Abstract
This study examines how research on smart water is contributing to climate-resilient municipal water systems around the globe. We identify smart water research trends over time, relationships with climate adaptation and mitigation goals, and applicability to places with developed or developing water and electrical infrastructure. To do so, we systematically review the literature, identifying research on Information Communication Technology-enabled technologies related to water supply, wastewater, and stormwater management. We assess the relationship between each study and climate adaptation and mitigation objectives: managing greater variation in water quantity, leading to scarcity and increased stormwater; managing declining water quality; and low-carbon water systems. We find 96 relevant studies and identify five major categories of research addressing climate adaptation and mitigation: monitoring, modeling, system design, system feedbacks, and uptake and implementation. We find there is a recent acceleration in smart water research, with a concentration of studies focused on modeling. There is an emphasis on water efficiency using data from Advanced Metering Infrastructure, which is most applicable to cities with developed water grids and consistent electrical supplies. Secondarily, there is a concentration of work using distributed sensors for early detection of water quality degradation, which is being done in all municipal contexts. There is far less research on uptake and implementation of smart approaches, especially at the institutional level. In addition, there is relatively little work that explicitly relates smart water technologies to reducing greenhouse gas emissions. While smart water approaches are applicable everywhere, there is a need to for expanded focus on areas without developed water grids or consistent electricity for smart water to meaningfully contribute to Sustainable Development Goal 6.
Introduction
Municipal water service providers around the globe are managing the combined challenges of outdated infrastructure and insufficient funding or revenue for maintenance and upgrades. This is the case in places with both developed and developing infrastructure. In the United States, the American Water Works Association (2012) suggests that US$1 trillion of investment in water infrastructure is needed through 2035, which dwarfs recent spending. Lagging even further behind is global water and wastewater provision. One-third of the population lacks access to safe drinking water and wastewater treatment (U.N. Water, 2019). On the current trajectory, UN member states are not on pace to reach Sustainable Development Goal 6, universal water coverage, by 2050 and doing so will require a threefold increase in investment in water, sanitation, and hygiene services (U.N. Water, 2019). Increasing the reach of water and wastewater services while minimizing costs is critical goal.
Layered atop these existing problems is the challenge of adapting or constructing water systems to accommodate a changing climate. Climate change will alter water supply quantities and quality and may change patterns of water demands. With respect to quantity, the average annual availability of freshwater, the timing of precipitation, and the intensity of precipitation may shift (Bates et al., 2008). With respect to quality, issues like increasing temperatures, urban runoff, flooding, and seawater intrusion will contribute to challenges like eutrophication, increasing turbidity, and salinization of water sources (Bates et al., 2008). With respect to demand, changing patterns of precipitation and temperature may lead to increased consumption (especially outdoor water use) in some locations and decreased consumption in others.
It is possible that one method for creating climate-resilient water systems, increasing the reach of water and wastewater services, and minimizing costs will be through employing emerging, Information Communication Technology (ICT)-enabled solutions associated with smart cities. Smart water technologies are being introduced at many different points of the water supply, distribution, and treatment chain using approaches like sensors, wireless communication, machine learning algorithms, real-time controls, and feedbacks aimed to alter human behavior. Proponents suggest smart water is associated with: improving water system flexibility, efficiency, and reliability; enhancing system maintenance and repairs; improving transparency; distributing risk; enabling solutions tailored to local conditions; improved revenue collection; and supporting behavioral change of water consumers (Kuisma et al., 2020). Critics claim smart water tweaks the system at the margins rather than fundamentally updating the system to build resilience, contributes to the narrative that emerging technology can solve all problems, introduces security and privacy concerns, introduces labor challenges, and leads to a brittle system lacking interoperability (Kuisma et al., 2020). In addition, there are doubts on whether smart technologies are relevant everywhere, as upfront implementation costs may be too high (Gupta et al., 2020).
We seek to understand the potential of municipal smart water technologies for contributing to climate adaptation and mitigation in both developed and developing contexts by evaluating and characterizing scope and trends in this multi-disciplinary literature. As of yet, smart water has not been broadly analyzed as a tool for enabling climate adaptation and mitigation within municipal water utilities. There are several reviews that focus on an aspect of smart water. This includes reviews of cyber-physical water networks (Adedeji and Hamam, 2020), cyber-security and water (Tuptuk et al., 2021), urban hydroinformatics (Makropoulos and Savić, 2019), domestic wastewater (Du., et al., 2019), smart metering (Cominola et al., 2015), water security (Su et al., 2020), resilience to failure (Marchese et al., 2020), and sustainability (Wang., et al., 2015). To evaluate municipal smart water and its contributions to climate adaptation and mitigation, we conduct a systematic review of the recent academic literature. We characterize the municipal smart water literature by the analytic focus of the study, assess the literature relative to climate adaptation and mitigation objectives, evaluate the necessity of a developed water grid or consistent electricity source to implement the technology, and chart trends over time. Finally, we discuss gaps and opportunities.
Reviewing the smart water literature
This study implements a systematic literature review, which contrasts with an ad hoc, expert-selection approach. Expert-selection has the potential to introduce bias in article selection (Noordzij et al., 2011). Mora et al. (2017: 3) note systematic reviews are suited to the smart cities domain, which require “the connection of many academic disciplines.” To conduct this review, we follow guidelines put forward by The Preferred Reporting Items for Systematic Reviews (PRISMA) (Moher et al., 2009), methods described by Xiao and Watson (2019), and methods demonstrated by both Clifton et al. (2020) and Tuptuk et al. (2021). First, we discuss the scope and definitions we use in the study. Next, we define a set of research questions devised to analyze and evaluate the relevant publications. Then, we identify a set of electronic databases and a search strategy, denoting inclusion and exclusion criteria to assess the eligibility of each publication. Finally, we manually inspect the content of the eligible publications to extract relevant data for analysis.
Definitions and scope
To define the appropriate search terms, we explicate: the functions of a water agency, managing for climate adaptation and mitigation in the context of municipal water management, and assessment criteria for whether a smart water solution is relevant to developed or developing municipalities.
