 2016

Click here

Close the valve in 1 sec and see the difference.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 85 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 86 OF 180 REVISION 2.3, FEB 2016

4.2. Complex Forces:

The main difference here is that complex forces are 3 dimensional forces. It is necessary
to create nodes at the positions where the flow direction swerves.
Complex forces are less commonly used. This is because typically pipe stress analysis
programs work with simple forces (unbalanced forces in the direction of the pipe) even if
such forces are applied to one end of a pipe. Complex forces would be useful if the finite
element analysis is applied for local stress analysis.
Complex forces again are based on defining control volumes.

Control
volume

Control
volume
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 87 OF 180 REVISION 2.3, FEB 2016

As complex forces are 3 dimensional forces they need to be calculated with reference to a
defined co-ordinate system. The user can freely choose the co-ordinate system he would
prefer to use. We can choose the following co-ordinate system, with the Z dimension
coming vertically out of the paper.

There are two types of data for each intersection between the control volume and the pipe.
(i) Position with reference to the input node of the pipe.
(ii) The unit normal coming out of the control volume in the direction of the pipe. In
fact this is what defines the 3D direction of the force.
(iii) Whether the end points are elastic or rigid. The typical end points please refer to
the summary in the simple force section.
Let us consider one of the control volumes above.
If we take the first elbow on its own we can define the force as follows.

Unit normal 1

Unit normal
2
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 88 OF 180 REVISION 2.3, FEB 2016

The following table shows the data for the unit normal
Unit Pipe label Position, x- y- z-
normal m component component component
number
1 1 99 -1 0 0
2 2 1 0 -1 0

PIPENET automatically infers the direction of the unit normal with reference to the
direction of pipe. This is necessary for assigning the sign of the force.
The end points should also be set to elastic. This is because the force on the elbow must
be calculated without the restraining effect of the pipe.

Elastic
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 89 OF 180 REVISION 2.3, FEB 2016

Elastic

As before, forces are not components in the schematic and so it is not possible to point
and select them for graphical output.

Click on the double arrow to move to the next line.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 90 OF 180 REVISION 2.3, FEB 2016

Click here

In the graph Viewer it is possible to see three components for the forces and also the
magnitude of the three dimensional force.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 91 OF 180 REVISION 2.3, FEB 2016

Change the valve closure time to 1 sec and see the effect on the complex forces.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 92 OF 180 REVISION 2.3, FEB 2016

4.3. Total Force and Dynamic force

PIPENET can output the force in two formats: (1) total force represents the absolute
magnitude and (2) dynamic force is the difference between the current time and the
reference time. In the figure below, the hydraulic force is output in the format of dynamic
force. The reference time is 0 sec, i.e. the start time of simulation.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 93 OF 180 REVISION 2.3, FEB 2016

Generally engineers are more interested in total force because it presents the real
hydraulic force that flow transfers to structure. The hydraulic force is always zero at steady
state if both the end points are rigid. Therefore, the total force is equal to the dynamic force
in this case, see the comparisons below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 94 OF 180 REVISION 2.3, FEB 2016

4.4. Forces Output File

PIPENET can generate an ASCII file specifically for the purpose of interfacing with pipe
stress analysis programs. The time step for this can be individually specified. The time
history for all the forces defined in the simulation will be output to this file.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 95 OF 180 REVISION 2.3, FEB 2016

An extract from the .frc file for one of the simple forces case is shown below.

Time units : seconds

Force units : Newtons

Output interval : 0.010 seconds

Number of forces tabulated : 4

Number of points per table : 1001

DIFF : F/1

.220000E+1 0.115313E-02

.221000E+1 0.114627E-02

.222000E+1 2.17615

.223000E+1 7.89779

.224000E+1 14.1450

.225000E+1 20.4487
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 96 OF 180 REVISION 2.3, FEB 2016

