A unified probabilistic energy and carbon footprint appraisal for intermittent water supply systems in arid regions

Abstract

A probabilistic methodology assessed hydraulics, water consumption, distribution pumping, and households pumping from the ground to rooftop storage for energy consumption in 24/7 and 2-day supply scenarios in two distribution networks, each with 379 and 959 service connections and diverse topographies. A local market survey gathered the characteristics of thirty-two household-level pumps used in the area. Considering a per capita daily water consumption of 250 L, probabilistic analysis revealed that the daily operating time of household pumps varied from 0.155 to 0.17 h at the median and from 0.37 to 0.47 h at a 0.9 probability, highlighting variation in household sizes within the study region. The study observed 26% higher energy consumption with a 24/7 supply compared to a 2-day supply for the average household size of 5.6 persons. The 2-day supply exceeded the 24/7 energy requirement for households larger than 8.7 (exceedance probability of 0.1), due to the longer operational hours. Sensitivity analysis identified the household pump’s operating time as the most critical contributor to the intermittent supply’s energy. Considering the consumers’ concerns on water quality in the realistic case, the study found ~ 60% higher energy (0.069 kWh/person/day) and carbon emissions (18.45 kg CO₂ eq/person/year) for 35% consumers using a household-level purifier, 60% bottled water, and 5% trusting in supplied water in comparison to a 24/7 supply. Instead of controlling water loss through intermittent supply, a shift to continuous supply with proactive maintenance can compensate for the additional costs by providing better water quality, satisfied customers, and a reduced global energy and carbon footprint for water supply systems in arid regions.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-18698-0.

Introduction

Approximately 60% of the electricity is generated by fossil fuels, including coal, gas, and oil, resulting in around 10 billion tons of CO₂ emissions worldwide¹. With around 7–8% of global energy usage, water supply systems require attention to minimize energy consumption². Water suppliers are diligent in enhancing the energy efficiency of their systems due to increasing energy prices and regulations on carbon adaptation^3,4(Aghazadeh and Attarnejad, 2020; Brahimi, 2019).

Energy consumption in a water supply system (WSS), operating with four primary components: abstraction, transmission, treatment, and distribution, depends on several factors, including source location, raw water quality, topography, and the size and age of the distribution network^5,6. Energy consumption for groundwater extraction from deep confined aquifers is higher than for surface water extraction⁷. Pumping energy for transmission is directly linked to topography and the distance between the source and the treatment plant or distribution network. The energy requirement for treatment is a function of raw water quality, ranging from chlorination for fresh groundwater to desalination for seawater^8,9. Among the various components, distribution system pumping is the Most energy-demanding operation, accounting for 50–70% of the total energy consumption of a WSS^10,11.

In low-income countries, municipalities often supply water intermittently for a few hours a day, with low residual heads, to limit water consumption and conserve energy. The household arrangement, which includes storage tanks and in-house pumps, ensures 24/7 water availability in the plumbing system, thereby transferring the energy burden to the consumer end. A recent study by Zhu et al.¹² based on econometric analysis for the Pune Metropolitan Region (India) reported 30% higher electricity consumption in households receiving a 12-hour supply compared to those with continuous access. In other cases, water is supplied for a few days a week; for instance, municipalities supply water for 2 days per week in some regions of the Kingdom of Saudi Arabia (KSA), the context in which this study was conducted. Due to the larger family size, long intervals (3 to 5 days) between two supplies, and the possibility of supply failure, households in KSA often contain large underground storage tanks. Extended storage in households with intermittent water supply frequently leads to microbiological water quality failures due to the proliferation of pathogens¹³.

In addition to the age, size, and efficiency of the pumps, ground elevation, family size, per capita consumption, and the type and duration of supply considerably influence the energy demand of the distribution system⁸. Ground elevation affects the placement of water sources and pumping requirements, as it influences the excess pressure head that travels through the pressure zone¹⁴. In contrast to flat and hilly terrains, rolling terrain exhibiting diverse ground elevation differences results in varying hydraulic gradients¹⁵. Rolling terrains significantly affect the delivery heads of supply pumps in smaller distribution networks without booster pumps.

Up to 40% of the water supplied to metropolitan areas in KSA is lost through leakage in water mains and inefficiencies in the system¹⁶. In addition to the loss of high environmental value water, leaking networks with intermittent supply lead to deterioration in water quality in both the distribution mains and underground storage tanks, resulting in a lack of consumer confidence in the portability of the supplied water¹⁷. Seeking to reduce water loss is an ongoing issue for water utility systems and a source of frustration for management. Most distribution systems in KSA operate with plastic pipes; however, pressure transients, the possibility of iron-oxidizing bacteria, and the rusting of metal fittings used at service connections on plastic pipes may lead to structural breakdowns and leakage issues^16,18. Damage to pipes and water loss increases energy usage to meet demand¹⁹.

The water directorates and municipalities in the KSA intermittently supply water for 2 to 5 days per week to minimize pumping energy and distribution system losses through leakage (Fig. 1). Before supply, brackish groundwater or seawater meets the KSA’s drinking water quality standards for physical, chemical, and biological parameters through reverse osmosis or thermal desalination. Small-sized household pumps lift the supplied water from the underground storage to the rooftop tanks installed on the 2nd or 3rd floor in residential settings. The intermittent supply dilemma, which can potentially contaminate the supplied water, forces residents to rely on mineral water for drinking purposes due to the need for household storage¹³. A continuous supply operates at higher residual heads (~ 14 m) to deliver water up to the 3rd story of the house, in comparison to an intermittent case, which delivers water to the household’s underground tanks. Nevertheless, the energy footprint of intermittent supply encompasses the energy utilized by the household pumps as well as the energy consumed for the production and transportation of mineral water for drinking (Fig. 1). To the extent of the authors’ knowledge, the energy and carbon footprints of oil-based energy in this setting of intermittent supply have not been investigated in the KSA or elsewhere.

Fig. 1.

Water-energy-carbon nexus for continuous and intermittent water supply systems in arid regions.

In the case of intermittent supply, the additional energy for household pumps depends on pumping hours, which are a function of the number of supply days per week, household size, and per capita daily water consumption. Residents with varying energy requirements use small pumps with different flow rates and heads available in local markets. Due to a lack of confidence in the supplied water quality, there is a common reliance on bottled mineral water consumption and household-level water purifiers²⁰. These practices also contribute to the energy and carbon footprint of the water supply, particularly in cases where sources are intermittent. All these variables add several uncertainties in the energy analysis for intermittent supply systems.

