COGNIZANT 20-20 INSIGHTS · PDF filePredictive Analytics to Deliver Smart Power Outage ......
Transcript of COGNIZANT 20-20 INSIGHTS · PDF filePredictive Analytics to Deliver Smart Power Outage ......
Applying Predictive Analytics to Deliver Smart Power Outage Communications
By building proactive outage communication that provides accurate, detailed and real-time outage statuses through customers’ preferred channels, power utilities can improve overall customer experience, customer satisfaction and operational efficiency.
Cognizant 20-20 Insights | November 2017
COGNIZANT 20-20 INSIGHTS
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 2
Executive Summary
Today’s energy utility customer is likely a user
of Amazon, UPS and Uber, and has similar high
experience expectations – especially before,
during and after power outages. This means
utilities must understand and address a wide
array of customer concerns, including:
• What: providing the most up-to-date outage
information.
• When: delivering updates in real-time.
• How: using channels such as social media,
phone calls, SMS and e-mail.
• Where: managing crew status and response.
• Why: understanding the root cause of an
outage.
Doing this requires energy utilities to overcome
two major hurdles: One, how can they be early,
accurate, proactive and detailed in communi-
cating the power outages to their customers?
Second, how effectively can they manage mul-
tiple information constraints simultaneously?
Power outages can result both from unplanned
reasons (weather, equipment malfunctions,
third-party actions like digging mishaps, car-
pole accidents or vandalism etc.) and from
planned reasons such as maintenance, so dif-
ferent communication channels are needed.
Communicating an electrical power outage
effectively during these cases in our view is
hypercritical for overall customer satisfaction.
Many energy utilities are upgrading their exist-
ing outage management ecosystem to improve
outage communications. However, utilities face
some major hurdles, including:
• The lack of an automated and intelligent
platform to collect and process all outage
inputs.
• Severe limitations of legacy outage man-
agement system (OMS) to filter all outage
inputs.
• No single source of data to store historical
and current/forecasted data.
• No intelligence mechanism for determining
which customers are affected by outages.
• Inaccurate outage messages sent to cus-
tomers.
Based on our experience working with many
leading energy utilities on similar initiatives,
the above hurdles can be addressed with a
smart outage communication ecosystem,
one that is is machine-learning-enabled. We
recommend that energy utilities build this envi-
ronment holistically and use advanced digital
tools that comprise three stages/sub-ecosys-
tems: upstream, midstream and downstream
components. This model would definitely help
utilities during emergencies or outages caused
by natural disasters such as hurricanes by
effectively providing the most up-to-date infor-
mation to customers.
Many energy utilities are upgrading their existing outage management ecosystem to improve outage communications.
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 3
HISTORICAL CONTEXT
From the dawn of electricity, utility custom-
ers were typically left in the dark concerning
the cause of an outage or when power will be
restored. Traditionally, customers needed to
be proactive and communicate with the utility
service provider to learn when the lights would
come back on. That’s the way utilities operated;
customers were conditioned to expect little in
the way of communication from their power pro-
viders.
However, in today’s age of information and social
media, customers expect much more from their
utility provider. Utilities are trying to keep pace
with the changing times to provide customers
with up-to-date outage alerts.
As illustrated in Figure 1, 75% of utility companies
are providing outage alerts to their customers.
Even though this is encouraging, these alerts
have limitations:
• Utilities are often unable to identify which
customers have lost power. For example, in
the case of a feeder outage, not all custom-
ers would be affected but alerts may be sent
to customers who still have power. This leads
to confusion and frequent calls to the utility’s
customer service department.
• Alerts may be based only on the ticket status
of reported outages. For example, alerts such
as crew status, outage causes, etc. (see Figure
2, next page) are rarely sent.
• Customers do not receive real-time outage
alerts. For instance, some customers may
have had their power restored due to auto-
matic feeder switching, but the outage
restoration alert is provided to them only
when the outage ticket is actually closed.
In today’s age of information and social media, customers expect much more from their utility provider.
Outage Communications Industry Survey
33% 38%
12%
21%
45%
75%
2014 (n=52) 2015 (n=47) � Portion of utilities offering a preference portal that allows customers to manage alerts across
various channels.
� Portion of utilities offering customers two-way texting for outage communications, including the option for customers to notify the utility about their outage by text message.
� Portion of utilities providing proactive outage alerts to customers.
Source: Chartwell’s Outage Communications Industry Survey
Figure 1
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 4
To communicate in real time to customers, utility
providers must integrate their OMS with the dis-
tribution management system (DMS), advanced
metering infrastructure (AMI), supervisory con-
trol and data acquisition (SCADA) and the work
management system (WMS). Utility companies’
progress in developing such OMS capabilities is
shown in Figure 3, next page.
