In the Republic of Korea, multi-regional water supply pipelines were intensively laid during the rapid economic growth period; however, these pipelines are now aging. The cost of renewal of the aged pipelines would be enormous. Therefore, limited resources need to be distributed effectively in accordance with the priority of renewal. In this study, the risks involved in the multi-regional water supply pipeline of region K were calculated and assessed for setting the priority of renewal. The probability of failure of the pipeline was calculated through a quantification theory, and the effects of failure on the water supply pipeline were calculated by interpreting a pressure driven analysis. The risk considering the probability and consequence of failure comprehensively was found to be the largest at section A12. Three alternatives were considered for risk reduction, and the effect of each alternative on the risk reduction was analyzed. As a result, the construction of double piping by laying a new pipeline was found to be the most effective measure for risk reduction. The results of this study can be utilized as the basic data for establishing an investment plan for the renewal of the water supply pipeline and asset management plan.

INTRODUCTION

In the Republic of Korea, multi-regional water supply pipelines were intensively laid during the rapid economic growth period between the 1970s and 1980s. Although those facilities made a great contribution to the development of Korea, they have been gradually deteriorating with time. As of 2013, it is expected that 1,381 km, which accounts for 27.17% of the entire multi-regional water pipes, will need renewal from 2014 to 2024. Further, the length of the pipe for renewal is expected to increase to 2,666 km, which accounts for 52.45% of the entire pipeline, between 2024 and 2034 (Koo et al. 2014). If an administrator allows water supply pipelines to age without appropriate countermeasures, the safety of the pipeline will be sharply decreased. Because the decreased safety can lead to water leakage, increased impacts of disasters, degradation of water quality and other relevant problems, there are concerns for the need of measures to reduce the risk of pipelines and increase safety (Kim et al. 2014). In particular, a majority of the multi-regional water supply pipelines consists of transmission pipelines with large diameters, which can cause large-scale accidents. In addition, the cost of renewal of the aged multi-regional water supply pipelines would be enormous. Therefore, limited resources need to be distributed effectively in accordance with the priority of renewal. To do so, the priority of renewal must be established (Park 2007).

To set the priority for renewal, the risk of failure of the water supply pipeline may be considered. Risk assessment methods can be categorized into connectivity/topological risk and hydraulic risk (Ostfeld 2004). The connectivity/topological risk is associated with the measurement of the probability that the components are not physically continuous in a fixed water pipe network. The hydraulic risk is related to the measurement of the probability that water cannot be supplied with consistent water quantity and pressure. Wagner et al. (1988) proposed an analytical methodology to evaluate the connectivity, and Goulter et al. (1990) defined system risk as the probability that a failure occurs in the system using dis-connectivity and the concept of topological risk. Al-Bahar (1988) classified risks into three categories as follows: risk events refer to inappropriate events (suspension accident, etc.) which are inherent to a system, uncertainty of risk factor refers to the probability of occurrence of risk factor (the failure rate of a pipeline), and the potential failure of risk factor refers to the results of an inherent accident (time spent on the repair of the pipeline, the level of consumption of water, redundancy, etc.). According to these, a risk event is defined by multiplying the uncertainty of the event by the potential loss or gain. The Environmental Protection Agency (EPA) (2012) also proposed a similar measure to calculate risk as Al-Bahar (1988), defining risk as the product of the multiplying probability of failure (PoF) and the consequence of failure (CoF) and proposed risk mitigation. Choi (2013) estimated the failure rate of a pipeline through logistics regression analysis and calculated risk by assessing the effect of failure on the pipeline in terms of water suspension quantity. Rahman et al. (2014) studied the PoF and a variety of influences of the failure to calculate risks of water supply pipelines in the subject region of their study. Further, they marked the result of the risks on a map. Choi et al. (2014) analyzed the water suspension using a fault tree analysis and estimated the amount of risk reduction when an administrator replaces the old pipe with a new one and connects it to the water district. Koo (2015) calculated the failure rate through safety rate and calculated shortage quantity on each demand node as the CoF to calculate risks.

