906

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 25, NO. 5, SEPTEMBEWOCTOBER 1989

The Application of a Continuous Leak Detection

System to Pipelines and Associated Equipment
CHET SANDBERG, SENIOR

MEMBER, IEEE,

JIM HOLMES, KEN MCCOY, AND HEINRICH KOPPITSCH

damage to the environment, especially in undersea and remote

land application. These leaks are, however, the easiest to
detect since they are accompaniedby easily measured pressure
drops or volume discrepancies.
A more difficult and equally dangerous leak is the s m a l l
hard-to-detect leak. Corrosion, erosion, weld or joint failure,
and fatigue can all lead to small leaks. Leaks as small as 1 gall
h can build up large lost volumes before they are noticed. Up
to now, these types of leaks have been almost impossible to
detect with the conventional means described in the following
INTRODUCTION
section. The hydrocarbon sensor described later in this paper
is
the first economically feasible distributed system available
EAKS of hazardous fluids such as crude oil, gasoline, or
chlorinated solvents can result in very serious to monitor pipelines for small or large leaks.
environmental pollution if the leak is not quickly detected and
repaired. Even domestic drinking water may become contami- Current Leak Detection Methods
nated with toxic chemicals if the groundwater supplying
Leak detection systems are classified into two main categoaquifers and wells become contaminated with leaking hydro- ries: static and dynamic. Dynamic systems are preferred since
carbons. The U.S.Environmental Protection Agency (EPA) they can be used while the pipeline is operating. Static leak
has proposed regulations designed to prevent contamination of detection methods are useful after a leak has been detected in
groundwater from the estimated 1 . 4 - d i o n underground order to find its location. Continuous monitoring of pipelines
storage tanks in service. This focus on leak detection is just as gives rise to a number of leak detection techniques. Meter
serious in the area of transmission pipelines. In one instance variance or compensated volume balance is the predominate
approximately 30 0o0 barrels of No. 2 fuel oil was lost as a method. This method is limited by the accuracy of the volume
result of a leak in a buried pipeline in suburban New Castle measurement and the variations associated with it. Pressure
County, DE. When the leak was investigated, it was found to monitoring and rate of flow can detect large leaks, and overbe less than 1 gal/min. The fuel oil seeped into the water table and-short calculations can detect smaller leaks. Differential
and threatened a downgradient stream and irrigation pond. pressure transmitters across sectionalizing valves can continuEven though recovery wells were drilled, only about 10 ously monitor for a negative pressure, which indicates a large
percent of the fuel oil was recovered and significant economic leak. Another method was attempted by wrapping an oilresources were expended [l]. Continuous leak detection of soluble tubing around certain critical sections of pipeline. The
small leaks has been difficult in the past, but a new technology internal pressure of this tubing was then monitored for loss of
is available to continuously monitor pipeline transmission of pressure. However, this system was a complete failure and
hydrocarbons.
subsequently abandoned [2].

Abstract-In recent yeam the problem of leak detection in pipelines,

tanks, and process vessels has been the focus of many man-hours of
effort. Some examples of leaks occurring in pipelines, an overview of
classicalleak detection systems, and the engineeringbasis of a new type of
detector system are examined. This system is a flexible hydrocarbon
sensing cable that can be installed dong pipelines, in double containment
tanks and piping, or in trenches to detect and locate leaks of common
industrial hydrocarbon solvents or fuels while ignoring the presence of
water. The simple electrical circuit is also described, which locates and
detects a leak anywhere along the length of the sensor.

Types of Leaks in Pipelines and Piping Systems

Various types of leaks can occur in pipelines and piping
systems, and there are various types of leak detectors to
analyze each type. The rupture leak is the least common but
is very dangerous. Catastrophic failure can cause significant
Paper PID 89-29, approved by the Petroleum and Chemical Industry
Committeeof the IEEE Industrial Applications Society for presentationat the
1988 Petroleum and Chemical Industry Committee Technical Conference,
Dallas, TX, September 12-14. Manuscript released for publication February
221, 1989.
The authors are with Raychem Corporation,300 ConstitutionDrive, Menlo
Park, CA 94025.
IEEE Log Number 892875 1.