CURRENT STATE LIMITATIONS
Customers who experience a power outage cur-
rently have two options for reporting it to the
utility provider:
• Calling customer service.
• Via online communication (mobile applica-
tion/utility website).
If the customer calls to report the outage, he can
speak with a customer service representative,
who will in turn create a ticket in the OMS. The
customer can also choose to interact with the
voice response unit (VRU) to submit the relevant
outage information.
Similarly, information can also be provided on
the company’s mobile application or website (if
the utility provider offers such support). Once
the information is collected, by any of the above
methods, an outage ticket is generated in the
OMS. The customer will be provided with the
ticket number, as well as an estimated time of
restoration. Some utility companies offer cus-
tomers the option to receive updates during the
restoration process. Customers must provide a
contact number where they can be reached with
the outage update information.
When Outage Communications Occur
14%
17%
20%
20%
40%
60%
60%
When crew status is updated
When the outage cause is established
When an outage is reported by the customer
When the outage cause is updated
When an outage is first verified by systems
When the estimated restoration time (ERT) is updated
When the power is thought to be restored
Source: Chartwell’s Outage Communications Industry Survey, n=47
Figure 2
To communicate in real time to customers, utility providers must integrate their OMS with the distribution management system, advanced metering infrastructure, SCADA and the work management system.
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 5
Major drawbacks of the current approach include:
• Inability to proactively communicate the
outage updates to the affected customers.
• The lack of real-time situational awareness
and visibility for accurately determining the
affected customers.
• The lack of integration among OMS, AMI and
SCADA systems, and inefficiencies caused by
the inability to seamlessly transition from a
legacy communication system to new digital
applications.
CREATING A SMART POWER OUTAGE COMMUNICATION ECOSYSTEM
A smart power outage communication ecosystem
should be able to resolve the following business
cases:
• Case 1: Predict the impact of special weather
events on the power grid and customers.
• Case 2: Predict the impact of near-term asset
failures on the power grid and customers.
• Case 3: Detect possible outages based on
smart meter events.
• Case 4: Detect outages at specific premises/
regions/areas based on the data from social
media channels.
• Case 5: Filter outage inputs from various tra-
ditional and smart channels in real time and
predict the outage type.
• Case 6: Accurately confirm an outage for
a given customer and send a personalized
outage message to the customer.
We recommend that energy utilities build such
an environment to address upstream outage
Status of Utility Company Efforts to Develop Outage Management System Capabilities
14
10
6
6
8
8
3
7
2
2
3
2
1
1
4
3
1
3
3
3
4
1
2
10
2
3
3
3
2
4
4
3
1
5
3
3
5
10
12
14
14
18
21
22
27
21
18
11
17
25
13
15
8
9
Interface with damage assessment system
Link to DMS/ADMS or other SCADA system
Interface with advanced metering infrastructure (AMI)
Interface with work management system (WMS)
A dynamic model that can handle off-nominal config.
Outage prediction capability
Electric model with accurate customer linking based on GIS
Interface with SCADA
Interface with customer information system (CIS)
� Considering � Will be developed within 6 months � Will be developed within 12 months
� Will be developed within 18 months � Will be developed within 2+ years � Already developed
� Don't know
Source: Chartwell’s Outage Communications Industry Survey, n=47
Figure 3
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 6
detection, midstream outage filtration and data
storage, and downstream outage communication
sub-ecosystems.
Figure 4 provides a high-level framework of our
proposed smart power outage ecosystem.
Stage 1 (Upstream): Outage Detection Sub-Ecosystem
The goal here is to continuously collect, process
and provide unplanned outage predictions.
In this stage, continuous outage feeds are collected
from multiple sources such as asset/smart device
sensors, weather forecasting systems, smart-meter
head-end systems/AMI, and also via various cus-
tomer information channels such as customer calls,
IVR, web reporting and social media.
The core component of this stage is the outage
prediction engine (OPE), which comprises the fol-
lowing engines:
• Weather-based outage prediction engine to
predict the impact of special weather events
on the power grid and customers.
• Predictive asset failure/fault detection engine
to predict the impact of near-term asset fail-
ures on the power grid and customers.
