SERA
United States
Environmental Protection
Agency
Design! ng Customer Complaint Surveillance
For Water Quality Surveillance and Response Systems
Office of Water (MC 140)
EPA 817-B-17-002
November 2017
-------
Disclaimer
The Water Security Division of the Office of Ground Water and Drinking Water has reviewed and
approved this document for publication. This document does not impose legally binding requirements on
any party. The information in this document is intended solely to recommend or suggest and does not
imply any requirements. Neither the U.S. Government nor any of its employees, contractors or their
employees make any warranty, expressed or implied, or assumes any legal liability or responsibility for
any third party's use of any information, product or process discussed in this document, or represent that
its use by such party would not infringe on privately owned rights. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
Questions concerning this document should be addressed to WSD-Qutreach@epa.gov or one of the
following contacts:
Nelson Mix
U.S. EPA Water Security Division
MC 4608T, 1200 Pennsylvania Ave, NW
Washington, DC 20460
(202)564-7951
Mix.Nelson@epa.gov
or
Steve Allgeier
U.S. EPA Water Security Division
26 West Martin Luther King Drive
Mail Code 140
Cincinnati, OH 45268
(513) 569-7131
Allgeier.Steve@epa.gov
-------
Acknowledgements
The document was developed by EPA's Water Security Division, with additional support provided under
EPA contract EP-C-15-012.
Nelson Mix, EPA, Water Security Division
Adam Bucher, CSRA
Adam Haas, CSRA
Amy Posner, CSRA
Peer review of this document was provided by the following individuals:
Steve Allgeier, EPA, Water Security Division
Yves Mikol, New York City Department of Environmental Protection
Connie Schrepal, Mohawk Valley Water Authority
David Travers, EPA, Water Security Division
-------
Table of Contents
List of Figures i
List of Tables
Abbreviations >
Section 1: Introduction
Section 2: Overview of CCS Design
2.1 CCS Approach
2.2 CCS Design Elements, Design Goals, and Performance Objectives
Section 3: Complaint Collection
3.1 Communicating Water Quality Concerns
3.1.1 Leveraging the Consumer Confidence Report
3.1.2 Communicating with Customers through Direct Marketing
3.1.3 Implementing a Media Campaign
3.2 Consolidating Water Quality Complaints
3.2.1 Implementing Procedures to Manually Route Complaints to a Single System
3.2.2 Establishing a Single Phone Number for Reporting Complaints 1
3.2.3 Automatically Forwarding Complaints to One Location within a Utility 1
3.2.4 Electronically Integrating Systems to Capture all Complaints 1
Section 4: Information Management and Analysis 1
4.1 Complaint Categories 1
4.1.1 Capturing Descriptive Data from Free Text 1
4.1.2 Establishing Complaint Categories 1
4.2 Alert Generation 1
4.2.1 Establishing Thresholds for Water Quality Complaints 1
4.2.2 Methods for Generating Alerts 1
4.2.3 Spatial Clustering Analysis Techniques 1
4.3 Alert Notifications 1
4.3.1 Notifications through Direct Messaging 1
4.3.2 Notifications through a Routinely Monitored System 1
Section 5: Alert Investigation Procedure 2
5.1 Developing an Effective Alert Investigation Procedure 2
5.1.1 Defining Potential Alert Causes 2
5.1.2 Establishing an Alert Investigation Process 2
5.1.3 Roles and Responsibilities 2
5.2 Developing Investigation Tools 2
5.2.1 A lert Investigation Checklists 2
5.2.2 Record of Alert Investigations 2
5.2.3 Quick Reference Guides 2
5.3 Preparing for Real-time Alert Investigations 2
5.3.1 Training 2
5.3.2 Preliminary Operation 2
5.3.3 Real-time Operation 2
Section 6: Preliminary CCS Design 2
Resources 3
References 3
Glossary 3
Appendix A: Spatial Clustering Analysis Techniques 3
-------
List of Figures
Figure 1-1. Surveillance and Response System Components 1
Figure 2-1. Funnel, Filter, and Focus Approach to Customer Complaint Surveillance 4
Figure 3 -1. Process Summary for Funneling Customer Complaints to a Central Location 9
Figure 4-1. Example Complaint Category Tiers 14
Figure 4-2. Example Dashboard Displaying a CCS Alert 20
Figure 5-1. Example CCS Alert Investigation Process 23
Figure 5-2. Example of Alert Investigation Records 25
Figure A-l. Hydraulically Connected Mega Pressure Zones within the DWU Distribution System 40
iv
-------
List of Tables
Table 2-1. Design Elements for Customer Complaint Surveillance 5
Table 2-2. Common CCS Design Goals 5
Table 2-3. Example CCS Performance Objectives 6
Table 3-1. Design Sub-elements for Complaint Collection 7
Table 3-2. Potential IT Systems to Integrate into a CCS Information Management System 10
Table 4-1. Design Sub-elements for Information Management and Analysis 12
Table 4-2. Target Attributes of CCS Data 17
Table 5-1. Common Causes of CCS Alerts 22
Table 5-2. Example of Roles and Responsibilities for CCS Alert Investigations 24
v
-------
Abbreviations
AET
Alarm Estimation Tool
AWWA
American Water Works Association
CCR
Customer Confidence Report
CCS
Customer Complaint Surveillance
CSR
Customer Service Representative
DWU
Dallas Water Utilities
EPA
United States Environmental Protection Agency
GCWW
Greater Cincinnati Water Works
GIS
Geographic Information System
IT
Information Technology
IVR
Interactive Voice Response
MMS
Multimedia Messaging Service
O&M
Operation and Maintenance
SCADA
Supervisory Control and Data Acquisition
TAT
Threshold Analysis Tool
SFPUC
San Francisco Public Utilities Commission
SMS
Short Messaging Service
SRS
Water Quality Surveillance and Response System
-------
Designing Customer Complaint Surveillance
Section 1: Introduction
The United States Environmental Protection Agency (EPA) designed a Water Quality Surveillance and
Response System1 (SRS) that employs multiple components to detect water quality anomalies that could
have potential public health and economic consequences. Figure 1-1 shows the components of an SRS
grouped into two operational phases, surveillance and response. Procedures guide the systematic
investigation of anomalies detected by the surveillance components in order to identify causes of
contamination. If distribution system contamination is detected, response plans guide actions intended to
minimize consequences, and ultimately, to return the distribution system to normal operations. EPA
intends the design of an SRS to be flexible and adaptable based on each utility's goals and the resources
available to support implementation and operation of the system.
Online Water
Quality Monitoring
Consequence
Management
TaKe corrective
action it necessary,
men resum e routine
surveillance.
Can distribution
system contamination
tie ruled out?
It unusual water
quality is detected,
an alert is generated
and investigated.
A
Surveillance
Enhanced Seciaity
Monltortng
Customer Complaint
Surveillance
Sampling and
Analysis
Public Health
Surveillance
Figure 1-1. Surveillance and Response System Components
The purpose of this document is to provide guidance for designing the Customer Complaint Surveillance
(CCS) component of an SRS. It is written for drinking water professionals who are involved with the
planning, design, or operation of CCS. Hie document is organized into the following major sections:
Section 2 provides an overview of CCS and a description of the design elements that define the
component. Guidance on developing each design element is presented in subsequent sections.
Section 2 also introduces the concepts of design goals and performance objectives and explains
how they inform the design of CCS.
Section 3 provides guidance on topics related to complaint collection and the sub-elements that
define this design element. Complaint collection determines how the water utility educates
customers about appropriate methods of contacting the utility, identifies systems to receive water
quality complaints, and ensures that complaints are directed to a central location for analysis.
Section 4 provides guidance on topics related to information management and analysis and the
sub-elements that define this design element. Information management and analysis determines
how the water utility systematically tracks water quality-related complaints from initial receipt to
closure.
^ Words in bold italic font are terms defined in the Glossary at the end of this document.
1
-------
Designing Customer Complaint Surveillance
Section 5 provides guidance on investigating CCS alerts. It describes attributes of an effective
alert investigation procedure, explains utility roles in a CCS alert investigation, describes tools to
support the investigation, and provides guidance on investigating alerts in real-time.
Section 6 describes the process for developing a preliminary design for the CCS component of an
SRS.
Resources presents a comprehensive list of documents, tools, and other resources cited in this
document, including a summary and link to each resource.
References presents a comprehensive list of published literature cited within the document.
Glossary presents definitions of terms used in this document, which are indicated by bold italic
font at first use in the body of the document.
This document is written in a modular format in which the guidance provided on a specific topic is largely
self-contained, allowing the reader to skip sections that may not be applicable to their approach to CCS or
that include capabilities that have already been implemented. Furthermore, this document was written to
provide a set of core guidance principles that are sufficient to design the CCS component, while pointing
the reader to additional technical resources useful for a specific design task.
2
-------
Designing Customer Complaint Surveillance
Section 2: Overview of CCS Design
CCS relies on customer feedback for effective monitoring of distribution system water quality. Customers
throughout a utility's distribution system can provide near real-time input regarding changes in the taste,
odor, or appearance of drinking water. In fact, water quality complaints may provide one of the earliest
warnings of a water quality incident if an effective system is in place to detect anomalous trends in water
quality complaints. CCS provides additional benefits to utilities such as enhancing customer service.
2.1 CCS Approach
CCS is based on a "Funnel, Filter, and Focus" approach that allows a utility to separate water quality
complaints that could indicate a water quality incident from other types of customer contact. The
approach is described below and illustrated in Figure 2-1:
Funnel: All customer contact should be directed to a central location.
Filter: Utility employees who routinely handle calls, such as customer service representatives
(CSRs), should be able to respond to billing and meter reading concerns, as well as general water
quality complaints related to benign issues, such as degassing, subtle pressure changes, or
temporarily switching from chloramine to free chlorine.
Focus: Based on complaint descriptions, more serious water quality complaints should be
forwarded to water quality personnel, who will gather in-depth information from the customer
and make a determination if sampling is needed.
3
-------
Designing Customer Complaint Surveillance
PP
H
Customer Service Representative
CUSTOMER CONTACT THROUGH:
Utility Call Centers
Other Utility Departments
311
Emafl Messages
Social Media
Water Quality
Complaint Related
to Benign Issue
Degassing
Pressure
Water Quality Staff
Water Quality
Complaint Could
Indicate
Contamination
Figure 2-1. Funnel, Filter, and Focus Approach to Customer Complaint Surveillance
4
-------
Designing Customer Complaint Surveillance
2.2 CCS Design Elements, Design Goals, and Performance Objectives
Design elements are the functional areas which comprise each component of an SRS. CCS consists of
three design elements, which are described in Table 2-1.