The nexus between water and energy has garnered significant attention worldwide in recent years, particularly among the member countries of the Gulf Cooperation Council²¹. Almost all countries in the Gulf primarily depend on brackish or saline groundwater¹³. Several studies have investigated the water-energy-carbon nexus (WECN) for treatment and distribution components in water supply systems^6,8. However, past studies have not considered variations in water consumption, household size, and pump types at the household level for energy assessment in intermittent supplies. Probabilistic methods have been employed in drinking water health risk assessment studies that involve multiple dynamic factors, such as intake rate, body weight, and skin area^22,23. A probabilistic methodology is necessary to account for variations in the above factors that affect energy consumption, enabling a comparative evaluation of intermittent and continuous water supplies.

The primary objectives of this research are to: (i) evaluate the impact of topography and household size on water consumption and household-level pumping energy for intermittent supply, (ii) assess the impact of water loss on energy consumption, (iii) appraise the contribution of bottled water use and household-level water purifiers to the overall energy and carbon footprint of intermittent supply systems, and (iv) compare continuous and intermittent systems based on energy and carbon indicators and develop guidelines for the preferred supply type for improving operational performance. An inclusive energy and carbon assessment framework is developed and applied to two distribution networks in the Qassim Region of KSA for pragmatic purposes. The proposed framework will facilitate policymakers and water suppliers in evaluating energy requirements for different types of supply, enabling effective decision-making to achieve the long-term sustainability of water supply systems in arid regions.

Methodology

Energy assessment framework

Figure 2 illustrates the proposed framework of WECN in water distribution systems operating with continuous and intermittent supplies in arid regions. The framework comprises a probabilistic energy assessment that considers variations in municipal supply, domestic consumption, pump types, and operating conditions at the household level. First, the required data for two water distribution systems, one operating with continuous and the other with intermittent supplies, was obtained from the National Water Company (NWC). The data for household-level pumps was obtained from the local market. The various probability distributions for water consumption patterns and household pump usage were generated using @Risk Software (8.8.1 Industrial Edition). The study examined variations in metered water consumption in terms of household size. Various scenarios were developed to capture the impact of topographical variations and duration of supply on pumping energy consumption in the selected distribution networks. The study performed both actual and hypothetical baseline scenario analysis, assuming no water loss for the comparative evaluation. Later, the study assessed the impact of water loss on energy consumption in distribution networks for the actual scenarios. Sensitivity analysis using the Monte Carlo method identifies the most significant factors contributing to energy consumption in intermittent and continuous supply systems. Energy and carbon indicators were evaluated based on probabilistic energy analysis for cross-study comparison. Finally, the framework considers a realistic case, taking into account consumers’ concerns about water quality, by assessing the impact of using mineral water bottles and household-level purifiers on the energy and carbon footprint of intermittent water supply. The details of the various framework components are as follows.

Fig. 2.

WECN assessment framework for distribution systems with continuous and intermittent supplies.

Study region, baseline data, and assumptions

Al-Qassim Region is one of the 13 administrative regions in Saudi Arabia, located in the north-central part of the country. The population of 678,000 in the city of Buraydah, the capital of the Qassim area, receives treated groundwater from the municipality²⁴. The Qassim region generally experiences hot summers and cold winters, with limited rainfall during the spring and winter seasons. Figure 3 illustrates the location of the study area in the south of Buraydah, the capital of the Qassim area in the Kingdom of Saudi Arabia. Figure 3 also illustrates the two distribution networks serving Area 1 and Area 2, which were evaluated in the present study. The study area consists of rolling terrain with an average ground elevation of 600 m above sea level. A water treatment plant in Al Shimasiyah, located 35 km away, started supplying the study region in June 2022. The region’s location was selected based on various criteria, including days of supply, diverse topography, and data availability. Figure 3 illustrates the study area, showing the different types of supplies received per week.

Fig. 3.

The study region shows three districts with different types of supplies. A1: Khub AlUshar and Elnasryah with 2-days’ supply, A2: Elkhodor with 7- days’ supply.

Area 1, spanning over 2.59 km², serves 379 households with a 2-day intermittent supply, while Area 2, covering 4.94 km², serves 959 households through a 24/7 continuous supply. Consumers in Area 1 store the supplied water in underground storage tanks and use it for the remainder of the week. The pipe materials are HDPE (high-density polyethylene) and UPVC (unplasticized polyvinyl chloride), with an average age of around 10 years. The ground elevation from mean sea level (MSL) in Area 1 varies between 604 m and 652 m, and between 603 m and 619 m in Area 2. Table 1 presents the characteristics of water distribution systems in the study region. The total length of pipes in A1 is 7,716 m. Approximately 90% of the pipes are made of high-density polyethylene (HDPE) with diameters ranging from 75 to 180 mm, while the remaining 10% are made of unplasticized polyvinyl chloride (UPVC).

Table 1.

Characteristics of area 1 and area 2 are shown in Fig. 3.

Parameters/characteristics	Area 1	Area 2
Type of supply	Intermittent	Continuous
Number of supply days per week	2 days	7 days
Population density (persons/ha)	10.7	10.9
Size (hectares)	259.13	494.77
No of households	379	959
Length of pipes (m)	7716	37,514
Pipe material	HDPE (90%), UPVC (10%)	HDPE (70%), UPVC (30%)
Pipe diameter (mm)	75–180	100–180
Elevation	604–652	603–619

Water supply data for September 2023 to February 2024 were obtained from NWC. For energy analysis, the detailed water consumption data per household for February 2024, reflecting the average case, were obtained from the NWC Office in Buraydah, Qassim, KSA. The study assumes an average of 5.6 persons per household, which is the average household size in the KSA²⁵. In the absence of information on per capita consumption variations, the study assumed an average water consumption of 250 L per capita per day (lpcd), as reported for the City of Buraydah by the Ministry of Municipal and Rural Affairs in KSA²⁶. The study also assumed residents consume tap water for drinking in the case of a continuous supply.

Water supply scenarios

The distribution network of Area 1 comprises 76 nodes and 85 pipes, serving 379 households. The network in Area 2 contains 282 nodes and 341 pipes, serving 959 households. Figure 4a and b show the hydraulic Models for Area 1 and Area 2. In addition to the actual conditions of a 2-day intermittent supply in Area 1 and a 7-day continuous supply per week in Area 2, six additional scenarios, described in Table 2, were evaluated to capture variations in topography and supply duration.

Fig. 4.

