Professional Documents
Culture Documents
1 © IWA Publishing 2016 Water Science & Technology: Water Supply | in press | 2016
ABSTRACT
Amin Abo-Monasar
Delivering water in sufficient quantity and acceptable quality is the main objective of water
Department of Civil Engineering,
distribution networks (WDN) and at the same time is the main challenge. Many factors do affect the University of Hafr Al-Batin,
P.O. Box 479,
delivery of water through distribution networks. Some of these factors are relevant to water quality, Hafr Al-Batin 31991,
Saudi Arabia
quantity and the condition of the infrastructure itself. The deterioration of water quality in the WDN
Muhammad Al-Zahrani (corresponding author)
leads to failure at the water quality level, which can be critical because it is closest to the point of Department of Civil and Environmental
Engineering, Water Research Group,
delivery and there are virtually no safety barriers before consumption. Accordingly, developing a
King Fahd University of Petroleum and Minerals,
powerful monitoring system that takes into consideration water demand distribution, the P.O. Box 686,
Dhahran 31261,
vulnerability of the distribution system and sensitivity of population to the deterioration of water Saudi Arabia
E-mail: mzahrani@kfupm.edu.sa
quality can be very beneficial and, more importantly, could save lives if there was any deterioration of
water quality due to operational failure or cross-contamination events. In this paper, a framework for
water quality monitoring system that considers water demand distribution, the vulnerability of the
system and sensitivity of population using fuzzy synthetic evaluation and optimization algorithms is
developed. The proposed approach has been applied to develop a monitoring system for a real WDN
in Saudi Arabia.
Key words | fuzzy rule-based modelling, monitoring stations, optimization, water distribution
network, water quality
INTRODUCTION
Water transported by distribution network usually passes the network. To prevent, control and/or reduce any poten-
through a complex treatment process before its distribution. tial deterioration of water quality and to ensure that the
Once the treated water is pumped into the network, there WDN is functioning the way it should be, constant and com-
are usually no defensive lines that could protect consumers prehensive assessment and monitoring of the WDN should
from any deterioration of water quality due to any be established.
unexpected consequence that happened during the transpor- In 1990 and 1992, Lee and Deininger were the first to
tation. These consequences might be gradual such as the develop a scientific criterion to determine representative
variation of chlorine levels or rapid such as contamination water quality monitoring locations by developing a coverage
caused by intentional criminal or terrorist attacks. Many fac- method concept. Since then, several improvements have
tors could affect water quality within the water distribution been made to the original approach developed by Lee & Dei-
networks (WDN) directly or indirectly. Some of these fac- ninger () in order to reduce the complexity by considering
tors are relevant to the infrastructure of the system, such steady and extended period simulations for water demand,
as pipe materials, age and breaks. Some are related to the adding key parameters, including water age and pipe diam-
operational practices of the network such as water age in eters, and considering the rapid water quality deterioration
doi: 10.2166/ws.2016.174
Uncorrected Proof
2 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
rather than considering only the gradual change in water deterioration of water quality. The second step is determining
quality. Although most of the studies that dealt with identify- the optimal locations of MSs that maximizes the covered
ing the optimal locations of monitoring stations (MSs) in demand (monitored demand) and takes into consideration the
WDN have used water demand as the key parameter for regional vulnerability and sensitivity risks. Detailed descriptions
locating the MSs (Lee & Deininger ; Kumar et al. ; of the two phases are explained in the following paragraphs.
Kessler et al. ; Liu et al. , ), other approaches
used other parameters such as water age rather than consid- Step I: Risk prioritization
ering the demand (Tryby & Uber ). In addition, there
were several objectives used to determine the optimal Risk prioritization requires an aggregation of measurable and
locations of MSs, such as minimizing the percentage of popu- non-measurable factors to estimate vulnerability, sensitivity
lation at risk due to sudden contamination attack, minimizing and overall risk at specific points in the WDN using routinely
the exposed population, minimizing the detection time of the collected data. Figure 1 provides the prioritization framework
contaminant, minimizing the contaminated water consumed, for determining vulnerability, sensitivity and overall risk indi-
minimizing the number of times the system fails to detect a ces. The framework consists of four levels or generations of
contaminant and minimizing the length extent that the con- factors aggregated in a hierarchical fashion to develop differ-
taminant would reach in the pipelines (Berry et al. , ent risk indices. The top of the pyramid represents the
). In general, integer programming was applied to overall risk that depends on two factors, i.e. vulnerability and
locate the MSs (Lee & Deininger ; Kumar et al. ; sensitivity in the first level. These two factors are determined
Berry et al. ; Liu et al. ); however, other optimization through aggregation of various factors in the previous levels.
