ABSTRACT
Under carbon peaking and carbon neutrality goals in China, many public institutions have started to decarbonise their energy consumption sectors, saving water and indirectly reducing energy consumption by controlling leaks in water supply networks. Universities receive more attention for their huge water consumption. The campus water supply system of a university in Xi'an was selected as the study object. Through real-time monitoring of the water consumption of the water supply network system, flow trends and leakage characteristics were analysed in conjunction with changes in water consumption, and WB-Easy Calc (water balance analysis software) proposed by the International Water Association was used to calibrate and evaluate the system for leakage. The results show that: with each building as a separate metering zone, the minimum night-time flow rate of the university occurs between 2:00 and 5:00 a.m., and the WB-Easy Calc analysis can obtain more accurate apparent leakage; therefore, the combination of the water balance method and the minimum night-time flow rate method can effectively assess the water use and leakage of the university, which is a good inspiration for assessing the water-saving potential of the university and formulating the corresponding water-saving measures.
HIGHLIGHTS
Each building as a district metering area unit can estimate water leakage.
The value of μ–σ can be approximated to obtain real water leakage at night.
Combining minimum night flow and WB-Easy Calc, more accurate apparent leakage can be obtained.
Monitoring applied to two water supply leakage assessment means can improve accuracy.
INTRODUCTION
Water is essential and irreplaceable for human livelihood, social economic development, and eco-environment protection (Cosgrove & Loucks 2015; Vorosmarty et al. 2000). However, the world is facing increasingly severe water resource challenges, and although countries have implemented water-saving measures, waste caused by per capita water consumption exceeding the quota, overpressure outflow, and pipeline leakage is still widespread. Especially during the water supply process, pipeline leakage not only causes huge loss of water resources, but also indirectly leads to energy waste (Talpur et al. 2020). In September 2020, China announced to achieve peak carbon emission by 2030 and carbon neutrality by 2060 (Xu et al. 2021). Under the carbon peaking and carbon neutrality goals of China, low-carbon management for enterprises has become crucial. Among them, reducing water resource waste caused by water supply network leakage can effectively reduce energy consumption and thus reduce carbon emission.
In public buildings, due to the high mobility of their personnel, water conservation awareness varies greatly, and the subjective initiative of water conservation needs to be improved. Although the public sector has introduced water-saving enterprise standards, most of them are limited to the level of norms and standards, and the support of water-saving technology system is not yet perfect. Although the public sector has introduced water conservation and leakage management enterprise standards, most of them are limited to the level of norms and standards, and the support of the management technology system is not yet perfect. More seriously, most enterprises lack a refined, information-based management platform and leakage control system, as well as specific systems for water consumption control, leakage detection in pipeline networks, and assessment of water conservation, resulting in sloppy management and low water-use efficiency.
Worldwide, the leakage level of water supply networks fluctuates between 10 and 55% (Ávila et al. 2021). Currently, leakage detection and assessment methods for water supply networks can be broadly divided into two types: one is the top-down approach, and the other is the bottom-up approach. The top-down approach is based on water balance calculation for leakage assessment, and the bottom-up approach is the minimum night flow. Tsitsifli & Kanakoudis (2010) developed water balance software more suitable for Greek water utilities, which allows a clear representation of the components of non-revenue water, actual leakage, and apparent leakage (Setianingsih & Karnaningroem 2019). Apparent leakage includes illegal water use, metering errors, and water volume generated by data processing errors. This method requires a large amount of water data and water information. The apparent leakage is estimated mainly from these water-use data, and finally, the actual leakage is obtained by counting the apparent leakage, which has a significant leakage assessment error (Amoatey et al. 2018). The bottom-up approach is the minimum night flow, which is an important indicator to assess the actual leakage in the district metering area (DMA). The minimum night flow occurs between 1:00 and 5:00 a.m. most of the time, when most water points are closed, and the minimum value during this time is closest to the actual leakage amount. Al-Washali et al. (2018) found that minimum night flow analysis could not be performed in intermittent supplies and that impacted the estimation of legitimate water use during the minimum night flow period. Farah & Shahrour (2017) improved the application of the minimum night flow method based on the determination of flow thresholds, above which a leak alarm is generated. This