Review of research on performance indicators for water utilities

The efficient and sustainable management of water services poses key challenges worldwide. Water utilities play critical roles in community and environmental wellbeing; they ensure safe and dependable water supply and sanitation services. Therefore, it is essential to assess and optimize water utility performance, with a focus on sustainability.

Performance indicators are useful tools for assessing the operational, financial, environmental, and social aspects of water systems. Water utilities are expected to use performance indicators to quantitatively assess their water services, identify operational problems, set targets and measures, and ensure accountability. A manual of performance indicator has been published by the IWA (IWA 2016), for use by water utilities themselves. Performance indicators include common indicators, such as non-revenue water (NRW), employees per connection, and water tariff. These performance indicators are versatile, and not only serve as self-assessment tools for water utilities, but are also widely utilized for academic research by researchers around the world. At the same time, new indicators are being developed, especially in academic studies. The ultimate goal of these research efforts is to enhance the efficiency and effectiveness of water utilities. While numerous studies have utilized performance indicators, a gap still exists between the actual management of utilities and the outcomes of those research works. Put simply, while several investigations have employed indicators that align with the performance indicators of water utilities, the results of these studies have not necessarily been incorporated into improving water utility efficiency and effectiveness.

Water utilities of today are impacted by factors such as population growth, urbanization, climate change, technological advancements, and an aging population. It is crucial to focus not only on enhancing the management of water utilities by using pre-existing performance indicators but also by incorporating the findings of research on innovative performance indicators. To this end, a thorough review of research conducted using performance indicators is necessary, along with an overview of findings that can potentially be incorporated into the improvement of water utilities. Therefore, this study focuses on examining studies on performance indicators within the water sector that have been published as research papers including case studies, and presents a comprehensive summary of the prevalent trends. The text outlines the characteristics and limitations of the present research and proposes lines of inquiry necessary to adapt to changes in social and environmental conditions.

A number of studies have been carried out on indicators for water utilities. A Web of Science search using the terms ‘water utility’ and ‘performance indicator’ yielded only five relevant articles. In addition to these five articles, International Water Association (IWA) journals were searched using the same terms, resulting in 130 full-text available articles out of a total of 220 articles found. Articles related to wastewater and stormwater were excluded, leaving a total of 108 articles.

Based on the statistics, the author decided to focus this review on four main topics, namely (i) distribution pipes, (ii) leakage management, (iii) asset management, and (iv) utility management. The next section reviews the performance indicators used in these studies and summarizes the key findings of the research.

Performance indicators in distribution pipe surveys express the condition or state of the Water Distribution Network (WDN), as summarized in Table 1. Work in developing countries highlight the topics for distribution pipe performance indicators, which range from hydraulic indicator to system performance. A case study in a South African city (Strijdom et al. 2017) employed the minimum pressure head as a performance indicator, whereas a case study in Ethiopia (Tekile & Legesse 2023) evaluated performance in terms of reliability, resilience, and vulnerability, and assessed water velocity, water pressure, and water quality as Water Quality Index (WQI). Another study in Ethiopia (Beker & Kansal 2023) evaluates location of isolation valves by hydraulic performance indicators such as the number of valves needed to successfully isolate an incident area. Meanwhile, a study in Bangkok, Thailand (Lipiwattanakarn et al. 2019), uses energies as performance indicators, which refer to input energy, outgoing energy through leaks, and energy delivered to users. Naamani & Sana (2021) focused on the WDN of an urban area in Oman, evaluated using IWA performance indicators ranging from inspection and maintenance to calibration, rehabilitation, water loss, and water quality. The performance criterion in each category has several indicators. For example, the inspection and maintenance criterion have six indicators: (i) pumping inspection and maintenance, (ii) Storage tank cleaning, (iii) network inspection, (iv) gate valve inspection, (v) leakage control repair, and (vi) hydrant inspection. The study concludes that most of the 37 performance indicators for the system show good results. As listed in Table 1, performance indicators target different levels of phenomenon from hydraulic pressure to inspection rate.