Municipal water management
We define a municipal water management to include water supply, wastewater treatment, and stormwater management. We do not consider all aspects of flood management to be within the purview of water utility. As such, we include pluvial floodwaters from a stormwater management perspective, but not a disaster preparedness or search-and-rescue perspective since these functions are not typically the direct responsibility of a water utility. We acknowledge the boundary of water utility functions are gray and some regions practicing integrated water resource management may have folded together a wider array of management functions. Other regions may maintain a more siloed approach to water management.
Municipal water management for climate adaptation and mitigation
To adapt, water systems will need to accommodate changes in quantities of precipitation—including scarcity and increased intensity of precipitation—and decreasing water quality. To mitigate, water systems must reduce greenhouse gas (GHG) outputs. As a result, we identify smart water solutions contributing to climate adaptation and mitigation against the following four management goals: (1) responds to water scarcity, (2) responds to increased stormwater, (3) responds to declining to water quality, and (4) contributes to GHG reductions by improving energy efficiency and/or integrating renewables.
Relevance to developed or developing municipalities
To determine development context for a smart water solution, we consider if the technology requires a developed water grid and/or a consistent electrical supply. With some exceptions, studies do not commonly indicate whether they require a water grid, consistently electricity, or are designed for intermittency. As a result, we infer the conditions most likely to support the smart water solution based on how the technology is being used within the study and the study sites for which it is developed. For example, advanced metering infrastructure (AMI) will most likely be implemented in places with piped household water and consistent electricity. An exception is when metering is specifically discussed in the context of disaggregating household-level use in a multi-family building sharing a water tank. In this case, a developed water grid may not necessary, but a consistent electrical supply is still required. In contrast, an ad hoc source water quality monitoring network created with battery operated sensors requires neither. While this categorization is coarse, it seeks to identify research that may be relevant in efforts to meet SDG 6.
Research questions
This study examines how ICT-enabled, smart water solutions may contribute to climate-resilient municipal water systems around the globe through four research questions. These questions are RQ1. What is the domain of recent smart water research (within the last 10 years) that relates to municipal water management (water supply provision, wastewater treatment, or stormwater management)? RQ2. Is municipal smart water research supporting climate adaptation and mitigation, defined as: (1) responding to water scarcity (including supply-side efficiency, demand-side efficiency, and supply augmentation), (2) responding to increased stormwater, (3) responding to declining water quality, and (4) contributing to GHG reduction (including energy-efficiency or introduction of renewables)? RQ3. How many research articles discuss smart water solutions that require a developed water grid and consistent electrical supply? How many do not require a developed water grid or consistent electricity? RQ4. Are patterns in smart water research for climate adaptation and mitigation changing over time?
Search strategy
The dataset of smart water articles in the analysis is identified using inclusion and exclusion criteria. A diagram adapted from PRISMA illustrating the process of inclusion and exclusion leading to the final dataset is available in Figure 1. Diagram of systematic review. EB = EBSO, GS = Google Scholar, WOS = Web of Science (Classic). If a record appeared in more than one database, it is noted by a combination of those terms. We review all results returned by EBSCO and Web of Science Classic. For Google Scholar, we review a maximum of 500 results per search.
Inclusion criteria
We identify published, peer-reviewed research articles using three search engines: EBSCO, Web of Science (Classic), and Google Scholar. We select these databases because they are common across a range of disciplines. In the first two databases, we filter for research articles. Google Scholar does not make it possible to do so and thus, in addition to research articles, the search results include books, chapters, theses, industry reports, white papers, conference abstracts, conference proceedings, article pre-prints, commentaries and editorials. We manually filter Google Scholar results, focusing on peer-reviewed, published research articles.
In each search engine, we use the following keywords as inclusion criteria: “smart water”; “smart wastewater”; “smart sewer”; “smart stormwater”; “smart flood”; “smart city” AND water; ICT AND water; IoT AND water; “intelligent water” AND urban; “intelligent water” NOT drop (there is an intelligent water droplet algorithm unrelated to water management that confounds results); water AND smartphone AND urban. We limit search results to studies that are available in English in the last 10 years in scholarly, peer-reviewed journals. In EBSCO, this returns a total of 2219 records. In Web of Science, the search terms return 1801 records. In Google Scholar, owing to a volume of gray literature, these search terms return almost 227,000 records (ranging from 82 to 156,000 results per search term). For EBSCO and Web of Science, we evaluate all results. For Google Scholar, we evaluate a maximum of 500 results per search term, leading to a total of 3637 records.
For inclusion in the database, the paper must be a published, peer-reviewed research article. The title or abstract must specifically reference ICT, IoT or smart technologies (such as ICT-enabled technologies like high resolution sensor data, crowd-sourced data, and real-time controls (RTC) clearly mitigated by software) and clearly relate to municipal water management (drinking water, municipal wastewater, or municipal stormwater) as a primary objective of the study. The study must evaluate empirical data or simulated data. In addition, each full research article must be digitally accessible in English from University of Pennsylvania libraries or a final copy of the study openly available online at the time of the review (February to April 2021).
After applying the inclusion criteria, we identify 226 studies from EBSCO, 297 studies from Web of Science (Classic), and 219 studies from Google Scholar.
Exclusion criteria
After the initial pass for inclusion based on search terms and basic content review (titles and skims of abstracts), we read abstracts thoroughly and skim papers. We exclude articles based on the following exclusion criteria: (1) Studies that do not directly relate to water supply, wastewater, or stormwater management for a municipality, such as: strategies targeted toward agriculture or ecological applications that do not specifically tie to municipal water management; end-use technologies (e.g., efficient washing machines, smart irrigation apps, solar water heaters) or building-scale technologies (e.g., cooling a building through smart water management) that do not connect to a larger, municipal-scale water management framework; studies that focus on a specific aspect of a water system (e.g., sensor design, leak detection algorithms, comparisons of methods segmenting smart meter data, and flood mapping), but do not contextualize the problems or findings in water management. If the abstract is ambiguous, we skim the paper for verification. If the connection is still ambiguous after skimming, then we exclude it. (2) Smart city studies relevant to multiple infrastructures that do not discuss water infrastructure as a primary theme (e.g., focus was on broader smart metering technology for electricity, with brief mention that it could be applied to water). (3) Studies that do not use empirical or simulated data. For example, editorials, commentaries, viewpoints, literature reviews, surveys, concept papers, descriptive case studies, and some theoretical proposals (e.g., for hardware architectures) that included a sketch but no models. If methods or data sources are unclear, even if the study self-identifies as empirical, we exclude it. We cite several of these papers in the introduction and discussion, but they are not included in the analysis of current research.