Hydraulic Model for the study region using EPANET 2.0, (a) Area-1 spanning over 259.13 hectares, (b) Area-2 spanning over 494.77 hectares.

Table 2.

Operating conditions of pumps in the study region.

No.		Scenario	Type	Flow (L/s)	Head (m)
S1	S11	2-day intermittent supply with actual ground elevations in Area 1	Actual	24.5	15.2
	S12	7-day continuous supply with actual ground elevations in Area 1	Hypothetical	10.5	22
S2	S21	2-day intermittent supply with flat ground in Area 1	Hypothetical	24.5	11.6
	S22	7-day continuous supply with flat ground in Area 1	Hypothetical	10.5	12.4
S3	S31	2-day intermittent supply with actual ground elevations in Area 2	Hypothetical	77.91	30
	S32	7-day continuous supply with actual ground elevations in Area 2	Actual	33.39	40
S4	S41	2-day intermittent supply with flat ground in Area 2	Hypothetical	77.91	25.5
	S42	7-day continuous supply with flat ground in Area 2	Hypothetical	33.39	31

Energy Estimation for continuous supply

Water supply systems aim to deliver water to people in an adequate amount at acceptable pressures. Pumping is one of the most critical functions of a water supply system. Pumps are stationary engineering equipment that convert mechanical energy from a revolving shaft into hydraulic pressure²⁷. Common types of pumps used in water distribution systems include centrifugal, vertical, and submersible⁶. The efficiency curve for larger pumps indicates an increase in energy consumption with increasing flow and head, which typically operates within a desired range, depending on variations in water demand.

The NWC is using centrifugal pumps for supply in the study region. Equation (1) estimated the power consumed by the pumps used for water distribution in the study region²⁸.

1

Where is pump power (kW), is flowrate (m³/h), is density (kg/m³), is the total pump head (m), is the acceleration due to gravity (9.81 m/s²), and is the efficiency of the distribution pump (%).

The flow rates for the distribution pumps were estimated for peak-hour demand in scenarios with a 24/7 continuous supply, and for maximum daily demand in scenarios with intermittent supply, considering both pumping and household storage. Based on personal communication with the supplier in the study region, the peak hour factor of 2.25 and the maximum day factor of 1.5 were used as the common values to estimate flow rates from average daily demand.

Equation (2) calculated the distribution system’s pump energy for a continuous supply.

2

Where represents the pumping energy for continuous supply (kWh), represents the pumping energy for water distribution (kWh), and denotes the pump operating hours (h) of distribution pumps. In the case of a continuous water supply .

Energy Estimation for intermittent supply

A detailed shop-to-shop local market survey was conducted to gather information on the types of pumps used in the study region through personal communication and site visits. Table 3 summarizes the technical specifications for 32 pumps used to lift water from underground tanks to rooftop tanks at the households in the study region. The data show significant variation in the pumps’ flows and heads among different manufacturers, accommodating varying user choices, household sizes, and building heights. Small centrifugal pumps at the household level operate at average efficiencies ranging from 45 to 70%, while submersible pumps operate at slightly higher efficiencies²⁹. The personal communication with professionals involved in installing these pumps revealed that the efficiency of the pumps varies between 50% and 70%, or even higher when they are new. Due to a lack of information (inadequate records of buyers’ addresses and the absence of a disclosure control protocol) regarding sales volume, all the pumps listed in Table 3 were assumed to be used at households in the same proportion.

Table 3.

Technical specifications for the pumps used at the household level.

No	Supplier	Manufacturer	Pump name	Pump type	Origin	Flow “Q” (l/min)	Operating head “H” (m)
1	Al Zamil Industry and Trade	Grundfos	SCALA2	Surface (centrifugal) Pump	Serbia	10–60	15–45
2	Al Zamil Industry and Trade	Grundfos	SBA3-45 A	Submersible Pump	Italy	20–90	10–40
3	Al Zamil Industry and Trade	Grundfos	SB 3–45 A	Submersible Pump	Italy	20–100	15–35
4	Al Zamil Industry and Trade	Grundfos	NS5-33	Surface (centrifugal) Pump	Italy	30–120	20–32
5	Al Zamil Industry and Trade	Grundfos	JP 4–47	Surface (centrifugal) Pump	Hungary	10–60	19–44
6	Al Zamil Industry and Trade	Zetalia	CT 150	Submersible Pump	Italy	20–135	15–66
7	Al Zamil Industry and Trade	DAB	K 40-100 M	Centrifugal Pump	Italy	40–175	18–45
8	HATCH	DAB	ESYBOX	Multi-stage centrifugal pump	Italy	120	Up to 65
9	HATCH	DAB	ESYBOX MINI 3	Multi-stage centrifugal pump	Italy	80	Up to 55
10	HATCH	DAB	Esybox Diver	Submersible Pump	Italy	120	Up to 55
11	HATCH	DAB	Esybox Max	Multi-stage centrifugal pump	Italy	240	Up to 113
12	HATCH	DAB	DIVERTRON	Submersible Pump	Italy	106	Up to 45
13	HATCH	DAB	DIVERTEK	Submersible Pump	Italy	106	Up to 45
14	Al Zayed Trading Establishment	Sabar	Gland (Single Phase)	Centrifugal Pump	India	165	Up to 47.5
15	Al Zayed Trading Establishment	Sabar	Gland (Three Phase)	Centrifugal Pump	India	130	Up to 50
16	General Home & Roxell	Roxel	ROX-944	Submersible Pump	China	130	32–36
17	General Home & Roxell	Dali	QB80	Centrifugal Pump	China	125	Up to 60
18	Al Nakheel Trading Company	Lowara	Horizontal Pump (PBK)	Peripheral Pump	Italy	62	Up to 82
19	Al Nakheel Trading Company	Lowara	SCUBA DRY	Submersible Pump	Italy	180	Up to 100
20	Al-Mu’tak Trading Establishment	QIFENG	QDX	Submersible Pump	China	133	Up to 32
21	Al-Mu’tak Trading Establishment	SHIMGE	QD	Submersible Pump	China	510	Up to 120
22	Al Wabel Pumps Company	Nauti	VN 5/4	Submersible Pump	Italy	150	Up to 70
23	Al Wabel Pumps Company	SHIMGE	PW370Z	Surface (centrifugal) Pump	China	40	Up to 35
24	Al-Ghamas Company	GrandMAS	T50	Intelligent control pump	Italy	310	Up to 31
25	Al-Ghamas Company	SHIMGE	NK	Submersible Pump	China	90	Up to 88
26	Semnan	Samnan Pump	1 HP	Centrifugal Pump	KSA	125	Up to 29
27	Semnan	Samnan Pump	1 HP	Centrifugal Pump	USA	208	Up to 34
28	Semnan	Samnan Pump	1 HP	Centrifugal Pump	China	95	Up to 31
29	Semnan	Samnan Pump	Goulds	Centrifugal Pump	Italy	160	Up to 24
30	Semnan	Samnan Pump	Goulds	Submersible Pump	Italy	73	Up to 51
31	Semnan	Samnan Pump	0.9 HP	Submersible Pump	China	60	Up to 52
32	Semnan	Samnan Pump	1 HP	Submersible Pump	Spain	73	Up to 51