Table 2-1. Design Elements for Customer Complaint Surveillance
Design Element
Description
Complaint Collection
Program to educate customers on communicating water quality concerns and
an established system to funnel complaints from all datastreams into a
central location.
Information Management and
Analysis
Established complaint categories and thresholds for complaints that
generate alerts and notify designated personnel when thresholds are
exceeded.
Alert Investigation Procedure
Documented procedure for the timely and systematic investigation of CCS
alerts, with clearly defined roles and responsibilities for each step of the
process.
An effective CCS component should have capabilities for each of the design elements listed in Table 2-1.
Sections 3 through 5 of this document define a target capability for each of these design elements, which,
if achieved, will result in a fully functional CCS component. However, the implementation of each design
element can vary for different utilities, and it is possible to substantially improve CCS capabilities
without fully achieving the target capability for each design element. Likewise, utilities can implement
CCS in a manner that exceeds the target capability.
The decision regarding how to implement each of these design elements and build CCS is informed by
design goals, which are the specific benefits a utility hopes to realize through implementation of an SRS.
Table 2-2 presents examples of common design goals for CCS.
Table 2-2. Common CCS Design Goals
CCS Design Goal
Description
Detect contamination incidents
Provide an early indicator of water contamination incidents that alter
the aesthetic characteristics of the water and may impact the health of
customers or damage utility infrastructure.
Monitor the impact of operational
changes noticed by customers
Detect a change in water quality resulting from operational changes by
monitoring for changes in the nature or volume of customer water quality
complaints.
Increase the level of customer
service, through improved
communications
CCS provides early detection of aesthetic degradation of water quality,
resulting in situational awareness that can be shared with other
customers. Early detection promotes faster resolution of problems and
increases customer service by reducing the number of customers
affected, and the duration of the problem.
Improve response to water quality
complaints
Develop procedures to streamline and standardize a utility's decision-
making process for investigating water quality complaints, thereby
reducing the utility response time.
Additional factors to consider when designing CCS are performance objectives, which are metrics used
to gauge how well the SRS or its components meet the established design goals. While specific
performance objectives must be developed in the context of a utility's unique design goals, general
performance objectives for an SRS are defined in the Water Quality Surveillance and Response System
Primer. CCS performance objectives and recommended targets are described in Table 2-3.
5
-------
Designing Customer Complaint Surveillance
Table 2-3. Example CCS Performance Objectives
CCS Performance
Objectives
Description
Recommended Target
Spatial coverage
The percentage of the distribution system service area
monitored by CCS, which is dependent on customer
awareness of how to contact their utility and communicate
water quality concerns
100% of the distribution
system service area
Timeliness of
detection
The time between when a water quality complaint is received
and when a CCS alert is generated, which is dependent on
how quickly water quality complaint data is available for
analysis and how often the analysis is performed
15 minutes or less
Operational
reliability
The percentage of time that equipment, personnel, and other
support functions are available to support collection and
analysis of CCS data and the investigation of a CCS alert
Availability of surveillance
capabilities and coverage of
CCS procedures 24/7/365
Alert occurrence
The ability to reliably indicate water quality incidents (through
generation of a valid alert) with a minimum number of
invalid alerts
95% confidence that an alert
is related to a water quality
incident
Sustainability
The ability to maintain and operate CCS using available
resources, which is dependent on the benefits derived from
the component relative to the costs to maintain it
All aspects of CCS are
incorporated into routine
utility operations within one
year of transitioning to real-
time operation
The design goals and performance objectives provide the basis for designing an effective CCS
component, within any constraints. Sections 3 through 5 present guidance on potential approaches to
enhance capabilities for each of the three CCS design elements described in Table 2-1. Additional
background on the design elements, design goals, and performance objectives for CCS can be found in
the Customer Complaint Surveillance Primer.
6
-------
Designing Customer Complaint Surveillance
Section 3: Complaint Collection
CCS requires that customers know how to communicate complaints to their utility and that all complaints
are captured and funneled into one location, regardless of how they are received. This improves data
completeness and the reliability of CCS for detecting water quality incidents. The Complaint Collection
design element contains two design sub-elements, which are described in Table 3-1.
Table 3-1. Design Sub-elements for Complaint Collection
Design Sub-element
Description
Communicating Water Quality
Concerns
Ensure customers understand how to communicate all water quality concerns to
their drinking water utility.
Consolidating Water Quality
Complaints
Funnel all complaints to a central location to facilitate analysis of water quality
complaint data.
Considerations for implementing the Complaint Collection design element are described in the following
subsections:
Subsection 3.1 provides guidance on communicating water quality concerns.
Subsection 3.2 provides guidance on consolidating water quality complaints.
3.1 Communicating Water Quality Concerns
This design sub-element addresses educating customers on how to communicate water quality concerns
effectively with their utility. Information on how to contact the utility can be provided through the annual
Consumer Confidence Report, email messages, bill inserts, or a media campaign and should be displayed
prominently on the utility website home page. Methods of communication should include all applicable
utility contact information customers can use to register a water quality complaint, such as utility phone
numbers, email addresses, links to web forms, links to instant message services, numbers for short
messaging services (SMS), and social media accounts. Utilities should coordinate with a public
communications specialist to ensure that any information provided to customers is non-technical and easy
to understand.
Target Capability
Customers are fully aware of how to communicate water quality concerns to their utility.
3.1.1 Leveraging the Consumer Confidence Report
Consider placing utility contact information in a prominent area of the Consumer Confidence Report and
clearly explaining how to report water quality issues. Also, consider providing examples of the types of
information that customers should provide in a water quality complaint, such as a description of the taste,
odor, and appearance of the water. Additionally, provide tips on how to recognize and report suspicious
activity.
Example Language for a Consumer Confidence Report (CCR)
provides examples that can be modified and included in a CCR to
instruct customers how to report water quality concerns. The
document can be opened in Word by clicking the icon in the callout
box.
| J
This document includes
example language for
inclusion in a CCR.
7
-------
Designing Customer Complaint Surveillance
3.1.2 Communicating with Customers through Direct Marketing
Utilities can communicate with customers through direct marketing with email or bill inserts. When
utilizing direct marketing, the frequency of customer contact should be carefully considered, as frequent
messaging can result in communication fatigue. The ease and relatively low cost of email allows for
regular contact with all customers that provide an email address to a utility. A current email distribution
list may be generated from customer billing information. A utility should inform customers of the
potential for mass electronic communication when they enroll in electronic billing or create online
accounts, as an email message may be ignored if it looks unfamiliar, non-urgent, or is automatically
routed to a spam folder.
The Email Message Template can be used as a guide for crafting
communications to customers about reporting water quality
concerns. The template can be opened in Word by clicking the
icon in the callout box.
ฆ I J ^
I m j
This document includes an
example email message
communication.
For customers who do not enroll in electronic billing, a utility can provide inserts, such as flyers or
promotional materials, in their billing statements. Bill inserts reach customers each billing cycle and allow
a utility to describe important issues or announce specific utility programs. Flyers should be easily
distinguishable from the billing material. Promotional materials, such as refrigerator magnets that
advertise utility contact information, can be distributed with bill inserts or sent in a separate mailing.
These physical materials can also be delivered to customers who enroll in electronic billing in addition to
emails, for greater impact.
The Generic Bill Insert Template provides an example of the
text that can be included in a bill to instruct customers about
how to report water quality concerns. The template can be
opened in PowerPoint by clicking the icon in the callout box.
23
This document includes
example bill insert text.
3.1.3 Implementing a Media Campaign
Implementing a media campaign involves the use of radio, television, billboards, internet, and social
media to promote widespread customer education and awareness of reporting water quality concerns. A
media campaign may require a significant expenditure but can be an extremely effective way to ensure
that customers are knowledgeable about how to report water quality concerns to their utility.
Utilities should develop customized materials and
use different media during the campaign. The
campaign should target all geographic areas,
customer demographics, and prevalent languages
used in the service area. An important
consideration is the timing of communications,
such as non-business hours for television
advertising or rush hour for radio. In a city with a
large university, many customers may be out of
town during the summer months when school is not
in session. Billboards may be most useful if located
in high volume areas, such as city centers or major
traffic arteries.
Media Campaign Case Study
San Francisco Public Utilities Commission
(SFPUC) implemented a media campaign to
encourage customers to utilize the new 311
citywide call center to report water quality
issues. The campaign increased customer
awareness while improving data collection,
detection, and response to possible water
quality issues in the distribution system. Details
of the SFPUC media campaign can be found in
Summary of Implementation Approaches and
Lessons Learned from the Water Security
Initiative Contamination Warning System Pilots.
8
-------
Designing Customer Complaint Surveillance
3.2 Consolidating Water Quality Complaints
All customer contact should be routed to one point
of contact within a utility and integrated into a
consistent format, as shown in Figure 3-1.
Technology has evolved with the growth of the
internet and emergence of social media, and call
centers have increasingly been superseded by
contact centers. In addition to traditional phone
services, contact centers allow customers to
communicate through email, web forms, instant
message, text message, and social media. These
forms of complaints are CCS datastreams that
should be used to detect water quality issues. As
the number of complaint datastreams increase, the
chances of complaints getting trapped in data silos
increases. If water quality complaints are received
by multiple, disconnected sources, such as a 311
city call center, utility field crews, or a utility
contact center, then a utility could be unaware of
total complaint volumes.
Organizations that might receive water quality
complaints should have procedures in place to
direct all water quality complaints to the correct
point of contact. If water quality complaints are
managed outside of the utility, priority should be
given to funneling water quality complaints to the
utility in a timely manner. Funneling complaints
to a central location ensures that no water quality
complaints are lost or overlooked and facilitates
data analysis. Utilities can use a CCS Information
Management System that is either a manual
tracking system, such as a notebook, or an automated system that digitally and consistently formats all
information. Methods for consolidating complaints are described in the following subsections.
Target Capability
All water quality complaints, regardless of the format or location where they are received, are collected and
integrated into a single CCS Information Management System.
3.2.1 Implementing Procedures to Manually Route Complaints to a Single System
All personnel that receive complaints, such as the 311 call center, other city departments, or utility field
crews, should be trained to route complaints to a single location. Modifying existing practices is a cost-
effective way to improve water quality complaint funneling and help standardize information.
Additionally, a utility may prioritize water quality complaints over other customer issues, ensuring water
quality complaints are quickly routed to a CCS Information Management System and available for
analysis.