.226000E+1 26.7713

.227000E+1 33.1121

.228000E+1 39.4712

.229000E+1 45.8497

.230000E+1 52.2479

.231000E+1 58.6670

.232000E+1 65.1072

.233000E+1 71.5698

.234000E+1 78.0550

.235000E+1 84.5641

.236000E+1 91.0973

.237000E+1 97.7186

.238000E+1 104.382

.239000E+1 111.074

.240000E+1 117.795

.241000E+1 124.545

.242000E+1 131.324

.243000E+1 138.135

.244000E+1 144.977

.245000E+1 151.851

.246000E+1 158.759

.247000E+1 165.701

.248000E+1 172.679

.249000E+1 179.692

.250000E+1 186.743

.251000E+1 191.401

.252000E+1 193.652

.253000E+1 195.222

.254000E+1 196.699

.255000E+1 198.187

.256000E+1 199.699

.257000E+1 201.236

.258000E+1 202.800

.259000E+1 204.391

.260000E+1 206.008

.261000E+1 207.653

.262000E+1 209.326
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 97 OF 180 REVISION 2.3, FEB 2016

.263000E+1 211.027

.264000E+1 212.757

.265000E+1 214.517
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 98 OF 180 REVISION 2.3, FEB 2016

5. CONTROL SYSTEMS

Control systems can be built by using the following components:

Sensors – Pressure, Flow and Differential pressure. The signal can be Analogue or
Digital.
Transfer function – Zeroth, First, Second order and Linear power ramp.
The output signal responds to the input signal instantly in the zeroth transfer function.
In the first order transfer function, the output signal responds to the change of input
signal quickly at the early stage and then gradually approaches the new steady state.
The second order transfer function introduces damping component to restrict the
change speed of output signal. The linear power ramp is usually used to set linear
operation of valves and devices.
PID Controller – Proportional, Integral and Derivative terms in any combination.
The input set point is the control target which represents the desired parameter at the
steady state. The output set point does not affect the simulation results if the controller
has I-term (I, PI, DI and PID) but it does affect in the other controllers (P, D and PD). A
reasonable output set point can save much run-in time to reach initial steady state.
The gain can be positive or negative. The negative gain is used when the operation of
the controlled device has opposite tendency with the input signal, i.e. negative
feedback control loop. In the control system below, the PID’s gain is negative because
the bypass valve tends to open further when the measured flow rate is below the set
point.

If the flow sensor is moved to the main pipeline, the PID’s gain should be positive
because the bypass valve tends to open further when the measured flow rate is above
the set point, see the figure below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 99 OF 180 REVISION 2.3, FEB 2016

The network may need extremely long time to settle down if the gain is too small. On
the contrary, the control system may cause unstable oscillation if the gain is too big.
The correct scale should base on the sensitivity of control signal to the operation of the
control element. In the above example, the flow rate at the main pipeline increases
about 334 l/min when the bypass valve changes the position from opening to closure.
Therefore, the gain can be approximately set as (1-0)/334 = 0.003 1/l/min.
Reset time is used to scale I-term (integral) controller whilst rate time is used to scale
D-term (differential) controller. The output control curve tends to be smooth with big
reset time but the system needs more time to settle down. However, small reset time
or big rate time may lead oscillations. Generally the rate time should not more than
25% of the reset time.
Anti-windup introduces a negative-feedback signal to offset the saturated I-term signal,
which makes PID controllers quickly respond to the input signal after the saturation is
over. Tracking time is required when the anti-windup function is activated. Ideally the
tracking time should be larger than the rate time but smaller than the reset time.
Cascade Controller – Set point of the PID controller itself is variable and is often set
by the signal from another control loop. The input parameter setting is same as the
conventional PID controller, except the input set point. The cascade controller is
usually used to implement quick control but its complexity and cost restrict the
application.
Switch – Can be pressure, flow, differential pressure or timer type. The parameters
hysteresis and dead time delay can avoid oscillation phenomenon.
Signal selector – selects between two input signals based on minimum, maximum or
time; signal can be pressure, flow, differential pressure or information. It is usually
used to build complex control logic.
It is very important that the following points are noted.
(i) The values of the control system parameters, including their signs, are important
and attention must be paid to the fact that they must have reasonable values.
(ii) Control loops have a direction of signal flow and the items on the schematic
should be input in the same direction.
(iii) Control systems may not have steady state solutions which can be directly
calculated. So, typically it is better to work with the initial state being determined
by extending the run-in time and unchecking the initial steady state switch. An
example can be seen in the dialog box below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 100 OF 180 REVISION 2.3, FEB 2016

5.1. The Importance of Run-in Time:

It is also a good idea to check if the run-in time is long enough for the system to reach
steady state. (The principle of run-in was described in section 1.2. of this chapter.)