data analysis method provides the ability to quickly detect pipeline ruptures, thereby reducing the run time of leaks. Haibo (2016) used confidence intervals of (μ − 1.5, μ + 1.5) to analyse the night water consumption of 176 residential water users in a DMA and obtained the minimum value of legal water consumption for that night period. Yu et al. (2021) developed an integrated water balance model by analysing the difference between apparent and actual losses to achieve quantification of different water consumption events and detection of water leakage events in water supply networks. This approach allows us to determine the proper water leakage of the corporate network more precisely, but ignores the apparent leakage volume. Al-Washali et al. (2020) studied all methods of apparent and real losses. They proposed that at least two ways should be used to assess the water loss components of an entire network to reasonably model and monitor the loss reduction in the distribution network. These two assessment methods studied the leakage of the networks in a particular area, but lacked studies of leakage of the network inside buildings. Smart metering systems can be used not only for the entire water distribution networks, but also for public utilities and individual customers. This provides a more adequate understanding of water leaks within the house building and becomes a reliable tool for leak analysis and assessment of indoor and outdoor water supply networks (Serov 2020). Based on the real-time monitoring system, identification and prevention of water-use anomalies can be achieved in both public utilities and inside buildings (Stewart et al. 2018). Kudva et al. (2015) designed a real-time water balance monitoring system for a campus, and the result showed that this system helps to better manage water resources on a campus. Visser et al. (2021) analysed a case study in the city of Cape Town (South Africa) where smart meters were installed in 105 schools, allowing the schools’ management to have a complete monitoring of both water supply and consumption. The water usage in the schools was reduced by 15–26% after proper interventions. GST4Water proposed a user-level leak identification as the goal of a real-time monitoring and treatment system for household water use, by using advanced digital or intelligent water meters, which may lead to water savings and more effective demand management (Luciani et al. 2019). Water conservation is promoted by analysing the data generated by digital meters and providing feedback to consumers and water utilities (Rahim et al. 2020). Applying real-time monitoring techniques to the water balance method and the minimum night flow method, the analysis of real-time data can verify the water balance and estimate the level of water loss in the water supply networks (Farah & Shahrour 2017). Smart meters also enable greater access and interaction with the collected data, thus providing real-time data to monitor supply networks, and respond to issues such as leakages (Idris 2006).
Graphical summary of leakage assessment for university pipeline networks.
METHODOLOGY
Using a university in Xi'an as a study site, each building in the university was viewed as a separate zoned unit for metering (DMA). A DMA is a specially defined area of a distribution network (usually defined by closing a valve) (Brothers 2003). The concept of DMA was first introduced to the water supply industry in the early 1980s (Farley & Trow 2003). Flow analyses are used to determine the extent of leakage in an area. Water-use efficiency and leakage in higher education institutions (HEIs) were analysed and assessed by combining water balance analysis and minimum night flow rates using water-use data from the school's intelligent real-time monitoring platform. According to the water consumption characteristics of colleges and universities, the water consumption pattern of colleges and universities is analysed through the intelligent detection platform, and the week with the least water consumption during the winter holidays is selected, so that the real leakage loss of the pipeline network of the school is approximated through the method of normal analysis, and then the data of the intelligent monitoring platform are used to verify it, and finally a more accurate apparent leakage loss is obtained by combining with water balance software WB-Easy Calc. In the following, the introduction of the specific application method is briefed.
Study area description
Monitoring
In 2019, the university installed remote water meters in each building, completed the construction of an intelligent real-time monitoring platform, and built an energy consumption monitoring system for extensive data analysis. Considering the whole campus as a region, there are three water inlets and three primary water meters installed. Each building as a DMA and one secondary remote water meter is installed at the water inlet of each building, with a total of 68 secondary water meters. The data of each water meter were uploaded every 15 min. The intelligent monitoring platform provides information technology support for the accuracy and effectiveness of analyzing nighttime water flow. At the same time, the collection of real-time data is helpful for water balance calculation, production and sales difference analysis, etc., avoiding empirical errors and achieving accurate estimation.