Local markets provide a variety of sizes for overhead tanks. The Most installed units in residences had a capacity of 3,000 to 4,000 L. In this study, the tank capacity used for calculations was 3,000 L (3 m³), as it is the Most widely used size. The height of the overhead tank is contingent upon building type and the required water pressure. Installing the overhead tank on a support at a minimum distance of 0.5–1 m from the roof slab enhances cleaning and maintenance procedures while safeguarding the structural integrity against damage from water leaks.

Equation (3) estimated the time required to fill the rooftop tank.

3

Where is the flow rate of the pump (m³/min), is the volume of the rooftop tank (m³), and is the time to fill the tank (min).

Equation (4) estimates the number of rooftop tanks () required for domestic consumption per day () for each household.

4

Equation (5) estimated daily pumping time for each household (.

5

The present study developed a probability distribution for the 32 household pumps presented in Table 3 and used that to estimate the pumping time required for each household. An example of t_H estimation is given in Appendix A of the supplementary material.

The total pumping energy for the intermittent water supply system up to the delivery point was estimated using the following Equation:

6

Where represents the pumping energy consumption for intermittent supply (kWh), is energy for household pumping, is the pump power for household pumping (kW), represents the operating hours (in hours) of household pumps.

7

Where is the flow rate of the household pump (m³/h) and is the head of the household pump (m).

Water-energy-carbon indicators

The water-energy-carbon nexus (WECN) for a water distribution system can be assessed using suitable indicators that are simple to measure, understand, and comparable with the findings of other studies. The denominator of an indicator generally makes it comparable, for example, estimated energy or CO₂ emissions per unit volume of water consumed or per population served. Comparing a study’s results for a geographical region with those of other areas identifies overperformance, average performance, or underperformance, which in turn leads to the development of improvement plans.

For an effective comparison, the present study selected “kWh per m³ of metered consumption” (kWh/m³⁾ as the energy indicator, which has been frequently used in several past studies^11,30–32. The analogous carbon indicator is “kg of CO₂ equivalent emissions per m³ of metered consumption” (kg CO₂ eq/m³)^11,33. Another understandable and comparable carbon indicator is “kg of CO₂ equivalent emissions per person per year” (kg CO₂ eq/person-year), adopted by Smith et al.³⁴.

Carbon emissions factors are required to calculate the above carbon indicators. The values of these factors primarily depend on the type of energy source, e.g., crude oil, diesel, natural gas, solar, and wind. The carbon emission factor for electricity generation using crude oil ranges from 0.65 to 0.80 kg CO₂ per kWh, depending on the specific crude oil type, the technology used for electricity generation, and the combustion efficiency of the process. Studies in KSA also reported a carbon emission factor within the same (0.65–0.8) range^35–37.

Energy utilized for drinking water with intermittent supply at the user end

A past study identified the recurrence of suspended solids and iron, as well as the complete absence of chlorine (potentially indicating the presence of pathogens), at the consumer end for intermittent supply systems in the study region¹³. The residents of the study region are well aware of the potential for water quality issues in underground or rooftop tanks due to intermittent supply. A study in Jeddah, KSA, reported that 60% of the samples collected from 36 residential districts were contaminated with coliform bacteria³⁸. Consequently, they use bottled water, generally 20-liter bottles available in local markets, or three-step household-level water purifiers installed at the kitchen tap. The first step in these purifiers is pre-treatment using polypropylene filters (commonly known as sediment filters) to remove sand, silt, or clay that may have entered through cracks in the walls of underground tanks. The second step eliminates organics or chlorine to resolve taste or odour issues by using activated carbon filters. The final step is reverse osmosis (RO), which removes total dissolved solids and pathogens. In some cases, UV sterilization is the fourth step in the disinfection process³⁹.

Assuming a 2-liter per capita daily consumption of mineral water, the average energy consumption for a standard 20-liter bottle of drinking water was found to be 0.108 kWh per person per day. For the water purifier installed at the kitchen tap, the average energy consumption was 0.0425 kWh per person per day. Table 4 presents the step-by-step calculations for both cases.

Table 4.

Energy Estimation for 20-liter mineral water bottles and multi-stage purifiers at the household level.

No.	Item description	Energy estimation	Reference
1.	20-liter mineral water bottle with recycling
1.1	Production energy for producing bottled mineral water using reverse osmosis (RO)	3–4 kWh/m3 (0.06–0.08 kWh for a 20-liter bottle)	Orfi and Ali40
1.2	Transportation energy	3–7 kWh per metric ton per km (1 kWh per 20-liter bottle, assuming 5 kWh per metric ton per km for 100 km)	Ikhries et al.41
1.3	Recycling energy	0.01 kWh for a 20-liter bottle	Aversa et al.42
1.4	Total energy for a 20-liter bottle	= 0.07 + 1 + 0.01 = 1.08 ± 0.001 kWh per 20-liter bottle
1.5	For a per capita daily consumption of 2 L, the energy consumption per person per day	= (1.08/20) x 2 = 0.108 + 0.001 kWh/person/day
2.	Multi-stage household-level purifier at the kitchen tap
2.1	Energy consumption per hour	25–60 watts per hour Average = 42.5 W/h	Livpure43
2.2	Assuming one hour of operations per day produces 11.2 L (5.6 persons per household x 2 L per person)	= (42.5/1000) kW x 1 h = 0.0425 ± 0.015 kWh
2.2	For a per capita daily consumption of 2 L, the energy consumption per person per day	= 0.0425 ± 0.015 kWh/person/day

Therefore, the total energy for the intermittent water supply system up to the consumer end was estimated using Eq. (8):

8

Where represents the pumping energy consumption for intermittent supply (kWh/person/day), is energy for bottled water consumption (kWh/person/day), and represents the energy consumption for household-level water purifiers installed at the kitchen tap (kWh/person/day). The estimated values of and were transformed into similar units for comparison.