See the following graphs. We will look at this network in detail later. At this point we are
merely considering the effect of the run-in time.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 101 OF 180 REVISION 2.3, FEB 2016

Run-in time of 50 secs:

Run-in time of 300 secs:

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 102 OF 180 REVISION 2.3, FEB 2016

5.2. Pressure Control System.

System schematic, description and data
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 103 OF 180 REVISION 2.3, FEB 2016

Our aim is to maintain a pressure of 8 bar G on node 7. The event we are considering is
the closure of valve 2.

In the above schematic let us first consider the function of the various parts of the control
loop.

PID Controller
Operating
valve

Pressure
transmitter

Pressure Transmitter: The physical pressure transmitter is modelled by two components

in PIPENET. The pressure sensor picks up the node pressure. The transfer function
models the dynamics of the physical pressure transmitter. A pressure transmitter takes a
finite time to respond and the transfer function models that.

Operating valve: The operating valve in this system has the role of a control valve. A
control valve also takes a finite time to respond to a signal to change its position. This is
modelled by a transfer function. If the operating valve is closed linearly, say 5 seconds
from fully open to closed position, the transfer function can be set as linear power ramp
order with 0.21/sec ramp speed. Note that valve 1 does not have a specification because
its information node is connected to a control loop.

PID Controller: This is the conventional PID controller. The P, I and D terms can be used
in any combination.

The input data and dialog boxes are shown below:

Module Options:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 104 OF 180 REVISION 2.3, FEB 2016

For units choose Metric

For fluid choose water at 20 °C (default)
For pipe schedule choose ANSI 36.10 schedule 40.

Pump Performance Curve:

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 105 OF 180 REVISION 2.3, FEB 2016

Pipe data:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 106 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 107 OF 180 REVISION 2.3, FEB 2016

Input the same pressure specifications for the two output nodes.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 108 OF 180 REVISION 2.3, FEB 2016

Base Case: Steady State

Steady state conditions: It is usually a good idea with control systems to perform a dry run
to make sure that the starting point is indeed the steady state conditions. However, it is
not always necessary to do this because the dynamic simulation graphs would also
indicate whether the initial steady state has been reached. This is one reason why it is
generally a good idea a dynamic event after simulation has started. This enables the
steady state solution to be seen before any change takes place.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 109 OF 180 REVISION 2.3, FEB 2016

Dynamic Case: Response of the system when the second valve closes completely:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 110 OF 180 REVISION 2.3, FEB 2016

 Effect of gain:

The system needs longer time to reach new steady state when the gain of the PID
controller is decreased, e.g. the gain decreases to 0.0075 l/bar in the above network, see
the figure below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 111 OF 180 REVISION 2.3, FEB 2016

On the contrary, the system may oscillate with big gain, see the figure below in which the
gain is increased to 0.75 l/bar.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 112 OF 180 REVISION 2.3, FEB 2016

 Effect of reset time:

Change the integral term in the PID controller to 0.01 sec and see what happens.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 113 OF 180 REVISION 2.3, FEB 2016

The instability occurs because the integral term becomes very large as a result of the reset
time being set very small.

Case 2: Valve Closure Followed by Pump Shutdown:

In this simulation, valves 1 and 2 close at the same time creating a relatively large
pressure surge. A pressure switch shuts down the pump on high pressure.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 114 OF 180 REVISION 2.3, FEB 2016

Strictly speaking a system of this type will have a check valve after the pump and the
pump itself should be modelled as an Inertial Pump. However, the main aim is the control
system model and so the network has not been changed.

Graphical output has been selected for the items shown in the schematic below.

The dialog boxes for the relevant items are shown below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 115 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 116 OF 180 REVISION 2.3, FEB 2016

Note the following points about the data for the switch.

(i) The choice of negative means that the pump will switch off above 12 bar G.
(ii) The initial setting of ‘On’ means the pump will run until the pressure on node 7
reaches 12 bar G.
(iii) The dead time of 0 sec means there is no explicit delay in the switch itself. The
pressure transmitter and the pump have dynamics introduced by the transfer
functions.
(iv) The setting for ‘Switch on below’ is set to -1 bar G in order to prevent the pump
restarting as the pressure decreased when the pump shut down.

