Modelling and Transient Simulation of Water Flow in Pipelines Using Wanda Transient Software
Modelling and Transient Simulation of Water Flow in Pipelines Using Wanda Transient Software
Modelling and Transient Simulation of Water Flow in Pipelines Using Wanda Transient Software
MECHANICAL ENGINEERING
a
Mechanical Engineering Department, University of Nigeria, Nsukka, Nigeria
b
Department of Offshore, Process and Energy Engineering, Cranfield University, UK
KEYWORDS Abstract Pressure transients in conduits such as pipelines are unsteady flow conditions caused by a
Pressure transients; sudden change in the flow velocity. These conditions might cause damage to the pipelines and its
Water hammer; fittings if the extreme pressure (high or low) is experienced within the pipeline. In order to avoid
Rigid column; this occurrence, engineers usually carry out pressure transient analysis in the hydraulic design phase
WANDA Transient; of pipeline network systems. Modelling and simulation of transients in pipelines is an acceptable
Pipelines and cost effective method of assessing this problem and finding technical solutions. This research
predicts the pressure surge for different flow conditions in two different pipeline systems using
WANDA Transient simulation software. Computer models were set-up in WANDA Transient
for two different systems namely; the Graze experiment (miniature system) and a simple main water
riser system based on some initial laboratory data and system parameters. The initial laboratory
data and system parameters were used for all the simulations. Results obtained from the
computer model simulations compared favourably with the experimental results at Polytropic
index of 1.2.
Ó 2015 Faculty of Engineering, Ain Shams University. Production and hosting by Elsevier B.V. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Nomenclature
pump(s) for supplying the pressure required to overcome static 1.2. WANDA Transient simulation tool
head and losses (frictional and minor) and valve(s) used to
control the rate, direction and pressure of fluid flow are usually WANDA Transient is manufactured by Deltares in the
required. Netherlands for the hydraulic design and optimization of pipe-
Transient flows in pipeline maybe initiated by systems line systems ranging from small central heaters to hydrocarbon
operator, be imposed by an external event, be caused by a transport pipelines. WANDA has been used for diverse appli-
badly selected component or be developed as a result of poor cations related to drinking water systems, oil or liquefied nat-
maintenance [2]. Some of the events that cause transient ural gas loading and offloading systems, fire fighting systems,
flows include the following: uncontrolled pump trip, valve process plant piping, district heating systems, hydropower
movement, check valve slam, pump start problem, excitation industry, aircraft hydraulics, etc. WANDA’s intelligent com-
of resonant vibrations by pumps with an unstable head against ponents and user-friendly interface enable its user to carry
discharge curve, and air in lines [1–8]. out any hydraulics analysis in the most efficient way [13].
The study of true transient flows must include fluid inertia The software is operated in two main modes (the engineering
and may also include the elasticity or compressibility of mode and the transient mode). The engineering mode supports
the fluid and the conduit. The rigid column theory and water most common activities in the hydraulic design of any piping
hammer theory are the two approaches adopted in pressure system. In this mode, steady state flow condition (stationary
transient analysis [1,2,9]. In both cases, the analysis requires flow) analysis can be carried out. The transient mode allows
the application of Newton’s second law which leads to the unsteady flow conditions (water hammer) to be investigated.
Euler equation. In the rigid column analysis approach, the [13].
elastic effects of the fluid and conduit material are ignored The objective of this work was to predict the pressure surge
without compromising the accuracy of the analysis. When for water flow in two different pipeline systems using WANDA
the velocities in a pipe system change so rapidly that the elastic transient simulation software.
properties of the pipe and liquid must be considered in an
analysis, then we have an elastic theory of hydraulic transient
commonly known as water hammer [10,11]. A water hammer 2. Methodology
is a pressure surge or wave caused by the kinetic energy of a
fluid in motion (especially in conduit systems) when it is forced Computer modelling and simulation were done for two
to stop or change direction suddenly [12]. systems using the WANDA Transient simulation tool. The
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
Modelling and transient simulation of water flow 3
description of the two systems investigated (system A and diameter of the vertical and section of galvanized steel pipe
system B) is explained in Sections 2.1.1 and 2.1.2. d2 diameter of the plastic tube (air vessel). HA is the constant
reservoir head. HY is the head in horizontal pipe section at a
2.1. Description of investigated systems distance LY from the reservoir, HZ is the head in vertical pipe
section at a distance LZ from the horizontal pipe, H is head in
2.1.1. System A: Graze laboratory set-up the air pocket, v1 is the velocity in the galvanized steel pipe,
System A (see Fig. 1) is a geometric model of Graze’s labora- and v2 is velocity at the interface between air and water in
tory set-up [14]. A pressure transient situation is triggered by the plastic tube (air vessel).
the sudden closure of the quick acting valve. The consequent
oscillation of the liquid column as soon as the valve closes is
2.1.2. System B: Simple water main riser system
what the laboratory set-up seeks to replicate.