There are some categories of studies which require careful evaluation to determine exclusion. For example, there are many studies discussing water quality. If the study broadly evaluates water quality in a river, we exclude it, but if it discusses water quality within a municipal water distribution system (e.g., drinking water quality) we include it. If a study discusses flooding in the context of emergency management, we exclude it, but if it was in the context of municipal stormwater management, we include it.
There are many studies using high resolution, novel data sets to create new models. If a study focuses on algorithms for disaggregating or segmenting and identifying end-use behaviors in residential water data but does not link to management, we exclude it; if it identifies implications for management, we include it. For example, we exclude high resolution modeling of precipitation and flooding unless it specifically relates to implications for stormwater management. We also exclude optimization studies (e.g., optimized valve or pump placement), if models do not have direct links to implications for ICT-enabled water management and smart infrastructures.
If a paper, or a close derivative of a paper, was published in multiple outlets, we include the most recent version of the manuscript and exclude the others.
Content analysis and data extraction
From each study, we extract information on (1) the primary analytic focus of the study, (2) the key relationship to the identified climate adaptation or mitigation goals, and (3) whether the technology is likely most suited to a developed water grid and consistent electrical source. There are many papers that have more than one theme, and we choose the theme that we find to be most central to the study. When possible, we identify these factors from the abstract. When it is ambiguous, we reference the full text.
During the process of data extraction, we exclude several more papers based on quality and content. Previously, the only filter on quality was through the inclusion criteria of publication in a peer-reviewed journal and, consequently, we observe large variation in quality. We elect to keep many smaller, regional journals to ensure the most geographically inclusive review possible, most likely to reflect smart water needs and research in a wider diversity of municipal contexts. However, if the language quality (spelling or grammar) or intent of the paper is sufficiently unclear that we are unable to identify primary themes of the paper, we exclude the study from the analysis.
After data extraction, we also exclude studies that not directly related to our four defined climate adaptation and mitigation objectives (scarcity, stormwater, water quality, GHG reduction). For example, we exclude studies that focus on cost recovery and billing, even though ensuring adequate system revenue is may be indirectly related to adaptation and mitigation.
After applying exclusion criteria, our dataset contains: 33 studies from EBSCO, 48 studies from Web of Science (Classic), and 62 studies from Google Scholar. Forty-one articles appeared in more than one database, leading to a total of 91 unique studies.
Final dataset
Finally, we conduct additional backward and forward searching and manual searching. While the systematic review was performed by one researcher, additional searching was conducted independently by two researchers. These results were compared and adjudicated. Through this process, we identify five additional peer-reviewed studies that meet inclusion and exclusion criteria, which we add to the final database. Ultimately, the final database contains 96 studies.
Analysis of results
What is the domain of recent smart water research?
Counts of studies by analytic focus from 2011–2021 (April). There are a total of 96 studies in the sample.
Note: ICT: Information Communication Technology.
Monitoring
Monitoring physical and natural water infrastructure is the first digital layer in a smart water system. We find that monitoring can be interpreted through three categories of sensors: AMI, other ICT-enabled sensors, and citizens-as-sensors.
Advanced metering infrastructure
Also known as smart meters, AMI provide near real-time digital readouts of water consumption that typically link to a database owned or accessed by a water utility. While there are many studies that use AMI data as the basis of algorithmic development, we only identified three studies that focus on AMI monitoring, as opposed to the data product produced by AMI. Muhammetoglu et al. (2020) examine smart water meters with a low sampling rate for a 6 month period to monitor and identify leakage in public places (schools, graveyards, public toilets, parks, and universities). Li and Chong (2019) suggest the design of a wireless smart meter that does not require an external power supply, instead using a water turbine generator that both senses flow and generates electricity. More disaggregate than the most common household-level AMI devices, Yang et al. (2017) monitor domestic water consumption at the appliance level, wirelessly transmit data, and then provide near real-time feedback to consumers on water-saving activities via an app.
Other ICT-enabled sensors
Beyond AMI, sensors are most commonly used to collect data on water quality (e.g., temperature, pH, turbidity, and conductivity), flow rates, and pressure.
There are many of studies that monitor water quality. Pantjawati et al. (2020) design a wireless system to evaluate river water quality before and after a factory sewer outfall. Mamun et al. (2019) implement a system linked to Geographic Information Systems (GIS) and powered using solar energy sources to monitor water quality in real-time. Postolache et al. (2014) monitor water quality with low-cost sensing nodes, a multi-parameter sensing probe, and a wireless sensor network. Martínez et al. (2020) develop a low-cost water quality monitoring device, including a nitrite analyzer that uses ion chromatography detection. Williamson et al. (2014) report on early detection of two water quality events using optical sensors and a refractive index. Levin et al. (2016) discuss a method of detecting high concentrations of a contaminant, fluoride, in groundwater by analyzing the color in a smartphone’s photograph of water samples mixed with a reagent. Lambrou et al. (2014) develop low-cost, electrochemical and optical sensor nodes and a companion algorithm for monitoring water quality in pipes, with an emphasis on lightweight implementation and lifespan. Jindal et al. (2017) suggest a low-cost, multi-parametric sensor for water quality monitoring that uses a smartphone and an ad hoc network. Pasika and Gandla (2020) suggest a water tank monitoring and water quality sensor system that wirelessly transmits data to ThinkSpeak, an open IoT analytics platform.