Note that the time constant is not the same as the time it takes for the pump to spin. It is a
measure of the rate at which it will spin down.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 117 OF 180 REVISION 2.3, FEB 2016

5.3. Liquid Level Control System

Base Case: Conventional PID controller

The network below presents a typical control system to control the tank’s liquid level.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 118 OF 180 REVISION 2.3, FEB 2016

Two control loops are designed to maintain the liquid level. The PID loop regulates the
control valve 2 to keep the liquid level at the middle position. The switch loop is used at
emergency situations to open the drain valve 1 when the liquid exceeds the warning level.
In the discussed scenario, the inlet valve 3 is fully opened from half opening.
The input data and the dialog boxed are shown below:

The fluid choose water at 20 °C (default).

For pipe schedule choose ANSI 36.10 schedule 40.
Pipe data:

Valve data:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 119 OF 180 REVISION 2.3, FEB 2016

Tank data:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 120 OF 180 REVISION 2.3, FEB 2016

PID control loop data:

The liquid level is measured by a differential pressure senor. The controller is PID type and
the transfer function is first order to simulate the dynamic character of the control valve 2.
This control loop is trying to keep the liquid level at the middle of the tank, i.e. the input set
point of the PID controller is 4 m fluid. The gain of the PID controller is positive because
the valve 2 tends to open further when the liquid level exceeds the set point.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 121 OF 180 REVISION 2.3, FEB 2016

Switch control loop data:

The switch starts to open drain valve 1 when the liquid reaches 5 m warning level and
closes the valve when the liquid level drops back to 4.5 m. The drain valve is linearly
opened in 5 seconds and closed in 10 seconds. Therefore, the transfer function is set as
power ramp order with 0.2 1/sec ramp up limit and -0.1 1/sec ramp down limit.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 122 OF 180 REVISION 2.3, FEB 2016

The scenario is to study dynamic response of the system when the inlet valve is opened
further, see the figure below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 123 OF 180 REVISION 2.3, FEB 2016

The run-in time is 200 seconds and the option “Calculate initial steady state” is ticked on
because the system is very close to the steady state after 200 seconds run-in time.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 124 OF 180 REVISION 2.3, FEB 2016

We can expect the following phenomena based on the calculation results.

 The control valve 2 is opened further to cope with the opening of the inlet valve 3.
 The drain valve 1 is opened when the liquid lever exceed the 5 m warning level and
then closed when the liquid level drops below 4.5 m.

Case 2: Cascade PID controller

In the above scenario, the liquid level is controlled by a conventional PID controller.
Conventional PID controllers are widely used in industry because of their simplicity and
cost efficiency. However, their response speed is relatively slow because the control
device is driven by the strength of input signal. In the above example, the control valve 2 is
less activated initially until the liquid level changes significantly. On the other hand, fast
response with a simple PID controller system could result in an unstable operation.
Cascade PID controllers can implement quick response by introducing two dependent
control loops, i.e. primary control loop and secondary control loop. The primary control
loop is designed for fine control purpose whilst the secondary control loop can achieve
quick response. The system below is modified from the network discussed in the previous
section. A cascade PID control system which controls the tank’s inflow replaces the
conventional PID control system which controls the tank’s outflow.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 125 OF 180 REVISION 2.3, FEB 2016

The input data is same as the model in the base case except the following.
The run-in time increase to 1000 seconds.
The outlet valve 2 keeps at constant 70% opening.
Valve data:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 126 OF 180 REVISION 2.3, FEB 2016

Primary PID controller data:

The primary PID controller is designed for fine control purpose. The input signal comes
from the differential pressure sensor which represents the liquid level of the tank. The
output signal is sent to the secondary PID controller to reset the input set point. As shown
in the figure below, the gain is set as -200 l/min/m fluid which indicates the set point of the
secondary PID controller decreases 200 l/min when the liquid level increases 1 meter. The
reset time and rate time is longer than those in the secondary PID controller because the
primary PID controller is used to control liquid level finely.

