a review of pipeline integrity · pdf filea review of pipeline integrity systems a report for...

A REVIEW OF PIPELINE INTEGRITY SYSTEMS

A Report for

National Measurement System Directorate Department of Trade & Industry 151 Buckingham Palace Road

London, SW1W 9SS

Project No: FEOT16 Report No: 2005/257 November 2005

The work described in this report was carried out under contract to the Department of Trade & Industry (‘the Department’) as part of the National Measurement System’s 2002-2005 Flow Programme. The Department has a free licence to copy, circulate and use the contents of this report within any United Kingdom Government Department, and to issue or copy the contents of the report to a supplier or potential supplier to the United Kingdom Government for a contract for the services of the Crown. For all other use, the prior written consent of TUV NEL Ltd shall be obtained before reproducing all or any part of this report. Applications for permission to publish should be made to: Contracts Manager TUV NEL Ltd Scottish Enterprise Technology Park East Kilbride G75 0QU E-mail: [email protected] Tel: +44 (0) 1355-272096 © TUV NEL Ltd 2005

NEL

Project No: FEOT16 Report No: 2005/257 Page 1 of 26

NEL East Kilbride

Glasgow G75 0QU Tel: 01355 220222 Fax: 01355 272999

A REVIEW OF PIPELINE INTEGRITY SYSTEMS

A Report for

National Measurement System Directorate Department of Trade & Industry 151 Buckingham Palace Road

London, SW1W 9SS

Prepared by: Dr N F Glen

Approved by: Mrs J Sattary Date: 30 November 2005 for Michael Valente Managing Director

NEL


EXECUTIVE SUMMARY The visible evidence of a leak is often the first sign that the integrity of a pipeline has been breached. In an ideal world, one in which cost did not matter, pipelines would be designed to have guaranteed integrity, but in the real world it is inevitable that leaks will occur. Whether carrying water, gas or oil, there are financial, political, regulatory, safety and environmental issues which must be addressed when designing and operating pipelines. Pipeline integrity monitoring and leakage detection systems therefore have a key role to play in minimizing the occurrence of leaks and their impact. The requirement to monitor the integrity of extensive and inaccessible pipelines is clearly widespread and there is a belief that some of these issues can be addressed by sharing experiences and practices across sectoral boundaries. As part of the 2002-2005 Flow Programme, NEL undertook a project to review the state of the art in monitoring systems for pipeline integrity across all relevant industrial sectors. The overall review covered pipeline integrity in general, with a focus on documenting leakage detection methodologies. Although such techniques are used across a range of industries, the different operational and safety requirements have led to a variety of implementations. Simple volume or mass balance approaches are widely used, despite their limitations in terms of the minimum leak size detectable and their inability to handle system transients. Modified approaches using additional sensors, transient modelling or statistical analysis can greatly improve the performance of these systems. Although there is no single ideal approach, several of the currently available technologies can work in a complementary fashion, greatly expanding the range of leaks that can be detected. In terms of the most important criteria (response time, leak size and ability to handle transient conditions), real-time transient modelling and statistical analysis appear to be the most successful but this must be balanced against their data requirements, complexity and cost. It is clear that practices differ from sector to sector and the very different operational regimes of, and potential hazards from, oil, gas and water pipelines, mean that transferring best practice from one sector to another is not simply a matter of implementing an identical solution. The key recommendation with regard to the selection of any leak detection system therefore is that it must be made within the context of an overall pipeline integrity management plan. This will ensure that, as far as possible, the system or systems chosen address the specific requirements of the pipeline and its interactions with its environment. Even within each sector, there is considerable variation in current practice, both in terms of implementation by pipeline operators and the solutions offered by leakage detection system providers. Whilst the decision-tree-based approach can provide an indication of systems that meet the technical criteria, it is important to speak to other pipeline operators, in other sectors if appropriate, but certainly within the sector. Industry bodies such as UKOOA, UKOPA, British Water and Water UK therefore have a role to play in ensuring best practice and it recommended that this report is disseminated through these routes, in addition to the Flow Programme website.

NEL


C O N T E N T S

Page EXECUTIVE SUMMARY .................................................................................... 2 1 INTRODUCTION ................................................................................................ 4 2 BACKGROUND .................................................................................................. 4 3 LEAK DETECTION SYSTEMS........................................................................... 7 3.1 Evaluation of Leakage Detection Systems ......................................................... 7 3.2 Hardware-Based Systems .................................................................................. 11 3.3 Software-Based Systems.................................................................................... 15 3.4 System Selection ................................................................................................ 21 4 CONCLUSIONS AND RECOMMENDATIONS................................................... 24 REFERENCES ................................................................................................... 25

NEL


1 INTRODUCTION The visible evidence of a leak is often the first sign that the integrity of a pipeline has been breached. In an ideal world, one in which cost did not matter, pipelines would be designed to have guaranteed integrity, but in the real world it is inevitable that leaks will occur. Whether carrying water, gas or oil, there are financial, political, regulatory, safety and environmental issues which must be addressed when designing and operating pipelines. Pipeline integrity monitoring and leakage detection systems therefore have a key role to play in minimizing the occurrence of leaks and their impact. However, it is currently believed that in the offshore oil industry the systems used to detect leaks are not sufficiently sensitive for the prompt identification of leaks. On the rare occasions when a pipeline leak has occurred, the early warning detection systems have been ineffective and the first warning of a leak has been when an oil slick, say in the North Sea, is detected visually from the air or from the sea. This delay in detection has huge environmental implications, and the longer the leak goes undetected the greater the environmental impact. As a result, the Environmental Department within the DTI has an increasing requirement for more robust pipeline integrity systems to be implemented in this industry. The requirement to monitor the integrity of extensive and inaccessible pipelines is clearly widespread and there is a belief that some of these issues can be addressed by sharing experiences and practices across sectoral boundaries. As part of the 2002-2005 Flow Programme, NEL undertook a project to review the state of the art in monitoring systems for pipeline integrity across all relevant industrial sectors. NEL made use of its extensive contacts in the water and offshore oil and gas industries to review current integrity monitoring practices in these sectors. To provide additional information about current practice in the onshore gas distribution industry, a subcontract was placed with Advantica; their findings are described in a detailed report(1). The overall review covered pipeline integrity in general, with a focus on documenting leakage detection methodologies. Although such techniques are used across a range of industries, the different operational and safety requirements have led to a variety of implementations. By identifying best practice in each industrial sector and promulgating it across all sectors, the project will help companies to realize the benefits of efficient and cost-effective leakage detection. 2 BACKGROUND Pipeline integrity can be assured by appropriate design, construction and operation; the use of a pipe-in-pipe system with annular-space leak sensing would, for example, significantly reduce or entirely eliminate the possibility of fluid release to the general environment. Whilst this approach can be applied to new pipelines, it is much more difficult to “retrofit” to an existing pipeline to ensure inherent integrity. Most integrity systems are therefore based on specific instrumentation and methodologies to reduce the likelihood of pipeline failure and minimize the consequences of such an event. Pipeline integrity systems can therefore be divided into ‘Before-the-event’ and ‘After-the-event’ systems, as summarized in Table 1 and Figure 1. Before-the-event systems are aimed at ensuring the integrity of a pipeline and use a combination of operational procedures, maintenance procedures, and dedicated hardware and software as part of an overall pipeline integrity management system (PIMS) to provide advance warning of any events or changes in the physical state of the pipeline which may lead to a loss of integrity. After-the-event

