Localizing pipe leaks is a significant challenge for water utilities worldwide. Pipe leaks in water distribution systems (WDSs) can cause the loss of a large amount of treated water, leading to pressure loss, increased energy costs, and contamination risks. What makes localizing pipe leaks challenging is the underground location of the water pipes and the similarity in impact on hydraulic properties (e.g., pressure, flow) due to leaks as compared to the effects of WDS operational changes. Physical methods to locate leaks are expensive, intrusive, and heavily localized. Computational approaches such as data-driven machine learning models provide an economical alternative to physical methods. Machine learning models are readily available and easily customizable to most problems; therefore, there is an increasing trend in their application for leak localization in WDSs. While several studies have applied machine learning models to localize leaks in single pipes and small test networks, these studies have yet to thoroughly test these models against the different complexities of leak localization problems, and hence their applicability to real-world WDSs is still unclear. The simplicity of the WDSs, the oversimplification of leak characteristics, and the lack of consideration of modeling and measuring device uncertainties adopted in most of these studies make the scalability of their proposed approaches questionable to real-world WDSs. Our study addresses this issue by devising four study cases of different complexity that account for realistic leak characteristics and model- and measuring device-related uncertainties. Two established machine learning models-multilayer perceptron (MLP) and convolutional neural network (CNN)-are trained and tested for their ability to localize the leaks and predict their sizes for each of the four study cases using different simulated hydraulic inputs. In addition, the potential benefit of combining different types of hydraulic data as inputs to the machine learning models in localizing leaks is also explored. Pressure and flow, two common hydraulic measurements, are used as inputs to the machine learning models. Further, the impact of single and multiple time point input in leak localization is also investigated. The results for the L-Town network indicate good accuracies for both the models for all study cases, with CNN consistently outperforming MLP.

Water distribution systems (WDSs) face a significant challenge in the form of pipe leaks. Pipe leaks can cause loss of a large amount of treated water, leading to pressure loss, increased energy costs, and contamination risks. Locating pipe leaks has been a constant challenge for water utilities and stakeholders due to the underground location of the pipes. Physical methods to detect leaks are expensive, intrusive, and heavily localized. Computational approaches provide an economical alternative to physical methods. Data-driven machine learning-based computational approaches have garnered growing interest in recent years to address the challenge of detecting pipe leaks in WDSs. While several studies have applied machine learning models for leak detection on single pipes and small test networks, their applicability to the real-world WDSs is unclear. Most of these studies simplify the leak characteristics and ignore modeling and measuring device uncertainties, which makes the scalability of their approaches questionable to real-world WDSs. Our study addresses this issue by devising four study cases that account for the realistic leak characteristics (multiple, multi-size, and randomly located leaks) and incorporating noise in the input data to account for the model- and measuring device- related uncertainties. A machine learning-based approach that uses simulated pressure as input to predict both location and size of leaks is proposed. Two different machine learning models: Multilayer Perceptron (MLP) and Convolutional Neural Network (CNN), are trained and tested for the four study cases, and their performances are compared. The precision and recall results for the L-Town network indicate good accuracies for both the models for all study cases, with CNN generally outperforming MLP.

The primary goal of the project is to leverage recent developments in smart water technologies to detect and reduce water leakages in large water distribution networks with the aid of neural networks. A cost effective, non-invasive solution to detect leakages in transmission pipelines is needed by many water utilities, as it will lead to significant water savings and reduced pipe breakage frequencies, especially in older infrastructure systems. The eventual goal of the project is to test the ANN model on a real network using field measured pressure and pipe breakage data after tuning and developing the model with simulated data. In this project, we propose building a regression model, based on multi-layer perceptron (MLP) algorithm, which is a class of feedforward artificial neural networks (ANNs) to detect the leak locations within a proposed network. The model should be able to learn the structure, i.e., mapping of various leak nodes and sensor nodes in an area, such that it can detect the leak nodes based on the pressure values with significant accuracy.

In many modern water networks, an emerging trend is to measure pressure at various points in the network for operational reasons. Because leaks typically induce a signature on pressure, these routine measurements can be used to develop nonintrusive leak detection approaches. This research employs successive linear approximation methods, based on linear programming and mixed integer linear programming, in a simulation-optimization framework to explore an alternative leak detection methodology for urban water distribution networks based on pressure measurements. The methods attempt to minimize the absolute differences between observed and simulated pressure values at the sensors to determine a linear combination of leaks that most closely approximates the observed pressure pattern. Steady-state and time-varying models of differing complexity (from small published networks to a 27,000-node network for a U.S. utility) were used to test the method. Results are presented to illustrate the method's effectiveness under different conditions. The methods are shown to work well when pervasive pressure data and hydraulic models representing true operational conditions are available. The methods developed in this work are not intended to replace traditional leak detection methods; rather, they are meant to work in concert with available methods to more accurately and efficiently isolate leak locations and reduce water loss. (C) 2017 American Society of Civil Engineers.

