UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Data assessment and utilization for improving asset management of small and medium size water utilities Wood, Andrew 2007

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
ubc_2007-268223.pdf [ 29.15MB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0063271.json
JSON-LD: 1.0063271+ld.json
RDF/XML (Pretty): 1.0063271.xml
RDF/JSON: 1.0063271+rdf.json
Turtle: 1.0063271+rdf-turtle.txt
N-Triples: 1.0063271+rdf-ntriples.txt
Original Record: 1.0063271 +original-record.json
Full Text
1.0063271.txt
Citation
1.0063271.ris

Full Text

DATA ASSESSMENT AND UTILIZATION FOR IMPRO VING ASSET MAN A GEMENT OF SMALL AND MEDIUM SIZE WATER UTILITIES by ANDREW WOOD B.A.Sc, The University of British Columbia, 1987 M.Eng., The University of British Columbia, 1998 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Civil Engineering) THE UNIVERSITY OF BRITISH COLUMBIA January 2007 © Andrew Wood, 2007 ABSTRACT Data regarding water main breaks are essential for undertaking informed and effective infrastructure asset management. This thesis reports on the findings of a survey regarding water main break data collection practices across North America and develops an approach for constructing databases and integrating the data with break prediction models to improve the asset management practices of a utility. The survey determines the amount and type of data collected by water utilities, the level of comfort with the amount of data collected and the availability of alternate sources of data. The responses provide insight into the strategies and data collection practices of small to mid-size utilities and show that the amount of data collected by utilities can be classified by the degree of data richness and defined as either an expanded, intermediate, limited or minimal data set. Utilities can implement recommended practices to increase the amount of data they collect, increase effectiveness of data collection and processing and consider additional sources of data for water main breaks to improve their data sets. The thesis also introduces an approach for constructing a water main break and general network database that relates data from multiple sources to augment the amount of data available for asset management analysis while maintaining existing data warehousing practices. When used, managers may gain insight into current and future performance of the distribution network and develop future asset management strategies. The approach is flexible, uses commonly available software tools and anticipates the evolution of data collection, verification and storage capabilities within the utility. Finally, a framework is presented that guides small to medium water utilities in identifying key data to be used in asset management and pipe break prediction modeling ii and in selecting appropriate water main break prediction models. The framework may be used to identify the magnitude of a utility's pipe burst problems today and in the future, enhance the development of pipe replacement priorities based on forecasted breaks and identify key data to collect in future data acquisition programs. Water utilities with varying amounts of data can easily implement it with their existing data management and analysis tools. iii TABLE OF CONTENTS Abstract ii Table of contents v List of tables ........viList of figures viii Acknowledgements x Co-authorship statement xii 1 Introduction 1 1.1 Background 2 1.2 Thesis objectives and organization 7 1.3 Asset management 9 1.4 Predicting water main breaks 13 1.5 Research methods 18 1.5 References 29 2 Assessment of water main break data for asset management 37 Preface 38 2.1 Introduction '. 40 2.2 Design of the survey '• 42 2.3 Survey results 43 2.4 Discussion and recommendations 51 2.5 Conclusions 56 2.6 Acknowledgements 8 iv 2.7 References 59 3 Constructing water main break databases for asset management 75 Preface 76 3.1 Introduction 8 3.2 State of data in utilities .' 80 3.3 Water main break data for asset management 1 3.4 Constructing water main databases 84 3.5 A water main break database for Maple Ridge, BC 92 3.6 Discussion 98 3.7 Conclusions 100 3.8 Acknowledgements 102 3.9 References 103 4 Using water main break data to improve asset management for small and medium utilities 119 Preface , 120 4.1 Introduction 122 4.2 Water main breaks 123 4.3 A framework for using prediction models to improve asset management 128 4.4 Break prediction models for Laity View, Maple Ridge, BC 132 4.5 Conclusions 139 4.6 Ackno wledgem ents 140 v 4.7 References... 141 5 Conclusions and recommendations 153 5.1 Summary of research goals ..154 5.2 Conclusions 155 5.3 Observations 7 5.4 Closing remarks 165 References 167 Appendices 179 Appendix A Comparison of survey questions for Deb et al. (2002) and Wood and Lence (2006) '. 179 Appendix B Statistical analysis for guiding the interpretation of survey results 183 Appendix C Description of Maple Ridge water system 187 Appendix D Water main break survey form 193 Appendix E List of organizations that were sent a copy of the water main break survey 203 Appendix F Survey responses 215 Appendix G Water main break and system data and statistics of models 393 Appendix H Degree of accuracy of predictions for pipe groups 437 vi LIST OF TABLES i Table 2.1 Service population of respondents 62 Table 2.2 Percentage of respondents that record location data 63_ Table 2.3 Additional sources of break-related physical data for respondents .64 Table 2.4 Suggested sources and approaches for collecting physical data on water main breaks 66 Table 3.1 Water main break data availability for Maple Ridge 107 Table 3.2 Water main breaks for a given year of installation 109 Table 3.3 Pipe breaks for pipes of a given diameter (1983-1999) 110 Table 4.1 Typical data used in models and factors for which they are a surrogate 146 vii LIST OF FIGURES Figure 1.1 Twenty-year total per household infrastructure cost estimates for different sized water systems 36 Figure 2.1 Percentage of respondents that collect general information ..68 Figure 2.2 Percentage of respondents that record physical data 69 Figure 2.3 Percentage of respondents that record failure causes 70 Figure 2.4 Percentage of respondents that record repair activities 71 Figure 2.5 Percentage of respondents that collect different types of environmental data : 72 Figure 2.6 Percentage of respondents that expressed confidence in data collected 73 Figure 2.7 Classes of data richness among water utilities 74 Figure 3.1 A schematic for constructing and using water main break data for knowledge discovery Ill Figure 3.2 A process for digitizing and creating data from archival geographical data 112 Figure 3.3 Linking hydraulic model data with network data 113 Figure 3.4 Buffering data to create data relationships using GIS 114 Figure 3.5 Maple Ridge water main break analysis data web 115 Figure 3.6 Cumulative breaks in Laity view area: 1983-1999 116 Figure 3.7 Number of years in service when break occurred in pipes (1983-1999) 117 Figure 3.8 Number of breaks for pipes in a given soil type installation (1983-1999) 118 viii Figure 4.1 Improving asset management using pipe break prediction models 147 Figure 4.2 Degree of accuracy of time-linear and time-exponential predictions for material groups 148 Figure 4.3 Degree of accuracy of time-linear and time-exponential predictions for material and diameter groups 149 Figure 4.4 Degree of accuracy of time-linear and time-exponential predictions for material, diameter and age groups 150 Figure 4.5 Degree of accuracy of time-linear and time-exponential predictions for material, diameter, soil and age groups 151 Figure 4.6 Degree of accuracy of time-linear and time-exponential predictions for material, diameter, age and surface material groups 152 ix ACKNOWLEDGEMENTS I believe that behind any work are people and events that supported, influenced and shaped it and this is indeed true for my thesis adventure and journey. ' I am grateful to Barbara Lence, my academic supervisor who encouraged and supported my ideas and work, gave me important advice and was patient throughout my program. You demonstrated by many long hours, a dedication to mentoring and I thank you for that and for igniting my passion for research, academia and success. My other committee members who mentored and gave me timely, succinct, gracious and kind advice and guidance were Alan Russell and Jim Atwater. My journey was successful because you pointed out or rolled stones for me to step on as I crossed the doctoral river. Thank you for inspiring me to connect work, research and community service. I can say of my supervisory committee, as Isaac Newton penned, "If I have seen further it is by standing on the shoulders of giants." My family has been instrumental, by caring, praying and encouraging throughout the journey. Clare, I could not have done this without your endless support, patience and belief in me. I love you. You shared the weight of the studies in so many ways and also freed me to journey through unimagined waters. Mark and Julia, who have only known their father as a graduate student, having you both along with me has been a joy, and I am truly blessed. I have also been supported by my mother, Betty by her steadfast prayers and confident and patient hope. I am grateful for the support of my employers, Maple Ridge and Coquitlam who tasked and then entrusted me with the quest of improving their asset management practices. x I trust that this work will support their efforts to sustain their water infrastructure, protect public health and support their local economy. Finally, I have had the pleasure of working with many colleagues and fellow practitioners. Of these, I wish to specifically acknowledge Wilson Liu. xi CO-AUTHORSHIP STATEMENT Andrew Wood was the lead and principal researcher of the work contained in the thesis titled "Data Assessment and Utilization for Improving Asset Management of Small and Medium Size Water Utilities". Dr. Barbara Lence, the Research Supervisor of the thesis, provided inspiration and supported the writing of the papers and Wilson Liu assisted with the preparation of figures used in the paper submitted as Chapter 3. xii CHAPTER 1 INTRODUCTION 1.1 BACKGROUND Reliable, efficient and effective water distribution systems are crucial to public health and safety. These systems are also essential to the economic well-being of many municipalities since manufacturing, industry and commerce rely to a large degree on obtaining reliable and economically-priced water delivered through a network of water pipe lines, more commonly referred to as Water mains. Many of these water mains installed over the decades in North America are now beginning to break and fail. The traditional public works emphasis on managing water main breaks has been directed toward minimizing the loss of water to key businesses and critical facilities and the damage to built and natural infrastructure. However, breaks are also potential gateways to contamination of the water distribution system (AWWA and EES, 2002) and are identified as a high priority in the assessment of water supply health risks by the National Research Council of Academy Sciences (2005). By replacing the pipes just before they fail, utilities reduce their risks and costs of water main breaks. This thesis is focused on helping small and medium size utilities by providing information on water main break data collection, construction, compilation, and management in support of asset management. The research presented includes an approach for these utilities to use water main break prediction models on a pipe by pipe basis for prioritizing the replacing of water mains. The work is presented as a series of three manuscripts that are published in peer-reviewed journals or are in the process of being peer-reviewed. When I embarked on this research program my goal was to assist small and medium size utilities with managing their pipe networks to address the risks arid costs associated with replacing water mains. Specifically, I wanted to assist them with identifying, 2 . collecting and constructing relevant pipe break data to analyze their pipe network, prioritizing their water main replacements by using break predictions and guiding their data acquisition and analysis programs. During my almost twenty years of experience as a professional engineer in public utilities I observed that my colleagues in other utilities were not using break prediction models in their practice. Within the organizations with which I was familiar, collection of water main break data was undertaken on an ad-hoc basis and storage of such information was not typically comprehensive. Among those utilities, there were no common practices or standards for data collection and very few utilities, if any, were aware of the data collected by other utilities. Best practices for Canadian water utilities regarding which data to collect were introduced in 2002 (NSGMI, 2002). As a manager, I also wanted to compare the information I was collecting with that of other utilities, to understand how data were informing us and to share this knowledge with my colleagues. During this time, I also became aware of the general belief held by many utility managers that asset management is a panacea for solving our aging water system problems. We needed to know when, where and how many pipes to replace; but we were focusing on the entire network, not on specific pipes, to obtain funding for programs. I found that among my colleagues, once funding was obtained, decisions on prioritizing pipe replacements were based on historical data, experiences and management policies rather than on the expected performance of specific pipes though there are models for predicting water main breaks. In fact, many of my colleagues were not aware of these models and none, if any, were using them. Thus, I was convinced that we needed an approach that applied pipe break models to inform our decisions regarding the prioritization and scheduling of specific pipe replacements. 3 As a result of my observations, my thesis identifies data that are collected and available for analysis across utilities in North America, develops a methodology for creating and linking data across various data sources and develops a framework for assisting in the prioritization of water main replacements and data acquisition based on predictions of future water main breaks within a given water distribution system. The framework may be applied across the range of data available in typical water utilities, acknowledges existing industry needs and practices and Will help managers identify and use key available break data. Our water mains are aging and failing. The water mains of many municipalities are predominantly of the 1960s to 1990s vintage because of extensive urban development during those years, with some smaller amounts prior to that period. However, the replacement needs of aging public works infrastructure has become the focus of governmental and academic attention in recent years. Because few water utilities have actively pursued aggressive water main rehabilitation or replacement programs, they are now facing the problem of replacing or rehabilitating their aging systems (Deb et al., 2002). The United States Environmental Protection Agency (USEPA) identifies that while United States (US) communities spent one trillion US dollars in 2001 on drinking water treatment, supply and waste water treatment and disposal, this expenditure level may not be sufficient to keep pace with future infrastructure needs (USEPA, 2002). They also identify in a 2003 survey that the 53,000 community water systems and 21,400 not-for-profit non-community water systems in the US will need an estimated $276.8 billion to continue to provide their services (USEPA, 2005). Similarly, the Ontario Ministry of Public Infrastructure Renewal (PIR) identifies that over the next 15 years, $25 billion will be needed for capital renewal of Ontario's $72 billion water and waste water assets (Ontario PIR, 2005). The Federation of 4 Canadian Municipalities (FCM) state that key investments must be made in core public infrastructure to manage waste and water systems, to meet pressing environmental and air quality needs and to maintain the economic health of Canada's communities (FCM, 2001). Given the current level of expenditures and the increasing age of systems, utilities need to predict breaks and use this information to prioritize and plan when pipes should be replaced. The ability to respond to the challenge of knowing when to replace a pipe varies among water utilities. Small to medium size utilities may find it particularly challenging due to their limited resources and technical capacity and there are many of these utilities across North America. In the US, small water systems make up 90 percent of those 53,000 community water systems (American Society of Civil Engineers, ASCE, 1999). No analogous data exist for Canada, although the Canadian National Research Council (CNRC) identifies that there are 3500 municipalities serving fewer than 5000 people and that there are only 63 municipalities serving populations greater than 100,000 (Vanier and Rahman, 2004). Most small to medium size water utilities and municipalities share similar characteristics. They purchase bulk water from a regional or larger supplier and require some treatment. The treatment processes are simple processes such as ozonation, ultra violet (UV) or chlorine disinfection. Small utility systems are typically comprised of small diameter pipes and.as noted by Kettler and Goulter (1985), smaller pipes break have been observed to break more frequently than larger ones. This may be because their beam strength and wall thickness are generally less than those of larger pipes. 5 Smaller utilities have scarce resources and technical expertise is typically riot available because staff are generally expected to undertake a number of different functions and do not have the luxury of developing in-house expertise. Compounding this is the fact that within these systems, there is little or no reliable documentation of the location, capacity, condition and adequacy of pipe network elements for meeting present or future needs (Myers, 2001). These utilities may also lack the resources, financially and organizationally, to implement a complex information management system program, nor the historical data or tools to fully analyze their system. They require capital to rehabilitate, upgrade and install infrastructure, but face an economic challenge in paying for these costs given a small revenue base. For example, although the costs of a small water system may be modest compared with those of large systems, the per household costs are significantly higher than those of larger systems (USEPA, 2001). The data shown in Figure 1.1 are the per household costs for different sized water systems for meeting anticipated needs. These data show that households in small systems face costs that are over three and one half times those of households serviced by large systems. Utilities are viewing asset management as an approach for addressing this dilemma of planning pipe replacements by understanding how much rehabilitation will cost, what to do first and when to rehabilitate their systems. Asset management is a business administration approach to decision-making that covers an extended time horizon, draws from economics as well as engineering science and considers a broad range of assets. The approach incorporates the economic assessment of alternative investment options and uses this information, to help make cost-effective investment decisions (United States Federal Highway Administration, USFHWA, 1999). The benefits of applying an asset management 6 approach include the ability to link user expectations and needs and to identify the means of assessing value, system condition, performance, service life and management and investment strategies. However, for smaller utilities to adopt asset management, they need portable, readily useable approaches that require little modification, are likely to be met with little organizational resistance and serve to incrementally improve water main replacement analysis and planning. The need for planning is urgent and the need to improve the planning approaches will only increase with time. 1.2 THESIS OBJECTIVES AND ORGANIZATION This thesis: • assesses the state of data available to small arid medium size water utilities, • develops an approach for constructing and compiling water main break data for analysis and • develops a framework that guides the identification of key data for asset management and the selection of the most appropriate data and models for predicting water main breaks. Organization of thesis. The thesis is presented as a series of three manuscripts that are published in peer-reviewed journals or are in the process of being published or peer-reviewed. Each of the three manuscript chapters (Chapters 2 to 4) contains a preface, introduction, overview of the literature, body, conclusion and references. Tables and figures referenced in each chapter are included at the end of that chapter and all appendices are provided at the end of the thesis. Finally, Chapter 5 summarizes the major conclusions of this work. 7 There are many benefits of a manuscript-based thesis over the traditional thesis and the primary benefit is that research findings are published and shared with the research community much earlier than in the production of a traditional thesis. However, there are also significant disadvantages. Literature reviews are often repeated, some continuity is lost and much of the data, results, and observations that would be in a traditional thesis are condensed or eliminated. This results in an abbreviated thesis and a perception by the reader of a lesser effort than was the case in the research. This thesis attempts to address these limitations by including a section in this chapter on the approach, methodology and context of the research. Each chapter preface provides the context for the work presented in that chapter (paper) with respect to the overall research and thesis and the preceding chapter. Data not provided in the papers are also included in the appendices. Chapter 2 provides the results of a survey that investigates the state of data collection practices, the commonly available water main break data and the data sources within North American water utilities. The paper was published in the July 2006 issue of the Journal of the AWWA. A comparison of the survey questions and those of Deb et al. (2002) is presented in Appendix A. Chapter 3 presents an approach for building and relating data from various sources for analysis, based on observations drawn from the survey responses reported in Chapter 2. This work demonstrates how to construct water main data and databases which are based on linking, relating, extracting and compiling data from sources internal and external to a utility for the purpose of analysis, or in other words relating relational databases. Issues related to creating, linking, transforming, cleansing, scrubbing and integrating data are identified and approaches for addressing them are presented. A number of databases were created in this portion of the research. These are summarized in Appendix G. This paper 8 was submitted to the Journal of the A WW A in February 2006, revised in October 2006 and is scheduled for publication in January 2007. Chapter 4 develops a framework that guides utilities in identifying key data to be used in asset management in general and specifically in pipe break prediction modeling and in selecting the most appropriate model for predicting water main breaks. It provides the utility with a method for considering future pipe breaks in the analysis of pipe prioritization strategies. It incorporates existing tools for data management and analysis that are widely available and easy to implement by small to medium size utilities with varying amounts of data. This paper was submitted to the ASCE Journal of Infrastructure Systems in July, 2006. The fifth and final chapter summarizes the conclusions of and discussions arising from the research. A discussion of future research is also presented. 1.3 ASSET MANAGEMENT Six components of an asset management program have been suggested (Vanier, 2001). These are inventory, asset value, deferred maintenance, condition, service life prediction, and prioritization of rehabilitation, replacement and renewal. Inventory. The first building block of asset management is to determine the type, composition, quantity and extent of the assets. The primary capital assets that typical water utilities own include water supply reservoirs, dams and supporting hydraulic structures, water treatment plants, water distribution systems including pipes, valves, fire hydrants, distribution reservoirs, pump stations, pressure reducing valves and meters to measure distributed and purchased water. Utilities are typically maintaining inventories using tools such as Geographic Information Systems (GISs), Computer Aided Drafting (CAD) systems and relational databases. These systems require significant effort and resources to P implement and maintain but for small systems, a simple spreadsheet may be all that is required. Wood and Lence (2G06) found that archival records are still the predominant sources of data for utilities that serve populations fewer than 50,000. In recent years, the use of Computerized Maintenance Management Systems (CMMSs) has also grown in popularity and these systems usually interface with or use GIS data. These systems are expensive to implement and require significant resources to maintain. Many utilities still do not have CMMSs. Asset value. To plan for replacement, water utility managers must know the total value of their assets. Managers must have a basic understanding of system worth and how much is required to replace it. According to Vanier (2001), the six commonly used terms to describe asset value are historical, appreciated historical, capital replacement, performance-in-use, deprival cost and market value. Historical value is defined as the original or book value, appreciated historical value is the historical value calculated in current dollars and capital replacement value is the cost of the asset in current dollars. The performance-in-use value is the prescribed value of the actual asset to the user. Deprival cost is the cost that would be incurred if deprived of the asset but still required to deliver the service. Finally, market value is the value if the asset is sold on the open market. Utilities may not use all of these values. Few municipalities have specific asset values. In most cases, utilities inherit their new infrastructure from developers who install it as part of a housing subdivision and then transfer the ownership and maintenance of the assets to the municipality. Only the book value of the assets constructed directly by the municipality is typically recorded. For many utilities, the value of their assets are simply calculated by multiplying an average cost of construction per metre of pipe by the total length of the network. Condition. Asset management requires knowledge of the condition of the assets. In the past, a significant amount of data were captured for evaluating operational values and objectives and a simple spreadsheet was used to record these data. In recent years, CMMS have been used to warehouse and process information and the management of maintenance activity planning. However, asset managers face the problem of determining how and what to evaluate, of defining what constitutes condition and indices, of providing for data storage and of determining how to use condition data. Not only are there are no standard condition indices for water mains (Grigg, 2004), but the challenge for many water utilities is that condition assessments have to be on-going. They can be costly and only provide information for that asset at a specific time. There is also a need for a significant improvement in the level of accuracy in pipeline condition assessment and accurate prediction of pipeline failures (DeSilva et al, 2002). Deferred maintenance. Deferred maintenance is maintenance that has not been performed or has been deferred. For utilities, managing deferred utility maintenance requires identifying and managing the maintenance that is to be deferred and the compounding effect of maintenance that has not been performed such as not cleaning or repairing water mains. The challenge for water utility asset management lies in managing and integrating the amount of deferred maintenance between the various components that make up a system with the subjective nature of how a maintenance backlog is calculated. For example, most pipe networks are segmented and of differing age, condition, value and consequence of failure. Determining quantitatively the maintenance backlog of that 11 network is very difficult without accounting for reservoirs, pump stations and other assets. Thus there is very little information regarding deferred maintenance of water distribution systems and few if any municipalities can determine their deferred maintenance. Remaining service life. Managing water utility assets and planning replacement requires estimates of technical and economic service life. Asset managers face funding constraints and competing needs, thus optimizing the expenditures from technical and economic perspectives is important. While technical life may be difficult to predict, determining economic service life is much simpler since it compares the immediate costs of repairs with the costs of renewal. So the question that is often raised as part of the understanding and analysis of pipe deterioration is when does a pipe reach the end of its useful life? One definition of when a pipe reaches the end of its useful life is when the pipe is replaced by a utility. This can be misleading since a utility's decision to replace a pipe may be based on other factors such as political choice, perception of the reliability of the pipe or the need to increase the hydraulic capacity of the pipe. Rajani and Makar (2000) define the time of death of a pipe as the time at which its mechanical factor of safety falls below an acceptable value set by the utility. Kleiner and Rajani (1999) propose that the useful life of a pipe is a function of the economic costs of deterioration rate and replacement and suggest that pipe death coincides with the optimal time of replacement. Statistics Canada defines the service life of an asset as its useful life at the time of its acquisition which generally ends at demolition (Statistics Canada, 2006). According to Deb et al. (1998), there are no standardized methods for predicting the life of distribution systems. 12 Prioritizing. Finally, effective management of water utility assets requires the prioritization of replacement needs. That is, managers must be able to determine what to replace and when to replace it. This issue is closely tied to resolving financial and technical challenges as to whether to maintain, repair or renew an asset or to choose an alternative ) such as twinning a water main. Because utility managers are usually required to plan annual capital projects, this component of asset management is usually performed. In doing so, they have to overcome obstacles to effective prioritization such as how to address uncertainty when longer-term planning horizons are considered or how to balance the various needs among an organization. Traditionally, utilities have prioritized pipe replacements based on a combination of current management practices and historical pipe breakage data. Management practices include directives based on general guidelines, consequence assessments, legislative requirements and other utility priorities. Rudimentary analyses that interpret historical pipe break data, including location, time and date of break and pipe diameter and material have provided information regarding where and how many breaks are occurring and what pipes are experiencing breaks (Kleiner and Rajani, 1999). 1.4 PREDICTING WATER MAIN BREAKS A primary goal of prioritizing pipe replacements is to identify investment strategies that, on one hand, avoid premature replacement of pipes (i.e., unnecessary pre-investment of funds), and on the Other hand, avoid water main breaks, interruptions in service, potential contamination of water and the costs of damage. If utilities can predict when and where water mains may break, this information is also useful for assisting with optimizing 13 crew efforts and minimizing the results of loss of water to key businesses and critical facilities. Thus asset management decisions can be improved by an ability to determine the future performance of water mains by predicting water main breaks and possibly identifying how such breaks may occur. Causes of breaks. A number of authors have analyzed the causes of breaks, * including O'Day (1982), Male et al. (1990), Savic and Walters (1999), Rajani and Makar (2000), Rajani and Kleiner (2001) and Dingus et al. (2002). According to Rajani and Tesfamariam (2005), a combination of circumstances leads to pipe failure in most cases and different factors cause failure in different pipe networks. The causes of breaks include deterioration as a result of use (e.g., internal corrosion), physical loads applied to the pipe (e.g., traffic, frost), limited structural resistance of the pipe because of construction practices during installation and declining resistance over time (e.g., corrosion, aging factors). Dingus et al. (2002) surveyed the 46 largest American Water Works Association Research Foundation (AWWARF) member utilities in 1997 and noted multiple common failure modes for cast iron piping systems. Corrosion, improper installation and ground movement were the three most common causes of pipe failure. Break prediction models. Break prediction models have been developed to help the water industry understand how pipes deteriorate and when pipes will break in the future. These models are typically grouped into two classes - statistical and physical-mechanical models (Kleiner and Rajani, 2001). Statistical models use historical pipe break data to identify break patterns and extrapolation of these patterns to predict future pipe breaks, or degrees of deterioration. Physical-mechanical models predict failure by 14 simulating the physical effects and loads on pipes and the capacity of the pipe to resist failure over time. Statistical models have been used to analyze large distribution systems (e.g., Kleiner and Rajani, 1999) and are typically characterized as either statistical deterministic or statistical probabilistic equations (Kleiner and Rajani, 2001). Under the statistical deterministic models, the pipe breakage is estimated based on a fit of pipe breakage data to various time-dependent equations, which may represent the cumulative pipe breaks as a function of time from date of installation or from the earliest date of available break data, and most commonly are time-linear (Kettler and Goulter, 1985) or time-exponential functions (Shamir and Howard, 1979; Walski, 1982 and Kleiner and Rajani, 1999). Deterministic models were developed as early as 1979. Statistical probabilistic models predict not only the failure potential, but the distribution of failure. These models are more complex than deterministic models and require more data. Examples of these include cohort survival, such as KANEW (Deb et al, 2002), Bayesian diagnostic, break clustering, semi-Markov Chain and data filtering methods. Survival analysis has been demonstrated by Mailhot et al (2000) as useful if there are adequate histories of pipe failures. Statistical models can use available historical' data on past failures to identify breakage patterns and are useful if the data are limited. Statistical methods require some technical expertise in developing the models and interpreting results, but not to the degree of expertise required of staff if physical-mechanical models are used. Kleiner and Rajani (2001) suggest that statistical models based on fewer data may be used to gain insights for future performance. 15 Physical-mechanical models typically fall into one of two classes: deterministic models which estimate pipe failure based on simulation of the physical conditions affecting the pipe (Doleac et al., 1980 and Rajani and Makar, 2000), and probabilistic models that use a distribution of input conditions, such as rate of corrosion, to predict the likelihood and distribution of pipe failure (Ahammed and Melchers, 1994). Physical models have been developed primarily for cast iron and cement pipes and have significant data needs. Kleiner and Rajani (2001) suggest that only larger diameter mains with costly consequences of failure may justify the required data collection efforts for these models. r Other developments in recent years include the use of Artificial Neural Networks (ANNs) by Sacluti (1999) and fuzzy-based techniques by Kleiner et al. (2004). The ANN model predicts the number of water main breaks based on a seven day weather forecast and is applied only to homogeneous groups of water mains for short-term maintenance work. The fuzzy-based techniques are applied to large transmission water mains. In addition, an agent-based system for predicting water main breaks is proposed by Davis (2000). While much work has been undertaken toward developing deterioration models, the use of these models is not common among utilities. Utilities face obstacles such as a lack of key data, limited institutional capacity within their organization to understand and use models and the complexity in managing work that involves a number of groups. There is no common model that may be applied to every water system. Because the literature suggests that breaks and causes of breaks for any particular water distribution system are system-specific (Rajani and Tesfamariam, 2005), a utility must create its system-specific model based on factors of deterioration that are relevant for the utility. 16 Small and medium size utilities typically have the capacity to use statistical deterministic models, but the implementation of more sophisticated models is not practical due to the data collection efforts and model maintenance required. As this research demonstrates, statistical deterministic models show promise for use by these utilities because they can be applied using commonly used software, they do not require specialized tools or expertise to operate, and may by their degree of accuracy provide managers with insights into their system on future pipe breaks. Ultimately, prediction models need data and in particular, data that are relevant for the models and are explanatory for accurate predictions. Data collection regarding water main breaks is not a simple exercise nor is the practice consistent across utilities. For all utilities data collection can have significant costs if performed at a comprehensive level. These costs include not only out of pocket costs but also organizational effort and human resources. While a collection of best practices have been recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb at al., 2002), the current state of knowledge of data collected by small and medium size utilities is limited. Recent water utility surveys include those initiated by the National Water and Wastewater Benchmarking Initiative (Earth Tech, 2004), the American Water Works Association, (A WW A, 2004) and AWWARF (Deb at al., 2002). The National Water and Wastewater Benchmarking Initiative, a partnership of a number of Canadian utilities, examines water main breaks as a performance measure but does not examine water main break data in detail. The AWWA database (A WW A, 2004), compiled as a joint effort between AWWA and AWWARF, focuses on generalized data for distribution systems, 17 utility revenue, treatment practices and finances and does not include data on water main breaks except the number of breaks. The survey by Deb et al. (2002) reports on responses from five utilities serving populations fewer than 100,000 and 32 with populations greater than 100,000. Thus knowledge of what small and medium size utilities have is needed. ( Small and medium size utilities need help with making predictions of the remaining service life of water mains and guidance for integrating data acquisition and analysis programs to improve the prioritization of their water main replacements. They lack knowledge regarding how to develop databases that support their efforts in predicting future breaks and what to do with break predictions. In particular, research is needed to determine the extent of data collected, how the data should be compiled and what can be done to support the use of models within these utilities. 