Water Distribution Systems Analysis Symposium 2006

Australia has some of the most variable rainfall patterns of any continent in the world. In addition to this it is experiencing a nation wide drought, which in some areas is the worst on record. Adding to the already problematic nature of our natural water supplies, are some longer-term trends in Cities such as Perth showing significant reductions in annual rainfall and runoff. While unproven, it does raise the spectre of human induced climate change. Hence managing climate variability is the quintessence of water resource management in Australia and it is this variability that creates the greatest challenge for water resource managers across the country. In addition to climate variability, our cities face other significant challenges going forward including population growth, water demand hardening, community expectations for increasing levels of service, a need for improved waterway health, and more sustainable decision-making. Traditionally the urban water needs of Australia's major cities have been met via surface water storages. However with waterways across Australia already stressed, few new future surface water storages are planned. To meet these challenges, major urban water utilities across Australia have been reviewing their water supply planning. In all cases the new strategies are underpinned by diversity as this is regarded as the best means of managing the major risks to our urban water supplies. More specifically, this entails reducing our almost exclusive reliance upon surface water storages and hence climate dependent water sources. The municipality of the Gold Coast in the State of Queensland, Australia has developed a water supply strategy that is underpinned by diversity, adaptability and sustainability. Furthermore this strategy has taken an integrated approach to managing water across the entire water cycle and focused on providing water that is fit for purpose, rather than using potable water for all uses within the home. This strategy will deliver benefits such as significant reductions in potable water consumption, nutrient run off to adjacent waterways and per capita reductions in greenhouse gas emissions. This strategy is well advanced in its implementation to meet the demands of growth in the region along with the current drought. As a consequence, change management issues are being resolved in parallel with the delivery of the strategy. This makes the water industry in Australia at present a challenging and most times rewarding place to be. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water distribution system analysis did not begin with the development of digital computer programs. Engineers were successfully designing, constructing and operating water distribution systems long before the coming of the computer age. This paper traces the development of analysis from Archimedes and the Roman aqueducts, through the development of principles of fluid flow by Newton, Bernoulli and Euler to the development of head loss equations by Chezy, Darcy, Weisbach, Hazen, Williams and Moody. It then looks at how principles developed for individual elements where combined to solve network problems by Cross and the subsequent development of analog computer methods. The paper will show how our understanding of hydraulics did not come about quickly but through an unraveling of one problem after another by some brilliant individuals. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Computation of flows and pressures in networks of pipes has been of great value and interest for those involved with the design, construction and maintenance of public water distribution systems. Many methods have been used in the past to compute flows in networks of pipes. Such methods range from graphical methods to the use of physical analogies and finally to the use of mathematical models. This paper will attempt to catalog and review those methods that have been developed and applied since the dawn of the "computer age" in 1957 when the original Hardy Cross method was first adapted for solution using a computer in analyzing the water distribution system of the city of Polo Alto, California. Subsequent methods have included the "simultaneous node: method, the "simultaneous loop" method, the "simultaneous pipe" method, and the "simultaneous network" method. A brief review of the theoretical framework of each method will be presented along with a critique of the relative advantages and/or limitations of each method. With the advent of computer models of water distributions systems, it has now become possible to analyze more complex network components (e.g. pumps, tanks, PRVs, etc) of the water distribution system as well as to investigate more complex issues associated with their design and operation. Such additional applications have included 1) extend period simulations, 2) transient analysis, 3) water quality analysis, 4) optimal model calibration and design, 5) optimal operations, 6) optimal sampling plans, etc. The utility and feasibility of these additional types of applications can be demonstrated to be largely dependent upon the underlying simulation methods that were first developed beginning in the early 1960s. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Hydraulic modeling of water distribution systems has had a long history dating back to the work of Hardy Cross in the 1930's and earlier researchers. On the other hand, though influenced by a wide range of water quality studies in distribution systems, water quality modeling has its formal roots only about 25 years ago. Over these past 25 years, developments in water quality modeling can be classified by time periods and important milestones. The 1980's was a period of initial development of steady state and dynamic models and initial field studies. In 1991, AwwaRF and EPA sponsored a technology transfer conference in Cincinnati that brought together experts from around the world that had been involved in some manner in the area of water quality analysis and modeling in the distribution system. The conference set the stage and pointed the direction for the next 15 years. In fact, examination of the research agenda shows that many of the topics identified then are still active areas of research and needs today. The 1990's are characterized by the development of user friendly models, research on chlorine dynamics, development of models of storage facilities, and the beginning of DBP and microbial modeling. In the 21^st century, emphasis has been placed on water security modeling, modeling water age as a surrogate for water quality, use of models to improve water quality operations, expansion in modeling of processes and transformations, and a re-examination of some of the underlying assumptions in water quality modeling. At this time, the field of water quality modeling is expanding in breadth, complexity, and general acceptance. However, there are many areas where work is still needed before water quality modeling will become a routine tool that water utilities can use with confidence to make important design and operational decisions. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water distribution systems present many interesting and challenging optimization problems. For example in the pipe sizing problem, the interactions between decisions are complex and not always easily understood or expected. Also, assumptions about decision and problem form alter the problem structure and permit a range of methods to be applied. Constraints and the objective can be linear or nonlinear. Some problems can be posed in a stage-wise structure. Thus, an abundance of optimization applications are found in the literature for the range of water distribution problems. This talk attempts to provide context of the maturity of optimization techniques applied in the area over time. Clearly, computer power is central to the improvement of methodologies and movement toward more complex problems. From the initial consideration of branched systems, other key causes of change include the evolution to considering looped systems, the development of fast hydraulic simulation models that led to the linking of optimization and simulation models, the emergence of water quality as a central issue in distribution systems, and the movement of water distribution oriented optimization software from research to practice through commercialization. Timelines of advances in the application of optimization methods to water distribution system problems are presented. They relate the types of problem considered over time, the methods applied, and the general formulation structures with the changes listed in the previous paragraph and others that altered the way problems were posed and solved. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Urban water facility improvements must be planned to optimize distribution system performance, minimize impacts to the community, and minimize cost. Past practices typically utilized sizing standards with an underlying presumption that more was better (more storage, more pumping). The size and location of major facilities with respect to demands and pipeline capacity drew little attention from all but the most prescient engineers and utilities. Changes to water quality requirements, ever improving monitoring and control technology, continuously improving modeling technology, combined with aging, existing infrastructure in built-out urban areas, community resistance to new infrastructure, and continuous pressure for efficiency and lower costs are driving the requirement for non-traditional distribution system changes. Current and emerging water quality requirements alone demand an optimized and economical system. Water quality will not tolerate storage that is too large or poorly located with respect to demands. Modern SCADA and hydraulic modeling provide the tools for accurately analyzing the conveyance aspects of distribution systems while aging infrastructure, community resistance, and pressure to reduce costs provide both a need for change and a need to get the system improved the first time. System modeling requires a clear understanding of not only existing and future demands, but also a relatively precise understanding of the spatial locations. It also requires a complete engineering understanding of the distribution system operations (which is often based on manual decisions), water qualify requirements, and desired operational standards (typically service pressures, maximum daily demands, and fire flow capacity which drive the perceived need to keep storage maximized). The East Bay Municipal Utility District (EBMUD) Distribution System Master Plan provided an opportunity to examine older engineered systems that no longer serve flow, pressure and water quality efficiently resulting in customer level of service difficulties. A key problem in many of the EBMUD pressure zones is water quality due to excess and poorly located storage. The poorly located storage also results in high or low pressure issues associated with poor hydraulics combined with excess pumping capacity or large localized demands. The case study will demonstrate the problems and the modeling and data steps needed to understand the problem and develop the range of modern solutions to the practical challenges of large existing distribution challenges. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Louisville Water Company (LWC) is in the process of enhancing its distribution system water quality program. The purpose of this effort is to conduct distribution water quality activities in a proactive versus a reactive matter. The key components of this enhanced program includes: distribution system modeling, innovative booster pumping strategies, distribution system modifications, and storage facility surveillance and maintenance program. Other aspects of the distribution water quality program were also utilized to improve the overall program. These aspects include treatment plant operations, the water compliant procedures, and the cross connection program. Ultimately, the result of the program will allow LWC to meet the ever growing challenges of distribution water quality with shrinking available resources. LWC is in the process of developing a hydraulic model. In limited use the model has already pinpointed areas of low flow and high water age. These areas were targeted for flushing before water quality problems developed. LWC predicts that results from a fully developed model will be an integral part of scheduling preemptive flushing. Two new techniques that LWC is using within its distribution system are innovative booster pumping strategies and the installations of by-valves. The first technique involves changing pumping strategies to increase both the storage facility turnover rate and increase mixing within a storage facility. This technique attempts to eliminate satisfaction within a storage facility and reduce water age. The second technique eliminates areas of poor water quality resulting from dead ends caused by pressure boundary valves. A by pass system is installed around the boundary valve. A small volume of water then continuously flows from the high pressure side to the low pressure side eliminating stagnation. As part of this program LWC is also enhancing its storage facility surveillance and maintenance program. The new surveillance program includes the formulation of a storage facility sampling plan that evolves based on current conditions within the distribution system, routine distribution sampling and analysis, and other data analysis. From the data generated by the surveillance program immediate, short term, and long term activities are planned. Storage facilities containing water with unacceptable water quality, if operationally feasible, are drained and refilled with higher quality water. This data can also be used to adjust pumping strategies to help improve storage facility water quality. At the end of a calendar year, LWC uses a newly developed spreadsheet that assesses current year data to determine additional required storage facility maintenance for water quality. Using a variety of factors, such as water quality within the storage facility, operational accessibility, potential for sludge formation, age of the storage facility, cost of cleaning or inspection, and time since last cleaning a system wide inspection/cleaning schedule is developed. Future year inspection/cleaning strategies are also developed based on the results obtained from this spreadsheet. Typically during off-line chemical cleaning an inspection by a certified storage facility inspector is also completed. Results of these inspections are provided to both the LWC plant maintenance coordinator and the plant engineering coordinator at scheduled quarterly meetings. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A hydraulic network model was constructed for the City of Sidney, Ohio water distribution system. It consists of 104 miles of water mains, some of which are well over a hundred year old and serve the original part of the system. The City intends to use the model for available fire flow analysis and a variety of many other applications. It is essential that the hydraulic model be able to accurately simulate the real system conditions. Model calibration is the critical step to achieve this goal. To calibrate the model, Sidney's Public Utility Department collected the pressure data at twenty locations, along with the boundary conditions of the observed tank levels and pump operating status. The calibration proceeds by applying a well-integrated hydraulic simulation and genetic algorithm optimization modeling platform in two phases including the systemic demand calibration to establish an extended period flow balance model and hydraulic grade calibration under multiple loading conditions. The model calibration performed for the Sidney water distribution model improves the confidence in the modeling results. It assists the engineers to better understand the insight into the system hydraulics and enhances modeling quality for evaluating the available fire flow and system improvement design. This paper illustrates the critical steps of the calibration process and the applications of the calibrated model for the City of Sidney. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper describes the use of the City of Chilliwack's (British Columbia, Canada) water distribution system model to validate their proposed network upgrades. The City's water system supplies a population of 62,000 from seven water production wells. In total, over 1,500 pipes (approx. 410 km ranging in diameter from 25 mm to 750 mm), 17 pumps, 12 valves, 2 reservoirs, and 9 elevated tanks are included in the hydraulic network model. The hydraulic model was first calibrated as it is required for fire flow modeling applications. The objective of this study was then to review, validate and prioritize the proposed network upgrades which will improve fire flow capabilities of one particular area of the Chilliwack water system. Proposed improvements consisted of building a new tank, building a new pipe loop, and upgrading existing pipes. Four different scenarios were then defined and fire flow simulation runs were performed to compare the available fire flow at each junction node. The computed available flow which can be delivered to afire was compared with the required fire flow to determine the adequacy of the overall system. The critical node with the minimum pressure, for each fire flow calculation was also identified. The optimal pipe diameter was then determined which met the required available fire flow. This study demonstrated that a complex hydraulic network model can be used as a practical tool for the optimal planning of network upgrades. Enhancement of water distribution infrastructure planning, operation and management is a principal benefit of this study. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Butler County Department of Environmental Services (BCDES) is a regional drinking water supplier with approximately 34,000 customer accounts in southwest Ohio. The service area has been consistently developing over the last 20 years and growth is expected to continue at a rapid pace in the future. BCDES currently purchases water from two different regional suppliers. As a wholesale water purchaser and wholesale provider, BCDES faces unique operational and planning challenges. BCDES has undertaken a Water System Master Plan to identify and plan for improvements through 2025. Factors included in the master plan analysis cover a range of issues from water purchase agreements to design of future improvements for maximum redundancy. To deal with this variety of challenges, the master planning effort is using advanced hydraulic modeling and GIS applications as well as genetic algorithm optimization of distribution system improvements. While optimization is often used to design distribution system improvements for the ultimate demand conditions only, BCDES is pursuing a phased approach to optimization. The optimized distribution system improvements for each 5-year planning horizon will be coordinated to produce a phased capital improvements plan that addresses population growth overtime. This presentation will outline the application of modeling and optimization tools for the case study of the BCDES system. General procedures for incorporating optimization into a phased planning approach will be discussed. The results will provide water utilities with practical options for distribution system master planning using optimization to reduce costs. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Green River, Rock Springs, Sweetwater County, Wyoming — Joint Powers Water Board (JPWB) is in the process of updating their Master Plan including a significant upgrade of their hydraulic model. The resulting detailed model is intended to serve a variety of purposes including water quality analysis. In order to support these diverse applications, an extensive field investigation and calibration/validation program is being performed including three tracer tests and multi-hydrant flow tests using continuous pressure gages. To overcome a lack of previous experience in the use of these advanced technologies, a methodical engineering approach was used including detailed planning, testing and implementation procedures. This resulted in a highly successful field program though the labor costs exceeded expectations. The hydraulic model was developed with the aid of a GIS with demands calculated by assigning meters to model nodes using geo-coded meter data. Historical SCADA data was used to determine representative diurnal demand curves for use in the extended period simulation. The use of these advanced techniques yield a higher degree in confidence in the resulting model, however it comes at a cost. Our intent is to share our experiences, knowledge and the costs associated with this project. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The City of Phoenix, Arizona recently updated its Water System Master Plan (WSMP). During this master plan update project, the City determined that it can not use previously calculated demand factors to project water demands for various planning years as the Sand use patterns used in the previous master plan had been changed significantly. This is very typical of most south western cities because of the explosive growth over last several years. Phoenix used a comprehensive process of developing demands to make sure that the demand is representative. This paper describes the processes involved in developing the demand database from land use and population projections, and also examines the process of validating the projected demands. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Mutual Aid is a concept that has worked well for fire departments, police departments, and other public health agencies for years. This is not so for water utilities, which historically have operated more independently. With real security threats now present, water utilities can no longer afford the luxury of indulging in turf battles and operating in an independent manner. Mutual aid between water utilities can have a profound effect on reliability of service by providing each utility a measure of redundancy in its most critical area, source of supply. By providing options to each water utility during an emergency, Mutual Aid attacks the security issue from a totally different perspective—some would argue a more practical perspective—compared to Vulnerability Assessments. How does Mutual Aid come about? It requires a champion who can lay out a compelling case to surrounding utilities that it would be in everyone's best interest to work collaboratively in order to improve the reliability of service of each utility and hence the greater "regional" water system. The first step is to conduct an Emergency Interconnection Study, which provides answers to the following questions: Is there excess capacity to share? How much excess capacity does each water utility have on an emergency basis? How much can be conveyed through existing interconnections? What chemical compatibility issues may arise? What capital expenditures would be needed to achieve the desired water conveyance? What additional interconnections would make sense? Can emergency aid be accomplished without jeopardizing compliance? What are the emergency scenarios that create the most vulnerability for each utility? Woodard & Curran completed the first large scale Mutual Aid Study examining five water utilities in the Greater Bangor area in 2002. Based on the success of that effort, we were hired to work with ten communities in coastal New Hampshire (Portsmouth, Dover, Durham, Rochester, Hampton, Rye, Seabrook, Rollinsford, Somersworth, and Newmarket) to evaluate the ability of each water utility to provide emergency aid to surrounding utilities. The goal of the study was to provide each utility with water supply options during a man-made or natural emergency in order to maintain uninterrupted service to customers and improve overall security of the regional water system. A skeletonized network model was developed incorporating the ten water systems and proposed interconnections between systems. A 30-day EPS model was applied to separately model each interconnection, and the results were evaluated in terms of the ability of the interconnection to deliver the needed demand, to provide adequate pressures in the two systems, and to assess potential water quality blending problems. The study rated fifteen (15) proposed interconnections with respect to likelihood of success from both a hydraulic and water quality standpoint. The final report was issued in December of 2005 and included recommendations on each of the fifteen potential interconnections. Each of the utilities will be using the report as the technical basis for funding applications to implement the emergency interconnections. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