Compensated Volume Balance

The current major method of leak detection is the compensated volume balance method. This method essentially
measures the volume in and subtracts the volume out.
There are meters that are guaranteed repeatable to within -05
percent. In a typical system such as the one in operation in
Alaska from the Tesoro Nikiski refinery on the Kenai
Peninsula to Anchorage [3], an alarm will sound when there is
a significant difference in volume. The pump station management will determine if the difference in the two measured
volumes is the result of an operational change or if the pipe is
leaking. Operational changes can result from a change in

OO93-9994/89/O9OO-O$01.oO O 1989 IEEE

907

SANDBERG er al.: APPLICATION OF CONTINUOUS LEAK DETECTION SYSTEM

product grade, change of pumps or pumping pressure, or a

change in temperature because of storage tank changes.
If there has not been a change in operational criteria, the
pipeline will be shut down and sections blocked with valves to
isolate the possible leak. If before shutting down the pipeline
the discrepant volume should increase by a factor of 2, the
pumping stations will automatically be shut down. Should a
rupture of the pipeline occur, a negative surge will travel back
to a pumping station and will be detected by a suction pressure
switch, and the pipeline will be shut down. Vibratory sensors
are located at the pumping stations and will detect an
earthquake of 3.0 on the Richter scale and shut down the
pipeline.
The problem with the compensated volume method is that
there usually needs to be a large volume before conclusive leak
detection can be made. Leaks of 2 percent for liquids and 10
percent for gases are needed for detection, according to some
authorities. Smaller leaks can be detected with a nonlinear
adaptive state basis on pressure and flow measurements at the
inlet and outlet [4]. However, based on flows of lo00 barrels/
h, the leaks can be significant before they are detected.

Acoustic Traveling Pig

A novel leak detector is the Texaco pig, which detects a
leak by sensing the acoustic energy generated by escaping
fluid. The location is recorded by an on-board odometer and
the data are correlated to the ultrasonic data, thereby detecting
the leak. The leak detector instrument is articulated in four
sections: two for rechargeable batteries, one for a printer, and
one for the hydrophone and circuit cards. The articulated
pig is used on pipe sizes from 6-10 in. Twelve inch pipe is
checked by a model having all components in a single section
[5]. Leaks were successfully detected and located by the pig in
field pipelines in Canada and Texas.

Thermal and Electrooptical Methods

An airborne remote sensor can detect leaks from natural gas
pipelines by monitoring the methane and ethane gas in the
atmosphere above the pipeline. The sensor is a passive
electrooptical system, operating in a downward-lookingmode.
It is designed to detect low levels of methane and ethane by
looking at their infrared spectral signatures using a nondispersive gas filter correlationtechnique. This system is experimental, and the results of the field trials are unknown to the authors
[61*
A similar airborne system using thermal infrared images has
been tested on water distribution systems in South Dakota. A
typical system has from 640-2400 km of pipeline, and the
n o d line water loss is 10-15 percent. Major leaks,
generally from 10-200 m3/day, are easy to identify because
they usually pond. Minor leaks, called seeper leaks, which
generally range from 2-10 m3/day are more common and are
difficult to detect using conventional ground surveys. Fifteen
possible leaks were identified from thermal images. Five of
these sites were eventually confirmed as leaks [7].
HYDROCARBON
DISTRIBUTED
SENSOR
CABLE
The hydrocarbon sensor system [8] consists of an alarm
module and sensing cable. This sensing cable provides

continuity 1
wire

conductive
polymer layer

\Hala?
braid

Fig. 1 . Hydrocarbon distributed sensor cable.

