Kütük F., Durmuş A., Karaköse E., Kurban R.
2nd International Conference on Engineering, Natural Sciences, and Technological Developments (ICENSTED 2025), Bayburt, Türkiye, 20 Haziran - 23 Ekim 2025, ss.352-360, (Tam Metin Bildiri)
-
Yayın Türü:
Bildiri / Tam Metin Bildiri
-
Basıldığı Şehir:
Bayburt
-
Basıldığı Ülke:
Türkiye
-
Sayfa Sayıları:
ss.352-360
-
Açık Arşiv Koleksiyonu:
AVESİS Açık Erişim Koleksiyonu
-
Kayseri Üniversitesi Adresli:
Evet
Özet
2nd International
Conference on Engineering, Natural
Sciences, and Technological Developments (ICENSTED 2025)
June 20-
23, 2025
352
Scheduling Pressure Reducing Valves for Reducing Water
-Leakage in
Urban
Networks
Using
K-Means Algorithm
Fatih Ku
tuk
1
, Ali Durmu
s
1
, Ercan Karak
ose
2
, Rifat Kurban
3
1
Department
of Electric
al & Electronics Engineering
, Kayseri
University,
Kayseri
, Türkiye
2
Department of Natural Sciences
, Kayseri University, Kayseri, Türkiye
3
Department of C
omputer Engineering, Abdullah Gul University, Kayseri, Türkiye
Corresponding
author
: Rifat Kurban (rifat.kurban@agu.edu.tr)
Abstract
Water losses are a vital
problem
to sustainable water management in urban water distribution networks. This study
presents an approach using K
-Means clustering algorithm to reduce water losses and leakages through
programming of pressure reducing valves
(PRVs). The research uses 24
-hour
water
network pressure and flow
rate
data obtained from a supervisory control and data acquisition (
SCADA)
system monitoring a district
metered
area
(DMA) located in a metropolitan municipality in the Central Anatolia Region of Türkiye. The proposed
methodology divides the temporal flow
rate
data into 12 different time periods. This is achieved by applying K
-
means
clustering to
2nd International Conference on Engineering, Natural
Sciences, and Technological Developments (ICENSTED 2025)
June 20-
23, 2025
352
Scheduling Pressure Reducing Valves for Reducing Water
-Leakage in
Urban
Networks
Using
K-Means Algorithm
Fatih Ku
tuk
1
, Ali Durmu
s
1
, Ercan Karak
ose
2
, Rifat Kurban
3
1
Department
of Electric
al & Electronics Engineering
, Kayseri
University,
Kayseri
, Türkiye
2
Department of Natural Sciences
, Kayseri University, Kayseri, Türkiye
3
Department of C
omputer Engineering, Abdullah Gul University, Kayseri, Türkiye
Corresponding
author
: Rifat Kurban (rifat.kurban@agu.edu.tr)
Abstract
Water losses are a vital
problem
to sustainable water management in urban water distribution networks. This study
presents an approach using K
-Means clustering algorithm to reduce water losses and leakages through
programming of pressure reducing valves
(PRVs). The research uses 24
-hour
water
network pressure and flow
rate
data obtained from a supervisory control and data acquisition (
SCADA)
system monitoring a district
metered
area
(DMA) located in a metropolitan municipality in the Central Anatolia Region of Türkiye. The proposed
methodology divides the temporal flow
rate
data into 12 different time periods. This is achieved by applying K
-
means
clustering to a dataset that maps the flow data to the time of day. Then, the optimal pressure setting values
are determined for each time period
by mapping the lowest and highest flow values at the obtained cluster centers
to the user
-defined minimum and maximum desired network operating pressures.
Simulation studies
were
performed using the K
-means algorithm both the
squared
euclidean
(dSE) and
city
block
distance
(dCB)
metrics
over 50 runs. Performance was evaluated against an error metric that combines the flow rate and hourly mean
absolute error. The results revealed that the dCB
distance metric outperformed the dSE
metric in terms of mean
and best error values. Clustering successfully identified distinct daily flow patterns and provided a dynamic
pressure management strategy: lower pressures (e.g. 3.0 bar) were assigned during low demand periods, while
higher pressures (up to 5.0 bar) were set during peak consumption times. This data
-driven
PRV
scheduling
significantly demonstrates the potential to reduce water leakage and optimize energy consumption in water
distribution systems, thus increasing operational efficien
cy and sustainability while ensuring adequate service
levels
2nd International Conference on Engineering, Natural
Sciences, and Technological Developments (ICENSTED 2025)
June 20-
23, 2025
352
Scheduling Pressure Reducing Valves for Reducing Water
-Leakage in
Urban
Networks
Using
K-Means Algorithm
Fatih Ku
tuk
1
, Ali Durmu
s
1
, Ercan Karak
ose
2
, Rifat Kurban
3
1
Department
of Electric
al & Electronics Engineering
, Kayseri
University,
Kayseri
, Türkiye
2
Department of Natural Sciences
, Kayseri University, Kayseri, Türkiye
3
Department of C
omputer Engineering, Abdullah Gul University, Kayseri, Türkiye
Corresponding
author
: Rifat Kurban (rifat.kurban@agu.edu.tr)
Abstract
Water losses are a vital
problem
to sustainable water management in urban water distribution networks. This study
presents an approach using K
-Means clustering algorithm to reduce water losses and leakages through
programming of pressure reducing valves
(PRVs). The research uses 24
-hour
water
network pressure and flow
rate
data obtained from a supervisory control and data acquisition (
SCADA)
system monitoring a district
metered
area
(DMA) located in a metropolitan municipality in the Central Anatolia Region of Türkiye. The proposed
methodology divides the temporal flow
rate
data into 12 different time periods. This is achieved by applying K
-
means
clustering to a dataset that maps the flow data to the time of day. Then, the optimal pressure setting values
are determined for each time period
by mapping the lowest and highest flow values at the obtained cluster centers
to the user
-defined minimum and maximum desired network operating pressures.