NEL


systems are aimed at detecting and locating leaks caused by a loss of integrity and can be based on dedicated sensors or a combination of existing sensors and modelling techniques. Operational considerations and safety requirements differ from sector to sector, with key differences summarized in Table 2. For product value and safety reasons, oil and gas pipelines make more use of before-the-event systems than water pipelines. For instance, gas distribution, oil and chemical pipelines are subject to observational surveys made by air, road vehicle and by foot in order to detect the presence of excavating activities close to the pipeline; this helps prevent failure due to impact. The pipeline may also be inspected in-line using a range of inspection tools to detect the presence of metal loss defects, such as those caused by corrosion and external impact. Pressure cycling can be monitored and controlled to prevent the growth of construction defects due to fatigue. Table 3 provides a summary of a range of in-line inspection and pipeline survey techniques. More detailed information is given in the Advantica report(1). In principle, these survey techniques can be applied to pipelines carrying any fluid but many of the techniques require access over the whole length of the pipeline and thus would not be suitable for offshore applications or buried onshore pipelines. The remainder of this review is therefore concerned with after-the-event, leakage detection, systems.

Figure 1 Pipeline integrity systems

Table 1 Pipeline Integrity Systems

System Before-the-event After-the-event

Aim ensure integrity detect leaks

Approach appropriate design and operation hardware and / or software-based monitoring systems

hardware and / or software-based detection systems

Reduce effects of an

event

Before-the-event system

After-the-event system

Pipeline integritysystems

Avoid an event

NEL


Table 2

Sector Intercomparison

Sector Pipeline pressure Issues

Gas transmission

up to 70 bar High value product

Leakage unacceptable on safety grounds

Oil and petrochemicals

up to 100 bar High value product

Leakage unacceptable on safety and environmental grounds

Water up to 10 bar Low value product Non-hazardous product but gross leakage unacceptable on political grounds

Table 3 In-line Inspection and Pipeline Survey Techniques

Technique Purpose

Axial magnetic flux leakage

Detection of metal loss and certain types of defects

Transverse magnetic flux leakage

Detection of metal loss and certain types of defects including longitudinally oriented seam corrosion

Ultrasonic compression wave

Detection of metal loss and internal/external defect discrimination

Ultrasonic shear wave Crack detection, including defects such as lack of fusion, hook cracks, stress corrosion cracking, and voids, as well as narrow axial corrosion

Direct Current Voltage Gradient

Detection and accurate location of coating defects

Pearson survey Detection of coating defects (including pin-point defects)

Current attenuation techniques e.g. C-Scan, Pipeline Current Mapper

Location of coating defects and provision of comparative assessment of coating quality

Cathodic protection surveys e.g. Close Interval Potential Survey

Assessment of the effectiveness of the cathodic protection system

Guided ultrasonic torsional wave

Detection of metal loss due to corrosion

Skin-effect electric current Detection of metal loss

NEL


3 LEAK DETECTION SYSTEMS Leak detection systems can be categorized based on where the measurements are made(2) or the methods used(3). In terms of where the measurements are made, this classification differentiates between internal measurements, which examine flow in the pipeline in an attempt to infer leakage, and external measurements, which look to detect fluids that have exited the pipe. Classification based on the methods used differentiates between systems that use sensors available in normal pipeline operations (pressure, temperature, flow rate) and those that require special sensors. Systems that use special sensors can be regarded as hardware-based; although software will be involved, it is dedicated to processing the data from these sensors. Systems that use existing sensors can be regarded as software-based since they depend on additional software to process data from different types of sensors and extract extra information from which leaks can be inferred. Table 4 summarizes this classification scheme.

Table 4 Categorization of Leak Detection Systems by Detection Method

Hardware-based Software-based Acoustic emission Mass or volume balance

Cable sensors Real-time transient modelling Fibre-optic sensors Rate of change

Soil monitoring Statistical analysis Vapour monitoring System identification with digital

signal analysis 3.1 Evaluation of Leak Detection Systems The key criteria for assessing any leak detection system are: • What is the minimum leak size that the system is capable of detecting? • What is the time needed to detect a leak of a given size? A leak is detectable only when its effect rises above uncertainties in the variables being monitored. The size of a leak is usually expressed as a percentage of the throughput of the pipeline. Leak size is a function of the size and shape of the opening (leak area) and the pipeline pressure. A leak can be either constant in size, such as a pre-existing small leak, or variable over time, such as a sizable leak that diminishes as the pipeline is depressurized. Depending on the leak detection methodology used, the response time can vary over a wide range. For algorithms based on volumetric or mass balance, the response time is related to the leak size because of the uncertainties in the variables involved. Reducing the uncertainty on the measurements improves the detection threshold(4), as illustrated in Figure 2. For leak detection methods based on discrepancy patterns generated from a real-time transient flow model, the response time is not a function of leak size. Instead, it is a function of the propagation speed of a pressure disturbance and the distance between the leak and the nearest pressure or flow sensors. Depending on the location of the pipeline and the fluid it is carrying, the rate of leakage or the total volume lost over a given period may be more significant. For example, a very slow leak

NEL


from an oil pipeline may go undetected for a long period. Whilst this is unlikely to have a major financial impact on the operation of the pipeline, the environmental impact may be significant. On the other hand, the environmental impact of a long-term, slow leak from a gas pipeline would probably be insignificant but the safety implications huge; in the case of an underground gas main, the gas could remain trapped around the pipe, building to a volume which, if ignited, could cause very serious damage.