A water supply pipeline system should be capable of supplying a certain level of water to the regions the pipeline affects, even during abnormal situations such as water leakage and pipeline failure. The situation in which a suspension cannot be used by consumers is regarded as a risk. Here, stochastic figures including uncertainty factors such as leakage and failure, the degree of damage caused by suspension, and redundancy to compensate for the damage in suspension accidents may be taken into consideration (Goulter 1987). That is to say, the stochastic figure for unstable supply of water to consumers as a result of various uncertainties, the effects of the damage caused by suspension time (required for repair of pipe, water consumption, etc.) and indexes including redundancy rates can be defined as water suspension risks.

In this study, therefore, the risks of multi-regional water supply pipelines of region K, which is the subject of this study, were calculated and assessed for setting the priority of renewal. The PoF of the water supply pipeline was calculated through a quantification theory, and the effects of the failure on the water supply pipeline were calculated by interpreting a pressure driven analysis (PDA). Three alternatives (replacement of the existing pipeline, increasing the level of retention time in a reservoir, and double piping by laying a new pipeline) were taken into consideration, and the effect of each alternative on the reduction of risk was analyzed.

METHODS

Study area

In this study, the multi-regional water supply pipeline of region K, in which the pipeline was laid more than 30 years ago, with a number of failure cases, was selected. Table 1 shows the summary of study areas. Figure 1 illustrates the study area and the water supply pipeline under study. The target water supply pipelines were classified into 14 major systems, by dividing them into 232 sections, depending on laying conditions. Further, each section was named in a regular sequence (e.g. A1, D3, E8).
Table 1

Description of multi-regional water supply of region K

Multi-regional supply of K region
Installation year 1972–2001 
Pipe type steel pipe, ductile cast iron pipe 
Pipe diameter 200 mm–900 mm 
Break history 32 (from 1980 to 2013) 
Pipe length 69.967 km 
Supply area and demand KJ city (68,732 m3/d) 
Multi-regional supply of K region
Installation year 1972–2001 
Pipe type steel pipe, ductile cast iron pipe 
Pipe diameter 200 mm–900 mm 
Break history 32 (from 1980 to 2013) 
Pipe length 69.967 km 
Supply area and demand KJ city (68,732 m3/d) 
Figure 1

Study area-multi-regional water supply of region K.

Figure 1

Study area-multi-regional water supply of region K.

Risk calculation method

In this study, the risk of failure of the water supply pipeline is defined as the product of the PoF of the pipeline and the CoF. The risks were calculated using formula (1). 
formula
1

PoF (quantification theory)

Quantification theory is a multi-dimensional data analysis method developed by Chikio Hayashi in the 1950s. In Japan, Hayashi's methods of quantification are well known and widely used in various fields, such as social and marketing surveys and psychological and medical research, where information is obtained mainly in the form of qualitative categories (Arai et al. 2009; Jiang et al. 2010). In particular, the quantification theory type II, a quantitative concept, is a statistical technique to find independent variables that can discriminate or describe the measured dependent variable to deduce a discriminant function. Further, similarly to a discriminant analysis, the quantification theory predicts the group of dependent variables to which new data will belong. Variables including the existence of a failure in a pipeline, the type of a pipeline, and laying environment cannot be quantitatively interpreted. Therefore, the quantification theory, which converts the variables of a qualitative concept to a quantitative one, was applied to calculate the failure rate of a pipeline.

The deduction of a discriminant function for the calculation of the failure rate of a pipeline by using the samples of the multi-regional pipeline and the failure history of region K was not useful for the deduction of statistical significance due to a small number of samples. In this study, therefore, the data of the multi-regional pipeline of the entire nation were utilized to deduce a discriminant function. The water pipelines were classified with respect to the changes in the installation year of the water pipeline, the types of pipes, pipe diameter, and topographic conditions. As a result, the water pipelines nationwide were classified into 8,170 groups. The history of failures that occurred on the pipeline for the last 34 years from 1980 to 2013 was used. The case with two or more failures was set to 1 and the case without failure occurrence was set to 0, because the pipeline with two or more failures was considered to be affected by only the laying environment rather than the construction or other conditions. There were 193 pipelines in which two or more failures occurred. The independent variables set to figure out dependent variables are listed in Table 2.

