Pipeline Leak Detection Techniques

Published on May 2017 | Categories: Documents | Downloads: 40 | Comments: 0 | Views: 339
of 10
Download PDF   Embed   Report

Comments

Content









Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


25





P Pi ip pe el li in ne e L Le ea ak k D De et te ec ct ti io on n T Te ec ch hn ni iq qu ue es s


Assist.Prof. Timur Chis, Ph.D., Dipl.Eng.
~Andrei Saguna¨ University, Constanta, Romania


ABSTRACT. Leak detection systems range Irom simple, visual
line walking and checking to complex arrangements oI hard-
ware and soItware. No one method is universally applicable
and operating requirements dictate which method is the most
cost eIIective. The aim oI the paper is to review the basic
techniques oI leak detection that are currently in use. The
advantages and disadvantages oI each method are discussed
and some indications oI applicability are outlined.


1. Introduction

Our need to transport Iluids Irom the point oI production to the area oI end use has
led to a rapid increase in the number oI pipe lines being designed and constructed.
Many oI these carry toxic and hazardous products, oIten close to centers oI high
population or through areas oI high environmental sensitive. With the need to
saIeguard these lines, on-line monitoring is becoming routine and in some case 24
hours surveillance is mandatory. With the increase in Rumanian pipeline
deterioration by thieI and terrorism, the need Ior rapid and reliable pipe line
measurement and control systems will increase.
The review begins with a summary oI causes oI leaks and the implications oI
the Iailure to detect them. Basic techniques are covered and the Ieatures oI each are
brieIly discussed. The bulk oI the paper deals with modern computer based
techniques. Basic Ilow equations are covered and the on-line dynamic calculation
required are listed together with the impute data required to enable the monitoring to
be accomplished with the minimum oI downtime. The latest systems are capable oI
resolving down to 1° oI maximum rated Ilow with a response time oI a Iew
minutes. Practical experience veriIies this Iigure but the total cost oI such a system
could be high. The system thereIore becomes a compromise, between response,







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


26
perIormance, and alarm availability, and choice oI instrumentation is crucial. The
integration oI good quality instruments with advanced real time models seems to be
the current trend and the paper closes with some personal thoughts on Iuture trends.
There are Iour main categories oI pipe line Iailures. These are:
-pipeline corrosion and wear;
-operation outside design limits;
-unintentional third party damage;
-intentional damage.
Many pipelines are operated Ior a number oI years with no regard to any
possible mechanical changes occurring in the line. Some oI the products may be
corrosive, the pipe line may be leIt partially Iull Ior periods oI time, or atmospheric
eIIects may cause external damage. These three reasons are responsible Ior pipe line
corrosion and this may give rise to corrosion 'pits¨ developing along the line. These
are small in nature and could be responsible Ior material imbalances over a period oI
time. Very accurate Ilow metering can be used to detect this as discussed in the next
section. Abrasive Iluids or dust-laden gas streams can give rise to pipe line weir.
Again, this is a slow process, but should a weak spot develop (more oIten than not
close to a change in direction or section) then a pipe break may occur very rapidly
and totally unexpectedly.
Operation outside design guidelines is more common than is realized, as
operators seek to use the line Ior as many Iluids as possible. II the line is deigned Ior
a certain maximum temperature and pressure, then operation at higher pressure
and/or high temperature could lead to spontaneous Iailure. The problem could be
compounded iI the line has a large but unknown amount oI corrosion. Unintentional
third party damage may occur iI excavation or building occurs near buried lines.
More oIten than not the right-oI-ways are not clearly marked and lines are sometimes
broken by bulldozers or similar plant machinery possibly with Iatal results.
Intentional damage unIortunately is on the increase and pipe line carrying
Ilammable or high value products make ideal targets. Alarm systems linked to block
valves can help to minimize the amount oI products reels as a result oI sabotage, so
again certain lines are instrumented with the intention oI reducing the eIIects oI
planned terrorism.
The cost oI Iailure to detect leaks also Ialls into Iour main areas:
-loss oI liIe and property;
-direct cost oI loss products and lie downtime;
-environmental cleanup cost;
-possible Iines and legal suits.
These are all selI explanatory, with the most costly oI these being the last,
although any oI Iour areas could be very expensive. The size oI claims can run into







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


27
many millions oI dollars, so the cost oI Iitting and operating leak detection systems is
oIten insigniIicant compared with the cost oI Iailure oI the line.
It is this background that is causing operators and designers to turn to on-line
program integrity monitoring systems.