0

20

40

60

80

100

Response time

Leak

siz

e as

per

cent

age

of th

roug

hput

Threshold with less uncertainty

Threshold with more uncertainty

Figure 2 Detectable leak size versus response time for mass balance detection

For comparable pipeline conditions, the response time is also, broadly, a function of the detection method used, as illustrated in Figure 3.

0.01

0.1

1

10

Seconds Minutes Hours

Response time

Leak

siz

e as

per

cent

age

of th

roug

hput

Mass balance

Hardware

Pressure loss

Figure 3 Detectable leak size versus response time for various detection methods

NEL


Whilst the key criteria are minimum leak size and response time, there are a number of other criteria that should be considered when assessing a leak detection system(4), as summarized in Table 5.

Table 5 Assessment Criteria

Criterion Description

1 Leak size / leak flow rate

What is the minimum leak size that the system is capable of detecting?

2 Response time What is the time needed to detect a leak of a given size? 3 Location estimation Can the system locate a leak and what is the accuracy of the

location estimate? 4 Release volume

estimation Does the system have the ability to determine the volume of liquid released?

5 Pre-existing leaks Does the system have the ability to detect pre-existing leaks, as well as the onset of a new leak?

6 Shut-in condition Does the system have the ability to detect the onset of a leak in a shut-in pipeline segment?

7 Slack condition Does the system have the ability to detect a leak in pipelines under a slack condition during transients?

8 False alarms What is the rate of false alarms and misses for the system? 9 Sensitivity to flow

conditions How sensitive is the system to operational transients (such as those caused by pump startups or valve swings)?

10 Robustness Will degradation or malfunction of a system component cause catastrophic loss of leak detection ability?

11 System self-checks Does the system have the capability to automatically check and possibly rectify parameters that affect leak detection performance?

12 Complex configurations

Can the system handle complex pipeline configurations (e.g. multiple injection and delivery points) as well as complex operations (e.g. multiple modes of operation, bi-directional operation)?

13 Availability Is the system available full-time or only during steady-state operation?

14 Ease of retrofitting What is required to install a new leak detection system and/or methodology on an existing pipeline?

15 Ease of testing How easy is it to test the system during commissioning and at regular intervals thereafter?

16 Cost What is the cost of the system including capital and operational expenses, as well as, data and equipment requirements? What is the cost of tuning the system to match current operation of the pipeline?

17 Ease of use Is the system easy to use? How much operator training is required?

18 Ease of maintenance

What are the maintenance requirements for the system? Will the system degrade with improper or missed maintenance tasks? How frequently does it need to be tuned?

NEL


The remainder of this section reviews a range of leak detection techniques, summarizing their advantages and limitations. Table 6 provides an overview of the techniques, following the classification scheme given in Table 4.

Table 6 Leak Detection Systems

Hardware-based

Acoustic emission A leak generates noise which can be picked up by acoustic sensors installed outside the pipeline.

Cable sensors These sensors use polymer materials that swell in the presence of hydrocarbon thus changing their electrical properties.

Fibre-optic sensors Leaks can be identified through the identification of temperature changes in the immediate surroundings using fibre-optic cable or through change in the optical property of the cable itself induced by the presence of a leak.

Soil monitoring Leaks are detected by analyzing the concentration of the vapour phase or tracer substances in the soil surrounding the pipeline.

Vapour monitoring If the product inside a pipeline is highly volatile, this system sucks the vapours into a low-density polyethylene (LDPE) sensor tube and runs this gas stream past specialized sensors that can detect trace concentrations of specific hydrocarbon compounds.

Software-based

Mass or volume balance

This method checks for leak by measuring the mass or volume at two sections of the pipeline.

Real-time transient modelling

This method mathematically models the fluid flow within a pipeline. The equations used to model the flow are conservation of mass, conservation of momentum, and an equation of state for the fluid.

Rate of change Rapid depressurization, rapid inflow increase, rapid outflow decrease, and rapid increase in the difference between inflow and outflow are associated with the onset of a leak.

Statistical analysis This method detects a leak by undertaking statistical analysis of pressure and/or flow at multiple locations. Leak alarm generation is based on a set of consistent patterns of relative changes of the mean data at different locations.

System identification with digital signal analysis

This method relies on the occurrence of a leak changing the pipeline-fluid system in some characteristic way but does not use a mathematical model for the transient pipeline hydraulics.

NEL


3.2 Hardware-Based Systems 3.2.1 Acoustic Emission Acoustic emission is based on the principle that the leaking liquids are in turbulent flow and create a detectable acoustic signal. For nearly all subsea oil and gas pipelines the operational flow regimes are highly turbulent and, for leaks from pressurized subsea gas pipelines, the leak will almost certainly act as a noise source. However, this may not be the case for heavy oil lines in a shut-in condition (non-flowing). Acoustic sensors are located on the outside of the pipe to monitor internal pipeline noise. The acoustic sensor is a transducer that converts the sound waves associated with leaks in the pipe to an electrical signal. Leak noise is generally wideband, ranging from about 1MHz to below 1kHz. The acoustical sensor is the most important component of detection and must have sufficient sensitivity and low intrinsic noise. Typical sensors operate in the range from 400kHz to 10MHz. Once the acoustical sensors are attached to the pipeline, a baseline acoustic map of the pipeline is developed that serves as a reference. Deviations from the acoustic profile result in an alarm. The acoustic signals can also be used to determine the location of the leak.

Figure 4 Acoustic emission leak detector (from Reference 5)

Several case studies have been performed by Physical Acoustics Corporation in both Russia and the United States(6). These have demonstrated that acoustic emission is able to detect and locate leaks in buried pipelines. However, although the technique offers very good performance with respect to the minimum leak detectable, there are a number of practical factors that limit its applicability. Whilst it is possible to achieve leak detection at the level of 10-6 m3s-1 for liquid and 10-5 m3s-1 for gas, this limits the maximum span between sensors to a few hundred metres. Corrosion protection coatings for pipelines can also attenuate the noise produced by a leak. Finally, for high flow rates, the background noise will mask the sound of a leak. 3.2.2 Cable Sensors Cable sensors can detect a leak either by completion of an electrical circuit (electrochemical sensing) or by measuring changes in cable impedance (time-domain reflectometry). Electrochemical sensing wire cables all function on the same general principle. The cable contains at least two circuit loops; a continuity circuit and an alarm circuit (Figure 5a). In normal operation the continuity circuit provides confirmation that the cable has not been

NEL


physically damaged; the alarm loop is open circuit. A leak is detected when the alarm loop is completed (Figure 5b), which can be caused by several mechanisms, depending on the cables used. For example, the short can occur when a leak of conductive fluid facilitates current flow between the cables. Shorting can also occur by direct wire contact; this can result when the material separating the wires degrades, allowing the wires to touch or when an outer coating of the cables swells when brought into contact with the leak, forcing the two wires together to complete the alarm circuit. By monitoring the voltage drops in the circuits it is possible to pinpoint the position of the leak. Conductive fluid cables can generally be dried and reused, while the swelling and degradation types must be replaced.