Results

Water consumption in the study region

Figure 5 illustrates the cumulative probability distributions for monthly water consumption per household in the study region. @RISK Software (8.8.1 Industrial Edition) fitted a Gamma distribution to the data, yielding a median and standard deviation of 25 and 24.14 m³/household for A1, and 30 and 33.37 m³/household for A2. Figure 5 shows that 90% of the households consume approximately 63 m³/household in the case of a 2-day supply serving 379 service connections in area A1 and around 83 m³/household for A2, serving 959 customers with a 24/7 supply. These results indicate an increasing trend in water consumption, accompanied by an increase in the number of supply days. Nevertheless, these variations in consumption can be attributed to differences in household size or per capita consumption between the two areas. Assuming a constant per capita consumption in the present study, a considerable variation in consumption patterns can be noticed in both cases by comparing the median with the 90th percentile values. For instance, considering 250 lpcd water consumption in the study region²⁶, the median (0.5 probability) Monthly consumption of 25 corresponds to 3.5 persons per household in A1 and 4.1 persons per household in A2. In contrast, the household size increases to 8.7 for A1 and 11.5 for A2, with a 0.9 probability of monthly consumption, as shown in Fig. 5. These variations in water consumption, along with the use of household pumps of varying sizes, support the need for probabilistic energy evaluation in distribution systems within the study region.

Fig. 5.

Water consumption in the study region, (a) Area-1 with 379 households, (b) Area-2 with 959 households.

Household pumping

Figure 6a and b illustrate the probability distributions for the characteristics of pumps (see Table 3 for details) used in the household. @RISK Software (8.8.1 Industrial Edition) fitted a lognormal distribution to the flow rate (Q_H) and a Laplace distribution to the household pumps’ head (H_H). The distributions showed median and standard deviation values of 0.075 and 0.063 for the flow rate (m³/min) (Fig. 6a) and 14.99 and 3.3 for the head (m) of the household pump (Fig. 6b). Figure 6a and b show that household pumps operate at approximately 0.165 m³/min for Q_H and 18.75 m³ for H_H, with a 0.9 probability. The Pert distribution effectively fitted the efficiency (η) of household pumps between 50% and 70%, with a median and standard deviation of 0.628 and 0.062, respectively.

Fig. 6.

Probability distributions for characteristics of pumps used at the household level, (a) pumping flow rate (Q_H), (b) delivery head (H_H), (c) household pump operating time for Area 1, (d) household pump operating time for Area 2.

Figure 6c and d illustrate the probability distributions (Pearson 5) of daily household pump operating time (t_H) for Area 1 and Area 2, as calculated using Eqs. (3)-(5) in the methodology. Figure 6c shows that the 0.9 probability is t_H ≤ 0.37 h/day in A1. In the case of A2, the 0.9 probability is t_H ≤ 0.47 h/day (Fig. 6d). These results align with the discussion above on water consumption and household size, which is higher in the case of A2. Households with zero consumption represent zero occupancy during the assessment period.

Impact of household size and topography on energy consumption

Figure 7a-d illustrates four scenarios, each for intermittent and continuous supplies with actual and flat topography in A1 and A2. Figure 7a shows that at median probability, the household receiving a 24/7 supply consumes approximately 37% more energy (6.84 kWh/household/month) compared to the 2-day intermittent case (4.33 kWh/household/month). These results correspond to a household size of 3.5, based on the Monthly consumption for Area 1, as shown in Fig. 5. In contrast, considering the average household size of 5.6 persons in the study region²⁵, which corresponds to a Monthly consumption of 40.6 m³ with a 0.725 probability, the hypothetical continuous supply case (S₁₂) consumes 26.3% more energy than the actual intermittent case (S₁₁). As shown in Fig. 7a, the total energy for the 2-day supply exceeds that of the 24/7 supply at a 0.9 probability, highlighting the impact of household size on energy consumption, which is 8.7 persons per household, as indicated in Fig. 5 for Area 1.

Fig. 7.

Probabilistic pumping energy assessment results for intermittent vs. continuous supplies, (a) Area 1 - actual topography, (b) Area 1 - flat topography, (c) Area 2 – actual topography, (d) Area 2 – flat topography.

Figure 7b highlights the impact of topography by comparing the 2-day supply (S₂₁) with a hypothetical scenario of 24/7 supply (S₂₂) by assuming flat ground in Area 1. For the case of smaller households at 0.5 probability, S₂₂ shows 11.8% more consumption than S₂₁. In contrast, for larger household sizes (with a 0.9 probability), the household pump operates for less than half an hour per day (Fig. 6c). For the intermittent supply case, the energy consumption exceeds that of the continuous supply case by 27.6%. The required delivery head was 22 m for 24/7 supply (S₁₂) for actual ground levels in Area 1, varying between 604 m and 652 m, while it was reduced to 12.4 m for flat ground (S₂₂). Such a reduction (~ 44%) in delivery head highlights the impact of topography on distribution system pumping energy requirements.

Figure 7c illustrates that a 24/7 supply consumes approximately 38.5% more energy (14.06 kWh/household/month) compared to the 2-day intermittent case (8.64 kWh/household/month), for an average household size of 4.1, which corresponds to the 50th percentile for Area 2 in Fig. 5. Conversely, the average household size of 5.6 persons in KSA corresponds to a Monthly consumption of 50.2 m³ at a probability of 0.725, indicating that the actual case of continuous supply (S₃₂) consumes 31.3% more energy than the hypothetical intermittent supply (S₃₁). Figure 7c illustrates that the total energy for the 2-day supply exceeds that of the 24/7 supply for the top 10% of households, with an average of 11.5 persons per household, as shown in Fig. 5 for Area 2.

Figure 7d demonstrates the influence of topography by hypothetically contrasting the 2-day intermittent supply (S₄₁) with a 24/7 continuous supply (S₄₂) by assuming flat ground in Area 2. For the case of average households at 0.5 probability, S₄₂ shows 30.2% more consumption than S₄₁. In contrast, for larger household sizes, the daily operational time of the household pump, approximately thirty minutes (Fig. 6d), exceeds the energy consumption by 17% compared to the continuous supply. The necessary delivery head for actual ground levels in Area 2, ranging from 603 m to 619 m, was 40 m for 24/7 supply (S₃₂) and 31 m for flat ground (S₄₂). This reduction of ~ 23% in delivery head underscores the influence of topography on the energy demands for pumping inside the distribution system.

