Leak Detection

Published on May 2017 | Categories: Documents | Downloads: 69 | Comments: 0 | Views: 815
of 48
Download PDF   Embed   Report

Comments

Content

 

 

Principles of Leak Detection Prof. Dr.-Ing. Gerhard Geiger

 

 

Introduction

Fundamentals of Leak Detection

KROHNE Oil & Gas

2

 

 

Introduction

Contents



Introduction



2

Regulatory Framework

8

2.1  2.1.1  2.1.2  2.1.3  2.1.4 

TRFL (Germany) Installations according to TRFL a) and b) Installations according to TRFL c) Installations according to TRFL d) Installations according to TRFL e)

8  8  8  8  9 

2.2 

 API 1130 (USA)



2.3 

 API 1155 (USA)





Pressure/Flow Monitor ing

10 

3.1 

Pressure Pre ssure Monitoring

10 

3.2 

Flow Monitoring

10 

3.3 

Summary

10 



Negativ e Pressu re Wave

12 

4.1 

Summary

12 



Balancing Methods

14 

5.1  5.1.1  5.1.2 

Mass Balance Mass Uncompensated mass balance Compensated mass balance

14  14  15 

5.2 

Use of volumetric flow mete meters rs

15 

5.3 

Summary

17 



Statistic al Leak Detection Systems

19 

6.1 

Probabil ity Ratio Test

19 

6.2 

Sequential Probabili ty Ratio Test (SP (SPRT) RT)

20 

6.3 

Summary

20 



Leak Locati on

22 

7.1 

Gradient Intersection Method

22 

7.2 

Wave Propagatio n Method

22 



RTTM – Real Time Transi ent Model

23 

8.1 

Compensation Compe nsation Approach

25 

8.2 

Head Hea d Stations Residual or Differenti al Appro ach

26 

8.3 

Substations Residual Residual or Differe Differential ntial Approach

27 

8.4 

Flow Calculation

27 



PipePatrol PipePatrol Statistic al Ma Mass ss Balance (SM (SMB) B)

28 

9.1 

Summary

28 

10

PipePatro l Extend ed Real-Time Transi ent Model (E-R (E-RTTM TTM))

31

10.1 

Leak Signatur e Analysis

31 

 

 

Fundamentals of Leak Detection

KROHNE Oil & Gas

 

 

3

 

 

Introduction

10.2 

Leak Locati on

31 

10.3  10.3.1  10.3.2  10.3.3  10.3.4 

PipePatrol E-RTTM PipePatrol E-RTTM/PC /PC – Leak Dete Detectio ctio n in Pumping Condit ion Head stations monitoring Substation monitoring without flow measurement Segment monitoring for substations with flow measurement Leak detection with substations and virtual flow monitoring

32  32  33  34  35 

10.4  10.4.1 

PipePatrol E-RTTM PipePatrol E-RTTM/SC /SC – Leak Dete Detectio ctio n in Shut-in Condit ions Head Station Monitoring

37  37 

11 

Comparison of all methods Comparison

41 

Index of figures Index Figure 1: Calculation Calc ulation of mass flow from volumetric flow ......................................................................................... .............................................................. ........................... 16   Figure 2: Conditional probability density functions ......................................................................... ................................................................................................. ........................ 19  Figure 3: Leak location by gradient intersection method ........................................................................................ ...................................................................... .................. 22   Figure 4: RTTM to calculate local profiles; model using pressure readings ........................................................... ......................... .................................. 24  Figure 5: RTTM to calculate local profiles; model using flow readings ..................................................... ................................................................... .............. 24  Figure 6: Compensated mass balance with RTTM based compensation .............................................................. .................................... .......................... 25   Figure 7: Residual or differential approach for head station monitoring ................................................................. 26   Figure 8: Residual or differential approach for pressure substation ...................................................................... ................................................................... ... 27   Figure 9: Functionality of PipePatrol SMB .............................................................................................................. ............................................................ .................................................. 28   Figure 10: PipePatrol E-RTTM/PC: pumping conditions, head station monitoring ................................................. 32  

 

Figure 12: 11: PipePatrol PipePatrol E-RTTM/PC: E-RTTM/PC: pumping pumping conditions, conditions, PTF pressure at substations ................................................. .............................. ................... 33 Figure substations............................................. substations............................................................. ................ 34   Figure 13: PipePatrol P ipePatrol E-RTTM/PC: pumping conditions, PT substations ............................................................... 35   Figure 14: PipePatrol P ipePatrol E-RTTM/PC: pumping conditions, P substations .......................................................... ................................................................. ....... 36   Figure 15: PipePatrol E-RTTM/SC: shut-in conditions, no substations .................................................................. 37   Figure 16: PipePatrol P ipePatrol E-RTTM/SC: shut-in condition, P substations ...................................................................... ........................................................... ........... 38  

Index of tables Table 1: Symbols, labelling, units - part 1...................................................... 1.............................................................................................................. ........................................................ 1-6   Table 2: Symbols, labelling, units - part 2...................................................... 2.............................................................................................................. ........................................................ 1-6   Table 3: Functionality and instrumentation of Pressure and Flow Monitoring . .......................................... ...................................................... ............ 10   Table 4: Fields of application of Pressure- and Flow Monitoring. ........................................................................... 10   Table 5: Performance P erformance Parameters of Pressure and Flow Monitoring. .................................................................... ....................................................... ............. 11  Table 6: Functionality and instrumentation for negative pressure wav wave. e. ................................................................ 12   Table 7: Fields of application for negative pressure wave. ......................................................... ..................................................................................... ............................ 12   Table 8: Performance P erformance Parameters of Negative Pressure Wave W ave ............................................................................. ..................................................... ........................ 13 Table 9: Functionality and instrumentation for balancing methods.................................................... methods. ........................................................................ ..................... 17    Table 10: Fields of application for balancing methods. ................................................................. .......................................................................................... ......................... 17   Table 11: Performance parameters for balancing methods..................................................... methods. ................................................................................... ............................... 18   Table 12: Functional summary of statistical LDSs. ....................................................... ................................................................................................. .......................................... 20   Table 13: Fields of application for statistical LDSs. ................................................................................................ .................................................. .............................................. 20   Table 14: Performance parameters for statistical LDSs. ........................................................................................ 21  Table 15: Functionality and instrumentation of PipePatrol SMB. ............................................................................ ....................................................... ..................... 28   Table 16: Possible fields of application of PipePatrol SMB. ................................................................................... 29  Table 17: Performance parameters of PipePatrol SMB. ........................................................................................ ........................................................... ............................. 30   Table 18: Functionality and instrumentation of PipePatrol E-RTTM. ........................................................... ...................................................................... ........... 39  Table 19: Possible fields of application of PipePatrol E-RTTM. ............................................................................. .................................................. ........................... 39  Table 20: Performance P erformance parameters of PipePatrol E-RTTM. ................................................................ ................................................................................... ................... 40   Table 21: Comparison of all methods for leak detection. .................................................... ....................................................................................... ................................... 41 

Fundamentals of Leak Detection

KROHNE Oil & Gas

4

 

 

Introduction

Symbols, Labelling, Labelling, Units Symbol

Synonym

SI-Unit SI-Unit

 A  

Cross section of the pipeline

m2 



Speed of sound

m/s



Mass flow or volumetric flow

kg/s, m3/s

 L  

Length of pipeline

m

 M   

Gas specific molar mass

kg/mol

 M   

Mass

kg

 M  Leak   

 Actual drained leak-mass

kg

 M Pipe  

Mass stored in pipeline

kg

&    M 

Mass flow in general

kg/s

&    M   I 

Mass flow inlet

kg/s

&  M   Leak   

Leak flow

kg/s

&    M  O

Mass flow outlet

kg/s



Number of substations

 N (  μ , σ )   Gaussian distribution

 p  

Pressure



Probability in general

P0  

Probability for correct decision under leak free conditions

P1

Probability for correct decision under leak conditions

PFA  

Probability for false alarm

P M   

Probability for false decision under leak conditions

 R  

Gas constant (8.314472)

J/(mol K)

 RS   

Specific gas constant

J/(kg K)



One-dimensional coordinate along the pipeline

m

t  

time

s

T , T F   

Temperature of fluid

K

T G  

Temperature of ground

K



Velocity of fluid

m/s

v I   

Velocity of fluid at inlet

m/s

vO  

Velocity of fluid at outlet

m/s

V   

Volume in general

m3 

&   V 

Volume-flow in general

m3/s

&   V   I 

Volume-flow at inlet

m3/s

& V  ref   

Volume-flow at reference conditions

sm3/s

&   V  O

Volume-flow at outlet

m3/s

VCF   

Volume correction factor

 x  

Flow residual inlet

kg/s

 y  

Flow residual outlet

kg/s

 z  

Pressure residual in general

Pa

Fundamentals of Leak Detection

Pa

KROHNE Oil & Gas

5

 

 

Introduction

Table 1: Symbols, labelling, units - part 1.

Symbol

Synonym

α   

Level of significance

ε   

Coefficient of SPRT

SI-Unit SI-U nit

γ    λ   

Smallest detectable leak rate Coefficient of SPRT

μ   

Mean of the Gaussian distribution

 ρ   

Density of the fluid

σ   

Standard deviation of the Gaussian distribution

 Z   

Compressibility factor

kg/s

kg/m3 

Table 2: Symbols, labelling, units - part 2.

Fundamentals of Leak Detection

KROHNE Oil & Gas

6

 

1

1

Introduction

Introduction

Pipelines are the most economic and safest transport systems for mineral oil, gases and other products. As a means of long-distance transport, pipelines have to fulfil high demands of safety, reliability and efficiency. Leak Detection Systems (LDSs) are therefore an important aspect of pipeline technology. Modern LDSs such as the PipePatrol family from KROHNE Oil and Gas monitor pipelines continuously, by testing whether a leak has occurred or not (leak detection). In case of a leak they also calculate the leak flow and the leak position (leak tracking). Some countries formally regulate pipeline safety, for example the German rules are laid down in “Technische Regel für Fernleitungen” [TRFL]. [TRFL]. Other countries provide standards, for example Leak Detection is specifically addressed by [API 1130] and 1130]  and [API 1155] in 1155] in the USA. Details on this subject can be found in Chapter 2. API1155 defines the following important requirements of an LDS: Sensitivity: a LDS must ensure that the loss of fluid as a result of a leak is as small as possible. This places two requirements on the system: it must detect small leaks, and it must detect them quickly. PipePatrol’s Extended RealTime Transient Model E-RTTM (Chapter 10) 10) is able to detect leakage below 1% of nominal flow rate within less than a minute. Reliability: the user must be able to trust the LDS. This means that it must correctly report any real alarms, but it is equally important that it does not generate false alarms.  Accuracy: the LDS must report the leak location accurately. Robustness: the LDS must to continue operate in circumstances. For example, in case ofsuch a transducer failure the system must detect the continue failure and tonon-ideal operate (possibly with necessary compromises as a s reduced sensitivity). Universal applicability: a modern LDS is expected to be universally applicable. For example, PipePatrol can be used with equal success on liquid or gas pipelines. PipePatrol operates effectively on multiproduct pipelines with or without separation pigs. Drag Reducing Agent (DRA) makes life a little more interesting, but creates no real problem. Wide operating range: leak detection is important throughout the whole operating range. This includes start-up and shutdown conditions as well as steady state. Some leak detection systems also provide a model that works under shut-in conditions, such as PipePatrol E-RTTM - see Chapter 10. 10. This survey describes the principles of leak detection and leak location as follows: •  Chapter 3:

Pressure/Flow Monitoring

•  Chapter 4:

Negative Pressure Wave

•  Chapter 5:

Balancing Methods

•  Chapter 6:

Statistical Leak Detection Systems

•  Chapter 7:

Leak Location

Chapter 8 describes in detail the use of powerful modern computer systems to implement Real-Time Transient Models( RTTMs). Chapter 9 describes PipePatrol Statistical Mass Balance (SMB), (SMB) , which combines the mass balance with statistical methods and RTTM technology. Chapter 10 describes 10 describes PipePatrol Extended Real-Time Transient Model E-RTTM, the Premium leak detection solution of KROHNE Oil & Gas. Chapter 11 provides 11 provides a comparison of available leak detection methods.