The input set point of the primary PID controller is input as constant 4 m fluid, see the
figure below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 127 OF 180 REVISION 2.3, FEB 2016

Second PID controller data:

The second PID controller aims to achieving quick response by controlling the valve 4
directly to maintain constant inflow rate. The gain, reset time and rate time are set as big
as possible. The input set point is finally controlled by the liquid level but the control signal
is treated by the primary PID controller. 0.5
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 128 OF 180 REVISION 2.3, FEB 2016

The calculation results show the system is much stable after the inlet valve is fully opened.
The liquid level only oscillates about 0.2 m which is much smaller than that using
conventional PID controller in the base case. The emergency drain valve keeps at its
closed position in this scenario, as can be seen in the figure below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 129 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 130 OF 180 REVISION 2.3, FEB 2016

6. CHECK VALVES:

There are three types of check valve models in PIPENET.

Non-return Valve: non-return valve is an ideal check valve which allows
unrestricted flow of fluid in a positive direction, and prevents all flow in the reverse
direction instantly.
Check Valve: check valve considers flow resistance in the positive direction and
the valve closure is not instantaneous. It depends on the flow and pressure across
the valve. An additional abstract parameter, namely time constant, is introduced so
that the dynamic characters of the valve can be taken into account.
Inertial Check Valve: An inertial check valve is a swing-type check valve. The
operating speed depends on the flow/pressure in the vicinity of the valve. The
dynamic characteristics are calculated based on its physical parameters, for
example door mass, door radius, spring constant, angle of closed door, inclination
of valve, damping and velocity exponent.
For the examples in this section we will start with the simple pump example which was set
up earlier and insert the different types of check valve.
In order to insert a check valve we first have to create a break in the system. This can be
done by using the Explode Node by pointing to the relevant node and right clicking.

Certain changes are made to the system in order to enable the simulation to run faster.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 131 OF 180 REVISION 2.3, FEB 2016

Reduce the simulation time to 30 sec. Also the vapour cavitation option is chosen in order
to introduce cavitation effects.

Vapour
Cavitation

Change the graphical output time step to 0.01 sec.

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 132 OF 180 REVISION 2.3, FEB 2016

Close the valve in 1 sec by changing the specification as shown below.

All other data is the same as the simple pump example we saw before.

6.1. Non-return Valve (Simple check valve)

The advantage of this type of check valve is that no data is required. It will close instantly
if reverse flow occurs. This should be good enough for a majority of cases because the
operation of the check valve itself will not affect the conclusion from a design point of view.
The results may be slightly different but the design conclusions will be the same.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 133 OF 180 REVISION 2.3, FEB 2016

6.2. Check Valve (Dynamic check valve)

Check valve considers pressure loss and it needs time to close the valve when the reverse
flow occurs.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 134 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 135 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 136 OF 180 REVISION 2.3, FEB 2016

6.3. Inertial Check Valve

The inertial check valve model is a mechanistic model which starts from first principles. It
considers many parameters which would influence the behaviour of the check valve. The
parameters are listed below.
The characteristics of the valve as Cv or k-factor and area just as for any other
valve.
The mass of the door of the valve
The actual radius of the door of the valve
The spring constant if the valve is spring assisted
The angle of the closed door with reference to the vertical
The inclination of the valve with reference to the horizontal
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 137 OF 180 REVISION 2.3, FEB 2016

The damping constant if the valve has damping action

The exponent of the velocity for the damping action

All parameters are the same as before except the check valve characteristics. They are
shown below for the different cases as appropriate.
Case 1:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 138 OF 180 REVISION 2.3, FEB 2016

Case 2:
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 139 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 140 OF 180 REVISION 2.3, FEB 2016

7. CAVITATION MODELLING:

There are three options when it comes to cavitation.

No cavitation
Vapour cavitation
Channel cavitation
No cavitation option is provided because the calculation will be faster and the numerical
method is most robust with this option. This model can allow the pressure to go below the
vapour pressure or even the absolute pressure.
Vapour cavitation model is for relatively small cavities where the volume occupied by the
cavity does not affect the behaviour of the system. In certain cases the vapour cavitation
which is based on the technique outlined in the classic text book on fluid transients can
lead to infinitely growing vapour volumes. In order to avoid that we have developed a
unique channel cavitation model in PIPENET. This model considers the physical size of
the cavity. This is suitable for bigger cavities and where the size of the cavity itself can
affect the results significantly.

