Keywords: GIS model, leakage control, water main rehabilitation, repair or replace,
In the region of Drammen – Norway, increased focus on leakage control on potable water mains has been initiated, and a number of activities to reduce the current water losses are taken. One is the establishment of a new GIS based tool for selecting water mains in distribution networks that are in need of rehabilitation. The new tool is based on multi criteria methods for comparing water mains, using several input parameters. Using this tool the municipal engineers get a tool that shifts the way of selecting water mains for rehabilitation from being based on selecting water mains due to the number of leakages or to follow the rehabilitation of other infrastructure to a more proactive approach based on a number of criteria such as leakage frequency, materials and importance of the water mains.
The Quality Water Project in the region of Drammen (the GVD project) is based on cooperation between nine municipalities in the region of Drammen in southern Norway.
The focus area of cooperation is water and sewerage projects where there are mutual interests beneficial for the participating municipalities
The cooperation itself is unique in a Norwegian context (Moen et al, 2008)
The region of Drammen has about 170.000 inhabitants. 150.000 have public water supply.
Figure 1: The region of Drammen (GVD) – Norway
The total water production is about 24 million m3 of water/year and there are about 1500 km of public water mains. The water production equalises to 400 l per capita per day.
The project is a result of a common master plan for water supply in the region, initiated by the municipalities and Glitre Water Works in 2004 (The municipalities of the Drammen region , 2004). The major topic to be addressed in the period covered by the master plan is reduced water demand and increased leakage control.
As indicated in Figure 2, the master plan forecasts that the capacity of the water source will be exceeded in 2035 if the increase in population continues, and no further actions are taken to reduce the water loss and the demand for water.
Figure 2:Expected growth in water production, with expected growth in population and no action is taken to reduce the water losses and the demand for water. (Master Plan 2005) Investigations indicate that the amount of water losses in the distribution network of the region amounts to about 50 % of the total production.
In order to avoid costly investments in new water resources, with all the provisions and restrictions that might follow from this, on the use of the resource and the neighbouring land, it was decided to reduce the water losses and the water demand.
The target to reach is: “Stabilising water demand on the level of year 2004”. To accomplice this, two targets have to be reached: 1) A reduction in water loss from 50 % in 2004 to ≤ 30 % in 2020. 2) All customers must have water tariffs based on meter reading of water consumption no later that 2013. (Currently about 60 % of the customers have water meter)
The general strategy for water loss control follows the four basic methods recommended by IWA (Farly & Trow, 2003). 1: Active leakage control. 2: Infrastructure management. 3. Speed and Quality of repair. 4: Pressure management.
The work so far has primarily focused on active leakage control. 83 district metered areas (DMAs) are established using more than 100 district water meters (whereof about 25 are new). Two leakage control teams are established, using the latest technologies in noise loggers, leakage noise equipments and correlators.
I order to also focus on infrastructure management, especially the rehabilitation of water mains, a new project was started: How to select the right water mains for rehabilitation?
During the last decades, selecting water mains for rehabilitation traditionally has followed either of two approaches. 1) Rehabilitate water mains when another infrastructure needs rehabilitation. For instance sewers, road work etc. 2) Rehabilitate water mains when the number of leakages makes it obvious that rehabilitation can not be avoided. Following these strategies have resulted in too low renewal rate and thus an overall deterioration of the water distribution network, and an increase in the number of leakages and water loss.
Therefore, the GVD project decided to develop a tool that can support the engineers in selecting water mains for rehabilitation in a more sofisticated way, taking into account not only the leakage frequency, but also the probability for a new leakage to occur and the consequence of a leakage.
Project description: The objective of the project was to develop a tool, that based on various parameters could point out the water mains in the distribution network which should be given priority for rehabilitation in water mains rehabilitation plans.
An important target for the project was that it should be feasible and easy to implement. The tool should, as far as possible be based on easily available data. Further data acquisition or digging for hardly available information in old archives should be limited to a minimum.
The software tool should focus primarily on the water distribution network, leaving out coordination with sewer, road works or other infrastructure initiatives.
It was decided that the tool should be tightly connected to the GIS system, as this is the database from where most information easily can be withdrawn.
