Bayesian Optimization of Booster Disinfection Scheduling in Water Distribution Networks.

Chlorine remains the most widely used disinfectant in drinking water treatment and distribution systems worldwide. To maintain a minimum residual throughout the distribution network, chlorine dosage needs to be regulated by optimizing the locations of chlorine boosters and their scheduling (i.e., chlorine injection rates). Such optimization can be computationally expensive since it requires numerous evaluations of water quality (WQ) simulation models. In recent years, Bayesian optimization (BO) has garnered considerable attention due to its efficiency in optimizing black-box functions in a wide range of applications. This study presents the first attempt to implement BO for the optimization of WQ in water distribution networks. The developed python-based framework couples BO with EPANET-MSX to optimize the scheduling of chlorine sources, while ensuring the delivery of water that satisfies water quality standards. Using Gaussian process regression to build the BO surrogate model, a comprehensive analysis was conducted to evaluate the performance of different BO methods. To that end, systematic testing of different acquisition functions, including the probability of improvement, expected improvement, upper confidence bound, and entropy search, in conjunction with different covariance kernels, including Matérn, squared-exponential, gamma-exponential, and rational quadratic, was conducted. Additionally, a thorough sensitivity analysis was performed to understand the influence of different BO parameters, including the number of initial points, covariance kernel length scale, and the level of exploration vs exploitation. The results revealed substantial variability in the performance of different BO methods and showed that the choice of the acquisition function has a more profound influence on the performance of BO than the covariance kernel.

Benchmarking dataset for leak detection and localization in water distribution systems.

This paper presents a dataset with two hundred and eighty sensory measurements for leak detection and localization in water distribution systems. The data were generated via a laboratory-scale water distribution system that included (1) three types of sensors: accelerometer, hydrophone, and dynamic pressure sensor; (2) four leak types: orifice leak, longitudinal and circumferential cracks, gasket leak, and no-leak condition; (3) two network topologies: looped and branched; and (4) six background conditions with different noise and demand variations. Each measurement was 30 s long, and the measurement frequencies were 51.2 kHz for the accelerometer and dynamic pressure sensors, and 8 kHz for the hydrophone. This is the first publicly available dataset for advancing leak detection and localization research, model validation, and generating new data for faulty sensor detection in water distribution systems.

Human-Infrastructure Interactions during the COVID-19 Pandemic: Understanding Water and Electricity Demand Profiles at the Building Level.

When engineers design and manage a building's water and electricity utilities, they must make assumptions about resource use. These assumptions are often challenged when unexpected changes in demand occur, such as the spatial and temporal changes observed during the coronavirus (COVID-19) pandemic. Social distancing policies (SDPs) enacted led many universities to close their campuses and implement remote learning, impacting utility consumption patterns. Yet, little is known about how consumption changed at the building level. Here, we aim to understand how water and electricity consumption changed during the pandemic by identifying characteristic weekly demand profiles and understanding how these changes were related to regulatory and social systems. We performed k-means clustering on utility demand data measured before and as the pandemic evolved from five buildings of different types at the University of Texas at Austin. As expected, after SDPs were enacted both water and electricity use shifted, with most buildings seeing a sharp initial decline that remained low until the university partially reopened. In contrast to electricity use, we found that water use was tightly coupled with SDPs. Our study provides actionable information for managers to mitigate negative impacts (e.g., water stagnation) and capitalize on opportunities to minimize resource use.

Implications of Social Distancing Policies on Drinking Water Infrastructure: An Overview of the Challenges to and Responses of U.S. Utilities during the COVID-19 Pandemic.