7.1. Vapour Cavitation Model

This takes a tanker loading system as the basic model. The jetty is made
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 141 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 142 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 143 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 144 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 145 OF 180 REVISION 2.3, FEB 2016

The calculation results show the cavity volume increases up to a maximum 10 litres. This
is much smaller than the liquid volume in the pipe. Therefore, the vapour cavitation model
is appropriate for this scenario.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 146 OF 180 REVISION 2.3, FEB 2016

Vapour Cavitation Model with Loading Arm

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 147 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 148 OF 180 REVISION 2.3, FEB 2016

It can be seen that the cavity size is growing indefinitely. Furthermore, the flowrate
through valve 3 does not go to 0. That is unrealistic. This shows the inadequacy of using
the vapour cavitation model under certain circumstances. So, it is necessary to use the
channel cavitation model in this case.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 149 OF 180 REVISION 2.3, FEB 2016

7.2. Channel Cavitation Model:

Apart from the choice of channel cavitation option all the data is the same.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 150 OF 180 REVISION 2.3, FEB 2016

It can be seen that the cavity length reaches at around 22 m and stays there.
Furthermore, the flowrate through valve 3 goes to 0. So, we can conclude that the
channel cavitation model predicts the cavity formation phenomenon well.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 151 OF 180 REVISION 2.3, FEB 2016

8. DRY PIPE MODELLING:

The pipe model in PIPENET is designed to simulate the following three flow conditions.
Dry condition: pipe is occupied by air entirely or partly.
Wet condition: pipe is filled with liquid.
Cavitation condition: Liquid and vapour coexist in pipe.
The dry pipe model can simulate both filling up and discharge processes. The typical
applications include: (1) estimating the prime time of dry system; (2) calculating the
pressure surge when dry pipeline fills up; (3) improving the simulation accuracy.
The air in dry pipes may come from four sources: (1) original air in dry pipes; (2) dry
specification nodes; (3) nozzles; (4) surge tanks or receiving vessels. In other side the air
in dry pipes can discharge to anywhere with atmospheric conditions. The dry pipe model is
a simplified model so there are a few restrictions..

 A dry pipe can only be modelled by an elastic pipe (not a short pipe).
 We assume that there are no wet pipes or dead ends downstream of the dry pipes
which block air flow. The back pressure of dry pipes is assumed to be 0 bar G, i.e.
the pressure drop in a dry pipeline is negligible.
 Similarly we assume that there are no short pipe; pipe bundle; tanks and caissons
downstream of the dry pipes to block air flow.
 Dry state cannot coexist with channel cavitation state in a single pipe. Therefore,
the channel cavitation model for that pipe is disabled during filling up and enabled
after the pipe is filled up.

8.1. Base network:

 Estimate prime time

The following network presents a simple dry deluge system.

Dry pipe
symbol

Uncheck “Treat all pipes as wet” to enable the dry pipe function.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 152 OF 180 REVISION 2.3, FEB 2016

For units choose Metric

For fluid choose water at 20 °C (default)
For pipe schedule choose ANSI 36.10 schedule 40.
Pipe Data is shown as below
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 153 OF 180 REVISION 2.3, FEB 2016

Valve Data

Nozzles data

Specification data
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 154 OF 180 REVISION 2.3, FEB 2016

The prime time of nozzles is summarized in the calculation report, data window and
property window. Please note that the priming time here means the time at which the
nozzle discharges initial flow; it is not necessarily the time at which full steady state flow is
established.

 Calculation report
PRIMED TIME OF NOZZLES
----------------------

Label Primed Time

seconds
-----------------------------------

1 0.219E+02
2 0.259E+02
3 0.299E+02

 Data window

 Property window
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 155 OF 180 REVISION 2.3, FEB 2016

 Calculate pressure surge and flow capacity:

Generally there is a pressure surge when the most remote nozzle is primed because the
flow resistance increases so that flow rate decreases suddenly, see the figures below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 156 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 157 OF 180 REVISION 2.3, FEB 2016

The calculated maximum pressure is 26 bar G and the nozzle capacity is 69.66 lit/min at
final steady state.