algorithms were also used, such as genetic algorithm, robust The vulnerability index is determined by aggregating factors
optimization and Bayesian belief network (Al-Zahrani & related to hydraulics, structural integrity and quality of the
Moied ; Grayman et al. ; Watson et al. ). water pumped through the WDN. The population sensitivity
In most of the studies that dealt with identifying the opti- index is linked to the presence of certain groups of consumers
mal locations of MSs in WDN, water demand was the only served by the WDN, who may suffer from various conse-
or the key parameter. Moreover, factors such as the vulner- quences if the water quality in the WDN deteriorates. These
ability of the system were not considered, despite its indices were developed for different regions in the WDN
importance and relation to the contamination events. Fur- based on the local data for each region (Francisque et al. ).
thermore, it was assumed that the population will be Due to the complexity of the system, a variation of the
exposed equally to any deterioration consequences in the collected data (measurable and non-measurable) and diffi-
WDN. However, the sensitivity of the population to con- culty in obtaining precise data for some of the factors, an
tamination events varies based on several factors such as index-based approach using fuzzy synthetic evaluation
age, standard of living and population density. In this (FSE) was used to describe vulnerability, sensitivity and
study, an approach for determining the optimal locations overall risk. An estimate of risk at a given location in the
of a monitoring system was developed considering water WDN will provide a representative value for a predefined
demand, the vulnerability of the system and population sen- geographical region. A region represents an influence zone
sitivity to water quality deterioration. in which the values of all contributing factors are assumed
to be fixed. Once the risk values are determined in various
sectors of the WDN, they can be ordered, ranked or priori-
METHODOLOGY tized based on the overall risk index.
making. In FSE, variables are used as inputs for decision classified using wide range of classes (for example: High,
making in which numeric or non-numeric data can be eval- Med and Low) when using fuzzy logic. Therefore, for crisp
uated using natural linguistics such as ‘low’, ‘med’ and ‘high’ logic a single input data can be a member of only one
(Ross ). In this study, FSE is used to quantify and esti- class while it can belong to more than one class in fuzzy
mate indices for hydraulics of the system, structure logic. This advantage gives more freedom in classifying
integrity, water quality, vulnerability, sensitivity and, conse- different parameters and considering fuzziness or uncertai-
quently, risk index. The FSE follows the following steps nity involved. Accordingly, due to the nature of input
(Sadiq & Rodriguez ; Sadiq et al. ; Khan & Sadiq parameters used in this study, fuzzy logic was used.
; Francisque et al. ). Triangular, Gaussian and trapezoidal fuzzy subsets are
usually used to fuzzify input data (Wang ; Lee ;
Fuzzification Sadiq et al. , ; Francisque et al. ) as shown in
Figure 2. Using these fuzzy subsets, input data can be
Fuzzification can be defined as a process by which measur- mapped to fuzzy sets. In this study, triangular and trapezoi-
able and non-measurable input data are transformed into a dal membership functions were used. For example, suppose
homogeneous scale (0 to 1), or the process of changing Figure 3 shows a fuzzy set for TDS level in water. There are
crisp values into fuzzy values (Francisque et al. ; Ross three fuzzy subsets, low (μlow), med (μmed) and high (μhigh). If
). When using different types of input data, which TDS in water is 200 ppm, then it is said that TDS level has a
could be a mix of measurable (i.e. pH, TDS, … etc) and membership of 1 to ‘low’ fuzzy subset, and if the level is
non-measurable (i.e. standard of living, activity, … etc) par- 800 ppm, then it is said that TDS level has a membership
ameters, fuzzy logic is preferred over crisp logic, since of 1 to ‘high’ fuzzy subset. If TDS level is 450 ppm, then it
crisp logic is limited to two classes only; True or False, 0 is said that TDS level has a membership of 0.25 to ‘low’
or 1, High or Low…etc. However, input data can be fuzzy subset and 0.75 to ‘med’ fuzzy subset.