With the increasing concerns over the security of water systems, there is considerable interest in accurately calculating the potential human exposure resulting from a contamination event in a distribution system. Currently, these calculations are performed at a macro-scale based on the contaminant concentrations estimated by modeling at the nearest node in the distribution system and gross estimates of human consumption of water associated with the node. In this study, the exposure of humans to water-borne contaminants is calculated at the building fixture level based on the movement of contaminated water from the street main and into and through the building. Examples of two different methods for estimating time-varying water demands and routing the contaminants to the fixtures are presented. This proof of concept illustrates the sensitivity of the amount and timing of human exposure to the dynamics of movement of water into and through a building. Future work is in progress to automate the routing procedure, to represent water use by a stochastic demand generator, and to develop demographic-based consumer exposure profiles for use in water network simulations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Several methods have been developed so far which attempt to detect leaks in water distribution systems by taking into account pressure (and possibly other available) measurements. As in many previous approaches, the leak detection problem is formulated and solved here as an inverse problem with unknown leak areas being the calibration parameters. However, unlike the previous approaches, a stochastic (rather than deterministic) leakage detection methodology is developed and used here. This methodology is based on the Shuffled Complex Evolution Metropolis (SCEM-UA) algorithm and is capable of estimating the posterior probability density functions of unknown leak areas in a single model run. Baring in mind the accuracy of similar leak detection approaches, it is believed that the stochastic context used here is more appropriate than the deterministic one used before. To achieve the above goal, the Epanet2 hydraulic network modelling software is linked to the SCEM-UA software in the Matlab programming environment. The SCEM-UA algorithm parameters are tuned to achieve an effective and efficient search for unknown leak areas. The methodology developed is tested on two literature case studies. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents an attempt to analyze the complex unsteady flow behavior of water distribution networks subject to demand pulses. Relevant elementary concepts from the Poisson Rectangular Pulse model and the elastic water column unsteady flow theory, used later in the analysis, are first presented. Engaging an appliance causes a rapid flow raise and related pressure drop and that propagates with high speed (celerity) along the pipes. Similarly, disengaging an appliance causes a propagating pressure raise. Each of those propagating pressure changes is transformed when a junction node is encountered. Depending on the relation of the areas of the pipes that join at the junction, a portion of the moving pressure change enters into the upstream pipes, and the rest is reflected. On this basis, the unsteady flow behavior following opening or closing of an individual water-using appliance inside a home is discussed, concluding that it may be important in in-home piping and service lines. Arriving at the connection of the service line to the street water main, however, only a small portion of the pressure and flow changes enters the main, since its cross sectional area is normally much larger compared to that of the service line. Because of that, the operation of a single water appliance inside a home is almost imperceptible in water mains and larger distribution network pipes. The subsequent analysis of the unsteady flow in pipe networks shows that the demand pulses deform in their path from the demand point to the pipe supplying the entire network, or the corresponding part of it. This way the commonly used approach to obtain the flow in pipes supplying a group of water users as a simple sum of the downstream demand pulses, is not always justified. The actual effect of the unsteady flow on a particular network depends on the length of the network pipes, the distance from the supplying pipe, the wave celerity and the number of branching points in the path. Since water users are located in different points throughout the network, the travel time and the network topology need to be considered too, as well as the effect of simultaneous demand pulses. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Drinking water quality sampling in distribution systems has typically been approached in an ad-hoc manner, based on simple metrics such as geographic coverage that may, or may not, relate to sampling goals. Sampling goals will depend on the envisioned uses for the data: compliance monitoring, operation and control, or research (including model calibration). This paper develops a methodology for quantifying the worth of alternative sampling plans for compliance monitoring, where samples are collected to efficiently represent some feature of the true underlying distribution of space- and time-varying water quality. Conceptually, the evaluation of sampling plans could proceed through prior collection of an exhaustive historical data set. One could then retrospectively examine, for example, the frequency with which critical water quality conditions were identified by a particular sparse sampling of the data. Of course this procedure is impractical due to time and economic constraints, but such a procedure could be implemented using simulated water quality data, so long as the simulated data were thought to represent the true field conditions. The sampling design evaluation methodology developed and applied here is based on Monte Carlo simulation and advanced methods for simulating chemical reaction dynamics in water distribution systems. We have used a general Monte Carlo simulation tool built on an extension of the Epanet network model that allows for interactions between multiple species or components. This core capability allows for generation of alternative water quality scenarios that may represent the true variability and uncertainty in water quality, and reflect the best available knowledge about water quality models and processes for any given application. Significant work is needed, however, on probabilistic models that mimic variation in system boundary conditions, water demands, and network operation. The Monte Carlo framework will be applied to two illustrative situations: sampling of chlorine residuals in the distribution system to ensure adequate residual maintenance, and sampling of chlorination by-products to ensure consumer protection from these potential carcinogens. Both these applications can involve uncertainty and variability in kinetic model parameters, source water quality, water demands, and system operation. The latter application is timely in the USA as it relates to promulgation of the Initial Distribution System Evaluation (IDSE) rule, which requires utilities to develop sampling plans for disinfection by-products that are more likely to capture critical water quality conditions. These results will be presented at the WDSA2006 conference. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The costs of maintenance, repair and replacement of transmission mains and distribution system piping that serve urban and rural areas of the United States is expected to cost approximately $75 billion over the next two decades to ensure that public water systems are capable of delivering safe drinking water. Unfortunately, there is a significant gap between the funds available and the projected costs of infrastructure replacement or rehabilitation. Infrastructure Management Systems (IMS) have been developed to assist utilities and decision-makers in determining how to allocate resources for infrastructure. In this case study, a tool for prioritizing infrastructure replacement was developed based on the existing water infrastructure managed by the City of Tampa Water Department (TWD). Currently, approximately 500 miles of water mains have been slated by the TWD for replacement or rehabilitation. The TWD has a Geographic Information System (GIS) that has been used to map and integrate information on main breaks, service line breaks, customer complaints and modeled water age. Information on fire hydrant spacing and line flushing dates are also integrated into the GIS. Following development of the GIS based infrastructure replacement prioritization system, approximately 3,000 pipe segments were identified and queries were performed to help develop cost-benefit analyses. The results were used to develop a prioritized list of potential capital projects and incorporate the time value of money and event forecasting. The GIS was also used to develop indicators of the overall infrastructure condition. From this analysis it was possible to develop an approach to categorize projects and identify the resources needed to address high priority problems associated with undersized mains, unlined cast iron mains, asbestos cement mains and hydraulic looping projects. As water infrastructure ages, increasing resources will need to be allocated for pipeline rehabilitation and replacement. The availability of reliable and easy-to-use methods for prioritizing capital spending can help utilities identify critical projects and develop cost-effective protocols for addressing rehabilitation, replacement and repair projects. In this study, long-term and short-term infrastructure projects were prioritized using an adaptable GIS platform in conjunction with databases and spreadsheets. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The condition of a water distribution system has strong correlations with community health and economic development. However, studies indicate an urgent need to upgrade the nation's aging and deteriorating distribution systems if they are to continue to provide customers with reliable and safe water supplies. In response to this need, water utilities are applying various performance measurement initiatives including pipeline asset management. A key component of this process is condition assessment of each individual pipeline in order to identify failure-prone pipes and prioritize their renewal. However, the below ground location of pipes combined with lack of standard guidelines or tools to assist in assessment makes pipeline assessment and renewal decisions difficult. In this research, a pipe failure assessment model was developed to assist water utilities with their pipe renewal decisions. A conceptual model was created from a review of case studies, theories and asset management tools with careful consideration of the distinct properties of buried infrastructure and unique operating and maintenance characteristics of the industry. The model consists of several modules (components) written in Visual Basic for Application (VBA) within a Microsoft Excel platform. Excel was selected due to its commercial availability and the use of VBA user-forms simplifies the data entry and analysis features of the program. Rather than requiring extensive field data to determine the cause of breaks, the model's failure prediction and Multicriteria Decision Analysis (MCDA) modules use pipe inventory and break data compiled from the utility's existing operation and maintenance records. Recognizing that pipe renewal decisions are based on risk avoidance as well as on failure probabilities, a unique feature of the model is a consequence module so the decision maker can compare "what-if" infrastructure investment scenarios. The model was refined, through collaboration of water utility professionals. By drawing on the knowledge and experience of these experts, the review process added several features that facilitate the model's use and its responsiveness to the industry 's needs. Initial evaluations from the participating utilities indicate that the pipe failure assessment model will enhance the industry's ability to prioritize pipe renewal decisions and improve their return on investment. Utility personnel indicated that the model's use of routine pipeline operation and maintenance records, combined with its consequence modeling features, addresses the data limitations and risk avoidance characteristics of the industry in a way that is intuitive and understandable to utility staff. Utility personnel also commented that the model adds knowledge and transparency to the decision process, which is critical in an environment in which decisions will have to withstand scrutiny from various interest groups. Lastly, the investigation illustrates the need for better inventory and break data since this data plays such an important role in buried infrastructure planning programs. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The U.S. Environmental Protection Agency (EPA) awarded a grant to Polytechnic University in order to conduct an independent evaluation of the European program CARE-W (Computer Aided Rehabilitation of Water networks.) Purposes included acquiring proficiency at using the various modules, disseminating knowledge, evaluating the resources deployed, and conducting an assessment of the modules. This review presents an overview of the CARE-W program, the development and testing processes and results, a review of the modules themselves, and an evaluation of the impact and penetration in Europe. CARE-W has the commendable merit to have set new standards in water Asset Management (AM) practices. It also recognizes the fact that water AM issues are complex and that an attempt at their realistic and scientific resolution requires to tap into advanced tools already commonly used in other industries. CARE-W has paved the way toward a comprehensive and flexible framework of analysis; clear steps have been defined; tools can be added on, modified, improved. In addition, a community of researchers from many different countries was able to work collaboratively and train future professionals to think creatively about AM within the CARE-W analytical spirit. In its current condition, each module can already be used by trained professionals to help make decisions for future rehabilitation projects. There is a need for further full scale demonstration projects. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Rehabilitation is nowadays a hot topic of the agenda of most European water supply stakeholders. Many existing water mains have reached their expected lifetime, and high capital investments are required to renovate them. Well-devised methodologies for rehabilitation planning are needed, in order to define what, when and how to rehabilitate water transmission and distribution networks. However, rehabilitation can be motivated not only by asset ageing, but also by a combination of other concurrent factors, such as unsatisfactory hydraulic performance, water quality problems (e.g. color, taste, smell, lead), need for expansion, insufficient reliability, high level of water losses, etc. Comprehensive and well-devised methodologies for rehabilitation planning are therefore needed. This paper proposes a general approach to network rehabilitation and uses the case study of Brandoa (Amadora, Portugal) to illustrate it. This network was selected because the identification of a good rehabilitation solution in this case is far from obvious. The network presents a combination of problems, such as excessive pressure problems in some areas, insufficient pressure in other areas, material ageing symptoms in some pipes and very high service connection failure rate. Conversely, Brandoa is a problematic area in social and urban organization terms, having started as an illegal settlement in the 50's. The implementation of the recommended rehab methodology is supported on the use of CARE-W tools, complemented with network simulation and with a Technical Performance Assessment tool. The CARE-W system (Computer-Aided Rehabilitation of Water Networks) was developed under a project of the 5th Framework Program of the European Union and aims for assisting water utilities in setting up strategic and tactical rehabilitation plans. The project ended in 2004, and since then a number of applications took place in various countries. Challenges, constraints and objectives depend on country to country and case to case, and the experience achieved is vital to allow for transforming the current prototype system into a fully developed and professional tool. In Portugal, we aim for enhancing CARE-W prototype, transforming it into a professional software application that is easier to use, is better adapted to the Portuguese context, and is complemented with some new features. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Pipe bursts typically result in a deficient supply in terms of quantity and pressure or the introduction of pollutants. In January 2004 an incident took place in the Netherlands where a dike of a main transportation canal was leaking and almost collapsed. Leakage stopped after closing valves in a water main. This incident raised the awareness of the risks of pipe bursts to third parties (or so-called external effects). The Dutch drinking water branch initiated an inventory of water mains at risky locations and prepared to take corrective measures if deemed necessary. This article discusses how mains at risky locations are assessed based on the likelihood of a pipe burst and the possible consequences. It also includes a list of measures to limit the risk of pipe bursts. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Nowadays water network design is more and more equivalent to network rehabilitation. Design of completely new systems is less and less frequent, whist there is an increasing need for reinforcement, replacement or renovation of existing infrastructures. Multiple design optimization techniques methodologies have been developed, but sometimes these are either too simplistic to be relevant in practical terms, or not easy to apply without the support from the experts who developed them, not only due to the complexity of the background theory, but also for the programming skills and computing capacity required. Engineers can use a number of software packages to assist them in applying some optimization tools. However, algorithm transparency and confidence on the results are key requirements for most designers, and with these packages they frequently feel insufficient awareness and control of the whole process. An added factor for the lack of willingness to use the existing optimization tools is the complexity of the real life problems: the viewpoints that need to be taken into account are frequently conflicting, difficult to express mathematically, and very different in nature. As a reaction to the referred difficulties, designers end up adopting very subjective criteria, based on their own experience and professional knowledge. The current availability of geographic information systems and water distribution simulation models suggest that an intermediate type of approach is appealing. This approach is based on the use of easy to understand and use tools that allow for the assessment and understanding of the systems behavior for the relevant viewpoints to be taken into account. This paper presents how technical performance assessment of water distribution systems can contribute to fill in the current gap between research and practice, specially if adequately integrated into a more comprehensive rehabilitation decision support tool, as presented in Alegre et al. (2006). The same case study, Brandoa distribution system, is used to illustrate how the methodology can be applied. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Pipe ageing leads to an increment both in breakage frequency and leakages, and to a decrement in the system efficiency. Strategies are then necessary to define when and where to operate in order to improve the water distribution system reliability and to reduce maintenance costs. A decision-making tool based on a multi-objective genetic algorithm technique for optimizing water distribution management strategies is proposed. Funds to be spent over a predefined time interval on pro-active actions (i.e. leakage detections and pipe replacements) are assumed available. The aim is to define how to spend this budget, i.e. where and when to make leakage detections and/or pipe replacements, in order to minimize a) passive costs (repair costs), b) leakages and c) to maximize the water distribution reliability by minimizing the volume of the unserved demands over a long predefined time horizon. The constraint of the funds allocated for pro-active actions to be spent in full within the planning time interval is selected to respect the standard way of operating of a water agency. In fact, once an amount for a given purpose is allocated, i.e. an item is defined, the water agency operates to spend this amount in full for that item within the time interval defined, since residual amounts are not necessarily reallocated to the same item, and return to the global budget of the entire set of activities to be planned for the subsequent period. Results regarding a real water distribution system show that the proposed procedure has the potential to be a useful decision-making tool for scheduling leakage detections and rehabilitation actions. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Deterioration of water distribution networks causes a progressive increase of main failures with both a growth of maintenance cost and a general reduction of service performance indicators. For this reason, maintenance activities represent one of the most important items for a water utility budget and therefore, any decision on possible alternative actions, such as pipes repair, rehabilitation or renewal, must be based on economical and technical issues that must be taken into account in both short and long-term planning. Cost of such actions are realty considerable and thus any strategy must be planned on the basis of a careful analysis of failures data. This paper presents a comparison between the statistical distributions of water main breakage data recorded within some districts of the distribution network of an Italian municipality recently subjected to a complete pipes replacement action within a large area. The analysis has been performed considering the breakage patterns over time in relation to several factors such as pipe length, material, age, water pressure, operational conditions, spatial location, type of failure, etc. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The new performance indicator Substandard Supply Minutes (SSM) is an extension of the performance indicator Customer Minutes Lost (CML). CML measures the average interruption time due to failures. SSM measures the average time that no drinking water is supplied. Substandard water supply is defined as supplying water that does not meet the quality standards or water that is supplied in too little quantity (too little pressure or an interruption of the water supply); the standards for water quality and pressure are defined in the Dutch Drinking Water Act. SSM includes interruptions due to failures of the distribution system as well as due to planned repairs and replacements. From September 2004 to March 2005 SSM was measured in a pilot project. Of the 13 Dutch water companies 7 joined this project. In 2006, SSM has become part of the benchmark of the Dutch water companies and most water companies use SSM in their asset management processes. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water distribution systems must be efficiently maintained, rehabilitated, and replaced to guarantee customers a sufficient supply of high quality and affordable water. Assessment of the system reliability and time to failure is required to support long-term planning of the renewal of pipes in water distribution systems. Internal and external corrosion in aging cast iron pipes can lead to mechanical failure in terms of water leakage and loss of hydraulic capacity due to buildup of corrosion products. While external corrosion has been shown to significantly affect the likelihood of mechanical failure, the risk of failure may be further heightened if internal corrosion is occurring. Chlorine consumption due to internal corrosion in aging cast iron pipes is a major factor in such systems, and has been used as an indicator of the rate of internal corrosion. This paper develops a methodology for estimating the probability of mechanical failure and the Factor of Safety (FOS) of cast iron pipes due to internal corrosion using a probabilistic analysis that incorporates the relationship between chlorine consumption and the rate of internal corrosion in a pipe. In addition to the length and diameter of the pipe, the relationship between chlorine consumption and the rate of internal corrosion is a function of the fraction of chlorine consumed within the pipe, the chlorine decay rate, and the velocity of the water and these factors may be considered uncertain at any given time. In this work, the probability of failure and FOS of pipes over a range of exposure times is estimated using the first order reliability method (FORM) to account for these uncertainties. Here mechanical failure is defined as the point at which the corrosion depth is more than the maximum acceptable decrease in pipe wall thickness set by the utility manager. The likelihood of failure increases to nearly 50% by the 100^th year of pipe age. The results indicate that the affect of internal corrosion on mechanical failure is lower than the effect of external corrosion, where the likelihood of failure is estimated at 50% in the 70^th year of pipe age. A sensitivity analysis is carried out to determine the sensitivity of the estimates of the failure probability to each of the uncertain factors. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The paper presents a study on performance evaluation in water distribution systems by using the Value Engineering (VE) methodology. It is composed of three sections—the theory of the Value Engineering, application of the Value Engineering theory in performance evaluation of water distribution systems and a case study. By building a data systems to represent present situation of a water distribution information, the mathematical model of water distribution function systems is studied and proposed. The key part of the paper is analyzed the function structure and the value structure of water distribution systems, and established the Value Engineering method to evaluate performance of water distribution systems. As a new parameter, the value index of water distribution systems can give some good advice to operation, management, rehabilitation of water supply system from economic and technology view. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A key component of water supply infrastructure asset management is the establishment of investment priorities: how to decide among investments of different types, values, functions, and physical conditions? The decision may be straightforward in some cases, but in most situations managers find themselves in the situation of having to adopt subjective criteria. Nowadays, decision making needs a comprehensive approach that takes into consideration cost of building, operating, maintaining and disposing of capital assets over their life cycles. Conversely, factors such as improvement of customer level of service, improvement of system reliability, or minimization of negative impacts of the infrastructures on the society and on the environment should be incorporated into the investment prioritization process. This was the motivation for a project promoted by the water utility EPAL, S.A. and jointly developed with a research team. The objective was to define a conceptual model and develop an operational tool able to assist decision-makers in the prioritization of water distribution system investments. In this context, an integrated decision support tool was developed taking into account direct, indirect and external costs during the life cycle of the assets. The model includes different points of view and service goals important for water infrastructure asset management (e.g., quantity and pressure, water quality, reliability, water losses, energy, operation and maintenance, and environment) and was developed for different infrastructure components (water mains and service connections, treatment facilities, pumping stations and service reservoirs). The model was implemented into an MS Excel-based application that allows for input data and report production. Detailed tests were made for the three most common types of investments identified by the involved water utility. Sensitive analyses were carried out and conclusions drawn about the most important and critical points of view for each case. Based on the results of the model applied to this set of investments, three different alternative approaches for investment prioritization are analyzed and discussed. The paper describes the principles of the conceptual model, summarizes the main features of the computer application, discusses its applicability, strengths and weaknesses and presents some recommendations for further development. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Valves are essential for isolating incidents such as pipe bursts and contaminations. In older networks where incidents occur most, valves have aged and their reliability has decreased. An international workshop on valves in 2001 [Rosenthal et al, 2001] showed that design rules or guidelines for placing valves in a network, to ensure continuity of supply and to minimize effects on customers, are generally lacking. Within the framework of the joint research program of the Dutch water companies a methodology was developed to analyze the criticality of valve locations for their effects on customer supply, characterized by performance indicators such as Customer Minutes Lost (CML). The methodology enables water companies to optimize designs of valve locations in a network, by comparing the designs for several performance indicators. The methodology can also be used to set priorities within valve maintenance strategy, based on critical performance indicators, costs and maintenance frequency. The methodology has been translated into software and applied to a number of case studies. In two case studies for the Dutch water companies Evides and Hydron, the original design of a network was compared to a new design with different valve locations. In both cases, 25 to 50% of the original valves would be removed based on new design rules. To retain the original number of CML, the reliability of the remaining valves had to be raised through increased maintenance. The two case studies prove that the methodology and the software work well. Specific changes such as (re) allocation and enhancing the reliability of valves improve both the performance and the reliability of the network. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Identifying and quantifying the critical elements in a water distribution system has traditionally involved a great deal of judgment. With the coming of computerized hydraulic analysis, it became easier to "fail" a pipe in a distribution system to assess its impact on service. However, when a failure occurs in a real system, it does not remove a single pipe from a distribution system but rather a "segment" which can be isolated using valving. A segment will often include several nodes, portions of pipes, and other elements. In this paper, the critical segments in a real system are identified based on existing valving. Then different rules for valve installation (e.g. N valves per junction, N-1 valves per junction) are used to add or remove isolating valves from the system and determine the performance of the system as a function of the density of valving. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The aim of the "Valve Criticality" project undertaken by Thames Water, UK is to identify critical valves within the Company's water supply network, in terms of their potential impact on the customer levels of service, either when they are shut or fail to shut, during planned or emergency work. The consequence of each valve operation can be measured in terms of the number of properties being disconnected or receiving an unacceptable level of service (e.g., low pressure). An additional benefit of the project is increased knowledge on the system operation, which will enable maintenance and replacement work to be prioritised to the most critical valves. As it is not always obvious which valves are the most critical within a water supply network, the decision was taken to make the assessment using an all mains hydraulic model of each water pressure zone. However, with many thousands of valves to analyse, each requiring separate model runs, even a small zone could take over a week to analyse using conventional modelling software. A decision was therefore taken to approach MWH Soft to enhance their InfoWater Protector modelling software to automate the process so that thousands of valves could be analysed in one batch of model runs. A careful specification of the output results was required to avoid vast quantities of data being produced. Network trace techniques were used to identify each section of the network to be closed in, and the next available valve, should the targeted valve fail to close. The InfoWater Valve Criticality Modeling (VCM) module was released to users on the 13^th May, 2006, ahead of schedule. MWH Soft has now successfully provided Thames Water and all InfoWater users with the capability to assess the Criticality of the many thousands of valves within major supply networks, enabling maintenance and replacement to be targeted to critical valves. The module will help Utilities develop an optimal investment regarding infrastructure maintenance, minimising the impact on customer levels of service and improving knowledge on the operation of the network. The tool is being developed alongside a risk assessment module, also using Info Water, that assesses the impact of valve operation on water quality within the network. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The United States Geological Survey estimates daily water loss at 6,000,000,000 gallons. In 2000, an American Water Works Association (AWWA) sponsored "States Survey" revealed that no standard definition exists for the term "unaccounted for water", there were no consistent definitions for the various components of use or loss and worldwide, and percentage indicators have been found to be suspect in measuring technical performance. This presentation by the A WWA Water Loss Committee is intended to educate industry professionals concerning new water loss control terminology and procedures recommended by the Committee in its 2003 Report to the AWWA. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The current paper focuses on the assessment of real (physical) and apparent losses in a District Metering Area (DMA) of a Portuguese water utility based on collected data during a leak detection and location survey. Water losses have been assessed by means of bottom-up calculations based on the analysis of minimum night flows in the DMA and extended period hydraulic simulations of the hydraulic system. The analysed DMA is part of a Portuguese water distribution network, covers a 0.7 km² area, is predominantly residential (mainly, with domestic consumption), with approximately 6200 inhabitants and a large non-domestic consumer (representing one-third of billed consumption). The DMA has been temporarily isolated by closing ten boundary valves. It has 9.4 km of pipes with diameter from 50 to 300 mm, 589 service connections and 3400 billed consumers. Flow and pressure data have been collected in the DMA input section, during a leak detection survey carried out during an eight-weak period (from 13th October to 13th December of 2004). Data collection frequency was 30 measurements per hour (every 2 minutes). A brief description of the leak detection survey is presented. Water losses have been assessed before and after the leakage detection survey by means of a bottom-up approach based on minimum night flow analysis and dynamic hydraulic simulation of the system. Minimum night flow data, between 2:00 and 4:00, were used for estimating real losses (leakage) based on several assumptions and considerations referred in "Report F — Using Night Flow Data" (WRC, 1994). In order to estimate daily volume of leakage and apparent losses, hydraulic simulations of the system behaviour have been carried out using collected flow and pressure during 24 hours. EPANET was used to carry out the extended period simulation of the system. The bottom-up approach followed can be summarised in a six-step procedure thatwill be described in the paper. Water losses are assessed before and after the survey. Results have shown that: (i) the leak detection survey reduced real losses (leakage) in approximately 27%, which corresponds to an annual water volume of 130 000 m³, despite the seasonal demand variation of 22%: and (ii) apparent losses and unbilled authorised consumption, even after the survey, still represent 33% of total water volume. The balance of the survey was very positive as it reduced real losses in 14.5 m³/h which represents a pay-back of 63 500 €/year. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The evaluation of the leakage level in a water network by minimum night/lows is probably the most commonly used method in practice. The method represents a simple approach to a complex problem, and only requires a limited set of data which may be obtained in a reasonable period of time, making it very attractive. The method, which represents a way to verify a water balance, is even useful in systems where universal metering is in place. This paper analyzes the accuracy of the procedures used to calculate the daily leakage rate assuming that the leakage flow rates at the hour of minimum flow has been properly determined. It explores the influence of the most relevant factors on the leakage level error and proposes guidelines to improve the assessment of daily leakage rates by means of the minimum night flow method. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In England and Wales, water companies are required to report to OFWAT annually on the performance of their networks to demonstrate the current and future level of serviceability to customers. In support of this, a Common Framework has been adopted for the necessary capital maintenance planning, which focuses on the asset condition assessment and risks associated with the probability and consequence of failure. Unfortunately conventional water distribution analysis, supported by computer packages such as EPANET-2, may not easily quantify the actual severity of such failures as most adopt Demand Driven Analysis (DDA) in that the water withdrawal from the system is predetermined and independent of the local pressure available to achieve it, hence the need for Pressure or Head Driven Analysis (HDA) routines. In this study a simple iterative approach known as the Sequential Solution Seeking DDA based HDA (or `SSS' hereafter) routine is employed to compute the actual outflows according to the chosen head outflow relationship(s), within the pressure deficient zones. A range of asset failure scenarios are run on a number of representative UK water distribution network configurations, of special interest being the consequential Zone Of Deficiency (ZOD). Both pipe (link) isolations and bursts incidents have been investigated, in an attempt to appraise the `robustness' of these real systems. ZOD size and the internal Demand Satisfaction Ratio (DSR) are quantified and the security of supply to sensitive customers such as hospitals has also been considered, as a precursor to the formal reliability estimation and the `consequence of failure' appraisal necessary for `risk' based system management. Finally, over the past decade or so UK companies have favoured pressure management schemes to reduce leakage and relieve stress on networks. This, however, reduces the residual head on the system which acts as a buffer when partial failure occurs. The consequences of asset failure before and after Pressure Reducing Valve (PRV) installation on a set of representative UK networks are also compared to evaluate the potential repercussions of this strategy. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Modeling rural networks and intermittent water supply systems is a challenging task because these systems are not fully pressurized pipeline networks but networks with very low pressures, with restricted water supply hours per day, and with thousands of ferrule points and roof tank connections. The alternate emptying and refilling of water pipelines makes it problematic to apply standard EPANET based hydraulic models because of low pressures and pipes without water. EPANET source code was adjusted to allow for modeling pressure dependent demands, for dealing with low pressure and "dry pipe" situations. A configurable tool was developed for incorporating roof tanks into the water supply analysis and for better formulation and schematization of the system hydraulics. Two cases studies, water distribution model of Shillong in India and detailed water distribution model of Dhaka in Bangladesh are used to illustrate the practical use of this approach. The experience from using and adjusting the EPANET engine for the modeling of intermittent water supply systems is discussed in this paper. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A Pressure Driven Model is presented for inclusion within a slow transient model (rigid water column) in water distribution networks. Special emphasis is made to consider the available pressure in calculating the resulting consumption. Nevertheless, integrity pressure constraints are also discussed and solved. Differently from conventional approaches, a formulation involving not only inertia terms for head at free level tank nodes and flow rates at links but also for head at fixed demand nodes, is proposed. With this extension, it is not possible anymore, as in a conventional demand driven analysis, to reduce the system so that it is loop-based nor even node-based. Nevertheless, the approach can be related to hybrid methods for steady states and extended period simulations. It is well known that numerical instability can appear due to initial conditions far from working mode, small diameter pipes, valve closures, pumping turn off/on, deficient water resources. This paper presents an efficient method for numerically integrating this stiff ordinary differential equation. A convenient Lyapunov function allows the reduction of time stepsize when possible. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Conventional water distribution models are formulated under the assumption that water consumption or demand defined at nodes is a known value so that nodal hydraulic head and pipe flows can be determined by solving a set of quasi-linear equations. This formulation is well developed and valid for the scenarios that the hydraulic pressures throughout a system are adequate for delivery the required nodal demand. However, there are some scenarios where nodal pressure is not sufficient for supplying the required demand. These cases may include the planned system maintenances, unplanned pipe outages, power failure at pump stations, and insufficient water supply from water sources. In addition, some water consumptions like leakages are pressure dependent. In this paper, a robust and efficient approach for pressure dependent demand analysis is developed for simulating a variety of low pressure scenarios. A set of element criticality evaluation criteria is also proposed for quantifying the relative importance of the elements that may be out of service. The results are presented for the applications of the approach to the trivial systems and also to a large water system. It is demonstrated that great modeling performance and convergence rates are achieved for modeling pressure dependent demand conditions and evaluating the element criticality of the large water distribution systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Modeling of water quality in water distribution systems can be improved significantly when using stochastic demands. The stochastic demand pattern generator SIMDEUM is based on stochastic information on end users as retrieved from surveys. Stochastic demand patterns from SIMDEUM were applied in a small network model of 550 demand nodes in a residential area in the Netherlands. An EPANET based network solver was used to assess velocities and travel times. The actual measurements agree well with the simulations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A better knowledge of the residential water consumption and the maximum water requirement allows for a more effective design or management of a Water Distribution System (WDS). The analysis of the behaviour of a small number of users in the request of water might be important also to understand the flow demand of numerous clusters of users supplied by a WDS. Indeed, each flow demand node can be seen as the aggregation of several consumers and, consequently, it represents the sum of several water requirements. An empirical study has been developed by monitoring a WDS in a small town, in southern Italy of about 1,200 inhabitants, in order to investigate water demand patterns and which probabilistic model could be better suitable to represent the residential water consumption. In particular, analysis on the experimental data collected from the system under consideration, has allowed to define probabilistic models which characterise the daily residential water demand. This study has highlighted that the goodness of those models depends on the number of users supplied and on the probability that a tap is opened. By means of statistical inferences on a large data sample, it has been shown that — at least for the range of users herein investigated — the Gumbel and Log-Normal distributions best represent the peak water demand; the mean water demand can be considered Normally distributed; the night flow requirement is well characterised with the Poisson model. In addition, in relation to the number of users supplied, the parameters of the probabilistic models have been estimated. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

End-use water demands are one of the most important and vexing inputs to a network hydraulic model. Accurate water demands are essential for meaningful network calibration, validation and simulation exercises as well as for design, analysis and performance assessment. However, assigning water use to network nodes is fraught with many difficulties, including questions about the demand magnitude, the temporal pattern and the cross (i.e., spatial) correlation. It is not unusual in practice to assume that, during extended period simulations, all demand nodes follow the same synchronized hourly demand pattern. In essence, the demands across the distribution system are often taken to be perfectly cross correlated in space. Clearly, the degree of cross correlation between nodal demands (if any exists) could have a profound effect on the results of subsequent hydraulic and water quality simulations. Despite the importance of the water use cross correlation issue, there has been relatively little work on this problem — perhaps due to a lack of field data. This paper seeks to quantify the degree of cross correlation that exists in 365,000 residential water demands collected from 21 single family homes in Milford, Ohio, during a seven month period. This collection of field data represents one of the most extensive high resolution records (one-second interval) of continuous residential water use ever assembled. In particular, the paper investigates two important research questions: (i) How does cross correlation between indoor water demands change with the level of spatial aggregation (e.g., two groups of five or ten homes)? (ii) How does cross correlation between water demands change with the averaging time step (i.e., one second, one minute, one hour, or longer)? This investigation offers empirical insights into the degree of cross correlation between water demands that exists in a small residential neighborhood, including a picture of how the cross correlation between demands is affected by changes in temporal averaging (time step) and spatial aggregation. Principal findings show that cross correlation tends to increase with an increase in spatial and temporal aggregation. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Estimating future demands with a high degree of accuracy in water distribution network design remains an elusive goal. The desired outcome is to match the design demands to the demands that are eventually "realized" in the built system. To this end, this paper explores the cross correlation between demands in an existing system in order to gain a better picture of the representative spatial and temporal patterns of design demands. The aim of the paper is to analyze the cross correlation in the residential demand data collected in the city of Milford, Ohio. More specifically, the paper begins to answer five important questions concerning the cross correlation of the Milford demand data: how strongly cross correlated are indoor, residential demands? How strongly correlated is the deterministic, diurnal component of residential demand? How strongly correlated is the random noise component of residential demand? To what extent does the choice of time step influence the strength of correlation between these 3 demand components? Does the correlation between these 3 demand components differ significantly between weekdays and weekends? To answer these questions, aperiodic regression model was used to isolate the deterministic and the random noise components from the residential demand data collected in Milford. Correlation indices were formulated to measure the cross correlation of residential demand, its deterministic, diurnal component, and its random noise component. The Milford results pointed to a number of preliminary findings: (1) both residential demand and its deterministic, diurnal component had a positive and moderate to high correlation, while the random noise component of demand had a low level of correlation for the cases investigated; (2) increasing the time step length (from 600 s to 3,600 s) did increase the strength of the correlation in residential demand and its deterministic, diurnal component. This suggests that a longer time step increases both the coherence in diurnal demand patterns and their synchronicity. It is unclear whether time step length had any influence on the correlation of the random noise component of demand; (3) both residential demand and its deterministic, diurnal component were more strongly correlated during weekend periods than during weekday periods. This finding suggests that weekend periods may be characterized by less erratic water use patterns between customers leading to more coherent and synchronous diurnal patterns. It is unclear whether the random noise component was influenced by day-of-week effects. The implications of these preliminary results are discussed in the context of extended period simulation (EPS) and water quality modeling as they pertain to cost-effective design. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Understanding the factors that influence daily and weekly water consumption patterns in distribution systems is fundamental for both long term tasks, such as planning, design, expansion or rehabilitation of the systems, and short term tasks, such as routine system operation or emergency management. This paper describes the development of a Portuguese nationwide program for characterizing residential water consumption, through the use of a specific consumption data analysis protocol and the build-up of a database of demand patterns; and for associating it with a range of easily obtained, potential relatable factors that include the technical features of the network (e.g., service pressure, network state and maintenance level, use of household tanks), billing and customer statistics and a range of social-demographic variables, such as age, social-economic level and mobility of consumers, age and type of buildings, and economic activities. The program is currently being test-run on a set of 20 metering districts from 9 water utilities, comprising a variety of network types and demand characteristics across representative urban areas of Portugal. The metering districts range in size from 2,000 to 12,000 connected properties, and have mostly been monitored since 2005 (in some cases, 2004). Total consumption is continuously monitored during the main seasonal scenarios (typically winter and summer) in the areas selected. Standardized daily demand patterns and consumption statistics are produced for each day of the week, as well as for each season, in case there is seasonality in the area. Social-demographic information is based on statistics obtained from the 2001 National Census for the basic statistical units, which are around 300 dwellings in size. After the initial run, the program is to be disseminated and made accessible to all water utilities willing to participate, on a voluntary basis and with data anonymity. This will be done mostly through a dedicated website containing the consumption database, together with query and analysis tools, as well as a supervised mechanism to submit new cases in standard format. It is believed that the publication of the result database and its continued growth will provide the water industry designers, consultants and managers with much more reliable and updated data on water consumption than previously available. It will also represent a means to estimate demand through correlation and profiling, in cases when there are no records to work from. The paper discusses the demand analysis performed, as well as the development of the set of potential explanatory factors (technical features of the network and social-demographic variables), and illustrates with the results obtained so far. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In this paper, analysis results of household water consumption are presented for each component. The water use components are metered by classifying into toilet, washbowl, bathing, laundry, kitchen, etc. in each residential type, such as apartment, multi-household, tenement house and individual house. The data are collected by web-based data collection system from 145 sampling sites in Korea for 2 years from the year 2003 to 2004, considering occupation, revenue, family members, housing types, age, floor area, water saving devices, education, etc. Reliable data are selected by frequency analysis for each observed water use component and statistical characteristics are estimated for each residential type to determine liter per capita per day. Estimated household water consumption per capita day show an indoor water use with the range from 141 L/day to 164 L/day for each housing type with the order of apartment, multi-household, tenement house and individual house. As the order of consuming amount among water use components, it is investigated that bathing is the first, and the second is toilet water, the third is laundry, and the next is kitchen. The results are compares with other countries' water use for each component for the validity of estimated waster use. The effect of life-style and residential types on water use has had a marked impact on household occupancy levels with the rate of economic development, particularly in the more urbanized areas. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