8.2. Optimize network:

There are three key parameters to evaluate the performance of a deluge system, i.e. prime
time, maximum pressure and flow capacity. In the above scenario, the prime time of the
most remote nozzle is 29.91 sec, the maximum pressure is 26 bar G and the nozzle
capacity is 69.6 lit/min. If any allowed time limit is exceeded, the system has to be
optimized in order to satisfy the design requirements. We consider two options below.

 Solution 1: decrease pipe size.

Small pipe size can significantly increase the flow velocity and resistance in the pipeline.
As the results, the prime time is shortened and the pressure surge is depressed. The side
effect is that the flow velocity and flow capacity must be checked carefully.
Let us see what will happen if the pipe size in the above network is decreased to DN50.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 158 OF 180 REVISION 2.3, FEB 2016

The prime time of the most remote nozzle shorten to 18.6 sec.
PRIMED TIME OF NOZZLES
----------------------

Label Primed Time

seconds
-----------------------------------

3 0.186E+02
2 0.154E+02
1 0.126E+02

The maximum pressure surge decreases to 18.07 bar G.

However, the flow capacity of the nozzle decreases to e 63.7 lit/min and the flow velocity in
the pipeline increases to 1.48 m/s at final steady state.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 159 OF 180 REVISION 2.3, FEB 2016

 Solution 2: add a closed end dry pipe which will act like an accumulator
If it is only necessary to depress the pressure surge, we can install a blocked dry pipe near
the remote nozzle to absorb the pressure surge when the nozzle is primed. This blocked
dry pipe can be modelled by an accumulator or a caisson, see the network below.

The caisson can be defined as

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 160 OF 180 REVISION 2.3, FEB 2016

The prime time of the most remote nozzle is unchanged 29.92 sec.
PRIMED TIME OF NOZZLES
----------------------

Label Primed Time

seconds
-----------------------------------

1 0.219E+02
2 0.259E+02
3 0.299E+02
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 161 OF 180 REVISION 2.3, FEB 2016

The flow capacity of the nozzle is also the same 63.7 lit/min, but the maximum pressure
drops to 9.64 bar G

8.3. Looped network:

The dry pipe model requires at least an exit for air flow out. However, this requirement may
not be satisfied in some networks, such as the looped network shown below.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 162 OF 180 REVISION 2.3, FEB 2016

The pipe 8 cannot be modelled by a dry pipe because there is no an exit for the air in the
pipe 8 after both the nozzles 3 and 6 are primed. This network may predict a non-existent
pressure surge when the dry pipe 8 is filled up, which is impossible in a real network.
We recommend modelling this dry pipe without a path for air to get out by an accumulator
or caisson model to consider the effect of air cushion in the pipe. Please refer to the
network below.

8.4. Improving Calculation Accuracy:

In the network below, the pipeline’s exit is open to the atmospheric conditions. The
ambient air may refill the pipeline after the pump trip. What we concern is to make sure the
pump’s reverse rotation cannot exceed the allowed maximum limit. The way the dry pipe is
modeled can affect the accuracy of the results in this scenario.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 163 OF 180 REVISION 2.3, FEB 2016

For units choose Metric

For fluid choose water at 20 °C (default)
For pipe schedule choose ANSI 36.10 schedule 40.
The model options are set as below, where the dry pipe model is activated.

The turbo pump library is defined as below

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 164 OF 180 REVISION 2.3, FEB 2016

The input data

PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 165 OF 180 REVISION 2.3, FEB 2016

 Scenario 1: the network exit is specified as wet node.

There is no air source from the exit of this network in this scenario so that the pipe 1 is
always filled with water. The model may overestimate the reverse flow and rotation
speed because of unreal high static pressure at the pump outlet. The calculated
reverse flow and rotation speed are -16041 lit/min and -6.6 revs/sec respectively.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 166 OF 180 REVISION 2.3, FEB 2016

 Scenario 2: the network exit is specified as dry node.

Air can flow into the pipeline from the dry node specification. The water level in the
pipe drops gradually after the pump trip. The reverse flow and rotation speed cannot
reach the values expected in the scenario 1 because of smaller static head at the
pump outlet. The calculated flow rate finally tends to be 0 lit/min and the maximum
reverse speed is about -3.5 revs/sec.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 167 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 168 OF 180 REVISION 2.3, FEB 2016

9. RELIEF/SAFETY VALVE MODELS:

We use the same network as we did in the previous section namely the loading system.
We assume that the maximum pressure the system is allowed is 15 bar G. This occurs in
front of the operating valve which closes.

