S. R. Mounce

Development and Verification of an Online Artificial Intelligence System for Detection of Bursts and Other Abnormal Flows

Water lost through leakage from water distribution networks is often appreciable. As pressure increases on water resources, there is a growing emphasis for water service providers to minimize this loss. The objective of the work presented in this paper was to assess the online application and resulting benefits of an artificial intelligence system for detection of leaks/bursts at district meter area (DMA) level. An artificial neural network model, a mixture density network, was trained using a continually updated historic database that constructed a probability density model of the future flow profile. A fuzzy inference system was used for classification; it compared latest observed flow values with predicted flows over time windows such that in the event of abnormal flow conditions alerts are generated. From the probability density functions of predicted flows, the fuzzy inference system provides confidence intervals associated with each detection, these confidence values provide useful information for f...

Sensor-fusion of hydraulic data for burst detection and location in a treated water distribution system

Abstract This paper presents research into analysis and data fusion for sensors measuring hydraulic parameters (flow and pressure) of the pipeline water flow in treated water distribution systems. An artificial neural network (ANN) based system is used on time series data produced by sensors to construct an empirical model for the prediction and classification of leaks. A rules based system performs a fusion on the ANNs’ outputs to produce an overall state classification for a set of zones. Results are presented using data from an experimental site in a distribution system of a UK water company in which bursts were simulated by hydrant flushing. The ANN system successfully detected events and a study of the pressure gradient across the zone provided a more precise location within the zone.

Burst detection using hydraulic data from water distribution systems with artificial neural networks

This paper presents research into the application of artificial neural networks (ANNs) for analysis of data from sensors measuring hydraulic parameters (flow and pressure) of the water flow in treated water distribution systems. Two neural architectures (static and time delay) are applied for time series pattern classification from the perspective of detecting leakage. Results are presented using data from an experimental site in a distribution system of a UK water company in which bursts were simulated by hydrant flushing. Field trials have shown how ANNs can be used effectively for a leakage detection task. Both static and time delay ANNs learned patterns of leaks/bursts. The time delay neural network showed improved performance over the static network. It is concluded that the effectiveness of an ANN in discovering relationships within the data is dependent upon two key factors: availability of sufficient exemplars and data quality.

Field testing of an optimal sensor placement methodology for event detection in an urban water distribution network

This paper presents a method to identify ‘optimal’ locations of pressure sensor instruments for the detection of leak/burst events and the results of a set of field trials conducted to evaluate the approach. The identification method is based on complete enumeration studies using hydraulic model simulations of a wide range of burst events and evaluating the response to each event at all possible monitoring points. The field trials simulated leak/burst events through the opening of fire hydrants within a selected District Metered Area (DMA), five different hydrants were opened systematically in the DMA to simulate different leak/burst events. By installing pressure instrumentation at different locations in the DMA, an understanding of how accurately the model methodology can determine sensitivity of instrument location can be obtained. Prior to and during the hydrant openings pressure data was collected at eight different instrument locations within the DMA. These pressure instruments were installed to cover different model predicted sensitivities and to provide good spatial coverage. The results show that pressure instrumentation location is crucial to sensitivity and that the modelling methodology is able to predict instrument location sensitivity to leak/burst events and thus offer an improvement over current industry practice for instrument deployment. It should be noted that this field application made use of current UK standard models, with no additional calibration or updating.

Development and Field Validation of a Burst Localization Methodology

Reducing water loss through bursts is a major challenge throughout the developed and developing world. Currently burst lifetimes are often long because awareness and location of them is time- and labor-intensive. Advances that can reduce these periods will lead to improved leakage performance, customer service, and reduce resource wastage. In water-distribution systems the sensitivity of a pressure instrument to change, including burst events, is greatly influenced by its own location and that of the event within the network. A method is described here that utilizes hydraulic-model simulations to determine the sensitivity of potential pressure-instrument locations by sequentially applying leaks to all potential burst locations. The simulation results are used to populate a Jacobian matrix, quantifying the different sensitivities. This matrix may then be searched to identify different instrument locations to achieve required goals: maximising overall sensitivity to all potential events or selective sensitivity to events in different network areas. It is shown here that by searching this matrix to optimize such selective sensitivity, while minimising instrument numbers, it is possible to provide useful burst-localization information. Results are presented from field studies that demonstrate the practical application of the method, showing that current standard network models can provide sufficiently accurate quantification of differential sensitivities and that, once combined with event-detection techniques for data analysis, events can effectively be localized using a small number of instruments.

Predicting combined sewer overflows chamber depth using artificial neural networks with rainfall radar data

Combined sewer overflows (CSOs) represent a common feature in combined urban drainage systems and are used to discharge excess water to the environment during heavy storms. To better understand the performance of CSOs, the UK water industry has installed a large number of monitoring systems that provide data for these assets. This paper presents research into the prediction of the hydraulic performance of CSOs using artificial neural networks (ANN) as an alternative to hydraulic models. Previous work has explored using an ANN model for the prediction of chamber depth using time series for depth and rain gauge data. Rainfall intensity data that can be provided by rainfall radar devices can be used to improve on this approach. Results are presented using real data from a CSO for a catchment in the North of England, UK. An ANN model trained with the pseudo-inverse rule was shown to be capable of predicting CSO depth with less than 5% error for predictions more than 1 hour ahead for unseen data. Such predictive approaches are important to the future management of combined sewer systems.