South Africa is a water scarce country that constantly strives to apply its available water resources in the most efficient and equitable manner. Different users including industry, domestic, agriculture and the environment, vie for the available resources and have to be awarded an equitable and adequate share to ensure sustainability in the long term. Proper water demand management in South Africa is thus of critical importance. Researchers, municipal engineers and engineering consultants require good and reliable data and design guidelines on water demand patterns and factors influencing water demand to achieve this aim. The most commonly used South African guideline for municipal water demand was first published in 1983, and is currently included in the document titled "Guidelines for human settlement planning and design". The guideline provides upper and lower limits for domestic consumer demands as a function of stand area. In this paper, the authors analyze measured water demand data of more than a million stands (domestic and non-domestic) contained in the treasury data for a cross-selection of South African towns and cities. The main objectives of the study were to evaluate the existing South African domestic demand guidelines and to evaluate the parameters that influence domestic and non-domestic water demand. The main results of the study indicate that the domestic demands of most of the users fall well within the upper and lower limit envelope curves of the South African design guideline although for larger stands the design guideline was found to be conservative. Stepwise multiple variable regressions were applied to domestic and non-domestic consumption data to determine the most significant variables influencing water demand. In a large majority of cases, either the stand area or stand value had the greatest significance. There is strong evidence that both domestic and non-domestic demand increase with increasing, stand area and stand value. In, the case of domestic demand, stand value was used as an indicator for household income level. The study found that inland domestic water demand is significantly and consistently higher than coastal demand. This trend was not observed for non-domestic demand. However, the development level (city vs. small town) was found to have a marked impact on non-domestic demand. The study also found that the frequency distributions of annual average daily demand (AADD) for specific non-domestic user categories seem to have a lognormal distribution. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Athens Water Supply and Sewerage Company (EYDAP) serves the capital city of Greece, its suburbs and nearby areas, with a total population of 3.7 million people. Total water consumption for 2002 was 421.7 million m³. Four water treatment plants with a total capacity of 2.050.000 m³ per day are in operation. The total length of the water supply network exceeds 7000 km. More than 200 pressure zones are defined due to a wide variation in ground levels within the area of supply. Since 1992, a simulation model of the EYDAP principal mains network (usually of diameter >300 mm) has been developed and applied to improve operations and indicate future capital investment decisions. In developing the model, information from a large number of paper drawings and sketches had to be assembled and processed. In order to calibrate the model, 483 hydraulic parameters were monitored using field measurements as well as data obtained from the existing telemetry system. Due to the extensive area covered by the network, 9 sub-networks were defined for each of which a separate field program was implemented. In view of the additional challenge of operating the network so as to cope with the demands and requirements of the Athens 2004 Olympics, the model was updated and used to simulate the operation of the principal mains network during the period of the Games. In order to obtain water demand and its spatial distribution during the Olympic Games, population estimates were made for each municipality within the supply area, Permanent (resident) population growth, seasonal (holiday) population migration patterns within and beyond the supply area and the additional effect of Games related visitors were considered. Average per capita consumptions for each municipality were obtained by extrapolation from available recorded data; a peak daily factor was then applied based on available records of daily water treatment plant flows into supply. In analyzing available recorded data, municipalities were grouped based on their urban and social characteristics. Demands specifically associated with Olympic venues and facilities (stadiums, press centers, Olympic village, etc) were also added. Two alternative population and demand scenarios were developed with peak daily demands of 1.73 and 1.61 million m³/day for the entire supply area. The network model comprises 3020 pipes and 2250 junction nodes. It was used to evaluate the ability of the network to supply predicted demands during the Olympics, to identify areas of weakness and to help propose capital works and operational control actions in response to those weaknesses. Proposed capital works include new lengths of principal mains, new pumping stations, increasing capacity of existing pumping stations, as well as installing new control valves. Proposed operational responses include modified control valve settings, modified pumping station operation and re-allocation of zonal demands to less vulnerable parts of the principal mains system. The network modeling software used is WaterCAD Version 3.0 by Haestad Methods Inc., USA. GIS data processing has been used to provide advanced data visualisation and thematic mapping and also to automate data processing and avoid tedious manual tasks, particularly with regard to the calculation and assignment of elevation and demand values per network node. The software selected for GIS implementation is ArcInfo Version 8.1 by ESRI Inc., USA. The model has since been implemented on the more up-to-date WATERGEMS software, also by Haestad Methods (now Bentley Systems), USA, which is directly integrated with ArcGIS. It is hoped that the Paper illustrates the value of network modeling in helping secure an adequate level of service from complex distribution systems under demanding conditions of operation. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Statistical water demand models are usually developed as time series coefficients using historically available water demand data, together with any other relevant variables. But structure identification turns out difficult for most of the applications. This study would count on the artificial neural networks (ANN) to forecast the water demand patterns. The ANN model may exhibit a nonlinear feature learned from historical data, in the same way as humans learn from experience. The nonlinearity, high complexity, and uncertainty associated with water demands may favor the potential use of ANNs to compete with or outperform the conventional time series methods for forecasting the similar topics. If the ANNs model is learned correctly, as verified by the accuracy of the predictions using input data not used during the training, the ANN algorithm can be robust with low computation time requirements, even if there are some errors or noise in the input data. Two types of cities, including Oviedo (fast growth) and Winter Springs (slow growth) in the Great Orlando Metropolitan Area in Florida were investigated with respect to monthly data. Such pattern recognition practices would help water utilities identify the expansion and operation strategies in water distribution systems in the long run. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper describes the use of Artificial Neural Networks (ANN) for short-term water demand forecasting. Statistical techniques are used to analyze and identify relevant input variables for the ANN models. A case study is presented in which ANN is used to forecast water demand with a two-day lead time for the Louisville Water Company supply system, Louisville, Kentucky. Previous studies separate the daily water use into non-seasonal use and seasonal use. This paper differs from this method by separating the whole year into summer and winter season. The partition is carried out according to the significantly different climate effects on short-term water demand. In the winter season, climate factors have little effects on water demand. Therefore, the ANN forecast model is simplified by using only historical demand as inputs. The forecasting accuracy reaches 97.21%, which is adequate for water supply system management in the winter. In the summer season, readily available selected climate factors, such as temperature, relative humidity, dew point, wind speed and rainfall are incorporated in ANN models. The difference of weekday and weekend demand is also considered. The optimal summer forecast ANN model uses maximum temperature, relative humidity, rainfall, and historical demand as inputs to forecast the demand in the next two days. The forecasting accuracy is 95.89%. A persistence mode] is employed in both winter and summer season demand forecast for the purpose of comparison. The comparison of results shows that ANN models demonstrate a strong capability in extracting the nonlinear relations between water demand and climate parameters. The ANN approach simplifies the short-term demand forecasting modeling process significantly compared with conventional regression models, and the forecasting results are promising. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The ability to accurately forecast water demand is an integral part of water utility planning for water production levels and the associated finances. Although experienced people are capable of predicting water demand with a reasonable level of accuracy, systematic prediction methods are required to minimize the randomness introduced by human subjectivity, define the quantifiable factors that affect water consumption, and pave the way for further analysis and understanding of these factors by providing the power of what if capability. Statistical models can be used to explain fluctuations in water use and, subsequently, the gained knowledge can be applied to forecasting total water usage. For more accurate studies, the forecasting models may be further refined, in particular by time frame. Short term water forecasting is for water consumption on a daily basis and is useful for decisions related to the daily operation of water treatment plants and distribution systems. Long term forecasting looks at water consumption on a monthly basis, or even longer span of time. These longer term forecasts are useful for revenue forecasting and water treatment plant expansion planning. The factors influencing water consumption are complex. The relationships between known factors and water consumption are often inconsistent and indirect. Furthermore, consumption analysis is limited to the information that is available and quantifiable. There are other factors that are difficult to quantify and difficult to forecast, such as economic factors, social factors, water usage habits, accidents, leaks, fires and climate changes. These factors contribute to the incidence of deviations between historic and current water usage patterns. At EPCOR Water Services Inc. (EWSI), several water demand forecast techniques have been investigated to facilitate better planning of the entire water utility cycle. Among the techniques investigated, pattern recognition has proven to be useful, in particular because it takes advantage of available historical data. This paper closely examines pattern recognition technology, and how its advantages are exploited in both short term and long term forecasting. Furthermore, this paper discusses other techniques used for short and long term forecasting and compares them to pattern recognition. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

On finer time scales, real residential water demand is a non steady stochastic process. The Poisson Rectangular Pulse (PRP) method can be used to model this process, provided the following basic parameters are given: intensity, duration and frequency of the demand pulses, each of them represented statistically by its mean value, variance and probability distribution. Known techniques for characterizing these parameters are based on registering the instantaneous water demand every second, separating the individual demand pulses, and processing statistically the resulting series of pulses. Unfortunately, that "direct" technique generates an enormous amount of data, and because of that is expensive and time-consuming. In this paper, a method for estimating the instantaneous water demand parameters from coarser time resolution (one minute or larger) demand readings is presented. The estimation is based on comparing the theoretical statistical moments of the demand series with the moments of the observed coarser time resolution demand data. The theoretical statistical moments are expressed by the Rodriguez-Iturbe's (Neyman-Scott process based) model for temporal rainfall processes. A least-squares objective function is then formulated for the differences between the theoretical and the observed statistical moments. The minimization of that objective function by a non linear programming technique provides the intensity, duration and frequency of the instantaneous demand pulses. Real residential water meter readings are used to validate the proposed method by comparing the observed water demand series with PRP model generated time series. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A computational tool for leakage, detection in a real water distribution system was developed by the Water Distribution and Sewer Systems Research Center at the Universidad de los Andes in Bogotá - Colombia, supported by the Colombian Institute for Science and Technology Development "Francisco José de Caldas" — COLCIENCIAS. In this research a methodology was developed focused in leakage detection in a real water distribution system. The methodology can simulate non steady flow and steady oscillatory flow. This methodology is based on a recent approach known as inverse transient analysis with genetic algorithms and uses observed pressure data, collected during the occurrence of transient events in a real network, generated by a sudden valve closure, to find the optimal location of the leak with genetic algorithms. This optimal location must minimize the difference between calculated and observed data. The methodology was tested and verified with experimental data collected in a real network located in the town of Chía, near Colombian District Capital Bogotá. The Chía network was selected because of its relatively small size and homogeneity of pipe material (PVC). This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A time-domain analytical solution for the transients in a pipeline with a small leak has been developed under variable boundary conditions. The solution shows that the presence of a small leak reduces the amplitudes of the resonant frequencies while its influence on the non-resonant frequencies is negligible. The amplitudes of the resonant components are a function of input signal, measurement locations, damping factor of the pipeline and leak size and location. As pipe section in a water distribution network can be considered as a single pipe with two variable boundary conditions at both ends. A new leak detection technique, which is able to detect and locate leaks in pipe networks, has been developed based on the analytical solution. The application of the new leak detection method is similar to the widely used acoustic leak detection techniques involving two monitoring locations across a leak. The new leak detection technique uses fully controlled and purposely generated fluid transient signals—coded transients. Compared to the acoustic based leak detection methods, the proposed method may be applied with a much greater measurement interval and is less influenced by the background noises. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The presence of a leak within a single line system alters the dynamic behavior of the system when subjected to transient pressure fluctuations. This change in dynamic behavior is seen in the time-domain as an increased rate of energy loss (damping rate) of the pressure response and in the frequency-domain as an attenuation in the magnitudes of the eigenfrequencies (harmonics) of the system. Two recent developments in the literature on leak detection, based on exploiting the time- and frequency-domain behavior are the transient damping method, and the frequency response method respectively. These methods are both utilize linearized approximations of the original nonlinear water hammer equations. The frequency-domain methods are essentially based on the transfer matrix method (TMM) of modeling the transient pipeline system, whereas the transient damping method utilizes a modal damping method (MDM) of modeling the system. The methods adopt the same approach to linearizing the original nonlinear water hammer equations, however their main difference lies in the manner in which they incorporate the internal boundary condition associated with the leak. These different approaches give rise to different linear systems. The focus of this paper is the comparative analysis of these different systems and the nature of how their dynamics change for different leak parameter values. As is convenient for linear systems, the TMM and the MDM are analyzed in the Laplace-domain, where this analysis takes the form of deriving the Laplace-domain pressure response functions for each method and studying the dependency of the pole locations of these functions on the leak parameters. The Laplace-domain is a useful way to consider and compare the methods as it is shown that the location of the poles of the pressure response functions are directly related to the magnitude of the eigenfrequencies (as used by the frequency-domain methods to estimate the leak properties) and are also directly related to the modal damping rates (as used by the transient damping method to estimate the leak properties). Both methods behave similarly for small leaks, however, for large leaks vastly different behaviors are observed. The TMM yields a solution that is consistent with the underlying equations for the asymptotically large leak case, but the MDM does not, therefore a difference in the behaviors is attributed to limitations on the MDM. Additionally, from the study of the TMM, insight is given into a contention that has existed in the literature on the impact that a leak has on the frequency response of a system. Through studying the structure of the poles for the TMM, it is shown that a leak only effects the eigenfrequencies of a system and that no new eigenfrequencies are introduced as a result of the leak. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In this paper, a periodical pipeline leak diagnosis technique based on transient response difference monitoring is presented. During past two decades, a considerable amount of research effort has been dedicated to the application of controlled hydraulic transients for pipe leak detection and location. Leak reflection method (LRM) and inverse transient analysis (ITA) are the most popular techniques described in the literature. Significant improvements of the theoretical part of the two methods have been presented. However, available experimental results indicate that both LRM and ITA approaches suffer from model precision-related errors. As a result, only large leaks can be detected and located. The fundamental principle of the proposed methodology is the assumption that leak diagnosis will be performed periodically. Leakage detection is considered to be a repetitive procedure that is a part of the active failure monitoring system, which is permanently installed on the pipeline. The technique does not use a numerical model of a pipeline to detect and locate leaks and, therefore, eliminates the model-related errors. A transient response of the pipeline system is measured periodically and the difference between transient responses is monitored to identify the presence of a leak/leaks. Leaks of a relatively small size can be successfully detected and located. The performance of the proposed method was evaluated on the real large water transmission pipeline and positive results were observed. The lower limit of the detectable leak diameter was estimated to be as small as 0.31% of the pipeline diameter. The observed precision of the derived leak location was less than 0.3% of the total pipeline length. Proposed periodical leak diagnosis system allows for a quick and inexpensive leak detection and location in water transmission mains. The technique is also able to detect and locate other pipeline abnormalities, such as blockages and air pockets. The approach can be installed separately or be integrated with the continuous burst monitoring technique that was earlier developed by the authors to form a multi-type pipeline failure detection and location system. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