WDN studies are reviewed also from theoretical perspective. Partitioning applies linear programming to optimize a large WDN by dividing it into smaller DMAs; this approach optimizes the transport function by boundary pipes and DMA uniformity (Creaco et al. 2022). Performance indicators in WDN research range from sensor data, hydraulic parameters to network modeling parameters. Sensor data research targets interpretation or handling of data from pressure or flow sensors (Laucelli et al. 2016), while hydraulic parameters consist of water age or water velocity in pipes (Armand et al. 2018), satisfactory rate, calculated as the actual water demand divided by the expected demand (Liu & Kang 2022). Network modeling parameters research targets resilience, vulnerability, and reliability (Beker & Lal Kansal 2022), energy use minimization (Campbell et al. 2016), characteristics or performance of the network, such as demand coverage, time to detect contamination, or volume of water consumed (Rath & Gupta 2015), or neuro-fuzzy controller to regulate pressure within a WDN (Augusto et al. 2021). One early report focused on contamination caused by backflow (Di Nardo et al. 2015) and techniques for water network sectorization and protection against network contamination. District Metered Area (DMA) is one of the concepts for sectorization and is expected to work in case of malicious attacks. In a recent review, Wang et al. (2022) divided the research scene into four distinct categories: pipe burst mechanism, pipe burst detection, pipe burst prediction, and sensor utilization. A case study in Turkey (Bozkurt et al. 2022b) evaluated the performance of four local administrations that used indicators including water balance monitoring conditions. While some recent studies have incorporated additional perspectives in their evaluations, most studies have focused primarily on the implementation and refinement of mathematical techniques. Therefore, it is recommended that future research consider the actual conditions of the water distribution pipeline network of water utilities, including societal changes such as depopulation and carbon neutrality.

Optimal DMA design is another research objective, and hydraulic or network modeling parameters are their research target. Galdiero et al. (2016) used resilience deviation index, which was calculated as the ratio of the sum of the difference between the start and partitioning of the total head divided by the sum of the difference between the starting head and minimum head. Statistical indicators such as root mean square error (RMSE) are commonly employed to assess the model fit (Barrela et al. 2017). A review of statistical models (Barton et al. 2022) for water distribution networks indicates the importance of data collection process to improve data quality and quantity. However, it is difficult to obtain continuous data; real-world continuous measurements always involve missing data (Laucelli et al. 2016). A study comparing different methods for filling concluded that the proposed model based on a weighted function for forecast and backcast had a good performance (Barrela et al. 2017). A recent case study in Istanbul (Kizilöz 2022) has attempted to use artificial neural networks (ANNs) to make predictions. Similarly to the studies on WDN, DMA design studies focus on proposing mathematical methods and statistical evaluation. However, it should be pointed out too-fine parameters with new methods are not always necessary, because water pipeline data are usually periodic data with a steady or constant pattern in nature.

One study employed indicators such as failure rate, corrosion protection, and condition of pumps, for the purpose of failure mode and effects analysis (Tuhovčák et al. 2018). Kakoudakis et al. (2018) used weather conditions as explanatory factors to construct an ANN. A study conducted in the Netherlands (Wols et al. 2019) also evaluated weather conditions and concluded that temperature affected the performance of asbestos cement and cast iron pipes. A recent study presented an integrated water management system, using both operational and external data as input (Bettin 2023). The ultimate goal of these research efforts is to enhance the efficiency and effectiveness of water utilities. Water utilities are facing challenges from social and environmental issues. Future research should be oriented toward such direction, as with some of above research.

Overall, review of studies in developing countries highlight the use of performance indicators at multiple levels. From reviews of WDN and DMA studies, it can be concluded that research tend to place too much emphasis on the development and refinement of mathematical methods. Future research should seek improvements by exploring relationships among indicators of multiple levels including societal and environmental conditions.

Leakage Management studies are always related to NRW. These studies can be divided into several categories, as described in Table 2. The daily NRW totals 346 million m³ worldwide (Liemberger & Wyatt 2019). Thus, comprehensive frameworks have been developed to minimize leakage (Bozkurt et al. 2022a). The framework includes different levels of components such as performance indicators ranging from main length, the number of valves, operating pressure, and the number of unreported leaks. Wright et al. (2014) used an adaptive WDN with dynamically reconfigurable topography to minimize leakage; the pressure head and flow were monitored.