Table 2

Variable and category for quantification analysis

VariableCategoryConditionNumber of groupRatio of group (%)Length (km)
Installation year Installation 1 Before 1983 1,019 12.47 487.42 
Installation 2 1983–1993 1,448 17.72 732.11 
Installation 3 1993–2003 2,635 32.25 2,216.08 
Installation 4 After 2003 3,068 37.55 1,544.69 
Pipe type Type 1 Steel pipe 4,114 50.36 2,756.40 
Type 2 Iron pipe 3,341 40.89 1,889.89 
Type 3 Others 715 8.75 333.99 
Coating Coating 1 Epoxy 2,807 34.36 1,951.40 
Coating 2 Enamel 1,325 16.22 829.79 
Coating 3 No coating 4,038 49.42 2,199.10 
Pipe diameter Diameter 1 Below 500 mm 2,186 26.76 1,062.87 
Diameter 2 500 mm–1,000 mm 2,309 28.26 1,495.40 
Diameter 3 1,000 mm–1,500 mm 1,777 21.75 1,096.20 
Diameter 4 1,500 mm–2,000 mm 1,015 12.42 661.19 
Diameter 5 Above 2,000 mm 883 10.82 664.63 
Pipe length Length 1 Below 250 m 3,928 48.08 306.19 
Length 2 250 m–500 m 1,103 13.50 396.90 
Length 3 500 m–750 m 755 9.24 466.42 
Length 4 Above 750 m 2,384 29.18 3,810.78 
Topography Topography 1 Road 2,413 29.53 1,648.49 
Topography 2 Land 4,968 60.81 2,732.16 
Topography 3 River or others 789 9.66 599.63 
Depth Depth 1 Below 1.5 m 1,503 18.40 651.28 
Depth 2 1.5 m–2.0 m 2,030 24.85 1,330.53 
Depth 3 2.0 m–2.5 m 1,943 23.78 1,341.41 
Depth 4 2.5 m–3.0 m 1,150 14.08 776.88 
Depth 5 Above 3.0 m 1,544 18.90 880.18 
VariableCategoryConditionNumber of groupRatio of group (%)Length (km)
Installation year Installation 1 Before 1983 1,019 12.47 487.42 
Installation 2 1983–1993 1,448 17.72 732.11 
Installation 3 1993–2003 2,635 32.25 2,216.08 
Installation 4 After 2003 3,068 37.55 1,544.69 
Pipe type Type 1 Steel pipe 4,114 50.36 2,756.40 
Type 2 Iron pipe 3,341 40.89 1,889.89 
Type 3 Others 715 8.75 333.99 
Coating Coating 1 Epoxy 2,807 34.36 1,951.40 
Coating 2 Enamel 1,325 16.22 829.79 
Coating 3 No coating 4,038 49.42 2,199.10 
Pipe diameter Diameter 1 Below 500 mm 2,186 26.76 1,062.87 
Diameter 2 500 mm–1,000 mm 2,309 28.26 1,495.40 
Diameter 3 1,000 mm–1,500 mm 1,777 21.75 1,096.20 
Diameter 4 1,500 mm–2,000 mm 1,015 12.42 661.19 
Diameter 5 Above 2,000 mm 883 10.82 664.63 
Pipe length Length 1 Below 250 m 3,928 48.08 306.19 
Length 2 250 m–500 m 1,103 13.50 396.90 
Length 3 500 m–750 m 755 9.24 466.42 
Length 4 Above 750 m 2,384 29.18 3,810.78 
Topography Topography 1 Road 2,413 29.53 1,648.49 
Topography 2 Land 4,968 60.81 2,732.16 
Topography 3 River or others 789 9.66 599.63 
Depth Depth 1 Below 1.5 m 1,503 18.40 651.28 
Depth 2 1.5 m–2.0 m 2,030 24.85 1,330.53 
Depth 3 2.0 m–2.5 m 1,943 23.78 1,341.41 
Depth 4 2.5 m–3.0 m 1,150 14.08 776.88 
Depth 5 Above 3.0 m 1,544 18.90 880.18 

