EPA/600/R-12/017 | March 2012 | www.epa.gov /nrmrl
United States
Environmental Protection
Agency
Condition Assessment Technologies
for Water Transmission and
Distribution Systems
Office of Research and Development
National Risk Management Research Laboratory -Water Supply and Water Resources Division
-------
Condition Assessment Technologies for
Water Transmission and Distribution Systems
by
Zheng Liu, Yehuda Kleiner, and Balvant Rajani
Institute for Research in Construction
National Research Council Canada
Ottawa, Ontario K1A OR6 Canada
and
Lili Wang and Wendy Condit
Battelle
Columbus, OH 43201
Contract No. EP-C-05-057
Task Order No. 0062
for
Michael Royer
Task Order Manager
Urban Watershed Management Branch
Water Supply and Water Resources Division
National Risk Management Research Laboratory
2890 Woodbridge Avenue (MS-104)
Edison, NJ 08837
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
March 2012
-------
DISCLAIMER
The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development,
funded and managed, or partially funded and collaborated in, the research described herein under Task
Order (TO) 0062 of Contract No. EP-C-05-057 to Battelle. It has been subjected to the Agency's peer
and administrative review and has been approved for publication. Any opinions expressed in this report
are those of the author(s) and do not necessarily reflect the views of the Agency, therefore, no official
endorsement should be inferred. Any mention of trade names or commercial products does not constitute
endorsement or recommendation for use. The quality of secondary data referenced in this document was
not independently evaluated by EPA and Battelle.
-------
ABSTRACT
As part of the U.S. Environmental Protection Agency's (EPA's) Aging Water Infrastructure Research
Program, this research was conducted to identify and characterize the state of the technology for structural
condition assessment of drinking water transmission and distribution systems. The broad definition of
structural condition assessment of water mains encompasses physical modeling of the pipe in the soil,
understanding of pipe failure modes, empirical/statistical modeling of historical failures, inspection of a
pipe to discern distress indicators, interpretation of distress indicators into pipe condition rating and
modeling deterioration to forecast future failures and pipe residual life.
Any asset management program must start with a thorough review of available historical data about pipe
performance and failure. Once the necessary data is gathered, deterioration models can go a long way in
providing insight into the condition of these assets. A well-defined and cost-effective inspection program
that complements the historic data can then be used to fill in gaps that remain. This report provides a
comprehensive inventory of both condition assessment technologies and decision support systems applied
to water mains and identifies capability gaps that need to be addressed. A comprehensive list is provided
of existing non-destructive evaluation technologies and techniques that are currently used for buried pipes
or that have the potential of being adapted to pipe inspection. Scientific principles, advantages, and
limitations of each technique are described. A review is also provided of physical models,
statistical/empirical models, and decision support software tools available to facilitate water main renewal
decisions.
To date, there has been a substantial amount of work and effort that has been invested in developing
approaches and tools for the condition assessment of water mains. However, there are still a number of
technology gaps and research needs including: the need for live internal insertion and retrieval of
inspection tools; the need to assess joint condition in metallic pipes; the need to develop technologies for
asbestos cement and plastic pipes with few options currently available; and the need for low cost
inspection methods to conduct screening for high risk locations in all pipe types for further assessment.
To overcome the barriers and challenges identified in this research, field demonstrations and further
research efforts are warranted in order to test promising technologies that could fill these gaps against
well defined performance criteria and to identify the critical performance, cost, and/or value added
attributes of emerging and innovative technologies for water main inspection.
in
-------
EXECUTIVE SUMMARY
The aging of water mains, coupled with the continuous stress placed on these systems by operational and
environmental conditions, has led to their deterioration. This deterioration can be classified into two
categories: (1) structural deterioration, which diminishes the structural resiliency of the pipes and their
ability to withstand various types of stress, and (2) deterioration of pipe inner surfaces, resulting in
diminished hydraulic capacity, degradation of water quality and even diminishing structural resiliency in
cases of severe internal corrosion. This deterioration manifests itself in the following ways:
• Increased rate of pipe breakage due to deterioration in pipe structural integrity. This, in turn,
causes increased operation and maintenance (O&M) costs, increased loss of (treated) water
and social costs such as property damage, loss of service, disruption of traffic, disruption of
business and industrial processes, disruption of residential life, public safety hazard, and loss
of landscape vegetation. In addition, pipe breakage events increase the risk of water quality
failure through intrusion of contaminants into the system.
• Decreased hydraulic capacity of pipes in the systems, which results in increased energy
consumption and disrupts the quality of service to the public. This includes drinking water as
well as fire extinguishing needs.
• Deterioration of water quality in the distribution system due to the condition of inner surfaces
of pipes, which may result in taste, odor, and aesthetic problems in the supply water and even
public health problems in extreme cases.
The structural deterioration of water mains and their subsequent failure are complex processes, which are
affected by many factors, both static (e.g., pipe material, size, soil type) and dynamic (e.g., age, climate,
cathodic protection, pressure zone changes). The physical mechanisms that lead to pipe breakage are
often very complex and not completely understood. The facts that most pipes are buried, and that
relatively little data are available about their breakage modes also contribute to this incomplete
knowledge.
It appears that while physical modeling of the structural deterioration of water mains may be scientifically
more robust, it is, to date, limited by existing knowledge and available data. Some of the data that are
required for the physical models can be obtained albeit at significant costs. These costs may currently be
justified only for major transmission water mains, where the consequences of failure are significant. In
contrast, statistically-derived empirical models can be applied with various levels of input data and may
therefore be useful for small diameter water mains for which low cost of failure does not justify expensive
data acquisition campaigns. The statistical analysis of breakage patterns of water mains has thus been a
cost effective way to model this deterioration, particularly when available data are scarce. However, this
effectiveness is higher at high-level planning (i.e., regional or network level) and diminishes to a certain
degree when applied to individual water mains. Information on the current structural condition of the
individual water main, combined with good understanding of failure modes and deterioration models, will
greatly enhance the ability of water utilities to manage these assets in a cost-effective manner.
Task Order 0062 (TO 0062) of the Environmental Protection Agency (EPA) contract addresses the
condition assessment of installed drinking water transmission and distribution mains. This report was
prepared under Task 2.2 of TO 0062, which focused on the assessment of structural condition (not
hydraulic capacity and water quality aspects) of pipes. The definition can vary to encompass different
elements. The broad definition of structural condition assessment of water mains encompasses physical
modeling of the pipe in the soil, understanding of pipe failure modes, empirical/statistical modeling of
IV
-------
historical failures, inspection of pipe to discern distress indicators, interpretation of distress indicators into
pipe condition rating and modeling deterioration to forecast future failures and pipe residual life. This
report focuses on direct and indirect inspection of pipes to discern their structural condition, interpretation
of distress indicators into condition rating and modeling structural deterioration to forecast future failures
and decision making about pipe renewal. The report covers extensively cast and ductile iron pipes, pre-
stressed concrete cylinder pipes (PCCP), asbestos cement (AC) and polyvinyl chloride (PVC) pipes.
However, some of the described technologies apply also to welded steel (WS), glass-fiber reinforced
polyester (GRP), concrete pressure pipe (CPP), un-plasticized polyvinyl chloride (uPVC), and
polyethylene (PE) pipes.
The report describes a comprehensive inventory of technologies, techniques, and methods that are
actually or potentially employed in the field of condition assessment of water mains, as follows:
• Section 1 describes the objective and scope of the task, as well as provides some background
information.
• Section 2 provides a primer on general issues related to the deterioration of buried pipes,
including distress indicators, known modes of failure and a general introduction to the classes
of nondestructive evaluation (NDE) technologies and methods to discern distress indicators
leading to failure.
• Section 3 provides a comprehensive list of existing NDE technologies/techniques that are
currently being used for buried pipes or that have the potential of being adapted to pipe
inspection. Each technology/technique is provided with a description of scientific principles,
advantages, and limitations. Where available, data on the breadth and manner of usage of the
inspection technologies are presented.
• Section 4 provides a comprehensive description of computational methods used to translate
inspection data (or discerned distress indicators) into pipe condition rating.
• Section 5 provides a comprehensive compilation of mathematical models that have been
proposed in the literature to model the deterioration of buried water mains. These include
both physical/mechanistic models and statistical/empirical models.
• Section 6 provides a comprehensive compilation of mathematical models intended to support
decisions related to the renewal planning of water mains. This includes theoretical models
from the literature as well as brief descriptions of currently available decision support
software tools.
• Section 7 identifies current technological gaps requiring further research between desired and
available capabilities of inspection techniques.
• Section 8 provides a summary and concluding remarks.
• Section 9 presents the references cited throughout the document.
-------
ACKNOWLEDGEMENTS
During the course of this study, the following individuals are acknowledged for their kind assistance (in
alphabetic order):
Advanced Engineering Solutions, Ltd.
AECOM
Echologics Engineering, Inc.
EPCOR Water Services
Fiber Optic Systems Technology, Inc. (FOX-TEK)
Halifax Regional Water Commission
Halma Water Management
NOT Corporation
NRC Institute for Research in Construction
NRC Industrial Material Institute
Pure Technologies, Ltd.
RapidView IBAK North America
Russell NDE Systems, Inc.
The City of Edmonton
The Pressure Pipe Inspection Company, Ltd.
Virginia Tech University
Craig Johnson, Malcom Wayman
Garry Mak
Chris Gates, Marc Bracken
Doug Seargeant
Sam Cauchi
Jammie Hannam
Vincent Favre
Paul Fisk
Dennis Krys, Yafei Hu, Osama Hunaidi
C.K. Jen
Chris Carroll
Matthew W. Sutton
Vincent Shen
Ken Chua
Xiangjie Kong, Hugh Leavens
Sunil Sinha and his research team
VI
-------
CONTENTS
DISCLAIMER ii
ABSTRACT iii
EXECUTIVE SUMMARY iv
ACKNOWLEDGEMENTS vi
FIGURES ix
TABLES x
ACRONYMS AND ABBREVIATIONS xii
1.0: INTRODUCTION 1
1.1 Objective, Scope, and Background 1
1.2 Organization of this Report 4
2.0: PIPE DETERIORATION, DISTRESS INDICATORS AND FAILURE MODES 5
2.1 Overview of Distress Indicators and Failure Modes 5
2.2 Distress Indicators for Major Pipe Types 6
2.3 Inferential Indicators for Major Pipe Materials 9
3.0: TECHNOLOGIES FOR CONDITION ASSESSMENT OF WATER MAINS 14
3.1 Nondestructive Testing and Evaluation 14
3.2 Pit Depth Measurement 16
3.3 Visual Inspection 16
3.3.1 Man Entry and Visual Inspection 17
3.3.2 Closed Circuit Television Inspection 17
3.3.3 Videoscope 19
3.3.4 3D Optical Scanning 20
3.3.5 Laser-Based Pipe Surface Profiling 21
3.3.6 Handyscan3D 21
3.4 Electromagnetic Inspection 23
3.4.1 Magnetic Flux Leakage 23
3.4.2 Remote Field Eddy Current 25
3.4.3 Broadband Electromagnetic 28
3.4.4 Pulsed Eddy Current System 30
3.4.5 Ground Penetrating Radar 31
3.4.6 Ultra-Wideband Pulsed Radar System: P-Scan 32
3.5 Acoustic Inspection for Structural Condition 32
3.5.1 Sonar Profile System 32
3.5.2 Impact Echo 33
3.5.3 Acoustic Emission 35
3.6 Acoustic Inspection for Leak Detection 36
3.6.1 SmartBall® 36
3.6.2 LeakfmderRT™ 38
3.6.3 Permalog® 40
3.6.4 MLOG™ 41
3.6.5 STARZoneScan™ 42
3.6.6 Sahara® 43
3.7 Ultrasonic Testing 44
3.7.1 Guided Wave Ultrasonic Testing 44
3.7.2 Discrete Ultrasonic Measurement 46
3.7.3 Phased Array Technology 47
vn
-------
3.7.4 Combined UT Inspection 48
3.8 Seismic Pulse Echo 49
3.9 Pipeline Current Mapper 50
3.10 Radiographic Testing 50
3.11 Thermographic Testing 52
3.12 Using Soil Properties to Infer Pipe Condition 52
3.12.1 Linear Polarization Resistance of Soil 52
3.12.2 Soil Characterization 53
3.12.3 Pipe to Soil Potential Survey 55
3.13 Emerging Sensor Technologies and Sensor Networks 55
3.13.1 Corrosion Rate Sensor 55
3.13.2 Magnetostrictive Sensor 57
3.13.3 Conformable and Flexible Eddy Current Array 59
3.13.4 Flexible Ultrasonic Transducer 61
3.13.5 Damage Sensor 61
3.13.6 Microwave Back-Scattering Sensor 63
3.13.7 Fiber Optic Sensor for Corrosion Monitoring 64
3.13.8 Fiber Optic Acoustic Monitoring Network 65
3.13.9 Wireless Sensor Network for Pipe Condition Monitoring 68
3.13.10 Multi-Sensor Approaches 69
3.13.11 Smart Pipe 69
3.14 Additional Leak Detection and Monitoring Methodologies 70
3.14.1 Hydraulic Transient-Based Methods 70
3.14.2 Measurement-Based Leak Monitoring Methods 71
3.14.3 Model-Based Leak Monitoring Methods 72
3.14.4 Information Fusion 73
3.15 Supplemental Information on Inspection Platforms, Intelligent Pigs, and Robotic
Survey Systems 74
3.15.1 Computer-Aided Approach: Augmented Reality 74
3.15.2 Intelligent Pigs and Robotic Survey Systems 74
3.16 Current Use of NDE and Condition Assessment 76
3.16.1 NRC Survey 76
3.16.2 Virginia Tech Survey 78
3.16.2.1 EPCOR Water Services Inc 79
3.16.2.2 Las Vegas Valley Water District 80
3.16.2.3 Newport News Waterworks 81
3.16.2.4 Seattle Public Utilities 82
3.16.2.5 Sydney Water 83
3.16.2.6 Washington Suburban Sanitary Commission 83
3.16.2.7 City of Hamilton Public Works Department 85
3.16.2.8 Louisville Water Company 85
3.16.2.9 Philadelphia Water Department 85
4.0: FROM DISTRESS/INFERENTIAL INDICATORS TO CONDITION RATING 86
4.1 Background 86
4.2 Point-Score Protocols for Sewers 88
4.3 Fuzzy Theory Based Techniques 89
4.3.1 Fuzzy Synthetic Evaluation 89
4.3.2 Fuzzy Composite Programming 91
4.4 Data Fusion and Data Mining 92
4.4.1 Hierarchical Evidential Reasoning 92
Vlll
-------
4.4.2 Incremental Learning 93
4.4.3 Genetic Algorithm 93
4.5 Data Driven Approaches to Predict Condition Rating Based Only on Inferential
Indicators 93
5.0: WATER MAIN DETERIORATION MODELS 95
5.1 Physical/Mechanistic Models 95
5.2 Statistical/Empirical Models 99
6.0: DECISION SUPPORT FOR WATER TRANSMISSION AND DISTRIBUTION SYSTEMS 104
6.1 Decision Support Models 104
6.2 Publicly Available Decision Support Software Tools 108
6.2.1 Computer Aided Rehabilitation of Water Networks (CARE-W) 108
6.2.2 KANEW 110
6.2.3 Pipeline Asset and Risk Management System (FARMS) 110
6.2.4 Pipe Rehabilitation Management (PiReM) Ill
6.2.5 Water Main Rehabilitation Planner (WARP) Ill
6.2.6 WilCO 112
7.0: TECHNOLOGY GAPS AND RESEARCH AND DEVELOPMENT NEEDS 113
8.0: SUMMARY AND CONCLUDING REMARKS 117
9.0: REFERENCES 120
FIGURES
Figure 1-1. Schematic for Inspection, Condition Assessment, and Failure Risk Evaluation of
Pipes 3
Figure 2-1. Factors Contributing to Water System Deterioration 5
Figure 3-1. Nondestructive Inspection Technologies for the Condition Assessment of Water
Mains 15
Figure 3-2. CCTV Inspection 18
Figure 3-3. The PANORAMO® System 19
Figure 3-4. PANORAMO® SI 3D Optical Manhole Scanner 20
Figure 3-5. Creaform Handyscan 3D 22
Figure 3-6. The Principle of Magnetic Flux Leakage Inspection 24
Figure 3-7. Remote Field Eddy Current Testing 26
Figure 3-8. The Breaking Wire Results in a Decrease in the Detector Signal 26
Figure 3 -9. The See Snake Tool for Inspection of Pipe Internal and External Flaws 27
Figure 3-10. P-Wave® System for Manned and Robotic PCCP Inspection 28
Figure 3-11. Impact Echo Testing 34
Figure 3-12. Pictures and Illustrations of SmartBall® 36
Figure 3-13. Principle of LeakFinderRT™ 38
Figure 3-14. Picture of Permalog® 40
Figure 3-15. Picture of MLOG™ 41
Figure 3-16. Picture of STAR™ Zone Scan™ 42
Figure 3-17. The Sahara System 43
Figure 3-18. Continuous and Discrete Ultrasonic Measurement 46
Figure 3-19. Sound Beams Generated by Phased Array of Composite Sensor Elements 48
-------
Figure 3-20. Radiographic Testing 51
Figure 3-21. Corrater® Acuamate™ Portable Instrument with Soil Corrosion Rate Probe 53
Figure 3-22. Picture of Corrosion Rate Senor with Embedded Metallic Coupon 56
Figure 3-23. Pipe Inspection with MsS 57
Figure 3-24. The MsS System for Pipe Corrosion Monitoring 58
Figure 3-25. (a) Inspection of Pipeline with Flexible Eddy Current Array, (b) the Sensor Array,
and (c) Samples of Inspection Results 60
Figure 3-26. The Flexible Ultrasound Transducer Array 62
Figure 3-27. FOX-TEK Coil Sensor 64
Figure 3-28. Monitoring Corrosion and Bending of Pipelines with Fiber Optic Sensors 65
Figure 3-29. (Left) Installation of Fiber Optic Sensor in a Dewatered Pipeline. (Right) Fiber Optic
Sensor Installed on a Stainless Steel Hoop with a Strain Relief Device 66
Figure 3-30. (Left) Installation Parachute Used to Install an AFO Cable in an in-Service Pipeline
and (Right) Parachute Caught and Extracted at Two Miles Downstream from
Insertion Point 66
Figure 3-31. The System Architecture of PipeNET 68
Figure 3-32. Summary of the Technologies for Leak Detection and Monitoring 70
Figure 3-33. Sensor Data Fusion for Leak Detection 73
Figure 3-34. The Mileages of RFEC-TC Inspection for PCCP 77
Figure 4-1. Optimal Renewal Frequency for Distribution Mains (top) versus Transmission Mains
(bottom) 87
Figure 4-2. Pipe Condition Assessment Indicators 91
Figure 4-3. The Framework of Hierarchical Evidential Reasoning 93
TABLES
Table 2-1. Distress Indicators that Influence Pipe Condition for Cast and Ductile Iron Pipes 6
Table 2-2. Distress Indicator that Influence Pipe Condition for PCCP Water Mains 7
Table 2-3. Distress Indicators that Influence Pipe Condition for AC Pipes 8
Table 2-4. Distress Indicators that Influence Pipe Condition for PVC Pipes 9
Table 2-5. Inferential Indicators for Cast and Ductile Iron Pipes 10
Table 2-6. Inferential Indicators for PCCP 11
Table 2-7. Inferential Indicators for AC Pipes 12
Table 2-8. Inferential Indicators for PVC Pipes 13
Table 3-1. Summary of Condition Assessment Technologies Applicable to Different Pipe
Materials 14
Table 3-2. Template Used for Description of Technologies 15
Table 3-3. Pit Depth Measurement 16
Table 3-4. Man Entry and Visual Inspection 17
Table 3-5. Closed Circuit Television Inspection 18
Table 3-6. Videoscope 20
Table 3-7. 3D Optical Scanning 21
Table 3-8. Laser-Based Pipe Surface Profiling 22
Table 3-9. Handy scan 3D 23
Table 3-10. Magnetic Flux Leakage 24
Table 3-11. Remote Field Eddy Current 25
Table 3-12. Broadband Electromagnetic 29
Table 3-13. Pulsed Eddy Current System 30
Table 3-14. Ground Penetrating Radar 31
Table 3-15. UWB Pulsed Radar System: P-Scan 32
-------
Table 3-16. Sonar Profile System 33
Table 3-17. Impact Echo 34
Table 3-18. Acoustic Emission 35
Table 3-19. SmartBall® 37
Table 3-20. LeakfinderRT™ 39
Table 3-21. Permalog® 41
Table 3-22. MLOG™ 42
Table 3-23. Sahara® Leak Detection and Sahara® Condition Assessment 44
Table 3-24. Guided Wave Ultrasonic Testing 45
Table 3-25. Discrete Ultrasonic Measurement 47
Table 3-26. Phased Array Technology 48
Table 3-27. Seismic Pulse Echo 49
Table 3-28. Radiographic Testing 51
Table 3-29. Thermographic Testing 52
Table 3-30. Comparison of Soil Corrosivity Rating Approaches Based on Soil Properties 54
Table 3-31. Corrosion Rate Sensor (Probe) 56
Table 3-32. Magnetostrictive Sensor 59
Table 3-33. Conformable and Flexible Eddy Current Array 61
Table 3-34. Flexible Ultrasonic Transducer 62
Table 3-35. Damage Sensor 63
Table 3-36. Microwave Back-Scattering Sensor 64
Table 3-37. Fiber Optic Sensor Corrosion Monitoring 65
Table 3-38. Fiber Optic Acoustic Monitoring Network 67
Table 3-39. Wireless MEMS Sensor Network 68
Table 3-40. Multi-Sensor Technology 69
Table 3-41. Robot Systems for Pipe Inspection 74
Table 3-42. Use of Condition Assessment Technologies for Water Mains by Five Water Utilities 78
Table 3-43. Summary of Utility Inspection Methods and Models 79
Table 3-44. PWD Point Score System 85
Table 4-1. Comparison of Two Point Scoring Protocols 89
Table 4-2. Distress Indicators and Their Assigned Scores (Deduct Values) 90
Table 5-1. Physical/Mechanistic Models For Pipe Deterioration 96
Table 5-2. Statistical/Empirical Models for Pipe Deterioration 99
Table 6-1. Decision Support Methods and Approaches Found in the Literature 104
XI
-------
ACRONYMS AND ABBREVIATIONS
3D
AC
AE
AFO
AMI
AR
ARP
AWWA
BEM
BPA
CARE-W
CCTV
CCD
CI
CMOS
CP
CPP
CSIRO
DBP
DC
DCVG
DFT
DI
D-S
DS
ECP
EIS
EMAT
EPA
EPR
ER
FCP
FOX-TEK
FSE
PUT
GA
GIS
GPIR
GPR
GRP
GTI
three-dimensional
asbestos cement
acoustic emission
acoustic fiber optic
advanced metering infrastructure
augmented reality
Annual Rehabilitation Project
American Water Works Association
broadband electromagnetic
basic probability assignment
Computer Aided Rehabilitation of Water Networks
closed circuit television
charge-coupled device
cast iron
complementary metal oxide semiconductor
cathodic protection
concrete pressure pipe
Commonwealth Scientific and Industrial Research Organization
disinfection byproduct
direct current
direct current voltage gradient
discrete Fourier transform
ductile iron
Dempster-Shafer
decision support (tool or software)
embedded cylinder pipe
electrochemical impedance spectroscopy
electromagnetic acoustic transducer
U.S. Environmental Protection Agency
evolutionary polynomial regression
electrical resistance
fuzzy composite programming
Fiber Optic Systems Technology, Inc.
fuzzy synthetic evaluation
flexible ultrasonic transducer
genetic algorithm
geographic information system
ground penetrating imaging radar
ground penetrating radar
glass-fiber reinforced polyester
Gas Technology Institute
xn
-------
HER
IWA
KARO
KURT
LCP
LEYP
LPR
LTP
LVVWD
LWC
MAKRO
MBS
MEMS
MFL
MIC
MLP
MsS
NASSCO
NDE
NOT
NRC
O&M
OP
FARMS
PCCP
PCM
PE
PI
PIRAT
PHM
PoD
PPIC
PRF
psi
PV
PVC
PWD
RFEC
RFEC/TC
RPV
SCADA
SI
hierarchical evidential reasoning
International Water Association
Kanalroboter
Kanal-Untersuchungs-Roboter-testplatform
lined cylinder pipe
linearly extended yule process
linear polarization resistance
long-term planning
Las Vegas Valley Water District
Louisville Water Company
Multi-segmented autonomous sewer robot
microwave back-scattering
microelectromechanical system
magnetic flux leakage
microbial induced corrosion
multilayer perceptron
magnetostrictive sensor
National Association of Sewer Service Companies
nondestructive evaluation
nondestructive testing
National Research Council of Canada
operation and maintenance
operating pressure
Pipeline Asset and Risk Management System
prestressed concrete cylinder pipe
pipeline current mapper
polyethylene
performance indicator
Pipe Inspection Real-Time Assessment Technique
proportional hazard method
probability of detection
Pressure Pipe Inspection Company
pulse repetition frequency
pounds per square inch
present value
polyvinyl chloride
Philadelphia Water Department
remote field eddy current
remote field eddy current/transformer coupling
replacement priority value
supervisory control and data acquisition
saturation index
xin
-------
SPU Seattle Public Utilities
SSET side scanning evaluation technology
TC transformer coupling
TDR time domain reflectometry
TO task order
UMP Utility Master Plan
UT ultrasonic testing
UWB ultra-wideband
WARP Water Main Rehabilitation Planner
WRc Water Research Centre
WRF Water Research Foundation
WS welded steel
WSN wireless sensor network
WSSC Washington Suburban Sanitary Commission
xiv
-------
1.0: INTRODUCTION
As part of the U.S. Environmental Protection Agency's (EPA's) Aging Water Infrastructure Research
Program, which supports the Sustainable Infrastructure Initiative, scientific and engineering research is
being conducted to evaluate and improve promising innovative technologies that can reduce costs and
improve the effectiveness of operation, maintenance, and replacement of aging and failing drinking water
distribution and wastewater conveyance systems (EPA, 2007). This research was conducted under Task
2.2 of Task Order (TO) 0062 (EPA STREAMS Contract No. EP-C-05-057), which is being conducted by
Battelle, in collaboration with National Research Council of Canada (NRC), to identify and characterize
the state of the technology for condition assessment of drinking water transmission and distribution
systems.
1.1 Objective, Scope, and Background
The objective of Task 2.2 is to compile a comprehensive inventory of condition assessment technologies
and decision support systems applied to water transmission and distribution mains and identify gaps that
need to be addressed by the research and development community.
EPA (2007) defines pipe condition assessment as "the collection of data and information through direct
and/or indirect methods, followed by analysis of the data and information, to make a determination of the
current and/or future structural, water quality, and hydraulic status of the pipeline." This task focuses on
the structural aspect of condition assessment. Pipe condition assessment may be undertaken by water
utilities with specific objectives, which include, but are not limited to:
• Monitoring and detecting critical indicators to prevent or mitigate catastrophic failures
• Implementing appropriate and timely repair/rehabilitation measures
• Early detection of accelerated deterioration for timely implementation of preventive measures
(e.g., retrofit cathodic protection [CP]) and for anticipation (and, where possible, mitigation)
of spikes in failure rate during extreme conditions (e.g., abnormally cold winters or drought)
• Setting inspection schedules and frequencies
• Screening and prioritizing assets to focus detailed, expensive inspections on critical sections
• Estimating remaining service life for pipe cohorts for mid- or long-term financial planning
and rate setting
• Detecting and reducing leakage to reduce water losses and water main breaks
• Determining whether structural vs. non-structural rehabilitation is suitable
• Providing insight into new pipe selection decisions - this could come from break histories,
forensic evaluations, screening inspections, or detailed inspections
The aging of water mains, coupled with the continuous stress placed on these systems by operational and
environmental conditions, has led to their deterioration, which has structural, hydraulic and water quality
manifestation, as implied in the EPA definition above. More specifically:
• Structural deterioration diminishes the structural resiliency of pipes and their ability to
withstand various types of stress, resulting in an increased rate of breakage. This, in turn,
causes increases in operation and maintenance (O&M) costs, loss of (treated) water, and
-------
social costs such as loss of service, disruption of traffic, disruption of business and industrial
processes, disruption of residential life and loss of landscape vegetation. In addition, pipe
breakage events increase the risk of water quality failure through intrusion of contaminants
into the system.
• The deterioration of pipe inner surfaces decreases the hydraulic capacity of pipes, which
results in increased energy consumption and disrupts the quality of service to the public. This
includes drinking water as well as fire extinguishing needs.
• The deterioration of pipe inner surfaces also causes deterioration of water quality. This
deterioration of water quality may manifest itself in taste, odor, and aesthetic problems in the
supply water and sometimes even in public health problems, such as higher risk of
disinfection byproduct (DBF) formation due to a higher need for chlorination.
The structural deterioration of water mains and their subsequent failure are complex processes, which are
affected by many factors, both static (e.g., pipe material, size, age, soil type) and dynamic (e.g., climate,
CP, pressure zone changes). The physical mechanisms that lead to pipe breakage are often very complex
and not completely understood. The facts that most pipes are buried, and that relatively little data are
available about their breakage modes (due to historical lack of awareness at water utilities of the
importance of collecting such data, as well as the time and cost involved in collecting and analyzing these
data) also contribute to this incomplete knowledge. It appears that while physical modeling of the
structural deterioration of water mains may be scientifically more robust, it is, to date, limited by existing
knowledge and available data.
Information on the current structural condition of individual water mains, combined with a good
understanding of failure modes and deterioration models, can greatly enhance the ability of water utilities
to manage their assets in a cost-effective manner. Some of the data required for physical models (e.g.,
detailed soil properties and detailed pipe material properties, data obtained by inspection of the pipe
current condition) can be obtained albeit at significant costs. These costs may currently be justified only
for maj or transmission water mains, where the consequences of failure are significant. In contrast,
statistically derived empirical models can be applied with various levels of input data and may therefore
be useful for small diameter water mains for which the low cost of failure does not justify expensive data
acquisition campaigns. The statistical analysis of breakage patterns of water mains has thus become a
cost-effective way to model this deterioration, particularly when available aforementioned data are scarce.
However, this effectiveness is higher at high-level planning (e.g., regional or network level) and
diminishes to a certain degree when applied to individual water mains.
The assessment of the structural condition of water mains and decision making on pipe renewal include
several elements:
(1) Physical modeling of the pipe in the soil.
(2) Understanding of pipe failure modes and their associated frequencies, including observable or
measurable signs (or distress indicators) that point to these modes, as well as inferential
indicators that point to potential existence of deterioration mechanisms.
(3) Inspection of the pipe to discern distress indicators.
(4) Interpretation of distress indicators to determine pipe condition.
(5) Empirical/statistical modeling of historical failures (mainly in small diameter distribution
mains).
(6) Modeling deterioration to forecast future failure rates and pipe residual life.
-------
(7) Assessment of failure consequences (direct, indirect and social costs).
(8) Scheduling pipe renewal so as to minimize life-cycle costs while meeting or exceeding
functional objectives of water distribution (quantity, quality, reliability, etc.)
Rajani and Kleiner (2004) described these elements schematically (Figure 1-1). Note that the
determination of pipe condition should not rely on distress indicators alone. Relevant information, such
as soil properties, environmental loads (climate, groundwater, overburden, etc.), operational practices
(CP, leak detection), sensor monitoring data, and pipe geometry, could also augment the understanding of
pipe condition.
In this report, all of the elements listed above (except item 7) will be discussed, but the main focus is on
elements 3 and 4, as well as elements 5, 6 and 8, in the context of making decisions about pipe renewal.
The American Water Works Association (AWWA) estimated, based on a survey of 337 water utilities,
that in the U.S. about two thirds (66%) of water mains are metallic (about 40% cast iron [CI], 22% ductile
iron [DI] and 4% steel), about 16% are asbestos cement (AC), 13% polyvinyl chloride (PVC) and 3%
various concrete pipes (Lillie et al., 2004). Rajani and McDonald (1995), in a survey encompassing 21
Canadian cities (about 11% of the population of Canada), revealed a similar distribution of pipe material
types. Consequently, pipe materials covered in this report include CI and DI, prestressed pressure
cylinder pipes (PCCP), AC, and PVC.
Physical model
• Material properties
• Pipe dimensions
• Internal pressure
• Temperature changes
• Loss of bedding
_s_u_P_P_°rtj _?'?;
I
Time (years)
1 — u lii —
Distress Indicators
• Size of corrosion pit
• # broken wires, damaged,
• Coating.
• Delamination.
T
I
— u u —
Interpretation and
Condition Assessment
• Failure modes
• Current factor of
safety, etc.
j|
0
-u u—
Failure
consequence
(cost of failure]
• Direct
• Indirect
• and social
I
Failure risk
Decision -making:
risk vs. renewal
__!
Figure 1-1. Schematic for Inspection, Condition Assessment, and Failure Risk Evaluation of Pipes
(Rajani and Kleiner, 2004)
-------
1.2 Organization of this Report
The objective, scope and background information is presented in Section 1. Section 2 provides a primer
on general issues related to the deterioration of buried pipes, including distress indicators, known modes
of failure and a general introduction to the classes of nondestructive evaluation (NDE) technologies and
methods to discern distress indicators leading to failure. Section 3 provides a comprehensive list of
existing NDE technologies/techniques that are currently used for buried pipes or that have the potential of
being adapted to pipe inspection. Scientific principles, advantages, and limitations of each technique are
described. Data about the extent of usage of many of these technologies cannot be easily obtained. Some
information is provided on usage, including the results of a limited survey conducted for this research.
Section 4 provides a comprehensive description of computation methods used to translate the inspection
data (or discerned distress indicators) into pipe condition rating. Section 5 provides a comprehensive
compilation of mathematical models that have been proposed in the literature to model the deterioration
of buried water mains. These include both physical models and statistical/empirical models. Section 6
provides a comprehensive compilation of mathematical models intended to support decisions related to
the renewal planning of water mains. This includes theoretical models from the literature as well as brief
descriptions of currently available decision support software tools. Section 7 identifies current
technological gaps that require further research. Section 8 provides summary and concluding remarks.
Section 9 presents the references cited throughout the report.
-------
2.0: PIPE DETERIORATION, DISTRESS INDICATORS AND FAILURE MODES
2.1
Overview of Distress Indicators and Failure Modes
Pipe condition is the cumulative effect of many factors acting on the pipe. Al-Barqawi and Zayed (2006)
classified these factors into three categories: physical, environmental, and operational, as depicted in
Figure 2-1. The factors in the first two classes can be further divided into static and dynamic (or time-
dependent). Static factors include pipe material, pipe geometry, and soil type, while dynamic factors
include pipe age, climate, and seismic activity. Operational factors are inherently dynamic.
Physical factors
Environment factors Operational factors
Pipe age and material
Pipe wall thickness
Pipe vintage
Pipe diameter
Type of joints
Thrust restraint
Pipe lining and coating
Dissimilar metals
Pipe installation
Pipe manufacture
Pipe bedding
Trench backfill
Soil type
Goundwater
Climate
Pipe location
Disturbances
Stray elecrical currents
Seismic activity
Internal water pressure,
transient pressure
Leakage
Water quality
Flow velocity
Backflow potential
Operation and maintenance
practices
Figure 2-1. Factors Contributing to Water System Deterioration
(Al-Barqawi and Zayed, 2006)
Many of the factors listed in Figure 2-1 are not readily measurable or quantifiable. Moreover, the
quantitative relationships between these factors and pipe failure are often not completely understood.
Consequently, contemporary practices of pipe condition assessment use two types of indicators, namely
distress indicators and inferential indicators.
-------
2.2
Distress Indicators for Major Pipe Types
Rajani et al. (2006) defined distress indicators as the observable/measurable physical manifestations of
the aging and deterioration process. Distress indicators are a result of some or all of the factors listed
above. Each distress indicator provides partial evidence for the condition of specific pipe components. It
is practical to refer to distress indicators by the respective pipe material, as provided in Tables 2-1 through
2-4, for CI and DI pipes, PCCP, AC, and PVC pipes, respectively. It is noted that leakage could also be
considered as a universal distress indicator regardless of pipe type (although the presence of a leak often
indicates that failure has already occurred). Leakage out of pressurized water mains is not an acceptable
public health risk and short-term pressure surges may pull contaminants into the pipe.
Table 2-1. Distress Indicators that Influence Pipe Condition for Cast and Ductile Iron Pipes
Category
External coating
(poly wrap/ tar/
zinc)
External pipe
barrel/bell
Inner lining/
surface
Joint
Distress Indicator
Crack/tear/holiday
Remaining wall thickness
Graphitization (pit) areal
extent
Crack (pit)f type
Crack (pit)f width
Cement lining (epoxy)
spalling (blistering)
Remaining wall thickness
Tuberculation
Change in alignment
Joint displacement
Comments
State of external coating will dictate how external corrosion
is likely to encourage damage to the pipe.
Remaining pipe wall thickness is usually obtained from NDE
tests or from spot exhumations and sand blasting samples.
Casting defects (voids or inclusions) can be of significant
size in CI pipes.
Areal extent as percentage of pipe diameter times unit length
indicates the size of affected area. Severe graphitization may
not always mean the pipe should have failed. In practice,
graphitized area can still provide some resistance - it acts as
a form of sticky plaster. In CI, graphitization is typically in
the form of graphite flakes, while in DI it is in the form of
nodules.
A pit is a manifestation of an electro-chemical process, while
a crack is a mechanical response to stress. Circumferential
cracks indicate some type of longitudinal movement, loss of
bedding support, or increase in vertical load (frost) has taken
place. Longitudinal cracks occur due to low hoop resistance,
typically coupled with high internal pressure.
Crack width is another indicator of corrosion. A wide crack
together with a deep pit will be more detrimental to the pipe
than a narrow, but shallow crack.
Inner lining deterioration is often due to incompatible water
chemistry or abrasion due to the presence of high water
velocities and sediments.
Occasionally, closed circuit television (CCTV) scans can
give estimates of internal corrosion pits when NDE tests are
not done to get an overall picture of the pipe wall status.
Heavy tuberculation (blockage) can significantly reduce
water delivery and produce red water condition.
Changes in joint alignment (rotation) indicate pipe
susceptible to ground movement. Large changes can lead to
leakage and eventually joint failure.
Joints can displace without undergoing joint misalignment
and hence is also an indicator of other forces at play.
(Rajani etal., 2006)
t Cracks and pits are common in CI pipes, while DI pipes usually only have pits. Small diameter CI pipes may also
be susceptible to ring fractures in shrink/swell soil conditions.
-------
Table 2-2. Distress Indicators that Influence Pipe Condition for PCCP Water Mains
Category
Mortar coating
Prestressed wire
Concrete core
Pipe geometry
Joint
Distress Indicator
Spalling
Crack type
Crack width
Crack density (frequency)
Coloration
Wire breaks
Delamination
Crack type
Crack width
Crack density (frequency)
Hammer tapping sound
Hollow area
Out of roundness
Change in alignment
Joint (internal) displacement
Joint diaper crack size
Joint ring degradation
Comments
Spalling is often a first indicator of corrosion. Large
spalling area may indicate that corrosion is taking
place over a significant surface area of the pipe
exterior.
Circumferential cracks indicate some type of
longitudinal movement has taken place. Longitudinal
cracks occur due to low hoop resistance.
Crack width is another indicator of severity of
spalling. Large widths mean that spalling is imminent.
Closer crack spacing usually means the pipe is under
higher stress.
Signs of color/stains on concrete exterior indicate that
corrosion is taking place. Often stains are precursors
to spalling, i.e., corrosion products have built up.
As the number of wire breaks increase, the factor of
safety decreases and eventually leads to pipe failure.
Delamination occurs when there is poor bonding
between concrete/wire or steel/steel cylinder. This can
also occur when prestressing is lost due to wire
breaks.
Circumferential cracks indicate some type of
longitudinal movement has taken place. Longitudinal
cracks occur when prestressing is lost due to wire
breaks.
Crack width is another indicator of severity of
delamination. Large width means that delamination is
imminent.
Closer crack spacing usually means the pipe is under
higher stress.
Hammer tapping sounds can indicate delamination. It
can be as simple as tapping a hammer or using the
pulse echo method.
Areal extent of hollow sound can give an idea of the
seriousness of the delamination (in comparison to pipe
surface area).
Out-of-roundness is another indicator of wire loss that
may not be evident from concrete spalling or presence
of corrosion products, etc.
Changes in joint alignment (rotation) indicate pipe
susceptible to ground movement. Eventually it can
lead to weld failures and joint failure.
Joints can displace without undergoing joint
misalignment and hence are also an indicator of other
forces at play.
Crack of external diaper can give an idea of joint
quality.
Joint failure due to microbial degradation of the
natural rubber joint rings.
(Kleiner et al., 2006a)
-------
Table 2-3. Distress Indicators that Influence Pipe Condition for AC Pipes
Category
External coating
(tar or bitumen)
External pipe
barrel
Internal pipe
surface
Joint
Distress Indicator
Holiday
Remaining wall thickness
Corrosion^ area! extent
Crack type
Crack width
Remaining wall thickness
Corrosion areal extent
Change in alignment
Joint displacement
Joint ring degradation
Comment
State of external coating will indicate how external
soil properties encourage damage to the pipe.
Remaining pipe wall thickness (includes both external
and internal walls) is usually obtained from spot test
samples and performing phenolphthalein test (to
measure cement softening) or on-site measurements
using the georadar technique.
Areal extent as percentage of pipe diameter times pipe
segment length indicates the size of affected area.
Severe corrosion may not always mean the pipe
should have failed.
Circumferential cracks indicate bending or significant
longitudinal movement has taken place. Longitudinal
cracks occur due to exceedance of hoop resistance,
due to occurrence of very high operational loads or
due to low remaining wall thickness as a result of
sulfate attack.
Crack width is another indicator of corrosion. Wide
crack together with a deep softening of asbestos
cement matrix will be more detrimental to the pipe
than a narrow but shallow crack.
See above for external pipe barrel category.
See above for external pipe barrel category.
Changes in joint alignment (rotation) indicate pipe is
susceptible to ground movement. Large changes can
lead to leakage and eventually joint failure.
Joints can displace without undergoing joint
misalignment (axial movement) and hence are also an
indicator of other forces at play.
Joint failure due to microbial degradation of the
natural rubber joint rings.
^Corrosion is meant to indicate leaching/depletion of cement within the AC matrix due to some chemical
mechanism.
-------
Table 2-4. Distress Indicators that Influence Pipe Condition for PVC Pipes
Category
External pipe
barrel surface
Service
connection
Joint
Distress Indicator
Remaining wall thickness
Scratch type
Scratch depth
Split at tap
Change in alignment
Joint displacement
Comment
Cavities or unfilled air bubbles introduced during
manufacturing (and not detected upon installation) can
be of significant size in PVC pipes.
Longitudinal scratches are formed due to improper or
rough handling. Circumferential scratches can form if
lifted or handled using rough slings (e.g., chains).
Also sharp scratches have more detrimental effects
than blunt scratches. Longitudinal scratches can
eventually lead to longitudinal split failures.
Fatigue failure becomes an important consideration
for deeper scratches, especially when they exceed
10% of pipe wall thickness.
Inadequate tapping procedure or thin pipe wall can
lead to a split in the PVC mains, usually on the pipe
inside. This type of failure is commonly referred to as
a fitting failure.
Changes in joint alignment (rotation) indicate pipe is
susceptible to ground movement. Large changes can
lead to leakage.
Joints can displace without undergoing joint
misalignment and hence are also an indicator of other
forces at play.