Uncorrected Proof
4 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
Defuzzification
The process by which fuzzy sets [μlow, μmed, μhigh] are con-
verted to a representative crisp value is called
defuzzification (Wang ; Francisque et al. ). There
are several methods to defuzzify fuzzy sets; however,
weighted average method or scoring is preferred by many
Figure 4 | TDS Membership function: (a) Triangular 1 (b) Triangilar 2 (c) Triangular 3 and (d) Tranpezoidal.
Thresholds
between 0 and 4.58 m/s. Although very few pipes have a highest and lowest risk indices for all regions is shown in
velocity of 2 m/s or higher, the most frequent violations Table 2.
were velocities lower than 0.4 m/s. Unlike pressure and Infrastructure and structure integrity of the WDN were
velocity, water age is within the recommended range in judged based on pipe breaks, age and material, in addition
all regions for all demand scenarios, which implies that to potential intrusions of wastewater or industrial waste in
water age will have no effect in the prioritization of risk case of the occurrence of pipe breaks. Historical records
and developing the hydraulic index. A summary of the of Al-Khobar municipality indicate that most pipe breaks
Uncorrected Proof
8 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
occurred in the central and southern regions of the city have been widely reported in asbestos pipes. According to
which are the oldest in the city (44 years). The main pipe Al-Khobar water authority, approximately 65% of the total
material in the central and southern regions is asbestos, breaks occurred in asbestos pipes. On the other hand, asbes-
and PVC in the northern regions. Cracks and breakages tos is considered as a carcinogenic material, although its risk
Uncorrected Proof
9 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
Figure 7 | Al-Khobar WDN: (a) sub-regions (b) vulnerability risk index for water quality (c) sensitivity of the regions and (d) total risk index for each region.
relevant to drinking water is not substantial. Accordingly, regions of the city have multi issues. Structure integrity
the infrastructures in the central and southern regions are risk index ranges between 0.18 and 0.63 (Table 2). As
not in good condition and might be a potential source of expected, regions in the southern part of the city have a
intrusions. higher risk index compared to the city center and the north-
Pipe material, age, breaks, sanitary coverage, industrial ern part. Regions in the north, such as 98 and 119, have PVC
and wastewater intrusions, water table levels and soil sur- pipes, relatively newer pipes, low pipe breakage ratio and no
rounding pipes give a clear view about the infrastructural industrial activity or potential intrusions (either industrial or
condition of the WDN. It is obvious that the southern wastewater), which explains the relatively low risk index in
Uncorrected Proof
10 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
Table 2 | Indices for regions with highest and lowest total risk
Regions
Water quality 0.21 0.14 0.18 0.19 0.32 0.26 0.27 0.28 0.20 0.19 0.15 0.20 0.23 0.25 0.24 0.45
Structure integrity 0.59 0.59 0.61 0.59 0.60 0.61 0.63 0.59 0.18 0.18 0.30 0.18 0.18 0.18 0.37 0.43
Hydraulics 0.21 0.80 0.29 0.21 0.21 0.00 0.21 0.75 0.29 0.21 0.21 0.00 0.21 0.22 0.31 0.30
Vulnerability 0.34 0.51 0.36 0.33 0.38 0.29 0.37 0.54 0.22 0.19 0.22 0.12 0.20 0.21 0.31 0.40
Sensitivity 0.75 0.68 0.35 0.26 0.17 0.03 0.06 0.86 0.46 0.21 0.39 0.10 0.10 0.14 0.60 0.59
Total risk 0.54 0.60 0.35 0.30 0.28 0.16 0.21 0.70 0.34 0.20 0.30 0.11 0.15 0.18 0.45 0.49
these regions. Regions in the south, such as 79 and 82, have intrusion to the network. Standard of living, population den-
asbestos pipes, older pipes, high pipe breakage ratio and sity, activity, number of beds in hospitals and number of
industrial potential intrusions, which explains the higher students are measures for regional sensitivity in case of
risk index compared to the northern regions. For other par- any possible deterioration of water quality within the
ameters (TDS, pH, free chlorine, temperature and turbidity), WDN. There are groups in the community who are more
the analysis indicates that they are within the acceptable sensitive to any deterioration of water quality than the
standards. Accordingly, a water quality risk index for each other groups. Patients and school students, as an example,
