• Cognizant 20-20 Insights

How to Generate Greater Value

from Smart Meter Data
By managing and analyzing smart meter event data,
utilities can improve customer experience, grid reliability,
operational efficiency and revenue assurance.

Executive Summary operations, we believe that information captured

from events can be used to derive useful insights
Utilities have made significant investments in
to vastly improve customer experience, grid reli-
smart meter roll-out programs and are now
ability, outage management and operational
looking for ways to get a return on this investment.
efficiency. The challenge lies in managing the
In addition to ROI, regulators are pushing utilities
high volumes of event data and applying logical
to show how these investments are helping to
and predictive analytics to it, such as filtration,
improve operational efficiencies and deliver
association, correlation, factor analysis and
enhanced levels of customer service.
regression, as these are relatively new concepts
Industry-led efforts such as Green Button1 are for most utilities.
utilizing smart meter read data to provide
This white paper discusses the numerous logical
customers with visibility into their energy usage
and statistical techniques that utilities can utilize
data and consumption and billing patterns, as
to tap the potential of events information. It also
well as tools for “what-if” scenarios. However, the
illustrates how these techniques can be applied
other category of data generated by smart meters
to improve the outage management process
— meter events — is a relatively new concept for
(outage detection, verification and restoration)
utilities, and its true value is largely untapped.
and enhance operational efficiency and field crew
Some utilities in North America are just at the
optimization.
early adoption stage of gaining insights from
event data. Meter Event Data:
Event information relayed from smart meters Beyond Interval Reads
includes real-time device status, power quality Smart meters are well known for their ability
information and meter status information, all of to provide meter read data at smaller intervals,
which provides a very powerful source of informa- such as every 15, 30 or 60 minutes, as well as bi-
tion to improve utilities’ core business processes. directional communication and remote operating
Based on our experience with and observations capabilities. In addition to these features, smart
of the changing nature of utilities’ industry meters also generate hundreds of meter events.

cognizant 20-20 insights | april 2012

An event is information that originates from the Deriving Business Value
meters’ endpoints and can have several attributes,
By now, many utilities are broadly aware of the
including source and proxy information, severity
possible areas where they would like to leverage
level and event category. The source is normally
information from events. However, the real
the device that originates the event, while the
challenge lies in how to develop the processes
proxy is the device responsible for detecting
and systems to continuously convert data into
and communicating the event. Severity levels
actionable information and then further refine
include emergency, information, error, warning
the models based on the results.
and clear. The event category provides informa-
tion regarding the process to which the event is This challenge arises because of the nature of
related. There are four basic event categories: event data, both status and exception. Event
data is a raw data stream and is also associated
• Meter or device status events, such as with high volumes because there are hundreds of
“power restore” and “last gasp.”
events generated for normal operations, as well
• Power quality events, such as voltage sag, as for changed conditions. These events also need
swell and high/low voltage alarms. to be validated with other relevant information,
• Meter or device tamper flags, such as as they basically manifest the conditions of the
reverse energy flow. network (meter or grid) and also some aspects of
customer behavior.
• Meter hardware information, such as low
battery alarms and battery critical alerts. To manage the above needs, we believe that
Potential Business Areas for utilities need to focus on two key dimensions:
Events Data Insights • Systems to manage large volumes of events
Some of the potential business areas where infor- data, both real-time and batch.
mation from meter events can be used to derive
useful business insights are:
• Logical and statistical techniques that will help
identify the right events and correlate with
various conditions, both event- and business-
• Customer experience: Events like last gasp
related, and, finally, predict the outcomes.
and power restore, which can identify field
outages and take proactive action even before Key logical and statistical techniques that could
the customer calls, as well as alerts and notifi- be used include:
cations to customers regarding power outages.

