Adaptation pathways for intermittent water supply systems under limited resources Open Access

Water scarcity and power outages are two main causal factors for the intermittency of water supply, especially in developing countries (Farmani et al. 2021). Water scarcity can be caused by natural (climate change and natural disasters) and human (population growth, inefficient water uses, and over-extraction of groundwater) factors, or a lack of clean freshwater. Power outages caused by infrastructure challenges, limited resources, and environmental issues can contribute to water intermittency in developing countries (Simukonda et al. 2018). The intermittent water supply (IWS) can present challenges and problems for individuals, households, and communities. These challenges include inequitable water supply, high-utility operating costs, and health problems. Increasing water consumption and shrinking water resources make balancing demand and supply even more challenging (Galaitsi et al. 2016). Limited electricity supply, driven by fossil fuel dependency and climate impacts on hydropower, is a major cause of IWS in developing countries (Simukonda et al. 2018). High connectivity between critical infrastructures, such as power and water networks, risks their resilience (Helbing 2013). In the context of IWS systems, resilience refers to the system's ability to cope with disruptions in water distribution. Several negative consequences result from intermittent water systems for utilities, consumers, and society, including rapid asset degradation, water quality issues, inequity, the financial burden on consumers and utilities, and public health problems (Klingel 2012; Galaitsi et al. 2016; Farmani et al. 2021). There is a well-known problem of inequity (among users) in IWS systems. Gottipati & Nanduri (2014) developed an index, the so-called Uniformity Coefficient (UC), to measure the global equity of IWS systems. Ilaya-Ayza et al. (2017a) performed a multicriteria optimization of supply schedules to improve equity. Ilaya-Ayza et al. (2017b) organized intermittent water networks into sectors (also called district-metered areas, DMAs) to improve the efficiency of supply schedules. Few studies have examined low-cost operational strategies for improving the equity of water distribution in existing water distribution networks (WDNs) (Nyahora et al. 2020; Gullotta et al. 2021; Souza et al. 2022). Using optimization techniques, Ameyaw et al. (2013) determined the optimal location and size of water supply reservoirs under water scarcity. In addition, Ayyash et al. (2024a) proposed an approach for sectorizing and optimizing IWS systems under water scarcity conditions. Gullotta et al. (2021) addressed the equity issue in intermittent WDNs by proposing the following two intervention strategies: (1) optimizing valve placements and (2) optimizing both valve locations and settings. The non-dominated sorting genetic algorithm (NSGA-II) multiobjective optimization algorithm was used to generate results for the two-objective (the UC and the number of valves) optimization problem, demonstrating that simple, low-cost interventions can significantly enhance equity. Hybrid methods, such as combining genetic algorithms (GAs) with hydraulic simulations (Gupta et al. 2018), further underscore the versatility and relevance of NSGA-II in addressing equity challenges in WDNs. Many external factors affect decision-making for water systems, such as climate change, population growth, technological developments, economic developments, and their impacts (Galaitsi et al. 2016; Farmani et al. 2021). Climate change has significant impacts on water resources, as shifting climatic patterns contribute to declining groundwater levels and increasing water demands. There is an urgent need for adaptive and sustainable water resource management strategies to address the growing pressures of climate change and ensure long-term water security (Choudhary & Singh 2024; Sunil et al. 2024). In addition, environmental conditions may change over time as well as societal perspectives and preferences, including stakeholders’ interests and their evaluation of plans (Van der Brugge & Loorbach 2005; Offermans 2010). Adaptation pathways can be viewed as sequences of actions that can be implemented progressively, depending on the dynamics of the future (Werners et al. 2021). In an adaptive plan, adaptation pathways describe how the implementation process will be carried out by defining measures to be taken now and those that are intended to be implemented once particular conditions are met (Kwakkel et al. 2016). Adaptation pathways explicitly consider uncertainty and incorporate flexibility into planning. As well as avoiding lock-in, threshold effects, and maladaptive consequences, other benefits might include identifying interventions that have ‘no or low regrets’ (Butler et al. 2016a). Initially, adaptation pathways have been designed in situations where stakeholders’ goals are uncontested and remain constant over time (Haasnoot et al. 2012; Rosenzweig & Solecki 2014). In recent years, they have been applied in uncertain, resource-constrained environments, where multiple decision-makers are involved, and adaptation outcomes and goals are unclear (Wise et al. 2014; Butler et al. 2016b; Gajjar et al. 2018). With the concept of adaptation pathways increasingly entering formal planning practice, and more people looking to apply them, it is becoming increasingly important to understand their applicability in various decision-making contexts (Werners et al. 2021). While adaptive pathways planning has advanced from theory to practice, challenges remain in managing complexity and governance, highlighting the need for transformative approaches and improved long-term planning and funding (Haasnoot et al. 2024; Muccione et al. 2024).

