|
|
HYDRAULIC TRANSIENTS AS A
MONITORING DEVICE
LENNART JÖNSSON
Department of Water Resources Engineering
University of Lund, P.O.Box 118, S-221 00 Lund,
Sweden
Tel: +46-46-2228993, Fax: +46-46-2224435,
Email: Lennart.Jonsson@tvrl.lth.se
ABSTRACT
Hydraulic transients occur at rapid flow changes in pressurized water
conveying pipelines. Normally such transients (waterhammer) are considered to
be a problem as they might damage the pipeline due to strong pressure peaks,
subatmospheric pressures or fatigue. However, normally occurring hydraulic
transients have also a potential to be utilized in an advantegeous manner. The
idea is that transients could be looked upon as a "probe" that
propagates through a pipeline and at the same time being affected by certain conditions
of the pipeline. Measurement and subsequent analysis of transients might thus
provide information on some aspects of the hydraulic status of a pipeline. This
paper discusses the analysis of measured transients in order to extract
information on hydraulic conditions of pipelines. This is done by means of two
specific measurements on pipelines concerning existence and location of a local
gas pocket and a leak respectively.
Keywords: Pipeline, hydraulic transient, leak, air pocket, location,
monitoring
INTRODUCTION
Hydraulic transients or waterhammer occur in pressurized conduits for
water conveyance due to rapid flow change, for instance when stopping pumps or
when operating valves. The flow change is accompanied by pressure changes which
propagate as a pressure wave through the conduit. The conventional view of
transients is that they constitute a problem for a conduit and its hydraulic
components as the pressure waves are characterized by strong peaks or periods
with very low pressure. Besides the transients are often oscillatory. These
properties could cause the conduit to break or lead to other harmful
consequences due to fatigue.
The more or less unavoidable transients in pipelines could, however, not
only be regarded as a problem - they might also be used in a positive sense.
The basic idea is that hydraulic transients actually are pressure waves
propagating back and forth in the conduit and that they are at the same time
affected by different characteristics of the conduit and of the flow. A measurement
and a subsequent analysis of a normal hydraulic transient - for instance at a
pump stop and a subsequent valve closure - can thus provide valuable
information on the conduit, on the functioning of its hydraulic components, on
the water flow. Hydraulic transients can thus be considered as a kind of a
"probe" which is transported in the conduit. The purpose with this
paper is to illustrate the possibility of using measured hydraulic transients
in order to derive certain properties of a conduit. Due to the restrictions on
paper length only two specific examples can be dealt with - detection of a gas
pocket and detection of a leak. There are, however, other possibilities too,
such as detection of a change of pipe material, detection of a change in pipe
diameter, detection of a check valve closing too late.
AIR/GAS POCKET IN A SEWAGE WATER PIPELINE
Fig 1 shows the profile for a PVC sewage water pipeline, 4400 m long,
inner diameter 203.4 mm and with a wall thickness of 10.8 mm. The pumping
station is equipped with four pumps, Flygt 3152-HT, with two pumps in series in
each of the two parallel branches of the station. On the basis of the curve for
one pump the following data are obtained for two pumps in series for
steady-state operation:
H0=43 m H2O, Q0=0.038
m3/s, NN=1450 rpm, I»0.2 kgm2
The geodetic height is 27.5 m. A swing check valve is located
immediately downstream of each pair of pumps in series. It is important to
notice that there are two local peaks of the pipe profile, one at L=2400 m and
one at L=3000 m. Transient pressure measurements were performed just downstream
the check valve - see Jönsson [1] for more details. Fig 2 shows an example of
the measured transient when one pair of pumps is in operation and when these
two pumps are stopped simultaneously. At pump stop the pressure decreases to
the atmospheric pressure approximately and after
Figure 1. Pipe profile for PVC sewage water conduit. Discharge end at
+37 m. Pipeline length 4400 m
a short while the check valve closes when the flow tends to reverse and
the typical pressure oscillation is obtained. According to basic hydraulic
transient theory the pressure wave velocity a is given by:
where L = pipe length
Thus
This is, however, too high a value for a PVC conduit. If, on the other
hand, a realistic value of a is chosen, say a»390 m/s, a corresponding
length L of the pipeline can be calculated:
This value of L agrees well with the location of the local peak at +41
m. The most probable explanation to the shape of the hydraulic transient is
that a complete reflexion of pressure waves occurs at the local peak due to a
relatively large gas pocket which could occur naturally at a local peak -
especially when sewage water is transported.
Figure 2. Measured
transient immediately downstream the check valve for the PVC sewage water
conduit. Simultaneous stop of two pumps in series
DETECTION OF A LEAK
A leak on a pipeline will affect the shape of a hydraulic transient in
several ways. If an oscillatory pressure arises, for instance as the case in
Fig 1 shows, a leak somewhere along the conduit will cause the pressure
oscillations to be attenuated faster - the larger the leak the faster will the
attenuation be. A fairly large leak will cause the oscillatory pressure to
disappear more or less completely, see Jönsson [2]. A fast attenuation could
thus be an indication of a leak.