1.5 RESEARCH METHODS Institutional factors affecting asset management. The research goals were initially focused on designing and conducting experiments to determine key data available in small and medium size utilities and to assist these data with using break prediction models that are applied on a pipe by pipe basis. As the research progressed, it became apparent that an understanding of utility organizational structures, behaviors and practices, as well as a willingness on the part of organization was needed if this research is to influence practice. This is because water utilities are complex organizations with many staff that have various roles, responsibilities and accountabilities. In addition, these organizations face a significant amount of staff retirements and "institutional memory loss" 18 in the coming years. Practices that are easily taught and documented will be important for business continuity. Furthermore, utilities also cannot rely solely on oral corporate knowledge and need data that are shared among the organization as well as systems that facilitate data sharing. Most utilities are typically comprised of operations, financial, information technology and engineering departments. Operations departments are responsible for operation and maintenance activities, financial departments manage the financial aspects of the utility business, information technology departments maintain and manage data systems, while engineering departments are responsible for the expansion or construction of new network assets. In many utilities, operating departments are responsible for determining which pipes should be replaced. Commonly activities, roles and responsibilities evolve to fit a particular person, their strengths and interests, though these may be inconsistent with the general organizational structure. Data are usually collected and managed on an ad-hoc basis throughout the organization by those who are involved in a specific activity or who have specific interest in the data. No one single person typically coordinates or manages all the data (typically asset management initiatives stall in many organizations due to the inability to dedicate or create a position for this role). Furthermore, within any utility, data may be oral or documented and the format and storage of such data may differ throughout that utility. In addition, budgets constrain activities in these organizations. Decisions and changes in these organizations are usually based on dialogue among and acceptance by those affected. While practices require approval by senior managers, public policy decisions are referred to municipal councils. The prioritization and scheduling of asset replacement projects are usually performed by groups of individuals or in some cases by 19 one individual and submitted to their municipal council for approval. These factors required developing research methods that could be applied to a variety of organizational structure, technical discipline, knowledge, skill, ability and size. N The District of Maple Ridge. The District of Maple Ridge, the case study for the research is a typical water utility. Maple Ridge was incorporated as a municipality in 1874 and experienced significant urban development over the past three decades. It was used to test the practicalities of applying the framework and approaches developed herein. Maple Ridge was interested in learning and improving its practices. But it was a challenge as a researcher to know, not only for Maple Ridge but for utilities in general, what data were available, to identify potential sources of data, to determine how to access the data, to develop approaches to create useful data from various sources, and to develop a framework for associating, accessing, and managing the data across all functional lines respecting responsibility, management and territorial issues. For example, in Maple Ridge, data are "owned" and managed by various departments. Water main break data are collected, stored and managed by those operating crews that respond to breaks. Other pipe related data such as pipe material, diameter and age are stored in the Engineering department. The operations department manager who oversees the superintendents and operating crews prepares and submits a list of replacement projects to the individual tasked as the corporate capital program coordinator who then schedules those projects along with all other capital projects (that are submitted by all other departments). The scheduling respects the available annual funding and is based on the submitted information and the coordinator's view of the future needs of the organization. The scheduled projects are then reviewed by the senior managers and the Chief Administrative Officer (more commonly known as "City Manager" in many municipalities) 20 prior to submission to municipal council for approval. Thus thelise of Maple Ridge as a case study required the research methods that could identify data in multiple sources, validate historic projected and estimated data, determine the basis of collaborated decisions and distinguish policy and practice. Research time frame and tools. The research in Maple Ridge began in early 2002 and was completed by mid-year 2006. It is based on three components: firstly, on conducting a survey of North American water utilities, secondly, on creating, linking and mining data and finally, on developing the framework for using prediction models. The data used in the experiments are actual water system data from the District of Maple Ridge. Analytical software used in the research include Microsoft Access® and Excel®, (Microsoft Office 2000 versions), Adobe Acrobat Reader v6, KANEW by Weston Solutions (based on Microsoft Access® Office 97 version), S-Plus v6 and V7 (Insightful Corporation) and Arcview GIS 3.2A and Arcmap v9 by ESRI. Survey of North American utilities. There were a number of challenges involved in determining the data collection practices of North American water utilities. The best practices indicated by Deb et al. (2002) and NSGMI (2002) are fairly recent observations and there are no published reports regarding industry awareness or use of the recommended best practices. Because the best practices recommended in these documents are not exactly the same, it was necessary to reconcile the differences and develop the questions in the survey questionnaire to reflect both sets of best practices. Another challenge in developing the survey was the need to balance the degree of detail in which I was interested with the effort required of the respondent to voluntarily provide such information. I soon recognized that in order for the survey to fulfill my research goal of identifying the data available to mid-size utilities, the questionnaire would 21' have to be detailed and that the information would not be readily available in one place or with any one person within any organization except within very small organizations. The survey would have to be easy to read, and not discourage utilities from responding to it. It would have to create interest and encourage persistence within those organizations attempting to complete it. The survey would need a layout and format that was easy to read and complete (regardless of the ability of the person completing it), and yet easy for me to compile the responses. While a few electronic formats were examined, it became apparent that the survey would be most easily completed if the forms were sent out as a spreadsheet workbook in order that they could be printed, completed by hand and faxed back or that they could be completed and sent back electronically to me by users (because spreadsheet software is commonly used by utilities). A draft version of the survey was sent to four colleagues, two in Maple Ridge, one in the City of West Vancouver and one in the City of Burnaby for their comments and this allowed me to test methods of electronically compiling the results. Minor editorial and clarification changes were made and incorporated in the final version. The final version included text control (for compiling and analyzing the data) using drop down boxes which also made the survey easier for users to complete. Another challenge was how to distribute the survey and how to encourage people to complete the survey. I considered giving honoraria but decided against that due to costs. To encourage responses, I noted in the distribution that, the results would be shared. I selected email distribution to minimize the distribution costs and delivery time. Not all professional associations I approached to help me distribute the survey were helpful, though most were. Over 400 surveys were directly sent out and about thirty personal requests and specific follow up telephone calls were made. In all, three mailings of the same survey were sent out. The first mailing was sent to members of the Canadian Water and Waste Association 22 (CWWA). The second to the members of the American Public Works Association (APWA) and third to the City Engineers in the Greater Vancouver Regional District. The responses were received gradually and the initial deadline for receiving surveys was extended because of a low number of timely responses. The first set of responses was from the CWWA members. The APWA distribution was made through the APWA to its members who were water resources professionals and this mailing generated a large response from a number of U.S. utilities. The CWWA distribution included the spreadsheet file while the APWA email distribution contained a link to the file. The link affected the submission of some utilities as it added another step to the response process and it created confusion on the part of the respondents. In many APWA cases, only a portion of the survey was returned initially and the utilities had to re-send the survey. While a majority of surveys were completed and sent back electronically, a number of responses were returned by fax. The faxed responses were then inputted manually into spreadsheets for processing. When the responses were received, they were reviewed for errors and text control. The raw data were exported into a Microsoft Access® database. The database was queried and then inputted to a Microsoft Excel® file for data manipulation, analysis and graphing. Most respondents spent about 45 minutes completing the survey. Of the 70 responses received, 11 responses were rejected as incomplete. These usually had only one or more of the seven spread sheets completed due to the confusion caused by the linking requirement or did not realize that the file contained several spreadsheets. Those surveys were sent back and follow up telephone calls were made where appropriate. The design and development of the survey began in 2003 and the surveys were finally distributed, compiled and analyzed in 2004. A manuscript reporting on the survey, Chapter 2, was submitted to Journal AWWA in 2005 and was published in July 2006. 23 Data creation. The creation of data described in Chapter 3 was inspired by the survey results that showed a gap between data presently collected and those available (see Figure 2.2 of Chapter 2), as well as by the fact that utilities noted the locations of different data within their various departments. It was also driven by the discovery that the data set purchased from the City of Seattle which was thought to have breadth and history was actually quite incomplete. After repeated correspondences and a site visit to Seattle, it was determined that the data from Seattle could not be used for the experiments. I then contacted four municipalities in the Greater Vancouver Regional District (Burnaby, Richmond, West Vancouver and Coquitlam and reviewed the breadth and history of their data. Yet again I determined that the data provided by these municipalities was insufficient. Thus I decided to use Maple Ridge data, though limited, for input for the experiments because of my ability to access these data, and to collect data from other internal and external sources for this work and for other municipal projects beyond the focus of my research. Because of its limited data, Maple Ridge is representative of many utilities and is ideal for demonstrating the concepts of finding and gathering data within and external to an organization and of constructing and linking data for asset management. The Laity View area of Maple Ridge was selected to be a case study area because it represents an urban area and experienced construction practices typical for the municipality. It comprises 13 percent of the utility's pipe network. Determining if data were available required meeting with and exploring activities of various work units throughout the organization. The research effort of creating data was significant and the approach and techniques used in doing so are presented in this thesis. These tools are general and may be used by all utilities. New data constructed for Maple Ridge in this research include soil, surface 24 material, water pressure and flow in pipes, bedding and backfill material and traffic loading. Additional activities were carried out to verify data including field surveys of soil and preliminary pipe sampling and analysis. With different departments collecting and storing data, the process of scrubbing data was not a straightforward exercise. During the process of linking the data, inconsistencies and missing data in the various databases were discovered (e.g., missing material or diameter data or discrepancies between the hydraulic model and GIS data). Typically, I had to delve into the details of those databases to see which data were more current or in some cases, to verify those data using another source. For example, if a pipe had different ages in the different databases (e.g., the hydraulic model and the GIS), I . checked the age in each database to ensure that an input error had not been made, and checked the capital works program to see if the pipe had been replaced but that the age of the new pipe had not been updated in one of the databases. I also reviewed the data for consistency with the knowledge I have about Maple Ridge's practices (e.g., policies such as that there should not be asbestos cement pipes installed after 1985). Where discrepancies existed, as-built drawings were reviewed to determine if data were correct. It was unfortunate that data were sometimes lost or deleted (e.g., at Maple Ridge when a pipe is replaced, all records pertaining to that pipe are typically eliminated). In some cases, I was able to verify some data for these pipes, but I was not successful in most cases. As well, the research required the development of a technique to facilitate creating on-demand databases for analysis and updating. This was required to address the need to obtain data for experiments and to be able to update the original database but also to respect the ownership of the data and the organizational structure. 25 An example of how creating data. How pipe break records were created for this research and how they are planned to be managed in the future by Maple Ridge is an example of the activities and efforts described above. The break records for Maple Ridge are not stored in any one location but in various files and physical locations and with different staff of the operations department in the operations building (which is separate from City Hall and the engineering department). Because of my experiences with other utilities, I was certain that Maple Ridge had some records on water main breaks. In the course of conversations with various staff over a number of months, I made many inquiries and requests of various operations and engineering staff (with suggestions of who may have the data or where they may be stored, e.g., the archive files, in someone's previous files or i in work request logs), I was able to obtain all water main break data for the period 1983-2004. Information regarding water main breaks for certain years was only available from field staff break forms, in paper form, and for other years, it was available on printed computer reports stored in separate files. Once all the records were obtained, the data were reviewed and scrubbed to separate out records of service connection failures (e.g. saddle failures, leaking copper service lines) from those of pipe breaks. Subsequently, an electronic database of the breaks was then created by inputting the data by hand into spreadsheets, identifying the specific identification number (see Chapter 3) of the pipes that broke by locating those pipes on maps using the location data recorded, and adding the GIS pipe identification number as a pipe attribute. The spreadsheets were then exported to a Microsoft Access® database where a new input form was created to allow the addition of future break data from revised field record forms (as discussed in Chapter 3). The storage location of the new water main break database was identified and mapped by those responsible in the engineering department for data warehousing. The database is available 26 to the operations department staff for searching, finding and sharing the raw data. The break data for the Laity View case study was then extracted from the new water main break database for the experiments described in Chapter 4. While the creation of the database became more mechanical once the paper records were obtained, creating the database required discussion and resolution regarding i) future data collection resources, efforts, quality control and training, and ii) database building, ownership, management and maintenance. A number of discussions and negotiations transpired between the engineering, operations and information technology departments regarding the goals and potential results resulting from pipe break data collection and analysis, the roles and responsibilities for data collection, storage, management and access, and the contribution by each department of staff time and finances. These discussions and my desire to improve data warehousing practices, yet respecting data ownership issues within Maple Ridge, inspired the development of the distributed but related databases for on-demand analysis described in Chapter 3. This portion of the research took from early 2004 until summer 2005. Framework for using prediction models. This portion of the research experiments was initially based on analyzing the created data to determine if there were key data that are currently available to most small and medium size utilities that could be used in prediction models for improving asset management. The Laity View data set with break history from 1983 to 2004 was used for the experiments and is described in Chapter 4. ^ The experiments were designed to use statistical deterministic time-exponential and time-linear regression models, survival analysis models and KANEW (Deb et al., 1998). Time-linear and time-exponential models can be used by small to medium size utilities because these utilities typically have the institutional capacity to use spreadsheets or simple 27 statistical software packages. The survival analysis and KANEW (Deb et al, 1998) techniques were applied to determine the suitability of those applications for more adept utilities, even though most utilities may not have the ability to use these more sophisticated techniques or operate the more complex software required. While not reported in Chapter 4, the application of survival analysis was not successful due to the low number of breaks in the Laity View case which is a young system, the limited breaks in older pipes and the loss of data regarding pipes that were replaced. Survival functions were derived for all the pipe groups described in Chapter 4 but were incomplete for the most part because survival analysis is dependent in knowing the failure history across a large range of pipe ages. There were not enough failures in the older pipes to determine a complete survival function for any type of pipe. As such, it was concluded that survival functions cannot be expected to be commonly used by small and medium size utilities with limited data. Similarly, the use of KANEW is constrained by;the limited break history to derive cohort specific survivals. KANEW, a Microsoft Access® 97-based program was developed based on interviews with utility staff regarding their experience and can be very conservative. I had a number of discussions with the developers of KANEW regarding the program, its development, use, limitations and application. Again, because of the incomplete failure history of pipes in Maple Ridge, KANEW could not be applied with success and was not reported on in Chapter 4. The strength of KANEW is in allowing a utility to develop replacement budgets. Using expected pipe lives from the experiences of other utilities has some value for developing network replacement budgets but is not useful for pipe by pipe replacements. 28 The experiments reported on in Chapter 4 were conducted during summer and fall 2005, and the analysis was completed in 2006. Contribution of research. This thesis offers a number contributions for small and medium size utilities and future researchers. It assesses the data collected by utilities, it develops and demonstrates how to create and link data for asset management in general and it explores the underlying causes of water main breaks. Finally, the research develops a framework for utilities to use break prediction models on a pipe by pipe basis to improve their method of prioritizing the replacing of water mains for asset management and to inform their future data acquisition and storage programs. Managers may gain insights from the lessons learned from conducting this research on organizational considerations such as institutional memory, staff training, engineering and operations responsibilities and strategic thinking for asset management. In particular, researchers will benefit from this thesis because it identifies which data are available for developing future asset management tools and how to access and construct water main break data. 29 1.6 REFERENCES Ahammed, M. and Melchers, R.E., 1994. Reliability of underground pipelines subjected to corrosion. Journal of Transportation Engineering, 120:6: 989-1003. . ASCE (American Society of Civil Engineers), 1999. American Society of Civil Engineers report card and issue briefs. Public Works Management and Policy, 4:1: 58-76. AWWA (American Water Works Association), 2004. WATER:\STATS - The Water Utility Database. 2002 version [CD-ROM]. American Water Works Association, Denver, CO. 80235. AWWA and EES, Inc., 2002. New or repaired water mains. Available on-line at http://www.epa.gov/safewater/tcr/pdf/maincontam.pdf., Accessed February 12, 2006. Canada. Statistics Canada, 2006. The age of public infrastructure in Canada, by Valerie Gaudreault and Patrick Lemire. January 2006. (Catalogue No. 11-621-MIE2006035). Ottawa, Minister of Industry. Davis, D.N., 2000. Agent-based decision support framework for water supply infrastructure rehabilitation and development. Computers, Environment and Urban Systems, 24:2000: 173-190. 30 Deb, A. R., Grablutz, F.M., Hasit, Y.J., Synder, J.K., Longanathan, G.V. and Agbenowski, N., 2002. Prioritizing Water main Replacement and Rehabilitation. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Deb, A.K., Hansit, Y. J. and Grabultz, F.M., 1998. Quantifying Future Rehabilitation and Replacement Needs of Watermains. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. DeSilva, D., Davis, P., Burn, L. S., Ferguson, P., Massie, D., Cull, J., Eiswirfh, M. and Heske, C, 2002. Condition Assessment of Cast Iron and Asbestos Cement Pipes by In-pipe Probes and Selective Sampling for Estimation of Remaining Service Life. Proceedings of the 20th International No-Dig Conference, Copenhagen, Denmark. Dingus, M., Haven, J. and Russell, A., 2002. Nondestructuve, Noninvasive Assessment of Underground Pipelines. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Doleac, M. L., Lackey, S. L. and Bratton, G. N., 1980. Prediction of time-to-failure for buried cast iron pipe. Proceedings of A WW A Annual Conference, Denver, CO. Earth Tech, 2004. National Water and Wastewater Benchmarking Initiative Final Results -June 2004. Earth Tech, Burnaby, BC. 31 FCM, 2001. Backgrounder: Protecting Canada's Quality of Life. Federation of Canadian Municipalities, (http://www.fcm.ca/infra/throne-a.html). Accessed December 2002. Grigg, N. S., 2004. Assessment and renewal of water distribution systems. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Kettler, A. J. and Goulter, I. C, 1985. An analysis of pipe breakage in urban water distribution networks. Canadian Journal of Civil Engineering, 12:2: 286-293. Kleiner, Y. and Rajani, B.B., 2001. Comprehensive review of structural deterioration of water mains: statistical models. Urban Water, 3:3: 131-150. Kleiner, Y. and Rajani, B.B., 1999. Using limited data to assess future needs. Journal AWWA, 91:7:47-62. Kleiner, Y., Rajani, B. B. and Sadiq, R., 2004. Management of failure risk in large diameter buried pipes using fuzzy-based techniques. 4th International Conference on Decision Making in Urban and Civil Engineering, Porto, Portugal. Mailhot, A., Pelletier, G., Noel, J.-F. and Villeneuve, J.-P., 2000. Modeling the evolution of the structural state of water pipe networks with brief recorded pipe break histories: Methodology and application. Water Resources Research, 36:10: 3053-3062. 32 Male, J. W., Walski, T. and Slutsky, A. H., 1990. Analyzing Water Main Replacement Policies. Journal of Water Resources Planning and Management, 116:3: 362-374. Myers, C, 2001. Rural Areas gain high tech infrastructure plan through partnership, Utility Executive, 3:2: 5-8, Water Environment Federation, 601 Wythe Street, Alexandria, VA, 22314-1994. NGSMI (National Guide to Sustainable Municipal Infrastructure - Inffaguide), 2002. Deterioration and Inspection of Water Distribution Systems. Infraguide - Potable Water. Ottawa, ON, Canada. National Research Council of the National Academies - Committee on Public Water Supply Distribution Systems: Assessing and Reducing Risks, 2005 . Public Water Supply Distribution Systems: Assessing and Reducing Risks - First Report. O'Day, D. K., 1982. Organizing and analyzing leak and break data for making main replacement decisions. Journal AWWA, 74:11: 588-594. Ontario PIR. 2005. WATERTIGHT: The case for change in Ontario's water and wastewater sector. Report of the Water Strategy Expert Panel. Ontario Ministry of Public Infrastructure Renewal. Rajani, B.B. and Kleiner, Y., 2001. Comprehensive review of structural deterioration of water mains: physically based models. Urban Water, 3:3: 151-164. 33 Rajani, B.B. and Makar, J., 2000. A methodology to estimate remaining service life of grey cast iron water mains. Canadian Journal of Civil Engineering, 27:6: 1259-1272. Rajani, B.B. and Tesfamariam, S., 2005. Estimating time to failure of ageing cast iron water mains under uncertainties. Water Management for the 21st Century, University of Exeter, UK., University of Exeter, UK., 1-7. Sacluti, F., 1999. Modeling Water Distribution Pipe Failures Using Artificial Neural Networks, Master of Science, University of Alberta, Edmonton, Alberta. Savic, DA. and Walters, G.A., 1999. Hydroinformatics, Data Mining and Maintenance of UK Water Networks. Anti-Corrosion Methods and Materials, 46:6: 415-425. Shamir, U. and Howard, C, 1979. An analytic approach to scheduling pipe replacement. Journal AWWA, 71:5: 248-258. USEPA, 2005. 2003 Drinking Water Infrastructure Needs Survey and Assessment, United States Environmental Protection Agency, Drinking Water Division, Washington, DC. USEPA, 2002. Clean Water and Drinking Water Infrastructure Gap Analysis, Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. 34 USEPA, 2001. Drinking Water Infrastructure Needs Survey. Second Report to Congress. United States. Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. USEPA, 1999. Drinking Water Infrastructure Needs Survey. United States Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. tJSFHWA, 1999. Asset Management Primer, Federal Highway Administration, US Department of Transportation, Office of Asset Management, 400 7th Street, S.W. Washington D.C. 20590. Vanier, D.J., 2001. Asset Management: "A" to "Z", American Public Works Association Annual Congress and Exposition - Innovations in Urban Infrastructure Seminar, Philadelphia, U.S. September, 2001. 1-16. Vanier, D.J. and Rahman, S., 2004. Municipal Infrastructure Investment Planning (MIIP) MlfP Report: A Primer on Municipal Infrastructure Asset Management. Report B-5123.3, National Research Council Canada, Ottawa, ON. Walski, T.M., 1982. Economic Analysis of Water Main Breaks. Journal of Water Resources Planning Management Division, 108:3: 296-308. Wood, A. and Lence, B.J., 2006. Assessment of Water Main Break Data for Asset Management. Journal AWWA, 98:07. 35 Figure 1.1 Twenty-year total per household infrastructure cost estimates for different sized water systems $3,000 Large Medium Small systems systems systems Source: USEPA 1999 Drinking Water Infrastructure Needs Survey 36 CHAPTER 2 ASSESSMENT OF WATER MAIN BREAK DATA FOR ASSET MANAGEMENT A version of this paper was published in the July 2006 issue of Journal A WW A. The paper is titled as Assessment of Water Main Break Data for Asset Management by A. Wood and B.J. hence. 37 PREFACE Much research has been conducted over the years regarding the development of water main break models. But as a manager, I found that none of the medium size utilities with whom I was familiar, used break prediction models to prioritize their pipe replacements. Instead of developing more models, I wanted to find a way to use break prediction models and existing data in asset management and in prioritizing specific pipes and to help other utilities to do likewise. As a manager, I knew that the utility that I worked for did not have a large amount of data which would limit their ability to use models and I wanted to know if the amount of data available was also a hindrance for other similar size utilities. In addition, as a manager, I was interested in comparing our data collection performance with other utilities and in improving our collection practices. Though best practices for water main break data collection were recommended in 2002 (see National Guide To Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb et al, 2002), I could not find reports regarding the use of the best practices, although Deb et al. (2002) had surveyed some large utilities (see Appendix A for a summary of the size of utilities surveyed and the questions posed). Therefore, I developed a survey, the results of which would serve as a foundation for my research and distributed it to small and medium size utilities in order that I might use it to guide my research in developing tools and techniques for improving utility asset management. I have received feedback on the results such as it is timely and provides managers with a benchmark of what is currently being done by utilities against best practices. Some of the results were presented at the 2006 British Columbia Public Works Association Conference, Qualicum Beach, BC and questions by attendees include: which data are most 38 important to collect, what methods should utilities employ to collect data and how should data be stored and analyzed for asset management? The raw data regarding survey responses are appended to this thesis. 39 2.1 INTRODUCTION Water main breaks can result in loss of water to key businesses and critical facilities, and lead to damage of infrastructure. Such events highlight the deterioration, of aging infrastructure, which is at the forefront of policy and program discussions between national and provincial or state governments. The need for aging water main rehabilitation is increasing, the costs of repairs and replacement can be high, and the impact to customers potentially significant (USEPA 2001). For utility managers, collecting, recording and monitoring water main breaks is important, not only because such events may result in significant public impact and destruction of private property, but also because the data obtained from breaks may provide insights for management of the system as a whole. This information is important in developing the tradeoffs between expenditures and the level of service provided, and in managing rehabilitation programs to achieve a desired level of service. This paper reports on the results of a 2004 survey of water utilities in the U.S. and Canada that determines the degree and types of field information that are currently collected and other data available within different departments of the utility regarding water main breaks. Queries were posed with reference to the best practices recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb et al, 2002). The survey results identify the data that are deemed important and the approach undertaken for recording and storing these data by different sizes of utilities. The level of confidence that respondents have in some of the data being collected and their level of comfort with the amount and types of data collected, is also identified. The survey provided the utility managers with a list of data recommended as best practices against which they could assess their own data collection practices. The survey 40 findings are important for the broader infrastructure management community, because knowledge of the data available underpins the development of relevant data acquisition and storage strategies and the advancement of asset management approaches. They also provide utility managers with a basis for gauging their practices relative to those of other utilities. The survey was developed considering other recent water utility surveys, including those initiated by the National Water and Wastewater Benchmarking Initiative (Earth Tech, 2004), the American Water Works Association, (AWWA, 2004), and AWWARF (Deb et al, 2002). The National Water and Wastewater Benchmarking Initiative (Earth Tech, 2004), a partnership of Canadian utilities, considers water main breaks as a benchmarking measure but does not examine water main break data in detail. The AWWA (AWWA, 2004) database, compiled as a joint effort between AWWA and AWWARF, is based on a 2002 and 2003 survey of 337 small, medium, and large U.S. and Canadian utilities and focuses on a broad range of potable water distribution characteristics including general information regarding service population, pipe material, valves, fire hydrants and flushing, finished water storage facilities, corrosion control, customer metering, water, supply auditing, and leakage management. It includes the reported number of water main breaks of the utilities surveyed but does not include data recorded on water main breaks. The survey reported on in this paper builds on previous AWWA work (Deb et al, 2002), but queried in greater detail the data that are collected by utilities, such as failure causes and the general physical characteristics of the pipe and soil, and had more responses from small to medium size utilities. In this survey, 37 utilities serving populations below 100,000 and 22 utilities serving populations greater than 100,000 responded, compared 41 with five utilities serving populations below 100,000 and 32 utilities serving populations greater than 100,000 that responded to the 2002 survey. 2.2 DESIGN OF THE SURVEY The survey was designed for ease of completion and compilation into a data base for analysis. Users were able to print the survey and complete it by hand, or complete it electronically using personal digital assistants and laptops. Microsoft Excel® work sheets were chosen due to their common use. Design of the questionnaire. Based on a review of the literature (NGSMI, 2002; and Deb et al, 2002), the questions posed in the survey were grouped into the following six Categories: • General information such as time of break, customers affected, response personnel and equipment, and cost of repairs; • Location data such as nearest property address and geographical coordinates; • Physical data such as pipe material and depth of cover; • Environmental data such as depth of frost and soil and air temperature; • Failure information such as type and cause of failure; and • Repair information such as components replaced, repaired or installed. The survey respondents were queried regarding how they currently store data related to water main breaks, their estimated level of confidence in selected data, their comfort with the amount of data and the percentage of events for which data are recorded. They were also given the opportunity to list the additional data elements that they collect. 42 Survey distribution. The survey was distributed by direct email to members of the Canadian Water and Waste Association (CWWA), and members of the American Public Works Association (APWA) who identified in,their member profile that they were either directly responsible for, or interested in, water distribution and treatment. 2.3 SURVEY RESULTS Seventy responses were received and after a review of each response for completeness, 59 surveys were deemed complete and analyzed herein as respondents. The general feedback from the respondents was that, while water main break data collection has been identified as important and a strategic initiative (Grigg, 2004), utilities are just beginning to collect data comprehensively, collection practices vary widely, and most utilities view their efforts as evolving. The sizes of the respondents range from one utility serving 2,600 people to two utilities serving a population of over one million. Most of the respondents serve populations of between 50,000 and 500,000. The service populations of the respondents are shown in Table 2.1. General water main break information. General water main break information includes the date and time that a break is reported, the response time and resources used to address the break, and the impact on customers. The percentages of all respondents that collect different general information are also shown in Figure 2.1. All the respondents indicated that they record the date and time when a break was reported. The data related to resources expended on the repair are generally reported in terms of the amount of crew hours spent responding to a reported break. However, only about 70 percent of respondents 43 record labor, materials, and equipment costs and less than 30 percent record the cost of property damage or the effect of the break on customers. From a management perspective, water main breaks are a liability and thus clear documentation of a report of the break or the request for assistance, and the utility response to the information, are required should claims be filed against the utility. Lengthy break response times may result in significant private or public property damage and can be costly and embarrassing for utility managers. Only 70 percent of respondents record costs. While this is surprising, it may be due to the difficulty associated with recording some costs (e.g., some costs may only be determined at a later time, such as costs from a claim or for re-paving a road). Some costs may be determined using other sources of data such as through payroll or work order systems. In addition, while managers need to track expenditures for cost control, accountability of resources and planning for future management decisions, they may use different reporting systems and may be required to report the costs for responding to water main breaks only on an aggregate basis. The survey responses show that few respondents record the effect of the breaks on customers in terms of damages incurred and the number or type of customers that experience a water stoppage as a result of the break or repair. However, the data may be available or calculated if necessary by determining if and where valves were operated to shut off water to sections of a water main and if services were turned off, how long it took to restore the water service. This would allow a utility to calculate the number of customers affected and the length of the interruption as a surrogate for the direct impact on customers. In addition, claims data were also referenced as a potential source of information. 44 Location data. Location data include the nearest property address, cross street, and geospatial coordinates. The percentage of respondents that record various location data is shown in Table 2.2. Location data along with maps can aid response crews in determining the time and effort of response, the size of main to be repaired, the supplies needed to repair the break and the location of valves that need to be closed in order to facilitate repairs. In addition, recording the location of the break allows future managers to determine if there are repeated breaks in a particular section of water main and may help to assess if rehabilitation or replacement of the main is required. Most respondents record the nearest property address and cross street name, but few record additional information that can identify the exact location of the break. The survey responses suggest that address and cross street data come from the initial report of the break (i.e., the customer) but are not revised when the actual break location is determined. The fact that data are collected suggests that dispatching a response is the primary motivation for the collection, but that collecting location details that would assist future break analyses is not as high a priority. The lack of actual break location data may suggest that either respondents find it difficult to record and store spatial data or do not consider the specific location of a break important. The fact that few respondents record whether isolation valves are operated can . make it difficult for utility managers to determine the number of customers affected by a break and a repair. The use of Geographic Information Systems (GISs) for storing information and visualizing break data, and the low cost of Global Positioning System (GPS) survey equipment, may change this practice in the future. GIS-stored location information may ultimately enhance decision-making regarding rehabilitation and replacement. 45 Physical data. The data typically classified as physical include: the age, material and characteristics of the pipe, and surrounding soil. The results reported in Figure 2.2 show that most respondents collect a narrow variety of physical data. Most respondents record data that include pipe diameter, pipe material, water service type, cover depth, and whether the surface is a roadway or other surface. The lack of physical data collected by respondents may be attributed to the fact that -water main breaks are typically treated as emergency situations in which the goals are to contain damage, repair the break, and restore lost water service to customers or that some information on the physical characteristics of water mains may be available elsewhere in the data warehouse of the utility, such as in as-built records, or in GIS form. When other data sources are accounted for, the percentage of respondents that possess the various data elements increases. Analysis of the survey results suggests respondents and utilities can be classified, according to the richness of the data recorded, into four groups. These groups include utilities possessing: i. basic data consisting of pipe size and material; ii. basic material and diameter data plus limited information such as age of the pipe, use of the surface at ground level, operating pressure, and type of pipe joint. If pipes in a network are failing at joints, details about joints can be used to develop a strategy to anticipate, prevent, and repair breaks; iii. in addition to the data described in (i) and (ii), data on construction of the water main; and iv. in addition to data described in (i), (ii), and (iii), detailed pipe information such as pipe wall thickness and pipe fracture toughness (typically as a result of pipe testing). 