IWS systems present critical challenges, including inequitable water access, limited supply hours due to pressure variability, and the compounded impacts of climate change and resource uncertainty, yet they remain underexplored in the context of adaptation planning. While adaptation pathways are increasingly adopted for managing uncertainty, existing frameworks often assume stable water availability and consistent stakeholder goals, making them unsuitable for IWS contexts. This study addresses these gaps by proposing a tailored adaptation pathways framework that accounts for the unique dynamics of IWS systems under limited water and power availability. By integrating short-, medium-, and long-term actions, and using evolutionary multiobjective optimization to balance operational costs, equity, and supply hours, the framework provides a structured approach to navigating uncertainty. Applied to the benchmark D_Town network, the study demonstrates significant improvements in performance, equity, and resilience. This research expands the applicability of adaptation pathways to IWS systems, offering decision-makers a practical tool to address complex challenges and promote sustainable, equitable water management under evolving resource conditions.

The water system of interest must be assessed as the first step in the water adaptive plan. Various factors contribute to IWS systems, including water shortages and power outages. When water resources are limited or insufficient to meet the demand of a growing population or various sectors like agriculture, industry, and households, IWS becomes a common response. Power outages disrupt the supply chain, leading to interruptions in the water supply. In response to intermittent supply, consumers’ behavior is shaped by the need to adapt to the irregular and often unreliable water supply.

In this study, 0 and 10 m pressure values are used for p_m and the p_req, respectively (Universitat Politècnica de València 2022). A typical value of 0.5 indicates the pressure function exponent (Muranho et al. 2020).

The performance indicators are used as measurable objectives to enhance the reliability of water supply even during outages and periods of scarcity. Improving key performance indicators involves investing in backup power systems such as generators or solar panels to ensure the continuous operation of water treatment plants and pumping stations. Furthermore, implementing measures to optimize water usage and distribution is crucial. This includes maximizing equity and the duration of supply hours and adopting efficient system operation to maximize output with minimal energy consumption.

In this study, the main objectives are improving equity among consumers and the proportion of effective supply hours for consumers with the least energy consumption.

UC is used to measure global equity in water distribution among network users (Gottipati & Nanduri 2014). It is based on the concept of minimizing the differences in the amount of water supplied among users, which in turn will maximize equity in the network (Ameyaw et al. 2013).

The planning of any water supply improvement project that will be implemented well into the future should consider several sources of uncertainties. The development and analysis of scenarios is one of the most effective methods of incorporating these uncertainties into planning (Kang & Lansey 2014). In the absence of complete knowledge of the future, scenarios are based on assumptions about what direction trends will take and how they will alter the system's state (Simukonda et al. 2022). Urban water management has also been analyzed using scenarios, with past studies showing their usefulness in accounting for uncertainties in key influencing factors that might affect the future of the sector (Sivagurunathan et al. 2022).