The time frame for the project was:
June – November 2007 Preliminary study
Spring 2007: Tendering for acquisition of IT / civil eng. consultant
June 2008: Project start
September 2008: Prof of concept model
November 2008: First version model
January 2009: Sensitivity test and test on pilot zone
February 2009: Final project report and acceptance.
GIS as the tool
As already mentioned, it was decided to use GIS as the foundation for the development of the tool. As part of the GVD project a common GIS based asset management system has been established.
Multi criteria selection: In the tool a kind of multi criteria selection is used. Multi criteria selection is not a new invention, neither in general nor in this particular field. The Care-W project running under the fifth framework program of the European Commission caused a thorough discussion around the multi criteria techniques (Le Gauffre, Laffréchine et al, 2002) and a survey of multi criteria techniques for use in water distribution network rehabilitation (Le Gauffre, Bauer et al, 2002).
The Care-W project takes into account a long range of criteria, as for instance: rehabilitation costs, coordination, disturbances induced by rehabilitation, water quality, water loss, water supply interruptions and more. As an important objective for our project was to maintain focus on feasibility, our software tool is less ambitious and is therefore more limited in the number of included criteria.
The selection criteria used, are classified in two groups. One group is containing criteria that define the probability for failure on water mains or fittings, and a second describing the consequence of a failure.
For each of the criteria in the two groups and for all the water mains in the distribution network, a score value between 1 and 6 is applied. Each of the parameters is then multiplied with a weight factor, and the result is used to calculate two values, The Probability Score (PS) and the Consequence Score (CS). The Probability Score and the Consequence Score thus multiplied with a weight factors and the result are used to calculate the final Rehabilitation Factor (RhF). As can be seen, this method resembles how risk evaluations often are calculated:
Risk = Probability * Consequence
Probability factor Leakage indicator (PLI) In today’s practise leakage is the single most important factor, when selecting which water mains are selected for rehabilitation. This should still be true, and therefore much labour was used to develop the leakage indicator. Just counting the number of leakage repairs on a water main section is too simple, and will lead to wrong decisions.
Information on leakage frequency is varying among the municipalities, except for the last 2 years where registration of leakages has been good due to the establishment of active leakage control in the region. However, for the pilot area registration of leakages has been done for the last 20 years.
The leakage indictor includes both the leakage frequency and the absolute number of leakages according to the formula:
LN : Number of leakages on the water main section. [0 ... 4]
If LN > 4 => PLI = 6. (In this case the water main should always be rehabilitated)
LF : Leakage frequency of the water main section, normalised t: no./100 m/10 year.
The leakage frequency is calculated for 3 periods:
The lifetime of the pipe (LF)
The lifetime – last 5 years (LF1)
The last 5 years (LF2)
If LF2 > LF1 Then FY= 1.5 else FY= 1
Table 1a: Score table for leakage frequency, normalised to: no./100 meter /10 year
From [no./100 m/10 year]
To [no./100 m/10 year ]
Age in it self is not considered to cause an increase in probability for leakages to occur. Age is anyway an important factor for the model and it must be determined for all water mains in the distribution network. In the software tool, age either serve as secondary information in order to classify another parameter (as is the case for the Pipe Material parameter), or is used indirectly to estimate an otherwise undetermined parameter (as is the case for the Excavation Method)
About 1/3 of the water mains in the pilot area lack information on age, therefore this information had to be established. This was done by gathering the public work veterans from the municipalities around a map. Thereby it was possible to determine with sufficient certainty the age of water mains for which this information was lacking.
Pipe material (PM) Pipe material is considered to be another important probability parameter for identifying water mains for rehabilitation. This parameter uses age as a secondary parameter to give an indirect indication on type of joints and corrosion protection.
For municipal water mains, information on material is generally available, while information on type of joints and corrosion protection are missing.
Table 2: Score table for material
1) When the project started the type of material for some of the water mains were unknown, but during the project a discussion and assumption was made, and at the end of the project, the material type was established for all water mains in the pilot area.