distributed coverage by detecting leaks of most fuels and

solvents at any point along its length. The cable can be used in
lengths of up to 2 lan with a resolution of approximately 1
percent of the length for leak detection. The core of the cable,
Fig. 1, is constructed of an alarm signal wire, a continuity
wire, and two sensor wires. It is encased in a conductive
polymer layer that swells during exposure to most hydrocarbon-based solvents and fuels. This material is surrounded with
a Halar [9] braid that restrains outward swelling. When a
solvent or fuel contacts the cable, the conductive polymer
swells inward and makes electrical contact with the two sensor
wires. The cable must be replaced once it has contacted a
solvent or fuel.
This sensor has combined unique radiation chemistry and
conductive polymer technologies to develop an industrial
system. The conductive polymer formulation is critical to the
success of the device. The formulation must swell rapidly and
remain conductive if the sensor is to detect the fluid of interest.
Additionally, the polymer must develop sufficient swelling
pressure to force the conductive formulation through the
apertures of the braided sensing cable to contact the sensor
wires. If the polymer is not crosslinked it will not swell but
simply dissolve. Therefore, the formulation is crosslinked
using radiation chemistry to form a solvent-swellable threedimensional polymer network. The expansion pressure generated by this network is a function of the crosslink density and
interaction between the polymer and solvent [ l l ] , [12].
Two formulations were developed to cover the range of
common industrial solvents and fuels. One formulation was
developed to swell rapidly in nonpolar hydrocarbons such as
diesel fuel and gasoline and was used to prepare a fuel sensing
cable. Another formulation was used to fabricate a solvent
sensing cable for the detection of polar fluids such as
chlorinated solvents.
Although the time taken for the sensing cables to respond in
a given fluid is a key property of the sensor, other factors must
also be considered. The cable must be rugged yet flexible so
that it can be easily pulled through double containment
enclosures. The formulations must not be so soft that any
accidental loading will cause the conductive polymer formulation to contact the sensor wires and trigger the circuit. Finally,
the conductive polymer layer must be continuous and act as a
water barrier to prevent the sensor from triggering.
Testing was conducted on several thousand feet of sensor
cable with a nominal outside diameter of 0.3 in. The
conductive polymer layer was crosslinked using a high-energy
electron beam. The cable was cut into convenient lengths. An
initial resistance of greater than 30 MQ was measured and

908

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 25, NO. 5, SEPTEMBERIOCTOBER 1989

leak and the instrument can be calculated using Ohms Law (R

= V/Z).Finally, if the resistance per meter of cable is known
(this is controlled by accurate manufacturing tolerances) then
the location in terms of distance is a simple scale conversion.
ltvdrocarbon Sensor Electrical Schematic

Fig. 2. Schematic of four wire circuit in hydrocarbon sensor.

m H i g h lmpedence Voltmeter

-~~

d/conPtant

Current source

Hydrocarbon Sensor Electrical Schematic

Fig. 3. Schematic of leak swelling polymer between two electrodes that

completes loop and current source. It can now generate current flow in
closed circuit.

specified to be equivalent to a sensor in its nontriggered state.

Five cm of the cable was completely immersed in a fluid, and
the time taken for the resistance to fall below 20 k62 was
measured. This time represents the response time of the
sensor since the alarm and locator module register the
presence of a fluid at this resistance value. Response time after
water immersion, bend diameter, and pressure triggering were
also measured. The tests showed that the resistance between
sensor wires remained greater than 30 M62 after water
immersion, a bend radius of less than 5 cm was needed to
trigger the cable, and a minimum trigger load of 20 lb/linear
inch was needed to trigger the alarm.
Fuel sensing cable response times are dependent on the
molecular weight of the target hydrocarbon. Gasoline takes 15
min, while diesel No. 2 and toluene take 60 min. Automatic
transmission fluid takes 4-6 h. This response time is based on
2 in of cable length immersed in the liquid whose temperature
is 20C. Response times are affected by operating temperature.

Electrical Circuitryfor Leak Detection

The hydrocarbon sensor makes use of a four-wire circuit
shown schematically in Fig. 2. The sensing electrodes are
represented by the zigzag lines, while the insulated telemetry
pair are indicated by the solid lines on the top and bottom of
the circuit. The key elements of the leak locating circuitry are
the constant current source and the high-impedance voltmeter.
Also note that the four conductors are joined to form two loops
at the far end of the monitored section.
If the cable has not been swollen by a hydrocarbon leak,
there is no path between the two electrodes and no current
flows from the constant current source. As shown in Fig. 3, a
leak that swells the polymer between the two electrodes
completes the loop and the current source can now generate a
current flow in the closed circuit.
The high impedance voltmeter is now used to measure the
voltage drop V between the leak location and the instrument.
Since the current is known (i.e., a constant current source) and
the voltage is measured, the resistance of the cable between the