For water systems, scenario paths are often used to project different future outcomes based on varying assumptions and conditions (Pierre et al. 2012). To project future changes in water supply, demand, stress, depletion, and variability, a global hydrology and water resources model (PCR-GLOBWB 2) could be employed (Kuzma et al. 2023). Climate projections are derived from CMP6 (Coupled Model Intercomparison Project Phase 6) and include three different scenarios, business-as-usual, optimistic, and pessimistic. As part of the optimistic scenario, the environment will be regulated strictly, effective institutions will be strong, technology will develop rapidly, resource efficiency will improve, and population growth will be low. The business-as-usual features regional competition, inequality, slow economic growth, weak governance, low investment in technology and the environment, high population growth, and moderate radiative forcing (Representative Concentration Pathways (RCP) 7.0), causing significant warming and water resource impacts. The pessimistic scenario assumes rapid economic growth, high energy demands, and fossil fuel reliance, leading to the highest radiative forcing. It represents a worst-case future with severe warming and its corresponding impacts on water availability (Kuzma et al. 2023).

Despite 30 years of efforts and some progress under the United Nations Framework Convention on Climate Change (UNFCCC), anthropogenic greenhouse gas emissions continue to rise. Few quantitative estimates have been available of the global aggregate impacts of warnings of 3 °C or above, according to the Intergovernmental Panel on Climate Change (IPCC) (Kemp et al. 2022). Erratic rainfall patterns and increased evaporation rates exacerbate the effects of climate change. Population growth and urbanization strain the already limited water infrastructure, resulting in frequent and prolonged water shortages. Combining all these factors will result in IWS systems experiencing increased water scarcity over the long term. So, the chosen scenario in this study is the Pessimistic scenario.

During this step, the current status of the system and possible future scenarios are compared based on specific performance indicators to identify any gaps. The possible future scenarios are ‘reference cases’ assuming no new policies are implemented, and they include (transient) scenarios that span the uncertainties for the different periods. The presence of a gap indicates that actions are needed. At this stage, a global resilience analysis (GRA) could be conducted to evaluate the system's resilience to different future scenarios (Diao et al. 2016). The GRA for pump failure under different water scarcity proportions, as implemented in this study, follows a systematic approach to assess the system's resilience under various failure scenarios (Ayyash et al. 2024b).

Possible actions can be specified in light of the vulnerability analysis previously identified and could be categorized as mitigation, adaptation, and coping actions. In this study, different possible adaptation actions can be implemented according to the plan period. Some of these actions can be implemented without changing the current infrastructure; they can be considered improvements to the existing system. These actions will likely be effective during the short term and at the beginning of the medium term. At the end of the medium-term and over the long-term periods, actions that change the infrastructure (capital investments) could be considered.

The performance of each action is assessed in reducing or removing a specified vulnerability. Ineffective actions are screened out, and only the promising actions are used in the next steps as the basic building blocks for the assembly of adaptation pathways.

NSGA-II introduced by Deb et al. (2002), is a popular evolutionary algorithm used for solving multiobjective optimization problems. It is especially suited for addressing conflicting objectives, aiming to identify a range of trade-off solutions rather than a singular optimal outcome. The algorithm utilizes a fast non-dominated sorting technique to categorize solutions into Pareto fronts according to dominance. Furthermore, it incorporates a crowding distance mechanism to maintain diversity among solutions within the Pareto front and to avoid premature convergence.

The computational efficiency of NSGA-II can be influenced by several key parameters: population size (PS), number of function evaluations (NFEs), crossover probability (Pc), and mutation probability (Pm) (Wang et al. 2019). The Platypus framework, created by Hadka (2022), is a Python-based tool for evolutionary computing, facilitating multiobjective optimization like NSGA-II. For hydraulic simulations, the Epanet-Python Toolkit (Epyt), an open-source tool developed by KIOS (2022), is commonly utilized, offering capabilities for modeling and analyzing water distribution systems.