Another set of studies concern methods of identifying leak detection from tap, to municipal grid, to longer pipelines. Mohapatra and Rath (2019) present a sensor-enabled water tap, which detect leaks and communicates to a central utility center. Fabbiano et al. (2020) propose detecting and localizing pipe leaks by monitoring radial vibrational status of critical system pipes. Zhang et al. (2020) develop a technique to automate detection of cracks in water pipes using acoustic sensors (accelerometers) in a noisy, urban environment. Karray et al. (2016) propose detecting and localizing leaks in long distance pipelines by combing algorithms and a low-power, energy efficient wireless sensor node architecture.
Finally, one study monitors to improve stormwater management. Parilla et al. (2020) address drainage clogging due to garbage with sensors that transmit data on garbage levels to an app that sends alerts to users via SMS.
Citizens as sensors
Crowd-sourced, volunteered geographic data is sometimes used as a component of municipal water management. Notably, Al-Bayari et al. (2020) develop a smartphone app for users to report water and wastewater complaints to the local water authority, automating the complaint process and geolocating the complaint.
Modeling
Almost all studies in the dataset include some degree of modeling, but studies in this category focus on algorithmic developing as the primary theme. Most interpret water data from ICT-enabled sensors with machine learning methods (e.g., long short-term memory (LSTM), artificial neural networks (ANN), and deep learning). There also example of hydraulic simulations and econometric methods. We do not create subcategories of models, however, because each study commonly tests more than one model type.
Many modeling studies relate to water efficiency and demand management programs, often using AMI data. These studies typically focus on segmenting residential end uses from the AMI data and/or isolating post-meter leakage. Cole and Stewart (2013) identify average hour, peak hour, peak day and peak month consumption of residential customers from an early (2006–2007) implementation of residential smart metering. Bethke et al. (2021) use an unsupervised segmentation algorithm and smart meter data to identify water end uses, peak demand times, and temporal trends. Nguyen et al. (2014) create an algorithm for residential water end-use classification and web-based interface designed for customers and water business managers. Luciani et al. (2019) develop a system that receives data from a smart meter, stores the data, and processes it to automatically identify leakage by benchmarking against non-consumption periods. Bennett et al. (2013) test three residential water end-use demand forecasting models that use an ANN, smart water meter consumption data, household demographics, and household physical data, and then forecast potential water savings from a citywide water appliance retrofit program. Cominola et al. (2019) use only smart meter readings (without complementary water end use surveys) to identify household water end use routines, main end use components, and temporal characteristics. Cominola et al. (2018a) examine accuracy of identifying residential end-uses from smart meter water data using variable sampling rates. Meyer et al. (2020) evaluate the possibility of extracting end use events from smart meter data with a low sampling rate, which is more common in municipalities seeking to minimize costs. Cominola et al. (2018b) segment customers based on combined water and electricity use profiles for combined water-electricity conservation or peak-shifting interventions. Niranjana et al. (2020) evaluate dynamic assignment of household water supplies based on optimal water use profiles and shutting off water access when it is not needed. For non-residential customers, Patabendige et al. (2018) develop an algorithm and a web-based software system to detect and visualize anomalous water use, presenting a daily anomaly score to users and a rationale for anomalous use.
There is another concentration of papers evaluating grid-level water pressure management and burst or leakage detection. Candelieri et al. (2013) analyze flow and pressure data measured at crucial network points to improve efficiency of leak localization. Abdelhafidh et al. (2020) develop a cognitive IoT-based architecture using a genetic algorithm for smart leak detection and localization. Sun et al., (2020) combine linear discriminant analysis, a neural network, and Bayes temporal reasoning for leak detection and localization in a District Metered Area. Farah and Shahrour (2017) develop a leakage-detection method combining a traditional water balance approach with a sensed data on minimum night flows. Rojek and Studzinski (2019) develop a leak detection and localization algorithm using a neural network, Geographical Information System, Supervisory Control and Data Acquisition (SCADA) system, and hydraulic model of the water supply network. Preciado et al. (2019) provide high resolution water demand forecasts (1 min time with a lead of 24 h) with pattern recognition and pattern-similarity techniques to reduce the needed data inputs, which they test in the context of leak detection. In the context of burst detection, Mounce et al. (2014) uses multi-sensor data and develop pattern matching techniques and binary associative neural networks to detect anomalies. Brentan et al. (2018) present a method for combining characteristic data of a water distribution system with operational data to improve pressure control and identify bursts. Wang et al. (2020) propose a two-stage method for burst detection in district metered areas: first clustering for outlier detection and then identifying the presence of bursts. To minimize system-level water loss through leakage, Rout et al. (2020) develop an algorithm to define a seasonal threshold constraint on water distribution, conserving water over the typical uniform supply.
A third cluster of papers concerns water quality, stormwater and combined sewer system flows. A focus within this set of papers is developing sufficiently fast processing speeds to scale the deployment of RTCs, which can be used to physically alter water systems. Bowes et al. (2020) evaluate reinforcement learning to create stormwater control policies in urban catchments with controllable valves. Similarly, Mullapudi et al. (2020) develop an algorithm trains a reinforcement learning agent to control RTC valves in a distributed stormwater system. Saliba et al. (2020) examine retrofitting passive stormwater systems with RTCs using a reinforcement learning algorithm, Deep Deterministic Policy Gradient, which is capable of handling noisy input data. Edmondson et al. (2018) prototype a smart sewer asset management model to monitor and evaluate real-time performance of the system and predict flooding. Zhang et al. (2018) test the accuracy of different neural network models for simulating and predicting water levels in a combined sewer system. Panchal et al. (2019) create an algorithm to detect floods from gait analysis of smartphone data, with the suggestion results can inform the construction of appropriate stormwater drains. Liu et al. (2019) predict the quality of drinking water using a LSTM deep neural network. Mariammal (2021) implement a cat swarm optimization (CSO) based neural network to forecast water quality.
One study was notable for its complexity, working across spatial scales and water management systems. (Sun et al., 2020) seek to minimize combined sewage overflows and by jointly modeling drinking water distribution, wastewater treatment, and stormwater in distribution networks operated by RTC.
System design
The next group of studies relates to smart water system architecture, and then prototypes and tests or creates theoretical simulations of the proposed system.