The real-time monitoring platform collects the water consumption data of each part of the water supply network system. It detects the accumulated flow of water meters online in real time. The platform adopts real-time communication and data collection technology. It enables managers at all levels to monitor and analyse the water consumption of each building through the form of web publishing, and in the early stage of the construction of the reserved port according to the need to achieve functional expansion, such as monitoring time setting module, cloud-computing analysis module, and alarm setting module.
Water balance method
The water balance system developed by the International Water Association (IWA) was used to provide a detailed division of the water supply system. The water balance method divides the water supply of the system into revenue water and non-revenue water. The revenue water volume consists of the billed metered water volume and the billed unmetered water volume. The non-revenue water volume includes the unbilled metered water volume and the unbilled unmetered water volume. The water losses include the real losses and the apparent losses.
In this study, WB-Easy Calc was used for data analysis to evaluate water-use efficiency in the water supply network and efficiency evaluation at the management level (Melato 2010).
Minimum night flow method
Minimum night flow period
The minimum night flow is the minimum value of the night water flow of a water supply network system, which usually occurs during the night period from 1:00 to 5:00 a.m. The true leakage occupies the largest proportion of the total water volume, and the water consumption has the least influence on assessing the real losses.
According to the recommendations of the IWA, the measurement of the minimum night flow is selected for the period 2:00–4:00 a.m. when users use the least amount of water at night and avoid the interference of other data (Marzola et al. 2021).
The water consumption from September 2022 to September 2023 monitored through the campus' smart monitoring platform is shown in Figure 3(a). It can be seen that the time when the campus uses the least amount of water is during the winter holidays, between 10 February and 20 February. Although in the holiday period, resident students may still study and rest in line with normal days, taking 7 days from 10 February to 16 February. The average water consumption from 1:00 to 5:00 a.m. is shown in Figure 3(b). The campus' water consumption did not fluctuate much during the winter vacation period. There were apparent fluctuations in water consumption from 1:00 to 2:00 a.m. on average days, indicating that some students were not asleep during this time and some people still used water. The water consumption gradually increased from 4:30 to 5:00 a.m., implying that students began to use water. The water consumption does not fluctuate much from 2:00 to 5:00 a.m., suggesting that students are asleep, and the water consumption is almost the minimum night flow of the campus.
Data statistical analysis methods
The normal distribution is denoted as N(μ, σ2). The normal distribution graph is symmetric at about μ. The closer the value taken to the central axis μ indicates, the higher the probability. Regardless of the values of μ and σ, 68.26% of the sampled data fall in the range of μ ± σ; 95.46% in the range of μ ± 2σ; and 99.73% in the range of μ ± 3σ (Navarro & del Águila 2017). The random variable x mainly takes discounts in the interval (μ − 3σ, μ ± 3σ).
RESULTS AND DISCUSSION
Minimum night flow estimation of real losses
(a) Comprehensive daily water volume in 2022–2023 and (b) average water consumption at night for 7 days.
(a) Comprehensive daily water volume in 2022–2023 and (b) average water consumption at night for 7 days.
From the figure, the flow rate of the water meter at this time is relatively stable, around 2 L/s. Moreover, on 10 February, there was a sudden increase in water consumption from 3:00 to 3:30 a.m., and the water consumption did not fluctuate much for the rest of the time, indicating that there were users using water in this period, so the water consumption in this time is not considered. Finally, 88 valid data were obtained.
The data were processed through mathematical statistics, and the frequency distribution table is shown in Table 1.