2 SIMPLE LEAK DETECTION SYSTEMS

The most basic method oI leak involves walking, driving, or Ilying the pipe line
right-oI-way to look Ior evidence oI discoloration oI vegetation near the line or
actually hear or see the leak. OIten unoIIicial pipe line monitoring is perIormed by
people living nearby who can inIorm the operator oI a problem with the line.
The most cost eIIective way to detect leaks in non-Ilammable products is to
simply add an odorant to the Iluid. This requires some care in selection, as Irequently
the odorant has to remove beIore the transport Iluid can be used. Organic compounds
make the most useIul odorizes, especially when the Iluid being carried has no natural
smell oI its own. The disadvantage oI such method is that iI the leak occurs in an area
oI no population the leak will go undetected unless the line is walked regularly by
pipe line surveillance crews carrying suitable 'stiIIer¨ detector. Thus to the apparent
low cost oI this method have to be added the cost oI removing the odorant and
maintaining staII to check the line at Irequent intervals. The location oI a leak is also
dependent on prevailing weather conditions. Strong winds may disperse the smell
and atmospheric inversion may give an incorrect location oI leak and the uncertainty
oI relying on this method alone is high. Nevertheless it is useIul method iI used in
conjunction with other techniques.
Simple line Ilow balances are Irequently used to check Irom gross imbalances
over hourly or daily based. This method may identiIy that a leak is present but Ilow
meters at each end oI the line will not identiIy the leak location. A line pressure
measurement system will be required in conjunction with the Ilow meters to
establish that the pressure gradient has changed Irom the on-leek situation. The
method is useIul, however in identiIying the existence oI corrosion pits as the outputs
oI Ilow meters at each end oI his line will consistently diverge iI Ilow in the line is
maintained constant. II line Ilow rate varies with time, that imbalances are more
diIIicult to detect, since the Ilow meter outputs ay vary nonlinearly with Ilow rate or
may have diIIerent Ilow characteristics Irom each other.
A loss oI product will be identiIied simply as the diIIerence between the steady
state inventory oI the system and the instantaneous inlet and outlet Ilows.
Mathematically this is:
l out in
J J J J ÷ ÷ = A
where:
(1)







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


28
J A ÷leakage volume;
V
in
÷meter inlet Ilow;
V
out
÷meter outlet Ilow;
V
l
÷pipe line Iluid inventory.
This last term can be calculates As the average oI the integrate inlet and
outlet Ilows in simple systems, but as will be seen later, the value oI this term can
be calculated more accurately and easily in real time as a Iunction oI several
variables.
Another method is based on detecting the noise associated with or
generated by a leak. There are many instances where Iluid Ilow can generate
vibration at Irequencies in excess oI 20 Hz. These Irequencies are in the
ultrasonic range but can be detected with suitable transducers. The device can be
made portable so that pipe line crews can clamp a transducer at any point along
the line to check Ior noise by noting the signal strength, the source oI the leak can
be pinpointed.
A similar technique, though base on a diIIerent principle, is the acoustic
,wave alert¨ monitor, more correctly called a negative pressure wave detector.
This a piezoelectric sensor that give an output went dynamically stressed. When
a leak Iollowed by rapid line repressurization a Iew milliseconds later. The low
pressure wave moves away Irom the leak in both directions at the speed oI sound.
The pipe walls act as a waveguide so that this rareIaction wave can travel Ior
great distances attenuating in amplitude as iI does so.
Sensors placed at distances along the line can be triggered as the wave
passes and location oI the leak can be calculated Irom the line conditions and the
internal timing devices in the instrument.
Such devices are particularly useIul in identiIying large breaks in lines
very rapidly since the transient wave typically moves 1 mile in 5 seconds in gases
and almost 1 mile per second in liquids. Response is thereIore on the order oI a
Iew seconds depending on the positioning oI the transducers.