From a long-term perspective all flows in a water system are unsteady in the sense that no single value persists throughout; all flows continuously change or adjust, either suddenly or gradually, over the system's life. From first filling to final decommission, the flow rate routinely changes from near zero to other, possibly quite high, values. Yet from either a performance or modeling perspective, a detailed or complete consideration not only of all flows, but of all possible transitions in flows, is both impossible and excessive. Thus, it is necessary to restrict attention to specific and targeted design regimes that capture in a reasonable way the key system performance issues in terms of cost effectiveness, hydraulic capacity, structural strength, as well as the system's overall reliability, robustness, vulnerability, and resilience. Yet even this is a tall order and immediately raises three key and intensely practical questions for the analyst: (i) How important is the unsteadiness inflow conditions to understand the system response? (ii) What mathematical models should be used to capture this unsteadiness? And (iii) what spatial and/or temporal discretization is needed to accurately simulate the selected model? A whole host of intriguing considerations arise from these questions that interconnect with issues including the purpose of the analysis, the physical and economic consequences of both conservatism and risk, and both the experiences and biases of the analyst. This paper briefly explores the related questions of modeling intent, physical and numerical approximations, and how these issues map to the physical behavior of the system. Several specific modeling options are overviewed including nearly steady (or extended period) approaches, rigid water column models, (i.e., including inertia effects), and water hammer models (i.e., also including small compressibility effects). This paper provides a brief summary of these considerations and attempts to provide some tentative and preliminary guidance into how these questions of unsteadiness need to be framed or posed. There are clearly issues of physics at stake, in terms of whether certain phenomena are dominant or trivial, but even this evaluation is contingent on the context of the modeling exercise and what specific questions are being answered. A couple of indices are helpful, at least in a limited way, for characterizing the system response, and should perhaps be routinely computed to help analysts gauge or judge the unsteadiness in a system. These include the time scale of boundary and flow adjustments relative to the water hammer time scale, and the size of two unsteady head terms (the acceleration head and the Joukowski head) relative to predicted pressure and flow changes. However, in complex systems the rules must always be supplemented with more detailed numerical explorations and exercises in grid refinement. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Hydraulic transients that naturally occur in pressurised pipe systems propagate back and forth and carry information about different features of the systems, like bursts or leaks, changes of pipe diameter or material, tee-junctions, local head losses, air pockets, closed valves or pipe branches. The aim of the current paper is the report and analysis of different characteristic pressure traces associated with different pipe system features, whose detection, in the transient pressure signal, allows its identification and approximate location. Transient events analysed herein were artificially generated in different experimental facilities and real life systems by the fast closure of side discharge or in-line valves. Transient pressure data were collected at several locations, with and without bursts, air pockets, changes in pipe diameter/material and branches. Different transient pressure traces were observed and are reported: a burst creates a pressure drop; a dead-end reflects totally incident waves; a diameter decrease induces a pressure peak; an air pocket creates a pressure drop followed by an increase; different pipe materials have different pressure wave shapes and damping. Analyses carried out were aided by two diagnosis support tools: a hydraulic transient simulator and an inverse transient solver. Collected data were compared with the results of numerical simulations obtained by running a transient solver incorporating the accurate description of these phenomena in multi-pipe systems, and an excellent agreement was obtained. An inverse transient solver was applied for the evaluation of the location and size of bursts and air pockets, and it successfully estimated them, providing that an accurate transient solver was used, the evaluation was carried out simultaneously with calibration, and the transient event was created by a fast linear flow change. Conclusions are drawn concerning the importance of considering these phenomena since the design phase until system operation and future expansion. Transient-based techniques are particularly important for the diagnosis, monitoring and control of existing water supply systems, not only to identify abnormal features and estimate their approximate locations, but also for a better understanding of the causes of pipe bursts induced by transient events. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Wenatchee Regional Water System was constructed in 1980 to supply all domestic water demands of the Greater Wenatchee, Washington service area. The regional source facility for the system consists of four deep well pumps of various capacity adjacent to the Columbia River, a 7800 gallon air chamber for surge relief, and 10 miles of transmission piping leading to the distribution system and various reservoirs. Included in the service area is the East Wenatchee Water District, which in 1980 opted not to participate and to continue using its own wells. As a result, all system demands have been satisfied by operation of a single large pump. In 2001, increases in demand required the East Wenatchee Water District to become a partner in the regional system. To satisfy the increase in demand, it was necessary to activate operation of two source pumps simultaneously. Transient analyses of the existing system revealed that the existing air chamber was of insufficient capacity to protect the system from a power failure with two pumps operating. A numerical model for transients was used for sizing a new air chamber and to match its operation in conjunction with the existing air chamber. During the course of the study, field measurements of normal pump shutdown were obtained and used to help calibrate the transient model, in particular the simulation of the pump operation and spin down (moments of inertia, etc.). Furthermore, after the new air chamber was installed and with the full service area operating, a series of pump failure tests were conducted. Measurements of water level drawdown and pressure versus time following failure, were taken to confirm the performance and adequacy of the dual air chamber operation. Pressure and discharge readings were taken at various locations throughout the system. Air valves were also monitored for activation during the tests, which would indicate undesirable subatmospheric conditions and potential inadequate surge protection. The tests indicated that the tanks were appropriately sized, with measured results and observations comparing favorably to the transient model predictions. Acquisition of real-world data is thus shown to be useful in model development and in confirmation of the results. The presentation will summarize the transient modeling and investigations, elements of the air chamber design (including air core vortex suppressors and throttling valves set to equalize drawdown in the two tanks), and present results and comparisons of the modeled results and the measured data. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Low or negative pressure transients (surges) can occur routinely in the distribution system (lasting less than one minute) even as adequate steady state pressures are maintained. These transients are most commonly caused by abrupt changes in water velocity when pumps shut down or when fire hydrants and valves open and close. Because backflow can occur when the pressure of water surrounding mains or when the pressure at customer connections exceeds the pressure inside of the distribution system, the potential for contaminant intrusion during periods of transient low and negative pressures exists. Surge models, developed from calibrated extended period simulation models, were used to determine vulnerable distribution system locations in sixteen distribution systems. The distribution systems ranged in maximum day system delivery capacity from 0.3 to 73 mgd and covered a range of operating pressures and configurations. Pressure monitoring (using high-speed, electronic pressure monitors) was conducted in three distribution systems to confirm the vulnerability of the areas determined to be susceptible to low and negative pressure transients via modeling. Surge modeling revealed that systems with less than 10 MGD system delivery, multiple groundwater sources, and few floating storage facilities were more susceptible to low and negative pressure transients. In addition, areas located near a pump station with downstream velocity greater than 3ft/s; greater than one mile away from elevated storage; elevated greater than 40 ft to 50 ft above surroundings; located near a dead end or located near a hydrant on a major main were found to have an increased susceptibility to low and negative pressure transients. By extension, increased vulnerability to low and negative pressures was an indication of increased vulnerability to backflow. Pressure monitoring in three of the sixteen systems studied produced one instance of a negative pressure reading (–0.3 psi) and several instances of pressures lower than 20 psi. The lowest positive pressure measured in these systems was 1.1 psi. Normal, static pressure across the three systems ranged from 25 to 150 psi. The pressure monitoring results confirmed that the models were useful for identifying vulnerable distribution system locations and demonstrated that transient low and negative pressure events were possible in systems where normal, static pressures were considered adequate. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Transient analysis has been shown to have applications for the detection of discrete anomalies such as leakage, blockage and air pockets within pipelines. This technique may be extended to the investigation and assessment of the condition of pipelines. The issues associated with the field implementation of this technique, such as field measurement equipment, synchronization of measurements, lack of reliable system information, low pressures, the presence of air valves and operational issues also require further investigation. This paper presents the analysis of a raw water pumping main using field data collected during transient testing. This composite main of Acrylonitrile-Butadiene-Styrene (ABS) and Ductile Iron Cement Lined (DICL) pipe was designed to deliver 105 L/s and at the time of testing was capable of delivering only 87 L/s. Online and offline testing was undertaken with transients generated by the fast closure of a valve connected to the system via a standpipe. Both openings and closures of the valve were recorded, as were varying nozzle sizes. Online transients were also initiated by pump failures. A combination of transient and steady state analysis is used to determine the presence and location of anomalies within the pipeline with the aim to discover possible causes of the reduced flow capacity and assess the condition of the pipeline. An emphasis on the analysis of transient reflections, from system components and anomalies, is made including investigation of the consistency of results between openings and closures and varying nozzle sizes. The correspondence between reflection size and timing as measured at different locations is used to determine anomaly presence. Issues associated with applying this technique to a real world system are also investigated and discussed. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents an optimization scheme for optimally sizing surge control devices in water distribution systems. The optimal sizing is accomplished through linking a genetic algorithm with the University of Kentucky surge control program. This study is part of an ongoing effort to simultaneously optimize both the system capacity (i.e., sizing of pipes, pumps, tanks, etc.) and its surge control devices, as well as exploring in a multiobjective evolutionary framework the tradeoffs between cost and the surge control devices reliability (i.e., the likelihood of the surge devices to perform their task during a surge event). An example application showing the algorithm performance potential capabilities is explored. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This study describes the methodology and application of a conjunctive hydraulic and water quality aggregation scheme for water distribution systems. The method outcome is a reduced (i.e., aggregated) network having fewer nodes and links which is equivalent to the original system in both the hydraulics (i.e., same nodes heads) and quality (i.e., same nodes concentrations). For the hydraulic part the approach is based on an existing methodology of reducing the linearized model of the full non-linear system while retrieving its non-linear properties, while for the water quality portion the depth-first search technique is invoked to select the minimum number of nodes and links which should retain at the reduced system. The methodology is demonstrated using two example applications. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Computer modeling of large water distribution systems has reached a remarkable stage of development. Applications to the time extended simulation of the network hydraulics and the propagation of quality parameters are widely-used as well as optimization algorithms facilitating the search for the most cost efficient network design and operation. Nevertheless, the maintenance and rehabilitation of real water supply systems being grown over decades requires additional tools for an efficient data management within the application. Additionally, today's data models, often originating from GIS databases, reach considerable sizes of ten-thousands of nodes and pipes. As a consequence, the management of the datasets becomes with increasing model accuracy more and more difficult for the modeler. Besides the well known hydraulic simulation intelligent network management tools are required to support the monitoring of the present state of the network, the processing of telemetry data, the identification of bottlenecks and vulnerabilities. In addition, the need for appropriate network simplification for visualization and efficient handling follows from the large and complex data models that are in use today. In this paper a new decomposition concept of the common network graph according to its connectivity properties is introduced which allows various applications on the field of systems analysis of water supply networks. In this case, the network graph consists of two main components: The outer branched network part, which is called forest, and its inner complement denoted by core of the network. The forest consists of single trees, the core network is composed of distinct biconnected blocks that are connected by bridges. Forest and core are overlapping in the so-called root nodes. Calculations for the simulation or the optimization of the single components can be done independently, thus decreasing matrix sizes and calculation time, but all using the common database of the entire network. With this approach also the controllability of state variables by distinct model parameters is derived which enhances the model calibration and systems operation. Based on graph decomposition applications on network simplification are also presented as well as statements regarding the vulnerability and the reliability of the supply network. Control actions that are necessary to operate a water supply system result in changes of the connectivity of the network graph. By using the presented decomposition approach the in time identification of the different supply areas according to the actual state of the network is straightforward. Integrated in the object oriented hydraulic simulation and optimization model KANET the presented graph decomposition approach allows a flexible processing of single network parts or supply areas. The different degrees of network simplification can help the modeler to get a better understanding of the main network components and their interaction. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The objective of this paper is to quantify the variation of water quality output with the level of skeletonization in a water distribution system model. Contaminant consequence assessment is defined as estimating the population and critical facilities (i.e., hospitals and schools) at risk from a water contamination event. An all-pipe model was used to study this problem. The all-pipe model was thinned of pipes and associated junctions at 2-inch diameter intervals to a maximum of 12-inches. Hydraulic and water quality model output was generated for six skeletonized models and the all-pipe model. The analysis was performed using a GIS-based water distribution model. The results show that velocity, water age, demands at each node, and population density varies with different levels of the skeleton model. Variation in velocity impacts time of arrival of contaminants and their concentration. Increased population density, due to the allocation of a given population to fewer nodes in a model, in turn, affects contaminant consequence assessment. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents the verification of sampling design problem for collecting data from a water distribution network. The aim is to compare the theory verification of model with the real one through developing a calibration procedure based on an optimization algorithm. At first, the multi-objective optimization model is assumed to be solved and the locations of measurement points are determined with different level of accuracy. Three approaches are considered for parameter uncertainty estimation in sampling design. With assuming the measurement points (i.e. the pressure heads) at each node, calibration procedure is made in order to adjust the best parameter (i.e. the pipe friction coefficients). Calibration is performed based on Genetic algorithm (GA) optimization approach. Determined Locations in the solutions of each scenario of sampling design are assumed to be the measurement points in each calibration procedure. Thus, the calibration results are compared in three scenarios with the same number of measurement. Comparison of the results is carried out based on the sum of squared deviations between all pressures calculated from the calibrated model and actual pressures with the real parameter. For the sets of the same measurement numbers, the best fitness solutions are selected based on the sum of squared error (SSE) criteria. The consistency of results for every specific number of measurements shows the robustness of methodology and its safe application to different cases. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Extended Period Simulation (EPS) hydraulic modeling is becoming increasingly popular for distribution system analyses, design applications, and water quality investigations. EPS models are useful because they can provide system performance information that may not be detectable with a steady-state model. Time variable system data, such as collected in a Supervisory Control and Data Acquisition (SCADA) network or registered on paper chart recorders, are used to configure and verify EPS models. Calibration is necessary to assure accurate EPS model results, and calibration procedures beyond those used for steady-state models need to be applied. It is important to develop accurate customer water usage profiles and employ appropriate model adjustment techniques to produce EPS simulated output that agrees with field measured system pressures, tank water levels, and metered flows. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Methods to automatically calibrate water distribution system models have been available for some time but it is very difficult to prove that any method is correct. Since at any one time the ability to know all the usage and flow conditions in a real system is impossible, obtaining all of the data needed in a real water distribution system to obtain an accurate and complete data set for model calibration is unrealistic. To test the ability of automated calibration methods to predict the actual conditions in a water system a laboratory scale physical model of a water distribution system was constructed and an automated water distribution model calibration program, employing genetic algorithms, was used to calibrate the model of that system. The results indicated that the automated calibration methods worked well in estimating pipe roughness, demands and locating closed valves. More specifically, the automated calibration model exactly matched the measured flows and pressures in the system. It was able to identify whether a valve was closed and where the demands were located. If given sufficient data, it was able to identify pipe roughness. The only problems occurred when the number of unknowns greatly exceeded the number of measurements. The model worked equally well regardless of whether the head loss equation used was the Hazen-Williams, Darcy-Weisbach or Manning equation. In all, automated calibration was successful. The paper describes how the lab data were collected, and how the calibration program matched the lab data and provides some suggestions for users of an automated water distribution calibration model. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The concept for this paper originated with the question of how to define acceptable calibration criteria for water distribution models. A 2001 report by the A WWA Engineering and Computer Applications Committee (ECAC) titled Hydraulic and Water Quality Modeling of Distribution Systems: What Are the Trends in the US and Canada reported that only 48% of the 174 survey respondents considered their hydraulic model to be calibrated. Respondents had various definitions of calibration, used various methods to achieve calibration, and accepted varying accuracies of calibration. The definition of calibration criteria will become increasingly important as a result of increasing regulatory focus on water quality in the distribution system. The recently promulgated Stage 2 Disinfection By-Products Rule (Stage 2 DBF Rule) is the first EPA regulation that specifically identifies computer modelling as a method for evaluating monitoring locations. It allows water age and source trace results from calibrated distribution system models to be used as an alternative to extensive field measurements in initial studies required by the rule. Prior to promulgation of the Stage 2 DBP Rule, a 1999 report by the ECAC titled Calibration Guidelines for Distribution System Modeling suggested that additional meaningful discussion is needed among the United States (US) modeling community regarding the need and validity of calibration standards. To further that effort, this paper presents extended period simulation (EPS) model calibration results for several medium and large utilities across the US. The results of the calibrations are evaluated relative to the guidelines in the ECAC report. In consideration of different practices and their impact on calibration accuracies, information on the development and calibration of a model in the United Kingdom (UK) is also presented. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Numerical models of water distribution systems are constructed for various purposes. This paper analyzes the value of data for calibrating a model for three purposes; analysis, design and pump operations for a demonstration system. Comparisons are made using model prediction uncertainty and the change in design and operations cost as more data is collected and provided to the calibration process. It is clear that a model developed for one purpose may not be acceptable for other goals. In particular, a hydraulic model for a system expansion in which the flow distribution increases and changes must be well calibrated. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The incident of attempted terrorist activity directed at the water supply of the US embassy in Rome that occurred in February 2002 has been well documented. The popular consensus is that this is just another example of terrorist ineptitude when dealing with chem-bio attacks. The author's conjecture is that this simplistic interpretation of the attempted attack is not correct and tends to underestimate the terrorists capabilities. An argument is presented that the terrorists were in fact not inept but were highly skilled and were in fact conducting an experiment to verify hydraulic models of the distribution system before the actual attack was launched. It is dangerous to underestimate the capabilities of terrorists as it can lead to complacency. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Disease occurrence in a population due to microbial intrusion in a water distribution system depends not only on hydraulic transport of contaminant in the system but also on factors such as age, gender, socio-economic status and preexisting disease conditions of the individuals. A Salmonella outbreak was identified in the municipal water distribution system in Gideon, Missouri in early December 1993, which affected around 486 of the 1104 residents and caused the death of seven nursing home residents. The Gideon outbreak provided the basis in this research for predicting the propagation of water-borne disease in a drinking water distribution system. It was assumed that if the amount of microbial contaminant ingested at any point in time by an individual exceeds the minimum infective dose for that pathogen, then there is a possibility of occurrence of disease. A model was developed, which utilizes Monte Carlo simulations as well as hydraulic modeling with EPANET, to obtain a range of possible disease outcomes at the network node, assuming different values of initial concentration of contaminant at the point of entry. Using the Kolmogorov-Smirnov D statistic, it was found that variation in the initial contaminant concentration has a significant effect on incidence of disease. Date of disease onset for each case was determined based on an assumption of normal distribution of incubation periods. The attack rates for the Gideon residents, estimated based on the probability of an individual falling ill from consuming one glass of contaminated water, were found to be comparable with the data obtained from a survey conducted in Gideon during the outbreak. The model developed in this research should be useful in predicting the most likely response of a population to a water-borne microbial contamination event. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The paper describes modelling undertaken in 2004 of a water pollution incident that took place in July 1988. The study was completed on behalf of a committee investigating the long term health implication for the public exposed to aluminium sulphate accidentally introduced into the contact tank at a treatment works supplying a rural area in the UK. The purpose of the analysis was to supplement the committee's understanding of the potential contaminant concentration and duration of exposure by modelling the two storage facilities through which the contaminant passed and to quantify the time variable concentration on pollutant leaving the works and then to model the hydraulics of the trunk main system downstream to assess the spread of the pollutant through the network and its change in concentration with time. The methodology used in this study was to analyse each component of the system in turn using the output of the upstream component as the input for the next component: 1) Model of contact tank, 2) Model of clear water tank, 3) Model of network (trunk mains only). The paper presents the background, methodology, analysis, results and conclusions of a practical application of water quality modelling (using CFD and hydraulic modelling techniques) to support the committee's investigations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper proposes a new method for snapshot analysis of water distribution based on the commonly used gradient method. The proposed method uses a secant (intersecting the headloss function in two points) instead of a tangent to approximate the pipe headloss function. A theoretical model is developed for the flow range in which the secant approximates the headloss function without exceeding a given allowable error. This scheme allows a trade-off to be made between the allowable error and the number of iterations required to achieve convergence. The proposed method is applied to three example networks to illustrate its application and benefits. It is shown that the number of iterations required to find a solution can be reduced significantly. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A water distribution system which consists of distribution network and pumping station is considered. The global energy performance and hydraulic power capacity of the water distribution network is analysed. The energetically maximum flows in pipes and networks are determined. The coefficient of output power efficiency and the surplus power factor of hydraulic systems are defined. The assemblage of characteristic curves of water distribution network is analysed from the aspect of energetically maximum flows in network. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Scope of this paper is to propose a unified insight to the problem of solving the looped water distribution systems non-linear hydraulic equations as well as to compare the convergence of the different resulting algorithms. After showing that the problem can be formulated as the solution of a system of equations, partly linear and partly non-linear, it will be demonstrated how most of the widely used algorithms for the analysis of looped pipe networks, can be formulated in terms of a gradient search approach and can be derived from the Global Gradient by simple linear transformations. For the sake of simplicity, the problem of analysing a looped pipe network will be defined on the absence of minor losses, pumps, valves and other concentrated dissipation devices, which may however be introduced without affecting the substance of the results. It will be shown how the different solution methods can be classified according to the space of the state variables used as unknowns (the pipe "flows", the nodal "heads", the mesh "residual flows"), and can be divided into two main categories: a first group incorporating the flow based algorithms and a second group based on the Nodal Heads. Using an extremely simple example, the paper shows that all the flow based algorithms have exactly the same convergence properties, in that they require the same number of non-linear Newton-Raphson steps to reach exactly the same result. Therefore, the computer time consumption only depends on the dimension of the space on which the problem is solved (the nodes, the pipes, the meshes) and on the symmetry and the sparsity of the resulting system matrix. This does not happens when the nodal head solvers are used. In this case the convergence properties of the Newton-Raphson algorithm are affected by the different concavity of the resulting functions, with the consequence of generating less convergent algorithms, which never reach the convergence properties of the flow based ones. even if corrected with relaxation coefficients. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper deals with the problem of extending several readily available packages, originally developed for the analysis of looped water distribution systems, such as for instance EPANET2, to perform more realistic "extended period simulations" (EPS). The EPS problem can certainly be solved by introducing a pressure-demand relationship at the nodes, as proposed by several authors, but apart from the difficulty of defining a meaningful relationship, this approach tends to reduce the convergence of the original algorithm due to the introduction of non-linear nodal head unknowns in the system matrix with a different concavity, which affects the convergence properties of the Newton-Raphson algorithm. Furthermore, the alternative use of "emitters", as for instance proposed in EPANET, does not solve the problem, since it may lead to inconsistent and/or incorrect solutions. The purpose of this paper is to show that in reality the definition and the introduction of a pressure-demand relationship is not strictly necessary. Taking advantage of the minimization properties of constrained convex functions and some logical reasoning, an iterative technique based on alternative nodal conditions, is here introduced as a simple procedure to allow most of the existing packages to perform more realistic EPS. The performances of the new approach are demonstrated on a numerical example also by comparing the results obtained with the proposed approach to the solution originating from EPANET2 with "emitters". This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents a method to solve the pump scheduling problem using Ant-colony optimisation. In the past, many researchers have attempted to solve the problem of optimal pump scheduling for complex water distribution systems. Despite the practical importance of the problem, many of the methods published have shortcomings range from lack of generality in the representation of state variables to excessively high computational effort. The present work considers the minimization of cost of pumping as the objective and incorporates the constraint on number of pump switches into the representation. The main emphasis of this work is on restricting the search space and reducing the number of function evaluations. Solutions to pump scheduling problem are, usually, encoded using a binary string showing the state (on/off) of each pump during each time interval. Using binary pump schedules and Ant-colony optimisation, a new method for optimal pump scheduling is developed. The representation of a solution is modified to restrict the number of pump switches to a specified value and to suit the application of Ant-colony optimisation. A comparison of the results obtained using the proposed method was compared with those obtained using Evolutionary algorithms. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Derceto pump scheduling optimization software package is a complex real-time and fully automatic pump scheduling system designed to minimize energy cost while maintaining water quality. It is now in use in four large US municipal operations. One application operates in the real-time electricity market and the optimizer is capable of making pump schedule decisions in real-time to optimize energy costs while not compromising operational requirements or water quality. The challenges overcome include generating 48-hour ahead schedules for over 80 pumps in less than 2 minutes using nothing more than standard office PC hardware and overcoming operator resistance to automation of critical distribution systems. Results showing the benefits of optimization including actual energy savings results and water quality improvements from four case studies are given. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Existing optimal scheduling packages for water distribution systems (WDS) calculate the time schedules for control elements such as pumps, valves and treatment works following the idea of open loop control. The calculations take into account the hydraulic model of WDS, operational constraints and the time dependent electrical tariff. The time horizon is typically one day or one week depending on the network storage capacity and the structure of the electrical tariff. If such a schedule is applied to a physical system the predicted and the physical reservoir levels diverge due to uncertain demands and as a result the calculations need to be repeated at some time using updated demand prediction and the current measured reservoir levels. The aim of this paper is to investigate the feasibility of synthesising continuous feedback from reservoir levels to pump and valve operation, taking into account the operational constraints and the electrical tariff. This attempt can be seen as a generalisation of a simple practical rule where a pump is controlled by the level of an associated reservoir, e.g. the rule `if the tank is full switch the pump OFF, if the tank is empty switch the pump ON'. The proposed methodology relies on preparing a control law in an off-line mode which is subsequently used in a real time situation. The traditional optimal scheduling problem is solved many times for typical initial reservoir levels in the system and a family of reservoir and corresponding control trajectories is generated. These trajectories are approximated by a hyper-surface which represents an optimal control law ready to use in the field. The control action will follow the sequence: log the time, measure the current reservoir levels, evaluate from the hyper-surface the corresponding operation for pumps and valves and finally apply the control action to the pumps and valves. The approach avoids time-consuming online-recalculation of pump schedules, and additionally the feedback control is much more robust with respect to uncertain demands. The limitation of the approach is dictated by the number of reservoirs and number of initial reservoir levels which need to be considered. In the worst case the computational complexity may increase exponentially with the number of reservoirs.^ Initially this approach was applied to a simple system containing a single reservoir fed by one pump station with a time varying tariff and subsequently to a medium-size system with four reservoirs and four pump stations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The efficient operation of a system is a fundamental tool to postpone the system's service life as much as possible, thus ensuring a good service to the consumer while keeping electrical energy and maintenance costs at acceptable levels. Efficient operation requires knowledge of the system, for this knowledge, supported by tools such as models for hydraulic simulation, optimization, and definition of rules, provides the operator with proper conditions for the rational operating of the system's units without depending exclusively on personal experience while maintaining the system's reliability. The purpose of this work is to present a methodology to achieve the optimal operation of water distribution systems, essentially macro systems (skeleton), concerning to the costs of operation and hydraulic benefits. It represents an attempt to provide adequate operation rules in order to minimize cost and maximize hydraulic benefits. Based on the knowledge of the system supported by technical and commercial geo-referenced records, the intention is hydraulically simulate the system using EPANET 2, optimize its operation though multi-objective genetic algorithms (MOGAs) and produce operational rules through machine learning (ML) and the preference ordering routine (POR). In this study an elitist multi-objective genetic algorithm, called strength Pareto evolutionary algorithm (SPEA) is used to obtain a Pareto front. The optimal operation analyses were conducted on the macro water system of the city of Goiânia in Brazil. This was in order to minimize the cost of the electrical energy in the pump stations and to maximize the hydraulic benefits in terms of the required pressure at the demand nodes and of adequate reservoir levels. The results show that solutions for satisfactory operation can be quickly produced as a substitute to the personal judgment of the operator. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Since the publication of the famous Anytown benchmark problem as part of the "Battle of the Networks: Epilogue" by Walski et al., researchers have been discussing and investigating different approaches to the network design problem. The search methods being used include hill climbing algorithms, genetic algorithms, nature-inspired methods such as ant colony optimization, simulated annealing, etc. and also hybrid algorithms combining two or more of the aforementioned methods. The fact that researchers continue to investigate the network design problems proves, that a general approach which would satisfy practitioners and researchers still needs to be developed. Recently the research emphasis in the system design problem has been considerably shifted from the original pipe selection problem to more complex problems of multi-objective optimization including cost, reliability and even water quality aspects. This paper discusses a general question of whether the design process will be ever fully automated and a specific question, which has not been sufficiently explored by the water community, of a relationship between system design and scheduling problems. This paper will discuss the dependence between system design and scheduling of control elements simultaneously. It will show that network layouts differ significantly depending on whether scheduling control elements (i.e. pumps, valves) were part of the original design problem or not. Therefore, initially a descriptive approach to the design problem is taken; which will be translated into an appropriate mathematical formulation at a latter stage. In this first stage all possible elements (i.e. pumps, valves, pipes) as well as nodes (reservoirs, sources) to be designed or scheduled are considered. A decision tree will be presented to facilitate the design process taking into account the scheduling and hydraulic feasibility. The approach will be able to handle component sizing (i.e. pipe diameters, reservoir volumes) as well as deliver mixed-integer schedules for the respective control elements. New objectives like reliability will be included in the considerations as part of the multi - objective problem formulation. Another difficulty when designing networks is the lack of complex benchmark problems, apart from Anytown, to test newly developed algorithms. Most of the commonly used benchmark problems are concerned exclusively with the pipe selection problem for gravity-fed networks. In this situation one may challenge the applicability of successful methods to more complicated cases. In this context, the authors try to investigate the applicability of the current state-of-the-art methods from simple benchmark networks to more difficult (complex) real-life scenarios and estimate the technical boundaries for any given approach. Their findings will complete the aforementioned decision tree for solving combined design and scheduling problems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Genetic Algorithm (GA) optimization is a proven technique for determining the best design for Water Distribution Systems (WDSs). Recent advances in WDS analysis have included water quality issues and uncertainty in system parameters. Water quality modeling requires an extended period hydraulic simulation to be executed, generally requiring a shorter time step and is therefore more computational intensity than steady-state modeling. Similarly, dealing with uncertainty is also more computationally intensive, when using a Monte Carlo (MC)-based method to calculate risk measures, such as reliability and vulnerability. The incorporation of both water quality and uncertainty into GA-based WDS optimization therefore results in impractically long run times. This paper uses metamodels to significantly reduce these run times. A metamodel is an approximation of an existing model, which takes less time to run, making it much more computationally efficient upon repeated use, such as in MC-based analyses or GA-based optimizations. The specific type of metamodel used in this research is an Artificial Neural Network (ANN), as it is capable of approximating any function without specifying, beforehand, its form. Previous research has illustrated the ability of ANN metamodels to approximate pressure heads, chlorine residuals, and both hydraulic and water quality risk measures. This research extends that work by linking those metamodels with a GA to enable risk-based optimization of WDSs by including system uncertainty. The approach developed was applied to an adapted New York Tunnels problem that incorporates water quality in the form of chlorine dosing for disinfection, and uncertainty in the demand. A series of single-objective optimization runs showed that there is a trade-off between minimizing cost, maximizing reliability and minimizing vulnerability. Using ANN metamodels for the risk-based optimization of WDSs that includes water quality issues had run-times of 30 hours, whereas the traditional approach (using EPANET rather than ANN metamodels) is estimated to take in the order of 10,000 hours. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A significant amount of research has been carried out into the optimization of water distribution systems (WDS). In recent times evolutionary algorithms such as genetic algorithms (GA) and ant colony optimisation have been used to optimize the design and operation of such systems. Most optimization studies have focused on minimizing cost as the sole objective, although there have been a few exceptions to this. The cost objective can be expressed as capital cost, operating cost or whole-of-life-cycle cost. In view of increasing global concerns about sustainability issues, it is appropriate that a number of additional objectives should be considered in future optimisation studies of WDS. The sustainability objectives most directly relevant to WDS include energy consumption, CO₂ production and the consumption of non-renewable resources. This paper outlines an approach that considers these other objectives and demonstrates it with a case study. The case study is the design of the Darling Anabranch Water System (DAWS) in New South Wales, Australia. DAWS supplies water to 70 agricultural properties for domestic, irrigation and livestock needs. It includes over 300 km of pipelines, two major pump stations and two booster pump stations. Although DAWS has a fairly simple layout, the problem lends itself to optimization due to the large number of options available including two possible water sources, choices among pipe sizes, materials and classes and primary and booster pump selection. The objectives considered in the study include the present value of total costs, energy consumption, CO₂ production and the total mass of pipe material used. An optimum design that emphasizes the sustainability objectives is identified. It turns out that this design is better on sustainability grounds and also cheaper than the Original Design. The potential application of this optimization approach to other WDS could result in significant reductions in the consumption of non-renewable resources and generation of CO₂ emissions and hence assist in the goal of developing sustainable systems for future generations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

During last five years, harmony search algorithm has been successfully applied to various optimization problems in civil engineering, such as structural design, vehicle routing, and environmental parameter calibration. The algorithm was also frequently applied to water network problems for layout geometry, pipe diameter, and pump switching. This study focused on pipe diameter optimization while minimizing design cost, then compared the results of HS with those of other meta-heuristic algorithms such as genetic algorithm, simulated annealing, tabu search, ant colony algorithm, shuffled frog leaping algorithm, and cross entropy. The HS performed better in terms of design cost or computing efforts. For the fair comparison, identical hydraulic coefficients were used by standard network simulator EPANET. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Large amount of money is being spent every year on piped water supply systems. The water distribution systems share nearly 70 to 85% of the total cost of a water supply project. Providing a solution merely satisfying the functional requirements is not enough. The extent to which the cost can be reduced, whilst ensuring all functional requirements, should be an integral part of the design. In this paper cost structure of pipe network system has bee discussed for constructing an objective function based on the cost. The objective function or cost function is a function of decision variables, which are also commonly known as design variables like pipe diameters and pumping heads. Such a function has to be minimised subject to safety constraints like minimum pipe diameter, minimum terminal pressure head and maximum velocity of flow; and system constraints like nodal discharge summation and total loop headless summation. These constraints along with essential parameters for network sizing like water demand, rate of water supply, peak factor and design period are described in this paper. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The experience achieved by the Water and Sewerage Systems Research Center (CIACUA) at Universidad de los Andes, in the application of the methodology that establish the optimal water head pressure at the entrance of a hydraulic sector in a drinking water distribution system, has allowed to obtain conclusions and results for the Colombian case. The methodology permits an efficient management of the resources and a suitable cost planning of capital and operation of the network. It also helps to reduce the commercial and technical water losses as it decreases the amount of burst pipes and raises their life time due to a less operational pressure. As well it identifies zones with high rates of technical losses to detect leaks and illegal connections, and spatial localization of users with consumption zero. All of these are done to develop renewal and rehabilitation strategies as well as long term investment in the networks. An essential part of the methodology is the calibration of the systems that returns a model that can be used as an operational tool. This article summarizes the differences between the results when the methodology is applied, comparing the calibrated models obtained and making recommendations for the different networks and future investigations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The City of Ottawa has a need to update its water supply system expansion with a plan that identifies the most cost-effective future growth-related projects needed to meet future demands for water, with appropriate water system operating rules. To achieve this result the City commissioned an optimization study, employing structured optimization search technology, aiming to select the best combination of watermains. pumping stations and storage facilities to achieve target levels of service at least overall cost, considering capital investment, energy, and operations and maintenance costs. The study was carried out by Delcan Water, with specialist optimization technology support from British-based Ewan Group plc. The design was subject to a number of performance constraints related to level of service, water quality and system reliability. Targets in these areas were to be met through the optimal selection of elements within various categories of infrastructure such as new reservoirs, reservoir expansions, new pumping stations, new watermains, and treatment plant capacity expansions. The hydraulic success of these selections involved a number of complex pressure regime changes, requiring specific combinations of options to be selected in order to meet the level of service and operational constraints. This paper describes the key objectives of the study, including the various targets to be met by the design. It describes how suitable potential schemes were identified, with supporting cost functions; and how these details were interpreted to be suitable for input to the optimization model. The application of the GAnet optimization model to this complex problem presented a broad range of design objectives; these, are set out, together with an account of how the optimization tool was modified to address them. The essential partnership between technology, sound engineering and strong system knowledge has been a fundamental element of this study which, at the time of writing, is progressing towards increasingly better solutions as the optimization problem definition evolves. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Niagara County Water District provides drinking water to a population of approximately 150,000 in the Western New York counties of Niagara, Erie, and Orleans. In 2002, the Niagara County Water District (NCWD) identified several improvements at the Shawnee Road Pump Station that would reduce excessive energy consumption and increase overall station efficiency. They looked to Wendel Duchscherer Architects & Engineers P.C., to provide professional consulting and engineering services related to the feasibility evaluation, design services, construction services, start-up services, operating enhancements, troubleshooting, and energy grant services in connection with the upgrade of the Pump Station. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A water quality (WQ) model of the distribution system is a tool to more efficiently and effectively perform distribution system related activities. Greater Cincinnati Water Works (GCWW) has been using the detailed skeletonized WQ model consisting of 1700 miles of mains, contained in 10000 pipes. The model represents 63% of the system length and 91% of the volume. GCWW operates a multi-divisional team to utilize WQ model in most effective manner. First, the team identifies and prioritizes system concerns and potential applications. Second, the Water Quality and Treatment Division implements selected applications. This strategy will result in improved delivered water quality and elevated safety in the distribution system against security threat. The WQ model helps GCWW currently to determine infrastructure improvement requirements, to test the impact of operational changes on quantity and quality, to predict water quality such as age, disinfectant and DBFs, to assess the ability to comply with developing regulations such as the DBP Rule, locating sample points, and to plan security measures in the distribution system. Examples of these activities from GCWW's current practice will be provided. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Storage reservoirs in water distribution systems have historically been designed to meet hydraulic requirements and not water quality issues. With initiation of the Stage 2 Disinfection By-products Rule, the negative impacts of reservoir operation on water quality have become an issue that can require costly mitigation activities. Through modifications to the operations of reservoirs a utility can maintain the traditional hydraulic capacity of storage reservoirs while improving water quality in the reservoir. Through the use of hydraulic modeling, the operation of reservoirs in two distribution systems were evaluated to determine a relatively low cost solution to existing water quality problems in reservoirs and the accompanying distribution system. This presentation will discuss the water quality issue, steps taken in the hydraulic analysis of the situation, and a discussion of the recommended alternative. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Current practice for the management and extraction of meaningful information from available data, to inform understanding of water supply system performance, is limited, time consuming and of inadequate accuracy. This is primarily due to reliance on human data analysis and interpretation, which is unfeasible for the volume and complexity of data involved. Further, it is often only practical and economically justifiable to instigate such analysis in a reactive manner to improve problem definition such as the determination of the magnitude of a burst once it has been recognised through a breach of serviceability limits. It is not cost effective to rely on human data analysis and interpretation and the inherent inefficiency contributes to lowered regulatory compliance and standards of service. A new automated approach that applies Artificial Neural Network (ANN) technology has been developed to provide more efficient and consistent analysis of large data volumes, ANNs are trained to infer the future behaviour of a system, or to classify current behaviour, by analysing data describing its past performance, so that problem solving is a matter of learning by example rather than programming. This is especially attractive for domains where an understanding of the problem to be solved is limited, but where training data is readily available, for example, flow and pressure time series from a water supply system. The system developed can be applied to historic flow time series data to construct a grey box model for the detection and classification of burst mains. Initial ANN data analysis to construct a probability density model of the future flow profile was followed by application of a Fuzzy Inference System for classification such that confidence intervals could be assigned to the detected burst events. This artificial intelligence analysis system was applied to four month periods of sample flow data (provided by a major UK water company) from district meter areas of various size, complexity and connectivity and successfully identified a number of burst events within the data; some of which could not have been detected by manual analysis. When integrated with real time communications, the `intelligent' analysis system developed can be readily applied to an online environment providing the capability to proactively manage water supply systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