Figure 3.: Top:Steel pipe set-up for the detection of a leak using
transients. The pipe is discharging to the atmosphere via a ball valve. A leak
was simulated at a point 43 m upstream of the valve
Bottom: Measured transient with a simulated leak. Notice pressure change
at t=3.734 s due to the leak
Another effect of a leak is that pressure waves are partly reflected and
will thus also affect the transient pressure in a measurement point compared to
the no-leak case. A large number of measurements were performed on a steel
conduit, 134 m long in total, consisting of two parts - 98 m section, diameter
50 mm and a 36 m section, diameter 40 mm. The small section was connected to a
main conduit acting as a reservoir. The 50 mm section discharged to the
atmosphere via a ball valve, Fig 3, top. Leaks were simulated at different
points along the larger pipe section by means of small openings on the steel
pipe. The leak flow was possible to measure by means of a small turbine
flowmeter attached to each opening. The pressure transient was generated by
rapid valve closure and the pressure transducer was located immediately
upstream of the valve. The study comprised investigations of the effect of the
leak on the pressure transient for different leak flow/pipe flow ratios and for
different locations of the leak. In this paper one case will be discussed
concerning a simulated leak located 43 m upstream of the valve and with a valve
closure causing the pipe flow to decrease from 0.83 l/s (mean velocity v=0.42
m/s) to 0 l/s almost instantaneously and with a leak flow amounting to 0.09
l/s in steady state conditions, corresponding to 11% of the steady state pipe
flow. The initial phase of the transient is shown in Fig 3, bottom. At t1=3.661
s the pressure starts rising very rapidly with the Joukowski pressure. At t2=3.734
s, however, the pressure starts to decrease due to the reflected wave from the
leakage point which now has reached back to the measurement point at the valve.
A calculation of the distance L1 to the leak gives:
which agrees well with the real value L1=43 m. There is thus
a potential to indicate the existence of a leak as well as to locate it. The
reflected wave from the main conduit reaches the measurement point at t4=3.889
s giving for the total length L of the conduit system L=142 m calculated in the
same way as for the leak point described above. The value of the wave velocity
a was determined from transient measurements on the pipeline without any leaks.
DISCUSSION AND CONCLUSION
The paper has shown two examples where measured hydraulic transients in
pipelines are affected by factors related to the pipeline. This influence could
be more or less distinct - from small changes requiring an experienced observer
to discover up to drastic modifications of the transient. This fact has given
rise to the idea that hydraulic transients could not only be considered as a
problem for a pipeline. Transients might also have the potential to be utilized
for obtaining information on the pipeline. The basic idea is that hydraulic
transients are waves propagating in a pipeline, simultaneously being affected
by different changes in the conduit - change of pipe material, change of pipe
diameter, branching pipes, leaks, air pockets etc. A hydraulic transient could
thus be regarded as a "probe" propagating through a pipeline. Such
transients occur naturally in pipe systems at rapid flow change. The
"probe" is thus easy to "provide". Measurement, analysis
and possibly a comparison with a calculated transient have then got the
potential of providing valuable information on the pipeline. In order to arrive
at a wellfounded interpretation of a measured transient at least two conditions
should be met. Firstly it is of course of primary importance that there is a
thorough understanding of hydraulic transients as a physical phenomenon.
Secondly one should use existing knowledge about the pipeline - profile,
material, branches etc - as a help at the analysis of the transient and in
order to reduce the possibility for erroneous interpretation of the shape of
the transient; the latter since different factors can have similar influence on
the transient.
A visual inspection of a measured transient has the potential of
revealing or at least indicating certain conditions in a pipeline. A further
analysis could utilize a comparison between a measured transient with a
computed one where the latter could be based on a normal functioning of the
pipeline and normal assumptions about the operation of the components of the pipeline.
If such a comparison reveals significant differences this fact could point at
the requirement of more detailed investigations of the pipeline.
Another aspect of the possible use of transients concerns a more
systematic monitoring of a pipeline - for instance a main for sewage transport.
The idea would be that transients are measured for supposedly identical
operational conditions at different times after commissioning. The measurements
should be stored in a database and compared with recently obtained data. If the
latest measurement would deviate significantly from earlier measurements this
fact would indicate a hydraulic change in the pipeline requiring further
investigations. Thus a deteriorating functioning of a pump would imply a
decreased flow and consequently smaller pressure oscillations at pump stop and
a change in the closure procedure of an automatically closing shut-off valve
would affect the initial pressure rise.
ACKNOWLEDGEMENT
I would like to thank the VA-FORSK Foundation, Sweden for the support
making it possible to carry out measurements related to leak detection. Mr
Anders Svensson, RUST VA-TEKNIK, Lund, Sweden was also involved in these
measurements. I would also like to thank the Carl Trygger Foundation making it
possible to perform laboratory measurements.
REFERENCES
1. Jönsson, L., Pressure transients in a PVC sewage conduit - Klagerup.
Influence of gas/air pocket, Report on transient flows carried out during an
appointment as a researcher at the University of Lund, Department of Water
Resources Engineering, 1991
2. Jönsson, L., Leak detection in pipelines using hydraulic transients -
laboratory measurements. Department of Water Resources Engineering, Univ of
Lund, Sweden, Report to the Carl Trygger Foundation, 1994