region was developed by aggregating the physico-chemical are more sensitive to waterborne illnesses and their
and microbial factors as shown in Table 2. It is obvious immune system is either weak or not fully developed (Fran-
that the water quality risk is low in the WDN since water cisque et al. ). Aggregation of all these factors
quality parameters are generally within the recommended determines the sensitivity of each region. It is obvious that
standards. Water quality indices range between 0.14 and regions 74, 75, 94, 120 and 121 are more sensitive than
0.43, which is a relatively low risk (Table 2). the other regions. These factors were aggregated to develop
The vulnerability index of the WDN is developed by the sensitivity risk index for the different regions in the city
aggregating the hydraulic properties, water quality and struc- as shown in Figure 7(c). The sensitivity index ranges
tural integrity. The major factors affecting vulnerability are between 0.03 and 0.86. Regions in the center, such as 74,
hydraulic properties and structure integrity since the water 75, 94 and 121, are more sensitive to any deterioration of
quality risk index is low all over the city. Regions which water quality in the WDN. Regions in the north and in the
have a high risk due to hydraulic properties and structure south, such as 81, 82, 102 and 119, have the least sensitivity
integrity, i.e. regions 75 and 94, are the regions in the city mainly because they have low population density, low
center which have high vulnerability. Regions in the north, number of students and less residential areas.
such as 98 and 102, have a low risk in terms of hydraulic Total regional risk is based on the aggregations of vul-
properties and structure integrity, which is reflected in the nerability and sensitivity of each region. Total risk indices,
vulnerability index. Regions in the extreme south of the as shown in Figure 7(d), range between 0.11 and 0.70. In
city, such as regions 77 and 82, have a low risk index due general, the results indicate that northern and southern
to hydraulic properties, but they have a high risk index regions have the least risk index, which is a reflection of
due to structure integrity, which explains why the vulner- the generally low vulnerability and sensitivity of these
ability indices in that zone are higher than the northern regions. Regions in the city center are the most sensitive
regions but at the same time less than the regions in the regions in the city, as shown in Figure 7(d). This explains
city center, as shown in Figure 7(b). the relatively high risk in the city center compared to the
Sensitivity of the region gives a sense of possible health other regions in the city. Table 2 shows the detailed risk indi-
risk consequences in case of any possible contaminant ces for all the regions.
Uncorrected Proof
11 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
Locating MSs Risk prioritization using FSE showed that each region
has different characteristics than the other regions, which
Currently, there are 15 MSs distributed over Al-Khobar makes the expected consequences in case of water quality
WDN as shown in Figure 8. In general, most of the MSs deterioration vary for the different regions. Severe conse-
are located close to major pumping locations such as quences are expected in some regions more than the
Makkah tank, central pumping stations and pumping wells others since these regions are having higher risk (vulner-
to monitor water quality directly after pumping to ensure ability and sensitivity). Therefore, demand and regional
that chlorine levels and TDS are within the acceptable risk were used to optimally locate the MSs.
limits. However, this distribution of MSs ignores the fact It was noticed that the demand coverage of the monitor-
that water quality can be deteriorated during delivery ing system increases as the number of the proposed MSs
through contaminant intrusions, cracks or due to intentional increases, as shown in Table 3. Based on the current prac-
criminal and/or terrorist acts. tices with 15 MSs, only 19.56% of the demand can be
In Al-Khobar WDN, there are 871 non-zero nodes, monitored; however, the coverage increased to 59.18%
water demand nodes, which can be considered as PMSs when 50 MSs were used. Figure 9 shows the optimal distri-
and have an equal chance of being chosen during the optim- bution of MSs when 15, 20, 30 and 50 MSs were proposed.