2. PIG BASED MONITORING SYSTEMS

Pipe lines are Irequently used Ior pipe line commissioning, cleaning, Iilling,
dewaxing, batching, and more recently pipe line monitoring. This last type oI pig
can be designed to carry a wide range oI surveillance and monitoring equipment
and can be used at regular intervals to check internal conditions rather than
continuously monitoring the line. Data, however, can be built up over a period oI
time to provide a history at the line. This inIormation can be used to predict or
estimate when maintenance, line cleaning, or repairs are required. II a leak is







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


29
detected, Ior example, by Ilow meter imbalance, the location can be Iound by
using a pig with acoustic equipment on board. This will alarm when the detection
equipment output reaches a maximum and the precise location oI pig can be
conIirmed by radio transmitters also mounted on board. Pigs require tracking
because they may become stuck, at a point oI debris build-up. Pigging should be
carried out at a steady speed, but occasionally the pig may stop and start,
particularly in smaller lines. InIormation on when and where the pig stop is
thereIore important in interpreting the inspection records. Pig tracking is not new
and many such proprietary systems exist. In the best systems, however, a picture
oI the line is oIten programmed in so that outputs Irom junctions, valves,
crossovers ad other geometries act as an aid to location. Pig tracking can make
use oI the acoustic methods discussed earlier. When the sealing cups at the Iront
oI the pig encounter a weld, vibrational or acoustic signal are generated. Each
pipe line thereIore has its characteristic sound pattern. When a crack occurs this
pattern changes Irom the no-leak case and the location can be Iound Irom direct
comparison. The technology has become so advanced that inIormation on dents,
buckles, ovality, weld penetration, expansion and pipe line Iootage can be
generated.
The equipment s oIten simple, consisting oI sensor, conditioning, and
ampliIier circuits and suitable output and recording devices. Such a device
developed by British Gas. The range oI detection is dependent on the pipe line
diameter and the type oI pig.


3. COMPUTER BASED MONITORING SYSTEMS

It is computer-based systems that the greatest amount oI data can be gathered,
processed, analyzed, and acted upon in the shortest period oI time. Programs can
calculate the inventory oI the line at any time and compare this with accurate
measurements at any section in the system. The eIIects oI pressure and
temperature on line dimensions Ior example, can be calculated to provide an
accurate estimate oI the mass oI Iluid in the line. Data Irom a wide range oI
instruments can be transmitted by telemetry, radio, or phone links to a central
computer which monitors the ,health¨ oI line continuously. By changing
programs and subroutines, a astound oI Iunctions and task can be accomplished
very easily and cost eIIectively.
The many Iunctions that can be perIormed by computer-based systems include
not only leak detection but also:
-pig tracking;
-back tracking oI Iluid;







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


30
-inventory accounting;
-on line Ilow compensation;
-instrument data and malIunctional checking.
Such systems have very rapid response and have the advantage oI multiple inputs
being required beIore leaks are declared. Thus some systems can run Ior short
outage periods with no loss oI integrity. They are oIten complex and costly to
install but once the initial capital investment has been made running cost are low.
The Iirst section in the detailed review oI such systems looks at the phenomena
that need to e modeled when a leak occurs.


4. PIPE LINE LEAK PHENOMENA

When a leak occurs in a pipe line measured pressure downstream oI the leak Ialls
but the pressure at the same location is predicted to rise. The Iirst is not diIIicult
to understand as the line is depressurizing as mass leaves through the lea. The
second eIIect can be explained as Iollows. The equations predict pressure based
on measured Ilow based on measured pressure. As mass leaves the system
through the leak hole, a reduced Ilow at the downstream end is compared to the
inlet Ilow. This may not have changed and so to balance the system, the
equations predict a downstream pressure rise. In physical terms the model thinks
to line is ,packing¨ and total system inventory is increasing. There is thereIore a
divergence between measurement and modeled pressure.
The same is true oI Ilow changes, but here the inlet Ilow could increase
due to lower pipe Ilow resistance between eater and leak while the section outlet
Ilow will Iall as mass leaves through the leak instead oI passing through the
meter. Thus a real imbalance will result. The model however will show an
inconsistency since the pressure comparison will indicate line packing and the
Ilow comparison a line unpacking. II selected pressure and Ilow imbalance limits
are exceeded a leak is declared. The magnitude oI the leak is predicted Irom the
Ilow imbalance and the location Irom the pressure proIile imbalance and the Ilow
leak indicators. The impact oI instrument accuracy is important Irom the leak
detections. It is vitally important to good leak sizing and location to have the best
pipe line instrumentation possible to minimize uncertainty.