Frequency distribution table
Flow range (L/s) . | Number . | Frequency . | Cumulative percentage . |
---|---|---|---|
1.72–1.76 | 1 | 0.0114 | 1.1 |
1.77–1.82 | 15 | 0.1705 | 18.2 |
1.83–1.87 | 11 | 0.125 | 30.7 |
1.88–1.92 | 12 | 0.1364 | 44.3 |
1.93–1.97 | 6 | 0.0682 | 51.1 |
1.98–2.02 | 13 | 0.1478 | 65.9 |
2.03–2.07 | 8 | 0.0909 | 75 |
2.08–2.12 | 7 | 0.0795 | 83 |
2.13–2.17 | 5 | 0.0568 | 88.6 |
2.18–2.22 | 5 | 0.0568 | 94.3 |
2.23–2.27 | 2 | 0.0227 | 96.6 |
2.28–2.32 | 0 | 0 | 96.6 |
2.33–2.37 | 2 | 0.0227 | 98.9 |
2.38–2.42 | 1 | 0.0113 | 100 |
Flow range (L/s) . | Number . | Frequency . | Cumulative percentage . |
---|---|---|---|
1.72–1.76 | 1 | 0.0114 | 1.1 |
1.77–1.82 | 15 | 0.1705 | 18.2 |
1.83–1.87 | 11 | 0.125 | 30.7 |
1.88–1.92 | 12 | 0.1364 | 44.3 |
1.93–1.97 | 6 | 0.0682 | 51.1 |
1.98–2.02 | 13 | 0.1478 | 65.9 |
2.03–2.07 | 8 | 0.0909 | 75 |
2.08–2.12 | 7 | 0.0795 | 83 |
2.13–2.17 | 5 | 0.0568 | 88.6 |
2.18–2.22 | 5 | 0.0568 | 94.3 |
2.23–2.27 | 2 | 0.0227 | 96.6 |
2.28–2.32 | 0 | 0 | 96.6 |
2.33–2.37 | 2 | 0.0227 | 98.9 |
2.38–2.42 | 1 | 0.0113 | 100 |
There were 88 valid data. Only one flow datum appears in the range of 1.72–1.76 L/s with a probability of only 1.14%, which is considered a small probability event and can be excluded. In the range of 1.77–2.12 L/s, 72 flow data appeared, accounting for 82% of the total data. Among them, the data frequency in the range of (1.77–1.82 L/s) is the highest, reaching 17.05%, and it is in the low value range in the entire flow distribution. Therefore, the actual leakage can be estimated to be more reliable in the range of (1.77–1.82 L/s).
Figure 5(a) shows a p–p plot reflecting the degree of conformity between the actual cumulative frequency probabilities of the variables and the expected theoretical cumulative probabilities. From the figure, it can be seen that the data points overlap with the theoretical straight lines, indicating that the predicted cumulative probabilities and the actual cumulative probabilities match very well and obey normal distribution. From Figure 5(b), the plot of the cumulative frequency residuals, it can be seen that the residuals are basically within the upper limit of Y = 0 uniformly distributed, and the absolute values of most of the residuals are within 0.03, which indicates that the normality of these data is good.
From the measured flow data, the μ was calculated to be 1.9697 L/s with a standard deviation of 0.1508. The true leakage of the campus is determined to be in the range of 1.77–1.82 L/s. Considering the need to verify whether the choice of confidence is related to the amount of data, the confidence intervals corresponding to 68.3, 95.5, and 99.74% were selected at the confidence level of (μ − kσ, μ + kσ, where k is taken as 1–3), with a range of statistical expectation of 1–3 standard deviations, where μ − σ is 1.8189 L/s, μ − 2σ is 1.6681 L/s, and μ − 3σ is 1.5173 L/s. Then the confidence intervals using (μ − σ, μ + σ) are optimal. The flow data for μ − σ are within the range of 1.77–1.82 L/s, regardless of whether all or some of the data are used.
The results show that the number of individual data volumes collected has little effect on analysing the real losses from 2:00 to 5:00 a.m. in the campus. It is more credible to analyse the flow data from 2:00 to 5:00 a.m. in the district with a 68.3% confidence. Because the leakage is becoming more serious, leading to a gradual increase in the minimum flow rate at night, the minimum flow rate at night should be taken as a smaller value. When taking two times the standard deviation, it contains many larger data values, contrary to the principle of taking a smaller value of the minimum flow rate at night. When taking three times the standard deviation of the data, the calculated flow rate value does not exist in the actual measured value.