L11 is the horizontal section of galvanized steel pipe, L12 is The simple main water riser system in Fig. 2 is a much larger
vertical section galvanized steel pipe, LY is length along L11 system when compared to the Graze Laboratory set-up. For
measured from the reservoir to location Y; LZ is length along instance, the volumetric size of the air vessel of this system is
L12 measured from pipe L11 to location z, L21 is the length of about 440 times bigger than that of system A. A pump
plastic tube, LTC is the distance between upper end of plastic discharges flow at a rate of 381 m3/h (1400 gpm) to a constant
pipe to location of thermocouple, X is the length of air pocket head reservoir with a static lift of 64 m. The transient condition
measured from upper end of plastic pipe, and d1 is the is triggered by a sudden pump trip.
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
4 P.U. Akpan et al.
2.2. Computer modelling and simulation using WANDA 2.2.1.2. Governing equations of the computer model. Some of
Transient the components used in this model are represented mathemat-
ically in WANDA Transient as follows:
Computer models were set up for systems A and B. These BOUNDH (Reservoir A): It is a constant pressure head
models are discussed in Sections 2.2.1 and 2.2.2 respectively. boundary condition. It prescribes the head (HGA Þ (see Eq. (1))
The following assumptions were incorporated while setting up at the main supply.
the computer models in WANDA: No cavitation, dynamic
HGA ¼ HA hatm ¼ Constant ð1Þ
friction (steady friction), no dissolution of gases, Newtonian
fluid (water was used), Incompressible flow and inelastic pipe where: HA ¼ absolute pressure head at the main supply (m);
for the rigid column analysis, no column rupture occurs, hatm ¼ atmospheric pressure head (10.3022 m as in [14]); and
Instantaneous opening of Sty. Gate valve, and the discharge HGA = gauge pressure head at the main supply (m).
area of the valves is the same with the cross sectional area of
the adjacent pipes. Pipe Lambda: The head loss between two points say 1 and 2
along the pipe during transient state analysis is given as
2.2.1. System A: Graze laboratory set-up L dv1 fL
H1 H2 ¼ þ 1 v1 jv1 j ð2Þ
2.2.1.1. Model formulation. A computer model was set-up in g dt 2gd1
WANDA Transient version 3, as shown in Fig. 3 to model
where H1 ¼ upstream head at point 1 (m); H2 ¼ downstream
the experimental set-up of Graze (see Fig. 1) A BOUNDH
head at point 2 (m); v1 ¼ flow through pipe (m/s); L ¼ pipe
reservoir was used to model the constant head (HA Þ main sup-
length between point 1 and 2 (m); d1 ¼ inner pipe diameter
ply. The quick-acting valve, galvanized steel pipe (vertical and
(m); and f1 ¼ friction factor in the pipe (–).
horizontal sections) and plastic pipe (containing the air pocket)
were modelled with a STy: gate (ideal standard valve), a pipe 2-way resist quadr.xi: specifies the overall loss co-efficient,
lambda and a vertical non-vented air vessel respectively. C, (except friction) in both directions of flow.
The two-way resist.quadr. xi component was used to model
Air vessel (vertical non-vented) is modelled using Eq. (3).
the overall hydraulic loss co-efficient (except friction) from
This describes the behaviour of the enclosed air:
the main supply to the air vessel.