Long Pipeline Telemetry

Since the effective range of the hydrocarbon sensor is about
2 km, continuous surveillance of a long pipeline needs some
sort of telemetry to relay leak detection information to a
central location. The type of telemetry equipment chosen is
commonly used in the telephone industry to monitor pressure
transducers attached to pressurized telephone cable. The
variations in air pressure are converted to a variable resistance
by the pressure transducer. The variation in resistance is in
turn used to modulate a signal on the two-wire telemetry
circuit so that the equipment located at the end of the line is
able to determine the air pressure at each of the transducer
locations. The transducers are each assigned an individual
identity at the time of installation. The multiplexing scheme is
based on a time splice technique so that only one of the
many transmitters on a given line is transmitting its measured
value at any given time. The master unit at the head of the line
starts the reporting sequence by a voltage step monitored by all
Units.
For detection of a hydrocarbon leak, the pressure transducer
is replaced by a loop formed from the sensor pair in the
hydrocarbon sensor cable and an end-of-line resistor. If the
cable is in a nontriggered state, then the transducer circuit sees
a predicted resistance value and the signal returned to the
monitoring equipment is within the normal range. If the
cable has been soaked by a hydrocarbon, the resistance seen
by the transducer on the monitoring equipment is shifted
downward in frequency and the monitor recognizes the
leak occurrence. If the sensing loop is broken, the
transducer sees a sudden rise in resistance. The signal returned
to the monitor is shifted up in frequency and the monitor
recognizes an open in the sensor.
When a leak is detected, maintenance personnel must go to
the monitor location and hook up a portable test box (PTB) to
determine the exact location in the 2-km section of leak
detection cable. Three different situations can occur to trigger
an alarm. There can be a continuity fault, a single contamination site on the sensing cable, or multiple contamination sites
on the sensing cable. The PTB can be used to determine which
of the three possibilities are present.

CONCLUSION
The hydrocarbon sensor has shown itself to be a significant
advancement in the technology available for leak detection.
Distributed systems have in the past not been economically
feasible for pipelines and other large systems, leaving volume
balance systems to determine leaks. As discussed, the volume
systems have limitations on low leak rates, which the sensor
cable system does not have. The hydrocarbon sensor can be
used for tank farms, ponds, and double containment piping
systems [13]. The same electronic systems can be used for
water sensor cables and acid and base sensor cables.