Alarm Alarm

continuity circuit continuity circuit

alarm circuit alarm circuit

a - no leak b - with leak

leak-inducedshort circuit

Figure 5 Electrochemical sensing wire principle Time-domain reflectometry measures an electromagnetic pulse sent down a coaxial cable to detect an impedance change or discontinuity. Pulse reflections, or echoes, are generated which are specific to the actual installation of the sensor cable. The echoes are processed and stored by a microprocessor to create a baseline reference map. In the event of a leak, the hydrocarbons penetrate the cable and alter the impedance of the cable at the leak site. The change in impedance alters the echoes returning to the microprocessor and triggers an alarm. The change in signal is used to detect the location of the impedance change and thus the leak location. The advantage of this type of system is that, once a leak occurs, the reference map can be updated and the system can continue to be used to detect leaks.

Figure 6 Examples of electrochemical sensing wires (from Reference 7) and a time-domain reflectometry detection cable (from Reference 4)

3.2.3 Fibre-Optic Sensors Fibre-optic cables can be used to detect leaks using a range of sensing techniques. The optical characteristics of fibres alter with temperature changes, mechanical stress and surface coating / absorption of chemicals. The resultant change in refractive index of the fibre can be detected either by measuring total signal attenuation or using optical time-domain techniques (reflectrometry (OTDR) or transmission (OTDT)). Fibre-optic sensors can be installed as point sensors and as distributed sensors. Point sensors may either be discrete or clustered. A discrete sensor system uses a single fibre-optic sensor at a single location, whereas a clustered sensor system uses multiple

NEL


individual sensors tied to a single detector system. For pipeline applications however it is more common to use either discrete distributed or continuous distributed sensor systems. Discrete distributed systems make use of optical fibre sensors placed at intervals along an optical fibre whereas continuous distributed systems make use of the fibre itself as both the sensing and communication element along its entire length. OTDR has been applied to detecting leaks from gas pipelines using distributed temperature sensing (DTS). High-pressure gas escaping from a leak site causes localized cooling as a result of the Joule-Thomson (adiabatic expansion) effect. By deploying a fibre-optic cable along the length of the pipeline, a localized cold spot detected by the DTS system indicates a gas leak. The DTS provides a temperature profile of the entire pipeline and is thus capable of identifying, not only the existence but also the location of a leak. This technology requires the pipeline to be buried or enclosed in a containment wrapper. Table 7 shows the characteristics of DTS systems in terms of location and time to detect a leak.

Table 7 Typical Fibre-Optic Temperature Sensor Performance

Range metres

Resolution metres

Measurement time seconds

Temperature resolutionK

500 1000 4000 10000 20000

1 1 1 1 10

5 5 10 60 60

1 1 1 2.5 2.5

As an alternative to DTS, leaks can be detected by their mechanical effects on optical fibres. In this method, when a leak occurs, optical fibres develop micro bends in the presence of hydrocarbons. This can be detected and located with an OTDR system. Detection of a leak by chemical effects is based on the use of an optical fibre core surrounded by a coating or cladding that is reactive to hydrocarbons. When the coating or cladding contacts hydrocarbon, the refractive index is altered and affects the transmission of light through the optical fibre. In a typical system, a light-emitting diode transmits light through a chemically coated optical fibre cable. When the cable comes in contact with hydrocarbons, the chemical coating is altered and allows some of the light to escape. A reference detector used in conjunction with a sensor at the other end of the cable measures the loss of light. The loss of light and the inferred change in refractive index are used to estimate the concentration of hydrocarbons and, depending on the optical detection system, the location of the leak. 3.2.4 Soil Monitoring Soil gas detection systems rely on the movement of volatile chemicals through the soil surrounding the pipeline and into a collection system. The collection system can be passive or active. Passive systems rely on diffusion of the chemicals into the detector or sample collection container; active systems assist the transport process using vacuum. If a collection container is used, it must be removed periodically and its contents analysed (typically using gas chromatography / mass spectroscopy (gc/ms)); such systems would be of limited use for pipeline applications.

NEL


A variant of this method involves inoculating pipelines with a unique, nontoxic, and highly volatile “tracer” compound. The tracer compound is added at a concentration of a few parts per million to pipeline contents and has no measurable impact on their physical properties but is easily detectable in the soil surrounding a leak site. 3.2.5 Vapour Monitoring This system detects leak by placing a sensor tube parallel to the pipeline. In the event of leak, the hydrocarbon vapours will diffuse into the sensor tube. The sensor tube is periodically pumped to the base station where the air in the tube passes through a hydrocarbon detector. This system has been extensively developed by Siemens (LEOS) and detects leaks by means of a low-density polyethylene (LDPE) tube which is highly permeable to the substances to be detected in the particular application. The LDPE tube fits around a perforated stiff core to provide strength (Figure 8). The LDPE tube is pressure tight at installation so air can only enter at the opening of the tube during purging. During a purge cycle, a pump pulls the contents of the tube through a detector that provides total concentrations and displays them in a format similar to a gas chromatogram. Before a purge cycle begins, an electrolytic cell injects a specific volume of test gas into the end of the tube. This gas acts as a marker and its appearance at the detector indicates that the entire tube has been purged. Based on the ratio of the travel time of the leak peak to the marker peak, the leak location can be calculated.

Figure 8 The LEOS vapour monitoring system (from Reference 8)

3.2.6 Summary of Hardware-Based Systems There are many factors that affect the performance of external leak detection methods and these should be considered as part of the selection process. Table 8 provides a comparison of the systems discussed previously, in terms of key criteria given in Table 5.

