In this paper, the risk of water supply systems as a combination of hydraulic analyses (consequence) and failure probability of pipelines based on the actual condition is discussed. On the one hand, a stepwise closing of pipelines calculates the influence on customer's supply which represents the importance of a pipe. On the other hand, based on existing failure statistics, age and material distribution, a failure probability of a pipe segment is calculated. The two factors represent a risk matrix which is crucial for optimising the system.

failure probability, hydraulic analysis, risk management, system optimisation

According to national standards, water utilities have to guarantee the supply of water to their consumers in appropriate quality, quantity and pressure. Therefore, combining of information about net service, net condition and additional assets is the basis for developing a customized strategic asset management and derivate an operative asset management.

Supply guarantee, supply quality and supply reliability are the cornerstones of a professional asset management. Periodic or condition based inspection, information from monitoring systems and the documentation of occurred failures and the implemented rehabilitation measures form the basis for the development of a state-of-the-art strategy. It is essential to combine the experience of the on-site operational staff and the ‘theoretically’ available data. For a supply system the information of the hydraulic performance implemented in a software based hydraulic model should be state for the art.

The actual hydraulic performance of a supply system can be analysed by a hydraulic net-evaluation for several loading cases. A calculation of a defined hydraulic load case (night flow, firefighting, … ) calibrated by on-site measurements will lead to significant statements for closed or semi-closed valves, sufficient supply pressure or hygienic problems according to raised hydraulic residence time.

According to the German standard DVGW W 400-1 (2004) a minimum supply pressure of 2.0 bar is required at the customers connection. In case of firefighting a minimum supply pressure of 1.5 bar is temporary tolerable.

To set up a hydraulic model with software basic information about the network should be available. Beside a plan of a map including diameter of pipes, installations like valves, booster stations, reservoirs also information about elevation head, customer consumption and special bulk consumer are necessary. Additional information of the existing water resources or information about planned housing should also be taken into consideration.

According to the experience of the author a main step in a hydraulic analysis is the calibration of the network. At a network calibration, pressure loggers a spread over the system and a pressure reduction caused by a defined measured withdrawal are recorded (Figure 1). According to the German standard DVGW GW 303-1 (2006) this pressure reduction should be at least 1.5 bar or 20% of the static pressure. For a detailed calibration a factor of 1 pressure per km is necessary.
Figure 1
Figure 1. Sample of a result of a network calibration (Gangl et al. 2012).
View largeDownload slide

Sample of a result of a network calibration (Gangl et al. 2012).

Figure 1
Figure 1. Sample of a result of a network calibration (Gangl et al. 2012).
View largeDownload slide

Sample of a result of a network calibration (Gangl et al. 2012).

Close modal

Figure 1 shows a deviation between the calculated and recorded pressure in a single pressure logger. This deviation might be the consequence of a wrong pipe diameter, a closed or semi-closed valve or a wrong pipe connection. In the last years RBS wave as an engineering company was responsible for more than 60 hydraulic network calibrations of German water supply system per year. There was no supply system which was free of relevant network failures which had an influence of the hydraulic capacity.

In an urban mashed supply network a singular pipe break may have no or less influence on the customers supply, in some cases the pressure will drop marginal. In other cases when the single transport pipe to the network will have a burst the whole are cannot be supplied.

When a calibrated hydraulic model is available the customer consumption will lead to a specific flow of water inside the pipe network. By a theoretical stepwise closing of pipes the influence on customers can be calculated weather the customer is supplied with a pressure below a defined minimum value or is not supplied. According to the percentage of influence of the single pipe with respect to the total supply volume the importance or impact on the system is defined.

Figure 2 shows the result of a relevance-calculation of a supply system with a net length of 130 km. As a boundary condition a minimum pressure for customers of 1.0 bar was defined. The relevance of pipes is set into relation to values where nodes with customer consumption are below 1.0 bar pressure, to nodes below 1.0 bar pressure but still with supply and nodes without any supply caused by closing of the pipe.
Figure 2
Figure 2. Result of a relevance-calculation of a water supply network.
View largeDownload slide

Result of a relevance-calculation of a water supply network.

Figure 2
Figure 2. Result of a relevance-calculation of a water supply network.
View largeDownload slide

Result of a relevance-calculation of a water supply network.

Close modal

The red pipes have a high relevance, followed by the blue and green pipes. For the rest of the network a break of a single pipe has no or only less influence on the pressure at the customers side. For these pipes also a repair-strategy is possible as the customers are not influenced.

The ongoing deterioration of the net and the change of the net condition caused by several influencing factors have to be merged and evaluated in a risk management system. An aging function which is influenced by the main factors like year of construction, material, diameter should reflect the utility-specific situation.

To describe this deterioration process, some well-known theoretical probability distribution functions can be fitted to the recorded failure data by statistical analysis (Gangl et al. 2007). Hence, a failure or risk probability for each defined pipe group depending on the ongoing age can be calculated. It is relevant that the utility-specific failure situation will be represented by an aging function. When a standard aging model is used which is not fitted to the actual situation a difference in the rehabilitation need will strain the condition of the network and the necessary investment budget. Figure 3 shows on the left side a fitted aging function to the recorded failure data of a water supply utility for old steel pipes without a sufficient corrosion protection. In the German regulation, for this type of material an aging function is published with obvious longer life time expectancy. The right side of Figure 3 shows the difference of the rehabilitation need by using the literature lifetime expectancy and the fitted aging function. Based on 300€/m for rehabilitation of a pipe the calculation lead to a difference of 7 Mio € for the investment budget!
Figure 3
Figure 3. Fitted age function to failures and difference in investment need to literature function.
View largeDownload slide

Fitted age function to failures and difference in investment need to literature function.