5. BACKGROUND PHILOSOPHY OF PIPE LINE MODELING

Real time modeling is a technique that uses the Iull data gathering capabilities oI
modern digital systems and the computational power oI small computers to give
accurate ,snapshots¨ oI the pipe line. The whole system is under the control oI







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


31
SCADA package oI programs, which poll the data stations on the line, process
the data, control the running oI the transient pipe line model and activate the
alarm and leak location routines. In addition to these basic soItware modules,
more complex systems might include a predictive model to analyze ,what iI`
operating scenarios, provide an optimization routine Ior least cost operating
strategies or include a separate man/machine interIace Ior the model system. The
SCADA interIace is responsible Ior acquiring the data Irom the SCADA system
and relating them to the model representation oI the line. As a point in the system
where two large and independently developed systems join, this can be the
source oI many problems in the implementation oI the real time modeling. Once
the measurement data have been obtained, noise altering and plausibility
checking can be perIormed prior to running oI the model. The model is the
mathematical representation oI the pipe line and will include such Ieatures as
elevation data, diameters, valve and pump locations, changes oI direction and the
location or cross-over and junctions. The model provides data on the Ilow
conditions within the line at intervals between seconds and minutes, depending
on operational needs. Whit the data available Irom both the measurement system
and the pipe line model, the real time applications modules are run. These are the
leak detection and location routines in the context oI integrity monitoring. He
leak detection module Iunctions by computing the diIIerence between the
modeled Ilows and pressures and the measured values at all points where
measurements not already used as boundary conditions are located. Because the
model accounts Ior normal transient operations, these diIIerences will be small
under normal condition. When a leak is present, the diIIerences become larger
since the model system does not account Ior leakage. When these diIIerences
exceed preselected values, a leak alarm is declared. Sophisticated voting schemes
which require multiple leak indicators to be in alarm Ior several consecutive time
intervals are used to reduce Ialse alarms while maintaining low thresholds. OIten
a simple pie line balance oI the type discussed earlier is used as a back-up to
veriIy the transient model. Response characteristic are, however, much slower
than the real time model.
Once the leak detection module declares a leak, the location routine is activated.
The location is calculated Irom the magnitude and distribution oI the leak
indicators. As an example, in a straight pipe line with an upstream Ilow
discrepancy and a downstream pressure discrepancy as leak indicators, it is an
easy calculation to determine where the leak must be such that the leak Ilow
when added to the modeled Ilow will produce the additional pressure drop
observed at the downstream end. Solutions Ior pipe networks are more
complicated and unique locations do not always exist. This might be the case







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


32
with parallel looped lines. In this case all the calculated locations should be
checked.
The components oI real time modeling work together to reduce the large volumes
oI raw data Irom the data acquisition system to a much smaller number oI
parameters and alarm that are more meaningIul to pipe line operations. In the
case oI integrity monitoring, this means leak event that could not be detected by
inspection oI the measured data can be Iound and isolated quickly and reliably.


6. BASIC PIPE LINE MODELING EQUATIONS

The transient pipe line Ilow model is the heart oI a pipe line modeling system.
The model computes the state oI the pipe line at each time interval Ior which data
are available. The state oI the pipeline is deIined as a set oI pressures,
temperatures, Ilows, and densities that describe the Iluids being transported at all
points within the system. These quantities are Iound as the solution to a set oI
equations which describe the behavior oI the pipe line system. These basic
equations are the Continuity equation, the Momentum equation, the Energy
equation and an equation oI state.
The continuity equation enIorces the conservation oI mass principle. Simply
stated. It requires that the diIIerence in mass Ilow into and out oI section oI pipe
line is equal to the rate oI change oI mass within the section. This can be
expressed mathematically by the relation:
0
) ( ) (
= +
dx
AJ d
dt
A d µ µ

The momentum equation describes the Iorce balance on the Iluid within a
section oI pipe line. It requires that any unbalanced Iorces result in an
acceleration oI the Iluid element. In mathematical Iorm, this is:
0
2
1 ) (
=
×
+ × + × + × +
D
J fJ
dx
dH
g
dx
dP
dx
J d
J
dt
dJ
µ

The energy equation states the diIIerence in the energy Ilow into
and out oI a section equals the rate oI change oI energy within the section.
This equation is:
0 ) (
4
2
) (
3
= ÷ + ÷ × ×
×
+ × +
g
T T
cD
U
cD
J f
dx
dJ
dT
dP
c
T
dx
T d
J
dt
dT
µ µ