COMPLAINTS SUBMITTED
BY CUSTOMER
0
PHONE INSTANT MESSAGES
EMAIL TEXT MESSAGES
WEB FORMS SOCIAL MEDIA
COMPLAINTS RECEIVED BY
311
FIELD CREWS
UTILITY CONTACT CENTER
r>
i
CAPTURED IN A
CONSISTENT FORMAT BY
CCS INFORMATION
MANAGEMENT SYSTEM
Figure 3-1. Summary of Process to Funnel
Customer Complaints to a Central Location
9
-------
Designing Customer Complaint Surveillance
3.2.2 Establishing a Single Phone Number for Reporting Complaints
Utilities should consolidate phone numbers and phase out those not frequently used. This may involve
updating contact information distributed to the public, such as information listed in bill inserts, reports, or
on the utility's website. Consolidating phone numbers will reduce confusion, response time, and training
needs. The single phone number should be widely publicized.
3.2.3 Automatically Forwarding Complaints to One Location within a Utility
A utility can use automated call distributors to forward calls from internal and external sources to one
system. A utility can also use an electronic routing process for complaints. Using existing methods of
communication and information systems eliminates the need to educate utility personnel on new
procedures, since customers can continue to use all existing methods of communication. Electronic
routing processes can also prioritize water quality complaints. Storing water quality complaint records in
one CCS Information Management System allows for automated analysis of the datastreams.
3.2.4 Electronically Integrating Systems to Capture all Complaints
Electronic integration involves development of a CCS Information Management System to merge
complaint records from phone calls, email messages, web forms, instant messages, text messages, SMS,
social media, work orders from a work management system, and other datastreams. For example, a utility
could design forms with data fields identical to fields in other utility applications for CSRs or field crews
to complete during customer interactions. These forms can then be automatically uploaded into the CCS
Information Management System and seamlessly integrated with data from other datastreams. Integrating
these datastreams will allow for easier analysis and better surveillance of water contamination incidents.
Integrating datastreams using information technology (IT) may involve implementation costs; however,
these costs can be minimized by implementing the integration of electronic records during routine life
cycle upgrades for the system. Table 3-2 lists different IT systems that can be leveraged for CCS
information management. These systems may support multiple CCS datastreams, which allows the
technical requirements, coding language, alerting algorithms, and operating systems to be streamlined.
These systems may also share a common user interface.
Table 3-2. Potential IT Systems to Integrate into a CCS Information Management System
IT System(s)
Description
CCS Datastreams
Interactive Voice
Response (IVR) System
Facilitates customer communication with the utility through
phone calls and messages. Data from this system can be one
of the earliest records of a complaint. CSRs can also use this
system to provide information to customers about ongoing
concerns.
Phone calls
Web Form
Submissions, Utility
Email Inbox, or Social
Media Accounts
Receives water quality complaints. Manual triage or
automated searching and sorting may occur.
Email message, web
forms, instant
messages, text
messages, and social
media
Customer Information
System/Customer
Relation Management
System/Meter Data
Management System
Tracks customer billing information, complaint information,
customer history, or frequency of reported issues. CSRs or
water quality staff review complaint information stored in
these systems to identify if the complaint could be indicative
of a water quality incident.
Data fields imported
from information
system or input by
CSRs
Work Management
System/Asset
Management System
Generates work orders and may be used to investigate or
address complaints received from the customer contact
center or from field crews. CSRs, field crews, and other utility
staff may track responses to a confirmed customer issue.
Work orders
10
-------
Designing Customer Complaint Surveillance
The Water Quality Complaint Processing Form Template
includes an example form that CSRs or other utility personnel
can use to capture data about a water quality complaint and
track datastreams associated with the complaint. The document
can be opened in Word by clicking the icon in the callout box.
The document also contains a checklist for CSRs to document threats and intrusions reported by
customers. The Proposed Customer Feedback Checksheet (Whelton et al., 2007) is another example of
how complaint data could be captured.
CCS Is A Century Old
Checksheets are consistent with "Complaint Books" that have tracked customer complaints for over 100 years.
Assigning the complaint a serial number, plotting the complaint on a map, and following up with laboratory
analysis as needed has been standard practice for over a century (Mercer, 2017).
This document includes an
example form to capture
information on complaints
and suspicious activity.
11
-------
Designing Customer Complaint Surveillance
Section 4: Information Management and Analysis
Once customer water quality complaints have been captured, the data needs to be stored, displayed, and
analyzed to determine if there is a water quality incident. Data analysis methods should be capable of
detecting an anomaly, such as unusually high numbers or spatial clustering of water quality complaints
with similar problem descriptions. CCS anomaly detection systems are a vital part of the CCS
Information Management System and usually operate continuously behind the scenes. The Information
Management and Analysis design element contains three design sub-elements listed in Table 4-1.
Table 4-1. Design Sub-elements for Information Management and Analysis
Design Sub-element
Description
Complaint Categories
Capture descriptive data from complaints using free text or pre-defined
categories.
Alert Generation
Establish thresholds and generate an alert when water quality complaints exceed
a threshold value indicative of a possible water quality incident. Processes used
to generate alerts may also include spatial analysis of water quality complaint
data.
Alert Notifications
Implement a system to notify designated utility personnel when a CCS alert is
generated.
Considerations for development of the CCS Information Management and Analysis Design Element are
described in the following subsections:
Subsection 4.1 provides guidance on developing complaint categories.
Subsection 4.2 provides guidance on alert generation.
Subsection 4.3 provides guidance on alert notifications.
4.1 Complaint Categories
Customers contact a utility for a variety of reasons, from questions about billing to concerns about water
quality. Complaints also arise from common distribution issues that are usually benign, such as degassing,
pressure changes, or turbidity caused by flushing or water main breaks. Water quality complaint
categories can be useful for quickly identifying the nature of the issue (e.g., cloudy water, rusty water,
musty odor) and allow for a more precise analysis of water quality complaints within each category.
Utilities may have pre-defined categories for complaints that customers can self-select through an
automated menu prompt or that utility personnel can identify after communicating with the customer.
Categories can also be determined from descriptive data if free text is used to capture water quality
complaints from customers or CSRs.
Target Capability
Categories are established to track customer water quality complaints and descriptive information about each
complaint is captured.
4.1.1 Capturing Descriptive Data from Free Text
Utilities can design their CCS Information Management System to receive information as free text,
entered either by customers through a web form or by CSRs during customer calls. If a utility is using an
automated system for analysis, all complaint records must be readable by the anomaly detection system.
Data entry fields can collect descriptive information about the taste, odor, or appearance of the water. An
investigator will look for similar problem descriptors in the free text to determine whether the complaints
are related.
12
-------
Designing Customer Complaint Surveillance
Allowing customers to provide information as free text has advantages and disadvantages. Some
customers will be descriptive and provide important information to their utility, but long entries could
include information that is incorrect or difficult for an automated system to interpret. Descriptive data
should be categorized using a structured methodology or a library of terms to prevent the grouping of
unrelated water quality complaints. The taxonomy can aid in the classification and prioritization of
complaints and can provide clues to the sources of potential contamination.
Utilities should train CSRs to capture information in free text when fielding customer calls, so the
information will be compatible with any CCS Information Management System taxonomy. Utilities
should consider allowing CSRs or water quality chemists to modify or validate water quality complaints
submitted directly by customers through email messages, web forms, instant messaging, text messaging,
SMS, or social media.
4.1.2 Establishing Complaint Categories
Utilities should identify complaint categories that correlate
with common water quality issues and other reasons that
customers may contact a utility. Customers or CSRs should
select a specific category in a data entry field when submitting
a complaint. This system of categorization separates calls
related to billing, main breaks, or water quality concerns. The
categories must be transparent to the customer, and technical
terms should be avoided. For example, a category related to
turbidity issues could be listed as "rusty" or "brown" water.
The water quality complaint categories can be grouped into
different tiers or priority levels based on potential severity.
Each tier should have an established threshold level that
generates an alert if exceeded. Figure 4-1 provides an example
of a tiered approach for water quality complaint categories,
adapted from EPA SRS program pilot utilities, with Tier 1 representing the highest priority complaints
and lowest threshold. Utilities should determine the most appropriate order of the tiers and whether
subcategories are necessary for their own automated systems.
r Complaint Category Case
Study
Greater Cincinnati Waterworks
(GCWW) added a sub-menu to the
phone system's IVR water quality
menu selection to filter rusty or
cloudy water and water pressure
issues from taste, odor, and
appearance complaints. Previously,
GCWW analyzed a general water
quality IVR menu prompt, which
captured many water quality
complaints that were not related to
contamination.
13
-------
Designing Customer Complaint Surveillance
TIER 1
ILLNESS
TIER 2
TASTE OR ODOR
TIER 3
DIRTY OR DISCOLORED
TIER 4
UNUSUAL APPEARANCE, PARTICLES, OILY/GREASY
Figure 4-1. Example Complaint Category Tiers
For additional information and guidance on defining water quality complaints, as well as short- and long-
term utility actions, see the following American Waterworks Association (AWWA) resources:
Taste and Odor: An Operator's Toolbox - Video that provides methods for determining the
causes of taste and odor in potable water and treatment responses.
Diagnosing Taste and Odor Problems Field Guide - Field guide that provides all the information
water utilities need to diagnose and remove objectionable tastes and odors at the tap.
Early Warning and Management of Surface Water Taste-and-Odor Events - Report that focuses
on strategies to head off taste and odor problems before they can cause complaints, offering a set
of practical guidelines and tools that any utility can use.
Taste at the Tap: A Consumer's Guide to Tap Water Flavor - Informative booklet that assists
utilities in helping their customers understand what can cause tastes and odors in tap water and
why tastes and odors do not necessarily indicate a problem with the water.
14
-------
Designing Customer Complaint Surveillance
4.2 Alert Generation
CCS alerts are generated when one or more thresholds are exceeded. Alerts can be generated through a
manual review of complaint records or automated algorithms that calculate the frequency of similar water
quality complaints in real time. Alerts may also factor in thresholds based on the spatial clustering of
complaints. Ultimately, the goal is to identify patterns in water quality complaints that indicate a
significant change in water quality.
Target Capability
A process is in place to detect volumes of water quality complaints that exceed threshold levels derived from
historical data. This capability may be further enhanced through spatial analysis to detect high complaint volumes
within defined spatial areas.
4.2.1 Establishing Thresholds for Water Quality Complaints
Establishing appropriate thresholds for water quality complaints is a key step in implementing CCS, since
alerts will only be generated when complaint levels rise above the threshold. Any threshold exceedance
could indicate poor water quality in the distribution system. Establishing thresholds based on historical
data can be more accurate and robust than basing thresholds on knowledge of utility personnel. For
efficient analysis, historical data should be captured electronically in a format conducive for analysis,
such as an Excel spreadsheet. Thresholds generated from historical data can be validated through
employee experience and optimized after implementation.