Since it is required to set-up an initial pressure head (Hi ) in PVn ¼ C ð3Þ
the air vessel before the start of transient state simulation, two
additional components were attached for this purpose namely; 2.2.1.3. Model simulation (steady and transient state). The sys-
A second BOUNDH reservoir and valve were attached to tem parameters used for the simulations are presented in
node E. The hydraulic node level C of this new reservoir is Appendix A. Seven simulation runs were carried out: Runs
the same as node E (even though the diagram suggests 1AB, 2AB, 3AB, 5A, 6A, 21A and 23A. For each of them a
otherwise). rigid column and a water hammer analysis was done for poly-
And once the initial pressure head (Hi Þ) in the air vessel is tropic index values of n ¼ 1:0; 1:2 and 1:4: and these runs were
set through a steady state simulation to approximate the initial broadly classified as the positive pressure input and the nega-
laboratory value given in [14], then this additional BOUNDH tive pressure input. For the positive pressure input Runs
reservoir is isolated from the system by closing the second (1AB, 3AB, 5A, 21A and 23A) the absolute pressure head at
valve close to node C. It should be noted that the geometric the mains ðHA Þ was greater than the initial absolute pressure
height shown in Fig. 3 is not an indication of the true height head ðHi Þ of the air in the vessel. These runs depict typical
of the physical system. This is just done for proper visualiza- valve closure scenarios, where the sudden pressure rise due
tion as the elevation above datum for each system component to the obstruction of flow causes the fluid to flow backwards
was well incorporated in the model. The transient is initiated to the air vessel first before subsequent oscillations occur. Runs
by suddenly opening the Sty: Gate. 21A and 23A were for vessels with a larger diameter (0.096 m).
Valve
E
D B A Airvessel vertical non-vented
Constant Head STy: Gate Pipe lambda 2-way resist quadr. xi
BoundH (reservoir)
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
Modelling and transient simulation of water flow 5
BoundH (reservoir)
A Airvessel vertical non-vented E
B
BoundQ (reservoir) Checkvalve (ideal) Pipe lambda
Figure 4 Computer model for the simple water main riser system.
A typical pump trip scenario is depicted by the negative Ideal check valve: The loss – coefficient was taken as zero,
pressure runs (Runs 2AB and 6A), where the initial absolute and continuity equation applies. It closes when discharge is less
pressure of the air, ðHi Þ in the vessel causes flow out of the ves- than zero i.e. V_ < 0.
sel first before at the event of a pressure transient. This allows In addition to the assumptions (i)–(vi) made in Section 2.2,
for the study of the air behaviour undergoing temperature the following simplifications were also made for this system:
changes below ambient conditions. The time for simulation
was set to five seconds for the smaller vessel and ten seconds (i) The discharge area of the valves is the same with the
for the larger vessel runs. cross sectional area of the adjacent pipes.
(ii) The pump discharge drops from 381 m3/h to zero within
2.2.2. System B: Simple water main riser system 0.01 s as soon as the pump is tripped.
2.2.2.1. Model formulation. The water main riser system is mod- (iii) The pipe connecting the pump and the check is so short
elled in WANDA as shown in Fig. 4. The pump was modelled when compared to the overall pipe length that the
using a discharge boundary BOUNDQ (reservoir). At steady hydraulic node condition of the pumps’ discharge is
state, it discharges the specified flow rate into the system. How- the same as the inlet of the check valve.
ever, during transient analysis, the discharge is time dependent (iv) The air vessel is located close to the check valve and
and a linear variation has been adopted for this model. An ideal pump—the hydraulic condition of the check valve exit
check valve with zero loss co-efficient at the fully opened posi- is the same as the inlet condition in the air vessel.
tion was used. A vertical non-vented air vessel, pipe lambda
and a constant pressure head boundary condition (BOUNDH
reservoir) were used to model the air vessel, the main riser and 2.2.2.3. Model simulation (steady and transient state). The
the overhead reservoir respectively. system parameters used for the simulations are presented in
Appendix B. Simulation was done for only two cases (Case
2.2.2.2. Governing equations and assumptions. The governing B and Case D). In each case, a rigid column and a water
equations for each of the system component used for this com- hammer analysis was done for Polytropic index values of
puter model are the same as discussed in Section 2.2.1.2. The n ¼ 1:0; 1:2 and 1.4. Both cases depict a typical pump trip
only new component that was introduced here is the time scenario where the initial absolute pressure of the air, Hi , in
dependent discharge boundary reservoir (BOUNDQ) which the vessel causes flow out of the vessel first at the event of a
has been discussed in Section 2.2.1.2. pressure transient. The major difference between both is
1.5
30
1.3
25
1.1
20
0.9
0.7 15
0.5 10
0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s)
(a) (b)
Figure 5 Comparing simulation results with experiment for RUN 2AB.