NEL


With the exception of the acoustic emission-based approach, most hardware-based systems are relatively difficult to retrofit even for on-shore pipelines, as they generally require access over most of the length of the pipeline; this would clearly limit their potential application to off-shore pipelines. However, developments in cable-based systems, in particular fibre-optic systems, may make retrofitting easier. Although the response times given in Table 8 range down to seconds, in practice many of these systems require much longer when detecting very small (less than 0.1%) leaks. However, as they are capable of detecting existing leaks and dealing with pipelines in shut-in or slack conditions, they are potentially very useful. Their use should therefore be considered as part of an overall pipeline integrity strategy when even low-level leaks are unacceptable. 3.3 Software-Based Systems 3.3.1 Mass or Volume Balance Mass or volume balance relies on the principle of conservation of mass. For each pipe section the mass of fluid entering the section either remains in the pipe section or leaves it. A leak is identified when less fluid leaves the pipeline section than is expected from the measurements of the input flow and estimates of the pipeline section contents. At its most basic, the technique is implemented by measuring the volume of products entering and leaving a pipeline section over a specified time period and expressing the results in terms of standardized volumes (volume at 60°F or 15°C and 0 psig). Although simple, this method gives credible results when the flow in the pipeline is at, or close to, a steady state (i.e. the pressure, temperature, and flow along the pipeline do not change rapidly over time), or when the time period is sufficiently long. The leak threshold depends on the accuracy of the volume measurements, the length of the time period, the pipeline volume, and the state of flow in the pipeline.

Table 8 Comparison of Hardware-Based Technologies

SystemCriterion

Acoustic emission

Electro-chemical

cable

Fibre-optic cable

Soil monitoring

Vapour monitoring

2 Response time real time seconds to minutes

seconds to minutes

minutes to hours minutes

3 Location estimate

4 Released volume estimate limited limited estimate estimate

5 Existing leak 6 Shut-in condition

7 Slack condition

8 False alarms frequent less frequent

less frequent

less frequent

less frequent

14 Ease of retrofit moderate difficult difficult difficult difficult

NEL


This approach is more effective for pipelines with a smaller volume (since the linepack is less affected by the state of flow) but obviously it cannot detect a leak in a shut-in pipeline since both the inflow and outflow at the ends of a pipeline segment are zero at all times and yield no useful information. The ability to detect a leak in a slack pipeline (one where low pressures cause localized vaporization) depends on the state of flow. If the flow is at a steady state, a leak can be detected by this method. However, when the flow is in a transient state, the vapour volume changes appreciably and cannot be modelled with accuracy. Consequently, the ability to detect even a significant leak is greatly diminished. The ability of this approach to detect a small leak is highly dependent on the combined nonrepeatability of flow meters at the ends of a pipeline segment. Whilst the simple volume-based approach can give reasonable results under certain conditions, it must be borne in mind that the physical basis of the technique is conservation of mass. However, most flowmeters are volumetric devices and so, strictly, it is necessary to know the density of the fluid throughout the section of the pipeline being monitored. For gas, the temperature and pressure in the pipeline will have a significant effect on the density. Furthermore, the composition of the gas (and hence the effect of temperature and pressure on its density) between two flowmeters within any section of pipe can change significantly over comparatively short periods of time, as gas from alternative sources is introduced into the distribution system. Although the effects of temperature and pressure on the density of oils are lower than for gases, they must still be taken into account. On oil pipelines, flowmeters for fiscal and custody transfer applications generally have an uncertainty of 0.25% (see Table 9) and, depending on the operational regime of the pipeline, it may be possible to resolve imbalances (and hence leaks) of the order of 0.1%.

Table 9 Flowmeter Accuracy

Metering type Accuracy (mass) Comments

Fiscal ±0.25 – 0.5% Single phase, well known conditions

Wet gas (>90% by volume gas phase)

±0.4 – 1.5% When test separation is regularly used to establish gas and liquid phase fractions

±3 – 5% for gas

±10 – 20% for liquids

With no test separation

Multiphase metering (<90% by volume gas)

±5 – 10% for each phase Accuracy deteriorates markedly if one phase is less than 5% by volume of total flow

Inference metering ±5 – 10% for gas

±20% for liquids

The limitations of the basic volume balance can be overcome by the use of additional sensors. By placing additional pressure and temperature sensors along the pipeline, real-time pressure and temperature can be measured at a set of selected locations along the pipeline. The change in the standardized volume over the line balancing period can be estimated using volume correction factors for pressure and for temperature, rather than using an assumed linepack. The accuracy of this correction improves as the spacing between adjacent sensors is reduced(9).

NEL


An extension of this approach is to use a valid transient flow model to compute changes in linepack from measured pressure and flow at the ends of a pipeline segment. When appropriate, the effects of temperature can be included in the model. Although this approach is more complicated and the data requirements can be significant, it may be the only choice when it is not feasible to obtain pressure and temperature data at a sufficient number of the interior points of a pipeline segment. For either of these extensions to the basic approach, the pressure-temperature-specific volume relationship of the liquid must be known, to enable linepack changes to be calculated from pressure and temperature. Hence, the type of liquid in the pipeline needs to be identified, usually by its specific gravity or degree-API. For products pipelines, the position and the specific gravity for each product needs to be tracked. The accuracy of batch tracking may be verified or enhanced by densitometers along the pipeline. For multiphase systems, the uncertainty in flowrate measurement for each phase is such that leaks above about 10% of the pipeline throughput can be detected(1). Figure 9 shows the results of a simulation of a multiphase pipeline with wet gas metering offshore and fiscal metering onshore. A leak of 10% of the pipeline throughput can be clearly identified (Figure 9a) but the flowrate difference seen for a 2.5% of throughput leak is far smaller than the possible meter errors (Figure 9b).

Figure 9 Simulation of the effect of a leak on flowrates in a multiphase pipeline

3.3.2 Rate of Change Rapid depressurization, rapid inflow increase, rapid outflow decrease, and rapid increase in the difference between inflow and outflow are associated with the onset of a leak. In principle, each of these criterion, or several in combination, can be used for leak detection. However, since pipeline operation transients can also cause rapid changes, alarms need to be inhibited for a time period following an operation, such as a pump start-up or a change in the set point of a control valve, limiting the limits the usefulness of this method. In addition, this approach is effective for large leaks only. 3.3.3 Real-Time Transient Modelling A volume or mass flow-based model with additional sensors pressure and temperature sensors along the pipeline can be extended by the use of a transient flow model or a simulation model. In this approach, a subset of the measured pressure and flow data recorded by the SCADA system is used to drive a simulation model. The model results are then compared with the remaining measured data. Since the measured data are affected by leaks while the model assumes the pipeline to be intact, leak-specific discrepancy patterns between the measured