An important tradeoff exists when determining a threshold value. Each CCS alert must be investigated,
which requires time and effort by utility personnel, resulting in a cost to the utility. A threshold that is set
too low will result in a high number of invalid alerts that may be caused by the day-to-day variation in
complaint frequency. A low threshold is also more likely to detect real water quality problems. A
threshold that is set too high can result in not detecting true water quality problems that do not generate
alerts and are missed. Utilities need to balance a minimum frequency of invalid alerts against the ability
of CCS to detect real water quality incidents.
Tools to Help Set Thresholds
EPA tools assist utilities in analyzing historic water quality data to establish a baseline for water quality complaint
data that can inform the development of threshold values. The tools apply a configurable scan algorithm to
historical data input by the user, and output when an alert would have been generated. These tools allow the user
to easily:
Analyze the performance of different thresholds using historical complaint data
Re-assess thresholds by analyzing new complaint data collected during real-time operation of CCS
Compare CCS alert occurrence during real-time operation to estimated CCS alert occurrence
EPA's Alarm Estimation Tool (AET) is a simple Excel spreadsheet with macros that generate alerts based on
historical complaint data. This data must be properly formatted and input either by hand or copied and pasted into
the spreadsheet. The alerts can be viewed in a list or on a time-series plot for visual interpretation. For users
unfamiliar with statistical methods, the AET provides a straightforward approach to establishing thresholds using
historical data with the visual interpretation of alerts.
EPA's Threshold Analysis Tool (TAT) is an application that analyzes a variety of data file formats, including .txt,
.csv, .xls, .xlsx, and .tab. The TAT offers three different statistical methods (percentile, standard deviation, and
recurrence intervals) to determine thresholds for CCS. Users can also set thresholds for temporal or spatial units
(e.g., week days, weekends, treatment source, pressure zone, etc.), if the complaint data contains those details.
15
-------
Designing Customer Complaint Surveillance
Three common statistical methods can be used for establishing thresholds: percentile, standard deviation
from the mean, and recurrence interval. Any of these methods can generate acceptable thresholds, and the
choice of method can be based on a user's familiarity or comfort with a specific method.
The AET and TAT use prospective scan statistic algorithms to determine alert occurrence in historical
CCS data. A prospective scan statistic allows for a comparison of the number of customer complaints
received by the utility within a rolling time window, or the "scan window," to a preset threshold. The scan
window is typically between one and seven days. These algorithms can be implemented using either a
reset or continuous mode. In reset mode the algorithm begins counting from zero after any alert and
ignores preceding data when an alert occurs. In continuous mode the algorithm will not report a new alert
until the number of complaints within the scan window falls below the threshold. The reset algorithm will
issue multiple alerts if complaint volume remains high for an extended time; whereas, the continuous
algorithm will only issue one alert. Multiple alerts can signal that the issue still persists, but may also be a
nuisance. In most cases, reset and continuous modes will generate a similar number of alerts, except when
the threshold is low, in which case reset mode could generate many more alerts compared with
continuous mode.
Alerting logic may need to consider historical complaint patterns, such as cyclical trends in customer
complaints. For example, a utility could use a daily parameter to establish thresholds for weekdays and
use two-day, seven-day, and monthly parameters to establish weekend, weekly, and monthly thresholds. It
is important to consider how customer activities may differ during weekends, holidays, and other events
when establishing thresholds. Additionally, parameters for the algorithms should reflect a utility's
distribution system characteristics, such as service areas and residence times. For example, if the
residence time in the distribution system is typically less than four days, it may not be necessary to have
an algorithm that uses a seven-day rolling time window.
Regularly Review and Update Thresholds
Because the water quality complaint baseline may shift over time, thresholds should be periodically re-
evaluated. A shift in the complaint baseline can occur due to changes in business processes, technology,
or utility operations. Unless thresholds are adjusted to accommodate the shift in baseline, CCS may
generate too many alerts or may not alert when there is a water quality incident. Utilities should re-
evaluate thresholds following any activity that can impact call volume or the geographic distribution of
water quality complaints, such as:
Adding new complaint categories to data entry record logs
Media campaigns that promote a single utility phone number
Updates to, or replacement of, software that processes complaint data
Integrating new IT systems and datastreams into a CCS Information Management System
Extension of the distribution system into a new geographical area
An annual review of thresholds is optimal, but if thresholds are re-evaluated every two years, a utility
would be able to capture most gradual changes in normal complaint volume.
4.2.2 Methods for Generating Alerts
Alerts are generated by a CCS Information Management System that uses algorithms, most commonly the
prospective scan statistic, to detect when complaints exceed thresholds. The algorithms used to evaluate
thresholds in the AET and TAT may also be used to generate alerts during real-time data analysis. The
most effective methods for generating alerts utilize an automated anomaly detection system that promptly
16
-------
Designing Customer Complaint Surveillance
notifies personnel when complaint thresholds are exceeded. Alternatively, alerts can be generated
following a manual review of complaint data. Both of these methods are described in more detail below.
CCS data can have attributes related to descriptive information, temporal information, spatial information,
and timeliness. Table 4-2 describes different target levels for each attribute.
Table 4-2. Target Attributes of CCS Data
Good
Better
Target
Descriptive
Information
Descriptive text
Complaint categories
Water quality complaint
categories with detailed
descriptions
Temporal Information
Date of complaint
Date and time of complaint
Date, time, and distinction
between business/non-
business hours or day of the
week
Spatial Information
Zip code of complaint
Hydraulic area of complaint
Precise location
Timeliness
24 hours from complaint
< 24 hours from complaint
< 15 minutes from complaint
The attributes of the data to be analyzed impact the types of anomaly detection systems that can be used.
For example, if the data does not contain spatial information or if there are no Geographic Information
System (GIS) capabilities, maps or spatial statistical models cannot be used. However, all data must have
at least the date, preferably the date and time, available for analysis.
Conducting Regular Manual Reviews of Water Quality Complaint Data
If resources are not available to implement an automated anomaly detection system, a utility can establish
a manual process to review water quality complaints. A periodic, manual review of existing records that
track water quality complaints can be conducted. Many contact center and work management applications
have the ability to generate reports on complaint volume and activity. Alternatively, a utility may use
simpler recordkeeping methods, such as paper call logs, a spreadsheet, or a whiteboard. Regardless of the
form of these records, designated utility personnel, such as a contact center manager or a shift manager,
can conduct regular reviews of the records to detect anomalous volumes or patterns of water quality
complaints.
Manual reviews should be conducted on an established schedule, at least daily. Utilities can review water
quality complaint data at the beginning or end of the day, or reviews could be incorporated into shift
change procedures. This provides utilities with a consistent process for establishing baseline complaint
volumes and reliably detecting when a threshold is exceeded. To increase the timeliness of detection,
complaints can be reviewed more frequently.
Implementing an Automated Anomaly Detection System
Automated anomaly detection systems analyze water quality complaint data and generate alerts when
thresholds are exceeded without manual intervention. Automated anomaly detection systems are preferred
over manual review of water quality complaints, since they can aggregate data from multiple IT systems
and do not require active participation in the surveillance. Automated anomaly detection systems only
require activity by utility personnel when anomalous complaint volumes are detected and an alert is
generated. By deploying an automated anomaly detection system, CCS can be implemented with
minimum interruption in day-to-day business operations.
17
-------
Designing Customer Complaint Surveillance
Automated anomaly detection systems should execute every 15 minutes or less to take full advantage of
the early warning capabilities of CCS. However, a utility may not have the ability to implement an
anomaly detection system that executes every 15 minutes, due to technical limitations. For example,
complaint data may only be available at certain times (hourly or daily), or frequent execution of the
anomaly detection system may strain IT system resources. In this case, the anomaly detection system
should execute as frequently as is technologically feasible.
4.2.3 Spatial Clustering Analysis Techniques
During a water quality incident, water quality complaints will cluster in space and time. Complaints that
are both temporally and spatially clustered are more likely to have a common cause. Spatial clustering of
complaints can be identified manually by plotting the complaint location on a map using push pins (or
dropping pins in free, web-based maps). More accurate automated anomaly detection systems can
perform spatial analysis by examining the frequency of complaints within hydraulically related areas,
such as pressure zones or service areas. Anomaly detection systems can also analyze complaints using
GIS functions.
Spatial data can be analyzed using (in order from least to most effective):
Administrative Areas (city quadrants, zip codes)
Hydraulic Areas (water source, pito zone, pressure zone, pressure district)
Hydraulic Areas and Spatial Algorithms (proximity measurements using GIS functions)
Administrative areas may be used as surrogates or approximations of hydraulic areas, if this is the only
data available to identify clusters. Hydraulic boundaries offer a better degree of certainty than
administrative boundaries or zip codes because hydraulically defined regions share common water quality
within a distribution system.
See Appendix A for more guidance on spatial clustering analysis techniques.
4.3 Alert Notifications
An anomaly detection system will generate alerts when thresholds are exceeded. Personnel responsible
for investigating CCS alerts must be notified when alerts are generated to ensure that all CCS alerts are
acknowledged and that the alert investigation begins in a timely manner. Automated notifications can be
electronically configured so utility personnel do not need to manually send a notification. An optimal
notification system brings the alert to the attention of the investigator immediately and provides selected
details from the alert to support the investigation. Notification methods through direct messaging and a
routinely monitored system are described in the subsections below.
Target Capability
An automated process is in place to notify designated personnel when a CCS alert has been generated.
4.3.1 Notifications through Direct Messaging
A utility can have automated phone, email, or text message notifications sent electronically when their
CCS anomaly detection system recognizes that alert thresholds have been exceeded. Direct message alerts
require investigators to be present at a workstation or have access to a smart phone with internet
connectivity to receive the alert.
18
-------
Designing Customer Complaint Surveillance
A utility should determine if "dedicated" versus ""on-call. after hour personnel" will receive the
notifications. This decision impacts the choice of technology deployed such as phone autodialers, email
messages, text messages, or any combination thereof.
The CCS Email Alert Template is an example of an email alert
message, with important details about the alert and instructions for
alert investigation. The document can be opened in Word by
clicking the icon in the callout box.
The CCS Text Message Template is an example of a text message or
a Short Messaging Service (SMS) alert notification, which contains
only basic text information about the alert. This short
communication does not provide instructions for investigation or
details about the alert; therefore, the investigator will need to go to another source to get information
about the alert. SMS alert notifications do not contain pictures, video, or audio, unlike a Multimedia
Messaging Service (MMS). A MMS can also be integrated into a CCS Information Management System.
The document can be opened in Word by clicking the icon in the callout box.