Fundamentals of Leak Detection

KROHNE Oil & Gas

7

 

2

Regulatory Framework

2

Regulatory Re gulatory Fra Framework mework

2.1

TRFL (Germany )

TRFL stands for „Technische Regel für Rohrfernleitungen“ [TRFL], [TRFL], which was published in 2003 in Germany and applies to all Pipelines that transport flammable and/or dangerous liquids or gases. Chapter 11.5 of the TRFL requires leak detection systems all such pipelines. It demands: •  Two autonomous, continuously operating systems that can detect leaks in steady state conditions. •  One of these systems, or a third one, able to detect leaks in transient conditions. •  One system to detect leaks in shut-in conditions. •  One system to detect gradual leaks. •  One system to detect the leak position.

2.1.1

Installations according to TRFL a) and b)

[TRFL] requires two autonomous, continuously operating systems that can detect leaks in the steady state. Either of [TRFL] requires these systems, or both, or a third one, must be able to detect leaks in transient conditions. Redundant instrumentation is required in principle, but in practice the requirement for redundant equipment is frequently relaxed. This may happen either because the risk of damage to life and property is relatively low, or because instruments at substations effectively provide back-ups back-ups for each other. Redundant signal paths and commu communication nication are always required, however. The leak detection system itself must always be redundant, for example using multiple techniques techni ques including: •  Pressure and Flow monitoring

Chapter 3

•   Acoustic/negative pressure wave

Chapter 4

•  Line balance methods

Chapter 5

•  Statistical LDS

Chapter 6

KROHNE Oil & Gas offers with PipePatrol state-of-the-art leak detection systems as: •  PipePatrol Statistical Mass Balance (SMB)

Chapter 9, and

•  PipePatrol Extended Real-Time Transient Model (E-RTTM)

Chapter 10. 10.

2.1.2

Installations according to TRFL c) 1

[TRFL] requires that each pipeline has one system to detect leaks in shut-in conditions [TRFL] requires c onditions . Chapter 11 lists 11 lists the methods that fulfil these needs. PipePatrol E-RTTM Chapter 10 uses 10 uses a model-based pressure-temperature method, which can be applied to liquids and gases. In shut-in conditions, valves will lock a pressure into one or more sections of the pipeline. It is possible for considerable pressure changes to occur in this case as a result of thermal effects, but any rapid or unexpected fall in pressure indicates that a leak has occurred. 2.1.3

Installations according to TRFL d)

Typically these systems utilise a sensor cable installed along the pipeline. Leak detection is either by change in temperature (fibre optics) or change in gas concentration (semi permeable sensor cable). The following should be taken into account: •  The operating pressure and temperature must be suitable •  Not all fluids can be monitored.

  1

This point is relevant to liquid pipelines, but flow in gas pipelines is normally continuous. Nevertheless, the TRFL requirement applies to all pipelines in principle. Fundamentals of Leak Detection

KROHNE Oil & Gas

8

 

2

Regulatory Framework

If accurate flow, pressure and temperature readings are available, availabl e, PipePatrol SMB Chapter 9 can be applied. If adequate, leak tight valves are present, the PipePatrol PipeP atrol E-RTTM/SC model based pressure-temperature method can also be applied, Chapter 10.4 10.4.. 2.1.4

Installations according to TRFL e)

The TRFL requires the LDS to locate the position of a leak as fast as possible. This function can be integrated into i nto one of the systems installed to comply with section a) – for example, PipePatrol E E-RTTM, -RTTM, Chapter 10. 10. Details of leak localisation are described in Chapter 7.

2.2

API 1130 (USA)

The second edition of API (American Petroleum P etroleum Institute) standard 1130 “Computational Pipeline Monitoring (CPM) for Liquid Pipelines” was released in 2002 [API 1130] 1130].. API 1130 does not directly iimpose mpose legal requirements on pipeline operators in the same way as TRFL, but it provides the necessary technical information for conscientious operators to operate their pipelines safely. [API 1130] covers 1130] covers liquid pipelines only. onl y. It describes design, implementation, test and operation of Computational Pipeline P ipeline Monitoring (CPM) systems, based on an algorithmic approach to leak detection. It also gives recommendations for (self) test and operator training. LDSs are divided into two groups: •  External systems use dedicated measurement equipment, such as a sensor cables •  Internal systems use existing measurement sensors providing flow or pressure readings. All LDSs introduced in this survey are part of this group.

2.3

API 1155 (USA)

The [API 1155] “Evaluation 1155] “Evaluation Methodology for Software Based Leak Detection Systems” was first published iin n 1995, and defines methods of comparing LDSs from different manufacturers. These criteria are defined: Sensitivity

Reliability

 Accu  Ac cu rac y

Robustness

The sensitivity is a composite measure of the size of a leak that a system is capable of detecting, and the time required for the system to issue an alarm in the event that a leak of that size should occur. PipePatrol E-RTTM typically detects leakage below 1% (relating to nominal flow rate) in less than one minute, resulting in a leak volume that is typically less than 50 litres. Reliability is a measure of the ability of a leak detection system to render accurate decisions about the possible existence of a leak on the pipeline, while operating within an envelope established by the leak detection It follows that reliabilityof is incorrectly directly related to theaprobability detecting leak, given that a leaksystem does indesign. fact exist, and the probability declaring leak, givenofthat no leakahas occurred.  Accuracy covers estimation of leak parameters such as leak flow rate, total volume lost, type of fluid lost, and leak location within the pipeline network. The validity of these leak parameter estimates should be as accurate as possible. Robustness is a measure of the leak detection system’s ability to continue to function and provide useful information even under changing conditions of pipeline operation, or in conditions where data is lost or suspect. A system is considered to be robust if it continues to function under such non-ideal conditions.

Fundamentals of Leak Detection

KROHNE Oil & Gas

9

 

3

Pressure/Flow Monitoring

3

Pressure/F Pre ssure/Flow low Monitori ng

 A leak changes the hydraulics of the pipeline, and therefore changes flow or pressure readings after some time [Krass/Kittel/Uhde].. Local monitoring of pressure or flow at only one point can therefore provide simple lleak [Krass/Kittel/Uhde] eak detection. It requires no telemetry, for example to compare flow rate at inlet an outlet, as local monitoring of pressure or flow rate is sufficient. It is only useful in steady s teady state conditions, however, and its ability to deal with gas pipelines and multi-product liquid pipelines is extremely limited. It does not provide good sensitivity, and leak localisation is not possible.

3.1 3.1

Pressure Pressure Monitoring

Δ p . As pressure sensors are almost always installed, it If a leak occurs, the pressure in the pipeline will fall by an amount is natural to use them for leak detection. The pressure in the pipeline is simply compared c ompared against a lower limit after reaching steady state conditions. When the pressure p ressure falls below this lower limit, a leak alarm is raised. This method is also called Pressure Point Analysis.

3.2 3.2

Flow Monitoring

The sensitivity of the pressure monitoring method depends on the leak location. Near the inlet and the outlet of the pipeline a leak leads to little or no change in pressure. This can be compensated by flow monitoring, where the flow is measured for change. The two methods can be combined.

3.3

Summary

The following table gives an overview of functionality and requirements (for more details, see Chapter 11). 11).

Instrumentation

Method

Function Complexity

Demands

Pressure Monitoring

LD

1xP

Low

Flow Monitoring

LD

1xQ

Low 2

Table 3: Functionality and instrumentation of Pressure and Flow Monitoring   .

Both methods provide leak detection, but no leak localisation. For pressure monitoring only pressure sensor is required, and for flow monitoring only one flow meter is required. Demands on instrumentation are low. The possible fields of application are:  Ap pl ic ati on Method

Pressure Monitoring

Medium

TRFL

Steady

L/G

(a) (c)

Steady

L/G

(a)

Pumping

Dynamics

PC, SC PC

Flow Monitoring

 3

Table 4: Fields of application of Pressure- and Flow Monitoring  .

Pressure Monitoring is able to detect leaks in shut-in conditions as well as in pumping conditions. This will be true if the pipeline valves seal tightly enough. In contrast, flow monitoring is only able to detect leaks in pumping conditions. Both

2

LD = Leak detection, P = Pressure sensor, Q = Flow sensor

3

 PC = Pumping conditions, SC = Shut-in conditions, L = Liquid, G = Gas

Fundamentals of Leak Detection

KROHNE Oil & Gas

10

 

3

Pressure/Flow Monitoring

methods are restricted to steady state, as small changes in pressure or flow will cause a false alarm. Either method is capable of monitoring gas and liquid pipelines. Pressure monitoring meets the following requirements of TRFL: •  TRFL a), one autonomous, continuous working system that can detect leaks in steady state conditions and •  TRFL c), one system to detect leaks in shut-in conditions.

In contrast, flow monitoring only achieves: •  TRFL a), one autonomous, continuous working system that can detect leaks in steady state conditions The following table lists the associated performance parameters. Sensitivity Method

 Al arm Threshold Thre shold

Tim e to D etec etectt   Liquid Ga Gas s

Leak Types

Pressure Monitoring

High

Short

Long

Both

Flow Monitoring

High

Short

Long

Both

Table 5: Performance Parameters of Pressure and Flow Monitoring.

Both methods will work without malfunction if i f pressure and flow stay constant in daily operation. This is true for some liquid pipelines, but never for gas pipelines. pipelin es. These simple methods normally do not use statistical methods to prevent false alarms (Chapter 6). The only way to avoid false alarms is therefore to set wide alarm a larm limits. This causes a short time to detect a leak within liquid pipelines. p ipelines. In gas pipelines pressure changes are rather slow, so leak detection is slow. Both methods detect sudden leaks as well as gradual leaks of adequate size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

11

 

4

4

Negative Pressure Wave

Negativ Ne gativ e Pressure Wa Wave ve

 A sudden leak caused, for example, by careless use of an excavator, leads to a negative pressure wave propagating at the speed of sound (c ) up- and downstream through the pipeline. Such a wave can be recognised using installed pressure transmitters, giving a leak alarm. It is also possible to calculate the leak location by timing the arrival of the pressure wave at two or more points on the pipeline (Chapter 7).

4.1

Summary

The following table gives an overview of functionality and requirements (for more details, see Chapter 11). 11).

Instrumentation Method

Function Complexity

Demands

LD

1xP

Medium

LD+LL

2xP

Medium

Negative Pressure Wave (no leak location) Negative Pressure Wave (with leak location)

 4

Table 6: Functionality and instrumentation for negative pressure wave .

One pressure transmitter allows leak detection only. At least two transmitters are needed for lleak eak localisation. In either case, the selected transmitters must be capable of detecting rapid changes in pressure. The possible fields of application are:  Ap pl ic ati on Method

Medium

TRFL

Pumping

Dynamics

Negative Pressure Wave (no leak location)

PC, SC

Steady

L

(a) (c)

Negative Pressure Wave (with leak location)

PC, SC

Steady

L

(a) (c) (e)

5

Table 7: Fields of application for negative pressure wave .