NEL


and the calculated parameters will develop. These discrepancy patterns provide the basis for leak detection, leak location, and release volume estimation. A model simulates transient flows in the pipeline based on the conservation principles of mass and momentum. An energy equation is not involved when the liquid temperature along the pipeline is known. When the temperature is not known, a separate temperature model based on the energy equation may be necessary. For products pipelines, a separate batch tracking model can be used to provide the transient flow model with valid batch positions. The advantage of this approach is that a leak occurring during all flow conditions (including operational transients) can potentially be detected. Figure 10 shows laboratory verification of one version of this approach(10). The measured head and flow at the inlet of a pipeline were used to drive a transient flow model that computed the head and flow at the outlet. At the same time, the measured head and flow at the outlet were used to compute the head and the flow at the inlet. A 6.5% leak was imposed while the pipeline was experiencing transients caused by a sudden 37% flow reduction due to a partial valve closure at the outlet. For this example, this gave rise to: • immediate and simultaneous increases in the discrepancy (measured minus calculated)

in inlet head and inlet flow, and • an immediate increase in the discrepancy of the outlet head and an immediate and

simultaneous decrease in the discrepancy of the outlet flow. Depending on how the transient model is driven by the measured data, the discrepancy patterns may vary but for each case they are specific to a leak and false alarms are claimed to be rare(11). In addition, the difference in the timing of the sudden changes can be used to indicate the location of the leak(12).

Figure 10 Effect of a leak in a system with real-time transient modelling This approach is data intensive and the SCADA system’s data scan rate needs to be fast. Thus, although this approach is classified as software-based, it requires extensive hardware. The model parameters must also be tuned regularly, to ensure that the model accurately reflects reality, thus requiring higher maintenance than for a simple mass or volume balance model approach.

NEL


3.3.4 Statistical Analysis In the simplest form of this approach, statistical analysis is performed on a measured pressure to discern a decrease in the mean value over a threshold. However, simple statistical methods are prone to false alarms; to reduce the frequency of false alarms, more sophisticated statistical analysis methods use pressure and/or flow at multiple locations. Leak alarm generation is based on a set of consistent patterns of relative changes of the mean data at different locations. For example, a leak alarm is generated only if the mean inlet pressure drops and the mean inlet flow exceeds the mean outlet flow. Statistical methods rely on trends consistent with the principle of mass conservation for corroborating mean data values at multiple locations. In this sense statistical methods are physically-based but they do not use a mathematical model for the transient hydraulics in the pipeline to compute pressure and flow. Consequently, the data requirement is not as demanding as for model-based approaches. The basic principle used for the probability calculations is mass conservation and hypothesis testing: leak against no-leak. Although the flow and pressure measurements in a pipeline fluctuate due to operational changes, statistically the total mass entering and leaving a network must be balanced by the inventory variation inside the network. Such a balance cannot be maintained if a leak occurs in a network. The deviation from the established balance is then detected by an optimal statistical test method(13). Leak thresholds are established only after a prolonged period of tuning to establish the underlying probabilistic distribution, the mean, and the variance of the parameter(s) to be tested under different states of no-leak flow (i.e. steady, drifting, or transients). The tuning process is necessary to reduce the occurrence of false alarms. 3.3.5 System Identification with Digital Signal Analysis System identification with digital signal analysis relies on the occurrence of a leak changing the pipeline-fluid system in some characteristic way. For example, a leak will alter the impulse response of a pipeline. This response can be extracted by digital signal processing techniques in real time. An alarm can then be generated when the impulse response changes in a leak-specific way(14). This approach, like the statistical methods, does not use a mathematical model for the transient pipeline hydraulics. Dealing with data noise and extracting information from noisy data is the main focus of this approach and it can, in principle, be used with other software-based techniques. 3.3.6 Summary of Software-Based Systems For each of the categories listed previously, the implementation of the various algorithms can vary considerably. As a result, the performance of a particular method may be significantly different from another one in the same category. Furthermore, the boundary between categories can be blurred by hybrid approaches. For example, statistical analysis can be applied to volume balance, with pressure sensor-based linepack correction. Table 10 provides a summary of the categories in terms of key criteria given in Table 5, using the basic volume balance approach as the basis for the comparison. More detailed information on the design, implementation, testing and operation of software-based systems are covered in the relevant API document(15). From Table 10 it can be seen that none of the methods is perfect. However, unlike the hardware-based approaches, most of the software-based systems are relatively easy to retrofit and in general do not require access to the pipeline over its full length, thus making

