The socio-technical short-term implications of drinking water hoarding on supply reliability

Water supply systems (WSS) play a pivotal role in urban infrastructure by ensuring a reliable supply of drinking water for domestic, industrial, tourist, and agricultural usage (Quitana et al. 2020). WSS are vulnerable to disasters and failures due to, e.g., natural hazards (Quitana et al. 2020), contamination (Zulkifli et al. 2018), and climate change (Frederick & Major 1997). Not only during such disasters, the withdrawal quantities are very much dependent on user behaviour and especially the water consumption in households is typically linked to factors such as lifestyle and behaviour (Dzimińska et al. 2021).

Currently, most responses and actions tend to focus on dealing with incidents or post-incident scenarios. A more effective approach to crisis management and WSS disruption lies in pre-incident planning and policy development, identifying problems and weak points in advance (Spearing et al. 2021). Pre-incident management planning and policy formulation are critical to helping utilities continually assess and enhance their preparedness, response, ability to recover, and future planning against different failures, hence increasing the resilience of the WSS.

Threats from climate change (Gosling & Arnell 2016), droughts (Koop et al. 2019), storms (Landsman et al. 2019), or pandemics (Lüdtke et al. 2021; Berglund et al. 2022; Vizanko et al. 2024) can lead to change in consumption patterns potentially resulting in water shortages. In uncertain situations, people often feel fear and anxiety about shortages of crucial needs (Dulam et al. 2021), and human nature tends to take proactive actions in response to heightened emotions (Yuen et al. 2020). The most common reaction to these emotions is to hoard goods for essential needs like portable water to mitigate the risk of shortage (Dulam et al. 2021), resulting in uncoordinated and simultaneous withdraws from the WSS. Also, government policies and announcements for emergency preparedness often recommend storing water for drinking, hygiene, and washing in large containers such as bathtubs. For instance, emergency plans from the governments of Canada (Govt. of Canada 2024), the state of Alaska (State of Alaska Drinking Water Program 2024), and Germany (Federal Office of Civil Protection & Disaster Assistance 2024) all advise collecting water in large containers in case of emergencies. Additionally, past disaster scenarios have led to people hoarding water in large storage units, such as during the drought period in Shanghai, China (Reuters 2022), and during storms or extreme heat in Texas, USA (Guilbeau 2024). Although real-life case studies on water hoarding behaviour, such as filling bathtubs, are undocumented for continuous WSS, it's been reported to have a negative impact on intermittent WSS (Klingel 2012).

Such uncoordinated and simultaneous withdrawals of water from the WSS for storage in storage units (large containers, tanks, or bathtubs) could arise from the fear of shortages. Subsequently, these withdrawals/hoarding could negatively impact the WSS and lead to short-term temporary unavailability or shortages in water supply from WSS. As shown by the literature, the perception of a potential threat is intensified even further in case of temporary unavailability or due to the increased information spread in social media caused by emotional triggers (Leung et al. 2021). If the water supply becomes the focus of such panic behaviour, such activities could result in uncoordinated and simultaneous withdrawals of water from the WSS for the purpose of storage that could lead to a (short-term and temporary) negative impact on the system. Such a negative impact (reduced outflow rates due to increased friction losses) can trigger greater parts of the population to hoard water. The short-term or temporary impacts could be significantly worse if such events occur during a crisis or when the network is being utilized for firefighting purposes.

Therefore, this research aims to:

• Analyse the impacts of water hoarding on the supply reliability of a WSS.

• Investigate the impacts of false information and negative feedback.

• Identify the critical threshold that governs the WSS's ability to handle simultaneous water hoarding and establish the limit at which the transmission of feedback and news must be contained to maintain operational performance.

