K. THOMPSON ET AL. 77
charge of each WTP or at wholesale connection interties
to indicate baseline conditions entering the distribution
system for comparison with downstream measurements.
These stations also indicate results of changes to the
treatment process and warn of conditions in the treatment
process that may otherwise go undetected. Some OWQM
process measurements may already be made at the WTP,
and these existing measurements can be used for OWQM
purposes. When adding other OWQM parameters that
are not already monitored at the WTP, some utilities use
installation of OWQM stations as an opportunity to up-
grade older instruments or to convert to reagentless
technologies to reduce labor and maintenance costs.
OWQM stations are frequently installed at the dis-
charge of distribution system reservoirs and chlorine
boosting pump stations. Measurement may be upstream
of chlorine addition to provide a measure of the quality
of the water in the reservoir and that further upstream.
The OWQM station could alternately be installed down-
stream of chlorine addition at a booster station to provide
a baseline measurement for comparison to OWQM sta-
tions further downstream.
OWQM stations can also be installed at critical nodes
in the distribution system, and different approaches may
be taken in selecting these locations. Distribution system
managers generally have a good understanding of the
operation of the piping network and can often identify
the nodes of interest based on experience. A more scien-
tific approach to OWQM siting is frequently conducted
through the use of the Threat Ensemble Vulnerability
Analysis and Sensor Placement Optimization Tool (TEVA-
SPOT) software [10]. This analytical package, developed
by EPA, Sandia Laboratories and others, analyzes a dis-
tribution system network and identifies critical nodes that
will represent water quality impacting the largest number
of consumers. TEVA-SPOT analysis results often vali-
date the operational understanding expressed by distribu-
tion system managers, but also frequently identify nodes
otherwise not understood to be critical or recommend
subtle location changes compared to the managers’ rec-
ommendations.
The analysis enables prioritization of a selected num-
ber of OWQM stations to meet the budget available for a
project and identify stations to be added as a monitoring
system expands over a period of years.
6. Requirements for Data Communication
and Analysis
Several products are available for analysis of OWQM data
and alarming of unusual conditions. The most common
software event detection systems (EDSs) are commer-
cially available through s::can, Whitewater, and Hach.
Also available is the Sandia National Laboratories free-
ware system Canary, which was developed as part of the
WS initiative. Each EDS has strengths and weaknesses
associated with its performance under distribution system
operations. Additional information can be obtained through
the EPA Water Security Division.
OWQM alarms are generally based on more than sim-
ple alarm setpoints for parameter measurements. Typi-
cally the alarms or alerts are associated with pattern
alarms, where multiple parameters change in a manner
that is atypical of their normal relationship. As an exam-
ple, if TOC increased, it would also be expected that
DOC would increase in a proportional manner. When a
utility implements enhanced coagulation, the TOC-to-
DOC relationship changes, generating an alert at the wa-
ter quality monitoring stations. Broadband UV/visible
systems also produce a spectral alarm that is initiated if
the normal spectral fingerprint displays an unusual shift.
OWQM data are collected more frequently than typi-
cal Supervisory Control and Data Acquisition (SCADA)
monitoring data, and therefore OWQM data usually
cannot be communicated over traditional low-bandwidth
SCADA networks. Additionally, spectral data cannot be
communicated over the typical SCADA data collection
network, so separate communication pathways must be
established. Many utilities, therefore, include all OWQM
data on the alternate path and keep monitoring and
maintenance of this information separate from opera-
tional SCADA data, although there is no requirement to
maintain this separation. If T-1 or optical connections to
the central monitoring facility are available, these path-
ways may be used, but typically the OWQM measure-
ments are still transmitted as a separate data stream from
SCADA parameters.
Water quality analysis is conducted locally at the OWQM
station, and measured values and alarms are communi-
cated to a central historian and display. Typically a long-
term database is used for storage and retrieval of data and
a short-term cache for short-term (30-day) trending.
For OWQM stations at water utility locations, the data
may be communicated over a virtual private network
(VPN) set up on the existing utility network, if available.
For locations that do not have access to the utility’s net-
work, data are frequently communicated over commer-
cial digital radio or cable service connections.
7. Fabricated On-line Water Quality
Monitoring Stations
Installed OWQM stations take several forms, depending
on the parameters and analyzers selected for use. Out-
door installations are generally fabricated in enclosed
cabinets for protection and security. Because water flows
inside the cabinet with an open drain, ventilation of the
cabinet is required to dissipate moisture that may accu-
mulate. In hot southern climates, temperatures inside the
enclosed cabinets are also of concern, so ventilation is
Copyright © 2013 SciRes. CWEEE