stream with other metadata was critical to
accurately interpreting the data stream.
Data Processing and Analysis for Online
Distribution System Monitoring (2008,
order #91226) provides the following
practical recommendations for online
distribution system monitoring:
• An overall description of the distribution
system is essential.
• Metadata, defined
recorded in a
for each measured
• Off-line sensor should be set to read a
known error state or a specific value that
is not a credible reading.
• Strong quality control procedures should
• A copy of the original recorded data
should be used for pre-processing.
The original recorded data should
not be altered.
• A table of monitored variables should
be produced using standard abbreviated
names for variables and include
the methods, units, and mode of
• For each variable, the acceptable values
should be determined.
• Pre-processing of the data should aim
to have time-order variables.
• An exploration of past data and the
identification of typical problems and
events are important.
• Variables should be measured and
recorded to the greatest number of
significant figures possible.
• Control charts applied to first differences
of each variable can be used to provide
a useful start to identify changes worthy
• Individual water utilities should
determine alarm levels suitable for their
own particular circumstances.
As illustrated in Data Processing and
Analysis for Online Distribution System
Monitoring (2008, order #91226), metadata
are necessary to
interpretation of the
sensor data. A new
for Water Quality Event Detection”
(project #4182), aims to integrate water
quality sensor data with subsets of
metadata, namely hydraulic data and
operation information, so that operators
can effectively identify water quality
abnormalities while minimizing false
alarms. This project will utilize the pattern
matching, outlier detection, and case-based reasoning algorithms developed
for Distribution System Security and
Water Quality Improvement Through
Data Mining (2008, order #3086) to
build an automated real-time monitoring
and decision aide (ARTMADA). A
demonstration version of ARTMADA is
expected to be available along with the
published final report.
Individual water utilities
should determine alarm
levels suitable for their own
Effective online monitoring of distribution
system water quality involves a number
• Reliable monitoring sensors. Currently,
the measurements of flow rate, turbidity,
pH, water temperature, fluoride, chlorine,