46 Failure causes and modes. An important aspect of water main break data is the nature of the failure. Though the mode of failure cannot be definitively correlated with specific causes of the failure, they may indicate a failure mechanism and suggest a cause of failure for analysis by asset managers. The survey queried the respondents regarding eleven common failure modes and the responses to the survey show that 85 percent of respondents record leaking joints, valves, hydrants and service connections and between 7Q and 83 percent record the remaining seven failure modes. These failures modes are: leaking joint, leaking service connection, leaking valve, leaking hydrant, longitudinal break, blow-out, split bell, corrosion pit hole, curb stop failure, tap failure and failed blow-off (i.e., air release valve). Respondents could also select an "other failure modes" category in the event that the failure is different from the list of failure modes provided. No other modes were reported. While utilities are able to determine and record the failure mode, only 25 percent of respondents record the cause in 100 percent of their records, 37 percent of the respondents record the cause in 75 percent of their records, and only 40 percent of respondents record a cause in at least 50 percent of their records. This suggests that few respondents have a consistently high level of information on the causes of breaks to their water mains. The different causes of failure and the percentage of respondents that record a specific cause of failure are shown on Figure 2.3. The analysis of data shown in Figure 2.3 suggests that it is difficult for respondents to determine the cause of failure. Managers may find that in'their organization, breaks are treated as emergency situations and the utility staff at the scene of the break focus on controlling the extent of collateral damage from the break. Also, some specialized 47 engineering background is required to confidently determine the cause of breaks under response situations. Repair activities. While it may be difficult to determine and record the cause of breaks, the response regarding the recording of repair activities is high. The data in Figure 2.4 show the percentage of respondents that record a specific repair activity. Repair activities typically include repairing clamps and joints or replacing sections of pipes, valves, hydrants, and connections. This could be explained by the fact that utilities can more readily record response actions in the field than determine the cause of a water main break. Environmental data. Environmental data related to water main breaks include information such as air temperature, soil acidity, and moisture, and other antecedent conditions. The environment is important for water mains as the National Research Council of the National Academies (2005) identifies that water main breaks and their repairs are also potential gateways to contamination of the water distribution system. The survey responses show that very little environmental information is collected by respondents. Only 12 percent of the respondents record environmental data for all of their water main breaks and only 27 percent record any environmental data at all. In fact, environmental data is the least recorded group of information for all of the respondents. If a respondent collects environmental information, most likely it is information on the depth of frost (27 percent of respondents). Less than 10 percent of the respondents take soil samples. This is surprising as it is well publicized that there is a significant relationship between external corrosion of water mains and soil conditions. Figure 2.5 shows the percentage of the respondents that record a given type of environmental data, or can obtain the data elsewhere either within the utility or from sources external to the utility such as 48 other agencies. In some cases, a significantly greater percentage of respondents indicated that more data are available from other sources than are recorded. Hydraulic models. When asked whether the utility has a water model and whether it is used, 78 percent of the respondents report having a hydraulic model. Only 71 percent of the respondents use these models for some purpose. The most popular uses of models in descending order of popularity (i.e., based on the percentage of respondents) are for capital planning (69 percent), development planning (61 percent), operations (56 percent), and maintenance purposes (42 percent ). Confidence in data. A key survey question regards the level of confidence that a utility places on the physical data that it collects. Utilities were asked if their confidence in the data collected is high. The intent was to determine not only the level of confidence but also if this level varies by data type. The confidence of the respondents in selected parameters is summarized in Figure 2.6. As shown in Figure 2.6, at least 72 percent of utilities have high confidence in pipe diameter and material, but only 48 percent are highly confident about the year of installation. Level of comfort with the amount of data collected. Utilities were asked to indicate their level of comfort with the amount of data collected. Sixty-four percent of the respondents were comfortable with the amount of data collected. However, 22 percent were not comfortable and 14 percent had no opinion regarding their level of confidence in the data collected. The level of comfort with the amount of data collected varies with utility size. Between 50 and 60 percent of respondents with service populations between 10,000 and 100,000 and approximately 70 percent of those serving between 100,000 and 500,000 are comfortable with the amount of data collected. Both respondents with service populations greater than one million were also comfortable with their data. 49 Sources of data available to utilities. The additional sources of physical data for the respondents are summarized in Table 2.3. The number of respondents that indicated that data were available from the other sources is also listed. For example, 46 percent of respondents identified other sources for obtaining normal operating pressures. Managers may use this table to consider similar data sources within their organization. They may also be encouraged to determine if this information is available from other sources that are not as yet identified. Storage of data. Different sizes of utilities have different means of data storage. The use of a GIS as a data management system is greatest for respondents that serve a population of between 50,000 and 100,000 (four of these ten communities use GISs), and archival records are the predominant source of data for utilities that serve a population of between 10,000 and 50,000 (eight of these fourteen communities use archival records). Of the fifteen respondents that serve a population of between 100,000.and 500,000, seven use archival records (such as as-builts and other paper-based historical records) as a source of water main break data, and four use GISs. Statistical confidence of the survey results. Although the number of respondents is low, the survey results can be used to draw some inferences regarding the practices of water utilities in general. A statistical analysis of the significance of the survey sample for providing observations for the general Canadian-U.S. utility population was undertaken. The accepted measure range of confidence limits is calculated using Wild and Seber (2000), to be between 11 percent and 13 percent for this survey. This is to say that if 63 percent of the respondents indicate that they record when water service is restored, we may expect that more than 50 percent (i.e., the lower confidence limit) but less than 75 percent 50 (i.e., the higher confidence limit) of all water utilities in the general population would record the same. Appendix B summarizes the statistical confidence of these results. 2.4 DISCUSSION AND RECOMMENDATIONS While all respondents are collecting data, they clearly do not collect all data suggested by best practices recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb at al, 2002). However, based on the responses it is apparent that data acquisition is evolving and that there is a strong interest in comparing their practices with others and with best practices. Feedback from the respondents indicates that the average respondent spent about 30 to 45 minutes completing the survey. This is in addition to the time that respondents spent determining the appropriate people within the organization to complete the survey. Such decisions are common among larger organizations where responsibilities for data collection, management and analysis are shared. The amount of data that respondents and utilities in general have can be categorized into general classes of data richness. Based on the survey results, it is suggested that four classes exist. These are: expanded, intermediate, limited and minimal data set classes. As shown in Figure 2.7, the proposed data set classes are based on the breadth and record length of data irrespective of the traditional data categories such as break, pipe inventory, and operational data because the survey results indicate that many respondents collect data in varying amounts in each of the traditional data categories. Expanded data sets are comprised of information, and records over a long period of time. These may include: an inventory of pipes that is correlated to pipe information such as diameter, material, year of installation, type of joint, surface cover, type of failures, probable cause of failure, type of 51 repair, pipe testing information such as pipe wall thickness at time of break, type of pipe lining, soil testing information such as corrosivity, and pH. Intermediate data sets are comprised of information on inventory and pressure zones, a break history of some length of time, and some amount of information regarding pipe diameter, material, age, exterior surface condition, and installation, surface cover, and pipe protection. A utility possessing information with this degree of data richness has generalized descriptions regarding the type of failures encountered and probable cause of those failures. This case is similar to the expanded data case but does not include pipe and soil testing data although installation details may be available. Limited data sets have only limited information on the pipe network, surface uses and loads. This would be an inventory, information regarding pressure zones and generalized problem areas, a break history of a minimal length of time, and a nominal amount of information regarding pipe diameter, pipe material, year of installation, and surface.cover. Under this case, the information comprising the intermediate data set is significantly reduced. Minimal data sets are comprised of pipe lengths, identification of pressure zones, generalized problem areas, and nominal information regarding pipe diameters. Given that asset management is becoming important for utility managers, more attention and effort should be given to improving data collection in the following areas: Collect for the future. Managers need to consider the long view when assessing their data collection strategy. Water main break data will be useful for future asset managers who will have to make difficult investment decisions regarding which pipes to replace. Estimates for the U.S. (USEPA, 2002) suggest that the capital needs for drinking water in the U.S. for the period between 2000 and 2019 range from 154 to 446 billion U.S. dollars. There are no similar estimates for Canada. Future managers will need to make wise 52 decisions to take into account not only the ability of the public to afford replacing these pipes but also the scarcity of human resources and capital to actually perform the work. Because the practice of infrastructure asset management will evolve, data requirements for making critical decisions are important but can also be expected to grow. Thus, the data collection strategy for utilities should be to gather the data recommended as best practices keeping in mind that much of the data will be useful for future management decisions rather than for today's decisions. It will likely take some time to establish a data record of sufficient length to support such decisions. Many managers may resist this approach because they may prefer to know what is needed and why before embarking on data collection programs. In some cases, such a strategy may be undertaken with little additional effort. For example, a utility may be able to accumulate more information by collecting additional information as part of existing tasks, such as a description of the failure mode, bedding, backfill and depth of cover. To facilitate asset management, data collection efforts should also include, where possible, the types and causes of breaks, and the physical data of all mains so as to provide a deeper understanding of the failure modes and mechanisms. Record data effectively. For collection practices to be effective, utilities need to record data continuously, consistently, and accurately. Data should be collected on all -breaks to ensure that the records portray the state of the whole system. Missing records may create a level of uncertainty regarding the records that are collected and undermine the confidence in conclusions drawn from technical analyses and lead to organizational frustration. In order to record data in a consistent manner, the use of standard operating practices, procedures, and forms as well as training for those who collect data are important. If a 53 utility is already collecting data, a review of all data collection forms is useful to ensure consistency in methods as well as to expand the data collected if appropriate. An audit of the records for quality and accuracy should be undertaken periodically. A suggested model for an audit exercise is the formation of a quality control group within the utility to audit the data and ensure quality across the organization. This group could also develop guidelines of practice for educating staff. The most efficient approach for recording data will vary among utilities and should reflect business and work flow processes and organizational structure of the utility. Many utilities may have different departments, such as laboratory, maintenance, technical, and financial departments, and response teams that collect or generate information. In such cases, it may be beneficial to bundle or group data based on organizational departments and develop a process for ensuring that the data are collected, assessed for relational links related to other data, recorded and stored in an effective manner. For example, data forms could be circulated among selected departments within the organization to collect input data comprehensively and to improve the sharing of information among departments, or depending on the data and organization, the data collection process between departments could be independent but linked or related via an asset index. The storage of data can be undertaken in various ways (see Table 2.3), or in a comprehensive data warehouse. Some utilities prefer a comprehensive data warehouse to facilitate the synthesis and visual representation of data and view GIS technology as the ideal tool for this purpose. Alternate sources of data and relating data. The survey results clearly show that alternate sources of data related to water main breaks exist in many organizations. Table 2.4 summarizes the potential range of information sources identified in this work for different 54 data elements. For example, current and future capital project designs, as-builts, or asset pro-formas can be used to capture information such as specifications, test pit logs, and inspection records. Utilities may wish to consider the concept of a data web, which may be thought of as a relational structure of the utility's data sources, including break records, maintenance reports, pipe and soil samples, customer information, archival systems, hydraulic model output, capital rehabilitation planning data, GISs, and maintenance management systems. The links between these data sources, support the web, and are keyed to some index of the asset, for example, a pipe identification number. The use of the links reduces the need to convert data when a new information management system is developed and implemented and facilitates the synthesis of existing corporate data for specific analyses. A data web is not a data storage application nor software, but a concept of relating or linking data and separate data sources in an organization to each other. While the development of a data web for a utility reduces the need to convert and store data, the addition of data elements in each data system, such as the asset index (e.g., corresponding to a pipe identification number) may be required. A well-designed data web would enhance utility management, particularly for small communities who cannot afford wholesale data management system implementation, conversions and upgrades. Design considerations may include identification of the architecture of the web, the characteristics and number of required links, and the appropriate connections between the elements to be "webbed". Tacit information or knowledge that is currently not recorded can also be linked within the web. Decision support tools. In developing decision tools for supporting asset management, utilities should consider a range of systems that are flexible and easily accessible for both present and future uses. The systems should be reliable and maintained 55 over time. Data that have been collected are useless if the decision makers within the organization cannot access them or are unaware of their existence. Analysis of the survey results suggests that decision support tools for prioritizing the replacement or rehabilitation of water mains should be tailored to the degree of richness of the data available to a utility. Utilities that have a minimal amount of data cannot use sophisticated tools such as physical (Rajani and Makar, 2000) or statistical pipe deterioration models (Shamir and Howard, 1979; Jacobs and Karney, 1994; Andreou et al, 1987; Kleiner and Rajani, 1999), or life cycle costing. These tools, and most recent research in water distribution asset management have focused on utilities that possess large amounts of data and more research is required to develop robust approaches for utilities with limited or minimal data records. These tools should also be flexible enough to adapt as utilities increase the amount and types of data collected. 2.5 CONCLUSIONS Studies, even as recent as in 2004, have identified the need for standardized main break databases and terminology and continued research regarding database development as strategic for informed infrastructure management (e.g., Grigg, 2004; O'Day et al, 1986). The results of the survey reported herein indicate that water main break data collection is evolving, that industry practices do not match best practices recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb et al, 2002) at this time, and that most respondents recognize the need for a strategy for data quality improvement. The data-related challenges that all utilities face include difficulty in mobilizing financial and human resources, absence of historical data, lack of knowledge of 56 current organizational practices, low reliability of previously collected data, difficulty in prioritizing data collection, and the need to develop effective data storage programs. While it is generally accepted that the use of reliable data regarding asset inventory and condition will enhance the management of municipal infrastructure (Vanier, 2001), the feedback from the respondents is that data collection practices regarding both inventory and condition vary widely. General data regarding customer location, time of break, and emergency response actions are typically available, but information regarding specific pipe location and physical attributes is inconsistent. Most respondents do not have a consistently high level of information regarding causes of failure and soil and pipe sampling are generally not undertaken. Confidence in and comfort with the amount of data collected varies; mid-size respondents expressed the least level of comfort with the amount of data collected. Many respondents identified additional sources of information including archival, operation and maintenance and GIS information, and hydraulic models. Moreover, having a.hydraulic model does not guarantee that it is used. This may depend on the staffs' ability to operate and maintain the model and on the reliability and currency of the input data. While both physical and statistical models have been developed for predicting pipe deterioration and for developing water main rehabilitation plans, it is evident that the choice and application of these models are limited by the data that utilities have regarding water main breaks (Rajani and Kleiner, 2001; Kleiner and Rajani, 2001). In general, utilities can be classified as those possessing expanded, intermediate, limited or minimal data. Characterization of these classes may be used to inform the development of new asset management techniques such as condition models or heuristics and new ways of effectively 57 collecting, storing, combining, and representing water main break data. This is a subject of a forthcoming paper by the authors. 2.6 ACKNOWLEDGEMENTS The authors gratefully acknowledge those utilities who invested their time and effort in completing the survey and Professors A. D. Russell and J. Atwater at the University of British Columbia and Dr. D. Vanier at the National Research Council Canada, who guided the development of the survey. Mr.s G. Phillips (ret.), W. Liu and G. Irwin of the District of Maple Ridge assisted in survey development, distribution and compilation; Ms. K. Fehrman of the Canadian Water and Waste Association and Mr. A. Gall of the American Public Works Association promoted and distributed the survey to their respective organizations. We are also grateful to Dr. N. Grigg at the Colorado State University for his thoughtful review of this manuscript. c 58 2.7 REFERENCES Andreou, S., Marks, D. and Clark, R., 1987. A new Methodology for Modeling Break Failure Patterns in Deteriorating Water Distribution Systems: Theory. Advances in Water Resources, 10:1:2-10. Elsevier Science, B.V. ASCE (American Society of Civil Engineers), 1999. American Society of Civil Engineers report card and issue briefs. Public Works Management and Policy, 4:1: 58-76. AWWA (American Water Works Association), 2004. WA TER: \STA TS - The Water Utility Database. 2002 version [CD-ROM]. American Water Works Association, Denver, CO. 80235. Deb, A. R., Grablutz, F.M., Hasit, Y.J., Synder, J.K., Longanathan, G.V. and Agbenowski, N., 2002. Prioritizing Water main Replacement and Rehabilitation. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Earth Tech, 2004. National Water and Wastewater Benchmarking Initiative Final Results -June 2004. Earth Tech, Burnaby, BC. Earth Tech, 2003. National Water and Wastewater Benchmarking Initiative- Water Utility Definitions 2003 Release B., Earth Tech, Burnaby, BC. Grigg, N. S., 2004. Assessment and renewal of water distribution systems. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. 59 Jacobs, P. and Kamey, B., 1994. GIS development with application to cast iron water main breakage rates. 2nd International Conference on Water Pipeline Systems, Edinburgh, Scotland, BHR Group. 53-62. Kleiner, Y. and Rajani, B.B., 2001. Comprehensive review of structural deterioration of water mains: statistical models. Urban Water, 3:3: 131-150. Kleiner, Y. and Rajani, B.B., 1999. Using limited data to assess future needs. Journal AWWA, 91:7: 47-62. NGSMI (National Guide to Sustainable Municipal Infrastructure - Infraguide), 2002. Deterioration and Inspection of Water Distribution Systems. Infraguide - Potable Water. Ottawa, ON, Canada. National Research Council of the National Academies - Committee on Public Water Supply Distribution Systems: Assessing and Reducing Risks, 2005 . Public Water Supply Distribution Systems: Assessing and Reducing Risks - First Report. O'Day, D. K., Weiss, R., Chiavari, S., and Blair, D., 1986. Watermain Evaluation for Rehabilitation / Replacement. AWWA Research Foundation (AWWARF) and US Environmental Protection Agency (USEPA), AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. 60 Rajani, B.B. and Kleiner, Y., 2001. Comprehensive review of structural deterioration of water mains: physically based models. Urban Water, 3:3: 151-164. Rajani, B.B. and Makar, J., 2000. A methodology to estimate remaining service life of grey cast iron water mains. Canadian Journal of Civil Engineering, 27:6: 1259-1272. Shamir, U. and Howard, C, 1979. An analytic approach to scheduling pipe replacement. Journal AWWA, 71:5: 248-258. USEPA, 2002. Clean Water and Drinking Water Infrastructure Gap Analysis, Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. USEPA, 2001. Drinking Water Infrastructure Needs Survey. Second Report to Congress. United States. Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. Vanier, D.J., 2001. Asset Management: "A" to "Z", American Public Works Association Annual Congress and Exposition - Innovations in Urban Infrastructure Seminar, Philadelphia, U.S. September, 2001. 1-16. Wild, C. J. and Seber, G.A.F., 2000. Chance Encounters - A First Course in Data Analysis and Inference. John Wiley and Sons Inc., New York, N.Y. 61 Table 2.1 Service population of respondents Service Population >1,000 but <5,000 <10,000 <50,000 <100,000 <500,000 <1 million >1 million Total Canadian utilities a lb 2 8 9 6 3 1 30 U.S. Utilitiesc 1 13 3 11 1 29 Total responses 1 3 21 12 17 3 2 , 59 a) Canadian respondents were from Alberta (5), British Columbia (9), Manitoba (2), New Brunswick (2), Newfoundland (1), Nova Scotia (1), Ontario (6) and Saskatchewan (4). An incomplete survey was received from one Quebec municipality so its response was not included in the analysis. Admittedly, the survey was in English and may be a reason for the low number of responses from Quebec. b) Population served is approximately 2600. c) U.S. respondents were from Alaska (1), Arizona (1), California (3), Colorado (1), Florida (1), Illinois (1), Kansas (2), Massachusetts (1), Maryland (1), Michigan (1), Minnesota (1), Mississippi (1), North Carolina (1), Nevada (1), Pennsylvania (1), South Dakota (1), Tennessee (1), Texas (3), Utah (1), Washington (4) and Wisconsin (1). 62 Table 2.2 Percentage of respondents that record location data Location Data Percentage of respondents that record location data Nearest property address 93% Cross street name 78% Distance from cross street 44% Isolation valve operated 36% Distance from nearest property line 34% Coordinates (northing and easting) 10% 63 Table 2.3 Additional sources of break-related physical data for respondents Number of respondents having additional sources Archival O&M7 GISs2 Other Normal operating pressure 27 X X X Models,.fire flow tests Traffic classification or type of road usage 23 X X X External sources and traffic management systems Year of installation 22 X X X Typical flow in area of break 21 X X X Models Pipe wall thickness / classification 19 X X X External sources Type of pipe lining 16 X X X External sources Length of pipe segment containing the repair 14 X X X3 Pipe protection (wrapped / anodes) 13 X X X Under boulevard or roadway 12 X X External sources Surface material 11 ' X X Pavement management systems Depth of cover 9 X X Type of joint 8 X X X Bedding material 8 X X Type of water service 7 X X Category of native soil 7 X X X Pipe fracture toughness 4 X X External sources Backfill material 4 X X Pipe modulus or rupture 3 X External sources Pipe sample collected 2 X Main tappings 64 Condition of cement lined pipe interior 2 X4 1 Models Pipe material 1 X Pipe diameter 1 X Condition of unlined pipe interior 1 X Condition of bedding 1 X Condition of pipe exterior 1 X 1 Operation and Maintenance records (O&M) 2 Geographic Information Systems (GISs) 3 GISs are the most popular source of length ofpipe segment containing repair data. 4 Evaluated with swabbing. 65 Table 2.4 Suggested sources and approaches for collecting physical data on water main breaks i Physical data Suggested sources of information Pipe diameter Best captured in the field during break repairs and available from as-builts Depth of cover Best captured in the field during break repairs and available from as-builts Pipe material Could be captured in the field during break repairs, available from as-builts and analyzed off-site, e.g., in a laboratory for determining the different types of cast iron pipes Condition of bedding Could be captured in the field during break repairs and also obtained with extra field work Category of native soil Could be captured in the field during break repairs, determined from other records and also could be analyzed in a laboratory Condition of unlined pipe interior Could be captured in the field during break repairs, and also could be analyzed in a laboratory Condition of pipe exterior Could be captured in the field during break repairs and slo could be analyzed in a laboratory Type of water service Could be captured in the field during break repairs or from other sources of information (e.g. land use plans) Surface material Could be captured in the field during break repairs or from other sources of information (e.g. models) Surface use (under boulevard or roadway) Could be captured in the field during break repairs or from other sources of information (e.g. model) Traffic classification or type of road usage Could be captured in the field during break repairs or from other sources of information (e.g. transportation plan, traffic models) " 1 Length of pipe segment containing the repair Could be captured in the field during break repairs Or obtained with additional field work, or from as-builts or GIS Bedding material Could be captured in the field during break repairs, may be available from as-builts and other sources and/or bundled with information gained through additional field work Pipe protection (wrapped/ anodes) Could be captured in the field during break repairs, may be available from as-builts and other sources and/or bundled with information gained through additional field work 66 Physical data Suggested sources of information Type of joint Could be captured in the field (bundled with information gained through additional field work or from other sources of information (e.g., archives or construction inspection records) Year of installation May be available from as-builts or other archival records such as construction inspection reports Backfill material May be available from as-builts or other archival records such as construction inspection reports Typical flow in area of break Technical tools (e.g., models) Normal operating pressure Technical tools (e.g., models) Condition of cement lined pipe interior May be available from as-builts and better analyzed in a laboratory for current condition Pipe modulus of rupture May be available from as-builts and better analyzed in a laboratory for current condition Type of pipe lining May be available from as-builts and better analyzed in a laboratory for current condition Pipe wall thickness/classification May be available from as-builts and better analyzed in a laboratory for current condition Pipe fracture toughness May be available from as-builts and better analyzed in a laboratory for current condition 67 Figure 2.1 Percentage of respondents that collect general information X X Mfy. fc \\\\ s0 'fc «&. ^ v i>3 °o- % \ •\ x So <5-w ft ^ V '-fc x. %3 ^p ^p ^p ^p sO ^p cv*" o"^ cf^ cT^ cr^ cf^ DOOOOOOOOOO :O)OON(DIT)^COCMT-So 68 Figure 23 Percentage of respondents that record failure causes s 70 Figure 2.4 Percentage of respondents that record repair activities 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% i4 # ^ y / 71 Figure 2.5 Percentage of respondents that collect different types of environmental data 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% • Data recorded • • Data recorded and/or available from other sources <0* ,0s II 72 Figure 2.6 Percentage of respondents that expressed confidence in data collected Figure 2.7 Classes of data richness among water utilities Increasing data collected or available for analysis (regardless of data categories, e.g., pipe inventory, break data, operational data) 74 CHAPTER 3 CONSTRUCTING WATER MAIN BREAK DATABASES FOR ASSET MANAGEMENT A version of this paper has been accepted and scheduled for publication in the January 2007 issue of Journal AWWA. The paper is titled as Constructing Water Main Break Databases for Asset Management by A. Wood, B. J. Lence and W. Liu. 75 PREFACE The survey results as reported in Chapter 2 confirm that a number of sources of data exist for use in analyzing water main breaks. In fact, the figure shown below (reference Figure 2.2) inspired my interest in developing an approach for obtaining and maximizing data from different sources internal and external to the utility. If utilities can obtain and integrate data from a wide variety of sources for analysis, they should be able to improve the breadth and richness of their water main break data for asset management. Getting data from other sources can help utilities ggo/o _ JLg^B^^ • Data available from break records (Inn H Data available from break records and other sources 80% " ~~H~~ 60% - J J-| -50% - J 40%- I I I I I n n 0 A key consideration for utilities is storage and management once the data are created. The focus of how to manage data in the Asset Management Systems currently being marketed is on consolidating all data into one large database (e.g., MAXIMO and 76 Hansen). Converting and consolidating data into one database is expensive, resource intensive^ and creates data ownership conflicts. It is unrealistic to expect small and medium water utilities to make the financial and organizational investment to achieve such as database artd thus this is a significant barrier to improving asset management. The research presented in this chapter describes an approach for accomplishing the acquisition and integration of data from other sources for analysis and may be used to foster discussion between various departments on how asset management data can be managed and coordinated within an organization. The research provides techniques and approaches for creating data, constructing databases and relating those databases to each Other. Most importantly, this work can be applied to practice, and has been the case of the Laity View area of Maple Ridge. For Maple Ridge, the research resulted in the construction of a water main break database, the compilation of soil type data with which to analyze the relationship between soil and water main breaks, a review of the current water main replacement policy (of solely replacing asbestos cement pipes), a survey of soil and corrosion potential and a renewed staff interest in improving the process and framework of deciding why, which and when pipes should be replaced. In addition, the approach of constructing databases from different sources and relating databases has been adopted by staff for dealing with all water system data and is being explored for use with Maple Ridge's sewer system. 77 3.1 INTRODUCTION The traditional public works emphasis on managing water main breaks has been directed toward minimizing the loss of water to key businesses and critical facilities (such as hospitals and industrial plants) and minimizing the damage to built and natural infrastructure. However, breaks are also potential gateways to contamination of the water distribution system and have been identified as a high priority in the assessment of water supply health risks by the National Research Council of Academy Sciences (2005). Predicting water main breaks to reduce such risks and optimize the investment in aging infrastructure requires reliable pipe data. These data include age, diameter and material for the subject distribution system and the number and nature of breaks that occur in the water mains. A recent study identifies a need for standardized main break databases and continued research regarding database development (Grigg, 2004). Based on a 2004 survey of North American utilities, Wood and Lence (2006) observe that water main break data collection is evolving and industry practices do not match best practices recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb et al, 2002). For utility managers, collecting, recording and monitoring water main breaks is also important because such events may be used to gain insights for the management of the entire network. This information is important in developing the tradeoffs between expenditures and level of service provided, and in managing rehabilitation programs to achieve a desired level of service. For example, many utilities focus their water main replacement program on tolerating a certain level of breaks in the water distribution system, or perhaps through targeting replacement of water mains of a particular vintage, or of a material that is prone to breaks. 78 Wood and Lence (2006) suggest a number of sources of data that may be used for improving a utility's data breadth and richness for analyzing water main breaks. This paper introduces an approach for constructing a water main database which is based on linking or "relating" data from sources internal and external to a utility for the purpose of knowledge discovery, or in other words relating relational databases. Issues related to creating, linking, transforming, cleansing, scrubbing and integrating data are identified and approaches for addressing them are presented. This approach may assist utilities in developing data acquisition and management strategies and guiding knowledge discovery. Recognizing that data collection and storage is organizationally driven, the approach is designed to be easily adapted. The approach links information among multiple databases and respects decentralized data input in order to maintain ease of data storage and management. This reduces the need for wholesale conversion of data to a central database and is likely to be cost effective for small to mid-size utilities. The decentralized approach to database construction requires coordination of data acquisition, personnel and a clear strategy for encouraging departmental cooperation. To demonstrate the efficacy of this approach for use in analyzing and predicting breaks in a water distribution system, a data schematic is created for analyzing water mains and predicting future breaks for the Laity View area of the District of Maple Ridge, BC. The development of a data schematic is not merely a data storage exercise. Rather it is the creation of linkages that can be used to aggregate information for analysis and that allow data to be updated over time. The linkages are created among primary and secondary data sources. For example, a municipality may not have a transportation plan that specifies traffic volumes, but if knowledge of the volume of vehicle traffic over a water main is 79 desired, it may be able to use volume data from a traffic management system database. The use of data from different sources provides flexibility for utilities to focus on collecting data for future analyses without having to commit to specific application software. 3.2 STATE OF DATA IN UTILITIES Water main break data collection practices vary across utilities and for many utilities, data collection can have significant costs if performed at a comprehensive level. These costs include direct financial costs as well as organizational effort and human resources. The challenges of collecting data include difficulty in mobilizing financial and human resources, absence of historical data, lack of knowledge of current organizational practices, poor reliability of previously collected data, complications due to the emergency-oriented collection conditions when breaks occur, difficulty in prioritizing collection efforts, and the need to develop effective data storage programs (Wood and Lence, 2006). In recent years, many utilities have been developing improved data acquisition and management strategies for water main breaks and in some cases using third parties for analytical tasks for obtaining specialized data such as soil conductivity. Most municipalities do have some information regarding their water pipes and conditions, but few have been maintaining records of pipe breaks for longer than a decade, and very little information is available regarding individual pipes in a given network (Pelletier et al, 2003). In a case study of three municipalities in Quebec, they found that only six parameters (diameter, length, type of material, year of installation, type of soil and land use above the pipe) were available for analysis. In recent years, best practices have been identified for water main break data collection (Deb et al, 2002; NSGMI, 2002). Wood and Lence (2006) surveyed North American utilities and conducted detailed 80 interviews to determine the richness of data available to utilities for analyzing water main breaks. They found that utilities in general can be categorized into general classes of data richness. These are the: expanded, intermediate, limited and minimal data set class. The data set classes are based on the breadth and record length of data irrespective of the traditional data categories such as break, pipe inventory, and operational data because the survey results indicate that many respondents collect data in varying amounts in each of the traditional data categories. The results of the survey also showed that while many utilities do not typically have a common break and water main database or the appropriate data, they may find relevant information available elsewhere in their organization and use this information to expand their database. 3.3 WATER MAIN BREAK DATA FOR ASSET MANAGEMENT Two challenging asset management issues are the determination of remaining service life and the prioritization of rehabilitation efforts. Both of these issues rely on knowledge discovery. To determine remaining service life, one must be able to assess the condition of a pipe and the expected remaining life or some measure of how long a given pipe can be expected to last from the date of installation. It is difficult to determine the exact conditioned buried pipes because they are difficult to comprehensively inspect. As a minimum, utilities should have a database of water main breaks because the occurrence of breaks may reflect the condition of a pipe and typically the number of annual water main breaks is used as a surrogate for the condition of the network. However, breaks do not necessarily reflect pipe condition as there are many causes of breaks such as damage from adjacent construction and frost heave. 