Theoretical network design
The majority of studies we observed on theoretical design of smart water networks offered flow charts or schematics of suggested designs without evaluating with data and were thus excluded based on the exclusion criteria. There are exceptions, however, and they are included in this category. Kim (2019) develops a scenario model for testing technology placement and alternative smart water grid configurations. Ibrahim (2020) suggests the barrier to upscaling RTC deployment for stormwater management is a lack of a theoretical operational framework, modeling a hybrid static and RTC approach for system-wide operation. Ramsey et al. (2020) suggest a peer-to-peer non-potable water market, where households use, sell, and buy rainwater within a network of water users. Gautam et al. (2020) suggest a system that monitors water levels in a residential building with a roof tank, forecasts water demand, detects leakage, and makes data viewable via a web interface. Mounce et al. (2015) present a distribution network equipped with flow and pressure sensors, smart meters, and data analysis using an anomaly detection system. Pérez-Padillo et al. (2020) develop smart pressure monitoring and alert system with low-cost hardware and open-source software to detect leaks. Priya et al. (2019) present a low-cost water quality monitoring system using a wireless sensor network to predict risks to water quality. Ramadhan et al. (2020) present a low-cost water monitoring system to provide real-time feedback on water quality with a wireless sensor network, SMS and emails. Saravanan et al. (2018) propose a new SCADA system for real-time water quality monitoring.
Hardware/component/system optimization
Optimization of water systems includes many studies seeking to reduce water leakage and improve system energy consumption. To reduce leakage, Fantozzi et al. (2014) optimize network sectorization and pressures. Brentan et al. (2018b) couple a predictive water demand model and an optimization algorithm to manage pumps and pressure-reducing valves in a water distribution network. Giudicianni et al. (2020) propose adaptive, dynamic district metered areas using an algorithmic to aggregate/desegregate based on optimizing the production of energy during the day (via micro-hydropower) and reduction of water leakage at night. Pan et al. (2015) seek to improve pressure management, introducing a web-based decision support system alongside two multi-objective optimization algorithms and a hydraulic simulation model. Narayanan and Sankaranarayanan (2020) use demand forecasts to optimize water distribution for reduction of leakage in distribution. Studziński and Ziółkowski (2020) test different algorithms for controlling water supply pumps to determine which best reduces water loss while ensuring water quality.
To reduce system energy use, Alshehri et al. (2021) optimize energy consumption of a desalination facility by sensing treated water quality parameters and matching treatment level with end uses (potable, non-potable, and agriculture). Pointl and Fuchs-Hanush (2021) compare different communication technologies, seeking to minimize energy consumption of pressure sensors.
In addition, three studies address stormwater optimization problems. Behzadian et al. (2018) optimize flow management in a rainwater harvesting tank for both stormwater capture and non-potable water demands, proactively controlling the levels of the water tank using rainfall forecasts. Di Matteo et al. (2019) optimize opening and closing of a network of ICT-enabled rainwater tanks to reduce peak stormwater flows. Maiolo et al. (2020) test different RTC configurations in an urban drainage network to optimize flood reduction.
Communication
Most research on communication standards is not included in this analysis, since ICT/IoT communication studies typically lack direct connection to climate adaptation and mitigation, and these studies are filtered out through the inclusion and exclusion criteria. We find two exceptions, however. In the context of reducing water system leaks, Hsia et al. (2020) suggest a method for integrating distributed water meter readings in a control center through a wireless system. In the context of managing wastewater networks to reduce overflows, Raza and Salam (2020) examine path loss analysis of wireless underground communications through soil and asphalt.
System feedbacks
Data and algorithms feed back into smart water systems through two channels: RTCs that physically manipulate systems or through behavioral change in water users and managers.
Real-time control
Using RTCs is especially common within stormwater management. Three studies focus on RTCs at the watershed-level for downstream management of municipal stormwater. Mullapudi et al. (2018) coordinate releases from two upstream stormwater basins enabled with RTCs to achieve downstream stormwater management goals. Sharior et al. (2019) propose a basin-level smart stormwater system that enables RTCs through with real-time water quality data and a stochastic rainfall time-series to improve downstream water quality. Shishegar et al. (2021) present a network of devices to dynamically generate set-points for system actuators at a remote control center where global optimization algorithms calculate real-time target values and activate local controllers at the outlets of spatially distributed detention basins.
Other studies examine the use of RTCs within a municipality for stormwater management and wastewater management. Lund et al. (2019) focus on reducing CSOs from short, intense rainfall events called cloudbursts through real-time control of stormwater inflow, integrating sewers, green infrastructure and the urban landscape. (Joseph-Duran et al., 2015) seek to reduce CSOs by improving performance of RTCs in a sewer network with Receding Horizon Control with Moving Horizon Estimation. Sadler et al. (2020) assess the utility of RTCs for reducing flooding from backflow of a combined sewer system in a coastal city, where tides enter the outlets of the combined sewer system. Persaud et al. (2019) evaluate the effectiveness of augmenting bioretention basins (e.g., green infrastructure) with RTCs for water quality improvements. Wei et al. (2021) examine methods of better controlling CSOs by controlling a double-gate based on water quality parameters. Troutman et al. (2020) suggest that an ICT-enabled sewerage system can be viewed as an extension of the water resource recovery facility and storage can be dynamically controlled to benefit water quality dynamics at inflow to a treatment plant.
In the context of water scarcity, Mutepfe et al. (2013) suggest RTCs to manage tap-level water distribution. At a shared tap, users are assigned privileges and use an access pass for water allocations.
Behavioral change
The studies in this sub-category propose or examine system feedbacks through user behavioral changes. Some studies observe behaviors through data, while some conduct interviews or surveys with users. One study focused on developing a web-based Decision Support Tool targeted at water managers conveying data on leak localization (Meseguer et al., 2014).