2.3
Inferential Indicators for Major Pipe Materials
Inferential indicators point to the potential existence of a pipe deterioration mechanism without actual
knowledge if this potential has actually been realized. Many of the environmental indicators are
inferential in nature, such as soil type, groundwater fluctuations, etc. It is important to note that
inferential indicators do not provide direct evidence about pipe deterioration, but rather indicate the
potential thereof. However, these indicators are usually easier and cheaper to discern since they can be
obtained by nondestructive and nonintrusive methods, and are often used to pre-screen pipes for more
expensive direct inspection or to obtain supplemental information in conjunction with distress indicators.
Pipe age could, in some context, be viewed as a universal inferential indicator. However, the age of the
pipe is only a measure of the pipe exposure to its surrounding environment and operating conditions (i.e.,
to other inferential indicators), therefore it does not appear as an explicit indicator in the following tables.
Tables 2-5 through 2-8 present the inferential indicators for CI and DI pipes, PCCP, AC, and PVC pipes,
respectively.
Some distress indicators can be discerned by direct observation, while others require the application of
more elaborate technologies. In either case, discerned distress indicators usually require interpretation,
aggregation and/or some other type of techniques (methods) to fuse data (data fusion) from different
sources to obtain the condition rating of the pipe. In this context, pipe condition rating is understood to
mean a grade or a score (or a rating) on some consistent ordinal scale (e.g., good, fair, poor) that enables
the condition rating of pipes relative to each other as well as to "quantify" and track the amount of
deterioration over time in a given pipe.
-------
Table 2-5. Inferential Indicators for Cast and Ductile Iron Pipes
Category
(Level 1)
Pipe vintage
Pipe joint
Water quality
Water pressure
Location
Soil
Agent
(Level 2)
Material type, historic
standards, and
installation practices
Joint type
Water pH
Operating pressure
(OP)
Pressure change
amplitude
(% OP)
Pressure change
frequency
Pipe embedment
Surface loads - traffic
type
Wet/dry cycle(s)
Water table level
Soil type / backfill
Soil resistivity
Soil pH
Redox potential
Soil chloride
Soil sulfate
Soil sulfide
Comment
Pipes of specific vintages can experience a higher breakage rate.
This can be a manifestation of manufacturing processes and
standards (e.g., pit vs. spun cast, pipe wall thickness, etc.), or
installation practices (e.g., internal lining, polywrap onDI pipes,
etc.). Knowledge of the installer could also help to identify poor
vs. adequate installation practices.
Historically, three main joint types: (1) rigid, e.g., bored bell and
turned spigot; flanged; (2) semi-rigid, e.g., lead-yarn; and (3)
flexible, e.g., rubber-gasket push in joint. Pre-mid 1930s, most
joints were semi-rigid type (lead-yarn combination). "Leadite"
(brand name for sulfur based compounds - mixture of iron, sulfur,
slag, and salt) also was used in North America between early
1900s and late 1940s, however, lead was often the jointing
material of choice in North America and in the UK. Rubber
gasket push-on or roll-on joints introduced in mid 1950s.
Anecdotal reports indicate that leadite joints have performed
poorly over the years.
Water with low pH can leach the internal cement lining or pipe
wall itself if lining is absent.
High pressure subjects pipe to high stress and hence higher
propensity to failure.
Large pressure changes (% of OP) can induce higher stresses than
expected by design.
Either slow or fast fatigue mechanism can induce early failure.
Pipes exposed to wet/dry conditions have higher failure rate than
pipes totally below water table or pipes totally exposed to
atmosphere.
Heavy surface loads will subject the pipe to high stresses and
hence faster deterioration in the long term.
Changing environment can promote corrosion.
Water table position will indicate if wet/dry cycle is likely to
occur.
Non-draining backfill leads to moisture retention and promotes
corrosion; also, poor backfill can lead to development of out-of-
roundness condition as soil side (springline) support is not
available as required by design of DI pipes.
Low resistivity soil leads to higher corrosion rates. Soil chlorides
(e.g., from de-icing salts) reduce soil resistivity.
Low pH (< 4) means soil is acidic and likely to promote
corrosion; high alkaline conditions (pH > 8) can also lead to high
corrosion.
High availability of oxygen promotes microbial induced corrosion
(MIC) in the presence of sulfates and sulfides.
Low chloride levels in high pH (> 1 1.5) environments can lead to
serious corrosion.
Accounts for MIC and possible food source for sulfate reducing
bacteria in anaerobic conditions under loose coatings.
Sulfate reducing bacteria give off sulfides that are excellent
electrolytes.
10
-------
Table 2-5. Inferential Indicators for Cast and Ductile Iron Pipes (Continued)
Category
(Level 1)
Corrosion
Agent
(Level 2)
Frost susceptibility
(load)
Cathodic protection
Stray current
Comment
CI and DI pipes are not designed for frost loads. If conditions
exist to develop significant frost loads, then pipe will be subjected
to additional stresses (annual) and lead to pipe failure if already
significantly corroded. These conditions are: high water table;
thermal gradient; right soil type to develop suction (i.e., silt or
clayey silt).
Cathodic protection (galvanic as well as impressed current) is
likely to reduce corrosion.
Stray current is known to accelerate corrosion unless adequate
measures have been taken.
(Kleiner etal., 2005)
Table 2-6. Inferential Indicators for PCCP
Category
(Level 1)
Pipe vintage
Water quality
Water pressure
Location
Soil
Agent (Level 2)
Material type, historic
standards, and
installation practices
Water pH
Operating pressure (OP)
OP change amplitude
(% OP)
OP change frequency
Pipe embedment
Surface loads - traffic
type
Wet/dry cycle(s)
Water table level
Soil type / backfill
Soil resistivity
Soil pH
Soil chloride
Comment
Some early vintage PCCP suffered from inadequate design and
manufacture and has a record of failure and increased wire
breakage rates. This can be a manifestation of manufacturing
processes and standards or installation practices. Knowledge of
the installer could also help to identify poor vs. adequate
installation practices.
Water with low pH can leach the internal cement/concrete lining.
High pressure subjects pipe to high stress and hence higher
propensity to failure.
Large pressure changes (% of OP) can induce higher stresses than
expected by design.
Either slow or fast fatigue mechanism can induce early failure.
Pipes exposed to wet/dry conditions have higher failure rate than
pipes totally below water table or pipes totally exposed to
atmosphere.
Heavy surface loads will subject the pipe to high stresses and
hence to faster deterioration in the long term.
Changing environment promotes corrosion of wires if chloride
concentration exceeds 140 mg/kg (140 ppm).
Water table position will indicate if wet/dry cycle is likely to
occur.
Non-draining backfill leads to moisture retention and hence
promotes corrosion; also, poor backfill can lead to development
of out-of-roundness condition as soil side (spring line) support is
not available as required by design.
Low resistivity soils lead to higher corrosion rates of prestressing
wire and steel cylinder. Soil chlorides (e.g., from de-icing salts)
reduce soil resistivity.
Low pH (< 4) means soil is acidic and likely to promote
corrosion; high alkaline conditions (pH > 8) can also lead to high
corrosion of prestressing wire and steel cylinder.
Mortar coating usually creates a pH environment of >12.4. Low
chloride levels in high pH (> 1 1.5) environments can lead to
serious corrosion as noted by Bianchetti (1993).
11
-------
Table 2-6. Inferential Indicator for PCCP (Continued)
Category
(Level 1)
External
coating
Prestressed wire
Corrosion
Agent (Level 2)
Soil sulfate
Soil sulfide
Frost susceptibility (load)
Coating type
Concrete chloride
concentration
Absorption capacity
Wire class
Cathodic protection
Stray current
Comment
Accounts for MIC and possible food source for sulfate reducing
bacteria in anaerobic conditions under loose coatings.
Sulfate reducing bacteria giving off sulfides, which are excellent
electrolytes.
PCCP pipes are not designed for frost loads. If conditions exist
to develop significant frost loads, then pipe will be subjected to
additional stresses (annual) and prematurely lead to development
of cracks. These conditions are: high water table; thermal
gradient; right soil type to develop suction (i.e., silt or clayey
silt).
Cast coating was applied prior to the mid 1960s which is prone to
spalling. After 1970, this coating has since been superseded by
mortar coating, which cracks but does not spall.
Chloride levels higher than 1,000 ppm promote corrosion.
Mortar absorption greater than 8% leads to higher corrosion rates.
Interpace pipe manufactured prior to 1985-1988 have Class IV
wire or Class III wire. These high strength wires are susceptible
to hydrogen embrittlement.
Too strong CP currents (especially impressed current systems)
may lead to hydrogen embrittlement, especially with Class I and
II prestressing wires.
Stray current is known to accelerate corrosion unless adequate
measures have been taken.
(Kleiner etal., 2005)
Table 2-7. Inferential Indicators for AC Pipes
Category
(Level 1)
Pipe vintage
Water quality
Water pressure
Location
Agent (Level 2)
Material type, historic
standards, and
installation practices.
Water pH
Water saturation index
(SI)
Operating pressure (OP)
Pressure change
amplitude (% OP)
Pressure change
frequency
Surface loads - traffic
type
Wet/dry cycle(s)
Comment
Pipes of specific vintages have experienced a higher breakage rate,
(e.g., AC pipes of types I and II [free lime < 1%]). This can be a
manifestation of manufacturing processes and standards or
installation practices. Knowledge of the installer could also help
to identify poor vs. adequate installation practices.
Water with low pH can leach the cement within the AC matrix.
Water with SI < 0.25 can leach the cement within the AC matrix.
High pressure subjects pipe to high stress and hence higher
propensity to failure.
Large pressure changes (% of OP) can induce higher stresses than
expected by design.
Fatigue mechanism not observed or documented for AC pipes.
Heavy surface loads will subject the pipe to high stresses and
hence to faster deterioration in the long term.
Changing environment promotes higher expansion of matrix than
unchanging environment. AC type II offers better resistance to
sulfate induced swelling.
12
-------
Table 2-7. Inferential Indicators for AC Pipes (Continued)
Category
(Level 1)
Soil
Agent (Level 2)
Water table level
Soil type / backfills
Soil pH
Soil sulfate
Frost susceptibility (load)
Comment
Water table position will indicate if wet/dry cycle is likely to
occur. Soil sulfate attack only occurs if sulfate is in solution.
Non-draining backfill leads to moisture retention and hence
promotes external corrosion.
Low pH (< 5) means soil is acidic and likely to promote corrosion.
Soils with high sulfate (> 1000 ppm) can attack AC pipes with
high free lime (type I AC pipes).
AC pipes are not designed for frost loads. If conditions exist to
develop significant frost loads then pipe will be subjected to
additional stresses (annual) and lead to pipe failure if already
significantly corroded. These conditions are: high water table;
thermal gradient; right soil type to develop suction (i.e., silt or
clayey silt).
Table 2-8. Inferential Indicators for PVC Pipes
Category
(Level 1)
Pipe vintage
Water pressure
Location
Soil
Agent (Level 2)
Material type, historic
standards, and installation
practices.
Operating pressure (OP)
Pressure change amplitude
(% OP)
Pressure change frequency
Surface loads - traffic type
Hydrocarbons
Frost susceptibility (load)
Comment
Most PVC pipes used in North America are of the unplasticized
PVC type. Newer modified PVC and oriented PVC have recently
appeared on the market. Failures could be tied to certain
manufacturing processes and standards or installation practices.
Knowledge of the installer could also help to identify poor vs.
adequate installation practices.
High pressure subjects pipe to high stress and hence higher
propensity to failure. Time to failure can be substantially reduced
in PVC pipes under high pressure since PVC is a visco-elastic
material.
Large pressure changes (% of OP) can induce higher stresses than
expected by design.
Fatigue mechanism is primary mechanism of PVC pipes if
scratches or gouging are present.
Heavy surface loads will subject the pipe to high stresses and
hence to faster deterioration in the long term especially if PVC
pipes have been previously scratched or gouged.
PVC pipes are impervious to high-octane gasoline and gasoline
saturated water for periods of up to 2 years.
PVC pipes are not designed for frost loads. If conditions exist to
develop significant frost loads, then pipe will be subjected to
additional stresses (annual) and lead to pipe failure if already
significantly scratched.
13
-------
3.0: TECHNOLOGIES FOR CONDITION ASSESSMENT OF WATER MAINS
3.1
Nondestructive Testing and Evaluation
As described earlier, there are two types of observations to be made in the course of pipe condition
assessment, namely observation of distress indicators and observation of inferential indicators. This
report addresses both. While the observation of inferential indicators is always nondestructive and
nonintrusive, the observation of distress indicators can be destructive or nondestructive as well as
intrusive or nonintrusive. Destructive testing entails the removal of a sample from pipe wall to analyze
remaining thickness, defects, damages, and residual strength. These types of tests are not addressed in
this report. Nondestructive testing (NOT) techniques (also commonly referred to as NDE) include the
direct visual observation of defects such as cracks, corrosion pits or holes, as well as techniques that
provide signals or signatures that are interpreted into distress indicators.
Descriptions of NOT technologies can be found in several published reports (Dingus et al, 2002; Reed et
al., 2004; Lillie et al., 2004; Marlow et al., 2007; Thomson and Wang, 2009; Feeney et al., 2009). This
report makes maximum use of published reports and input from water utilities, vendors, and consultants
to provide the most up-to-date information. The descriptions have been sent to technology vendors for
comments. Detailed technical information for some of the technologies is not available from the vendors.
Therefore, the information collected from their Web sites and publications will be used in this report.
Figure 3-1 lists the inspection technologies covered in this section. Table 3-1 shows the potential to apply
an inspection technology to different pipe materials. Each technology is described briefly in the main text
followed by a short table summarizing the purpose, status, source of information, advantages, limitations,
performance, breadth of use, and other available information. With few exceptions, all technologies are
presented using the template in Table 3-2, for simplicity and ease of comparison. It should also be noted
that whenever a vendor/developer of a technology was identified, a copy of the entry related to this
technology was sent to them for review and verification. Consequently, the vast majority of the relevant
entries have received vendor/developer feedback. Cost data were provided wherever available, but it was
not available for most of the technologies covered in this section.
Table 3-1. Summary of Condition Assessment Technologies Applicable to Different Pipe Materials
Technology
Pit depth measurement
Visual inspection
Electromagnetic inspection
Acoustic inspection
Ultrasonic testing
Pipeline current mapper
Radiographic testing
Thermographic testing
Pipe condition assessment
from soil properties
Sensor technologies1^
Metallic Pipes
CI DI f WS
A/
A/
A/
A/
A/3
A/
A/
A/
A/
A/
Concrete Pipes
CPP/PCCP AC
-
A/
A/
A/
-
-
-
-
7
A/
Poly Pipes
GRP PVC/uPVC PE
-
?
-
A/
?
-
-
-
?
?
(a) Ultrasonic thickness methods may be less accurate for pit cast iron pipes because of the larger grain structure.
(b) Emerging sensors and sensor networks and their applicability to various pipe types are described in Section 3.13.
A/: available; ?: may/may not work; CI = cast iron, DI = ductile iron, WS = welded steel, CPP/PCCP = concrete
pressure/pre-stressed concrete cylinder pipe, AC = asbestos cement, GPJ3 = glass-fiber reinforced polyester,
PVC/uPVC= polyvinyl chloride/un-plasticized PVC, PE = polyethylene
14
-------
Pitting depth
measurement .
f Visual inspection H
^/ Electromagnetic \
inspection
~-
; AcOUStic
inspection
w / ... . .
^, ultrasonic testing i~
^ . Pipeline current
mapper
.•• Radiographic ,
testing
/ Thermographic \
testing
^/ Assessment from
soil properties
Emerging sensor
— ^-; technologies and .
'• sensor networks.
t
Man entry
inspection
t
Magnetic flux
leakage
t
Sonar profile
t
Continuous
measure
t
LPR
t
CCTV
inspection
t
Remote field
eddy current
t
Impact echo
t
Discrete
measure
t
Soil
properties
t
Video
endoscope
t
Broadband
EM
y
Acoustic
emission
t
Phased array
t
Soil
corrosivity
t
3D optical
scanning
t
Pulsed eddy
current
t
Leak
detection
t
Combined UT
inspection
t
Soil
resistivity
t t
Laser Handyscan
profiling 3D
t t
Ultra-wideband 1
pulsed radar |
t
Seismic pulse
echo
t
Pipe to soil
potential survey ;
Figure 3-1. Nondestructive Inspection Technologies for Condition Assessment of Water Mains
Table 3-2. Template Used for Description of Technologies
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Common name of technology
Intended purpose or scope
Commercially available/experimental/in development, etc.
Identifies sources of information.
Relative to similar or to other technologies
Relative to similar or to other technologies
Accuracy, false positives, false negatives, etc.
If currently used for water mains - to what extent?
If not currently used for water mains, where is it used? What is the potential
for use in water mains?
If available
15
-------
3.2
Pit Depth Measurement
Pit depth can be measured with a pointed micrometer or needle-point depth gauge. Other methods
include a grid with ultrasonic spot measurement, automated ultrasonic scanner, and laser range
measurement. The pit depth measurement can be carried out in the field on exposed sections of the pipe
(for external corrosion) or in a laboratory on pipe samples (for external and internal corrosion). Before pit
measurement, pipe samples are sand/grit blasted to remove corrosion products. Pitting depth
measurement can be applied methodically, within a general survey, or opportunistically when a pipe is
exposed (e.g., upon breakage repair). Table 3-3 provides more information on pit depth measurement.
Table 3-3. Pit Depth Measurement
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Pit depth measurement
Measure the pit depth of ferrous pipes due to corrosion. Can help to evaluate
historical pipe corrosion rate (subject to some fundamental assumptions that often
cannot be verified). This rate, in conjunction with deterioration modeling, can be used
to assess pipe remaining life.
Various pit depth measurement devices are commercially available. Some devices
(e.g., laser range finder) have been developed for research purposes only.
SwRI, 2002; Marlow et al., 2007; many others available
• Direct measurement, no need for interpretation
• Provides good indication of sample condition.
• Does not require special skills, easy to train personnel.
• For external corrosion, exposed pipe does not need to be taken out of service.
• Can be practically applied only to samples, therefore requires some sophisticated
statistical analysis to infer general condition of the entire pipe (or pipe segment).
• Need to expose the pipe or to cut coupon (destructive testing). When exposing a
pipe, care needs to be taken to adequately protect the exposed pipe segments from
future corrosion.
• Existing coating needs to be removed.
• Original pipe wall thickness must be available for corrosion rate estimation.
• For internal corrosion, pipe needs to be taken out of service.
Manual measurement does not need highly-skilled operators. Simple to implement.
Only provides information that is specific to the sample. No issue with false positives,
false negatives, etc.
No direct information about breadth of use, but because of its simplicity, it is likely
used by many to varying degrees.
Pit depth measurement of samples along the pipe can be used in a statistical analysis to
infer pipe condition. Also, can be used as an input to pipe deterioration models to
estimate time to failure.
3.3
Visual Inspection
The condition of the internal surfaces of the pipe can be assessed by a visual inspection. It may be done
without specialized equipment or a variety of vision aids (e.g., closed-circuit television, videoscope, or
laser-based surface profiler) may be employed to augment human vision. It is generally used in
conjunction with a library of defects/deficiencies.
16
-------
3.3.1 Man Entry and Visual Inspection. Inspectors can record defects/deficiencies, including
size, location, and extent, with hand-held video or still cameras. Acoustic tests are often performed
concurrently to provide non-visible information about the pipe. By striking the pipe wall with a hammer,
the sound, either dull or solid, provides qualitative information about the condition of the pipe wall. In
the office, defects/deficiencies can be coded, assigned scores and aggregated to provide the overall
condition of the pipe. Table 3-4 provides more information on man entry and visual inspection.
Table 3-4. Man Entry and Visual Inspection
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Man entry and visual inspection
Man entry inspection is suitable for relatively large diameter pipes. Visual inspection
can also be applied to the external surface of an exposed pipe.
Currently being applied mainly in sewers, but also in large transmission water mains.
Marlow et al., 2007; many others available
• Relatively simple, no special equipment necessary and training courses are widely
available.
• The exposure of a buried pipe also allows the assessment of the quality and
condition of the backfill.
• The assessment can provide an indication of the cause of the deterioration and the
likelihood of being more widespread.
• Internal inspection suitable only for relatively large diameter pipes.
• External inspection involves exposing of pipes — expensive.
• Not very effective to discover defects/deficiencies that are not manifested on the
pipe surface.
• Water mains need to be taken out of service.
Depends on skills of personnel.
Widely applied in water and sewer mains.
Visual inspection can be a precursor to other condition assessment techniques.
3.3.2 Closed Circuit Television Inspection. The CCTV inspection records a close-up observation
of the pipe surface. The CCTV system comprises a CCTV camera and lighting apparatus, mounted on a
carrier. A winch and pulley system moves the CCTV module through the pipe. Larger modules can use
an umbilical cord system, which can provide power and communication from and to the ground station, as
well as serve to retrieve the device. The basic steps of CCTV inspection include:
• Introduce a carrier with the CCTV camera into the pipe via access points;
• The carrier travels along the pipe and the camera captures and transmits the images to a
ground station (inspection truck);
• Analyze images in the field or office.
The procedure is illustrated in Figure 3-2 and more details are provided in Table 3-5. The live video
images are sent back to the ground control center via coaxial or twisted pair cables so the operator can
remotely control the CCTV module. Most CCTV modules are equipped with panned and tilted cameras,
which can implement a close-up observation of the pipe surface. Local storage devices can also save
image data on a hard drive, DVD disk or VHS tape.
17
-------
Cfcv
Figure 3-2. CCTV Inspection
Table 3-5. Closed Circuit Television Inspection
Name
Purpose/Scope
Status
Source of Information
Advantages
Limitations
Performance
Breadth of use
Other Information
CCTV inspection
Visual inspection without man-entry. Particularly suitable for smaller diameter pipes.
Applied mainly to sewers and stormwater pipes, but can also be applied to water
mains for the inspection of inner surfaces after the line is emptied. Generally used in
conjunction with a library of defects/deficiencies.
Several CCTV systems are commercially available.
Hydro max, 2006; RapidView, 2007a; many others available
• Simple, relatively inexpensive, suitable for small and large pipes
• New systems with multi -camera and/or fish-eye technology can record a full view
of a pipe and allow relatively high scanning speed as well as full off-line
inspection.
• Digital recording is convenient for data storage, as well as future developments in
automatic data interpretation.
• Provides information only on defects that are manifested on the pipe inner surface;
• Inspection results are qualitative and need interpretation.
• Quantitative rating requires trained inspectors.
• Limitations of traditional CCTV inspection include the need to pan and tilt to see
sides and laterals, the camera has to stop at each defect's location for a closer look
and identification, and to ensure an acceptable video quality; the carrier's speed is
limited to 150 mm/s (5.9 in./s).
• Tuberculated pipes may need to be scrubbed and cleaned prior to inspection.
• Currently not available for in-service water main inspection.
• Requires a special launching and retrieval chamber in water mains.
Depends on skills of personnel.
CCTV systems have been widely used for sewers. Usage in water mains is limited
mainly due to the last three limitations listed above.
Not available
An improvement on the traditional CCTV is the side scanning evaluation technology (SSET), which
provides both frontal and 360° images of the interior surface of the pipe wall (Hydromax, 2006). Two
cameras simultaneously capture a forward view and a perpendicular view of the pipeline. The SSET
system can travel through the pipeline at a constant speed without stopping to observe defects. A pan or
18
-------
tilt camera is not needed. A key benefit of the SSET is that it lends itself better to comparison of data
from one year to the next. An advanced version of SSET is the DigiSewer system marketed by
Envirosight (Envirosight, 2010). With DigiSewer side-scanning, pipe footage can be captured at a speed
up to 70 ft/min without stopping to pan, tilt or zoom. The detailed flat scan can be further reviewed and
annotated.
Another improvement on the traditional CCTV is the PANORAMO® optoscanner, which uses two
integrated scanning units, one at the front end and one at the rear end as shown in Figure 3-3 (RapidView,
2007a). Each scanning unit consists of a 185° fish-eye lens and a high resolution digital camera. The two
units take hemispherical images and create 360° spherical images. An unfolded, two-dimensional view of
the entire section and a three-dimensional view of the pipe allow the viewer to pan the angle of view in all
directions. This pan and tilt scanning of details can be done in the office without actually operating the
camera during inspection. The operator can pan and rotate a virtual camera like a real one. Another
advantage of the PANORAMO® system is that it can operate at a relatively high speed of 300 mm per
second.
The inspection results need to be interpreted. This interpretation is currently done manually, but
machine-vision techniques are likely to be developed in the future. In the office, defects/deficiencies can
be coded, assigned scores, and aggregated to provide the overall condition of the pipe.
Figure 3-3. The PANORAMO® System
(Reprinted with permission of RapidView)
3.3.3 Videoscope. A borescope is an optical device consisting of a rigid or flexible tube with an
eyepiece on one end and an objective lens on the other, linked together by a relay optical system.
Videoscope is an advanced type of borescope that houses a very small charge-coupled device (CCD) chip
embedded in the tip of the scope. Videoscopes are normally 10 mm (0.4 in.) or less in diameter and come
in lengths up to 15.24 m (50 ft). Several integral features include the insertion probe section, the
articulated tip, articulation controls, lighting bundle, high intensity external light source and cable
interfaces, and external media recording device. The video image is relayed from the distal tip and
focusable lens assembly back to the display via internal wiring.
This technique is used for the visualization of areas that are otherwise inaccessible. Videoscopes are easy
to operate. The user can get a full control of the scope position with articulated controls and the captured
video sequences/images can be analyzed with software. Table 3-6 provides more information on the
videoscope.
19
-------
Table 3-6. Videoscope
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Videoscope
Remote visual inspection. Suitable for the inspection of objects to which
normal access is difficult or impossible.
Equipment is commercially available.
http://www.fiberscope.net/; many others available
• Visualization of hidden areas with limited access
• High quality images
• Enables high quality control of inspected devices
• High-speed video capturing ability
• Provides information only on defects that are manifested on the pipe
inner surface.
• Inspection results are qualitative and need interpretation.
• Quantitative rating requires trained inspectors.
• Usage is limited to short-length and small diameter pipe.
Same as other visual inspection techniques.
Videoscopes have been used for gas/oil pipeline inspection and many other
applications such as aircraft engine, automotive transmission, concrete,
security, as well as drinking water and wastewater pipes.
Not available
3.3.4 3D Optical Scanning. The three-dimensional (3D) optical scanner in Figure 3-4 contains
two high-resolution digital cameras with distortion-free, wide-angle lenses (RapidView, 2007b). Image
data are captured and transmitted to a control vehicle for processing and storage. Maximum speed is 350
mm/s (about 14 in./s). Inspection results need to be interpreted by trained personnel. Similar to the
optoscanner, a 360° pan/ zoom as well as unfolded view of the inner surface of the manhole can be
obtained offline with software tools. Table 3-7 provides more information on 3D optical scanning.
Figure 3-4. PANORAMO® SI 3D Optical Manhole Scanner
(Reprinted with permission of RapidView)
20
-------
Table 3-7. 3D Optical Scanning
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
3D optical scanning
Inspection of manholes, drilled shaft, and boreholes.
Commercially available (Panoramo® SI)
RapidView, 2007b; http://www.rapidview.com/panoramosi.htm
• Fast inspection
• Suitable for vertical pipes or pipe-shaped structures
• There are no known limitations except those generally associated
visual inspections.
with all
Same as other visual inspection techniques
Used mainly for manholes from 400 mm (16 in.) diameter upwards.
Not available
3.3.5 Laser-Based Pipe Surface Profiling. Distance measurement by laser can be done using one
of four principles, including triangulation, time-of-flight, pulse-type time-of-flight, and modulated beam
systems. In a triangulation system, the detecting element measures the laser spot within its field of view.
Usually, this type of laser measurement is used for distances of a few inches. Time-of-flight sensors
derive range from the time it takes light to travel from the sensor to the target and back (Acuity, 2008).
This technology is typically used for relatively long distance measurements. For very long distances, a
pulsed laser beam is used. A modulated beam system also uses the time duration for light to travel to the
target and back; however, in this case, time is not measured directly. Instead, the strength of the laser is
varied to produce a signal that changes overtime. The time delay is indirectly discerned by comparing
the signal from the laser with the delayed signal returning from the target. Modulated beam sensors are
typically used in intermediate range applications.
To acquire the pipe inner profile, a spinning apparatus is needed to control the laser beam. Such a laser
range measurement does not require any special illumination and can be carried out in complete darkness.
The speed of spinning, sampling rate, and carrier moving velocity determine the accuracy and resolution
of the scanning. The inspection is affected by the roughness as well as the color of the pipe surface.
Another method makes use of a ring of laser light projected onto the pipe inner surface (Duran et al.,
2003). The ring must be strong enough to be "seen" by a camera. The camera is used to capture the
images of this projected ring. The laser device moves with the camera through the pipe. The analysis
software extracts the laser ring from captured images and reconstructs a digital pipe profile. This profile
can be easily unfolded or manipulated for review and analysis. The setup requires that the laser ring fall
in the field of view of the camera. The accuracy depends on the fineness of the laser ring and the
resolution of the camera.
The laser ring and camera are typically mounted on a carrier or robotic platform. Gyroscopic position
data (i.e., pitch, yaw, and roll) of this platform are needed to achieve the required precision (Dettmer,
2007). Currently available laser profiling systems are only used in de-watered pipes. To date there is no
known report on underwater laser profiling for in-service water mains. Table 3-8 provides more
information on the laser-based pipe surface profiling technique.
3.3.6 Handyscan 3D. This portable device is a combination of laser and stereo vision (two
cameras) for fast creation of an object surface profile with high resolution as shown in Figure 3-5 and
summarized in Table 3-9. By tracking the laser beam (pattern) and positioning targets (marks on the
surface to match images), separate images acquired by the two cameras are stitched together with the help
of special software. Table 3-9 provides more information on Handyscan 3D.
21
-------
Table 3-8. Laser-Based Pipe Surface Profiling
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other Information
Laser-based pipe surface profiling
Acquire the topography of the pipe surfaces, from which pitting corrosion
can be inferred.
Commercial systems are available. An advanced technique is in
development.
Acuity, 2008; Duran et al., 2003; Dettmer, 2007
Potential to show the early signs of pipe degradation by corrosion
Provides exact geometric dimensions for rehabilitation options.
Enables inspection with minimum lighting requirements.
Measures cross-sectional area.
Can be applied in a wide range of pipe sizes.
Tuberculated pipes need to be scrubbed and cleaned prior to inspection.
Pipeline needs to be de-watered.
Data analysis combines measurements with software and automated
processes.
No documentation is available on capability to detect cracks.
The laser profiling is accurate, but still needs data processing to
compensate for errors introduced during scanning.
• Report on performance study is not available.
The laser profiling technique has been applied to generate pipe inner
surface profile and can also be used for quantifying the outer-surface metal
loss of metallic pipes.
Not available
Figure 3-5. Creaform Handyscan 3D
(Reprinted with permission from Creaform)
22
-------
Table 3-9. Handyscan 3D
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other Information
Handyscan 3D
Scanning process used for industrial design, manufacturing, and inspection
(Creafrom, 2008). It is a non-contact inspection to acquire the geometric
dimension of objects in various environments.
Commercially available device. This technique is still under development
for industrial inspection. A third modality (i.e., laser ultrasound) is being
introduced to this scanner for detecting subsurface conditions by a
company in Belgium (SDCorrosion, 2005).
Creafrom, 2008
• This technique provides more efficient scanning than laser alone.
• No limitation on scan orientation
• Easy to set-up and operate
• The scanner is a portable device, which needs an operator; therefore it is
suitable only for large pipes that allow man-entry or for external
inspection. For the same reason, it does not appear to be a convenient
alternative to scan long stretches of pipe.
• For pipe inspection, the scanning requires a clean surface to map the
corrosion pits. Tuberculated pipes need to be scrubbed and cleaned
prior to inspection.
• Need to set up positioning targets.
Comparative study on the scanning of helicopter tail rotor blades with a 3D
laser scanner from NRC Institute for Information Technology was carried
out. The results were confidential and not available to the public.
The device was exhibited at the 17th World Conference on Non-destructive
Testing (Shanghai, China, 2008). No information is available on its use in
water mains or any other type of pipe.
The scanner, setup, and maintenance cost is low. Learning curve is short.
3.4
Electromagnetic Inspection
3.4.1 Magnetic Flux Leakage. The magnetic flux leakage (MFL) method uses large magnets to
induce a saturated magnetic field around the pipe wall. If the pipe is in good condition, a homogeneous
distribution of magnetic flux is obtained. Anomalies such as metal loss will alter the distribution of the
magnetic flux. Flux leakage is recorded by a detector coil as shown in Figure 3-6. The pipe surface
needs to be cleaned for direct contact with the MFL detection device.
MFL inspection can be used inside the pipe (de-watering required) or outside an exposed pipe (pipe can
be in service). However, it is not possible to inspect small diameter pipe internally due to the mass of the
magnets and steel backups required. The MFL tool provides raw data that need to be interpreted. In the
software developed by Advanced Engineering Solutions, algorithms to identify and characterize the metal
loss are implemented. The raw data are interpreted to defect sizes at a known level of confidence. Table
3-10 summarizes more information on the MFL technology.
23
-------
Magnetic Flax Lines
Sleel
Pipe Wall
Magnetic Sensor
Corrosion Pit
Figure 3-6. The Principle of Magnetic Flux Leakage Inspection
(Makar and Chagnon, 1999)
Table 3-10. Magnetic Flux Leakage
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other Information
Magnetic flux leakage (MFL)
Identify and measure metal loss due to corrosion in ferrous pipes. MFL inspection
can be used inside a pipe or outside an exposed pipe.
MFL systems are available for the oil and gas industry. Commercial MFL system
for wall thickness measurement from outside is also available for metallic water
pipes. More advanced technique, namely pulsed MFL, is being developed.
http://www.ndt-ed.org: Makar and Chagnon, 1999; Wilson et al., 2008; Marlow et
al., 2007
• High degree of accuracy for wall thickness measurement
• External surface inspection does not require a service interruption.
• Using MFL in metallic water pipes requires maintaining close contact with the
pipe wall (Makar and Chagnon, 1999). This contact is strengthened by the
magnetic forces between the tool and the wall, which pull the two together.
• Direct contact with the pipe wall is required and the surface of the pipe must be
clean. Thus, for in-line inspection, MFL is limited to cleaned, unlined metallic
pipes (otherwise, the tool is likely to damage interior coating and slough off
tuberculation).
• It is not possible to develop internal tools to suit small diameter distribution pipes
since the mass of the magnets and steel backups need to be greater than the pipe
wall. Tools for external examination are available for small and large diameter
pipes; however, excavation of buried pipes and replacement of coating or
insulation are required, which make it economically questionable.
• The MFL test needs to be calibrated to interpret the acquired signal.
• It is mainly used for detecting corrosion pits and small defects.
• The detection of pipe wall remaining thickness is quite accurate.
• MFL techniques are generally used in the oil and gas industry for metal loss
detection and are not suitable for internal inspection of small diameter pipes due
to the size of the probes.
• The use of in-line MFL in water industry is limited to cleaned, unlined ferrous
pipes which are accessible.
• Although anecdotal information is available about its use for water mains, in-line
MFL is not widely used due to the high costs associated with it.
The pulsed excitation for MFL has been reported to extract depth information of
defects in rolled steel water pipeline (Wilson et al., 2008). More information will
probably be available from the response of a wider frequency band.
24
-------
3.4.2 Remote Field Eddy Current. A remote field eddy current (RFEC) system consists of an
exciter coil and one or more detectors (see Table 3-11 for more information). The exciter coil is driven
by a low-frequency alternating current signal. The interaction region can be divided into three zones
(Grouse, 2009; Mergelas and Kong, 2001):
(a) Direct coupled zone: in this zone magnetic field from the exciter coil interacts with the pipe
wall to produce a concentrated field of eddy current;
(b) Transition zone: just outside the direct couple zone. In this zone, there is much interaction
between the magnet flux from the exciter coil and the flux induced by the eddy current;
(c) Remote field zone: the region in which direct coupling between the exciter coil and the
receiver coil is negligible.
Table 3-11. Remote Field Eddy Current
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other
Remote field eddy current (RFEC)
Inspect ferromagnetic pipes as well as ferromagnetic components of
composite pipes (Mergelas et al, 2001).
Various proprietary commercial systems are available. Different inspection
systems have been developed for different types of pipe.
Grouse 2009; Mergelas and Kong, 2001; Russell, 2009; Thomson and
Wang, 2009
• Can be applied to different applications (e.g. detect broken wire,
measure corrosion pits).
• Can be operated in wet or dry conditions; therefore, inspection of in-
service pipes is possible.
• Can be used for inspecting lined pipe; direct contact with pipe wall not
required.
• Inspection systems are available for different pipe sizes.
• Data interpretation needs experience and skill.
• Some tools require pipe cleaning and/or dewatering before inspection.
Proprietors do not publish information about false positives/false negatives;
however, RFEC seems to be the prevailing technology in the drinking
water industry for inspection of ferromagnetic pipes and ferromagnetic
components in composite pipes (e.g., PCCP).
• The RFEC/TC technique and P-Wave® are widely used for detecting
broken wires in prestressed concrete pipes.
• The See Snake tool is applied to small diameter ferromagnetic pipes.
• The PipeDiver™ RFEC tool can be used to inspect large diameter, full,
ferromagnetic pipes.
Not available
Two paths exist between the exciter and detector as shown in Figure 3-7. The direct electromagnetic field
inside the pipe is attenuated rapidly by circumferential eddy currents induced in the conducting pipe wall
(Mergelas and Kong, 2001). The indirect field diffuses radially outward through the pipe wall. This field
spreads rapidly along the pipe with little attenuation. These two fields re-diffuse back through the pipe
wall and are dominant at the remote field zone. Any discontinuities in the indirect path will cause
changes in signal magnitude and phase. This is the principle of RFEC testing. This technology does not
require the sensors to be in close contact to the pipe wall.
25
-------
Indirect energy transmission path
ID
2D 3D
Figure 3-7. Remote Field Eddy Current Testing
(Rajani and Kleiner, 2004)
Remote field eddy current/transformer coupling
PCCP pipes have two metallic elements, namely a steel cylinder and steel prestressing wire that is
wrapped tightly around the core concrete to provide it with resistance to tensile stresses. Both metallic
elements interact with the induced magnetic field. The interaction between the indirect transmission path
and the prestressing wire is known as transformer coupling (TC). Thus, the received signal consists of
two components, a remote field component and a TC component. The presence of broken wires will
reduce the response of the transformer coupling component, thus allowing their detection (Figure 3-8).
The remote field transformer coupling technique was developed by the Applied Magnetic Group in the
Department of Physics at Queen's University in Kingston, Ontario, Canada.
Exciter Coil''
Pain
Direct Path
Detector Coil
Direction of Movement
,- Transformer Coupling Path
Figure 3-8. The Breaking Wire Results in a Decrease in the Detector Signal
(Reprinted with permission from PPIC)
The proprietor of the commercial system is the Pressure Pipe Inspection Company (PPIC). The technique
requires analyses and interpretation (proprietary) of the amplitude and phase signals. The amplitude
represents the strength of the transmitted signal while the phase represents the time that the signal takes to
arrive at the detector. According to PPIC, the technique, which is named RFEC/TC can:
26
-------
• Detect broken wires in PCCP;
• Quantify the number of breaks along the length and anywhere around the circumference of
PCCP;
• Quantify the wire breaks in embedded cylinder pipe (ECP), lined cylinder pipe (LCP), and
noncylinder pipe;
• Quantify and spot wire breaks in pipes with or without shorting straps and in pipes with or
without bonding straps.
• Services based on the manned (PipeWalker), tether robotic (PipeCrawler), and free
swimming (PipeDiver™) tools are all commercially available from PPIC.
The free-swimming robotic tool PipeDiver™ was developed by PPIC to carry out in-service inspection of
pipelines with diameters of 600 to 2,000 mm (23.6 to 78.7 in.). The tool is inserted into the live main
using proprietary launch and retrieval devices, which attach to any full-bore tap of least 12 in. in diameter
(18 in. for pipe diameters over 1,000 mm). The tool travels at roughly 90% the flow speed of the water,
and is held in the center of the pipe by flexible fins.
See Snake Tool
See Snake Tool is an RFEC-based technology, developed by Russell NDE Systems Inc., to measure
internal and external corrosion pits in ferromagnetic pipes (Russell, 2009). Tools are available for 50 to
400 mm (2 to 16 in.) and 500 to 700 mm (20 to 28 in.) diameter pipes in wall thicknesses up to 25.4 mm
(1 in.). A picture of the See Snake system is shown in Figure 3-9. The tool can be free swimming or
tethered on a wire line. Lengths up to 3,000 ft can be inspected from one launch point when wire line
tethered (Thomson and Wang, 2009). The free swimming version can inspect lengths up to 15,000 ft
from the launch point. The features of the See Snake tools include:
• Can negotiate multiple 90° welded elbows.
• Completely water and pressure proof for water and wastewater line pressures.
• Inspection speed is approximately 0.2 to 0.5 km/hour.
• Can measure remaining wall thickness, surface area (length and width), and stress.
• Can be tracked and detected from above ground.
Figure 3-9. The See Snake Tool for Inspection of Pipe Internal and External Flaws
(Reprinted with permission from Russell NDE Systems Inc.)
27
-------
The technique is able to distinguish internal defects from external ones with the addition of extra sense
coils (not included in standard tools). Currently, See Snake tools are used to inspect water, wastewater,
oil and gas pipelines with and without internal liners, downhole casings and large diameter raw water
pipelines.
P-Wave®
P-Wave® is an RFEC-based system, developed by Pure Technologies, for the condition assessment of
PCCP. P-Wave® detects breaks in prestressed wire and estimates the total number of breaks for each pipe
section. The P-Wave® system can traverse the pipeline in either a manned or robotic manner. In the
manned inspection, the P-Wave® system is pushed through a dewatered pipeline. In the robotic
inspection, the pipe must be depressurized. In general, a manned inspection is preferred to facilitate a
close-up visual inspection of interior surface of the pipe wall. However, if manned entry is not feasible
due to pipe diameter and confined space entry requirements, a robotic inspection is applied (Figure 3-10).
The system has a variety of configurations to accommodate all diameters of PCCP. The number of
broken wires is derived from the acquired inspection data and will be used further as an input for a
structural model.
Figure 3-10. P-Wave® System for Manned and Robotic PCCP Inspection
(Reprinted with permission from Pure Technologies)
3.4.3 Broadband Electromagnetic. Unlike the conventional eddy current technique, which uses a
single frequency for testing, the broadband electromagnetic (BEM) technique transmits a signal that
covers a broad frequency spectrum (Hazelden et al., 2003). A transient input signal generates multiple
frequencies, typically ranging from 50 Hz to 50 kHz. The recorded signal from a broadband transmission
contains more information, and allows detection and quantification of various wall thicknesses as well as
28
-------
the effective conductivity of the complex through-wall components of the pipe. Changes in conductivity
reflect changes in material properties.
A transmitter coil passes an alternating current to the pipe surface, which generates an alternating
magnetic field. Flux lines from this magnetic field pass through the metallic pipe wall, generating a
voltage across it. This voltage produces eddy currents in the pipe wall, which induce a secondary
magnetic field. Wall thickness is indirectly estimated by measuring signal attenuation and phase delay of
the secondary magnetic field.
External scanning requires excavation of buried pipes. Internal scanning can be carried out with an inline
inspection pig, which is driven by hydraulics or by push/pull rod devices (Thomson and Wang, 2009).