Impact of water losses on energy consumption

The present study conducted an energy assessment, assuming 0% water losses in hydraulic simulations based on metered consumption records for the households in A1 and A2. The national benchmark for acceptable current annual real losses (CARL) in KSA is 8–10%⁴⁴. Water distribution systems losses between 10% and 30% in Saudi Arabian water supply networks have been reported in past studies¹⁶. The simulations assessed the impact of water loss on the energy requirements for pumping in the distribution systems.

Figure 8a illustrates the results for energy consumption to pump 1 m³ of water to the distribution network, expressed in kWh/m³. The analysis shows that a 24/7 continuous supply consumes around 55% More energy than a 2-day supply under all scenarios in Area A1. For instance, considering 10% water loss, the 7-day pumping consumes (0.244 kWh/m³) compared to the 2-day supply (0.107 kWh/m³). Figure 8a also shows, at a 0.9 probability, a 15.5% increase in energy consumption, accompanied by a corresponding increase in water loss from 10 to 30% in both 2-day and 7-day supplies in A1.

Fig. 8.

Impact of water loss on energy consumption for distribution system in continuous (7-day) and intermittent (2-day) supplies, (a) Area 1, (b) Area 2.

Figure 8b illustrates the results for energy consumption to pump 1 m³ (kWh/m³) to the distribution network, considering 0%, 10%, 20%, and 30% losses for Area 2. The 24/7 supply consumes 50% More energy than the 2-day intermittent case for all the scenarios shown in Fig. 8b for A2. For example, considering 10% water loss, the 7-day pumping consumes (0.399 kWh/m³) compared to the 2-day supply (0.199 kWh/m³). Figure 8b also shows, at a 0.9 probability, a 16.1% increase in energy consumption, accompanied by a corresponding increase in water loss from 10 to 30% in both 2-day and 7-day supplies in A2.

Sensitivity analysis

The energy utilized by a 2-day intermittent supply depends on several variables, including distribution system pump flow, head, and efficiency, as well as the operating hours of the household pump. Additionally, the overall energy use depends on the characteristics of the distribution pumps. The study performed 10,000 Monte Carlo simulations using @Risk Software to identify the factors contributing to the energy variance. Figure 9 demonstrates the simulation results for A1 and A2. The operating hours of household pumps (h/day) have been identified as the Most significant contributor to energy requirements for intermittent supply, accounting for 38.1% contribution to variance (CoV) in A1 (Fig. 9a) and 43.9% in A2 (Fig. 9b). Household pump flow (m³/h) was found to be the second Most important factor affecting energy demand, with a CoV of 32% in A1 and 25.4% in A2. These results highlight the influence of household pumping energy on the total energy required for intermittent supply.

Fig. 9.

Contribution of distribution and household systems variables to the pumping energy variance: (a) distribution pumping energy for 2-day supply A1, (b) distribution pumping energy for 2-day supply A2.

In the case of a 2-day supply, the delivery head (1.2% for A1 and 0.8% for A2) can be seen as the most critical contributor to CoV compared to the flow (0.3% for A1 and 0.1% for A2) and efficiency (0.8% for A1 and 0.7% for A2) of the distribution system pump. For the 7-day continuous supply, the pump head was found to be the Most critical variable, with a CoV of 50.3% in A1 due to significant variations in ground elevations between 604 m and 652 m. In A2, the CoV is 48% for pump efficiency, followed by 37.7% for pump head (results not shown).

Water-energy-carbon indicators

Figure 10a and b illustrate the energy indicators for 2-day intermittent and 7-day continuous supply cases, respectively, for actual scenarios of A1 and A2. The study assumed a 10% water loss for WEC indicator calculations, expecting systems to operate at the allowable water loss benchmark of 8–10% in KSA⁴⁴. Figure 10a shows that the continuous supply consumes 37% more energy per m³ of water than the intermittent supply in A1 at the median (0.5) probability, with ~ 29% at a 0.75 probability, corresponding to the average size in the study region. In contrast, it’s only ~ 3% higher at 0.9 probability, showing the impact of household size on energy consumption. In Fig. 10c, CO₂ emissions estimated as “kg CO₂ eq per m³” of water consumed are 38% higher in the case of a 7-day supply (0.177 kg CO₂ eq/m³) compared to a 2-day supply (0.109 kg CO₂ eq/m³) in A1 at median probability and around 7.3% at 0.9 probability. Figure 10e demonstrates a similar difference between continuous and intermittent supplies for annual CO₂ emissions per person for A1.

Fig. 10.

Probabilistic assessment results for water energy carbon indicators: (a) energy consumption per m³ in A1, (b) energy consumption per m³ in A2, (c) CO₂ emissions per m³ in A1, (d) CO₂ emissions per m³ in A2, (e) annual CO₂ emissions per person in A1, (f) annual CO₂ emissions per person in A2.

Figure 10b shows that the continuous supply consumes 40% more energy per m³ of water than the intermittent supply in A2 at the median (0.5) probability, and it’s almost 22% higher at a 0.9 probability, indicating higher water consumption and less variation in household size compared to A1. These results can also be verified from Fig. 6c and d, which show higher household pump operating times in A2, indicating increased water consumption per person for a continuous supply. In Fig. 10d, CO₂ emissions estimated as “kg CO₂ eq” per m³ of water consumed are 39% higher in the case of a 7-day supply (0.176 kg CO₂eq/m³) compared to a 2-day supply (0.289 kgCO₂eq/m³) in A2 at median probability and around 22.8% at 0.9 probability. The results in Fig. 10f complement the findings shown in Fig. 10d for A2.

Past studies report that up to 60% of Saudi households rely on mineral (bottled) water, while up to 35% use household-level water purifiers^45,46. To assess the carbon footprint of intermittent supply facing water quality challenges, the present study considers the following three scenarios for energy consumption till the consumers end: (i) all households use 20-liter bottled water (BW), (ii) all household use water purifiers (WP), and (iii) realistic case − 60% of households use BW, 35% use WP installed at the kitchen tap, and 5% consider municipal tap water safe for drinking and consume it directly.

The energy and carbon emissions were assessed for the realistic case of A1 using the factors 0.108 kWh/person/day for BW and 0.0425 kWh/person/day, as given in Table 4. Figure 11a illustrates that one person with complete (100%) reliance on bottled water for drinking consumes 3.2 times More daily energy at 0.9 probability than a person with a 24/7 supply providing safe drinking water at the point of use. Furthermore, a comparison between a continuous supply and a 2-day supply with all households equipped with water purifiers at the kitchen tap reveals 1.8 times More energy consumption in the intermittent case. In the realistic case, the potential energy consumption is 2.6 times higher than in the 24/7 supply case. Figure 11b illustrates a similar comparison between the continuous and intermittent supplies in terms of annual carbon emissions generated per person.