Most behavioral change studies focus on the effectiveness of smart meters in reducing water use. Britton et al. (2013) examine households that have post-meter water leakage and test the effectiveness of communication interventions for reduction in leaks. Schultz et al. (2018) evaluate if customers with smart meters respond to leak detection notifications and conduct repairs. Fielding et al. (2013) study the effectiveness of household water demand management interventions derived from smart metering data. Liu et al. (2017) examine the differential effectiveness of household-level water consumption feedback via paper end-use reports or an online portal. Ribeiro et al. (2015) develop tailored recommendations to customers for demand management based on their smart meter data. Rougé et al. (2018) evaluate the possibility of using dynamic pricing through smart metering to control water demand.
Two studies examine the effectiveness of other smart water programs. Hsu et al. (2020) evaluate changes in water quality related to a citizens-as-sensors project in China, where an app allows citizen to log water quality complaints. Stephens et al. (2020) reports on findings of the effectiveness of a predictive leak and localization system underpinned by distributed acoustic sensors (accelerometers) with customized analytics and IoT technologies.
Uptake and implementation
Uptake and implementation is the least researched category. These studies focus on perceptions of technology.
Government, politics, and institutional perceptions
While reviewing the literature we noted descriptive case studies on implementing smart water programs, but only two studies pass through our inclusion and exclusion criteria. Kumpel et al. (2015) analyze the potential of ICT for facilitating the use of water quality data, given institutional structures, across six African countries. Meng and Hsu (2019) conduct a national, stated-preference survey of water utility officials to understand interest in smart green infrastructure for stormwater management.
User-level perceptions
There are more studies that focus on user-level perceptions than institutional perceptions. Most concern the design of household demand management programs. Beal et al. (2013) use smart meter data, surveys, audits, and water diaries to identify households overestimating and underestimating water use, with implications for targeted demand management programs. Bermejo-Martín et al. (2020) survey households’ knowledge of the urban water cycle, water values, opinions on reclaimed water new technologies to classify users and implement customized demand management strategies. Liu et al. (2016) conduct a mixed methods study on the appeal and effectiveness of customized, paper-based “Home Water Updates” derived from smart meter data. Cahn et al. (2020) conduct focus groups of water users to analyze incentives to conserve and preferences for online feedback applications. Finally, Montginoul and Vestier (2018) investigate why there is low participation in adopting free smart meters provided to French households through a natural field experiment and a survey.
Is municipal smart water supporting climate adaptation and mitigation goals?
Analytic focus of the reviewed studies by climate adaptation or mitigation goal.
Note: GHG: greenhouse gas; AMI: advanced metering infrastructure; RTC: Real-Time Controls.
While many studies have a single analytic focus, others aim to address multiple goals simultaneously. For instance, several stormwater solutions also help prevent combined sewer overflows, which addresses water quality. In studies that relate to multiple climate adaptation and mitigation goals, we focus on the primary intent the study articulates. There is one exception, however. Cominola et al. (2018a) use the combination of water and energy data to segment customers for a water-energy demand management program, with equal focus on water and energy efficiency. We score Cominola et al. (2018b) as addressing both scarcity and low-GHG goals. As a result, we identify a total of 97 goals in the 96 studies. As a whole, we find that 51 studies focus on scarcity/efficiency, 17 on stormwater/flooding, 22 on water quality, and 7 on low-GHG systems. These figures are broken out by sub-category in Table 2.
Goal 1. Responds to water scarcity
Over half of the studies relate to scarcity, dwarfing the other categories. Strategies are primarily targeted toward household efficiency and grid-level efficiency. Exceptions are two studies which examine leakage in non-residential municipal water users (Muhammetoglu et al., 2020; Patabendige et al., 2018) and a study which discusses alternative water supply development in the form of smart rainwater harvesting (Ramsey et al., 2020).
At the household level, studies focus on implementing and interpreting household smart water meter data and identifying household-level leaks. Many researchers are working on models to disaggregate or segment end-uses from smart meter data (Bennett et al., 2013; Bethke et al., 2021; Cole and Stewart, 2013; Cominola et al., 2018a; Cominola et al., 2019; Meyer et al., 2020; Nguyen et al., 2015). While we found one paper that focuses more on leak detection, end-use and leak detection often overlap (Gautam et al., 2020). One study looks at a method of monitoring and managing demand for specific end-uses (Mohapatra and Rath, 2019; Mutepfe et al., 2013; Yang et al., 2017) and one for implementing a smart meter without an external electricity supply (Li and Chong, 2019).
Also at the household level, many studies focus on translating smart water data into demand management programs. Some segment customers to form the basis of targeted demand management programs (Bermejo-Martín et al., 2020; Cominola et al., 2018b). One study suggests localized, dynamic pricing as a tool to drive demand management (Rougé et al., 2018). Another study develops customized demand management recommendations (Ribeiro et al., 2015). Several studies test different water data communication interfaces and recommend conservation actions to identify customer behaviors and responses (Britton et al., 2013; Cahn et al., 2020; Fielding et al., 2013; Liu et al., 2016, 2017; Schultz et al., 2018). One study examines why some customers do not adopt a free smart meter (Montginoul and Vestier, 2018). The focusing on testing interfaces and adoption are notable for incorporating more qualitative data on human perceptions in smart water management.
At the grid-level, studies primarily focus on distribution to reduce water loss, through both pressure management and leak detection. Many of these studies examine two or more of these subjects at once. Several studies examine water distribution based on demand forecasting (Brentan, et al., 2018a; Mounce et al., 2014; Narayanan and Sankaranarayanan, 2020; Preciado et al., 2019; Rout et al., 2020). Two studies specifically focus on measuring and managing pressures (Pan et al., 2015; Pérez-Padillo et al., 2020). Many studies focus on detecting and localizing pipe leaks (Abdelhafidh et al., 2020; Al-Bayari et al., 2020; Candelieri et al., 2013; Fabbiano et al., 2020; Farah and Shahrour, 2017; Hsia et al., 2020; Karray et al., 2016; Luciani et al., 2019; Mounce et al., 2015; Pointl and Fuchs-Hanusch, 2021; Rojek and Studzinski, 2019; Stephens et al., 2020; Sun et al., 2020; Wang., et al., 2020; Zhang., et al., 2020). One study focuses on helping water operators identify leaks with a web-based tool (Meseguer et al., 2014).