• Smart meter event processing engine to
detect possible outages at the premises level
Smart Power Outage Communication Ecosystem
Weather-Based
Outage
Prediction Engine
Predictive Asset
Failure
Prediction Engine
Social Events
Processing Engine
Smart Meter
Event
Processing Engine
Outage/Restoration
status communicated
to enrolled customers
as per their
communication preferencesOutage NotificationProcessing Engine
(ONPE)
Damage Assessment
System(DAS)
Drone/Air/Manual
Surveillance
Weather Forecasting
System(WFS)
SCADA Inputs
Advanced Distribution Mgmt. System (ADMS)
Enterprise Asset &
Work Management
System(EAWMS)
Geographic
Information System
(GIS)
Field Service
Mgmt. System(FSMS)
*SCADA: Supervisory control and data acquisition, *DMS: Distribution Management System, *OMS: Outage Management System
Mobility
Collaboration
MessagingPlatforms &
Desktops
United
Communications Customer
InteractionRestoration Restoration Status
Potentially Impacted Premises Historical Patterns
Data Lake: ADMS+ CCS+CIS+ GIS
+EAWMS+FSMS
+ DAS+OPE
Outage Prediction Algorithm
Pla
nned
Out
ages
Outage Prediction Engine (OPE)
Substation Substation
Last Gasp/
Power Restore
Outage Inputs
Head EndSystem
Generation Transmission Distribution Retail/Consumer
Smart Sensors/IEDs/Switches Smart Devices-FCIs/ALS/AFS
Social Events
Real Time
Check
Meter Ping
• Social Media Events
• Customer Calls/
IVR/Web
• Smart Meter
Customer
Preferences
Customer
Care SystemCustomer
OutagePreference
Portal
Retrieve listof enrolled customers
Retrieve thecommunicationpreferences
DMS OMS SCADA
Communication Channel Platform Communication Channels
Figure 4
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 7
based on meter events – last gasp (LG) and
power restoration (PR).
• Social media event processing engine to
detect outages at specific premises/regions/
areas based on social-media data.
Illustrated in Figure 5 are the key building blocks
for building each of the aforementioned engines.
The above engines leverage predictive modeling
and machine learning algorithms to proactively
detect outages. The OPE, tightly coupled with
the advanced distribution management system
(ADMS), feeds potential outage inputs. In addi-
tion, it is bidirectionally integrated with a data
lake to get insights on real-time and historical
distribution grid information, as well as feedback
on potential outage inputs for future reference.
Stage 2 (Midstream): Outage Filtration and Data Storage Sub-Ecosystem
The goal here is to collect planned and
unplanned inputs, gather damage inputs,
remove duplicates, and maintain historical and
current data.
In this stage, planned and unplanned outage
inputs, restoration status inputs and damage
inputs are collected by the ADMS, which
comprises integrated SCADA, outage manage-
ment and distribution management system
data from multiple sources, such as the OPE,
damage assessment systems, enterprise asset
and work management systems, and field ser-
vice management systems.
Moreover, the data lake – a storage repository
that holds a vast amount of mirage image data
from multiple systems such as ADMS, OPE,
weather forecasting system, damage assess-
ment system, customer information system
The Anatomy of a Prediction Algorithm
Potentially ImpactedPremises
Enterprise AssetManagement
(EAM ) System
GIS
GIS
Potentially Impacted Devices
Smart Devices/SensorsIEDs
SCADA
Potentially ImpactedPremises
Potentially ImpactedPremises
HeadEndSystem
Social MediaEvents –Twitter,Facebook, etc.
Validate Simulate Assess
Forecast & Actual Data
Outage Prediction Algorithm
STEP 2 STEP 1 STEP 3
WeatherForecasting
System
Weather-based Outage Prediction Engine (WOPE)
Monitor Condition
AssessHealth/Risk Score
AnalyzePatterns
Historic Failure Pattern Library
Asset Failure Prediction Algorithm
STEP 2 STEP 1 STEP 3
Predictive Asset Failure/Fault Detection Engine (PAFDE)
TextPreprocessing
TextTransformation
PatternDiscovery
Social Events Data
Outage Prediction Algorithm
STEP 2 STEP 1 STEP 3
Social Event Processing Engine (SEPE)
Filter Process Identify
AMI + GIS Data
Outage Prediction Algorithm
STEP 2 STEP 1 STEP 3
Smart Meter Event Processing Engine (SMEPE)
GeographicalInformation System (GIS)
Figure 5
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 8
(CIS), enterprise asset and work management
system, field service management system and
geographic information system (GIS) – will
maintain historical as well as near-real-time
data such as ticket patterns, actual damages,
ERT and average restoration times. The fil-
tered outage inputs from ADMS and historical
outage-specific data from the data lake will be
then fed into the downstream ecosystem.
Stage 3 (Downstream): Outage Communication Sub-Ecosystem
The goal here is to further refine the informa-
tion, accurately identify affected customers
and then feed outage messages into multiple
communication channels.
In this stage, feeds from the ADMS, data lake,
meter ping and customer preference portal are
processed by the outage notification process-
ing engine (ONPE), which runs on the machine
learning and predictive analytics algorithm.
It is used to accurately confirm the premises
affected based on outage types (such as feeder,
lateral, transformer, etc.) and real-time meter
pings to confirm its findings, deliver estimated
time of restoration (based on the historical
data/status restoration crew data) and provide
alerts based on weather/damage data.