Social distancing policies (SDPs) implemented throughout the United States in response to COVID-19 have led to spatial and temporal shifts in drinking water demand and, for water utilities, created sociotechnical challenges. During this unique period, many water utilities have been forced to operate outside of design conditions with reduced workforce and financial capacities. Few studies have examined how water utilities respond to a pandemic; such methods are even absent from many emergency response plans. Here, we documented how utilities have been impacted by the COVID-19 pandemic. We conducted a qualitative analysis of 30 interviews with 53 practitioners spanning 28 U.S. water utilities. Our aim was to, first, understand the challenges experienced by utilities and changes to operations (e.g., demand and deficit accounts) and, second, to document utilities' responses. Results showed that to maintain service continuity and implement SDPs, utilities had to overcome various challenges. These include supply chain issues, spatiotemporal changes in demand, and financial losses, and these challenges were largely dependent on the type of customers served (e.g., commercial or residential). Examples of utilities' responses include proactively ordering extra supplies and postponing capital projects. Although utilities' adaptations ensured the immediate provision of water services, their responses might have negative repercussions in the future (e.g., delayed projects contributing to aging infrastructure).

Spatiotemporal trends of recovery from lead contamination in Flint, MI as revealed by crowdsourced water sampling.

In the aftermath of the lead contamination crisis that plagued the water system in Flint, MI, more than 35,000 water samples were collected from the city's premises. The majority of these samples (>85%) were collected through a voluntary crowdsourced sampling campaign. The samples were analyzed for lead and copper concentrations by the Michigan Department of Environmental Quality (MDEQ). In this study, the crowdsourced sampling data was analyzed by means of spatial autocorrelation analysis to reveal the locations of statistically significant hotspot regions of high water lead levels (WLLs), and to track the spatiotemporal evolution of WLLs as the system recovered from lead contamination. The results showed that hotspot regions that experienced high WLLs were consistent with the areas where lead service line (LSL) density was the highest. Additionally, galvanized service lines and other lead-containing plumbing components could have also contributed to lead release in hotspot regions. The temporal trend exhibited by the crowdsourced sampling data did not reflect a consistent decrease in WLLs despite the interventions implemented by MDEQ and EPA. Instead, sampled WLLs remained high for several months after boosting the orthophosphate dose and launching a city-wide residential flushing campaign. The findings of this study suggest that this could be partially attributed to disproportionate sampling from premises in hotspot regions of high WLLs and LSL density.

Review of Modeling Methodologies for Managing Water Distribution Security.

Water distribution systems are vulnerable to hazards that threaten water delivery, water quality, and physical and cybernetic infrastructure. Water utilities and managers are responsible for assessing and preparing for these hazards, and researchers have developed a range of computational frameworks to explore and identify strategies for what-if scenarios. This manuscript conducts a review of the literature to report on the state of the art in modeling methodologies that have been developed to support the security of water distribution systems. First, the major activities outlined in the emergency management framework are reviewed; the activities include risk assessment, mitigation, emergency preparedness, response, and recovery. Simulation approaches and prototype software tools are reviewed that have been developed by government agencies and researchers for assessing and mitigating four threat modes, including contamination events, physical destruction, interconnected infrastructure cascading failures, and cybernetic attacks. Modeling tools are mapped to emergency management activities, and an analysis of the research is conducted to group studies based on methodologies that are used and developed to support emergency management activities. Recommendations are made for research needs that will contribute to the enhancement of the security of water distribution systems.

Unsteady pressure patterns discovery from high-frequency sensing in water distribution systems.

Pressure transients have been identified as one of the major contributing factors in many pipe failures in water distribution systems (WDSs). The behavior of these pressure transients is largely unknown and cannot be fully assessed by numerical simulation or modeling. This study investigates the behavior of pressure transients in WDSs as measured by high-frequency pressure sensors. A Time Series Data Mining (TSDM) approach is proposed to detect and cluster pressure transients to reveal recurrent and consistent patterns. The proposed technique, based on a modified two-sided cumulative sum (CUSUM) algorithm, is used to detect pressure transients. Dynamic Time Warping (DTW) is adopted to measure the similarity between the detected pressure transients, and k-means clustering algorithm is used to discover the characteristic patterns. Several performance scores are suggested to evaluate the quality of the clustering results, including sum of squared error, Silhouette index, and Calinski-Harabaz index. Results demonstrate that the proposed approach is able to reveal consistent and unique patterns across multiple sensing locations. The proposed approach provides a fast and efficient way to discover the hidden information in WDSs by analyzing high-frequency pressure signals from distributed sensors.