Water shortages due to various scenarios, combined with government policies and recommendations to hoard water, could lead the population to engage in this activity. These activities could trigger uncoordinated and simultaneous water withdrawals from WSS for household storage, potentially disrupting the system and causing shortages in different areas. To evaluate the impact of water hoarding on the WSS, we assume varying percentages of households simultaneously filling their storage units (large containers, tanks, or bathtubs). The filling rate is systematically examined within the range of 0.1–0.4 L/s. This systematic analysis seeks to identify the filling critical threshold at which the network can maintain normal operations and examine how the filling rate and the percentage of households hoarding water affect the WSS. Additionally, the percentage of households filling their storage units varied between 5 and 100%. A reliability analysis is then performed to measure the impact of water hoarding behaviour on the WSS. Each scenario analysis was conducted at least 50 times to find the range of impact due to change in spatial distribution and reliability analysis was done using EPANET 2.2 (Rossman et al. 2020). In each scenario, random households are assigned to fill their storage units. Once these households are assigned, a new demand is then assigned in EPANET to reflect the increased water usage. The filling critical thresholds for various filling rates were identified when, after a minimum of 50 analyses, the average percentage of demand not supplied (DNS) exceeded 5% in a specific scenario. The 5% filling critical threshold provides a margin of flexibility to account for uncertainties arising from input data quality, such as pipe topological data, model calibration, model complexities, and pipe roughness coefficients.

In critical scenarios, such as in scenario 1, where households hoard water above critical thresholds, there is the potential to initiate the dissemination of news or negative feedback about water shortages. This information may spread throughout the population via channels like social media feeds, telephone conversations, and official announcements from authorities. Multiple studies have suggested that social media has been used or could be utilized for communicating risks to the public (Panagiotopoulos et al. 2016; Wolkin et al. 2019). The continuous presence of social networks in daily life serves as a prime method for spreading information. They enable users to create and share information and allow messages (including rumours) to reach a wide audience efficiently, even among individuals who do not know each other directly (Sahafizadeh & Ladani 2018). As a result, they can trigger cascading effects, leading to new scenarios arising due to the influence of others. These situations may rapidly increase the number of households hoarding water in their storage units like bathtubs. The initial impact point (impacted households) for information transmission is determined by utilizing the maximum number of households impacted at filling critical thresholds of each corresponding filling rate.

The reliability focuses on the short-term impact of water hoarding. The central storage facilities (tanks) in WSS are designed to balance at least daily demand fluctuations. Therefore, it is assumed that these short-term water hoarding scenarios do not stress the balancing tanks. However, it is checked that the overall withdrawal volumes do not exceed the available tank volume, and maximum filling rates are determined, respectively.

A less than 42% participation in water hoarding for a filling rate of 0.1 L/s results in negligible impact on WSS, with the percentage of DNS below 5% (Figure 3(a)). As can be noticed, there is an increase in DNS above 42%, representing a critical threshold for a filling rate of 0.1 L/s. As the filling rate increases, the critical threshold decreases from 42% at 0.1 L/s to 20% at 0.2 L/s, and further down to 13% at 0.3 L/s, and 10% at 0.4 L/s (Figure 3(a)–(d)). Additionally, there is a sharp increase in the DNS after the filling critical thresholds for higher filling rates (0.2–0. 4 L/s).

The critical factor for the filling rates is also evident in the number of AH, similar to the DNS graph due to simultaneous water hoarding, as shown in Figure 3(e)–(h). Initially, with 5–42% of households hoarding water for a filling rate of 0.1 L/s, only minimal impacts are seen, with 1–200 households being affected. Then, there is a steady increase in AH, with a maximum of 300 households affected at 42%. From this point, the number of impacted households increases rapidly to 720 at 50%, 1,300 at 75%, and finally 1,800 households at 100%, as depicted in Figure 3(e). The increased variability in the graphs of AH (Figure 3(e)–(h)) arises because, in cases where the household node has a DNS greater than zero, all households in the node are considered as impacted. In contrast, for graphs illustrating DNS (Figure 3(a)–(d)), the DNS highlights the precise differences between normal conditions and hoarding conditions. Similar patterns appear in Figure 3(e)–(h)), where higher filling rates reduce the filling critical threshold: 20% for 0.2 L/s, 13% for 0.3 L/s, and 10%.

These analyses have been conducted under a single time step with a focus on the performance of the water supply network neglecting tank re-filling which is a conservative assumption. However, to better mimic reality, the re-filling processes of the tank and water availability should be considered. Considering the tank volume, the elevation tank could not get empty for filling rates of 0.1 and 0.2 L/s but would become empty for filling rates of 0.3 L/s and 71% of households and 0.4 L/s with 59% of households hoarding water. The blue-dotted line in Figure 3(c), (d), (g) and (h) represents the capacity of the elevation tank.

