Assessing impacts of climate crisis on water distribution systems though service inefficiency indicator

Climate change can lead to complex effects on water systems, necessitating adaptive strategies for ageing infrastructure and maintenance. Key approaches in recent literature include implementing dynamic replacement schemes that consider both pipe-intrinsic factors and climate data to predict failure rates and prioritize interventions (Fan et al. 2023). Water utilities are adopting more sophisticated failure prediction models incorporating climate variables like temperature and precipitation, allowing for targeted maintenance efforts, optimizing at the same time pipe material selection relaying on regional climate projections, with ductile iron pipes showing better performance in colder regions and cast iron pipes in warmer areas (Bruaset & Sægrov 2018; Fan et al. 2023). Additionally, the impact of soil-related factors is gaining attention, as climate change affects soil moisture patterns and ground movement (Żywiec et al. 2021). Utilities are considering soil shrink–swell potential, especially in clay-rich soils, when planning pipe replacements and maintenance schedules (Barton et al. 2019; Żywiec et al. 2021). The seasonal variation in failure rates, with higher occurrences during winter for certain materials and summer for others, is informing more nuanced maintenance strategies (Kakoudakis et al. 2018; Żywiec et al. 2021). In this context, adaptive management approaches using scenario planning are becoming crucial for projecting future climate impacts and developing flexible strategies. Operational adjustments, such as advanced pressure management systems and leak detection technologies, are being implemented to reduce stress on ageing pipes during extreme weather events, as well as long-term planning horizons which are being extended to 50–80 years to account for prolonged climate change impacts, with new infrastructure designed for greater climate resilience (Fan et al. 2023).

Advanced statistical and machine learning models, such as Artificial Neural Networks and Evolutionary Polynomial Regression, are also being employed to predict both short-term failure occurrences and long-term annual failure rates based on weather conditions, allowing for more proactive management of water distribution networks (Kakoudakis et al. 2018; Barton et al. 2019).

Furthermore, climate variability presents a significant global challenge, impacting water availability, quality, groundwater resources, and public health, underscoring the critical need for adaptive strategies to address these interconnected issues in the context of climate change (Ahmed et al. 2020).

Water utilities are proactively adapting to climate change by employing strategies to enhance the resilience and sustainability of water systems. These measures address hazards such as rising temperatures, droughts, extreme precipitation, and shifting weather patterns, which significantly affect water infrastructure (Lyle et al. 2023). Adaptation plans focus on improving operational efficiency, reducing water leakages, maintaining supply quality, and achieving cost savings while ensuring resilience against future climate impacts (EPA 2015). By addressing these challenges through targeted planning and innovative maintenance approaches, water utilities can ensure the continued provision of safe and reliable services in a changing climate.

In order to do this, it is fundamental for water managers to understand and quantify impacts of climate change and water crises on their water supply systems in terms of system failure or service inefficiency due to water scarcity. The quantification of water shortage impacts on distribution pipes employs a multifaceted approach in recent scientific literature, integrating various methodologies to assess system resilience and potential risks. Advanced modeling techniques, such as Time Domain Reflectometry (TDR) inversion (Scarpetta et al. 2023), enhance leak detection capabilities, while probabilistic frameworks enable large-scale network analysis. Risk assessment strategies incorporate multi-hazard Bayesian Network models (Bozorgi et al. 2021) and segment-based evaluations (Hernandez & Ormsbee 2021) to comprehensively gauge system vulnerabilities. Chemical and biological analyses, including adsorption–desorption studies of contaminants (Somer et al. 2021) and invertebrate monitoring (Gunkel et al. 2022), provide crucial insights into water quality and secondary contamination risks during shortage periods. Economic modeling, exemplified by the agroeconomic water scarcity indicator, offers a broader perspective on regional water scarcity hotspots and their economic implications (Schmitz et Al. 2013). Probabilistic framework for parametric modeling of leakages in water distribution networks shows good performances in large-scale applications (Serafeim et al. 2022).

