Search

Using a system to promptly detect anomalous water quality levels in a water distribution system (WDS) is a critical task to ensure security of a public water supply. Using continuous monitoring stations placed at strategic locations throughout a WDS has shown to be an effective method to detect potential contamination or low water quality; however, the performance of these monitoring stations is highly sensitive to the specific locations at which they are placed throughout the network. As a result, a large amount of research has explored how to determine the locations at which to place monitoring stations in a WDS, which may be composed of tens of thousands of junctions and pipes. These studies have typically used explicit simulations of network hydraulics, and contamination events imposed on a water distribution system, to compare how effectively a network of monitoring stations detects simulated contamination events. Building off these previous studies, the work herein proposes a framework to place fixed monitoring stations and input inline mobile sensors to best detect contamination events under uncertain water quality conditions. An adaptive-noisy-multiobjective-messy genetic algorithm is used to efficiently determine the locations at which to place monitoring stations in two sample water distribution systems for minimum cost. Results show that monitoring stations and sensor networks designed within a demand uncertain framework outperform the solutions designed in a deterministic demand framework when evaluated under more realistic demand uncertain conditions. [ABSTRACT FROM AUTHOR]

This paper focuses on evaluating a scenario-based multiobjective evolutionary algorithm for real-world design problems in which the environment where a system will operate is dynamic, and uncertain. Subsequently, the performance of a stochastic scenario selection scheme, inspired by methods to reduce overfitting in genetic programming, is investigated for scenario-based optimization. Using a scenario-based scheme to address uncertainty in a real-world system's operational environment, system designs are developed via aggregating the performance of a solution evaluated across many scenarios. Within each generation of the evolutionary algorithm the evaluation suite is resampled and evaluated by the current generation's solutions. This scheme is evaluated on two historical noisy test problems and two real-world water resources design problem instances. For each case, the stochastic scenario selection scheme is compared to a static selection scheme at various evaluation suite sizes. Results show the proposed scenario selection scheme to outperform static sampling schemes and increase efficiency of a multiobjective evolutionary algorithm for robust optimization objectives. [ABSTRACT FROM AUTHOR]

Incorporating a system of monitoring stations to insure high quality water is being delivered to consumers has been acknowledged a crucial component required by any public water distribution system (WDS). Extensive studies have acknowledged the risk posed to large populations by an accidental or intentional contamination intrusion within a WDS; failure of an early warning system (EWS) to report a contamination event carries profound economic and public health consequences. Dynamic, stochastic conditions exist in municipal WDSs and a monitoring system needs to be designed according to a robust protocol that incorporates the inherent uncertainty in WDS operation, including: demand variability, and contamination event characteristic variability. This work composes the problem of locating the best junctions within a WDS to place fixed monitoring stations, and the best junctions to input innovative inline mobile sensors, in a multi-objective framework that incorporates uncertainty in the network's demands and EWS operation. Mobile sensors are carried by flow within pipes sampling and monitoring water quality in real time, and wirelessly uploading data to fixed transceiver beacons, providing an implicit preference towards demand dense regions. A multi-objective noisy messy genetic algorithm is structured to the problem at hand and employed on a small, medium, and large-scale model WDS to calculate near-optimal solutions from the large solutions space. This multi-objective framework provides high performing trade off (Pareto) sets comparing an EWS's system cost to numerous performance objectives incorporating non-deterministic objective functions to provide a high performing and resilient EWS. Results show a large trade off surface between the cost and the respective system's performance, with large diminishing returns. Although implementing a more expensive solution may provide little to no benefit from a traditional performance standpoint, implementing a system of higher cost can increase the systems resiliency, highlighting the importance of incorporating proper objective measures in optimization procedure. [ABSTRACT FROM AUTHOR]

Prompt detection of intentional or accidental contamination of the public water supply is vital to maintain public health in any centralized water distribution system. Being able to quickly detect a system contamination event may be the single most influential factor to reduce possible contamination fallout. Consequently, major research has explored how to best protect a water distribution system (WDS) through strategic placement of fixed water quality monitoring stations. Although fixed monitoring stations within a wireless sensor network (WSN) are robust with respect to hydraulic conditions, the stations are expensive to place, and may not provide the highest spatial and temporal resolution of contamination detection. This work sets to build the understanding of a mobile wireless sensor network (MWSN) where inline mobile sensors function within water in pipes to monitor water quality and to wirelessly transmit data to fixed transceivers in real time. Mobile sensor behavior was modeled alongside contamination simulations and the deployment of fixed and mobile sensors was together optimized to minimize the affected population prior contamination event detection constrained by a total system cost. Results show a MWSN to be highly sensitive to sensor battery life, transceiver network coverage, and total system cost. Future obstacles for implementation of a MWSN are highlighted and discussed to be address in future work. [ABSTRACT FROM AUTHOR]