The Dividing Rectangles (DIRECT) search is a deterministic, derivative-free, global search algorithm. The algorithm searches for the global minimum by recursive space partitioning, essentially grouping similar regions within the decision space and selecting a sample from each group. The DIRECT algorithm was initially designed for continuous problems, but has since been modified to allow for integer variable types. Though DIRECT has been previously used for discrete numbers, the algorithm has not been extended to other discrete variable types, such as graph nodes or multi-dimensional points. This research further extends the DIRECT algorithm to use a mix of continuous and discrete variables, including connected graph nodes. In this paper, the algorithm is applied to leak detection problems in water distribution systems (WDSs), which involve both discrete network nodes and continuous leak magnitudes. In addition, the DIRECT algorithm is parallelized using a master-worker paradigm and tested using cluster resources for a moderate number of processors. The generalization and abstraction of the DIRECT algorithm presented in this research will enable the application of DIRECT to a wider class of problems than previously possible.

Evolution of a Flipped Engineering Economy CourseThis paper would fit into the Engineering Economy Division tracks at ASEE 2015At the ISERC 2013 conference in Puerto Rico the paper entitled "Flipped Out EngineeringEconomy: Converting a Traditional Class to Hybrid Model" was presented. That paper andpresentation focused on the history of the CE 390: Engineering Economy course at theinstitution, its conversion to a flipped model, and early student results (1 semester) of convertinga traditional lecture class to a hybrid model. The present paper describes the evolution of thehybrid model that has been implemented in this course over the last five semesters, discussingthe lessons learned and advantages and disadvantages of our approaches over this evolution.Included will be a description of changes made to the structure and pedagogy used as well as theimpact over time on student learning and assessment of the course. Statistical analysis of studentresults will be included. Faculty interested in flipping courses or incorporating hybrid techniquesin their courses should be interested in this work.

Leaks in water distribution systems waste energy and water resources, increase damage to infrastructure, and may allow contamination of potable water. This research develops an evolutionary algorithm-based approach to minimize the cost of water loss, new infrastructure, and operations that reduce background leakage. A new design approach is introduced that minimizes capital and operational costs, including energy and water loss costs. Design decisions identify a combination of infrastructure improvements, including pipe replacement and valve installment, and operation rules for tanks and pumps. Solution approaches are developed to solve both a single-objective and multiobjective problem formulation. A genetic algorithm and a nondominated sorting genetic algorithm are implemented within a high-performance computing platform to select tank sizes, pump placement and operations, placement of pressure-reducing valves, and pipe diameters for replacing pipes. The evolutionary algorithm approaches identify solutions that minimize water loss due to leakage, operational costs, and capital costs, while maintaining pressure at nodes and operational feasibility for tanks and pumps. Solutions are compared to identify a recommended design. The framework is demonstrated to redesign a water distribution system for an illustrative case study, C-Town. (C) 2015 American Society of Civil Engineers.

Regional water supply systems undergo surplus and deficit conditions due to differences in inflow characteristics as well as due to their seasonal demand patterns. This study proposes a framework for regional water management by proposing an interbasin transfer (IBT) model that uses climate‐information‐based inflow forecast for minimizing the deviations from the end‐of‐season target storage across the participating pools. Using the ensemble streamflow forecast, the IBT water allocation model was applied for two reservoir systems in the North Carolina Triangle Area. Results show that interbasin transfers initiated by the ensemble streamflow forecast could potentially improve the overall water supply reliability as the demand continues to grow in the Triangle Area. To further understand the utility of climate forecasts in facilitating IBT under different spatial correlation structures between inflows and between the initial storages of the two systems, a synthetic experiment was designed to evaluate the framework under inflow forecast having different skills. Findings from the synthetic study can be summarized as follows: (a) inflow forecasts combined with the proposed IBT optimization model provide improved allocation in comparison to the allocations obtained under the no‐transfer scenario as well as under transfers obtained with climatology; (b) spatial correlations between inflows and between initial storages among participating reservoirs could also influence the potential benefits that could be achieved through IBT; (c) IBT is particularly beneficial for systems that experience low correlations between inflows or between initial storages or on both attributes of the regional water supply system. Thus, if both infrastructure and permitting structures exist for promoting interbasin transfers, season‐ahead inflow forecasts could provide added benefits in forecasting surplus/deficit conditions among the participating pools in the regional water supply system.