Additionally, researchers have proposed diverse approaches to quantify the impact of water scarcity on water distribution networks in terms of service inefficiency. These approaches include the ‘water scarcity footprint’ concept, which measures the impact of additional water consumption on availability and network efficiency (Lee et al. 2019). Data Envelopment Analysis (DEA) and other efficiency measurement tools provide insights into operational efficiency, highlighting areas for improvement (Palomero-González et al. 2021). Quantifying the level of service in water distribution networks (Khan et al. 2009) also helps identify gaps and inefficiencies. Water distribution network (WDN) sectorization is utilized to evaluate efficiency by analyzing the real sectorization of the network (Gomez et al. 2014), helping to assess current structure and identify areas for improvement. Combining sectorization with pressure management can minimize water losses (Klosok-Bazan et al. 2021), as dividing networks into smaller sectors aids in controlling pressure, ensuring uniformity, and reducing leaks and unprofitable water usage (Giffoni et al. 2021). These diverse methods collectively contribute to a more nuanced understanding of water shortage impacts, facilitating informed decision-making in water resource management and infrastructure planning. By synthesizing data from these varied approaches, water management professionals can develop more effective strategies to mitigate the consequences of water shortages on distribution systems, ultimately enhancing the resilience and sustainability of water infrastructure.

Sicily region (Italy) is experiencing a severe water crisis due to drought and climate change, with significant impacts on agriculture and urban water supply and inhabitants experiencing water scarcity and restrictions (UNESCO & UN-Water 2020; Ide 2024; Monforte et al. 2024; Pecorino et al. 2024).

In March 2024, the Sicilian Water Regional Authority declared a water scarcity emergency in six provinces, including Messina. The drought, deemed the worst in decades, has led to water rationing in over 93 municipalities, affecting 850,000 residents. Compared to February 2023, the region has seen a 70% reduction in agricultural water availability and a 23% decrease in drinking water supply. Messina city specifically has experienced a 15% reduction in drinking water supply (D'Orazio 2024; James 2024).

The main purpose of this study is to assess the impact of current water crisis on Messina City's water distribution system performance. The research proposes a service Inefficiency Indicator (II) that combines three factors: unsatisfactory water distribution amounts, pressure regime disservices, and WDN resilience.

Researchers evaluate water network performance through various metrics tailored to specific goals. Studies focus on balancing efficiency and resilience (Li & Yang 2011), trade-offs in sustainable systems (Piratla & Ariaratnam 2012), and vulnerability assessment (Hamouda et al. 2009). Hydraulic performance is also important, with Tabesh & Saber (2011) using pressure, velocity, and reliability indices. Loucks (1997) introduced a Sustainability Index (SI) combining reliability, resilience, and vulnerability. Aydin et al. (2014) further refined this by incorporating both pressure and water quality into SIs for network zones. This variety highlights the complexity of performance assessment and the need for context-specific metrics.

This study examines the WDN of Messina City, utilizing recent partitioning solutions to optimize supply. The network was divided into sub-areas, each modeled using EPANET software (Rossman 2000). A methodology for calculating the service II was applied to four distinct scenarios: the pre-crisis working scenario, the current water scarcity situation, a water rationing plan with intermittent supply, and a scenario with reduced water leakage. Results compare the II results across these scenarios to evaluate the effectiveness of different water management strategies in addressing the city's water crisis.

This section contains first a description of the case study of Messina city WDN. It describes the network's characteristics and how it's modeled mathematically using EPANET software. Then, the proposed methodology is explained, with its mathematical formulations and the process for calculating the II for each network node. Pressure values, demand and supply at each node are used to assess the occurrence of dissatisfaction in the amount of distributed water, with the Unsatisfaction Indicator (UNS), the disservices in pressure regime, with the Disservice Indicator (DISS) and the resilience of water distribution network, with the Resilience Indicator (RES). These individual indices are then combined to create a comprehensive II for every node. To analyze the entire network in more detail, it's further divided into sub-networks based on tank locations. An overall II score, namely Zonal Inefficiency Indicator (ZII) is then calculated for each zone and each scenario simulation. This approach allows us to compare the effectiveness of different network management strategies in coping with water crisis situations.

Messina, a coastal city in Sicily, Italy, boasts a population of 250,000. Its drinking water supply relies on a network of three main aqueducts. The Santissima Aqueduct, the oldest, taps into the nearby Nebrodi Mountains, providing roughly 150 liters per second (l/s), (Borzì & Bonaccorso 2021). The remaining two, the Alcantara and Fiumefreddo aqueducts, source their water from Mount Etna's groundwater aquifer, supplying Messina with approximately 300 and 1,000 l/s, respectively (Borzì et al. 2019, 2020, Borzì 2023).

In the whole WDN, the elevation of supply junctions ranges from a minimum of 3 meters above sea level (m.a.s.l) at the coast to over 120 m.a.s.l. in higher regions. Maintaining adequate pressure across a network with such substantial elevation disparity presents a significant challenge for water management personnel. Low pressure in high-altitude areas can disrupt water service for residents, while excessive pressure near the coast can lead to increased water loss through leaks. Identifying practical and cost-effective solutions to address these pressure-related issues is critical.