Fig. 11.

The impact of bottled water use and drinking water purifiers at the household level on energy consumption for water supply. Abbreviations used in the figure are Bottled Water (BW), Water Purifier (WP), Pumping Energy (PE), and Carbon Emissions (CE).

Discussion

The study assessed the influence of topography and household size on energy and carbon emissions for intermittent supply systems with a 2-day supply in arid regions. The study found an increase in energy demand with an increase in household size, requiring longer operational hours for the household pump to lift water from underground storage tanks to the rooftop tanks. Considering an average household size of 5.6 persons for the City of Buraydah²⁵, the 7-day supply consumed ~ 26% More energy than a 2-day supply. Interestingly, the energy demand for a 2-day supply exceeded the 7-day energy demand for households larger than 8.7, highlighting the impact of water consumption on household pumping energy consumption. These findings are consistent with past studies indicating that larger households, particularly those in India¹², South Africa⁴⁷, and China⁴⁸, experience higher energy consumption due to intermittent water supply. Sanjuan-Delmás et al.⁴⁹ stated that areas with fewer people use significantly more power (up to seven times) than cities with higher populations. Filion⁵⁰ demonstrated that increasing population density from 10 persons/ha to 275 persons/ha could result in a 10% decrease in overall per capita energy use. These critical findings highlight the substantial carbon footprint associated with intermittent supplies, particularly in densely populated areas with a high concentration of apartment buildings or rental properties with shared storage tanks.

The study appraises the impact of topography on the energy requirements of the distribution system for pumping. The distribution networks in the study region operate on rolling terrain with ground elevations ranging between 604 and 652 m in A1 and 603 and 619 m in A2; the distribution pump must operate 9–10 m higher with a delivery head compared to flat terrain. Consequently, the energy consumption for a 2-day supply with rolling terrain is 22% higher in A1 and 14% in A2, while it is 40% higher in A1 and 22% in A2 for a 24/7 supply than the flat terrain scenario. These findings align with the study by the American Council for an Energy-Efficient Economy (ACEEE), which reports a 35% increase in energy and carbon emissions, along with a 45 m difference in ground elevation⁵¹.

The present study assessed energy consumption for the water distribution system using kWh/m³ of supplied water for both intermittent and continuous supplies. The energy consumption was found to be 0.153 to 0.24 kWh/m³ at a 0.5 probability and 0.26 to 0.34 kWh/m³ at a 0.9 probability for a 2-day supply with 10% water losses and an 18–33 m delivery head in the study region. Alresheedi et al.¹¹ reported 0.73 kWh/m³ for combined conveyance (transmission) and distribution in an extensive distribution network in the city of Buraydah. The estimated values in this study are generally consistent with those of Alresheedi et al.¹¹, as their research was based on actual data, which may have resulted in water losses potentially higher than those reported in the present study. Also, the results are aligned with studies conducted on energy requirements for water distribution worldwide, such as 0.1–0.26 kWh/m³ in Taiwan⁵², 0.015–0.41 kWh/m³ in California⁵³, 0.15 kWh/m³ in Japan⁵⁴, and 0.079 0.093 kWh/m³ in China⁵⁵. The energy requirements of a distribution system pumping depend on factors such as pumping duration, population density, elevation changes, and residual head requirements. The global average energy consumption of the water supply system is 0.942 kWh/m³⁵.

Long-term storage at the household level due to intermittent supply results in the absence of residual chlorine, which increases the likelihood of bacteriological water quality failure. Intermittent supply systems have a higher potential for physical, chemical, and microbial water quality failures than continuous supply systems through cracks and poor joints, as they remain unpressurized for longer intervals⁵⁶. Additionally, biofilm growth on the walls of pipes and in household storage tanks can further contribute to water quality failures. Gonzalez et al.⁵⁷ reported that 20% of the samples collected from household storage were bacteriologically contaminated due to a prolonged storage period, resulting in a lack of chlorine in the Central American region. Similarly, Haider et al.¹³ reported microbiological contamination in the distribution networks operating with intermittent supplies in the Qassim Region. Consumers often have no alternative but to rely on bottled water or household-level water purifiers for safe drinking water.

Saudi Arabia is concerned about carbon emissions as most of its electricity is generated from fossil fuels⁵⁸. As of 2020, the Kingdom of Saudi Arabia was among the countries with the highest per capita CO₂ emissions worldwide, with 588 million tons of CO₂ emissions resulting from the burning of fossil fuels⁵⁹. All agencies involved in the production and supply of clean water to the public are eager to comply with Saudi Arabia’s 2030 Vision, which aims to reduce CO₂ emissions to 30 million tons per year⁶⁰. The production and transportation of bottled water, as well as the installation of household-level water purifiers at the kitchen tap, consume energy and contribute to the carbon footprint of the water supply at the consumer end. The unified probabilistic energy analysis of the intermittent supply systems in an arid region of Qassim in KSA shows around 60% increase in energy (0.069 kWh/person/day) and carbon emissions (18.45 kg CO₂ eq/person/year) for the realistic case of mixed use of supplied water, household-level purifier, and 20-liter bottle water, in comparison to 24/7 supply (0.042 kWh/person/day and 11.12 kg CO₂ eq/person/year). Considering 100% household-level purifiers and bottled water for drinking at a median (0.5) and 0.9 probabilities, the intermittent supply case, respectively, revealed nearly 40% and 70% higher energy consumption and CO₂ emission generation.

The typical carbon emissions produced by water distribution systems typically range from 0.05 to 0.08 kg CO₂eq/m³ of supplied water^61,62. Direct comparisons of emissions must consider the local context and operational efficiencies within different water supply systems. As of 2022, More than 99% of Saudi Arabia’s energy comes from the burning of oil and natural gas⁶³. With the diverse potential of sustainable energy sources, such as solar, wind, and geothermal, Saudi Arabia’s water administrations are working diligently to transition from fossil-fuel-dependent energy sources to renewable sources¹¹.