The size of the data of contrasting groups should be the same for the reliability of the discriminant analysis (Kim et al. 2013). To set the same size of data groups, systematic sampling was used. Further, it was applied after transforming the scale type variables including installation year, length of the pipeline, and depth to ordinal data. When the sample was selected, a random number table was utilized. Meanwhile, for the independent variable, which is to be included in discriminant, only a statistically significant variable was selected through a stepwise method. SPSS ver. 21.0 (IBM) was employed for statistical analysis. When the statistical analysis was conducted through a stepwise method, generally used in many cases, 0.05 of significant probability of F value, which includes variables in discriminant function, and 0.10 of significance probability of F value, which eliminates variables from discriminant function, were used (Koizumi et al. 2012).

The Wilk's lambda test, used to test the significance of the discriminant function, employing chi-squared distribution and Box's M with a corresponding F statistic, is a procedure for testing homogeneity of variance and covariance. These two tests were used to determine the validation of the discriminant analysis (Katos 2007).

The results of the discriminant analysis estimate a discriminant score, which determines where independent variable should belong according to the dependent variable groups. In this study, the discriminant score for the estimation of a qualitative result was converted to a quantitative failure rate to calculate the risk quantitatively. The failure rate was calculated through the discriminant score deduced from each pipeline and regression analysis between failure rates of each pipeline.

CoF (PDA)

The hydraulic analysis of a pipe network can be divided into two techniques according to the major subject to be interpreted: demand driven analysis (DDA) and PDA. DDA is used as a basic principle for most of the commonly used pipe network analysis programs such as EPANET. This is the method for the calculation of the head loss of each node under the assumption that the water demand of each node is always satisfied. In the case of the normal status of the water supply system, the application of DDA is not problematic. However, if the water demand increases rapidly due to fire or insufficient hydraulic pressure of each node because of a pipeline failure, an unrealistic problem, in which negative pressure is calculated, may occur. If the hydraulic pressure of the node decreased, a demand shortage would occur. PDA is the method to calculate node runoff considering the node's hydraulic pressure. Unlike DDA, which calculates node head under an assumption that water demand of the node is always satisfied, the PDA calculates the water supply according to the changes in the hydraulic pressure of the node. Further, the amount of water supply available and head would be obtained at the same time (Mays 2003). In other words, PDA is a method for the calculation of the amount of water available in an actual node in accordance with head-nodal outflow relationship (HOR).

In this study, the CoF was defined as the water suspension quantity. The water suspension quantity was defined as the volume that cannot be supplied to the consumers, and the PDA hydraulic analysis was carried out to calculate the suspension quantity. For the PDA hydraulic analysis, WaterGEMS V8i of Bentley, a commercialized program, was used. The HOR proposed by Wagner et al. (1988) in the form of a power function was used as a basis for the PDA module of the WaterGEMS model. 
formula
2
 
formula
3
 
formula
4

Even if the required amount of water is not supplied during the repairing time, the required water can be supplied for certain times using the holding quantity in a reservoir or water storage. The suspension quantity was calculated by considering ease quantity, i.e., the holding quantity of the reservoir or water storage at the demand break point of the multi-regional water supply pipeline of region K. Table 3 lists the time required for the repair corresponding to the pipe diameter utilized in this study (K-water 2015). At this point, an additional five hours were added for the calculation of the time required to repair a failure, assuming that additional repair time would be required for road excavation if the pipe is laid beneath the road. Table 4 lists the demand of water and holding quantity of the reservoir or water storage.