The same method was used to analyse water leakage in the buildings to verify the accuracy of the minimum night flow, considering μ − σ values. There are 35 buildings on the campus. The data revealed fluctuations in the flow of water used at night from 2:00 to 5:00 a.m. in 10 buildings. The water consumption at night in the remaining buildings fluctuates occasionally. Still, the water consumption is zero most of the time, indicating no leakage inside the remaining buildings. The data processing table and the real leakage values for the 10 structures are shown in Table 2.
School water data processing for leaky building
Building . | . | A . | B . | C . | D . | E . | F . | G . | H . | I . | J . |
---|---|---|---|---|---|---|---|---|---|---|---|
Number | Valid data | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 |
Invalid data | 0 | 0 | 6 | 6 | 0 | 6 | 12 | 12 | 6 | 6 | |
Standard deviation | 0.0304 | 0.0153 | 0.0362 | 0.027 | 0.0243 | 0.0474 | 0.0228 | 0.031 | 0.0081 | 0.0376 | |
Variance | 0.001 | 0 | 0.001 | 0.001 | 0.001 | 0.002 | 0.001 | 0.002 | 0 | 0.001 | |
Average | 0.1194 | 0.0282 | 0.0559 | 0.2194 | 0.0595 | 0.1001 | 0.0812 | 0.058 | 0.141 | 0.1678 | |
Minimum | 0.011 | 0.011 | 0.011 | 0.111 | 0.011 | 0.022 | 0.033 | 0.011 | 0.122 | 0.022 | |
Maximum | 0.2 | 0.078 | 0.122 | 0.289 | 0.133 | 0.244 | 0.122 | 0.222 | 0.167 | 0.289 | |
Real losses’ interval | 0.083–0.093 | 0.006–0.016 | 0.007–0.015 | 0.184–0.198 | 0.028–0.038 | 0.049–0.059 | 0.05–0.06 | 0.017–0.027 | 0.129–0.138 | 0.123–0.143 | |
μ − σ | 0.089 | 0.0129 | 0.0197 | 0.1924 | 0.0352 | 0.0528 | 0.0584 | 0.027 | 0.133 | 0.1302 | |
μ − 2σ | 0.0585 | – | – | 0.1654 | 0.0108 | 0.0054 | 0.0356 | – | 0.1249 | 0.0926 |
Building . | . | A . | B . | C . | D . | E . | F . | G . | H . | I . | J . |
---|---|---|---|---|---|---|---|---|---|---|---|
Number | Valid data | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 | 91 |
Invalid data | 0 | 0 | 6 | 6 | 0 | 6 | 12 | 12 | 6 | 6 | |
Standard deviation | 0.0304 | 0.0153 | 0.0362 | 0.027 | 0.0243 | 0.0474 | 0.0228 | 0.031 | 0.0081 | 0.0376 | |
Variance | 0.001 | 0 | 0.001 | 0.001 | 0.001 | 0.002 | 0.001 | 0.002 | 0 | 0.001 | |
Average | 0.1194 | 0.0282 | 0.0559 | 0.2194 | 0.0595 | 0.1001 | 0.0812 | 0.058 | 0.141 | 0.1678 | |
Minimum | 0.011 | 0.011 | 0.011 | 0.111 | 0.011 | 0.022 | 0.033 | 0.011 | 0.122 | 0.022 | |
Maximum | 0.2 | 0.078 | 0.122 | 0.289 | 0.133 | 0.244 | 0.122 | 0.222 | 0.167 | 0.289 | |
Real losses’ interval | 0.083–0.093 | 0.006–0.016 | 0.007–0.015 | 0.184–0.198 | 0.028–0.038 | 0.049–0.059 | 0.05–0.06 | 0.017–0.027 | 0.129–0.138 | 0.123–0.143 | |
μ − σ | 0.089 | 0.0129 | 0.0197 | 0.1924 | 0.0352 | 0.0528 | 0.0584 | 0.027 | 0.133 | 0.1302 | |
μ − 2σ | 0.0585 | – | – | 0.1654 | 0.0108 | 0.0054 | 0.0356 | – | 0.1249 | 0.0926 |
As seen from Table 2, the analysis of the buildings where leakage occurred using the standard distribution method revealed that the values of μ − σ were all within the interval of real losses. It is further demonstrated that whether the confidence is selected independent of the number of data is generalisable, and the value of μ − σ can be directly approximated as the actual losses of the campus. Then the leakage loss of each building under secondary metering is 0.7505 L/s. The above shows that the total leakage of the primary metering campus network is 1.81 L/s. The losses of the campus outdoor water supply network are 1.0595 L/s.