Excavation method (PEM) When the use of excavator was introduced, the handling of the water mains became less careful, resulting in increased damage to the pipe and the pipe protection. New directions and improved pipe protection have reduced this problem, and it is believed that this problem is minor today. Generally information about excavation method is not registered in the GIS system or any other easily available source. Age is therefore used as an indirect method to determine the method of trenching that had been used.
Table 3: Score table for excavation method
Digging by hand
Digging by excavator
Digging by excavator
Digging by excavator
Placement method (PPM) It is generally considered that placement method is an important parameter in forecasting the probability for leakage. Unfortunately very little information is registered over time about placement method. Furthermore age is not a good indirect parameter, because varying techniques have been used during most of the times. However, for a single and important t
Figure 3: Pipes supported by wooden planks
echnique, some information may be drawn from the information of cause for the leakages, namely the use of wooden planks to support the pipe-ends before backfilling the trench.
Table 4:Score table forPlacement method
Pipes supported by wooden planks
It is clear that further development of this parameter may improve the model. Indicating where wooden planks have been used would improve the model considerably. The use of wooden planks has two serious consequences: a) Source of organic material that promote SRB corrosion under anaerobic soil conditions, b) When the wood deteriorates and rots the pipe foundation is destroyed and the risk of pipe fractures increase.
These parameters are used to define the importance of a water main section. It is clear that the more important a water main is with regard to how many people that are served, the type of customers, the possibility for rerouting the water supply etc. the sooner the water main should be rehabilitated.
Pressure, flow and flow direction It is the intension that the selection tool should use hydraulic modelling for the calculation of the consequence of bursts on pressure and water flow to customers. Unfortunately the hydraulic model is not yet fully established (this is an ongoing project), and therefore it has not been possible to get reliable data on pressure and flow to customers affected by broken water mains. In the current version of the tool, the hydraulic model have only been used to calculate how many people have interruptions in their water supply caused by pipe bursts.
Number of people loosing their water supply due to pipe burst (CP) Table 5: Score table for number of people losing water supply in case of pipe bursts.
People from [no.]
People to [no.]
Supply to vulnerable customers. (CVCF) Vulnerable customers are defined as customers to whom continuous water supply is crucial. Vulnerable customers are divided in 3 categories
Category1: Very vulnerable customers (Hospitals etc.)
Given the score tables, the rehabilitation factor for each water main is calculated according to the following formulas:
Calculating probability score (PS):
Table 7: Default weights for calculating Probability Score (PS)
Pipe material weight
Calculating consequence score (CS):
Table 8: Default weights for calculating Consequence Score (CS)
Vulnurable Customer Factor (VCF)
Calculating rehabilitation factor (RhF)
Table 9: Default weights for Rehabilitation factor (RhF)
Results and experiences: The pilot area The area “Åssiden” in the city of Drammen was selected as pilot area for testing the water main selection tool. As shown on the map in Figure 4,the water mains were installed in the period for 1950 to 1980. It is known that there are problems in this area with both the water mains and the sewers. This opens up for coordination when the rehabilitation programme is planned.
Figure 4: Pilot area for the selection tool. “Åssiden” In the city of Drammen”
Most pipes are installed between 1950 and 1980.
The result The result for the calculation is a map Figure 5 with water mains in colours representing the Rehabilitation Factor (RhF) calculated for each water main by the selecting tool and. The results can also be found in a table that in addition includes more details about individual parameter scores.
The rehabilitation factor fall in 5 classes:
1: Very low Rehabilitation is not needed
3: Middle Rehabilitation should be included in long term planning
5: Very High Rehabilitation should seriously be considered on a short term view
Figure 5: Pilot area with water mains coloured according to the calculated Rehabilitation factor. Red circles show areas were the selection tool has identified water mains with a high Rehabilitation Factor (Indicating that the water main should be rehabilitated in a short horizon)
A detailed view Figure 6 on the content of the larger red circle in Figure 5, show an area with 5 water mains with an RhF larger than 4, and therefore candidates for rehabilitation.
Figure 6: Detailed view from the finale calculation. It shows 5 water mains with an RhF score ≥ 4.