Event detection is one of the current most challenging topics in water distribution systems analysis: how regular on-line hydraulic (e.g., pressure, flow) and water quality (e.g., pH, residual chlorine, turbidity) measurements at different network locations can be efficiently utilized to detect water quality contamination events. This study describes an integrated event detection model which combines multiple sensor stations data with network hydraulics. To date event detection modelling is likely limited to single sensor station location and dataset. Single sensor station models are detached from network hydraulics insights and as a result might be significantly exposed to false positive alarms. This work is aimed at decreasing this limitation through integrating local and spatial hydraulic data understanding into an event detection model. The spatial analysis complements the local event detection effort through discovering events with lower signatures by exploring the sensors mutual hydraulic influences. The unique contribution of this study is in incorporating hydraulic simulation information into the overall event detection process of spatially distributed sensors. The methodology is demonstrated on two example applications using base runs and sensitivity analyses. Results show a clear advantage of the suggested model over single-sensor event detection schemes. [ABSTRACT FROM AUTHOR]

This paper explores two applied classification models alerting for contamination events in water distribution systems. The models perform multivariate analysis of water quality online measurements for event detection. The developed models comprise an outlier detection algorithm and a following sequence analysis for the classification of events. The first model is an unsupervised minimum volume ellipsoid (MVE), which utilizes only normal operation measurements but requires calibration. The second is a supervised weighted support vector machine, which utilizes event examples and performs data-driven optimized calibration. The models were trained and tested on real water utility data with randomly simulated events that were superimposed on the original database. The models showed high accuracy and detection ability compared to previous studies. All in all, the MVE model achieved preferable results. [ABSTRACT FROM AUTHOR]

The problem of contamination event detection in water distribution systems has become one of the most challenging research topics in water distribution systems analysis. Current attempts for event detection utilize a variety of approaches including statistical, heuristics, machine learning, and optimization methods. Several existing event detection systems share a common feature in which alarms are obtained separately for each of the water quality indicators. Unifying those single alarms from different indicators is usually performed by means of simple heuristics. A salient feature of the current developed approach is using a statistically oriented model for discrete choice prediction which is estimated using the maximum likelihood method for integrating the single alarms. The discrete choice model is jointly calibrated with other components of the event detection system framework in a training data set using genetic algorithms. The fusing process of each indicator probabilities, which is left out of focus in many existing event detection system models, is confirmed to be a crucial part of the system which could be modelled by exploiting a discrete choice model for improving its performance. The developed methodology is tested on real water quality data, showing improved performances in decreasing the number of false positive alarms and in its ability to detect events with higher probabilities, compared to previous studies. [ABSTRACT FROM AUTHOR]

Abstract: The presented study features an event detection model alerting for contamination events in water distribution systems. The developed model comprises a minimum volume ellipsoid (MVE) classifier, detecting outlier measurements, and a following sequence analysis utilizing the MVE binary output, for the classification of events. The model is updated continuously and exploits a constantly growing data base. The MVE enables simultaneous analysis of the water quality parameters. The multivariate analysis explores the relations between water quality parameters and detects changes in their common patterns. The suggested model applied an un-supervised classification method, eliminates the need for simulated events examples in the classifier construction. In the absent of satisfying information regarding the influence of contamination event on the parameter measurements, eliminating the use of any assumption contributes to the model reliability and generality. The model was trained on a real water utility data, and tested on randomly simulated events that were superimposed on the original data base. The model showed high accuracy and detection ability compared to previous studies. [Copyright &y& Elsevier]

Abstract: This study describes a decision support system, alerts for contamination events in water distribution systems. The developed model comprises a weighted support vector machine (SVM) for the detection of outliers, and a following sequence analysis for the classification of contamination events. The contribution of this study is an improvement of contamination events detection ability and a multi-dimensional analysis of the data, differing from the parallel one-dimensional analysis conducted so far. The multivariate analysis examines the relationships between water quality parameters and detects changes in their mutual patterns. The weights of the SVM model accomplish two goals: blurring the difference between sizes of the two classes' data sets (as there are much more normal/regular than event time measurements), and adhering the time factor attribute by a time decay coefficient, ascribing higher importance to recent observations when classifying a time step measurement. All model parameters were determined by data driven optimization so the calibration of the model was completely autonomic. The model was trained and tested on a real water distribution system (WDS) data set with randomly simulated events superimposed on the original measurements. The model is prominent in its ability to detect events that were only partly expressed in the data (i.e., affecting only some of the measured parameters). The model showed high accuracy and better detection ability as compared to previous modeling attempts of contamination event detection. [Copyright &y& Elsevier]

This study presents a methodology for the inclusion of hydraulics uncertainty in contamination source identification. Current research normally considers the system hydraulics as deterministic and the water quality sensors as ideal. In reality however only a small portion of the hydraulic data is known and most likely only Boolean sensor information of a contamination existence. There is a need to incorporate these considerations in contamination source identification models and to explore their influence on the modelling ability to correctly detect the characteristics of a contamination intrusion. This problem is addressed in this manuscript. The proposed method is based on a previous contamination source detection model developed by the authors which is further embedded in a statistical framework for quantifying the uncertainty of a contamination source detection outcome. The methodology is demonstrated on three example applications of increasing complexity through base runs and sensitivity analyses. [ABSTRACT FROM AUTHOR]