NEL


Bas

ic v

olum

e ba

lanc

e

Volu

me

bala

nce

with

pre

ssur

e se

nsor

s

Volu

me

bala

nce

with

dyn

amic

m

odel

Rea

l-tim

e tr

ansi

ent

mod

ellin

gR

ate

of c

hang

eSt

atis

tical

ana

lysi

sSy

stem

ID w

ith

digi

tal s

igna

l an

alys

is

1, 2

Leak

siz

e / r

espo

nse

time

1-5%

/m

inut

es to

hou

rs1%

/m

inut

es1%

/m

inut

es1%

/se

cond

s5%

/m

inut

es1%

/se

cond

s to

min

utes

1% /

seco

nds

to m

inut

es

3Lo

catio

n es

timat

ion

4R

elea

sed

volu

me

roug

h es

timat

ion

for

long

-term

ave

rage

impr

oved

est

imat

ion

for l

ong-

term

av

erag

e

impr

oved

est

imat

ion

for l

ong-

term

av

erag

e

can

prov

ide

leak

flo

wra

te a

s a

func

tion

of ti

me

can

estim

ate

the

volu

me

for l

arge

le

aks

5Ex

istin

g le

aks

6Sh

ut-in

con

ditio

n7

Slac

k co

nditi

on8

Fals

e al

arm

sfre

quen

tle

ss fr

eque

ntle

ss fr

eque

ntm

ore

frequ

ent

frequ

ent

less

freq

uent

less

freq

uent

9Sy

stem

tran

sien

tsno

tole

ranc

ebe

tter t

oler

ance

bette

r tol

eran

cebe

st to

lera

nce

som

e to

lera

nce

best

tole

ranc

ebe

st to

lera

nce

10R

obus

tnes

sav

erag

eav

erag

epo

orer

poor

erpo

orer

bette

rbe

tter

9, 1

2,

13Av

aila

bilit

ypa

rt tim

efu

ll tim

efu

ll tim

efu

ll tim

epa

rt tim

efu

ll tim

efu

ll tim

e

14Ea

se o

f ret

rofit

easy

not e

asy

easy

easy

easy

easy

not e

asy

12C

ompl

ex c

onfig

urat

ion

adap

tabl

ead

apta

ble

less

ada

ptab

lele

ss a

dapt

able

not a

dapt

able

easi

ly a

dapt

able

easi

ly a

dapt

able

14, 1

7Si

mpl

icity

sim

ple

mor

e co

mpl

exm

ore

com

plex

mos

t com

plex

com

plex

com

plex

com

plex

15Ea

se o

f tes

ting

easy

easy

diffi

cult

mor

e di

fficu

ltm

ore

diffi

cult

diffi

cult

*di

fficu

lt17

Ease

of t

rain

ing

easy

easy

diffi

cult

mor

e di

fficu

ltdi

fficu

ltea

syea

sy18

Ease

of m

aint

enan

ceea

sydi

fficu

ltdi

fficu

ltdi

fficu

ltea

syea

syea

sy16

Cos

tm

oder

ate

high

erhi

gher

high

est

high

erhi

gher

high

er

*

Syst

emC

riter

ion

(Tab

le 5

)

Tabl

e 6

Com

paris

on o

f Com

puta

tiona

l Pip

elin

e M

onito

ring

Leak

Det

ectio

n M

etho

ds

Rec

ent w

ork

has

allo

wed

the

inje

ctio

n of

"sta

ndar

d" d

ata

to s

tatis

tical

sys

tem

s, s

impl

ifyin

g te

stin

g

NEL


them suitable for off-shore applications. In practice, to obtain the best performance in terms of response time and leak size detection, it may be necessary to upgrade existing sensors (flowmeters, pressure transducers etc) or add new ones (additional pressure transducers, temperature sensors, densitometers etc). In terms of the most important criteria (response time, leak size and ability to handle transient conditions), real-time transient modelling and statistical analysis appear to be the most successful but this must be balanced against their data requirements, complexity and cost. 3.4 System Selection The selection of one or more leak detection systems must be made within the context of an overall pipeline integrity management plan(1). For example, for on-shore pipeline systems constructed with ferrous materials and transporting gas, the ASME B31.8S standard(16) is specifically designed to provide the pipeline operator with the information necessary to develop and implement an effective integrity management programme utilizing proven industry practices and processes. In this case the pipeline system means all parts of physical facilities through which gas is transported, including pipe, valves, fittings and other attachements, compressor units, metering stations, regulator stations, delivery stations, holders and fabricated assemblies. Whilst this standard applies specifically to gas pipelines, the principles and processes embodied in its integrity management approach are applicable to all pipeline systems. A pipeline integrity management plan should contain full details of the pipeline, including associated facilities and their operations. The plan should contain (or reference) a full description of the technical methods and analyses of the threats to the integrity of the pipeline and the risks to the surrounding population and environment. Details of existing and any proposed new prevention, detection and mitigation practices should also be included, along with a justification for any scheduling applied, demonstrating that pipeline segments with the highest risk are prioritized accordingly. The plan should also address the responses to information collected during assessments, mitigation activities and other integrity-related activities. This will ensure that the plan is dynamic rather than static and so can be updated periodically to reflect any new information that may affect the pipeline integrity systems. This includes damage incidents or pipeline leaks, improved understanding of integrity threats, any changes in the operation of the pipelines or changes in the environment of the pipeline such as new mining/quarrying activity in the vicinity of the pipelines or changes to population density that may result in new pipeline segments covered by regulation. At the detail level of selecting one or more leak detection systems, the first stage of the process is to address the questions given in Table 11. The answers to these questions will help to define the weighting to apply to the criteria listed in Table 5 and allow a decision-tree approach to be used for selecting the best available technique. Figure 11 illustrates this approach for single-phase sub-sea applications. The depth of the tree can be extended to address more of the criteria in Table 5.

NEL


Table 11

System Selection Issues

1 What are the expected operational features of the pipeline?

• steady state / frequent transients• single product / batch mode • single phase / multi-phase • shut in (no-flow) conditions • slack (localized flashing)

conditions • variable pressure operation

2 What are the regulatory requirements? • safety case • environmental impact

assessment

3 What risks apply in the case of a leak? • explosion / fire - release of explosive or flammable product

• poisoning - rapid release of poisonous (to man / other species) substance

• poisoning - slow release of bio-accumulating toxins

• environmental damage • structural damage – rapid

release of large volume of fluid

4 What is the role of leak detection in minimizing the consequences of a leak?

• detection of low-level leaks • rapid detection and location of

large-scale leaks • estimation of release volumes

5 Where is metering located and what type is it?

• basic meter / smart meter • process / fiscal accuracy

6 What other instrumentation is available?

• temperature sensors • pressure transducers • densitometers • composition analysis

7 What will be the responsibilities of the operator / user?

• operator training • system optimization • system maintenance / testing

NEL


Can

it b

e us

ed

subs

ea?

Can

it b

e re

trofit

ted?

D

oes i

t int

erfe

re

with

cat

hodi

c pr

otec

tion

syst

ems?

Tran

sien

t op

erat

ion

Is lo

catio

n es

timat

ion

requ

ired?

Stea

dy-s

tate

op

erat

ion

Is re

leas

e vo

lum

e es

timat

ion

requ

ired?

Is d

etec

tion

of e

xist

ing

leak

s re

quire

d?