NSGA-II was selected for this study due to its effectiveness in solving multiobjective optimization problems with conflicting objectives, such as equity, efficiency, and cost in water distribution systems. It offers several advantages, including computational efficiency, flexibility in parameter customization, and the ability to generate a diverse set of Pareto-optimal solutions, providing decision-makers with trade-off options. Widely recognized and validated in the research community, NSGA-II integrates seamlessly with simulation tools like the Platypus framework and the Epanet-Python Toolkit, making it a robust and reliable choice for optimizing complex hydraulic systems.

Operation optimization is a short-term action to improve the performance of the system. Following three objective functions are considered in this study: (1) minimizing the operational cost (Equation (4)), (2) maximizing equity among consumers (Equation (2)), and (3) maximizing the proportion of effective supply hours that consumers are served (Equation (3)). The decision variables for the optimum operation problem include (1) the pairs of trigger-on and trigger-off of water levels for the pumps and valves; and (2) the setting of the valves. Three constraints are set in the optimization: (1) Hydraulic errors should be prevented during the simulation. (2) Pump switches should not exceed a certain limit. A pump switch is defined as turning on a pump or valve not operating in the previous period (Lansey et al. 1994). (3) The tank level at the end of the simulation should be equal to the initial level to ensure a continuous system operation.

Two medium-term actions include optimizing sectorized network operation and rehabilitation (tank expansion). Sectorization involves dividing the network into sectors based on connectivity, direct access to the source, and minimizing cut size. A detailed description of the sectorization of an IWS system can be found in Ayyash et al. (2024a). In addition to short-term actions, bridge pipe statuses (on/off) are introduced as decision variables, limited to a maximum of four switches. The daily cost of pumping operation could be computed according to Equation (4), and then the weekly cost can be computed. This cost is then multiplied by 52 weeks in a year, divided by 1.3 peak-hour factor to account for demand variability throughout the year. In this study, capital costs include both costs of tank expansion and upgrading of the existing pumping station. Details of the capital and operational costs were explained in the Battle of the Water Networks II instructions (Adelaide South Australia 2012).

The optimum operation and rehabilitation strategies are considered to improve the network resilience to both power outages and water scarcity over the long-term period because the infrastructure is needed to enhance and ensure the system's sustained functionality. In addition to the objective functions and decision variables that were considered in the previous section, the capital cost of the pumps and addition of new pumps in parallel to the existing pumping stations were included in the optimization problem.

Figure 2 shows that Actions A and D should be able to achieve the targets for the long-term scenarios. If Action B is chosen after the short-term period, a tipping point is reached at the start of the medium-term. Choosing Action B may be ineffective as additional actions are needed. If Action C is chosen after the medium-term period, a shift to Actions A, B, or D will be required (follow the solid green lines). Choosing option C involves taking a risk, as additional actions may be needed in case scenario X becomes a reality.

To mitigate the risks associated with pessimistic scenario paths in IWS systems, it is essential to select and implement cost-effective pathways. These pathways should focus on short-, medium-, and long-term actions that enhance the resilience of the IWS system to both water scarcity and power outage scenarios.

In this stage, one of the cost-effective pathways will be chosen and assessed as an example. The adaptive plan should answer the following questions: Given a set of pathways and uncertainties about the future, what actions/decisions should be taken now (and which actions/decisions can be postponed)?

An adaptive plan is characterized by its flexibility and foresight, meaning that it addresses current challenges while keeping the system flexible enough to adjust to future changes. As a result of these immediate actions, short-term benefits can be obtained, but several future pathways can also be opened for potential adjustments in the future. Then, time starts running, signpost information related to the triggers is collected, and actions are started, stopped, or expanded in response to this information. After implementation of the initial actions, activation of other actions is suspended until a trigger event occurs.