81 Once pipe condition is determined, the calculation of remaining service life can be made using deterioration models that predict when failure or future breaks will occur. These models may be either statistically or physically based. The concept of remaining service life is that, given use and time, all pipes will reach a point when they are replaced for reasons such as poor condition, perception of poor reliability or the need to increase hydraulic capacity. Rajani and Makar (2000) define the time of death of a pipe as the time at which its mechanical factor of safety falls below an acceptable value. Kleiner and Rajani (1999) propose that the useful life of a pipe is a function of the economic costs of deterioration and replacement and suggest that pipe death coincides with the optimal time of replacement. Prioritization of rehabilitation efforts involve the timing and scheduling of repairs or replacement of pipes. This is closely tied to resolving financial and technical challenges as to whether to maintain, repair or renew an asset or to choose an alternative such as twinning a water main or constructing an alternative water main. Prioritization is compounded by uncertainty regarding available funds and organizational trends when longer-term planning horizons are considered. Many engineers have faced circumstances where short-term solutions may not be the most economical in the long term but are the most expedient, for instance when a main is repeatedly repaired instead of replaced because capital replacement funds are difficult to obtain, but emergency operating funds are available. Deb et al. (2002) describe four general approaches for prioritizing pipes for replacement: the Deterioration Point Assignment method (DPA), break-even analysis, failure probability and regression methods, and mechanistic methods. Kleiner and Rajani (2001) suggest that only larger diameter mains with costly consequences of failure may justify the data collection efforts and costs required to calibrate mechanistic models. Deb et 82 al. (2002) suggests that break-even analyses be augmented with predictive techniques for pipe breaks, such as failure probability, regression and mechanistic methods. Other considerations for prioritizing water main replacements also include reliability (Xu and Goulter, 1998), consequence of failure (Cooper et al, 2000), consideration of other assets (Grigg, 2004; Vanier, 2001), and on-going engineering and management processes. Davis (2000) suggests that for analyzing the impact of changes in a water main rehabilitation strategy, an agent-based approach is promising. Agents are defined as information-processing systems and are based on artificial intelligence approaches. Davis proposes a loosely coupled generic agent-based decision support framework for water utilities. In this framework, agents are used to extract data from infrastructure, Geographic Information Systems (GIS) and strategic databases. Agents are also used to cleanse data, interface with other databases, predict pipe deterioration and assist in developing rehabilitation strategies. Small to medium size utilities often do not have staff with the skills to implement agent-based approaches. For these utilities, an approach that employs engineering expertise and common data processing systems is likely to be more feasible. Knowledge discovery is the process of identifying valid, useful and ultimately understandable patterns in data (Torra et al, 2004). The analysis of water main breaks is limited by the challenges faced in constructing databases such as limited personnel and resources, missing and conflicting data, and non-computerized information (Pelletier et al, 2003; Habibian, 1992; and O'Day, 1982). Furthermore, baby-boomer staff of many utilities will retire over the coming years and data mining will be overshadowed by the issue of data creation and storage because much water distribution data are orally recorded or are stored in discrete departmentally managed databases rather than in a central database. 83 3.4 CONSTRUCTING WATER MAIN DATABASES The approach for sourcing, constructing and linking water main break data proposed herein is to create a connected data schematic and undertake knowledge discovery based on these data as shown as Figure 3.1. Identifiers that are shared among databases are used to loosely associate and relate data across databases, as represented by the wavy lines in the figure. Data from different databases such as pipe break data files, soil characteristics maps, ortho-photographs, as-builts and orally recorded information are linked by creating identifiers in each data source that are common to one or more sources. For example, specific pipe segments, surface material and break data can each be assigned a common pipe identification number (pipe ID). Information that are only orally recorded and transmitted between staff or no longer documented within an organization may be used to "create" data that can be verified and integrated for analysis. Construction standards and practices and manufacturer data can be identified through interviews. Data from external sources such as the US Department of Agriculture (USDA) and Canadian Soil Survey (CSC) can also be related. The database is created using processed and blended data. The processing and blending of data is performed by technical specialists using tools such as GIS, spreadsheets and databases. Analysis of the data can then be performed to identify network performance including the occurrence of water main breaks, complaints and pressure deficiencies. These can be accomplished with various approaches such as statistical, physical and neural network modeling and risk analysis. From the analysis and knowledge discovery, utilities can develop strategies to predict the remaining service life and prioritize rehabilitation efforts. 84 Identifying the Purpose of Analysis. The first step is to identify the purpose for which the data are to be analyzed. In addition to asset management, objectives such as , performance measurement and improvement, research and development, cost recovery of services, transparency and public accountability may be identified. After the purpose is established, the data required may be determined, i.e., if the purpose of the analysis is to determine factors associated with breaks that lead to service life estimates, the data should include pipe material, age and diameter. Asset management analyses of pump stations require horsepower rating, pump run times, vibration levels and maintenance details. Pavement asset management requires data on pavement thickness, sub-base material, Annual Average Daily Traffic (AADT) and type of axle loading. The purpose of the analysis affects the design of the schematic, the reliability of the relationships between the elements and the degree of information transferred across a link. Developing a data schematic. A data schematic identifies the data required, their potential sources and if and how they are related. The schematic should be constructed considering the characteristics of the data elements including the availability of data, whether sources are primary or secondary or require interpretation, whether data can be obtained from parallel or through a series of sources, the confidence in and explicitness of the data, and how data may be linked. In addition, data can be obtained from multiple sources and some sources may be more readily available than others. Parallel sources contain the same or related data and data can be drawn in parallel, while sources that are in series contain data that in some way can be related to each other and are drawn in series. Once sourced, data can be linked across various databases using identifiers that are unique for specific databases or are common across a number of databases (e.g., a common pipe ID number). Linking data from various sources also allows for the updating of the various 85 . sources of data as a result of analysis, knowledge discovery or further research. For instance, if the purpose of the analysis is to determine if water mains under heavily traveled roads experience higher rates of water main breaks, data of interest include the surface material, road function and traffic volumes. If current network pipe data do not include these attributes, surface ortho-photographs may be used to determine the surface material. The road function on the surface over a water main may be determined using a transportation plan. Alternatively, if the transportation plan does not contain road use data, but traffic counts or a traffic network plan exist, this information may be used to determine traffic loading. Identifying Sources of data. Wood and Lence (2006) provide a list of alternate secondary sources for break-related data based on survey respondents' written observations and personal interviews. On their own, these data are not very useful for analyzing breaks, but when combined, they can provide useful information for developing rehabilitation and operation and maintenance strategies. The following is a detailed discussion of the alternative sources that utilities should consider in building a water main break database. Network data. Network data may be available from GISs, Asset Management (AM) / Facilities Management (FM) systems, databases, spreadsheet systems and other analytical models, e.g., hydraulic models. Other sources of data include Supervisory Control and Data Acquisition (SCADA) systems and monitoring systems. AM/FM and SCADA systems typically are enterprise systems, used by many departments throughout an organization, store large amounts of data on large server databases and operate on a client-server environment. AM/FM systems collect and store maintenance, repair and financial information while SCADA systems collect and store operating data such as pump-run times, pressure and hydraulic data. 86 If GISs are being used for network mapping, each pipe in the system is given an individual pipe ID and may have corresponding information such as year of installation, length of pipe segments, diameter and material. The length of pipe segment is typically obtained from the geometry of the spatial file and are usually collected from as-builts as part of the construction of the GIS. Many utilities use hydraulic models for their water, sewer and drainage infrastructure. Hydraulic models contain information such as flow, pressure and pipe layout that can be useful for analysis of water main breaks and asset management. For example, roughness, mass balance discrepancies and hydraulic losses calculated in a model can be indications of pipe condition. As-built drawings, previous construction standards, staff experience, purchasing records, manufacturer specifications, inspector and surveyor field notes, main tappings, flushing records, swabbing, laboratory, hydrant fire flow testing, inspections, water quality testing results, and customer complaints may be used to indicate information such as the bedding and backfill material, depth of cover, type of joints, pipe protection, original pipe wall thickness, current wall thickness, condition of pipe interior and level of internal corrosion. Surface data. Sources for surface material, land use and traffic loading data include high resolution (e.g., 0.15 metre / 6 inches) ortho-photographs and as-built drawings. Land use and transportation plans, GIS maps of roads and orally recorded knowledge of types and ratios of road users, and traffic patterns provide information about surface use, traffic loading, and potential sources of damage from surface activities. Ortho-photographs may be loaded into GIS as a "background theme" and overlain with the pipe network to generate drawings to determine surface material data for each pipe. This may reduce the need for field surveys. V 87 Soil data. Soil data are available from both provincial and federal government databases in Canada and analogous sources in the US. The CSC National Soil Database contains the soil types for all of Canada and is the national repository for survey information from the broad (1:1 million scale) to the detailed level (1:10,000 to 1:250,000 scale). The USDA Natural Resources Conservation Service is responsible for the National Cooperative Soil Survey which includes the efforts of federal, state, and academic institutions (Rossiter, 2005). The Washington Suburban Sanitary Commission has developed soil corrosivity maps based on USDA soil survey results (Habibian,1992). While agricultural surveys may be limited in terms of their usefulness in corrosion analysis, when combined with a soil sampling or survey program, they may be used to provide a general understanding of the soil conditions. Utilities should also review boreholes and logs from previous studies, well drilling data and information gathered on construction projects. Soil permeability data and water table information can be useful for drainage and sanitary sewer inflow and infiltration analysis. Other sources of data. Utilities should explore the data collected by others or used within the organization such as mapping of environmentally sensitive areas, pavement management systems and infrastructure plans. Environmental information such as water and ambient temperature and the frequency of frost, may be available from water quality testing, weather and environment agencies. Gas pipeline, electrical, telecommunication or other utilities operating in the area may have information on soil data, stray currents and the corrosivity of soils. Information that are not recorded but available orally from staff can be obtained by using interviews of staff to determine, for example, the type of bedding and backfill material, pipe lining and pipe protection. Whenever possible, verification of observations 88 is recommended. Broad surveys and spot testing are other techniques to consider when obtaining data. For example, a program that samples soil for corrosivity and relates the agricultural classifications to soils can yield insights into potential corrosion hot spots and focus monitoring efforts. Linking data. The objective of connecting or linking data is by relating relational databases to support on-demand data analysis. Data from multiple sources may be linked in series, or parallel or remain unlinked. This approach can facilitate the updating of the analysis data set as data sources are updated and thereby support future knowledge discovery. Information processing tools, such as GISs, spreadsheets and databases can be used to create electronic records, manipulate and analyze data. Output from these tools is also easily exported to other, comprehensive databases. GIS shape files can contain a myriad of information that may be viewed and exported for analysis using other analytical software such as spreadsheets and databases. Depending on the expertise and skills of those analyzing the data for knowledge discovery, utilities may choose to use a spreadsheet file as the main data analysis file. An example of which is described in the case study presented in this paper. J If break records are stored only on paper, an electronic break database should be created. Spreadsheets are easy to use and the data can be easily transferred to GIS format for viewing while being retained as the key database. Pipe ID numbers that are assigned for GIS network maps can be used as the connection between the break records and pipe characteristic data. If records do not specify the exact location of the break along a pipe segment, some extrapolation may be required to identify the damaged pipe. For example, a house number may need to be used to define the location along a pipe where the break occurred. . •• 89 A process for converting geographical archival information into electronic data for relating data is shown in Figure 3.2. In the first step, information in the form of paper maps is scanned to create Tag Image File Format (TIFF) images. Then the images are digitized to create polyline drawings and GIS are used to build polygon coverage. Attributes are attached for each polygon and then related to a common identifier, such as a pipe ID. Using the intersect function of the GIS, the attributes of the network pipes and the attributes of the polygons are combined and exported as a spreadsheet file. All the files form a portion of the data warehouse that is used to create the main analysis file. For example, soil image maps may be digitized to create a closed polyline drawing of soil types, fitted to the cadastral drawings of the water network in AutoDesk Map® using the 2D transformation process, and then used to build polygon coverage of the soil conditions with ArcMap®. If each polygon is assigned a soil type identification number that corresponds to a soil type, then using the intersect function of ArcMap®, the pipe and corresponding soil attributes can be generated and exported to a shape file, and ultimately exported as a spreadsheet file of pipe IDs and corresponding soil type. Hydraulic models may be used to calculate the flow through model links, and the pressure and heads at nodes within the water distribution system. A process for estimating pressure data along each link within the network is shown in Figure 3.3. Hydraulic model outputs (e.g., flow and friction coefficient estimates in links and pressure estimates at nodes) are determined for the corresponding pipe ID. The pipe link and node shape files are combined using GIS tools, exported to a spreadsheet file, and thereafter combined with the other pipe network data, such as pipe material, diameter, date of installation, and length. Often, the hydraulic model and the pipe network do not have a one to one relationship, and commonly the hydraulic model is a skeleton of the network where a link 90 in the model may actually represent a number of pipes in the real system. For example, the City of Toronto recently skeletonized their 307,956 pipe network to a 76,989 link hydraulic model (Schick, 2005). In this case, the pressure and flow may be related by buffering the hydraulic model output to the pipe network data. Buffering creates a relationship between the links in the model and the well documented pipes in the pipe network; a process shown in Figure 3.4. Here, the pipe network data (a .DWG file) is converted to a shape file with identification numbers (usually the pipe ID) and the hydraulic model data are prepared as a shape file. The two shape files are buffered using GIS analysis tools. Buffering the two shape files creates a relationship between each pipe and the model pressure and flow in the links. Using this process, all pipes can be associated with corresponding flows and pressures. This approach may also be used for combining sewer or drainage hydraulic models or other spatial data. Processing Data. Prior to the analysis stage, data should be processed to reduce errors and inconsistencies and to produce a coherent data set. Data processing activities include simple transformations (e.g., detecting and removing outliers), cleaning and scrubbing, blending data from the various sources and summarizing the data to reduce the number of records (Torra et al, 2004). In the process of relating the data, analysts may discover inconsistencies in the databases and may have to determine the reliability of the data and how to treat incomplete data, e.g., pipes that do not have an installation date. Analysts may also find pipes that are missing material or diameter data or that hydraulic model output and GIS data differ. It falls on the analyst and data schematic builder to carefully consider the impacts of data use approaches and how these approaches affect the ability to achieve the purpose of the analysis. 91 Knowledge discovery. Once data processing is complete, utility managers may analyze and extract knowledge and gain insights regarding their system. Typical knowledge discovery techniques include spatial and statistical analyses using GIS, spreadsheets, databases, physical and statistical pipe deterioration models and artificial intelligence techniques. Here, patterns in the data may be determined such as break patterns, and rates for differing pipe materials, diameters, and ages and the spatial distribution of the breaks over time. As an example of the use of such observations, knowing the proportion of pipe materials of a distribution system and the failure rates within each material class can help determine the vulnerabilities and failures that may be expected of the system. Determining the past, present and future failure rates for pipes of different vintages, diameters, soil conditions, and surface loadings is important for guiding a utility's rehabilitation program. 3.5 WATER MAIN BREAK DATABASES FOR MAPLE RIDGE, BC The construction of a data schematic for analyzing water main breaks is demonstrated here for the area of Laity View in the District of Maple Ridge. This area comprises 13 percent of the 335 kilometer distribution system for the District and represents an urban area. The Laity View area experienced construction practices and has , soil types typical for Maple Ridge. The District has information residing in various formats (e.g., electronic, archival, and oral) that are distributed and managed across the organization. For example, the Operations Department has paper copies with limited information on the break history of water mains from 1983 to 2004, and no database, while the Engineering Department has a skeletonized hydraulic model of the system constructed in 2001 in a .DWG format and a spatial representation of the pipe network in GIS. 92 The purpose of this analysis was to determine relationships among water main breaks and factors such as pipe age, material, soil and depth of cover so as to aid in asset management (NSGMI, 2003). The information processing tools used were Microsoft Excel®, Microsoft Access®, Arc View® GIS and Autodesk® Map. These software tools were selected because they are available in-house, and District staff are trained to use them. The data set for the Laity View area of Maple Ridge represents an intermediate water main break data set with a 20 year break history (1983 to 2004). The data included pipe material, pipe diameter, whether the pipe is under a boulevard or roadway, the year of installation, depth of cover, length of pipe segment, surface material, normal operating pressure, bedding material, whether the pipe is wrapped or anode protected (wrapped/anodes), backfill material, type of road function, traffic classification, type of pipe lining and typical flow in the pipe. Constructing and linking data. The availability and location of the data, level of confidence in them, and information regarding how they were created are identified and listed in Table 3.1. Data associated with a low level of confidence are candidates for a data verification program. Based on the various sources of the data and the goal of aiding asset management, the schematic of how data can be related as shown in Figure 3.5 was created. The schematic identifies potential links and inputs for analysis and major data of interest. Primary sources of data typically aggregate data and are readily available but do not contain all the data deemed of interest for asset management. Secondary sources that contain additional information were identified. Many of these use various data formats (e.g., databases and paper records) and are distributed across the organization. Some data are available from parallel sources and one or all of the sources may be used for analysis. For example, traffic volumes may be obtained from a transportation plan or from a traffic 93 network plan, traffic volume database and pavement management system. The transportation plan, traffic network plan and pavement management system use AADT for defining the volume of traffic while the traffic network plan contains 24-hour vehicle volumes which may be transformed to equivalent AADT. Where possible, data were linked using the pipe ID as the common link across the databases in order to facilitate updating of all data sources using the results of the knowledge discovery process and additional field investigation. Like many utilities, the electronic water distribution network information for Maple Ridge includes year of installation, diameter and material and these are available from the GIS and drawing files. The length of the pipe segment was obtained from the geometry of the pipes in these files. Autodesk Map® tools were used to define and create the shape file for the region. The study area was defined for extraction from the water network map, and exported to a shape file with the object data for each pipe. The shape file was viewed and plotted in Arc View® GIS 3.2A and exported as a spreadsheet file as the main data file for analyses. The District's hydraulic model data are stored in a .DWG format but the model uses links and nodes which do not completely relate to specific pipes. In many cases, links represent a number of pipes in series and required buffering. The break history for Maple Ridge is limited in detail, but is relatively long. Breaks are recorded by operation and maintenance staff and stored in paper form. The recording of break locations as well as the amount of environmental and break information varies among the records within a given year and during the period of record. Field crews typically record locations with respect to the nearest cross street but do not record the exact location of the break. As a result, if a pipe experienced multiple breaks but the exact locations of these breaks are not identified, it cannot be determined if the pipe failed at the same 94 location or at multiple sites along the pipe. However, identification of the exact location was not necessary because the purpose of this analysis was to determine if and when a pipe broke and not the exact location of the break along each pipe. To create the electronic break database, the data from the field crew reports were entered into an electronic spreadsheet. Pipe ID from the water network map were added to the break record. The data worksheet was then converted to a database that was then exported as a .DBF file to GIS where it was plotted for visualization. The District does not have detailed soil information. To construct these data, a number of secondary sources including federal, provincial and local government reports (Golder, 2002; Luttmerding, 1981; Luttmerding, 1980) were reviewed. While the provincial reports use agricultural soil classifications, soil groups based on the parent material (i.e., the upper stratigraphic unit) better represent the soil types at installation depth. The three main classes of parent material are: marine clay, marine sand or eolian silt. The provincial soil maps were converted into an electronic form and intersected with the water network maps as described previously in the linking data section. To obtain information regarding surface material, the ortho-photographs of surface features were plotted as background to the water distribution network. The possible surface material types are asphalt, gravel (typically representing a road shoulder) and landscaping. Whether the water main is under a boulevard or roadway was similarly determined and recorded. A GIS-based map of the roads, overlaid on the water pipe network and knowledge based on the District of Maple Ridge Transportation Plan was used to classify the traffic on the ground surface above the water main. Categories of traffic loads that were used include those attributed to local, collector, arterial, and commercial roads, and lanes and those 95 experiencing no traffic, such as boulevards. In the case where a pipe is under two or more road classifications, the classification corresponding to the heavier traffic road function was selected. As is the case for many utilities, the quality and amount of construction details provided in as-builts varies depending on the designer, contractor and time period during which the main was constructed. Thus, the confidence in these details also varies. However, in spite of the variability of records, utilities may find that construction standards, in place for years, may provide reliable information. For example, the standard cover used for construction in Maple Ridge is generally 900 millimetres (three feet) and has been verified by interviews with staff (Thain, 2005). Data regarding bedding and backfill material, and pipe lining and protection practices were determined through interviews with staff. Relationships between the bedding and pipe material were established and a data set of bedding material was created. Similarly, whether pipe protection exists for pipes in the study area was determined using the soil type for each pipe. Where the soil type indicates clay and the pipe material is ductile iron, the pipe was considered on the basis of practice to be wrapped. All other pipes are considered to be unprotected. This assumption is identified as one which may require further verification. While some data were obtained using secondary sources, it was not possible to obtain or construct all data. Also, while minimal scrubbing of the data was required, there were some gaps in data. For example, when the District replaces a pipe or segment of a pipe, it does not retain any history of the pipe. Consequently, there is no information regarding the former history of the age, material or diameter of pipes that have been ' replaced unless the data are recoded on the break record form (typically this would only be diameter and material data). 96 Preliminary analysis and knowledge discovery. Knowledge discovery results for this case are reported in Figures 3.6 to 3.8 and Tables 3.2 and 3.3. In the study area, 69 percent of the pipes are ductile iron, 26 percent are asbestos cement, five percent are cast iron and less than one percent is steel pipe. The number of breaks over time,, plotted in Figure 3.6, may help managers develop strategies and budgets. While data indicate a stable number of annual breaks over the past years, a prediction model would be useful to estimate if and when the number of breaks will increase. Of the 47 breaks that occurred, there were 32 first-time breaks. During 1983-1999, ten breaks occurred in nine pipes and these pipes were replaced. Because Maple Ridge does not retain data on pipes that are replaced except what is on the break records (which may record the pipe diameter and material), only 37 breaks had associated pipe diameter, material and age data. The usefulness of knowledge discovery is in determining patterns of pipe breakage and insights based on these patterns. One example pattern is illustrated in Figure 3.7 and Table 3.2 which show that the majority of first breaks occur when pipes are between 15 to 19 years old, and that pipes of the 1960-1974 vintage have a significantly higher rate of failure than others, and thus attention should be given to these pipes. In addition, other relationships such as that between breaks and pipe size (as shown in Table 3.3) and breaks and soil conditions (as shown in Figure 3.8) can provide insights, such as the amount of failures in ductile iron pipes that are installed in clay soils. Currently, the water main replacement program of Maple Ridge is targeted at replacing asbestos cement pipes. As a result of this project, Maple Ridge initiated a soil corrosivity survey and plans to correlate those results with the soil data created in this study. They are also examining ways to improve prioritization of the rehabilitation of water mains using the newly gained knowledge. Practices have also been revised including: data 97 on replaced pipes are now retained, field data are now being collected when new pipes are installed, and new water main break forms have been implemented. Communication on data collection is improving as departments work together to collect and share data and have a better understanding of each other's role in achieving the common goal of effective asset management. 3.6 DISCUSSION The process of creating a data schematic among multiple databases for data compilation, analysis and knowledge discovery may yield valuable insights for improving asset management practices. It expands the data available for analysis for both present and future applications and respects decentralized data input and management. The use of decentralized data storage reduces problems related to data ownership among units within organizations regarding data collection, management and dissemination of water, sewer, drainage or other infrastructure information. As in any business process, it is important to document the purpose for constructing, structuring, sourcihg and linking the data schematic. A clearly documented outline of the relationships between the key data sets is important for future managers and those utilizing the data. NSGMI (2003) suggests some key relating information such as asset number, asset location and work order numbers. The work undertaken herein suggests that those collecting data within an organization should share information regarding the purpose for collecting and selecting the amount and type of information to avoid duplication and focus efforts. Data managers must also be aware of disclosure control in which important or confidential data may be inadvertently connected and unintentionally released to third parties (Torra et al., 2004). For public utilities, the issue of private and proprietary 98 i information is important and should be considered (e.g., considering the Freedom of Information Act in British Columbia.). Processing of data is a significant step in data schematic construction and relating data. For example, in the Maple Ridge case study, use of agricultural soil classifications resulted in too many soil zones for analysis but reducing the number of classifications and using the parent material categories improved the understanding of the relationship between soil type and the number of water main breaks. Summarizing and aggregating data requires expertise and knowledge of the intended analyses. In seeking data sources, utilities should consider data that other agencies are collecting even if they ate doing so for different purposes and determine whether the data are useful as a secondary source. Application of new technologies, such as infrared, electro magnetic surveys and Light Detection and Ranging (LIDAR) may be employed to capture relevant data. Utilities face the challenge of capturing undocumented institutional information and knowledge over the next decades as their baby-boomer staff retire. As demonstrated herein, interviews are useful in capturing these data, but the processes of interviewing staff can be awkward. The authors found that staff shared more information when interviewers began the session by making observations and asking about the validity of these observations, and then asking specific questions. Maps and other visual cues usually triggered the interviewee's memory. It is important for interviewers to pose questions and present the exercise as soliciting information to aid in knowledge discovery rather than criticism of past practices. It is equally important to, where appropriate, validate the observations and experiences of staff. It is noted that retired staff seem uncomfortable with 99 general observations unless they could also provide exceptions. Though there may be confidence in orally recorded data, verification should be undertaken where possible. A limitation of constructed databases is the possibility that created data are more correlated than data that are collected independently. For example, in the case study, all ductile iron pipes are defined as cement mortar lined and the pipe material determined the backfill and bedding data. This may affect the amount of explanatory variables in an analysis. Other limitations include the fact that data may not always be accurate or quantifiable. The limitations may be overcome by a program to verify the data over time, which is a long-term exercise. For predicting future water main breaks, it may be difficult to determine if variables are significant if there is a limited history of breaks. It is important that the data are stored and that future break data be added to increase the history and data set for further analysis. 3.7 CONCLUSIONS Grigg (2005) has identified the need for standardized main break databases and continued research regarding database development as strategic for informed infrastructure management though the topic of databases has been the subject of other authors such as Deb et al, (2002); Habibian (1992), and O'Day, 1982). However, surveys and literature indicate that data are typically scarce. Compounding this, water main break data in centralized databases are not common and approaches and techniques to relate and manage data for analysis are needed. The approach presented herein expands the sources (including, in particular, oral transmission) and the amount of data for asset management. Researchers and managers may gain insight into a system by systematically sourcing, relating, processing, blending 100 and analyzing data. The framework is flexible, anticipates the evolution of data collection, building, verification and storage and allows for a variety of users. It does not abruptly disrupt data collection and warehousing practices. A key benefit of this approach of relating data is that it allows managers to continue to expand data collection because databases are decentralized. It also fosters a dialogue of data development, knowledge discovery and information processing across an organization. It is flexible and can be adapted to all utilities, whether they are small, medium or large, and regardless of the uniqueness of the data collected and organizational framework. The recording of future breaks and verification tasks are important for building confidence in the data and for augmenting it. A long-term strategy should be developed that verifies and improves the breadth of data and confidence in the database. The tasks can be opportunistic, undertaken when other repairs of the system are occurring, or systematic. When pipes are replaced or new service connections are installed, opportunities arise for obtaining physical samples and verifying the nature and condition of pipe protection, bedding, backfill and the pipe exterior. An example systematic program is a utility-wide survey of soil corrosivity in various soil classifications and an exercise that improves the understanding of the relationships among soil classifications, corrosivity and breaks. This approach can also be used for sewerage, drainage and other systems for improving asset management, operation and maintenance analysis, performance and public accountability. For example, by relating and analyzing grease build-up areas, complaints, slope, flow and model flushing velocities, a sewer system manager can analyze performance and resource efforts and develop plans to address previously "un-connected" problem areas. As well, for a road network, constructing relationships between complaints, ( 101 traffic volume, speed and crash data, lighting level, signage, road condition and geometry, managers may be able to improve road safety and predict high crash locations. As observed in the Laity View area case study, the process and results improve communication and the basis for decision making. New insights can assist managers in focusing effort and resources and discovering unanticipated issues and challenges. While the focus of utilities has been on collecting and storing data, the next step will be in the application of data mining and knowledge discovery. As more tools become available to analyze data, managers need to consider the business intelligence that it should employ to wisely invest resources to meet future demands. 3.8 ACKNOWLEDGEMENTS N The authors gratefully acknowledge the District of Maple Ridge for providing data and support in the form of employee resources, Messrs. D. Thain (ret.), H. You (ret.) and J. Scherban from the District for providing tacit information, and Professors A. D. Russell, J. Atwater and H. Schrier at the University of British Columbia (UBC) for their insights and suggestions. Mr. A. Malyuk of the District of Maple Ridge assisted in data processing and compilation and Messrs B. Kampala, M.A.Sc. Candidate, UBC, and W. Johnstone, Principal, Spatial Vision Group assisted in reviewing the manuscript. 102 3.9 REFERENCES 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, 2:2000: 97-103. Davis, D.N., 2000. Agent-based decision support framework for water supply infrastructure rehabilitation and development. Computers, Environment and Urban Systems, 24:2000: 173-190. Deb, A. R., Grablutz, F.M., Hasit, Y.J., Synder, J.K., Longanathan, G.V. and Agbenowski, N., 2002. Prioritizing Water main Replacement and Rehabilitation. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Golder Associates Ltd., 2002. Geotechnical Input to the seismic vulnerability assessment for the District of Maple Ridge, B.C. 500-4260 Still Creek Drive, Burnaby, BC, V5C 6C6. Grigg, N.S., 2005. Assessment and Renewal of Water Distribution Systems. Journal AWWA, 97:2: 58-67. Grigg, N. S., 2004. Assessment and renewal of water distribution systems. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. 103 Habibian, A., 1992. Developing and utilizing databases for water, main rehabilitation. Journal AWWA, 84:7: 75-79. Kleiner, Y. and Rajani, B.B., 1999. Using limited data to assess future needs. Journal AWWA, 91:7: 47-62. Kleiner, Y. and Rajani, B.B., 2001. Comprehensive review of structural deterioration of water mains: statistical models. Urban Water, 3:3: 131-150. Luttmerding, H.A., 1981. Soils of the Langley-Vancouver map area: Report NoJ5, Volume 6. B.C. Ministry of Environment Assessment and Planning Division, Kelowna, BC. Luttmerding, H.A., 1980. Soils of the Langley-Vancouver map area Report No.15 Volume 1. B.C. Ministry of Environment Assessment and Planning Division. Kelowna, BC. NGSMI, (National Guide to Sustainable Municipal Infrastructure - Infraguide), 2003. Best Practices for Utility-Based Data. Infraguide - Potable Water. Ottawa, ON, Canada. NGSMI (National Guide to Sustainable Municipal Infrastructure - Infraguide), 2002. Deterioration and Inspection of Water Distribution Systems. Infraguide - Potable Water. Ottawa, ON, Canada. I 104 National Research Council of the National Academies - Committee on Public Water Supply Distribution Systems: Assessing and Reducing Risks, 2005 . Public Water Supply Distribution Systems: Assessing and Reducing Risks - First Report. O'Day, D. K., 1982. Organizing and analyzing leak and break data for making main replacement decisions. Journal A WW A, 74:11: 588-594. O'Day, D. K., Weiss, R., Chiavari, S., and Blair, D., 1986. Watermain Evaluation for Rehabilitation / Replacement. AWWA Research Foundation (AWWARF) and US Environmental Protection Agency (USEPA), AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235." Pelletier, G., Mailhot, A. and Villeneuve, J.-P., 2003. Modeling water pipe breaks - three case studies. Journal of Water Resources Planning and Management, 129:2: 115-123. Rajani, B.B. and Makar, J., 2000. A methodology to estimate remaining service life of grey cast iron water mains. Canadian Journal of Civil Engineering, 27:6: 1259-1272. Rossiter, D.G., 2005. A compendium of on-line soil survey information - Soil survey institutes and activities. Accessed May 31, 2005, http://www.itc.nl/ rossiter/research/ rsrch ss sources.html. 105 Schick, S., 2005. Toronto bridges data needs in water network project. ITBusiness.ca July 21, 2005. Accessed July 22, 2005 http://www.itbusiness.ca/index.asp7theaction =61&sdi= 59540. Thain, D., 2005. Water mains in the District of Maple Ridge. A. Wood, ed., Maple Ridge, Interview notes. Torra, V., Domingo-Ferrer, J. and Torres, A., 2004. Data mining methods for linking data coming from several sources. 