Critical link analysis in InfoWorks WS has been combined with an assessment of pipe failure rates to develop an effective asset management regime for the water supply system in the northern Italian province of Ferrara.
Utility company HERA Group is modeling its entire water supply network in InfoWorks WS. The detailed model covers every aspect, from the smallest pipes to the region’s treatment plants, and there are plans for it to be integrated into HERA’s real-time operational decision-making by 2012.
The model has already been used to reduce leakage and to explore the consequences of failure in key pipes in order to prioritize work that will reduce the risk to customers’ supply.
About HERA
HERA Group is a major utility based in Bologna in the north of Italy. It came into being in 2002 through the merger of 11 companies and now provides wide-ranging services across the water cycle as well as in the energy supply and environmental sectors. HERA’s integrated services for the water cycle cover water treatment and distribution, conveyance of the wastewater in its sewers and wastewater treatment. Water is supplied to about a million customers.
Adopting a standardized approach
One of the companies now forming part of HERA had adopted hydraulic modeling in the early 1990s, using a system developed at Bologna University. By 2005, various modeling packages were in use across the Group.
Engineers from HERA Bologna carried out a research exercise to identify the best software to adopt for its water supply and wastewater networks. The technical analysis showed InfoWorks WS and InfoWorks CS to be the best systems. Similar research that was subsequently carried out by the Group’s research and development department confirmed InfoWorks WS and CS as the best products for HERA.
All of the Group’s operating companies have since bought InfoWorks licenses and begun to convert their models into the InfoWorks format.
InfoWorks model
Eleven engineers are focused on model development and about 69% of HERA’s 25,000km of water supply pipes have been modeled to date. A series of targets has been set for model building or conversion, with the aim of covering the Group’s entire water supply system by 2010. Development will continue once the modeling is complete, as the aim is to achieve integration by 2012 with the Regional Network & Plant Remote Control Centre, which is one of the largest in Europe.
The integration will include automatic updating with live data and real-time simulation of incidents such as network breaks or plant failures. HERA aims to use InfoWorks to implement procedures for semi-automated control of its water supply system.
Ferrara’s water supply
Operating company HERA Ferrara manages services for 250,000 people in the western part of Ferrara province. Its system includes two water treatment works, both of which take water from Po River. They are markedly different in size, with one supplying almost 95% of the system’s input volume with an average flow of about 1,000 l/sec (264 gps).
Ferrara’s system also includes 2,200km (1,367 miles) of distribution pipes, 266km (165km) of trunk mains and 17 reservoirs. Two booster stations and three connections to surrounding systems ensure supply to the more remote customers, located up to 80km from the treatment works.
A leakage reduction program has proved successful in bringing the infrastructure leak index down from 12.5 to 10.1, with further improvements planned.
The InfoWorks hydraulic model supports the leakage reduction program, which is based primarily on three approaches, none of which had previously been widely used in Italy. District metering areas (DMAs) have been introduced, which involves dividing the network into small areas whose individual consumption can be monitored. This makes it easier to identify the locations of leaks. Two types of specialist equipment have also been purchased: Permalog leak detectors and state-of-the-art correlators.
Detailed modeling
HERA’s previous model had been skeletonized, whereas an all-pipes model has been built in InfoWorks. The model includes every pipe, valve, hydrant, pump, tank and reservoir within the network. No simplification has been applied: plant modeling incorporates full detail of all aspects such as variable speed pumps, floating valves and control devices. Every control device has been accurately modeled, including the most complex ones.
All 107,000 customer meters have been geo-referenced and linked to supply pipes. The network and data can be kept current, thanks to InfoWorks’ ease of updating from the GIS system. The customer database has been imported from the billing system.
The model has been fully calibrated which is relatively uncommon in Italy. Leakage and demand have been simulated using the pressure-related demand functions within InfoWorks.
Determining leakage
Leakage has been modeled separately from other consumption. Data on bursts and leaks had been collected between 2000 and 2006. This data was geo-referenced by the University of Ferrara, imported into the model and allocated to the appropriate pipes and nodes. In addition, the total level of leakage from the water supply system was distributed based on calculated values.
The base run modeled leakage for a 24 hour simulation and then the pressure-related demand function was activated to consider leakage that was 100% dependent on pressure. Just 20% of domestic consumption was considered to be pressure dependent.
Two equations were used to model domestic consumption and leakage. For domestic consumption, the modelers adopted the default demand equation within InfoWorks. For the leakage equation, the value was based on average night zone pressure with an exponent value of 2.5, which was confirmed by field data.
The most important application of the InfoWorks model to date has been in pressure management, using the leakage thematic view. It has been used successfully at the Pontelagoscuro plant which supplies Montagnone Reservoir. InfoWorks showed that reducing the pressure set point from 3.1 bar to 2.9 bar would reduce losses by almost 50 l/sec (13 gps). The reduction in pressure was carried out in the field and the 50 liter figure was achieved, which confirmed the model’s abilities and will save an estimated €50,000 Euros a year at this location alone.
Critical link analysis
Implementation and calibration of the model of Ferrara’s water supply system was followed by an automated critical asset analysis of the most important pipes in the network. The analysis was made possible by the tools included in InfoWorks WS.
Production of a simple risk index has assisted in prioritizing asset management and replacement.
HERA has developed a practical procedure to calculate the risk associated with particular assets, based on meeting service level criteria. The work makes use of the critical link analysis within InfoWorks which identifies pipes where failure would leave customers isolated. This is combined with the results of statistical and probability analyses to establish which pipes are most likely to fail.
Calculations take account of failure rates for particular pipe diameters and materials. Diameters of less than 60mm have been found to exhibit the highest failure rates and the material with the greatest breakage rate was found to be polyethylene. The information has been used to calculate mean times to failure.
The analysis achieves answers to several key issues. It determines the number of customers who would be isolated or would receive water below the minimum pressure threshold. It can also be used to identify vulnerable customers such as hospitals and dialysis patients, or schools. Another application is to calculate the percentage of demand reduction, and it can also highlight negative network pressures, which can cause an increase in contaminants.
However, criticality analysis is a time-consuming process, requiring considerable hard disk space. After performing tests on a sample area, it was estimated that the whole process would have taken a year for the whole system and would have required some 600 gigabytes of hard disk space.
This led to a decision to consider only the larger diameter pipes - the trunk mains in the diameter range from 400mm to 900mm (15” to 35”) - and pipes that had failed at least once in the 2000-2006 data. This reduced the hard disk requirement to 30 gigabytes and brought the time down to a week.
The work was used to calculate a risk index for each pipe. A straightforward approach was adopted, taking account of the likelihood of failure and its consequences. The consequences of failure were rated according to a criticality index that considered the number of general or key customers who would be isolated by a failure and how many would be supplied with water below the minimum pressure. The pipe with the highest value of the risk index was in the Bondeno municipality. It had a mean time to failure of 149 days and would leave 200 customers isolated, including one key one. About 400 would have a low-pressure supply.
Further analysis and research will be carried out to determine the costs of asset failures and to take account of the costs of repair or replacement.
This article is by Claudio Meoli and Alessio Benini from HERA Group’s research and development department. It is based on a presentation they gave at Wallingford Software’s 2008 International User Conference.