The methodology is applied to the D-town water distribution system (University of Exeter's (2020) benchmark network library) as shown in Supplementary material, Figure A1, Appendix A. Water is primarily supplied from a single reservoir (R1) with a head of 59 m supplemented by groundwater through pumping station S1. The system includes seven tanks (T1-T7) for water storage and distribution, alongside 11 pumps spread across five stations (S1–S5). The network is divided into six supply areas (A1–A6), each varying in elevation, baseline demand, and tank capacity (Supplementary material, Table A1, Appendix A). A flow-limiting valve (FCV) at the reservoir outlet simulates water scarcity with a set flow below the system's average demand of 264 L/s. Identifying these areas and simplifying the system is crucial for developing adaptation pathways to manage water supply over different time frames. The variation of the electricity tariff is shown in Table A2, Appendix A.

According to the IPCC 2022 report, human activity has been responsible for approximately 1.1 °C of warming since 1850–1900. However, recently the IPCC warned that rising temperatures, driven by human activities, are intensifying global water scarcity, projected to increase by 10–20% every two decades. Exceeding 1.5 °C warming will exacerbate risks like reduced freshwater availability, droughts, and evaporation, disproportionately impacting vulnerable populations and arid regions. Immediate and significant emissions reductions, coupled with improved water management and increased adaptation investments, are essential to mitigate these effects and address the growing water crisis. This assessment is based on improved observational datasets to assess historical warming. Furthermore, it is based on scientific research into how human-caused greenhouse gases affect the climate system. Three percentages, 10, 20, and 30%, are considered to demonstrate a progressive increase in water scarcity in the future for short-term (1–10 years), medium-term (10–20 years), and long-term (beyond 20 years) periods respectively. Therefore, the fixed flow values for the FCV are 238, 212, and 185 L/s, respectively, for modeling 10, 20, and 30% water scarcity.

Stress testing incorporates uncertainty in electricity supply by simulating pump failure scenarios and evaluating system performance across numerous plausible future scenarios, assessing each investment option under various combinations of uncertainties. The system is operated under five demand patterns (DP) that are classified into four daily periods: off-peak period (12:00 a.m. to 5:00 a.m.), mid-peak period (6:00 a.m. to 11:00 p.m.), and two peak periods (12:00 p.m. to 5:00 p.m. and 6:00 p.m. to 11:00 p.m.). So, under a single failure mode (i.e., one pumping station being out of work which is equal to 20% failure), there are 16 scenarios throughout the day. These scenarios arise from the combination of the four pumping stations and the four daily periods (off-peak, mid-peak, and two peak periods). Since the failure of any one of the four stations can occur during any of the four time periods, the total number of possible combinations is calculated as 4 (stations) × 4 (periods) = 16, accounting for all potential station failures across the different time intervals. In this study six failure modes are considered (0, 20, 40, 60, 80, 100%), therefore, the number of scenarios that are considered for each term is 96.

For instance, a 60% pump failure (equivalent to three stations) results in highly inequitable water distribution, especially at 30% water scarcity. Minimum equity values drop to 1, 0.6, 0.38, and 0.28 for 0, 10, 20, and 30% scarcity, respectively. With 20% pump failure, mean equity levels reduce further under increasing scarcity, with values of 0.63, 0.54, 0.34, and 0.15 for each respective scarcity level. Similarly, maximum equity values decline to 0.78, 0.59, 0.36, and 0.2 across these scarcity levels.

The GRA also identifies critical pumps within the system, where the failure of specific combinations has a greater impact. For example, the failure of station S3 under 20% scarcity substantially reduces equity, with a minimum UC of 0.31. The analysis also finds that power outages between 6:00 p.m. and 12:00 a.m. are particularly impactful. When two pump stations fail (40% failure), the combination of stations S3 and S4 is especially critical due to their role in high-demand areas (A6 and A4) with high consumer counts and elevated base demand, while also having the smallest tanks in the system (500 m³ each). This combination leads to significant service disruption during peak hours.

In the short-term period, optimizing network operations could be implemented. The performance of the system can also be increased through another operational intervention, which is optimizing the operation of sectorized networks at the start of the medium-term period. Repairing deteriorating pipes is an action that could reduce water loss and improve the efficiency of the system in the medium period. Additional capital investments (tank capacities) will increase the resilience of the network until the end of the medium-term period. Long-term, significant infrastructure development like increasing the capacity of pumps and tanks enhances the overall water supply system's ability to meet future demands. Under water scarcity conditions and unreliable electricity supply, alternative water and power resources, including desalination, water recycling, and renewable energy integration, must be developed. As large-scale interventions, they require considerable planning, investment, and implementation time, but they are essential to ensuring long-term sustainability.

