Page 155 - Water Loss Control

Modelling W ater Losses 133

meter population, including the average response time to correct accounts that chroni-
cally register zero consumption.
For this type of analysis to be accurate, it is necessary that there is a large enough
test sample of data from field inspections of zero consumption accounts in order to
properly represent the total customer account population.
Improper Size or Type of Meter
Many brands of customer meters are known to become appreciably inaccurate when
very high or very low flows (relative to the design range of the meter) are registered. If
the size or type of meter in a given application results in the majority of flow occurring
in these extreme ranges, then the meter will fail to register a large portion of the cus-
tomer flow. Section 12.4 provides a detailed discussion on meter sizing impacts and the
best practices to employ to ensure that losses due to poor sizing or typing are mini-
mized. With direct-feed pressure systems as are typical in North American water utili-
ties, customer meters need to be selected and sized to record a wide range of flow rates.
Any underregistration of metered consumption is considered an apparent loss in the
water audit, as the lost water is reaching the customer, but a portion of the consumption
is not being registered or billed. A number of software models have been developed for
this type of loss analysis.
Similar modeling techniques can be applied for the apparent loss components of
data transfer error, systematic data handling error in customer billing systems, and
unauthorized consumption. Detailed spreadsheet models for these components are not
as common as those modeling customer meter inaccuracy; however, it is up to the water
auditor to assess the occurrences of these losses and attempt to model their extent in
their utility operations.

10.4 Modeling Components of Real Losses Using Breaks
and Background Estimates Concepts 1
In the early 1990s, during the U.K. National Leakage Control Initiative, a systematic
approach to modeling components of real losses (leakage and overflows) was devel-
oped by Allan Lambert.
Recognizing that the annual volume of real losses is the result of numerous leakage
events, each individual volume loss being influenced by flow rate and duration, Lam-
bert considered leakage events in three categories:

• Background (undetectable) leakage: Small flow rate, runs continuously
• Reported breaks: High flow rate, relatively short duration
• Unreported breaks: Moderate flow rates, duration depends on intervention
policy
For each separate component of the distribution system—mains, service reservoirs,
service connections (main to curb stop), service connections (curb stop to meter) —the
value for each component of annual losses can be calculated using the parameters in
Table 10.6 below for some given standard pressure. The effect of operating at different
pressures can then be modeled by applying FAVAD principles to each of the individual
components of real losses, using appropriate specific N1 values. FAVAD is discussed in
more detail in Sec. 10.6.3 of this chapter.