A previous study (Borzì 2022) investigated, in terms of pressure regime optimization and sustainability, the potential benefits of partitioning Messina's WDN based on the strategic placement of its main tanks, following established methodologies outlined in prior research (Morrison et al. 2007). The findings demonstrate that water network partitioning (WNP), implemented using a tank location criterion, significantly improves network pressure ratio (NPR) in districts characterized by homogenous topography. However, for districts with significant elevation variations in supply junctions, WNP alone proves insufficient in achieving sustainable NPR. For the latters, introduction of pressure regulating valves proves to be an effective solution for enhancing NPR sustainability at both the district and WDN-wide level.

This section details the modeling and analysis of Messina's WDN to evaluate pressure distribution throughout the system. EPANET (Rossman 2000), a widely recognized software program developed by the US Environmental Protection Agency (EPA) for water distribution system modeling, was employed for this purpose. The analysis of water distribution networks using EPANET has become a cornerstone in scientific literature, offering researchers a powerful tool for modeling and simulating various aspects of these complex systems. EPANET's versatility allows for comprehensive hydraulic modeling, enabling the simulation of flow rates and pressure distributions across pipe networks (Jawale et al. 2022; Hari & Praveen Chari 2024). This capability is crucial for identifying areas with insufficient or excessive pressure, which can lead to service issues or leaks. Beyond hydraulics, EPANET facilitates water quality modeling, allowing researchers to simulate chlorine residual levels and incorporate multi-component kinetic models for analyzing disinfection by-products (Dong et al. 2024). The software's utility extends to leak detection and assessment, where pressure-based methods and water loss quantification play pivotal roles in maintaining system integrity (Hari & Praveen Chari 2024). Researchers often integrate EPANET with optimization algorithms, such as genetic algorithms, to enhance network design and operation, balancing multiple objectives like cost minimization and reliability maximization. The software's application in reliability analysis is noteworthy, with studies utilizing EPANET to calculate resilience indices and simulate failure scenarios (Moghaddam et al. 2022).

This study presents four distinct scenario simulations for the Messina's WDN, each representing different conditions related to water supply and distribution. The simulations are conducted using the EPANET model, which incorporates carefully designed water demand patterns based on real-world data from Messina's city center.

The second scenario simulates the impact of the current water scarcity on the Messina WDN. In this simulation, the system experiences a 15% reduction in water supply from aqueducts and urban wells. However, the actual system configuration maintains a 24-h distribution pattern. This scenario allows for the assessment of how the existing infrastructure copes with reduced water availability without implementing any mitigation measures.

The third scenario represents short-term measures implemented to address water scarcity. In this simulation, the Messina WDN operates with a 15% reduction in water supply from aqueducts and urban wells, like in Scenario 2. However, a water rationing plan is applied through intermittent water supply, resulting in an 18-h distribution pattern (Figure 4(b)). This scenario demonstrates how the system performs under restricted supply conditions and allows for the evaluation of the effectiveness of short-term mitigation strategies.

The fourth scenario simulates long-term measures to cope with water scarcity or potential future water crises. In this simulation, the Messina WDN incorporates water leakage reductions while maintaining the 24-h distribution pattern. This scenario enables the assessment of infrastructure improvements and their impact on the system's resilience to water scarcity.

The 24-h water demand pattern used in the EPANET model (Figure 4(a)) was carefully designed to reflect real-world water usage in Messina. This pattern was derived from data collected at an experimental site in Messina's city center that recorded actual water demand levels throughout a 24-h period (Borzì 2022). This approach ensures that the simulations accurately represent the typical water usage patterns within the city. The water demand patterns employed within the EPANET model exhibit a key distinction between the 18-h and 24-h scenarios, as illustrated in Figure 4(b) and 4(a). Notably, the 18-h pattern reflects a constant water demand from 3:00 AM to 10:00 PM. This specific modeling choice is necessitated by the prevalence of private water tanks utilized by a significant portion of Messina's residents. These private tanks serve as a crucial stopgap measure to address the absence of overnight water supply. They function essentially as individual buffer systems, accumulating water resources during the day for release at night when the primary supply is unavailable. The continuous demand exerted by these private tanks on the WDN necessitates the implementation of a constant water demand pattern within the EPANET model. This approach is necessitated by the software's inherent limitations in simulating the behavior of individual private tanks. Telemetry data from urban tanks confirms this approach. Figure 4(c) presents a representative example of the output flow trend observed from a single tank within Messina's WDN.