p–p test for normal distribution: (a) normal p–p plot of flow and (b) detrended normal p–p plot of flow.
p–p test for normal distribution: (a) normal p–p plot of flow and (b) detrended normal p–p plot of flow.
Table 3 shows that the average campus network leakage is 0.94 L/s, which is 0.1195 L/s different from that of 1.0595 L/s obtained from the typical distribution analysis, and the difference is little. In turn, the feasibility of normal distribution modelling and the accuracy of taking μ − σ for the leakage values of the pipe network can also be verified on the basis of monitoring data at 15 min intervals.
Real losses’ volume of campus outdoor water network
Time . | Primary metering (L/s) . | Secondary metering (L/s) . | Outdoor water network leakage (L/s) . | Leakage average value (L/s) . |
---|---|---|---|---|
2:00 | 1.983 | 1.157 | 0.826 | 0.94 |
2:15 | 1.984 | 1.05 | 0.934 | |
2:30 | 2.019 | 1.21 | 0.809 | |
2:45 | 1.968 | 1.002 | 0.967 | |
3:00 | 2.256 | 1.029 | 1.226 | |
3:15 | 2.324 | 1.168 | 1.156 | |
3:30 | 2.13 | 1.022 | 1.108 | |
3:45 | 1.9 | 1.033 | 0.867 | |
4:00 | 1.951 | 1.131 | 0.82 | |
4:15 | 1.956 | 1.129 | 0.827 | |
4:30 | 2.014 | 1.116 | 0.898 | |
4:45 | 1.952 | 1.105 | 0.847 | |
5:00 | 2.043 | 1.113 | 0.93 |
Time . | Primary metering (L/s) . | Secondary metering (L/s) . | Outdoor water network leakage (L/s) . | Leakage average value (L/s) . |
---|---|---|---|---|
2:00 | 1.983 | 1.157 | 0.826 | 0.94 |
2:15 | 1.984 | 1.05 | 0.934 | |
2:30 | 2.019 | 1.21 | 0.809 | |
2:45 | 1.968 | 1.002 | 0.967 | |
3:00 | 2.256 | 1.029 | 1.226 | |
3:15 | 2.324 | 1.168 | 1.156 | |
3:30 | 2.13 | 1.022 | 1.108 | |
3:45 | 1.9 | 1.033 | 0.867 | |
4:00 | 1.951 | 1.131 | 0.82 | |
4:15 | 1.956 | 1.129 | 0.827 | |
4:30 | 2.014 | 1.116 | 0.898 | |
4:45 | 1.952 | 1.105 | 0.847 | |
5:00 | 2.043 | 1.113 | 0.93 |
Water balance leakage analysis based on WB-Easy Calc
Campus primary metering and secondary metering from 2:00 to 5:00 a.m. on 7 days.
The billed metered water quantity includes water consumption for bathing and retail stores. The amount of water used for bathing is 5,366.52 m3, and the amount used for commercial stores is 14,086.21 m3.
The billed unmetered water volume includes 7,151.1 m3 of direct drinking water in each campus building.