However, the pipeline needs to be out of service, emptied, and cleaned of loose deposits in order to run
the pig. In CI, BEM can identify and locate metal loss and cracks. It does not require contact with bare
metal to detect pits and other metal loss. It is possible to apply BEM for pipeline assessment through
keyholes (GTI, 2005). Refer to Table 3-12 for more information on BEM.
Table 3-12. Broadband Electromagnetic
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Broadband electromagnetic (BEM)
Detect and quantify wall thickness, as well as the effective conductivity of the
complex through-wall components of the ferrous pipes.
Commercially available from Rock Solid Pry. Ltd. A hand-held tool based on
the same principle is also available from the same company to measure
corrosion pits.
Feeney et al., 2009; Hazelden et al., 2003; Thomson and Wang, 2009
• Does not require contact with the metallic pipe wall and is not sensitive to
the corrosion products.
• Can scan through coatings, linings, and insulation with a penetration depth
of 2.5 times the transmitter diameter.
• Measures average thickness in the area under the sensor's footprint; the
resolution of the scan depends on the size of the sensor; unable to detect
pin-hole failures or isolated pits.
• For in-line inspection, pipe needs to be emptied and cleaned.
• The inspection process is time consuming because the scanning process is
not continuous.
• The mean value of wall thickness is measured for a square grid.
• A surface scratch or an isolated pit smaller than the square grid will not be
detected.
• BEM technology has been primarily used for condition assessment of water
mains.
• It can only be used on ferrous materials.
• BEM can be used to measure wall thickness, quantify graphitization, and
locate broken wires in PCCP (Feeney et al., 2009).
• Inspection of a 760 mm (30 in.) cast iron and steel lines was reported
(Hazelden et al., 2003).
• Information about the limits on pipe size is not available.
The BEM system is being further modified to facilitate the inspection of pipes
exposed in keyhole excavations. This will help acquire information about pipe
condition without disrupting service or full access excavations.
29
-------
3.4.4 Pulsed Eddy Current System. Pulsed eddy current is a successful method to determine wall
thickness of insulated and non-insulated steel pipelines from external inspection (Waters, 2005). A
rectangular shaped eddy current is generated by a transmitter coil. Each cycle consists of one positive and
one negative pulse. The strength of the eddy currents is measured at some distance from the pipe wall
(e.g., due to liftoff or insulation thickness) by quantifying the magnetic reaction field picked up by the
receiver coil (Waters, 2005). The strength is related to wall thickness. It computes the average thickness
of the metal by comparing the transient time of certain signal features with similar calibrated signals
(Waters, 2005). The contact between the magnetic field and the inspected component produces a
footprint that represents the area inspected for wall thickness calculation. The diameter of the footprint
varies between 25 and 150 mm (1 to 6 in.), depending on wall thickness, insulation thickness and sensor
size. The inspection tool is compact and can be easily deployed by remotely operated vehicles. See
Table 3-13 for more information on the pulsed eddy current system.
Table 3-13. Pulsed Eddy Current System
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Pulsed eddy current system
Primarily an external method for detecting corrosion in ferrous pipes and
vessels without removing insulation, fireproofing concrete or similar coatings
(MB Inspection, 2008).
Commercial available (from Applus+RTD, formerly PNDT).
MB, 2008; Waters, 2005; httD:/www.pndt.com.au
• Unaffected by the presence of insulating coatings and no need to remove
them.
• All commonly used insulating materials like glass wool, rock wool,
asbestos, polyurethane foam, scales of silicate, concrete and all kinds of fire
proofing have no influence on the magnetic field and induced eddy
currents. However, the binding ties, fitting supports, fixing materials and
composition of the weatherproofmg have influences on the examination.
These influences can be compensated by properly tuning the measurement
parameters.
• Can operate submerged (sub-sea inspection).
• Interpretation of the signal requires a high level of skill. The pulsed eddy
current data needs to be analyzed carefully because results are highly
sensitive to variations in factors such as lift off and air gap.
• The measurement result is affected by a number of factors including
variations in metallurgy and temperature.
• The size of the instrument's footprint will mask small areas of localized
steel loss and appropriate selection of the sensor head (from 30 to 200 mm
in diameter) is essential.
PNDT claims that the instrument is capable of high accuracy and good
repeatability.
• Used for inspection of insulated pipe/vessels in chemical plants and the oil
and gas industry. Actual numbers were not reported.
• In-line operation is possible with battery supply without disrupting service.
Not available
30
-------
3.4.5 Ground Penetrating Radar. Ground penetrating radar (GPR) antennae transmit
electromagnetic wave pulses into the ground. These pulses propagate through the ground and reflect off
sub-surface boundaries. The reflections are detected by a receiving antenna and subsequently interpreted
(Costello et al., 2007). See Table 3-14 for more information on GPR.
Table 3-14. Ground Penetrating Radar
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Ground penetrating radar (GPR)
Acquire subsurface information. It can be used to locate buried assets, such as
plastic or clay pipes.
Conventional GPR systems are commercially available. A prototype for
ground penetrating image radar was recently developed.
Costello et al., 2007; Makar, 1999; Marlow et al., 2007
• Able to locate pipes of all materials
• Inspection can be performed from the surface non-intrusively or from
within the pipe for more detailed information. Antenna does not have to
touch the pipe surface.
• Relatively high inspection speed
• A GPR survey also provides information on the condition of the soil
surrounding the pipe and details of voids.
• Air gap and variations in soil conditions will affect the GPR result;
• The pulses lose strength very quickly in conductive materials, such as clay
and saturated soils, which is a limitation for these soil types.
• Limited ability to detect assets below the water table.
• Data interpretation needs highly skilled operators.
• The performance of GPR is highly dependent on soil conditions.
• No evidence of consistent ability to detect voids with GPR.
• Substantial operator interpretation of results is necessary (Makar, 1999).
Limited use for locating non-metallic pipes and detecting pipe leakage.
Significant work needs to be done to process GPR data and signals.
Conventional GPR systems are operated from the ground surface. In-pipe GPR systems were also
reported (Costello et al., 2007). Such systems use two or three antennae with different frequencies to
investigate the structure of the surrounding soil, the interface between the soil and pipe, and the structure
of the pipe. GPR can potentially identify leaks in buried water pipes either by detecting underground
voids created by the leaking water or by detecting anomalies in the depth of the pipe as the radar
propagation velocity changes due to soil saturation with leaking water (Hunaidi and Giamou, 1998). The
GPR technique was also applied to determine the degree of internal leaching of hydroxides in AC pipes
(Slaats et al., 2004).
A prototype ground penetrating imaging radar (GPIR) was recently developed within a European
Commission supported project "WATERPIPE" (WATERPIPE, 2009b). This high resolution GPIR is
designed to detect leaks and image damaged regions in pipes. The capabilities of this high resolution
GPIR reportedly include:
• Locate water pipe of all types of materials;
• Detect leaks and damages in water pipelines of all types of materials;
• Penetrate the ground to a depth of up to 200 mm (78.74 in.);
• Achieve an image resolution of less than 50 mm (1.96 in.);
• Survey velocity at approximately 0.36 km/hr (0.22 mi/hr).
31
-------
The measurement results currently available were obtained in a laboratory environment. The inspection
results were used to assess the structural reliability, leakage, and conformity to water quality standards of
the pipes (WATERPIPE, 2009b).
3.4.6 Ultra-Wideband Pulsed Radar System: P-Scan. P-Scan is based on ultra-wideband
(UWB) antennae capable of transmitting and receiving electromagnetic pulses in the nano- and pico-
second ranges (see Table 3-15). For the inspection of buried pipes, it is desirable to operate in the
picoseconds range because pulse widths in this region are equal to or less than the wall thickness of most
non-ferrous buried pipes. The pulse repetition frequency (PRF) ranges from thousands to several billion
pulses per second. Numerical experiments demonstrated the potential of this technique for pipe condition
assessment. The use of ultra-short duration pulses makes it possible to obtain relatively high resolution
results.
Table 3-15. UWB Pulsed Radar System: P-Scan
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
UWB pulsed radar system: P-Scan
Detect below surface defects, corrosion, and out-of-pipe voids in non-metallic
buried pipes (Allouche, 2007; Jaganathan et al., 2006). The UWB inspection is
capable of providing higher resolution images of the pipe wall and a greater
penetration depth than high-frequency GPR.
Numerical simulation for P-Scan has been carried out and a pre-commercial
prototype is not available yet. The system is still under development.
Allouche, 2007; Jaganathan et al., 2006
• Accurate measurement of wall thickness of pipes
• Increased resolution of images
• The pipe wall thickness and other distinct layers can be measured in a
continuous manner. Forward processing algorithms can be used to back
calculate the dielectric constant of the various materials.
• Capability to inspect not only the pipe wall, but also the pipe liner.
Not yet determined.
Not yet determined.
Not yet determined.
Not yet determined.
3.5
Acoustic Inspection for Structural Condition
3.5.1 Sonar Profile System. Sonar is an acoustic detection technology designed to operate under
water (see Table 3-16). In the pipe inspection field, it has been adapted to provide information about
elements in the pipe that are submerged below the water line. These may include submerged debris in the
pipe (sewers), grease level (sewers), differential settling and other submerged deformations and defects.
A sonar system may consist of an underwater scanner unit, collapsible sonar siphon float, sonar
processor/monitor, skid set, and all necessary interconnect cables (CUES, 2008). It typically travels in
pipes at velocities in the range of 0.1 to 0.2 m/s and sends a pulse approximately every 1.5s. Each pulse
provides an outline of the cross-section of the submerged part of the pipe (CUES, 2008). Accurate
measurements can be performed based on these outlines.
The sonar profiling system can be used with different frequencies to achieve different goals (RedZone,
2008). High frequency sonar can provide a higher resolution scan, but a high resolution pulse attenuates
quickly and therefore has a relatively low penetration capability. In contrast, low frequency sonar has a
32
-------
high penetration capability but is limited in its scanning resolution. Consequently, whereas high
frequency sonar can be suitable for clear water conditions, turbid water with high concentrations of
suspended solids may require a lower frequency signal. Small defects are more likely to be observed by a
high frequency signal. Some systems are capable of a multi-frequency scan to obtain maximum
information.
Table 3-16. Sonar Profile System
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Sonar profile system
Provides visual profile, profile comparison, and dimension data of significant
items or defects on internal pipe below waterline.
Commercially available.
RedZone, 2008; CUES, 2008
• Can be operated on a robotic platform in both fully charged and partially
charged lines without disrupting the service (sewers).
• Can work in conjunction with a CCTV system in the inspection of semi-
submerged pipes.
• Must be operated under water.
• Limited by the operating frequency
Can generate precise pipe cross-section via dwell scan.
• Applied widely to the inspection of sewers
• No data found about its use in water mains.
• A system that integrates sonar and video for use in submerged and large
semi-submerged pipelines is also available.
• The cost of sonar inspections varies depending on the diameter of the pipe
to be inspected.
3.5.2 Impact Echo. Impact echo testing is based on the use of impact-generated stress waves that
propagate through and are reflected by the object under test (see Table 3-17). The impact echo equation
is (Sack and Olson, 1998):
= V/(2Fp)
where
T is thickness;
V is wave speed
Fp is peak frequency.
The time domain test data of the impulse hammer and accelerometer are transformed to the frequency
domain as illustrated in Figure 3-11. A transfer function is computed between the hammer and receiver
as a function of frequency. Peaks in the transfer function reflect the thickness of the pipe wall at the test
location. A more complicated model would be required to discern other properties of the object under test
from frequency responses.
The test can be performed on concrete, stone, plastic, masonry materials, wood and some ceramics.
Testing is conducted by hitting the test surface at a given location with a small instrumented impulse
hammer or impactor and recording the reflected wave with a displacement or accelerometer receiver
adjacent to the impact location (Sansalone and Streett, 1998). The receiver is mounted to or pressed
against the test surface.
33
-------
Table 3-17. Impact Echo
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Impact echo
Determine the location and extent of flaws such as depth and width of surface
cracks, delamination, voids and other damages. Application suitability
depends on the properties and internal structure of the material being tested
(Marlowetal.,2007).
Various instruments are commercially available.
Marlow et al., 2007; Sack and Olson, 1998; Sansalone and Streett, 1998
• The impact echo test can be applied to varied materials.
• The test is easy to carry out.
• Works through paints, coatings, and tiles.
• Only one side of the structure needs to be accessible for testing.
• Frequency domain analysis is complicated when information other than
thickness and geometry is needed and experience is required.
• Embedded items may affect wave behavior and test results.
• This method is not limited by pipe size and can be applied both internally
and externally only if the testing is executable.
• Not applicable to metals
• Accuracy is typically 2% at high resolution when properly calibrated on a
known thickness location (Marlow et al., 2007).
• The typical thickness for the impact echo testing ranges from 66 mm to 1.8
m (2.6 to 70.9 in.).
• Extensively used on flat surfaces (concrete slabs, bridge decks, etc.)
• Also used for inspection of water and sewer PCCP and concrete pipes,
usually large diameter pipes with man access.
Not available
.
Figure 3-11. Impact Echo Testing
34
-------
3.5.3 Acoustic Emission. Acoustic emission testing (see Table 3-18) is based on the detection of
sound waves generated from within the material itself (e.g., when a crack propagates). The monitoring
sensors are placed in or on the pipe to monitor acoustic activity. Signals obtained by the monitors are
typically compared to a library of acoustic signatures of known events (e.g., a wire break in PCCP) to
identify activities. The sensors used for acoustic monitoring include (Higgins and Paulson, 2006):
• Hydrophone arrays: multiple hydrophones are mounted on a cable with specific spacing.
• Hydrophone station: single hydrophones are inserted into the water flow at convenient
locations.
• Surface mounted sensor: piezoelectric sensors are placed on the surface of the pipe or
appurtenances along the pipe.
• Fiber optic sensor: long jacketed cable containing glass fiber sensor is inserted into the pipe.
• Microelectromechanical system (MEMS) acoustic emission sensor: four resonant sensors of
frequency range 100 to 500 kHz are integrated on a 5 mm (0.196 in.) square chip (Grevea et
al., 2008).
• Wire-guided transducer: wire-guided transducer uses a steel wire to acoustically couple a
piezoceramic wafer to a test structure (Neilla et al., 2007).
The two important variables for acoustic monitoring are sensor spacing and monitoring duration. The
acoustic sensor should be spaced close enough to ensure two sensors detect the acoustic event and have
sufficient acoustic information to identify the source. The short-term monitoring installs the acoustic
sensors temporarily, while the long-term monitoring needs a permanent installation of the sensors for
continuously tracking the performance of a pipe with time.
Table 3-18. Acoustic Emission
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Acoustic emission
Monitor the acoustic emission when a sudden appearance or propagation of a
microscopic crack occurs within a material under load or the break of
prestressed wire in PCCP (Marlow et al., 2007).
Acoustic emission sensors (e.g., hydrophone, surface mounted sensor, and
fiber optic sensor), are commercially available. New sensors are being
designed and tested (e.g., MEMS acoustic emission sensor and wire-guided
transducer).
Grevea et al., 2008; Neilla et al., 2007; Higgins and Paulson, 2006; Holley
and Buchanan, 1998
Implement real-time online monitoring.
• Can only detect what is happening during monitoring period (no indication
about past deterioration);
• Installation of sensors may need interruption of service;
• Quantitative information (e.g., size) about the crack is not available.
Not available
Acoustic emission test is being applied to detect wire breaking and leakage of
pipelines.
Not available
35
-------
3.6
Acoustic Inspection for Leak Detection
3.6.1 SmartBall®. SmartBall® comprises a range of acoustic sensors, as well as an accelerometer,
magnetometer, ultrasonic transmitter, and temperature sensors, which travel with the water flow down a
pipe and detects, locates, and estimates the magnitude of leaks as it rolls. The acoustic sensors are
encased in an aluminum alloy core with a power source and other electronic components (Fletcher, 2008;
Pure Technologies, 2009). The core is encapsulated inside a protective outer foam shell or sphere (see
Figure 3-12). The outer foam shell provides additional surface area to propel the device and also
eliminates the noise that the device might generate while traversing the pipeline. The diameter of the
outer sphere depends on the pipe diameter and flow conditions. See Table 3-19 for more information on
the SmartBall® technology.
Figure 3-12. Pictures and Illustrations of SmartBall®: internal view (left) and external view (right)
(Reprinted with permission from Pure Technologies)
The SmartBall® is deployed into the water flow of a pipeline and captured at a downstream point. It
continuously records acoustic data and emits an acoustic pulse every 3 seconds for tracking purposes
while the device traverses the pipeline. A SmartBall® Acoustic Receiver, which is patented by Pure
Technologies, is used to track the location of the ball. The above-ground markers can be laid at 2 km
36
-------
intervals and leak locations can be determined within 1 m. The recorded acoustic data are analyzed to
identify air pockets and leaks. Other sensory data are used to determine the location of air pockets and
leaks. The severity of leaks is estimated by calibrated baseline data. Frequency analysis needs to be
carried out to confirm that an acoustic anomaly is actually a leak.
A resilient elastomeric coating is placed around the ball to minimize background noise, while the ball
rolls through the pipe. The inspection route needs to be carefully planned to ensure that the ball does not
block bypass lines. The effect of offtakes should also be considered. As the ball is smaller than the inside
diameter of the pipe, with the required amount of fluid, the ball can traverse the pipe without any
difficulties.
Table 3-19. SmartBall8
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
SmartBall®
Detect leaks and air pockets in medium and large diameter (8 in. and greater)
water and wastewater pipes.
Commercially available (from Pure Technologies) since 2006.
Fletcher, 2008; Pure Technologies, 2009;,
http://www.puretechnolosiesltd.com/html/smartball water.php
• Can be used for any pipe material (concrete, steel, PVC, GRP, etc).
• Can be applied to detect air pockets and leaks on medium and large
diameter pipe (> 8 in.).
• Can survey long pipelines with a single deployment. The total length of
survey capacity depends on flow rates in the pipeline and battery life. The
longest water line survey presently is 15 miles (25 km) under 2 f/s flow.
For higher flow rates, longer surveys could be performed.
• Can detect very small noise disturbances along the pipeline.
• Inspection is performed while a pipeline remains in service.
• The conventional SmartBall® cannot be used for pipelines with very high
water pressure (> 400 pounds per square inch [psi]).
• If the survey involves long pipe lengths, the surface sensor used for
monitoring the pulses being emitted from the SmartBall® has to be moved
along the pipe length.
• The estimation of the leak magnitude is qualitative.
• As reported by Pure Technologies, the device can detect leaks of less than
0.026 L/hr (0. 1 gal/hr) under ideal conditions (high pressure and low levels
of ambient noise) (Pure Technologies, 2009).
• Location accuracy depends on how well the configuration of a pipeline is
known. Typically, the location accuracy of the device is within 3 ft (1 m).
• SmartBall® is a relatively new technology and has seen significant entry
into the marketplace.
• It has been used in many countries, including the U.S., Canada, and Mexico,
on a wide range of pipe materials.
• It was commercially introduced in late 2006 and, as of August 2009, has
been deployed through more than 900 miles of pressure pipe.
Further development of SmartBall® technology for natural gas pipeline
applications is being supported by research funding from the U.S. Department
of Transportation Pipeline and Hazardous Safety
Administration. http://primis.phmsa.dot.sov/matrix/PrjHome.rdm?pri=234
37
-------
3.6.2 LeakfinderRT™. As illustrated in Figure 3-13 and summarized in Table 3-20, the
LeakfinderRT™ system is composed of leak sensors, a wireless signal transmission system, and a
personal computer. Acoustic sensors, such as accelerometers or hydrophones, are attached to two contact
points on the pipe, such as a fire hydrant. Accelerometers are used to sense leak-induced vibration, while
hydrophones are used for sensing leak-induced sound in water column. Accelerometers are sensitive to
background noise and hydrophones are often used together with accelerometers to achieve a better signal
to noise ratio.
Ssnsor
Computer
\^
y a
--RF
Receiver
Leak
Pipe
Figure 3-13. Principle of LeakFinderRT™
(Hunaidi et al., 2004)
The computer calculates the cross-correlation function of the two leak signals to determine the time lag
(Vax) between the two sensors. Then the location of the leak can be derived from the equation below:
D-C-T
and L2=D-Ll
where
Lj and L2 are the positions of the leak relative to sensors 1 and 2, respectively;
c is the propagation velocity of sound in the pipe;
D is the distance between location 1 and 2.
Propagation velocity is determined experimentally or estimated based on the type and size of the pipe.
LeakfinderRT™ uses a patented, enhanced cross-correlation function that is calculated indirectly in the
frequency domain using the inverse Fourier transform of the cross-spectral density function rather than
using the shift-and-multiply method in the time domain (Hunaidi et al., 2004). The enhanced correlation
function provides improved resolution for narrow-band leak signals. This is very helpful for plastic pipes
(low frequency sound emission), small leaks, multiple leaks and situations with high background noise.
Moreover, a major advantage of the enhanced function is that it does not require the usual filtering of leak
signals to remove interfering noises (Hunaidi et al., 2004).
38
-------
Table 3-20. LeakfinderRT
.TM
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
LeakfinderRT 1M
LeakfinderRT 1M is a computer-based system for locating leaks in all types of
water and other fluid transmission and distribution pipes (Echologics, 2009;
Hunaidi et al., 2004).
LeakfinderRT 1M is commercially available (from Echologics) since 2002.
Hunaidi, 2006b
• Non-intrusive tool used to locate leaks in pipes.
• Uses a proprietary, enhanced correlation method, which improves the
effectiveness of locating leaks in all types of pipes including plastic pipes.
• Effective for small leaks and for situations with high background noise
• Uses a low-frequency vibration sensor to locate leaks in plastic pipes.
• Correlation implemented using software rather than hardware.
• Information about the leak size is not available from the test.
• Sensor spacing is influenced by both the pipe diameter and pipe material
due to the attenuation of the acoustic signal. More signal attenuation is
experienced the larger the diameter of the pipe and the less rigid the
material. This effect is present in all pipe types, but is most pronounced in
PVC and PCCP due to their material properties.
• Maybe susceptible to interference from low-frequency vibrations (e.g.,
pumps and road traffic).
The performance of the LeakfinderRT 1M system has been successfully tested
for the following scenarios (Hunaidi et al., 2004):
• Narrow-band leak noise in PVC pipes
• Small leaks in PVC pipes under a very low pressure of 20 psi
• Locating small leaks in metal pipes
• Effective for situations with high background noise
• Improved peak definition for resolving multiple leaks
• The smallest PVC pipe leaks detectable with LeakfinderRT™ 's low
frequency vibration sensors (1.7 L/min) and hydrophones (0.85 L/min).
• Theoretical leak location error is less than 10 cm. Actual error depends on
accuracy of sensor spacing and propagation velocity; The distance between
acoustic sensors is determined by the pipe materials and size.
• The monitoring duration depends on the quality of the signal. More signal
with much noises need a longer monitoring time.
LeakfinderRT 1M is used for locating leaks in all types of water and other fluid
transmission and distribution pipes.
Based on principles similar to LeakfinderRT 1M, a technique was developed
(WallThicknessFinder) and patented (but not yet commercialized) to estimate
the average pipe wall thickness between two 'listening' points on the pipe
(Hunaidi, 2006b). The average thickness of the pipe section between two
acoustic sensors can be back calculated from a theoretical model, which
incorporates the acoustic velocity, pipe diameter, Young's modulus of the pipe
wall, and the bulk modulus of elasticity of water (Hunaidi, 2006a). Velocity
measurement can be performed with the same hardware as LeakfinderRT™ by
using the cross-correlation method.
Leak signals are measured using either vibration sensors or hydrophones. Accelerometers, which sense
the acceleration of vibration induced by leak signals in the pipe wall or fittings, are normally used to
measure leak signals in metal pipes (Hunaidi et al., 2004). Sensors can be attached to the pipe directly; if
not, they can be attached to fire hydrants or to the underground valves. Hydrophones are used through
39
-------
fire hydrants to pick up the leak signals propagating through water. It is good for non-metallic pipes.
LeakfinderRT™ has a special low-frequency vibration sensor, which is more effective than
accelerometers.
Signals from leak sensors can be transmitted wirelessly to a computer for processing. Leak sounds are
recorded and correlated by LeakfinderRT™ in a few minutes, but for noisy signals a longer duration is
required. The cross-correlation results are displayed on screen and are continuously updated in real time,
while leak signals are being recorded.
3.6.3 Permalog®. Permalog® is a semi-permanently or permanently installed system for detecting
and logging leak noise in water distribution systems (Fluid Conservation Systems [PCS], 2011). The
loggers (Figure 3-14) are installed on pipe fittings and valves and are retained in place by magnets and
powered by replaceable batteries. The logger is 4.85 in. tall by 1.95 in. wide, weighs 1.5 Ibs, and operates
between 902 to 928 MHz.
Figure 3-14. Picture of Permalog (Courtesy of www.hwm-water.com)
The noise loggers typically operate during the night when background noise is lowest and pressure is
highest. If no leak is present, a radio signal transmits to indicate normal background conditions, but as
soon as a possible leak is detected, the unit sounds an alarm and transmits a radio signal to indicate a leak
condition (Butler, 2009). The logger has changeable alarm threshold settings and the data can be
accessed by three methods: (1) lift and shift - the loggers are removed from the ground and the data is
manually retrieved; (2) drive by - the data is transmitted via radio to a moving patrol vehicle using a
patroller system; and (3) PermaNet - the data is transmitted directly to an office computer via radio
network. See Table 3-21 for more information on the Permalog® technology.
The Permalog® system has been deployed by several water utilities such as West Virginia American
Water, Birmingham Water Works Board, and Las Vegas Valley Water District to locate leaks.
Permalog® has been deployed by West Virginia American Water as part of an advanced metering
infrastructure (AMI) system for an area serving around 12,000 customers (Hughes, 2011). Birmingham
has used Permalog® since 2004 and the devices have located more than 700 leaks and helped to reduce its
non-revenue water rate by 57% (Birmingham Water Works Board, 2009). Las Vegas has used the
technology since 2004 to locate more than 1,300 leaks and estimates they have saved more than 109
million gallons of water (Las Vegas Valley Water District, 2011).
40
-------
Table 3-21. Permalog® Technology
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Breadth of use
Permalog®
Continuous monitoring and leak detection for water distribution
systems.
Developed by PCS and commercially available from Raima Water
Management
www.hwm-water.com/leakDetectionPermalog.htm
www.datamatic.com/product docs/Permaloa.pdf
www.fluidconservation.com/permaloa+AMR.htm
Hughes, 2011; BWWB, 2009; LVVWD, 2011; Butler, 2009
• Can be permanent, semi -permanent, or survey (as required by area).
• Responds to new leaks and breaks in a timely manner.
• Automated leak surveying
• Non-invasive method with no detrimental effects on the customer
supply
• Can be quickly deployed and used repeatedly without disruption to
the surrounding area.
• Low cost battery replacement with minimum maintenance (battery
lasts 5 or more years depending on mode of operation)
• Monitoring length varies based on pipe material, with plastic pipe
requiring closer spacing than metallic pipe.
• Background noise can create issues in finding leaks.
Over 200,000 units in use worldwide and used by more than 200 U.S.
water utilities (PCS, 2011).
3.6.4 MLOG™. MLOG™ is a permanently installed acoustic monitor used for locating leaks in
water distribution systems. The monitoring device, contained in a black polycarbonate and brass hosing
(Figure 3-15), is installed near the water meter and powered by an AA lithium battery with a battery life
of 10 years or more. The device is 4.8 in. tall by 2.58 in. wide and operates at a frequency of 915 MHz.
Once the sensors are installed near the water meters every 500 ft, readings are taken each night and the
data are sent for analysis. The network monitoring system then computes a leak index for each MLOG
sensor and assigns a leak status as either: no leak; possible leak; probable leak; or out of status. Next, a
communication module generates reports to direct leakage investigations and pinpointing activities. See
Table 3-22 for more information on the MLOG™ technology.
Figure 3-15. Picture of MLOG™ (Courtesy of www.itron.com)
41
-------
Table 3-22. MLOG™ Technology
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
MLOG™
Continuous monitoring and leak detection for water distribution systems
Developed by Flow Metrix and commercially available from Itron
www.itron.com/na/productsAndServices/Paaes/MLOG.aspx?market=water
Hughes, 2011;
• Can help reduce water loss in the distribution system.
• Can optimize system maintenance by locating pipeline leaks.
• Can improve effectiveness of water conservation.
• Low cost battery with minimum maintenance (battery last 10 or more
years)
• Sensor spacing is limited by metal covered meter pits (up to 100 ft) and
obstructed views (up to 300 ft).
• Background noise can create issues in finding leaks.
American Water has successfully piloted the MLOG technology at multiple locations including
Connellsville, PA in 2005 where the non-revenue water was reduced from 25% to 12% in the first year,
resulting in an estimated savings of $175,000 (Malone and Morgan, 2006). New Jersey American Water
tested the technology in Irvington, NJ on a system which serves 9,000 customers. American Water
largest deployment of MLOGs is California American Water's Monterey system, where 4,100 devices
have been installed (Hughes, 2011). MLOG devices were also deployed in Clayton County, GA in March
2008 and the 585 sensors identified 11 leaks in the oldest part of the network, which totalled to a savings
of more than 54,662,400 million gallons per year at a production cost of $41,000 (Itron, 2011).
3.6.5 STAR ZoneScan™. STAR™ ZoneScan™ (Figure 3-16) is an acoustic leak detection
system that is installed on the operating nut of water valves via a magnetic bottom. The system, which
can be deployed permanently or temporarily, analyzes noise on water lines at scheduled times to pinpoint
the location of leaks. The battery lasts 10 or more years.
Figure 3-16. Picture of STAR™ ZoneScan™ (Courtesy of www.aclaratech.com)
42
-------
3.6.6 Sahara®. The Sahara® system uses a hydrophone tethered to an umbilical cable, which
travels inside in-service water mains to record leak noises (Costello et al., 2007; Mergelas and Henrich,
2005). A locator beacon can be tracked on the surface, enabling leaks to be marked for excavation and
subsequent repair (PPIC, 2006).
Sahara® locates leaks through identifying the distinctive acoustic signals generated by leaks in the pipe
wall, the joints or steel welds. The magnitude of the leaks can also be estimated from the acoustic signal
(PPIC, 2006). Gas pockets in the pipeline are also detected by their unique acoustic signature. Figure 3-
17 shows the Sahara® system in use. See Table 3-23 for more information on the Sahara® system.
A video and lighting sensor is also available on the Sahara® platform to provide CCTV inspection of in-
service potable water pipelines. Wastewater force mains have also been successfully inspected by
flushing the line with clean water during the inspection.
An average wall thickness calculation across a set interval of pipe (typically 30 ft) can be provided based
on speed of sound measurements taken with the Sahara® system (in developmental stage).
The Sahara® sensors are launched into in-service water mains through a launching chamber that is
mounted on a 2 in. (50 mm) or larger access hole. A small parachute uses the flow of water to draw the
sensor through the pipeline; alternatively, a pre-installed pull-tape can be used to draw the sensor through
the line when no flow is available, such as on pre-commissioned pipelines. The sensor is tethered to the
surface control unit. The sensory data are displayed in real time.
Surface Locator
•i
Cable Drum &
Operations Unit
Figure 3-17. The Sahara® System
(Reprinted with permission from PPIC)
43
-------
Table 3-23. Sahara Leak Detection and Sahara Condition Assessment
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Sahara*1 leak detection and Sahara18 condition assessment
Inspect in-service water mains for leaks, gas pockets, visible defects, and wall
thickness of metallic pipe (with acoustic technology).
Commercially available for leak detection since mid-1990s. First developed
by the Water Research Centre (WRc) in the UK. PPIC acquired the worldwide
right for Sahara® leak detection in 2008.
Costello et al., 2007; Mergelas and Henrich, 2005;
http://www.ppic.com/services/sahara.shtml
• Can be used for in-service pipe inspection.
• Can use existing 2-in. taps.
• Sensitive to small leaks
• Surface tracking can map the pipeline under inspection.
• Can be used for small mains (4 in.) and equally effective in large diameter
mains because of the proximity of the sensor to the leak.
• Tether control allows withdrawal of the sensor when unexpected flow
conditions are encountered. It also allows extending the listening period at
a particular location, if needed.
• Intrusive technology
• Requires access points at a frequency that is determined by bends and flow
rates in the pipeline.
• Buried unknown leaks as small as 0.25 gal per hour have been successfully
located.
• The accuracy of locating a leak is generally less than 1 m (40 in.).
The Sahara® system is used for detecting leaks, pockets of trapped gas, and
structural defects in large mains. In 2007, Sahara® live CCTV inspection was
introduced to the market. It was also reported that the Sahara® in-line platform
had been used to identify wall thickness loss levels of a 48 in. cast iron pipe.
The potential for using the Sahara® system as a platform for other sensors is
being explored.
3.7
Ultrasonic Testing
Ultrasonic testing (UT) is carried out by sending high frequency sound into the object under inspection
and analyzing the received echo. UT has been widely applied for thickness measurements, corrosion
monitoring, delamination checks, and flaw detection on forgings, castings, and pipes.
3.7.1 Guided Wave Ultrasonic Testing. The guided wave ultrasound technique is based on the
capability of propagating waves for a long distance (Rose et al., 2008). Depending on the type of guided
wave, the number of transducers can range between two and four. Torsional waves require two
transducers, while longitudinal waves require three to four transducers. Torsional or longitudinal guided
waves are induced into the pipe and propagated along the length of the pipe segment. A torsional wave
system can be used in pipes filled with water, while the longitudinal system cannot. In a longitudinal
system, three transducers can only operate on a single frequency. Multiple frequencies can be applied if
four transducers are used; this arrangement leads to an improved inspection result.
When these guided waves meet an anomaly or pipe feature, waves reflect back to the transducer's original
location. The time-of-flight for each signature is calculated to determine its distance from the transducer.
The amplitude of the signature determines the size of the defect. See Table 3-24 for more information.
44
-------
A probe in the form of a ring array of piezoelectric transducers is clamped around the pipe and an
ultrasound is sent simultaneously in both directions along the pipe. The acquired signal is similar to
conventional UT A-scans. The horizontal axis represents the distance along the pipe while the vertical
axis represents signal magnitude, which can be used to characterize metal loss due to corrosion. This
technique is suitable for pipes above 50 mm (1.97 in.) in diameter and wall thicknesses up to 40 mm (1.57
in.). Inspection for an elevated pipe can be conducted for a range of up to 30 m (98.4 ft) in either
direction from a specific spot where the probe is placed. This technology is generally applicable to steel
and iron pipe materials. Trials of guided wave systems on steel water mains are described in Reed et al.
(2004). For CI and DI pipe, the most prominent pipe feature is the bell and spigot joint, which would
reflect the propagating wave and therefore limit the inspection to one pipe length for external inspection
tools. The U.S. EPA is sponsoring a grant to research the use of ultrasonic guided waves (using in-situ
magnetostrictive sensors) to establish the feasibility for buried water pipe inspection. Magnetostrictive
sensors are an alternate configuration of this technology as presented in Section 3.13. The types of pipe
being tested in this research grant are steel and CI (with cement mortar lining). Both an external tool and
an internal tool (to scan the entire pipe length from the inside) are being tested. The use of an internal
tool that travels through the pipe would potentially help to overcome the attenuation of the signal at pipe
joints (FBS, 2011).
Table 3-24. Guided Wave Ultrasonic Testing
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Guided wave ultrasonic testing (continuous ultrasonic measurement)
Implement rapid screening of pipes for material loss due to corrosion/erosion.
Commercially available from many vendors and consulting companies.
Rose et al., 2008; Marlow et al., 2007; Moore, 2007
• Inspection from a single probe position is possible. The initial screening only
needs exposure of a small section of buried pipe to attach the probe.
• It is also possible to inspect hidden structures under coating, insulations and
concrete.
• The range of inspection is limited to 30 m (98 ft) for aboveground pipe with
continuous joints. It has been applied to buried pipes, but with an even shorter
range of inspection due to the rapid attenuation of the signals.
• Pipes with bell and spigot joints will limit the range of inspection to one pipe
segment for external inspection.
• It is not applicable to heavily coated pipes due to wave attenuation.
• It cannot distinguish between internal and external corrosion.
Sensitivity can be as good as 1% loss of cross-section in ideal conditions (but is
typically set at 5%).
The guided wave system was originally designed for use on above-ground exposed or
insulated pipes.
Electromagnetic acoustic transducer (EMAT) is a couplant-free transducer based on a
different physical principle (Marlow et al., 2007). It generates ultrasound waves in
electric conductive materials by Lorentz force known as the electro-magnetostrictive
effect (Marlow et al., 2007). It can provide relatively consistent results in comparison
to piezoelectric transducers.
The labor cost to perform guided wave ultrasonic inspections is expected to be the
major cost. Equipment costs are estimated to range from $1,000 to $10,000 (Jolley et
al., 2010; Marlow et al., 2007).
45
-------
3.7.2 Discrete Ultrasonic Measurement. Discrete ultrasonic measurement transmits a high-
frequency short wave through a couplant to the material being tested (Figure 3-18). The wave can be
generated by several methods, including piezoelectric ceramics, electromagnetic acoustic transducer,
magnetostrictive sensor, laser and piezoelectric polymers. The waves propagate to the back wall of the
specimen and are reflected back towards the transducer. Transit time is recorded and used in combination
with the velocity of the wave propagating in the material to compute the travel distance of the wave.
Materials with known thicknesses are used to calibrate the sensor.
A typical UT system consists of a pulser/receiver, transducer, and display unit. Driven by the pulser, the
transducer generates a high frequency ultrasonic energy that propagates through the materials in the form
of waves. When an object is encountered in its path, part of the energy is reflected back from the object's
surface. The reflected wave is transformed into an electrical signal, from which information on the
reflector's location, size, orientation, and other features is inferred.
Types of ultrasonic system displays include:
• A - scan: discontinuity depth and amplitude of signal;
• B - scan: discontinuity depth and distribution in cross sectional view;
• C - scan: discontinuity distribution in plane view.
Operation may need extensive skill and training. The UT inspection for pipe can be done both externally
and internally. Usually, UT inspection needs couplant or water to transmit the wave between the
transducer and the pipe wall. However, the EMAT does not need couplant. See Table 3-25 for more
information on discrete ultrasonic measurement.
transducer
continuous ultrasonic measurement
pipe
\
transducer
region of pipe inspected
discrete ultrasonic measurement
pipe
\ region of pipe inspected
Figure 3-18. Continuous and Discrete Ultrasonic Measurement
46
-------
Table 3-25. Discrete Ultrasonic Measurement
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Discrete ultrasonic measurement
Used externally or internally for screening of pipes for corrosion/erosion at discrete
locations.
Commercially available from many companies, for example GE Inspection
Technologies, Olympus NDT, etc.
Grouse, 2009; Moore, 2007; Marlow et al., 2007
• Sensitive to both surface and subsurface discontinuities
• Provides instantaneous results.
• Probes of different sizes and frequencies are available for different applications.
• Supply shutdown is not necessary when using external tools. Water can be the
coupling medium.
• Surface of object to be inspected must be accessible.
• Coupling medium is required.
• Difficult to inspect materials that are rough, irregular in shape, or not homogeneous,
such as concrete
• CI and other coarse grained materials are difficult to inspect due to low sound
transmission and high signal noise.
• Calibration is required.
• Requires pipe cleaning prior to inspection.
Can achieve a reasonable degree of accuracy for the remaining wall thickness
measurement.
Used for thickness measurement, corrosion monitoring, delamination checks, and flaw
detection in welds, forgings, castings, and ferrous pipes.
UT is relatively inexpensive for conventional applications. Manual discrete ultrasonic
testing is estimated to cost $1,200 per day with an inspection rate of 200 ft per day.
Ultrasonic pigs for pipeline inspections are expensive (Jolley et al., 2010).
3.7.3 Phased Array Technology. Phased array ultrasonic has been used for medical imaging for
over 20 years and has recently been adapted for industrial applications. An array transducer contains a
number of individual sensor elements in a single package. With phased array technology, it is possible to
detect wall thickness, corrosion, or cracks with one multi-element transducer. The phased array
transducer is built up of composite sensor elements that are controlled individually by the ultrasound
electronics (Bosch et al., 2004). The sound beam and its direction are determined by the time sequencing
of the individual sensor elements. The sound beams are formed by shifting the phase of the signal
emitted from each radiating sensor element. Constructive interference of the waves amplifies the signal
in the desired direction, while destructive interference of the waves improves the sharpness of the sound
beam.
Phased arrays use an array of sensor elements, all individually wired, pulsed, and time shifted (Moore,
2007). The elements can be organized as a linear array, a two-dimensional matrix array, a circular array
or in more complex forms. Any set of sensor elements can be used as a virtual sensor. For a wall
thickness measurement, all of the elements are triggered simultaneously and a sound beam perpendicular
to wall surface is generated (as illustrated in Figure 3-19). For crack detection, the neighboring elements
are triggered with a certain time shift from element to element and an angular sound beam is generated (as
illustrated in 3-19). See Table 3-26 for more information on the phased array technology. Phased array
ultrasonic technology has been used in the nuclear industry to inspect coarse grained stainless steel
materials, where conventional UT methods were found to have significant limitations.
47
-------
sound
beam
wavefront
••••••••••••••••••••••I
ttttttttttttttttttttttt
wavefront
phased array of composite sensor
elements for wall thickness
measurement
phased array of composite sensor
elements for crack detection
Figure 3-19. Sound Beams Generated by Phased Array of Composite Sensor Elements
(after Bosch et al., 2004)
Table 3-26. Phased Array Technology
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Phased array technology
Phased array technique offers significant technical advantages over
conventional single-probe UT: the phased array beams can be steered, scanned,
swept, and focused electronically.
Commercially available from Olympus NDT and GE Inspection Technologies;
application to water mains not reported.
Bosch et al., 2004; Moore, 2007;
http://www.olvmpus-ims.com/en/ndt-mtorials/plmsed-arrav/
• Scanning is faster than single probe.
• Scanning can be done from different angles to obtain a better understanding
of the geometry of defects.
• A wide variety of test angles can be used to distinguish complex defect
types.
• Cost may be higher than single-channel systems.
• Setups for three-dimensional applications are complex.
Phased array technique can optimize discontinuity detection while minimizing
test time.
Used in a wide variety of industries including aerospace, nuclear power plants,
steel mills, pipe mills, petrochemical plants, and pipeline construction.
Inspection of water mains with the phased array technique has not been
reported.
Phased array technique is undergoing further development.
3.7.4 Combined UT Inspection. A combined UT technique, which can simultaneously quantify
metal loss and detect cracks, was reported by Beller and Barbian (2006). This technique uses a newly
designed and optimized sensor carrier to perform both inspections in a single run. A sufficient number of
UT sensors are placed to cover the pipe circumferentially. These sensors work in a pulse-echo mode with
a high repetition frequency. Straight incidence of the ultrasonic pulses is used to measure the wall
thickness and 45° incidence is used for the detection of cracks (Beller and Barbian, 2006). This is a
pigging technology developed for oil and gas pipelines.
48
-------
3.8
Seismic Pulse Echo
The seismic pulse echo technique uses impact from a metal sphere to generate ultrasonic compression,
shear, and surface waves in the pipe wall. The combination of wave velocity and thickness resonance
values can be used to determine the condition of the pipe.
A sensor array is used to pick up the response signals. From the recorded data, multiple measurements
can be carried out, including the compression and shear wave velocity and resonant frequencies. The
wave transmission velocity can be measured directly from the energy point of impact to the sensors at
distances larger than the thickness of concrete being tested (Fisk and Marshall, 2006). Reflectors are
measured individually or by examining the resonant frequency contents. Resonant frequency values are
used to identify thin areas where the mortar coating is delaminated or missing. Loss of resonant
frequency is an indication of micro-cracking and weakening of the core concrete as a result of prestressed
wire breaking, poor manufacturing, or overloading (Fisk and Marshall, 2006). Low velocity is an
indication of weakening of the core concrete.