The chosen adaptation pathway must account for the current state of the infrastructure, such as the condition of pipes, pumps, and tanks, and the context of water scarcity and power outages. For IWS systems with newer pipes, short-term adaptation should focus on operational optimization using algorithms to enhance water flow, pressure, and distribution timing for efficiency. Medium-term actions should include increasing tank capacity to manage demand during intermittent supply periods, while long-term efforts should expand both pump and tank capacities to ensure reliability and address higher demand. Additionally, alternative water sources like desalination and renewable energy will be essential for addressing prolonged shortages. For systems with older pipes, immediate stabilization through repairs and replacements is critical, complemented by operational optimization in the short- and medium-term to improve reliability. Long-term measures should focus on expanding pump and tank capacities to build a more sustainable and resilient system against water scarcity and power outages. Across both scenarios, prioritizing water and power efficiency is crucial. This can include upgrading to energy-efficient pumps, which helps to reduce operational costs and minimizes stress on power resources. Once the system's efficiency is maximized, backup power sources can be introduced to provide further resilience, especially in areas facing frequent outages. Options for backup power include diesel or natural gas generators for immediate, reliable power, or solar power generators, which offer a renewable, long-term solution. These adaptations ensure the IWS system remains robust under water scarcity and power instability, supporting sustainable and equitable water distribution over time.

Optimizing the network operation is the first action that needs to be taken during the short-term period (under 10% water scarcity). The number of decision variables for the optimum operation intervention is 36. This includes 22 variables for the operation of 11 pumps, with each pump controlled by two trigger levels (trigger-on and trigger-off) tied to tank levels, resulting in 11 × 2 = 22 variables. Additionally, two decision variables are allocated for Valve V2, which is also managed by two trigger levels. Lastly, 12 decision variables are assigned to two Pressure Reducing Valves (PRVs), V1 and N15, with each PRV's settings divided into 4-h time steps, requiring 6 variables per PRV for 24 h (24/4 = 6); thus, 6 × 2 = 12 variables are considered for the PRVs. The optimization of valve settings for V1 and N15 in a 24-h IWS simulation utilized NSGA-II, with six decision variables per valve (constant over 4-h intervals) and a range of 0–40 m. Key algorithm parameters included a population size of 100, 100,000 function evaluations, a crossover probability of 0.95, and a mutation probability of 0.01. To ensure robust and reliable results, several independent runs (15 runs) should be performed to minimize the influence of random initialization and ensure consistency in the optimization process. After 15 trials, no significant improvements in results were observed, confirming that this number of runs was sufficient to achieve convergence and validate the robustness of the results. Constraints included a maximum of four pump switches to minimize energy-saving inefficiencies and wear-and-tear costs (Lansey et al. 1994).

A scenario analysis of the power outage period indicates that the most critical time is between 6:00 p.m. and 12:00 a.m. In addition, according to GRA, the failure of the pumping station (S3) results in a minimum equity of 20% pump failure. The failure of pumping stations (S3 + S2) and (S3 + S2 + S4) caused the lowest equity among consumers, with 40 and 60% pump failures, respectively.

The selected operational intervention solution for the analysis has an operational cost of $980 per day, a highest min (UC) of 0.53, and an EH of 0.87. By comparing this investment with the current system, there is some saving of energy (from $996 to $980 per day) and a considerable increase in the min (UC) during the day (from 0.28 to 0.53) as well as an increase in EH (from 0.70 to 0.87) under 40% pumping failure (two pumping stations failed) and 10% of water scarcity. During the short-term period (10% water scarcity), actions without altering existing infrastructure can improve the IWS system's performance under both water scarcity and power outage conditions.