Syst

em ty

pe

1 ac

oust

ic e

mis

sion

2

cabl

e se

nsor

s 3

fib

re-o

ptic

sens

ors

4 so

il m

onito

ring

5 va

pour

mon

itorin

g 6

basi

c vo

lum

e ba

lanc

e 7

volu

me

bala

nce

with

pre

ssur

e se

nsor

s 8

volu

me

bala

nce

with

dyn

amic

mod

el

9 re

al-ti

me

trans

ient

mod

ellin

g 10

ra

te o

f cha

nge

11

stat

istic

al a

naly

sis

1

2, 3

, 4, 5

9 10

9 - 12

7 8 7 8

7 - 10

7 - 12

9 10

9 - 12

6 8 6 8

6 - 10

6 - 12

Figu

re 1

1 B

est a

vaila

ble

tech

nolo

gy fo

r lea

k de

tect

ion

in su

bsea

app

licat

ions

NEL


5 CONCLUSIONS AND RECOMMENDATIONS The overall review covered pipeline integrity in general, with a focus on documenting leakage detection methodologies. Although such techniques are used across a range of industries, the different operational and safety requirements have lead to a variety of implementations. For gas, oil and chemical pipelines, the hazardous nature of the fluids being transported and the potential consequences from leaks place the emphasis on prevention rather than detection. Systems aimed at ensuring the integrity of a pipeline use a combination of operational procedures, maintenance procedures, and dedicated hardware and software as part of an overall pipeline integrity management system (PIMS) to provide advance warning of any events or changes in the physical state of the pipeline which may lead to a loss of integrity. However, it is inevitable that leaks will occur. Leakage detection therefore forms a key part of the total strategy and the selection of one or more leak detection systems must be made within the context of an overall PIMS. Simple volume or mass balance approaches are widely used, despite their limitations in terms of the minimum leak size detectable and their inability to handle system transients. Modified approaches using additional sensors, transient modelling or statistical analysis can greatly improve the performance of these systems. Although there is no single ideal approach, several of the currently available technologies can work in a complementary fashion, greatly expanding the range of leaks that can be detected. In terms of the most important criteria (response time, leak size and ability to handle transient conditions), real-time transient modelling and statistical analysis appear to be the most successful but this must be balanced against their data requirements, complexity and cost. Although the safety and environmental implications of water leaks are much less severe than gas or oil leaks, there are still significant financial, political and regulatory pressures which should drive water companies to adopt good pipeline integrity and leakage detection systems. In principle, simple volume or mass balance approaches should be much easier for water; the composition is essentially constant and its variation of density with temperature and pressure is small across the normal range of operating conditions. However, the complex nature of water distribution (networks rather than simple point-to-point pipelines), typical meter uncertainties (between 1 and 5 per cent) and existing leaks make the task far from trivial. In recognition of this, many water companies are now adopting more advanced techniques. At the detail level of selecting one or more leak detection systems for any pipeline, the first stage of the process is to address a series of questions that define the operation of the pipeline and its interaction with the environment. The answers to these questions define the weighting to apply to the criteria that characterize the performance of each system and allow a decision-tree approach to be used for selecting the best available technique. It is clear that practices differ from sector to sector and the very different operational regimes of, and potential hazards from, oil, gas and water pipelines, mean that transferring best practice from one sector to another is not simply a matter of implementing an identical solution. The key recommendation with regard to the selection of any leak detection system therefore is that it must be made within the context of an overall pipeline integrity management plan(1,16). This will ensure that, as far as possible, the system or systems chosen address the specific requirements of the pipeline and its interactions with its environment.

NEL


Even within each sector, there is considerable variation in best practice, both in terms of implementation by pipeline operators and the solutions offered by leakage detection system providers. Whilst the decision-tree-based approach can provide an indication of systems that meet the technical criteria, it is important to speak to other pipeline operators, in other sectors if appropriate, but certainly within the sector. Industry bodies such as UKOOA, UKOPA, British Water and Water UK therefore have a role to play in ensuring best practice and it recommended that this report is disseminated through these routes, in addition to the Flow Programme website.

References 1. Rowe, S., Hodgkinson, J., Williams, A. and Hicks, R. “Review of Gas Pipeline

Integrity Systems”. Advantica Report 6843, prepared for TUV NEL under contract FEOT16, March 2004.

2. Stafford, M. and Williams, N. “Pipeline Leak Detection Study”. HSE Report OTH 94

431, 1996. 3. Scott, S.L. and Barrufet, M.A. “Worldwide Assessment of Industry Leak Detection

Capabilities for Single and Multiphase Pipelines”. US Minerals Management Service, Project Report for Research Agreement 1435-01-9-CA-3103, August 2003.

4. Liou, J. C. P., Hall, R. J. and McMahon, M. C. “Hazardous Liquid Leak Detection

Techniques and Processes”. Report No. DTRS56-02-D-70037-01, US Department of Transportation, Research and Special Programs Administration, Office of Pipeline Safety, April 2003.

5. Lefave, J.P. and Karr, L. “Underground Pipeline Leak Detection and Location

Technology Application Guide, User Guide”. US-2028-ENV, Naval Facilities Engineering Center, Port Hueneme, California.

6. Vahaviolos, S.J., Miller, R.K., Watts, D.J., Shemyakin, V.V.and Strizkov, S.A.

“Detection and Location of Cracks and Leaks in Buried Pipelines using Acoustic Emission: Progress in Acoustic Emission X”. Tokyo, Japan (September 11-14, 2000).

7. “Survey of Technologies for Monitoring Containment Liners and Covers”. EPA 542-

R-04-013, June 2004. National Service Center for Environmental Publications (NSCEP), P.O. Box 42419, Cincinnati, OH 45242-0419.

8. Siemens. LEOS Sales Brochure. http://www.de.framatome-

anp.com/anp/e/foa/anp/products/a-z/leos.htm. 9. Liou, C. P. “Leak Detection by Mass Balance Effective for Norman Wells Line”. Oil

and Gas Journal, 94 (17), pp. 69-74, 1996. 10. Liou, C.P. “Pipeline Leak Detection and Location”. Proceedings of the International

Conference on Pipeline Design and Installation, pp.255-269, American Society of Civil Engineers, Las Vegas, Nevada, March 1990.

11. Liou, C. P., and Tian, J. “Leak Detection - A Transient Flow Simulation Approach”.

Journal of Energy Source Technology, 117 (3), pp. 243-248. American Society of Mechanical Engineers, 1995.

NEL


12. Liou, C.P. “Pipeline Variable Uncertainties and Their Effects on Leak Detectability,” API Publication 1149, American Petroleum Institute, Washington, D.C., November 1993.

13. Beushausen, R., Tornow, S., Borchers, H., Murphy, K. and Zhang, J. “Transient Leak

Detection in Crude Oil Pipelines”. J. Proceedings of IPC 2004, International Pipeline Conference, October 4 - 8, 2004 Calgary, Alberta, Canada.

14. Liou, C. P. “Pipeline Leak Detection by Impulse Response Extraction”. Journal of

Fluids Engineering, 120, (4), pp. 833-838. American Society of Mechanical Engineers, 1998.

15. API 1130. “Computational Pipeline Monitoring for Liquids Pipelines”, 2nd Edition.

American Petroleum Institute, 01-Nov-2002. 16. ASME B31.8S-2004 Managing System Integrity of Gas Pipelines American Society of

Mechanical Engineers 14-Jan-2005

a review of pipeline integrity · pdf filea review of pipeline integrity systems a report for...

Documents