SANDBERG et al. APPLICATION OF CONTINUOUS LEAK DETECTION SYSTEM

REFERENCES
J . McNally et al., Containment and removal of fuel from groundwater beneath a densely populated housing development, in 1985 A P /
Oil Spill (Prevention, Behavior, Control Cleanup), Conf. Proc.
Los Angeles, CA, Feb. 25-28, 1985.
F. J. Berto, Computer aids line-leak detection, Oil and Gas
Journal, vol. 71, no. 49, pp. 41-47, Dec. 3, 1973.
R. C. Ewing, Alaskas nikishi product line completed. Oil and Gas
Journal, vol. 74. no. 39, pp. 76-82, Sept. 27, 1976.
L. Billmann et al., Leak detection methods for pipelines, Proc.
Ninth Triennial World Congress of IFAC, Bridge Between Control
Science and Technology, vol. 4: Process Industries, Power Systems,
IFAC Proceedings Series, 1985, no. 4, pp. 1813-1818.
N. E. Flournoy et a/., Development of a pipeline leak detector,
Journal of Canadian Petroleum Technology, vol. 17, no. 3, JulySept. 1978, p. 33-36.
H. S . Lee et al., Gas pipeline leak sensor (gaspils) for airborne use,
International Conf. on Pipeline Inspection, Edmonton, Alberta,
CAN, June 1983, pp. 555-565.
J. C. Eidenshink, Detection of leaks in buried rural water pipelines
using thermal infrared images, Photogrammetric Engineering and
Remote Sensing, vol. 51, no. 5. pp. 561-564, May 1985.
TraceTek Fuel and Solvent Sensing Cables, Raychem Corporation,
Menlo Park, CA, no. HS2858, Aug. 1987.
L. R. G. Treloar, The Physics of Rubber Elasticity, London:
Oxford University Press, p. 123, 1985.
S . D. Gehman, Rubber Chem. Technology, vol. 40, no. 21, p. 632,
1967.
New pipe system stops leaking fluid lines, National Engineer, vol.
91, no. 4, Apr. 1987. p. 15.
R. I. Williams, Fundamentals of pipeline instrumentation, /SA
Trans. Automation, and Supervisory Control, vol. 21, no. 1. pp.
45-54, 1982.
Torstein K. Fannelop, Flow processes and leak rates associated with
broken underwater oil pipelines, Norwegian Maritime Research,
vol. 5 , no. I , pp. 6-13, 1977.
R. E. Nicholas, Leak detection by model compensated volume
balance, 1987 Pipeline Engineering Symposium, Dallas, TX, Feb.
15-18, 1987, pp. 13-20.
Pipes and Pipelines International, New pipeline leak detection
pig, vol. 21, no. 4,pp. 26-28. Aug. 1976.
E. 0. Butts, Detecting leaks in pipelines and storage tanks,
Engineering Journal (Montreal). vol. 60, no. 3, pp. 45-47, May-June
1977.
T . Fukuda et al., Leak detection and localization in a pipeline system
based on time series analysis techniques, Journal of Fluid Control,
vol. 15, no. 4, pp. 5-17, Dec. 1983.
T . Lippitt, Williams pipe leak detection system. 1987 Pipeline
Engineering Symposium Dallas, TX, ASME PD. vol. 6, pp. 35-4 1.
David Atherton, Magnetic detection of pipeline corrosion, Proc.
International Congress on Metallic Corrosion, Toronto. ON, Canada, vol. 4, pp. 151-156.
Thomas Lippitt, Williams system signals advance in leak detection.
Oil and Gas Journal, vol. 8 5 , no. 29, pp. 42-47, July 20, 1987.

909
He is a Registered Professional Electrical Engineer in the State of California.
He also holds five patents and has published over 15 technical papers on fiber
optics, sensor technology, and electric heat tracing.
Mr. Sandberg is a member of ASME and ISA, and is chairman of the IEEE
622 Working Group on Heat Tracing. In addition, he is a ounder of several
small startup companies in the Palo Alto, CA area.

Jim Holmes received the B S. degree in product

design and the M S degree in engneering management from Stanford University, Stanford, CA
He is the Technical Director o i Raychem Corporations TraceTek Products Group He started with
the company in 1977, has workvd in development
and technical positions with Raychems telecommunications products and pipeline corrosion protection
products prior to his current position

Ken McCoy received his B S E.E degree from

Northwestern University, Evanston, IL, and the
M B A from the University of Santa Clara, Santa
Clara. CA, in 1971 and 1976, r,:spectively
After serving with the United States Navy and a
period of consulting work with S R I. International,
he joined Raychem in 1977. He held various
positions with the Petroleum Tc chnology Division
and was involved with a number of corrosion
control products used as pipelire coatings and for
cathodic protection In 1984, he was part of a small
start-up group that formed the TraceTek Division Presently, he IS the
Operations Manager for TraceTek and combines his man1 facturing responsibilities with sales support and marketing in the Pacific Rim countries

Chet Sandberg (M77-SM89) received the B S

degree from the Massachusetts Institute of Technology, Cambridge, MA, and the M S degree from
Stanford University, Stanford, CA, in 1967 and
1972, respectively
In 1972 he joined the Raychem Corporation,
Menlo Park, CA, in the Chemelex Heating Systems
Division He has held various positions at Raychem
including Technical Strategy Manager, Electrical
Engineering Manager of the Chemelex Division and
Consumer Product$ Division Manager He has
managed and participated in research and development of fiber optic sensors,
conductive polymer heating elements, and process instrumentation systems

Heinrich Koppitsch received the Dipl.Eng. degree

from Technical University, Graz, Austria, and the
M.B A degree from Insead, the European Institute
of Business Adrmnistration, Foittainbleau, France.
He then worked at Waagner-Biro in technical
sales of turnkey power plants He joined Raychem
in 1980 and has held various sales and marketing
positions with the Chemelex Division in Germany,
France, and the United State, He is currently
Marketing Manager for the IraceTek Products
Group.