Table 3

Failure repair time according to pipe diameter

Diameter (mm)<350400450500600700800900
Time (h) 10 11 11 12 13 14 15 16 
Diameter (mm)<350400450500600700800900
Time (h) 10 11 11 12 13 14 15 16 
Table 4

Supply area and holding quantity of reservoir

Water supply areaWater demand (m3/d)Holding quantity of reservoir (m3)Water supply areaWater demand (m3/d)Holding quantity of reservoir (m3)
SD1 area 2,515 750 D industry 3,577 7,000 
SH area 23,929 7,000 D hospital 151 160 
S industry 6,405 6,000 D dormitory 2,782 2,500 
HC area 3,000 1,375 DM area 5,664 2,500 
YJ area 1,032 940 Y apartment 771 1,035 
SW area 1,706 1,140 N apartment 290 585 
SD2 area 6,148 3,113 DB area 5,640 3,330 
AJ area 3,005 2,000 IW area 2,177 1,700 
Water supply areaWater demand (m3/d)Holding quantity of reservoir (m3)Water supply areaWater demand (m3/d)Holding quantity of reservoir (m3)
SD1 area 2,515 750 D industry 3,577 7,000 
SH area 23,929 7,000 D hospital 151 160 
S industry 6,405 6,000 D dormitory 2,782 2,500 
HC area 3,000 1,375 DM area 5,664 2,500 
YJ area 1,032 940 Y apartment 771 1,035 
SW area 1,706 1,140 N apartment 290 585 
SD2 area 6,148 3,113 DB area 5,640 3,330 
AJ area 3,005 2,000 IW area 2,177 1,700 

RESULTS AND DISCUSSION

Result of risk assessment

PoF

The results of the quantification type II analysis conducted in 193 pipelines with two or more failures and 193 ones without failures derived from the systematic sampling method are shown in formula (5). 
formula
5
The hit ratio of the discriminant function was 81.1%, which is favorable. Further, Wilk's lambda was 0.547 and the significant probability of chi-squared was 0.000, which means the discriminant function was statistically significant. The results of the Box's M test indicated that the significant probability of F was 0.132, which satisfied the assumption that the covariance of each group was the same. The coefficient of formula (5) is one of the standardized canonical discriminant functions, of which the size of an absolute value indicates the relative importance between variables. The hit point of formula (5) is Z = 0; the results with a negative value and a positive value belong to low-risk and high-risk groups, respectively. In other words, the risk of failure increases with increase in the number of years after the installation of a pipeline. If epoxy was used as the coating material, pipelines were laid in general land, and the pipe diameter was large, the risk of failure was found to be decreased.
The results of the quantification analysis indicate the qualitative results of groups with low and high risk of failure, respectively. At this point, the tendency of belonging to each group would be increased with the increase in the distance from the hit point, i.e., the increase in absolute value of the discriminant score. The results of the regression analysis between the discriminant scores derived from each pipeline and the failure rates of each pipeline to convert the qualitative result into the quantitative failure rate are shown in formula (6). As a result, formula (7), derived from formulas (5) and (6), was used to estimate the PoF. 
formula
6
 
formula
7
The results of the average failure rates of the 14 systems of region K indicate that the failure rate was highest in system N, followed by systems G and D.

CoF

For the threshold pressure of water demand break point for the PDA analysis, 39.1 m, a hydraulic pressure for consumers indicating the minimum pressure at normal conditions, was applied. If the pressure head of the water demand break point is higher than 39.1 m, the entire water demand can be supplied. In addition, if the pressure head is between 0 and 39.1 m, only parts of the water demand can be supplied by a power function. Further, if the pressure head of the water demand break point is less than 0 m, the required water cannot be supplied.