Modeling studies for NRW reduction always utilize performance indicators about water network structure, such as system input volume (SIV) and number of junctions (NJs). A model by Kızılöz & Şişman (2021), found that a serial triple diagram model, which employed SIV, NJ, and mean pressure network (MPN), mean age of pipes (MAP) along with errors has a better performance compared to conventional models. Another recent study (Kizilöz 2021) used an ANN that considered MPN and age as references, that results suggested water utilities to replace old pipes and apply pressure management. Another study in Turkey (Şişman & Kizilöz 2020) compared an ANN with Kriging model for NRW prediction by using conventional performance indicators such as SIV, NJ, MPN, MAP. As a result, the latter model performed better and required fewer inputs.

Concept of water balance is important for NRW study and there are several performance indicators originally published in Lambert & Hirner (2000). The study considers total as SIV, which consists of authorised consumption and water losses. Authorised consumption can be divided into billed and unbilled, further broken down into metered and unmetered. Water losses is composed of apparent losses and real losses. Real losses account for leakage on transmission and distribution pipes, leakage at storage tanks, and leakage on service connections. Apparent losses are unauthorised consumption and customer metering in accuracy, therefore good performance of water meters is essential for NRW management. Various meters have been tested at different flow rates and water pressures (Ethem Karadirek 2020); volumetric meters were found to have the highest accuracy. Meter accuracy was also investigated in Ecuador (Szilveszter et al. 2017); 44% of meters were found to have optimal performance, with weighted error derived using two different consumption patterns differing by 0.95%, which significantly impacted water balance and performance. Recent study by Wyatt (2020) revisited several those indicators and concluded % based indicators have a flaw.

There are some recent studies using additional performance indicators or an aspect. One study (Cassidy et al. 2021) used cloud-based tools to reduce real losses, using various monitoring data including flow and meter data as key indicators. Another study incorporated the financial aspect into a model (Yilmaz et al. 2022), then calculated the optimum economic water loss level as 8.62% in the case. Similarly, Malm et al. (2020) investigated economic leakage levels in Nordic countries by calculating the water balance and leakage based on the minimum night flow, from both top-down and bottom-up perspectives. A water leakage evolution model was constructed by Guo et al. (2021) for the purpose to predict background leakage. The model assumed that the leak flow rate was initially low but became reportable after some time, utilizing indicators such as age, diameter, material, and length of pipes. Created motel was evaluated by statistical indicators such as the mean absolute error were used for evaluation. An unique study in Korea (Jun et al. 2020) developed a fuzzy function-based model targeting on consumer satisfaction. Various scenarios were compared by calculating the performance measure of fuzzy membership for the whole area.

Other NRW reports are case studies. A case study in Malaysia (Lai et al. 2020) used a systems-thinking approach to represent key relationships in causal loop diagrams to reduce NRW. A case study in India (Bandari & Sadhukhan 2023) employed data envelopment analysis (DEA) to assess the efficiency of urban water utilities using NRW as the input and per capita water supply and cost recovery as the outputs. A case study in Turkey (Akdeniz 2022) examined the impacts of water meter replacement and passive/active leakage management, which reduced NRW from 53 to 37%. A case study in Brazil (de Santi et al. 2021) listed leakage management practices and applied them to 76 municipalities to discuss influencing factors. In Ontario, Canada, Jenks et al. (2023) used a mobile testing unit to collect minimum night flow and pressure data in the DMA.

Overall, review of leakage management research shows it ranges from studies that comprehensively assess an entire network to case studies of individual initiatives for reducing NRW. The latter group includes works on meter replacement and how to measure minimum night-time flow. In contrast, network-wide studies focus on water pressure, pipe age, meter accuracy, and budgets. All of these topics have been researched intensively. However, as far as the author has reviewed, pipeline material and ground information, such as corrosive soil, have not been well studied except a study by Guo et al. (2020). In other words, current leakage management studies are primarily concerned with improving NRW in the present, without adequate consideration of maintaining the improved state over time. In terms of maintenance, whole-network evaluations that incorporate budget considerations are necessary. Additionally, comprehensive evaluations of individual aspects, such as the replacement of old pipes and their effect on leakage, are important. To ensure effective leakage management, future studies should include sustainability parameters, including the deterioration of aged pipelines and their replacement, while taking budget and carbon neutrality into consideration.