Goal 2. Responds to increased stormwater
There are 17 studies that focus on smart, municipal stormwater management. The bulk of this research addresses dynamic control of stormwater flows through RTCs from the watershed to the city (Bowes et al., 2020; Ibrahim, 2020; Joseph-Duran et al., 2015; Maiolo et al., 2020; Mullapudi et al., 2018, 2020; Sadler et al., 2020; Saliba et al., 2020; Shishegar et al., 2021). Related to this work, there is also some research around monitoring and optimizing wastewater systems to prevent combined sewer overflows (Edmondson et al., 2018; Lund et al., 2019; Zhang et al., 2018). Also, in the vein of improving stormwater drainage is a smart approach to managing trash accumulation blocking stormwater systems (Parilla et al., 2020).
There is also some work on integrating new infrastructures to reduce peak flows. This includes rainwater collection (Behzadian et al., 2018; Di Matteo et al., 2019) and green stormwater infrastructure (Meng and Hsu, 2019).
With an intent of identifying new infrastructure needs, one study uses crowd-sourced data to map street-level flooding (Panchal et al., 2019).
Goal 3. Responds to decreasing water quality
We identified 22 water quality studies. Most focus on drinking water and wastewater monitoring (Jindal et al., 2017; Lambrou et al., 2014; Levin et al., 2016; Mamun et al., 2019; Martínez et al., 2020; Pantjawati et al., 2020; Pasika and Gandla, 2020; Postolache et al., 2014; Priya et al., 2019; Ramadhan et al., 2020; Raza and Salam, 2020; Williamson et al., 2014). Two focus on water quality prediction (Liu et al., 2019; Mariammal, 2021). Two examine how monitoring data is used in management (Hsu et al., 2020; Kumpel et al., 2015).
There are four studies that focus on water quality in stormwater retention and combined sewer systems (as opposed to volumes, the focus of the stormwater section), all using RTCs (Persaud et al., 2019; Sharior et al., 2019; Troutman et al., 2020; Wei et al., 2021). One study combines drinking water volumes, wastewater volumes, combined sewer overflows, and RTCs, primarily focusing on water quality (Sun et al., 2020).
Goal 4. Reduces greenhouse gasses
Thus far, few studies directly address low-GHG water systems, either through energy efficiency or integrating renewables. There is a substantial amount of literature concerning energy efficiency with respect to increasing the lifespan of the batteries used to power distributed smart infrastructure, but these are not often explicitly linked to climate mitigation. We identified three studies seek to improve system-wide energy efficiency through water grid-level distribution and pressure management (Brentan, Meirelles, Luvizotto Jr and Izquierdo, 2018b; Fantozzi et al., 2014; Giudicianni et al., 2020; Studziński and Ziółkowski, 2020). Also to improve system-wide efficiency, one study tests the different grid configurations (Kim, 2019) and one study matches water treatment levels with the needs of end-users (Alshehri et al., 2021). At the household level, one study seeks to improve energy efficiency by implementing a water-energy demand management program (Cominola et al., 2018b)
Does the approach require a developed water grid and/or consistent electrical supply?
To answer RQ3, we examine the distribution of research for each climate adaptation and mitigation goal by its infrastructure needs (Figure 2). There are many studies which address scarcity through AMI monitoring and data sourced from AMI, which creates a density of research on efficiency for cities with developed infrastructure. Across stormwater management and water quality, however, there is a more even distribution of research relative to developed water grid and consistent electricity needs. In total, we identify: 74 out of 96 of the climate research areas examine technologies most suited to developed water grid and 81 out of 96 studies examine technologies most suited to a consistent electrical supply. Smart water climate adaptation and mitigation objective by need for developed water grid and consistent electrical infrastructure.
Have patterns in smart water research changed over time?
To answer RQ4, we examine the number and percent of papers published in each category by year. In total, the number of smart water articles increases over the study period (Figures 3 and 4). Notably, we find no articles on municipal smart water for climate adaptation and mitigation in 2011 or 2012, the first 2 years of the study period. This number grows to 35 articles in 2020. Count of papers by analytic focus over time. (Top) Count of papers by climate adaptation and mitigation goal over time. (Bottom) Percentage of studies by climate adaptation and mitigation goal over time.
Figure 3 shows the distribution of studies by analytic focus over time. During the study period, we do not observe any notable trends in the analytic focus of the research.
Figure 4 shows the distribution of studies by climate adaptation and mitigation goal, both by count and percentage of all studies evaluated. There are no studies in our dataset in 2011 and 2012, and in 2013 all identified research focuses on water scarcity. Beginning 2014, water quality becomes a more dominant focus of research. In 2015–2017, all identified smart water research focused on scarcity or water quality. Beginning in 2018, there is a distribution of research across all four climate goals. Though a longer study period is needed to anticipate future trends, it is possible that distribution of these research focuses is beginning to converge.
Discussion
The identified research questions evaluate scope and trends in municipal smart water for climate adaptation and mitigation. RQ1 reveals concentrations of research and areas that are less studied. Analytically, most of the research (31 out of the 96 studies) focuses on model development, while only 7 studies focus on uptake and implementation. RQ2 builds on this question, examining the relationship between analytic focus and climate adaptation or mitigation goal, finding the majority of research concerns water scarcity (51 studies). There is a concentration of research on modeling with AMI data targeted toward efficiency applications, which is most applicable in municipalities with a developed water grid and consistent electricity. A distant second is a concentration of research on water quality monitoring and modeling, applied to places with both developed and developing infrastructure. RQ3 determines that most research is suited to a developed water grid (74 out of 92 studies) and a consistent electrical supply (81 out of 92 studies), though there are studies applicable to all contexts. RQ4 indicates that the smart water literature is accelerating and moving toward addressing all four identified climate adaptation and mitigation goals.