A simple, straightforward, modified genetic algorithm scheme for contaminant source characterization using imperfect sensors is presented and demonstrated in this study. Previous work on this subject concentrated on developing source-inversion models using sensors that provide accurate, unbiased, contamination concentration measurements. The developed contamination source-detection model is implemented using three sensor types: (1) perfect sensors providing accurate, unbiased, contamination concentration measurements; (2) sensors transmitting fuzzy measured information (i.e., high, medium, and low contamination); and (3) '0-1' (Boolean) sensors indicating only a contamination presence. A comparison between the three sensor types is explored taking into consideration thesystem's response time (i.e., the time elapsed between a contaminant detection and a decision-maker's response action). The methodology capabilities are demonstrated using two example applications of increasing complexity, showing the trade-offs between the sensor types and the model abilities to receive a unique solution to the source-detection problem. [ABSTRACT FROM AUTHOR]

This study deals with the integration of contamination simulations and a spatial event detection model. The simulation of contaminant intrusion includes detailed chemical-specific reactions within a multi-species water quality model. This set-up generates a scenario of contaminant distribution and produces a continuous multiple sensor stations database. Three organophosphates pesticides, Chlorpyrifos, Malathion, and Parathion, are modeled as possible contaminants. The event detection model comprises both local and spatial data analysis. The local model applies a previously developed single-sensor event detection model with a higher alert threshold that reduces false alarm rates. The spatial model considers upstream sensor datasets which are examined for their uniqueness and mutual resemblance in a sliding time window. The model utilizes outlier detection, data analyses, and network hydraulics for the detection of suspicious spatial trends. The proposed algorithm is capable of detecting events with low contamination signatures and spatial influence. Two case studies are explored and compared to the single sensor model. The proposed methodology resulted in a lower number of false alarms compared to the previous single sensor event detection modeling approach. [ABSTRACT FROM AUTHOR]

Placement of water quality sensors in a water distribution system is a common approach for minimizing contamination intrusion risks. This study incorporates detailed chemistry of organophosphate contaminations into the problem of sensor placement and links quantitative measures of the affected population as a result of such intrusions. The suggested methodology utilizes the stoichiometry and kinetics of the reactions between organophosphate contaminants and free chlorine for predicting the number of affected consumers. This is accomplished through linking a multi-species water quality model and a statistical dose–response model. Three organophosphates (chlorpyrifos, malathion, and parathion) are tested as possible contaminants. Their corresponding by-products were modeled and accounted for in the affected consumers impact calculations. The methodology incorporates a series of randomly generated intrusion events linked to a genetic algorithm for minimizing the contaminants impact through a sensors system. Three example applications are explored for demonstrating the model capabilities through base runs and sensitivity analyses. [ABSTRACT FROM AUTHOR]

Optimal sensor placement for detecting contamination events in water distribution systems is a well explored problem in water distribution systems security. We study herein the problem of sensor placement in water networks to minimize the consumption of contaminated water prior to contamination detection. For any sensor placement, the average consumption of contaminated water prior to event detection amongst all simulated events is employed as the sensing performance metric. A branch and bound sensor placement algorithm is proposed based on greedy heuristics and convex relaxation. Compared to the state of the art results of the battle of the water sensor networks (BWSN) study, the proposed methodology demonstrated a significant performance enhancement, in particular by applying greedy heuristics to repeated sampling of random subsets of events. [ABSTRACT FROM AUTHOR]

Abstract: In this study, a dynamic thresholds scheme is developed and demonstrated for contamination event detection in water distribution systems. The developed methodology is based on a recently published article of the authors (Perelman et al., 2012). Event detection in water supply systems is aimed at disclosing abnormal hydraulic or water quality events by exploring the time series behavior of routine hydraulic (e.g., flow, pressure) and water quality measurements (e.g., residual chlorine, pH, turbidity). While event detection raises alerts to the possibility of an event occurrence, it does not relate to origins, thus an event may be hydraulically-driven, as a consequence of problems like sudden leakages or pump/pipe malfunctions. Most events, however, are related to deliberate, accidental, or natural contamination intrusions. The developed methodology herein is based on off-line and on-line stages. During the off-line stage, a genetic algorithm (GA) is utilized for tuning five decision variables: positive and negative filters, positive and negative dynamic thresholds, and window size. During the on-line stage, a recursively Bayes'' rule is invoked, employing the five decision variables, for real time on-line event detection. Using the same database, the proposed methodology is compared to Perelman et al. (2012), showing considerably improved detection ability. Metadata and the computer code are provided as Supplementary material. [Copyright &y& Elsevier]