In addition to compromised consumer confidence in the quality of supplied water, intermittent supply has several socio-economic repercussions, placing an extra burden on residents. Wu et al.⁶⁴ highlighted that the lack of time and money required for the maintenance of underground and rooftop tanks is a significant constraint. Dragging such maintenance increases the vulnerability to waterborne diseases in low-income countries and a lack of confidence in middle- to high-income countries in adopting bottled water⁶⁵. A study in low-income areas of Uganda by Ssemugabo et al.⁶⁶ reported that only ~ 32% of the residents effectively maintain the safe water chain after receiving an intermittent supply. Maintenance of household pumps for lifting water from underground tanks to the rooftop is another time- and cost-involving activity for customers with intermittent supply.

Electricity bills are also higher for customers with intermittent supply. Zhu et al.¹² reported that customers receiving a 12-hour supply had 30% higher electricity bills than those with a 24-hour supply in India. Although the existing electricity prices in KSA are relatively lower than those in the rest of the world due to government subsidies, Vision 2030 may result in pricing reforms that promote long-term equity and sustainability⁶⁷. A possible increase in electricity prices would transfer the energy burden, to supply water with adequate head, from the supplier to the user in KSA as well. Despite concerns about energy and water quality, buying and transporting bottled water regularly poses a challenge for most consumers, particularly women, elderly individuals, and those with physical disabilities.

Despite the numerous water quality, human health, and socio-economic issues, water suppliers in KSA and most other arid regions primarily use intermittent supplies for leakage control. Haider et al.¹⁶ reported that the water loss of up to 35% caused by leaks (mostly at service connections) in the distribution networks is the primary concern of KSA’s leaders, politicians, specialists, and municipal administrators. A study by Haider et al.⁴⁴ reported an infrastructure leakage index (ILI) of 20 before the implementation of an active leakage control program for a water distribution network in Buraydah city, Saudi Arabia, with a 2-day supply. Instead of controlling water loss with a few days’ supply every week, asset management through proactive maintenance planning of service connections, valves, and water mains can significantly reduce water losses. The pump’s efficiency generally decreases with increasing motor size and age, which can be improved by regular monitoring and maintenance⁸. A well-managed continuous supply can restore customer confidence and reduce the need for household-level treatment. Water loss control through proactive maintenance planning will justify the cost of transitioning to a continuous supply, resulting in improved water quality, customer satisfaction, and a reduced energy and carbon footprint for water supply systems in arid regions.

Limitations and recommendations

The study assumed a constant water consumption of 250 lpcd to estimate water demand and household size in the study area, which can be different depending on income level, number of appliances (dishwashers and washing machines), lifestyle, attitude towards water consumption, presence of green areas and swimming pools, use of smart and low flow fixtures, and cultural norms. Secondly, the present methodology assumed a constant proportion of household pump use due to the lack of available information on sales volumes. Third, the use of household purifiers is considered to be absent in the continuous supply scenario due to a lack of information. However, some users may have installed such purifiers to ensure the safety of their water.

Future studies on variation in per capita consumption across residential neighborhoods in the KSA and other Middle Eastern countries can improve the reliability of results, establishing the probabilistic behavior of this critical parameter. More detailed future surveys on the proportional use of household pumps can certainly enhance the credibility of the probability function of operational hours. Further research on the use of home purifiers in cases of continuous supply, considering study areas with diverse topography (e.g., the use of booster pumps in large, flat areas) and land uses, will significantly contribute to the accuracy of energy appraisal for water supply systems.

Conclusions

A novel probabilistic framework demonstrates the influence of topography, household size, and type of supply on energy and carbon emissions in arid regions. Water meter data analysis of 1,338 connections in two distribution networks revealed a significant variation in Monthly consumption, with a median of 25 m³ per household (SD = 24.14) in Area 1 and 30 m³ per household (SD = 33.37) in Area 2. Data for thirty-two small household pumps revealed notable variation in flow rate (median = 0.075 m³/min, SD = 0.063 m³/min) and delivery head (median = 15 m, SD = 3.3). The daily household pump operating times of 0.155 h in Area 1 and 0.17 h in Area 2 at the median (0.5) probability, and 0.37 h and 0.47 h at a 0.9 probability demonstrate considerable variation in consumption patterns and household size in the study region.

The study’s novel contribution to appreciating the impact of household pumping from ground to rooftop storage demonstrates captivating findings. The energy analysis revealed 26% More energy consumption by a 24/7 supply compared to a 2-day supply for an average household size of 5.6 persons, demonstrating the direct impact of continuous supply on energy consumption. Notably, the 2-day supply exceeded the 7-day energy requirement for households larger than 8.7, due to the longer operational hours, highlighting the impact of water consumption and household size on energy consumption. All additional energy used at the household-level to get a sufficient head from rooftop tank is a transferred burden from supplier to user. The analysis also showed 14–22% higher energy consumption for a 2-day supply with rolling terrain and an elevation difference of 16–48 m compared to flat topography at the median (0.5) probability.

Monte Carlo simulations identify household pump’s operating time (h/day) as the most significant (CoV = 38% in Area 1 and 44% in Area 2) contributor to energy requirements for intermittent supply, followed by household pump flow (m³/h) with a CoV of 32% in A1 and 25.4% in A2, signifying the influence of household pumping energy on the total energy required for intermittent supply.

Due to contamination in distribution mains, household storage, and the absence of chlorine, 60% consumers rely on bottled water and 35% on household-level purifiers in intermittent supply case. The study shows ~ 60% increase in energy and carbon emissions for the realistic case in comparison to 24/7 supply. The study identifies 40% more energy consumption and CO₂ emission generation when considering 100% household-level purifiers, and approximately 70% in the scenario of complete reliance on bottled water.

In addition to a larger carbon footprint and health risks, the existing strategy of controlling water loss through intermittent supply exposes consumers to several socio-economic challenges, such as the maintenance of household-level purifiers and the transportation of bottled water. Hence, suppliers in arid regions should implement proactive maintenance to control water loss, which justifies the additional cost of transitioning to a continuous supply by improving water quality, enhancing customer satisfaction, and reducing the energy and carbon footprint of water supply systems.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

H.H. conceptualized and developed methodology, performed modeling, supervised the research, and wrote the main manuscript. K.H. performed data curation and analysis and reviewed the main manuscript. M.S. and M.A. verified the results and reviewed the main manuscript. J.M. and A.R.G. validated the results and reviewed the main manuscript.

Funding

This study did not receive any funding.

Data availability

Data cannot be shared due to the confidentiality agreement between the research and data-sharing organizations.

Declarations

Competing interests

The authors declare no competing interests.

Data sharing

Data cannot be shared due to the confidentiality agreement between the research and data-sharing organizations.