The negative pressure wave method is able to detect leaks in steady state as well as in shut-in condition. It is only able to detect leaks in steady state conditions, and small variations in pressure can easily lead to false alarms. Negative pressure wave methods are most useful in liquid pipelines, p ipelines, as pressure waves are quickly attenuated in gas pipelines. This technique meets the following TRFL requirements: •  TRFL a), one autonomous, continuous working system that can detect leaks in steady state conditions and •  TRFL c), one system to detect leaks in shut-in conditions. •  TRFL e), one system to detect the leak position.

4

 LD = Leak detection, LL = Leak location, P = pressure sensor

5

PC = Pumping conditions, SC = Shut-in conditions

Fundamentals of Leak Detection

KROHNE Oil & Gas

12

 

4

Negative Pressure Wave

The following table lists the associated performance parameters. Sensitivity Method

 Al arm Threshold Thre shold

Tim e to Detect Detec t   Liquid

Leak Types Ga Gas s

Pressure Monitoring

High

Short

Long

Sudden

Flow Monitoring

High

Short

Long

Sudden

Table 8: Performance Parameters of Negative Pressure Wave

This technique will work without malfunction if i f pressure and flow stay constant in daily operation, which is true for some liquid pipelines but never for gas pipelines. Statistical methods to prevent false alarms (Chapter 6) normally will not be used. The only way to avoid false alarms is therefore to set wide alarm limits. This causes a short time to detect a leak within liquid pipelines. In gas pipelines pipel ines pressure changes are rather slow, so leak detection is also slow. This method only detects sudden leaks of adequate size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

13

 

5

Balancing Methods

5

Balancing Bala ncing Me Methods thods

5.1

Mass Balance

The mass balance method is based on the equation of conservation of mass. In the steady state, the mass entering a leak-free pipeline (MI) will balance the mass leaving leavi ng it (MO). In the more general case, the difference in mass at the two ends must be balanced against the change of mass inventory of the pipeline ( ∆Mpipe). Over any given period of time, we can therefore say

Δ M  I  − Δ M    O = ΔM  pipe

 

If there is no leak. In principle, the mass in the pipe depends on the density of the product multiplied by the volume of the pipeline. Both are functions of temperature and pressure and the density is also a function of the composition of the product. None of these values is necessarily constant along the pipeline.  Any addition mass imbalance indicates a leak. This can be quantified by rearranging the equation and adding a term for leak mass ( ∆Mleak):

Δ M leak  = Δ M  I  −  Δ M  O − ΔM  pipe

 

These equations are valid in any consistent mass units. 5.1.1 Uncompensated mass balance Supposing that a leak were allowed allo wed to continue for a long period, the mass entering and leaving leavi ng the pipeline would increase indefinitely. The mass inventory of the pipeline, on the other hand, remains within a fixed range – and in reasonably steady conditions that range is quite narrow.  ∆M  pipe therefore becomes negligible over a sufficiently long period, and the equation above reduces to:

Δ M leak  ≈ Δ M     I  − ΔM O   Over a finite period ( T ), ), this equation is an approximation. We must therefore set a detection limit, below which an apparent imbalance may the result of neglecting the inventory. This is the smallest detectable lleak eak rate (γ ). ). A leak is declared if:

( Δ M  I − Δ M O ) > γ  ⋅ T   The time period T must be sufficiently long lo ng for the flow in and out of the pipeline to be large in comparison with the change in pipeline inventory. In the following cases, a very large value will be required: •  Start-up of a pipeline •  Change of pressure at inlet or outlet, even the change is small •  Product change •  Most gas pipelines, most of the time

Fundamentals of Leak Detection

KROHNE Oil & Gas

14

 

5

Balancing Methods

5.1.2

Compensated mass balance

Unlike the uncompensated mass balance, the compensated mass balance takes account of changes in pipeline inventory. The mass inventory of a short section of pipeline with length  ∆s and cross-sectional area A containing a product of density ρ is given by:

Δ M pipe =  ρ ⋅ A ⋅ Δs   Both density and pipe area may vary v ary along the pipeline. To calculate the exact inventory i nventory of a pipeline of length L, it is necessary to integrate the density profile:

 M  pipe  =

  L



0

ρ ( s ) A( s ).ds

It is not possible to determine the density densit y profile along the pipeline directly. All practical methods are based on initially determining the temperature and pressure profile, and then applying and equation of state – an equation of state allows the density to be calculated as a function of temperature and pressure. For products with multiple components such as crude oil and natural gas, additional variables such as molecular weight or density at reference conditions are required. The density of crude oil and common refined products can be calculated according to Manual of Petroleum Measurement Standards Chapters 10 and 11, also known as [API 2540]. 2540]. The density of gas can be calculated from pressure ( p), temperature (T ), ), molecular weight ( M ) and compressibility factor ( Z ) according to the gas law:

 ρ  =

 Mp

 

 ZRT  The value R is the universal gas constant, equal to 8.314472 J (  mol ⋅ K ) . The compressibility factor represents the deviation of the gas from ideal. For temperature and pressure well below the critical point, it often can be assumed to be close to unity. Three main methods are used to determine the pressure and temperature profile: 1.

Direct measurement of pressure and temperature. A qu quantity antity (n) of pressure ( p  pi ) and temperature (T i i ) transmitters must be installed sufficiently closely. The pipeline is then split into segments of known volume ΔV i   at each transducer pair, and the total inventory calculated using:

 M  pipe =

n

∑ ΔV ρ ( p T  ) i

i =1

i, i

 

2.

Determination with the help of a6  simple, simple, steady state model. In liquid pipelines a linear decrease in pressur pressure e can be assumed along the pipeline ; temperature of the fluid can be assumed to equal ground temperature for long pipelines.

3.

Determination with the help of a Re Real-Time al-Time Transient Model (RTTM). The most accurate method is to use a pipeline model that covers transient as well as steady state conditions. This allows the temperature and pressure to be determined at every point – see Chapter 8.

Chapter 9 describes PipePatrol SMB from KROHNE Oil & Gas. PipePatrol SMB is a mass balance system, which offers the possibility to use one of all introduced methods, a), b), and c), to take account on the change of mass in inventory Δ M Pipe in the monitored pipeline.

5.2 5.2

Use of volumetri c flow meters

It is not always practical to measure the mass flow in and out of the pipeline directly – for example, direct mass meters are only available in a limited range of sizes. It is possible to substitute volumetric flow meters, but the indicated volume flow must be multiplied by line density to derive the mass. Depending on the application, the options for obtaining density include:   6 

 This assumes a horizontal pipeline of constant internal roughness and cross-sectional area. Other cases require a modified approach.

Fundamentals of Leak Detection

KROHNE Oil & Gas

15

 

5

Balancing Methods

•  The density of liquids of known composition can be stored in a lookup table •  The density can be directly measured •  The density for crude oil and its products can be determined with the help of pressure and temperature using [API 2540],, provided that a reference density is available 2540]

lo w •  The density of gas can be calculated according to the gas law introduced in the previous section. For pure gas at low pressure, a very simple approach is possible. For natural gas it will be necessary to measure the molecular weight, for example using a gas chromatograph. At high pressure it will be necessary to calculate the compressibility factor. Where volumetric flow meters are used, it can be convenient to express the pipeline balance in the form of standard volume instead of mass. The standard volume ( V s) is defined as the mass (M ) divided by the density at standard conditions ( ρs):

V s =

 M 

 ρ s

 

The standard density is simply the product density at some fixed and agreed temperature and pressure, such as 1.01325 bar and 15°C. In principle, conversion to standard volume simply involves dividing the mass balance equation by a constant. The conversion can be more complicated in the case of a multi-product pipeline, where the product ent entering ering the pipeline can be different from the product leaving it. For liquids, the standard form of the API equation according to [API 2540] 2540] allows  allows standard volume to be derived from actual volume (V  A) using two coefficients:

V s = V  A  ×   C tl × C  pl In this case, the mass flow is apparently bypassed – though it is, in fact, still hidden in the derivation of the coefficients.

Q

(V& , v )

Q

&  M 

&  M 

  Figure 1: Calculation of mass flow from volumetric flow

Fundamentals of Leak Detection

KROHNE Oil & Gas

16

 

5

5.3

Balancing Methods

Summary

The following table gives an overview of functionality and requirements (for more details, see Chapter 11. 11. Instrumentation Method

Function Complexity

Demands

Uncompensated mass balance

LD

2xQ

High

Compensated mass balance Direct p and T measurement

LD

2 x (Q,P,T) n x (P,T)

High

Compensated mass balance Steady state model

LD

2 x (Q,P,T) TG 

High

Compensated mass balance RTTM

LD

2 x (Q,P,T) TG 

High

7

Table 9: Functionality and instrumentation for balancing methods .

 All balancing methods require at least two flow meters, one at the inlet, the other at the outlet. They provide leak detection, but no leak location. When the change in pipeline inventory is compensated, additional pressure and temperature sensors are also needed. Demands on the accuracy acc uracy of the flow meters are high, because their error limits are also the detection limits. The possible fields of application are:  Ap pl ic ati on Method

Medium

TRFL

Steady

L/G

(a)

PC

Steady Low Transient

L/G

(a)

Compensated mass balance Steady state model

PC

Steady Low Transient

L/G

(a)

Compensated mass balance RTTM

PC

Steady Transient

L/G

(a) (b)

Pumping

Dynamics

Uncompensated mass balance

PC

Compensated mass balance Direct p and T measurement

8

Table 10: Fields of application for balancing methods .

Balancing methods can be used only in pumping conditions: use in shut-in conditions is not possible. Uncompensated mass balance is only able to monitor steady state conditions. Compensated mass balance is able to monitor for leaks in the presence of moderate transients, but the detection time will be increased. Uncompensated mass balance is limited to liquid pipelines; compensated mass balance can monitor gas pipelines more or less. This technique meets the following TRFL requirements: •  TRFL a), one autonomous, continuous working system that can detect leaks in steady state conditions.

Only the RTTM-compensated mass balance meets this TRFL requirement: •  TRFL b) one of these systems, or a third one, has to be able to detect leaks in transient conditions.



 LD = Leak detection, Q = Flow sensor, T= Temperature sensor, P = Pressure sensor, T G = Ground temperature sensor

8

PC = Pumping conditions, L = Liquid, G = gas

Fundamentals of Leak Detection

KROHNE Oil & Gas

17

 

5

Balancing Methods

The following table lists the associated performance parameters. Sensitivity Method

 Al arm Threshold Thre shold

Tim e to Detect Detec t   Liquid

Leak Types Ga Gas s

Negative Nega tive Pressure Wave

Uncompensated mass balance

Medium

Long

Very Long

Both

Compensated mass balance Direct p and T measurement

Medium

Medium

Long

Both

Compensated mass balance Steady state model

Medium

Medium

Medium

Both

Compensated mass balance RTTM

Medium

Short

Short

Both

Table 11: Performance parameters for balancing methods.