Table 5 lists the predicted water suspension quantity during the failure repair in case of a failure on each of the 232 multi-regional water supply pipelines of region K, which indicates the sum of suspension quantity at each demand break point. The results of the estimation of suspension quantity indicate that over 10,000 m3 of suspension occurs if a failure occurs in zones A and B. Because suspension quantity is different for failures occurring in time with low and high water consumptions, the average suspension quantity is indicated. The results of the calculation of risk with comprehensive considerations of the PoF and CoF are shown in Table 5. As shown in Table 5, CoF is 0 in systems F, G, H, I, J, K, L, and N, and the risk is also calculated as 0. It is possible to satisfy the water demand during repairing time even if a failure occurs in the water pipelines because the small pipe diameter of the system does not require much time to recover and the reservoir in the system has abundant average water capacity.

Table 5

Result of risk assessment

System IDPipe length (km)PoF (failures/km/yr)CoF (m3/failure)System risk (m3/yr)
5.027 0.0608 17,921 5,477 
6.799 0.0460 11,118 3,477 
8.196 0.0565 2,133 988 
2.601 0.0712 74 14 
5.889 0.0529 430 134 
7.461 0.0529 
4.088 0.0736 
2.702 0.0437 
0.562 0.0230 
3.575 0.0564 
8.287 0.0646 
5.226 0.0658 
4.757 0.0526 264 66 
2.727 0.0797 
System IDPipe length (km)PoF (failures/km/yr)CoF (m3/failure)System risk (m3/yr)
5.027 0.0608 17,921 5,477 
6.799 0.0460 11,118 3,477 
8.196 0.0565 2,133 988 
2.601 0.0712 74 14 
5.889 0.0529 430 134 
7.461 0.0529 
4.088 0.0736 
2.702 0.0437 
0.562 0.0230 
3.575 0.0564 
8.287 0.0646 
5.226 0.0658 
4.757 0.0526 264 66 
2.727 0.0797 

Risk assessment and prioritization of renewal

The PoF and CoF refer to the weighted averages of the calculated values of each section within the pipelines for the length of the pipelines. System risk can be seen as suspension quantity risk for an entire line.

Table 5 lists PoF and CoF of each pipeline section, and Figure 2 illustrates four zones classified according to the type of risk. The first tolerable zone indicates that, although failure occurs, 80% of daily average water consumption can be supplied. The case where the daily average water can be supplied at 80% of the quantity was judged as a tolerable risk, and a separate zone was set.
Figure 2

Separated risk zone.

Figure 2

Separated risk zone.

In the risk map of Figure 3, the deeper color of pipeline indicates a higher level of risk, and the light gray color indicates the pipeline without a risk.
Figure 3

Risk map.

Figure 3

Risk map.

The priority of renewal can be set in accordance with the order of the quantitative risk of failure of each pipeline zone. At this point, the pipelines predicted to have no risk and the ones inducing tolerable risk were excluded from the priority. Then, the results of the prioritization of renewal are listed in Table 6.

Table 6

Prioritization of renewal

RankSection IDIndividual risk (m3/yr)RankSection IDIndividual risk (m3/yr)
A12 1,251 15 B20 150 
B3 1,039 16 B19 131 
A11 987 17 B2 121 
A5 795 18 B15 119 
A7 763 19 C1 114 
B21 582 20 A2 92 
A10 566 21 A6 87 
A9 425 22 C2 79 
A3 413 23 B2 73 
10 B12 344 24 C3 66 
11 B17 259 25 C4 62 
12 B14 214 26 C5 57 
13 B22 161 27 C6 56 
14 B7 158 Others Tolerable risk  
RankSection IDIndividual risk (m3/yr)RankSection IDIndividual risk (m3/yr)
A12 1,251 15 B20 150 
B3 1,039 16 B19 131 
A11 987 17 B2 121 
A5 795 18 B15 119 
A7 763 19 C1 114 
B21 582 20 A2 92 
A10 566 21 A6 87 
A9 425 22 C2 79 
A3 413 23 B2 73 
10 B12 344 24 C3 66 
11 B17 259 25 C4 62 
12 B14 214 26 C5 57 
13 B22 161 27 C6 56 
14 B7 158 Others Tolerable risk  