SCA-Red is a software application developed under the environment of the SCADA systems in charge of the real time control of water distribution systems, whose aim is to use the hydraulic model of the network as a background in order to help network operators to take more reliable decision by increasing the amount of information on the system, by anticipating the results for immediate actions, by playing back past scenarios for training or by suggesting the future actions to take in the next hours. One of the more outstanding features of SCA-Red is its general purpose conception, in order to be installed as a complement of any real existing SCADA, thanks to the component oriented conception, the client-sever architecture, and the connection with the main SCADA via OPC. So, multiple managers or operators can take benefit simultaneously of this assistant. Currently SCA-Red is being implemented for Aguas de Valencia water company, to manage a network of 1,200 Km of pipe length, which supplies water to a population of 1.5 million, regulated by near 40 valves remotely operated. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The sensor placement problem (SPP) in contaminant warning system (CWS) design for water distribution networks involves maximizing the level of protection afforded by a limited number of sensors. In existing SPP formulations, the protection level is typically quantified as either the expected impact of a contamination event, weighted by occurrence probability, or the proportion of events that are detectable. In these formulations, the issue of how to mitigate against potentially high-impact events is either handled implicitly or ignored entirely. Consequently, any solutions of these formulations run the serious risk of failing to protect against any number of high-impact, 9/11-style attacks. This risk is further amplified by the fact that reliable estimation of contamination event probabilities is extremely difficult, such that existing SPP formulations may significantly discount the potential of high-impact events. In contrast, robust formulations of the SPP directly address these concerns by focusing strictly on a subset of high-impact contamination events, and placing sensors to minimize the impact of such events. We introduce several robust formulations of the SPP that are distinguished by how they quantify the potential damage due to high-impact contamination events. These include minimization of the worst-case impact, the Value at Risk (VaR), and the Tail-Conditional Expectation (TCE). The worst-case formulation is equivalent to the p-center problem in facility location theory. VaR and TCE are standard measures of robustness in the financial literature; the corresponding robust formulations of the SPP respectively minimize the (1-α)% largest impact and a weighted sum of the α% largest impacts. All formulations can be expressed as Mixed-Integer Programs (MIPs), which can be solved using both commercial MIP solvers and specialized heuristics. Additionally, we develop computational methods for exploring the performance trade-offs between robust and expectation-based SPP formulations. We use this framework to explore the nature of robust versus expectation-based solutions to the SPP on three real-world water distribution networks, ranging in size from 400 to over 10,000 junctions. We observe that robust SPP formulations are one or more orders of magnitude more difficult to solve than expectation-based SPPs. Our results indicate that simple heuristics yield optimal solutions to the smaller test problems in shorter run-times than MIP solvers, and yield higher-quality solutions for larger test problems. For realistic sensor budgets, solutions with low expected impact fail to protect against large numbers of high-impact contamination events (with impact 5–10 times larger than the expectation). In contrast, we show that solutions to robust SPPs yield 10–25number and magnitude of high-impact events. In general, our results indicate that it is possible to trade off mean impact versus high impact performance in real-world water distribution networks, exposing a key, unexplored dimension in the design of sensor placements for CWSs. Further analysis indicates that the performance of solutions to the worst-case, VaR, and TCE formulations is strongly correlated. Consequently, it may be possible in the future to restrict focus to the worst-case robust SPP formulation, which is significantly easier to solve than the VaR and TCE variants. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Cities without an early warning system of indwelling sensors can consider monitoring their networks manually, especially during times of heightened security levels. We consider the problem of calculating an optimal schedule for manual sampling in a municipal water network. Preliminary computations with a small-scale example indicate that during normal times, manual sampling can provide some benefit, but it is far inferior to an indwelling sensor network. However, given information that significantly constrains the nature of an imminent threat, manual sampling can perform as well as a small sensor network designed to handle normal threats. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Accurate modeling of chemical transport in water distribution systems depends on accurate knowledge of temporally and spatially variable water demands. Typical network models would include water demands that are allocated from billing or census data, and thus may not be appropriate for specific operational analysis, such as emergency events arising from intentional or accidental contamination. During such an event, water consumption patterns may be significantly different from those assumed when developing the hydraulic model, and may change significantly over short time periods due to the unusual circumstances of the event. To allow accurate hydraulic and water quality prediction in real-time, the water demands should be updated continuously to reflect current conditions. The development of such a real-time water demand calibration method poses many complex issues such as identifiability and uncertainty of the water demand estimates. Given the sparsity of data that are likely to be available in real time, prior statistical information about water demands must be incorporated in the calibration procedure. In this paper, a method and algorithms are proposed for a real time water demand calibration process. A predictor-corrector methodology is proposed to predict statistical hydraulic behavior based on prior estimation of water demands, and then correct this prediction using new, real-time measurements. The problem is solved using the extended Kalman filter, which is a linear algorithm that calculates the estimate of water demands and their uncertainty. As part of the Kalman filter calculation, we calculate direct sensitivities of system hydraulic responses with respect to water demands. Results of numerical experiments illustrate the impacts of statistical demand variability, hydraulic measurement accuracy and sampling design on demand estimation. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The coupling between EPANET and SCADA system was developed to operate in three different modes. The real-time mode receives real time data from the SCADA database and performs an on-line analysis of the system status and response to changing conditions. The hindcasting mode can model any past event within a defined period based on the historical SCADA database and the forecasting mode can predict the system behavior. EPANET 2.0 toolkit was linked with both historical and real-time SCADA database through middleware software. This software is filtering the input data and filling the gaps, preparing initial and boundary conditions, adjusting demand balance in every time step, running EPANET simulation fragmented in time, and retrieving simulated results. The system initial conditions are retrieved from the SCADA real-time or historical database and the system boundary conditions are developed to mimic the behavior of the modeled network. System control includes the possibilities for PID (Proportional Integral Differential Control) control of control valves and pumps during an extended period analysis. The state of the art in available couplings between an EPANET based hydraulic model and a SCADA system, as well as the experience from existing implementations is discussed in this paper. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In order to provide effective early warning of a contamination event in drinking water, the technology used must produce credible and timely information that supports the decision making process of the operator. Recent testing and evaluation by the United States Environmental Protection Agency (USEPA) has shown that it is feasible to integrate technology solutions to monitor multiple parameters that define water quality. Extrapolating this concept to practical use in the field has resulted in a complete integrated water quality remote monitoring system that detects and reports to an operations center the evidence of contamination in real time on a continuous basis. In drinking water systems with a limited number of sources of water, the water quality should remain predictable, but where the system is served by multiple and diverse sources, it may be normal for the water quality in certain areas to vary daily. The value of the information gleaned from a baseline monitoring approach will be greatly enhanced by saturating (providing complete coverage of) the distribution system with water quality monitoring systems. There is no single device that can provide all the desired contamination information (chemical, biological, and radiological), so an integrated, economical, and modular "early warning" system to monitor and analyze water quality remotely has been designed. This cost-effective solution will provide complete coverage of the area of concern by designing and installing the remote monitoring system in "tiers". Tiers 1, 2, and 3 will function with the following progressive capabilities: (i) Detect and in some cases classify chemical, biological, and radiological contaminants in drinking water distribution systems; (ii) Operate in a continuous, on line fashion that provides an immediate warning directly to a decision maker; (iii) Identify where the contamination occurs within the system and track the movement of it throughout the system; and (iv) Verify the contamination event, and, potentially identify the contaminants. The system can be modified to meet each utility's need versus budget. Tier 1 provides detection and notification of a potential contamination. This module is designed for general and pervasive, monitoring. Tier 1 monitoring sites have inexpensive, but sensitive sensors that react to a broad range of contaminants, and, because of the low cost, the sensors can be deployed to cover the distribution system. A group of Tier 1 sites will make up a "domain", based on hydraulics of the system, geography, and criticality of sensitive areas. The Tier 1 sites report water quality data to the Remote Data Acquisition and Processing Unit (RDAPU) at the Tier 2 site. The RDAPU monitors and "governs" a specific domain. The Tier 2 site acquires the data from Tier 1 sites in its domain, communicates critical information to the operations center, and provides greatly enhanced detection capability and a level of verification. Tier 3 is designed to verify the contamination and identify the contaminant. Tier 3 will provide confirmation of the event and identify the contaminant within 2 hours of the initial detection. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The City of Ann Arbor, Michigan's water system provides drinking water for about 130,000 customers and operates a distribution system that provides service to about 40 square miles of the city and surrounding customer communities. Increasing concerns about the water quality within this distribution system as well as concerns about potential contamination events within the distribution system are making additional monitoring of the quality of the finished water a higher priority within the utility. To understand the issues, and develop solutions, the City's water utility has undertaken an evaluation of how to most effectively locate water quality monitoring equipment to meet these multiple goals. This paper describes the following: (1) the interest of the utility in designing a multi-objective monitoring system; (2) the use of established hydraulic/water quality modeling based techniques in the form of TEVA (Threat Ensemble Vulnerability Assessment) and PipelineNet; (3) identification of likely locations for monitors based on established site criteria (accessibility, power availability, etc.) and (4) integration of this information through facilitated workshops involving Ann Arbor staff and the consultant's project team. The paper describes an evaluation process that was undertaken using these analysis tools, how the results were interpreted and utilized, and the different hurdles that were addressed in the process. In addition, water quality parameters and instruments were reviewed and preliminary selections made to meet both project goals of assessing water quality changes and detecting potential contamination events. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Charleston Water System (Charleston Water) in Charleston, South Carolina and the American Water Works Association Research Foundation (AwwaRF) have been engaged in a tailored collaboration research project to address concerns related to inadvertent or deliberate attacks on water distribution systems involving chemical contaminants. Colorado State University (CSU) and Advanced Data Mining International are also involved in this research effort. The thesis of the research is that a combination of "field-grade" water quality monitoring sensors and intelligent software can provide reliable, relatively low cost security monitoring for water distribution systems. The types of sensors being evaluated are already widely used to measure water chemistry properties — conductivity, pH, chlorine residual, turbidity, total organic carbon, and UV-254 absorption. The intelligent software is called a Decision Support System (DSS), which employs an artificial intelligence technology called Case-Based Reasoning. The DSS will automatically learn and retain "normal" patterns of water quality behavior; identify unprecedented patterns (outliers) that indicate a potential problem; generate alarms; and make recommendations to operators about what to do. Different sensor arrays composed of multi-parameter devices and individual sensors have been installed at three sites in Charleston Water's distribution system with varying degrees of operational success. A problem that has affected the reliability of the sensors is pressure surges caused by pumping. Similarly, additional sensors such as those deployed by Charleston Water were installed in an experimental pilot loop operated by CSU in which the dynamic behaviors of water chemistry variables could be measured. Seven organic and inorganic chemical contaminants were tested in the pilot loop, each at increasing concentrations up to their LD50s. The contaminants were sodium arsenate, sodium cyanide, sodium fluoroacetate, nicotine, cycloheximide, dicrotophos, and aldicarb. This paper will present intermediate findings that describe practical knowledge acquired in the deployment of the various field-grade sensors in Charleston Water's distribution system, results from the experimental studies conducted in the pilot loop at CSU, and details of the DSS presently under development. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In recent years, research in wireless sensor networks (WSN) has been undergoing a quiet revolution, promising to have significant impact on a broad range of applications relating to environmental monitoring, structural health monitoring, security and water safety. The convergence of the Internet, telecommunications, and novel information technologies with techniques for miniaturisation now provides vast opportunities for the development and application of low-cost monitoring solutions which could drastically increase the spatial and temporal resolution of environmental data. The paper describes the development of a prototype monitoring system which bridges advances in wireless sensor networks with advances in hydraulic and water quality modeling. The prototype monitoring system was deployed at Boston Water and Sewer Commission (BWSC) in December 2004, and it has been successfully collecting and charting near-real time hydraulic and water quality data as well as data from combined sewer outflows (CSO). The remote monitoring system has unique functionalities in terms of sampling rates (up to 1000 S/s), time synchronization (up to 1 ms) and in-network processing. These features create novel opportunities for wirelessly collecting data for applications such as hydraulic pressure transients, remote acoustic leak detection together with low-duty cycle applications such as monitoring water quality parameters and water levels in CSOs. The trial with BWSC has been tremendously useful to prototype hardware and software tools, and to identify deployment and operational challenges in using sensor networks for monitoring and management of large scale water supply systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The objective of this paper is to present an optimal sensor placement methodology to assist in the effective and efficient detection of accidental and/or intentional contaminant intrusion(s) in water distribution systems. The work presented here is done in response to call for papers for the Battle of the Water Sensors Networks (BWSN), at the Water Distribution Systems Analysis Symposium (2006). The above problem is formulated and solved as a constrained multiobjective optimisation problem. The four objectives are: (1) minimisation of the expected time of detection, (2) minimisation of the expected population affected prior to detection, (3) minimisation of the expected demand of contaminated water prior to detection and (4) maximisation of the detection likelihood. The constraint modelled is the pre-specified number of detection sensors used in the sampling design. Decision variables are the sensor network locations. The solution methodology proposed is based on the novel Noisy Cross-Entropy Sensor Locator (nCESL) algorithm. This algorithm is applied to the two competition networks under four base contamination scenarios (A, B, C and D) and two different numbers of sensors available (5 and 20). The results obtained demonstrate the effectiveness and efficiency of the sensor placement methodology proposed. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

There is interest in optimally locating water quality sensors in a water distribution system due to possible terrorist injections of a contaminant. This paper is part of a contest for the 8^th Water Distribution System Analysis Symposium entitled "Battle of the Water Sensor Networks" (BWSN), which asks contestants to propose sensor designs for two networks and two numbers of sensors. The method of sensor location selection described here is a demand-based approach in which sensors are located at the junctions with the highest demands. The rationale behind this method is that an injection of a contaminant will be advected by the flow, which goes toward the high-demand locations. Therefore, demand is a surrogate for contaminant concentration. In addition, the mass of contaminant released from the system is a function of demand. The 5-sensor and 20-sensor designs are presented in the paper. Designs were identical for all of the contamination scenarios contained in the contest, thereby demonstrating the simplicity and uniformity of the sensor location designs. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In this entry for the "Battle of the Water Sensor Networks (BWSN)," the authors develop a closed-loop algorithmic process for the optimal design of water sensor placement in water-distribution systems. The proposed solution, the simulation-optimization methodology, focuses on the relation between the input and output of the water-distribution system and not on the topological structure of the system. The proposed model is based on a single objective function approach as opposed to a multi-objective case. However, unlike conventional single objective models, the proposed objective function incorporates multiple factors such as time of detection, contaminated water volume, population affected, and reliability of the optimal system—in this sense it mimics a multi-objective approach. An improved genetic algorithm is proposed for the solution of the model. The algorithm works on a subset of the complete set of junctions present in the system (junction subdomain) and the final solution is obtained through the evolution of subdomains. The proposed algorithm is applied to two test networks submitted by the BWSN committee. The results indicate that the proposed model is effective in solving this problem. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A metaheuristic approach is proposed to design the optimal placement of monitoring stations, for aiming an early detection of the intentional water distribution networks contamination. The approach is based on the use of a predator-prey model, that is applied to multi-objective optimization. The proposed algorithm is used to solve the sensor constrained contamination detection problem, which is polynomially equivalent to the asymmetric k-center problem, so it is NP-Hard In particular the predator-prey model is applied to find the optimal sensors placement evaluated, according to the four design objectives which are described in the Battle of Water Sensor Networks (BWSN) manifesto. Both predators and preys are subjected to an evolution process. The competing coevolution approach has been chosen to avoid the problem of designing the fitness function or, in other words, to avoid the problem of locating the most representative contamination events. The candidate solutions and tests, which are used to evaluate these solutions, evolve simultaneously to find the optimal evaluation set, and as a consequence to minimize the number of needed checks, during the selection of the optimal solution. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Placement of water quality sensor has received an increasing concern/or timely providing the warning of possible contamination in a water system. Due to the large dimension of water distribution network and the difficulty for predicting where a contamination event occurs, it is a great challenge for engineers to come up with good sensor locations with any confidence to effectively detect possible contamination events. The problem is complicated by the fact that sensor location is evaluated against a number of objective criteria that may include the detection likelihood, the expected detection time, affected population and contaminated water consumption. A design that improves one objective may deteriorate another. In this paper, sensor placement is formulated as a multi objective optimization problem that is solved by using a competent genetic algorithm while the contamination events are simulated by the latest development of Monte Carlo method. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water utilities are challenged to meet increasingly stringent regulatory requirements for potable water quality with the additional burden of maintaining consumer confidence in the face of potential threats to the security of their water distribution systems. Raising the performance bar from "in compliance" to an appropriately proactive protection against contamination by accident or malevolent acts involves deployment of more sophisticated monitoring technologies and knowledge of what to do when the warning alarm sounds. These technologies are relatively expensive now. Until greater confidence can be placed in these technologies and related analytical tools, utilities should leverage existing knowledge of their systems and customer needs, work within budgetary and other constraints, and demonstrate "due diligence" in protecting public health. Industry and government-financed research programs are working toward development of cost-effective technologies and analysis methodologies. In the short term, water utilities can take a common sense approach to significantly improve their ability to detect contamination, quickly respond, and mitigate the consequences. It may also be possible with this approach, to some extent, for water utilities to warn their customers of imminent exposure to contaminated water. The "strawman" common sense approach offered may enable water system operators to obtain support from stakeholders in moving cost effectively from a compliance posture toward a protective capability. As recently suggested in AWWA's guidance to water utilities, water system operators must consider whether they will "Detect to treat, to warn, or to protect". The results of previous research on monitoring technologies and analytical methodologies are considered in this paper together with practicalities faced by water systems. Aside from "due diligence" in protecting public health, budgetary constraints and organizational considerations, key drivers for elements of the "strawman" approach offered include: 1) Defining "normal" distribution system water quality, 2) Monitoring to achieve compliance with the Stage 2 D/DBP Rule among others, 3) Managing the distribution system to achieve optimal water quality, 4) Detecting contamination in a timely manner with a high degree of confidence, and 5) Responding operationally for prompt and effective mitigation of contamination consequences. The "strawman" approach considers population distribution, system pressure and flow patterns (derived from operating data), critical customer locations, design basis threat evaluation, system facilities, site installation characteristics (ownership, access, security, availability of utilities, etc.), cost, and organizational capability for sensor maintenance and contamination event response as important factors for planning sensor deployment. The approach is offered as a baseline for comparison with sensor site selections derived from more sophisticated approaches. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents a multiobjective model to optimal sensor design in water distribution systems as part of the battle of the water sensors networks (BWSN). Previous work on optimal sensor design for water distribution systems (WDS) focused on one objective (e.g., maximizing the detection likelihood of contamination events). In this study the Non-Dominated Sorted Genetic Algorithm-II (NSGA-II) is implemented to tradeoff the following four conflicting objectives: (1) maximizing the detection likelihood; (2) minimizing the detection time; (3) maximizing the detection instrumentation redundancy; and (4) maximizing the contamination source identification likelihood (i.e., the likelihood to provide a unique solution to the inverse problem of contamination source identification for a given layout of sensors). The effectiveness of the multiobjective approach is demonstrated through using the two BWSN network examples, where Pareto fronts are plotted for each two objectives; for each three; and finally for all four. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This study presents a multiobjective solution approach to the Battle of the Water Sensor Networks (BWSN) initiative. The developed methodology tailors the algorithm of Ostfeld and Salomons for optimally placing sensors in a water distribution system with the NSGA-II multiobjective genetic algorithm of Deb et al. Pareto optimal fronts are shown and discussed for the two BWSN Networks, for selected BWSN cases. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Drinking water distribution networks represent complex systems. Water flow rates in a water distribution system vary with time, with periodic features that reflect temporal variations in water demand by consumers. The intentional introduction of a contaminant disrupts the system and could theoretically be detected by a sensor or network of sensors placed at nodes (pipe junctions, reservoirs, storage tanks, or even individual consumer taps) in the system. Determining the best locations for placement of these sensors represents a significant research question, because the system has multiple states, the number of possible intrusion points is large, and the likely high cost of these sensors limits the number that can realistically be deployed. The optimal placement of these sensors to minimize the effect of an introduced contaminant on the population is a critical issue. Sensor placement for intrusion detection exhibits an important diminishing returns property: adding a sensor to a sensor network improves the detection ability less than adding it to a subset of the sensor network. We prove that this submodularity property holds for the objective functions that we consider for placing sensors, and exploit it by applying algorithms for maximizing monotonic submodular functions. Unlike existing optimization algorithms for selecting sensor placements, our efficient optimization procedure has strong theoretical performance guarantees. In spite of the problem's complexity, our algorithm is guaranteed to always find a solution that is at least within 63% of the optimum, and will often find a (near-) optimal solution. This method is applied to two hypothetical distribution systems (129 nodes and 12,527 nodes) to determine optimal sensor placements for a sensor network of 5 or 20 sensors. Optimization was based on multiple criteria including: (1) minimizing time to detection, (2) minimizing population affected prior to detection, (3) minimizing expected demand for contaminated water prior to detection, and (4) maximizing detection likelihood. A base scenario and three derivative scenarios were used to test the sensor location optimization for the hypothetical systems. In order to compute accurately the objective criteria, we exhaustively simulated all possible attack scenarios, using distributed computation. Five optimization objective functions were considered (i.e., optimization on each of the four objectives independently and then an equally weighted multi-objective optimization). The two networks analyzed in this project illustrate how a sensor network of 20 sensors is more than "adequate" for the example distribution system of 129 nodes, while a much larger sensor network would be needed for "adequate" detection in the example large network of 12,527 nodes. The developed algorithms generalize to networks of arbitrary size and can be constrained by expert knowledge or rankings of scenario likelihood. Further, the optimization algorithms have potential applications for placement of sensors in other complex, dynamic systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

There is interest in optimally locating water quality sensors in a water distribution system due to possible terrorist injections of a contaminant. This paper is part of a contest for the 8^th Water Distribution System Analysis Symposium entitled "Battle of the Water Sensor Networks" (BWSN), which asks contestants to propose sensor designs for two networks and two numbers of sensors. The method of sensor location selection described here is a mass-based approach in which sensors are located at the junctions with the greatest mass released from the junctions. The rationale behind this method is that the mass of contaminant, in contrast to other variables, directly causes illness. The 5-sensor and 20-sensor designs are presented in the paper. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The general sensor placement problem (SPP) for contaminant warning system (CWS) design involves placement of a limited number of sensors such that the expected impact of an attack is minimized. We cast the SPP in terms of the well-known p-median problem from discrete location theory. The p-median formulation assumes a fixed number of attack scenarios, each specifying a probability of occurrence, a set of injection sites, injection strengths, and injection durations. The impact of each potential attack is determined via contaminant transport simulation. Specifically, EPANET is used to generate a time-series of contaminant concentration at each network junction for each attack. For each combination of attack and network junction, the resulting time-series are then used to compute the network-wide impact of the attack assuming detection via a hypothetical contaminant sensor placed at the network junction. In conjunction with the attack probabilities, the resulting "impact coefficients" completely specify the input to our p-median formulation of the SPP. We solve the p-median SPP using domain-specific heuristics based on a combination of GRASP and local search. Extensive computational tests indicate that the heuristics consistently locate globally optimal solutions, as verified via solution of the corresponding mixed-integer program (MIP) using commercially available solvers. Further, the heuristic executes in seconds to minutes for networks ranging from 100 to 10,000 junctions, respectively, subject to large numbers (on the order of the number of network junctions) of hypothetical attack scenarios. By isolating objective-specific information to the impact coefficients, our approach seamlessly allows for optimization of disparate performance objectives, e.g., detection likelihood and population exposed. The use of general-purpose contaminant transport simulators allows us to handle arbitrarily complex attack scenarios, e.g., multiple simultaneous injection sites with different contaminants at variable injection strengths and durations. Certain response strategies, such as delays incurred due to manual verification of sensor hits, can be incorporated via appropriate modification of the impact coefficients. Through the use of side constraints,we are able to generate solutions to then p-median formulation of the SPP that simultaneously yield high-quality performance with respect to a range of performance objectives. Finally, in contrast to competing approaches to solving the SPP, we are able to provide provable bounds on the quality of solutions generated by our heuristic via formulation and solution of the SPP as a p-median MIP. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A mixed-integer linear program is used to design the sensor location problem to detect potentially harmful contaminations in drinking water distribution systems. The size of data report time step is investigated as a mean to simplify problem formulation and solution identification. The uncertainty on the cost function value due to the size of the report time step can be explicitly calculated. This allows the designer (1) to estimate the theoretical cost function lower bound and thus to decide if a design is acceptable and (2) to establish a rational criteria for determining a meaningful size of the contamination event ensemble. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