 All balancing methods achieve (using accurate flow meters) a medium detection limit. Uncompensated mass balance has a long time to detect, while compensation for change of inventory helps to shorten the detection time. RTTMcompensated mass balance shows the best results. Leak detection time is longer for gases because of the dynamic inertia of pressure and flow. All balancing methods detect sudden leaks as well as gradual leaks of adequate size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

18

 

6

6

Statistical Leak Detection Systems

Statist Sta tist ical Lea Leak k De Detection tection Systems

Statistical Leak Detection Systems use statistical methods to detect a leak. This leads to the opportunity to optimise the decision if a leak exists in the sense of chosen statistical parameters. However it makes great demands on measurements. They need to be steady state (in a statistical sense) for example. Statistical LDSs have poor sensitivity in transient conditions unless they are adapted, for example using a Real-Time Transient Model Statistical methods can improve the performance of all leak detection methods introduced in this survey. This chapter describes statistical LDSs based on Uncompensated mass balance balance,, Chapter 5.1.1 5.1.1,, because these systems are common.  ATMOS PipeTM from ATMOS International [Zhang] [Zhang] is  is an example. Statistical Leak Detection Systems use methods and processes from decision theory  [Kay] [Kay].. The hypothesis-test for leak detection based on the Uncompensated mass balance balance,, Chapter 5.1.1 5.1.1,, uses either a single measurement, or multiple measurements made at different times. & = M & − M  &  can be used to decide between two hypotheses,  H  and  H  : One or more measurements of Δ M  0 1  I O

 H 0 : No leak   H 1 : Leak    & | H   for hypothesis  H  Every individual measurement is described by conditional probability density function  p ( Δ M 0 0)   & | H  (no leak) and  p ( Δ M 1 ) for hypothesis  H 1 (leak). In general a Gaussian distribution is assumed:

 

Figure 2: Conditional probability density functions   &  to one To assign a single measurement Δ M  one of the the both both hypo hypoth thes esis is  H 0 and and H 1 , an alarm alarm limi limitt γ is defined. The test than is defined as follows:

leak  k  ⎧≤ γ  ⇒  H 0 : No lea & Δ M    ⎨ Leak ak ⎩ > γ  ⇒  H 1 : Le

6.1

Probabil ity Ratio Ratio Test

The art of the Probability Ratio Test is to choose a value of γ so that: l eak free conditions ( PFA = P ( ΔM > γ  | H 0  true ) → min ), and •  No false alarm is given under leak

Fundamentals of Leak Detection

KROHNE Oil & Gas

19

 

6

Statistical Leak Detection Systems

•   An alarm is always given under leak conditions ( P M   = P ( ΔM ≤ γ  | H1 t rue ) → min )

This problem is solved throughout the family of Probability-Ratio-Tests. Details can be found in [Kroschel] [Kroschel] for  for example.

6.2

Sequential Probabil ity Ratio Test (SPRT) (SPRT)

& . Statistical methods are more The Probability Ratio Test bases its classification result on one single measurement Δ  M  powerful when testing a whole collection of data:

&  = Δ M  & K   & ΔM 1   ΔM  N    & is described by the conditional probability density function  p ΔM The characteristic of the collection of data ΔM ( & | H 0 )   & | H   for hypotheses  H  (leak). [Wald] published for hypotheses  H 0 (no leak)   and p ( ΔM [Wald] published the Sequential Probability 1 1)

Ratio Test (SPRT) in the early 40’s, which leads to a recursive algorithm that can c an be used for online-testing.

6.3

Summary

The following table gives an overview of functionality and requirements (for more details, see Chapter 11). 11). Instrumentation Method

Function Complexity

Demands

2xQ

Medium

Statistical LDS

Uncompensated mass balance

LD 9

Table 12: Functional summary of statistical LDSs .

Statistical methods need two flow meters at least, one at the inlet, the other one at the outlet. They provide leak detection, but no leak location. The use of statistical methods can reduce the demands on accuracy of the flow meters. Possible fields of application are:  Ap pl ic ati on Method Pumping

Dynamics

PC

Steady Low Transient

Medium

TRFL

L/G

(a)

Statistical LDS

Uncompensated mass balance 10

Table 13: Fields of application for statistical LDSs .

Statistical LDSs can be used in pumping conditions, but not under shut-in conditions. Statistical LDSs are able to operate in moderate transient conditions, but with increased leak detection time. Statistical LDS provides moderate performance on gas pipelines. They tick the following boxes: •  TRFL a), one autonomous, continuous working system that can detect leaks in steady state conditions.

The following table lists the associated performance parameters.

9

 LD = Leak detection

10 

 PC = Pumping conditions

Fundamentals of Leak Detection

KROHNE Oil & Gas

20

 

6

Statistical Leak Detection Systems

Sensitivity Method

 Al arm Threshold Threshold

Tim e to Detect Detec t   Liquid

Leak Types Ga Gas s

Statistical LDS

Uncompensated mass balance

Low

Long

Very Long

Both

Table 14: Performance parameters for statistical LDSs.

Statistical LDS have are very sensitive with a low alarm limit, but time to detect is rather long. Statistical methods detect sudden leaks as well as gradual leaks of adequate size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

21

 

7

Leak Location

7

Leak Lea k Location

When a leak is detected, it i t is important to locate it. An exact leak location gives the opportunity to take swift containing action to minimise harm to people and the environment. Localilsed repairs can then be carried out cost-effectively.

7.1

Gradient Intersect ion Method

The gradient intersection method is based on the fact that the pressure profile along the pipeline with its length L will change significantly if a leak occurs.

& V   I 

 

dV  Leak  dt 

& V  O

 

Figure 3: Leak location by gradient intersection method 11

Pressure drop in a leak free pipeline is linear   (dashed, green line in Figure 3). If a leak occurs, the pressure profile develops a kink at the leak point – (continuous, red line). The leak location can be determined by calculating the intersection point of the pressure profiles upstream and downstream of the leak. The classic gradient intersection approach calculates the gradient of both lines using two pressure readings near the inlet and two t wo pressure readings near the outlet. The model-based gradient intersection method as used by PipePatrol E-RTTM LDS Chapter 10, 10, calculates the two gradients with the help of the real time transient model, computed by flow and pressure pressu re measurements at in- and outlet. Direct use of pressure measurements achieves accurate results, but only onl y if pipeline is in steady state. The origin or development of the leak (sudden or gradual) does not matter. PipePatrol E-RTTM uses the RTTM based gradient intersection method, which compensates transients leading to good results result s even under highly transient conditions.

7.2

Wave Propagatio n Method

 A sudden leak caused, for example, by careless use of an excavator, leads to a negative pressure wave propagating at the speed of sound (c) up- and downstream through the pipeline of given length (L). Such a wave can be recognised 12 

using installed pressure transmitters, giving a leak alarm. The leak position can be determined  if the moment t 

(downstream) and up (upstream), when this negative wave passes the transmitters is measured. Setting the leak location is: sˆ Leak  =

t down

Δt = tdown − t up

,

1 ⋅ ( L − c ⋅ Δt )   2

The wave propagation method needs an identifiable negative pressure wave. Results will be good, if a leak is sufficiently large and sudden. Small and/or gradual leaks cannot be located by this method. In practical use, it is limited to steady state conditions. It is able to locate leaks in pumping or in shut-in conditions. PipePatrol E-RTTM uses the RTTM based gradient intersection method, which compensates transients leading to good results even under highly transient conditions..  

 

11

This is true must for liquid pipelines with constant wall friction coefficient k R, a constant cross-section A, and a horizontal built pipeline. The method be slightly modified in otherlocal cases. 12

The speed of sound is not constant in liquid pipelines under multi-product condition or in gas pipelines. The method has to be modified slightly. Fundamentals of Leak Detection

KROHNE Oil & Gas

22

 

8

8

RTTM – Real Time Transient Model

RTTM – Rea Reall Time Trans Transient ient Model

RTTM means “Real-Time Transient Model”. Some LDSs of the PipePatrol-LDS-Family by KROHNE Oil & Gas are based on RTTM, also known as the “Pipeline Observer”. Chapter 9 introduces PipePatrol Statistical Mass Balance (SMB), a mass balance system using RTTM for calculating the change in inventory. It also uses statistical methods introduced in Chapter 6. The KROHNE “flagship” is PipePatrol Extended Real-Time Transient Model (E-RTTM), which combines RTTM technology used for the residual-method (Chapter 8.2) 8.2) with leak signature analysis to prevent false alarms, Chapter 10. 10. RTTM systems build mathematical models of the flow within a pipeline using basic physical laws such as: •  Conservation of mass •  Conservation of momentum •  Conservation of energy

When combined with an equation of state, introduced in Chapter 5, RTTM systems easily model transient and steady state flow in a pipeline. A transient state means a large change in short time, so flow, pressure, temperature and density may all change rapidly. The changes propagate like waves through the pipeline with the speed of sound (c ) of the fluid. Transient state conditions occur in a pipeline for example: •   At start-up

ev en the change is small •  If the pressure at inlet or outlet changes, even •  When a batch changes or when multiple products are in the pipeline

Gas pipelines are almost always in transient conditions, because gases are very compressible. Even in liquid pipelines transient effects cannot be disregarded.  An RTTM makes it possible to calculate mass flow, pressure, density and temperature at every point along the pipeline pipeline in real-time with the help of mathematical algorithms. These solutions are called local profiles. The outputs of the RTTM are ˆ  ˆ   (s ) , for example. The "^" is used to indicate that the values are not shown in the diagrams that follow using usi ng the format:  p measured, but calculated. The addition of (s) indicates that these are not simple point values, but profiles and therefore functions of the distance along the pipeline (s). Calculation of the local profiles needs process measurements at the inlet (subscript I) and outlet (subscript O) of the pipeline – these points are known together as the “head stations”. Various combinations of measurement are possible, as we shall see in a moment. A value of ground temperature, T G is also needed, assuming that the pipeline is underground. As it does not vary much along alon g a pipeline in practice, one sensor is used to measure a representative "assumed constant" value.

Fundamentals of Leak Detection

KROHNE Oil & Gas

23

 

8

RTTM – Real Time Transient Model

The simplest and lowest-cost possibility for RTTM R TTM is shown in Figure 4. In this case only temperature and pressure at the head stations are fed into the RTTM, RTT M, along with ground temperature. Flow (F) Pressure (P)

Inlet

Outlet

Temperature (T) of Fluid Temperature (T) of Ground

F

P

T

T

P I  T FF,I  ,I 

 

T

T G,I->O G,I->O

 

P

F

T F,O F,O   PO

RTTM (Pipeline Observer)

ˆ

& s  M ( )

 

 pˆ ( s )

 

ˆ  ( s )  ρ 

 

Tˆ ( s )

 

13

Figure 4: RTTM to calculate local profiles; model using pressure readings  

It is also possible to implement the model using flow at the head stations instead of pressure: Flow (F) Pressure (P)

Inlet

Outlet

Temperature (T) of Fluid Temperature (T) of Ground

F

v I 

P

T

T

T

T FF,I  ,I 

  T G,I->O G,I->O

T F,O F,O

RTTM (Pipeline Observer)  

ˆ

& s  M ( )

 

 pˆ ( s )

 

ˆ  ( s )  ρ 

 

P

F

vO

Tˆ ( s ) 14

 

Figure 5: RTTM to calculate local profiles; model using usi ng flow readings  

Especially in shut-in conditions (chapter 10) 10) this method of RTTM is used.

13

Subscripts are used as follows: "I" = inlet, "O" = outlet, "G" = ground

14

The RTTM is needs the flow as a velocity v in m/s. Usually, flow is given in terms of mass or volume flow, so velocity has to be calculated. Details can be found in section 8.4 Fundamentals of Leak Detection

KROHNE Oil & Gas

24

 

8

RTTM – Real Time Transient Model

Compensation Compe nsation Approach

It was mentioned in 5.1.2 that 5.1.2 that a compensated mass balance calculation n needs eeds to integrate the density ( ρ) at every point along the pipeline in order to determine the mass inventory of the pipe. The RTTM provides the necessary information to do so accurately, as shown in Figure 6.