Asset management is an important aspect of water utility management and covers a wide range of topics. The WDN and leakage management are closely related to asset management. However, the concepts are not aligned in the same class, and since asset management is a higher level concept, it is discussed separately in this section. In particular, those that include asset management in their titles are discussed here. Performance indicators used in those studies are summarized in Table 3.

Asset management is important for achieving Sustainable Development Goal 6 (Alegre et al. 2020). This growing need was highlighted in a report on asset management activities of past 20 years in the Netherlands (Beuken et al. 2020). The report contains several important lessons. In particular, it emphasizes that the current drinking water supply paradigm will change; asset managers should not rebuild existing assets but rather redesign them to fit the set of future states. Objective, representative, complete datasets are crucial for asset management. Reports on nationwide asset management activities are useful. For example, one Japanese case study (Sakai et al. 2020) assessed the current condition of water distribution pipes nationwide. Old pipes and ineffective water were surveyed; the situation varied according to the population served. A French study discussed resource management strategies for the year 2070 (Rulleau et al. 2020) under four scenarios that considered water supply conditions and long-term strategies that should be adopted by stakeholders.

Data quality is also essential for asset management. In a Swedish case study (Okwori et al. 2021), asset management indicators included static pipeline data (age and length) and inspection, failure, and maintenance data. It was emphasized that poor-quality pipe network datasets and non-interoperability of asset management tools created inappropriate data silos.

Several criteria have been used to evaluate water supply assets. A study in Spain (Pardo & Valdes-Abellan 2019) questioned whether pipe replacement should be based on age alone. The study prioritized methods that focused on unit losses, economic factors including energy and water savings, and economic factors for the replacement plan. On the other hand, a Portuguese literature (Ramalho et al. 2020) warns against focusing on efficiency, as inefficiencies in the current system may have long-term effects. In the study, indicators such as NRW volume, and real loss per input volume are selected. Their behaviour is compared in different years and the study concludes that NRW projects need a certain time frame and intangible benefits should be considered in the decision-making process. The use of a single indicator may ignore important information. A degradation index or infrastructure histogram facilitates communications among stakeholders (Estruch-Juan et al. 2020), as they contain information on all pipes within a network. However, the choice of indicators is sometimes difficult. A case study in Egypt employed 25 performance indicators (Abdelghany & Abdel-Monem 2019), including number of bursts, employee-to-customer ratios, and NRW. Twenty of these indicators have been used by the Egyptian Water and Sewerage Regulatory Authority in the long term, whereas five are new.

Overall, asset management helps water utilities to define the necessary information and direction for their operations. Review of current studies revealed data quality and selection of indicators are important. At each stage, attention needs to be paid to the quality of the input data, the input method, the input elements, the interpretation of the output and the actions to be taken. As mentioned in several studies, the acquisition and integration of good quality data, is a prerequisite for asset management. Future research should not only focus on solving specific problems, but also on generalizing the procedures. In addition, current asset management research is weighted toward solving specific problems, such as pipeline damage, and does not focus on the management part of the problem, such as human and financial resources. Multifaceted discussions on each of these points are needed to improve the quality of asset management. In addition, many previous studies have focused on pipelines. Although pipelines are important, future research should also focus on asset management of treatment plants.

It is important to evaluate water utilities by using performance indicators. Performance indicators used in those research works are summarized in Table 4. The principles of benchmarking are basically similar, but the methods of evaluation are diverse. Traditional management studies take a comprehensive approach, using a few key performance indicators. A case study in China evaluated water supply services using several performance indicators, including utilization rate of water plant capacity, or call-center connection rate (Li & Han 2020). Performance scores of four utilities in Jiangsu province were calculated in several ways: the service, operation, resource, asset, finance, and personnel scores differed. In India, an integrated management information system is used to monitor rural drinking water. Different indicators such as the coverage rate are appropriate for evaluating water supply in distinct regions (Wescoat et al. 2016). Processing data obtained from information systems can be challenging. A case study in Portugal (Carriço et al. 2020) applied a stepwise approach to integrate data from different systems. First, all necessary files were imported and metrics were calculated; this process is demanding in terms of human and financial resources. The usefulness of existing performance indicators for small water utilities were examined by Haider et al. (2015). Indicators were selected via outranking and group decision-making. Examples included the number of days with water restrictions, average daily per capita domestic water consumption. These studies will be useful to identify the cause and solve current problems of a particular water utility.