This research minimizes the impact of leaks on the operation of the system to reduce lost water while meeting typical management goals. A genetic algorithm approach is implemented within a high-performance computing platform to select tank sizes, pump placement and operations, placement of pressure control valves, and pipe diameters for replacing pipes. It identifies solutions that minimize water loss, operational costs, and capital costs, while maintaining pressure at nodes and operational feasibility for tanks. Multiple problem formulations are solved that use alternative objective functions and allow varying degrees of freedom in the decision space. The methodology is demonstrated to identify a water distribution system re-design for the C-Town case study.

The Battle of the Water Networks II (BWN-II) is the latest of a series of competitions related to the design and operation of water distribution systems (WDSs) undertaken within the Water Distribution Systems Analysis (WDSA) Symposium series. The BWN-II problem specification involved a broadly defined design and operation problem for an existing network that has to be upgraded for increased future demands, and the addition of a new development area. The design decisions involved addition of new and parallel pipes, storage, operational controls for pumps and valves, and sizing of backup power supply. Design criteria involved hydraulic, water quality, reliability, and environmental performance measures. Fourteen teams participated in the Battle and presented their results at the 14th Water Distribution Systems Analysis conference in Adelaide, Australia, September 2012. This paper summarizes the approaches used by the participants and the results they obtained. Given the complexity of the BWN-II problem and the innovative methods required to deal with the multiobjective, high dimensional and computationally demanding nature of the problem, this paper represents a snap-shot of state of the art methods for the design and operation of water distribution systems. A general finding of this paper is that there is benefit in using a combination of heuristic engineering experience and sophisticated optimization algorithms when tackling complex real-world water distribution system design problems. (C) 2014 American Society of Civil Engineers.

Population based heuristic search methods such as evolutionary algorithms (EA) and particle swarm optimization (PSO) methods are widely used for solving optimization problems especially when classical techniques are inadequate. A parallel optimization framework using multiple concurrent particle swarms is developed and applied to water distribution problems. Details of the enabling framework that couples the optimization methods with a parallel simulator built around EPANET will be discussed. In addition, algorithmic and computational performance results using ORNL's and ANL's leadership class parallel architectures will be presented for leakage detection and contaminant source characterization problems for two water distribution networks with 1,834 and 12,457 nodes respectively.

This paper presents a simulation–optimization-based method for identification of contamination source characteristics in a water distribution system using filtered data from threshold-based binary water quality signals. The effects of quality and quantity of the data on the accuracy of the source identification methodology are investigated. This study also addresses the issue of non-uniqueness in contaminant source identification under various data availability conditions. To establish the robustness and applicability of the methodology, numerous scenarios are investigated for a wide range of contamination incidents associated with two different networks. Results indicate that, even though use of lower resolution sensors lead to more non-unique solutions, the true source location is always included among these solutions.

The modification of land and water resources for human use alters the natural hydrologic flow regime of a downstream receiving body of water. The natural flow regime is essential for sustaining biotic structure and equilibrium within the ecosystem. Best management practices mitigate the increased storm water runoff due to increased imperviousness and are typically designed and located within a watershed to match peak and minimum flows for a small set of targeted design storms. Ecosystems are, however, affected by all the characteristics of a long-term flow regime, including the magnitude, duration, frequency, and timing of flows. A more environmentally sustainable approach for watershed development is presented based on the minimization of differences in the characteristics of the flow regime between predevelopment and postdevelopment conditions. The indicator of hydrologic alteration (IHA) is a set of 33 hydrologic indices that characterize a flow regime and, coupled with the range of variability approach (RVA), can be used to evaluate a development strategy for its alteration of the long-term hydrologic flow regime. This paper presents a methodology to identify watershed management strategies that will have a minimal impact on the flow regime and downstream ecosystems. This methodology utilizes a metric that evaluates development strategies based on an IHA/RVA analysis implemented within a simulation-optimization framework. Continuous simulation of urban runoff for different land use strategies is enabled through the use of the storm water management model, and the resulting long-term hydrograph is analyzed using IHA/RVA. Development is allocated within subcatchments to maintain a predefined minimum level of total development while minimizing the hydrologic alteration. A hybrid optimization approach based on genetic algorithm and Nelder-Meade approaches is used to identify optimal land use allocation. Further analysis is conducted to identify alternative development patterns that allocate impervious development maximally differently among subcatchments while achieving similarly low alteration in the hydrologic flow regime.