4.3.2 Notifications through a Routinely Monitored System
A utility can use a system that is monitored 24/7, such as a dashboard or Supervisory Control and Data
Acquisition (SCADA) system, to send alert notifications and provide information about an alert to
investigators. Notification through a routinely monitored system is optimal, since investigators can access
complaint records, complaint descriptions and locations, operations and water quality data, and
information about other SRS datastreams. A routinely monitored system can be configured to require alert
acknowledgment.
Figure 4-2 is an example dashboard interface with CCS complaints displayed. The toolbar at the top of
the figure shows different data layers available for viewing. The dashboard uses a map that displays icons
to indicate different types of information. In this example, a user can access additional information on
CCS complaints, such as the complaint ID, status, tier, pressure zone, date and time, and comments. For
more information on building notifications into a dashboard, consult the Dashboard Design Guidance for
Water Qualify Surveillance and Response System.
Establishing IT Requirements
If CCS will require enhancements to existing IT systems or the implementation of new systems, it is important to
establish requirements for the work. The utility must clearly define requirements for an information management
system to ensure that it meets all of the needs for CCS. The Information Management Requirements Development
Tool guides a user through a series of questions that help define:
Expected Uses of the System: identification of users and the manner in which they will interact with the CCS
information management system
Data to be Managed: types and quantity of data that will need to be managed for CCS
Functional Requirements Rating: user ratings indicating the importance of system features commonly used
for CCS information management, which will be used to prioritize functional requirements for CCS
[jv
1
This template can be used
to craft an email alert.
1
This template can be used
to craft a text message or
SMS alert.
19
-------
Designing Customer Complaint Surveillance
Cluster of dirty
water calls during
flushing operations
in central
pressure zone
CC3 infotriKtom Sv*'*"*
CwnpwiHW
10ฉ
110
~rty or
LMOCNlnd
P**ซruet
AcHrwtodtfeci CCS
Ackrwtedgisci CCS
fffKtfOiQcXX NCKlh
Crtyw
QSMicfed
Cental
DilvTins
;oi rฎ7/Ki os <3
I7f3ซrfป 11 :*
-'G17,>0&ie 0400
*cww*ปsw X
Statu* Unoer rv
tJelซSiiiซh f ^rQ^3 04Q(|
Oy-TOIซic-in*rj >01 7ni/tB 04c.istD-nef Hog^l avr plain!
i ffTOugh wsbfofTT. reports acone-nl hiap vws!ct
i 2D1XฎWfQ 0432 JoNi Smih askncft-fcsdgenefil
2O17yqa/t0 04 OB M *1 SiihI'i c^whlna vnlli
r;jirtBaair>co far rwarta^ actrvtfy
icon Description
Displayed Icon
CCS complainl
P
Mam break
Hydrant
Maintenance activity
Unacknowledged alert
ฉ
Acknowledged atert
@
Figure 4-2. Example Dashboard Displaying a CCS Alert
CCS Dashboard
USA
20
-------
Designing Customer Complaint Surveillance
Section 5: Alert Investigation Procedure
Once a CCS alert is received by utility personnel, it should be promptly investigated to determine the
underlying cause of the alert. This process should be documented in an alert investigation procedure that
guides personnel through a consistent and methodical approach.
Target Capability
A procedure has been developed, documented, and put into practice to facilitate timely and efficient investigation of
CCS alerts to determine whether the alert indicates a possible water contamination incident. The procedure
provides a clear and comprehensive sequence of steps for the investigation of alerts and assigns responsibilities
for carrying out each step. The procedure is provided to personnel responsible for supporting investigations in a
user-friendly format, such as a checklist. Personnel are trained on proper implementation of the procedure and
tools.
This section describes considerations for development of a CCS alert investigation procedure and consists
of the following subsections:
Subsection 5.1 provides guidance on developing an effective alert investigation procedure.
Subsection 5.2 provides guidance on developing tools to support the investigation.
Subsection 5.3 provides guidance on preparing for real-time alert investigation.
5.1 Developing an Effective Alert Investigation Procedure
Utilities should follow a methodical process when developing a CCS alert investigation procedure. The
following steps of the process are described in the sections below:
Define Potential Alert Causes: develop a discrete list of alert causes used to classify each alert.
Establish an Alert Investigation Process: list detailed, sequential steps for investigating an alert.
Assign Roles and Responsibilities: identify all personnel who have a role in alert investigations
and summarize their responsibilities.
5.1.1 Defining Potential Alert Causes
The objective of the alert investigation process is to identify the cause of an alert. Pre-defined alert
categories can be used to classify alerts based on common causes at the conclusion of each investigation.
Awareness of the common causes of CCS alerts can be used to develop the steps of an investigation
procedure and identify information resources helpful in confirming or ruling out potential causes of an
alert. Table 5-1 lists and summarizes the most common causes of CCS alerts, based on experience from
utilities that have implemented CCS. The causes are grouped into invalid and valid alerts.
21
-------
Designing Customer Complaint Surveillance
Table 5-1. Common Causes of CCS Alerts
Alert Cause
Description
(0
c
_c
Background Variability
Normal variance occasionally resulting in higher than expected complaint
volumes.
(0
Change in System Operations
A water quality change caused by unusual system operations, such as a
change in pumping, valving, or treatment processes.
Contamination Incident
An accidental or intentional introduction of a foreign substance into the
distribution system, which may or may not be harmful.
5.1.2 Establishing an Alert Investigation Process
With potential causes of a valid CCS alert defined, the next step is to develop an alert investigation
process to guide investigators through a detailed sequence of steps to determine the cause of an alert. This
process will generally begin with an initial review of complaint characteristics and locations, followed by
a review of water quality data, treatment plant operations, distribution work in the area, and utility
operations that may impact water quality.
If the review of complaints shows dissimilar descriptions or dispersed locations, the alert is not likely
indicative of a localized deterioration of water quality in the distribution system. Instead, the alert may be
due to normal variations in the number of water quality complaints received by the utility. If the
complaints are related, i.e., similar in nature and clustered in one general area, the investigation may
reveal a likely cause for the alert, such as a main break or localized flushing activities. In this case, a
utility may choose to implement standard procedures for addressing water quality complaints. If the
investigation concludes that the CCS alert is not an indicator of contamination, the investigation is closed.
If, however, the alert investigation reveals clustered complaints that are similar in nature and are not the
result of known utility operations in the area, then the alert may be due to a more serious water quality
problem. In this case, the SRS Manager is contacted and additional investigative and response actions are
implemented under Consequence Management.
Utilities should document their CCS alert investigation procedure in a diagram that walks through the
entire process. This simplified representation of the alert investigation process allows individuals with
responsibilities for discrete steps to see how their activities support the overall investigation.
The CCS Alert Investigation Procedure Template includes an
editable table, process flow diagram, and alert investigation
checklist that can be used to develop a CCS alert investigation
procedure for a utility. The template can be opened in Word by
clicking the icon in the callout box.
Figure 5-1 provides an example of a CCS alert investigation procedure diagram. The major steps and
decision points are shown, as well as additional detail on the actions implemented. A range of estimated
times for properly trained personnel to complete groups of steps is provided to the left of the steps and
decision points. The total time for utility personnel to complete a CCS alert investigation could range
from 2 to 85 minutes, based on experience with SRS pilot utilities.
[wfj
This template includes an
editable alert investigation
process diagram and
checklist.
22
-------
Designing Customer Complaint Surveillance
ALERT INVESTIGATION PROCESS
ACTIONS IMPLEMENTED
E
o
r
Utility personnel
receive alert ,
Investigator reviews
underlying complaints
r
Are the complaints related? ^
k.
i
Investigator reviews
complaints for;
Similar problem descriptions
Spatial clustering
NO
n
YES
Close investigation,
log Incident, and
return to normal
operation
t *
o
to
Investigator reviews
un
utility operations and data
ป
A Utility reviews:
M ' Water quality data
M > Treatment plant operations and
monitoring data
Distribution system operations
V and monitoring data
V ป Work activities and main breaks
1 In distribution system
Can water
contamination
be ruled out?
A
NO
I
Contamination is
possible, notify
L
SfiS Manager
YES
~L
Close investigation,
log Incident, and
return to normal
operation
Figure 5-1. Example CCS Alert Investigation Process
23
-------
Designing Customer Complaint Surveillance
5.1.3 Roles and Responsibilities
Different utility personnel may be involved in investigating CCS alerts. Table 5-2 shows an example of
roles and responsibilities during CCS alert investigations. If utility water quality complaints are handled
by a 311 system or city call center, partners external to the utility may need to be involved in the
investigation.
Table 5-2. Example of Roles and Responsibilities for CCS Alert Investigations
Role
Alert Investigation Responsibilities
Water Quality Chemist
Receive notification of CCS alerts and lead the investigation in collaboration with
the CSR Supervisor.
Coordinate support from the Distribution System Operator, Distribution Dispatch
Supervisor, and CSR Supervisor during investigation of the CCS alert.
Review water quality data in the area of the CCS alert.
Document the alert investigation and close out.
Distribution Dispatch
Supervisor
Review distribution system work orders during investigation of a CCS alert.
Distribution System
Operator
Monitor SCADA alerts and review operational data to support the investigation of
alerts.
CSR
Collect detailed information from customers regarding water quality complaints
during normal business hours.
Advise customers about water quality incidents related to typical distribution
system issues (e.g., rusty water due to flushing, chlorine odor due to operations).
Provide details on specific water quality complaints to the CSR Supervisor.
CSR Supervisor
Receive and assist in the investigation of contact alerts in conjunction with the
Water Quality Chemist.
SRS Manager
Receive notification of possible water quality contamination.
5.2 Developing Investigation Tools
CCS alert investigation tools help guide implementation of the investigation procedure. The following
investigation tools are discussed in the sections below:
Checklists
Record of Alert Investigations
Quick Reference Guides
5.2.1 Alert Investigation Checklists
Alert investigation checklists are job aids that guide personnel through their investigative responsibilities
and document investigation findings. Checklists also ensure consistency among investigators and improve
recordkeeping. Checklists are derived from the process flow and serve to
prompt investigators to check resources, evaluate information, and
perform actions. They generally list the activities assigned to specific
roles; therefore, more than one checklist may be developed to support
the CCS alert investigation procedure.
The checklist should be presented concisely and require that only critical
information be documented. This will allow the user to focus on
conducting the investigation without needing to track or record
Consistency
If other SRS components are
being implemented, strive for
consistency in investigation
tools and roles and
responsibilities across
components.
24
-------
Designing Customer Complaint Surveillance
unnecessary details. A checklist is most useful when limited to one double-sided page, which should be
adequate for most utilities. Refer to the CCS Alert Investigation Procedure Template in Section 5.1.2 for
an editable checklist.