The score only indicates which water mains that should be target for rehabilitation on a short time horizon. It’s up to the municipal engineers to decide whether or not the water main should be subject to complete replacement (1) or to point repair (2). The municipal engineer must also decide on the method of rehabilitation (Traditional or No dig) The calculated rehabilitation factor do not by itself dictate which water main that is the first candidate for rehabilitation, but the rehabilitation factor can be used by the municipalities as an indicator when elaboration rehabilitation project plans. Whether or not a water main should be rehabilitated is still subject to coordination with other infrastructure projects in the same area, as for instance sewer projects or road projects. If no coordination is possible, investigation of the water main should be carried out to reveal whether a “no-dig” method for the rehabilitation is the right choice.
The dispersion of leakages on the water main can reveal whether the water main should be replaced in full, as for instance the water main in circle 1 in Figure 6, or be point repaired as could be the case for the water main in circle 2.
The tool has proved to be a help for the engineers in the water sector. The selection of water mains for rehabilitation has moved from a pure defensive strategy (selecting the water mains with the highest number of leakages) to a slightly more proactive strategy where also water mains with otherwise high probability for a leakage and strategic importance are selected for rehabilitation.
The input is based on readly available information. The only improvement needed in data quality, was to ensure that all water mains had an “age” attribute set, a task that at first might look difficult, but actually showed to be possible through the veterans “workshop”.
Thought we have established a tool that can be used today, plans exists for future improvements, such as:
Extending the model to cover the full area of the region of Drammen. This will need a quality increase of the GIS system, in order to ensure that the water main criteria data, such as age, type of material and dimension are available for all mains in the water distribution network. This will give us more experience with the use of the tool, and ensure that we over time will find the right parameters.
Improve the input from hydraulic models. In the current version of the tool, hydraulic modelling is used only to calculate the direction of flow. Calculation of the consequence of water main failures on pressure and flow might add new insight about the consequence factor which then can be used with greater confidence than it is the case today.
Improve the quality of the placement parameter. Collecting information about how water mains are actually placed in the trench may improve the predictions on where the next leakage will occur. For instance the mappings of pipes where wooden planks are used as pipe foundation are important. Also, information about the backfill material used might be important.
Long term improvements:
Another improvement which can be done in the future is to coordinate with other utilities as for instance sewer, traffic load, soil conditions, importance of the road etc. This could be done by developing a similar tool for the sewer mains, for roads etc. and then put the result from the different utilities together for a complete rehabilitation program. If then a sewer main is present, and the water main needs rehabilitation, it might be more practical to use a traditional excavation method. If no sewer exists or the sewer is not subject for rehabilitation one might consider using a no-dig method for the rehabilitation of the water main.
Farly, M., Trow, S. (2003) Losses in Water Distribution Networks. IWA Publishing. ISBN 1 900222 11 6,
Le Gauffre P., Laffréchine K., Baur R., Di Federico V., Eisenbeis P., König A., Kowalski M., Sægrov S., Torterotot J.P., Tuhovcak L., Werey C., (2002) Care-W: WP3 – Decision support for annual rehabilitation programmes. D6 – Criteria for the prioritisation of rehabilitation projects. Care-W ( Computer Aided Rehabilitation of Water networks), EU project under the 5th Framework Program, contract no. EVK1-CT-2000-00053. Lyon (F): INSA-URGC, June 2002, 72 pages.
Le Gauffre P., Baur R., Laffréchine K., Miramond, M., (2002) Care-W: WP3 – Decision support for annual rehabilitation programmes. D7 – Survey of multi-criteria techniques and selection of relevant procedures. Care-W ( Computer Aided Rehabilitation of Water networks), EU project under the 5th Framework Program, contract no. EVK1-CT-2000-00053. Lyon (F): INSA-URGC, June 2002, 30 pages.
Moen, A.D., Dupont, R. A., Skaret, J.E., Røren, T., (2008) Practical Regional cooperation in the Water Supply sector between 9 Local Municipalities/Councils. Paper presented to the 6th Nordic Water Supply Conference in Oslo june 2008.
The Municipalities of the Region Drammen (2005). Quality Water - The Master plan for the region of Drammen (2005-2020). 42 pages, (In Norwegian)