The system typically consists of an energy source, an array of sensors, signal conditioner, analog to
digital converter and a computer. The data acquired are a time-distance recording of the amplitude of a
stress wave produced by a projectile impact. Data are recorded by an array of four sensors spaced
approximately 1 foot apart. The data are interpreted by determining the time required for the compression
and shear wave to travel to each sensor and then calculating average wave velocities given the known
distances between the sensors. Fourier analysis is conducted on the time series to determine frequency
content and resonances. See Table 3-27 for more information on the seismic pulse echo technique.
Table 3-27. Seismic Pulse Echo
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Seismic pulse echo
Seismic pulse echo is also known as the sonic/ultrasonic technique for
evaluation of PCCP. This technique is used to assess the condition of PCCP
by determining the strength of the core concrete.
Commercially available (fromNDT Corporation).
Fisk and Marshall, 2006; Communications with Paul Fisk; Wardany,
2008; http://www.ndtcorDoration.com/
• Can be conducted either from inside (dewatered) pipes or from outside (in-
service) pipes. Repeating measurements at a later date provides two data
sets in time to determine the deterioration rate (if any) of each pipe section.
• Inspection is not affected by overlying coatings or wearing surfaces.
• The inspection of long distances is time consuming. The test is very local
in nature.
• Skilled operators are required for field inspection.
The accuracy for the detection of PCCP broken wires is not known (Wardany,
2008).
The results of sonic/ultrasonic testing provide baseline current condition and
deterioration rate data to prioritize repair and developing a management
program for large diameter PCCP lines. This technique inspects the strength of
the core concrete and determines whether the pipe is acting as a composite
structure. This technology can also be used to test the exterior of wastewater
pipe for hydrogen sulfide corrosion. Applications have also been found in
bridge deck evaluations.
Not available
49
-------
3.9 Pipeline Current Mapper
Pipeline current mapper (PCM) is a technology intended to locate leakage of electrical current in
cathodically protected pipes. These leakages typically correspond to phenomena such as coating faults,
shorts due to connection to other metallic structures, etc. CP is a technique to arrest/limit corrosion. It
can be used for any metallic structure, including buried pipes, where it is often used in conjunction with
external coatings for the protection of sensitive pipelines. In North America, CP is used heavily for oil
and gas pipelines and to a much lesser extent in water mains (Radiodetection, 2002).
The PCM system consists of a portable transmitter and a handheld receiver. The transmitter applies a
special near direct current (DC) signal to the pipe under investigation. Direct contact is required between
the transmitter and the pipe. The receiver is carried along the pipe above ground, reading the transmitted
signal remotely, and identifying the position and depth of the pipe, as well the magnitude and direction of
the protective CP current. These readings are interpreted to identify deteriorated coating, specific coating
faults and possible cross-connections with other metallic structures (Radiodetection, 2002). In a network
where all (or most) pipes are electrically continuous, this technique provides fast location of potential
problems while minimizing excavation.
3.10 Radiographic Testing
Radiographic testing uses a source of radiation, either gamma or x-rays, which passes through the
material and onto a photographic film (see Table 3-28). The density changes on the film indicate possible
imperfections. Nowadays, digital cameras have been used to replace film, but are limited by the size of
the complementary metal oxide semiconductor (CMOS) photodiode array in the image sensor. Gamma
rays emitted from isotopes are used for ferrous and cementitious materials. X-rays created by cathode-ray
tubes are used for plastic materials. Details of the material structure can be seen on the radiograph and
darker areas correspond to thinner or less dense material. It has technical limitations in that pipes of 38.1
cm (15 in.) inside diameter and greater must be emptied. Typical defects that can be detected include:
• Pits in ferrous materials. Corrosion byproducts are less dense and appear darker on the
radiograph.
• Voids in cementitious materials.
• Inclusions or manufacturing voids.
There are basically three setups for radiographic testing as illustrated in Figure 3-20. Gamma or x-rays
are used to penetrate a weld, valve, or pipe wall to create a latent image on a radiographic film. The
radiation can pass through a single object onto the film (single wall-single image) or it can pass through
two sections of the pipe wall onto the film (double wall-single image). The third configuration is referred
to as double loading where two films of different speeds are used (one fast film and one slow film) to
document the condition of two adjacent objects between the film and source. For the same exposure
period, the slow film records the features of the first object closest to the source, while the fast film
records the features of the second object (Randall-Smith, 1992).
50
-------
single wall single image
radiation
source
\ radiation
detector/film
double wall single image
radiation
source
, radiation
detector/film
fast film
radiation
source
slow film
double loading
Figure 3-20. Radiographic Testing
Table 3-28. Radiographic Testing
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Radiographic testing
Show variants and thickness changes in material and structures, also applicable
to inspection of valves.
Commercially available
Marlow et al., 2007; Galbraith et al., 2009; http://www.yxlon.com
Can be applied to most materials.
• The setups in Figure 3-20 may not be practical for field inspection of buried
pipes.
• Examines only a small area at a time.
• Access is required to both sides of the inspected object.
• Radiation safety issues exist and inspection requires specialist operators.
• Can provide accurate measurements, but experience is required to interpret
the inspection results.
Radiography has been introduced to the water sector to examine pipe
conditions and valves in situ. In the U.S., it is used widely in petrochemical
processing plants, but also on water mains outside the U.S.
A recent development is the x-ray backscatter technique, which does not
require film on the other side of the inspected object. This technology is
currently being applied to thin structures, such as aircraft lap joints. No
application on pipes is reported.
51
-------
3.11 Thermographic Testing
Thermographic testing is a non-contact method of detecting thermal anomalies (see Table 3-29). Infrared
radiation has a longer wavelength than visible light (>700 nm). Any object above 0°K radiates infrared
energy and the amount of radiated energy is a function of the object's temperature and emissivity, which
is a measure of the surface efficiency in transferring infrared energy. Areas with different thermal masses
have different rates for heat absorption and radiation.
The infrared radiation is converted into a visible image and objects under test can be viewed on the basis
of their heat emission. In thermographic testing, an external heat source is typically used to heat the
inspected object. Subsequently, the object's cooling characteristics are monitored by an infrared camera
and these characteristics are then interpreted to provide object properties (Grouse, 2009). Varied active
thermographic testing methods have been developed for different applications. These methods include
pulse thermography, stepped heating thermography, lock-in thermography, and vibro-thermography.
Table 3-29. Thermographic Testing
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Thermographic testing
Detect material loss of relatively thin structures.
Commercially available.
Grouse, 2009 ; Marlow et al., 2007; htto://www. flir.com
• Allow rapid scanning of objects;
• No direct contact and intrusion is required;
• The thermographic system is easy to operate.
• In order to identify the anomalies, a temperature difference is necessary.
The infrared sensor is sensitive and reliable.
Thermographic testing has been used for leak detection of oil pipelines and
many other applications. Its use for water mains has been limited to less
accessible water pipes (Thomson and Wang, 2009).
Not available
3.12 Using Soil Properties to Infer Pipe Condition
3.12.1 Linear Polarization Resistance of Soil. An electrochemical reaction with a weak electrical
current is produced when a metal is immersed in an electrolyte solution, which leads to the corrosion of
metal. The rate of corrosion is directly proportional to this current and inversely proportional to the
electrical resistance (polarization resistance) of the metal/electrolyte pair. The direct measurement of
corrosion current in the soil (electrolyte) is very difficult. Instead, it can be inferred by imposing a weak
electrical potential (10 to 20 mV) between two electrodes. This potential produces small currents that are
linearly proportional to actual corrosion current. The ratio between the imposed electrical potential and
the resulting current provides the property known as the polarization resistance which, at low potential
values, is nearly linear to the corrosion current.
Several methods are available to measure linear polarization resistance (LPR) in order to estimate
corrosion rate. In the lab, a soil sample is brought to its wilting point and a small potential is
subsequently applied across two identical electrodes in a cell containing the prepared soil sample. The
current at each electrode is measured over a range of potentials. The resulting relationship between
current and applied potential is called the polarization curve. LPR is independent of the corrosion
potential of a specific metal in the soil (Marlow et al., 2007). This technique allows the assessment of
52
-------
corrosion rate in real time. It should be noted that lab-assessment of LPRto predict corrosion rate of
water mains is common mainly in Australia, where it is believed that soil moisture at wilting point is a
good approximation for long-term soil moisture (hence, no correction is made in the analysis for 'true'
soil moisture [Ferguson, 2010]). It appears that more research might be needed to ascertain this
assumption, especially if lab-LPR is to be adopted outside of Australia. Portable LPR instruments are
commercially available from several companies, including Metal Samples Corrosion Monitoring
Systems, Caproco, Rohrback Cosasco Systems, Inc. (Figure 3-21) and others. Multiple readings can be
taken at different locations to check the consistency of the soil corrosivity. For the device shown in
Figure 3-21, two carbon steel electrodes are contained within a single probe head, which has a pointed tip
that is used to facilitate pushing the probe head into the soil.
TWO-ELECTRODE
Figure 3-21. Corrater® Aquamate™ Portable Instrument with Soil Corrosion Rate Probe
(Reprinted with permission from Rohrback Cosasco Systems, Inc.)
3.12.2 Soil Characterization. Soil characterization is used to explore the soil parameters relevant
to the deterioration of buried pipes. Samples from the locations near the pipe are collected for lab
characterization or in-situ testing. The following is a list of the main soil parameters of interest (Marlow
et al., 2007):
(a) Soil resistivity: Low resistivity is likely to have high corrosion rates.
(b) pH value: Low pH value (pH <4) is generally associated with corrosion of ferrous materials
and deterioration of cementitious materials. However, high alkalinity soils (pH > 8) can also
lead to corrosion of metallic pipes as well as the prestressing wire and steel cylinder in PCCP.
(c) Redox potential: The redox potential of soil is a measure of soil aeration and provides an
indication of the suitability of conditions for sulfate reducing bacteria. High availability of
oxygen promotes MIC in the presence of sulfates and sulfides.
(d) Sulfates: Sulfates react with cementitious materials, forming gypsum and ettringite. Sulfate
attachment only occurs where the sulfate salts are in solution.
(e) Chloride content: chloride ions in moist soil act as electrolyte and reduce soil resistivity,
which encourages corrosion in CI and DI pipes, where the metal is in contact with the soil. In
the case of PCCP, if there are cracks in the outer mortar layer, ingress of chlorides in the
presence of oxygen will promote corrosion in the prestressing wire and steel cylinder.
(f) Moisture content: Soil moisture acts as the electrolyte in electrochemical corrosion of ferrous
pipes. It also defines the degree of soil saturation.
(g) Shrink/swell capacity: High shrink/swell capacities are known to have an increased failure
rate due to the stresses imparted by the soil during the shrink/swell cycle.
53
-------
(h) Buffering capacity: A soil's buffering capacity is the degree to which it is able to resist
changes in pH in particular acidification.
(i) LPR: High LPR indicates low corrosion rates. The corrosion rate can be roughly estimated
from LPR measurements.
(j) Contaminants: Soil contaminants can have negative effects on polymeric materials. High
levels of acidic constituents can also cause environmental stress cracking of polymers,
dramatically reducing lifetime.
(k) Soil compaction: The susceptibility of the trench filling and the surrounding sediments for
compaction.
Soil corrosivity is not a directly measurable parameter and there is no explicit relationship between the
soil corrosivity and pipe deterioration rate. Consequently, Table 3-30 provides a number of empirical
approaches that have been proposed in the literature to consider some or all of the above listed parameters
in the determination of soil corrosivity and potential pipe deterioration.
Table 3-30. Comparison of Soil Corrosivity Rating Approaches Based on Soil Properties
Methods
10-point
scoring method
12-factor
evaluation
25-point
scoring method
Fuzzy -based
method
Fuzzy inference
Factors
resistivity, pH value, redox potential,
sulfide, and soil type
Soil type, soil resistivity, water content, pH
value, buffering capacity, sulfide, chloride
and sulfate concentration, groundwater level,
horizontal and vertical soil homogeneities,
and electrochemical potential
pH value, sulfate content, redox potential,
soil type, resistivity, sulfides, moisture, pipe
size, pipe maximum design surge pressure
factor, pipe minimum design life factor, pipe
location and leak repair difficulty factor,
potential interference sources, pipe zone
back fill materials, and additional factors to
consider.
Same as in 10-point scoring method
Same as in 10-point scoring method
Classification
Results
Binary
Four
categories
Four
categories
Three
categories
Numerical
value between
[0,1]
Corrosivity
Potential
Corrosive or non-
corrosive
Highly corrosive,
corrosive, slightly
corrosive,
virtually not
corrosive
Mildly corrosive,
moderately
corrosive,
appreciably
corrosive,
severely corrosive
Non-corrosive,
moderately
corrosive,
corrosive
Non-corrosive = 0
and most
corrosive = 1
References
AWWA,
1999
Metalogic,
2003
Spickelmire,
2002
Sadiq et al.,
2004
Najjaran et
al., 2006
54
-------
3.12.3 Pipe to Soil Potential Survey. Pipe-to-soil potential reflects the interaction between ferrous
pipes and the surrounding soil. The measurement can be done with a voltmeter and a reference electrode
(Marlow et al., 2007). There are two types of pipe potential survey. The first is the direct current voltage
gradient (DCVG) survey that can be used to determine the location of gaps in a pipe's protective coating.
A DC is introduced to the pipe and the difference between two reference electrodes is measured in the
pipe-to-soil voltage. The two electrodes are gradually moved along the whole length of the pipe. If a gap
exists in the coating, there will be a significant increase in voltage gradient compared with the gradient
found when the coating is intact. The second type of potential survey consists of using a single reference
electrode (Cu/CuSO4) without an imposed current to determine the pipe-to-soil potential along the pipe.
The pipe-to-soil potential can be used to estimate corrosion rate with calibration data. Calibration is
carried out by directly assessing the external conditions of mains in different soils. The soil is sampled
every 50 or 100 m and sections of the main located in different soil types are then exposed and their
external condition directly assessed to relate this information to a pipe-to-soil potential value. It should
be noted that the potential survey reflects a propensity for corrosion rather than actual corrosion.
3.13 Emerging Sensor Technologies and Sensor Networks
Advances in electronics, sensor technology, information science, electrical and computer engineering give
rise to emerging technologies and some of these advances could be applied to the inspection, monitoring,
and condition assessment of buried water mains. The applicability to buried pipes of the emerging
sensors described here has not yet been fully verified in the field.
In addition, the growing use of sensor networks (some of which are already commercially available) and
multi-sensor approaches may also improve the prospects for real-time data acquisition and monitoring for
water mains. Several examples of sensor networks are discussed below.
Tables 3-31 to 3-40 provide detailed technology information for emerging sensors, sensor networks, and
multi-sensor approaches.
3.13.1 Corrosion Rate Sensor. The corrosion rate sensor uses the electrical resistance (ER)
technique, which is one of the most widely used methods to measure metal loss due to corrosion in buried
DI pipes (Bell and Moore, 2007). A trench must be dug to expose the surface of the pipe in order to
install the sensor. An exposed ferric element in the ground will experience metal loss due to corrosion and
consequently see a change (increase) in its electrical resistivity. The ER method compares this change to
a sealed reference element (Khan, 2007). The probe is typically placed in close proximity to the exposed
element of interest so that this element is subjected to exactly the same temperature as the reference
element (metal resistivity is affected by temperature).
It is not practical to use an entire pipe as the exposed element. Consequently, a coupon from the pipe of
interest (or a coupon of the same type of material) is used (Figure 3-22). This type of probe can also
measure the effectiveness of pipe cathodic protection by measuring the metal loss (in terms of electrical
resistivity) of a coupon that is cathodically protected. The exposed element doesn't need to be a metal
coupon. It can also be the soil in the vicinity of a structure (pipe) of interest to provide changes in soil
resistivity (relative to the reference element). It should be noted, however, that pipes rarely corrode in a
uniform manner due to material heterogeneity and soil variability. Therefore, a single sensor is not likely
to provide a good representation for the condition of long pipes. Table 3-31 provides more information
on the corrosion rate sensor.
55
-------
Figure 3-22. Picture of Corrosion Rate Senor with Embedded Metallic Coupon
Table 3-31. Corrosion Rate Sensor (Probe)
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Corrosion rate sensor (probe)
Measures cumulative corrosion and calculates in-situ corrosion rates for ductile
iron pipes (as referenced to an embedded ductile iron coupon).
Commercially available.
Bell and Moore, 2007; Khan, 2007; httrj://www.tinker-rasor.com
• Compatible with all standard ER instruments.
• Low profile of element makes it easy to install under polyethylene
encasement.
• End user's calibration is not required.
• Provides point measurement. Multiple sensors required to provide a good
representation for the condition of long pipes.
• A period of monitoring is necessary to obtain reliable measurements of
corrosion rates.
• Not applicable in cast iron pipes or non-metallic pipes. The company does
offer a similar sensor with an embedded carbon steel coupon.
The accuracy in corrosion rate measurement has not been verified.
Used for monitoring the corrosion rates of DI pipes. This product is relatively
new and has not been widely applied.
Not available
56
-------
3.13.2 Magnetostrictive Sensor. Magneto strict! ve sensor (MsS) is based on the principles of
magnetostrictive (Joule) and inverse-magnetostrictive (Villari) effects (Kwun, 1991; Kwun, 2000). The
magnetostrictive effect refers to a small change in the physical dimension of ferromagnetic materials
caused by an externally applied magnetic field. The inverse-magnetostrictive effect refers to the change
in the magnetic induction of ferromagnetic material caused by mechanical stress or strain. The generation
and detection of guided waves are based on the Joule and Villari effect, respectively.
MsS typically consists of two magnetic fields. A bias magnet (Figure 3-23) establishes a magnetic field
in the pipe and the dimension changes due to magnetostriction. A short-duration pulse is sent to the
transmitting coil that produces a magnetic field that opposes the bias magnetization. Then a time-varying
magnetic field causes the pipe to change dimension, hence causing an elastic wave pulse (Bartels et al.,
1999). The generated waves propagate along the pipe in both directions. When the wave passes by the
receiving coil, the magnetic induction changes and an electric voltage signal is induced. The duration of
the pulse defines the frequency of the elastic wave, which is often in the ultrasonic range for pipe
inspection, on the order of a few hundred kilohertz. This signal is then amplified, filtered, and digitized.
There are many magnet and coil configurations for generating elasticity waves in pipe. The
magnetostrictive method for wave generation is an alternative to the piezoelectric method presented in
Section 3.7.1 on guided wave ultrasonic testing.
Transmitting
Receiving
coil
\j
~
1
V^Ull
/
~
Crack
t /
* J
Wave A
I
^
1
Bias magnet Bias magnet
Figure 3-23. Pipe Inspection with MsS
Figure 3-24 shows a typical MsS system setup. The MsS device is ring-shaped to encircle the inspected
pipe (Kwun et al., 2003). At the transmitting coil, the operation wave mode is controlled by the relative
alignment between the DC bias magnetic field and the time-varying magnetic field produced by the MsS.
Applicable guided wave modes include: longitudinal, torsional, and flexural wave modes for cylindrical
objects. The coil and magnet configuration defines the wave types and are selected to maximize the
inspection distance and sensitivity to defects. For longitudinal wave modes in cylindrical objects and
Lamb wave modes in plates, a parallel alignment is used. For torsional wave modes in cylindrical objects
and shear horizontal wave modes in plates, a perpendicular alignment is used. For buried pipelines, the
signal loss into the external coating and soil and internal fluid is prominent. The inspection distance is a
57
-------
challenge and limitation of the guided wave inspection technology. For CI and DI water mains, the bell
and spigot joint will reflect the propagating wave limiting inspection to one pipe length for external
inspection of pipes with this joint type.
The MsS can be implemented in two modes: survey mode and monitoring mode. With the survey mode,
the MsS strips are temporarily attached to a de-insulated pipe. Both inside diameter/outside diameter
defects and circumferential cracks (>2% cross-sectional area) can be detected. Once complete, the strips
are removed and the pipe section is reinsulated. In the monitoring mode, the MsS strips are permanently
bonded to the pipe outside diameter using epoxy-based compounds and protected by a sealed clamshell
cover. The survey mode is ideal for aboveground pipes, while monitoring mode is primarily for
underground pipes. See Table 3-32 for more information about the MsS technique.
MsS Instrument for Horizontally Polarized
Shear (torsional) guided wave mode
Figure 3-24. The MsS System for Pipe Corrosion Monitoring
(Reprinted with permission from SwRI)
The U.S. EPA is sponsoring a grant to research the use of ultrasonic guided waves (using in-situ
magnetostrictive sensors) to establish the feasibility for buried water pipe inspection. Magnetostrictive
sensors are an alternate configuration of the guided wave technology that was presented in Section 3.7.1.
The types of pipe being tested in this research grant are steel and CI (with cement mortar lining). Both an
external tool and an internal tool (to scan the entire pipe length from the inside) are being tested. The use
of an internal tool that travels through the pipe would potentially help to overcome the attenuation of the
signal at pipe joints (FBS, 2011).
58
-------
Table 3-32. Magnetostrictive Sensor
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Magnetostrictive sensor (MsS)
Long-range inspection and structural health monitoring using guided waves.
Commercially available from SwRI (http://www.swri.ors)
Kwun, 1991; Kwun, 2000; Bartels et al., 1999
• Does not need couplant.
• Can be operated with a gap between the sensor and material under test.
• Detects both ID/OD wall loss and circumferential cracks.
• Low cost sensor for long-term structural health monitoring.
• Pipes with bell and spigot joints will limit the range of inspection to one pipe
segment for external inspection.
• Guided wave technology cannot differentiate between ID and OD damage.
• Inspection capability past elbows is limited due to the distortion of the shape of the
wave.
• Coated and buried piping reduces the test range to less than 9. 14 m (30 ft).
• Can inspect inaccessible areas from a remote accessible pipeline location to detect
erosion, corrosion, and other defects for a full or empty pipeline. Sizes the area of the
defect in the radial circumferential plane. Detection of 2 to 5% change of cross-
sectional area using the survey mode or 1% using the monitoring mode.
• Accuracy of defect location is within 2.5 in.
• Test range varies depending on piping condition; up to 500 ft in straight,
aboveground piping. There is a limitation for buried pipes because of the rapid
attenuation of wave propagation.
MsS relies on magneto strictive effects, so it is applicable to ferrous materials such as
carbon steel, alloy steel, and ferritic stainless steel. For nonferrous materials, the MsS
can be operated over a thin strip of ferromagnetic material (such as nickel) bonded to
the material. The applications include boiler piping, piping crossing over roads or small
streams, and insulated pipes. MsS is currently being investigated for use with buried
water mains under a U.S. EPA grant.
Not available
3.13.3 Conformable and Flexible Eddy Current Array. The principle behind the conformable
eddy current sensor array is the traditional eddy current theory, i.e., the change of coil impedance (phase
and magnitude) reflects the properties of the conducting object under test. For the pipe pitting
measurement, the displacement between the eddy current probe and bottom of the pit is detected by using
an eddy current probe (Crouch and Goyen, 2003).
The conformable array was designed to transform discrete measurements into a two-dimensional scan
(image). A picture of such an array (implemented with the flexible printed circuit board technology) is
shown in Figure 3-25. Similar technology was reported by (Chen and Ding, 2007). The conformable
array can be easily adapted to the surface curvature of pipes, however, this entails excavation and
cleaning to expose the bare pipe. Table 3-33 provides more information about conformable and flexible
eddy current array technique.
59
-------
Conformable Array for Mapping Corrosion Profiles
(a)
(c)
Figure 3-25. (a) Inspection of Pipeline with Flexible Eddy Current Array, (b) the Sensor Array,
and (c) Samples of Inspection Results
(Reprinted with permission from Clock Spring)
60
-------
Table 3-33. Conformable and Flexible Eddy Current Array
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Conformable and flexible eddy current array
Generate contour map of corrosion, which identifies the depth and location of
the deepest pit.
A prototype was developed at SwRI and passed a field test (Anna, 2004).
Anna, 2004; Crouch and Goyen, 2003
• No special apparatus is needed for the scanning.
• Does not require sandblasting, but the cleaner the pipe surface, the better the
depth measurement.
• The measurement system can be easily operated.
• There is limitation in pit depth measurement. Improvement is needed for
measuring deeper pits.
• There are limits to sensor density.
• Pipe has to be excavated.
• Measurement is local.
• The typical accuracy of depth measurement is: ±0.020 in. (0.50 mm). The
scan area is about 6 by 6 in. (152 by 152 mm) (single scan).
• The eddy current response is non-linear and sensitive to liftoff. Any
material that creates liftoff appears to have a great depth. Coating left on
the pipe causes liftoff and results in higher depth readings.
• The deepest pit and smallest (in diameter) pit that can be measured are not
known. Extra care is suggested to verify depth above 0.250 in. The
measurement variation becomes large for depth above 0.300 in.
• Pit diameter less than 0.250 in. can result in low depth measurements (this
depends on proximity to a coil).
The conformable eddy current array targets the gas transmission pipelines and
is designed for hands-on use by field technicians at spots of interest.
Not available
3.13.4 Flexible Ultrasonic Transducer. The flexible ultrasonic transducer (PUT) consists of a
metal foil, a piezoelectric ceramic film, and atop electrode. Top electrodes can be fabricated of silver or
platinum paste, while the metal foil (e.g., stainless steel) serves as both the substrate and bottom
electrode. The porosity of piezoelectric film and the thinness of metal foil provide this sensor with
sufficient flexibility for application to curved and irregular surfaces (Figure 3-26). FUT can be easily
formed into an array by placing many electrodes into a desired configuration.
This type of transducer operates in the pulse-echo (i.e., the subject of interest needs to be excited by an
external source of energy), transmission and pitch-catch modes. They can be used as phased array for fast
electronic scanning and imaging. Table 3-34 provides more information about the flexible ultrasonic
transducer.
3.13.5 Damage Sensor. The damage sensor is a combination of distributed electrochemical
impedance spectroscopy (EIS) and time domain reflectometry (TDR). The EIS measurement provides
information about the effectiveness of a coating over a relatively small area. An alternating voltage is
applied between corroding material and a reference electrode. The impedance measurement reflects the
condition of the pipe coating. Modeling of the EIS system can provide the location and status of flaws in
the pipe. More information about the damage sensor is available in Table 3-35.
61
-------
78mm-
• ••••I
•••••••••••••••a
element:3mmx3mm Element spacing: 2mfei
Figure 3-26. The Flexible Ultrasound Transducer Array
(Reprinted with permission from NRC IMI)
Table 3-34. Flexible Ultrasonic Transducer
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Flexible ultrasonic transducer (PUT)
Monitor pipe structural health and perform on-line diagnostics.
Available from the NRC's Industrial Materials Institute.
License issued for aerospace structural health monitoring applications and
being negotiated for power plant and oil and gas industries.
Kobayashi et al., 2006; Kobayashi et al, 2007; Mrad et al, 2006; Kobayashi et
al., 2009
• Can operate through a wide range of temperatures (-80°C to 500°C).
• Can be self-aligned to object surface with curved or complex geometry.
• Can be used to excite and receive guided acoustic waves along pipes to
perform long distance defect evaluation.
• Can be used to perform electronic scanning and imaging.
• Has low cost.
• Can be miniaturized.
• Can be operated using a wireless network.
• One PUT covers limited area. A large area may need an PUT array with a
number of FUTs.
62
-------
Table 3-32. Flexible Ultrasonic Transducer (FUT) (Continued)
Performance
Breadth of use
Other information
Ultrasonic performance is better than commercial broadband ultrasound
transducers, but without their disadvantages (e.g., large size, small operation
temperature ranges [up to 70°C]). FUT has been tested on curved surfaces
with curvatures of 25 mm in diameter with no observable damage.
Applied to structural health monitoring of gas turbine engines, aircraft frames,
power plant and oil refinery pipes.
A variety of commercial applications in industrial plants for in-situ
characterization of materials and real-time process monitoring at high
temperatures.
Research and development contracts established to perform NDT and
structural health monitoring of pipes for oil-sand transportation, oil refinery
and electrical power and chemical plants.
Table 3-35. Damage Sensor
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Damage sensor
Defect location and characterization
In research and development
Juliano et al., 2005
• With the potential to integrate/fuse the information from
for a more accurate result.
• Only detects changes in the local environment. Need to
distributed sensing system along a full section of pipe.
multiple sensors
develop a
Not available
Not available
Not available
The TDR measurement sends an electromagnetic or sharp DC pulse and analyzes the reflected signal
from discontinuities. Discontinuities include pipe failures and accidental impact damage. The TDR
technique can identify fault locations with high precision.
The combination of the EIS and TDR sensors takes advantage of each sensor to obtain maximum
information regarding defect location and characterization. This is still in development and has not been
fully implemented yet.
3.13.6 Microwave Back-Scattering Sensor. The microwave back-scattering (MBS) sensor is
based on the principle of transmitting continuous electromagnetic microwaves at a frequency of 2.45 GHz
and receiving the back-scattered signals (Munser et al., 1999). It detects nonhomogeneities in terms of
dielectricity, such as holes caused by erosion and humidity changes due to leaking water. The inspection
with the MBS sensor is from the inside of the pipe.
The MBS sensor consists of four transmission patch antennae and four staggered receiving patch
antennae. The whole inner surface of the pipe is covered for inspection. The absolute amplitude and
relative phase for each signal channel are processed to characterize the detected anomalies. See Table 3-
36 for more information on the MBS sensor.
63
-------
Table 3-36. Microwave Back-Scattering Sensor
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Microwave back-scattering (MBS) sensor
Detect hidden objects or material anomalies by penetrating the pipe surface.
The MBS sensor had been tested in a sewerage test bed (Munser et al. 1999).
No commercial product is available.
Eiswirth et al. 2000; Munser et al. 1999
May provide complementary information to other types of sensors (such as
ultrasonic sensor, camera, etc.).
Not available
Not available
Used for inspecting sewerage systems experimentally. Usage in water mains is
not reported.
Not available
3.13.7 Fiber Optic Sensor for Corrosion Monitoring. Changes in pipe wall thickness will lead to
the change of the outer surface strain (for a given stress level). The fiber optic sensor monitors and
records changes in strain and the wall thickness can be derived from this measurement.
Three fiber optic sensors are needed to calculate pipe wall thickness. One is to measure the strain due to
wall thinning that depends on the internal pressure and the other two to compensate for the operational
variation in temperature and pressure. The sensitivity of the system depends on wall thickness, pressure,
and pipe materials, but can be as high as 0.002 in. (50.8 microns) as reported in Morison (2007). Up to
eight fiber optic sensors can work simultaneously with one monitoring unit. The monitor units can also
be networked together, making remote access possible. The client can access real-time data over a Web-
based application. Portable instrumentation that is battery powered is also available.
The fiber optic sensors can also be designed to measure pipe bending due to ground movement (Cauchi et
al., 2007). Linear and coiled fiber optic sensors were designed and used for monitoring gas transmission
pipes (Figures 3-27 and 3-28). See Table 3-37 for more information about the fiber optic sensor for
corrosion monitoring.
Figure 3-27. FOX-TEK Coil Sensor
(Reprinted with permission from FOX-TEX)
64
-------
Figure 3-28. Monitoring Corrosion and Bending of Pipelines with Fiber Optic Sensors
(Reprinted with permission from FOX-TEX)
Table 3-37. Fiber Optic Sensor Corrosion and Bending Monitoring
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Fiber optic sensor for corrosion monitoring
Direct measurement of pipe wall thickness.
Commercially available from Fiber Optic Systems Technology Inc. (FOX-
TEK).
Morison, 2007; Cauchi et al., 2007
• Suitable for monitoring many types of problems, including corrosion and
pipe bending
• Faster access to corrosion data from difficult-to-access locations
• Can maintain a database on direct measurement of the pipe wall thickness.
• Continuous Web-based monitoring
• Internal corrosion is an extremely slow process. It may take more than 30
days to separate the signal of wall thickness loss from background signals.
• Need to expose the pipe to attach fiber optic sensors from outside.
The strain sensitivity of the sensor will be dependent upon operating pressure,
initial wall thickness, pipe diameter, and pipe material.
• The fiber optic sensors have found their applications in the oil, gas,
pertrochemical, and chemical processing industries, where corrosion is
severe and failures are of high consequence.
• This technology also has potential for applications in water industry.
The use of fiber optics for distributed temperature monitoring has also found
its application for leak detection in oil and gas pipelines (Nikles et al., 2004).
The detection is based on the fact that when there is a leak, the surrounding soil
temperature changes accordingly. This technique uses a similar concept to
optical time-domain reflectometry for the localization.
3.13.8 Fiber Optic Acoustic Monitoring Network. Acoustic fiber optic (AFO) cable is installed
inside a PCCP main and is connected to a laser at a data acquisition system. Light is projected through
the AFO cable. When there is only ambient noise in the pipe, the reflected light is relatively constant and
the resulting signal does not have a significant dynamic component. When a wire break occurs in the
pipe, the sudden strain energy release generates pressure waves acting on the AFO cable. A dynamic
pattern of light is obtained and can be used to evaluate the acoustic properties of the event. Frequency,
65
-------
acoustic magnitude, attenuation characteristics, and other acoustic variables are analyzed to determine
when and where a wire break has occurred (Higgins and Paulson, 2006).
The AFO cable may consist of four or more long continuous glass fibers. A picture showing the
installation of the AFO sensors is provided in Figures 3-29 and 3-30. These fibers, together with a
strength fiber that provides strength to resist tension, are encased in a protective jacket. The monitoring
results for each pipe section are available on a secure Web site, where the pipes are mapped on Google
Earth for easy identification. See Table 3-38 for more information on the AFO monitoring sensor.
Figure 3-29. (Left) Installation of Fiber Optic Sensor in a Dewatered Pipeline. (Right) Fiber Optic
Sensor Installed on a Stainless Steel Hoop with a Strain Relief Device
(Reprinted with permission from Pure Technologies)
Figure 3-30. (Left) Installation Parachute Used to Install an AFO Cable in an in-Service Pipeline
and (Right) Parachute Caught and Extracted at Two Miles Downstream from Insertion Point
(Reprinted with permission from Pure Technologies)
66
-------
Table 3-38. Fiber Optic Acoustic Monitoring Network
Name
Purpose/Scope
Status
Source of information
Advantages
Limitations
Performance
Breadth of use
Other information
Fiber optic acoustic monitoring sensor
Identifies wire breaks as they occur in PCCP
Commercially available (Pure Technologies Ltd. and the Pressure Pipe
Inspection Company)
Higgins and Paulson, 2006;, Higgins et al., 2007
• The entire AFO cable acts as a sensor and is therefore acoustically sensitive.
This means that the acoustic sensor is not further than a pipe diameter from
a wire break, which results in negligible acoustic attenuation of the acoustic
activity associated with a wire break.
• Acoustic data are acquired continuously and wire breaks are identified and
reported in near real time. This allows a water/wastewater utility to know
on an ongoing basis where wire breaks are occurring in a pipeline and the
risks associated with each pipe section and they can intervene to mitigate
risk as needed.
• No electronics are placed in the water flow.
• Long lengths of a pipeline can be monitored with one data acquisition
system. Up to 12 miles (20 kilometers) can be monitored from one data
acquisition system.
• Electronic noise in monitoring is negligible.
• The monitoring sensor is always near a wire break.
• The AFO cable can be installed either while a pipeline is out of service or in
service. Installing it while a pipe is in service may result in an increased
need for tapping outlets into the pipeline.
• Transduction mechanisms may result in the light intensity change and thus
introduce errors. Pure Technologies reports that this has negligible effects
on data quality.
• Other potential errors may be introduced by variable losses due to
connectors and splices, and misalignment of light sources and detectors
(Gholamzadeh and Nabovati, 2008). Pure Technologies reports that this has
negligible effects on data quality.
• The monitoring system does not provide information on wire breaks that
occurred prior to the installation of the AFO cable.
• The AFO cable is installed while the pipe is out of service; it is periodically
attached to the invert of the pipe and is routed around inline valves.
• The AFO cable-based monitoring was compared with hydrophone-based
monitoring by Pure Technologies (Higgins and Paulson, 2006). Better
performance was observed from the experiments. An accuracy of +/- 5 ft to
locate wire breaks was reported by Pure Technologies.
• In addition, a second comparison was also performed during an experiment
on the Great Man-made River in Libya and also documented better
performance.
The AFO system is presently being used to monitor more than 150 miles of
PCCP mains in the U.S. In addition, the AFO cable is being installed on an
ongoing basis for the Great Man-made River Authority where 250 miles of
AFO cable have been installed as of August 2009 (NACE International, 20 1 1).
Not available
67
-------
3.13.9 Wireless Sensor Network for Pipe Condition Monitoring. The MEMS sensors that
measure the acceleration change at the pipe surface were connected through a wireless network. The
basic principle is that a sharp transient change in hydraulic (water) head in the pipe flow induces a
correspondingly sharp change in the acceleration of pipe vibration on the pipe surface. The inverse
analysis may locate the damage in a pipe segment between two neighboring sensors. Accelerometers
H34C and SD 1221 made by MEMS technology were integrated with two sensor units respectively
(Shinozuka et al., 2007). Three-axial vibration data were collected. The change in the water pressure due
to pipe damage can be identified by the change in acceleration on the pipe surface. However, the
algorithm to locate the damage through the captured transient signal has not been developed. Only this
concept is being developed.
A wireless sensor network (WSN) system called PipeNET was described in (Stoianov et al., 2007). The
architecture of this system is illustrated in Figure 3-31. Piezoresistive silicon sensors were used to
measure the pressure. Acoustic/vibration data were collected by accelerometers installed along the pipe.
The third function block included a different set of applications such as monitoring water quality in
transmission and distribution water systems, and monitoring the water level in sewer collection systems.
The WSN can increase the spatial and temporal resolution of operational data from pipeline
infrastructures and implement near real-time monitoring and control. Table 3-39 provides more
information on wireless MEMS sensor network.
Figure 3-31. The System Architecture of PipeNET
Table 3-39. Wireless MEMS Sensor Network
Wireless MEMS Sensor Network
Capabilities:
• Identify the damage locations from three-axial vibration data over a vast lifeline network.
• Real-time monitoring
• Data transmission via wireless network
Benefits:
• Can be easily integrated into a supervisory control and data acquisition (SCADA) system.
• MEMS sensor is power-efficient.
Limitations:
• Can only point to a problem occurring while monitoring is active.
• Installation of the MEMS sensor unit to existing pipes remains a challenge. Possible solution is to
install at hydrant locations.
68
-------
3.13.10 Multi-Sensor Approaches. Any individual non-destructive inspection technique may not be
able to fully characterize the condition of pipes. Multi-sensor technologies incorporate multiple sensors
for a comprehensive pipe inspection and assessment. This technology encompasses two aspects. One is
the sensor selection; the other is the sensor fusion. The choice of sensors depends on the particular
requirements of an application. The sensor fusion algorithms take care of processing the signals acquired
by heterogeneous sensors. A multi-sensor experimental platform called sewer assessment with multi-
sensors was developed with the support of the German Research Foundation. The sensors reported in
Eiswirth et al. (2001) include:
(a) Optical triangulation sensor: optical 3D measurement of a sewer pipe;
(b) Microwave sensor: inspect the soil state behind sewerage pipes;
(c) Geoelectrical sensor: detect leak points;
(d) Hydrochemical sensor: detect groundwater infiltration;
(e) Radioactive sensor (neutron and gamma ray probes): investigate soil density and soil
moisture content;
(f) Acoustic systems: detect leaks, cracks and determine the state of connections and pipe
bedding.
The basic steps include sensor data acquisition, signal processing, feature extraction, data fusion and
diagnosis. The sensor fusion algorithm is implemented with the fuzzy-logic method. However, a report
on the overall performance is not available.
The data acquired by multiple sensors need to be synchronized or registered so that the correspondences
between the data can be established. This is the first step in processing multi-sensor data. See Table 3-40
for more information on the multi-sensor technology.
Table 3-40. Multi-Sensor Technology
Multi-Sensor Technology
Capabilities:
• Can provide more reliable data and continuous profile of pipe walls.
Benefits:
• Higher benefit/cost ratio is anticipated compared to single sensor technologies.
• Can be implemented as a robotic module.
Limitations:
• The performance of overall system is not known.
3.13.11 Smart Pipe. The so-called "smart pipe" concept has been around for the last 15 years or so.
It is a loosely defined concept, whereby the pipe is equipped with a range of sensors (embedded or
otherwise) that provides a complete monitoring network of the pipe condition and performance. A smart
pipe project for deep-sea pipelines was initiated in Europe in 2006 and is slated for completion in 2012.
The objective is to develop a complete monitoring system for pipelines, integrating sensor technology,
data acquisition, data interpretation, and decision support for on-line, real-time management of pipeline
assets (SINTEF, 2008). The entire length of each pipeline is to be monitored by sensors throughout the
life of the pipe. The expected benefits include, but are not limited to, improved basis for decision making,
improved residual life prediction, and decreased need for inspection. In the U.S., Smart Pipe® is also a
registered technology commercially available from the Smart Pipe Company Inc. in Katy, TX. It is a
69
-------
reinforced thermoplastic pipe designed for use as a tight fit liner in high pressure applications such as
gas/liquid lines from 150 to 1,440 psi. Within the high strength fiber windings, a monitoring system is
secured, which consists of longitudinally oriented fiber optic sensors that send signals for strain and
temperature anomalies that can be used to detect damage and potential leaks.
3.14 Additional Leak Detection and Monitoring Methodologies
The main objectives of leak detection are the reduction (or elimination) of water losses through leaks, as
well as reducing the possibility of small leaks developing into pipe failures. However, while addressing
these two main objectives, information about leakage rates provides an important indication about the
condition of the pipe. This subsection provides additional information about leak detection and
monitoring methodologies to supplement information already provided in earlier subsections on acoustic
and non-acoustic inspection technologies for leaks (some overlap exists between this and previous
subsections). A summary of technologies and computational methods for leak detection and monitoring
is given in Figure 3-32, followed by brief descriptions of hydraulic transient-based methods,
measurement-based leak monitoring methods, model-based leak monitoring methods, and information
fusion with neural networks.
Figure 3-32. Summary of the Technologies for Leak Detection and Monitoring
(Misiunas, 2005)
3.14.1 Hydraulic Transient-Based Methods. Besides the NDT methods described in previous
sections of this report, hydraulic transient-based techniques are also available to detect and locate existing
leaks. The information regarding the presence of a leak is extracted from a measured transient trace.
Various computational approaches have been proposed to analyze the hydraulic information for both
detection and monitoring purposes. Techniques such as inverse transient analysis have been verified in
laboratory settings, but their feasibility and limitations under actual field conditions require further
70
-------
verification. NRC Canada is conducting a pilot study into the use of inverse transient analysis for the
detection of leaks in the City of Regina's municipal water distribution network (Karney et al., 2009).
Leak reflection method. The method is based on the principle of time domain reflectometry. A transient
wave is reflected at the leak and can be identified in a measured pressure trace. The location of the leak
can be calculated.
Inverse transient analysis. Least square regression is applied to the modeled and measured transient
pressure traces. The minimization of the deviation between the measured and calculated pressures gives
the leak location and size.
Impulse response analysis. The impulse responses of the same pipeline with and without a leak are
compared. The presence of a leak will introduce the change of the impulse response.
Transient damping method. A leak detection and location method was developed based on the rate of
leak-induced damping. This rate depends on leak characteristics, pressure, location of the transient
generation point, and the shape of the transient.
Frequency domain response analysis. The analysis of transient response in the frequency domain
compares the dominant frequencies of non-leaking and leaking pipelines. The leak location can be
obtained.