Outlet

Inlet

F

P

T

T

P I 

T F,I  F,I 

  T G,I->O G,I->O

T

P

F

T FF,O ,O   PO

RTTM (Pipeline Observer) ˆ  ( s )  ρ   L

∫  A ( s ) ⋅ ρ ( s ) ds

Line Pack Compensation

0

ˆ Pipe  M  d/dt 

+

-

ˆ dM Pipe dt   -

&  M  O

&  M 

 I 

ˆ & Δ M& = M   Leak  15

 

Figure 6: Compensated mass balance with RTTM based compensation  

In this implementation, the RTTM calculates the density densit y profile based on pressure and temperature at the head stations. Independent flow measurements at the head stations are combined with the calculated pipeline inventory to perform a complete mass balance. PipePatrol Statistical Mass Balance (SMB) (Chapter 9) combines the mass balance with statistical methods (Chapter 6) and RTTM-technology.

15

The flow needs to be mass flow here; if volume flow given, mass flow has to be calculated – see chapter 5.2

Fundamentals of Leak Detection

KROHNE Oil & Gas

25

 

8

RTTM – Real Time Transient Model

Head Stations Residual or Differential Approach

The flow at head stations in Figure 7 is not necessary to calculate the local profiles, as pressure is used for this purpose. The RTTM calculates flow at the end points of the pipe as well as everywhere else. It is therefore possible check the difference between measured  and  and calculated  flow.  flow. A difference between the two t wo indicates a change in the dynamics of the pipeline – in other words, a suspicion that there may be a leak.

Outlet

Inlet

P

T

T

P I 

T F,I  F,I 

T G,I->O   G,I->O

F

T

P

F

T FF,O ,O   PO

RTTM (no leak) = Pipeline Observer &  M   I 

-

ˆ  M   I 

ˆ  M  O -

 x

&  M  O

 

y

Flow-Residuals 16

 

Figure 7: Residual or differential approach for head station monitoring   

Both of the Flow-Residuals can be used as leak indicators

ˆ

&  x ≡ M&  I − M  I 

 y ≡

M&

ˆ&

O

 

− M O

The no-leak hypothesis H0 is true if the indicated flows agree sufficiently closely with the model. m odel. The leak-present hypothesis H1 is true if there is a positive residual at the inlet and/or a negative residual at the outlet. Mathematically:  H 0 : No leak  H1 : Leak

⇒ x ≈ 0, y ≈ 0 ⇒ x > 0, y < 0

We insist on the appropriate signs for the residuals because a positive residual at the outlet, for example, would indicate that more fluid was leaving the pipeline than expected. In other words, the cases x < 0 and y > 0 would indicate i ndicate a “negative leak”. This tells us something interesting about the performance of the meters or the validity of the RTTM, but it is not a physically realistic basis for declaring a leak alarm. PipePatrol E-RTTM (Chapter 10) 10) uses this technology together with statisti statistical cal methods (Chapter 6) and RTTMtechnology.

  16 

 The flow here needs to be mass; if volume flow given, mass flow has to be calculated – see section 5.2.

Fundamentals of Leak Detection

KROHNE Oil & Gas

26

 

8

RTTM – Real Time Transient Model

Substations Residual Residual o r Differential Approach

If a pipeline is long enough, substations with pressure sensors will often be included. The indicated pressures can be compared with those calculated using the RTTM method, giving pressure residuals as follows:  zi = pi − pˆ i , 1 ≤ i ≤ n  

 A significant residual leads to a suspicion of a leak, although “negative leaks” are once again ignored for the purposes of leak monitoring.

 p1

 pˆ 1

 pˆ 2

 p2

17 18

 

Figure 8: Residual or differential approach for pressure substation    

Note that temperature and flow measurement at the substations are unnecessary.

8.1

Flow Calculati on

Flow Q  can be expressed in three ways: &  e.g. in kg/s or  t /h •  Mass flow  M  /h is needed for Line-Pack-Compensation ( Figure 5) or flow-residuals (Figure 6).

&  e.g. in m3/s or m3/h. •  Volume flow V 

•  Flow velocity v  in m/s. This way is needed n eeded for RTTM as shown in Figure 4.

The relation between those values is given by  M& = A ⋅V&  = A ⋅ ρ  ⋅ v , where A is the cross section of the pipeline in m2 and  ρ   the density of the fluid in kg/m3. Three options for calculating density have already been presented in Chapter 5.2. 5.2.

17 

 For a better point of view only two substations are shown. The method is able to handle as much substation as present at the pipeline.

18 

 P-substations provide pressure readings; P,T-substations provide temperature readings in addition; P,T,F-substations provide flow

readings in addition.

Fundamentals of Leak Detection

KROHNE Oil & Gas

27

 

9

PipePatrol Statistical Mass Balance (SMB)

9

PipePatro PipePa troll Statist ical Ma Mass ss Balance (S (SMB) MB)

PipePatrol Statistical Mass Balance (SMB) combines mass balance with statistical methods (Chapter 6) and RTTMtechnology (chapter 8). Figure 9 presents an overview. Outlet

Inlet

P

T

T

P I 

T FF,I  ,I 

T G,I->O   G,I->O

F

T

P

F

T F,O F,O   PO

Line Pack Compensation • • • •

Without compensation compensation Using measured p and T along the pipeline Using statio onary nary model Using RTTM

dMˆ Pipe dt  

Q &  M 

-

+ &  M   I 

Q &  M 

-

 & Δ M 

&  M  O

Leak Classification Classification: - Leak yes/no - if yes: leak flow

 

19

Figure 9: Functionality of PipePatrol SMB  

The estimated leak rate is analysed by a statistical leak classifier. This reliably prevents false alarms. Line pack compensation is possible with any of those methods presented in Chapter 5.1.2. 5.1.2. Leak classification provides the following advantages: •  Minimum probability of a false alarm ( PFA  → min ) •  Maximum probability of giving an alarm in leak conditions ( P M   → min )

9.1

Summary

 An overview of functionality and requirements is given in the following table (functionality and requirements in comparison to the other, introduced methods will be summarized s ummarized in Chapter 11). 11).

Method

Instrumentation

Function

Complexity

Demands

Medium

PipePatrol SMB

PipePatrol SMB uncompensated

LD

2xQ

PipePatrol SMB with direct P and T compensation

LD

2 x (Q,P,T) n x (P,T)

Medium Mediu m 20  High

PipePatrol SMB with steady state model

LD

2 x (Q,P,T) TG 

Medium Mediu m 20  High

PipePatrol SMB with RTTM

LD

2 x (Q,P,T) TG 

Medium High20

 

 

 

21

Table 15: Functionality and instrumentation of PipePatrol SMB .

19

The transformation of the volume flow Q to mass flow is done within PipePatrol, see chapter 5.2.

20

Medium when meeting TRFL (a), high when meeting TRFL (d)

Fundamentals of Leak Detection

KROHNE Oil & Gas

28

 

9

PipePatrol Statistical Mass Balance (SMB)

 All balancing methods need at least two flow meters, one at the inlet and the other at the outlet. They provide leak detection, but no leak location. If the change in pipeline inventory is compensated, the compensation method defines additional instrumentation such as pressure and temperature sensors. If PipePatrol SMB is combined with highly accurate flow meters, it can detect gradual leaks according to TRFL d). The accuracy requirement on the flow can be reduced if all that is required is an autonomous, continuously operating system that can detect leaks within steady state conditions. These demands are lower than when statistical methods are not used. The possible fields of application are:  Ap pl ic ati on Method

Medium

TRFL

Pumping

Dynamics

PipePatrol SMB uncompensated

PC

Steady Low Transient

L/G

(a)

PipePatrol SMB with direct P and T compensation

PC

Steady Low Transient

L/G

(a) (d)

PipePatrol SMB with steady state model

PC

Steady Low Transient

L/G

(a) (d)

PipePatrol SMB with RTTM

PC

Steady Transient

L/G

(a) (b) (d)

PipePatrol SMB

22

Table 16: Possible fields of application of PipePatrol SMB .

 All versions of PipePatrol-SMB can be used in pumping conditions, but use under shut-in conditions is not possible. All versions of PipePatrol SMB are capable of handling low transient conditions, even on gas pipelines. PipePatrol SMB based on RTTM technology is able to monitor in heavily transient conditions, for example start up and shutdown conditions, and shows outstanding performance on gas pipelines. All versions of PipePatrol SMB are capable of offering: •  TRFL a), an autonomous, continuous working system, which can detect leaks within steady state conditions.

If pipeline inventory compensation is used and the flow meters are accurate, PipePatrol SMB is also able to offer: •  TRFL d), a system to detect gradual leaks.

PipePatrol SMB combined with RTTM technology additionally offers: •  TRFL c), a system able to detect leaks in transient conditions.

21

 LD = Leak Detection, LL = Leak Location.

22 

 PC = Pumping Conditions, SC = Shut-in Conditions.

Fundamentals of Leak Detection

KROHNE Oil & Gas

29

 

9

PipePatrol Statistical Mass Balance (SMB)

The following table lists the associated performance parameters. As the performance varies according to whether gradual leak detection is required, two sets of sensitivity requirements are given in some cases:

Method

T  R F  L 

Sensitivity  Al arm Threshold

Tim e to Detect Detec t   Liquid Gas

Leak Types

PipePatrol SMB

PipePatrol SMB uncompensated

PipePatrol SMB with direct P and T compensation PipePatrol SMB with steady state model PipePatrol SMB with RTTM

Low

Long

Very Long

(a)

Low

Medium

Long

(d)

Very low

Long

Very Long

(a)

Low

Medium

Long

(d)

Very low

Long

Very Long

(a)

Low

Short

Medium

(d)

Very low

Medium

Long

Both Both Both Both

Table 17: Performance parameters of PipePatrol SMB.

 All versions of PipePatrol-SMB offer (very) sensitive alarming thresholds. Time to detect a leak is long without pipeline inventory compensation, but shortens significantly significantly if it is available. Use of highly accurate flow mete meters rs enables detection of gradual leaks – in which case very sensitive alarming thresholds are possible, but b ut time to detect a leak will rise. Time to detect a leak is longer for gas pipelines because of the dynamic inertia of pressure and flow. All versions of PipePatrolP ipePatrolSMB detect both sudden and gradual leaks of sufficient size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

30

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

10

PipePatroll Exten PipePatro Extended ded Rea Real-Tim l-Time e Trans Transient ient Model (E(E-RTTM RTTM))

PipePatrol E-RTTM is KROHNE’s flagship LDS, which fuses the RTTM technology described in chapter 8 with leak signature analysis described in chapter 10.1 10.1 in  in a unique manner. For this reason it is called “Extended RTTM” [Geiger/Werner/Matko].. PipePatrol E-RTTM is able to monitor pipelines in pumping conditions (PipePatrol E-RTTM/PC, [Geiger/Werner/Matko] Pumping Condition) and in shut-in conditions (PipePatrol E-RTTM/SC, Stand-still Condition or Shut-In Condition).

10.1 10. 1

Leak Signatu re Analy sis

 An LDS that generate false alarms cannot be trusted, so it is a key task to eliminate them. PipePatrol E-RTTM uses leak signature analysis, which executes after the pipeline pipelin e observer, to prevent them. In this second stage residuals are analysed for leak signatures: •  Sudden leak. This “classical leak” develops quickly by external damage of the pipeline. It causes a dynamic signature in residuals. When such a leak recognised, a leak alarm will be reported and the leak location and leak flow are determined. •  Sensor drift or gradual leak. These may occur occ ur by contamination of the flow meter or by sm small all leaks caused by corrosion. They result in indistinguishable, slow signatures. When drift is recognized, a sensor alarm is reported and the apparent leak flow is determined.

This boosts the reliability and the robustness of the system without compromise to sensitivity and accuracy. False alarms are prevented, even with low alarm thresholds.