These thee are the basic one dimensional pipe Ilow equations and are
present in one Iorm or another in all transient pipe models. What is needed
to solve them, however, is in relation between the pressure, density and
temperature Ior the Iluid-an equation oI state.
(2)
(3)
(4)







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


33
The state equation depends on the type oI Iluid being modeled, as no one
equation Iully describes the variety oI products that are shipped in pipe line.
Some oI the Iorms in use include a bulk modulus type oI relation oI the
Iorm:
(
¸
(

¸

÷ +
÷
+ = ) ( 1
0
0
0
T T
B
P P
o µ µ
This is normally used Ior liquids that can be regarded as incompressible.
The bulk modulus B and the thermal expansion coeIIicient o can be
constant or Iunctions oI temperature and/or pressure depending on the
application. For light hydrocarbon gases a basic equation such as:
T Z R P × × × = µ
Is appropriate, where Z (the compressibility) is a known Iunction oI
temperature and pressure. For reasonable ranges oI temperature and pressure
a Iunction oI the Iorm:
v
T
P
Z
~ ÷1
1

may be adequate. For conditions where Iluids are transported at or near the
critical point, a more sophisticated correlation is required to obtain the required
accuracy but there is still a large uncertainty in the true density under these
operating conditions are they should be avoided wherever possible. Many real
time systems have been installed on lines carrying ethylene, butane, propane and
LPG products are used the Benedict-Webb-Rubin correlation as modiIied by
Starling with reasonable results. A Iurther complication arises iI the product is
not uniIorm throughout the system. This can occur due to batching oI Iluid or
Irom varying inlet condition. The Iirst is more common where diIIerent products
are shipped in a common line. The properties are essentially discontinuous across
the interIace oI two Iluids, but can be considered as uniIorm within batches. The
basic problem here is to keep tack oI the location oI the interIace. Systems with
continuous variation in inlet conditions occur n both liquid and gas systems. The
variations can result Irom mixing oI Iluid oI slightly diIIerent composition or
large variations in supply conditions.
The governing equation s presented is non-linear partial diIIerential equations
which are not suitable Ior machine computation. They have to be solved by
implicit or explicit Iinite diIIerence techniques or the method seems the most
appropriate, as the over two methods could give rise to mathematical instabilities
iI the wrong time step or distance interval is used.



(5)
(6)
(7)







Anale. Seria InIormaticá. Vol. V Iasc. I - 2007
Annals. Computer Science Series. 5
th
Tome 1
st
Fasc. - 2007


34
7. SYSTEM DESIGN ASPECTS AND GUIDELINES

The availability o leak alarm uptime depends heavily on the system design and
the choice oI hardware. Generally the more complex the system, the greater the
risk oI leak indicator loss, but the more accurate the location oI the leaks. Design
is thereIore a compromise between cost, perIormance, and reliability. For the
three simple alarms oI Ilow imbalance, pressure imbalance, and acoustic alarms
between station A and B and C the Iollowing components are needed:
-MASS FLOW: 2 Ilowmeter, 2 pressure sensor, 2 temperature sensor, 2 RTUs, 2
communications links, 1 computer (11 elements);
-PRESSURE: 4 pressure sensors, 4 RTUs, 4 communications links, 1 computer
(13 elements);
-ACOUSTIC: 2 acoustic monitors, 2 RTUs, 2 communications links, 1 computer
(7 elements);
II combined hybrid alarms oI Ilow/acoustic, Ilow/pressure, or
pressure/acoustic are used then the number oI components in the chain is
increased.
By summing the component availabilities Ior each element, an uptime Ior
each alarm can be estimated. From such an analysis, the conclusion can be drawn
that the system should be made as simple as possible or instruments should be
made as simple as possible or instruments should be installed in duplicate to
maximize alarm uptime.


CONCLUSION

The paper cannot do justice in such a short space, to the complex and diverse
subject oI leak detection. Such systems have been in operation in many Iorms all
over the world, but it is only recently that environmental as well as economic Iactors
have inIluenced their development. Instrument selection is critical, as is the need to
develop better thermodynamic models, Ior the next generation oI systems to become
more reliable and accurate.


REFERENCES

|Tim05| Chis Timur, ,Modern Pipe Line Monitoring Techniques¨, First
International Symposium oI Flow measurement and Control,
Tokyo, Japan, 2005 .

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close