There are many limitations to this study which likely contribute to a downward bias in identification and evaluation of technical studies. Our inclusion criteria of explicitly relating to management of urban water supply, stormwater, or wastewater means studies were not included in the analysis that focus on very specific aspects of smart water, like sensor specifications and comparisons in the details of machine learning algorithms. Further, some technical papers may not be returned by our search terms, with titles and abstracts avoiding words like “smart” or “IoT” and instead using terms like “predictive optimization” or “real-time.” In addition, our inclusion criteria require explicit mention of one of the four climate adaptation or mitigation management goals. There were many studies that could be interpreted to have impact on climate adaptation or mitigation without an explicit mention. We required explicit mention to successfully filter the large volume of initial search results. The final dataset in the review should not be interpreted to infer that there is a lack of research in some technical areas, like cyber-security, that may not be explicitly related to climate adaptation and mitigation management goals.
Another limitation is that we choose not to include gray literature. There are approximately 227,000 results returned by Google Scholar, in contrast to the approximately 2000 results returned by EBSCO and Web of Science (Classic), and this difference is largely attributable to gray literature. In our review, we observe that many of the studies in the gray literature overlap in content and are of poor quality. To improve the quality of the final database, we choose to limit results to published, peer-reviewed articles. Gray literature has value, however. It is noted in having a role in systematic reviews seeking to overcome publication bias (Kitchenham and Charters, 2007); thus, our database will likely not surmount these biases. Further, we note that conference papers are an especially prevalent method of publication in computer science and the exclusion of conference papers may consequently lead to an underrepresentation of some relevant studies tied to this discipline.
Despite factors that might limit identification of some of the most technical papers, 80 of the 96 articles are purely technical. Among the 96 studies in the final dataset only 16 relate to sociotechnical aspects of smart water (behavior change: 9; uptake and implementation: 7). Of the 7 studies focusing on uptake and implementation, 5 concern user perceptions of smart meters and smart meter data (Table 1). Only 2 discuss institutional-level aspects of smart water. To facilitate greater penetration of smart technologies for climate adaptation and mitigation, it is likely that better understanding of institutional opportunities and barriers is necessary. Beal and Flynn (2015) suggest that metering technology may not have scaled due to utilities’ difficulty identifying and communicating the customer benefit. Stewart et al. (2018) suggest instead that, in addition to a barrier of high upfront capital costs, utilities have been slow to embrace digital transformation because its natural monopoly status disincentives innovation or competition. In contrast, Meng and Hsu (2019) find that water managers are interested in smart technologies, especially if they reduce ongoing operation costs. Kumpel et al. (2015) find that variation in internal institutional structure can facilitate or prohibit implementing smart technologies.
Of the examined climate management goals examined though RQ2, the most substantial gaps are in responding to decreased water quality and contributing to GHG reductions. With respect to water quality, we observe far more research on smart monitoring than system feedbacks. There are many early warning systems for declining water quality. Yet, it is often unclear what happens when there is an alert that water quality is low. This challenge relates to an issue raised by Therrien et al. (2020: 2613): “Collecting data without a proper strategy leads to data graveyards.” To make the most of this line of research, the full pathway between sensed water quality and RTCs or behavioral response will benefit from further development, especially in developing contexts. With respect to reducing greenhouse gases, some research envisions lower embedded energy through improving pumping or water distribution and pressure management within grids and microgrids. Yet, smart water literature that pertains to energy efficiency often is more closely related to battery lifespans in distributed smart devices (e.g., Du., et al., 2015). There is research being done in many arenas that relates to energy efficient water systems, but there is opportunity to explicate and develop the contribution of municipal smart water to climate mitigation.
To increase the impact of smart water and ability to meaningfully contribute to SDG 6, there is a need for expanded focus on areas without developed water grids and consistent electricity. In some cases, the value proposition of smart water may depend on the local cost of labor relative to the cost of smart technologies, however. In a study focusing on the Czech Republic, Sladek et al. (2020: 40) note “an isolated implementation of IoT technologies is much more expensive than the current solutions that are based on human labour…This is caused mainly by high prices of the… metering devices supporting IoT functionality.” Studying Harare, Zimbabwe, Gambe (2015: 236) remarks, “ …the city is struggling financially, thus its capacity to install and maintain the system is compromised. This is aggravated by the frequent power shortages…” and suggests testing smart water infrastructure in more affluent neighborhoods before extending smart systems across the city. It is well documented that places in the process of developing water and electrical infrastructure have the opportunity to leapfrog, however, creating infrastructure systems that are more contemporary than those in cities locked into existing infrastructures.
Conclusion
In this study, we systematically search the literature to identify research on municipal smart water supporting management for climate adaptation and mitigation. Focusing on studies using empirical or simulated data published in peer-reviewed journals, we use EBSCO, Web of Science (Classic), and Google Scholar to identify 92 studies. These studies address strategies to manage for greater variability in water quantity (scarcity, increased stormwater), declining water quality, and GHG reduction in both developed and developing contexts. We find that there is a concentration of research on modeling with AMI data for water scarcity applications. Secondarily, there is a concentration of research on water quality monitoring. Areas that have received less focus are: institutional perceptions, integrating water quality monitoring and modeling with system feedbacks, and GHG reduction.
The published literature on smart water is accelerating, but there is far more research on how data can be collected and analyzed than operationalized. Uptake and implementation of smart water technologies to successfully adapt to climate change and mitigate impacts will likely require social transitions. Better understanding institutional opportunities and barriers for smart water adoption may be especially useful. For example, it is likely the deployment of smart water technology will require engaging different groups of stakeholders than has been necessary in the past, such as local electricity and ICT providers or other agencies coordinating broader smart city strategies, and it may be possible to facilitate these novel collaborations by codeveloping technologies and system feedbacks with the institutions and users that will ultimately engage with these new products.
This systematic review reveals that smart water research is moving toward contributing to climate adaptation and mitigation goals in municipal regions with both developed and developing water and electrical infrastructures. As smart water technology matures and becomes more established, it will become possible to empirically evaluate the impact or benefits gained from implementing smart technologies. There is promise in the technology development, but still much to learn about uptake, implementation, and results.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by funding from the Integrating Sustainability Across the Curriculum (ISAC) Initiative at University of Pennsylvania's Office of Sustainability.