Simulation studies
were
performed using the K
-means algorithm both the
squared
euclidean
(dSE) and
city
block
distance
(dCB)
metrics
over 50 runs. Performance was evaluated against an error metric that combines the flow rate and hourly mean
absolute error. The results revealed that the dCB
distance metric outperformed the dSE
metric in terms of mean
and best error values. Clustering successfully identified distinct daily flow patterns and provided a dynamic
pressure management strategy: lower pressures (e.g. 3.0 bar) were assigned during low demand periods, while
higher pressures (up to 5.0 bar) were set during peak consumption times. This data
-driven
PRV
scheduling
significantly demonstrates the potential to reduce water leakage and optimize energy consumption in water
distribution systems, thus increasing operational efficien
cy and sustainability while ensuring adequate service
levelsa dataset that maps the flow data to the time of day. Then, the optimal pressure setting values
are determined for eachWater losses are a vital problem to sustainable water management in urban water distribution networks. This study presents an approach using K-Means clustering algorithm to reduce water losses and leakages through programming of pressure reducing valves (PRVs). The research uses 24-hour water network pressure and flowrate data obtained from a supervisory control and data acquisition (SCADA) system monitoring a district metered area (DMA) located in a metropolitan municipality in the Central Anatolia Region of Türkiye. The proposed methodology divides the temporal flowrate data into 12 different time periods. This is achieved by applying K means clustering to a dataset that maps the flow data to the time of day. Then, the optimal pressure setting values are determined for each time period by mapping the lowest and highest flow values at the obtained cluster centers to the user-defined minimum and maximum desired network operating pressures. Simulation studies were performed using the K-means algorithm both the squared euclidean (dSE) and city block distance (dCB) metrics over 50 runs. Performance was evaluated against an error metric that combines the flow rate and hourly mean absolute error. The results revealed that the dCB distance metric outperformed the dSE metric in terms of mean and best error values. Clustering successfully identified distinct daily flow patterns and provided a dynamic pressure management strategy: lower pressures (e.g. 3.0 bar) were assigned during low demand periods, while higher pressures (up to 5.0 bar) were set during peak consumption times. This data-driven PRV scheduling significantly demonstrates the potential to reduce water leakage and optimize energy consumption in water distribution systems, thus increasing operational efficiency and sustainability while ensuring adequate service levels.period
by
2nd International Conference on Engineering, Natural
Sciences, and Technological Developments (ICENSTED 2025)
June 20-
23, 2025
352
Scheduling Pressure Reducing Valves for Reducing Water
-Leakage in
Urban
Networks
Using
K-Means Algorithm
Fatih Ku
tuk
1
, Ali Durmu
s
1
, Ercan Karak
ose
2
, Rifat Kurban
3
1
Department
of Electric
al & Electronics Engineering
, Kayseri
University,
Kayseri
, Türkiye
2
Department of Natural Sciences
, Kayseri University, Kayseri, Türkiye
3
Department of C
omputer Engineering, Abdullah Gul University, Kayseri, Türkiye
Corresponding
author
: Rifat Kurban (rifat.kurban@agu.edu.tr)
Abstract
Water losses are a vital
problem
to sustainable water management in urban water distribution networks. This study
presents an approach using K
-Means clustering algorithm to reduce water losses and leakages through
programming of pressure reducing valves
(PRVs). The research uses 24
-hour
water
network pressure and flow
rate
data obtained from a supervisory control and data acquisition (
SCADA)
system monitoring a district
metered
area
(DMA) located in a metropolitan municipality in the Central Anatolia Region of Türkiye. The proposed
methodology divides the temporal flow
rate
data into 12 different time periods. This is achieved by applying K
-
means
clustering to a dataset that maps the flow data to the time of day. Then, the optimal pressure setting values
are determined for each time period
by mapping the lowest and highest flow values at the obtained cluster centers
to the user
-defined minimum and maximum desired network operating pressures.
Simulation studies
were
performed using the K
-means algorithm both the
squared
euclidean
(dSE) and
city
block
distance
(dCB)
metrics
over 50 runs. Performance was evaluated against an error metric that combines the flow rate and hourly mean
absolute error. The results revealed that the dCB
distance metric outperformed the dSE
metric in terms of mean
and best error values. Clustering successfully identified distinct daily flow patterns and provided a dynamic
pressure management strategy: lower pressures (e.g. 3.0 bar) were assigned during low demand periods, while
higher pressures (up to 5.0 bar) were set during peak consumption times. This data
-driven
PRV
scheduling
significantly demonstrates the potential to reduce water leakage and optimize energy consumption in water
distribution systems, thus increasing operational efficien
cy and sustainability while ensuring adequate service
levelsthe lowest and highest flow values at the obtained cluster centers
to the user
-defined minimum and maximum desired network operating pressures.
Simulation studies
were
performed using the K
-means algorithm both the
squared
euclidean
(dSE) and
city
block
distance
(dCB)
metrics
over 50 runs. Performance was evaluated against an error metric that combines the flow rate and hourly mean
absolute error. The results revealed that the dCB
distance metric outperformed the dSE
metric in terms of mean
and best error values. Clustering successfully identified distinct daily flow patterns and provided a dynamic
pressure management strategy: lower pressures (e.g. 3.0 bar) were assigned during low demand periods, while
higher pressures (up to 5.0 bar) were set during peak consumption times. This data
-driven
PRV
scheduling
significantly demonstrates the potential to reduce water leakage and optimize energy consumption in water
distribution systems, thus increasing operational efficien
cy and sustainability while ensuring adequate service
levels