3.14.2 Measurement-Based Leak Monitoring Methods. A brief description is provided below of
measurement-based methods used to detect leaks.
Acoustic monitoring. Through analyzing the acoustic signals with a leak and without a leak, the situation
is identified. A correlator is often used to locate the leaks. The cross-correlation methodology relies
upon detecting noise emitted by a leak from two sensor locations and analysis of the acoustic signature
from each location.
The noise emitted by a leak is detected by the two sensors and produces an acoustic signature in each.
These signatures are identical in shape, but offset from each other. The size of the offset is determined by
the difference in time at which the noise is detected by each sensor. Since sound velocity is constant, the
time gap imax is proportional to the respective distances of the sensors from the leak. The location of the
leak with respect to the sensors can then be computed by (Hunaidi et al., 2004):
max
L7-L
where LI and L2 are the positions of the leak relative to sensor 1 and 2, respectively, and c is the
propagation velocity of the leak sound in the pipe. The distance between the sensors (D) is equal to LI +
L2. Therefore, Lj can be expressed as:
The propagation velocity can be determined onsite or calculated based on pipe material and diameter.
However, the acquired signals are prone to distortion. LeakfmderRT™ uses an enhanced cross-
correlation function, which is implemented in the frequency domain using the cross-spectral density
function. Thus, a better resolution and definition of peaks can be achieved (Hunaidi et al., 2004).
71
-------
Another robust method was implemented by the Central Research Laboratories at Thames Water
(MathWorks, 2007). Complex discrete Fourier transform (DFT) is used to transform the input time
domain signal to the frequency domain. The echo of the signal is removed/cancelled by analyzing the
auto-correlation of each channel. Phase coherence analysis is used to determine which parts of the
frequency spectrum contain useful information. The output of the analysis constructed a weighted
frequency filter, which achieved an optimal performance in the detection of leak signals (MathWorks,
2007).
Volume balance method. The basic principle is that the amount of fluid that goes into the pipe should be
equal to the amount that comes out of the pipe. The flow measurements will calculate:
where:
^ : volume balance;
in : inlet volume;
out : outlet volume;
^ : volume of fluid contained in the pipe (line pack).
Any leak will give a positive value of VB.
Pressure-point analysis. This method is implemented by monitoring the leak-induced pressure drop.
Statistical techniques are applied to identify the leak signature in the measured pressure trace.
Negative pressure wave method. This method is based on monitoring the pressure for the leak-induced
pressure wave. The location of the leak can be determined from the wave arrival times and wave speed.
Statistical pipeline leak detection. The statistical method uses flow rate, pressure, and temperature
measurements to carry out a sequential probability ratio test.
Statistical data analysis-based methods. An autoregressive model, which uses two consecutive time
sequences of pressure gradients at both ends of the pipeline, was established to detect the leak. The
parameters and residual variance of the fitted models are dependent on the condition of the pipeline and
reflect the presence of a leak.
District meter areas. This method conducts a water audit in district meter areas. Flow and/or pressure
sensors are placed on the boundary of the district meter area. The collected data are analyzed for leakage
trends.
3.14.3 Model-Based Leak Monitoring Methods. A brief description is provided below of model-
based methods used to detect leaks.
Real-time transient model-based methods. Two techniques are considered: one is the deviation analysis
and the other is the model compensated volume balance method.
In the pressure-flow deviation method, the flow rate and pressure at one boundary can be calculated from
the flow rate and pressure values measured at another boundary using the transient simulation model.
72
-------
The calculated values should match the measured values if no leak is present. The discrepancy between
the measured and calculated values indicates a leak.
The model compensated volume balance approach implements the real-time comparison of the
measurement generated flow balances and model generated line packing rates, which are computed from
measured pressures and temperatures at the end points of a pipeline segment with the model. In the case
of leakage, the measured flow balance and the model generated line packing will diverge.
Steady-state inverse analysis. A leak is detected and located by solving an inverse problem using
measurements of pressure and/or flow rate.
Inverse transient analysis. This method can be applied to an unsteady flow situation. The responses of
transient events are measured and interpreted by calculating the model parameters using the inverse
method.
State estimation approaches. The flow in pipelines can be represented by a distributed parameter system,
which is implemented with a state estimator or a filter. An extended Kalman filter can be used to
estimated leaks (Misiunas, 2005).
3.14.4 Information Fusion. A framework for leak detection from multiple acoustic emission (AE)
sensors was proposed by Jiao et al. (2007). The idea is illustrated in Figure 3-33. The AE signal is
processed with a wavelet transform to extract signal features. Next, a neural network is trained to provide
a mass function. The Dempster-Shafer evidence theory is then employed to combine the mass function
values from multiple sensors. The fused result will identify the pipe leak.
Figure 3-33. Sensor Data Fusion for Leak Detection (Jiao et al., 2007)
The fusion of hydraulic data for burst detection and location in a treated water distribution system was
reported by Mounce et al. (2003). An artificial neural network is used to model the time series data
acquired by a flow sensor. A mixture density network was employed to predict the conditional
probability distribution of the target data. The actual observed value is analyzed in the context of the
predicted probability distribution. A normal (non-leak) or abnormal (leak) state is understood. The
classification results from various zones are fused by a rule-based expert system implemented by Mounce
et al. using PROLOG (a general purpose programming language).
73
-------
3.15 Supplemental Information on Inspection Platforms, Intelligent Pigs, and Robotic
Survey Systems
This section includes additional information on advances in inspection and monitoring platforms
including computer-aided augmented reality, intelligent pigs, and robotic survey systems.
3.15.1 Computer-Aided Approach: Augmented Reality. Augmented reality (AR) is a technology
that blends in real-time, real-world footage and computer-generated graphics. The AR system described
in Lawson and Pretlove (1998) consists of a stereo robotic head device, virtual reality graphics engine,
scan converters, head mounted display, and stereo monitor. The AR system itself does not introduce any
new method for pipe inspection, but it provides a human-computer interface, which facilitates advanced
data manipulation and enhanced visualization of faults and deficiencies in the pipe.
3.15.2 Intelligent Pigs and Robotic Survey Systems. Pigs and robots serve as platforms for the
introduction of one or more sensory payload into the pipe for assessing its conditions (Schemph, 2004).
The fundamental requirements of such systems include (Jamoussi, 2005): ability to traverse the entire
pipe in a reasonable time without getting stuck; ability to inspect the pipe with acceptable accuracy and
resolution, and ability to transmit the inspection data to the outside for reporting or save the data locally.
Most of the robotic systems for water and sewer mains are tethered for power and communications. A list
of available platforms is given in Table 3-41. Although not all platforms are intended for water mains, it
is still a good source of reference for the development of the robotic platforms.
An inspection system is preferred that can be operated on-line without an interruption of service. A
robotic system for internal inspection of water pipelines was presented (Moraleda et al., 1999). From
their research, the authors learned (Moraleda et al., 1999):
• No cost-effective system will be able to negotiate through all possible scenarios that may
exist inside a water pipeline network;
• A tethered solution could be adopted for recoverability, despite the greater autonomy that a
non-tethered vehicle could provide.
Table 3-41. Robot Systems for Pipe Inspection
System
PIRAT
(Kirham et
al., 2000)
KARO
(Kuntze and
Haffner,
1998)
Description
Pipe Inspection Real-Time Assessment Technique (PIRAT) is a non-
autonomous tethered robot for the quantitative and automatic
assessment of sewer conditions. A human operator can operate the
robot from a surveillance unit via a cable, with a length of 250
meters (maximum). An expert system running on a workstation was
responsible for data interpretation and damage classification.
Kanalroboter (KARO) is an experimental semi-autonomous platform
for sewer inspection. It is tethered via a cable to a surveillance unit.
With on-board inclinometers, KARO is able to correct for tilt in its
pose and wheel slippage.
Sensors
Video camera
and laser scanner
Standard video
camera,
ultrasound
transducer,
microwave
sensor, and 3D
optical sensor
Date
2000
1998-
2000
74
-------
Table 3-41. Robot Systems for Pipe Inspection (Continued)
System
Description
Sensors
Date
KURT
(Kirchner
and
Hertzberg,
1997)
Kanal-Untersuchungs-Roboter-testplatform (KURT) is a six-
wheeled autonomous un-tethered robot. A map of pipe net is needed
for the navigation.
Ultrasound
transducer,
inclinometers,
CCD camera
1997
KANTARO
(Nassiraei et
al., 2006)
KANTARO is a fully autonomous, untethered robot for pipes of
diameter 20 to 30 cm. It was designed to move in straight pipes and
pass different kinds of pipe bends without any special controller or
sensor.
Fish eye camera
and 2D laser
scanner
2006
MAKRO
(Rome et
al., 1999)
Multi-segmented autonomous sewer robot (MAKRO) is a fully
autonomous, un-tethered, self-steering articulated robot platform for
sewer inspection. It has six segments connected via flexible joints.
This enables MAKRO to crawl along narrow pipes. The on-board
batteries can support a two-hour operation of the robot.
Infrared sensors,
ultrasonic
sensors, camera,
laser crosshair
projector
2002
RoboScan
(Vradis and
Leary,
2004)
RoboScan is a modular platform for unpiggable gas distribution
pipelines. Each module has its own micro-controller, which is
connected through a network. A fiber optic cable is used to connect
RoboScan and the base station. A magnetic flux leakage module
was used for pipeline inspection.
Magnetic flux
leakage
2004
Explorer-II
(Schempf,
2006)
Explorer-II (X-II) is a modular robot platform for inspection of live
gas mains.
Digital camera
2008
Ultrasonic
inspection
robot
Ultrasonic inspection robots were developed by Inspector Systems
for use in refinery pipes, buried pipes, and pipes with long vertical
inclines. The robots are made of three modules connected with
flexible folding bellows. One of the three modules is the ultrasonic
element, which consists of an ultrasonic sensor unit for measuring
pipe wall thickness, a camera, and a positioning unit. The robot can
move both horizontally and vertically along pipes about several
hundred meters long. It can pass bends and turns of 1.5 diameter. A
fiber optic cable is used to connect the control unit for transmission
of control commands as well as inspection data. A special fluid is
used as the couplant for the inspection.
http://www.inspector-svstems.com/ultrasonic robots.html
Ultrasonic sensor
and camera
Information
retrieved in
2009
Robots for
video and
laser
inspection
These robots from Inspector Systems can be applied to nuclear
power industry, refineries, chemical plants, petrochemical plants,
offshore industry, gas pipelines, beverage industry, and other types
of pipes. The maximum distance that the robot can travel is about
500 m. A color camera with a ring of light emitting diode lights is
mounted on the head with pan and tilt functions for video inspection.
An adjustable point laser is used for internal measurement and
classification of defects and corrosion. The robot has three drive
elements and an inspection head as standard. The drive elements are
connected via flexible folding bellows and each of them contains
two direct current motors. A fiber optic cable is used to transmit
inspection data and control commands.
http://www.inspector-svstems.com/video robots.html
Color camera and
point laser
75
-------
Table 3-41. Robot Systems for Pipe Inspection (Continued)
System
Description
Sensors
Date
PipeDiver
At the 2009 International No-Dig Show (Toronto, Ontario, Canada),
the PPIC demonstrated a prototype of its modularized free-swim
platform, PipeDiver™, for the inspection of in-service PCCP with
diameters ranging from 0.6 (24 in.) to 1.5 m (60 in.). Driven by
water flow, the platform has three modules, which are used for
vehicle tracking, pipe inspection, and power supplying, respectively.
The RFEC/TC technique is integrated in the inspection module.
Two challenges for PCCP inspection were considered in the design
of the first generation of PipeDiver™, platform launch and retrieval,
passing pipe bend and butterfly valve. A version of PipeDiver™
was field tested at Louisville, KY in 2009 summer on 24-in. CI
pipe.
RFEC
2009
Super-Pig
(Clay,
2009)
Super-pig is a platform with an ultrasonic module to measure the
pipe wall thickness loss, longitudinal and circumferential cracks,
damage to linings, and leaks. The targeted mains are in the range of
200 to 300 mm. The super-pig can operate for water mains in
service. Special launch and retrieval facilities are needed.
Ultrasonic
transducer array
2009
3.16
Current Use of NDE and Condition Assessment
Although the subject of condition assessment of water mains has received increasing attention in the last
10 to 20 years, it is not clear at what rate and what type of condition assessment practices water utilities
are actually adopting. Anecdotal evidence and limited surveys (e.g., Deb et al., 2002; Deb et al., 1990;
Dingus et al., 2002; Grigg, 2007; Marlow et al., 2007; and others) suggest that many medium and large
water utilities have adopted some form of condition assessment and pipe renewal decision process, either
from the literature, developed in-house, or a combination of both. Many of these water utilities use some
form of inspection including visual, destructive or nondestructive techniques (the latter predominantly on
large transmission mains). Information about smaller water utilities is scant. Thomson and Wang (2009)
reported the existence of a number of barriers to effective use of condition assessment technologies,
including lack of data for comparison, lack of consensus on what are the key required data, the limited
availability of proven inspection techniques to discern pipe structural condition, cost of inspection and
condition assessment, lack of confidence in the adequacy of existing techniques and level of expertise
required for the implementation of various techniques and models.
In this subsection, the results of two limited surveys are presented to provide additional information about
current practices of condition assessment of water mains. The first part provides the results of an
anonymous survey conducted by NRC, while the second part describes a survey of nine utilities carried
out by Virginia Tech.
3.16.1 NRC Survey. The NRC collected information from technology vendors and water utilities in
order to assess the usage and application of inspection technologies from both perspectives.
A questionnaire was issued to ten technology vendors and six responses were received. The information
received has been incorporated into Section 3. For the technology vendors, the following information
was requested:
• Main features of the technology (what are the distinct features of this technology);
• About the inspection results:
o What is the inspection result, or what types of data are acquired during the inspection?
76
-------
o How are the data interpreted?
o Will the inspection data be transferred to condition rating? How?
• About the cost: what is the major cost of this technology, including equipment costs, the need
for well-trained operators or the operational cost?
• Technology usages (practitioners): how is this technology being used by municipalities and
water utilities?
• Other information.
Most vendors were reluctant to provide sales data (considered as confidential). Only PPIC provided some
relevant data pertaining to mileages of RFEC inspection for PCCP pipes. These data are presented in
Figure 3-34.
Inspection of PCCP (RFEC)
1998
Figure 3-34. The Mileages of RFEC-TC Inspection for PCCP
A separate questionnaire was sent to ten water utilities in the U.S. and Canada with five responses
received. The questionnaire included the following items:
• Did you use NDT to assess the condition of your water pipes?
• What NDT techniques/systems are currently being used?
• How many miles of pipes are inspected (roughly) and how often (frequency)?
• Did you purchase the NDT equipment for inspection or purchase the service from another
company?
• How do you use the inspection data? Was any decision made based on the inspection?
The responses from five utilities are summarized in Table 3-42. All respondents used the services from
contractors or third parties to carry out the inspection. The inspection results, to some extent, were used
in the decision making for the maintenance or rehabilitation of water mains, but the information regarding
the strategy for repairing and rehabilitating was not available. The frequency of inspection was not
explicitly defined or determined by any of the respondents.
77
-------
Table 3-42. Use of Condition Assessment Technologies for Water Mains by Five Water Utilities
Utility
A
B
C
D
E
Pipe
Type
PCCP
CI
PCCP,
steel
PCCP,
DI
DI,
thin-
wall CI,
PCCP
Use of
NDT
Yes
Yes (in
the
past)
Yes
Yes
Yes
NDT
Techniques
Currently
Used
RFEC, leak
detection,
impact echo,
acoustic, etc.
Hydroscope
SmartBall® ,
RFEC
RFEC,
acoustic
monitoring
and
correlators
for leak
detection
Hydroscope,
RFEC, fiber-
optic
monitoring,
hydrophone,
etc.
Miles of Pipes
Inspected
50
~O
miles/year;
Re-inspect
high-risk areas
50 miles
Few miles
6.2 miles of
PCCP in
history;
hundreds of
miles of leak
detection in DI
pipes
3 miles of
PCCP;
70 miles of DI
and CI pipes
Frequency
of
Inspection
Re-inspect
8% over the
last 3 years
Information
not available
Wall
thickness
along the
same steel
pipe sections
in 14 years
2.5
mile/year for
PCCP
inspection
(planned
from 2009)
Information
not available
Use of
Inspection
Data
Rate the
residual
strength and
damage with
models
Used in a
prioritization
process for
pipe
replacement
Used in a
repair or
replacement
strategy
Used for risk
assessment
and renewal
decisions
Inspection
data
integrated
into
CAD/GIS for
making
decisions on
pipe renewal
Inspection
Service
Third party
Third party
Third party
Third party
for PCCP
inspection;
In house
staff and
equipments
for leak
detection
Contractors
3.16.2 Virginia Tech Survey. The Virginia Tech survey was designed to understand issues related
to condition curves or any deterioration models used across the U.S., Canada, and Australia. The
participating utilities included:
• EPCOR Water Services Inc., (Edmonton, Alberta, Canada)
• Las Vegas Valley Water District (Las Vegas, Nevada)
• Newport News Waterworks (Newport News, Virginia)
• Seattle Public Utilities (Seattle, Washington)
• Sydney Water (Sydney, Australia)
• Washington Suburban Sanitary Commission (Laurel, Maryland)
• City of Hamilton Public Works Department (Hamilton, Ontario, Canada)
• Louisville Water Company (Louisville, Kentucky)
• Philadelphia Water Department (Philadelphia, Pennsylvania)
78
-------
In this report, only the information related to condition assessment of pipes is included, namely, the use of
NDT techniques and deterioration/failure models as summarized in Table 3-43.
Table 3-43. Summary of Utility Inspection Methods and Models
Utility
EPCOR
Water
Services, Inc.
Las Vegas
Valley Water
District
Newport
News
Waterworks
Seattle Public
Utilities
Sydney Water
Washington
Suburban
Sanitary
Commission
City of
Hamilton
Public Works
Department
Louisville
Water
Company
Philadelphia
Water
Department
Inspection
Methods
Cathodic Protection Program, Pipe Sampling,
Leak Detection, Uni-Directional Flushing
Program, Water Main Internal Lining Program,
Valve and Hydrant Replacement Program,
Neighborhood Program, and Hydroscope
Non-Invasive Technology, Cathodic
Protection, Forensic, Sahara®, SmartBall® , and
Echologics Acoustic Wave Technology
Hazen Williams C-Factor Test and Corrosion
Monitoring Stations
Spot checks of exposed pipes for general
exterior condition assessment
Linear Polarization Resistance, Magnetic Flux
Leakage, Ultrasonics
Internal Visual/Sounding Inspection,
Electromagnetic Inspection, Sonic/Ultrasonic
Pulse Echo, Sahara®, SmartBall®, LeakFinder
RT™, Acoustic Fiber Optics, and
Electrochemical Potential Survey
Unavailable
Unavailable
Unavailable
Models
1 . Reactive Renewal Program
2. Proactive Renewal Program
3. Hydraulic Model
1 . Computer Aided Rehabilitation of Water
Networks (CARE-W)
2. Linearly Extended Yule Process (LEYP)
1. Nessie Curve (Long -Term Economic
Forecast)
2 . Pipe Prioritization Replacement Model
3. Hydraulic Model
1 . Wave Rider (Long -Term Economic
Forecast)
2. Water Main Replacement Model
1. KANEW (Long-Term Capital Investment
Forecast Tool)
2. FARMS-PRIORITY (Water Main
Prediction Model)
1. Nessie Curve (Long-Term Economic
Forecasting Model)
2. UMP Condition Rating System
1 . Hansen Asset Management System
1 . Point-Score System referred to as LWC's
Pipe Evaluation Model (PEM)
1. Point-Score System
3.16.2.1 EPCOR Water Services Inc. EPCOR Water Services Inc. is a corporatized public utility
located in Edmonton, Alberta, Canada. Its governance structure is the same as a private utility; however,
it is wholly-owned by the municipality of Edmonton.
• Inspection and condition assessment:
Pipe sampling - Samples of various pipe materials (AC and PVC) are tested to evaluate the
condition and remaining life.
79
-------
Leak detection - Leak detection is to pinpoint the location of active leaks and breaks of CI
and steel pipes. This inspection aids in limiting the amount of surface disruption due to
excavation.
Hydroscope - Originally, the hydroscope was used to determine the remaining wall thickness
in CI pipes. However, the program was stopped because only a limited benefit was achieved
and many water quality complaints were received while utilizing this technology.
• Models:
EPCOR had experimented with failure models and artificial neural network analysis to
predict the residual life of water pipes. However, these efforts did not succeed and EPCOR
decided to continue using its renewal program, which consists of a reactive and a proactive
program. The reactive renewal program identifies the deteriorating distribution water mains
with a geographic information system (GIS) application that calculates the break frequencies
for candidate stretches between valves. EPCOR utilizes a ranking system within its proactive
renewal program to prioritize water pipes. The proactive renewal program consists of area
criteria rankings as well as candidate criteria rankings. The area criteria rankings are useful
in pinpointing locations where infrastructure requires more work, while candidate criteria
rankings help to choose a specific section of a pipe over another. The proactive renewal
program analyzes area and candidate criteria ranking for water pipes. It helped EPCOR
understand the interconnected piping system.
EPCOR's main focus is minimizing the impacts and response times to breaks, improving tools for
renewal candidate selection, and reducing the construction impact during actual renewal. Even though
validation of its program with respect to predictive effectiveness has not been a main focus, EPCOR has
evaluated the replacement priority value (RPV) renewal qualification criteria. It was shown that, once a
pipe reached the renewal qualification criteria and it was not renewed, the break rate would increase on
that particular section of pipe.
It is estimated that EPCOR spends approximately $50,000 to $100,000 per year on the analysis of
identifying pipes at risk, pipe inspection and data collection, data management, modeling software, and
interpreting results. Reactive and proactive renewal programs serve as EPCOR's prediction model in
determining water pipe replacement.
3.16.2.2 Las Vegas Valley Water District. The Las Vegas Valley Water District (LVVWD) is a
public utility located in Las Vegas, Nevada.
• Inspection and condition assessment:
LVVWD inspects AC pipes larger than 4 in. and steel pipes greater than 12 in. The
inspection and condition assessment techniques include:
Non-invasive technology - Acoustic wave velocity measurements in in-service water pipes
(AC and steel) is performed to estimate the percentage loss of pipe strength.
Cathodic protection - The assessment of steel pipes is implemented through analyzing CP.
Forensic - Forensic condition assessment is applied to AC pipes. Pipe samples are analyzed
and crush tests are performed on AC pipes in the lab.
Sahara® - Sahara® is used to detect leaks and structural defects in large mains.
SmartBalf - SmartBall® is used to detect and locate leaks.
80
-------
LeakfinderRT™- LeakfmderRT™ is used to locate leaks in water pipe.
Wall thickness assessment - The wall thickness assessment is implemented by measuring the
acoustic wave velocity within a pipe using sensors attached to two longitudinal points along a
pipe. This is still an experimental technology provided by Echologics (see "Other
information" in Table 3-20 in Section 3.6.2).
• Models:
The LVVWD utilizes several types of models/systems to analyze water pipes.
CARE-W- Computer Aided Rehabilitation of Water Networks is computer software that
includes fundamental instruments for estimating the current and future condition of water
networks. Detailed information about CARE-W is available in Section 6. LVVWD uses the
Annual Rehabilitation Plan (ARP) and Long-Term Planning (LTP) to identify the pipes that
should be considered for rehabilitation and obtain the information on the future investment
needs for the water network.
Linearly Extended Yule Process (LEYP) - The LEYP statistical failure model predicts break
rates. This model uses the software for analyzing break data and making break predictions.
Two types of data are input into the LEYP model: pipe data and break data. The analysis is
only applied to AC pipes.
The LVVWD does not perform any kind of statistical evaluation involving the CARE-W ARP and LTP
models; however, they do feel confident that their models are practical.
The average cost for non-invasive acoustic wave assessment is $2 per foot and increases exponentially the
more invasive the technology becomes. The LVVWD did incur a one-time cost for implementing the
CARE-W program. Currently, there are no ongoing annual fees for using the software. The main
advantages of LVVWD's models are the capabilities of prioritizing pipe condition assessment and
planning for long-term capital replacements. Disadvantages of the models include the need to acquire
reliable data and the cost of in-house analysis.
3.16.2.3 Newport News Waterworks. Newport News Waterworks is the public utility of Newport
News, Virginia.
• Inspection and condition assessment:
Hazen Williams C-Factor Test - The Hazen Williams C-Factor Test indicates the water pipe
wall roughness. The higher the C-factor, the smoother the pipe is. Newport News performs
two types of tests: one test isolates a section of a pipe and records the water flow per pressure
gradient along the pipe; the other test places up to 11 electronic pressure recorders on
hydrants around a flow hydrant.
Corrosion Monitoring Stations - Newport News installed the corrosion monitoring stations in
1994.
• Models:
Newport News utilizes several types of models/programs to analyze water pipes.
Long-Term Economic Forecast Model - This model, known as the "Nessie Curve"
(developed by the South Australian Water Company), projects replacement costs and "wear-
out" cost together to support the total lifecycle cost analysis. A "Nessie Curve" is a graph of
81
-------
estimated annual expenditure needed for replacement of pipe infrastructure. It reflects an
echo of demographic waves (i.e., cohorts of pipes with their respective installation years and
assumed end of life). The rising shape of this graph has caused it to be named a "Nessie
Curve" after the Loch Ness Monster. The Nessie curves help utilities estimate the long-term
pipe replacement budgets.
Prioritization Model - The initial priority program consisted of a point-score system
evaluating 10 different pipe criteria. However, it was found that several of these pipe criteria
did not affect the priority ranking over time. The revised and updated program assigns points
based only on the number of breaks, life expectancy, and maintenance cost.
Newport News does not perform any kind of statistical evaluation for these models. They feel confident
that their models are practical.
3.16.2.4 Seattle Public Utilities. The Seattle Public Utilities (SPU) is a private utility located in
Seattle, Washington.
• Inspection and condition assessment:
SPU does not utilize routine inspections and/or condition assessment techniques for water
transmission and distribution pipes. SPU's primary focus is on leak and break data in which
the leak rate is assumed to be a surrogate for condition assessment. A condition assessment
program was conducted for several years in the 1990s; however, it was discontinued since
costs exceeded the value of the information obtained. This program primarily consisted of
the opportunistic collection of samples (e.g., from a new tap or repair event) for analysis.
• Models:
Wave Rider (long-term capital planning) Model - The Wave Rider model forecasts the repair
and replacement expenditures by year for each of nine classes of pipe material and size,
where a Weibull distribution is assumed for the economic life of pipes in each class. The
Wave Rider model is very similar to the 'Nessie curves' described earlier, except it addresses
repair costs as well as replacement costs (repair rates are assumed to grow as the pipe
approaches end of life). The nine classes of pipe considered are DI, CI (divided into four
subcategories by size and vintage), steel, concrete, galvanized, and other. The calibration of
the model is implemented by comparing actual repair rates since the year 1990 to those
predicted by the model.
Prediction (pipe replacement) Model - The prediction model is for the repair/replacement
decisions of individual pipes. This model is primarily based on water pipe leak history and
standards for all pipe materials, sizes, and locations.
Water Main Replacement Model - The water main replacement model uses the deterioration
model described above to compare the expected cost of failure and repair against the
estimated replacement cost. Data classes used in analyzing the water main replacement
consist of pipe, construction, service, traffic, lost water, damage, fire risk, water quality, and
benefits.
No statistical analysis has been completed to evaluate the validity of the models. One reason for this is
that it is difficult for the SPU to validate the predictive effectiveness of the failure curve since the
majority of its pipes have remained in the flat part of the curve.
82
-------
In 2008, the SPU spent an estimated $43,000 on asset data, decision models, and related support. The
SPU is not considering any alternative methods and it periodically updates the water main failure rates.
The SPU plans to implement weighted factors to replace major transmission pipelines prior to expected
failure.
3.16.2.5 Sydney Water. Sydney Water is the public utility of Sydney, Australia, responsible for the
management of water and wastewater systems.
• Inspection and condition assessment:
Sydney Water mainly focuses on the condition assessment of CI water mains. For NDE, it
utilizes LPR to evaluate corrosion potential, and MFL and ultrasound to evaluate the extent of
corrosion in pipes. Sydney Water has conducted condition assessment for over 10 years.
• Models:
Sydney Water utilizes several types of models/programs to analyze water pipes for various
pipe materials, including CI, DI, AC, steel, and plastic.
KANEW- Sydney Water used the KANEW program to generate and analyze long-term
capital investment needs for renewal of water pipes. The long-term capital forecast tool is
based on the asset deterioration curve, which is deterministic. Detailed information about
KANEW is available in Section 6.2.
FARMS - Sydney Water is implementing FARMS to predict the condition of water pipelines.
It supports pipeline renewal prioritization focusing on the analytical assessment such as
pipeline replacement and pressure reduction in terms of associated risks.
Detailed information about FARMS is also available in Section 6.2.
Both software programs are described in detail in Section 6.
Sydney Water constantly validates and calibrates the failure curves based on analysis and failure history.
In its experience, the accuracy of the forecast from the FARMS model is close to the actual asset
performance.
Advantages of utilizing KANEW consist of benefiting from the long-term capital investment forecast as
well as the prediction of asset deterioration. Limitations of KANEW include that the failure curve
represents a cohort of pipes and not a specific individual asset that is based on limited variables. The
analysis only utilizes historical data to develop rates based on estimated averages. Furthermore, Sydney
Water feels that there is no explicit relationship between the asset performances versus the deterioration
curve. In the end, KANEW is not suitable for critical water mains and does not take risk into account.
3.16.2.6 Washington Suburban Sanitary Commission. The Washington Suburban Sanitary
Commission (WSSC) is a public utility located in Maryland.
• Inspection and condition assessment:
WSSC uses several inspection and condition assessment techniques. WSSC is primarily
focused on the inspection of large diameter PCCP transmission mains (greater than 48-in.
diameter). They expect to increase their inspection budget to allow for the inspection of up to
83
-------
12 mi per year of PCCP transmission mains with a target of every 6 years for the interval
between repeat inspection events (WSSC, 2011).
Internal Visual/Sounding Inspection - to detect visible cracks, damage, and delaminations
within the PCCP.
Electromagnetic Inspection - RFEC/transformer coupling is used to detect and quantify the
number of wire breaks in PCCP. WSSC has primarily used the P-Wave® system (Section 3).
Sonic/Ultrasonic Pulse Echo - The sonic/ultrasonic velocity and resonant frequency (pulse
echo) measurements are used to identify deficiencies in pipe condition, such as broken pre-
stressing wires and damaged or deteriorated concrete.
Long Term Acoustic Monitoring - Acoustic fiber optic cable (see Section 3) has been
permanently installed in 16.9 mi of PCCP transmission mains to listen for additional wire
break activity, as an early warning sign and to establish the rate of deterioration after a
baseline condition assessment is conducted (WSSC, 2011).
Other techniques used by WSSC include Sahara®, SmartBall®, LeakfinderRT™, and
electrochemical potential survey (see Section 3).
Models:
The WSSC prioritizes the inspection, maintenance, repair, rehabilitation, and replacement of
water pipes utilizing various methods for pipe material type and diameter. The WSSC is
currently in the process of developing a Utility Master Plan (UMP).
Nessie Curve Model - See description provided earlier for the Newport News System.
Risk Model - developed in house and includes six parameters to aid the WSSC in prioritizing
rehabilitation, repair, replacement or inspection of pipes. The risk factors include:
(1) Land use factor
(2) Repair history
(3) Operational needs
(4) Known manufacturing defects
(5) Last inspected
(6) Diameter
Each of these risk factors has a defined set of ratings per specific description of the risk
factor.
Risk Model for Large Diameter PCCP (>48-in.) - WSSC has developed a specific program
for condition assessment of its PCCP transmission mains (48-in. diameter or greater). A risk
rating is generated based factors including the pipe size, pipe age, pipe design standards and
manufacturer, land use, operational criticality, repair history, and date of last inspection
(WSSC, 2011).
84
-------
3.16.2.7 City of Hamilton Public Works Department. The City of Hamilton Public Works
Department Water and Wastewater Division is a public utility located in Hamilton, Ontario, Canada.
Hamilton's water infrastructure program consists of an asset management program that includes a GIS.
Hamilton has developed a replacement profile for water mains utilizing the Hansen asset management
system, which is a key component for analysis required to estimate the timing of major interventions,
such as rehabilitation and replacement. The replacement profile is primarily based on the age of the asset.
3.16.2.8 Louisville Water Company. The Louisville Water Company (LWC) was established in 1860.
The LWC's Pipe Evaluation Model (PEM) is a comprehensive planning and decision support tool
designed to assess priorities for the replacement and rehabilitation of water pipes. The PEM is a detailed
scoring system that assigns points based on over 23 assessment factors that can take on different
weighting schemes to allow LWC to adjust its model based on annual priorities. The main assessment
factors included within this model are categorized as follows:
• Geographical (central business district, redevelopment areas, and roadway classifications)
• Hydraulic (main size, fire flow availability, number of parallel mains, high pressure
frequency, and low pressure frequency)
• Maintenance (main break frequency, joint leak frequency, material samples, corrosive soil
data, installation date, pipe type, joint type, and maintenance record)
• Quality of Service (taste and odor complaints, discolored water complaints, water quality
data, number of domestic/fire services, lead service frequency, dead-end water mains, and
paving age).
The renewal projects are scored according to all of these criteria and then the projects are ranked based
upon their degree of importance. LWC uses a criterion of 2 breaks per mile per year as the threshold for
replacement. Additional information about LWC's PEM approach can be found in Bhagwan (2009).
3.16.2.9 Philadelphia Water Department. The Philadelphia Water Department (PWD) has the
distinction of being one of the first water distribution systems in the U.S. with operations beginning in
1815. PWD collects data on the maintenance history, date and location of main breaks, installation year,
size of main, and other information that is compiled in a database. PWD has developed a point-score
system for water main replacement, which was first described in O'Day et al. (1986). Based on input
from PWD to the Virginia Tech Survey, the scoring system is currently comprised of a combination of the
age of the water main and its break frequency as shown in Table 3-44. The goal of the PWD is to further
assess mains with scores of seven or more points.
Table 3-44. PWD Point Score System
Year of Installation
pre 1854
1854-1877
1878-1900
1901-1938
1939-1966
1967 -present
Break Frequency (Block-by-Block Basis)
A. Two or more breaks in the most recent year OR
B. Three or more breaks within the past 5 years AND
C. Each break not accounted for in A or B above
Points Assigned
5
4
3
2
1
0
Points Assigned
2 per break
2 per break
1 per break
85
-------
4.0: FROM DISTRESS/INFERENTIAL INDICATORS TO CONDITION RATING
4.1 Background
As water mains age they deteriorate. As discussed in Section 1, this deterioration can be classified into
two categories: (i) structural deterioration, which diminishes the pipe's structural resiliency and ability to
withstand the various types of stresses; and (ii) deterioration of the pipe's inner surfaces, resulting in
diminished hydraulic capacity, degradation of water quality and even diminishing structural resiliency in
cases of severe internal corrosion.
The probability of a water main failure due to structural deterioration can be estimated using physical
(mechanistic) models (Rajani and Kleiner, 2001) and/or statistical (empirical) models (Kleiner and
Rajani, 2001). Statistical models develop empirical relationships between the pipe, its exposure to the
external and operational environments, and its observed failure frequency. Physical models attempt to
mimic realistic (albeit simplified) field conditions taking into account both the external environment and
internal pipe operational conditions. Empirical models typically over-simplify a complex reality in order
to (hopefully) achieve "80% of the answer with 20% of the effort." In contrast, physical models, because
they are based on universal physical/mechanistic principles, can theoretically be applied in any
circumstances provided all pertinent data are available. However, pertinent data usually comprise a
substantial amount of data to represent specific conditions and environments. These data are either
unavailable or very costly to obtain for even a modest portion of a distribution network. Therefore,
physical models are useful to gain good insight into deterioration and failure mechanisms, as well as to
explore small-scale critical cases, but are impractical for large-scale implementation.
The essence of asset management is the balance between system performance and cost. This balance
behaves differently in small distribution mains compared to large transmission mains, and this difference
leads to different forms of management for the two classes of assets. Figure 4-1 illustrates these
differences qualitatively. As a pipe ages and deteriorates (without renewal), its probability of failure (or
failure frequency) increases and the risk increases as well. Note that the risk is expressed as the present
value (PV) of expected cost (or consequences) of failure. At the same time, the discounted (or PV) cost
of the renewal declines as pipe renewal is deferred. The total expected life-cycle cost typically forms a
convex shape, where the minimum point depicts the optimal time of renewal (t*). The top part of
Figure 4-1 illustrates atypical case of small distribution mains, where the cost of failure is relatively low;
as a consequence, the optimal time of renewal corresponds to a relatively higher failure frequency. In
contrast, the bottom part of Figure 4-1 illustrates that for large transmission mains, where the cost of
failure is typically very high, the optimal strategy is to avoid failure altogether, i.e., failure prevention
rather than failure frequency management. It must be noted that there is no clear cutoff pipe diameter
below which a pipe is considered 'small' and above which it is considered 'large'. For a metropolitan
like New York City, for example, large transmission mains could be pipes of at least 30-in. (750 mm) in
diameter, while for a small town a 12-in. (300 mm) diameter pipe might be considered large. It appears
that in the context of asset management the relative importance of the pipe in the network, or even more
precisely, the relative magnitude of the consequences of its failure, are the prevailing factors in
considering the pipe as 'small' or 'large.'
In addition to the economic difference between small distribution and large transmission mains described
above, the reality is that failure frequency in large transmission mains is a rather rare event, whereas in
small distribution mains failure is much more frequent, which facilitates the ability (generally absent in
large mains) to use the statistical analysis of historical failure patterns to discern deterioration rate and
forecast future failure rates. This statistical exercise is a de facto condition assessment of small
distribution mains. In large transmission mains, this type of analysis is not practical because of the rarity
86
-------
Failure frequency (#/km)
Cost
(present
value)
min.
cost
Total expected
cost
Cost of
renewal
Cost
(present
value)
mm.
cost
Time of renewal
Failure frequency (#/km)
Total expected cost
Failure risk
Cost of renewal
Time of renewal
Figure 4-1. Optimal Renewal Frequency for Distribution Mains (top) versus Transmission Mains
(bottom). (Time scale not necessarily same in both graphs.)
of failures. Failure prevention in large transmission mains requires knowledge about the condition and
deterioration rate of the pipe before it fails. Distress and inferential indicators obtained from inspection of
these large pipes provide information about the condition of the pipe; however, in order to estimate
deterioration rate (i.e., changes in condition overtime) as well as prioritize assets for renewal (i.e.,
compare condition of different pipes), these distress/inferential indicators need to be translated into a
rating scale that is consistent over time as well as over different types of pipe.
87
-------
Note that Figure 4-1 illustrates idealized cases, where the minimum point on the convex curve is quite
clear. There are cases where this curve is not as well behaved. When the aging rate (i.e., the rate at
which failure frequency increases) is similar in magnitude to the discounting factor (the "interest" rate
used to compute PV), the convexity of this curve can become quite flat, and the point of minimum cost
becomes less crisp. When the cost of failure is relatively low compared to the cost of renewal and the
discounting factor relatively high, the curve can take the shape of the "hammock-chair" as described by
Herz (1999), with no definite minimum, indicating that renewal could perhaps be postponed indefinitely.
Currently, most NDT technologies intended to identify distress indicators are too costly to be justified for
small water distribution mains. Some of the technologies intended to identify inferential indicators (e.g.,
those related to soil properties) are used for both small and large pipes. However, the predominant
approach for assessing the condition of small diameter distribution mains is based on the observation of
historical failure frequency (in many publications also referred to as number of previous failures).
Historical failure frequency is, strictly speaking, neither exactly a distress indicator nor a pure inferential
indicator, but can be viewed as a little of both and interpreted as a surrogate measure. This issue is
discussed in more detail in the next section. Also, in this context the definition of pipe 'failure' is an
important issue. In most cases, for practical reasons pipe 'failure' is defined as a 'maintenance event' (for
which there is a work order recorded). However, if a utility endeavors to differentiate between the break
types, and if staff is sufficiently trained and experienced and willing to do so credibly, then any
statistical/empirical analysis could be refined to gain better insight into the true deterioration patterns of
the pipe inventory.
Two types of raw data can be obtained from the inspection technologies described in Section 3: visual-
based technologies provide direct information about actual, observable distress indicators (cracks,
delamination, etc.), while most other NDT technologies (UT, magnetic, etc.) provide signal patterns that
require interpretation into distress indicators. In the latter case, signal interpretation is almost always
proprietary knowledge and is not the focus of this section, which concentrates on techniques and methods
intended to interpret distress/inferential indicators into pipe condition ratings.
Techniques and methods to interpret distress indicators into condition ratings started predominantly in
sewer condition assessment (late 1970s, early 1980s) and are in the process of evolving into the realm of
large water transmission mains (since the mid 1990s). One obstacle to this evolution is the fact that large
diameter water transmission mains are inherently expensive components of the water supply system, and
due to their high cost, the system often does not have enough redundancy to function while they are off-
line for inspection. Some of the developers and/or providers of NDT services developed their own
methods to interpret distress indicators into condition ratings, but these are generally proprietary and often
appropriate for a specific NDT technology (e.g., RFEC/TC by PPIC in Section 3) and therefore are not
addressed here.
4.2 Point-Score Protocols for Sewers
Many water utilities use some type of point score method, usually developed in house, for the evaluation
of their water mains. These point score methodologies are typically customized for use by a single utility
and the scoring criteria and weighting factors (although published in the literature) have not been
standardized across the water industry. Examples of customized approaches implemented by the
Louisville Water Company and the Philadelphia Water Department are provided in Section 3.16. In
contrast, the point score method is a fairly established and consistent practice in the condition assessment
of sewer mains, therefore the sewer examples are provided here as an illustration.
A few protocols are available in the literature to record distress indicators in sewers and then translate
them into condition rating, e.g., WRc of UK, NRC of Canada, Water Services Association of Australia
-------
(2002), Cemagref (2003) and the National Association of Sewer Service Companies (NASSCO). The
WRc protocol is perhaps the most widely used protocol today. It was initiated in 1978 as a five-year
research project to investigate failures of sewer mains. Based on this research, the Sewerage
Rehabilitation Manual was developed (WRc, 1986; WRc, 1993; WRc, 1994; WRc, 2001). The latest
manual includes a computerized grading system compatible with European defect coding systems, and
new design methods for renovation techniques (WRc, 2001). The NRC's protocol is known as
Guidelines for Condition Assessment and Rehabilitation of Large Sewers (Zhao et al., 2001). These
guidelines were developed in partnership with several Canadian municipalities and consulting engineers
and are intended for large diameter sewers (>900 mm) only. NASSCO developed a similar set of coding
standards based on the WRc system (WRc, 2001).
Virtually all protocols largely use a point scoring approach, whereby each type of defect is assigned a
score ("deduct value"). After a score is assigned to each observed defect, all of the scores are tallied and
the totals are used to rate the condition of the pipe. While WRc uses a scale of 165 deduct values for
condition rating, NRC uses a 20-point scale. Table 4-1 illustrates the point score schemes of both
protocols. Note that both protocols contain separate scoring schemes for structural and operational
observed defects. This report addresses only the structural aspects of pipe condition. It is clear that while
WRc uses a five grade rating, NRC uses six, from zero to five where 0 = excellent, 1 = good (G), 2 = fair
(F), 3 = poor (P), 4 = bad (B) and 5 = imminent collapse (1C). Table 4-2 provides a summary of structural
distress indicators (defects) and their associated point scores (deduct values) in the two protocols.