As water distribution systems are vulnerable to a variety of accidental or deliberate contaminant intrusion events, efficient in-situ water quality monitoring is important in providing a robust water supply. To identify optimal placements of monitoring sensors in water distribution systems, a multiple-objective optimization method employing genetic algorithms (GA) in conjunction with data mining, is developed. The proposed methodology is capable of identifying an optimal set of monitoring stations based on three objectives: detection delay time, detection probability, and the affected population prior to detection. To apply the method, a database which stores data for intrusion events at each node, and the classified consequences of these intrusions at each node, is prepared. The initial solutions for multi-objective optimization are obtained from the database based on sensor coverage criteria. Pareto ranking is performed during the GA optimization. The effectiveness of the proposed method is illustrated by applying the methodology to the two networks, Networks 1 and 2, provided by the Battle of the Water Sensor Networks design competition. The final results in application to Networks 1 and 2 are also provided. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper investigates the issue of finding the best sensor locations in a drinking water distribution network for detecting harmful substances. The problem is formulated in a multi-objective optimization framework with five performance measures: time of detection, population affected prior to detection, demand of contaminated water, detection likelihood and demand coverage. In practical application, due to the large size of water distribution networks, the space of possible solutions expands dramatically, making it difficult or impossible to determine the optimal solutions. We propose the "Iterative Deepening of Pareto Solutions" search algorithm, for locating "good enough" solutions. The algorithm solves the problem by iteratively choosing the best non-dominant solutions, and expanding them by increasing the depth of the search tree until all the sensors have been used. Simulation experiments were performed on two water distribution networks, following the formulation defined in the "Battle of the Water Sensor Networks" design challenge (Ostfeld et al. 2005). Four contamination scenarios are considered and from the sets of possible solutions, the most appropriate designs are proposed. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A US pilot study revealed that regular analysis of customer feedback data has the potential to be an effective drinking water contamination event monitoring technique. The pilot study involved onsite interviews and data analyses at 13 water utilities that served 30,000 to >8,000,000 customers. A literature review and regional survey of 16 utilities that served 1,400 to >1,000,000 were also conducted. Project results revealed that customer feedback has been used to uncover >56 incidents (e.g., water main breaks, treatment failures, contamination events, and intrusion attempts). Utilities visited used between 10–197 different feedback categories, with appearance, illness, organoleptic, suspicious activity, and flow being the most useful. No uniform water industry standard exists for coding and handling drinking water customer feedback. Of the utilities visited, 83% used customer call centers, 50% of the call centers directly transferred the customer to another utility department without entering data, and 50% of utilities stored customer feedback data in a database. Employee experience (100%) determined whether or not a customer concern required immediate attention (n=13). Analyses of utility consumer feedback data tended to be monthly which could not detect an immediate problem. Few utilities visited (6/13) and surveyed (1/16) used charts or graphs. Analyses of one utility's records revealed that the number of customer calls varied based on time of year, month, and day of week. The amount of historical customer feedback data stored at each utility varied from 1–4 years. Operations and Maintenance (O&M) and Water Quality (WQ) departments were the most likely to be forwarded customers who had water system operation, drinking water quality, and health concerns. Onsite actions conducted by O&M personnel would include flushing of nearby hydrants, while WQ representatives would typically interview the customer, enter the customer's building, and collect and test water samples. 92% of the utilities visited dispatched only one representative to the customer's building. Onsite response time depended on the type of concern, time it was received, day of week, personnel workload, and resources. 60% of the utilities permitted employee entry into the customer building. Disinfectant residual (free and combined chlorine) was the most common onsite analyses parameter and significant changes in it were considered to indicate a problem and/or the presence of a contaminant. Drinking water illness complaints were considered to be unfounded if disinfectant residual and pH results were within acceptable limits and bacteria test results were negative. Before customer feedback monitoring can be effectively implemented, utilities must optimize information flow, communication, storage, investigative procedures, and consistent terminology. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A team led by Michael Baker Jr., Inc. is conducting a study of the state-of-the-art in water security for the U.S. Department of Homeland Security, Homeland Security Advanced Research Projects Agency, in coordination with EPA. The Baker Team will is developing a "roadmap" to help guide DHS's future R&D investments in infrastructure security for the water sector in the short and mid-term (1–5 years). Among several tasks, the project is examining intentional contamination with chemical or biological agents and long-lived, highly active radiological isotopes, as applied to post treatment distribution component of water systems. For the Symposium, the team will present initial findings in each of these areas. Chemical, biological, and radiological agents have been prioritized, considering whether they can be hydrolyzed in water, their availability, required quantities, and whether they can be transported and introduced before they kill the perpetrator. The ultimate goal of the project is to identify technical and operational gaps in our ability to prevent, respond, and recover from attacks. The preliminary set of R&D recommendations will be presented. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

With emerging security issues, drinking water utilities are facing new challenges. New security strategies and emergency response plans should be implemented to face a range of possible threats. While not the most likely, drinking water contamination may be the most worrying in terms of public health, socio-economic and psychological impacts. In many previous studies, hydraulic and water quality simulation tools have been used to address security issues related to contamination. However, to our knowledge, no studies have yet focused on the elaboration of a strategy defining the field operations to implement, after contamination has been detected, to protect public health. This paper presents ongoing work defining an operational strategy based on optimizing a sequence of field operations. Three main objectives were defined in the context of this research project: 1) minimize the risk that contaminated water is consumed; 2) identify the valves to be closed to safely contain the contaminated water and proceed, as quickly as possible, to isolation operations; 3) define a set of operations to efficiently flush contaminated water from the network to quickly and safely return to a normal operation. It is assumed that a utility has installed a Contamination Early Warning System (CEWS) to physically secure the distribution network, and that a reliable public notification system is in place as well. A simplified version of a static sensor placement optimization model is used to locate contaminant detectors. Depending on which detector gives the first alarm, a potentially contaminated zone can be delineated and the spatial expansion of this zone can be traced through time. A heuristic algorithm based on a set of pragmatic, operational and safety rules to isolate contaminated zones is introduced. Considering simultaneous field manipulations and no limit on the number of response teams (2 persons, 1 specialized vehicle), a solution is sought to ensure that every operation required for isolation can be executed before contamination reaches isolation valves. An application example is presented for the small network of Valcourt (Quebec, Canada). Although the current work is based on simplifying assumptions regarding hydraulic and contaminant transport simulation, our isolation algorithm remains general and straightforward enough to be implemented under various modeling schemes. Future works will address the operational issues related to flushing contaminated water from previously isolated zones. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A system designed to address the problem of distribution system monitoring is described here. The developed system employs an array of common analytical instrumentation, such as pH and chlorine monitors, coupled with advanced interpretive algorithms housed in an event monitor to provide detection/identification-response networks that are capable of enhancing system security and quality. A variety of real world venues and testing protocols were used to verify the efficacy of the system. Deployed systems are shown to demonstrate the capability of learning base line in a rapid timeframe while being capable of detecting and characterizing system anomalies related to security and basic water quality operations. Included are data generated from several real world events including caustic overfeeds, rain events, street work and major line breaks among others. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In 2000 Essex and Suffolk Water (a UK water services company located east of London) invited tenders for the construction of `All Mains' models for their water distribution network. A key requirement of the project was sustainability (due to the costs associated with field tests), the models had to be updated when new GIS, CIS and telemetry data became available. Additional requirements included calibration targets, the use of a set of standard demand profiles based on SIC codes and the support of the Safege Piccolo modelling package. Crowder Consulting were chosen to undertake this project by developing a new modelling module for our existing management information application NETBASE. The outcome was a model management system that could link to any GIS, CIS and Telemetry system, allow the models to be built, edited and validated. The models could then be exported to any modelling application (currently EPANET and Piccolo are supported) for analysis and then loaded back into the NETBASE database to analyse the model results. Rather than developing a custom data warehouse, an Oracle Database was used enabling the development team to concentrate their efforts on the analysis and interface tools. The data sources for the project were the clients GIS (Geographical Information System) system, CIS (Customer Information System) system and SCADA (Supervisory Control and Data Acquisition) systems. To create the flow and demand profiles a Profiles Library was developed. The profiles library enabled the profile creation process to be automated so that it was possible to quickly and easily create demand profiles for individual users and monitoring points. Demand profiles were created using the customers annual consumption, a standard profile based on the customers SIC code and seasonal factors. Profiles were created for peak, summer, winter and typical demands. Custom profiles were created for users with unusual demand patterns using logged data captured as part of the field test. Model calibration was also automated using a RMS goal-seeking algorithm to automatically roughen groups of pipes until a calibration target was reached. Although not all the models could be calibrated using this method it provide an effective first pass solution. When calibration was completed it was validated (by the modelling team) and a report generated for the client. Models were calibrated on a sector by sector basis and then finally combined into a single model for a demand area. Sustainability was achieved by using the database as a buffer between the source data and the models and developing a set of import applications that enabled the incoming data to be `quarantined' while it was validated before the model was updated. This proved particularly effective at safeguarding the changes that had been made to the system to model features such as pumping stations and reservoirs (typically not included in GIS systems in the UK) as is was possible to load new GIS data around features while retaining changes. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Modern modeling software supplies the tools to help utility engineers design, simulate, manage, optimize, and operate every aspects of a water distribution system. Countless hydraulic models have been developed to address water distribution problems ranging from the network extension or rehabilitation through unaccounted-for-water and unidirectional flushing analysis to pump-cost or leakage optimization. It comes as a surprise to see how rarely these models have been maintained and utilized to address other problems in the water distribution system. Is it possible to break out of this cycle of creating models from scratch for every new issue that needs to be resolved? Can water utilities move forward to sustainable modeling so an up-to-date model is always available to support operations? The recent developments in the modeling process make sustainable modeling a practical proposition; the utilities should expect to see an increased return on investment (ROI) in modeling. Until recently, building a hydraulic model was a complex and time-consuming process; maintaining a model was even harder. Now, with modeling software that provides advanced model building and maintenance features, this is no longer the case. A sustainable model is a model developed with continuous use in mind, used for a variety of problems encountered by the utility. It reflects the current state of the water distribution system, in terms of hydraulic network, system operation, and boundary conditions. In order to maintain the model a utility needs the following tools: 1) automated network updates, 2) integration with the real-time data, and 3) multi-user environment. The automated network update is performed using persistent links to GIS or other database systems. To perform the update, one needs persistent connection configurations, key mapping between GIS and model elements, and import scripts. The update can be initiated by the user or scheduled to occur in regular intervals. Integration with real-time data (SCADA) ensures that the initial conditions and operational settings of the model correspond to the current state of the system. An automatic SCADA update should provide: 1) model initial conditions state (e.g. tank levels), 2) model controls (e.g. pump settings), 3) demand and model calibration, and 4) model validation Finally, different players should be able to use the sustainable model through a multi-user environment with access rights, a log of model changes, and the data source identification. Building a hydraulic model requires expenditure of both time and money; building a sustainable model does not greatly increase the development costs and significantly increase the ROI. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

For many years, modeling and GIS applications have been used as separate systems for planning, design and operation of water distribution systems. As digital information becomes increasingly available, the potential benefits of an integrated model and GIS system become more obvious. These benefits of integration and resulting automation include: better service to internal and external customers, improved work processes and higher productivity, reduced data entry errors and duplication, improved data security and reliability, and improved emergency response. Las Vegas Valley Water District (LVVWD) is developing an integrated system that includes a hydraulic and water quality model. Geographic Information System (GIS), Supervisory Control and Data Acquisition (SCADA), and other database systems. Model data is dynamically linked to the GIS information with unique identification features. The model utilizes GIS and other linked data to compute hydraulic and water quality results for master planning, operational planning, facility outage scenarios, and other what-if scenarios. The integrated system enables the data process to be automated to the maximum extent possible so that manpower and time is saved. This paper describes the significance of such integration and demonstrates its value with example applications. The examples encompass master planning, water quality planning and operational planning for rapid-growing communities. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Techniques to build models of water distribution networks have advanced significantly over the last years. EPANET has been the standard to identify the hydraulic components that usually compose a model, its properties, as well as all those additional operation data that allow to configure an extended period simulation model. Data used to be entered through a graphic interface by means of some property dialogs. However, GIS are becoming an important tool to manage water distribution systems and their connection to other applications oriented to configure the network model has supposed a remarkable advance, mainly due to the reduction in the effort to accomplish it allowing to build detailed models with a large number of nodes. Nonetheless, models obtained from a GIS still need to be debugged and completed in order to make up the whole data set of a simulation model. Obtaining the whole model directly from the different databases that compose the information system of a water utility company would suppose an important milestone. Namely, the corporative GIS as the physical network component manager, the customer information database as the demand and registered consumption support system, the operation and maintenance database as the support system of all daily changes and the SCADA as the support tool of the evolution of the control variables that regulate the system. All of them are necessary to assemble the model. The database presented on this paper enables to collect all the aforementioned information in a unique data structure oriented to build an accurate model to reflect the real network behaviour for any time period selected by the user by means of a simple query. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Accidental drinking water contamination has long been and remains a major threat to water security throughout the world. Consequently, contamination source identification is an important and difficult problem in the managing safety in water distribution systems. This problem involves the characterization of the contaminant source based on observations that are streaming from a set of sensors in the distribution network. Since contamination spread in a water distribution network is relatively quick and unpredictable, rapid identification of the source location and related characteristics is important to take contaminant control and containment actions. As the contaminant event unfolds, the streaming data could be processed over time to adoptively estimate the source characteristics. This provides an estimate of the source characteristics at any time after a contamination event is detected, and this estimate is continually updated as new observations become available. We pose and solve this problem using a dynamic optimization procedure that could potentially provide a real-time response. As time progresses, additional data is observed at a set of sensors, changing the vector of observations that should be predicted. Thus, the prediction error function, is updated dynamically, changing the objective function in the optimization model. We investigate a new multi population-based search using an evolutionary algorithm (EA) that at any time represents the solution state that best matches the available observations. The set of populations migrates to represent updated solution states as new observations are added over time. At the initial detection period, non-uniqueness is inherent in the source-identification due to inadequate information, and, consequently, several solutions may predict similarly well. To address non-uniqueness at the initial stages of the search and prevent premature convergence of the EA to an incorrect solution, the multiple populations in the proposed methodology are designed to maintain a set of alternative solutions representing different non-unique solutions. As more observations are added, the EA solutions not only migrate to better solution states, but also reduce the number of solutions as the degree of non-uniqueness diminishes. This new dynamic optimization algorithm adoptively converges to the best solution(s) to match the observations available at any time. The new method will be demonstrated for a contamination source identification problem in an illustrative water distribution network. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The development of water quality sensors and sensor technologies makes it feasible to establish water quality monitoring networks in drinking water distribution systems. The collected water quality data can be used to determine the location and time of the source contamination in the case of terrorist attack or accidental contamination intrusion. If the contaminant reaction can be modeled reliably within the pipe network and the sensors can measure water quality quantitatively, minimization of the difference between modeled and measured water quality is one approach to solution of the contaminant source determination problem. The underling mathematical problem is, however, inherently ill-posed, due to the shortage of measurements compared to source parameters, and regularization methods are required to force identification of a unique solution. Further, it is usually the case that the contaminant reaction dynamics are unknown, and/or the sensor can only detect the presence or absence of the contaminant and not the quantitative concentration. Even if the source determination problem can be formulated mathematically and optimization algorithms can be applied to solve the problem, the decision variable dimension can be huge since contamination can happen anywhere and anytime in the network. An alternative practical method is developed in this paper to identify all possible locations and times that explain contamination incidents detected by the water quality sensors. It is assumed that contaminant injections occur at network junctions and over discrete time intervals. The method only requires the positive/negative sensor status over time, and knowledge of network hydraulics. A particle backtracking algorithm is used to identify the water flow paths leading to each sensor measurement and the travel time from the junctions along the flow paths to the measurement. Those locations and times that are connected to positive sensor measurements — and are not connected to negative measurements — are the possible sources, assuming no false positive/negative readings and an accurate hydraulic model. This method can also be used as a pruning step for solving the source identification problem using optimization algorithms, as it reduces the number of decision variables by eliminating locations and times that are inconsistent with the sensor responses. The method also forms the basis for incorporating important concerns about hydraulic and sensor uncertainty, which are likely to enlarge the set of possible sources. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Being the quality of drinking water a great concern in many countries, many researches have recently focused their attention to improve drinking water security threatened by deliberate or accidental intrusions in water distribution systems. In this contest, methods for detecting and controlling contaminations are required. If a contamination is detected by monitoring stations installed on the network or directly by users, the individuation of the pollutant source location is very important to stop the event as soon as possible and propose a control strategy. In this work is used a methodology to identify source location of an accidental pollution in a municipal water distribution network, using time-varying concentration measurements. Since source identification is an inverse problem, it may be ill-posed and then its results may depend strongly on the input data quality. This aspect can be crucial in a water quality inverse problem, in which, among the input data, nodal demands uncertainty and errors in concentration measurements determine a high level of uncertainty. In order to verify the robustness of the methodology respect to water demand variability and measurement errors, an analysis based on a Monte Carlo procedure is performed. The methodology effectiveness is demonstrated through an application to a midsize sample network and considering different levels of uncertainty. The results show a good frequency of identification of the right pollution source node also at high uncertainty levels. It is also observed that results depend on number and the location of water quality measurements. The maximum coverage criterion appears a good method for selecting measurement location. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The source of contamination in a water distribution system may be identified through a simulation-optimization approach. The optimization method searches for the contaminant source characteristics by iteratively estimating the contaminant plume concentrations until they match observations at sensors. The amount of information available for characterizing the source depends on the number and spatial locations of the sensors, as well as on the temporally varying stream of sensed data. The accuracy of the source characterization depends on the amount of observations available. A major factor affecting, this accuracy is the degree of non-uniqueness present in the problem, which may cause misidentification of the source characteristics. As more sensors are added to the network, the non-uniqueness may be reduced and a unique solution may be identified. Thus, a key consideration when solving these problems is to assess whether the solution identified is unique, and if not, what other possible solutions are present. A systematic search for a set of alternatives that are maximally different in solution characteristics can be used to address and quantify non-uniqueness. For example, if the most different set of solutions that are identified by a search procedure are very similar, then that solution will be considered as the unique solution with a higher degree of certainty. Alternatively, identification of a set of maximally different solutions that vary widely in solution characteristics will indicate that non-uniqueness is present in the problem, and the range of solutions can be used as a general representation of the amount of non-uniqueness. This paper investigates the use of evolutionary algorithm (EA)-based alternatives generation procedures to quantify and address non-uniqueness present in a contaminant source identification problem for a water distribution network. As additional sensors may decrease the amount of non-uniqueness, several sensor configurations will be tested to investigate and quantify the improvement in uniqueness as more information is used in the source characterization. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents a methodology for optimally allocating sensors for solving the contamination source identification problem in water distribution systems (i.e., finding the optimal layout of a given number of sensors which maximizes the likelihood of contamination source identification). The model is comprised of two stages: at the first stage the water network is divided into influence zones based on the network configuration and hydraulics. Thereafter, all possible combinations of placing the given number of sensors at the different influence zones are tested for their ability to identify contamination sources for a set of pollution events (i.e., a set of contamination injections from different parts of the network, with different injection mass, duration, and starting times). A genetic algorithm framework is used for the contamination source identification. The GA is coupled with EPANET and applied to disclose pollution event characteristics (i.e., injection time, duration, and concentration) using the sensors measured data. The GA fitness function is of a least square type measuring the Euclidean distance between computed and measured concentrations at the sensor locations. The genetic algorithm decision variables (i.e., each genetic algorithm string) incorporate: (1) the contaminant injection node; (2) the injection mass rate; (3) the injection starting time; and (4) the injection duration. The global minimum for a least square minimization problem is known (i.e., zero) and thus when obtained, the contamination source is identified. At the second stage the combination that maximizes the contamination source identification likelihood is used as the search space at which only nodes from this combination can be selected as possible sensors locations. The result of the two search processes is the optimal sensors layout that maximizes the contamination source identification likelihood. The effectiveness of the method is demonstrated through two example applications. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A general method is proposed to identify appropriate network locations for placing water quality monitoring sensors to identify the sources of accidental or deliberate contaminations. First, the contamination source identification problem is discussed for an assigned set of monitored locations. If the contaminant is conservative or its reaction rate in water is first order, the problem can be expressed as a general under-determined linear system of equations. The set of infinite solutions can be found as the summation of a particular solution and the product of the null-space matrix with an arbitrary parameter vector. Optimal sensor locations are then determined by minimizing the dimension of the null-space for a statistical significant ensemble of contamination events. Simulated annealing is employed to identify these locations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

We consider the problem of optimally placing water quality sensors in municipal water networks under the assumption that sensors may fail. We give a non-linear formulation of the problem, then a linearization of this formulation in the form of a mixed-integer program (MIP). We explore the scalability limits of this formulation, then use it as a bounding procedure for a local search heuristic that optimizes the same objective: minimizing the expected impact of a contamination event. This heuristic can find optimal or near-optimal solutions on networks with over ten thousand junctions. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The ability to detect deliberate or accidental contamination of a water distribution system is of real concern to the safety and security of our nation's drinking water. To address these concerns, increased attention has been placed on sophisticated monitoring of water distribution systems and the use of robust statistical analysis. Using existing data from in-situ water quality sensors, this paper explores the ability to detect anomalies in water quality using multivariate techniques. The algorithm developed in this study uses a multivariate distance measure between the current water quality measurement and the closest observation in multivariate space within a moving window of previous observations. To discriminate between normal and anomalous water quality, the distance measure is compared to a constant threshold. To test the algorithm, we utilize both simulated anomalous events and laboratory based events that correspond to real contaminants. These events are superimposed onto in-situ water quality recorded at four different locations within a single utility network. Measured water quality parameters include free chlorine, pH, temperature and electrical conductivity. Robust discrimination methods have a high probability of detecting anomalies with a low false alarm rate. Here, receiver operating characteristic (ROC) curves are used to test the ability of the multivariate classification algorithm to detect anomalous water quality while keeping false alarms low. This analysis explores the false alarm rate associated with detecting a range of anomalous water quality observations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The recent heightened awareness of the possibility of terrorist attacks on municipal water systems has prompted the development of the necessary hardware and software for implementing a Contamination Warning System (CWS) to detect if an attack has been made in a water distribution system. Much progress has been made in this area, including the development of improved water quality sensors, algorithms for placing the sensors most effectively in a water system, and algorithms for analyzing the data that comes from these sensors. This work focuses on this last component of analyzing sensor data. Several algorithms have been developed to tackle this problem. However, the various methods have never been tested by an independent party or compared to each other. This paper describes the preliminary work that has been done to design and implement a fair and thorough evaluation of algorithms to determine which is most effective in detecting a water contamination event. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Rapid detection of anomalous operating conditions within a water distribution network is desirable for the protection of the network against both accidental and malevolent contamination events. In the absence of a suite of in-situ, real-time sensors that can accurately identify a wide range of contaminants, we focus on detecting changes in water quality through analysis of existing data streams from in-situ water quality sensors. Three different change detection algorithms are tested: time series increments, linear filter and multivariate distance. Each of these three algorithms uses previous observations of the water quality to predict future water quality values. Large deviations between the predicted or previously measured values and observed values at future times indicate a change in the expected water quality. The definition of what constitutes a large deviation is quantified by a threshold value applied to the observed differences. Both simulated time series of water quality as well as measured chlorine residual values from two different locations within a distribution network are used as the background water quality values. The simulated time series are created specifically to challenge the change detection algorithms with bimodally distributed water quality values having a square wave and sin wave time series, with and without correlated noise. Additionally, a simulated time series resembling observed water quality time series is created with different levels of variability. The algorithms are tested in two different ways. First, background water quality without any anomalous events are used to test the ability of each algorithm to identify the water quality value at the next time step. Summary statistics on the prediction errors as well as the number of false positive detections quantify the ability of each algorithm to predict the background water quality. The performance of the algorithms with respect to limiting false positives is also compared against a simpler "set point" approach to detecting water quality changes. The second mode of testing employs events in the form of square waves superimposed on top of modeled/measured background water quality data. Three different event strengths are examined and the event detection capabilities of each algorithm are evaluated through the use of receiver operating characteristic (ROC) curves. The area under the ROC curve provides a quantitative basis of comparison across the three algorithms. Results show that the multivariate algorithm produces the lowest prediction errors for all cases of background water quality. A comparison of the number of false positives reported from the change detection algorithms and a set point approach highlights the efficiency of the change detection algorithms. Across all three algorithms, most prediction errors are within one standard deviation of the mean water quality. The event detection results show that the best performing algorithm varies across different background water quality models and simulated event strength. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