• Outage management: Events to detect • Data filtering: This refers to the analysis of
outages at the right device level and create events and intelligent filtration of redundant
proactive tickets, as well as “power restore” data based on predefined conditions from
to identify nested outages after large-scale the event data stream. This technique uses
outage restoration. Boolean logic.2 Based on our experience, events
like last gasp and power restore are relayed
• Power quality: Events like “voltage sag” and multiple times from the smart meters due to
“voltage swell,” in correlation with other device reliability considerations. These kinds of events
status information to proactively identify open have the same event occurrence intervals but
neutrals and flickering lights. different event insertion times. Hence, in such
• Revenue assurance: Events like meter cases, duplicate traps could be filtered from
inversion and reverse energy flow, along processing using timing conditions.
with meter reads to identify power theft and
abnormal usage/demand patterns.
• Association rules: Algorithms or business
rules to enable the discovery of relationships
• Smart meter network operations and between events and other variables. Inputs
monitoring: Events and meter ping commands received from other systems, such as work
to identify damaged/defective meters, access management systems (WMS), customer infor-
relays and other devices, as well as hardware mation systems (CIS) and supervisory control
events to provide information regarding and data acquisition (SCADA) systems, may be
device hardware such as battery information, associated with event information to determine
firmware version, etc. device-level issues before rolling out to the field
crews. Also, events received from the smart

cognizant 20-20 insights 2

 meters can be logically segregated based on analysis and regression will be required to obtain
the inputs received from such systems. the correct results.

• Point-of-detection algorithms: These algo- Improving Outage Management

rithms can help develop patterns of their
through Meter Events
occurrence, which can help in taking proactive
actions. For instance, time-wise and day-wise Smart meter events such as last gasp and power
patterns for events can be developed. Further, restore that provide meter off/on status can be
filtration criteria can be applied to remove all used for improving outage management. Being
patterns caused by electric, communication near-real-time, these events have an advantage
or network issues, and then the remaining over outage information coming from customers
patterns can used to explain occurrences of and field staff. Event information generated by
certain business outcomes, such as outages, smart meters is raw data with duplicate traps and
power quality or device tampering. high volume due to:

• Data clustering: This is an unsupervised • Momentary outages and restoration-related

model that uses data similarity to group the events.
data points. Similar categories of events can • Communication and network interface issue-
be clustered together, with analysis performed related events
to extract business value from the clusters of
events. For example, we can identify clusters
• Events due to planned outages, outages at the
lateral, feeder or transformer level, customer
among all event types and then develop rela- disconnects, etc.
tionships between outcomes and clusters of
events. Device status, meter tamper and power Hence, it is practically not possible for outage
quality events can be a cluster to determine management systems3 to process raw event data
issues such as open neutrals or flickering lights. in the same way as they currently process inputs
from SCADA systems, customers and field staff.
• Correlation: This measures the association
Many utilities realized this when they integrated
between two variables, while assuming there is
event information from head end systems (HES)
no causal relationship between the two. We can
directly into their outage management systems.
develop a correlation among various events
and other outcomes to determine future In order to effectively use events data, an event
behavior. For example, correlation between processing and analytics engine is required.
event type and consumption fluctuation can This engine needs to have the capabilities of
help with revenue assurance. logical filtration based on uniqueness of events,
• Factor analysis: This allows variables to be momentary and existing outages and capabilities
grouped into common sub-groups in order to of association based on physical network hierar-
reduce the number of factors to be initially chies. It also needs to have pattern analysis or
analyzed. For example, by performing factor regression capabilities to predict the outages.
analysis, we can identify dominating factors
A multistage event processing and analytics
that contribute to events or a set of events or
framework identifies confirmed cases of outages
an outcome.
that can be passed to the outage management
• Regression: This refers to the statistical rela- system for restoration (see Figure 1).
tionship between two random variables to
predict the outcome. Commonly used for fore- • Stage 1: A set of conditions is used to filter
duplicates from last-gasp events to identify
casting purposes, regression examines the
unique cases of outage events. Such events
causal relationship between two variables. An
are then correlated with power-restore events
example is using regression to analyze the
to remove the cases of momentary outages
relationship between equipment conditions in
(outages with a duration of less than 60
the field, such as a prediction of transformer
seconds).
failure, based on the demand from meters
associated with it. Further, inputs from other systems such as CIS
and WMS are considered to segregate outage
Usually, more than one technique might be events that have occurred due to existing
required to solve the problem. For example, to planned maintenance, meter exchange or
develop a relationship between device status customer disconnect. The remaining outage
and outage, a combination of correlation, factor events are considered as realized events.