Table 4-1. Comparison of Two Point Scoring Protocols
Protocol
WRc scores
NRC scores
Condition states (structural)
0(E)
N/A
0
1(G)
<10
1-4
2(F)
10-39
5-9
3(P)
40-79
10-14
4(B)
80 - 164
15-19
5 (1C)
>165
20
Note: N/A = not applicable
The process of applying the protocol to real situations is inherently imprecise and subjective. Often, two
different evaluators will provide different scores to the same observed defects. Moreover, from Table 4-
1, it can be seen that a score of 80 (WRc protocol) is in fact equivalent to 164 because both scores would
translate to the same condition state of 4.
4.3
Fuzzy Theory Based Techniques
The interpretation of pipe distress indicators (observed through NDT) into a condition rating involves a
certain amount of subjective judgment. Fuzzy sets with their notion of membership functions are very
well suited to accommodate this subjectivity. Further, practitioners have an intuitive understanding of the
deterioration process in buried pipes, although many of the relationships between cause and effect are not
well understood, let alone quantified. Fuzzy techniques seem well suited to capture this intuition.
4.3.1 Fuzzy Synthetic Evaluation. The fuzzy synthetic evaluation (FSE)-based approach
comprises three steps: fuzzification of raw data, aggregation of the various types of observed distress
indicators, and de-fuzzification that adjusts the fuzzy condition rating to a practical crisp format (Kleiner
et al., 2005; Rajani et al., 2006).
89
-------
Table 4-2. Distress Indicators and Their Assigned Scores (Deduct Values)
Distress indicator
(defect)00
Longitudinal crack
Circumferential crack
Diagonal crack
Longitudinal fracture
Circumferential
fracture
Diagonal fracture
Deformation
Surface damage
(spalling)
Joint displacement
Broken pipe
Collapse
Distress level(b)
• Light (up to 3 cracks, no leakage)
• Moderate (> 3 cracks, leakage)
• Light (up to 3 cracks, no leakage)
• Moderate (> 3 cracks, leakage)
• Light (up to 3 cracks, no leakage)
• Moderate (> 3 cracks, leakage)
• Severe (multiple cracks, leakage)
• Light (< 10 mm)
• Moderate (10 - 25 mm or more than one)
• Severe (> 25 mm)
• Light (< 10 mm)
• Moderate (10 - 25 mm or more than one)
• Severe (> 25 mm)
• Light (< 10 mm)
• Moderate (10 - 25 mm or multiple)
• Severe (> 25 mm)
• Light (< 5% change in diameter)
• Moderate (5% - 10% change in diameter)
• Severe (11% - 25% change in diameter)
• Light
• Moderate
• Severe
• Light (< % pipe wall thickness)
• Moderate (1A - 1A pipe wall thickness)
• Severe (> 1A pipe wall thickness)
-
-
Unit
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
m
each
each
each
each
each
Scores
NRC
3
5
3
5
3
5
10
5
10
15
5
10
15
5
10
15
5
10
15
3
10
15
3
10
15
15
20
WRc
10
40
10
40
N/A
N/A
40
40
80
N/A
40
80
N/A
40
80
N/A
20
80
165
5
20
120
N/A
1
2
60
165
(a) This is a partial list, based on the cited references.
(b) Definitions sometimes vary between the two protocols.
For example, if a distress indicator shows loss of 50% of pipe wall thickness, given a pre-defined fuzzy
scale of mild, medium, severe, or critical for wall loss, the observed 50% loss could be fuzzified to
somewhere between severe and critical, say 0.7 membership to severe and 0.3 membership to critical. It
follows then that if multiple distress indicators are provided, each is fuzzified in this way into an
appropriate, pre-defined fuzzy scale.
In the next step, the fuzzified values of the various distress indicators are aggregated with appropriate
weights (weights are assigned according to the importance of a given distress indicator to the
determination of the overall condition rating) to provide a fuzzy condition rating. Rajani et al. (2006)
proposed a seven-state condition rating (excellent, good, adequate, fair, poor, bad, and failing).
On such a scale, an example of a fuzzy condition rating could be 0, 0.1, 0.3, 0.4, 0.15, 0.05, 0, where
these values denote membership values corresponding to the seven condition states. As practitioners
intuitively understand that realistically the condition rating of a pipe cannot have non-zero membership
values to more than three contiguous states, the proposed method provides a process to re-distribute
membership values (wherever needed) to conform to this practical constraint.
90
-------
In the final step, the fuzzy condition rating can be de-fuzzified into a representative (equivalent to mean)
value. This is done to enable comparisons as needed.
This method was proposed to facilitate a deterioration model based on a so-called fuzzy Markov process,
which is described in the next section. It should be noted that fuzzy-based methods require mathematical
training that is typically not provided to practitioners, therefore these methods do not lend themselves to
easy in-house implementation. However, computer software (e.g., T-WARP - see Section 6.2) can make
the technique available through an easy-to-understand user interface.
4.3.2 Fuzzy Composite Programming. Fuzzy composite programming (FCP) is a mathematical
programming technique that employs a single level normalized/non-normalized distance-based
technology to rank a discrete set of solutions based on their distances from an ideal solution. Pipe
condition assessment needs to combine completely different variables into an overall condition indicator.
This problem is actually making decisions based on multiple criteria (often formally known as multiple-
criteria decision making). Vairavamoorthy et al. (2006) applied this method to the condition rating of
pipe, with condition indicators considered in the FCP method shown in Figure 4-2. The following steps
are involved (Vairavamoorthy et al., 2006):
• Identify the pipe condition indicators;
• Prepare the hierarchical structure of pipe condition indicators;
• Obtain the weightings for each indicator and decide the balance factor (balance factor
determines the degree of compromise between indicators of the same group);
• Normalize all of the indicators into scale [0, 1];
• Obtain a fuzzy member by using the FCP-based hierarchical aggregation process for each
pipe;
• Rank the fuzzy numbers.
Pipe
condition
assessment
Physical
indicators
Pipe
' indicators
Material decay
Diameter
Length
Internal proection
External protection
Bedding condition
Installation s^—i Workmanship
indicators \l—' Joint method
No. of joints
Operational
indicators
• Corrosion
indicators
Load/strength
indicators
Intermittency
indicators
Failure
' indicators
Year of install
/I—| Soil corrosivity
Surface permeability
Groundwater condition
Buried depth
<* I Traffic load
Hydraulic pressure
No. of valves
No. ofwater supply periods per day
ofwater supply per day
<^ | Breakage history
Figure 4-2. Pipe Condition Assessment Indicators
(Vairavamoorthy et al., 2006)
91
-------
It should be noted that some of the raw criteria (Figure 4-2 right column) such as bedding condition and
workmanship are not easily quantified. Also, some of these raw criteria (no. of supply periods) may be
appropriate only for developing countries. Furthermore, the FCP method can be sensitive to the weights
and balance factors. An example of the calculation is available in Vairavamoorthy et al. (2006).
This methodology was tested for utilities in India and Uganda. In these countries, water supply can be
intermittent leaving periods of time where the water mains are unpressurized, which increases the risk of
contaminant ingress. For this reason, the results of the FCP method for water main condition were
compared to seepage envelopes from foul water bodies (ditches, sewers, etc) to create maps of regions
with high risk for cross contamination. For example, a pipeline predicted to be in poor condition via the
FCP model and in close proximity to seepage sources was given a high risk for cross contamination.
Although significant data was collected from each utility as shown in Figure 4-2, these were largely
indirect indicator data on the pipe condition and the results of the FCP model were not verified to pipe
condition in the field (Vairavamoorthy et al., 2006).
4.4 Data Fusion and Data Mining
The purpose of data fusion is to combine the capabilities of each sensor modality with historic data to
provide more accurate and complete information (Juliano et al., 2005). Three factors should be
considered:
(a) Redundancy of information presented in the sensor modalities;
(b) Diversity in the sensor modalities;
(c) Complementary sensor modalities.
Data fusion is not limited to sensory data. The analysis benefits from multi-source information to
diminish the uncertainties and inaccuracies in the data.
Data mining is defined as a technique to identify useful patterns or trends from data. Where pipe
condition assessment is concerned, the data mining technique is applied to predict the residual life, burst
rate, and/or leakage based on historical records, or other attributes, e.g., pipe age, diameter, soil type, etc.,
(Savic and Walters, 1999).
4.4.1 Hierarchical Evidential Reasoning. A hierarchical evidential reasoning (HER) model was
proposed to combine different distress indicators at different hierarchical levels using the Dempster -
Shafer's (D-S) rule of combination (Bai et al., 2008).
The framework of the HER model is illustrated in Figure 4-3. The attribute at a higher level is evaluated
based on the assessment of its associated lower-level factors. In the HER model, elements of basic
evidence are referred to as factors, which are essentially distress indicators. Attributes are the categories
(Rajani et al., 2006). The distress indicators (factors) are aggregated to evaluate categories (attributes).
The overall condition is obtained by the aggregation of categories.
The most important part of applying the D-S fusion rule is the definition of basic probability assignment
(BPA). The BPA for each factor is derived based on a degree of confidence assigned to these condition
states as well as the associated importance and reliability of the data.
92
-------
ffl
»iU-)=S(ei')x/; m(e?< )=S(e{' )x / ' ' m(e't )=S(e| >< /it
Attribute level
Factor level
Model inputs for
assigning BPA to
different factors
Figure 4-3. The Framework of Hierarchical Evidential Reasoning
(Bai et al., 2008)
Note: where m represents the bodies of evidence, S the condition rating, and e the contributing factors.
4.4.2 Incremental Learning. LEARN++ is a supervised learning algorithm that makes it possible
for a classifier to learn incrementally from new data without forgetting what has been learned in earlier
training sessions (Polikar et al., 2001). The current LEARN++ algorithm is implemented for the
classification with multilayer perceptron (MLP) neural networks. The idea is to assemble weak classifiers
to achieve an improved performance in classification. This makes LEARN++ very useful for the
interpretation of pipe inspection data. The inspection data may not be sufficient or good enough when a
classifier is being trained. Nevertheless, the classifier can be further improved when new data become
available. Fusion of MFL, thermography, and ultrasonic data for gas transmission pipeline was described
in a technical report (Mandayam et al., 2006). Improved performance for defect identification and
characterization was reported.
4.4.3 Genetic Algorithm. A genetic algorithm (GA) is a search technique that can be applied in
large, complex, and multimodal search spaces. It emulates biological principles, such as inheritance,
mutation, selection, and crossover, to solve complex optimization problems. The GA has the ability to
locate regions that potentially contain an optimal solution for a given problem by searching the solution
space (Shaw et al., 2004).
Three researchers were identified in the literature as having applied this technique to water mains
including Babovic et al. (2002), Vitkovsky et al. (2000), and Dandy and Englehardt (2001). GA was used
to search the best scoring model to determine the risks of pipe bursts (Babovic et al., 2002). The scoring
model that is a function of associated characteristics of bursting pipe can be established by analysis of
burst events that have already occurred. GA was also employed as a search method in the inverse
transient technique for leak detection (Vitkovsky et al., 2000). GA has also been applied to identify the
schedule of pipe replacement in a deteriorating water distribution system (Dandy and Engelhardt, 2001).
4.5 Data Driven Approaches to Predict Condition Rating Based Only on Inferential
Indicators
The high cost of thorough inspection to observe distress indicators has motivated researchers to try and
predict the condition of pipes based on a reduced set of indicators, or just on inferential indicators, which
are generally easier and cheaper to obtain.
93
-------
The data driven approaches comprise multiple input data, the corresponding observed output data, and a
mathematical relationship that is supposed to link a set of inputs with a corresponding output. This
mathematical process is usually not a-priori known. However, given sufficient data, it can be discerned
by a process known as 'training'. Whereas the mathematical relationship (sometimes also called 'model')
comprises mathematical operators and coefficients, the training process in essence finds the coefficients
that, given a set of inputs, computes outputs that match the observed outputs as closely as possible. In the
context of interpretation of inferential indicators to condition ratings of pipes, the input is a set of
observed distress indicators and the output is the condition rating of the pipe. The general steps involved
are:
(a) For a sufficiently large and diverse inventory of pipes, perform full inspection, including
distress and inferential indicators. The inferential indicators will serve as input data.
(b) Based on these distress indicators, obtain condition ratings for these pipes. These condition
ratings will serve as a set of output data.
(c) Model training (or calibration): calibrate the proposed mathematical relationship (by varying
the various coefficients) so that given a set of inputs (inferential indicators), the model
computes an output (condition rating) that is as close as possible to the observed condition
rating that corresponds to the same inputs. Repeat for all pipes.
(d) Validation: once the model has been calibrated, examine its ability to predict condition rating
(based on inferential indicators only) on a set of pipes that were not used in the calibration
process.
It should be noted that so far, these efforts have been applied only to sewers and not to water mains.
Some of the mathematical relationships or models documented that have been used for this purpose
include the following (non-exhaustive) list:
• Logistic regression (e.g., applied by Ariaratnam et al. [2001] to the sewer system in
Edmonton Canada). Distress indicators were largely obtained by CCTV inspection (Davies
etal.,2001).
• Artificial neural networks (e.g., Najafi and Kulandaivel, 2005; Tran, 2007; Moselhi and
Fahmy, 2008; Al-Barqawi and Zayed, 2006; Achim et al., 2007).
• Bayesian statistics (e.g., Fenner and Sweeting, 1999).
• A metaheuristic linear classifier model (Wright et al., 2006).
• A fuzzy-based method to estimate soil corrosivity from soil properties (Sadiq et al., 2004).
• A fuzzy expert system to estimate the soil corrosivity potential from soil properties (Najjaran
etal., 2006).
• Fuzzy PROMETHEE (preference ranking organization method for enrichment evaluation)
(Zhou et al., 2009).
The benefits of this class of methods would depend on three main issues: (a) the ability of the model to
predict asset condition in a credible manner, (b) the ratio between the cost of obtaining inferential data
and that of actual inspection, and (c) the general state of the asset network. Given a large network and
limited resources to inspect its entirety in a reasonable time period, and given a relatively credible model
and inexpensive inferential data, the use of such a model could be quite beneficial in screening assets for
detailed inspection.
94
-------
5.0: WATER MAIN DETERIORATION MODELS
Deterioration models of water mains can be classified into two main categories, physical (or mechanistic)
models and statistical/empirical models. While physical models are scientifically more robust and widely
applicable (i.e., they could be applied to various pipe materials and external conditions, given data
availability), they are limited by existing knowledge and available data. Some of the data required for the
physical models are very costly to obtain if at all available at a resolution of an individual pipe. These
costs may be justified only for major transmission water mains, where the cost of failure is significant and
failure needs to be prevented. In contrast, the empirically derived statistical models can be applied with
various levels of input data and may thus be useful in relatively small water mains for which the low cost
of failure entails failure frequency management (not prevention), and therefore expensive data acquisition
campaigns cannot be economically justified.
5.1 Physical/Mechanistic Models
The physical mechanisms that lead to pipe breakage are often very complex and are not completely
understood, but recent efforts look promising. These physical mechanisms involve several aspects
including pipe-intrinsic properties (e.g., material type, pipe geometry, type of joints, quality of
installation); loads (including internal loads due to operational pressure and external loads due to soil
overburden, traffic, frost and third party interference); and finally material deterioration (due to external
and internal chemical, bio-chemical and electro-chemical environment). The principles of the structural
behavior of buried pipes are, for the most part, fairly well understood. However, the understanding of
issues such as frost loads and structural deterioration due to chemical processes and fatigue, is still quite
varied, but substantial progress has been made in this direction in recent years.
In North America in the 1990s, a water mains survey concluded that about two thirds of the installed
water mains inventory was CI and DI pipes, about 15% was AC pipes, and the rest was plastic, concrete,
steel, and others (Kirmeyer et al., 1994; Rajani and McDonald, 1995). In the last decade and a half, these
proportions have likely changed to some degree (as the shares of PVC and concrete pressure pipes have
increased at the expense of metallic and AC pipes); however, the majority (>60%) of existing water mains
is still CI and DI pipes.
The predominant deterioration mechanism on the exterior of CI and DI pipes is electro-chemical
corrosion causing damage in the form of corrosion pits. The damage to grey CI is often disguised by the
presence of graphitization. Graphitization is a term used to describe the network of graphite flakes
remaining behind after the iron in the pipe has been leached away by corrosion. Either form of metal loss
represents a corrosion pit that will grow with time and eventually may lead to a water main break. The
physical environment that surrounds the pipe has a significant impact on the deterioration rate. Factors
that accelerate corrosion of metallic pipes are stray electrical currents, and soil characteristics such as
moisture content, chemical and microbiological content, electrical resistivity, aeration, redox potential,
etc. The interior of a metal pipe may be subject to tuberculation, erosion and crevice corrosion resulting
in a reduced effective inside diameter, as well as a breeding ground for bacteria. Severe internal
corrosion may also impact pipe structural deterioration. The supply water affects the internal corrosion in
pipes through its chemical properties, e.g., pH, dissolved oxygen, free chlorine residual, alkalinity, etc., as
well as temperature and microbiological activity.
The long-term deterioration mechanisms in PVC pipes are not as well documented mainly because these
mechanisms are typically slower than in metallic pipes and also because PVC pipes have been used
commercially only in the last 35 to 40 years. A recent WaterRF research report (Burn et al., 2005)
provides a rather comprehensive account of available pertinent information on the structural properties of
95
-------
PVC pipes as well as on the impact of water quality on the pipe material. Additionally, while
hydrocarbons such as high octane gasoline do not affect PVC, other contaminants such as toluene can
permeate the pipe wall if present in sufficiently high concentration levels (Ong et al., 2008).
AC and concrete pipes are subject to deterioration due to various chemical processes that either leach out
the cement material or penetrate the concrete to form products that weaken the cement matrix. The
presence of inorganic or organic acids, alkalis, or sulfates in the soil is directly responsible for concrete
corrosion. In reinforced and prestressed concrete, low pH values in the soil may lower the pH of the
cement mortar to a point where corrosion of the prestressing or reinforcing wire will occur, resulting in
substantial weakening of the pipe (Dorn et al., 1996).
There are four general classes of pipe failures: holes due to corrosion; circumferential breaks caused by
longitudinal or bending stresses; longitudinal breaks caused by hoop stresses; and split bells. Split bells
can be the result of pipe rotation due to differential movement or small cracks introduced during
transportation and/or installation which, over the years, were subjected to cyclical loads until fatigued to
failure (Raj ani, 2010).
Circumferential breaks are typically the result of thermal contraction acting on a restrained pipe, bending
stress (beam failure caused by soil differential movement or prolonged leaks creating large voids in the
bedding), inadequate trench and bedding practices, or third party interference or a combination of one or
more of the above. The contribution of operating pressure to longitudinal stress, although small, may
increase the risk of circumferential breaks when occurring simultaneously with one or more of the other
sources of stress. In some circumstances (e.g., power failure in a pumping station, fast closure of a valve
on a pipe that is subject to high flow velocity), transient pressures can introduce large stresses in the pipe.
Longitudinal breaks caused by transverse stresses are typically the result of either hoop stress due to
pressure in the pipe, ring stress due to soil cover load, ring stress due to live loads caused by traffic,
increase in ring loads when penetrating frost causes the expansion of frozen moisture in the ground, or a
combination of one or more of the above.
Raj ani and Kleiner (2001) provided a comprehensive review of the main physical models found in the
literature through the end of the 1990s. It is not the intent of this report to duplicate their review. Instead,
Table 5-1 provides the main points of each model reviewed by Rajani and Kleiner (2001), along with
some additional models published in the last 10 years. Table 5-1 is by no means an exhaustive list of all
relevant models. Table 5-1 is presented in chronological order of year of publication of reference.
Table 5-1. Physical/Mechanistic Models for Pipe Deterioration
Reference
Issues Addressed
Data Requirement
Comments
Spangler, 1941;
Watkins and
Spangler, 1958
Pipe-soil interaction
analysis
Pipe elastic modulus, internal
pressure, pipe geometry, trench
geometry, some soil/backfill
properties, vehicle impact factor
and wheel load on surface.
Assumes in-plane action only
- appropriate for large
diameter pipes, but not for
small diameter pipes. Thermal
issues not addressed, as well
as material deterioration and
soil shrinkage effect.
96
-------
Table 5-1. Physical/Mechanistic Models for Pipe Deterioration (Continued)
Reference
Rossum, 1969;
Doleac, 1979; Doleac,
etal, 1980
Kumar etal, 1984
Kiefner and Vieth,
1989
Randall-Smith etal.,
1992
Ahammed and
Melchers, 1994
Rajanietal., 1996
Rajani and Zhan,
1996; Zhan and
Rajani, 1997
Pandey, 1998; Hong,
1997
Rajani and Makar,
1999
Rajani and Makar,
1999
Issues Addressed
Predict remaining
wall thickness of pit
cast mains
Corrosion status
index
Residual structural
resistance
Estimate remaining
service life of water
mains
Estimate the
probability of failure
in steel pipelines
Pipe-soil interaction
analysis of jointed
pipe
Frost loads
Estimate the
probability of failure
in steel pipelines
Residual structural
resistance
Estimate the
remaining service
life of grey cast iron
mains
Data Requirement
Soil properties such as pH,
resistivity and redox potential.
Pipe age, type, wall thickness,
diameter, joints, soil properties -
resistivity, chlorides, sulfides, pH,
moisture, year of first leak (if
available).
Material properties, 3D
characteristics of pipe corrosion
pits.
Current age and maximum pit
depth.
Mechanical properties of steel
and constants for power-law
corrosion model.
Same as Watkins and Spangler
(1958) plus thermal properties of
pipe and special soil properties to
simulate pipe-soil adhesion.
Continuous freezing index, soil
backfill porosity, segregation
potential, unfrozen water content,
thermal gradient at the freezing
front, frost depth.
Mechanical properties of steel
and constants for power-law
corrosion model.
Similar to that proposed by
(Kiefner and Vieth, 1989).
Pipe geometry and mechanical
properties of cast iron and soil
properties used in the Rossum
(1969) model or empirical
parameters used to define two-
phase corrosion model.
Comments
Time to failure estimated
using a power law as a
function of soil properties and
time.
Power-law model used for
corrosion rate.
Empirical/statistical
formulation (time-exponential
to forecast breaks).
Developed for oil and gas
steel pipelines. Appropriate
for ductile materials, e.g., DI
and steel, but not CI.
Time to failure estimated
assuming that corrosion is a
linear function of time.
Probability of failure was
found to be most sensitive to
constants for the corrosion
model
Longitudinal bending is
considered as primary action -
appropriate for small diameter
pipes. No account for material
deterioration and for soil
shrinkage effects.
Provides frost load as a
function of time
Pipeline reliability estimated
with a probabilistic analysis
framework that incorporates
impact of inspection and
repair activities.
Addresses fracture toughness.
Appropriate for brittle
material for CI. Needs large-
scale validation.
Time to failure estimated
assuming that corrosion pits
grow as predicted by Rossum
or two-phase corrosion
models.
97
-------
Table 5-1. Physical/Mechanistic Models for Pipe Deterioration (Continued)
Reference
Issues Addressed
Data Requirement
Comments
Hadzilacos etal.,
2000
Reliability-based
prediction of pipe
residual life.
Pipe elastic modulus, internal
pressure, pipe geometry, trench
geometry, some soil/backfill
properties, vehicle impact factor
and wheel load on surface, and
information of loss of bedding
support.
Probability of failure
determined for different
failure modes.
Deb et al., 2002
Mechanistic models
to rank deterioration
of cast iron pipes.
Pipe geometry and mechanical
properties of cast iron, soil
properties, and climate data.
Rossum's pit growth model was
used to determine pit growth from
soil properties.
Mechanistic model is
essentially the same as
developed by Rajani and
Makar (1999) except that out-
of-plane response was
included besides the in-plane
response. Modifications to the
parameters in Rossum's pit
growth model were not
explained.
Rajani and
Tesfamariam, 2004;
Tesfamariam et al.,
2006
Pipe-soil interaction
analysis of partially
supported jointed
pipes.
Pipe elastic modulus, internal
pressure, pipe geometry, trench
geometry, some soil/backfill
properties, vehicle impact factor
and wheel load on surface,
thermal properties of pipe and
special soil properties to simulate
pipe-soil adhesion and
information of loss of bedding
support.
Longitudinal bending is
considered as primary action -
appropriate for small diameter
pipes. Does not account for
material deterioration and for
soil shrinkage effects.
Davis etal., 2007
Prediction of failure
in PVC pipes.
Pipe geometry, material
properties (Young's modulus,
yield strength, other coefficients),
Soil properties, internal pressure,
burial depth, crack growth
parameters.
Fracture mechanics used to
model failure stress as a
function of internal pressure,
external loads, residual
stresses in the pipe, pipe and
crack geometry. Paris law
assumed for crack growth.
Probabilistic treatment of
variability in material
properties and degradation
rate.
Mogliaetal., 2008
Failure prediction in
cast iron pipes.
Pipe geometry, age, operating
pressure, corrosion rate, pipe
material properties (to determine
tensile strength), soil properties
(to determine dead load), dynamic
external loads.
Deterministic limit state
(based on fracture mechanics),
with probabilistic inputs.
Monte-Carlo simulation used
to compute failure time, where
corrosion rate is assumed
linear over time.
98
-------
Table 5-1. Physical/Mechanistic Models for Pipe Deterioration (Continued)
Reference
Rajani and
Tesfamariam, 2007
Davis etal., 2008
Issues Addressed
Estimate the
remaining service
life of grey cast iron
mains with
consideration of
partially supported
jointed pipes.
Prediction of failure
in asbestos cement
pipes.
Data Requirement
Pipe elastic modulus, internal
pressure, pipe geometry, trench
geometry, some soil/backfill
properties, vehicle impact factor
and wheel load on surface,
thermal properties of pipe and
special soil properties to simulate
pipe-soil adhesion and
information of loss of bedding
support and corrosion model
constants.
Pipe geometry and material
properties extracted by testing of
specimens, soil/bedding
properties, burial depth, operating
pressure.
Comments
Longitudinal bending is
considered as primary action -
appropriate for small diameter
pipes. Corrosion, loss of
bedding support, temperature
differential and pipe material
toughness are identified as the
most important parameters to
influence pipe longevity.
Model assumes linear rate of
strength loss. Probabilistic
treatment of variability in
material properties and
degradation rate.
5.2
Statistical/Empirical Models
Statistical/empirical models quantify the structural deterioration of water mains by analyzing historical
performance data. In small distribution water mains, this historical performance is manifested in
observed breakage frequency. Historical performance of large transmission mains is usually measured on
an ordinal condition rating scale. Kleiner and Rajani and (2001) provided a comprehensive review of the
major statistical/empirical models found in the literature through the end of the 1990s. Table 5-2 provides
the main points of each model reviewed by Kleiner and Rajani (2001), along with some additional models
published in the last 10 years. Table 5-2 is by no means an exhaustive list of all relevant models.
Table 5-2 is presented in chronological order of year of publication of reference.
Table 5-2. Statistical/Empirical Models for Pipe Deterioration
Reference
Shamir and Howard,
1979
Clark etal., 1982
McMullen, 1982
Kettler and Coulter,
1985
Type
Deterministic
Deterministic
Deterministic
Deterministic
Type Of
Deterioration
Breakage
frequency
Breakage
frequency
Breakage
frequency
Breakage
frequency
Data Required
Pipe length, installation
date, breakage history.
Time of installation,
breakage history, type
and diameter of the pipe,
operating pressures, soil
corrosivity and zoning
composition of area
overlaying pipe.
Saturated soil resistivity,
soil pH,
redox potential.
Same data as Shamir and
Howard (1979).
Comments
Time exponential model.
Analysis more effective on
homogenous cohorts,
therefore data on pipe
diameter, material, soil type,
break type, etc. very useful.
Mixed time-linear and time-
exponential model.
Additional types of data such
as the type of breaks and pipe
vintage required to enhance
model.
Model predicts time to first
break.
Time linear model.
99
-------
Table 5-2. Statistical/Empirical Models for Pipe Deterioration (Continued)
Reference
Kulkarnietal, 1986
Walski, 1987
Andreou et al.,
1987a; 1987b;
Marks et al., 1987;
Bremond, 1997;
Eisenbeis, 1994;
Rostum, 2000
Constantino and
Darroch, 1993;
Miller, 1993;
Constantino et al.,
1996; Rostum,
2000; Economouet
al., 2008
Jacobs and Karney,
1994
Lietal., 1997;
1996; 1995
Madanat et al.,
1995;Mauchand
Madanat, 2001
Herz, 1996; Kropp
and Baur, 2005
Lei, 1997; Eisenbeis
etal., 1999
Hong, 1998
Type
Probabilistic
Deterministic
Probabilistic
Probabilistic
Deterministic
Probabilistic
Probabilistic
Probabilistic
Probabilistic
Probabilistic
Type Of
Deterioration
Breakage
frequency
Breakage
frequency
Breakage
frequency
Breakage
frequency
Breakage
frequency
Condition
rating
Condition
rating
Survival
analysis
Breakage
frequency
Condition
rating
Data Required
Pipe length, breakage
history, data to create
pipe cohorts (more data
more refined analysis).
Same data as Shamir and
Howard (1979) plus
information on the
method of pipe casting
and pipe diameter.
Pipe length, installation
year, operating pressure,
% low land
development, breakage
history, soil corrosivity.
Mean static pressure,
overhead traffic
conditions, pipe
diameter, material length
soil type.
Pipe length, age,
breakage history.
Asset condition rating
and age.
Asset condition rating.
Data to create pipe
cohorts, installation
year, historical
replacement year, expert
opinion on pipe life
expectancy.
Pipe age, diameter,
length, material, traffic
loading, soil acidity, soil
humidity, breakage
history.
Pipe operating pressure
and remaining strength.
Comments
Based on Bayesian analysis
of relative breakage
frequencies in the various
cohorts.
Time exponential model.
Data to enable homogenous
cohorts very useful.
Proportional hazards model
for inter-break duration. Not
all listed data are essential.
Other data types could be
incorporated if available.
Time-dependent Poisson-
based models.
Not all listed data are
essential. Other data types
could be incorporated if
available.
Time linear model. Data to
enable homogenous cohorts
very useful.
Markov-based model with
non-homogeneous transition
probabilities.
Markov deterioration
processes, with underlying
latent continuous
deterioration process. Models
were developed for general
infrastructure assets, not
specifically pipes.
Cohort survival model base
on the Herz probability
distribution.
Accelerated life-based
models. Not all listed data
are essential. Other data
types could be incorporated
if available.
Markov-based model.
Condition states defined as
ratio between pressure and
strength.
100
-------
Table 5-2. Statistical/Empirical Models for Pipe Deterioration (Continued)
Reference
Rostumetal., 1999
Gustafson and
Clancy, 1999a
Abraham and
Wirahadikusumah,
1999;,
Wirahadikusumah et
al., 2001
Kathula and
McKim, 1999;
McKim et al., 2002
Ariaratnametal.,
1999; 2001; Davies
etal., 2001; Cooper
et al., 2000
Dandy and
Engelhardt, 2001
Kleiner, 2001
Micevski et al.,
2002
Park and
Loganathan, 2002
Type
Probabilistic
Probabilistic
Probabilistic
Probabilistic
Probabilistic
Deterministic
Probabilistic
Probabilistic
Deterministic
Type Of
Deterioration
Condition
rating
Breakage
frequency
Condition
rating
Condition
rating
Condition
rating
Breakage
frequency
Condition
rating
Condition
rating
Breakage
frequency
Data Required
Pipe condition ratings,
age.
Detailed breakage
history, data to create
pipe cohorts.
Pipe condition rating.
Pipe condition rating.
Various inferential
indicators.
Breakage history, data to
create pipe cohorts.
Pipe condition ratings
preferably based on two
or more consecutive
inspections.
Alternatively - expert
opinions.
Pipe condition rating.
Breakage history, data to
create pipe cohorts.
Comments
Time duration between
deterioration states a random
variable with a Herz
probability distribution.
Breakage history modeled as
a semi-Markov process in
which each break order (e.g.,
1st, 2nd, 3rd break, etc.) is
considered a "state" in the
process and the inter-break
time it is considered the
"holding time" between state
(i - 1) and state i.
Markov chain process
applied to sanitary sewers.
Four phases considered in
pipe life. Transition
probabilities stationary
within each phase.
Markov chain process
applied to sewers. Expert
opinion used to derive
transition probabilities. Later
introduced risk ratios.
Models base on multi-
covariate logistic regression,
where one of the covariates
is pipe age.
Model based on power law
increase in breakage
frequency. Coefficients
extracted by pure regression.
Semi-Markov based model.
Markov-based model for
deterioration of storm water
sewers. Transition
probabilities assumed
homogeneous over time.
A mixed time-
exponential/time-linear
model.
101
-------
Table 5-2. Statistical/Empirical Models for Pipe Deterioration (Continued)
Reference
Mailhotetal., 2003;
Dridi et al., 2005
Watson et al., 2004
Kleiner and Rajani,
2004
Kleiner et al., 2006a
Giustolisi and
Berardi, 2007;
Berardi et al., 2008
LeGat, 2008a
LeGat, 2008b
Kleiner and Rajani,
2009
Type
Probabilistic
Probabilistic
Deterministic
Probabilistic.
Deterministic
Probabilistic
Probabilistic
Probabilistic
Type Of
Deterioration
Breakage
frequency
Breakage
frequency
Breakage
frequency
Condition
rating
Breakage
frequency
Breakage
frequency
Condition
rating
Breakage
frequency
Data Required
Breakage history, data to
create pipe cohorts.
Breakage history, data to
create pipe cohorts.
Breakage history, data to
create pipe cohorts,
history of cathodic
protection, climate
history.
Distress indicators from
at least one inspection -
interpreted into
condition rating.
Breakage history, data to
create pipe cohorts.
Breakage history, pipe
data (material, diameter,
etc. that could have
potential influence on
breakage rates.
Asset condition
rating (preferably more
than one), and age.
Covariates can be
considered if known.
Breakage history, data to
create pipe cohorts,
history of cathodic
protection, climate
history.
Comments
Distinguish between inter-
break time in lower order
breaks (Weibull or gamma
distribution) and higher order
breaks (exponential
distribution). Assume linear
or power law relationship
between mean duration and
age in the higher order
breaks.
Deterioration model based on
the nonhomogeneous Poisson
process. No explicit time-
dependency is assumed,
rather it is implied based on
Bayesian updating.
D-WARP. Time exponential
model. Can consider
dynamic (time-dependent)
covariates (e.g., climate,
cathodic protection).
T-WARP. Deterioration of
large water transmission
mains modeled as fuzzy
Markov deterioration
process.
Model(s) based on
evolutionary polynomial
regression (EPR): fit a
parsimonious polynomial to
observed historical breakage
rates, using genetic
algorithm.
Model based on Yule (pure
birth) process, with a linear
extension.
Model for drainage pipes,
based on nonhomogeneous
Markov chain. Transition
probabilities are derived
from Gompertz survival
probabilities.
I-WARP. Model based on
the nonhomogeneous Poisson
process, with capability to
address dynamic (time-
dependent) covariates.
102
-------
Tables 5-1 and 5-2 provide models found in the literature for pipe deterioration. This compilation was not
intended to provide practitioners with sufficient information to decide which of the models is most
suitable for their own circumstances; as such a decision would require a level of details that is beyond the
scope of this report. Further, it would have been useful to know which of these models have actually
been used in practice, by whom and under what circumstances, and to what degree of success. However,
such information is usually not available. For example, Grigg (2007) reported on a survey that comprised
45 water utilities, where only a few employed break prediction methods from the literature and some
developed their own method.
As a general observation, it may be safely assumed that the likelihood of a model to be used in practice
increases significantly if the model is implemented in a software program that is publicly available. In
Section 6.2, a list of publicly available software programs is provided, some of which are based on
models described in Table 5-1 or 5-2. Consequently, these models are likely to have been (or to be) used
in practice, while others are likely to have been used sporadically at best.
103
-------
6.0: DECISION SUPPORT FOR WATER TRANSMISSION AND
DISTRIBUTION SYSTEMS
There are a vast number of models in the literature intended to optimize or near-optimize decisions related
to the renewal of water mains, addressing issues such as prioritization of water mains for renewal,
scheduling water mains for renewal, and also selection of renewal alternatives for water mains. These
models may differ from each other by the number and nature of objectives addressed, the data required,
the solution method, and the constraints considered. The detailed description of all of these models is
beyond the scope of this report. The objective of this section is therefore twofold: to provide the reader
with a general overview of most of the relevant decision support models (Section 6.1), and to provide a
general description for those models that have been transformed into actual decision support tools in the
form of software products that are publicly available either as a commercial product, a research tool or a
prototype computer program (Section 6.2). Additional information on decision support tools for
predicting the performance of water distribution and wastewater collection systems can also be found in
Stone et al. (2002).
6.1
Decision Support Models
Table 6-1 provides an extensive (though not exhaustive) list of decision-support approaches/models found
in the literature. Many of the entries in this table correspond to entries in Table 5-1, as the decision-
support approach was offered as a natural continuation of the pipe deterioration models. Some entries in
Table 6-1 correspond to (usually components of) some software packages.
Table 6-1 is presented in chronological order of year of publication of reference.
Table 6-1. Decision Support Methods and Approaches Found in the Literature
Reference
Shamir and
Howard, 1979
Clark etal.,
1982
Walski, 1987;
Walski and
Pelliccia,
1982
Lansey et al.,
1992
Objectives
Minimize cost
Minimize cost
Minimize cost
Minimize cost
Constraints
No
No
No
Hydraulic,
(minimum
pressure,
continuity,
mass
conservation)
Optimization
Method
Calculus
Calculus
Calculus
Coupled
network solver
(KYPIPE) and
General
Reduced
Gradient
(GRG2)
Comments
Optimal pipe replacement timing is that
which minimizes the discounted costs of pipe
replacement and breakage. Simplified
approach laid basis for numerous extensions
and enhancements.
Similar to Shamir and Howard (1979).
Optimal pipe replacement timing is that
which minimizes the discounted costs of pipe
and valve replacement, leakage and leak
detection, pipe and valve breakage.
Alternatively, corresponding critical break
rate can be computed.
Define planning periods, each with associated
breakage frequency, demand flows, pipe
friction coefficients. For every pipe find the
period in which to replace/reline/reinforce,
and for every pump the period in which to
replace/reinforce, so as to minimize total
discounted cost of replacement, repair and
pumping energy.
104
-------
Table 6-1. Decision Support Methods and Approaches Found in the Literature (Continued)
Reference
Li and Haims,
1992a; 1992b
Kim and
Mays, 1994
Halhal et al
A Adllldl ^ I dl . ,
1997
Kleiner et al.,
1998a;
Kleiner et al.,
1998b
Gustafson and
Clancy, 1999b
Cooper etal.,
2000
Kleiner, 2001
Objectives
Ivliixiimzc
availability,
allocate funds
to maximize
overall system
availability
Minimize cost
Multi-
objective
(cost and
improved
service)
Minimize cost
Minimize cost
Prioritize
pipes for
renewal based
on failure risk
scores
Minimize risk
Constraints
Available
funds
Hydraulic
(minimum
pressure,
continuity,
mass
conservation)
Cost
(pressure
shortfall is
considered an
objective)
Hydraulic
(minimum
pressure,
continuity,
mass
conservation)
N/A
N/A
N/A
Optimization
Method
Calculus
Implicit
enumeration
scheme using a
branch and
bound
algorithm along
with a
generalized
reduced
gradient
procedure
Structured
Messy genetic
algorithm
Dynamic
programming
& partial
enumeration
coupled with
network solver
EPANET
Calculus
Risk-based
ranking
Calculus
Comments
Two-stage decision-making process based on
Andreou et al.'s (1987a; b) proportional
hazard method (PHM) deterioration. A. Semi-
Markovian model applied to individual water
mains to optimize repair/replace decision
while maximizing the availability of the water
main. B. Multilevel decomposition approach
to optimally distribute available funds among
the distribution network components, so as to
maximize overall system availability.
Finds which pipe to replace/reline/
rehabilitate/continue repair, while minimizing
total cost of replacement, rehabilitation,
reline, repair and pumping energy. Time
dimension not considered.
Produce a Pareto front of non-dominated
solutions with tradeoff between cost and level
of service, defined in four dimensions,
including improved pressure, improved
maintenance, improved operations and
improved water quality. The four dimensions
are combined using weights.
Each pipe in the network assumed to have
exponential increase in break frequency and
logarithmic decline in hydraulic capacity.
Schedules pipe replacement/relining to
minimize total life-cycle discounted costs.
Life-cycle costs consider perpetual
deterioration/replacement cycles.
Based on their pipe deterioration model,
generate potential breaks history, using
Monte Carlo simulations. Optimal
replacement timing is that which minimizes
total discounted cost of replacement and
breakage.
Failure probability determined by logistic
regression, using multiple covariates (soil,
bus and car traffic, peak pressure, etc).
Failure consequences determined by various
factors (affected properties, repair cost, etc.)
that are discerned from GIS.
Optimal time for intervention is that which
minimizes the expected cost of failure.
Expected cost of failure is calculated as
product of probability of failure and its
consequence.
105
-------
Table 6-1. Decision Support Methods and Approaches Found in the Literature (Continued)
Reference
Dandy and
Engelhardt,
2001
Loganathan et
al, 2002; Park
and
Loganathan,
2002
Hahn et al.,
2002
Burnetal.,
2003;Moglia
et al., 2008
Kleiner and
Rajani, 2004
Watson et al.,
2004
Kleiner et al.,
2006b;
Kleiner, 2005
Alvisi and
Franchini,
2006a; 2006b
Objectives
Minimize
costs
Minimize cost
(by
replacement
after the
threshold
break)
See comments
Prioritize
pipes for
replacement
Minimize
cost, analyze
scenarios
Minimize
costs
Minimize cost
given
acceptable
risk level
Minimize
cost, leakage,
unserved
demand
Constraints
Hydraulic,
(minimum
pressure,
continuity,
mass
conservation),
budget
N/A
N/A
N/A
N/A
Hydraulic
constraints
only in the
discrete event
simulator
N/A
Hydraulic,
(minimum
pressure,
continuity,
mass
conservation)
Optimization
Method
Genetic
algorithms
coupled with
network solver
EPANET
Calculus
N/A
Calculus
Calculus
Calculus +
Monte-Carlo
simulations to
propagate
uncertainty
Fuzzy
mathematics
and calculus
Multi-objective
GA
Comments
Assumes breakage frequency follows power
law; minimize the present value of
replacement, repair, and damage costs, by
scheduling pipe replacement, including
selection of appropriate diameters.
Total cost of pipe when it is replaced after the
nth recorded break includes n breaks since
installation + pipe replacement. The nth break
is a threshold break if total discounted cost
associated with it is smaller than that
associated with the (n + l)th break.
Expert opinion pooled to build knowledge
base 'SCRAPS' to support an expert system
intended for the prioritization of the
inspection of sewers.
FARMS. Pipe deterioration modeled as
Nonhomogeneous Poisson Process.
Alternatively, in some cases, physical models
are used. Probability of failure is combined
with consequences to obtain risk. Whole life
costing can be considered. Decision based on
scenario-generation and analysis.
D-WARP. For a cohort, find optimal renewal
time. Also examine scenarios that combine
mixed strategies of replacement and cathodic
protection.
Power law deterioration model (derived from
his nonhomogeneous Poisson process).
Optimal pipe replacement timing is that
which minimizes costs of pipe replacement
and breakage (no discounting). Incorporates
some MCS to consider uncertainties in the
model coefficients.
T-WARP. Combine fuzzy Markov
deterioration with fuzzy failure consequences
to define fuzzy risk over pipe life. If risk
tolerance is exceeded then renew pipe,
otherwise schedule next inspection.