Alternatives for risk mitigation

If pipeline A12, which was estimated to be the top priority for renewal through risk assessment, was replaced with a new one, the PoF would be reduced from 0.0937 breaks/km/yr in the current level to 0.0718 breaks/km/yr, decreasing about 382 m3/yr of risk. Further, if all pipelines in system A were replaced, the entire system risk would be decreased from 5,477 to 2,761 m3/yr. The changes in the type and diameter of pipes, if any, when replacing the existing pipelines with new ones, may cause a hydraulic problem. Therefore, the existing pipelines were replaced on the assumption that new water pipelines should have the same type and diameter as the previous ones.

If the holding quantity is increased by enlarging the size of the reservoir in the current system, the redundancy will be increased as well, and therefore, the risk will be reduced. In pipeline A12, which showed the highest level of risk, the holding quantity of the reservoir that needs to be increased to reduce the risk is as follows. If the holding quantity of the reservoir is as described in Table 7, the risk of failure of any pipeline of multi-regional pipelines of region K was found to be tolerable.

Table 7

Reservoir retention quantity for tolerable risk level

Holding quantity of SD1 reservoir (m3)Holding quantity of SH reservoir (m3)Holding quantity of SD2 reservoir (m3)A12 risk (m3/yr)
Before increase 750 7,000 3,113 1,251 
After increase 2,000 21,000 5,000 49 
Holding quantity of SD1 reservoir (m3)Holding quantity of SH reservoir (m3)Holding quantity of SD2 reservoir (m3)A12 risk (m3/yr)
Before increase 750 7,000 3,113 1,251 
After increase 2,000 21,000 5,000 49 

On the other hand, in the case of the construction of double piping by laying new pipes with the same pipe size in zones A, B, and C, a certain level of water suspension would occur due to the hydraulic pressure in some zones. However, the level of risk would be tolerable. In conclusion, when compared with the replacement of existing pipes with new ones and the increase in the reservoir size, laying of additional pipes with the same sizes as the existing ones in zones A, B, and C would have the largest risk reduction.

For the multi-regional pipeline of K, the utilization of all measures for risk reduction is the most effective way to prevent risk despite the failure on any pipeline. However, because of the high cost of implementation, the most practical measure needs to be utilized.

CONCLUSIONS

In this study, the risk of failure was calculated for the multi-regional water supply pipelines of region K. From the results, the priority of renewal was set for each risk, and the measures for the risk reduction were proposed.

For the PoF, the quantification theory type II converts the qualitative results to the failure rate of a pipeline. The discriminant formula, which was constituted by utilizing the data from the national multi-regional water supply pipelines, resulted in 81.1% hit ratio. Further, the regression equation, which indicates the discriminant score and failure rate, was obtained. The results obtained from the application of the regression formula of the failure rate to the multi-regional water supply pipelines of region K indicated that the zones N, G, and D have the highest failure rates. The CoF was calculated by the amount of water that cannot be supplied in case of failure using the PDA. For this analysis, the threshold pressure was set as a minimum hydraulic pressure of the demand break point for a normal condition. During the repair of a failed pipeline, the results of the PDA and the average holding quantity of the reservoir were considered to calculate suspension quantity. In the case of a failure in zone A of the multi-regional pipeline of region K, the effect of failure was found to be the largest. The risk considering the PoF and CoF comprehensively was found to be the largest in section A12. Meanwhile, the case where 80% of the daily average water consumption can be supplied was defined and expressed as the tolerable risk.

Further, as alternatives to reduce the risk, the replacement of a certain pipeline with a new one, increasing the holding quantity of reservoir until a tolerable level of risk is achieved, and the construction of double piping by laying a new pipeline were proposed. As a result, the construction of double piping by laying a new pipeline was found to be the most effective measure for the risk reduction. In particular, the construction of double piping with the same pipe size as that of the existing pipelines of zone A, B, and C would result in a tolerable risk of failure of the entire region K. Therefore, this measure was inferred to be suitable.