The pressure in front of the valve exceeds 15 bar G without any protection. The graph is
shown below.

9.1. Using a Bursting Disc:

These are also known as rupture discs. As the pressure must not exceed 15 bar G, the bursting
pressure is set at 14 bar G. The bursting disc is fully opened once the inlet pressure exceeds 14
bar G and keeps at open condition until a new disc is replaced. The bursting time is the time it
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 169 OF 180 REVISION 2.3, FEB 2016

takes for the disc to reach its final area and the Cv reaches its final value. The opening profile
should not make a significant difference to the results.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 170 OF 180 REVISION 2.3, FEB 2016

9.2. Using a Regulator:

A regulator can also be used as a relief valve if its downstream pressure is set to 0 bar G.
The time response refers to the dynamics of the regulating valve and the time constant is a
measure of the speed of its response.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 171 OF 180 REVISION 2.3, FEB 2016

Effect of the regulator dynamics

Change the time constant to 0.1 sec while keeping the Cv at 750 (m3/hr, bar) and see its
effect.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 172 OF 180 REVISION 2.3, FEB 2016

This is because the regulating valve does not have hysteresis.

Effect of the Regulator Size
Now keep the time constant at 0.1 sec but reduce the valve Cv to 50 (m3/hr, bar)
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 173 OF 180 REVISION 2.3, FEB 2016

The pressure profile is much stable in this scenario because the operation of small valve
has less impact on the flow in the main pipeline.

9.3. Using a Liquid Surge Relief Valve

Liquid surge relief valves are relief valves which operate very fast and are assisted in their
operation by a gas, typically nitrogen.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 174 OF 180 REVISION 2.3, FEB 2016

Smaller Surge Relief Valve

Liquid surge relief valves are extremely expensive and there is a lot to be gained by
reducing the size. It is clear that in the above case the valve is too big because it opens
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 175 OF 180 REVISION 2.3, FEB 2016

only around 14%. There is a lot to be gained by reducing its size. The result is very
encouraging because the valve opens around 55% and is able to control the maximum
pressure quite well.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 176 OF 180 REVISION 2.3, FEB 2016

10. MILE POST DATA

This is particularly useful in long pipelines without branches.

Typically when a valve closes in a system the pressure will settle down to the static (0
flow) value. This does not of course mean that the pressure is the same throughout the
network. In particular the elevation changes in the system will make a difference to the
pressure.
The pressure surge which the valve closure creates will move back and forth along the
pipework at the wave speed. Even after the fluid reaches the static flow conditions this
pressure wave will keep moving back and forth. The mile post data (sometimes called
chainage) is a convenient way of inputting a pipe which has several changes in the
elevation. Instead of inputting several pipe segments, the user can simply input one pipe
along with a definition of the change in the elevation of the pipe in the form of a table. This
can save a lot of time when it comes to inputting long pipelines which may have a large
number of elevation changes. Please note the channel cavitation function is deactivated
for the mile post pipes.
See the dialog box below for the method of inputting the milepost data.

In the following example, the same pipeline has been input in two different ways in parallel. Pipes
3,4,5,6 and 7 have been input as straight segments of pipe. The pipe with the label LONG/1 has
been input with the milepost data as shown.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 177 OF 180 REVISION 2.3, FEB 2016
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 178 OF 180 REVISION 2.3, FEB 2016

Examples of output from the graph viewer are shown below

It can be seen that the pressure profile on pipe LONG/1 is identical to the profiles on pipes
3 – 7 if a single graph is built by pacing the piece-wise graphs side by side. So, we can
conclude that the results will be the same regardless of whether the long pipe is made up
of individual pipes or a single pipe using milepost data.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 179 OF 180 REVISION 2.3, FEB 2016

It can be seen that the pressure against distance graph for pipes 3,4,5,6 and 7 are pieces
of the above graph for pipe LONG/1.
PIPENET VISION TRAINING MANUAL TRANSIENT:CHAPTER 11
PAGE 180 OF 180 REVISION 2.3, FEB 2016