10.2 10. 2

Leak Loc ation

PipePatrol E-RTTM locates leaks using two methods, both introduced in chapter 7: •  Model-based gradient intersection method •  Model-based wave propagation method

The model-based gradient intersection method ([Billmann] ( [Billmann])) is calculated using the mass residuals at the head station ( x,  x, y ). ). For a pipeline of length L:  zˆ Leak  =

− y ⋅L   x − y

The model based wave propagation method analyses residuals x and y for the appearance of a step. If a step is recognized downstream at time t down  in y, and recognized upstream at time t up  in x, the leak location can be determined   by the runtime difference Δt = tdown − t up : sˆ Leak  =

1 ⋅ ( L − c ⋅ Δt )    2

Fundamentals of Leak Detection

KROHNE Oil & Gas

31

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

10.3 10. 3

PipePatro PipePatro l E-RTTM E-RTTM/PC /PC – Leak Detecti Detecti on in Pumpin g Condit ion

10.3.1

Head stations monitoring

Inlet

F

Outlet

P

T

T

P I 

T F,I  F,I 

T G,I->O   G,I->O

T

P

F

T FF,O ,O   PO

RTTM (no leak) = Pipeline Observer ˆ

ˆ

Q &  M 

&  M   I 

&  M  O

&  M   I 

& -  M 

Q &  M 

O

 x

y

Leak Signature Analysis

Leak  Signatures

(Head-End Station)

Leak-Alarm

Sensor-Alarm

Leak flow and location 23

 

Figure 10: PipePatrol E-RTTM/PC: pumping conditions, head station monitoring   

The example in Figure 10 shows head station monitoring based on the residual approach described in chapter 8.2. 8.2. With the help of the RTTM, PipePatrol E-RTTM E -RTTM compares the measured flow at inlet and outlet with the calculated flow assuming a leak free pipeline. The flow residuals, which are used by the leak signature analysis, are:

ˆ

 x ≡ M&  I − M& I

ˆ

& y ≡ M& O − M  O

Use of the RTTM Pipeline Observer compensates the transient behaviour of the pipeline. Even under heavy transient conditions (for example during pipeline start-up) residuals stay close cl ose to zero in leak-free conditions. Sensitive leak detection is therefore possible in transient conditions The dynamic-free residuals are now passed to the second stage, the leak signature analysis. Its tasks according to Chapter 10.1 and 10.1 and 10.2 are 10.2 are to: •  Manage alarms •  Determine leak rate and leak location.

  23

The transformation of the volume flow Q to mass flow is done within P ipePatrol, see Chapter 8

Fundamentals of Leak Detection

KROHNE Oil & Gas

32

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

10.3.2

Substation monitoring without flow measurement

Substation 1

Inlet

Substation 2

Section I->1

F

P

T

P I 

T FF,I  ,I 

Section 1->2

Section 2->O

T

P

Outlet

T

P

T G,I->O   G,I->O

P

F

T FF,O ,O   PO

RTTM (no leak) = Pipeline Observer  p1

-

 pˆ1

 pˆ 2

 

 z1 Leak  Signatures

Leak Signature Analysis (Substation 1)

-

 p2

z2 Leak Signature Analysis (Substation 2)

Leak  Signatures

Substation Evaluation

Le Leak ak-A -Ala larm rmss S Sec ecti tion on (i)(i)->( >(i+ i+1) 1)

Le Leak ak fflo low wa and nd lloc ocat atio ion n

Sensor-Alarms Section (i)->(i+1)

 

Figure 11: 11: P ipePatrol ipePatrol E-RTTM/PC: pumping conditions, pressure at substations (Substation monitoring without flow by pressure 24

residuals) .

The example in Figure 11 shows a pipeline with pressure measurement at substations, an idea already introduced in Section 8.3 8.3.. The RTTM Pipeline Observer uses pressure a and nd temperature sensors at the head stations to calculate the local profiles, including the pressure profile along the pipe. Any discrepancy between the calculated and the observed pressure at the substations indicates a change in the pipeline dynamics: in other words, a leak. The pressure residual  at  at each station is:  z = p − pˆ i

i

i

≤i≤n , 1

These are used by the leak signature analysis anal ysis to detect a leak and find its location. This kind of pipeline monitoring is called substation monitoring. The Pipeline Observer compensates for any transient behaviour of the pipeline. The compensated residuals are now passed to the second stage, the leak signature analysis. Its task is to manage alarms for each individual substation. Results of leak signature analysis are combined in a Substation Evaluation stage, which groups substation alarms and determines leak flow rate and leak location.

  24

Two substations are shown for clarity. The method is able to handle any number of substations

Fundamentals of Leak Detection

KROHNE Oil & Gas

33

 

10

10.3.3

PipePatrol Extended Real-Time Transient Model (E-RTTM)

Segment monitoring for substations with flow measurement

ˆ&  ( I   M   I   

Q

&  M 

1)



 

ˆ

&  ( I   M  1  

1)

 



&  M   I 

Q

&  M 

ˆ (1

&1  M 

  →

2)

 

&  M  1

 

ˆ

&2  M 

 

&  M   I 

(  1→2 )

&ˆ (2   M  2  

Q

&  M 

&  M  2



O )

&ˆ (2   M  O  

 



O)

Q

&  M 

&  M  O

 

25

Figure 12: PipePatrol E-RTTM/PC: pumping conditions, PTF substations (segment monitoring by measured flow with flow residuals) .

The example in Figure 12 shows a special case where flow measurement is available at substations in addition to pressure and temperature. This configuration permits Segment Monitoring , where the pipeline is split into independent segments as shown in the diagram. Only two substations are shown for clarity, but the method can be scaled to cover as many as required. Independent RTTM Pipeline Observers and E-RTTM Leak Classifiers Classifi ers may be applied in parallel to every segment, each using methods in earlier sections. The shorter length of the monitored sections compared to the overallthe length of thealready pipelineintroduced leads to several advantages: •  Significantly lower smallest detectable leak rate •  Significantly shorter time to detect a leak •  Significant improvement in accuracy of leak location

The Segment Evaluation chooses the segment that shows sho ws the most significant leak signature, determines whether a leak alarm or a sensor alarm is present, and reports the leak location and flow if appropriate. This method achieves better performance then the method shown in Figure 11, especially on gas pipelines. The disadvantage is the complex instrumentation needed at the substations. KROHNE Oil and Gas is able to configure this method to assign pipeline segments dynamically. For example, in the event of a transmitter failure at substation 1 it is possible to skip this station and perform leak detection between the inlet station and substation 2.

25

Ground temperature is omitted for clarity

Fundamentals of Leak Detection

KROHNE Oil & Gas

34

 

10

10.3.4

PipePatrol Extended Real-Time Transient Model (E-RTTM)

Leak detection with substations and virtual flow monitoring

ˆ&  ( I   M   I   

Q

&  M 

ˆ  ( I 

&  M  1

 

1)



 

1)



ˆ (1

&1  M 

  →

2)

ˆ

(  1→2 )

&2  M 

 

&  M   I 

&  M   I 

&ˆ (2   M  2  

 



O)  

&ˆ (2   M  O  



O)

Q

&  M 

&  M  O

  Figure 13: PipePatrol E-RTTM/PC: pumping conditions, PT substations (segment monitoring by virtual flow with flow residuals).

The disadvantage of the complex and expensive instrumentation from chapter 10.3.3 10.3.3 can  can be eliminated by the virtual flow measurement, as shown in Figure 13. The functionality is nearly the same as shown in Figure 12, except that direct flow measurement is replaced by values calculated in the RTTM Pipeline Observer. Each RTTM Pipeline Observer calculates flow at every point along its associated segment, including the inlet and the outlet. The measured flow at the head stations is compared with the calculated flow in i n the usual way. At iintermediate ntermediate stations, the calculated outlet flow for the section upstream is simply compared with the calculated inlet flow for the section downstream. This method hasObserver no significant disadvantages to real flow measurement at intermediate stations. only one RTTM Pipeline calculates the flow atcompared the head stations, a real flow measurement is still needed at As these locations for comparison.

Fundamentals of Leak Detection

KROHNE Oil & Gas

35

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

ˆ&  ( I   M   I   

Q

&  M 

1)



ˆ

&  ( I   M  1  

 

1)



ˆ& (1

  →

 M 1

2)

&ˆ (2   M  2  

 

 

&ˆ (1  M 

  →

2

2)



O)  

&ˆ (2   M  O  



O)

Q

&  M 

&  M  O

&  M   I 

  Figure 14: PipePatrol E-RTTM/PC: pumping conditions, P substations (segment monitoring by virtual flow with flow residuals).

It is possible to take this approach even further. Figure 14 shows a pipeline with only pressure sensors at the substations. The functionality of configuration is nearly the same as shown in Figure 13, except that the temperature at substations is now calculated by an additional single RTTM Pipeline Observer. This single RTTM Pipeline Observer calculates temperature profile along the entire pipeline, using inputs from the head stations. In leak-free conditions, this gives exact e xact temperature values at every substation. In case of a leak, the dynamics of the pipeline are disturbed so in principle the calculated temperature values will show an error. As leak flow has little li ttle effect on pipeline temperature, these errors are negligible in practice. Using an RTTM Pipeline Observer, it is therefore possible not only to eliminate flow flo w measurement at substations but temperature measurements also. This is highly advantageous, as all intrusive process measurements are now eliminated. With the KROHNE Oil and Gas E-RTTM model, it is therefore possible to realise the advantages of Substation Leak Monitoring without compromising the ability to pig the pipeline from end to end.

Fundamentals of Leak Detection

KROHNE Oil & Gas

36

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

10.4 10. 4

PipePatro PipePatro l E-RTTM E-RTTM/SC /SC – Leak Detection in Shut-in Cond Cond iti ons

PipePatrol E-RTTM/SC is the abbreviated form of PipePatrol E-RTTM/Stand-Still or Shut-In Conditions. A model-based pressure-temperature method is used, which is valid for both bot h liquid and gas pipelines. In shut-in conditions conditio ns the pipeline is pressurised using pumps and valves, the fluid is sealed in the pipeline, and the pressure is monitored. The relevant valves must be leak-tight, and this should be considered when choosing them. 10.4.1

Head Station Monitoring

Inlet

Outlet

  F

P

T

T

T

T F,I  F,I 

T G,I->O   G,I->O

T FF,O ,O

RTTM (no leak) = Pipeline Observer

&  =0  M   I 

 p I  -

 pˆ O

 pˆ I 

 z I 

 

-

P

F

&   =0  M  O

 pO

zO

Leak Signature Analysis

Leak  Signatures

(Head-End Station)

Leak-Alarm

Leak flow and location

Sensor-Alarm

 

Figure 15: PipePatrol E-RTTM/SC: shut-in conditions, no substations - Head station monitoring

The simple case in Figure 15 shows leak monitoring in shut-in conditions. It is possible to deduce even without direct measurement thatincluding the flow the at the inlet andpressure outlet should RTTMItPipeline Obs Observer can use thiswith to calculate the local profiles, expected at the be twozero. headThe stations. is possible toerver compare these the measured values, giving pressure residuals z I and zO:  z I = pI − pˆ I

zO = pO − pˆ O  

The RTTM Pipeline Observer is able to compensate transient behaviour of the pipeline in shut-in co conditions. nditions. In addition, the equation of state in the RTTM compensates for temperature influence on pressure. The compensated residuals are passed to the E-RTTM Leak Signature Analysis, as introduced in chapter chapt er 10.3 10.3.. Leak rate and leak location will be determined when necessary.

Fundamentals of Leak Detection

KROHNE Oil & Gas

37

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

&    M   I 

=

&    M  O

0

 p I 

 pˆ I 

 p1

 pˆ1

 pˆ 2

 p2

 pˆ O

=

0

 pO

  Figure 16: PipePatrol E-RTTM/SC: shut-in condition, P substations 26 

(Substation monitoring with pressure residuals) .