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
6 P.U. Akpan et al.
2.3 40
Figure 6 Water hammer versus rigid column analysis for Run 2AB at n = 1.2.
30
3. Results and discussion
Profile of Pipe
25 Steady state Head (m)
Head min (Rigid)
Head max (Rigid)
3.1. Validating polytropic model simulation results from
20 Head min(Water hammer) WANDA transient
Head max (Water hammer)
Head (m)
15
In order to validate the results of the computer models, exper-
10 imental data available in [14] were used for comparison. The
experimental results available were for the dynamic behaviour
5 of the air in the air vessel. If the prediction of the dynamic
behaviour of the entrapped air in the air vessel is accurate then
0 the other WANDA Transient predictions for the flow along
the pipeline could be said to be accurate. Fig. 5(a) and (b)
-5 shows the time history of volume per unit area (air column)
X- Distance (m) and absolute pressure head for the entrapped air in the air ves-
-10
0 2 4 6 8 10 12 14 16 18 sel. The simulation results for n = 1.2 closely predict the
experimental result obtained by Graze [14]. This same observa-
Figure 7 Pressure head envelope along the pipeline (comparing tion was made for the other simulation runs and cases.
rigid column versus water hammer analysis for Run 2AB at It should be noted also that some results presented here
n = 1.2. such as the prediction of the pressure head envelope along
Case B Case D
Case B Case D 120
2.5 System B: Air vessel
Absolute Pressure head (m)
100
2.1
90
1.9
1.7 80
1.5 70
1.3 60
1.1 50
0.9 40
0 20 40 60 80 0 20 40 60 80
Time (s) Time (s)
(a) (b)
Figure 8 Effect of initial air volume on simple water main riser system.
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
Modelling and transient simulation of water flow 7
System B Profile of Pipe the runs. The rigid column and the water hammer analysis pre-
Minimum Head (Case B)
100 dict an identical behaviour for the enclosed air in the vessel for
Maximu head (Case B)
Minimum head(Case D) the air volume per unit area and absolute pressure head. How-
Maximum head (Case D) ever, this close proximity is not replicated for the pressure
80
envelope (lines showing the maximum and minimum pressure
values along the pipeline) in the pipe (see Fig. 7). The water
Head (m)
n= 1.2 100
Absolute Pressure Head (m)
90
1.6
80
1.4
70
1.2
60
1
50
0.8 40
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Time (sec) Time (sec)
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
8 P.U. Akpan et al.
Figure 11 Pressure envelope along the pipeline for Case D (the effect of polytropic index).
because the final volume of expansion of the air for the of the enclosed air in an air vessel during pressure transients
‘‘isothermal process” is much larger than the ‘‘adiabatic pro- and how it protects pipelines.
cess”. The rate of change of the pressure in Fig. 10(b) is much The Graze experiment simulation results obtained for poly-
rapid for the adiabatic process n ¼ 1:4 than for the isothermal tropic index value of n ¼ 1:2 tend to agree more with the
process n ¼ 1:0. This is to be expected in accordance with the experimental results for all the Runs in the experimental sys-
Polytropic relation (Eq. (3)). The consequence that this has on tem while simulation results obtained for polytropic index
pressure envelope of the pipeline is shown in Fig. 11. For a value of 1.4 for the simple main water riser system tend to
constant initial volume of air, the effect of designing an air ves- agree more with those from Graze’s results. The level of accu-
sel with a polytropic index value of 1.0 gives the pipeline a bet- racy of the predictions from the computer models for both sys-
ter protection as the minimum head predicted is much higher tems is sufficiently high.
than for other index values. Although this comes with increase
in air vessel size and hence increased cost, these effects were Acknowledgement
also observed for the other simulation runs and cases.
Thankfully, the authors do acknowledge Esso Exploration
4. Conclusions Nigeria Limited (EEPNL), a subsidiary of Exxon Mobil in
Nigeria, for the MSc scholarship award given to the principal
A computer modelling and transient simulation of water flow author and Deltares for providing WANDA Transient
in two different pipeline systems were considered in this work software used in this research.
using WANDA Transient software.
The simulation results shows that it is reasonably accurate Appendix A
to approximate the air behaviour in air vessel used for water
pipeline protection systems, by using a rigid column analysis, System parameters and initial conditions for System A (Graze
as this simplifies the problem and saves computational Laboratory Set-up) simulation [14].
time. The same cannot be said of the pipeline pressure head
envelope predictions as the water hammer analysis is still more Note:
accurate.