Alternatively, select desired risk/cost tradeoff
from a Pareto front of non-inferior strategies.
Assumes power law increase in breakage rate.
Considers multiple demand patterns to
calculate shortfall in supply (vs. demand)
when a pipe fails. Leak rate calculated under
the assumption that un-reported (and un-
repaired) breaks (or leaks) are a known
proportion of total reported (and repaired)
breaks.
106
-------
Table 6-1. Decision Support Methods and Approaches Found in the Literature (Continued)
Reference
Dandy and
Engelhardt,
2006
Hong ct al
2006
Berardi et al.,
2007
Renaud et al.,
2007
Cabrera et al
2007
LeGauffre et
al., 2007
Dridi et al.,
2008
Objectives
Multi-
objective
fpnct
\\S\Jj\,.,
reliability)
Minimize cost
Multi-
objective
(cost,
reliability)
Prioritize
pipes for
replacement
Minimize cost
Multi-
fYHi pr'tiAJ'p
U UJ CL- 11 V C
prioritization
of pipes for
replacement
Minimize cost
Constraints
Budget,
hydraulic
(minimum
pressure,
continuity,
mass
conservation)
N/A
Hydraulic,
(minimum
pressure,
continuity,
mass
conservation)
Hydraulic
(continuity,
mass
conservation)
N/A
N/A
Hydraulic,
(minimum
pressure,
continuity,
mass
conservation)
Optimization
Method
Multi-objective
GA
VJ^i.
Calculus
Multi-objective
genetic
algorithm
coupled with
network solver
EPANET
Point score
(weighted)
Calculus
Prioritization
process follows
the Electre-Tri
method
OP TIP tlP
VJ^/ ll^UV^
slgorithm
coupled with
nptwnrk" Qnlvpr
lit L WU1.1S. oUlvtl
EPANET
Comments
Assumes breakage frequency follows power
law. Reliability comprises total number of
customers affected by failure. This number
includes customers whose supply is cut off
(local interruption) and customers who
experience too low pressure (global
interruption).
Nonhomogeneous Poisson process assumed
for increase in pipe breakage frequency.
Minimize life-cycle costs (repair and
replacement), where life-cycle costs consider
perpetual deterioration/replacement cycles.
Cost includes pipe break and replacement.
Reliability is defined as the number of
customers affected by a broken pipe.
Breakage frequency discerned using
Evolutionary polynomial regression.
Part of SIROCO. Uses PHM for breakage
prediction. Pipe hydraulic criticality
calculated as the demand shortfall resulting
from the pipe failure. Scores are assigned to
each pipe based on hydraulic criticality,
impact of failure on traffic, on service level,
expected damage, and repair/replace costs. In
addition, two so-called opportunity criteria,
namely coordination with roadwork and need
for rehabilitation index.
Similar to Shamir and Howard (1979) but
considers also water loss during repair,
energy vested in this water loss, social and
other occasional costs that typically
accompany a breakage event.
Basis for Care-W ARP. Compile a list of
evaluation criteria, each with a quantitative or
qualitative rating scheme. Using these
criteria, define n (typically n = 2) threshold
profiles that define (n + 1) grades (e.g., poor,
adequate, good). Classify pipe inventory into
the (n + 1) grades using the Electre-Tri
process. Select the worst ranking pipes for
renewal. Threshold profiles can be re-
calibrated to given budget.
Based on deterioration of breakage frequency
model by Mailhot et al. (2000). Considers
also deterioration of hydraulic capacity. For a
given planning period, schedule for
replacement those pipes that yield minimum
discounted cost.
107
-------
Table 6-1. Decision Support Methods and Approaches Found in the Literature (Continued)
Reference
Davis etal.,
2008
Nafi et al.,
2008
Nafi and
Kleiner, 2009
Objectives
of cost benefit
Multi-
objective
(cost,
hydraulic
reliability)
Minimize cost
Constraints
N/A
Hydraulic
(continuity,
mass
conservation)
Budget
Optimization
Method
Calculus
Multi-
objective,
modified GA
Heuristics - GA
Comments
Probability of pipe failure based on their
model of AC pipe deterioration. Benefits
include the cost of breaks avoided (by pipe
replacement) through the period of pipe
physical existence. Costs include replacement
as well as pre-replacement inspection costs.
All pipes with at least three historical breaks
for which failure probability is greater than
0.5 within the planning horizon are candidates
for replacement. Two hydraulic reliability
indices, i.e., proportion of peak demand the
network can supply when a pipe fails, and the
proportion of the nodes the network can feed
with a pre-defined minimum pressure.
Given break predictions (e.g., with I-WARP),
perform medium-term replacement planning
while considering economies of scale, and
adjacent infrastructure.
N/A = information not available.
Table 6-1 provides approximately 29 models found in the literature for decision support of pipe renewal.
As in Section 5, this compilation was not intended to provide practitioners with sufficient information to
decide which of the models is most suitable for their own circumstances; such a decision would require a
level of detail that is beyond the scope of this report. Some of these models are simple enough for a
competent engineer to implement in a spreadsheet environment, while others require expertise and
resources that most water utilities do not have.
As for the deterioration models, it may be safely assumed that the likelihood of a model to be used in
practice increases significantly if this model is implemented in a software program that is publicly
available. In Section 6.2, a list of publicly available software programs is provided, some of which are
based on models described in Table 6-1. Consequently, these models are likely to have been (or to be)
used in practice, while others are likely to have been used sporadically at best.
6.2 Publicly Available Decision Support Software Tools
The available decision support (DS) software tools are described below in alphabetical order.
6.2.1 Computer Aided Rehabilitation of Water Networks (CARE-W). CARE-W was a
European Union-sponsored collaborative research effort (under the fifth Framework Programme of the
European Commission 2001 to 2004) intended to improve decision support tools in water supply systems.
The outcome of this project was a decision support toolbox carrying the same name. As of August 2009,
the software package is not commercially available, nor is it available for the public at large. It is
available for consultancy services through its principal developers and it may also be selectively available
for research (Saegrov, 2009).
CARE-W is a toolbox software package. It contains several independent decision support tools
(developed by several participating researchers) that are connected to the database module, which is the
only common link between them. Some of the tools described below, in particular CARE-W FAIL and
CARE-W LTP, have been further advanced in recent years by Cemagref of France and Baur and Kropp of
108
-------
Germany. CARE-W has been used for rehabilitation planning in several cities worldwide, including Las
Vegas, U.S., Lyon, France and Oslo, Norway (Saegrov, 2010).
The tools included are:
• CARE-W PI (Performance Indicator): used to estimate the current and future condition of
water network against a range of key performance indicators. It is based on the International
Water Association (IWA) Pis. There are a total of 49 Pis in five groups, including
operational, quality of service, financial, water resources, and physical indicators. There is an
allowance for additional Pis that are more difficult to define and quantify, such as network
reliability, remaining life, and others. It is noted that 153 single pieces of utility information
are required to assess the 49 Pis. In addition, 29 external indicators, not under utility control,
such as climate, soil, and topography, are considered in the evaluation (Batista and Alegre,
2002).
• CARE-W FAIL comprises five different models/tools to forecast pipe failure: (a) Failnet-
Stat, developed at Cemagref (France) based on the proportional hazards model (Table 5-2);
(b) Winroc, developed at NTNU (Norway), based on the non-homogeneous Poisson process
(Table 5-2); (c) AssetMapl, developed at INSA-Lyon (France) based on the Markov chain
(Table 5-2); (d) AssetMap2, developed at INSA-Lyon, which uses a technique called
"Poisson regression" to help users identify cohorts that are significantly different from one
another; and (e) Utilnets, developed at SINTEF (Norway) (Table 5-1) (Eisenbeis et al., 2002).
• CARE-W REL (Reliability) comprises three different models to compute network reliability:
(a) AquaRel, developed by SINTEF, Department of Water and Wastewater, Norway, couples
a hydraulic simulator (EPANET) and pipe failure rate to quantify the impact of pipe
condition on the reliability of the network. This impact is measured as the number of nodes
that suffer critical pressure reduction due to pipe failure; (b) FailNet-Reliab, developed by the
Hydraulic and Civil Engineering unit of Cemagref in France, also couples a hydraulic
simulator and pipe failure rate to quantify the impact of pipe condition on the reliability of the
network. However, this impact is measured as the shortfall of supply versus demand flows
due to the failure of one or more pipes; and (c) RelNet, developed at the Brno University of
Technology, Czech Republic, also couples a hydraulic simulator (ODULA) and pipe failure
rate; however, it does so probabilistically using Monte-Carlo simulations. Subsequently, a
probability distribution of entering into a deficient pressure state is computed for every node
in the network. These distributions are converted to nodal reliabilities, which can be
aggregated to provide network reliability (Eisenbeis et al., 2002).
• CARE-W ARP (Annual Rehabilitation Project) is a multi-objective decision support tool to
prioritize water mains for renewal based on CARE-W analysis tools, as well as on any
additional relevant information that is available to the user. A more detailed description is
provided in Table 6-1.
• CARE-W LTP (Long Term Planning) comprises three tools (Scenario Writer, Rehabilitation
Strategy Manager and Rehabilitation Strategy Evaluator), all developed at the Technical
University of Dresden. The Scenario Writer is a tool intended for the development of
consistent scenarios. This form of analyses is essential for a fair and robust comparison of
scenarios, which require many assumptions about the future. The Rehabilitation Strategy
Manager is largely based on the KANEW model (described below) and is intended to
simulate the long-term effects of specific rehabilitation options. The Rehabilitation Strategy
Manager is used to identify the best long-term rehabilitation strategy, using techniques based
on the "Formalized Weighting and Ranking Procedure" (Rostum et al., 2004).
109
-------
6.2.2 KANEW. KANEW is a cohort survival model for infrastructure assets that was developed
by Professor Herz while at Karlsruhe University in Germany. Later, after joining the Technical
University of Dresden (Germany), Professor Herz and his students continued to evolve the initial model
both independently as well as within the framework of CARE-W. In 1995-1998, the Roy Weston
Research Group, in collaboration with Professor Herz, developed a computer application based on the
model and applied it to several water utilities in North America (Deb et al., 1998). This project was
sponsored by the American Waterworks Association Research Foundation, which is now called the
Water Research Foundation (WRF). This early and limited software version is available in Microsoft®
Access format to Foundation subscribers. KANEW is based on a probability distribution proposed by
Professor Herz (1996). This probability distribution is fitted to (i.e., its parameters are discerned for) a
cohort of pipes (i.e., a relatively homogeneous group of pipes with the same age, same material, diameter,
etc.). Initially, this fitting was done based on the water utility's historical practices of pipe replacement,
or alternatively, based on expert opinion as to the proportion of pipes (or quantiles) expected to survive to
various age levels. Over the years the fitting techniques evolved to rely more on actual failure data rather
than perception. Since this is a three-parameter probability distribution, the knowledge (or guess) of three
quantiles (typically 10th percentile, median and 90th percentile) is sufficient to compute these parameters.
Once the parameters are computed, service life prediction of the cohort can be estimated. Typically
pessimistic, most probable, and optimistic scenarios are explored for each cohort to account for
uncertainties in the accuracy of the estimated parameters.
The current version of the KANEW software includes a module to manage pipe inventory, a module to
perform the cohort survival calculations, a failure and break forecasting module, a module to perform cost
calculation, a module to support decision making by running and comparing various scenarios, an
economic data module, and a strategy comparison module. In evaluating scenarios, the software enables
the consideration of short-, medium- and long-term impacts of a given policy on the population served, on
the structural condition of the network, and costs. KANEW is currently used mainly in Germany, but
also in the U.S. and Australia for long-term renewal planning of water and gas mains.
6.2.3 Pipeline Asset and Risk Management System (FARMS). FARMS is a suite of computer
applications based on models that have been developed by Commonwealth Scientific and Industrial
Research Organization (CSIRO) of Australia. Currently, two FARMS applications are publicly available
as commercial products, PARMS-Planning and PARMS-Priority, both used by several Australian water
utilities (Marlow et al., 2007). While PARMS-Planning forecasts the number of pipe failures and
assesses cost implications of various high-level, long-term pipe renewal scenarios, PARMS-Priority
allows prioritization of individual pipes for renewal and facilitates low-level planning of pipe replacement
and some aspects of network operations.
PARMS-Planning uses two types of pipe deterioration models, a non-homogeneous Poisson based model
(Table 5-2) (Jarrett et al., 2003) for high probability, low-consequence failure of pipes, requiring reactive
renewal strategies (management of failure frequency) and probabilistic/physical models (Table 5-1)
(Davis et al., 2007; Davis et al., 2008; Moglia et al., 2008) for low-probability, high-consequence failure
of pipes requiring proactive renewal strategies (failure prevention). After calibration using the recorded
history of failures, these models are used to forecast the number of failures expected in the planning
period. The cost of failure is computed based on user-input, including cost of repair, extent of the
network that experiences interruption of service, penalties and rebates. Based on future failure frequency
and cost, as well as pipe replacement cost, candidate pipes for replacement are identified. These are
either replaced or the costs associated with their failure are reduced by installing additional isolation
valves to reduce impact on the network.
PARMS-Priority enables the prioritization of candidate pipes for replacement or alternatively for the
installation of additional pressure reducing valves and/or additional isolation valves. It focuses mainly on
110
-------
high probability, low-consequence failure of pipes, requiring reactive renewal strategies (management of
failure frequency). It uses the same non-homogeneous Poisson-based model as PARMS-Planning for
pipe deterioration. Cost calculations are also similar to PARMS-Planning except for a more refined,
probabilistic approach to the computation of failure consequence. PARMS-Priority allows the
examination of complex scenarios, involving single and pipe cluster replacements, pressure reduction and
shutoff block reduction (Moglia et al., 2006).
6.2.4 Pipe Rehabilitation Management (PiReM). PiReM is a decision support tool for the
rehabilitation management of water supply systems (TuGraz, 2006). It is based on the doctoral thesis of
Daniela Fuchs-Hansuch at the Graz Technical University (Austria), which was not available at the time
this report was prepared. The PiReM software currently consists of two modules: long-term (20 to 50
years) rehabilitation management, and medium-term (5 years) rehabilitation management. From the
general description provided in TuGraz (2006), it appears that the underlying approach used in PiReM is
similar to the KANEW method, i.e., analysis of cohorts of pipes using the Herz distribution, and the
examination of rehabilitation scenarios subject to assumptions about the deterioration characteristics of
replacement pipes. The long-term planning module of the program is said to also consider environmental
influences but an explanation of how this is done was not available. The medium-term rehabilitation
planning module ranks individual pipes for renewal based on forecasted failure rate, risk of corrosion,
obsolescence of pipe material and diameter (old types that are difficult to repair and maintain) as well as
other technical, economical, and business management criteria that are not specified (TuGraz, 2006).
6.2.5 Water Main Rehabilitation Planner (WARP). WARP comprises a set of planning tools,
developed by NRC, for effective planning of water main renewal. WARP currently comprises four
individual (non-integrated) tools: D-WARP (Distribution mains-WARP), T-WARP (Transmission mains-
WARP), Q-WARP (Water Quality-WARP), and I-WARP (Individual [mains]-WARP).
D-WARP models the deterioration of water distribution pipe cohorts (in terms of the increase of their
breakage rates) as an exponential function of age (see Table 5-2). The analysis of water main breakage
patterns takes into consideration time-dependent factors such as temperature, soil moisture and rainfall
deficit, and CP strategies, including both hot-spot and methodical retrofit CP. D-WARP allows the user
to see the "optimal" time of pipe replacement (Table 6-1), as well as to generate, examine, and compare
complex scenarios that include combinations of replacement and CP strategies. D-WARP is currently a
stand-alone program, available for free download at the NRC Web site.
T-WARP models the deterioration of large diameter water transmission mains using a so-called fuzzy
rule-based Markov deterioration process (Table 5-2). It requires that the pipe be inspected at least once to
establish its condition rating. Future deterioration is forecasted to provide failure likelihood in the future
based on past condition rating(s). The pipe owner is required to rate the consequences of pipe failure on a
fuzzy scale. Given the likelihood and consequences of failure, a fuzzy risk of failure can be computed
and a rehabilitation strategy can be formulated (Table 6-1). T-WARP is currently a software prototype,
publicly available through WRF.
Q-WARP is a tool to predict the potential occurrence of various mechanisms of water quality
deterioration that lead to water quality failures in the distribution network. Q-WARP models the complex
nature of water quality processes in the distribution network as a set of so-called fuzzy cognitive maps. It
enables the evaluation of multiple strategies (e.g., pipe renewal, cross connection control program) for
reducing the risk of water quality failures, which leads to better-informed decision making. Q-WARP is
currently a software prototype, publicly available through the WRF.
I-WARP models the deterioration of individual water distribution pipes (in terms of the increase of their
breakage rates) as a nonhomogeneous Poisson process (Table 5-2). I-WARP is different from other
111
-------
nonhomogeneous Poisson process-based models in that it allows the consideration of time-dependent
factors such as temperature, soil moisture and rainfall deficit, CP strategies, including both hot-spot and
methodical retrofit CP, as well as user defined qualitative/quantitative factors (e.g., changes in operational
conditions, leak-detection campaigns, etc.). I-WARP is currently a software prototype, publicly available
through the WRF.
6.2.6 WilCO. WilCO is a modeling approach to manage resources or assets. It was developed by
Peter Skipworth and Mark Engelhardt, who also founded the SEAMS Corporation, which currently
markets WilCO in the form of a software package and/or a service. Although the approach was later
made into a generic tool, it was originally developed specifically for water mains, with the intent of
supporting water utilities in the UK in their quest to meet regulator's (Office of Water - OFWAT)
requirements (Engelhardt and Skipworth, 2005).
The heart of WilCO is the so-called "model builder." It allows the user to define the performance of the
asset (pipes) in terms of key performance indicators - KPIs (e.g., reliability, serviceability, customer
complaints, breakage frequency, etc.) as well as the associated whole life costs. The software does not
oblige the user to use a pre-defined deterioration model, but rather allows the users (for better or for
worse) to define their own models and train (or calibrate) the models using their own data or expert
opinion. Once the objective of the asset renewal planning is defined (e.g., maximize cost effectiveness or
alternatively maximize cost benefit), WilCO employs search algorithms to find an 'optimal' solution. For
example, to maximize cost-effectiveness subjected to predefined levels of KPIs, the user must provide the
desired levels of KPIs. Alternatively, if the objective is to maximize cost-benefit, the user must provide
the appropriate relationship between each KPI and its associated benefit. The software also includes
"add-ons" that enable users to examine and compare scenarios, as well as rank, and prioritize and
schedule individual assets for renewal action (Engelhardt and Skipworth, 2005).
112
-------
7.0: TECHNOLOGY GAPS AND RESEARCH AND DEVELOPMENT NEEDS
Technology gaps in the area of condition assessment technologies for water transmission and distribution
systems can be generally classified into four types:
• A complete absence of technology or method, capable of achieving a stated objective(s).
Such objectives could include effective detection of distress indicators, accurate interpretation
of distress indicators to condition rating, effective forecasting of deterioration, decision
optimization, etc.
• A technology exists, but is too costly to apply for the required objective.
• Promising technology exists in other domains, but development work is required for
adaptation to the domain at hand.
• Technology exists that is potentially useful, affordable, and valuable, but these attributes have
not been adequately demonstrated, documented, and justified so that utilities are convinced
that the investment in the technology will be worthwhile in the long run.
In the scope of water main NDE and condition assessment, many of the gaps are of the second and third
types, but there are also gaps of the first and fourth types. The following is a list of gaps identified in the
course of preparing this report. Although beyond the scope of this project, it would be a useful effort to
rank the gaps and future research identified below by the impact and value to the water community. This
could be accomplished through collaboration with the Water Research Foundation (WaterRF), water
utilities, and other key stakeholders in order to determine the most pressing research needs and future
technology investment efforts.
NDE technologies:
• NDE of Small Diameter (<12 in.) Metallic (CI and DI) Distribution Pipes. CI and DI are
the predominant pipe materials in distribution networks in North America and in most of
Europe, Australia, South Korea and Japan. As alluded to in Section 2, elaborate inspection
and condition assessment is economically justified only when it costs less than letting the
pipe fail. Failure consequences of small distribution mains are currently low compared to the
cost of most inspection technologies. Therefore, there is an urgent need to develop new, low
cost NDE technologies for small diameter CI and DI pipes. These technologies will need to
be reliable with operational costs low enough to justify wide usage for pipes that are
relatively inexpensive to replace and whose failure consequences are relatively low.
Existing technologies suitable for small diameter CI and DI pipes include RFEC, MFL, and
ultrasound. Hydroscope, which was based on the RFEC technology, was available
commercially in the late 1990s and early 2000s. Hydroscope could be launched into pipes
through fire hydrants, but required pre-cleaning of tuberculated pipes. Further, it appears that
it did not gain wide acceptance (probably due to its high cost). Recent developments indicate
that the See Snake Tool (described in Section 3), also based on RFEC technology, has
superseded Hydroscope. The See Snake Tool requires a dedicated launching chamber, but
appears to be less restrictive on pipe pre-cleaning requirements. In-line MFL has size
limitations (not suitable for small pipes) and external MFL requires costly excavation of
pipes. The ultrasonic-based Super-pig reportedly achieves good results, but is currently a
prototype (suitable for 10 to 12 in. diameter pipes only) that is not publicly available and the
costs involved in its acquisition and operation are not yet known. Non-intrusive technologies,
113
-------
such as WallThicknessFinder (Section 3), capable of evaluating a relatively long section of
pipe between two points, is a promising prospect, but is currently limited in providing the
average condition of a pipe section (cannot locate corrosion pits) and, as such, is limited to
function as a screening tool to identify candidates for intrusive (and expensive) inspection.
• NDE of Large Diameter (>16 in.) Metallic (CI and DI) Transmission Mains. Although
corrosion is still a dominant mode of failure in these pipes, other important modes of failure
include pipe rotation and cracks introduced in the pipe during transportation, installation and
lead/leadite caulking (Section 2). These cracks typically occur near the joint, either at the bell
or (less frequently) at the spigot end. Many of these cracks eventually develop into failure
due to fatigue. There is a need for NDE technology capable of identifying such cracks, as
well as joint rotation, preferably from the inside of the pipe to minimize pipe excavation with
all its associated costs and disruptions. Early knowledge about the presence of such cracks
and the prevention of excessive joint rotation could potentially result in tremendous savings
of losses due to catastrophic failures of these large mains. Several innovative inspections
technologies for CI pipes were tested under a separate task of this EPA TO 62 project to
inspect a 2,000-ft long, 24-in. cast iron transmission main at Louisville, KY (in progress).
This included a prototype PipeDiver™ and a custom See Snake Tool developed for a 24-in.
diameter cast iron pipe, as well as several leak detection and acoustic pipe wall inspection
technologies.
• NDE of PCCP. Current technologies for PCCP inspection are based on magnetic techniques
(RFEC and its derivatives, see Section 3.4.2) and sound emissions. Magnetic techniques are
capable of detecting discontinuities in the prestressing wire (i.e., wire breaks) but they cannot
detect deteriorated (corroded) wires where breakage is imminent. Additionally, they cannot
detect the presence of hydrogen embrittlement which, if present, can cause a sudden failure of
wire(s). NDE technologies capable of detecting these phenomena, or perhaps inferring other
signs of wire deterioration (e.g., delamination, concrete deterioration), would be beneficial
for the early detection of PCCP pipe failures. The AE techniques (hydrophones, fiber optics)
endeavor to capture the sound that a prestressing wire creates when it snaps. Properly placed
and spaced sensors will detect wire snaps, but are not capable of quantifying damage that
occurred prior to monitoring.
• NDE of AC pipes. There currently are no known NDE technologies for AC pipes. Early
trials by Echologics to test an acoustic-based method for the evaluation of the remaining wall
thickness of AC pipes (see Section 3) has shown some promise, but rigorous testing has yet to
be done (Bracken, 2009).
• NDE of PVC and PE pipes. The predominant failure mode of PVC pipes is associated with
scratches, voids and inclusions. NDE technology to detect these factors in buried water
mains does not yet exist. Further, although laser- and sonar- based techniques exist for the
detection of out-of-roundness in sewers, equivalent techniques have not been developed for
in-service plastic water mains. This out-of-roundness deformation is a useful indicator of
distress in plastic pipes.
114
-------
• NDE of Large Diameter Transmission Mains. Large transmission mains are typically the
backbone of water distribution systems. Based on input from water utilities, the sizes of their
water transmission mains range from 12-in., 16-in., 18-in., 20-in., and up in diameter.
Although there is no formal cutoff point between small distribution and large transmission
mains, larger than 16-in. diameter would be a typical cut-off value for mid- to large size
water utilities. Due to the high costs of large diameter water mains, networks often do not
have the redundancy required to take them offline. Consequently, water utilities are reluctant
to perform inspections, which would require pipe dewatering. Therefore, there is an urgent
need to develop technologies capable of performing NDE for in-service pipes. This requires
the development of sensors, as well as robotic platforms, to introduce these sensors into the
pipe. This also requires the development of launching and retrieval chambers that are not
prohibitively expensive.
• NDE technologies to estimate the loss of pipe bedding and support do not currently
exist. Deterioration of pipe bedding and surrounding backfill is an important indication of
distress leading to failure, especially for thermoplastic pipes since its strength is augmented
by soil support. In CI pipes, loss of bedding may lead to joint rotation and eventual failure.
GPR has been tried to detect voids around pipes but this application has not yet matured.
• The reliability of many of the available NDE technologies for buried water mains is not
known. There is a need to establish protocols for standard tests and ratings that would
address issues such as probability of detection (PoD), rate of false positives, false negatives,
etc. This will enable users to select appropriate technologies, with a robust understanding of
advantages and limitations under different conditions.
• There is a need for detailed protocols of forensic analyses of the failure of all pipe
materials, with special focus on pipes whose failure would be associated with high
consequences. This will enhance the understanding of failure modes and their associated
telltale signs and potentially lead to the development of improved NDE technologies capable
of detecting these signs. Adopting such practices will necessitate appropriate training of staff
and possibly a wide access depository of information and results.
Condition rating:
There is a limit to the accuracy in which deterioration models (both physical and
empirical) are able to predict failure. This limit can be overcome if these models were
combined with reliable data on the current condition of the pipe. Consequently, there is a
need to develop methods that are capable of fusing sensory data and historical performance
records/states. The multi-source data will enhance the reliability of any prediction effort.
Buried pipes have a useful life spanning many decades. Pipe inspection technologies
probably evolve and change during the life of a pipe. In large diameter mains, where
distress indicators are eventually transformed into an ordinal condition rating, historical
condition rating data that deterioration models use would then require appropriate updating or
normalization to account for the different types of technologies used to discern these distress
indicators during the pipe's lifetime. The intent of this updating or normalization would be to
make the (ordinal) condition ratings independent of the technology used to produce this
condition rating.
115
-------
Failure risk:
• More efforts should be invested in formalizing and automating the quantification of the
consequences of a failure.
• There are insufficient historical data to validate and refine the existing deterioration
models of large diameter pipes. Research efforts should be directed more towards
validation, calibration, and refinement of existing models (using real field data) than
developing new models.
Decision support:
• The ultimate goal of decision making is to provide service at stated levels (where levels
are defined for reliability, pressure, water quality, environmental impact, etc.) at the
lowest life-cycle cost (often with budget constraints). While the state of the art is still far
from formulating an all-encompassing model, achievable interim goals should include the
consideration of structural condition, hydraulic reliability, and impact of pipes on the water
quality in the network, while on the cost side, decision making should consider economies of
scale and interaction with adjacent infrastructure.
• Additional outreach is needed to assist water utilities to better understand where the
many available inspections tools and models fit within their investment decision making
process. Further outreach could help utilities to belter understand the following:
• The key inspection data required to help improve deterioration modeling, risk
assessment, and decision support tool outputs;
• The processes, tools, and costs to provide this data at the various levels for both
distribution and transmission mains; and
• What tools require development to fulfill any gaps in these data requirements (given
that the gaps identified above are primarily related to technology requirements rather
than utility data requirements)
Condition assessment data ultimately lays the foundation for decision making regarding repair,
rehabilitation, or replacement of deteriorated water mains. Currently, this decision making is based
largely on performance factors such as main break frequency or severity, water quality problems, or poor
hydraulic characteristics. As the state of the art in inspection technologies improves, this will also
improve the ability to incorporate valuable data on the host pipe's structural condition into the selection of
appropriate renovation techniques. The decision making steps involved in the selection of repair,
rehabilitation, and replacement techniques is beyond the scope of this research project and is explored in
other EPA research (Matthews et al., 2011).
116
-------
8.0: SUMMARY AND CONCLUDING REMARKS
As water mains age, they are increasingly exposed to continuous stress from operational and
environmental conditions. These mains deteriorate structurally and hydraulically, adversely impacting
water quality, leakage, and reliability. Effective management of these assets requires condition
assessment, which includes the collection of information about their condition, analysis of this
information, and ultimately transformation of this information into knowledge, leading to effective
decision about renewal. In the introduction, several key issues were identified for the assessment of the
structural condition of water mains and decision making on pipe renewal. The following summarizes the
manner with which these key issues have been addressed in this report:
(1) Physical modeling of the pipe in the soil.
This issue was addressed briefly in Section 2 through the description of the physical
manifestation of pipe performance in the soil. Also, brief descriptions were provided in
Section 5 of physical/mechanistic models found in the literature for the performance and
deterioration of buried pipes.
(2) Understanding of pipe failure modes, including observable or measurable signs (or distress
indicators) that point to these modes, as well as inferential indicators that point to potential
existence of deterioration mechanisms.
In Section 2, an overview of pipe deterioration mechanisms was provided with
comprehensive lists of how these mechanisms manifest themselves in different pipe
materials. Section 5 provided brief descriptions of physical/mechanistic models to describe
pipe deterioration in the ground. The list comprises a total of 17 such models from the
literature and is believed to be quite comprehensive, if not exhaustive.
(3) Inspection of the pipe to discern distress indicators.
Section 3 provided descriptions of approximately 70 technologies/techniques/methods for
inspection and evaluation of distress indicators in pipes. These include visual,
electromagnetic, ultrasonic, and laser-based technologies, leak detection, direct distress
indicators (indicators observed and measured on the pipe itself) and inferential indicators
(soil and environmental properties). Emerging sensor technologies with potential application
in the water supply industry, along with sensor networks were also reviewed.
(4) Interpretation of distress indicators to determine pipe condition.
Section 4 described a range of methods/approaches used to interpret distress indicators into
condition ratings, including point score, fuzzy-based techniques, data fusion, data mining and
data-driven approaches.
(5) Empirical/statistical modeling of historical failures (mainly in small diameter distribution
mains).
Section 5 summarizes statistical/empirical models that appeared in the literature in the last 30
years. Both deterministic and probabilistic models are included. In most models that address
small diameter distribution mains, deterioration is defined as the increase in breakage
frequency. In contrast, models addressing large diameter transmission mains define
deterioration in terms of condition ratings. This difference is inherent in the manner with
which these two classes of assets are managed (i.e., manage break frequency vs. failure
prevention).
117
-------
(6) Modeling deterioration to forecast future failure rates and pipe residual life.
Section 5 provided descriptions of 17 physically based deterioration models and 31
statistical/empirical deterioration models for water mains that have been proposed in the
literature over the years. These two lists are believed to be very comprehensive.
(7) Assessment of failure consequences (direct, indirect and social costs).
This element was not addressed in this report.
(8) Scheduling pipe renewal so as to minimize life-cycle costs while meeting or exceeding
functional objectives of water distribution (quantity, quality, reliability, etc.)
Section 6 provided descriptions of 29 decision support models that have been proposed in the
literature over the past few years. In addition, detailed descriptions were provided for
decision support software tools that are publicly available, either in a commercial or research
version.
Section 7 provided a list of identified technology gaps and research and development needs, addressing
aspects of NDE technologies, condition rating techniques and decision-making techniques including risk-
based techniques requiring the quantification of failure consequences.
This report reflects a substantial amount of work and effort that has been invested in developing
approaches and tools for the condition assessment of water mains. There are currently a number of
technologies that are commercially available for leak detection and structural integrity monitoring of
water mains. In particular, the development of inspection technologies for large diameter PCCP has been
a success story where the cost of gathering the inspection data was superseded by the benefit in improved
data that it provided on the pipe condition, which allowed detailed engineering analysis to determine if
pipes required repair or replacement. Because of the high consequence of large diameter PCCP failures,
the cost of inspection was readily justified and allowed utilities to make more informed and proactive
decisions on whether or not to renew a given PCCP segment based upon its likelihood of failure as
identified from inspection. In addition, the use of leak detection technologies is growing as water utilities
focus their efforts on reducing water losses in order to maintain or increase their revenue, conserve water
resources, and reduce public health risks (EPA, 2009).
Any asset management program must start with a thorough review of available historical data about pipe
performance and failure. Once the necessary data is gathered, deterioration models (some of which are
quite affordable) can go a long way in providing insight into the condition of these assets, especially for
small diameter pipes. A well-defined and cost-effective inspection program that complements the historic
data can then be used to fill in gaps that remain and/or to validate the results of modeling efforts for the
specific conditions faced by a water utility.
Currently, the relatively high cost of various NDE technologies justifies their use mainly on large water
transmission mains, where the consequences of failure are relatively high. However, it is foreseen that as
novel technologies develop and competition intensifies, prices will decline and NDE inspection will
become justified even for pipes with relatively moderate consequences of failure. This will result in
higher uptake rate, which in turn will drive unit prices down.
Further research and development by key stakeholders including the federal government, non-profit
research organizations, and industry could aid in the acceleration of this process. As described in Section
7, there are a number of technology gaps and research needs including: the need for live internal insertion
and retrieval of inspection tools for large diameter pipes; the need to assess joint condition in metallic
pipes; the need to develop technologies for asbestos cement and plastic pipes with few options currently
118
-------
available; and the need for low cost inspection methods to conduct screening for high risk locations in all
pipe types for further assessment. To overcome the barriers and challenges identified in Section 7, field
demonstrations and further research efforts are warranted in order to test promising technologies that
could fill these gaps against well defined performance criteria and to identify the critical performance,
cost, and/or value added attributes of emerging and innovative technologies for water main inspection.
119
-------
9.0: REFERENCES
SDCorrosion. (2005). http://www.3Dcorrosion.com.
Abraham, D.M., and Wirahadikusumah, R. (1999). "Development of Prediction Model for Sewer
Deterioration." 8th Conference Durability of Building Materials and Components, M.A. Lacasse
and D.J. Vanier, eds., Vancouver, BC, Canada.
Achim, D., Ghotb, D., and McManus, K.J. (2007). "Prediction of Water Pipe Asset Life Using Neural
Networks." Journal of Infrastructure Systems, 13(1), 26-30.
Acuity. (2008). "Principles of Measurement Used by Laser Sensor."
Ahammed, M., and Melchers, R.E. (1994). "Reliability of Underground Pipelines Sugjectedto
Corrosion." ASCE Journal of Transportation Engineering, 120(6), 989-1003.
Al-Barqawi, H., and Zayed, T. (2006). "Condition Rating Model for Underground Infrastructure
Sustainable Water Mains." Journal of Performance of Constructed Facilities, 20(2), 126-135.
Allouche, E.N. (2007). "Computer Aided Tomography: Inspecting Buried Structures from Inside Out."
7th Annual Tulane Engineering Forum, New Orleans, USA.
Alvisi, S., and Franchini, M. (2006a). "Near-optimal Rehabilitation Scheduling of Water Distribution
Systems Based on a Multi-objective Genetic Algorithm." Civil Engineering and Environmental
Systems, 23(3), 143-160.
Alvisi, S., and Franchini, M. (2006b). "Rehabilitation, Repairing and Leakage Detection Optimization in
Water Distribution Systems." 8th Annual Water Distribution Systems Analysis Symposium,
Cincinnati, Ohio, USA.
Andreou, S.A., Marks, D.H., and Clark, R.M. (1987a). "A New Methodology for Modeling Break Failure
Patterns in Deteriorating Water Distribution Systems: Theory." Advance in Water Resources, 10,
2-10.
Andreou, S.A., Marks, D.H., and Clark, R.M. (1987b). "A New Methodology for Modelling Break
Failure Patterns in Deteriorating Water Distribution Systems: Applications." Advance in Water
Resources, 10, 11-20.
Anna, D.J. (2004). "Prototype Pipeline Sensor Passes Field Test." Materials Performance, NACE
International, 33-34.
Ariaratnam, S.T., El-Assaly, A., and Yang, Y. (1999). "Sewer Infrastructure Assessment Using Logistic
Statistical Models." Annual Conference of the Canadian Society for Civil Engineering, Regina,
Canada, 330-338.
Ariaratnam, S.T., El-Assaly, A., and Yang, Y. (2001). "Assessment of Infrastructure Needs Using
Logistic Models." ASCE Journal of Infrastructure Systems, 7(4), 160-165.
AWWA. (1999). "American National Standard for Polyethylene Encasement for Ductile-Iron Pipe
Systems." American Waterworks Association.
Babovic, V., Drecourt, J., Keijzer, M., and Hansen, P.P. (2002). "A Data Mining Approach to Modelling
of Water Supply Assets." Urban Water, 4, 401-414.
Bai, H., Sadiq, R., Najjaran, H., and Rajani, B. (2008). "Condition Assessment of Buried Pipes Using
Hierarchical Evidential Reasoning Model." Journal of Computing in Civil Engineering, 22(2),
114-122.
120
-------
Bartels, K.A., Dynes, C., Lu, Y., and Kwun, H. (1999). "Evaluation of Concrete Reinforcements Using
Magnetostrictive Sensors." SPIE Conference on Nondestructive Evaluation of Bridges and
Highways III, S.B. Chase, ed., SPIE, Newport Beach, CA, USA, 210-218.
Batista, J., and Alegre, H. (2002). "CARE-W WP1 D2 - Validation of the Rehabilitation Performance
Indicators System." National Civil Engineering Laboratory, Lisbon, Portugal.
Bell, G.E.C., and Moore, C.G. (2007). "Development and Application of Ductile Iron Pipe Electrical
Resistance Probes for Monitoring Underground External Pipeline Corrosion." Recent
Development and Field Experience in Corrosion and Erosion Monitoring, NACE International, 1-
14.
Beller, M., and Barbian, A. (2006). "Combined In-Line Inspection of Pipelines for Metal Loss and
Cracks." European Conference on Nondestructive Testing, Berlin, Germany, 1-13.
Berardi, L., Giustolisi, O., Kapelan, Z., and Savic, D.A. (2008). "Development of Pipe Deterioration
Models for Water Distribution Systems Using EPR." Journal ofHydroinformatics, 10(2), 113-
126.
Berardi, L., Giustolisi, O., and Primativo, F. (2007). "Exploiting Multi-objective Strategies for Optimal
Rehabilitation Planning." Water Management Challenges in Global Change, B. Ulanicki, K.
Vairavamoorthy, D. Butler, P.L.M. Bounds, and F.A. Memon, eds., Talyor & Francis, De
Montfort Leicester, UK, 23-30.
Bhagwan, J.N. (2009). "Compendium of Best Practices in Water Infrastructure Asset Management."
Global Water Research Coalition, London, U.K.
Birmingham Water Works Board (BWWB). (2009). "BWWB Recieves International Recognition for its
Use of New Technology to Reduce Non-Revenue Water." Available at:
www.birminghamwaterworks.com/index.php?option=com_content&task=view&id=218&Itemid
—ss
—oo.
Bosch, J., Hugger, A., Franz, J., Falter, S., and Oberdorfer, Y. (2004). "Phase Array Technology for
Automated Pipeline Inspection." Business Briefing: Exploration & Production: The Oil & Gas
Review, 1-4.
Bracken, M. (2009). Personal Communication.
Bremond, B. (1997). "Statistical Modeling as Help in Network Renewal Decision." European
Commission Co-operation on Science and Technology (COST), Committee C3 - Diagnostics of
Urban Infrastructure, Paris, France.
Burn, L.S., Tucker, S.N., Rahilly, M., Davis, P., Jarrett, R, and Po, M. (2003). "Asset Planning for Water
Reticulation Systems - the FARMS Model." Water Science and Technology: Water Supply, 3,55-
62.
Burn, L.S., Davis, P., Schiller, T, Tiganis, B., Tjandraatmadja, G., Cardy, M., Gould, S., Sadler, P., and
Whittle, A.J.. (2005). "Long-Term Performance Prediction for Polyvinyl Chloride Pipe." AWWA
Research Foundation, Denver, CO.
Butler, D. (2009). "Leak Detection and Management: A Comprehensive Guide to Technology and
Practice in the WaterSupply Industry." Palmer Environmental, Cwmbran, UK. Available at:
http://aseiran.com/media/pdf/Leakage%20Detection%20and%20Management%20Book.pdf
Cabrera, E., Pardo, M.A.,E., Carbrera, J., and Cobacho, R. (2007). "Optimal Scheduling of Pipe
Replacement, Including Opportunity, Social, and Environmental Costs." ASCE International
Conference on Pipeline Engineering and Construction, Boston, Massachusetts, USA.
121
-------
Cauchi, S., Cherpillod, T., Morison, D., and McClarty, E. (2007). "Fiber-Optic Sensors for Monitoring
Pipe Bending due to Ground Movement." Pipeline & Gas Journal, January, 36-40.
Chen, X., and Ding, T. (2007). "Flexible Eddy Current Sensor Array for Proximity Sensing." Sensors and
Actuators, 135, 126-130.
Clark, R.M., Stafford, C.L., and Goodrich, J.A. (1982). "Water Distribution Systems: A Spatial and Cost
Evaluation." ASCE Journal of Water Resources Planning and Management Division, 108(3),
243-256.
Clay, M. (2009). "Super-pig for Cast Iron Pipe Inspection." personal communication.
Constantine, A.G., and Darroch, J.N. (1993). Pipeline Reliability: Stochastic Models in Engineering
Technology and Management. S. Osaki and D.N.P. Murthy, eds., World Scientific Publishing Co.
Constantine, A.G., Darroch, J.N., and Miller, R. (1996). "Predicting Underground Pipe Failure."
Australian Water Works Association.
Cooper, N.R., Blakey, G., Sherwin, C., Ta, T., Whiter, J.T., and Woodward, C.A. (2000). "The Use of
GIS to Develop a Probability-based Trunk Mains Burst Risk Model." Urban Water, 2, 97-103.
Costello, S.B., Chapman, D.N., Rogers, C.D.F., and Metje, N. (2007). "Underground Asset Location and
Condition Assessment Technologies." Tunnelling and Underground Space Technology, 22, 524-
542.
Creaform. (2008). http://www.creaform3d.com.
Crouch, A.E., and Goyen, T.H. (2003). "Defect Assessment Using Conformable Array Data." Southwest
Research Institute, San Antonio, USA.
Grouse, B. (2009). "NOT Resource Center." Iowa State University.
CUES. (2008). "Sonar Profiler System for Submerged and Semi-submerged Pipes."
Dandy, G.C., and Engelhardt, M. (2001). "Optimal Scheduling of Water Pipe Replacement Using Genetic
Algorithm." Journal of Water Resources Planning and Management, 127(4), 214-223.
Dandy, G.C., and Engelhardt, M.O. (2006). "Multi-Objective Trade-Offs between Cost and Reliability in
the Replacement of Water Mains." Journal of Water Resources Planning and Management,
132(2), 79-88.
Davies, J.P., Clarke, B.A., Whiter, J.T., Cuningham, R.J., and Leidi, A. (2001). "The Structural Condition
of Rigid Sewer Pipes: A Statistical Investigation." Urban Water, 3, 277-286.