Figure 16 shows a pipeline with substations. The pressure profile for the entire pipe is cal calculated culated using one RTTM Pipeline Observer. Calculated pressures can then be compared comp ared with measured values at the head stations and the substations. If the relevant valves are completely leak tight, even very small, gradual leaks will be recognised. In this case, PipePatrol PipeP atrol E-RTTM/SC meets the requirements of TRFL (d).

26 

 Only two substations are shown for clarity. The method is able to handle as many substations as required.

Fundamentals of Leak Detection

KROHNE Oil & Gas

38

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

Summary

 An overview of functionality and requirements is given in the following table (see Chapter 11 for 11 for more details). Instrumentation Method

Function Complexity

Demands

PipePatrol E-RTTM, KROHNE Oil & Gas

PipePatrol E-RTTM/PC Head station monitoring

LD+LL

2 x (Q,P,T) TG 

Medium

PipePatrol E-RTTM/PC Substation monitoring

LD+LL

2 x (Q,P,T) TG  nxP

Medium

PipePatrol E-RTTM/PC Segment monitoring

LD+LL

2 x (Q,P,T) TG n x (Q,P,T)

Medium

PipePatrol E-RTTM/PC Segment monitoring by virtual flow

LD+LL

2 x (Q,P,T) TG nxP

Medium

PipePatrol E-RTTM/SC Head station monitoring

LD+LL

2 x (P,T) TG 

Medium

PipePatrol E-RTTM/SC Substation monitoring

LD+LL

2 x (P,T) TG  nxP

Medium

27

Table 18: Functionality and instrumentation of PipePatrol E-RTTM  .

 All versions of PipePatrol-E-RTTM in pumping conditions (PipePatrol E-RTTM/PC) need measurements of flow, temperature and pressure at the head stations. Ground temperature is reasonably constant along the pipeline, so a single representative measurement at some point along the pipeline is sufficient. All versions of PipePatrol-E-RTTM in shut-in conditions (PipePatrol E-RTTM/SC) need measurements of temperature and pressure at the head stations. PipePatrol-E-RTTM provides leak detection and leak location l ocation in both pumping condition and shut-in condition. Use of statistical methods reduces the demands on instrumentation to medium. In particular, the absolute accuracy of the instruments does not matter. Possible fields of application are:  Ap pl ic ati on Method

Medium

TRFL

Steady Transient

L/G

(a) (b) (e)

PC

Steady Transient

L/G

(a) (b) (e)

PipePatrol E-RTTM/PC Segment monitoring

PC

Steady Transient

L/G

(a) (b) (e)

PipePatrol E-RTTM/PC Segment monitoring by virtual flow

PC

Steady Transient

L/G

(a) (b) (e)

PipePatrol E-RTTM/SC Head station monitoring

SC

Steady Transient

L/G

(c) (d) (e)

PipePatrol E-RTTM/SC Substation monitoring

SC

Steady Transient

L/G

(c) (d) (e)

Pumping

Dynamics

PipePatrol E-RTTM/PC Head station monitoring

PC

PipePatrol E-RTTM/PC Substation monitoring

28

Table 19: Possible fields of application of PipePatrol E-RTTM  .

27 

 LD = Leak Detection, LL = Leak Location.

28 

 PC = Pumping Conditions, SC = Shut-in Conditions.

Fundamentals of Leak Detection

KROHNE Oil & Gas

39

 

10

PipePatrol Extended Real-Time Transient Model (E-RTTM)

PipePatrol E-RTTM/PC provides leak detection in pumping conditions; PipePatrol E-RTTM/SC provides leak detection in shut-in conditions. All versions are able to monitor steady state, and transient conditions. Even gas pipelines can be monitored without problems. PipePatrol E-RTTM/PC performs the following functions: •  TRFL a), one autonomous, continuous working systems, which can ca n detect leaks within steady state conditions •  TRFL b), one of these systems, or a third one, has to be able to detect leaks in transient conditions •  TRFL e), one system to detect the leak position.

PipePatrol E-RTTM/SC performs the following functions: •  TRFL c), one system to detect leaks in shut-in conditions. •  TRFL d), one system to detect gradual leaks, •  TRFL e), one system to detect the leak position.

The following table lists the associated performance parameters.

Method

T  R F  L 

Sensitivity

Leak Types

 Al arm Threshold Thre shold

Tim e to Det ect   Liquid

Ga Gas s

PipePatrol E-RTTM/PC Head station monitoring

Low

Very short

Medium

Both

PipePatrol E-RTTM/PC Substation monitoring

Low

Very short

Lo Long ng

Both

PipePatrol E-RTTM/PC Segment monitoring

Low

Very short

Short

Both

PipePatrol E-RTTM/PC Segment monitoring by virtual flow

Low

Very short

Short

Both

(c)

Low

Short

Long

(d)

Very low

Long

Very Long

(c)

Low

Short

Long

(d)

Very low

Long

Very Long

PipePatrol E-RTTM/SC Head station monitoring PipePatrol E-RTTM/SC Substation monitoring

Both Both

Table 20: Performance parameters of PipePatrol E-RTTM.

 All versions of PipePatrol-E-RTTM provide a sensitive alarm threshold, PipePatrol E-RTTM/SC even a very sensitive alarming threshold if required. PipePatrol E-RTTM/PC provides short inertia times to a leak for liqui liquid d pipelines. Time to detect a leak becomes longer for gas pipelines because of thevery dynamic ofdetect the fluid. Instead of substation monitoring, section monitoring should be used here, because it is much faster. If PipePatrol E-RTTM/SC is used to detect very slow gradual leaks, time to detect a leak will raise.  All versions of PipePatrol-E-RTTM are able to detect and locate sudden leaks as well as gradual leaks of sufficient size.

Fundamentals of Leak Detection

KROHNE Oil & Gas

40

 

 

Bibliography

Bibliography [API 1130]

 API 1130: Computational Pipeline Monitoring for Liquid Pipelines. American Petroleum Institute, 2002.

[API 1155]

 API 1155: Evaluation Methodology for Software Based Leak Detection Detection Systems.  American Petroleum Institute, 1995.

[API 2540]

 API 2540: Volume Correction Factors. American Petroleum Institute, 2002.

[Baehr]

Baehr, H. D.: Thermodynamik. Springer, 1996.

[Billmann]

Billmann, L.: Methoden zur Lecküberwachung und Regelung von Gasfernleitungen. Fortschrittsberichte VDI Reihe 8, VDI-Verlag.

[Bohl]

Bohl, W.: Technische Strömungslehre. Vogel-Verlag, 12. Auflage, 2002.

[Geiger/Werner/Matko]

Geiger, G., Werner, T., Matko, D.: Leak Detection and Locating – A Survey. Surve y. 35th   Annual PSIG Meeting, 15 October – 17 October 2003, 2003, Bern, Switzerland.

[Hancock]

Hancock, John C.; Wintz, Paul A.: Signal Detection Theory. McGraw Hill, 1966.

[Kay]

Kay, Steven M.: Fundamentals of Statistical Signal Processing, Volume 2. Prentice Hall, 1998.

[Krass/Kittel/Uhde]

Krass, W., Kittel, A., Uhde, A.: Pipelinetechnik. Verlag TÜV Rheinland, 1979.

[Kroschel]

Kroschel, K.: Statistische Nachrichtentheorie. Springer-Verlag, 1996.

[RFVO]

Rohrfernleitungsverordnung. In TRFL - Technische Regeln für Fernleitungen. CarlHeymanns-Verlag, 2003.

[TRFL]

TRFL - Technische Regeln für Fernleitungen. Carl-Heymanns-Verlag, 2003.

[Wald]

Wald, A.: Sequential Analysis, John Wiley and Sons, New York, 1947.

[Zhang]

Zhang, X. J.: Statistical Leak Detection in Gas and Liquid Pipelines. Pipes & P ipelines International, July – August 1993, p. 26 – 29.

Fundamentals of Leak Detection

KROHNE Oil & Gas

42

 

 

Definitions

Definitions  Ac cu rac y  API 1130  API 1155

 API 2540

Balancing Methods

Batch-Separation-Pig Coefficient of Compressibilit Compre ssibilit y (z) Compressibility CPM System Gradual Leak Density Meter Detection Detection Li mit Difference Method DRAG Drift Error threshold E-RTTM-System False Alarm Flexibility Flow Monitoring Flow Flow Computer Fluid Forward Processing Ga Gaussian ussian Distributi on Gradient-IntersectionMethod Head Stations Head Monitoring Head Stations Head Hypothesis Test Inlet Inventory KROHNE Oil & Gas LDS Leak Alarm Leak Flow Leak Rate Leak Signature Analysis

Fundamentals of Leak Detection

Criterion of API 1155. This especially concerns leak location: details of the leak location must be accurate. Computational Pipeline Monitoring for Liquid Pipelines, American Petroleum Institute. Covers Design, Implementation, Test and processing of . Evaluation Methodology for Software Based Leak Detection Systems, American Petroleum Institute. Helps to compare different ldss. Defines Sensitivity, Reliability,  Accuracy and Robustness. Volume Correction Factors, American Petroleum Institute. Describes relations to calculate density of common crude oil and its products (such as gasoline) from temperature and pressure.   Also Mass Balance. Ldss that use the conservation of mass for leak detection. detection. These are compensated and uncompensated Mass Balance as well (e.g. PipePatrol SMB), In the broader sense also RTTM- and E-RTTM-Systems as PipePatrol E-RTTM.  Device to separate batches from each other in multi-product pipeline.   Dimensionless coefficient representing the departure of a gas from “ideal” behaviour. A coefficient of one indicates ideal behaviour.  Attribute of a liquid, representing the rate of change of density with respect to pressure. It is compensated by (e)-rttm-systems. Computational Pipeline Monitoring system  A slowly developing leak, often with a very low Leak rate. TRFL demands special Systems to detect gradual leaks  A sensor that monitors the actual density of a fluid Theoretical Value of the smallest, detectable leak rate. Synonym for: Residual Method Drag Reducing Agent. Added to liquids to reduce pipe wall friction Very low frequency disturbance in measurements. Maximal or guaranteed difference of a measured value from its true value.   LDS based on Extended RTTM-Technology. This technology combines RTTMTechnology with the Leak signature analysis.  A leak alarm that is raised when no real leak is present  An attribute of the wall for a pipeline, representing the rate of change of cross section with respect to pressure. Can be compensated by (e)-rttm-systems  A simple LDS, where flow is measured at one single location at the pipeline; flow changes in case of a leak. Method is out of date. Collective term for mass flow (e.g. In kg/s), volume flow (e.g. In m 3/h), and velocity (e.g. In m/s).  Device used to “pre-process” field signals and apply calculations    A substance that is capable of flowing, including all gases and liquids.  Pumping the fluid from the inlet to the outlet of the pipeline. Such flow is counted as positive. See also reverse processing. Well-known form of probability density function, published by Gauss. Many different procedures in nature are described by this function (approximately).   Method for leak location, where leak location specific change in pressure press ure profile along the pipeline is analysed. PipePatrol E-RTTM uses a model based version, which is able to locate a leak even under transient condition.   This Configuration of PipePatrol E-RTTM uses measurements from the Head Stations of a pipeline to calculate the flow-residuals and monitor the pipeline for leaks. Related Substations Monitoring.  Metering Station at Inlet or Outlet  Method of the statistical decision theory   The "left handed" beginning of a Pipeline, where the Fluid enters the pipeline in forward operation. Related Outlet.   Also: line-packing.  Netherlands subsidiary of KROHNE Messtechnik Duisburg gmbh & Co. KG. PipePatrol is the LSS-Family of KROHNE Oil & Gas.  Leak Detection System. A System to detect leaks in pipelines, additional leak location can be detected, too. In Germany LDS have to make grade to the TRFL.  Declaration of a leak event. Related to E-RTTM-Technology this alarm will raise in case of a spontaneous leak.  Lost fluid at leak location in units per time, e.g. M/s, t/h or m 3/h.  Quantitative value about the size of the leak. Absolute e.g. In m/s, t/h or m 3/h. Relative in % related to a reference value.  Method to avoid False Alarms related to E-RTTM-Technology. Residuals are analysed for different Leak signatures. I case of a leak Leak Alarm or Sensor Alarm will raise.   43

 

 

Definitions

Leak Signature

Specific signature in signals, which occurs in case of a leak.  