The unbilled metered water includes water used by students in the restrooms and water used for lawn landscaping. The amount of water used in the restrooms of each building is 293,533.7 m3, and the amount of water used for lawn greening is 3,421.7 m3.
The unbilled unmetered water mainly includes the water consumption for road cleaning, secondary water supply facilities’ cleaning, and some water consumption for firefighting. The amount of water used for road cleaning is 17,320 m2, and 3 m3 of water is sprinkled by the sprinkler once, which is three to four times a day in summer. In spring and autumn, it is two times a day. When the temperature reaches below zero, generally in winter, no sprinkling or occasional sprinkling occurs. By checking the national greenhouse data system, the lowest temperature in Xi'an is below zero in December, January, and February. To reduce slippery roads due to icing, the amount of water used for road cleaning is 2,340 m3 in these 3 months based on no sprinkling.
The campus' secondary water supply facilities have two underground water pools with 1,100 m3 and three high-level cisterns totalling 116 m3. The campus flushes the secondary water supply facilities twice a year. The cleaning volume of the water tank is approximately equal to the volume of the water tank, and the cleaning volume of the underground pool is estimated at 30% of the volume of each pool. The cleaning volume of the secondary water supply facilities is 892 m3.
The campus' fire protection pipe networks are connected to the roof-top domestic water tank of the teaching building and apartment no. 2, which can temporarily act as water supply units when the domestic water supply is insufficient, and the annual fire protection water consumption is 3,826 m3.
According to the above data, the invalid water supply rate of the campus is 14.3%, and the unmetered water consumption accounts for 3.6%, which indicates that the overall water supply network metering penetration rate of the campus is high. However, the real losses of water are still high. It is known from Section 3.1 that the leakage of the pipe network outside the building is higher compared with that inside the building because of the more extended construction period of this outside pipe network. There are 17 leakage points found in the whole campus in 1 year, and only 3 in the external water supply network, which means that the leakage inside the building can easily be found and repaired in time. While the external water supply network is all underground, which is not easy to find leakage and lasts for a long time, and the overall leakage is more extensive. The unbilled water consumption accounts for 76.4% of the total water consumption, while the water consumption in the bathroom of each building accounts for 96.6% of the unbilled water consumption. Chinese colleges and universities do not charge students for water used in public buildings, such as teaching buildings, dormitory buildings, and office buildings. In most of these buildings, water is used for students' teaching and living, so the unbilled water consumption accounts for a relatively large amount. To save water and build water-saving colleges and universities, we need to pay attention to this part of water consumption.
As mentioned above, in an in-depth exploration of new strategies for leakage assessment, an advanced technological framework based on sensor networks and monitoring data was used to digitally connect real water leakage scenarios with an advanced monitoring platform. An accurate model of normal distribution of water volume was successfully constructed through the refined night-time flow analysis method. This model not only accurately assesses water leakage, but also combines the principle of water balance with in-depth analysis of production and sales differentials, thus realising collaborative management and service optimisation of the water supply system. This innovative practice has achieved remarkable results in actual leakage control.
In particular, this methodological system has played a key role in the construction of energy and resource-saving campuses in universities. It not only provides a clear positioning, a clear roadmap, and a specific implementation path for campus water conservation, but also greatly enhances the practical utility and guiding significance of water conservation research. Further, the implementation of this methodology provides a valuable model for measuring and evaluating leakage in regional pipeline networks in universities and other areas of public utilities, which will help universities and other areas of public utilities to achieve the long-term goal of energy and resource conservation.
CONCLUSIONS AND FURTHER RESEARCH
Based on the water consumption of a university in Xi'an, this paper analyses the minimum night-time flow rate of the campus with the help of a normal distribution model by selecting the minimum flow rate for 7 consecutive days during the winter holidays, selecting a reasonable confidence interval to assess the actual leakage value of the school, and reviewing the monitoring data. Combined with WB-Easy Calc water balance software, the school's water-use situation and water-saving potential were accurately assessed. The following conclusions can be drawn.