Different filling rates could affect varying numbers of households, leading to different initial triggers for potential impacts due to the propagation of information. Table 1 presents the critical threshold levels for filling and the corresponding number of AH in the WSS at various filling rates. The data reveal that 5% DNS can influence a broad spectrum of households, ranging from 212 to 350 households, across different filling rates. This underscores the significant dependence of the AH on spatial distribution and the specific locations where households hoard water. Consequently, an impact on high-impact zones within the WSS network has the potential to magnify the effects of water hoarding, leading to the rapid propagation of negative feedback regarding water shortages. This, in turn, may accelerate the demand for water from the WSS.

Based on these results, it is important to maintain transmission rates below specified thresholds to ensure that the proportion of households hoarding water remains within the critical limit for appropriate filling rates. Furthermore, it is important to consider that the speed of information transmission can vary based on the source and medium of propagation. Additionally, initial impact points may deviate significantly from the researched values leading to a higher impact in the WSS. Consequently, subsequent analysis could be conducted by WSS operators, integrating a wide range of initial impact points and transmission rates, allowing a definition of an extensive range of thresholds. These thresholds could serve as valuable tools for operators and other agencies involved in WSS to assess impact, counteract the spread of negative feedback, and prevent its propagation, thereby maintaining the WSS under normal operating conditions and enhancing its resilience.

Based on the findings of this research, water hoarding behaviours can affect quantity constraints and system failures. This could have an imbalance in water resource distribution, resulting in shortages for certain user zones or areas. Subsequently, it imposes additional stress on WSS infrastructure, inducing short-term impacts such as system failures, reduced pressure, and capacity limitations. Notably, substantial hoarding can accelerate these issues, leading to both a reduction in supplied water quantity and a significant decline in water pressure and adversely impacting critical infrastructures like hospitals and firefighting facilities. This could accelerate the impact of the emergency situation and could have adverse effects like critical infrastructures lacking the water required and lacking firefighting water. Understanding the characteristics of demand can assist WSS companies in devising effective emergency responses and communication strategies that address the unique needs and concerns of the public, ultimately improving the WSS's resilience.

The study focused on the simultaneous hoarding of water in household storage units from WSS. The impact of these water hoarding scenarios was analysed on an Alpine WSS using reliability analysis. The reliability analysis was conducted by using performance indicators like the percentage of DNS that is the ratio between water supplied to required demand in the WSS and the number of AH impacted. The findings indicated that when multiple households engage in water hoarding, it can lead to a short-term performance decrease in water supply from the WSS. Thereby, filling critical thresholds (5% DNS) regarding households hoarding water can be identified depending on the filling rates, within which the WSS can maintain normal operating supply conditions. For instance, at a filling rate of 0.1 L/s, up to 42% of households can hoard water without any major impact on the case study. Beyond this point, there is a significant increase in the impact (DNS/ households impacted) on the WSS.

In the next step, a spread of information and negative feedback contributing to water scarcity in the WSS, originating from AH was investigated. The initially impacted households were determined based on the filling critical thresholds of each filling rate and were employed for the analysis of transmission rates. The results indicate that the spread of feedback due to failures can significantly affect the DNS in the WSS network. Thereby, the transmission rate (R₀) emerges as a pivotal factor influencing the impact on the WSS, with distinct transmission rate thresholds for various filling rates. The transmission rate is the rate of spread of news/feedback about water shortage from AH to hoard water in the next time step. The critical threshold R₀ values were identified where there is no impact (no households are impacted) on the WSS concerning the number of households hoarding water due to the transmission of information. The analysis with the initial impact points found these critical threshold R₀ values to be four for all filling rates in this research.

These impacts due to hoarding water could potentially lead to system failures and capacity constraints, posing significant challenges to the WSS's capacity to provide water. Therefore, understanding demand characteristics becomes important for WSS companies. It enables the development of effective emergency responses and communication strategies, playing an important role in improving the system's overall resilience. By embracing these insights, WSS operators could navigate challenges and increase the system's adaptability, ensuring sustained water supply reliability even during adverse social circumstances.

The project ‘RESIST’ is funded by the Austrian security research programme KIRAS of the Federal Ministry of Finance (BMF). Among others, the project RESIST aims to increase the resilience of water distribution networks by further researching socio-technical impacts that could impact water supply scenarios.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.