Cambridge Water supplies about 72,000 properties in the city and surrounding villages with water from chalk boreholes. Total demand is some 45,000m3 a day, with an additional 20,000m3 being transferred in bulk for supply elsewhere.
The ages of Cambridge Water’s 940km of pipes vary considerably, with some more than 150 years old. The company’s first-ever failure of Ofwat leakage targets prompted the decision to investigate the introduction of district meter areas (DMAs) in order to improve understanding of consumption in each zone for maximum efficiency in tackling the leaks.
EnlargeSetting DMAs up involves dividing the network into small areas, each ideally having a single feed supply. Every area is metered individually so that consumption is identified for that part of the network. 'The sum of all those small parts gives you your total consumption,' explains Cambridge Water Design & Standards Engineer John Brock, who is responsible for all aspects of network modelling.
Building the initial model
An all-mains model had been built and calibrated in 2002 in InfoWorks, using data from the company’s GIS system. 'I was fortunate in that the original digitization was done very well,' says Mr Brock.
Much field data had to be collected to validate the model and Parkman - now Mouchel - was brought in to help. A total of 160 pressure loggers and 16 insertion meters were installed. Consumption by 32 of the larger customers was also logged, representing about 10% of demand. In addition, data was collected from 48 points on the company’s telemetry (SCADA) system. “The model was found to calibrate extremely well,” says Mr Brock. 'There are some very good tools within InfoWorks which we used extensively as we went through the process of calibration.' In particular, he singles out data flagging, which helped the team to keep track of the changes made to the original assumptions. Updating of the model is carried out annually to incorporate new connections and any changes to operating practices. It is used extensively to evaluate enquiries from property developers, and to look at the potential impact of planned maintenance on customers.
Impetus for change
Cambridge Water failed its Ofwat leakage target for the first time in 2004/5. A leakage review was undertaken. 'What this revealed was that only 37% of our supply was covered by district meter areas, and they tended to be the rural areas supplied directly from small towers,' says Mr Brock. The majority of the supply area was still covered by quarterly waste tests. There was only 5% DMA coverage in the city zone.
'Our view on district metering had always been that it could be done fairly easily if necessary in the rural areas and the outer parts of the Cambridge zone,” he says. “But it was always deemed to be too difficult for the inner part, because it is a fully-interconnected network.” There had been a feeling that the network would not cope with the large number of valve closures required to create the district meter areas, but the need to tackle leakage gave the impetus to look afresh at the possibilities.
EnlargeCambridge Water started by using its own resources to implement DMAs around the periphery of the network, where the introduction was relatively straightforward. Mouchel was then invited in April 2006 to explore whether district metering might be a viable option elsewhere. 'The outcome of that study was quite favorable,' says Mr Brock. The consultant was therefore invited to carry out a detailed implementation study for district metering throughout.
The InfoWorks WS network model became the base model for the whole study and all subsequent field trials and pressure loggings have been based on it.
The starting point was to undertake a desktop review of the system, working with Cambridge Water operational staff, explains Mouchel Project Engineer Gareth Mallows. The feasibility study had developed an initial split into 36 DMAs and the boundaries for these then had to be confirmed.
The open part of the system had been geographically split into eleven areas during the feasibility stage, using local geographical features, such as the River Cam, the A14 and the rail network. The majority of these eleven areas were then subdivided into three or four district meter areas to enable manageable field trials to take place. Only the city centre was designated as a unitary DMA.
Each of the areas had to be individually modelled in InfoWorks ahead of field trials. Network models were run for average and peak day scenarios to flag up any apparent low pressures. 'The modelling results were then reviewed to ensure that customers within the proposed DMAs would not experience any issues in terms of level of service,' says Mr Mallows. Cambridge Water’s GIS system was also used to identify any critical information within each DMA, such as major and sensitive customers. The preparations also confirmed which valves would need to be closed to isolate the individual DMAs.
Preparing for testing
Modelling was an iterative process using two models - the original base version without DMAs and the working model, which was broken down into the constituent DMAs. 'The two models were continually compared throughout the study to ensure that the proposed DMAs would not be detrimental to the system,' explains Mr Mallows.
Controls were then set up for each individual area to look at the demand on both peak and average days. For each control, the boundary valves were closed for each DMA area. Modelling was then run with all the proposed meter location valves closed to ensure that the area was discrete. Pressure zero testing could then be undertaken to confirm the isolation from the rest of the network.
'The ‘pipe closed’ facility within InfoWorks was very useful here for enabling rapid representation of a closed valve,' says Mr Mallows. The InfoWorks ‘boundary trace’ tool was also very useful, he adds. 'This enabled us to check that the areas were discrete and that all the boundary valves and proposed meter location valves had been identified.' This tool was used in conjunction with Cambridge Water’s GIS system as a further check on the areas. 'Positive results from both pieces of software gave us confidence that no boundary valves had been omitted from the network model,' says Mr Mallows.
The model was then used to simulate the scenario of the field trials in each DMA. It enabled the pinpointing of the nodes and areas with the lowest pressures - these would be critical monitoring points. Any pressure anomalies could also be identified for logging in the field.
Carrying out the tests
The InfoWorks modelling was also used to create implementation packs for all eleven areas to facilitate the field trials. The packs were issued to the control room and key operational staff.
To prove that each DMA was separate and discrete, each area was closed in and the pressure was checked. Field trials were carried out over 14-day periods to allow assessment of different scenarios. Pressure loggers were installed on the first day, so that existing conditions could be monitored. Boundary valves were then closed on day 6, with the pressure zero test carried out on the seventh day.
A pressure zero test involves closure of all inlet and outlet points to ensure that all feeds have been identified. The first step was to identify and close all boundary valves, generally between 01:00 - 05:00hrs. Customers with special needs were informed, as were large metered consumers. A pressure gauge was set up at a standpipe at a convenient hydrant. The valve at the area’s inlet was then closed to isolate the DMA, and the pressure on the standpipe gauge should then drop.
However, pressures sometimes did not drop straightaway, and might not drop to zero as some residual head may remain in the system. If the pressure did not drop below 5m (the allowable figure deemed to represent residual head), then it was adjudged that not all feeds into the area had been identified.
On completion of the pressure zero test, the supply valve was opened and the pressure gauge was used on a nearby hydrant to ensure that pressure has been restored.
After a pressure zero test was undertaken, the boundary valves would generally be opened six days later and then the following day the pressure loggers were removed.
'The overall correlation between field test data and modelled data was excellent, and proved that Cambridge Water had a very robust model,” says Mr Mallows. “This gave us confidence throughout the study.'
On successful completion of all the pressure zero tests, five areas indicated potential problems with the supply. Additional model runs were undertaken to identify solutions such as mains reinforcement and rehabilitation. Potential pressure problems were also predicted in ten areas at times of peak demand. 'Due to the nature of transfer across the system, these are furthest from the source of supply,” says Mr Mallows. A potential solution was identified, involving opening the boundary valves to adjoining DMAs to create ‘super DMAs’ to maximize supply.
Implementing the proposals
'The study to design district meter areas was completed on time and on budget by Mouchel,' says Mr Brock. An information pack has been produced for each DMA to facilitate the meter installation. Each pack contains an overview plan and schematic, mains information, a property listing, a list of sensitive and major customers, boundary valve and network modelling information, pressure logging results and proposed meter locations.
The meters, valves and instrumentation required to implement the scheme are being installed throughout 2008.
This article is based on a paper presented by Cambridge Water Design & Standards Engineer John Brock and Mouchel Project Engineer Gareth Mallows at the Wallingford Software International User Conference, 2007.