Desalination & Reuse Handbook - Flipbook - Page 84
IDA
WATER SECURITY
HANDBOOK
Acoustic leak detection
The use of noise loggers and correlators is a more rapid and digitised version of the age-old listening stick used by
conventional leak detection teams. Noise loggers producing alarms or using correlation to locate leaks generally record
sound at night when ambient noise is at its lowest, meaning that smaller leaks can be detected. Acoustic technology
involves the following key distinctions.
Alarms vs correlation
Alarms can indicate a burst or leak, but they may also
result from background noise, such as nearby traffic or
construction work. As a result, acoustic measurements
are most effective when taken at night when background
noise is at a minimum. Alarms point to some leaks but
have a relatively high rate of false negatives – reporting
no abnormal data when a leak is in fact present.
Correlation involves comparison of sound levels at
different points in the network at the same point in
time. Probable leak locations are then calculated using
advanced algorithms, which allow for variables such as
distance between sensors, and pipeline material and
age. These systems are at their most accurate when
comprehensive information on the network is available.
Typically, a result is accurate to within 5% of the distance
between two sensors.
Utility server
Alarm
displayed
Utility server
Proposed
leak location
displayed
Cloud server
Advanced algorithms used
to locate leaks by
comparing readings from
logger pairs
Raw data
Alarm
Noise
logger
Noise loggers
Leak
control, giving a utility the ability to take real-time action on realtime data. He likens this automation of remote network control
to the autopilot of an airplane, in which a computer programme
can take over routine decision-making and an operator remains
on standby as a contingency measure. In the meantime, both i2O
and Visenti are seeking to improve the integration of other inputs
such as acoustic correlation into their respective platforms.
Acoustic leak detection
The delivery of leak reports directly to a utility’s SCADA
system comes as standard when using acoustic correlation for
leak detection. While standalone noise loggers can transmit
‘alarms’ to alert operators of potential leaks, the application of
algorithms to ‘correlate’ readings from multiple sensors in real
time can more accurately calculate the location of a suspected
breach and, instead of raw data, deliver a location directly to the
user. The precision of the leak is typically in the region of 5%
of the distance between two loggers, but varies depending on
operating conditions. A fixed-network acoustic leak detection
system compresses and transmits vast amounts of data, which is
analysed between large numbers of logger pairs, either deployed
permanently or (in the case of above-ground listening points)
76
Leak
deployed on a ‘lift and shift’ basis to monitor different areas
of the network for intervals of several months. The result is a
contract that is part hardware supply,and part data management
service. Owing to both the large amounts of processing power
required and the use of proprietary correlation algorithms, data
management services are an inherent part of packages offered
with fixed network correlators. When it comes to contract
structure, technology providers tend to adapt to suit client
preferences.
Correlators can be deployed as a mobile leak detection kit using
laptop- or smartphone-based software to run on-site analytics, or
as a fixed-network system installed onto underground assets or
attached to above-ground listening points such as fire hydrants.
Where a variety of providers offer correlation in its mobile form,
either as a service or a purchasable kit, and a few offer sensors
for license or purchase for fixed-network applications, only
Gutermann offer permanently installed, underground acoustic
network monitoring. Whilst the Swiss company’s 2016 project,
installing and managing over 5,000 acoustic sensors in Lyon, is
one of just a handful of smart water projects on such a scale, their
install-base is in the tens of thousands and growing fast. Head