Finally, it would be possible to utilize the results of this study as the basic data for the establishment of an investment plan for the renewal of water supply pipelines and asset management plan.

ACKNOWLEDGEMENT

This study was funded by the Korean Ministry of Environment (MOE) as ‘Public technology program based on Environmental Policy (2014000700001)’.

REFERENCES

REFERENCES
Al-Bahar
J. F.
1988
Risk Management Approach for Construction Project; A Systematic Analytical Approach for Contractors
.
PhD Thesis
,
University of California
,
Berkeley
.
Arai
Y.
Koizumi
A.
Umano
H.
Ashida
H.
Ozaki
M.
Yoshida
E.
2009
Statistical analysis of the corrosion of water distribution pipes under their environmental factors
.
Journal of Japan Society of Civil Engineering and Environment System Research
37
(
37
),
9
17
.
Choi
T. H.
2013
A study on the water supply risk assessment and optimal design for water distribution network
.
PhD Thesis
,
University of Seoul
.
Choi
T. H.
Kim
A. R.
Kim
M. C.
Koo
J. Y.
2014
A study on optimal planning of risk reduction for water suspension in water pipe system using fault tree analysis
.
Journal of Korean Society of Water and Wastewater
28
(
6
),
699
711
.
EPA
2012
Fundamental of asset management session 6–Determine Business Risk (‘Criticality’), http://water.epa.gov/infrastructure/sustain/am_training.cfm (accessed 15 September 2015)
.
Goulter
I. C.
Bouchart
F.
1990
Reliability constrained pipe network model
.
Journal of Hydraulic Engineering
116
(
2
),
221
229
.
Kim
M. C.
Inakazu
T.
Koizumi
A.
Koo
J. Y.
2013
Statistical approach for corrosion prediction under fuzzy soil environment
.
Environmental Engineering Research
18
(
1
),
1
7
.
Kim
M. C.
Inakazu
T.
Koizumi
A.
Koo
J. Y.
2014
A long-term plan for water pipeline rehabilitation considering preventive maintenance
.
Journal of Water Supply: Research and Technology-AQUA
63
(
2
),
131
138
.
Koizumi
A.
Hayashi
H.
Arai
Y.
Inakazu
T.
Tamura
S.
Ashida
H.
2012
A study on the corrosion prediction about surface of water distribution pipes using multivariate analysis
.
Journal of Japan Society of Civil Engineers Ser. G (Environmental Research)
67
(
6
),
135
142
.
Koo
J. Y.
Kim
K. B.
Kim
S. T.
Lee
S. C.
Moon
J. Y.
2014
Analysis of optimal renewal period of multi-regional water supply with economic valuation
.
Korean Society of Water and Wastewater & Korean Society of Water Environment 2015 Autumn Forum materials
,
Seoul
,
Korea
.
Koo
M. S.
2015
A study on strategic risk assessment model for water suspension accident alternative in water supply network
.
Master Thesis
,
University of Seoul
.
K-water
2015
Validation review of optimal renewal period for multi-regional water supply pipelines. Korea Water Resources Development Corporation, Korea.
Mays
L. W.
2003
Water supply systems security, McGraw-Hill, New York
.
Ostfeld
A.
2004
Reliability analysis of water distribution systems
.
Journal of Hydroinformatics
6
(
4
),
281
294
.
Park
S. W.
2007
Decision making system for stable management of waterworks
.
Magazine of Korea Water Resources Association
40
(
2
),
16
21
.
Rahman
S.
Devera
J.
Reynold
J.
2014
Risk assessment model for pipe rehabilitation and replacement in a water distribution system, Pipelines 2014: From Underground to the Forefront of Innovation and Sustainability
,
ASCE
, pp.
1997
2006
.
Wagner
J. M.
Shamir
U.
Marks
D. H.
1988
Water distribution reliability: analytic methods
.
Journal of Water Resources Planning and Management, ASCE
114
(
3
),
253
275
.