Water distribution systems are designed for fast and efficient transport of the drinking water and mixing of chlorine to maintain the required disinfectant levels in the system. Thus, any contaminant if injected in the system would also spread quickly through the network and can have serious impact on public health if consumed. Contaminant injected during any intentional contamination event can be chemical or biological, the nature of which may remain unknown. Practically it's not possible to monitor any system for the presence of all possible chemical or biological contaminants. However, the distribution systems are routinely monitored for several water quality parameters like chlorine, pH, etc. Any contaminant injected in the system would react with water and chlorine leading to the increased degradation of the chlorine levels in the system. In our past work we developed methodology to use routine chlorine measurements as a surrogate to identify a contamination event in a WDS. An evolutionary algorithm based approach simulation-optimization was developed to identify the contaminant source characteristics (i.e., location of the contaminant source, time of start of injection and injection pattern) during a contamination event under conditions of uncertainty about the reaction kinetics of the contaminant in the system. The investigation was extended to study the source characterization problem under different uncertain reaction conditions. We present here a detailed analysis of source characterization problem for the reactive contaminants and the simulation-optimization methodology developed. Case studies carried out on a number of water distributions systems will be reported.

Water distributions systems are primary means of safe drinking water supply to the public. Water produced and delivered to the distribution system is intended for the customer. However, a significant amount of the water is lost in the distribution system before even reaching the customers. Water customers are metered for the usage at end connection but a significant portion of water produced by the utilities never passes through the meters. This leads to wastage of valuable water and loss of revenues for the utilities. The occurrence of leaks depends on the factors like material, composition, age and joining methods of the distribution systems components. Due to the complex nature and vast spatial extent of a water distribution system it may be difficult for the utility personnel to identify and fix the leaks. Traditionally, the method of inverse transient analysis (ITA) has been used by the researchers for identifying the leaks in a distribution system. While transient analysis is an efficient method for leak detection, it often requires that a series of hydraulic transients (or pressure pulses) be injected into the system in order to detect the leaks (e.g., controlled opening/closing of a fire hydrant). In contrast to ITA, this work attempts to use routinely measured water quality and pressure measurements for the detection of leaks. Distribution systems are routinely monitored for several water quality parameters such as Chlorine, pH, and turbidity. Water loss due to any leaks present in the system would impact the flow characteristics of the system and would have an impact on the water quality. In this study a methodology has been developed to use the water quality data along with available pressure measurements for the improved detection of leaks in a water distribution system. Leak detection is formulated as an inverse problem and solved using a simulation-optimization approach.

Accurate knowledge of the characteristics of the contamination source during a contamination event is necessary for development of any mitigation and control strategy. Contaminant injected in a system is most likely to be reactive with chlorine; however, it is impractical for water quality monitoring systems to be able to monitor for the presence of all possible contaminants. In any distribution system, chlorine levels and other water quality parameters (pH, conductance, etc.) are routinely monitored to maintain the prescribed disinfection capacity. Any reactive contaminant would affect the chlorine levels resulting in deviations in the expected chlorine levels from those expected under normal operating conditions. Anomalies in the chlorine concentration from that of the expected value can be used as a surrogate to characterize the contaminant source in the system. In the absence of knowing the reactive characteristics of the contaminants, the location of injection, and injection pattern, source identification becomes a difficult problem to solve. Source identification can be posed as an inverse problem. In earlier work authors investigated the effect of the order of reaction kinetics of the contaminant with chlorine and its impact on source identification problem assuming the reaction kinetics to be known. That work is extended to investigate a methodology to address the source identification problem based on chlorine measurements, and the effects of different uncertain contamination conditions. Findings from a range of scenarios will be presented and discussed.

The rapid discovery of the contaminant source and its mass loading characteristics in a water distribution system (WDS) is vital for generating an efficient control strategy during a contamination event. Previous work on the Adaptive Dynamic Optimization Technique (ADOPT), which was developed as an Evolution Strategy (ES) based procedure, presents an approach to estimate the source characteristics adaptively, given dynamically updated observation data. Although this simulation-optimization approach is promising, it is computationally expensive, which poses challenges in the context of real-time solutions. This paper reports the findings of an investigation that builds upon the prior work by introducing a hybrid heuristic search method for the real-time characterization of a contaminant source. This new method integrates the ES-based ADOPT with a logistic regression (LR) analysis and a local improvement method to expedite the convergence and possibly solve the problem quickly. As a prescreening technique, a LR analysis step is performed prior to ADOPT; this step reduces the search space by eliminating unnecessary source nodes as potential source locations. Then, a local search (LS) approach is embedded into some of the algorithmic steps in ADOPT to serve as a postscreening step that potentially speeds up the convergence in localized regions in the solution space. Numerical experiments for the proposed hybrid approach are performed on an example water distribution network, and the results are compared with those of the standard implementation of ADOPT.

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).