This section details the proposed methodology for assessing the performance of Messina's WDN at the nodal level. The methodology centers on calculating a comprehensive II for each network node. Pressure values, water demand, and supply at each node are used to evaluate three key aspects of network performance:

• UNS: Reflects potential user dissatisfaction due to insufficient water delivery volumes at a particular node.

• DISS: Captures the occurrence of disservices arising from pressure regime fluctuations at a node.

• RES: Evaluates the resilience of the network at each node, considering its ability to recover from unexpected disruptions.

These individual indicators are mathematically formulated and then combined into a single, composite II for each node. This approach provides a holistic perspective on network performance, encompassing water delivery adequacy, pressure stability, and resilience.

The WDN is further subdivided into sub-networks based on the locations of water tanks. An overall II score is then calculated for each sub-network and for each scenario simulation. This multi-tiered evaluation allows for a comprehensive comparison of different network management strategies and their impact on the overall sustainability of Messina's WDN.

The proposed II is firstly evaluated for each node of the WDN. It is a combination of the three indicators UNS, DISS and RES which can vary over space and time and are evaluated for each node.

The RES indicator can assume values between 0, corresponding to a completely un-resilient WDN and 1, corresponding to a completely resilient system.

The II serves as a valuable tool for assessing network performance. It employs a clear and intuitive scale ranging from 0 (representing the highest level of efficiency and the lowest level of inefficiency) to 1 (indicating the highest level of inefficiency). This straightforward range facilitates the interpretation and comparison of results across the network. The specific II values used in this study, which translate these inefficiency levels into corresponding scores, are detailed in Table 1.

Table 1

Inefficiency indicator range definition

View Large

This section employs the previously described methodology to assess the performance of Messina's WDN under various water scarcity conditions. Four distinct scenarios are evaluated to simulate impacts of the current water crisis in terms of WDN service inefficiency and explore mitigation strategies.

The first scenario serves as a baseline, representing the WDN in its pre-crisis state with normal operating conditions. This scenario utilizes a 24-h water distribution pattern (Figure 3) and the regular water supply volume from aqueducts and urban wells. In this scenario all districts (San Licandro, Trapani, Torre Vittoria, Noviziato-Vascone, Mangialupi) boast relatively low ZII values (ranging from 0.07 to 0.27). This indicates efficient network operation before the onset of water scarcity. These low baseline values suggest that the WDN functioned well under normal circumstances, in terms of network pressure regime and amount of water supplied with respect to water demand during the 24 h. Table 3 reports details regarding UNS, DISS, and RES values for each of the five districts in the first scenario simulation.

The second scenario simulates the current water scarcity situation. Here, a 15% reduction in water supply from aqueducts and urban wells is modeled while maintaining the existing 24-h distribution pattern. This scenario allows us to isolate and evaluate the impact of reduced water availability on network performance. A significant rise in ZII values is observed across all districts compared to Scenario 1. For instance, San Licandro's ZII rises from 0.23 to 0.80, and Noviziato-Vascone's ZII rises from 0.27 to 0.83, reflecting a substantial deterioration in performance, both districts jumping from an acceptable efficiency of WDN to a total inefficient service. The latter two districts, which suffer more for water scarcity, are also the two districts with the most inhomogeneous topography. The Noviziato-Vascone district has a difference in altitude values of 66 meters between the highest and lowest district supply junction and the San Licandro district has a difference in altitude values of 82 meters between the highest and lowest district supply junction. These substantial elevation differences within Messina's WDN pose challenges for maintaining consistent water pressure and supply throughout the system, as highlighted in a previous study (Borzì 2022). Results in terms of ZII for the latter two districts highlight how water scarcity exacerbate issues related to pressure regime regulation and supply–demand mismatches. The other three districts (Mangialupi., Trapani and Torrevittoria) finds in this scenario configuration values of ZII corresponding to high inefficiency of the WDN. This highlights the negative impact of water scarcity on the network. With a reduced water supply, the WDN struggles to maintain pressure and deliver water effectively, leading to inefficiencies. Table 4 shows information on UNS, DISS and RES values for each of the five districts in the second scenario simulation.