3rd Joint Un/ECE-Eurostat Work Session on Statistical Data Confidentiality, Monographs in Official Statistics, Luxembourg, Eurostat. 143-150. Vanier, D.J., 2001. Asset Management: "A" to "Z", American Public Works Association Annual Congress and Exposition - Innovations in Urban Infrastructure Seminar, Philadelphia, U.S. September, 2001. 1-16. Wood, A. and Lence, B.J., Assessment of Water Main Break Data for Asset Management. Journal AWWA, 98:07. Xu, C. and Goulter, I.C, 1998. Probabilistic model for water distribution reliability. Journal of Water Resources Planning and Management, 124:4: 218-228. 106 V Table 3.1 Water main break data availability for Maple Ridge Data element Availability for "off the shelf analysis Secondary source Notes Confidence/reliability Pipe material Available in AutoCad in .DWG format High Pipe diameter Available in AutoCad in .DWG format and GIS High Type of water service Available in Ortho-photographs (0.5metre/1.6 feet * and0.15metre/0.5 feet intervals) Requires interpretation of data available Medium Under boulevard or roadway Available in 1 Ortho-photographs (0.5metre/1.6feet and0.15metre/0.5 feet intervals) Requires interpretation of data available High Year of installation /age Available in AutoCad in .DWG format and GIS Medium - high Depth of cover Orally recorded information Based on standards of the day Medium - but could be verified over time Length of pipe segment Available in AutoCad in .DWG format and GIS Medium Normal operating pressure Available in model and in .DWG format Need buffering to assign pipe with node data Medium - High Type of joint Not available Field investigation commenced for future pipe breaks Low Condition of pipe exterior Not available Field investigation commenced for future pipe breaks Bedding material Orally recorded information Data inferred from known pipe material, future field verification required Medium - but could be verified over time 707 Data element Availability for "off the shelf analysis Secondary source Notes Confidence/reliability Pipe protection (wrapped/anodes) Orally recorded information Use of soil maps to infer information, future field verification required Medium - but could be verified over time Backfill material Orally recorded information Data inferred from known pipe material, Future field verification required Medium - but could be verified over time Traffic classification or type of road usage Transportation plan designation and maps Need better understanding of traffic loading Type of pipe lining Orally recorded information Data inferred from known pipe material, future field verification required Medium - but could be verified over time Typical flow in area of break Available in hydraulic model and in .DWG format Some more modeling information (typical flow) Medium-high (from model confidence) 108 Table 3.2 Water main breaks for a given year of installation (1983-1999) Year of Installation Number of Pipes Length in metres (feet) Number of breaks as an age group Percentage of pipes in age group with breaks 1955-1959 14 1,687 (5,535) 4 29% 1960-1964 32 • 2,968 (9,738) 8 25% 1965-1969 12 1,511 (4,958) 7 58% 1970-1974 52 4,791 (15,719) 11 • 21% 1975-1979 56 4,030 (13,222) 3 5% 1980-1984 56 4,122 (13,524) 1 2% 1985-1989 56 4,584 (15,040) 3 5% 1990-1994 96 7,452 (24,450) 0 0% 1995-1999 51 3,003 (9,853) 0 0% Unknown 4 29 (95) 0 0% Total 429 34,177 (112,135) 37 9% 109 Table 3.3 Pipe breaks for pipes of a given diameter (1983-1999) Pipe diameter in millimeters Number of Pipes Number of breaks Percentage of pipes for each diameter group with breaks 100 2 0 0.0% 150 246 26 10.6% 200 117 7 6.0% 250 62 4 6.5% 300 2 0 0.0% 350 0 0 0.0% 429 37 9.0% 110 Figure 3.1 A schematic for constructing and using water main break data for knowledge discovery Verificalion"and:;upaating;of;aata';:;™s Soil maps transportation I plans , Ortho-photographsl reports AM/FM systerrf data Break data files Linking or relating data using identifiers common to two or more databases Processing Application Software: -GIS -CAD - spreadsheets Human analyst and application software 4 Data Analysis and Application Analysis tools -GIS -CAD - Spreadsheets - Databases Knowledge Discovery Strategies and Plans ill Figure 3.2 Process for digitizing and creating data from archival geographical data Scan hard copy image and create TIFF image Create closed polyline drawing Use GIS to create polygon coverage (closed polylines) and identify attributes to be linked Apply GIS processing tools (e.g., intersect function) to pipe network and the polygon to generate shape file Export attribute data of the shape file to analysis file (e.g., spreadsheet file) XZ1 = 5 Sde ndte%Mo%> 1 § for titisshfidkd I f = : :: 1 I i Note: Title of data columns Number Month Years Pipe of Pipe Material Year of Year Of Of of Years In length breaks ID Size Material Code installation Failure Failure Service Ground (metres) to 1999 Has pipe been Soil Number Under replaced since zone of soil boulevard Surface Traffic Pipe break? type zones or road material classification lining Bedding Backfill Q V C_number HGL Max demand day Pipe protection C factor Roughness (l/s) (m/s) (Hazen) (metres) pressure _(PSI) 112 Figure 3.3 Linking hydraulic model data with network data Determine hydraulic model input and output data and relationships of data. For example, are pressures calculated only for nodes or are they also assigned to model links? Do the pipe link data include all the data being sought for analysis? Combine hydraulic model output node pressures and link flow data for a given pipe ID. Determine pipe network information such as material, diameter, date of installation, length, etc. How does the pipe network relate to the model, i.e., are there one-to-one relationships or is the model skeletonized? Combine hydraulic model data with pipe network data. Export combined attributes to analysis file 113 Figure 3.4 Buffering data to create data relationships using GIS Link ID 1 is a link that defines connectivity in the hydraulic model. Each linkjs-~ identified as unique in the> model. Buffering the Link ID 1 data to Pipe ID 1 and Pipe ID 2 gives the attributes of Link ID 1 to the network pipes Pipe ID 1 and Pipe ID 2. Pipe ID 1 and Pipe ID 2 are network pipes in GIS. Each pipe identified as unique. ID 1 Pipe ID 1 with Link ID1 data Pipe ID 2 with Link ID1 data Steps to create buffered data 1. Create shape files of model data and network data. 2. Buffer data using capabilities of GIS. 3. Export the attributes of the buffered shape file as a spreadsheet file. 114 Figure 3.5 Maple Ridge water main break analysis data schematic Tacit data: bedding, construction standards, pipe protection, cover, etc. Project report: pipe materia], installation date, pipe length, etc. Hydraulic model: flow, pressure, pipe material, friction coefficient, etc. Geological Survey of Canada (GSC) data Ministry of Environment (MOE) soil data Pipe inspection record: pipe material, etc. Road as-built: surface material, road function, etc. Break report form: break data, date, costs and effort of repairs Primary sources Secondary sources Potential link (e.g. pipe ID) Potential input to analysis Accounting system: break activity costs, annual break expenditure, etc. Pavement management system: surface material, road function, etc. Traffic network plan: road function, traffic volume, etc. Traffic volume data base: traffic volume, road function, etc. 115 Figure 3.6 Cumulative breaks in Laity View area (1983-1999) 116 Figure 3.7 Number of years in service when break occurred in pipes (1983-1999) 10 Number of years in service Note: Of the 47 breaks in the area, no pipe age data were available for 10 of the 47 breaks. 117 Figure 3.8 Number of breaks for pipes in a given soil type installation (1983-1999) Asbestos cement pipes Cast iron pipes Ductile iron pipes • Marine Clay • Eolian Silt • Silty Clay 2 O Marine Clay a Marine Clay • Eolian Silt 118 CHAPTER 4 USING WATER MAIN BREAK DATA TO IMPROVE ASSET MANAGEMENT FOR SMALL AND MEDIUM UTILITIES A version of this paper has been submitted for publication to ASCE Journal of Infrastructure Systems as Using Water Main Break Data to Improve Asset Management for Small and Medium Utilities by A. Wood and B. J. Lence. 119 PREFACE In the previous chapter, I created data and a number of databases to demonstrate some techniques that utilities can employ to enrich their asset management data sets. In Chapter 4,1 demonstrate that once data are created and linked for analysis, utilities can use these data within a framework that I developed for improving their asset management practices. The intent of this research is not to create a new model, but to develop a framework that uses break prediction models that small and medium size utilities can apply. The soil and surface material data that were created in Chapter 3 were used in the experimental application of this framework and give'significant insights into the factors that influence pipe breaks. While the data created provided a demonstration case, there was not sufficient information to apply all the models that I initially proposed to investigate. The use of statistical deterministic time-linear and time-exponential models could be sufficiently demonstrated with the data created in Chapter 3, but there were insufficient data to obtain meaningful results from an application of the survival analysis and KANEW (Deb at al, 1998) models. Thus, all four types of models were applied in an effort to obtain proof of concept, but only two types of models are reported in the manuscript that comprises Chapter 4. It is important for utilities to continue to create and mine data. This research reinforces the notion that not all data may be useful for applying sophisticated models but that unsophisticated pipe break models can provide insights into the performance of a water main system, be used in identifying system specific factors that may cause breaks, guide the development of a water main break data collection strategy, be used to identify groups 120 of pipes, their historical and predicted break frequency to further investigate and prioritize for replacing and thus be a benefit to communities. 121 4.1 INTRODUCTION Water utilities have aging and deteriorating infrastructure and must prioritize the replacement of their water mains to minimize pipe breaks. Breaks result in loss of water to key businesses and critical facilities, may lead to damage of other infrastructure, and have been identified as a pathway for microbial contamination of distribution systems (AWWA and EES, 2002). The need for rehabilitating aging water mains is increasing, the costs of repairs and replacement can be high, and the impact on customers potentially significant (USEPA, 2001). Asset management practices are generally used to prioritize pipe replacements and thereby identify investment strategies that, on one hand, avoid premature replacement of pipes (i.e., unnecessary pre-investment of funds), and on the other hand, avoid water main breaks, commensurate interruptions in service and the costs of damage. An effective asset management decision is dependent on the ability to determine the future performance of water mains by predicting water main breaks, and identifying how such breaks may occur. Much research has focused on the development of models for predicting water main breaks and pipe deterioration, but the use of such models is not common among utilities. In addition, the amount and quality of water main break data available for developing or implementing these models varies among utilities (Wood and Lence, 2006). Many utilities lack data and are not confident in the data they have and this is generally an impediment to their investing in pipe prediction models. However, they can create and relate data that can be useful for asset management (Wood et al, 2007). This paper develops a framework that guides utilities in identifying key data to be used in asset management in general and specifically for pipe break prediction modeling v and selecting the most appropriate model for predicting water main breaks. This 122 information may then be used to enhance the development of replacement priorities based on forecasted breaks, the maintenance of the database, and the identification of future data acquisition programs. It provides the utility with a method for considering future pipe breaks in the analysis of pipe prioritization strategies, and it incorporates existing tools for data management and analysis that are widely available and easy to implement by small and medium size utilities. The framework is applicable to utilities with varying amounts of data, and it is demonstrated here with a case study based on the Laity View area of Maple Ridge, B.C and constructed data. The following sections review the available techniques for predicting pipe breaks, the factors that influence break predictions, the framework developed for assisting in asset management of pipe networks and the results of the Maple Ridge example implementation of this framework. The framework can be applied without creating and constructing data, but the usefulness without such efforts is limited. 4.2 WATER MAIN BREAKS A number of authors analyze and report on the causes of breaks, including O'Day (1982), Marks et al. (1987), Male et al. (1990), Savic and Walters (1999), Rajani and Makar (2000), Rajani and Kleiner (2001) and Dingus et al. (2002). According to Rajani and Tesfamariam (2005), a combination of circumstances leads to pipe failure in most cases and different factors cause failure in different pipe networks. The causes of breaks include deterioration as a result of use (e.g., internal corrosion), physical loads applied to the pipe (e.g., traffic, frost), limited structural resistance of the pipe because of construction practices during installation and declining resistance over time (e.g., corrosion, aging factors). Dingus et al. (2002) surveyed the 46 largest American Water Works Association Research Foundation (AwwaRF) member utilities in 1997 and note multiple common 123 failure modes for cast iron pipe systems. Corrosion, improper installation and ground movement are the three most common causes of pipe failure. According to Levelton (2005), corrosion is dependent on a number of factors including material, soil type, chemical characteristics of soil, soil bacteria and stray electrical currents. Prediction modeling of water main breaks. Break prediction models have been developed to help the water industry understand how pipes deteriorate and when pipes will break in the future. These models are typically grouped into two classes - statistical and physical-mechanical models (Kleiner and Rajani, 2001). Statistical models use historical pipe break data to identify break patterns and extrapolation of these patterns to predict future pipe breaks, or degrees of deterioration. Physical-mechanical models predict failure by simulating the physical effects and loads on pipes and the capacity of the pipe to'resist failure over time. Statistical models are typically characterized as either deterministic or probabilistic equations (Kleiner and Rajani, 2001). Under the deterministic models, the pipe breakage is estimated based on a fit of pipe breakage data to various time-dependent equations, which may represent the cumulative pipe breaks as a function of time from date of installation or from the earliest date of available break data, most commonly are time-linear (Kettler and Goulter, 1985) or time-exponential functions (Shamir and Howard, 1979; Walski, 1982 and Kleiner and Rajani, 1999). Prior to fitting these functions, pipes are partitioned into groups that have similar characteristics, and the functions are evaluated for these groups. The ' characteristics used to sort the pipes are based on the factors that are assumed to influence breaks such as pipe age, pipe material, diameter, or soil type. Probabilistic models predict not only the failure potential, but the distribution of failure. These models are more complex than deterministic models and require more data. Examples of these include 124 cohort survival, such as KANEW (Deb et al. 1998), Bayesian diagnostic, break clustering, semi-Markov Chain and data filtering methods. Physical-mechanical models typically fall into one of two classes: deterministic models which estimate pipe failure based on simulation of the physical conditions affecting the pipe (Doleac et al, 1980, and Rajani and Makar, 2000), or probabilistic models that use a distribution of input conditions, such as rate of corrosion, to predict the likelihood and distribution of pipe failure (Ahammed and Melchers, 1994). These models have been developed primarily for cast iron and cement pipes. Physical models have significant data needs. Kleiner and Rajani (2001) suggest that only larger diameter mains with costly consequences of failure may justify the required data collection efforts for these models, and that statistical models based on fewer data may be used to gain insights for future performance. Rajani and Tesfamariam (2005), in using a probabilistic approach, suggests that breaks and causes of breaks for any particular water distribution network are system-specific and that a utility must create its system-specific model based on the deterioration factors that are relevant for that utility. While small and medium utilities typically have the capacity to use statistical deterministic models, the implementation of physical models is not practical due to the data collection efforts and model maintenance required. For small and medium utilities, it is often most important to gain insights about the rate of pipe breakage, that is, whether the quantity of expected breaks is increasing linearly or exponentially. Knowing the rate of change for utility managers is important because budgets and performance are based in part on future needs such as the number and rate of breaks in the system. Factors for predicting water main breaks. A number of studies identify factors for predicting water main breaks, though what is considered relevant data appears to be 125 specific to the system investigated. O'Day (1982) reviews break studies in Manhattan and Binghamton, New York, and cites a number of studies that use age as an indicator for predicting break rates for cast iron pipes. He notes, however, that age alone is a poor predictor of main break patterns and identifies the major determinants of water main break rates, as localized factors such as corrosion conditions, construction practices and external loads. He also finds that soil type affects external forces on water mains, such as shrink-swell, frost penetration and external corrosion. According to Jacobs and Karney (1994), pipe age range is an effective basis for models because pipes of a given age range are typically uniform with respect to manufacture, installation and to a large extent, operating conditions. Moreover, pipe installed in geographically contiguous sections often share similar soil conditions, installation conditions and pressure regimes. In their study, Jacobs and Karney group pipes based on material, diameter and fairly broad age ranges and develop regression relationships for pipe breakage versus age and versus pipe length. Savic and Walters (1999) suggest that the causes of water main failures may be split into pipe quality and age, type of environment, quality of construction workmanship and service condition and find that age, length, and diameter are the most important variables in influencing pipe bursts. Kettler and Goulter (1985) find that break rate, age and material are related for asbestos cement and cast iron pipes. In their study, no single type of failure of asbestos cement pipes exhibited a marked change in the rate of failure with time, while there were distinct changes in the failure rate with time for some types of failure in cast iron pipes. According to Male (1990), different manufacturing processes of cast iron pipes can account for differences in durability. Cooper et al (2000) apply a probabilistic approach to estimate trunk main failure probability, based on four key variables: number of buses per hour, pipe diameter, soil corrosivity and density of pipes in a given area. They find that 126 \ pipe age and material are important factors contributing to the break probability. Rajani and Tesfamariam (2005) show that long-term performance of buried cast iron is dictated by pit growth rate, unsupported length, fracture toughness and temperature differential. Data typically used in models are surrogates for factors that can explain breaks. For example, as shown in Table 4.1, the age of a pipe may represent the method of pipe manufacture or particular construction standards, as well as deterioration over time. Bedding material may be an indicator of a particular construction practice that induces physical stress, of the structural resistance of the pipe, or of the soil type. For example, in some utilities where native soil is used as backfill, the soil may not screened for rocks and other objects or properly leveled. As a result, this construction practice results in circumstances where a stress is induced on pipes and ultimately causes failures. In some cases, fines migration of corrosive native soil through particular bedding types can occur and create the potential for external corrosion. Soil type can represent corrosivity and potential for external pipe corrosion, this is also dependent on the pipe material. Availability of water main data within utilities for models. The amount of water main break data needed for extensive model development is not commonly available in utilities (Wood and Lerice, 2006), in spite,of best practices recommended by the National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARF (Deb et al, 2002). Most municipalities only have limited recorded pipe breakage histories and do not have much data for analysis (Pelletier et al, 2003). However, in many instances utilities may have more available data than they realize. They can apply approaches such as constructing and relating available data from archives, models and other such sources (Wood and Lence, 2006) to construct and link databases for analysis. 127 A key to any data management strategy is identifying the purpose for which one is collecting and analyzing the data, whether it is for asset management, compiling an inventory of assets or discovering the magnitude and nature of pipe breaks. There is growing interest in using Knowledge Discovery techniques such as data mining for water main break data (Savic and Walters, 1999). Knowledge Discovery is the process of identifying valid, novel, potentially useful and ultimately understandable patterns in data (Torra et al, 2004). Such patterns may help to identify factors that are related to breaks. 4.3 A FRAMEWORK FOR USING DATA AND PREDICTION MODELS TO IMPROVE ASSET MANAGEMENT The framework developed herein may be used to guide a utility in identifying the magnitude of its water main break problems today and in the future, and thereby enhance the development of strategies for prioritizing pipe replacements and data collection. Its salient feature is that it integrates break prediction or deterioration models that provide an indication of future pipe conditions with existing data, and thereby uses enhanced estimates of vulnerability for each pipe. It is designed to accommodate systems with limited data but is sufficiently flexible to adapt for additional information that may be acquired over time. Other design considerations include ease and transparency of use and facilitation of a decision-making process that is repeatable and defensible. Traditionally, utilities prioritize pipe replacements based on a combination of current management practices and historical pipe breakage data. Management practices include directives based on general guidelines, consequence assessments, legislative requirements, and other utility priorities. Rudimentary analyses employed interpret historical pipe break data, including location, time and date of break, and pipe diameter and 128 material, and typically has provided information regarding where and how many breaks are occurring, and what pipes are experiencing breaks (Kleiner and Rajani, 1999). Considering this information, the priority of the utility may be to replace water mains of a certain material or size, those in a certain area due to previous failures, those under roads that are to be re-paved, those that are currently undersized, or those that have significant consequences if failures were to occur, such as mains that serve hospitals. Some utilities may use a multiple objective approach, weighting each of a number of prioritization criteria, and assigning points to each pipe that describe the degree to which it meets a given criteria (Deb et al, 2002; Sargeant, 2003). For each pipe, the sum of the product of the weight and assigned points for each prioritization criteria is obtained and used to prioritize candidate pipes. The framework developed in this research is shown in Figure 4.1. In order to forecast pipe breaks, the historical data set may need to be expanded with data available from other sources within the utility and from external agencies. In addition to historical pipe breakage data, data for factors that may be important for predicting water main breaks as previously described may need to be obtained, including soil type, surface, bedding, and backfill material, type of road usage, or typical flow in area of break. This information may be consolidated by creating a schematic of data, which does not establish a new database per se, but draws from available data for analysis when required, as described by Wood and Lence (2006) and Wood et al, 2007. These data may be used directly in the prioritization process and as input to break prediction and deterioration models. The input to the models is developed by grouping pipes in which breaks have occurred based on factors that contribute to breaks. Material and diameter data are available to most utilities and should be considered as the minimum 129 factors on which to base pipe groups. The decision of whether to use a physical-mechanical or statistical break prediction model may be made at this point, because the pipe material determines whether a physical model exists for a given pipe and the diameter influences the practicality of applying such a model. For small and medium size utilities, the practical starting point is deterministic statistical models, which may be developed with readily available commercial software, including spreadsheets. More capable utilities may consider more complex statistical or even physical-mechanical models, however, the long term use and maintenance of these models is a serious consideration for those who choose these models. To evaluate the accuracy of a given statistical model, a portion of the break data should be used to develop the equations and the most recent portion of the break data should be retained as a holdout sample for comparison. For example, if a utility has twenty years of break history, it may choose to develop models based on the first fifteen years of data, and compare the model predictions with the remaining five years of actual breaks to assess the accuracy of the predictive model. While five years is a reasonable length of holdout sample, this is a function of the length of record, and data required to generate the statistical models. In developing and using statistical models, one must determine the amount of data that are required, the level of detail to be modeled, and the knowledge that will be gained. In order to determine the length of record required to develop a credible statistical model, the pipe break record used to model the system may be varied to evaluate the sensitivity of the model accuracy to the length of record used. In order to evaluate the important factors for predicting pipe breaks, the pipe break data may be subdivided into different sub-groups and models for each of these sub-groups 130 may be developed and compared in terms of their relative accuracy. This process naturally reduces the number of breaks within each sub-group used in performing statistical analyses, but may yield more credible models. Considering data that are typically available to utilities (Wood and Lence, 2006), potential sub-groups of pipes for these models, include those of a specified i) pipe material and diameter, which indicate pipe strength; ii) pipe material, diameter, and age which indicate pipe strength and age effects such as deterioration and construction practices; and iii) pipe material, diameter, soil type, and age which indicate pipe strength, interaction of the pipe material and the soil, and age effects. Should the utility have access to information regarding surface conditions, this may also be considered in forming the pipe sub-groups. With the knowledge gained from the model results, managers can then target pipes that have the highest predicted breaks or rates of breaks for prioritization. This information is also useful for identifying future investigative programs such as soil and pipe condition assessments and data acquisition strategies such as changes in field collection practices. The utility may also choose to verify the data or conduct investigative assessments to understand the deterioration of pipes that have significant breaks but cannot be accurately modeled. From these activities, new data can be created to improve the understanding of pipe deterioration factors. Pipe network management practices may also be altered based on the model results. Examples of such changes include identification of new design specifications such as the type of joints required for certain pipes in a particular soil, and introduction of corrective measures such as cathodic protection programs. In order to maintain relevance, it is recommended that models be routinely reviewed and updated as part of the detailed capital plan of the utility and to account for changes in the rate at which breaks are occurring as a 131 result of the changes in pipe management practices. Finally, break data should be kept current. 4.4 BREAK PREDICTION MODELS FOR LAITY VIEW, MAPLE RIDGE, BC The application of the framework is demonstrated using the Laity View area of Maple Ridge, BC, Canada. This area comprises 13 percent of the 335 kilometer distribution system for Maple Ridge, is representative of the urban area, experienced the same construction practices and has soil types found in the rest of the municipality, and is home to a population of approximately 6,000. The pipe materials found in the area are asbestos cement, cast iron, ductile iron and steel, and in diameters of 150, 200 and 250 mm. Pipe installation records began in 1959 and few pipes in Maple Ridge were installed before this date. The soil types found in the Laity View area are clay, silty-clay, silt and sand. Break data are available from 1983 to 2004. A total of 54 breaks occurred in this period, and seven of these occurred after the year 2000. Preliminary analysis of these data indicates that breaks are occurring in asbestos cement, cast iron and ductile iron pipes, in pipes that are greater than 15 years old, and in clay and silty-clay type soils (Wood et al, 2007). Given the 20-year history of record, the final five years from 2000 to 2004 was selected as the holdout sample. To investigate the important factors for predicting pipe breaks, pipes in the area were grouped based on the four types of sub-groups previously described. Information for surface material which included asphalt, concrete, and gravel or grass, is available for this region and thus another sub-grouping was examined that included pipes of a specified pipe material, diameter, age and surface material. The 132 combination of knowing which factors are important for predicting breaks and the common failure types for a network can provide insight on pipe deterioration behavior. Pipe age sub-groups were created by examining the data and identifying time periods in which a meaningful number of breaks occurred. For asbestos cement and cast iron pipes, these sub-groups were comprised of pipes with installation dates before 1959, between 1960 and 1974, and between 1975 and 1984. Asbestos cement and cast iron pipes were not installed in Maple Ridge after 1984. Ductile iron pipes were sub-grouped into pipes with installation dates between 1970 and 1979, 1980 and 1989, 1990 and 1999, and subsequent to 1999. The only steel pipes were installed in 1978 and are approximately 24 metres in length. These have not broken. Statistical deterministic equations for each group of Laity View pipes were developed for time-linear and time-exponential functions. Statistical deterministic equations were selected as most appropriate for Maple Ridge because they do not have a sufficient amount of pipes (e.g. grey cast iron) and data (such as remaining pipe wall thickness) to use physical-mechanical models or the resources to maintain complicated models (Maple Ridge has only three engineers on staff and relies on technical support staff for much of the engineering department duties and responsibilities). Statistical deterministic models can be easily taught to and applied by technical staff and require fewer data. The results should provide insights for future performance and improve Maple Ridge's current practices of prioritizing water main replacements which are based on experience. The time-linear equations for the cumulative number of breaks at year t are based on Equation 1. N(t) = A(t-to) + C (1) 133 Where N(t) is the cumulative number of breaks for the year t, t0 is the reference year, in the case of Laity View, 1983, A is a coefficient and C is a constant. Time-exponential equations for the cumulative number of breaks at year t are based on Equation 2. N(t) = A e k(t"t0) (2) Where A and k are coefficients and all other variables are as described above. As noted earlier in this chapter, these equations and their coefficients are specific to the Laity View area pipes and their respective sub-groups. Utilities should develop their own equations using their system-specific data, selected sub-groups and estimated coefficients (though they may choose to also use time-linear and time-linear regression). For each sub-group which had sufficient data, equations were derived using S-Plus® and , spreadsheet software to solve for the coefficients. A minimum of two breaks is required in order to estimate these equations, and thus equations could not be derived for all sub groups analyzed. For each sub-grouping analysis, the percent of all pipes for which an equation could be derived was estimated, as this is an indication of the extent of the network that may be modeled. The accuracy of the derived equations, henceforth referred to as models, was calculated as the percent error of model predictions relative to the cumulative breaks in 2004. Finally for time-linear models, R- squared estimates are reported. Results of break prediction models. The accuracy of the prediction results for both time-linear and time-exponential models for the various sub-groups are shown in Figures 4.2 through 4.6. The different amount of breaks and the rate of breaks among the 134 various groups suggest that there are differences in behavior for deterioration and breakage. For the material sub-groups, three sub-groups could be modeled; those for asbestos cement, cast iron and ductile iron pipes and these represent approximately one hundred percent of the pipe length in the network. As shown in Figure 4.2, the time-linear models are more accurate than the time-exponential models for asbestos cement and ductile iron pipes; the percent error for the time-linear models was 29 and 34, and the percent error for the time-exponential models was 210 and 136, for the asbestos cement and ductile iron pipes, respectively. The range of R-squared statistic for all of the time-linear models was 0.81 to 0.92 and the average R-squared value was 0.84. The results for the cast iron pipes indicate that while few breaks have occurred in these pipes they are occurring at an increasing rate. The accuracy of predictions for material and diameter sub-groups is shown in Figure 4.3. Here, seven sub-groups had sufficient number of breaks to be modeled and these represent 99 percent of the pipe length in the network. Again, with the exception of the cast-iron pipes, the time-linear models are more accurate than the time-exponential models. While the most accurate model is the time-linear model for the ductile iron pipe with a diameter of 150 mm, in general the performance of time-linear models for the asbestos cement and ductile iron pipes is similar. The average R:squared statistic for all of the time-linear models was 0.84 and the R-squared statistic was between 0.75 and 0.95. When age was considered, seven sub-groups contained sufficient number of breaks to be modeled, and these represent 56 percent of the pipe length in the network. As shown in Figure 4.4, the accuracy of the time-linear models for ductile iron pipes improved dramatically, indicating that age effects are important in predicting break rates for these pipes, and should be investigated. With respect to the asbestos cement pipes, the age delineated sub-groups suggest that different age groups of 150 mm asbestos cement pipes 135 are behaving differently with respect to breaks, the ability to accurately predict breaks differ and that the number of breaks for pipes installed between 1975 and 1984 are increasing. The R-squared statistic for all of the time-linear models was between 0.75 and 0.94 and the average R-squared statistic was 0.84. When pipe-soil interactions were considered, eight sub-groups could be modeled, but this represented only 38% of the pipe length in the network. The accuracy of predictions for the material, diameter, soil and age sub-groups is shown in Figure 4.5. These results suggest that the accuracy of predictions for pipes of the same material differs in different soils, even when they are installed at the same time. When clay is considered as a factor in the analysis, the accuracy of the time-linear models stayed the same or improved relative to analyses that considered only material, diameter and age. The average R-squared statistic for all of the time-linear models for these sub-groupings was 0.78 and ranged between 0.63 and 0.94. ' The accuracy of the models for material, diameter, age, and surface material sub groups is shown in Figure 4.6. Here, eight sub-groups could be modeled but these represent only 40 percent of the pipe length in the network. While the average R-squared statistic for all time-linear models for this case was 0.84 and ranged from 0.72 to 0.95, the accuracy of these models is no better than the accuracy of the time-linear models for the material, diameter, and age sub-groups alone. This indicates that, in contrast to soil type, surface material may not be an important factor to consider in predicting pipe breaks for Maple Ridge. Observations. It is important to note that some of the data that were created as part of an earlier water main break database constructed (see Chapter 3) contributed to increasing the accuracy of the applications of the break prediction models. In particular, the 136 • soil data provided insights for engineering staff by indicating that soil characteristics may be an influence on the rate of corrosion in certain pipes or bedding and backfill used in installing the pipes and thus confirmed the value of creating, relating and processing data. As a result, though the effort was significant, the data creation and mining provided insights and is useful for management decisions and asset management. Without created data, the analysis and use of prediction models would be limited. For Maple Ridge, the pipe groups associated with models that accurately predict high break rates are: 250 millimeter diameter asbestos cement pipes installed in clay soil between 1960 and 1969 and 150 millimeter diameter asbestos cement pipes installed in clay soil between 1960 and 1969. As a result of these analyses, consideration of the break rates of various pipe groups and discussions with operations and maintenance staff,, asbestos cement pipes will be prioritized for replacement (along with cast iron pipes when opportunities arise) and further attention will be given to collecting data on ductile iron pipes. More importantly, because soil type was identified as an important factor in modeling breaks, a soil sampling program was undertaken to improve the utility's information regarding soil resistivity, pH, chlorides and soil type. A preliminary pipe sampling program was implemented at the same time to collect information on asbestos cement pipes and ductile iron pipes in the area. As a result of the sampling programs, the importance of bedding and backfilling practices and construction inspections was identified and changes in construction specifications and inspection practices are being developed. Plans are underway to apply this framework to the rest of the Maple Ridge network. Ultimately, scheduling of the pipe replacements and budget estimates will be undertaken in conjunction with other management considerations such as road rehabilitation. 137 An ongoing problem for utility managers is the allocation of scarce resources for both data collection and analysis. One approach for determining the value of the framework and a supporting data collection program is to evaluate the value of the additional information obtained. The value of additional information may be estimated by comparing the decision that would be undertaken without the additional information with the decision that would be undertaken with the additional information (Schuyler, 2001). For example, for Laity View, the value of developing statistical models that incorporate soil data may be determined by comparing the cost of replacing the group of pipes that has the highest break rates based on data for material, diameter and age (i.e., 250 mm diameter asbestos cement pipes) with that of a replacement strategy that considers replacing only those 250 mm diameter asbestos cement pipes in clay soils. If all 250 mm diameter asbestos cement pipes, with a total length of 646 meters were to be replaced, the total cost of replacement would be $193,800 (assuming a replacement cost of $300 per meter). By applying the framework it was determined that all the breaks in these pipes occurred in clay soil. If it is assumed that all future breaks of this pipe type will occur in clay soils, replacement of these pipes, with a total length of only 258 meters would cost $77,400. Thus the value of the analysis and the soil information is approximately $193,800 - $77,400 = $116,400. While this additional information may not always lead to savings in terms of reducing the cost of pipe replacement, for example in cases where the all pipes of a certain material, diameter and age were installed in the same soil type, the information regarding soil type may still be of benefit. This information could be used to improve the installation practices or justify corrective measures. 