5.2.2 Record of Alert Investigations
A record of alert investigations provides documentation of key information, such as date/time and name
of the investigator, including the actions implemented during the investigation and the likely cause of the
alert. This record may also serve as a resource during the investigation of future alerts and provides a
means to analyze the frequency of alert investigations by a variety of factors (e.g., alert cause,
investigator, time of day, season of year).
There are a variety of ways to document alert investigations. For example, a spreadsheet can be
maintained that can be accessed by the SRS Manager and all necessary investigators on a shared drive.
Electronic tools and mobile applications make it easy to standardize, synchronize, and compare data,
while increasing accuracy. These tools can be aggregated to capture response times and conclusions,
which may provide valuable insight into the implementation of the alert investigation procedure. Figure
5-2 provides an example of electronic alert investigation records.
Atafl lnV44Ugซ)ซl InlertmjlB"!
Alert Lwatlofl
InvaMiqaiof
Investigation
Start Oirtumnw
iMYillig.lHtjN
End OiWTi
Conclusion
Notes
4ffl2Ql7 8.i1
&2V2QW 17 02
Abbmoion Village
Monflo Court
Pal FpekJ
Joe Rqscis
V3f2Q17ป23
&24WV 17 12
V3mv 10.46
M40Q17 17-25
Invalid alert procedural e"(H
Invalid ilcrt Sackatiuntt vanafcHlilT
CSR miscoded two paaitfs
CcnifKainl* did not have
simila-descriptors
7^2017 8.03
5omce Ccnlur WUiaroi
TIMfXV &.1C
7/12O017SS&
Vis alert ouh! identified
dwttibubon *yiป(ri issuu
lia n jireak resgied j j;i,-sIht
kT rusiy wartc-t OomEtuinh;
Figure 5-2. Example of Alert Investigation Records
If a dashboard will be used to support the SRS, the electronic tracking of investigations may be
incorporated into the design. For example, electronic checklists can be developed that automatically enter
investigation records and updates into an SRS s information management system. See Dashboard
Design Guidance for a Water Quality Surveillance and Response System for more information.
5.2.3 Quick Reference Guides
While many alert investigation activities will become second nature to investigators, additional tools may
be useful for completing complex or less frequently implemented tasks. Key information can be
summarized using quick reference guides or factsheets to ensure investigators can easily get the
information they need. For example, quick reference guides could be developed for checking a call queue
volume, recalling additional CSRs, or using a reverse 911 system to notify customers about a potential
water quality issue.
5.3 Preparing for Real-time Alert Investigations
After the CCS alert investigation procedure is developed, a utility will need to develop a plan to put it into
practice. The benefits of CCS can be fully realized only if CCS alerts are investigated in real time and
responded to appropriately. The following topics are covered under this section:
Training
Preliminary operation
Real-time operation
25
-------
Designing Customer Complaint Surveillance
5.3.1 Training
Proper training on the alert investigation procedure ensures that all personnel with a role in investigating
CCS alerts are aware of their responsibilities and have the knowledge and expertise needed to execute
those responsibilities. Training on the alert investigation procedure could include the following:
An overview of the purpose and design of CCS
A detailed description of the alert investigation procedure and the role of each participant
A review of checklists, quick reference guides, information management systems, and other tools
available to support CCS alert investigations
Instructions for entering new alert investigation records and retrieving previous records
Section 6 of Guidance for Developing Integrated Water Quality Surveillance and Response Systems
provides information on implementing a training and exercise program. In general, classroom training is
used first to orient personnel to the procedure and their responsibilities during CCS alert investigations.
Once CCS personnel are comfortable with the procedure, exercises can be conducted to provide personnel
with an opportunity to implement their responsibilities in a controlled environment. The SRS Exercise
Development Toolbox is an interactive software program designed to help utilities design, conduct, and
evaluate exercises specific to CCS and the other SRS components.
5.3.2 Preliminary Operation
A period of preliminary operation should follow initial training, allowing utility personnel to practice
their responsibilities in test mode before the transition to real-time operation. For example, personnel can
be asked to investigate alerts in batches as they have time, not necessarily as the alerts are generated.
During this period, investigators may or may not receive alert notifications.
During preliminary operations, it may be useful to hold regular
meetings with all investigators to discuss recent data and alerts. It
is generally most effective if participants are asked to perform
specific analyses or alert investigations before each meeting and
then discuss conclusions, observations, insights, and challenges as
a group. These meetings can be held more frequently initially
(such as weekly) but become less frequent as proficiency increases
and issues are resolved. Meeting monthly during the period of
preliminary operation would be appropriate and sufficient for most
CCS applications.
Preliminary operation provides personnel with opportunities to refine the alert investigation procedure
and investigation tools. Based on feedback from investigators, responsibilities can be clarified,
unnecessary steps can be eliminated, existing tools can be refined, new tools can be developed, and
procedures can be better integrated into existing job functions.
5.3.3 Real-time Operation
During real-time operation, CCS alerts are investigated as they are generated, and Consequence
Management is activated if a contamination incident is considered possible. The transition from
preliminary operation to real-time operation should be clearly communicated to all utility personnel with
a role in CCS alert investigations. This includes establishing a date for the transition to real-time
operation and providing expectations for how alert investigations will be performed and documented.
Do Not Rush
Do not rush preliminary
operation. This period provides
an opportunity for personnel to
practice their responsibilities and
become familiar with the data
typically used during
investigations, improving the
efficiency of alert investigations.
26
-------
Designing Customer Complaint Surveillance
After transitioning to real-time operation, it is important to continue to oversee and support investigators.
The record of alert investigations should be regularly reviewed to ensure that personnel are accurately and
thoroughly carrying out their responsibilities, and instruction should be provided to individuals who are
not. Ongoing drills, exercises, and training are important to ensure that personnel remain familiar with
their responsibilities and to address any changes, such as updates to the procedure or investigation tools.
Maintenance of the alert investigation procedure during real-time operation may involve periodic review
to verify that it is working as intended. Because CCS alerts may be infrequent, refresher training may be
needed to maintain proficiency. Finally, it is important to thoroughly train new CCS personnel on their
responsibilities and alert investigation procedures.
Regularly Review and Update the Alert Investigation Procedure
Routine updates to the alert investigation procedure and investigation tools are necessary to maintain their
usefulness. Recommendations for procedure maintenance include:
Designate one or more individuals with responsibility for maintaining alert investigation materials
Establish a review schedule (annual reviews should suffice in most cases)
Review the record of alert investigations, conduct tabletop exercises, and solicit feedback from investigators
to identify necessary updates
Establish a protocol for submitting and tracking change requests
27
-------
Designing Customer Complaint Surveillance
Section 6: Preliminary CCS Design
The information presented in the previous sections of this document can guide the development of a
preliminary CCS design that supports a utility's SRS design goals and performance objectives. If CCS
will be a component in a multi-component SRS, the design of the integrated system will likely be guided
by a project management team. In this case, guidelines for design of the individual components should be
provided to the component implementation teams and should include:
Overarching design goals and performance objectives for the SRS
Existing resources that could be leveraged to implement the SRS components, including
personnel, procedures, equipment, and information management systems
Project constraints, such as budget ceilings, schedule milestones, and policy restrictions
Instructions or specific guidelines for the development of preliminary component designs
Regardless of whether CCS will be developed as a standalone component or as part of a multi-component
SRS, the preliminary CCS design should be documented in sufficient detail to assess whether or not it can
achieve the design goals established for the component within project constraints.
Ep
This template can be
used to develop a
preliminary CCS design.
A Preliminary CCS Design Template can be opened and edited in
Word by clicking the icon in the callout box. This template covers the
following aspects of CCS design:
Component implementation team: Identify personnel from the
utility that will have a role in the design and implementation of CCS. Document the role,
responsibilities, and estimated time commitment of each team member.
Design goals and performance objectives: Using the overarching design goals and performance
objectives established for the SRS, develop specific CCS goals and performance objectives to
guide the design process.
Customer complaint surveillance systems: Identify all datastreams and IT systems that will be
used to monitor water quality complaints. If existing systems will be enhanced or integrated,
describe the enhancements or integration. If new systems will be deployed, provide
specifications, including the datastreams that will be monitored.
Preliminary information management requirements: Identify all information management systems
that would be used during operation of CCS. This will likely include utility systems that will be
accessed during the investigation of CCS alerts. Develop an information flow diagram depicting
user-to-machine and machine-to-machine interactions. Document technical and functional
requirements for any new or modified information management systems. Note any data sharing
agreements that will need to be developed in order to implement the information management
system.
Initial training requirements: Develop a training plan to educate personnel about their
responsibilities during operation of CCS.
Budget: Develop a line item budget for the CCS component noting the method for covering each
cost item. It is recommended that the budget include implementation as well as operation and
maintenance (O&M) costs, which can be used to develop a lifecycle cost estimate. The budget
should indicate the year in which each cost is incurred. Contingencies should be included to avoid
cost overruns.
Schedule: Develop a schedule that shows the planned sequencing of activities and key
dependencies. The schedule may reflect a phased implementation over multiple years, which may
be advantageous or necessary to overcome resource (financial or personnel) limitations.
28
-------
Designing Customer Complaint Surveillance
It is also a good idea to develop a preliminary alert investigation procedure using the template provided in
Section 5. This procedure can help to identify information resources needed to conduct a CCS alert
investigation, which can inform CCS design, particularly information management requirements.
In some cases, multiple design alternatives may emerge. A benefit-cost analysis should be performed to
identify the preferred option. The resource Framework for Comparing. Alternative Water Quality
Surveillance and Response Systems provides an objective process for comparing design alternatives with
respect to their lifecycle costs and capability.
29
-------
Designing Customer Complaint Surveillance
Resources
Overview of CCS Design
Water Quality Surveillance and Response System Primer
This document provides an overview of SRS for drinking water distribution systems. It covers
possible applications of an SRS, provides information about the monitoring and surveillance
components, describes common design goals and performance objectives, and includes an
overview of the approach for implementing an SRS. EPA 817-B-15-002, May 2015.
https ://www.epa. gov/sites/production/files/2015-
06/documents/water quality sureveillance and response svstem primer.pdf
Customer Complaint Surveillance Primer
This document provides an overview of the CCS component and presents information about the
goals and objectives of CCS in the context of an SRS. EPA 817-B-15-002C, May 2015.
http://www.epa.gov/sites/production/files/2015-
06/documents/ciistomer complaint surveillance primer.pdf
Complaint Collection
Example Language for a Consumer Confidence Report (Word File)
The case study includes examples of how a customer can be directed to the best method for
communicating water quality concerns. November 2017.