The DEA approach is sometimes used to analyze efficiency employing the indicators. A DEA case study in Tanzania (Gidion et al. 2019) employed the NRW, metering ratio, and water quality compliance as performance indicators. Another study by the same group simulated best management practices (Gidion et al. 2022). DEAs of South African water utilities (Ngobeni & Breitenbach 2021) were used to construct four models; the best performance was shown by a model that included the undesirable output as a ratio to the desirable output. The DEA model was further explored in a report from Tanzania (Gidion 2023). The selected indicators included the NRW, staff per 1,000 connections, metering ratio and average hours of water supply. Absolute and ratio input data were compared; ratio data were found to improve model performance. DEAs are suitable for comparison of a large number of utilities. Therefore, it would be useful to identify poorly managed water utilities among many. However, it is important to note that DEA results may vary based on the indicators selected. Thus, when employing DEA, it is essential to carefully consider the selection of indicators taking into account social conditions.

Water utility analyses often require a variety of perspectives given their diversity and the complex issues that they address. Dealing with customer complaints is very important. One study set ‘listening to customer claims’ as the objective and developed a hierarchical model including odor, overbilling, and connection delays; the model was applied and evaluated for a water utility in Algeria (Bouchraki et al. 2021) from 2014 to 2018. The remunicipalization of water services in Naples, Italy and Paris, France has also been studied (Turri 2022). The Paris water authority Eau de Paris prepared a contract with the people (‘Contrat d'Objectifs’) that proposes meeting objectives based on 130 performance indicators. A report from Mozambique (Lima et al. 2021) examined public–private partnerships with a focus on contracts. Interviews with 15 people revealed that the integration of key risk indicators into public–private partnership contracts improved risk management. A regulatory impact assessment study in Brasilia informed policy-making (de Carvalho et al. 2019); Resolution Law No. 08/2016 attempts to maximize each policy option within the Brazilian water sector and facilitate the public understanding of decision-makers. Environmental aspects are also important. One review (Boldrin & Formiga 2023) suggested that Life Cycle Assessment was the most widely accepted method for measuring the environmental performance of water utilities. The World Bank uses a decision tree that is continuously updated to include climate change impacts, robustly and resiliently supporting climate-informed project investment decisions (Rodríguez et al. 2021). One investment study focused on sanitation projects (Hutchings et al. 2018) in an attempt to classify projects by investment area; more comprehensive indicators and measures were found to be required. A case study on subsidies explored water utility management using the water connection, connection rate, subsidy receipt as indicators (Abramovsky et al. 2020). Performance indicators are originally intended to improve business environment of water utilities. In that regard, it is recommended to utilize performance indicators including societal, environmental and monetary aspect, as depicted in those studies.

Overall, review of research on utility management shows various perspectives that reflect the diverse environments in which water utilities operate. Selecting appropriate indicators is therefore crucial for studying and evaluating utility performances. While a portion of the ongoing research employs perspectives such as PPP and life cycle assessment, a challenge arises from a lack of research utilizing perspectives necessary for future society, such as considerations for depopulation, aging, and carbon neutrality. Additionally, it is requested for academic research to conform to a uniform analytical methodology. For this purpose, a comprehensive DEA should be utilized. In the analysis, it is also essential to choose adaptable indicators within the analysis that can accommodate future changes in social conditions.

The author reviewed water supply performance indicators that evaluate distribution pipes, leakage, asset and supply management, and other aspects.

As a perspective, there are a lot of good mathematical and theoretical studies on distribution pipes and leakage management, but there is still a gap in the application to policy-making of water utilities. Therefore, future research should address the problems of water utilities, by using multiple levels of performance indicators including social and environmental context in the long term. Asset management and utility management studies address diverse and current problems faced by water utilities. However, there is still room for improvement in the collection of good quality data and the selection of indicators. Although a methodology such as DEA already exists, further research should aim at standardizing the methodology for data collection, processing, and integration.

As a suggestion to specific research area, carbon neutrality should be included in future research directions. In addition, pipeline materials and soil information should be included in leakage management. Furthermore, asset management studies should cover not only distribution pipelines but also treatment plants, with including additional indicators about human and financial resources.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.