Davis, P., Burn, S., Moglia, M., and Gould, S. (2007). "A Physical Probabilistic Model to Predict Failure
Rates in Buried PVC Pipelines." Reliability Engineering & System Safety, 92, 1258-1266.
Davis, P., Silva, D.D., Marlow, D., Moglia, M., Gould, S., and Burn, S. (2008). "Failure Prediction and
Optimal Scheduling of Replacements in Asbestos Cement Water Pipes." Journal of Water Supply
Research and Technology, 57(4), 237-252.
Deb, A., Grablutz, F., Hasit, Y.J., Snyder, J., Loganathan, G.V., and Agbenowsi, N. (2002). "Prioritizing
Water Main Replacement and Rehabilitation." AWWA Research Foundation, Denver, CO, USA.
Deb, A., Snyder, J.K., Chelius, J.J., Urie, J., and O'Day, O.K. (1990). "Assessment of Existing and
Developing Water Main Rehabilitation Practices." AWWA Research Foundation, Denver, CO,
USA.
Deb, A.K., Herz, R.K., Hasit, Y.J., and Grablutz, P.M. (1998). "Quantifying Future Rehabilitation and
Replacement Needs for Water Mains." AWWA Research Foundation, Denver, CO, USA.
122
-------
Dettmer, A. (2007). "Pipe Profiling Innovation at The Trenchless Technology Center." Underground
Construction, 42-43.
Dingus, M., Haven, J., and Austin, R. (2002). "Nondestructive, Noninvasive Assessment of Underground
Pipelines." American Water Works Association Research Foundation, Denver, CO, USA.
Doleac, M.L. (1979). "Time-to Failure Analysis of Cast Iron Water Mains." CH2M HILL, Vancouver,
BC, Canada.
Doleac, M.L., Lackey, S.L., and Bratton, G.N. (1980). "Prediction of Time-to Failure for Buried Cast Iron
Pipe." AWWA Annual Conference, Denver, CO, USA, 28-31.
Dorn, R., Houseman, P., Hyde, R. A., and Jarvis, M. G. (1996). " Water Mains: Guidance on Assessment
of Inspection Techniques." Construction Industry Research and Information Association, London,
UK.
Dridi, L., Mailhot, A., Parizeau, M., and Villeneuve, J.P. (2005). "A Strategy for Optimal Replacement of
Water Pipes Integrating Structual and Hydraulic Indicators Based on a Statistical Water Pipe
Break Model" 8th International Conference on Computing and Control for the Water Industry,
University of Exeter, UK, 65-70.
Dridi, L., Parizeau, M., Mailhot, A., and Villeneuve, J.P. (2008). "Using Evolutionary Optimization
Techniques for Scheduling Water Pipe Renewal Considering a Short Planning Horizon."
Computer-Aided Civil and Infrastructure Engineering, 23, 625-635.
Duran, O., Althoefer, K., and Seneviratne, L.D. (2003). "Pipe Inspection Using a Laser-based Transducer
and Automated Analysis Techniques." IEEE/ASME Transactions onMechatronics, 8(3), 401-
409.
Echologics. (2009). "PC Based Leak Noise Correlator."
Economou, T., Kapelan, Z., and Bailey, T. (2008). "A Zero-Inflated Bayesian Model for the Prediction of
Water Pipe Bursts." 10th International Water Distribution System Analysis Conference, Kruger
National Park, South Africa.
Eisenbeis, P. (1994). "Modelisation statistique de la prevision des defaillances sur les conduites d'eau
potable," University Louis Pasteur of Strasbourg.
Eisenbeis, P., Gat, Y.L., Laffrechine, K., Gauffre, P.L., Konig, A., Rostum, J., Tuhovcak, L., and
Valkovic, P. (2002). "CARE-W WP2 D2 - Description and Validation of Technical Tools."
National Civil Engineering Laboratory, Lisbon, Portugal.
Eisenbeis, P., Rostum, J., and Gat, Y.L. (1999). "Statistical Models for Assessing the Technical State of
Water Networks - Some European Experiences." AWWA Annual Conference, Chicago, USA.
Eiswirth, M., Frey, C., Herbst, J., Jacubasch, A., Held, I., Heske, C., Hotzl, H., Kuntze, H.B., Kramp, J.,
Munser, R., and Wolf, L. (2001). "Sewer Assessment by Multi-Sensor Systems." IWA 2, World
Water Congress, Berlin, Germany.
Eiswirth, M., Heske, C., Hotzl, H., Schneider, T., and Burn, L.S. (2000). "Pipe Defect Characterisation by
Multi-Sensor Systems." 18th International Conference on NO-DIG Perth, Australia.
Engelhardt, M.O., and Skipworth, P.J. (2005). "WiLCO - State of the Art Decision Support." 8th
International Conference on Computing and Control for the Water Industry, Exeter, UK, 27-32.
Envirosight. (2010). http://www.envirosight.com/.
EPA. (2007). Innovation and Re search for Water Infrastructure for the 21st Century: Research Plan,
U.S. Environmental Protection Agency, National Risk Management Research Laboratory,
Cincinnati, Ohio, USA.
123
-------
EPA. (2009). "Control and Mitigation of Drinking Water Losses in Distribution Systems." EPA 816-D-
09-001. U.S. Environmental Protection Agency, Drinking Water Protection Division, Office of
Ground Water and Drinking Water, Washington, D.C., USA.
FBS. (2011). "In-Situ Imaging of Water Pipelines Using Ultrasonic Guided Waves." Available at
http://cfpub.epa.gov/ncer_abstracts/index.cfm/fuseaction/display.abstractDetail/abstract/9472/rep
ort/0
PCS. (2009). "Permalog 3: The Versatile Acoustic Leak Noise Logger."
PCS. (2011). www.fluidconservation.com/permalog+AMR.htm
Feeney, C.S., Thayer, S., Bonomo, M., and Martel, K. (2009). "State of Technology Review Report on
Condition Assessment of Wastewater Collection Systems." EPA/600/R-09/049, U.S.
Environmental Protection Agency, National Risk Management Research Laboratory, Cincinnati,
Ohio, USA.
Fenner, R.A., and Sweeting, L. (1999). "A Decision Support Model for the Rehabilitation of'Non-critical'
Sewers." Water Science and Technology, 39(9), 193-200.
Ferguson, P. (2010). "Personal Communication." Y. Kleiner, ed.
Fisk, P.S., and Marshall, J. (2006). "Development of a PCCP Management Plan Using Sonic/Ultrasonic
Nondestructive Testing Results." Pipelines Division Specialty Conference, The American Society
of Civil Engineers, Chicago, Illinois, USA, 1-8.
Fletcher, R. (2008). "SmartBall - A New Approach in Pipeline Leak Detection." International Pipeline
Conference, Calgary, Alberta, Canada, 1-17.
Galbraith, J.M., Williamson, G.C., and Creech, M. (2009). "Digitial Pipeline Radiography Speeds
Detection of Corrosion and Other Anomalies." Pipeline & Gas Journal, 58-60.
Gholamzadeh, B., andNabovati, H. (2008). "Fiber Optic Sensors." Proceedings of World Academy of
Science, Engineering and Technology, 32, 303-313.
Giustolisi, O., and Berardi, L. (2007). "Pipe Level Burst Prediction Using EPR and MCS-EPR." The
Combined International Conference of Computing and Control for the Water Industry and
Sustainable Urban Water Management on Water Management Challenges in Global Change,
Leicester, UK, 39-46.
Grevea, D.W., Oppenheim, I.J., Wright, A.P., and Wua, W. (2008). "Design and Testing of a MEMS
Acoustic Emission Sensor System." SPIE Smart Structures/NDE Joint Conference, San Diego,
USA.
Grigg, N.S. (2007). "Main Break Prediction, Prevention and Control." AWWA Research Foundation,
Denver, CO, USA.
GTI. (2005). "Pipeline Assessments Through Keyholes."
Gustafson, J.M., and Clancy, D.V. (1999a). "Modeling the Occurence of Breaks in Cast Iron Water Mains
Using Methods of Survival Analysis." AWWA Annual Conference, Chicago, IL, USA.
Gustafson, J.M., and Clancy, D.V. (1999b). "Using Monte Carlo Simulation to Develop Economic
Decision Criteria for the Replacement of Cast Iron Water Mains." AWWA Annual Conference,
Chicago, IL, USA.
Hadzilacos, T., Kalles, D., Preston, N., Paul Melbourne, Camarinopoulos, L., Eimermacher, M.,
Kallidromitis, V., Frondistou-Yannas, S., and Saegrov, S. (2000). "UTILNETS: A Water Mains
Rehabilitation Decision Support System." Computers, Environment and Urban Systems, 24(3),
215-232.
124
-------
Hahn, M., Palmer, R.N., Merrill, S.M., and Lukas, A.B. (2002). "Expert System for Prioritizing the
Inspection of Sewers: Knowledge Base Formulation and Evaluation." Journal of Water
Resources Planning and Management, 128(2), 121-129.
Halhal, D., Walters, G.A., Ouazar, D., and Savic, D.A. (1997). "Water Network Rehabilitation with a
Structured Messy Genetic Algorithm." Journal of Water Resources Planning and Management,
123(3), 137-146.
Hazelden, G., Ragula, G., and Roubal, M. (2003). "The Use of Broadband Electromagnetic Technology
for Integrity Inspection of a 760 mm (30 in.) Cast Iron and Steel Line." 22nd World Gas
Conference, Tokyo, Japan, 1-6.
Herz, R.K. (1996). "Ageing Process and Rehabilitation Needs of Drinking Water Distribution Networks."
Journal of Water SRT-Aqua, 45(5), 221-231.
Herz, R.K. (1999). "Bath-tubs and Hammock-Chairs in Service Life Modelling." 13th European Junior
Scientist Workshop: Service Life Management of Water Mains and Sewers, R. Baur and R. Herz,
eds., Rathen, Germany, 11-18.
Higgins, M.S., Gadoury, P.J., LePage, P., Razza, R., Keaney, J., and Mead, I. (2007). "Technologies to
Assess and Manage of Providence Water's 102 in. PCCP Aqueduct." Pipelines 2007: Advances
and Experiences with Trenchless Pipeline Projects, 1-8.
Higgins, M.S., and Paulson, P. (2006). "Fiber Optic Sensor for Acoustic Monitoring of PCCP." Pipelines
2006 - Service to the Owner, A. Atalah and A. Tremblay, eds., Chicago, Illinois, USA.
Holley, M., and Buchanan, D. (1998). "Acoustic Monitoring of Prestressed Concrete Cylinder Pipe."
Pipelines in the Constructed Environment, J.P. Castronovo and J.A. Clark, eds., ASCE, San
Diego, California, USA, 468-476.
Hong, H.P. (1997). "Reliability Based Optimal Inspection and Maintenance for Pipeline Under
Corrosion." Civil Engineering Systems, 14, 313-334.
Hong, H.P. (1998). "Reliability Based Optimal Inspection Schedule for Corroded Pipeline." Annual
Conference of the Canadian Society for Civil Engineering, Halifax, NS, Canada, 743-752.
Hong, H.P., Allouche, E.N., and Trivedi, M. (2006). "Optimal Scheduling of Replacement and
Rehabilitation of Water Distribution Systems." ASCE Journal of Infrastructure Systems, 12(3),
184-191.
Hughes, D. (2011). "Investigating Infrastructure Integrity through Leak-Detection Technology." Utility
Products, Vol. 8, Issue 1, January.
Hunaidi, O. (2006a). "Condition Assessment of Water Pipes." EPA Workshop on Innovation and
Research for Water Infrastructure in the 21st Century, Arlington, Virginia, USA, 1-6.
Hunaidi, O. (2006b). "Non-Destructive Testing of Pipes." USPTO, ed., USA.
Hunaidi, O., and Giamou, P. (1998). "Ground-Penetrating Radar for Detection of Leaks in Buried Plastic
Water Distribution Pipes." 7th International Conference on Ground-Penetrating Radar, Lawrence,
Kansas, USA, 783-786.
Hunaidi, O., Wang, A., Bracken, M., Gambino, T., and Fricke, C. (2004). "Acoustic Methods for
Locating Leaks in Municipal Water Pipe Networks." International Conference on Water Demand
Management, Dead Sea, Jordan, 1-14.
Hydromax. (2006). "SSET Inspection: Side Scanning Evaluation Technology."
Itron. (2011). www.itron.com/na/productsAndServices/Pages/MLOG.aspx?market=water
125
-------
Jacobs, P., and Karney, B. (1994). "GIS Development with Application to Cast Iron Water Main
Breakage Rate." 2nd International Conference on Water Pipeline Systems, Edinburgh, Scotland.
Jaganathan, A., Allouche, E., and Simicevic, N. (2006). "Pipeline Scanning: Novel Technology for
Detection of Voids and Internal Defects in Non-Conductive Buried Pipes." Trenchless
International, 1-8.
Jamoussi, A. (2005). "Robotic NDE: A New Solution for In-line Pipe Inspection." 3rd Middle East
Nondestructive Testing Conference and Exhibition, Bahrain, Manama, 1-8.
Jarrett, R., Hussain, O., and Touw, J.v.d. (2003). "Reliability Assessment of Water Pipelines Using
Limited Data." 19th AWA Convention, Perth, 1-14.
Jiao, Y., Yang, Q., and Li, G. (2007). "Leak Identification Method for Buried Gas Pipeline Based on
Spatial-Temporal Data Fusion." IEEE International Conference on Control and Automation,
Guangzhou, China, 774-777.
Jolley, J., Boyd, G., and Sterling, R. (2010). "Advanced Integrity Monitoring and Assessment
Technology." U. S Environmental Protection Agency.
Juliano, T., Meegoda, J., Niver, E., and Watts, D. (2005). "A Conceptual Framework for Intelligent
Monitoring Systems." Nondestructive Detection and Measurement for Homeland Security III,
Proceedings ofSPIE, A.A. Diaz, A.E. Aktan, H.F. Wu, S.R. Doctor, and Y. Bar-Cohen, eds.,
SPIE, 162-170.
Karney, B., D. Khani, M. Halfaway, and O. Hunaidi. (2009). "A Simulation Study on Using Inverse
Transient Analysis for Leak Detection in Water Distribution Networks." NRCC-50452. National
Research Council Canada, Ottawa. March.
Kathula, V.S., and McKim, R. (1999). "Sewer Deterioration Prediction." Infra 99 International
Concention, CERIU, Montreal, QC, Canada.
Kettler, A.J., and Goulter, I.C. (1985). "An Analysis of Pipe Breakage in Urban Water Distribution
Networks." Canadian Journal of Civil Engineering, 12, 286-293.
Khan, H.A. (2007). "Using Coupons and ER Soil Corrosion Probes in Applying Pipeline CP Criteria."
Materials Performance, 26-30.
Kiefner, J.F., and Vieth, P.H. (1989). "Project PR-3-805: A Modified Criterion for Evaluating the
Remaining Strength of Corroded Pipe." Pipeline Corrosion Supervisory Committee of the
Pipeline Research Committee of the American Gas Association.
Kim, J.H., and Mays, L.W. (1994). "Optimal Rehabilitation Model for Water-Distribution Systems."
Journal of Water Resources Planning and Management, 120(5), 674-692.
Kirchner, F., and Hertzberg, J. (1997). "A Prototype Study of an Autonomous Robot Platform for
Sewerage System Maintenance." Autonomous Robots, 4, 319-331.
Kirham, R., Kearney, P.O., Rogers, K.J., and Mashford, J. (2000). "PIRAT- A System for Quantitative
Sewer Pipe Assessment." The International Journal of Robotics Research, 19(11), 1033-1053.
Kirmeyer, G.J., Richards, W., and Smith, C.D. (1994). "An Assessment of Water Distribution Systems
and Associated Research Needs." AWWA Research Foundation, Denver, CO, USA.
Kleiner, Y. (2001). "Scheduling Inspection and Renewal of Large Infrastructure Assets." Journal of
Infrastructure Systems, 7(4), 136-143.
Kleiner, Y. (2005). "Risk Approach to Examine Strategies for Extending the Residual Life of Large
Pipes." Middle East Water 2005, Manama, Bahrain.
126
-------
Kleiner, Y., Adams, B.J., and Rogers, J.S. (1998a). "Long-Term Planning Methodology for Water
Distribution System Rehabilitation." Water Resources Research, 34(8), 2039-2051.
Kleiner, Y., Adams, B.J., and Rogers, J.S. (1998b). "Selection and Scheduling of Rehabilitation
Alternatives for Water Distribution Systems." Water Resources Research, 34(8), 2053-2061.
Kleiner, Y., and Rajani, B. (2001). "Comprehensive Review of Structural Deterioration of Water Mains:
Statistical Models." Urban Water, 3(3), 131-151.
Kleiner, Y., and Rajani, B. (2004). "Quantifying Effectiveness of Cathodic Protection in Watermains:
Theory." ASCE Journal of Infrastructure Systems, 10(2), 43-51.
Kleiner, Y., and Rajani, B. (2009). "I-WARP: Individual Water Main Renewal Planner." Computing and
Control in the Water Industry 2009: Intergrating Water Systems, Sheffield, UK.
Kleiner, Y., Rajani, B., and Sadiq, R. (2005). "Risk Management of Large-Diameter Water Transmission
Mains." AWWA Research Foundation, Denver, CO, USA.
Kleiner, Y., Rajani, B., and Sadiq, R. (2006a). "Modeling Deterioration and Managing Failure Risk of
Buried Critical Infrastructure." Building Science Insight 2006 Sustainable Infrastructure:
Techniques, Tools and Guidelines, 1-13.
Kleiner, Y., Sadiq, R., and Rajani, B.B. (2006b). "Modelling the Deterioration of Buried Infrastructure as
a Fuzzy Markov Process." Journal of Water Supply Research and Technology, 55(2), 67-80.
Kobayashi, M., Jen, C.K., Bussiere, J.F., and Wu, K.-T. (2009). "High Temperature Integrated and
Flexible Ultrasonic Transducers for Non-Destructive Testing." NDT & EInternational, 42, 157-
161.
Kobayashi, M., Jen, C.K., and Levesque, D. (2006). "Flexbible Ultrasonic Transducers." IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 53(8), 1478-1486.
Kobayashi, M., Wu, K.T., Song, L., Liu, Q., and Jen, C.K. (2007). "Flexible Ultrasonic Array Sensors for
Health Monitoring." SPIE Conference on Health Monitoring of Structural and Biological
Systems, T. Kundu, ed., 1-8.
Kropp, I., and Baur, R. (2005). "Integrated Failure Forecasting Model for the Strategic Rehabilitation
Planning Process." Water Science and Technology: Water Supply, 5(2), 1-8.
Kulkarni, R.B., Golabi, K., and Chuang, J. (1986). "Analytical Techniques for Selection of Repair-or-
Replace Options for Cast Iron Gas Piping Systems - Phase I." Gas Research Instititute, PB87-
114112, Chicago, USA.
Kumar, A., Meronyk, E., and Segan, E. (1984). "Development of Concepts for Corrosion Assessment and
Evaluation of Underground Pipelines." CERL-TR-M-337, U.S. Army Corps of Engineers,
Construction Engineering Research Laboratory.
Kuntze, H.B., and Haffner, H. (1998). "Experiences with the Development of a Robot for Smart
Multisensoric Pipe Inspection." IEEE International Conference on Robotics and Automation,
Leuven, Belgium.
Kwun, H. (1991). "Back in Style: Magnetostrictive Sensors." Technology Today, Southwest Research
Institute, USA.
Kwun, H. (2000). "Technical Background on MsS." Southwest Research Institute. USA.
Kwun, H., Kim, S., and Light, G.M. (2003). "The Magnetrostrictive Sensor Technology for Long Range
Guided Wave Testing and Monitoring of Structures." Materials Evaluation, 61, 80-84.
127
-------
Lansey, K.E., Basnet, C., Mays, L.W., and Woodburn, J. (1992). "Optimal Maintenance Scheduling for
Water Distribution Systems." Civil Engineering Systems, 9, 211-226.
Las Vegas Valley Water District. (2011). www.lvvwd.com/about/facilities_transmission_leaks.html.
Lawson, S.W., and Pretlove, J.R. (1998). "Augmented Reality for Underground Pipe Inspection and
Maintenance." M.R. Stein, ed., 98-104.
LeGat, Y. (2008a). "Extending the Yule Process to Model Recurrent Failures of Pressure Pipes."
LeGat, Y. (2008b). "Modelling the Deterioration Process of Drainage Pipelines." Urban Water, 5(2), 97-
106.
LeGauffre, P., Haidar, H., Baur, R, and Poinard, D. (2007). "A Multi-criterial Decision Support
Methodology for Annual Rehabilitation Programs of Water Networks." Computer Aided Civil
and Infrastructure Engineering, 22, 478-488.
Lei, J. (1997). "Statistical Approach for Describing Lifetimes of Water Mains - Case Trondheim
Municipality." 22F007.28, SINTEF Civil and Environmental Engineering, Trondheim, Norway.
Li, D., and Haims, Y.Y. (1992a). "Optimal Maintenance-related Decision Making for Deteriorating Water
Distribution Systems 1. Semi-Markovian Model for a Water Main." Water Resources Research,
28(4), 1053-1061.
Li, D., and Haims, Y.Y. (1992b). "Optimal Maintenance-related Decision Making for Deteriorating Water
Distribution Systems 2: Multilevel Decomposition Approach." Water Resources Research, 28(4),
1063-1070.
Li, N., Haas, L.R., and Xie, W.C. (1997). "Development of a New Asphalt Pavement Performance
Prediction Model." Canadian Journal of Civil Engineering, 24, 547-559.
Li, N., Xie, W.C., and Haas, L.R. (1996). "Reliability-based Processing of Markov Chains for Modelling
Pavement Network Deterioration." Transportation Research Record, 1524, 203-213.
Li, N., Xie, W.C., and Hass, L.R. (1995). "A New Application of Markov Modelling and Dynamic
Programming in Pavement Management." 2nd International Conference on Road and Airfield
Pavement Technology, Singapore.
Lillie, K., C. Reed and Rodgers, M. (2004). "Workshop on Condition Assessment Inspection Devices for
Water Transmission Mains." AWWA Research Foundation, Denver, CO.
Loganathan, G.V., Park, S., and Sherali, H.D. (2002). "Threshold Break Rate for Pipeline Replacement in
Water Distribution Systems." Journal of Water Resources Planning and Management, 128(4),
271-279.
Madanat, S.M., Mishalani, R., and Ibrahim, W.H.W. (1995). "Estimation of Infrastructure Transition
Probabilities from Condition Rating Data." ASCE Journal of Infrastructure Systems, 1(2), 120-
125.
Mailhot, A., Paulin, A., and Villeneuve, J.P. (2003). "Optimal Replacement of Water Pipe." Water
Resources Research, 39(5), HWC2.1-HWC2.14.
Malone, P.and W. Morgan. (2006). "The Use of Emerging Technologies for Comprehensive Water
Conservation and Preservation." Mayors Water Council.
Makar, J.M. (1999). "Diagnostic Techniques for Sewer Systems." Journal of Infrastructure Systems, 5(2),
69-78.
Makar, J.M., and Chagnon, N. (1999). "Inspecting Systems for Leaks, Pits, and Corrosion." Journal of
AWWA, 91(7), 36-46.
128
-------
Mandayam, S., Polikar, R., and Chen, J.C. (2006). "A Data Fusion System for the Nondestructive
Evaluation of Non-Piggable Pipes." Rowan University, Glassboro, NJ, USA.
Marks, H.D., Andreous, S., Jeffrey, L., Park, C., and Zaslavski, A. (1987). "Statistical Models for Water
Main Failures." U.S. Environment Protection Agency (Co-operative Agreement CR810558)
M.I.T Office of Sponsored Projects No. 94211, Boston, MA, USA.
Marlow, D., Heart, S., Burn, S., Urquhart, A., Gould, S., Anderson, M., Cook, S., Ambrose, M., Madin,
B., and Fitzgerald, A. (2007). "Condition Assessment Strategies and Protocols for Water and
Wastewater Utility Assets." Water Environment Research Foundation, Alexandria, VA, USA.
MathWorks. (2007). "Thames Water Aims to Reduce Leaks by More Than 25% Using a Matlab-based
Leak-Location System."
Matthews, J., W. Condit, and McKim, R.. (2011). "Decision Support for Renewal of Wastewater
Collection and Water Distribution Systems." In Press. U.S. Environmental Protection Agency,
National Risk Management Research Laboratory, Cincinnati, Ohio, USA.
Mauch, M., and Madanat, S.M. (2001). "Semiparametric Hazard Rate Models for Reinforced Concrete
Bridge Deck Deterioration," ASCE Journal of Infrastructure Systems, 7(2), 49-57.
MB Inspection. (2008). "Pulsed Eddy Current System - Incotest."
http://www.mbinspection.co.uk/index/incotest
McKim, R., Kathula, V.S., and Nassar, R. (2002). "The Development of Risk Ratios for Sewer Prediction
Modelling." No-Dig Conference, Montreal, QC, Canada.
McMullen, L.D. (1982). "Advanced Concepts in Soil Evaluation for Exterior Pipeline Corrosion."
AWWA Annual Conference, Miami, USA.
Mergelas, B., and Henrich, G. (2005). "Leak Locating Method for Pre-Commissioned Transmission
Pipelines: North American Case Studies." Leakage 2005, Halifax, Canada, 1-7.
Mergelas, B., and Kong, X. (2001). Electromagnetic Inspection ofPrestressed Concrete Pressure Pipe,
AWWA Research Foundation and AWWA.
Mergelas, B.J., Atherton, D.L., and Kong, X. (2001). "NDE Inspection of PCCP Using Remote Field
Eddy Current Transformer Coupling." Proceedings of ASCE Pipeline Conference: Advances in
Pipeline Engineering and Construction, J.P. Castronovo, ed., San Diego, California, USA, 31-37.
Metalogic. (retrieved in 2003). http://www.metalogic.be.
Micevski, T., Kuczera, G., and Coombes, P. (2002). "Markov Model for Storm Water Pipe
Deterioration." ASCE Journal of Infrastructure Systems, 8(2), 49-56.
Miller, R.B. (1993). Personal Communications. CSIRO Division of Mathematics and Statistics, Glen
Osmond, Australia.
Misiunas, D. (2005). "Failure Monitoring and Asset Condition Assessment in Water Supply Systems,"
Ph.D., Lund University.
Moglia, M., Burn, S., and Meddings, S. (2006). "Decision Support System for Water Pipeline Renewal
Prioritisation." ITcon, 11, 237-256.
Moglia, M., Davis, P., and Burn, S. (2008). "Strong Exploration of a Cast Iron Pipe Failure Model."
Reliability Engineering & System Safety, 93(6), 885-896.
Moore, P.O. (2007). Nondestructive Testing Handbook: Ultrasonic Testing. American Society for
Nondestructive Testing Inc., Columbus, OH, USA.
129
-------
Moraleda, J., Ollero, A., and Orte, M. (1999). "A Robotic System for Internal Inspection of Water
Pipelines." IEEE Robotics and Automation Magazine, 6(3), 30-41.
Morison, D. (2007). "Fiber-Optic Sensors for Corrosion Monitoring." Sensors.
Moselhi, O., and Fahmy, M. (2008). "Discussion of Prediction of Water Pipe Asset Life Using Neural
Networks." Journal of Infrastructure Systems, 14(3), 272-273.
Mounce, S.R., Khan, A., Woodb, A.S., Dayb, A.J., Widdopb, P.O., and Machellc, J. (2003). "Sensor-
Fusion of Hydraulic Data for Burst Detection and Location in a Treated Water Distribution
System " Information Fusion, 4(3), 217-229.
Mrad, N., Liu, Z., Kobayashi, M., and Jen, C.K. (2006). "Exfoliation Detection Using Structurally
Integrated Piezoelectric Ultrasonic Transducers." Insight, 48(12), 738-742.
Munser, R., Robner, M., Hartrumpf, M., and Kuntze, H.B. (1999). "Microwave Back-Scattering Sensor
for the Detection of Hidden Material Inhomogeneities e.g. Pipe Leakages." 9th International
Trade Fair and Conference for Sensors, Transducers and Systems, Nuremberg, Germany, 1-7.
NACE International. 2011. "Acoustic Fiber Optic Monitoring Detects Water Main Problem Before
Failure Occurs,"Materials Performance, Vol. 50, No. 6, pp. 16-19.NASSCO. (2001). Pipeline
Assessment and Certification Program (PACP), v. 3.0. National Association of Sewer Service
Companies, Owings Mills, MD, USA.
Nafi, A., and Kleiner, Y. (2009). "Considering Economies of Scale and Adjacent Infrastructure Works in
Water Main Renewal Planning." Computing and Control in the Water Industry.
Nafi, A., Werey, C., and Llerena, P. (2008). "Water Pipe Renewal Using a Multiobjective Approach."
Canadian Journal of Civil Engineering, 35(1), 87-94.
Najafi, M., and Kulandaivel, G. (2005). "Pipeline Condition Prediction Using Neural Network Models."
Pipelines, ASCE, Hawaii, USA, 767-775.
Najjaran, H., Sadiq, R., and Rajani, B. (2006). "Fuzzy Expert System to Assess Corrosivity of
Cast/Ductile Iron Pipes from Backfill Properties." Computer Aided Civil and Infrastructure
Engineering, 21(1), 67-77.
Nassiraei, A.A.E., Kawamura, Y., Ahrary, A., Mikuriya, Y., and Ishii, K. (2006). "A New Approach to
the Sewer Pipe Inspection: Fully Autonomous Mobile Robot "KANTARO"." IEEE 32nd Annual
Conference on Industrial Electronics.
Neilla, I.T., Oppenheima, I.J., and Greveb, D.W. (2007). "A Wire-Guided Transducer for Acoustic
Emission Sensing." SPIE Smart Structures/NDE Joint Conference, M. Tomizuka, C.-B. Yun, and
V. Giurgiutiu, eds., San Diego, USA.
Nikles, M., Vogel, B., Briffod, F., Grosswig, S., Sauser, F., Luebbecke, S., Bals, A., and Pfeiffer, T.
(2004). "Leakage Detection Using Fiber Optics Distributed Temperature Monitoring." SPIE 11th
Annual International Symposium on Smart Structures and Materials, San Diego, California, USA,
18-25.
O'Day, O.K., Weiss, R, Chiavari, S., Blair, D., and Clark, R.M. (1986). "Water Main Evaluation for
Rehabilitation/Replacement." AWWA Research Foundation, Denver, CO and U.S. EPA,
Cincinnati, OH.
Ong, S.K., Gaunt, J., Mao, F., Cheng, C.L., Esteve-Agelet, L., and Hurburgh, C. (2008). "Impact of
Hydrocarbons on PE/PVC Pipes and Pipe Gaskets." Water Research Foundation (Report 91204),
Denver, CO, USA.
130
-------
Pandey, M.D. (1998). "Probabilistic Models for Condition Assessment of Oil and Gas Pipeline." NDT &
EInternational, 31(5), 349-358.
Park, S., and Loganathan, G.V. (2002). "Optimal Pipe Replacement Analysis with a New Pipe Break
Prediction Model." Journal of the Korean Society of Water and Wastewater, 16(6), 710-716.
Polikar, R., Udpa, L., Udpa, S.S., and Honavar, V. (2001). "Learn++: An Incremental Learning
Algorithm for Supervised Neural Networks." IEEE Transactions on Systems, Man, and
Cybernetics - Part C: Applications and Reviews, 31(4), 497-508.
PPIC. (2006). "Transmission Main Leak Location (Sahara)."
PureTechnologies. (2009). "SmartBall for Water Leak Detection."
Radiodetection. (2002). "Pipeline Current Mapper User Guide."
Rajani, B. (2010). "Unpublished Client Report. "National Research Council Canada.
Rajani, B., and Kleiner, Y. (2001). "Comprehensive Review of Structural Deterioration of Water Mains:
Physically Based Models." Urban Water, 3(3), 151-164.
Rajani, B., and Kleiner, Y. (2004). "Non-Destructive Inspection Techniques to Determine Structural
Distress Indicators in Water Mains." Evaluation and Control of Water Loss in Urban Water
Networks, Valencia, Spain, 1-20.
Rajani, B., Kleiner, Y., and Sadiq, R. (2006). "Translation of Pipe Inspection Results into Condition
Ratings Using the Fuzzy Synthetic Evaluation Technique." Journal of Water Supply Research
and Technology, 55(1), 11-24.
Rajani, B., and Makar, J. (1999). "A Methodology to Estimate Remaining Service Life of Grey Cast Iron
Water Mains." Canadian Journal of Civil Engineering, 27, 1259-1272.
Rajani, B., and McDonald, S. (1995). "Water Mains Break Data on Different Pipe Materials for 1992-
1993." A-7019.1, National Research Council Canada, Ottawa.
Rajani, B., and Tesfamariam, S. (2004). "Uncoupled Axial, Flexural, and Circumferential Pipe-Soil
Interaction Analyses of Partially Supported Jointed Water Mains." Canadian Geotechnical
Journal, 41,997-1010.
Rajani, B., and Tesfamariam, S. (2007). "Estimating Time to Failure of Cast-Iron Water Mains." Water
Management, 160(WM2), 83-88.
Rajani, B., and Zhan, C. (1996). "On the Estimation of Frost Load." Canadian Geotechnical Journal,
33(4), 629-641.
Rajani, B., Zhan, C., and Kuraoka, S. (1996). "Pipe-Soil Interaction Analysis for Jointed Water Mains."
Canadian Geotechnical Journal, 33(3), 393-404.
Randall-Smith, M., Russell, A., and Oliphant, R. (1992). "Guidance Manual for the Structural Condition
Assessment of the Trunk Mains." Water Research Centre, Swindon, UK.
RapidView. (2007a). "PANORAMO 3D Optical Scanner."
RapidView. (2007b). "PANORAMO SI 3D Optical Manhole Scanner."
RedZone. (2008). "Multi-Frequency Sonar: Fine Tuning Your Pipe Inspection System."
Renaud, E., JC, D. M., Bremond, B., and Laplaud, C. (2007). "SIROCO, a Decision Support System for
Rehabilitation Adapted for Small and Medium Size Water Distribution Companies." 2nd Leading
Edge Conference on Strategic Asset Management, Lisbon, Portugal, 17-19.
131
-------
Reed,C., Robinson, A.J., and Smart, D. (2004). "Techniques for Monitoring Structural Behaviour of
Pipeline Systems." AWWA Research Foundation, Denver, CO.
Rome, E., Hertzberg, J., Kirchner, F., Licht, U., and Christaller, T. (1999). "Towards Autonomous Sewer
Robots: The MAKRO Project." Urban Water,(1), 57-70.
Rose, J.L., Mu, J., and Cho, Y. (2008). "Recent Advances on Guided Waves in Pipe Inspection." 17th
World Conference on Nondestructive Testing, Shanghai, China.
Rossum, J.R. (1969). "Prediction of Pitting Rates in Ferrous Metals from Soil Parameters." Journal
AWWA, 61(6), 305-310.
Rostum, J. (2000). "Statistical Modelling of Pipe Failures in Water Networks," Norwegian University of
Science and Technology, Trondheim, Norway.
Rostum, J., Bauer, R, Saugrove, S., Horold, S., and Schilling, W. (1999). "Predictive Service Life
Models for Urban Water Infrastructure Management." 8th International Conferene Urban Storm
Drainage, Sydney, Australia, 594-601.
Rostum, J., Kowalski, M., and Hulance, J. (2004). "User Manual CARE-W Rehab Manager." WRc,
Trondheim, UK.
Russell. (2009). "See Snake Brochure." Russell NDE Systems Inc.
Sack, D.A., and Olson, L.D. (1998). "Impact Echo Testing of In-Situ Precast Concrete Cylinder Pipe."
Pipelines in the Constructed Environment, J.P. Castronovo and J.A. Clark, eds., American
Society of Civil Engineers, San Diego, CA, USA, 250-259.
Sadiq, R., Rajani, B., and Kleiner, Y. (2004). "Fuzzy-Based Method to Evaluate Soil Corrosivity for
Prediction of Water Main Deterioration." Journal of Infrastructure Systems, 10(4), 149-156.
Saegrov, S. (2009). Personal Communications. Norwegian University of Science and Technology,
Department of Hydraulic and Environmental Engineering, Trondheim, Norway.
Saegrov, S. (2010). Personal Communication. Y. Kleiner, ed.
Sansalone, M.J., and Streett, W.B. (1998). "The Impact-Echo Method." NDTnet.
Savic, D.A., and Walters, G.A. (1999). "Hydroinformatics, Data Mining and Maintenance of UK Water
Networks." Anti-Corrosion Methods and Materials, 46(6), 415-425.
Schempf, H. (2006). "Explorer-II: Wireless Self-powered Visual and NDE Robotic Inspection System for
Live Gas Distribution Mains." Carnegie Mellon University.
Schemph, H. (2004). "State-of-the-Art Review: In-Pipe-Assessment Robot Platforms." Carnegie Mellon
University.
Shamir, U., and Howard, C.D.D. (1979). "An Analytic Approach to Scheduling Pipe Replacement."
Journal AWWA, 71(5), 248-258.
Shaw, D., Mile, J., and Gray, A. (2004). "Genetic Programming within Civil Engineering." Organization
of the Adaptive Computing in Design and Manufacture 2004 Conference, Clifton, Bristol, UK, 1-
6.
Shinozuka, M., Feng, M., and Mosallam, A. (2007). "Wireless MEMS-Sensor Networks for Monitoring
and Condition Assessment of Lifeline Systems." Urban Remote Sensing Joint Event, Paris,
France.
SINTEF. (2008). "Smart Pipe Project."
132
-------
Slaats, P.O., Mesman, G.A., Rosenthal, L.P., and Brink, H. (2004). "Tools to Monitor Corrosion of
Cement-Containing Water Mains." Water Science and Technology, 49(2), 33-39.
Spangler, M.G. (1941). "The Structual Design of Flexible Pipe Culverts." Iowa Engineering Experimental
Station Bulletin, No. 53, Ames, Iowa, USA.
Spickelmire, B. (2002). "Corrosion Consideration for Ductile Iron Pipe." Materials Performance, 41, 16-
23.
Stone, S., E. Dzuray, D. Meisegeier, A. Dahlborg, and M. Erickson. (2002). "Decision Support Tools for
Predicting the Performance of Water Distribution and Wastewater Collection Systems."
EPA/600/R-02/029. U.S. Environmental Protection Agency, National Risk Management
Research Laboratory, Cincinnati, Ohio, USA. Available
at http://www.epa.gov/nrmrl/pubs/600r02029/600R02029.pdf
Stoianov, I., Nachman, L., and Madden, S. (2007). "PIPENET: A Wireless Sensor Network for Pipeline
Monitoring." The 6th International Conference on Information Processing in Sensor Networks,
Cambridge, Massachusetts, USA, 264-273.
SwRI. (2002). "Conformable Array for Mapping Corrosion Profiles." Natural Gas Infrastructure
Reliability Industry Forums, Morgantown, WV, USA.
Tesfamariam, S., Rajani, B., and Sadiq, R. (2006). "Possibilistic Approach for Consideration of
Uncertainties to Estimate Structural Capacity of Ageing Cast Iron Water Mains." Canadian
Journal of Civil Engineering, 33, 1050-1064.
Thomson, J., and Wang, L. (2009). "Condition Assessment of Ferrous Water Transmission and
Distribution Systems State-of-Technology Review Report." EPA/600/R-09/055, U.S.
Environmental Protection Agency, National Risk Management Research Laboratory, Cincinnati,
Ohio, USA.
Tran, H.D. (2007). "Investigation of Deterioration Models for Storm Water Pipe Systems," Victoria
University, Australia.
TuGraz. (2006). "PiReM: Decision Support System for the Rehabilitation Management of Water Supply
Systems."
Vairavamoorthy, K., Gorantiwar, S.D., Yan, J., Galgale, H.M., Mohamed-Mansoor, M.A., and Moban, S.
(2006). Risk Assessment of Contaminant Intrusion into Water Distribution Systems, Water,
Engineering and Development Centre, Loughborough University, Leicestershire, UK.
Vitkovsky, J.P., Simpson, A.R., and Lambert, M.F. (2000). "Leak Detection and Calibration Using
Transients and Genetic Algorithm." Journal of Water Resources Planning and Management
(July/August), 262-265.
Vradis, G.C., and Leary, W. (2004). "Development of an Inspection Platform and a Suite of Sensors for
Assessing Corrosion and Mechanical Damage on Unpiggable Transmission Mains." Northeast
Gas Association.
Walski, T.M. (1987). "Replacement Rules for Water Mains." Journal AWWA, 79(11), 33-37.
Walski, T.M., and Pelliccia, A. (1982). "Economic Analysis of Water Main Breaks." Journal AWWA,
74(3), 140-147.
Wardany, R.A. (2008). "Condition Assessment of Prestressed Concrete Cylindrical Water Pipes." 60th
Annual WCWWA Conference and Trade Show, Regina, SK, Canada, 1-9.
WATERPIPE. (2009a). "3rd Waterpipe Newsletter."
133
-------
WATERPIPE. (2009b). "Integrated High Resolution Imaging Ground Penetrating Radar and Decision
Support System for WATER PIPEline Rehabilitation."
Waters, C. (2005). "RTD-INCOTEST- For the Detection of Corrosion Under Insulation."
Watkins, R.K., and Spangler, M.G. (1958). "Some Characteristics of the Modulus of Passive Resistance
of Soil - A Study in Similitude." Highway Research Board, 576-583.
Watson, T.G., Christian, C.D., Mason, A.J., Smith, M.H., and Meyer, R. (2004). "Bayesian-based Pipe
Failure Model." Journal ofHydroinformatics, 6(4), 259-264.
Wilson, J.W., Kaba, M., and Tian, G.Y. (2008). "New Techniques for the Quantification of Defects
Through Pulsed Magnetic Flux Leakage." 17th World Conference on Nondestructive Testing,
Shanghai, China.
Wirahadikusumah, R., Abraham, D., and Isely, T. (2001). "Challenging Issues in Modelling Deterioration
of Combined Sewers." ASCE Journal of Infrastructure Systems, 7(2), 77-84.
WRc. (1986). Sewer Rehabilitation Manual, 2nd Edition. Water Research Centre Pic, Wiltshire, UK.
WRc. (1993). Manual of Sewer Condition Classification, 3rd Edition. Water Research Centre Pic
Wiltshire, UK.
WRc. (1994). Sewer Rehabilitation Manual, 3rd Edition. Water Research Centre Pic, Wiltshire, UK.
WRc. (2001). Sewer Rehabilitation Manual, 4th Edition. Water Research Centre Pic, Wiltshire, UK.
Wright, L.T., Heaney, J.P., and Dent, S. (2006). "Prioritizing Sanitary Sewers for Rehabilitation Using
Least-Cost Classifiers." ASCE Journal of Infrastructure Systems, 12(3), 174-183.
WSSC. (2011). "66-in. Water Main Break on River Road and PCCP Program." Available at
http://www.washingtonpost.com/wp-srv/metro/documents/wssc012709.ppt.
Zhan, C., and Rajani, B.B. (1997). "On the Estimation of Frost Load in a Trench: Theory and
Experiment." Canadian Geotechnical Journal, 34(3), 568-579.
Zhao, J.Q., McDonald, S.E., and Kleiner, Y. (2001). "Guidelines for Condition Assessment and
Rehabilitation of Large Sewers." Institute for Research in Construction, National Research
Council Canada, Ottawa.
Zhou, Y., Vairavamoorthy, K., and Grimshaw, F. (2009). "Development of a Fuzzy Based Pipe Condition
Assessment Model Using PROMETHEE." The 29th World Environmental & Water Resources
Congress, Kansas City, Missouri, USA, 1-10.
134
------- |