Click this link to open the template
Email Message Template (Word File)
The email message template is an example of how a utility can communicate directly with
customers. November 2017.
Click this link to open the template
Generic Bill Insert Template (PowerPoint File)
The generic bill insert template is an example of how a utility can communicate directly with
customers. November 2017.
Click this link to open the template
Summary of Implementation Approaches and Lessons Learned from the Water Security Initiative
Contamination Warning System Pilots
This summary report contains case studies from the Water Security Initiative, including each
pilot's implementation of CCS. EPA 817-R-15-002, October 2015.
https://www.epa.gov/sites/production/files/2015-
12/documents/wsi pilot summary report 102715.pdf
Water Quality Complaint Processing Form Template (Word File)
The template includes of a form that CSRs or other utility personnel can use to capture
information about a water quality complaint and track any work orders associated with the
complaint. November 2017.
Click this link to open the template
30
-------
Designing Customer Complaint Surveillance
Information Management and Analysis
Taste and Odor: An Operator's Toolbox [DVD]
This video provides methods for determining the causes of taste and odor in potable water, as
well as treatment responses. AWWA. (2001).
Diagnosing Taste and Odor Problems Field Guide
This field guide provides all the information water utilities need to diagnose and remove
objectionable tastes and odors at the tap. Burlingame, G., Booth, A., Dietrich, A., Gallagher, D.,
Khiari, D., Suffet, I.H., and Watson, S. (2011). Washington, DC: AWWA.
Early Warning and Management of Surface Water Taste-and-Odor Events
This report focuses on strategies to head off taste and odor problems before they can cause
complaints, offering a set of practical guidelines and tools that any utility can use. Taylor, W.,
Losee, R., Torobin, M., Izaguirre, G., and Sas., D. (2006). Washington, DC: AWWA.
Taste at the Tap: A Consumer's Guide to Tap Water Flavor
This informative booklet assists utilities in helping their customers understand what can cause
tastes and odors in tap water and why tastes and odors do not necessarily indicate a problem with
the water. Burlingame, G. (2010). Washington, DC: AWWA.
Alarm Estimation Tool (Excel File)
Estimates the number of CCS alerts generated at a given alert threshold using a spreadsheet
interface. EPA, January 2011.
https://www.epa.gov/waterqualitvsurveillance/customer-complaint-surveillance-resources
Threshold Analysis Tool
Supports the analysis of a variety of data file formats and offers three different statistical methods
(percentile, and standard deviation, and recurrence intervals) to determine thresholds for use in
identifying anomalous water quality complaint volumes. EPA 817-B-13-005, 2013.
https://www.epa.gov/waterqualitvsurveillance/customer-complaint-surveillance-resources
CCS Email Alert Template (Word File)
The CCS email alert template is an example communication which provides significant detail on
alert data and instructions for investigation. November 2017.
Click this link to open the template
CCS Text Message or SMS Alert Template (Word File)
The CCS text message or SMS alert template is an example communication which provides only
basic alert data. This short communication does not provide instructions for investigation.
November 2017.
Click this link to open the template
Dashboard Design Guidance for Water Quality Surveillance and Response Systems
This document provides information about useful features and functions that can be incorporated
into an SRS dashboard. It also provides guidance on a systematic approach that can be used by
utility managers and IT personnel engaged in the process of designing a dashboard to define
requirements. EPA 817-B-15-007, November 2015.
https ://www.epa. gov/sites/production/files/2015-
12/documents/srs dashboard guidance 112015.pdf
31
-------
Designing Customer Complaint Surveillance
Information Management Requirements Development Tool
This tool is intended to help users develop requirements for an SRS information management
system, thereby preparing them to select and implement an information management solution.
Specifically, this tool (1) assists SRS component teams with development of component
functional requirements, (2) assists IT personnel with development of technical requirements, and
(3) allows the IT design team to efficiently consolidate and review all requirements. EPA 817-B-
15-004, October 2015.
https://www.epa.gov/waterqualitvsurveillance/information-management-requirements-
development-tool
Alert Investigation Procedure
CCS Alert Investigation Procedure Template (Word File)
The alert investigation procedure template includes an editable table, flow diagram, and checklist
that can be used to document the utility's role in a CCS alert investigation process. November
2017.
Click this link to open the template
Guidance for Developing Integrated Water Quality Surveillance and Response Systems
This document provides guidance for applying system engineering principles to the design and
implementation of an SRS to ensure that the SRS functions as an integrated whole and is
designed to effectively perform its intended function. EPA 817-B-15-006, October 2015.
https ://www.epa. gov/sites/production/files/2015-
12/documents/guidance for developing integrated wq srss 1 I04l5.pdf
SRS Exercise Development Toolbox (EPA, 2016)
The SRS Exercise Development Toolbox assists utilities and response partner agencies to design,
develop, conduct, and evaluate SRS-related discussion and operations-based exercises. These
exercises help to develop, teach, refine, and improve SRS procedures. EPA, 2016.
https://www.epa.gov/waterqualitvsurveillance/water-qualitv-surveillance-and-response-svstem-
exercise-development-toolbox
Preliminary CCS Design
Preliminary CCS Design Template (Word File)
The preliminary CCS design template can be used to document aspects of CCS component
design, such as the component implementation team, design goals and performance objectives,
CCS systems, preliminary information management requirements, initial training requirements,
budget, and schedule. November 2017.
Click this link to open the template
Framework for Comparing Alternatives for Water Quality Surveillance and Response Systems
(EPA, 2015g)
This document provides guidance for selecting the most appropriate SRS design for a utility from
a set of viable alternatives. It guides the user through an objective, stepwise analysis for ranking
multiple alternatives and describes the types of information necessary to compare the alternatives.
EPA 817-B-15-003, June 2015.
https ://www.epa. gov/sites/production/files/2015-
07/documents/framework for comparing alternatives for water quality surveillance and resp
onse svstems.pdf
32
-------
Designing Customer Complaint Surveillance
References
Grabinski, J. and Hesner, R., 2011. Calls Taken Here: How a Joint Pilot Project Streamlined Operational
Response for Consumer Water Quality Issues. Proceedings ofAWWA Water Security Congress.
Nashville, TN.
Mercer, Kenneth L., 2017. Pages From the Past: Experience in Handling Bad Water Complaints and
Laboratory Control, by Earl T. Kirkpatrick, 1916. JAWWA, 109 (4):64-66.
Whelton, A.J., Dietrich, A., Gallagher, D.L., and Roberson, J.A., 2007. Using Customer Feedback for
Improved Water Quality and Infrastructure Monitoring. JAWWA, 99 (11): 62-76.
33
-------
Designing Customer Complaint Surveillance
Glossary
311. A municipal phone number used for non-emergency requests and information.
alert. An indication from an SRS surveillance component that an anomaly has been detected in a
datastream monitored by that component. Alerts may be visual or audible and may initiate automatic
notifications such as pager, text, or email messages.
alert investigation. The process of investigating the validity and potential causes of an alert generated by
an SRS surveillance component.
alert investigation checklist. A form that lists a sequence of steps to follow when investigating an SRS
alert. This form ensures consistency with an alert investigation procedure and provides documentation of
the investigation of each alert.
alert investigation procedure. A documented process that guides the investigation of an SRS alert. A
typical procedure defines roles and responsibilities for alert investigations, includes an investigation
process diagram, and provides one or more checklists to guide investigators through their role in the
process.
anomaly. A deviation from an established baseline in a monitored datastream. Detection of an anomaly
by an SRS surveillance component generates an alert.
anomaly detection system. A data analysis tool designed to detect deviations from an established
baseline. An anomaly detection system may take a variety of forms, ranging from complex computer
algorithms to a simple set of heuristics that are manually implemented.
autodialer. An electronic device or software application that automatically dials pre-determined phone
numbers. Once a call has been answered, an autodialer either plays a recorded message or connects the
call to a live person.
baseline. Values for a datastream that include the variability observed during typical system conditions.
benefit-cost analysis. An evaluation of the benefits and costs of a project or program, such as an SRS, to
assess whether the investment is justifiable considering both financial and qualitative factors.
CCS Information Management System. Used to manage customer and water quality complaints. The
system may integrate datastreams and parts of other IT systems including Asset Management Systems,
Customer Information Systems, Customer Relation Management Systems, Geographic Information
Systems, Interactive Voice Response, Meter Data Management Systems, and Work Management
Systems.
component. One of the primary functional areas of an SRS. There are four surveillance components:
Online Water Quality Monitoring; Enhanced Security Monitoring; Customer Complaint Surveillance; and
Public Health Surveillance. There are two response components: Consequence Management and
Sampling and Analysis.
consequence. An adverse public health or economic impact resulting from a contamination incident.
34
-------
Designing Customer Complaint Surveillance
Consequence Management. One of the response components of an SRS. This component encompasses
actions taken to plan for and respond to possible drinking water contamination incidents in order to
minimize the response and recovery timeframe and, ultimately, minimize consequences to a utility and its
customers.
constraints. Requirements or limitations that may impact the viability of an alternative. The primary
constraints for an SRS project are typically schedule, budget, and policy issues (e.g., zoning restrictions,
IT restrictions, union prohibitions).
contamination incident. The presence of a contaminant in a drinking water distribution system that has
the potential to cause harm to a utility or the community served by the utility. Contamination incidents
may have natural (e.g., toxins produced by a source water algal bloom), accidental (e.g., chemicals
introduced through an accidental cross-connection), or intentional (e.g., purposeful injection of a
contaminant at a fire hydrant) causes.
Customer Complaint Surveillance (CCS). One of the surveillance components of an SRS. CCS
monitors water quality complaint data in call or work management systems and identifies abnormally
high volumes or spatial clustering of complaints that may be indicative of a contamination incident.
customer service representative (CSR). Personnel at a utility or city contact center who receive
customer information or interact with customers. These personnel often resolve issues related to water
quality, service, or billing.
dashboard. A visually-oriented user interface that integrates data from multiple SRS components to
provide a holistic view of distribution system water quality. The integrated display of information in a
dashboard allows for more efficient and effective management of distribution system water quality and
the timely investigation of water quality incidents.
datastream. A time series of values for a unique parameter or set of parameters. Examples of SRS
datastreams include chlorine residual values, water quality complaint counts, and number of emergency
department cases.
design elements. The functional areas which comprise each component of an SRS. In some cases design
elements are divided into design sub-elements. In general, the information presented in SRS guidance and
products is organized by design elements and sub-elements.
design goal. The specific benefits to be realized through deployment of an SRS and each of its
components. A fundamental design goal of an SRS is detecting and responding to distribution system
contamination incidents. Additional design goals for an SRS are established by a utility and often include
benefits to routine utility operations.
design sub-element. Features, capabilities or attributes that comprise a design element. In general, the
information presented in SRS guidance and products is organized by design elements and sub-elements.
distribution system model. A mathematical representation of a drinking water distribution system,
including pipes, junctions, valves, pumps, tanks, reservoirs, and other appurtenances. A model predicts
flow and pressure of water through the system, and, in some cases, water quality.
functional requirement. A type of information management requirement that defines key features and
attributes of an information management system that are visible to the end user. Examples of functional
requirements include the manner in which data is accessed, types of tables and plots that can be produced
35
-------
Designing Customer Complaint Surveillance
through the user interface, the manner in which component alerts are transmitted to investigators, and the
ability to generate custom reports.