138 4.5 CONCLUSIONS Predictive modeling is useful for identifying replacement needs over time. However, utilities do not commonly use predictive modeling as part of their asset, management practices. There are no common databases for break analysis or common condition indices, and few utilities undertake condition assessment (Grigg, 2004), all of which hinders industry-wide use of predictive modeling. The framework presented in this paper improves upon the traditional pipe prioritization approaches that only look at the past history in aggregate and do not necessarily take into account trends and timing of future breaks. An advantage of the framework is that it can be applied by small to medium size utilities with limited information and commonly used analytical tools. For example, for Maple Ridge, while useful information was gained by investigating soil type, reasonable insights may have been drawn from analyses that considered only material, diameter and age, information that is typically available to most utilities. Because factors that cause breaks vary among utilities, utilities may find individually that creating more data (such as by collecting traffic loading, backfill and pressure data) and relating data for mining and analysis is worthy of the effort and expands the use of this framework. With this in mind, the framework is flexible and allows for consideration of any available data. In addition to guiding water main replacements, the framework may also be used to identify the key data for predicting water main breaks. Because there is variability in the causes of pipe breaks among different utilities, in order to understand the performance of their system,'utilities should collect data as' identified in recommended Best Practices; see National Guide to Sustainable Municipal Infrastructure (NGSMI, 2002) and AWWARP (Deb et al, 2002). Additional information may often be obtained efficiently at the time of the break repair by revising forms to collect 139 more information, such as bedding or backfill material (Wood and Lence, 2006). Training will often be required, and it is prudent to verify data. Convincing staff to collect data may be an obstacle, but involving them in decision making can be a way to gain support. By using models to predict future breaks, reviewing the accuracy of the predictions and updating the models, a utility can improve its asset management practices. 4.6 ACKNOWLEDGEMENTS The authors gratefully acknowledge the District of Maple Ridge for providing data arid support in the form of employee resources, and Professors A. D. Russell, J. W. Atwater at the University of British Columbia (UBC) for their insights and suggestions. Mr. W. Liu assisted with the preparation of break data and Mr. A. Malyuk of the District of Maple Ridge assisted in data processing and compilation.' 140 4.7 REFERENCES AWWA and EES, Inc., 2002. New or repaired water mains. Available on-line at http://www.epa.gov/safewater/tcr/pdf/maincontam.pdf., Accessed February 12, 2006 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, 2:2000:97-103. Deb, A.K., Hansit, Y. J. and Grabultz, F.M., 1998. Quantifying Future Rehabilitation and Replacement Needs of Watermains. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Deb, A. R., Grablutz, F.M., Hasit, Y.J., Synder, J.K., Longanathan, G.V. and Agbenowski, N., 2002. Prioritizing Water main Replacement and Rehabilitation. 6666 West Quincy Avenue, Denver, CO. 80235, AWWA Research Foundation: 200. Dingus, M., Haven, J., and Russell, A., 2002. Nondestructuve, Noninvasive Assessment of Underground Pipelines. AWWARFReport 90873, AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO 80235. Doleac, M. L., Lackey, S. L., and Bratton, G. N., 1980. Prediction of time-to-failure for buried cast iron pipe. Proceedings of A WW A Annual Conference, Denver, CO. 141 Grigg, N. S., 2004. Assessment and Renewal of Water Distribution Systems. 6666 West Quincy Avenue, Denver, CO, AWWA Research Foundation. Jacobs, P., and Karney, B., 1994. GIS development with application to cast iron water main breakage rates. 2nd International Conference on Water Pipeline Systems, Edinburgh, Scotland, 53-62. Kettler, A. J. and Goulter, I. C, 1985. An analysis of pipe breakage in urban water distribution networks. Canadian Journal of Civil Engineering, 12:2: 286-293. Kleiner, Y., and Rajani, B.B., 2001. Comprehensive review of structural deterioration of water mains: statistical models. Urban Water, 3:3: 131-150. Kleiner, Y., and Rajani, B.B., 1999. Using limited data to assess future needs. Journal AWWA, 91:7: 47-62. Levelton Consultants Ltd., 2005. Maple Ridge Water Works Corrosion Investigation. Report 2705-0380, Levelton Consultants Ltd. 102-19292 60th Avenue, Surrey, BC. Male, J. W., Walski, T., and Slutsky, A. FL, 1990. Analyzing Water Main Replacement Policies. Journal of Water Resources Planning and Management, ASCE, 116:3: 362-374. 142 Marks, D. H., Andreou, S., Jeffrey, L., Park, C, and Zaslavsky, A., 1987. Statistical Models for Water Main Failures. EPA/600/5-87/003, Water Engineering Research Lab, USEPA. Cincinnati, OH. NGSMI, 2002. Deterioration and Inspection of Water Distribution Systems. Infraguide -Potable Water. Ottawa, Canada, National Guide to Sustainable Municipal Infrastructure. O'Day, D. K., 1982. Organizing and analyzing leak and break data for making main replacement decisions. Journal A WWA, 74:11: 588-594. Pelletier, G., Mailhot, A., and Villeneuve, J.-P., 2003. Modeling water pipe breaks - three case studies. Journal of'Water Resources Planning and Management, ASCE, 129:2: 115-123. Rajani, B. B., and Makar J., 2000. A methodology to estimate remaining service life of grey cast iron water mains. Canadian Journal of Civil Engineering, 27: 1259-1272. Rajani, B. B., and Kliener, Y. 2001. Comprehensive review of structural deterioration of water mains: physically based models. Urban Water, 3:3: 151-164. Rajani, B. B., and Tesfamariam, S., 2005. Estimating time to failure of ageing cast iron water mains under uncertainties." Water Management for the 21st Century, University of Exeter, UK., 1-7. 143 Sargeant, D., 2003. Water Main Rehabilitation Prioritization. AWWA 2003 Seminar -Infrastructure: above and below ground, Anaheim, CA. 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. Schuyler, J., 2001. Risk and Decision Analysis in Projects. Project Management Institute, Four Campus Blvd. Newtown Square, Pennsylvania. Shamir, U. and Howard, C. D. D., 1979. An analytic approach to scheduling pipe replacement. Journal AWWA, 71:5: 248-258 Torra, V., Domingo-Ferrer, J., and Torres, A., 2004. Data mining methods for linking data coming from several sources. 3rd Joint Un/ECE-Eurostat Work Session on Statistical Data Confidentiality, Monographs in Official Statistics, Luxembourg, Eurostat, 143-150 USEPA, 2001. Drinking Water Infrastructure Needs Survey. Second Report to Congress. EPA 816-R-01-004, U.S. Environmental Protection Agency Office of Water, Washington, DC. Walski, T.M., 1982. Economic Analysis of Water Main Breaks. Journal of Water Resources Planning Management Division, ASCE, 108:3: 296-308. •J 144 Wood, A. and Lence, B.J., 2006. Assessment of Water Main Break Data for Asset Management. Journal AWWA, 98:07. Wood, A., Lence, B.J. and Liu, W., 2007. Constructing Water Main Break Data for Asset Management. Journal AWWA. 99:01. 145 Table 4.1 Typical data used in models and factors for which they are a surrogate Surrogate Factor Age Method of pipe manufacture, construction standards, deterioration over time Pipe material Construction practice, method of manufacture, failure mechanisms and causes, joint failures Pipe diameter Wall thickness and resistance to beam loading, pipe use, method of pipe manufacture, construction standards Type of pipe lining Method of pipe manufacture, resistance to corrosion Bedding and backfill material Physical stress on pipes caused by construction practices, structural resistance, soil type, fines migration Pipe protection (wrapped/anodes) Structural resistance, life expectancy, construction practice, method of pipe manufacture Pipe condition Remaining life Soil type Soil corrosivity, physical loading on the pipe such as swelling and frost, level of pipe protection, ground water effects such as draining ability or corrosion, construction practice, bedding and/or backfill material Under a boulevard or roadway Physical loading from surface loads such as traffic, road salt effects Depth of cover Physical loading on the pipe from the weight of soil Surface material/type Physical loading from surface use Normal operating pressure Internal pressure on pipe structure Typical flow in area of break Physical impact from factors such as accelerated internal corrosion from low flow mains Traffic classification Physical loading from surface loads such as traffic volumes and wheel loads Road/surface usage Physical loading from surface loads 146 Figure 4.1 Improving asset management using pipe break prediction models Management considerations Management strategies Master growth plans, construction of other infrastructure, risk assessments, legislative requirements, etc. "'Prioritization of pipe replaccmenl'for asset ^Ifl^wiagernent. Inputs'indude^he:,' ^consideration of managemehfrhistbiical and future predictions • [Historical data \ Stratify pipes into groups based on factors Data construction, linking and updating of data • Investigative programs '.- -• Verification program of constructed data • Pipe network practices assessments Predictions of , future breaks Apply models, compare results with holdout samples 'and evaluate . model performance Target pipes with critical or highest break rates as part of asset management replacement strategy 147 Figure 4.2 Degree of accuracy of time-linear and time-exponential predictions for material groups 250% 200% 150% 100% 50% 0% -50% -100% Asbestos cement pipes Cast iron pipes • Time-linear • Time-exponential Ductile iron pipes Note In 2004, there were a total of 32 breaks in asbestos cement pipes, 5 in cast iron pipes and 17 in ductile iron pipes. 148 Figure 4.3 Degree of accuracy of time-linear and time-exponential predictions for material and diameter groups 300% Notes AC denotes asbestos cement pipes, CI denotes cast iron pipes and DI denotes ductile iron pipes. Pipes are grouped by material and diameter (in millimeters). For example, AC 150pipes are asbestos cement pipes of 150 millimeters diameter. In 2004, there were 24 breaks in AC 150pipes, 5 in AC 200, 3 in AC 250, 4 in CI 150, 9 in DI 150 and 2 in DI 250 pipes. 149 Figure 4.4 Degree of accuracy of time-linear and time-exponential predictions for material, diameter and age groups 250% 200% 150% 100% 50% 0% -50% • Time-linear • Time-exponential AC 150 AC 150 AC 150- AC 200 AC 250 DI 150 1980- DI 200 1980-1960-1969 1970-1974 1975-1984 1970-1974 1960-1969 1989 1989 Notes AC denotes asbestos cement pipes, CI denotes cast iron pipes and DI denotes ductile iron pipes. Pipes are grouped by material, diameter (in millimeters) and age. For example, AC 150 1960-1969 pipes are asbestos cement pipes of 150 millimeters diameter installed between 1960 and 1969. In 2004, there were 12 breaks in AC 150 1960-1969pipes, 7 in AC 150 1970-1974, 5 in AC 150 1975-1984, 3 in AC 200 1970-1974, 3 in AC 250 1960-1969, 4 in DI 150 1980-1989 and 3 in DI 200 1980-1989 pipes. 150 Figure 4.5 Degree of accuracy of time-linear and time-exponential predictions for material, diameter, soil and age groups 400% 350% 300% 250% 200% 150% 100% 50% 0% • Time-linear • Time-exponential AC 150 AC 150 AC 150 AC 250 Dl 150 AC 150 AC 200 DI200 Clay Clay Clay Clay Clay Silt 1960- Silt 1970- Silt 1980-1960- 1970- 1975- 1960- 1980- 1969 1974 1989 1969 1974 1984 1969 1989 Notes AC denotes asbestos cement pipes, CI denotes cast iron pipes and DI denotes ductile iron pipes. Pipes are grouped by material, diameter (in millimeters), soil type and age. For example, AC 150 Clay 1960-1969 pipes are asbestos cement pipes of 150 millimeters diameter installed in clay soil between I960 and 1969. In 2004, there were 7 breaks in AC 150 Clay 1960-1969pipes, 7 in AC 150 Clay 1970-1974, 2 in AC 150 Clay 1975-1984, 2 in AC 250 Clay 1960-1969, 4 in DI 150 Clay 1980-1989, 3 in AC 150 Silt 1960-1969, 3 in AC 200 Silt 1970-1974 and 2 in DI200 Silt 1980-1989 pipes. 151 Figure 4.6 Degree of accuracy of time-linear and time-exponential predictions for material, diameter, age and surface material groups 200% 150% 100% 50% 0% -50% • Time-linear • Time-exponential *-' v." v.-^ ^ ^ ^ ^2 ^ %, 'S>± '<°> <9> ^ *3 \ \ V V V V \ \ \ \ \ * \ \ \ % \ % \ c?o °;o Notes AC denotes asbestos cement pipes, CI denotes cast iron pipes and DI denotes ductile iron pipes. Pipes are grouped by material, diameter (in millimeters), age and surface material. For example, AC 150 1970 - 1974 Asphalt pipes are asbestos cement pipes of 150 millimeters diameter installed between 1970 and 1974 under an asphalt surface. In 2004, there were 5 breaks in AC 150 1970-1974 Asphalt pipes, 7 in AC 150 1960-1969 Gravel/grass, 5 in AC 150 1970-1974 Gravel/grass, 3 in AC 150 1975-1984 Gravel/grass, 2 in AC 200 1 970-1974 Concrete, 2 in AC 250 1960-1969 Gravel/grass, 4 in DI 150 1980-1989 Gravel/grass and 2 in DI 200 1980-1989 Asphalt pipes. 152 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 153 5.1 SUMMARY OF RESEARCH GOALS Asset management of water systems involves assessing when to replace aging and deteriorating pipes. The goal of this research is to assist small to medium size utilities with identifying, collecting and constructing relevant pipe break data to analyze their pipe network, and using break predictions to inform their water main replacement strategy and guide their data acquisition programs. First, the data that are collected and available for analysis across utilities in North America are identified. Next, a methodology to create and link data obtained from various data sources that can be used in utilities of all sizes to construct databases is developed. Finally, a framework to assist the prioritization of water main replacements and data acquisition based on predicting future water main breaks within a given water distribution system is presented. The framework is applicable for the range of data available in typical water utilities, acknowledges existing industry needs and practices and may help managers to acquire and use available data. Much of the focus of water main break research has been on the data rich and technically sophisticated larger utilities. However, small and medium size utilities need the techniques developed in this research because they have scarce resources for asset management within their organizations. Compared with larger utilities, they do not have the staffing expertise or the capacity for training, monitoring or developing back-up systems (Ontario PIR, 2005). Within these utilities, there is little or no reliable documentation regarding the location, capacity, condition and adequacy of pipe network elements for meeting present or future needs (Myers, 2001). These utilities may also lack the financial and organizational resources to implement a complex asset management program or lack the historical data or tools to fully analyze their system. They need relatively inexpensive techniques. This research provides adaptable approaches for efficiently acquiring data 154 regarding water main breaks, compiling, analyzing and using the data to predict future water main breaks and improving prioritization of pipe replacements. Selecting the case study. The research in this thesis uses real data from the District of Maple Ridge, BC. Maple Ridge was selected because it is a medium size water utility, is similar to municipalities that have undergone urbanization over the past decades and possesses data that were made available to me. Moreover, Maple Ridge was interested in using break prediction models to improve its asset management practices and while it lacked a comprehensive water main break database, it was receptive to developing databases for analysis. It also needed a strategy to manage and maintain the data after they were used to predict water main breaks. Appendix C is a summary description of the Maple Ridge water system. 5.2 CONCLUSIONS Water main break data collection is evolving and industry practices do not match best practices recommended by NGSMI (2002) and Deb et al (2002) at this time. Utilities need a strategy for data quality improvement that will help them deal with challenges such as difficulty in mobilizing financial and human resources, absence of historical data, lack of knowledge of current organizational practices, low reliability of previously collected data, difficulty in prioritizing data collection, and the need to develop effective data storage programs. In general, utilities can be classified as those possessing expanded, intermediate, limited or minimal data. While both physical and statistical models have been developed for predicting pipe deterioration and for developing water main rehabilitation plans, it is evident that the choice and application of these models are limited by the data that utilities have regarding water main breaks (Rajani and Kleiner, 2001; Kleiner and Rajani, 2001). 155 For practitioners and researchers alike, characterization of these data classes may be used to inform the development of new asset management techniques that are tailored to the data limitations that utilities face. Utilities may also improve their data collection by modifying their current practices and more significantly, they can seek out alternative data sources from which to analyze breaks and ultimately predict future breaks. The alternate sources that are identified in this research can yield information for researchers and managers alike. The process of creating the data schematic as developed in this research is useful for linking multiple databases in order to compile and analyze data. It expands the data available for analysis for both present and future applications and allows decentralized data input and management. More importantly, it is a flexible approach that all utilities can employ without significant resource commitments. It reduces problems related to data ownership among units within organizations regarding data collection, management and dissemination of infrastructure information. The framework developed in this research for linking data is flexible, anticipates the evolution of data collection, building, verification and storage and allows for a variety of users. It does not abruptly disrupt data collection and warehousing practices and it allows managers to continue to expand data collection because databases are decentralized. It is flexible and can be easily adapted to all utilities, whether they are small, medium or large, and regardless of the uniqueness of the data collected and organizational framework. This technique can also incorporate tacit data. The importance of capturing tacit data will increase over the next decades as baby boomer staff retire. This thesis develops an approach for using break prediction models for identifying replacement needs over time and uses improvement in model accuracy as means of 156 identifying the key data for predicting future water main breaks and informing future data acquisition strategies. Traditional approaches for prioritizing pipe replacements do not incorporate break predictions. The framework developed in this research allows for the construction, assessment and use of any available data by any size utility. 5.3 OBSERVATIONS General observations. A number of observations arise from this research. While this research primarily focuses on engineering science, a number of the observations relate to management science and the relationships organizations and people have with data utilization. While these topics arise from the research, they are not addressed in this thesis and future research in these areas will be valuable. 1. As noted in the thesis, data are collected throughout an organization by various departments and staff for various purposes. For data to be used corporately, executives must acknowledge and address data ownership among departments and managers. 2. Because asset management is a corporate responsibility, proprietary issues and organizational compartmentalization can pose major challenges to implementing an asset management program. Corporate objectives of knowledge management (e.g., database development and maintenance) should be established and most importantly, be accepted by those responsible for collecting, managing and analysing the data. These objectives should be established and promoted by utility executives throughout the utility because in some organizations knowledge may be viewed as power and data may be interpreted as a surrogate for knowledge. 157 3. An important task for managers is to establish and encourage a culture of knowledge sharing (Connelly, 2000). Motivating and coaching staff to share knowledge can be a major challenge. As well, managers themselves are not immune to the habit of hoarding information and knowledge, though this tendency may be reflective of the corporate culture and the individual's relationship with their subordinates, peers and supervisor. 4. A significant need exists in utilities for capturing institutional memory and data that currently exist and which will be lost as employees retire. Commonly, large amounts of institutional memory are not recorded. For most utilities, there is a need to capture the memory of operating departments since they are more often characterized as "action oriented" and "hands on" rather than "paper loving". Managers must find solutions to address this important issue. 5. Water utilities also face recruiting, training and capacity building challenges. The view that staff will join an organization early in their career and spend thirty years in that organization is becoming extinct. For example, District of Maple Ridge recruiters consider obtaining five to ten years of service from a non-union manager (before he or she leaves the organization) as a valuable experience. Compounding this, currently, there is significant competition among BC employers to recruit and retain engineers. Utilities also require training and capacity building programs for new staff as well as existing staff. One challenge for many utilities is that training budgets are often the first to be cut in times of financial restraint (because budgets are primarily viewed as supporting expenditures and training as a discretionary expense). This was the case in Maple Ridge's history. The success of training programs is dependent on the marketing and implementation of the training as well as the quality of the training. This aspect of organizational development is 158 also important because according to Jacobson and Prusak (2006), organizations will receive greater value from information by developing strategies and training staff to help them use what they have rather than by searching for more data. 6. Utility managers also need to inspire and motivate positive change. During the course of the research, the author observed that managers find that support for change (from their supervisor or employees) is not automatic, and factors that influence the support or change include the career goals of their supervisor, their level of influence within the organization's politics, staff motivators and how the utility is governed. For example, changes in data collection practices because of legislation tend to be more rapidly implemented than other reasons for changes (NSGMI, 2003). 7. A final observation is that organizational leadership, structure and behaviour have a major influence on a utility's focus and practices. Based on what I have observed of various utilities, the manner in which engineering and maintenance departments function and in which responsibilities are distributed within a utility can lead to duplication of work (because of ambiguity or the desire to possess the information or responsibility) or the incompletion of work (because each department may deny responsibility and assume that the other is addressing the issue). Furthermore, utility executives need to provide strategic thinking and leadership of the organization, while balancing corporate and individual goals and strengths. Without this, staff efforts may be misplaced or frustrated or may flounder. Professional applications. This thesis has many practical applications for small and medium size utilities. It provides information on data collection and presents an approach that a water utility may adopt should they decide to use break prediction models. Given the work reported in Chapter 2, utilities can assess their practices against those of their peers and those recommended as best practices. They can identify and 159 improve their data collection practices and use the survey results to facilitate discussions within their organization regarding the availability and storage of data. They may find additional data available from other sources such as those identified in this thesis. If there are insufficient data for analysis, they may compile and link data and construct additional databases to extend the breadth of data as demonstrated in Chapter 3. Following that example, they can construct a data schematic for their organization to analyze their system for asset management in general and to explore the underlying causes of water main breaks. Once they have constructed and analyzed the data which is useful despite the effort required, they can then use the framework developed and demonstrated in Chapter 4 to . incorporate the use of break prediction models to improve asset management and to inform their future data acquisition and storage programs. The approaches developed in Chapters 3 and 4 address the problems faced by utilities, and are designed to be adaptable to their needs and the examples used are based on real data. Researchers will also benefit from the work reported in Chapters 2 and 3 which identifies the data that are available for developing future asset management tools and how utilities can access and construct data for research purposes. Lessons learned. While the research has a number of applications, a number of lessons were learned. When applying the research as described in Chapter 3, a utility requires staff and external help in two areas. Firstly, they require technical assistance in the following tasks: • Deciding on the objective of the exercise. A suggested objective is presented in the thesis and other purposes are also identified. These should be developed by managers and engineers. 160 • Creating data by reviewing different sources of data (internal and external to the utility). The thesis presents a number of techniques such as buffering a collection of data, interviewing people to capture tacit information, examining previous practice standards and conducting surveys. Some of this work can be undertaken by consultants and by training existing staff. • Deciding upon the databases to be related and the nature of the relationships. Some databases and relationships are suggested in this thesis, but these decisions should be determined jointly by utility and data managers. Secondly, the utility will require more technical resources to: • Review the extent of data availability throughout the organization. The thesis suggests sources to explore. • Create data from paper records. . • Establish links among various databases. • Map the links and various databases. This can be undertaken by line staff and this thesis provides guidance for this task. • Perform some exploratory analyses. It should be noted, that the technical work requires some understanding of the data and some ability to read design drawings. In addition, the project manager applying the research should have strong communication and interpersonal skills to determine where data may be available and to obtain consensus on data sharing among various departments. They should be tactful, diplomatic, persuasive, patient and persistent and should not be 161 easily discouraged because applying the work will take time within any utility. Sponsorship and support from the utility executive should be obtained which will help when working through departmental and corporate issues. Finally, it is not unrealistic to expect that a project such as described in Chapter 3 would take a 1.2 person years to two person years of effort and a project such as described in Chapter 4 may take as much as one person year. Validation of the work. While the proof of this work lies in the arguments and demonstration of the approaches developed, the validation of the contribution will be the acceptance of the methods proposed in practice and whether they are employed by utilities to improve asset management and data productivity. In using their own system-specific data and the thesis as guidance, this research may be applied and adapted to utilities of various sizes possessing a range of abilities for improving their asset management practices. Ultimate validation may be undertaken by evaluating the success of such applications. Further research. To make the application of this research more useful, further research is needed in the following areas. 1. The survey was developed in English and forwarded to those communities listed in Appendix E. Only one response was received from the Anglo-Quebec municipalities. Investigation of French-Canadian water main break data collection practices could inform these communities and the larger water utility industry, although the data available to Pelletier et al. (2003) suggest that French-Canadian practices are no different than those identified in this research. Regardless, translating and distributing the survey throughout Quebec is a potential future project. 2. Testing the framework of using water main break prediction models in a utility that has more water main breaks, a longer history of breaks or different climate would add 162 further confirmation of the applicability of the research to other utilities. The framework has been designed and demonstrated to be robust, though further confirmation would be useful. Furthermore, with a larger number of breaks to analyze, it is possible that other key data may be identified as important for determining the accuracy of break prediction models. 3. Application of a combination of statistical and physical-mechanical models should be explored within the framework provided herein. The accuracy of predictions for the different types of models could also be assessed to give further guidance for choosing models. In addition, combining the models in a case study could provide insights that can be used to potentially demonstrate or improve the robustness and flexibility of the framework developed herein. 4. Finally, the topics identified in the general observations section should be further explored to assist utilities in applying the research in practice. These issues include how utilities can increase knowledge sharing (of data and practices) within their organization and identifying key organizational structures, leadership attributes and motivators that are needed to improve water utility management. While small and medium size utilities are able to make operating policy and practice changes quickly because decision making typically rests with fewer staff than is the case in larger organizations, they are like most organizations. They are limited by resources including leadership and motivation skills and training to execute large changes. Research in determining and assessing the required skills, personality aptitudes, abilities, training and human resource needs for institutionalizing the approaches presented in Chapters 3 and 4 would be useful in extending the impact of this work. 163 Future research interests of the author. Three areas that I am considering for further research arising from the work in this thesis include applying additional approaches to analyzing break data, applying the data construction approach and framework developed herein to other infrastructure assets, and enhancing utility knowledge management practices. 1. Applying Knowledge Discovery tools such as artificial intelligence to analyse large water main break databases. This has potential to assist those water utilities fortunate enough to have a breadth of data and a long history of breaks and to reduce the necessity for having many experienced staff to manage infrastructure assets. 2. Applying the concept of linking and relating data (as shown in the water main break schematic shown in Chapter 3) and using statistical deterministic and other prediction models to sewerage and drainage systems. Extending the approach presented in this thesis to these infrastructure systems has several advantages. These systems are usually owned and managed within the same managerial unit and typically have similar replacement values as water systems. Nonetheless, there are differences among the systems. For example, sewer systems are easier to inspect using video cameras and a common condition rating system for them exists. Though drainage systems could use the same condition rating system, they are rarely evaluated and wear out at a faster rate. 3. Examining how to manage knowledge of a system within an organization. The data schematic and concept of relating relational databases as discussed in this thesis has potential to play a significant role in connecting data for various services among the members of an organization, but research is needed to assist utilities in capturing, managing, maintaining and using this knowledge. Example research questions include: what are the best techniques for interviewing and recording tacit data from retiring 164 employees, what is the optimal organizational structure for sharing knowledge among engineers and maintenance staff, and what are the most efficient and effective approaches that utilities can employ to make data accessible across an organization? 5.4 CLOSING REMARKS The approaches developed herein are easily applied. There are significantly more small and medium size utilities than larger utilities (Vanier and Rahman, 2004) and these utilities can benefit from applying the work. The approaches have been demonstrated with real data and may provide utilities with insights and guidance for allocating their scarce resources, considering their current limitations and future needs. The application is practical and can be implemented incrementally. The thesis chapters are at various stages of publication or review. Chapter 2 was published in the July 2006 issue of the Journal of the AWWA. Chapter 3 is published in the January 2007 issue of the Journal of the AWWA. Chapter 4 was submitted to the ASCE Journal of Infrastructure Systems in July, 2006. An invited presentation on the results of the first paper was made at the September 2006 Annual Public Works Association of BC (a chapter of the American Public Works Association) Conference in Qualicum Beach, BC. In addition, the work will be presented as part of a water system asset management workshop to engineers in the Greater Vancouver area. In closing, as a result of the work conducted in this thesis, Maple Ridge has implemented a number of the findings to improve its water main break data collection practices. It has undertaken soil surveys, tested a number of pipe samples and verified some of the linked and related data. It is also developing a utility wide data schematic that will form the basis of a data warehouse strategy that will be applied to the entire water system 165 and sanitary sewer system. Finally, the framework for determining the data that contribute to pipe failures will be implemented beyond the case study area to improve Maple Ridge's asset management practices. \ 166 REFERENCES 167 REFERENCES Ahammed, M. and Melchers, R.E., 1994. Reliability of underground pipelines subjected to corrosion. Journal of Transportation Engineering, 120:6: 989-1003. Andreou, S., Marks, D. and Clark, R., 1987. A new Methodology for Modeling Break Failure Patterns in Deteriorating Water Distribution Systems: Theory. Advances in Water Resources, 10:1:2-10. Elsevier Science, B.V. ASCE (American Society of Civil Engineers), 1999. American Society of Civil Engineers report card and issue briefs. Public Works Management and Policy, 4:1: 58-76. AWWA (American Water Works Association), 2004. WATER: \STATS- The Water Utility Database. 2002 version [CD-ROM]. American Water Works Association, Denver, CO. 80235. AWWA and EES, Inc., 2002. New or repaired water mains. Available on-line at http://www.epa.gov/safewater/tcr/pdf/maincontam.pdf., Accessed February 12, 2006. Canada. Statistics Canada, 2006. The age of public infrastructure in Canada, by Valerie Gaudreault and Patrick Lemire. January 2006. (Catalogue No. 11-621-MIE2006035). Ottawa, Minister of Industry. 168 Connelly, C. E. 2000. Predictors of knowledge sharing in organizations. Master of Science Thesis, Queens University, Kingston, Ontario, Canada. 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, 2:2000: 97-103. Davis, D.N., 2000. Agent-based decision support framework for water supply infrastructure rehabilitation and development. Computers, Environment and Urban Systems, 24:2000: 173-190. Deb, A. R., Grablutz, F.M., Hasit, Y.J., Synder, J.K., Longanathan, G.V. and Agbenowski, N., 2002. Prioritizing Water main Replacement and Rehabilitation. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Deb, A.K., Hansit, Y. J. and Grabultz, F.M., 1998. Quantifying Future Rehabilitation and Replacement Needs of Watermains. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. DeSilva, D., Davis, P., Burn, L. S., Ferguson, P., Massie, D., Cull, J., Eiswirth, M. and Heske, C, 2002. Condition Assessment of Cast Iron and Asbestos Cement Pipes by In-pipe Probes and Selective Sampling for Estimation of Remaining Service Life. Proceedings of the 20th International No-Dig Conference, Copenhagen, Denmark. 169 Dingus, M., Haven, J. and Russell, A., 2002. Nondestructive, Noninvasive Assessment of Underground Pipelines. AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. Doleac, M. L., Lackey, S. L. and Bratton, G. N., 1980. Prediction of time-to-failure for buried cast iron pipe. Proceedings of A WW A Annual Conference, Denver, CO. Earth Tech, 2004. National Water and Wastewater Benchmarking Initiative Final Results -June 2004. Earth Tech, Burnaby, BC. Earth Tech, 2003. National Water and Wastewater Benchmarking Initiative- Water Utility Definitions 2003 Release B., Earth Tech, Burnaby, BC. FCM, 2001. Backgrounder: Protecting Canada's Quality of Life. Federation of Canadian Municipalities, (http://www.fcm.ca/infra/throne-a.html). Accessed December 2002. \ Golder Associates Ltd., 2002. Geotechnical Input to the seismic vulnerability assessment for the District of Maple Ridge, B.C. 500-4260 Still Creek Drive, Burnaby, BC, V5C 6C6. Grigg, N.S., 2005. Assessment and Renewal of Water Distribution Systems. Journal AWWA, 97:2: 58-67. Grigg, N. S., 2004. Assessment and renewal of water distribution systems^ AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. 170 Habibian, A., 1992. Developing and utilizing databases for water main rehabilitation. Journal AWWA, 84:7: 75-79. Jacobs, P. and Karney, B., 1994. GIS development with application to cast iron water main breakage rates. 2nd International Conference on Water Pipeline Systems, Edinburgh, Scotland, BHR Group. 53-62. Jacobson, A. and Prusak, L., 2006. The cost of knowledge. Harvard Business Review, 84:11:34 Kettler, A. J. and Goulter, I. C., 1985. An analysis of pipe breakage in urban water distribution networks. Canadian Journal of Civil Engineering, 12:2: 286-293. Kleiner, Y. and Rajani, B.B., 2001. Comprehensive review of structural deterioration of water mains: statistical models. Urban Water, 3:3: 131-150. Kleiner, Y. and Rajani, B.B., 1999. Using limited data to assess future needs. Journal AWWA, 91:7:47-62. Kleiner, Y., Rajani, B. B. and Sadiq, R., 2004. Management of failure risk in large diameter buried pipes using fuzzy-based techniques. 4th International Conference on Decision Making in Urban and Civil Engineering, Porto, Portugal. 171 Levelton Consultants Ltd., 2005. Maple Ridge Water Works Corrosion Investigation. Report 2705-0380, Levelton Consultants Ltd. 102-19292 60th Avenue, Surrey, BC. Luttmerding, H.A., 1981. Soils of the Langley-Vancouver map area: Report No. 15, Volume 6. B.C. Ministry of Environment Assessment and Planning Division, Kelowna, BC. Luttmerding, H.A., 1980. Soils of the Langley-Vancouver map area Report No. 15 Volume 1. B.C. Ministry of Environment Assessment and Planning Division. Kelowna, BC. Mailhot, A., Pelletier, G., Noel, J.-F. and Villeneuve, J.-P., 2000. Modeling the evolution of the structural state of water pipe networks with brief recorded pipe break histories: Methodology and application. Water Resources Research, 36:10: 3053-3062. Male, J. W., Walski, T. and Slutsky, A. H., 1990. Analyzing Water Main Replacement Policies. Journal of Water Resources Planning and Management, 116:3: 362-374. Marks, D. H., Andreou, S., Jeffrey, L., Park, C. and Zaslavsky, A., 1987. Statistical Models for Water Main Failures. Water Engineering Research Lab, USEPA. Cincinnati, OH. Myers, C, 2001. Rural Areas gain high tech infrastructure plan through partnership, Utility Executive, 3:2: 5-8, Water Environment Federation, 601 Wythe Street, Alexandria, VA, 22314-1994. 172 NGSMI, (National Guide to Sustainable Municipal Infrastructure - Infraguide), 2003. Best Practices for Utility-Based Data. Infraguide - Potable Water. Ottawa, ON, Canada. NGSMI (National Guide to Sustainable Municipal Infrastructure - Infraguide), 2002. Deterioration and Inspection of Water Distribution Systems. Infraguide - Potable Water. Ottawa, ON, Canada. National Research Council of the National Academies - Committee on Public Water Supply Distribution Systems: Assessing and Reducing Risks, 2005 . Public Water Supply Distribution Systems: Assessing and Reducing Risks - First Report. O'Day, D.K., 1986. Watermain Evaluation for Rehabilitation / Replacement. AWWA Research Foundation (AWWARF) and US Environmental Protection Agency (USEPA), 6666 West Quincy Avenue, Denver, CO. 80235. O'Day, D. K., 1982. Organizing and analyzing leak and break data for making main replacement decisions. Journal A WWA, 74:11: 588-594. O'Day, D. K., Weiss, R., Chiavari, S., and Blair, D., 1986. Watermain Evaluation for Rehabilitation / Replacement. AWWA Research Foundation (AWWARF) and US Environmental Protection Agency (USEPA), AWWA Research Foundation, 6666 West Quincy Avenue, Denver, CO. 80235. 