Figure 3
Figure 3. Fitted age function to failures and difference in investment need to literature function.
View largeDownload slide

Fitted age function to failures and difference in investment need to literature function.

Close modal
A pipe segment will move along the defined aging function from a good condition to a time, where the requirements for a further use are not fulfilled any more. This lifetime expectancy is influenced by several factors. For a utility with high production costs the economic lifetime with respect to the volume of lost water caused by a failure or a leaking pipe connection will be earlier than for a utility with a sand soil, water with a high quality without treatment and no pumping costs. Depending on these influencing factors a water supply utility should set up diverse rehabilitation strategy depending on importance and failure probability (Figure 4).
Figure 4
Figure 4. Risk matrix based on importance and condition of pipes.
View largeDownload slide

Risk matrix based on importance and condition of pipes.

Figure 4
Figure 4. Risk matrix based on importance and condition of pipes.
View largeDownload slide

Risk matrix based on importance and condition of pipes.

Close modal

To set up a risk management for a water supply utility the definition of the considered area and the strict definition of the boundary is relevant. A risk management can have the focus on the resources (contamination of wells or springs … ), on the transport pipes and the consequence of a failure of this sensitive infrastructure or the pipe network itself. In the actual paper the risk management is focused on the supply network itself. For large water supply utilities like the water utility of the city of Stuttgart, the responsibilities for resources, transport and storage, and water supply is separated into three departments. It is for sure, that an overall strategy has to take into consideration the results of all three analysed fields.

For a City in Baden-Württemberg, Germany a relevance-calculation and a rehabilitation concept was evaluated. The analyses for the hydraulic calculation of the relevance of a pipe segment were done by the restriction of a minimum pressure of 1.5 bar at each node in the network. This is the minimum allowed pressure in case of fire-fighting in Germany; under normal condition a minimum pressure of 2.0 bar and additional 0.35 bar per every level in a house is required (DVGW W 400-1 2004). For the network also a rehabilitation strategy with fitted aging functions to several pipe groups was elaborated. The database for recorded failure data allowed an optimal fitting of mathematical functions to the available failure data. In that case, the failure probability of each pipe segment as well as the hydraulic relevance of each pipe segment was available.

Figure 5 shows the result of the relevance calculation as well as the failure risk calculation of each pipe segment of a water supply network with a length of 174 km. In the left picture the red coloured pipes represent pipes which lead to pressure of less than 1.5 bar when they are out of order. The volume of water in nodes with less than 1.5 bar is more than 1.0 m3/h. For the blue pipes the volume of water is more than 0.1 m3/h. Under normal condition, the pink coloured nodes have an average flow pressure of more than 4.0 bar pressure.
Figure 5
Figure 5. Relevance and risk based evaluation of pipe segments.
View largeDownload slide

Relevance and risk based evaluation of pipe segments.

Figure 5
Figure 5. Relevance and risk based evaluation of pipe segments.
View largeDownload slide

Relevance and risk based evaluation of pipe segments.

Close modal

The right side of Figure 5 shows the result of the risk based analysis with respect to an aging model. The red coloured pips segments show a high failure risk. The actual recorded failure data are also marked in the plan.

The black dotted ellipse in both maps represent an area with a high failure risk as well as a high relevance of pipes which will cause a pressure drop below 1.5 bar when the pipe segments have to be out of order in case of failure.

The method described is a valid approach to optimise water supply systems based on a risk-management approach. An overall strategy for a rehabilitation of a network should include as well the result of a hydraulic analysis and a failure based analysis. The presented case study of a city in Germany shows the overlapping of pipes with a high failure probability which have also a high relevance for the supply situation in the network. A rehabilitation strategy has to focus especially on those pipes. In a mashed supply structure the influence of a single pipe on the hydraulic situation is difficult to estimate – a software based analysis supports this analysis. The combination of the result of the relevance calculation and a failure probability can result in a better definition of an investment budget and a priority ranking for rehabilitation of pipe segments.

DVGW W 400-1
2004
Technische Regeln Wasserverteilungsanlagen, Teil 1
,
www.dvgw.de
.
DVGW GW 303-1
2006
Berechnung von Gas - und Wasserrohrnetzen - Teil 1: Hydraulische Grundlagen, Netzmodellisierung und Berechnung, http://www.dvgw.de
.
Gangl
G.
Fuchs-Hanusch
D.
Stadlober
E.
Kauch
P.
2007
Analysis of the failure behaviour of drinking water pipelines
.
Water Science & Technology: Water Supply
7
(
5–6
),
219
–
225
.
Google Scholar
Crossref
Search ADS
 
Gangl
G.
Kober
E.
Fischer
J.
2012
Virtual DMA – a practical example for sustainable water loss reduction. blue facts, 2012, 102–108
.
© IWA Publishing 2017