The trend of the results obtained from both systems’ com- (i) Gauge pressure head (H p Þ specified for the temporary
puter models agrees with already established knowledge of BOUNDH reservoir used to pressurize the air vessel ini-
effects of initial air volume, effect of polytropic index values tially was obtained by carrying out a steady state analy-
and rigid column approximations on the dynamic behaviour sis of the system in WANDA.
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
Modelling and transient simulation of water flow 9
Table A1 System parameters and initial conditions for graze experiment [14].
Parameters/variables Run no.
1AB 2AB 3AB 5A 6A 21A 23A
L11 (m) 16.4592 16.4592 16.4592 16.4592 16.4592 16.3678 16.3678
L12 (m) 1.0363 1.0363 1.0363 1.0363 1.0363 1.0363 1.0363
L21 (m) 3.0480 3.0480 3.0480 3.0480 3.0480 3.0480 3.0480
d1 (m) 0.0508 0.0508 0.0508 0.0508 0.0508 0.0508 0.0508
d2 (m) 0.0512 0.0512 0.0512 0.0512 0.0512 0.0960 0.0960
hatm (m) 10.3022 10.3022 10.3022 10.3022 10.3022 10.3022 10.3022
HA (m) 22.4333 22.4333 22.4333 15.3010 15.3010 22.4333 15.3010
HGA (m) 12.1311 12.1311 12.1311 4.9989 4.9989 12.1311 4.9989
Hp (m) 2.26 30.36 2.27 2.265 19.98 2.26 2.25
f1 0.029 0.029 0.029 0.029 0.029 0.029 0.029
f2 0.021 0.021 0.021 0.021 0.021 0.0207 0.0207
C 1.81 1.81 1.81 1.81 1.81 2.25 2.25
Initial conditions for transient simulation at vi = 0 m/s
Hi (m) 10.3022 37.1856 10.3022 10.3022 26.8224 10.3022 10.3022
Xi (m) 1.8288 0.6096 1.8166 1.8227 0.6218 1.8288 1.8349
Ti (°C) 20.1833 20.1833 19.0722 18.7940 18.7940 21.4610 21.4610
Time of valve opening (s) 0.0491 0.0446 0.0359 0.0394 0.0441 0.0388 0.044
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006
10 P.U. Akpan et al.
Mr. Akpan completed a BEng in Mechanical Dr. Eke completed his PhD and MEng in
Engineering (Energy and Power option), Mechanical Engineering in 2015 and 2002
University of Nigeria, Nsukka (UNN) in respectively from University of Nigeria. His
2006. In 2011, He completed an M.Sc. BEng was obtained in 1996 from Federal
programme in Process Systems Engineering University of Technology, Owerri. His
(Energy systems and Thermal processes research interests are in thermofluids, energy
option) at Cranfield, University, UK. He is a systems, Power plant Engineering etc.
member of American Society of Mechanical
Engineers (ASME) and Energy Institute (EI)
UK. He works as Lecturer I in the
Department of Mechanical Engineering,
UNN. His teaching responsibilities are on the following courses:
Measurement and Instrumentations Mechanical Engineering Labora-
tory, Heat and Mass Transfer. He has seven (7) Publications. Professor Hoi graduated from University of
Hong Kong with first class honours in
Mechanical Engineering and completed PhD
Sarah Jones completed a BA (Hons) in Engi- in turbomachinery from University of
neering Economics and Management at Newcastle (UK). After a period in GEC Gas
University of Oxford. She has worked with Turbines, he worked at the British Hydrody-
Cranfield University and BHR Group Lim- namics Research Association as a research
ited. At present, She is a Principal Hydraulic and then a senior engineer. Hoi lectured in
Engineer at Mott MacDonald. Her research Hong Kong Polytechnic University for a
interests are in Pressure surge analysis, couple of years before returning to the UK as
hydraulic design of pumping stations and principal engineer with a water industry
hydraulic structures (computer and physical consultant. He joined Cranfield University in the 90s as a Senior
scale modelling). Lecturer. Hoi has over 100 publications. He is the Head of the Process
Systems Engineering Group.
Please cite this article in press as: Akpan PU et al., Modelling and transient simulation of water flow in pipelines using WANDA Transient software, Ain Shams Eng J
(2015), http://dx.doi.org/10.1016/j.asej.2015.09.006