cognizant 20-20 insights 3

Event Processing and Analytics Framework

Stage 1 Stage 2 Stage 3

Event Processing Probable Outage Confirmed Outage

Event Event Outage Outage Outage Outage

Filtration Realization Escalation Comparison Verification Confirmation

Figure 1

• Stage 2: In this stage, the meter-level realized meter data management (MDM), WMS, distribu-
events from Stage 1 are escalated to a higher tion automation and SCADA (see Figure 2). This
level of device hierarchies (lateral, feeder, trans- will enable effective outage management and
former, etc.) and compared with other device crew optimization by focusing on “real” outage
inputs using association rules and conditions events from smart meters.
to identify an outage incident. These cases of
outage are considered to be probable cases The benefits of this approach include:
that need to be tested further. • Early and accurate outage detection, leading
• Stage 3: During this stage, the probable cases to improvement in power system reliability
of outages from Stage 2 are verified using indices such as CAIDI, SAIDI, etc.
remote meter ping functionality, and only • Early detection of momentary pnd planned
confirmed outage incidents results are com- outages to help avoid costly field visits.
municated to the outage management system
for further action.
• Outage and restoration verification to avoid
costly field crew movement.
The event processing and analytics engine • Improved intelligence due to inputs from appli-
needs to be integrated into the utilities system cations such as CIS, WMS and SCADA .
landscape, comprising the head end system, CIS,

Smart Meter Event Processing: Business Context Diagram

Distribution Area Applications

SCADA

Field Force
Automation
Smart Feeder
Equipment Data Telemetry Data

Field Work Execution

High-Quality
Head End Events Data Events Data
System/ Smart Meter Event Outage
Smart Meter Processing Solution Management
Real-Time Real-Time System
Status Check Status Check

Customer/ Planned
Premise Data Outage Data

Customer Information Work

System/Meter Data Management
Management System System

Figure 2

cognizant 20-20 insights 4

Cognizant Smart Meter Event In addition to the above features, SMEP has been
Processing (SMEP) Solution designed using the event-driven architecture
(EDA). EDA helps orchestrate the generation,
Our Utilities Practice has designed a smart
detection and consumption of meter events, as
meter event processing (SMEP) solution for
well as the responses evoked by them. It helps
improving the outage management process. The
effectively manage events and communica-
SMEP solution is configurable to meet dynamic
tion with various application processes using
business requirements and is based on multistage
messaging (see Figure 3).
processing and analytics.
Conclusion: From Data to Insights
Our SMEP solution is designed to provide the
functionality required to process huge volumes of The concept of leveraging meter events data
real-time outage meter events data. The following to gain business insights is at an early stage.
are the key features of the SMEP solution: To effectively convert raw data into meaningful
insights, utilities need to build state-of-the-art
• Near-real-time processing of a high volume of methods in logical and predictive reasoning with
meter event data. data management capabilities. The theory of inte-
• Business rules-based engine to configure the grating and exploiting logical and statistical data
algorithms and rules to process the events. relationships is quite new; most utilities are still
• Dynamic and flexible control based on require- at an early stage of the maturity curve, primarily
ments from other utility systems. reporting on and dashboarding the smart meter
analytics they gather.
• Businessprocess management to effectively
route and manage events/incidents. Analytics need a combination of sound business
• Integration with other utility applications for and statistical capabilities, which many utilities
validation, association and correlation. lack. Statistical capabilities include knowledge of
• Visualization and dashboarding tools. statistical methods, statistical tools such as SAS
and an ability to provide statistical inferences.

Smart Meter Event Processing Solution

Stage 1 Stage 2 Stage 3

Event Preprocessing Probable Outage Confirmed Outage
Event Event Outage Outage Outage
Filtration Refinement Escalation Comparison Verification
Enterprise Service Bus
Head End System

Meter
Events Outage management
system/other
applications

Visualization and
Dashboarding Database Event Log Entry

Smart Meter Event Processing Solution

Figure 3

cognizant 20-20 insights 5

Hence, utilities need to have a two-pronged needs of the enterprise and leveraging various
approach. In the short to medium term, sources of information (not limited to meter read
utilities can build solutions largely on logical or event data) based on the assessment of the
techniques where they have sufficient develop- current state of process and people skills. They
ment experience and can leverage vendors and should consider various approaches, including
partners that provide statistical capabilities. building analytics skills through a Center of
Excellence for Analytics or developing collabora-
For the longer term, utilities need to take a holistic tive models with vendors specializing in analytics.
approach toward analytics, keeping in mind the

Footnotes
1
Green Button is an industry-led effort in response to a White House call-to-action
http://www.greenbuttondata.org/greenabout.html.
2
Boolean logic consists of three logical operators: “OR,” “AND” and “NOT” http://booleanlogic.net.
3
Outage management systems develop alternate supply plans and create job orders for restoration.