Linking distribution system water quality issues to possible causes via hydraulic pathways

Our limited understanding and quantification of the variety and complexity of chemical, physical and biological reactions and interactions occurring within drinking water distribution systems currently prohibit the development of a deterministic model of water quality. The causes of known water quality anomalies can however be investigated through mining the large volumes of water quality, hydraulic and asset data currently being collected by utility companies. The data-driven methodology described here permits historical cause-effect linkages to be identified in a scalable, largely automatable fashion. Under Distribution System Integrated Modelling (DSIM), spatio-temporal searches within the set of pipes that typically lie upstream of a known water quality anomaly are used to identify possible causes. Understanding of the flow paths that connect causes and effects are derived from the results of hydraulic network simulations. DSIM was used to investigate contacts regarding discolouration and smell/taste issues from customers within a Water Supply Zone in England, UK, over a six-year period. 17.6% of discolouration issues and 17.4% of smell/taste issues were linked to maintenance jobs using the methodology, much smaller proportions than were identified using radial cause searches. The DSIM search results contained a greater proportion of one-to-one linkages and so are less ambiguous than the results of the radial spatio-temporal searches. DSIM was found to be a useful and informative tool for data mining multiple water quality related datasets.

ONLINE APPLICATION OF ANN AND FUZZY LOGIC SYSTEM FOR BURST DETECTION

Minimising the loss of treated water from water supply systems due to burst and leakage is an ongoing issue for water service providers around the world. Flow monitoring techniques are currently used by the water industry to monitor leakage, generally offline through the application of mass balance type calculations or through observations of change in night line values. The data for such analysis has, until recently, been at best collected 24 hourly via SMS technology. The objective of the study reported here was to assess the online application of an AI system to a real distribution system and the potential benefits of so doing. Specifically the application of Artificial Neural Networks (ANNs) and Fuzzy Inference Systems (FIS), which are computational techniques in the field of Artificial Intelligence (AI). The online hybrid ANN/FIS system developed uniquely uses DMA (District Meter Areas) level flow data for the detection of leaks/bursts as they occur. The ANN model (a Mixture Density Network) was trained using a continually updated historic database that constructed a probability density model of the future flow profile. A FIS, used for classification, compared observed flows with the probability density function of predicted flows over time windows such that confidence intervals could be assigned to alerts and further, an accurate estimate of likely burst size provided. A Water Supply System in the UK was used for the case study. The case study pilot area has near real-time flow data provided by General Packet Radio Service (GPRS). The online AI leak/burst detection system was constructed to operate along side an existing flat line alarm system, and continuously analyse every twelve hours a set of 50 DMAs of various size, complexity and connectivity within the case study area. Results are presented from a six month period. The new system identified a number of events and alerts were raised prior to their notification in the control room; either through flat line alarms or customer contacts. Examples are given of their correlation with burst reports and subsequent mains repairs. 56% of AI alerts were found to correspond to bursts confirmed by repair data or customer contacts reporting bursts. The study shows that the integration of the AI system with near real time communications can facilitate rapid determination (i.e. before customers are impact) of abnormal flow patterns. It is concluded from the study that the system is an effective and viable tool for online burst detection in water distribution systems.

Automated Data-Driven Approaches to Evaluating and Interpreting Water Quality Time Series Data from Water Distribution Systems

AbstractWater distribution networks are not inert transport systems. The high-quality water produced at water treatment works is subject to a variety of complex and interacting physical, chemical, and biological interactions within these highly variable, high-surface reactors. In particular, the aging and deteriorating asset condition in water distribution systems can result in a degradation of water quality delivered to the customer, often experienced as discoloration caused by increasing amounts of fine particulate matter. Here, it is proposed that by assessing measured turbidity over time, in particular its correlation with local hydraulics, an assessment of change in risk of fouling can be obtained and asset deterioration inferred. This paper presents a methodology for pairwise monitoring of a hydraulic parameter (flow or pressure) and turbidity using wavelet-based semblance analysis—a novel methodology from another domain, which is applied for the first time to water quality data in distribution syst...

FIELD VALIDATION OF 'OPTIMAL' INSTRUMENTATION METHODOLOGY FOR BURST/LEAK DETECTION AND LOCATION

Leakage is a commonly accepted feature of Water Distribution Systems (WDS). The UK claims to be at, or close to, an Economic Level of Leakage (ELL), where the value of water lost is marginal to the cost of finding and fixing leaks. However ELL is not necessarily fixed. Drivers such as energy, carbon and climate change are forcing re-evaluation of the economics and sustainability drivers are requiring inclusion of social (public acceptance, levels of service) and environmental (natural resources) ‘costs’. Hence there is a need for water companies to improve their awareness and location of bursts/leaks as they occur. This paper presents a methodology to determine, and the results of fieldwork to validate, the ‘optimal’ placement of pressure instruments within Distribution Management Areas /District Meter Areas (DMAs) to detect and locate new leak/burst events. While flow data is generally regarded as a superior indicator of leak/burst events, pressure instruments have the advantage of lower cost and ease of deployment. Hence it is feasible to justify the deployment of additional pressure instrumentation to improve leak/burst detection and location. However the placement of instrumentation remains a challenge. The methodology presented here is based upon complete enumeration studies using hydraulic model simulations of multiple leak/burst events and then evaluating the responses at all potential instrumentation points. This produces a sensitivity matrix of all possible instrument locations to all possible leak/burst events. The matrix can then be searched for combinations of instruments that provide sensitivities to different regions (discrete or selectively overlapping) to enable detection and location of new events. Results of extensive fieldwork are presented to validate the proposed modeling methodology. This fieldwork utilized fire hydrant opening and flushing to simulate the additional system flow of leak/burst events. Data presented confirms the ability of the approach to predict the instrument sensitivities across numerous points within a selected DMA, as well as the ability to provide location information. Results are from current UK industry standard data sampling and hydraulic models confirming the practicality of the method.