Geographic Information System (GIS). Hardware and software used to store, manage, and display
geographically referenced information. Typical information layers used by water utilities include utility
infrastructure, hydrants, service lines, streets, and hydraulic zones. GIS can also be used to display
information generated by an SRS.
historical data. Data that has been generated and stored, including recent data that is readily available in
an information management system and older data that has been stored or archived in a historian.
implementation costs. Costs to procure and install equipment, IT components, and other assets necessary
to build an operational system.
information management. The processes involved in the collection, storage, access, and visualization of
information. In the context of an SRS, information includes the raw data generated by SRS surveillance
components, alerts generated by the components, ancillary information used to support data analysis or
alert investigation, details entered during alert investigations, and documentation of Consequence
Management activities.
information management system. The combination of hardware, software, tools, and processes that
collectively supports an SRS and provides users with information needed to monitor real-time system
conditions. The system allows users to efficiently identify, investigate, and respond to water quality
incidents.
information technology (IT). Hardware, software, and data networks that store, manage, and process
information.
interactive voice response (IVR). An automated call management system that transfers utility customer
calls to designated customer service representatives, based on customer selected issues such as billing or
water quality concerns.
invalid alert. An alert from an SRS surveillance component that is not due to water quality incident or
public health incident.
lifecycle cost. The total cost of a system, component, or asset over its useful life. Lifecycle cost includes
the cost of implementation, operation and maintenance, and renewal.
operational change. A change in the way the distribution system is operated, including changes in
pumping or valving.
operation and maintenance (O&M) costs. Expenses incurred to sustain operation of a system at an
acceptable level of performance. O&M costs are typically reported on an annual basis and include labor
and other expenditures, such as supplies and purchased services.
performance objectives. Measurable indicators of how well an SRS or its components meet established
design goals.
possible. In the context of the threat level determination process, water contamination is considered
possible if the cause of an alert from one of the surveillance components cannot be identified or
determined to be benign.
36
-------
Designing Customer Complaint Surveillance
real-time. A mode of operation in which data describing the current state of a system is available in
sufficient time for analysis and subsequent use to support assessment, control, and decision functions
related to the monitored system.
remote access. The ability of a user to access an information management system from a location other
than the physical location of the hardware that hosts the system.
software. A program that runs on a computer and performs certain functions.
Supervisory Control and Data Acquisition (SCADA). A system that collects data from various sensors
at a drinking water treatment plant and locations in a distribution system, and sends this data to a central
information management system.
target capability. A level of performance or an outcome for a design element that is necessary for an
effective CCS component. Even if the target capability is not completely achieved, a design element can
be enhanced to improve performance of the component.
technical requirement. A type of information management requirement that defines system attributes
and design features that are often not readily apparent to the end user but are essential to meeting
functional requirements or other design constraints. Examples include attributes such as system
availability, information security and privacy, back-up and recovery, data storage needs, and integration
requirements.
threshold. A value that is compared against current or recent data to determine whether conditions are
anomalous or atypical of normal operations.
user interface. A visually oriented interface that allows a user to interact with an information
management system. A user interface typically facilitates data access and analysis.
valid alert. Alerts due to water contamination, verified water quality incidents (e.g., occurrence of rusty
water), intrusions at utility facilities, or public health incidents.
water quality complaints. Complaints received by a utility from a customer indicating that water quality
is not as expected. Traits such as an unusual taste, odor, or appearance can all indicate abnormal water
quality within the distribution system.
water quality incident. An incident that results in an undesirable change in water quality (e.g., low
residual disinfectant, rusty water, taste and odor, etc.). Contamination incidents are a subset of water
quality incidents.
Water Quality Surveillance and Response System (SRS). A system that employs one or more
surveillance components to monitor and manage distribution system water quality in real time. An SRS
utilizes a variety of data analysis techniques to detect water quality anomalies and generate alerts.
Procedures guide the investigation of alerts and the response to validated water quality incidents that
might impact operations, public health, or utility infrastructure.
Water Quality Surveillance and Response System Manager (SRS Manager). A role within an SRS
typically filled by a mid- to upper-level manager from a drinking water utility. Responsibilities of this
position include: receiving notification of valid alerts, coordinating the threat level determination process,
37
-------
Designing Customer Complaint Surveillance
integrating information across the different surveillance components, and activating Consequence
Management.
Water Security Initiative. A program developed by EPA to design, evaluate, and promote adoption of
Water Quality Surveillance and Response Systems within the drinking water sector.
work management system. Software used by a utility to schedule and track maintenance, repairs, or
other operations in the distribution system. The system may generate work orders or work requests that
can be leveraged as a CCS datastream.
38
-------
Designing Customer Complaint Surveillance
Appendix A: Spatial Clustering Analysis Techniques
This appendix expands on the topics discussed in Section 4.2.3: Spatial Clustering Analysis Techniques.
Spatial analysis reduces the frequency of invalid alerts by only generating an alert when a threshold of
complaints within a defined area is surpassed. Spatial data can be analyzed using (in order from least to
most effective):
Administrative Areas (city quadrants, zip codes)
Hydraulic Areas (water source, pito zone, pressure zone, pressure district)
Hydraulic Areas and Spatial Algorithms (proximity measurements using GIS functions)
Analyze Data within Administrative Areas
Analyzing the frequency of complaints within an administrative area can improve the ability of CCS to
detect water quality incidents. Administrative areas such as neighborhoods, zip codes, or school districts
can be used to spatially delineate and capture water quality complaints. To identify a cluster of spatially
related complaints, a utility should develop different thresholds for each administrative area. An alert will
be generated whenever the threshold within the area is surpassed, regardless of alert occurrence in
neighboring areas. While this results in alerts that are easily understood by the end user, it does not
incorporate information from neighboring areas. The complaints generated by water contamination could
be spread out over several adjacent areas. Thus, a rise in water quality complaints in a region containing
multiple areas may be missed by focusing on individual areas.
One method to account for the possibility that a water quality incident will extend beyond a single area is
to use spatial filtering methods. Spatial filtering, or "smoothing," techniques use data from surrounding
areas or regions to even out areas of high variance. A smoothing of values across regions can create one
estimated value. The area around the regions used for smoothing estimates can be defined by the user,
such as a one mile radius around the region. There are a number of spatial smoothing techniques such as
locally-weighted average, empirical Bayesian models, and the Head-bang algorithm. The Bayesian
models are more flexible than other methods, but can be more cumbersome to use and understand.
Visualizing spatially smoothed data with GIS can provide investigators with a real-time view of data
across the distribution system. These tools allow investigators to see if adjacent regions have more
complaints which may not have reached a threshold.
Analyze Data within Hydraulic Areas
Analyzing the frequency of complaints within a
hydraulic area can improve the ability of the CCS
anomaly detection system to detect a water quality
incident. Hydraulic areas correlate with distribution
system operations, so complaints within hydraulic
areas are more likely to be related to the same
underlying cause. A utility can program an automated
mapping function for plotting complaints within a GIS
layer showing hydraulic areas. This layer is created
using a utility's distribution system model.
Complaints that originate from the same hydraulic
area should be evaluated to determine whether there
are similar problem descriptions amongst spatially
related complaints. To determine whether there is a
hydraulic relationship between complaints, it is
Hydraulic Areas Case Study
Dallas Water Utilities (DWU) established
hydraulic areas consisting of four aggregated
"mega" pressure zones. The mega pressure
zones, shown in Figure A-1, consist of
multiple, smaller pressure zones that share a
common water source. Each mega-pressure
zone has its own set of characteristics in
terms of population and water demand. For
example, Central Low contains the business
district, which presents a unique monitoring
challenge because the weekday population
and demand during work hours is much
Vgreater than after work hours and on
weekends.
39
-------
Designing Customer Complaint Surveillance
recommended that source water data, service area data, pressure zone data, pressure district data, or any
other descriptor of normal distribution system hydraulics be captured in IT systems used to report water
quality complaints.
Meandering Way I
Whispering Hills
Lovers Lane
Intermediate
Central Low
Lone Star
Intermediate
Pleasant Grove
intermediate
/ kTVH
South High / _ Trinity-T
I ^ Heights^
\ I Intermedial
' - / pv.
Brooklyn Heights
Intermediate
Polk Street
Intermediate
Figure A-1. Hydraulically Connected Mega Pressure Zones within the DWU Distribution System
(Grabinski, 2011)
40
-------
Designing Customer Complaint Surveillance
Analyze Data Using Hydraulic Areas and Spatial Algorithms
CCS can apply many of the algorithms utilized by public health organizations to detect outbreaks. Spatial
scan statistics typically apply statistical measures of proximity to determine if complaint locations
constitute a cluster. Proximity measurements can result in the identification of very small clusters within
administrative areas or hydraulic areas, providing a clear focus for the investigation. Alternatively, these
algorithms can identify clusters which are spread across
adjacent spatial areas, which may get lost if examining
complaints only within a specific spatial area. Leveraging
a GIS system's analytical and data capabilities to
determine the relative distances between all water quality
complaint locations is more comprehensive than an
analysis on a regional basis.
Another algorithm available through the use of GIS is
network tracing. Many GIS applications have the
capability to identify hydraulically connected customers
through automated tracing along the distribution system
model network. Combined with a scan window parameter,
this algorithm could theoretically determine if the number
of complaints has surpassed a critical level and if the
complaints are hydraulically related.
Spatial Algorithms Case Study
The Philadelphia Water Department uses
a GIS-based work management system
in which CSRs enter complaints. The
anomaly detection system analyzes the
complaint data by using predetermined
hydraulic pressure district areas.
Additionally, the system searches a
2,000-foot radius around the complaints
to identify any service requests or work
orders for main breaks, distribution work,
hydrant activity, and valve operations in
the area. This expedites the response by
investigators, since common explanations
for customer complaints are displayed
with the alert.
41
------- |