References
“Electric Power Industry Overview 2007,” U.S. Energy Information Administration,
http://www.eia.gov/cneaf/electricity/page/prim2/toc2.html.
Deepal Rodrigo, Anil Pahwa and John E. Boyer, “Location of Outage in Distribution
System Based on Statistical Hypotheses Testing,” IEEE Transactions on Power Delivery,
Vol. 11, No. 1, January 1996, p. 546.
Deepal Rodrigo, Anil Pahwa and John E. Boyer, “Smart Grid Regional Demonstration Project:
Project Narrative,” DOE-FOA-0000036, August 2009.
“Deploy Smart Grid in Difficult and Varying Terrain,” Silverspring Networks,
http://www.silverspringnet.com/services/mesh-design.html.
Doug Micheel, “Smart Grid Implementation: The PHI Story,” Pepco Holdings, Inc.,
Presentation to the 2011 GreenGov Symposium, Nov. 2, 2011.
“1-210 Single phase Meter,” GE Energy,
http://www.geindustrial.com/publibrary/checkout/GEA13391?TN R=Brochures|GEA13391|PDF.
“1-210+c SmartMeter,” SmartSynch, http://smartsynch.com/pdf/i-210+c_smartmeter_e.pdf.
Krishna Sridharan and Noel N. Schulz, “Outage Management Through AMR Systems Using An Intelli­gent
Data Filter,” IEEE Transactions on Power Delivery, Vol. 16, No. 4, October 2001, pp. 669-675.
Lise Getoor and Renee J. Miller, “Collective Information Integration Using Logical and Statistical
Methods,” University of Pennsylvania.
Peter Yeung and Michael Jung, “Improving Electric Reliability with Smart Meters,” Silverspring
Networks, 2012, http://www.silverspringnet.com/pdfs/whitepapers/SilverSpring-Whitepaper-Improving-
Electric-Reliability-SmartMeters.pdf.
Yan Liu, “Distribution System Outage Information Processing Using Comprehensive Data and
Intelligent Techniques,” Ph.D. dissertation, Michigan Technological University, 2001.

cognizant 20-20 insights 6

About the Authors
Dr. Sanjay Gupta is Cognizant’s Director of Consulting within the Energy and Utilities Practice of Cognizant
Business Consulting. He has more than 20 years of global energy and utilities industry experience in
consulting, business development and business operations and has led and executed consulting engage-
ments with several large global customers. Sanjay is also responsible for developing industry solutions
and services, with a focus on smart grid/smart metering, asset optimization, analytics, renewable energy
and operations management. Sanjay holds a doctorate degree in energy and power and a master’s in
engineering. He can be reached at [email protected].

Ashish Mohan Tiwari is a Consultant within the Energy and Utilities Practice of Cognizant Business
Consulting, with six-plus years of experience providing consulting services in the implementation of
IT systems for the utilities industry. He has extensive experience in smart metering infrastructure,
smart grid data analytics solutions and enterprise asset management. Ashish has worked on numerous
transformation engagements in the areas of process consulting, package evaluation and solution
design for global utilities companies in regulated and de-regulated markets. He can be reached at
[email protected].

About Cognizant
Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process out-
sourcing services, dedicated to helping the world’s leading companies build stronger businesses. Headquartered in
Teaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep industry
and business process expertise, and a global, collaborative workforce that embodies the future of work. With over 50
delivery centers worldwide and approximately 137,700 employees as of December 31, 2011, Cognizant is a member of
the NASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked among the top performing
and fastest growing companies in the world. Visit us online at www.cognizant.com or follow us on Twitter: Cognizant.

World Headquarters European Headquarters India Operations Headquarters

500 Frank W. Burr Blvd. 1 Kingdom Street #5/535, Old Mahabalipuram Road
Teaneck, NJ 07666 USA Paddington Central Okkiyam Pettai, Thoraipakkam
Phone: +1 201 801 0233 London W2 6BD Chennai, 600 096 India
Fax: +1 201 801 0243 Phone: +44 (0) 20 7297 7600 Phone: +91 (0) 44 4209 6000
Toll Free: +1 888 937 3277 Fax: +44 (0) 20 7121 0102 Fax: +91 (0) 44 4209 6060
Email: [email protected] Email: [email protected] Email: [email protected]

­­© Copyright 2012, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is
subject to change without notice. All other trademarks mentioned herein are the property of their respective owners.