During the medium-term period, optimizing sectorized network is considered which does not require system infrastructure changes, then followed by increasing tanks’ capacities.

As emphasized in our previous work (Ayyash et al. 2024a), sectorization significantly enhances operational efficiency, improves effective supply hours, and promotes equity among consumers under water scarcity conditions. By dividing the network into distinct sectors, water distribution becomes more manageable and ensures a fairer allocation of resources among consumers within each sector and subsequently for the entire system.

The number of decision variables is 156, which includes 120 decision variables for the on/off of bridge pipes connecting sectors in addition to the other operational decision variables described in the previous section. Specifically, the bridge pipes connect the different sectors in the network, allowing operational flexibility. There are five bridge pipes to connect six sectors, each with an on/off status for every hour in 24 h, resulting in 5 × 24 = 120 decision variables. In addition, the model incorporates 36 operational decision variables as described in the previous section, bringing the total number of decision variables to 156. More detailed information on the sectorization process and the role of bridge pipes can be found in Ayyash et al. (2024a).

The operational cost of the selected solution is $961 per day, the min (UC) is 0.52, and the EH is 0.80. This solution was chosen because it is the best one regarding equity and effective supply hours simultaneously. By comparing this optimal solution with the current system under 20% water scarcity and 20% power outages, there is a saving of energy (from $1,010 to $961) and an improvement in the min (UC) during the day (from 0.33 to 0.52) as well as an increase in the EH (from 0.62 to 0.80).

In the medium term, improving equity among consumers can be achieved by sectorizing the network and optimizing pump and bridge pipe operations without requiring capital investment. Expanding tank capacity in each sector may further enhance water storage for use during power outages. As each sector has its tank, an increase in the tanks’ capacity allows more water to be stored for usage during the power outage period. Increasing tank capacities in each sector improves water storage, ensuring reliable supply during power outages and water scarcity. Larger tanks enable gravity-fed distribution, ensuring uninterrupted service that caters to the specific needs of each sector, even during disruptions to pumping operations. This approach enhances system resilience and reliability across the network. Regarding the operation and rehabilitation (tank expansion) actions for the sectorized network, the number of decision variables is 163 which includes 120 decision variables for the on/off of bridge pipes, 22 for tanks’ trigger levels to control the 11 pumps, two tanks’ trigger levels for one valve (V2), the settings of the two PRVs, V1 and N15 during each 4-h time step, and seven tanks’ volumes. An optimization process is applied to the network in the case of failure of one pumping station (20% pump failure) during the peak period (6:00 p.m.–12:00 a.m.). In addition to optimizing the operation of the sectorized network, the volume of tank ‘T4’ has been increased by 1,000 m³. The optimization selected this particular tank for expansion because it feeds the area with high demand and the current tank size is not sufficient. There are 74 users in this area, which is slightly higher than other areas, with a base demand of 57.21 L/s and high-elevation users (32–106 m above sea level). As a result, the min (UC) will increase from 0.33 for the current system to 0.65 for the chosen optimal solution and the EH from 0.62 to 0.83.

There will be an increase in water scarcity during the long-term period due to various factors, including climate change, population growth, and insufficient water use management. Additional actions will be taken to improve equity among consumers and the proportion of effective supplied hours. These actions include rehabilitation (increasing capacities) and the optimum operation of the sectorized network system. For this optimization process, the number of decision variables is 174, including the same decision variables considered in the previous section, only adding 11 upgraded existing pumps.

Table 1 shows the details of the pumps added parallel to the existing ones and the added capacity of the tanks. One of the least-cost solutions, which has been chosen for analysis, includes adding nine pumps in parallel to the existing ones as shown in Figure 10, as well as increasing the capacity of tanks T3, T4, T5, and T7 by 500, 1,000, 1,000, and 2,000 m³, respectively.