173 Ontario PIR. 2005. WATERTIGHT: The case for change in Ontario's water and wastewater sector. Report of the Water Strategy Expert Panel. Ontario Ministry of Public Infrastructure Renewal. Pelletier, G., Mailhot, A. and Villeneuve, J.-P., 2003. Modeling water pipe breaks - three case studies. Journal of Water Resources Planning and Management, 129:2: 115-123. Rajani, B.B. and Kleiner, Y., 2001. Comprehensive review of structural deterioration of water mains: physically based models. Urban Water, 3:3: 151-164. Rajani, B.B. and Makar, J., 2000. A methodology to estimate remaining service life of grey cast iron water mains. Canadian Journal of Civil Engineering, 27:6: 1259-1272. Rajani, B.B. and Tesfamariam, S., 2005. Estimating time to failure of ageing cast iron water mains under uncertainties. Water Management for the 21st Century, University of Exeter, UK., University of Exeter, UK., 1-7. Rossiter, D.G., 2005. A compendium of on-line soil survey information - Soil survey institutes and activities. Accessed May 31, 2005 http://www.itc.nl/ rossiter/research/ rsrch_ss_sOurces.html. Sacluti, F., 1999. Modeling Water Distribution Pipe Failures Using Artificial Neural Networks, Master of Science Thesis, University of Alberta, Edmonton, Alberta. 174 Sargeant, D., 2003. Water Main Rehabilitation Prioritization. AWWA 2003 Seminar -Infrastructure: above and below ground, Anaheim, CA. 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. Schick, S., 2005. Toronto bridges data needs in water network project. ITBusiness.ca July 21,2005. Accessed July 22, 2005 http://www.itbusiness.ca/index.asp7theaction =61&sdi= 59540. Schuyler, J., 2001. Risk and Decision Analysis in Projects. Project Management Institute, Four Campus Blvd. Newtown Square, Pennsylvania. Shamir, U. and Howard, C, 1979. An analytic approach to scheduling pipe replacement. Journal AWWA, 71:5: 248-258. Thain, D., 2005. Water mains in the District of Maple Ridge. A. Wood, ed., Maple Ridge, Interview notes. Torra, V., Domingo-Ferrer, J. and Torres, A., 2004. Data mining methods for linking data coming from several sources. 3rd Joint Un/ECE-Eurostat Work Session on Statistical Data Confidentiality, Monographs in Official Statistics, Luxembourg, Eurostat. 143-150. 175 USEPA, 2005. 2003 Drinking Water Infrastructure Needs Survey and Assessment, United States Environmental Protection Agency, Drinking Water Division, Washington, DC. USEPA, 2002. Clean Water and Drinking Water Infrastructure Gap Analysis, Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. USEPA, 2001. Drinking Water Infrastructure Needs Survey. Second Report to Congress. United States. Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. USEPA, 1999. Drinking Water Infrastructure Needs Survey. United States Environmental Protection Agency, Office of Water, Drinking Water Division, Washington, DC. USFHWA, 1999. Asset Management Primer, Federal Highways Administration, US Department of Transportation, Office of Asset Management, 400 7th Street, S.W. Washington D.C. 20590. Vanier, D.J., 2001. Asset Management: "A" to "Z", American Public Works Association Annual Congress and Exposition - Innovations in Urban Infrastructure Seminar, Philadelphia, U.S. September, 2001. 1-16. 176 Vanier, DJ. and Rahman, S., 2004. Municipal Infrastructure Investment Planning (MID?) MUP Report: A Primer on Municipal Infrastructure Asset Management. Report B-5123.3, National Research Council Canada, Ottawa, ON. Walski, T.M., 1982. Economic Analysis of Water Main Breaks. Journal of Water Resources Planning Management Division, 108:3: 296-308. Wild, C. J. and Seber, G.A.F., 2000. Chance Encounters - A First Course in Data Analysis and Inference. John Wiley and Sons Inc., New York, N.Y. Wood, A. and Lence, B.J., 2006. Assessment of Water Main Break Data for Asset Management. Journal AWWA, 98:07. Wood, A., Lence, B.J. and Liu, W., 2007. Constructing Water Main Break Data for Asset Management, Journal AWWA, 99:01. Xu, C. and Goulter, I.C., 1998. Probabilistic model for water distribution reliability. Journal of Water Resources Planning and Management, 124:4: 218-228. 177 PAGE LEFT BLANK 178 APPENDICES APPENDIX A 179 Appendix A summarizes the utilities surveyed and the data queried by Deb et al. (2002) and by the author as part of the survey reported on in Chapter 2. A comparison of the survey responses by service population is shown in Table A.l. A comparison of the questions asked by the two surveys is shown in Table A.2. Table A.l Comparison of North American utility responses by service populations Survey <100,000 >100,000 Survey by Deb et al (2002) 5 32 Survey by Wood and Lence (2006) 37 22 Table A.2 Comparison of questions posed by Deb et al. (2002) and Wood and Lence (2006) Survey question by Deb et al (2002) Question addressed by Wood and Lence Notes Utility size yes Water production no Total length of pipe yes Main failures Formal program for control of failure n/a Main inventory n/a Computerized main inventory of total n/a Failure records yes Wood and Lence survey focused on failure records Computerized failure records of total n/a General Information yes Date, address yes Temperature yes Wood and Lence also queried air, water, change in water temperature and soil temperature Time of detection and arrival yes includes repair date Impact on surroundings Services affected yes Wood and Lence survey includes type of services and length of outage 180 Blocks affected no Wood and Lence survey includes quantityof parcels and number of customers affected and property damage Hydrants affected no Proximity to buried objects no Water main information Material yes Location in street yes Wood and Lence survey includes surface use Diameter yes Depth yes Cover depth Installation date yes Cathodic protection yes Wood and Lence survey queried pipe protection and /or anode installation Joint type yes Type of repair yes Pressure Range yes Operating pressure Pump station status no Wood and Lence survey queried flow in area Failures Type of failure yes Wood and Lence survey queried on suite of types (11), Debet al.just queried if type of failure was recorded Probable cause yes Wood and Lence survey queried on suite of causes (13) Type of repair yes Wood and Lence survey queried on suite of repairs (7) and additional treatments Exterior/interior of pipe yes Wood and Lence survey queried on both exterior and interior, lining condition and other details Bell condition no Wood and Lence survey queried on Joint type Condition of valves Required for isolation yes Condition no 181 Reporting bedding conditions yes Wood and Lence survey queried on material and condition separately Reporting of seismic/geotechnical conditions Soil description yes Wood and Lence survey queried on native soil, soil pH, soil Moisture content while Deb at al. identified if soil description was recorded Geologic unit description no Groundwater depth no Seismic hazard unit no Collecting field samples Pipe samples yes Soil samples yes Use of automated systems Wood and Lence survey explored other sources of data Field portable computers see'note GPS see note GIS see note DBMS see note Formal renewal program Main replacement n/a Main rehabilitation n/a Costs records Direct labour yes Wood and Lence survey queried on crew hours Indirect labour see note Wood and Lence survey queried on all labour costs Materials yes Equipment, yes Surface repairs see note Wood and Lence survey queried on Property damage costs Damage yes-Wood and Lence survey queried on Property damage costs 182 APPENDIX B 183 s Appendix B provides the statistical analysis for guiding the interpretation of the survey results and their applicability to the general population of water utilities. Given a specified percentage of respondents that collect a particular type of data, one may wish to determine the percentage of all water utilities that are likely to collect the given data. Such questions are addressed using the confidence range of observations regarding the particular type of data for utilities in the general population, and these are based on the standard error of the sample population. Wild and Seber (2000) suggest that an accepted measure of the confidence in the behaviour of the general population (e.g., all water utilities) is equivalent to two standard errors of the sample proportion (e.g., the respondents). The standard error of the sample proportion is calculated using the equation: Se(p) =(p*(l-p)/n)05 (Bl) Where Se (p) = the standard error of a sample proportion, p = the proportion of the sample that collects a given data type or element, n = the sample size, in this case 59 utilities. For example, using Equation Bl, if 63 percent of the respondents (i.e., 37 responses of the 59 survey respondents), indicate that they record when water service was restored, we may expect that more than 50 percent (i.e., the lower confidence limit) but less than 75 percent (i.e., the upper confidence limit) of all water utilities in the general population would record the same. The lower confidence limit of 50 percent is calculated by the response proportion minus two times the standard error of the sample population, i.e., 0.50 = 0.63 - 0.13, where 0.13 = 2*Se(p) = 2 (0.63*0.37/59) °'5. 184 Because the standard error is calculated using the sample proportions, the standard error varies with the number of responses. For example, the higher the number of responses for ' recording a given type of data, the more certain we are of the general population recording that type of data. The range of standard error is between 5.5 percent and 6.5 percent for response proportions of between 33 percent (20 responses) and 76 percent (45 responses) A summary of values for the standard error for various levels of responses and the general population confidence limits corresponding to those responses is shown in Table B.l. Table B.l Confidence limits for the general population0 based on proportion of responses Upper limit of Lower limit of the confidence the confidence level for the level for the Number of Percentage of Standard error of general population (within 2 standard general population (within 2 standard responses respondents response - % errors) errors) 20 34% 6.2% 46% 22% 30 51% 1 6.5% 64% 38% 37 63% 6.3% 75% 50% 39 66% 6.2% 78% 54% 44 75% 5.7% 86% 63% 45 76% . 5.5% 87% 65% a) Based on 59 responses (i.e., the sample size). 185 PAGE LEFT BLANK 186 APPENDIX C 187 Appendix C describes the municipality of Maple Ridge, its water utility and the Laity View area. MAPLE RIDGE, B.C. General description. The District of Maple Ridge (BC) is a municipality located within the Greater Vancouver region. Of the 75,000 people that resided in the municipality in 2005, approximately 65,000 residents were served by the water utility and 12,000 were served by on-site private wells. The municipality is in transition from being predominantly rural to being a suburban community. It has a town centre and surrounding urban area. Outside of the urban boundary are lands that are zoned rural and agricultural. Distribution system. The distribution system of Maple Ridge has over 15,000 connections and is comprised of 5 pump stations, 7 reservoirs and approximately 350,000 metres of water mains. A summary of the utility's pipe inventory by material types for 2001 is shown in Table C.l A summary of the volume of water purchased in 2001 from the Greater Vancouver Water District (GVWD) and distributed daily is presented in Table C.2. A comparison of the age of Maple Ridge water mains with a number of municipalities across Canada that are part of the Earth Tech benchmarking initiative (Earth Tech 2004) is shown in Table C.3. The table shows the distribution of pipes by age cohorts of Maple Ridge compared with the entire set of communities in the study. The participants include Richmond Hill (ON), City of Delta (BC), City of Waterloo (ON), City of Calgary (AB), City of Ottawa (ON), Regional Municipality of Halton (ON), City of Saskatoon (SK), City of London (ON), City of Toronto (ON), City of Hamilton (ON), City of Thunder Bay (ON), City of St. Catherines (ON) and City of Victoria (BC). 188 Financial measures. Financially, the utility operates as a self-liquidating utility -i.e., annual revenues and expenditures must balance. A five year business plan is submitted to Maple Ridge Council each year and rates are established by Council bylaw. In 2003, residential water customers were charged $230 per household and metered customers were charged $0,395 per cubic metre. A financial summary of the 2001 major expenditure categories is shown in Table C.4. Laity View. The Laity View area of Maple Ridge represents approximately ten percent of the water distribution system of Maple Ridge and serves a population of approximately 6,000. The total length of water mains in this area is 36,300 metres. A regional water main forms the boundary on one side of the study area while larger pipes may be used to form the internal boundaries for six selected pipe zones. These zones are comprised of mainly 150 mm pipes. One zone has a few metres of 100 mm pipe serving a very short cul-de-sac. The zones are shown in Figure C.l. Because Maple Ridge has a minimal number of pressure zones (most of the urban part of the District is served by one water pressure zone), the study area has only one water pressure zone. Laity View is known to have multiple and routine breaks. Typical break repairs are approximately three metres in length. The number of connections and number of people affected by breaks are not recorded. A comparison of Laity View break rates against those of North American and European water utilities as identified in an AWWARF study (Deb et al, 2002) is shown in Table C.5 below. The break rates for Laity View are comparable with those of other North American utilities, and are less than those of European systems. 189 Table C.l - Maple Ridge water system inventory Asbestos cement (m) Cast iron (m) Copper (m) Ductile iron (m) Galvanized Iron (m) Poly Vinyl Chloride (m) Steel (m) Total length of mains (m) Total 67,422 13,183 59 252,501 180 2,968 11,856 348,169 Table C.2 Maple Ridge system water volumes Annual Average Daily Water Consumption (ML/d*) Peak Day (ML/d) Annual Total Volume Purchased (ML) 2001 29ML/d (6.4 MGD) 52 ML/day (11.5MGD) 10,400 (22,900 MG) * ML/d is Mega-litres per day, ML is Mega-litres and MGD is Million Imperial gallons per day and MG is Million Imperial gallons. Table C.3 Comparison of age cohorts of Maple Ridge pipes against Earth Tech National benchmarking utilities Maple Ridge Participating utilities 0-24 55 46 25-49 45 42 50-74 0 6 75-99 0 5 >100 0 1 Total 100 100 Table C.4 Summary of major annual expenditure categories for 2001 Total Utility Revenues Total Utility Expenditures Water Purchase Expenditures Debt Servicing 2001 S6.5M S6.5M S2.4M S2.6M 190 Table C.5 Comparison of Laity View pipe break rates against those of other North American and European systems Laity View average break rates North America break rates* European break rates * Break rate (breaks/m/year) 0.0001126 0.0001375 0.000313 * source: Deb et al. (2002) 191 Figure C.l Laity View area The following figure is of the approximate extent of Maple Ridge's water system and the boundaries of the Laity View system. There are some developing sections northeast of the figure. S.SO KM 192 ( ( APPENDIX D 193 Appendix D is the survey that was distributed to North American water utilities in 2004. The survey was comprised of seven spreadsheets. Survey of Water System ( mains and fittings) failure Please complete this page and the 6 subsequent pages of questions about the information that your organization maintains on water main and fitting failures. If you are not the best source of this information please forward the survey to the appropriate person(s) in your organization. Organization Department Position title Contact (name of person who completed this survey) E-mail address: Telephone number: Area code Extension Address What is the total population of your jurisdiction What is the population served by your water system? What is the total length of your water mains? General water system breakage data are collected by: Our data are managed using: 194 DATA CONFIDENCE INCIDENT GENERAL INFORMATION RECORDED LEVEL IN INBREAK DATA RECORD COLLECTED Date of reported break Time of reported break Reported by Address of reporter Ph. Number of reporter Date of repair start Time of repair start Date of repair finish Time when water service was resumed Total hours on site Quantity of parcels without service Was service to customers disrupted Estimated number of customers affected affected customers by residential type affected customers by commercial type affected customers by industrial type affected customers by institutional type Repaired by Indication of property damage y or n Property damage cost Employee researching pipe & completing form ID of water main feature being repaired ID assigned to break, leak etc Sketch of damaged facility Digital photo of damaged facility Length of unsupported pipe Job / work order number Dates of previous breaks at same location Equipment used Crew members Crew total hours Total labour cost Total materials cost Total equipment cost Please list any locational incident details that your organization records and also any comments that you would like to include: 195 INCIDENT LOCATION INFORMATION DATA RECORDED INBREAK RECORD CONFIDENCE LEVEL IN DATA COLLECTED Nearest property address Distance from nearest property line Cross street name Distance from nearest cross street Northing and Easting coordinates Isolation Valve operated ( Gate valve ID ) Isolation Valve operated ( Gate valve ID ) Please list any locational incident details that your organization records and also any comments you would like to include: 196 DATA INCIDENT PHYSICAL DATA RECORDED IN BREAK RECORD Pipe diameter Pipe material Length ofpipe segment containing the repair Year of installation Pipe wall thickness/ classification Type of pipe lining Pipe protection ( wrapped / anodes) Type of joint Type of water service Normal operating pressure Under boulevard or roadway Surface material Depth of cover Bedding material Condition of bedding Backfill material Category of native soil Pipe sample collected Condition of pipe exterior Condition of unlinedpipe interior Condition of cement lined pipe interior Traffic classification or type of road useage Pipe modulus or rupture Pipe fracture toughness Typical flow in area of break CONFIDENCE LEVEL IN DATA COLLECTED IS DATA AVAILABLE FROM OTHER SOURCES? YES or NO OTHER SOURCE OF INFORMA -TION COMMENTS Please list any locational incident details that your organization records and also any comments that you would like to include: 197 INCIDENT FAILURE DESCRIPTION DATA RECORDED IN BREAK RECORD CONFIDENCE LEVEL IN DATA COLLECTED COMMENTS Types of failure are recorded in this precentage of our incidents IBB The following failure modes are fields that are recorded: YES or NO NONE - HIGH Longitudinal break Blow out Split bell Corrosion pit hole Leaking joint Leaking hydrant Leaking valve Tap failure Curbstop failure Leaking service connection Failed blow-off Please list below any additional types of failure that your organization records and any comments you would like to include: SUSPECTED CAUSE OF FAILURE DATA RECORDED IN BREAK RECORD CONFIDENCE LEVEL IN DATA COLLECTED ARE DATA EASILY AVAILABLE FROM OTHER SOURCES? IF DATA ARE AVAILABLE, SOURCE OF INFORMATION IS COMMENTS Is your water corrosive to your watermains (yes or no) Causes of failure are recorded in Ms percentage of our records The following causes are fields that are recorded: YES or NO NONE - HIGH YES or NO Corrosion Traffic load Poor construction practices Ground frost Settlement Joint failure rock contact Construction disturbance High pressure Water temperature change Frozen pipe Errosion / unsupported pipe Unknown Please list any locational incident details that your organization records and also any comments you would like, to include: 198 INCIDENT REPAIR DETAILS DATA RECORDED INBREAK RECORD CONFIDENCE LEVEL IN DATA COLLECTED Repair activities are recorded in this precentage of our incidents The following activities are fields that are recorded: YES or NO NONE - HIGH Repair clamp Replace pipe section Replace valve Replace service connection Replace hydrant part(s) Replace entire hydrant Anode installed External protection installed Repair joint Surface restoration Dechlorination performed Please list any locaiional incident details that your organization records and also any comments you would like to include: 199 INCIDENT ENVIRONMENT CONDITIONS Environment information is recorded in this percentage of our incidents DATA RECORDED IN BREAK RECORD CONFIDENCE LEVEL IN DATA COLLECTED ARE DATA EASILY AVAILABLE FROM OTHER SOURCES? 111!!! IF DATA ARE AVAILABLE, SOURCE OF • INFORMATION lilS COMMENTS The following information are fields that are recorded YES or NO NONE - HIGH YES or NO PICK BOX lillllll Water temperature Air temperature No. of consecutive days below 32 for 0 c Depth of frost Water temperature change Soil temperature at pipe depth Soil sample taken SoilPH Soil moisture content Please list any locational incident details that your organization records and also any comments you would like to include: Does your organization possess a water model? Is your model used? For what reasons does your organization use your model? What is the name of your water system model? Do you feel comfortable that the amount of data you collect and manage, adequately serves your utility? Development planning Capital Works planning Operation of water system Maintenance of water system Specify other: • .  •' • How many minutes did you take to complete the survey? Comments on the survey : Would you like a copy of the survey results? Yes or No If so, by mail, fax or e-mail? ^ Please return the completed .xls file to Andrew W°°d by E-mail: , 200 aw00</@mapleridge.org or by fax : 604-467-7425 ' This survey is being conducted by the Corporation of the District of Maple Ridge -Engineering Department District of Maple Ridge 11995 Haney Place Maple Ridge, British Columbia Canada, V2X6A9 attention: Andrew Wood, Municipal Engineer Thank you very much for your participation. \ 201 PAGE LEFT BLANK 202 APPENDIX E 203 Appendix E lists the organizations to which the Wood and Lence 2004 survey was directly mailed, not including those sent by the Canadian Water and Waste Association (CWWA). A list of those communities contacted by the CWWA is not available. The total number of organizations to whom surveys were mailed was 411. Name of organization State/Province Country Abbotsford, City of British Columbia CAN Aberdeen, City of South Dakota USA Abilene, City of Texas USA Ada, City of Oklahoma USA Addison, Village of Illinois USA Aiken, City of South Carolina USA Alachua County Florida USA Albuquerque, City of New Mexico USA Alexander City, City of Alabama USA Allentown, City of Pennsylvania USA Anchorage, City of Alaska USA Antioch, City of California USA Arlington County Virginia USA Arroyo Grande, City of California USA Atlanta, City of Georgia . USA Auburn, City of Georgia USA Auburn, City of Alabama USA Augusta County Georgia USA Bakersfield, City of California USA Ballwin, City of Missouri USA Bangor, City of Maine USA Barrie, City of Ontario CAN' Barrington, Village of Illinois USA Bathurst New Brunswick CAN Bay City, City of Michigan USA Bedford, City of Virginia USA Bellingham, City of Washington USA Beloit, City of Wisconsin USA Beloit, City of Wisconsin USA Bexley, City of Ohio USA Birmingham, City of Alabama USA Bismarck, City of North Dakota USA Blair, City of Nebraska USA Bloomington, City of Indiana USA Bloomington, City of Minnesota USA Bolingbrook, Village of Illinois USA 204 Boone County Kentucky USA Bowling Green, City of Ohio USA Boyle County Kentucky USA Brampton Ontario CAN Brandon, City of Mississippi USA Bridgewater, Town of Nova Scotia . CAN Brookline, Town of Massachusetts USA Broward County Florida USA Brownwood, City of Texas USA Bryant, City of Arkansas USA Burlington, City of North Carolina USA Burlington, City of Iowa USA Butler, Township of Ohio USA Cabot, City of Arkansas USA Caledonia, Town of Wisconsin USA Cambridge Ohio USA Camden, City of South Carolina USA Campbell, City of California USA Camrose, City of Alberta CAN Capital, Regional District British Columbia CAN Carmel-by-the-Sea, City of California USA Carpinteria, City of California USA Cartersville, City of Georgia USA Cary, Town of North Carolina USA Cedar Rapids, City of Iowa USA Chamberlain, City of South Dakota USA Champaign, City of Illinois USA Chandler, City of Arizona USA Charlotte, City of North Carolina USA Chatham County Georgia USA Chatham-Kent, Municipality of Ontario CAN Chattahoochee, City of Florida USA Chattanooga, City of Tennessee USA Chersterfield, City of Missouri USA Chester Metropolitan District South Carolina USA Chetwynd, District of British Columbia CAN Chicopee, City of Massachusetts USA Chilliwack, City of British Columbia CAN City of University Texas USA Claremont, City of California USA Clarksville, City of Tennessee USA Clearwater, City of Florida USA Cleveland, City of Mississippi USA Cleveland, City of Ohio USA Clinton, City of Mississippi USA 205 Clovis, City of California USA Coconut Creek, City of Florida USA Collierville, Town of Tennessee USA Columbia Heights, City of Minnesota USA Columbia, City of Missouri USA Columbia, City of South Carolina USA Columbia, Govt of District of District of Columbia USA Columbus, City of Ohio USA Columbus, City of Nebraska USA Concord, City of California USA Cooper City, City of Florida USA Coral Gables, City of Florida USA Corner Brook, City of Newfoundland , CAN Coronado, City of California USA Council Bluffs, City of Iowa USA Crystal Lake, City of Illinois USA Cullman, City of Alabama USA Cumberland, Town of Maine USA Dalton, Georgia USA Danvers, Town of Massachusetts USA Dauphin Manitoba CAN Daytona Beach, City of Florida USA Dekalb County Georgia USA-Delta,.District Municpaliy of British Columbia CAN DesMoines, City of Iowa USA Dodge City, City of Kansas USA Dorion, Ville de Quebec CAN Dorval, Ville de Quebec CAN Dover, City of New Hampshire USA Downey, City of California USA Dublin, City of Ohio USA Dubuque, City of Iowa USA Durham, Regional Municipality of Ontario CAN Eagan, City of Minnesota USA Eau Claire, City of Wisconsin USA Edmundston, City of New Brunswick CAN Effingham, City of Illinois USA Elgin, City of Illinois USA Enterprise, City of Alabama USA EPCOR Water Services Inc. Alberta CAN-Escondido, City of California USA Esquimalt, Township of British Columbia CAN Eugene, City of Oregon USA Everett, City of Washington USA Fairhope, City of Alabama USA 206 Fairview Heights, City of Illinois USA Fenton, City of Michigan USA Flower Mound, Town of Texas USA Forsyth, Village of Illinois USA Fort Wayne, City Indiana USA Fox Point, Village of Wisconsin USA Franklin, City of Virginia USA Fredericton, City of New Brunswick CAN Fresno, City of California USA Ft. Myers, City of Florida, USA Gainesville, City of Florida USA Galesburg, City of Illinois USA Gander, Town of Newfoundland CAN Garland, City of Texas USA Gastonia, City of North Carolina USA Georgetown County South Carolina USA Germantown, City of Tennessee USA Glendale, City of Arizona USA Goffstown, City of New Hampshire USA Golden Valley, City of Minnesota USA Goodyear, City of Arizona USA Grafton, Town of Massachusetts USA Grand Falls-Windsor, Town of Newfoundland CAN Grand Prairie, City of Texas . USA Greater Chicago, District of . Illinois ' USA Greensboro North Carolina USA Greenville County South Carolina USA Guelph, City of Ontario CAN Halifax, Regional Municipality of Nova Scotia CAN Hallandale Beach Florida USA Halton, Regional Municipality of Ontario CAN Hamilton, City of Ohio USA Hampton, City of Virginia USA Harford, County of Maryland USA Harlan, City of Iowa USA Hartsville, City of South Carolina USA Hastings, City of Minnesota USA Hattiesburg, City of Mississippi USA Helix Water District California USA Henderson, City of Nevada USA Henrico County Virginia USA High Point, City of North Carolina USA Highland Village, City of Texas USA-Highline Water District Washington USA Hillsboro Oregon USA 207 Houston, City of Texas USA Huntsville, City of Alabama USA Huron, City of South Dakota USA Huron-Kinloss, Township of Ontario CAN Hutchinson, City of Minnesota USA Iles-de-la-Madeleine Quebec CAN Incline, Village of Nevada USA Independence, City of Missouri USA Iowa City, City of Iowa USA Irving, City of Texas USA. Jackson, City of Mississippi USA Jacksonville, City of Florida , USA Kansas City, City of Missouri USA -Kelowna, City of British Columbia CAN Kennewick, City of Washington USA Kerrville, City of Texas USA Killeen, City of Texas USA Kingsport, City of Tennessee USA Kingston, City of Ontario CAN Kirksville, City of Missouri USA . Knoxville, City of . Iowa USA Knoxville, City of Tennessee USA La Palma, City of California USA La Vista, City of Nebraska USA Lake Zurich, Village of Illinois USA Lakeland, City of Florida USA Lakewood, City of California USA Laporte, City of Texas USA Las Vegas, City of Nevada USA Lawrence, City of Kansas USA Leawood, City of Kansas USA Leduc, City of Alberta CAN Lenexa, City of Kansas USA Lewiston, City of Maine USA Lexington North Carolina USA Lexington-Fayette Urban County Kentucky USA Lincoln City, City of Oregon USA Lindsay, City of California USA Little Rock, City of Arkansas USA Lombard, Village of Illinois USA Lompoc, City of California USA London, City of Ontario CAN Longview, City of Washington USA Los Angeles, City of California USA Louisville Metro Government Kentucky USA 208 r Lowndes County Georgia USA Lubbock County Texas USA Macon, City of Georgia USA Madera, City of California USA Madison, City of Mississippi USA Madison, City of Wisconsin USA Maitland, City of Florida USA Mandeville, City of Louisiana USA Mankato, City of Minnesota USA Mansfield, Town of Connecticut USA Maple Ridge, District of British Columbia CAN Marana, Town of Arizona USA Markham Ontario CAN Marquette Michigan USA Maryville, City of Tennessee USA Mc Comb, City of Mississippi USA McAllen, City of Texas USA Medley, Town of Florida USA Medicine Hat, City of British Columbia CAN Memphis, City of Tennessee USA Middletown, City of Ohio USA Milwaukee, City of Wisconsin USA Miramichi, City of New Brunswick CAN Mission Viejo, City of California USA Mission, District of British Columbia CAN Missouri City, City of Texas USA Mobile, City of Alabama USA Moncton, City of New Brunswick CAN Monterey, City of California USA Montreal-Nord, Ville de Quebec CAN Morton, Village of Illinois USA Mount Vernon, City of Ohio USA Mountain View, City of California USA Mt. Zion, Village of Illinois USA Murray City, City of Utah USA Murrieta, City of California - USA Muskegon, City of Michigan USA Muskoka,District of Ontario CAN Nelson, City of British Columbia CAN New Albany Ohio USA New Lenox, Village of Illinois USA New Orleans, City of Louisiana USA Newington, Town of Connecticut USA Newport News, City of Virginia USA Newton, City of Kansas USA 209 Niagra, Regional Municipality of Ontario CAN Norfolk, City of Virginia USA North Battleford Saskatchewan CAN North Las Vegas, City of Nevada USA North Miami Beach, City of Florida USA North Miami, City of Florida USA North Redding, Town of Massachusetts USA. North Vancouver, District of British Columbia CAN Northbrook, City of Illinois USA Oak Park, Village of Illinois USA Oklahoma City, City of Oklahoma USA Oakland, County of Mississippi USA Olathe, City of Kansas USA Olympia, City of Washington USA Omaha, City of Nebraska USA Orlando, City of Florida USA Oromocto, Town of New Brunswick CAN Ottawa, City of Ontario CAN Overland Park, City of Kansas USA Owensboro, City of Kentucky USA Oxford, County of Ontario CAN Ozark, City of Alabama USA Palatine, Village of Illinois USA Palmdale, City of California USA Pasadena, City of California USA Pasadena, Town of Newfoundland CAN Pebble Beach California USA Penticton, City of British Columbia CAN Peterborough, Utilities Corp. Ontario CAN Phoenix, City of Arizona USA Pierre, City of South Dakota USA Pigeon Forge, City of Tennessee USA Pitt Meadows, District of British Columbia CAN Plainview, City of Texas USA Piano, City of Texas USA Plantation, City of Florida USA Plymouth, Town of Massachusetts USA Pointe Claire Quebec CAN . Pomona, City of California USA Port Angeles, City of Washington USA Port Coquitlam, City of British Columbia CAN Port Moody, City of British Columbia CAN Port St Lucie, City of Florida USA Portland, City of Oregon USA Portsmouth, City of New Hampshire USA 210 Presque Isle, City of Maine USA Prince Albert Saskatchewan CAN Punta Gorda, City of Florida USA Qualicum Beach British Columbia CAN ' Quispamsis, Town of New Brunswick CAN Ralston, City of Nebraska USA Rapid City, City of South Dakota USA Red Deer Alberta CAN Redmond, City of Washington USA Redwood City, City of California USA Regina, City of Saskatchewan CAN Reno, City of Nevada USA Revelstoke, City of British Columbia CAN Richardson, City of Texas USA Richmond, City of Virginia USA Rio Rancho, City of New Mexico USA Rockford, City of Illinois USA Rocky Hill, Connecticut USA Round Rock, City of Texas USA Royal Palm Beach Utilities, Village of Florida USA Sacramento County CA California USA Salmon Arm, District of British Columbia CAN San Antonio-DPW, City of Texas USA San Buenaventura, City pf California USA San Diego, City of California USA San Francisco, City & County of California USA San Jose, City of California USA San Luis Obispo, City of California USA Sandford, City of North Carolina USA Sanford, City of Florida USA Santa Ana, City of California USA Santa Clara, City of California USA Sarasota, City of Florida USA Saskatoon, City of Saskatchewan CAN Savoy, Village of Illinois USA Scottsdale, City of Arizona , USA Seattle, City of Washington USA Selkirk, City of Manitoba CAN Shaker Heights, City of Ohio USA Shawnee, City of Kansas USA Shelton, City of Connecticut USA Shorewood, Village of Wisconsin USA Sioux Falls, City of South Dakota USA Skowhegan, Town of Maine USA -Slidell, City of Louisiana USA 211 South Bend, City of Indiana USA South Daytona, City of Florida USA South Portland, City of Maine USA Southbridge, Town of Massachusetts USA Spartanburg County South Carolina USA Spokane, City of Washington USA Springfield, City of Illinois USA Springfield, City of Missouri USA Springfield, Town of Vermont USA St. Peters, City of Missouri USA St. Tammany Parish LA Louisiana USA Starkville, City of Mississippi USA Strathcona Alberta CAN Sturgis, City of South Dakota USA Sunset Hills, City of Missouri USA Surrey, City of British Columbia CAN Tacoma, City of Washington USA Tallahassee, City of Florida USA Thousand Oaks, City of California USA Thornton, City of Colorado USA Tifton, City of Georgia USA Topeka, City of Kansas USA Topsham, Town of Maine USA Toronto City of Ontario CAN Traverse City, City of Michigan USA Tucson, City of Arizona USA Tulsa, City of Oklahoma USA Tupelo, City of Mississippi USA Upper Arlington, City of Ohio USA University of Wisconson Wisconsin USA Urbana, City of Illinois USA Uxbridge, ToWn of Massachusetts USA Valdosta, City of Georgia USA Venice,' City of Florida USA Vernon Hills, Village of Illinois USA Virginia Beach, City of Virginia USA Visalia, City of California USA Wakefield, Town of Maine USA Washington County MD Maryland USA Watertown, City of South Dakota USA Wauwatosa, City of Wisconsin USA Wellington, Village of Florida USA Weslaco, City of Texas USA West Allis, City of Wisconsin USA West Vancouver, City of British Columbia CAN Westbrook, City of Maine USA Whitby Ontario CAN Wichita, City of Kansas USA Wickliffe, City of Ohio USA Williston, City of North Dakota USA Wilmette, Village of Illinois USA Windham1, Town of Maine USA -Winfield, City of Kansas USA Winnipeg, City of Manitoba CAN Winston Salem North Carolina USA Wood River, City of Illinois USA Woodridge, Village of Illinois USA Wyoming, City of Michigan USA Yarmouth, Town of Massachusetts USA York, Region of Ontario CAN 213 PAGE LEFT BLANK 214 APPENDIX F 215 Appendix F provides the number of respondents for the queries in the Wood and Lence 2004 survey and the data collected. The number of respondents are reported by country of respondent. Number of respondents that collect general information Canadian utilities U.S. utilities Address of reporter 21 22 Crew members 27 25 Crew total hours 26 24 Date of repair finish 28 26 Date of repair start 29 26 Date of reported break 30 27 Dates of previous breaks at same location 15 16 Digital photo of damaged facility 15 16 Employee researching pipe & completing form 15 17 Equipment used 26 24 Estimated number of affected customers by commercial type 5 5 Estimated number of affected customers by industrial type 5 5 Estimated number of affected customers by institutional type 5 4 Estimated number of affected customers by residential type 1 2 Estimated number of customers affected 6 6 ID assigned to break, leak etc 23 17 ID of watermain feature being repaired 22 19 Indication of property damage y or n 23 23 Job / work order number 24 20 Length of unsupported pipe 4 5 Ph. Number of reporter 19 23 Property damage cost 7 9 Quantity of parcels without service 10 9 Repaired by 28 26 Reported by 22 22 Sketch of damaged facility 19 9 Time of repair start 23 22 Time of reported break 23 23 Time when water service was resumed 18 20 Total equipment cost 23 19 Total hours on site 28 25 Total labour cost 22 20 Total materials cost 23 18 Was service to customers disrupted 18 17 216. Number of respondents that record physical data Canadian Physical data recorded by utilities utilities U.S. utilities Backfill 18 11 Bedding material 17 10 Native soil 9 7 Bedding condition 12 10 Cement lining condition 7 10 Pipe exterior condition 20 15 Pipe interior condition (unlined) 11 11 Cover depth 24 16 Pipe segment length 18 15 Operating pressure 6 9 Pipe diameter 29 29 Fracture toughness . 1 3 Pipe material 30 28 Pipe modulus 7 7 Pipe protection 13 9 Pipe sampled 14 8 Pipe wall thickness 2 7 Surface material 20 15 Traffic classification 3 5 Joint type 21 13 Pipe lining type 6 7 Water service type 21 23 Flow in area 1 4 Surface use class 21 18 Year of installation 14 11 Physical data recorded and/or available elsewhere for Canadian utilities utilities U.S. utilities Backfill 20 13 Bedding material 20 15 Native soil - 11 13 Bedding condition 12 11 Cement lining condition 9 10 Pipe exterior condition 21 15 Pipe interior condition (unlined) 12 11 Cover depth 28 21 Pipe segment length 24 23 Operating pressure 19 23 Pipe diameter 30 29 Fracture toughness 2 6 Pipe material 30 29 Pipe modulus 8 9 Pipe protection 20 15 Pipe sampled 15 9 Pipe wall thickness 10 16 Surface material 24 22 217 Canadian Physical data recorded and available elsewhere for utilities utilities U.S. utilities Traffic classification 14 17 Joint type 24 18 Pipe lining type 3 16 Water service type 4 27 Flow in area 14 13 Surface use class 26 25 Year of installation 4 23 Number of respondents that record failure causes Failure cause data recorded by utilities Canadian utilities U.S. utilities Construction disturbance 13 16 Corrosion 20 16 Erosion / unsupported pipe 13 13 Frozen pipe 14 13 Ground frost 13 7 High pressure 8 12 Joint failure 22 c' . .18 Poor construction practices 10 13 rock contact 14 15 Settlement •14 13 Traffic load 3 6 Unknown 11 12 Water temperature change 2 2 Canadian Causes are recorded in this percentage of records utilities U.S. utilities 0 percent 2 2 25 percent 6 4 50 percent 1 1 75 percent 4 3 100 percent 7 8 n/a 0 11 218 Number of respondents that record repair activities Repair activity data recorded by utilities Canadian utilities U.S. utilities Anode installed 24 10 Dechlorination performed 14 12 Extended protection installed 15 14 Repair clamp 29 27 Repair joint 27 25 Replace entire hydrant 24 27 Replace hydrant parts 23 25 Replace pipe section 29 26 Replace service connection 23 25 Replace valve 26 27 Surface restoration 22 26 Number of respondents that collect different types of environmental data Environmental data recorded Air temperature Depth of Frost < 32oC days Soil Moisture Content Soil pH Sample taken Soil temperature Water temperature Change in water temperature Canadian utilities U.S. utilities Environmental data recorded and/or available elsewhere for Canadian utilities utilities U.S. utilities Air temperature 16 15 Depth of Frost 2 12 < 32 degree days 3 14 Soil Moisture Content 3 SoilpH 7 ' 1Sample taken 3 5 Soil temperature 0 2 Water temperature 7 8 Change in water temperature .7 6 219 Number of respondents that expressed confidence in data collected Confidence in data collected Pipe diameter Pipe material Water service type Surface use classification Cover depth Year of installation High Good Fair Low 44 41 30 19 17 12 14 16 10 16 18 Number of respondents that record types offailures Types of Failure recorded by utilities Blow out Corrosion pit hole Curbstop failure Failed blow-off Leaking hydrant Leaking joint Leaking service connection Leaking valve Longitudinal break Split bell Tap failure Canadian utilities U.S. utilities 26 23 25 22 24 23 21 20 23 27 26 24 24 26 24 26 27 22 26 21 23 24 Number of respondents that record location data Location data recorded by utilities Cross street name Distance from cross street Distance from nearest property line Isolation valve operated Nearest property address Coordinates (northing and easting) Canadian utilities 24 16 15 12 28 2 U.S. utilities 22 . 