To date utilities have focused predominantly on the quality of treated water and the physical condition of distribution assets when improving the quality of water at the customer's tap. However, the quality of the water delivered is also affected by the time the water is retained in the different elements of the distribution network. As water passes through a distribution system, its quality, with respect to many chemical and biological parameters, will degrade as a result of chemical processes and biological reactions that occur in the bulk water and at the pipe wall. Pipe wall reactions are strongly influenced by the distribution system materials and the contact time of the water with them. AwwaRF published the findings of the research project, titled #2796: Managing Distribution Retention Time to Improve Water Quality in May 2006 as an electronic guidance manual. The software based decision support tool provides a structured approach and practical methodologies for managing water quality by controlling retention time. The presentation will introduce the session "Meeting Current and Future Drinking Water Standards: The Role of Retention Time Management" setting the background to the session using incidents and projects to illustrate some of the issues surrounding deteriorating water quality in distribution systems and how they could be managed efficiently. The presentation will conclude with an introduction to the electronic guidance manual developed during the research project. This paper therefore provides the background to the presentation. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The provision of safe drinking requires high quality water sources, treatment of the water, and its distribution. However, the first distribution systems in cities were designed for fire protection. In time they were extended to commercial and then to residential properties but their design then was the responsibility of the National Board of Fire Underwriters. Today, it is the Insurance Services Office, Inc., whose regulations are the text of AWWA's Manual M31, Distribution System Requirements for Fire Protection. Not one word relates to drinking water quality. Its only purpose is to assure adequate pressure and flows for fire protection. Apparently, drinking water quality is of no importance. Recent AWWA Water Quality Technology Conferences have had hundreds of papers on the degradation of water quality in distribution systems. Last year, AWWA compiled a 1,083-page volume. Water Quality in the Distribution System. The problems arise from defects in distribution systems designed for fire protection. Pipes and storage tanks need to be larger than necessary for drinking water systems, resulting in residence times that range to months. Long residence makes many water quality problems. Difficulties in maintaining disinfection, demanding higher doses of disinfectant, in turn result in excessive levels of disinfection by-products (DBPs), which are the most serious water quality health problems extant today. Another problem created by long residence times is the growth of biofilm on the insides of the pipes. The biofilms react with the disinfectants, reducing their and creating growths that harbor the pathogens This leads to tuberculation that in time reduces the hydraulic capacity of the pipes. Still another problem created by the minimum 6- and 8-inch diameter pipes, which make up the larger lengths of pipes in the systems, are heavy pipes laid on soil with some 350 joints per mile, many of which, in time, leak. Loss of water may not be serious, but the frequent negative transient pressures that occur in the pipes have been found to suck contaminated groundwater into the pipe. One solution is a distribution system that carries only drinking water. It would not have hydrants and in most of the city, the pipes would be less than one inch, and laid with few joints. It would be a drinking water distribution system with none of the problems we face today. What about fire protection? In the last half-century, dual water systems have spread over the country. They were created to conserve drinking water by using reclaimed wastewater for the many nonpotable demands, including water for fire protection. More than 2,000 cities and towns have dual systems and more are being created every day. Most conventional water treatment plants are far short of being adequate despite statements that they meet EPA regulations. If only drinking water needs to be treated, the far lower cost would permit the adoption of membrane treatment, assuring drinking water of high quality Without question, new communities adopting such dual systems would deliver much higher quality water at lower cost. For existing communities, distribution systems would need to be introduced gradually when retrofitting for growth and as funds are available. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Changes and deterioration in water quality occur during the passage of potable water through distribution systems. Resulting issues include taste and odour, high corrosion rates, precipitation, discolouration, disinfection by-product formation, and biological activity. While the direct simulation of the behaviour, reactions and interaction leading to such problems are currently beyond complete understanding and modelling capabilities, many of these processes and mechanisms have been directly related to the age of the water. Hence it is desirable to predict the distribution of age of water throughout water distribution systems as a surrogate for water quality characteristics, and as a basis for further calculation. Current mathematical simulation models calculate average water age, assuming flow weighted mixing. In reality however, age values can not be averaged in this manner. A litre of water with an age of 2 hours, mixed with a litre with an age of 10 hours, does not produce 2 litres with an age of 6 hours. The fact is that 50% of this water is 5 times as old as the other 50% and will have different quality characteristics; but current models disguise this information. To improve age representation, an approach which provides information about the component ages contributing to the mean age has been developed. The enhanced model tracks the volumes and ages of different flow contributions, within user defined bands, throughout water distribution systems. As the age of water within a network is related to flow patterns, it is postulated that age information derived from the model, in particular extreme age components, can be translated into operational strategies to manage flows, and hence residence times, thereby improving overall water quality and asset performance. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water age, or time elapsed since treatment and release of that water into a distribution system, is generally acknowledged an important criterion with respect to its degradation in quality within that system. It is significant in each of four main sources of such degradation; i.e., bacteriological growth, formation of such chemicals as disinfection-by-products (DBPs) and taste and odor compounds, leaching of pipe materials, and point contamination from either human or natural intrusions. For any network-derived water quality degradation phenomenon that varies as a function of time, a specific relationship between water age and extent of water quality degradation can be established. The work described here develops a closed-form mathematical relationship for approximating variations in water age throughout distribution networks. The approach results in a simple equation for relating water age and the cumulative fraction of connections within a distribution system, and provides a useful tool to examine the impacts of various time-dependent phenomena on water quality within a system. The model further provides useful insights regarding large-scale network parameters that influence system water age; e.g., the important role of network storage volume. The approach presents a simple tool for estimating the fraction of a service population receiving water above or below a certain water age and thus above or below any particular water quality degradation threshold. In its simplest form the model uses network storage, number of connections, and flow per connection to develop a relationship between cumulative fraction of population and water age. Extensions on the basic form of the equation can be introduced to account also for geographical variations in population density and network storage. The model approach idealizes flow patterns and storage capacities within a hypothetical distribution network representing a typically complex city as a smooth surface. A single centrally located water source is assumed to emit flow emitting radially outward towards consumer endpoints. The instantaneous increase in water age at any point is integrated over the entire distribution surface to obtain time since treatment as a function of position within the network.. The water age model calculations compare well with those obtained from a network model solution using nodes and links. Good correlations between the closed-form solution and network models created with as few as 20 connections are shown, and model accuracy is demonstrated to increase with system size due to the underlying assumptions of smooth flow and demand surfaces. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Source water has been deteriorating in China in the recent decades. Due to the increasing demand for higher quality of drinking water, water quality management is paid close attention to. Many maximum contaminant levels for drinking water have been lowered, and there are more and more contaminants which were not regulated are in the list of the Standard for Safe Drinking Water in China. However, drinking water from many municipal water works and distribution system is typically not suitable for direct drinking without further treatment in China. Due to economical limits to supply high quality drinking water for most cities and communities, dual water distribution systems have been expanding steadily since last decade in China. In this paper, the origin, current status, and trend of research and application in dual water distribution systems in China were reviewed and discussed. In a typical dual water distribution system, about 5 percent of drinking water demand is by-passed to an advanced water treatment station in residence or business zones. Effluent water from the treatment is of high quality and suitable for direct drinking. High quality drinking water is then supplied to each household, office, or public building through a separate pipe networks. Dual water distribution systems have many advantages over conventional water distribution systems in China. The water from a dual water distribution system is safe and convenience for drinking. It is also more economical and practical to supply high quality drinking water and common drinking water separately in a dual distribution system where the quality of source water is often low. In China, the first dual water distribution system was set up at Jinghua Residential Quarter of Pudong District of Shanghai in 1997. Since then, technologies related to dual distribution systems have been developed or established. A lot of technological problems including advanced treatment processes, pH regulation, sterilization, and pipe selection have been solved cost-effectively. Many dual distribution system projects for residential quarters have been constructed, and many engineering companies have been providing services to design and construct dual water distribution systems in the last decade in China. It has been demonstrated that dual distribution systems are cost-effective and in accordance with the current status of economic and environmental conditions in China. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Las Vegas Valley Water District (LVVWD) has experienced rapid expansion of its water distribution system for many years, and has observed water age related issues in the early stages of new development, where large scale water facilities have been constructed and few homes occupied. The increased water age not only results in low chlorine residual, it also contributes to DBP formation. The LVVWD has also observed that water age in development areas reduces over time as more homes are occupied and water demand increases. To address this issue, the LVVWD is investigating localized treatment technologies and is participating in an American Water Works Association Research Foundation Tailored Collaboration Project 3103, entitled "Localized Treatment for Disinfection By-products" (AWWARF Project). The LVVWD has also studied operating strategies to reduce water age in its distribution system. This paper describes a conceptual plan to use a combination of localized treatment and operating strategies as an alternative to using chloramination. Topics discussed include research status of localized treatment options and the feasibility of combining localized treatment with distribution system modeling. Future options are also discussed for treatment plant relocation, from areas where water age issues have resolved themselves, to new development areas. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The Initial Distribution System Evaluation (IDSE) requirement of the Stage 2 Disinfectant/Disinfection By-Products Rule requires all public drinking water systems to identify monitoring locations that are characterized as "near the entry point", and "average residence time", and locations of "high total trihalomethanes" (THM) and "high haloacetic acids" (HAA) concentration. While a distribution system network model is the preferred method to identify these locations, many public water systems to not maintain a calibrated model. In such cases, an alternative approach may be used that applies statistical analysis of existing chlorine residual data to the IDSE requirements. Moreover, statistical analysis can test hypotheses that chlorine residual is a surrogate for water age for individual systems, in cases where both THM and chlorine residual were analyzed for the same sample. If the hypothesis is true then average residence time, high THM locations for the IDSE can be identified using existing chlorine residual data. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Understanding the dynamics of a treated water distribution system is critical to its successful operation and maintenance. While hydraulic models of water distribution systems are often calibrated hydraulically to reproduce historic operational performance data (water treatment plant production, tank level fluctuations, pump station operations, as well as in-system pressure and flow data), confidence in predicted flow patterns and velocities within these distribution systems is limited, especially in robust systems where the piping network has excess conveyance capacity. Two fluoride tracer tests were performed by the City of Fort Collins, Colorado during the period of February 20, 2006 through March 17, 2006. The first-test was conducted with fluoride in decreasing mode (at time zero the fluoride feed was turned off). The second test was conducted immediately following the first test with fluoride in increasing mode (at time zero fluoride dosing was resumed at a rate that, when combined with background concentration levels, resulted in a fluoride concentration of 1 mg/L for distribution). The purpose of this study was to provide data to: 1) Verify and refine the hydraulic calibration of the City's existing H2OMAP model of its drinking water distribution system 2) Calibrate the City's H2OMAP drinking water distribution system water quality model 3) Increase understanding of factors influencing water quality in the water distribution system 4) Comply with Initial Distribution System Evaluation (IDSE) requirements 5) Help determine sampling locations in the water distribution system for disinfection byproducts and other water quality parameters 6) Evaluate the formation potential of disinfection byproducts 7) Determine an appropriate, water main flushing strategy and 8) Examine the influence of treated water supplied from neighboring water district interconnections on water quality. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The planning and execution of chemical tracer tests have been detailed in a recent reference guide developed by the U.S. EPA for water utilities. The objective of this paper is to extend the EPA guidance by presenting a simple, analytical technique to calculate mean residence time (MRT) from chemical tracer data. Case histories of tracer tests in three distribution systems in North Carolina (Raleigh, Greenville, and Cary) are used to illustrate the technique. Examples are given of the time required for sampling and the mileage driven. Tracer responses to positive and negative step input tracer tests are compared. Storage tanks are shown to affect the tracer response at nearby stations and complicate calculation of an MRT value. The MRT values obtained at 22 stations in the Raleigh distribution system compared fairly well to hydraulic model predictions of the average water age. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Detecting chemical contaminants in water distribution systems has traditionally been accomplished by grab sampling techniques or more recently, using high-end process analyzers such as TOC (total organic carbon). The method presented in this paper uses an online UV254 analyzer to look for chemical contamination events. The results of seven "particularly hazardous" chemicals and three baseline chemicals are presented, along with laboratory analysis of each, showing the viability of using UV254 for quickly investigating and detecting contamination events at a much lower cost than with traditional instruments or grab sampling techniques. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A number of improved on line detection systems that analyze for chemicals, organics, and particles in water and in real time are available. However, a major concern in safeguarding water purity is still real time detection of pathogens. Even though improved rapid methods are available that can shorten the analysis time, these are typically performed through intermittent grab samples and cannot provide an automated alert of an insidious biological contamination event. Through years of research and development, an emerging technology called MALS (multi-angle light scattering) can now achieve on line, real time detection and classification of waterborne microorganisms. The method utilizes a continuous slip stream of water passing through a flow cell in the instrument. A laser beam, focused perpendicular to the water flow, strikes particles as they pass through the beam generating unique light scattering patterns and are captured by photodetectors. Microorganisms produce patterns termed `bio-optical signatures' which are comparable to fingerprints. By comparing these bio-optical signatures to an on-board database of microorganism patterns, detection and classification occurs within minutes. If a pattern is not recognized in 'the data base, it is classified as an `unknown' and the unidentified contaminant is registered as a potential threat. In either case, if the contaminant exceeds a customer's threshold, the system will immediately alert personnel to the contamination event while extracting a water sample for further confirmation. A new system utilizing MALS technology, called BioSentry^TM, has been developed by JMAR Technologies and is now commercially available. The system is cost effective, uses no reagents, is easy to operate, and can be used for continuous microbial surveillance in many water treatment environments. The advantages for use in municipal water distribution systems will be presented along with data from field installations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Chlorination of treated drinking water has been widely used to prevent microbial contamination or waterborne diseases caused by reverse siphon or cross pipe connection in pipe networks. Continuous residual chlorine test is required to ensure the safety of drinking water against pathogenic microbes. With development of sensor technology, i.e., easy operation, lower operating cost, and no discharge of reagent, on-line chlorine sensors have been widely used in water supply facilities. Although the device has many advantages, it also has a little practical problems related with reliability in measurement reading due to disturbing materials and difference in N,N-diethyl-p-phenylenediamine (DPD) principle. The performance of on-line chlorine sensor differs significantly according to their manufacturers. The objective of the study was to evaluate the performance and the characteristics of 5 sensors which were used in water supply facilities in Seoul Metropolis at the moment. One was sensor with reagent and the other four sensors were no-reagent type. The effect of combined chlorine and pH to free chlorine was investigated with one reagent type sensor and four no-reagent type sensors. The authors concluded that the on-line free chlorine sensors were affected by combined chlorine and pH. No-reagent type sensors calibrated for pH were not affected pH while sensors not calibrated were affected significantly by pH fluctuation. With sufficient chlorination, all the sensors tested recorded similar free chlorine concentration. But insufficient chlorination with much combined chlorine shows different result according to the sensors which means the effect of combined chlorine. The important parameters to be checked in evaluating on-line chlorine sensor applied for water supply facilities are pH and combined chlorine. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The increased use of on-line sensors means that a number of physical and chemical parameters are often measured routinely in drinking water distribution systems. These include temperature, pH, turbidity, free and total chlorine, oxygen reduction potential (ORP), electrical conductivity (EC), and fluoride. Laboratory trials showed that real-time monitoring of such parameters following addition of suitable tracers could be used to measure the travel time of water in pipes. Inert tracers gave good estimates of travel time, while reactive tracers such as temperature pulses interacted with the pipe and were delayed. Laminar flow caused increased broadening of tracer pulses, so that turbulent flow was preferred. Inspection of on-line monitoring data also showed that these parameters fluctuate naturally with time giving the water flow a characteristic fingerprint without the need to add a tracer. Field trials showed that these fingerprints were preserved for significant periods and could be successfully detected in a local distribution system after transit times of more than one hour. Visual matching of output peaks was used to measure transit time, and some success has also been achieved using an automatic correlation routine to track water parcel movement. The fingerprint tracking technique is a unique way of following water movement without using added tracer, and it does not require detailed knowledge of network connectivity that is necessary when using a hydraulic modeling approach. This technique has considerable potential in water quality and water security applications, and should enable real-time measurements of water age and chemistry in distribution systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

It has become generally accepted that water quality can deteriorate in a distribution system through reactions in the bulk phase and/or at the pipe wall. These reactions may be physical, chemical and/or microbiological in nature. Perhaps one of the most serious aspects of water quality deterioration in a network is the loss of disinfectant residual that can weaken the barrier against microbial contamination. Data from studies conducted in a specially constructed pipe loop located at the U.S Environmental Protection Agency's (USEPA's) Test and Evaluation (T&E) Facility in Cincinnati, Ohio and from a study conducted in collaboration with the Burlingame California (CA) water utility is utilized to confirm the belief that chlorine and chloramine residual losses increase with increasing mass transfer in corroded metal pipes. Two models are compared. One is a wall reaction limited model such as the model embedded in EPANET and the other is a mass transfer limited model. In general it is concluded that the mass transfer limited model performs more effectively then the wall reaction limited or EPANET model. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Drinking water produced from desalinating coastal seawater is characterized by high bromides and low organic precursors. The main formed chlorination byproduct in this case is bromoform. A recent study evaluated such byproducts in the desalinated water of a coastal city in United Arab Emirates and reported low levels of most byproducts but bromoform. The levels of formed bromoform at locations far from the desalination plant were generally less than their levels in the desalination plant indicating bromoform attenuation in the distribution system. A series of lab experiments were conducted to verify such phenomena by evaluating the bromoform formed in the same desalinated water when transported in a simulated pipe environment. Results were obtained for different hydraulic conditions; such as flows and pipe diameters as well as for varying dosages of spiked chorine. The dynamic effects associated with the transported water in pipelines were found to clearly causing bromoform hydrolysis and attenuation. The initial growth of bromoform followed by its decay was successfully modeled using a first-order saturation growth rate considering a newly developed time-dependent ultimate concentration. The ultimate concentration was related to the prevailing hydraulic conditions using the well known wall and mass transfer coefficients. Such formulation allowed to favorably simulating the bromoform formation in the desalinated water under various dynamic conditions including the static condition of zero flow. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Decay of chlorine residuals within a water distribution system is a result of a number of interacting processes operating at multiple spatial and temporal scales. These processes are not easily combined into a single and simple predictive model. In the past, first-order decay models have been widely adopted to describe chlorine decay. However, as these models only entail two variables (initial chlorine level and time), these models are not representative of many of the important phenomena influencing chlorine residual decay. Numerous other factors have been identified, including pipe wall material, TOC, temperature and pH. As an alternative, multi-component chlorine decay models represent the aggregation of underlying processes contributing to overall chlorine consumption more appropriately. Clark first proposed a two-component second-order chlorine bulk decay model based on the concept of competing reacting substances. The research in this paper extends the two-component second-order model to incorporate recent findings of the involvement of NOM in the chlorine decay process. This model is further applied in the estimation of wall decay coefficients as used in EPANET. A novel procedure, a two-step parameter assignment method which employs a Bayesian statistical method and Monte Carlo Markov Chain (MCMC), is described to evaluate the parameters of both bulk decay and the wall decay components. A study conducted in Goderich, Ontario, Canada is used to demonstrate the application of the two-component second-order chlorine model. The effects of pipe diameter, pipe wall roughness, flow velocity. TOC, initial chlorine concentration on chlorine decay are investigated. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Constituent modeling was utilized to evaluate a "blending" strategy to achieve Gross Alpha (GA) concentration compliance at two of New Jersey American Water company's groundwater (GW) stations. An extended period simulation model measured the anticipated GA levels within the distribution system from two GW sources that exhibited elevated GA levels and two surface water supply sources that had minimal GA levels. Quarterly compliance sampling at the two GW stations' points of entries had revealed elevated GA results for the first two quarters of 2005. GA is indicative of naturally occurring radionuclides in the aquifer system. The federal and state maximum contaminant level (MCL) for GA is 15 pd/l calculated as a running annual average over 4 consecutive quarters. Although part of the increased levels may be attributed to new and/or improper laboratory analysis methods, MAW felt a plan of action was required in case the early sample results proved to be valid. The laboratory analysis issue is being pursued separately with the state Department of Environmental Protection. Following a review of GA results, NJAW decided in July 2005 to reduce the amount of GW supply by limiting the production from the two groundwater stations. System demands were met by increasing the surface water (SW) supply received through the two surface water interconnections. Based on the availability SW supplies, NJAW modified the supply operation by blending SW with each station's GW to reduce the concentration of GA that enters the distribution system and achieve compliance with the MCL. Minor piping, valve, and instrumentation modifications were undertaken at each station to achieve a SW:GW blend. Once both physical and instrumentation changes were made, NJAW also conducted pilot studies at each station to validate the blending process by varying the SW:GW ratio. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Corrosion in a water distribution system has many adverse impacts including the build-up of scale on the pipe wall, the loss of hydraulic capacity, compromised structural integrity that can lead to the onset and proliferation of leakage, and the deterioration of water quality. Yet, for all its familiarity, pipe corrosion is a complex phenomenon due to the inter-dependent set of reactions that can take place at the pipe wall and with chemicals in the bulk water. The principle objective of this study is to explore the impact of chemical species on an existing but basic 1-D transient model. The research is focused on studying the impact of various chemical species — dissolved oxygen, pH, and the carbonate system (CO₂, HCO₃^–, & CO₃^2–) on the formation of ferrous (Fe²⁺) and ferric (Fe³⁺) ions. The associated formation of various iron precipitates is also numerically explored. For the purpose of this study, the 1-D corrosion simulation model developed by Naser and Karney (2005) is extended by adding three sets of chemical reactions including reaction of the pipe wall with the bulk water and dissolved oxygen, reactions involving the carbonate system, and iron precipitation. Each chemical species is tracked using the advection-diffusion-reaction equation (ADRE) coupled to a hydraulic model involving the continuity and momentum equations. The numerical solution of ADRE is computed by the implicit finite difference method and the flow equations are resolved with the method of characteristics (MOC). The model is simulated for various flow conditions both with and without consideration of the carbonate system. Results show that the rate of iron release increases in the axial direction but varies with flow conditions and simulation time. When the carbonate system is neglected, the oxygen concentration steadily decreases, and both iron concentrations and pH increase along the pipe axis with time. However, the inclusion of the carbonate system directly affects the pH and dissolved oxygen concentration in the bulk water which then influences the rate of iron release. In particular, the decrease in pH values and the increase in dissolved oxygen concentrations along the pipe axis, which are observed by the inclusion of the carbonate system into the model, accelerate oxidation of the pipe wall, and a high rate of iron release is observed. This reveals that flow conditions, low pH, high dissolved oxygen concentrations, are the most significant factors that increase the corrosion rate. In a real water distribution system, the rate of internal corrosion is further influenced by other so-called secondary reactions, and also by the physical, chemical and biological characteristics of the water. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper summarizes the different aspects that play major roles in the function of pipe materials in drinking water distributions systems. World widely, there are different water qualities and different materials used in drinking water pipe systems. Different materials with different properties used in water distribution systems consist of metals (copper, cast iron, lead, brass, steel), plastics (PVC, PE, PB, latex), asbestos cement and concrete. Physical-chemical factors influencing functions of pipe material include temperature, pH, flow rate (–>retention time), alkalinity, hardness, conductivity, metals, organic compounds, nutrients for microbes and disinfection chemicals. The main microbiological hazard with pipelines is the formation of biofilms on the surfaces of pipes and other materials in contact with the corrosion of metallic materials can shorten their service life remarkable. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Discoloration is a major source of consumer complaints associated with a water distribution system (WDS). Depending on the pipe material, different colors can be attributed to varying concentrations and chemical oxidation states of iron or other metals. Overall, the discoloration materials are apparently entrained from particles at the pipe wall during the passage of water. The major material sources for discoloration materials are organic and inorganic materials entering a WDS at the source, materials from treatment processes, processes within a WDS such as corrosion and erosion, biological growth and chemical reactions, external contamination occurring during operations and repairs, intrusion and back flow. Traditionally, two mechanisms — hydraulic entrainment and iron uptake — have been used to described discoloration events. The hydraulic-entrainment mechanism is the physical motion of loose deposits of particles at the pipe wall into the bulk flow due to abrupt changes in the flow condition. In iron-uptake mechanism, some particles of ferrous hydroxide (produced by corrosion reactions) are oxidized to ferric hydroxide, which then can cause red bulk events in the bulk flow. Considering the hydraulic-entrainment a key main mechanism initiating discoloration event, this paper seeks in at least a preliminary way to simulate and approximate red water events in a water a WDS. Abrupt changes in flows can create aggressive shear forces, sometimes causing the accumulated particles at the pipe wall to be resuspended and be carried with the bulk flow. In this study, one- and two-dimensional (1-D and 2-D) transient simulation models are proposed, by which the turbidity of water is modeled with a transport equation. The transport equation, which is then coupled with the appropriate continuity and momentum equations for flow, considers advection and diffusion as the main mass transport mechanisms. Defining a self-cleaning threshold limit for wall shear stress, Boxall's approach is used here to model the release of the discoloration material from the pipe wall into the bulk flow. To overcome the complexity in the determination of the turbulent shear stresses and to reduce the computational burden, a five-region model is used to model the turbulence. In this model, molecular viscosity is responsible for damping turbulence in the region near the pipe wall (viscous sublayer) while eddy viscosity is the main damping-mechanism in the core region (nearer the centerline of the pipe). The turbulent diffusivity coefficient is accurately approximated from the eddy-diffusivity concept and the dimensionless turbulent Schmidt number; whereas the 1-D counterparts are determined using steady state Darcy-Weisbach formula (for wall shear stress) and Taylor's formula (for diffusion coefficient). An implicit/explicit finite difference method; namely an Alternating Direction Implicit (ADI) approach, is superposed with a fixed grid method of characteristic (MOC) to numerically integrate the governing equations for unknown flow parameters including velocity, pressure head, and water turbidity. Finally, the results of both 1-D and 2-D models for a straightforward reservoir-pipe-valve case study are compared with available field data reported in the literature by other researchers. The overall agreement of the two sets of results supports the basic validity of the proposed 1-D and 2-D models. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

This paper presents computational simulations and experiments of water flow and contaminant transport through pipes with incomplete mixing at pipe joints. The hydraulics and contaminant transport were modeled using computational fluid dynamics software that solves the continuity, momentum, energy, and species equations (laminar and turbulent) using finite-element methods. Simulations were performed of experiments consisting of individual and multiple pipe joints where tracer and clean water were separately introduced into the pipe junction. Results showed that the incoming flow streams generally remained separated within the junction, leading to incomplete mixing of the tracer. Simulations of the mixing matched the experimental results when appropriate scaling of the tracer diffusivity (via the turbulent Schmidt number) was calibrated based on results of single-joint experiments using cross and double-T configurations. Results showed that a turbulent Schmidt number between ∼0.001–0.01 was able to account for enhanced mixing caused by instabilities along the interface of impinging flows. Unequal flow rates within the network were also shown to affect the outlet concentration at each pipe junction, with "enhanced" or "reduced" mixing possible depending on the relative flow rates entering the junction. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Network simulation models for water distribution systems typically assume the mixing at pipe intersections is complete and instantaneous. More accurate characterizations of chemical or biological agent transport may be required to efficiently identify, control and mitigate the spread of harmful agents and to protect critical components within the network. Recent experimental data have shown that mixing is incomplete at pipe junctions (pipe crosses and tees). Numerical simulations of these experiments have been performed using simplified turbulence methods. In the present study, simulations of mixing in pipe junctions are performed using the high-fidelity Large Eddy Simulation (LES) approach to fully resolve the mixing behavior. Simulation results show unsteady mixing behavior at the fluid interface due to shear layer instabilities. These high-fidelity results will be used to develop a lower fidelity model for mixing at pipe junctions. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The movement of chemicals or biological agents in a water distribution system is examined via computational fluid dynamics simulations. A series of computational simulations using selected intersecting geometries are carried out at various Reynolds numbers. Boundary conditions, turbulence intensities, convergence criteria, and mesh sizes are thoroughly evaluated. The present parametric study focuses particularly on pipe intersections to characterize complex mixing phenomena in pressurized water distribution pipe networks. Selected computational results are compared with experimental results. The water quality model integrated with an existing computer program (EPANET) was re-evaluated based on the computational and experimental data. Initial computational fluid dynamics simulations consistently underestimated mixing, and experimental data are utilized to reexamine turbulent mixing by adjusting the turbulent Schmidt number. Corrections based on computational results are incorporated into the existing code as an example case study. The improvement of the existing code may be important not only to predict concentrations of chemical species such as chlorine in water distribution systems, but also to prepare for potential intentional and accidental contamination events. Computational results must be further calibrated and verified through lab- and field-scale experiments. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

An Advection-Dispersion-Reaction model, ADRNET, is developed to perform dynamic water quality simulation in complex pipe networks. The network hydraulic functions in the EPANET toolkit are incorporated and used as hydraulic engine to simulate the extended period pipes flows in ADRNET based on the time-averaged stochastic demands. An Eulerian-Lagrangian scheme is combined with a numerical Green's Function technique to numerically solve the ADR equation efficiently in network conditions. Verification with analytical solution shows that ADRNET achieves excellent performances of numerical convergence and stability. ADRNET is applied to an all-pipes network with stochastic water demands to compare against results obtained by EPANET advection-reaction water quality model and field observations. Both models achieve similar results at locations where turbulent flows prevail. However, the ADRNET provides better agreement with field observations than EPANET at locations where laminar flows dominate. The ADRNET proves to be a valuable tool for managing water quality practice in water distribution network. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Previous studies have monitored pressure transients and fluctuations in several distribution systems and low pressures (<20 psi) were commonly recorded, including a few negative pressure events. During a low-pressure event, microbiological contamination of the distribution system can result due to intrusion of contaminated water. However, no study to date has attempted to assess microbiological water quality in temporally-composited samples collected immediately following a pressure drop, nor have temporally-composited samples been collected to statistically characterize microbiological water quality in a distribution system or point of entry. The findings presented here represent preliminary results from an ongoing study monitoring a large distribution system for pressure fluctuations and assessing microbiological water quality in 100-L water samples collected immediately after the occurrence of low pressure events. Pressure monitoring by up to six pressure loggers located at the point-of-entry (POE) and throughout the distribution system for four months found 11 distinct low pressure events. Of these, three separate events resulted in minimum pressure readings in the distribution system of 0 psi. The data indicate that these events can be both isolated and system-wide, and that system-wide events were associated with pressure drops at the POE. In addition to providing additional data on the occurrence of low pressure events in the distribution system, these data will assist in determining appropriate locations of automated monitoring and sampling units designed to collect water samples following low pressure events. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Monitoring of drinking water distribution systems should undertake to capture the worst case water quality scenarios in order to provide the maximum protection for public health. Total coliform monitoring is expensive and time consuming relative to other chemical or physical measures of system integrity that may be monitored in real time. This paper proposes a methodology for mining these additional water quality parameters to inform coliform monitoring. Three machine learning algorithms are selected for application to a case study distribution system: i) gradient tree boosting, ii) decision trees, and in) distance-weighted nearest neighbor algorithm. In addition, the effect of expanding training data to include unlabeled data records will be explored. The performance of these data mining techniques should provide insight into the use of surrogate water quality parameters to indicate high coliform levels. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Earlier descriptions of water distribution systems (WDS) microbial communities have relied on culturing techniques. These techniques are known to be highly selective in nature, but more importantly, they tend to grossly underestimate the microbial diversity of most environments. The goal of this ongoing study is to compare the microbial composition of a WDS receiving different sources of water. To circumvent some of the problems associated with culture-based techniques, sequence analysis of total community 16S rDNA clone libraries was performed. A total of 260 16S rDNA clones were analyzed in this study. Sequence comparisons with existing databases revealed that α-Proteobacteria and Mycobacterium sp. represented nearly 43% and 48% of the total clones examined, respectively. Other bacterial groups identified included members of the genera Legionella, Pseudomonas, and Agrobacterium. Seventy percent of the clones analyzed in this study showed at least 97% sequence identity with sequences available in current databases. While α-Proteobacteria is a numerically dominant group in chlorinated drinking water systems, the abundance of mycobacterial sequences in this study indicates that there are significant differences in microbial community structure between the WDS analyzed. These differences could be attributed to differences in water treatment or receiving water sources between the distribution systems. While some of the genera identified in this study have been associated with some public health risks, it should be noted that analysis of 16S rDNA clones does not confirm the presence of pathogenic strains of any organisms identified in this study, as current methods for detection and identification require several steps including selective enrichment, isolation, and final confirmation via in vitro studies. Future studies will focus on expanding sequencing databases to more accurately characterize these potential differences. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