Based on the results, high capital investments are required to increase equity and the effective supplied hours under 30% of water scarcity and 20% of pump failure (equivalent to the failure of one pumping station). However, these investments (operation and rehabilitation) are not enough to address the higher water scarcity (≥30%) and power outages (≥40%).

Over the long term, IWS systems will face significant challenges due to increasing water scarcity and electricity shortages, which will strain infrastructure and operational capacity. Enhancing pumps, tank capacities, and network operations will be essential to mitigate these impacts. Additionally, alternative water sources such as rainwater collection, desalination, and wastewater reuse can supplement supply, while renewable energy options like solar, wind, hydropower, biomass, and geothermal power can address power outage conditions, ensuring more resilient and sustainable water and energy systems.

This study evaluates the performance of IWS systems under scenarios of water scarcity and pump failures, revealing significant declines in equity and effective supply hours as pump failures increase, especially under higher scarcity levels. Critical failures, such as those involving specific pumps (e.g., S3 and S4) during peak demand periods, exacerbate inequities, particularly in high-demand areas with limited tank capacities. The findings emphasize the importance of targeted interventions, including optimizing operations, expanding tank capacities, and rehabilitating pumps, to mitigate these impacts. Compared to prior research that focuses on either operational efficiency or equity, this study combines stress testing, multiobjective optimization, and adaptation pathways to systematically address changes in water scarcity and power outages over time, offering significant advancements in the field. By improving system equity and resilience through strategic interventions, this research provides a practical framework for enhancing the sustainability and reliability of IWS systems in resource-constrained environments, filling a critical gap in adaptation IWS planning literature.

This research presents an adaptation pathways approach for IWS systems under uncertain water scarcity and/or power outages. This approach assists in designing adaptive plans over different scenario paths that consist of preparing actions that might be triggered to keep a plan on track to meeting its objectives.

According to adaptation pathways, an example of a low-cost pathway was used to visualize sequences of possible actions through time (short, medium, and long-term periods). Every pathway meets a minimum performance level regarding the main target. Nevertheless, some pathways are more attractive than others based on costs or the ability to achieve the main objectives. This can be used to select a set of cost-effective pathways.

The approach has been illustrated and tested by using a benchmark network (D-Town) that is modeled as an IWS system under water scarcity and /or power outage conditions. In the implemented pathways map, there are different actions done over short, medium and long-term periods to meet the main objectives which are minimizing both operation and rehabilitation costs and maximizing the minimum equity and the proportion of effective supplied hours during the day. In the short-term (10% water scarcity), only optimizing network operation is considered. Based on the results, there is an increase in min (UC) and EH during the short-term period, from 0.28 and 0.7 for the current status to 0.53 and 0.87 for the optimum solution, under 40% pump failure (two failed pumping stations) scenarios. During the medium-term period (20% water scarcity), optimizing the network operation does not improve equity among consumers and effective supply hours. So, optimizing the sectorized network operation is considered at this stage. During this period under 20% power outage, the results of optimum operation of sectorized network show that the min (UC) increased from 0.33 for the current status to 0.52 for the optimum solution. During the same period, optimizing the operation and rehabilitation (tank expansion) of the system slightly increased the min (UC) to 0.65. Over the long-term period (≥30% water scarcity), optimizing the operation and rehabilitation (increasing pumps and tanks capacities) for the sectorized network improved both min (UC) and EH from 0.15 and 0.62 for the current status to 0.48 and 0.74 for the selected optimum solution, respectively.

In IWS systems, pipe filling, emptying, and household tanks are key features. The research included the calculation of equity and effective supply hours using EPANET 2.2 software. It is not possible to model the filling and emptying processes of pressurized systems with the current version of EPANET. The main steps of the proposed methodology will not be affected, but the estimated hydraulic models to account for filling and emptying processes and demand uncertainty (the presence of tanks at household levels) produce more effective results.

The authors gratefully acknowledge the Faculty for the Future program, Schlumberger Foundation for funding the PhD scholarship.

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

The authors declare there is no conflict.