The third scenario depicts potential short-term measures implemented in response to water scarcity. The water supply reduction of 15% from Scenario 2 is maintained. However, an intermittent water supply strategy is introduced, with water distribution limited to an 18-h period (Figure 3(b)). This scenario assesses the effectiveness of rationing as a short-term solution for managing water scarcity. Compared to Scenario 2, ZII values show a slight improvement for most districts. San Licandro's and Noviziato-Vascone's ZIIs, for example, decrease respectively from 0.80 to 0.77 and from 0.83 to 0.77, jumping from a total inefficiency of the service to a high inefficiency of WDN. Trapani and Mangialupi districts reach in this scenario configuration a moderate inefficiency in ZII values, while Torre Vittoria district maintains a high inefficiency ZII classification, even with a slight improvement in ZII score, from 0.79 to 0.71. This suggests that water rationing, while not a perfect solution, may offer some mitigation for the inefficiencies caused by water scarcity. By limiting water supply duration, rationing can potentially help regulate pressure and distribution patterns, leading to a modest improvement in efficiency. Finally, Table 5 reports UNS, DISS and RES values for each of the five districts in the scenario third scenario simulation.

The fourth scenario envisions long-term measures to address potential water scarcity or future water crises. Here, the WDN is modeled with a focus on water leakage reduction strategies, with a 24 h water distribution pattern (Figure 3(a)). This analysis investigates the potential benefits of proactive leakage management in enhancing network sustainability and mitigating the impact of water scarcity in the long term. Scenario 4 shows a more substantial decrease in ZII values compared to Scenario 3. A significant decrease in ZII values is observed across all districts even compared to Scenario 1. In this configuration, Torre Vittoria, Mangialupi and Trapani districts reach a ZII score of total efficiency, while San Licandro and Noviziato-Vascone reach a ZII score of acceptable efficiency. This underscores the significant potential of proactive leakage reduction strategies in enhancing network efficiency. By minimizing water losses, leakage reduction strategies decrease the overall water demand on the network. This potentially reduces the reliance on rationing measures, which can come with drawbacks like issues in network optimal pressure regime maintenance, especially in districts with non-homogeneous topography. Addressing leaks within the system helps maintain adequate pressure levels throughout the network, which finds evidence also in recent literature (Ortega-Ballesteros et al. 2022; Alsaydalani Majed 2024). This ensures more effective water delivery and reduces the risk of pressure fluctuations that can further exacerbate inefficiencies, ultimately leading to improved overall performance. This scenario emphasizes the importance of prioritizing leak detection and repair programs as a long-term solution to promote network sustainability and minimize inefficiencies, also finding confirmation in recent literature (Ahopelto & Vahala 2020). Table 6 reports information on UNS, DISS and RES values for each of the five districts in the fourth scenario simulation.

This study investigates the effects of water scarcity on the WDN of Messina, Italy. A novel ZII is introduced, accounting for dissatisfaction in delivered water volume, pressure disruptions, and network resilience to comprehensively quantify service inefficiency. ZII represents a useful and easy tool in evaluating water scarcity's impact on WDNs, with its comprehensive perspective, facilitating targeted interventions for water managers, data-driven decision-making regarding mitigation strategies, and improved performance monitoring.

The analysis on Messina's WDN reveals that ZII values across all districts were low under normal operating conditions, signifying efficient network performance. However, a simulated 15% reduction in water supply, representing the ongoing water crisis in Sicily, resulted in a significant increase in ZII values, highlighting network inefficiencies arising from water scarcity. This effect was most pronounced in districts with significant elevation variations, where maintaining adequate pressure throughout the network presents inherent challenges. While implementing short-term water rationing (18-h supply) yielded some improvement in ZII compared to the water scarcity scenario, inefficiencies persisted. The study identified proactive leakage reduction as the most effective mitigation strategy. When this scenario was modeled, ZII values across all districts substantially decreased compared to other scenarios, even surpassing pre-crisis levels. This underscores the long-term benefits of proactive leakage reduction in enhancing network efficiency and mitigating the impacts of water scarcity.

Based on these findings, the study recommends prioritizing leakage detection and repair programs as a foundation for Messina's water management plan. This approach directly addresses inefficiencies by minimizing water losses and fosters long-term network sustainability.

Future considerations for optimizing network performance include investigating the integration of pressure management strategies with leakage reduction. Furthermore, analysing the cost-effectiveness of various mitigation strategies can inform resource allocation decisions. By adopting these recommendations, Messina can ensure a more reliable water supply for its residents and safeguard the long-term sustainability of its WDN in the face of water scarcity challenges.

The author is grateful to Giuseppe Tito Aronica for inspiring this research, for supervising it at the early stage, and for providing information on the case study. The author is also grateful to A.M.A.M. S.p.a. (Azienda Meridionale Acque Messina) for having provided all the information and the data related to the case study.

This article has not received any external funding.

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

The authors declare there is no conflict.