(1) In exploring the time period during which the minimum flow occurs at night, there was diversity in the selection of time intervals due to the differences in water-use practices in different places. Initially, a broad time range of 1:00–5:00 a.m. was selected, and then, by comparing the average water consumption data for that time period over a 7-day period on a winter holiday day with that on a normal day, finally 2:00–5:00 a.m. was considered as the time period during which the minimum night-time flow occurs at the school.
(2) The frequency of the occurrence of flow from 2:00 to 5:00 a.m. in the primary meter on the campus is consistent with the law of normal distribution. The method, with a confidence level of 68.3%, and a confidence interval of (μ − σ, μ + σ), can be used to approximate the actual losses at night in this campus.
(3) Based on the analysis of the secondary water meter data, the values of μ − σ for water consumption in the building are in the interval of the real losses’ estimation, and the difference in the results is little. This indicates that after obtaining some night flow water consumption data for a campus, the μ − σ value can approximate the proper leakage water volume after verifying that it conforms to the normal distribution.
(4) Using the night-time minimum flow method to accurately calculate the real leakage volume, and then implementing the WB-Easy Calc water balance analysis for the campus, effectively reduces the bias of apparent leakage estimation, reduces empirical errors, and significantly improves the water-use efficiency and management effectiveness of the university network.
Based on this experimental study, the combination of independent metering zone technology and advanced metering tools shows excellent performance in leakage management of pipeline network in school districts, which not only has excellent economic and management benefits, but also has good promotion value in the management of production and sales differentials and in the reduction of loss and leakage control. However, DMA management still requires managers to invest efforts in analysing the data platform, and there is a risk of management lag, which may prolong the time of leakage sensing and thus affects the realisation of the overall benefits of DMA construction.
To improve the efficiency and accuracy of leakage assessment and management, the potential of near real-time monitoring smart metering for water management can be explored in depth. By optimising the monitoring platform settings, the data upload interval can be shortened to 5–10 min to ensure real-time data accuracy. Combined with the cloud-computing data analysis module within the platform, the rich monitoring data are used to present the night minimum flow characteristic curve in a visual way, accurately calculate the amount of water leakage, and deeply analyse potential leakage events. When the monitoring data reach the pre-set alarm threshold (which is based on the continuous analysis of leakage events after the sensor data are uploaded to the cloud-computing platform), the sensor immediately sends an alarm prompt to the operation and maintenance unit of the intelligent monitoring platform. This mechanism significantly reduces the management burden of water supply enterprises, reducing the dependence on the professionalism of personnel. Managers only need to verify the water volume and arrange leakage detection based on the water volume alarm information, which greatly improves the management efficiency of water enterprises, further shortens the time of perceived leakage, and reduces water loss.
In the future, leakage management will be based on the expansion of the functionality of the intelligent monitoring platform with the construction of DMA partitions to enable deeper applications. Comprehensive solutions to further reduce leakage will include, firstly, real-time collection of core data from the water supply network through the deployment of flow and pressure sensors, followed by undergoing a rigorous data processing process whereby, through algorithmic analyses, the flow and pressure data are able to reveal potential leakage patterns in the water supply network and instantly locate leakage locations. Based on these data, precise pressure management is implemented for regional DMAs. By installing pressure regulating valves at the inlet points of the DMAs, zone pressures are adjusted based on real-time flow and pressure data to achieve optimal operating pressure control. At the same time, the physical loss levels of the DMAs are accurately estimated by combining the top-down IWA water balance method and the bottom-up minimum night flow method to provide a comprehensive assessment of the network's performance and guide the identification of potential water-use pattern improvements.
Through this closed-loop leakage management measure, we aim to reduce and control actual losses and build an efficient and intelligent regional leakage management system. A truly integrated and intelligent DMA is achieved to ensure efficient use of water resources and continuous optimisation of management strategies.
ACKNOWLEDGEMENTS
This study was supported by the National Natural Science Foundation of China (No. 52070152), and the Shaanxi Innovative Provincial Water Resources Department Science and Technology Plan Project (No. 2021slkj-6).
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.