By including Mycobacterium avium Complex (MAC) on the Contaminant Candidate List, the U.S. EPA has initiated research to evaluate if MAC represent a significant threat to human health through occurrence in drinking water supplies. Currently, culture-based approaches to monitor for Mycobacterium require weeks before results are available. To reduce the time needed to screen for Mycobacterium, an interdisciplinary team including environmental and electrical engineers, microbiologists, and physicists worked together to create miniaturized culture-based devices to detect Mycobacterium rapidly. The first generation device relies upon the lipophilic properties of Mycobacterium to isolate target microorganisms in an abundance of non-target microorganisms. Water samples are re-circulated through a twenty-liter water distribution system simulator where paraffin chip devices are exposed to Mycobacterium present in the bulk water. After one day's exposure, the devices, are removed, prepared, and analyzed. An occurrence study of environmental MAC in southwestern Ohio hospitals is in progress. The two main questions being addressed by this study are: (1) is MAC in water distribution systems in health care settings, and (2) does the device provide a faster and more accurate way to detect MAC in environmental samples when compared with traditional methods. There is ecological, epidemiological, and pathological significance in rapidly screening for MAC in drinking water systems. It is possible, using the culture-based device described, to detect and identify MAC more quickly, and future developments aim to use this device for real-time MAC screening. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Six biofilm annular reactors were used in three duplicate pairs to simulate a 15.24cm diameter pipe with a flow velocity of 0.3m/s (Re=46,000). The mean residence time was set to be at least 3 hours. These conditions were chosen to reproduce the characteristics of an experiment by Gibbs et al. (2003). The main objective of this study was to check if a given pathogenic microorganism, Klebsiella pneumoniae, could colonize a mature water-distribution system biofilm. To achieve conservative results it was decided to choose parameters representative of the friendliest water-distribution system environment possible for the bacteria to penetrate the biofilm. This included dechlorinating the tap water used as feed and keeping the reactors at room temperature. The choice of Klebsiella pneumoniae as the microorganism of study was warranted by its ability to produce high amounts of polysaccharides, making the cells likely to stick to the biofilm. The study was split in two parts. In a first step, the spatial variability of the biofilm density and the hydraulic residence time distribution were determined for each reactor. The second study consisted of injecting three 100mL solutions of different concentrations of Klebsiella pneumoniae into respective duplicate reactors. The first experiment showed that at a given sample time the biofilm is not homogeneous within a reactor and this pattern is repeated throughout each reactor. These results were taken into account when designing the second experiment. Composite samples of two coupons were collected to decrease the impact of sampling variability. The second study showed that at injected concentrations of 10⁷ and 10⁵ microorganisms per milliliter, colonization of the biofilm seemed unlikely. The cell counts dropped very quickly and after 5 days most of the samples showed non-detectable cell counts. In the case of a 10⁹ cell/mL injected amount, the cell counts were steady during the first 24 hours before decreasing. This shows the potential for a threshold value in retention of the microorganism by the biofilm. After 5 days, the cell counts had decreased by 4 log units. However, it is not possible to reject the potential for regrowth, since the amounts of K. pneumoniae cells in the samples were still very high in the last sample event. The sampling period was limited by the number of available coupons. To check for regrowth, future experiments will require a sample design that includes a longer sampling period. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Basic research is needed to better understand the potential risk of dangerous biological agents that are unintentionally or intentionally introduced into a water distribution system. Included in this research is an assessment of how bio-pathogens will interact with biofilms growing on water-distribution system pipe-wall surfaces. Experiments are being run at Sandia National Laboratories to better understand the impact of biofilms on pathogen dissemination through water distribution systems, both before and after decontamination of the system. The hope is that these data can be fed into numerical tools that quantify the risk and associated uncertainty of pathogenic biological agents being introduced into water distribution systems and the effects of biofilms on this risk. In addition these data could be used in numerical tools that assess decontamination methods for water distribution system networks. Reproducible Pseudomonas fluorescens biofilms were grown in annular reactors with plate counts on the order of 10⁵ and 10⁶ CFU/cm². A series of pathogen-introduction experiments, where both 1-μm-diameter fluorescently labeled polystyrene microspheres and Bacillus cereus (as a surrogate for B. anthracis) were introduced in the annular reactor after the P. fluorescens biofilms have reached stationary phase. Integration of the pathogens in the biofilms was monitored by gram-specific plating techniques and epifluorescent microscopy. Fluorescent nucleic-acid stains were used to visually differentiate the gram-positive (B. cereus) and gram-negative (P. fluorescens) organisms. Fluorescence spectrophotometry was used to measure the microsphere concentrations. Variables examined in the experiments include initial pathogen concentration, rotation speed of the inner cylinder of the annular reactor (shear force), and biofilm plate counts prior to the start of the experiment. Chlorine was added to the system after the pathogen had at least 14 days in contact with the biofilms and the impact of the chlorine on pathogen concentration in the biofilms was monitored. Pathogens are observed to get integrated in the biofilms at a relatively constant concentration until the start of the chlorine treatment. The higher the initial pathogen concentration and the higher the initial biofilm plate counts prior the pathogen introduction, the more pathogen integrated into the biofilms. A decrease in pathogen concentration to below the detection limit in the biofilms and the reactor water was observed during the chlorine treatment. From the results of one experiment, it appears that the pathogens do not recover after termination of the chlorine treatment. However, more data need to be collected to confirm these results. These results indicated that biofilms may act as a safe harbor for bio-pathogens in drinking water systems, making it difficult to decontaminate the systems. Chlorination of the system appears to be effective in decontaminating the system in the annular reactor geometry at the concentrations that we used. Additional experiments are being run to further test this observation. As these experiments cannot assess the effectiveness of chlorination at pipe junctions and because annular reactors have a higher surface area to volume ratio than pipe systems, the results from these experiments should also be tested in a system with a more complicated geometry that more closely matches water distribution systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Based on a hydraulic simulation model, the operational optimization problem for the multi-source water supply network is proposed. The objective is to minimize the operation costs in the water distribution system including pumping cost and cost for water purification. In the case of a large-scale water network with many pump stations and tanks, there will be a huge computational load when a conventional genetic algorithm is used to solve the optimization problem. Therefore, an improved genetic algorithm is presented in this paper. Firstly, a mixed-coding methodology is adopted according to the characteristics of the decision variables. For fixed speed pumps, use 1 or 0 representing open or close operation; while for variable speed pumps, use a hybrid code to denote the running states. Secondly, a Pseudo-Parallel GA (PPGA) is designed and implemented, which is based on a coarse grained scheme. In this PPGA, several populations are firstly, generated and evolve separately, and the migration of individuals among the populations is performed through an exchanging tunnel in which the best individuals from different populations can be exchanged with each other. The PPGA is implemented on a PC using multithread programming techniques. The PPGA can not reduce the computer time obviously because the work is still performed on single personal computer rather than in a real parallel computing environment. However, the evolution process of those populations may be accelerated through the migration of individuals. Both SGA and PPGA were applied to solve an operational optimization problem for a case study water network. Compared to SGA, the PPGA is proved to be more efficient and able to find much better solutions. The results also show that substantial saving on operation costs could be achieved if the optimal solution was adopted. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

One common application of network water quality models is chlorine residual prediction. These applications typically use estimates of bulk chlorine decay kinetic parameters derived from bottle test experimental data. While various kinetic models have been developed to predict chlorine residual loss from these bottle tests, and thus in the water distribution system, previous work has not adequately addressed kinetic parameter uncertainty and consequent model prediction uncertainty. In this paper, Monte Carlo simulation is used to estimate chlorine kinetic parameter uncertainty and propagate the resulting uncertainty to network chlorine residual predictions. The consequences of choices regarding the type and quantity of chlorine bottle test experimental data and the type of chlorine kinetic model are also explored. Two kinetic models are investigated for the bulk decay of chlorine residuals: the widely used first-order model, and a second-order model with respect to chlorine and fictive reactant. Results show that the second-order model can perform better than the first-order model under rechlorination conditions performed in laboratory tests, but the difference in model predictions within the water distribution system depends on residence time, mixing and rechlorination options. Moreover, the parameter estimates for the second-order model can be highly uncertain when using single dose bottle tests data. This uncertainty is reduced when parameters are estimated using data from multi-dose experiments. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Early warning detection systems are currently being utilized to reduce the vulnerability of water distribution networks against accidental and intentional physical damage and/or biological and chemical contamination events. Following the identification of a damaging event or presence of a contaminant, a variety of response actions must be examined to successfully identify the most beneficial consequence management strategy. Multi-objective combinatorial optimization techniques can be employed to determine the cost/benefit ratio of reducing impacts to the network from physical damage or contamination by isolating and/or flushing system components. This provides water distribution system operators with additional information to remediate damage or contaminated sections. In this current effort, we evaluate employment of a genetic algorithm to minimize contaminant concentrations in a network while minimizing the cost of demand alteration. Realistic response delays are also considered. Application of this technique to a relatively simple network demonstrates the usefulness of this optimization method as consequence management strategy to reduce contaminant concentration. The genetic algorithm was able to determine the optimal demand in order to reduce the total network contaminant concentration following sensor detection. The nodal pair which resulted in the optimal demand is node 11 and node 31. The total contaminant concentration was reduced by 61.6% with a total demand change of 50%. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In the USA, about 90% of drinking water home plumbing systems use copper pipes. Pinhole leaks in copper plumbing pipes have become a concern nationwide. For majority of homeowners, their home is the most valuable asset. The possibility of losing home insurance, falling home value and health concerns have caused considerable anxiety among the affected homeowners. They are seeking a tool that would enable them to decide on whether to continue to repair or replace their plumbing system. They are also looking for advice on what factors are to be considered in a replacement decision. Plastic pipes such as PEX (Cross-linked polyethylene) and CPVC (Chlorinated Polyvinyl Chloride) are used along with the copper. Stainless steel is being considered but does not have a significant market share. In a focus group study, we elicited consumer preferences towards a variety of attributes associated with a plumbing system. We formalized the attribute ranking within the frame work of the Analytic Hierarchy Process (AHP). Multiple attributes including price, corrosion resistance, fire retardance, health effects, longevity, re-sale value of home, and taste and odor are considered. Participants' overall preference tradeoffs are reported in this paper. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

A common practice in water distribution systems management is to classify its problems into three hierarchical levels: (1) layout (analysis of system connectivity/topology); (2) design (system sizing given a layout); and (3) operation (system operation given a design), with reliability/risk/water quality considerations inherently related to all three. In most of the cases in reality the layout, design, and operation phases are solved separately with an implicit assumption that if certain constraints are fulfilled at a specific phase, then at the next stage the system will prevail (i.e., will be able to accomplish its tasks). This is the case at the design stage in which selected loading conditions are used instead of the full extended period demand cycle: it is implicitly assumed that if a system is designed for a number of chosen loading conditions (e.g., peak, low, average), then it will function properly for any other loading conditions sequence. This hypothesis is explored in this manuscript through comparing different designs with different loading conditions, to using the full extended period demand cycle. The Cross Entropy (CE) method for optimal design of water distribution systems, developed recently by the authors, is used as the optimization tool. Two example applications are explored. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The characteristics of disinfection by-products (DBPs) formation in the distribution system of Hiroshima City were investigated. Then a model was developed for predicting DBFs based on the formation kinetics. In previous studies, many factors, such as water temperature, pH, amount of chlorine residual (Cl₂), type and amount of organic substances, and residence time were confirmed to affect DBP formation experimentally. These factors, except for Cl₂ concentration, were observed to have a significant influence on DBP formation in the real distribution system. However, there was not a distinct relationship between DBPs and Cl₂ concentrations at the sampling location using a reaction time of over 30 hr. We attempt to predict DBPs in the distribution system from considerations of the above DBP formation kinetics that follow; DBPs = DBPs-eq x (1–exp(–k x (t + c))) Here, DBPs-eq is the DBP concentration at equilibrium, k is the parameter of the DBP formation rate after the effluent is in the distribution reservoir (after approximately 10 hr), t is the residence time in the water distribution system, and c is the parameter of the residence time based on the DBP formation pattern in the distribution system. The equation with the input variables of water temperature and residence time was developed from the relationship between the coefficients in the equation and the factors mentioned above. The ratios of the predicted DBP concentrations to the observed DBP concentrations in the case of dichloroacetic acid and trichloroacetic acid were 0.79 (0.60–1.23) and 0.82 (0.73–1.04), respectively. The model is helpful for managing the DBFs in water distribution systems. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Consumption of potable water is essential/or human survival. However, water distribution systems are vulnerable to a wide variety of contaminants which may enter a system accidentally or even by deliberate injection. Ingestion of this contaminated water can result in the risk of illness on the part of affected consumers and result in a waterborne outbreak. The outbreak will result in economic losses from disease-related mortality and morbidity; these losses consist of medical costs, lost productivity, and lost lives. These costs will vary with the characteristics of the outbreak, including the etiological agent involved, the number of people affected, including any sensitive subpopulations involved and the spatial and temporal extent of the outbreak. A simple method to calculate these economic costs is presented in this paper and used to estimate the medical costs of the Salmonella typhimurium outbreak in Gideon, Missouri, in 1993. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Leakage of water distribution systems can be reduced by rehabilitation or replacement of deteriorated water pipes. Rehabilitation or replacement planning should be established based on the evaluation of deterioration degree of pipes. In this study, advanced evaluation model of water main was proposed for decision making of replacements and rehabilitations. The methodology employed in this study is presented in Figure 1. The field investigation was performed to collect and analyze the influential factors on deterioration, such as pipe age, diameter, maintenance record, leakage history, pipe materials, and soil type and so on. All factors that have been used in conventional models were considered and selected to find out the determination stage of influential factors on the deterioration of water pipelines. The way to find those factors can be divided into two categories as direct and indirect method. Direct method for the evaluation of the deterioration degree of water pipeline requires excavation at laying sites to collect pre-determined factors. While, deterioration degree can be guessed or predicted roughly by indirect method using pipe history, soil condition data, and so on. Factors can be modified and corrected by the field inspectors. Also, experts in the area of pipeline management works can judge the deterioration degree of pipeline on their old experiences and accumulated know-how. Representative effectiveness of study area was analyzed with statistical experimental method to get confidence of surveying sites. The selected pipes were consisted of 3 types; renewal cast iron pipe (CIP), epoxy steel, stainless steel (STS) and polyethylene power fused pipe (PFP). The proposed deterioration models were validated against the direct measuring factors and the correlation coefficient (CC) was 0.74. The estimated 95% confidence interval and estimated interval for all pipe materials are presented and the results shown that the validation result of renewal CIP was slightly higher than the model for all pipe material (CC: 0.78). The model of epoxy steel pipe was developed for 10 field sample pipes and the correlation coefficient was 0.5. This result could be thought that the selected samples and the model factors have different characteristics. The validation result of STS and PFP was very poor and there were very low correlation between the direct factors and indirect factors. The validation results indicated that the proposed model is applicable to newel CIP but not for epoxy steel pipe and STS. For the rehabilitation plan of epoxy steel pipe, the leakage and breaking record would be more appropriate factors to evaluate the pipe condition. The proposed modeling approach can give decision makers reasonable results to establish a plan of rehabilitation and replacement of deteriorated water pipes. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In recent years, a number of field studies on the effect of pressure on leakage in distribution systems have shown that the leakage exponent can vary between 0.5 (the theoretical value for an orifice) and around 2.8. The median value is approximately 1.15. Various factors can play a role in this behavior including leak hydraulics, pipe material behavior, soil hydraulics and water demand. This paper will report results of an experimental study into the behavior of small circular holes in plastic pipes. Only turbulent flows (Reynolds number > 4000) were considered. Leak flow rates were measured at different pressures and the data used to estimate the leakage coefficients and exponents. Variables in the study included pipe material (uPVC and HDPE), hole diameter (1 and 2 mm), surrounding media (air, water and glass beads) and short term pressure fluctuations (present or not). Major findings include a clear transition between tube and orifice flow for large (wall thickness)/(hole diameter) ratios, leakage exponents very close to 0.5 under ideal conditions, a significant effect on the flow behavior when soil and water are present, and negligible effects of pressure fluctuations and pipe material. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Water distribution network sectorization projects, in process in Mexican cities in the last two decades, consist in dividing the large interconnected city distribution network in smaller networks with one (or two in exceptional cases) supply points. Known water distribution network models are routinely applied to revise any proposed sectorization. For large network sectorization projects, however, additional algorithmic capabilities are needed, such as connectivity and source contribution analysis. This paper presents algorithms of that type, based on graph theory, for obtaining the number of independent sectors in a network layout, the set of nodes belonging to each sector, the set of disconnected nodes, and the source to node contribution. The algorithms are implemented in a computer AutoCAD-based system. Real sectorization project in two cities, where the proposed algorithms are applied, are then commented. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The drinking water distribution system is vulnerable to deliberate or accidental back flow events. The large number of agents that could be potential threats and the innumerable access points make this a difficult scenario to defend against. This was clearly stated in a GAO report to Congress that listed the vulnerability of the distribution system to attack as the largest security risk to water supplies. A system designed to address the problem of distribution system monitoring is described here. The developed system employs an array of common analytical instrumentation, such as pH and chlorine monitors, coupled with advanced interpretive algorithms housed in an event monitor to provide detection/identification-response networks that are capable of enhancing system security. A variety of testing protocols were used to verify the efficacy of the system. Data obtained from a Battelle/EPA ETV study and a cooperative research and development agreement (CRADA) between Hach HST the EPA Office of Research and Development addresses issues such as long-term deployment and ability to detect and characterize contaminants. Information obtained from test loop studies carried out by Hach HST, the US Army Corp of Engineers Research Lab, and the Edgewood Biological and Chemical Command as the result of a 3-way CRADA demonstrate data collected when the system is exposed to actual warfare agents. The system is shown to be a practical measure to help detect and characterize backflow events. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Leakage is a problem that draws worldwide attention. In the United States and the United Kingdom leakage ranges from 10 to 30 %. In developing countries leakage may reach even 70 %. In international perspective, leakage in the Netherlands is low: approximately 3 to 7 % distribution input. This low leakage draws international attention. Recently, a research project has been completed where experiences from British and Dutch water companies have been exchanged. Part of this research was the application in the Netherlands of UK leakage measuring methods. Also a leak detection survey has been performed by UK staff. Both the measurements and the leak detection surveys confirmed the low leakage levels in the Netherlands. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

In water distribution systems, water is transported via underground pipes normally made of cast iron or, more recently, PVC or polyethylene. Breaks in pipes are caused by a number of factors — movement of surrounding soil (e.g. freeze/thaw events), pressure surges or as a result of poor installation(joint leaks etc.). For assessing leakage levels in a network, implementing pressure control or locating leaks it is important to understand the nature of leakage and how it varies with pressure. It is normally assumed that the leakage characteristic for a leak (the relationship between leakage flow and water pressure) follows a square root power law as predicted by physical principles. However, recent experiments show that the leakage characteristic can exhibit a higher power law depending on pipe material, defect type or the surrounding soil. The effects of pipe material and defect type on the leakage characteristic is the subject of a lot of current research and publications are routinely appearing. This paper purely addresses the effect that soil has on the leakage characteristic and will supplement the work describing the other factors. The true behaviour is obviously a combination of all these things and will ultimately be determined by pulling them all together. The research in this paper considers the idealised case of leakage from a circular orifice modelled using the Darcy-Weisbach equation and a two-dimensional model of soil, represented by the standard diffusion equation. Combining these two equations into one general steady state model allows the effects of soil diffusibility on the pipe leakage characteristic to be determined. This is shown in terms of the flow discharge over varying pressure for several different soil diffusibilities. The validity of the model has been confirmed with practical experiments performed on a test facility. A large box, with holes drilled into the side to allow drainage, was filled with different types of compacted soil and a leaking pipe placed in the centre. The water pressure in the pipe was increased from 10m–50m and the corresponding steady state flow measured (once soil was saturated). The main result of the work is that for high soil diffusibility (a sandy soil, for example) then the leakage characteristic is similar to that for discharge to air — i.e. a square root law. As the diffusibility of the soil decreases, then the exponent increases and the overall discharge level drops. This pattern is most pronounced for very heavy soil such as clay where the exponent approaches one and the flow is very low. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

The problem of Water Distribution System (WDS) rehabilitation is formulated here as a multiobjective optimisation problem under uncertain demands. The objectives are to minimise the total rehabilitation cost and to maximise the hydraulic reliability of the WDS. In this context, reliability is defined as a probability of simultaneously satisfying the minimum pressure head constraints at all nodes in the network. The total rehabilitation cost takes into account, not just the structural costs, but also the revenue lost due to the volume of water required by users but not supplied to them because of the structural inadequacy of the network. On this basis, safeguarding the hydraulic reliability of a WDS also implies a reasonable rate of return on the water company's capital investment. A probabilistic approach is used to characterise demand uncertainties within the optimisation model. The relevant probability density functions and their parameters are estimated through an experimental study conducted on a real-life WDS. The rNSGAII optimisation algorithm is used to solve the optimisation problem. The lost revenue estimation is first performed as a part of the post-optimisation procedure, and then as an integral part of the optimisation process. The results obtained in the two approaches are compared showing the better performance of the integrated type approach. The methodology presented allows the identification of the specific optimal solution of the Pareto front that corresponds to the minimal total rehabilitation cost (structural cost plus lost revenue) yet corresponding to an high level of reliability, — the Economic Level of Reliability (ELR). This value could be assumed as the optimal solution or, allowing for socio-economic considerations (if higher WDS performance standards are required), as a lower threshold of reliability which restricts the range of technically feasible solutions that could be adopted in a rehabilitation programme. This way, the proposed methodology contributes to the definition of the reliability threshold values with which studying or designing a WDS. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

It is an important task to real-time monitor the water quality in the water distribution networks. But it is difficult to select the representative locations from the whole water distribution networks because of the dynamical hydraulic status in the water distribution networks. The Lagrange method was used to analysis the water dynamic quality path of every node in every water quality time step during the water quality simulation in the networks. An objective function based on the node demand factors and the water dynamic quality path was represented to evaluate the water quality importance of each node in every water quality time step. So the optimal nodes that adapted to be fixed the sensors were selected from the networks. Fast Fourier Transformation and convolution method were used to analysis the signals collected by the water quality on-line sensors. The contamination source can be traced because the sensors impulse reaction matrix can be defined by the water dynamic path analysis. Dynamic water qualify path was calculated out in an example water distribution networks. The optimal water quality sensors' locations were defined by the water quality path calculation and the contamination sources were distinguished in a hypothetical condition through the sensors' signals analysis. It shows that the water dynamic path analysis based on the dynamic node percent information is a useful tool to optimize the water quality sensors locations and the contamination source identification. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Federal research initiatives and commercial products have shown the viability of using basic bulk parameter monitoring techniques coupled with advanced algorithms to enhance security in water distribution systems. These systems have been shown to be effective in detecting and classifying system incursions using advanced chemometrics. When classic on-line monitoring techniques are used as the sensors providing the data for the algorithms, there remain certain problems when deploying such units infield locations. There are constraints as to size, disposal of waste streams produced, reagent replacement, etc. The development of a new sensor suite that can be installed directly into the pipe is described. This system makes use of modified environmental data sondes to monitor the required parameters and provide data to the algorithms. While less sensitive than the traditional methods the low cost and ease of deployment of these systems offers a wider choice in site locations. A network approach can be utilized to optimize security by using the more traditional methods where applicable and the sonde based method in less optimum locations. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Todini (2000) defined the Resilience Index as the relation between the power per unit weight dissipated by the WDN and the power per unit weight available to be dissipated. The first term is the power per unit weight spent in form of friction and minor losses, by the configuration of pipes, to give the service consumed in each node of the WDN inform of pressure and demand. The second term is the power per unit weight given to the system by the reservoir (s) and/or the pump (s) minus the optimal one, which corresponds to the case of minimum pressure in each node giving the demanded discharge. The main objective of this research work is to look for that power per unit of weight dissipated by the WDN which would be equal to the power available by means of replacement of certain pipes in the WDN. In this way the reliability degree of the WDN is increased as well as the capacity of the WDN to transport the water under minimum conditions of pressure and demand. To obtain this in an optimal form, one must affect the power per unit of weight that is consumed by the WDN during the process; this power is the one dissipated by hydraulic friction and minor losses inside the pipes. The developed methodology can select, step by step, which single pipeline should be replaced to obtain the maximum increase in the Resilience Index. In doing so, the pressure at the entrance to the network can be reduced, increasing the pressure uniformity and decreasing water leaks. The process must stop either when there is a water quality problem due to the increase in retention time or when the replacement costs are greater than the savings due to leaks reduction. In that way the operator can optimize its infrastructure replacement investments. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).

Identifying how much water is being lost from water networks and where the losses are occurring is of great importance to water utilities both for operational and planning reasons as well as for reputation. In this paper, an optimization-based approach is presented for simultaneously quantifying and locating water losses via the process of hydraulic model calibration. The model calibration is formulated as a nonlinear optimization problem that is solved by using a genetic algorithm. The method is developed as an integrated framework of hydraulic simulation and optimization modeling. Case studies are presented to demonstrate how the integrated approach is applied to water loss detection. The results obtained show that the method is effective at detecting water loss as part of the hydraulic calibration of the network model. The accuracy of water loss detection is dependent on the quality of the field observed data and model granularity. However, it has been shown that the approach can be used for reducing the uncertainty of the water loss identification by locating water loss hotspots, which could lead to improved operating revenues at water utilities. This paper was presented at the 8th Annual Water Distribution Systems Analysis Symposium which was held with the generous support of Awwa Research Foundation (AwwaRF).