ization process based on the objective functions and Each region is monitored by at least one MS as shown in
constraints defined previously. Figure 9, with the exception of region 81, which is not
Table 3 | Percentage of demand coverage for the proposed MSs located in the central area of the city compared to the
extreme northern and southern parts. The central part of
Number of MSs 15 20 30 50 the city is the most populated zone in the city with higher
Demand coverage (%) 19.56 31.94 44.97 59.18 demand and higher sensitivity. In addition, the risk assess-
ment showed that the central regions are also vulnerable.
populated and, accordingly, it is not served by the distri- Accordingly, the higher number of MSs is justified by the
bution network. The figure indicates that more MSs are high demand and high risk in these regions.
Figure 9 | Locations of proposed MSs for: (a) 15 MSs (b) 20 MSs (c) 30 MSs and (d) 50 MSs.
Uncorrected Proof
13 A. Abo-Monasar & M. Al-Zahrani | Locating water monitoring stations based on vulnerability Water Science & Technology: Water Supply | in press | 2016
multiple demand patterns: a flaw of demand coverage method Silvert, W. Fuzzy indices of environmental conditions.
and modification. Frontiers of Environmental Science & Ecological Modelling 130 (1/3), 111–119. doi: 10.1016/
Engineering 6 (2), 204–212. doi: 10.1007/s11783-011-0364-9. S0304-3800(00)00204-0.
MathWorks Fuzzy Logic Toolbox: for Use with MATLAB: Tryby, M. & Uber, J. Representative water quality sampling in
User’s Guide. MathWorks, Incorporated. water distribution systems. In Bridging the Gap: Meeting the
Ministry of Economy and Planning (MEP) Population and World’s Water and Environmental Resources Challenges.
Housing Census 2010. Central Department of Statistics and ASCE, pp. 1–10. doi: 10.1061/40569(2001)404.
Information, Saudi Arabia. USEPA EPA Drinking Water Advice Note No. 5: Turbidity in
Ross, T. Fuzzy Logic with Engineering Applications. Wiley. Drinking Water. The Environmental Protection Agency. US
Sadiq, R. & Rodriguez, M. J. Fuzzy synthetic evaluation of Environmental Protection Agency.
disinfection by-products – a risk-based indexing system. Wang, L. Fuzzy systems are universal approximators. In IEEE
Journal of Environmental Management 73 (1), 1–13. International Conference on Fuzzy Systems 1992. IEEE, pp.
Sadiq, R., Husain, T., Veitch, B. & Bose, N. Risk-based 1163–1170. doi: 10.1109/FUZZY.1992.258721.
decision-making for drilling waste discharges using a fuzzy Watson, J., Murray, R. & Hart, W. Formulation and
synthetic evaluation technique. Ocean Engineering 31 (16), optimization of robust sensor placement problems for
1929–1953. doi: 10.1016/j.oceaneng.2004.05.001. drinking water contamination warning systems. Journal of
Sadiq, R., Kleiner, Y. & Rajani, B. Aggregative risk analysis Infrastructure Systems 15 (4), 330–339. doi: 10.1061/(ASCE)
for water quality failure in distribution networks. Journal of 1076-0342(2009)15:4(330).
Water Supply: Research and Technology— AQUA, 53 (4). WHO Guidelines for Drinking-water Quality, 3rd edn, Vol 1.
Sarbatly, H. & Krishnaiah, D. Free chlorine residual content World Health Organization, Geneva.
within the drinking water distribution system. International Zadeh, L. A. Fuzzy sets as a basis for a theory of possibility.
Journal of Physical Sciences 2 (8), 196–201. Fuzzy Sets and Systems 1, 3–28.
First received 23 January 2016; accepted in revised form 26 October 2016. Available online 11 November 2016