Likelihood-Ratio-Test

Statistical method, to decide for one of two predefined Hypotheses (e.g. Leak no/yes) by a collection of measured values. 

Line-Packing

International common definition for the change of inventory stored in a pipeline.  

Local Profiles

Flowand thermodynamic Values as e.g. Pressure and temperature, described along the pipeline.  

Mass Balance Method

 Also Balancing Methods. LDS-method, which uses the conservation of mass mass for leak detection. These are compensated and uncompensated Mass Balance as well (e.g. PipePatrol SMB), In the broader sense also RTTM- and E-RTTM-Systems as PipePatrol E-RTTM 

Measured Mea sured Sectio n

Section of a Pipeline terminated by Metering Equipment. Is (maybe the only) part of a Measured Section. 

Measurement Mea surement Statio n

 A station at one specific point of the pipeline, equipped with with metering sensors and/or clusters. At In- and Outlet these are called Head Stations, otherwise Substations.  

Monitored Section

Part of a pipeline, monitored by a LDS. Consists of one ore more Monitoring Sections.  Sections. 

Multi-Batch-Condition

Condition, where several different Batches are pumped through a pipeline. Related Single-Batch-Condition. 

Negative Nega tive Pressur e Drop

LDS method, where with Speed of sound propagating, negative pressure wave is analysed. If several pressure meters are used, leak location can be determined by Wave propagation method. Detects only spontaneous Leaks  

Nominal Flow

Flow at nominal conditions, e.g. In m 3/h. 

Nominal Flow

Flow, given at nominal conditions, e.g. In m 3/h. 

Outlet

The "right handed" end of a Pipeline, where the Fluid leaves the pipeline in forward operation. Related Inlet. 

Pig

System which is inserted into a pipeline on demand e.g. Separate of batches (Batch Separation Pig). 

Pipeline-Observer

SW-Module, which calculates Residuals based on measured values. Is used to eliminate Compressibility and Elasticity-Effects. 

Pipelin

Serve the purpose to transport Fluids.

PipePatrol E-RTTM

PipePatrol Extended Real-Time Transient Model; LDS by KROHNE Oil & Gas based on E-RTTM-Technology and Leak Signature Analysis. Provides Leak detection and location in Pumping Conditions (PipePatrol E-RTTM/PC) and Shut-in Condition (PipePatrol ERTTM/SC) for stationary state and transient state defined in API 1130 and TRFL 

PipePatrol SMB

PipePatrol Statistical Mass Balance; LDS by KROHNE Oil & Gas, based on the mass balance method and Leak-Classification. Provides continuous leak detection in Pumping Conditions for stationary state and limited transient state as defined in API 1130 and TRFL. 

PipePatrol

Family of LDS by KROHNE Oil & Gas 

Pressure Monitoring

Simple LDS, where pressure is measured at one single location at the pipeline; Pressure drops in case of a leak. Method is out of date.  

Probability Density Function

Function of probability theory, which enables calculation of probability for an event. Related Gaussian Distribution. 

Process-Conditions

 Actual conditions for pressure and temperature. Also values for density and (ProcessDensity) and flow (Process-Flow) belong to process conditions. These values differ from the values at Reference Conditions in general.  

Process-Density

Density of the fluid at Process-Conditions.  

Process-Flow

Flow of the fluid at Process-Conditions.  

P-T-Method

Pressure-Temperature-Method. In this method pressure within a tightly closed pipeline is analysed. The temperature is analysed by use of the thermodynamic equation law of teh fluid. PipePatrol E-RTTM/SC uses atoo model based version of this method.  

Fundamentals of Leak Detection

44

 

 

Definitions

Pumping Condition

Condition where fluid is pumped through the pipeline. TRFL assumes special LDS for this condition 

Reference-Conditions

Defined condition for pressure (e.g.. 1,01325 bar) and temperature (e.g. 15°C Also values for density and (Process-Density) and flow (Process-Flow) belong to process conditions. These values differ from the values at Process-Conditions in general.  

Reference-Density

Density of the fluid at Reference-Conditions 

Reference-Flow

Flow of the fluid at Reference-Conditions 

Reliability

Criteria by API 1155, e.g. Probability for a false alarm, most times related to one year.  

Residual

Difference in measured values (e.g. Pressure or flow) of by Pipeline-Observer calculated values assuming a leak free pipeline. Input for Leak Signature Analysis.  

Residual-Method

 A RTTM-based Method for leak detection, where where redundant measurement values are analysed. The measured values are compared to the calculated ones; their difference is called Residual. 

Reverse Processing

Processing, where the Fluid is pumped from Out- to Inlet. Per definition flow will be counted as negative in this case. Related Forward Processing.  

Robustness

Criteria by API 1155; Defines the processing of a LDS, if conditions are not ideal, e.g. The damage of a sensor  

RTTM-System

Real Time Transient Model System, International common for a real-time capable model based LDS. The mathematical model of the flow within the pipeline is simulated si mulated on

Section Section Monitoring

industry computers in real-time. It is capable to handle even transient conditions.   In this configuration of PipePatrol E-RTTM leak detection based on flow residual is processed for single Sections. Related Substation Monitoring.

Section

Synonym: monitoring section 

Section-Residual

The Flow Residual calculated by Section Monitoring of one monitored Section.  

Sensitivity

Criteria by API 1155; combined criteria, combines smallest detectable leak rate as well as time to detect a leak. Example: Lost Volume by leak rate from beginning till Leak Alarm.  

Sensor Alarm

Declaration of a leak event. Related to E-RTTM-Technology this alarm will raise in case of a gradual leak or Sensor Drift.  

Sequential Probability Ratio Test (SPRT)

Published by Wald, a sequential version of the Likelihood-Ratio-Test. Statistical base of PipePatrol SMB. 

Single-Batch-Condition

Process-Condition, where only one single batch is pumped through a pipeline. Related Multi-Batch-Condition. 

Speed of Sound

Propagation Speed of Pressure-, Density- and flow dynamics in Fluids. Leak Location by Wave Propagation method is based on the speed of sound. 

Spontaneous Leak

 A fast developing, step-like leak. 

Shut-in Condition

Condition, where pumps are switched off. TRFL assumes special LDS for this condition.  

Shut-in Conditions

Condition, where no fluid is pumped through the pipeline. TRFL makes demand for a LDS capable of monitoring shut-in conditions. 

Stationary

In statistical manner the time independence of the probability density function, e.g. Gaussian Distribution. 

Statistical LDS

LDS equipped with special statistical data processing e.g. Hypotheses Test (Leak no/yes) as SPRT. 

Steady State

Process Conditions for a pipeline, where physical values (e.g. Pressure) do NOT change over time. TRFL makes demand for LDS, which are capable to monitor pipelines for leaks under steady state conditions. Related transient condition. 

Substation Monitoring

This Configuration of PipePatrol E-RTTM uses measurements from the Substations of a pipeline to calculate the flow-residuals and monitor the pipeline for leaks. Substation Monitoring is proceeded additional to Head Station Monitoring. Related Section

Substations.

Fundamentals of Leak Detection

Monitoring.   All Measurement Stations except the In- and Outlet ones ones 

45

 

 

Definitions

Thermodynamic Equation Law

Relation between Pressure, temperature and density (or a spec. Volume), which is true for a Fluid  

Transient Transient Condition

Process Conditions for a pipeline, where physical values (e.g. Pressure) do change over time. TRFL makes demand for LDS, which are capable to monitor pipelines for leaks under steady state conditions. Related steady state 

TRFL

TRFL stands for “Technische Regel für Rohrfernleitungen”. Published in 2003 in Germany itChapter is applied toof allthe Pipelines, which transport dangerous liquids or gases. 11.5 TRFL instructs LDS areflammable necessary and/or for a pipeline.  

Volume Correction Coefficient

Coefficient describing the relation between Process-Density to pressure and temperature.

Volumetric Flow Measurement

Measurement, where Flow is measured as volume flow or flow rate. To determine mass flow the density of the fluids needs to be known.  

Wave Propagation Propagation Method

Method to locate a leak, where the difference in runtime of a wave-like propagating pressure drop is analysed at different locations along the pipeline. PipePatrol E-RTTM uses a model based version, which is able to locate a leak even under transient condition.  

Way through

Missing alarm in case of a leak  

Zero False-Alarm Methodology

Synonym: e-rttm-technology. 

Fundamentals of Leak Detection

46

 

Fundament als of Leak Detection

KROHNE Oil & Gas

 

From the well head, through massive pipelines, onto tankers and into the terminals and reneries; the ow of oil and gas products needs to be measured accurately and reliably. That is the world of KROHNE Oil & Gas. The scope of KROHNE Oil and Gas starts with custody transfer ow metering for oil, gas and liqueed gas and continues through tank management, loading and ofoading and leak detection and localisation systems. KROHNE Oil & Gas is one of the most important Companies in the KROHNE Group. KROHNE Oil & Gas's headquarters are located in Breda, the Netherlands, close to Europe’s major Oil and Gas centres.  

We have grown dynamically and now have over 120 engineers solely dedicated to the oil and gas industry. industry. KROHNE Oil & Gas now has 6 manufacturing facilities in the Netherlands, UK, Malaysia, USA, Brazil, Colombia, Middle East. The headquarters in Breda services the world’s oil industry through its own ofces and through the KROHNE group, in more than 60 countries worldwide. The parent company, KROHNE, has 42 owned subsidiaries and more than 45 representatives throughout the world. We make use of this network network to maintain a high level of service service for our customers. KROHNE Oil & Customer provide specialised knowledge rsthand with the backing of the world’s most knowledgeable concern in the eld of ow measurement technology. technology.

KROHNE Oil & Gas Overview Systems • Flow Meters for Custody Transfer • Liquid Flow Metering Systems • Gas Flow Metering Systems • Wet Gas Metering Systems • Provers & Master Meters • Flow Computing, Supervisory Software & Analyser Management • Calibration Systems • Tank Inventory & Management Systems • Analyser Houses and Shelters • Loading & Ofoading Systems • Leak Detection and Localisation Systems • Revamps & Upgrades

Products • Gas Utrasonic Flow Meters for Custody Transfer • Liquid ULtrasonic Flow Meters for Custody Transfer • Mass Flow Meters for Custody Transfer • Venturis for Wet Gas Metering • Flow Computers • Supervisory Systems • Meter Validation Software Packages • Electromagn Electromagnetic etic Flowmeters • Level Measuring Instruments • Variable Area Flowmeters • Temperature Measuring Instruments Pressure e Measuring Instruments • Pressur • Analysis

• esting, Installation, Commissioning, Service • T Training

• Flowmeters • Vortex Flow Controllers

KROHNE Oil & Gas Minervum 7441 · 4801 LH Breda · The Netherlands Tel.: +31-76.711 +31-76.711.2000 .2000 · Fax.: +31-76.71 +31-76.711.2001 1.2001 [email protected]

www.krohne-oilandgas.com

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