10 5 9 27 4 220 Number of respondents that express comfort with recording location data and use water models Comfort with level of data collected by utilities Yes No N/A Canadian utilities U.S. utilities 21 17 5 8 4 4 Number of respondents that have water models and what they use the models for Canadian utilities U.S. utilities Number of utilities that have a water model 24 22 Number of utilities that use their water model - 21 21 Number of utilities that uses a water model for: Capital planning 20 21 Development planning 19 20 Operations 16 "17 Maintenance planning 1 14 221 Population and Country data of survey responses City - ID Country IX^cnpiion Total 21 Can .What is the total population of your jurisdiction 115000 42 Can What is the. total population of your jurisdiction 7700 55 Can What is the total population of your jurisdiction 15669 56 Can What is the total population of your jurisdiction 380000 72 Can What is the total population of your jurisdiction 2600 428 Can What is the total population of your jurisdiction 70000 413 Can What is the total population of your jurisdiction 20000 100 Can What is the total population of your jurisdiction • 8085 103 Can What is the total population of your jurisdiction 102000 112 Can What is the total population of your jurisdiction 600000 120 Can What is the total population of your jurisdiction 884700 _ 134 Can What is the total population of your jurisdiction 48000 418 Can What is the total population of your jurisdiction 73000 227 Can What is the total population of your jurisdiction 222000 231 Can What is the total population of your jurisdiction 52000 240 Can What is the total population of your jurisdiction 60000 249 Can What is the total population of your jurisdiction 14000 260 (" an What is the total population of your jurisdiction 14500 401 Can What is the total population of your jurisdiction 86000 283 Can What is the total population of your jurisdiction 32000 284 Can What is the total population of your jurisdiction 72000 295 Can What is the total population of your jurisdiction 5 <000 301 Can What is the total population of your jurisdiction 40000 424 Can What is the total population of your jurisdiction 73000 433 Can What is the total population of your jurisdiction 190093 332 Can What is the total population of your jurisdiction 212000 435 Can What is the total population of your jurisdiction 65000 400 Can What is the total population of your jurisdiction 2385421 383 Can What is the total population of your jurisdiction 42000 392 • Can What is the total population of your jurisdiction 631200 403 USA What is the total population of your jurisdiction 106000 409 USA What is the total population of your jurisdiction 260000 10 USA What is the total population of your jurisdiction 98000 48 USA What is the total population of your jurisdiction 45000 404 USA What is the total population of your jurisdiction 200000 411 USA What is the total population of your jurisdiction 22950 76 USA What is the total population of your jurisdiction 13842 414 USA What is the total population of your jurisdiction 25500 141 USA What is the total population of your jurisdiction 220657 405 USA What is the total population of your jurisdiction 100001 402 USA What is the total population of your jurisdiction 14500 430 USA What is the total population of your jurisdiction (.5000 431 USA What is the total population of your jurisdiction -'400 415 USA What is the total population of your jurisdiction 48000 416 USA What is the total population of your jurisdiction )00 407 USA What is the total population of your jurisdiction 35000 417 USA What is the total population of your jurisdiction 32000 432 USA What is the total population of your jurisdiction 20600 419 USA What is the total population of your jurisdiction 20000 426 USA What is the total population of your jurisdiction 35000 420 USA What is the total population of your jurisdiction 100001 222 427 USA What is the total population of your jurisdiction 105000 422 USA What is the total population of your jurisdiction 343700 434 USA What is the total population of your jurisdiction 1300000 342 USA What is the total population of your jurisdiction 140000 41* I S\ What is the total population of your jurisdiction nrooo 423 USA What is the total population of your jurisdiction 123000 379 USA What is the total population of your jurisdiction 47000 389 USA What is the total population of your jurisdiction 27000 223 General data of survey responses fii\ ID ,, Description.-. '.' ii Recorded 420 •Date of reported break 100 427 Date of reported break 100 76 Date of reported break 100 430 Date of reported break 100 10 Date of reported break 100 301 Date of reported break 100 100 Date of reported break 100 416 Date of reported break 100 134 Date of reported break 100 414 Date of reported break 100 409 Date of reported break • 100 415 Date of reported break 100 55 Date of reported break 100 284 Date of reported break 100 249 Date of reported break 100 141 Date of reported break 100 . 260 432 Date of reported break 100 Date of reported break 100 42 Date of reported break 100 295 Date of reported break 100 428 Date of reported break 100 411 Date of reported break 100 405 Date of reported break 433 Date of reported break 100 342 Date of reported break • 100 332 Date of reported break 100 21 Date of reported break (- 100 48 Date of reported break 103 Date of reported break 100 434 Date of reported break 100 392 Date of reported break 100 407 Date of reported break 100 408 Date of reported break 100 404 Date of reported break 100 413 Date of reported break 75 402 Date of reported break 100 240 Date of reported break 100 ' 401 Date of reported break 100 72 Date of reported break 100 435 Date of reported break 100 422 Date of reported break 100 423 Date of reported break 100 419 Date of reported break 100 431 Date of reported break 100 389 Date of reported break - 100 224 227 Date of reported break 100 112 Date of reported break 100 403 Date of reported break 100 418 Date of reported break 100 231 Date of reported break 100 383 Date of reported break 100 283 Date of reported break 100 424 Date of reported break 100 426 Date of reported break 100 400 Date of reported break 100 379 Date of reported break 100 100 120 Date of reported break 417 i Date of reported break 100 56 : Date of reported break 100 283 1 Time of reported break 100 332 | Time of reported break 100 389 i Time of reported break 100 379 i Time of reported break 75 418 Time of reported break 100 . 433 Time of reported break 75 428 Time of reported break 100 112 Time of reported break 0 295 : Time of reported break 100 404 i Time of reported break 100 284 i Time of reported break 100 76 i Time of reported break 50 423 Time of reported break 100 424 Time of reported break 100 260 Time of reported break 100 342 Time of reported break 0 411 | Time of reported break 0 420 Time of reported break 100 21 Time of reported break 100 240 Time of reported break 75 120 j Time of reported break 100 407 | Time of reported break 100 383 I Time of reported break 100 392 Time of reported break 0 432 Time of reported break 0 416 Time of reported break 100 417 Time of reported break 100 430 Time of reported break 100 10 Time of reported break 75 100 Time of reported break 100 427 ! Time of reported break 100 415 i Time of reported break 50 249 i Time of reported break 0 141 Time of reported break 100 431 Time of reported break 100 225 J 134 Time of reported break 100 56 Time of reported break 75 409 Time of reported break 75 426 Time of reported break 100 422 Time of reported break 0 401 Time of reported break 100 435 Time of reported break 50 419 Time of reported break 100 402 Time of reported break 100 227 Time of reported break 0 403 Time of reported break 100 72 Time of reported break 100 400 Time of reported break 100 48 Time of reported break 434 Time of reported break 100 405 Time of reported break 55 Time of reported break J 0 231 Time of reported break 0 301 Time of reported break 0 103 Time of reported break 100 413 Time of reported break 75 408 Time of reported break 100 414 Time of reported break 100 42 Time of reported break 25 419 Reported by 100 426 Reported by 100 48 Reported by • 424 Reported by 423 Reported by 100 422 Reported by 0 417 Reported by 383 Reported by 379 Reported by • 75 418 Reported by 100 402 Reported by 100 55 Reported by 0 ' 227 Reported by 0 433 Reported by 75 112 Reported by 100 403 Reported by 100 240 Reported by 100 103 Reported by 100 134 Reported by 100 342 Reported by 0 21 Reported by 75 76 Reported by 100 301 Reported by 100 408 Reported by 100 407 Reported by 100 226 427 Reported by 100 420 Reported by 332 Reported by 50 411 Reported by 50 413 Reported by 405 Reported by 249 Reported by 25 432 Reported by 25 141 Reported by 100 404 Reported by 75 283 Reported by 100 414 Reported by 100 56 Reported by 0 415 Reported by 75 409 Reported by 416 Reported by 100 10 Reported by 75 120 Reported by 100 ' 435 Reported by 25 260 Reported by 75 434 Reported by . 100 100 Reported by 75 284 Reported by 100 100 430 Reported by 392 Reported by 100 401 Reported by 0 389 Reported by 100 72 Reported by 75 231 Reported by 100 400 Reported by 431 Reported by 100 42 Reported by 75 295 Reported by r 100 428 Reported by 100 420 Address of reporter 100 120 Address of reporter 0 76 Address of reporter 0 379 Address of reporter 75 10 Address of reporter 75 427 Address of reporter 100 260 Address of reporter 25 431 Address of reporter 100 249 Address of reporter 0 404 Address of reporter 75 141 Address of reporter 100 56 Address of reporter 0 284 Address of reporter 100 283 Address of reporter 100 409 Address of reporter 75 227 432 Address of reporter 0 433 Address of reporter 75 295 Address of reporter 100 301 Address of reporter 75 411 Address of reporter 100 428 Address of reporter 100 55 Address of reporter 0 405 Address of reporter 48 Address of reporter 413 Address of reporter 0 434 Address of reporter 25 392 Address of reporter 50 430 Address of reporter 100 100 Address of reporter 75 103 Address of reporter 407 Address of reporter 100 42 Address of reporter 1 100 21 Address of reporter 50 415 Address of reporter 100 408 Address of reporter 100 134 Address of reporter 75 414 Address of reporter 50 342 Address of reporter 0 332 Address of reporter 50 418 Address of reporter 100 422 Address of reporter 0 400 Address of reporter 50 419 Address of reporter 100 401 Address of reporter 0 383 Address of reporter ' 50 435 Address of reporter 25 402 Address of reporter 25 423 Address of reporter 100 416 Address of reporter 50 240 Address of reporter 100 426 Address of reporter 100 227 Address of reporter 0 . 417 Address of reporter 75 231 Address of reporter 0 112 Address of reporter 100 403 Address of reporter 75 389 Address of reporter 0 72 Address of reporter 75 424 Address of reporter 25 434 Ph. Number of reporter 100 112 Ph. Number of reporter 100 42 Ph. Number of reporter 100 408 Ph. Number of reporter IOO1 383 Ph. Number of reporter 50 228 435 Ph. Number of reporter 50 432 Ph. Number of reporter 25 21 Ph. Number of reporter _75_ 134 Ph. Number of reporter 25 249 Ph. Number of reporter 0 240 Ph. Number of reporter 100 403 Ph. Number of reporter 75 411 Ph. Number of reporter 25 389 Ph. Number of reporter 405 Ph. Number of reporter 72 Ph. Number of reporter 100 48 Ph. Number of reporter 227 Ph. Number of reporter 0 332 Ph. Number of reporter 75 428 Ph. Number of reporter 100 407 Ph. Number of reporter 100 392 Ph. Number of reporter 50 418 Ph. Number of reporter' 342 Ph. Number of reporter 0 100 Ph. Number of reporter 75 231 Ph. Number of reporter 0 402 Ph. Number of reporter 25 55 Ph. Number of reporter 0 103 Ph. Number of reporter 413 Ph. Number of reporter 0 431 Ph. Number of reporter 100 420 Ph. Number of reporter 100 423 Ph. Number of reporter 100 416 Ph. Number of reporter 75 414 Ph. Number of reporter 50 295 Ph. Number of reporter 100 415 Ph. Number of reporter 100 426 Ph. Number of reporter 100 284 Ph. Number of reporter 100 404 Ph. Number of reporter 75 433 Ph. Number of reporter • " 260 Ph. Number of reporter , 0 141 Ph. Number of reporter 100 283 Ph. Number of reporter 100 56 Ph. Number of reporter 0 424 Ph. Number of reporter 25 379 Ph. Number of reporter 75 400 Ph. Number of reporter 25 120 Ph. Number of reporter 0 417 Ph. Number of reporter 75 419 Ph. Number of reporter 100 427 Ph. Number of reporter 25 76 Ph. Number of reporter 0 10 Ph. Number of reporter 50 229 401 Ph. Number of reporter 0 301 Ph. Number of reporter 75 422 Ph. Number of reporter 0 409 Ph. Number of reporter 50 430 Ph. Number of reporter 100 426 Date of repair start 100 295 Date of repair start 100 419 Date of repair start 100 56 Date of repair start 100 414 Date of repair start 100 141 Date of repair start 100 134 Date of repair start 100 434 Date of repair start 100 403 Date of repair start 100 400 Date of repair start 50 405 Date of repair start 422 Date of repair start 0 100 Date of repair start 100 10 Date of repair start 100 435 Date of repair start 100 249 Date of repair start 100 401 Date of repair start 100 402 Date of repair start 100 408 Date of repair start 100 48 Date of repair start 430 Date of repair start 100 431 Date of repair start 100 413 Date of repair start 100 415 Date of repair start 100 427 Date of repair start 100 4T Date of repair start 100 103 Date of repair start 100 227 Date of repair start 100 72 4i"> Date of repair start 100 Date of repair start 100 332 Date of repair start 100 120 Date of repair start 100 260 Date of repair start 100 21 Date of repair start 100 432 Date of repair start 100 428 . Date of repair start 100 112 Date of repair start 0 392 Date of repair start 100 284 Date of repair start 100 283 Date of repair start 100 420 Date of repair start 100 42 Date of repair start 100 433 Date of repair start 100 407 Date of repair start 100 230 404 Date of repair start 100 379 76 Date of repair start Date of repair start 100 100 411 | Date of repair start 100 424 i Date of repair start 100 231 ; Date of repair start 100 301 | Date of repair start 100 416 | Date of repair start 100 55 : Date of repair start 100 342 1 Date of repair start 100 240 i Date of repair start 100 100 423 j Date of repair start 383 i Date of repair start 100 389 1 Date of repair start 100 418 i Date of repair start 100 418 ! Time of repair start 75 76 Time of repair start 100 402 Time of repair start 100 420 Time of repair start 100 427 i Time of repair start 100 409 i Time of repair start 100 10 Time of repair start 50 72 Time of repair start 0 249 Time of repair start 50 400 i Time of repair start 50 416 Time of repair start 100 422 Time of repair start 0 419 Time of repair start 50 231 Time of repair start 0 240 Time of repair start 100 227 i Time of repair start 100 423 Time of repair start 100 401 Time of repair start 100 383 Time of repair start 75. 417 Time of repair start 75 120 | Time of repair start 100 260 I Time of repair start 100 415 i Time of repair start 0 426 i Time of repair start 100 435 i Time of repair start 100 424 Time of repair start 100 379 Time of repair start 50 56 Time of repair start 0 301 Time of repair start 0 141 : Time of repair start 100 284 i Time of repair start 100 404 Time of repair start 100 43.1 Time of repair start 0 283 Time of repair start 100 231 48 Time of repair start 405 Time of repair start 55 ' Time of repair start 100 42 Time of repair start 50 407 Time of repair start 100 332 Time of repair start 100 432 Time of repair start 0 21 Time of repair start 100 408 Time of repair start 100 134 Time of repair start 100 392 Time of repair start 0 112 Time of repair start 0 430 Time of repair start 75 100 Time of repair start 0 413 Time of repair start 100 403 Time of repair start 100 414 Time of repair start 100 433 Time of repair start 100 295 Time of repair start ~ 100 342 Time of repair start 100 103 Time of repair start 100 428 Time of repair start 100 434 Time of repair start 100 389 Time of repair start 0 411 Time of repair start 100 400 Date of repair finish 50 240 Date of repair finish 100 392 Date of repair finish 0 411 Date of repair finish 100 426 Date of repair finish 100 417 Date of repair finish 75 433 Date of repair finish 100 100 Date of repair finish 100 424 Date of repair finish 100 416 Date of repair finish 100 430 Date of repair finish 100 55 Date of repair finish 100 383 Date of repair finish 75 422- Date of repair finish 0 260 Date of repair finish 100 403 Date of repair finish 100 295 Date of repair finish 100 231 Date of repair finish 100 283 Date of repair finish 100 112 Date of repair finish 0 227 Date of repair finish 100 423 Date of repair finish 100 405 Date of repair finish 419 Date of repair finish 50 232 48 Date of repair finish 434 Date of repair finish 100 408 Date of repair finish 100 42 Date of repair finish 100 249 Date of repair finish 100 72 Date of repair finish ' 100 10 Date of repair finish 100 402 Date of repair finish 100 141 Date of repair finish 100 432 i Date of repair finish 100 76 Date of repair finish 100 428 Date of repair finish 100 427 Date of repair finish 100 301 Date of repair finish 100 420 Date of repair finish 100 414 Date of repair finish 100 284 Date of repair finish 100 389 Date of repair finish 100 103 Date of repair finish 100 134 Date of repair finish . 100 413 Date of repair finish 100 56 Date of repair finish 100 100 409 Date of repair finish 332 Date of repair finish 100 342 Date of repair finish 100 120 Date of repair finish 100 401 Date of repair finish 100 404 Date of repair finish 75 407 Date of repair finish 100 21. Date of repair finish 100 431 Date of repair finish 100 418 Date of repair finish 75 415 Date of repair finish 100 379 Date of repair finish 100 43 S Date of repair finish 100 240 Time when water service was resumed 100 435 Time when water service was resumed 100 418 Time when water service was resumed 25 403 Time when water service was resumed 100 295 Time when water service was resumed 100 430 Time when water service was resumed 25 48 Time when water service was resumed 55 Time when water service was resumed 100 21 Time when water service was resumed 100 134 Time when water service was resumed 100 103 Time when water service was resumed 432 Time when water service was resumed 0 428 Time when water service was resumed 0 284 Time when water service was resumed 100 233 401 Time when water service was resumed 100 427 Time when water service was resumed 0 416 Time when water service was resumed 100 249 Time when water service was resumed 0 10 Time when water service was resumed 25 433 Time when water service was resumed 50 231 Time when water service was resumed 0 76 Time when water service was resumed 100 141 Time when water service was resumed 100 424 i Time when water service was resumed 100 426 Time when water service was resumed 0 227 Time when water service was resumed 0 419 Time when water service was resumed 50 283 Time when water service was resumed 100 420 Time when water service was resumed 100 423 Time when water service was resumed 100 407 Time when water service was resumed 50 260 Time when water service was resumed 100 415 Time when water service was resumed 0 75 417 Time when water service was resumed 404 Time when water service was resumed 75 332 Time when water service was resumed 75 100 Time when water service was resumed 0 434 Time when water service was resumed 100 409 Time when water service was resumed 100 392 Time when water service was resumed 0 301 Time when water service was resumed 0 42 Time when water service was resumed 25 414 Time when water service was resumed 100 120 Time when water service was resumed 100 400 i Time when water service was resumed 0 389 Time when water service was resumed 100 72 Time when water service was resumed 0 112 408 Time when water service was resumed 0 Time when water service was resumed 100 411 : Time when water service was resumed 25 56 Time when water service was resumed 0 383 Time when water service was resumed 75 431 Time when water service was resumed 0 379 i Time when water service was resumed 25 405 Time when water service was resumed 342 Time when water service was resumed 100 _4m 422 413 379 Time when water service was resumed 0 Time when water service was resumed 0 Time when water service was resumed 100 Total hours on site . 100 76 Total hours on site 75 415 Total hours on site 100 249 Total hours on site 100 234 424 Total hours on site 100 383 Total hours on site 100 423 Total hours on site 100 56 Total hours on site 100 141 404 Total hours on site 100 Total hours on site 100 55 I Total hours on site 100 134 1 Total hours on site 100 413 i Total hours on site 0 408 i Total hours on site 100 48 Total hours on site 103 Total hours on site 100 405 Total hours on site 433 Total hours on site 100 411 Total hours on site 100 432 Total hours on site 100 407 Total hours on site 100 402 Total hours on site 100 332 Total hours on site 100 21 Total hours on site 100 409 Total hours on site 100 427 Total hours on site 100 283 Total hours on site 100 301 Total hours on site 100 420 Total hours on site 100 342 Total hours on site 100 403 Total hours on site 100 414 Total hours on site 100 418 Total hours on site 100 417 Total hours on site 100 284 i Total hours on site 100 231 Total hours on site 25 10 Total hours on site 100 227 Total hours on site 100 428 Total hours on site 100 260 i Total hours on site 50 392 Total hours on site 0 120 Total hours on site 100 434 Total hours on site 100 '422 : Total hours on site 0 295 Total hours on site 100 . 100 Total hours on site 100 430 Total hours on site 25 419 Total hours on site 100 42 Total hours on site 100 400 Total hours on site 25 401 Total hours on site 100 426 Total hours on site 100 112 Total hours on site 100 235 240 Total hours on site 100 72 Total hours on site 100 389 Total hours on site 0 435. Total hours on site 100 416 Total hours on site 100 431 Total hours on site 100 407 Quantity of parcels without service 0 134 Quantity of parcels without service 100 435 Quantity of parcels without service , 50 72 j Quantity of parcels without service 0 434 Quantity of parcels without service 100 55 Quantity of parcels without service 100 405 Quantity of parcels without service 295 Quantity of parcels without service 25 431 Quantity of parcels without service 0 411 Quantity of parcels without service 0 10 Quantity of parcels without service 0 433 Quantity of parcels without service 0 100 Quantity of parcels without service 0 401 Quantity of parcels without service 0 48 Quantity of parcels without service 400 Quantity of parcels without service 0 231 Quantity of parcels without service 0 418 Quantity of parcels without service 0 423 Quantity of parcels without service 0 389 Quantity of parcels without service 0 413 Quantity of parcels without service 100 112 Quantity of parcels without service 0 240 416 Quantity of parcels without service 0 Quantity of parcels without service 0 383 j Quantity of parcels without service 0 424 Quantity of parcels without service 100 404 Quantity of parcels without service ; 75 0 260 Quantity of parcels without service 284 Quantity of parcels without service 100 120 1 Quantity of parcels without service 100 332 | Quantity of parcels without service 100 428 | Quantity of parcels without service 0 379 | Quantity of parcels without service 0 420 ! Quantity of parcels without service 100 342 Quantity of parcels without service 0 42 Quantity of parcels without service 0 432 Quantity of parcels without service 0 392 Quantity of parcels without service 0 283 Quantity of parcels without service 0 301 Quantity of parcels without service 0 430 Quantity of parcels without service 0 21 Quantity of parcels without service 0 76 Quantity of parcels without service 0 236 103 Quantity of parcels without service 227 Quantity of parcels without service 50 414 419 Quantity of parcels without service 100 Quantity of parcels without service 100 56 Quantity of parcels without service 0 422 141 Quantity of parcels without service 0 Quantity of parcels without service 100 417 Quantity of parcels without service 0 402 Quantity of parcels without service 0 415 Quantity of parcels without service 0 409 Quantity of parcels without service 75 403 Quantity of parcels without service 50 408 Quantity of parcels without service 100 0 427 Quantity of parcels without service 426 Quantity of parcels without service 0 249 Quantity of parcels without service 0 432 Was service to customers disrupted 0 389 Was service to customers disrupted 100 418 Was service to customers disrupted 0 301 Was service to customers disrupted 100 55 Was service to customers disrupted 100 379 Was service to customers disrupted 0 414 Was service to customers disrupted 100 427 Was service to customers disrupted 25 407 Was service to customers disrupted 100 42 Was service to customers disrupted 100 227 Was service to customers disrupted 100 56 Was service to customers disrupted 0 428 Was service to customers disrupted 0 383 Was service to customers disrupted 100 423 Was service to customers disrupted 0 112 Was service to customers disrupted 0 240 Was service to customers disrupted 50 409 Was service to customers disrupted 75 21 Was service to customers disrupted 75 231 ! Was service to customers disrupted 0 283 Was service to customers disrupted 100 260 Was service to customers disrupted 0 426 Was service to customers disrupted 0 392 Was service to customers disrupted 0 332 Was service to customers disrupted 75 417 Was service to customers disrupted 0 141 Was service to customers disrupted 0 120 Was service to customers disrupted 100 416 Was service to customers disrupted 0 10 Was service to customers disrupted 100 342 Was service to customers disrupted 0 404 Was service to customers disrupted . 75 401 Was service to customers disrupted 0 237 415 Was service to customers disrupted • 100 411 Was service to customers disrupted 100 402 Was service to customers disrupted 0 249 Was service to customers disrupted 100 413 | Was service to customers disrupted 0 431 Was service to customers disrupted 100 76 Was service to customers disrupted 50 72 Was service to customers disrupted 100 134 Was service to customers disrupted 100 408 Was service to customers disrupted 100 403 j Was service to'customers disrupted 100 435 1 Was service to customers disrupted 50 400 i Was service to customers disrupted 0 424 j Was service to customers disrupted 100 405 Was service to customers disrupted 48 Was service to customers disrupted 100 Was service to customers disrupted ' 0 284 Was service to customers disrupted 100 433 Was service to customers disrupted 100 430 | Was service to customers disrupted 50 422 i Was service to customers disrupted 0^ 100 434 Was service to customers disrupted 420 Was service to customers disrupted 100 103 Was service to customers disrupted 295 i Was service to customers disrupted 100 419 i Was service to customers disrupted 100 383 | Estimated number of customers affected 0 428 423 Estimated number of customers affected 0 Estimated number of customers affected 0 414 : Estimated number of customers affected 100 415 i Estimated number of customers affected 0 141 i Estimated number of customers affected 0 404 | Estimated number of customers affected 283 Estimated number of customers affected 0 56 Estimated number of customers affected 0 424 Estimated number of customers affected 100 76 Estimated number of customers affected 0 100 Estimated number of customers affected 403 ! Estimated number of customers affected 100 72 | Estimated number of customers affected 0 10 Estimated number of customers affected 0 400 Estimated number of customers affected 0 402 Estimated number of customers affected 0 435 I Estimated number of customers affected 50 401 i Estimated number of customers affected 0 431 i Estimated number of customers affected 0 2'H Estimated number of customers affected 0 434 Estimated number of customers affected 100 422 Estimated number of customers affected 0 238 120 Estimated number of customers affected 100 4I<> Estimated number of customers affected 100 41 S Estimated number of customers affected . 0 284 Estimated number of customers affected 100 392 Estimated number of customers affected 227 Estimated number of customers affected 50 42 Estimated number of customers affected 0 231 Estimated number of customers affected 0 417 Estimated number of customers affected 0 260 Estimated number of customers affected 0 416 Estimated number of customers affected 0 112 Estimated number of customers affected 0 240 Estimated number of customers affected 426 Estimated number of customers affected 0 389 Estimated number of customers affected 0 430 Estimated number of customers affected 0 301 Estimated number of customers affected 0 103 Estimated number of customers affected 48 Estimated number of customers affected 342 Estimated number of customers affected 0 413 Estimated number of customers affected 411 Estimated number of customers affected 0 433 Estimated number of customers affected 50 409 Estimated number of customers affected 55 Estimated number of customers affected 100 405 Estimated number of customers affected 432 Estimated number of customers affected 0 420 Estimated number of customers affected 100 -21 Estimated number of customers affected 0 249 Estimated number of customers affected 0 408 Estimated number of customers affected 427 Estimated number of customers affected 0 407 Estimated number of customers affected 0 134 Estimated number of customers affected 0 379 Estimated number of customers affected 0 332 Estimated number of customers affected 301 Estimated number of affected customers by residential type 295 Estimated number of affected customers by residential type 0 423 Estimated number of affected customers by residential type 434 Estimated number of affected customers by residential type 407 Estimated number of affected customers by residential type 103 Estimated number of affected customers by residential type 120 Estimated number of affected customers by residential type -1 260 Estimated number of affected customers by residential type 0 231 Estimated number of affected customers by residential type 112 Estimated number of affected customers by residential type 332 Estimated number of affected customers by residential type -1 422 Estimated number of affected customers by residential type 141 Estimated number of affected customers by residential type 239 419 Estimated number of affected customers by residential type -1 405 Estimated number of affected customers by residential type 100 Estimated number of affected customers by residential type 426 Estimated number of affected customers by residential type 0 284 | Estimated number of affected customers by residential type 100 240 Estimated number of affected customers by residential type 389 Estimated number of affected customers by residential type 416 Estimated number of affected customers by residential type 392 i Estimated number of affected customers by residential type 431 i Estimated number of affected customers by residential type 430 i Estimated number of affected customers by residential type 415 427 Estimated number of affected customers by residential type Estimated number of affected customers by residential type 0 134 Estimated number of affected customers by residential type 0 432 Estimated number of affected customers by residential type 424 Estimated number of affected customers by residential type 0 10 Estimated number of affected customers by residential type 100 418 Estimated number of affected customers by residential type 404 Estimated number of affected customers by residential type 342 Estimated number of affected customers by residential type 420 Estimated number of affected customers by residential type 100 72 Estimated number of affected customers by residential type 0 249 Estimated number of affected customers by residential type 76 Estimated number of affected customers by residential type 403 Estimated number of affected customers by residential type -1 408 Estimated number of affected customers by residential type 414 Estimated number of affected customers by residential type 0 435 Estimated number of affected customers by residential type 379 417 Estimated number of affected customers by residential type Estimated number of affected customers by residential type 0 48 | Estimated number of affected customers by residential type 21 Estimated number of affected customers by residential type 0 428 Estimated number of affected customers by residential type 0 227 Estimated number of affected customers by residential type 0 283 Estimated number of affected customers by residential type 433 : Estimated number of affected customers by residential type 55 Estimated number of affected customers by residential type 400 Estimated number of affected customers by residential type 42 Estimated number of affected customers by residential type 0 401 Estimated number of affected customers by residential type Estimated number of affected customers by residential type 413 Estimated number of affected customers by residential type 383 Estimated number of affected customers by residential type 411 Estimated number of affected customers by residential type 402 Estimated number of affected customers by residential type 0 409 Estimated number of affected customers by residential type 42 Estimated number of affected customers by commercial type 0 422 : Estimated number of affected customers by commercial type 231 Estimated number of affected customers by commercial type 0 240 428 Estimated number of affected customers by commercial type 0 407 Estimated number of affected customers by commercial type 0 55 Estimated number of affected customers by commercial type 434 Estimated number of affected customers by commercialtype 0 435 Estimated number of affected customers by commercial type 50 403 Estimated number of affected customers by commercial type 100 134 Estimated number of affected customers by commercial type 0 72 Estimated number of affected customers by commercial type 0 10 Estimated number of affected customers by commercial type 100 400 Estimated number of affected customers by commercial type 0 402 Estimated number of affected customers by commercial type 0 411 Estimated number of affected customers by commercial type 401 Estimated number of affected customers by commercial type 433 Estimated number of affected customers by commercial type 0 431 Estimated number of affected customers by commercial type 0 227 295 Estimated number of affected customers by commercial type 0 Estimated number of affected customers by commercial type 0 417 Estimated number of affected customers by commercial type 0 103 Estimated number of affected customers by commercial type 430 Estimated number of affected customers by commercial type 0 405 Estimated number of affected customers by commercial type 0 100 Estimated number of affected customers by commercial type 392 Estimated number of affected customers by commercial type 0 284 Estimated number of affected customers by commercial type 100 413 Estimated number of affected customers by commercial type 0 419 Estimated number of affected customers by commercial type -1 48 Estimated number of affected customers by commercial type 426 Estimated number of affected customers by commercial type 0 424 Estimated number of affected customers by commercial type 0 383 Estimated number of affected customers by commercial type 332 Estimated number of affected customers by commercial type 100 56 Estimated number of affected customers by commercial type 240 Estimated number of affected customers by commercial type 50 249 Estimated number of affected customers by commercial type 0 432 Estimated number of affected customers by commercial type 260 Estimated number of affected customers by commercial type II 301 Estimated number of affected customers by commercial type 112 389 Estimated number of affected customers by commercial type 0 Estimated number of affected customers by commercial type 0 423 Estimated number of affected customers by commercial type 141 Estimated number of affected customers by commercial type 0 427 Estimated number of affected customers by commercial type 0 408 Estimated number of affected customers by commercial type 418 Estimated number of affected customers by commercial type 0 120 Estimated number of affected customers by commercial type 100 283 Estimated number of affected customers by commercial type 0 415 Estimated number of affected customers by commercial type 100 379 Estimated number of affected customers by commercial type 0 416 Estimated number of affected customers by commercial type 0 241 414 Estimated number of affected customers by commercial type 0 420 Estimated number of affected customers by commercial type 100 409 Estimated number of affected customers by commercial type 0 404 76 342 21 Estimated number of affected customers by commercial type 25 Estimated number of affected customers by commercial type 0 Estimated number of affected customers by commercial type 0 Estimated number of affected customers by commercial type 0 431 Estimated number of affected customers by industrial type 0 76 Estimated number of affected customers by industrial type 0 434 | Estimated number of affected customers by industrial type 0 283 | Estimated number of affected customers by industrial type 0 295 Estimated number of affected customers by industrial type 0 379 Estimated number of affected customers by industrial type 0 10 Estimated number of affected customers by industrial type 100 48 : Estimated number of affected customers by industrial type 414 i Estimated number of affected customers by industrial type 0 408 Estimated number of affected customers by industrial type 435 Estimated number of affected customers by industrial type 75 427 Estimated number of affected customers by industrial type 0 420 : Estimated number of affected customers by industrial type 100 424 i Estimated number of affected customers by industrial type 0 56 Estimated number of affected customers by industrial type 411 Estimated number of affected customers by industrial type 403 Estimated number of affected customers by industrial type 100 404 Estimated number of affected customers by industrial type 25 383 Estimated number of affected customers by industrial type 432 Estimated number of affected customers by industrial type 423 Estimated number of affected customers by industrial type 342 Estimated number of affected customers by industrial type 0 400 : Estimated number of affected customers by industrial type 0 72 i Estimated number of affected customers by industrial type 0 249 Estimated number of affected customers by industrial type 0 401 Estimated number of affected customers by industrial type 134 Estimated number of affected customers by industrial type 0 402 Estimated number of affected customers by industrial type 0 417 | Estimated number of affected customers by industrial type 0 428 | Estimated number of affected customers by industrial type 0 392 Estimated number of affected customers by industrial type 0 416 Estimated number of affected customers by industrial type 0 332 Estimated number of affected customers by industrial type 100 21 i Estimated number of affected customers by industrial type 0 284 1 Estimated number of affected customers by industrial type 100 240 Estimated number of affected customers by industrial type 50 407 i Estimated number of affected customers by industrial type 0 55 | Estimated number of affected customers by industrial type 42 | Estimated number of affected customers by industrial type 0 419. | Estimated number of affected customers by industrial type -1 227 112 Estimated number of affected customers by industrial type 0 Estimated number of affected customers by industrial type 0 242 415 Estimated number of affected customers by industrial type 100 260 Estimated number of affected customers by industrial type 0 409 Estimated number of affected customers by industrial type 0 41? Estimated number of affected customers by industrial type 0 422 Estimated number of affected customers by industrial type 418 Estimated number of affected customers by industrial type 0 433 Estimated number of affected customers by industrial type 0 100 Estimated number of affected customers by industrial type 0 301 Estimated number of affected customers by industrial type 389 Estimated number of affected customers by industrial type ' 0 426 Estimated number of affected customers by industrial type 0 430 Estimated number of affected customers by industrial type 0 120 Estimated number of affected customers by industrial type 100 405 Estimated number of affected customers by industrial type 231 Estimated number of affected customers by industrial type 0 141 Estimated number of affected customers by industrial type 0 103 Estimated number of affected customers by industrial type 10 Estimated number of affected customers by institutional type 100 231 Estimated number of affected customers by institutional type 0 72 Estimated number of affected customers by institutional type 0 419 Estimated number of affected customers by institutional type -1 415 Estimated number of affected customers by institutional type 100 76 Estimated number of affected customers by institutional type 0 414 Estimated number of affected customers by institutional type 0 42 Estimated number of affected customers by institutional type 0 379 Estimated number of affected customers by institutional type 0 55 Estimated number of affected customers by institutional type 301 Estimated number of affected customers by institutional type 417 Estimated number of affected customers by institutional type _0_ 48 Estimated number of affected customers by institutional type 283 Estimated number of affected customers by institutional type 0 426 Estimated number of affected customers by institutional type 0 295 Estimated number of affected customers by institutional type 0 431 Estimated number of affected customers by institutional type 0 100 Estimated number of affected customers by institutional type 0 383 Estimated number of affected customers by institutional type 402 Estimated number of affected customers by institutional type 0 413 Estimated number of affected customers by institutional type 0 249 Estimated number of affected customers by institutional type 0 100 284 Estimated number of affected customers by institutional type 240 Estimated number of affected customers by institutional type 50 407 Estimated number of affected customers by institutional type 0 389 Estimated number of affected customers by institutional type 0 411 Estimated number of affected customers by institutional type 260 Estimated number of affected customers by institutional type 0 134 Estimated number of affected customers by institutional type 0 432 Estimated number of affected customers by institutional type 332 Estimated number of affected customers by institutional type 100 424 Estimated number of affected customers by institutional type 0 243 21 Estimated number of affected customers by institutional type 0 427 Estimated number of affected customers by institutional type 0 428 Estimated number of affected customers by institutional type 0 227 Estimated number of affected customers by institutional type 0 423 Estimated number of affected customers by institutional type , 420 Estimated number of affected customers by institutional type 100 409 Estimated number of affected customers by institutional type 0 433 Estimated number of affected customers by institutional type 0 434 Estimated number of affected customers by institutional type 0 141 Estimated number of affected customers by institutional type 0 400 Estimated number of affected customers by institutional type 0 401 Estimated number of affected customers by institutional type 408 Estimated number of affected customers by institutional type 404 Estimated number of affected customers by institutional type 0 342 Estimated number of affected customers by institutional type 0 435 ' Estimated number of affected customers by institutional type 75 405 Estimated number of affected customers by institutional type 422 Estimated number of affected customers by institutional type 120 Estimated number of affected customers by institutional type 100 416 Estimated number of affected customers by institutional type 0 112 Estimated number of affected customers by institutional type 0 418 Estimated number of affected customers by institutional type 0 403 Estimated number of a