Moreover, this engine utilizes customer prefer-
ences portal data to identify the list of enrolled
customers who wish to be contacted via any
of the preferred communications channels.
Finally, the communication channel platform
retrieves inputs from the ONP engine and pro-
cesses customized notifications to be sent out
via the customer’s preferred channel.
Key benefits from our proposed approach
include:
• Improved prediction of outages using the
OPE. For feeder-related outages, we estimate
an accuracy of more than 95% in identifying
the right set of affected premises.
• Reduction in the number of incoming calls
from customers enquiring about outages,
which will yield monetary benefits. By
proactively communicating to customers
regarding outage status, we estimate that
outage-related inbound calls can be reduced
by at least 50%.
Building a Smart Outage Communication Ecosystem
• Define the overall objectives and goals.
• Evaluate smart grid maturity regarding outage management.
• Create the business case justification.
• Conduct an impact analysis.
• Strategize tactics to be employed in the pursuit of each goal.
• Align technologies with strategic business objectives.
• Analyze vendor responses and facilitate scoring exercises.
• Optimize design of core components in the upstream outage management ecosystem and downstream outage communication ecosystem.
• Create seamless integration with key systems or applica-tions.
• Build key components of the outage prediction engine and outage notification processing engine.
• Integrate core components for outage detection and prediction of the interrupting device.
• Build predictive models and machine learning algorithms to predict future outages.
• Use optimization techniques for complex scenarios with multiple constraints and data to recommend optimal solutions.
• Build historical data on various patterns.
ASSESSMENTSTRATEGY &PLANNING DESIGN EXECUTE OPTIMIZE
Figure 6
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 9
• Accurate real-time outage alerts sent to the
right set of customers based on their pre-
ferred communication channel.
• Increased customer satisfaction.
• Enhanced compliance with established regu-
lations.
As our experience suggests, energy utilities
are not very well equipped with the required
resources and skills to build complete smart
outage communication ecosystems concurrently.
Figure 6 (previous page), however, can help utili-
ties to realize that vision.
LOOKING FORWARD
The concept of smart power outage communica-
tion is at a nascent stage in the utility industry.
By making proactive outage communication a
central part of the business strategy and by pro-
viding accurate, detailed and real-time outage
statuses through customers’ preferred channels,
energy utilities can improve overall customer
experience, customer satisfaction and opera-
tional efficiency. Moreover, energy utilities can
reduce the cost of operations by limiting out-
age-specific customer calls and can also comply
with regulations.
Energy utilities can reduce the cost of operations by limiting outage-specific customer calls and can also comply with regulations.
Cognizant 20-20 Insights
Applying Predictive Analytics to Deliver Smart Power Outage Communications | 10
Pankaj Galdhar Consulting Manager, Cognizant Consulting Energy and Utilities Practice
Vijin Varughese Vallelil Consultant, Cognizant Consulting Energy and Utilities Practice
Abhinav Johnson Consultant, Cognizant Consulting Energy and Utilities Practice
Pankaj Galdhar is a Consulting Manager within Cognizant Consult-
ing’s Energy and Utilities Practice. He has more than nine years of
experience in consulting, business process reengineering, business
development, commercial modeling, digital operations and program
management. Pankaj has led and executed consulting engagements
with multiple energy utility customers. He has extensive experience
in transmission, distribution and customer service, with a focus on
distribution automation/smart grid, asset and work management,
outage management, field service management, billing, payments
and customer experience management. Pankaj has an M.B.A. in
strategy and IT management and a degree in electronics and tele-
communication engineering. He can be reached at Pankaj.Galdhar@
cognizant.com.
Vijin Varughese Vallelil is a Consultant within Cognizant Consult-
ing’s Energy and Utilities Practice. He has more than eight years
of experience in IT and the business analysis space consulting in
the implementation of IT systems for the utilities industry. Vijin
has extensive experience in power delivery outage management,
as well as communications, field service management and cus-
tomer care center applications. He holds a master’s degree in
e-business and innovation, and a degree in electronics and com-
munication engineering. Vijin can be reached at Vijin.Varghese@
cognizant.com.
Abhinav Johnson is a Consultant within Cognizant Consulting’s
Energy and Utilities Practice. He has over eight years of experi-
ence in providing consulting services in implementing IT systems
for utility companies across the U.S. and UK. Abhinav has exten-
sive experience in field service management, smart metering,
smart grid automation and health monitoring, settings integration
and distribution load forecasting. He has an M.B.A. in operations
and a degree in mechanical engineering. Abhinav can be reached
ABOUT THE AUTHORS
World Headquarters
500 Frank W. Burr Blvd.Teaneck, NJ 07666 USAPhone: +1 201 801 0233Fax: +1 201 801 0243Toll Free: +1 888 937 3277
European Headquarters
1 Kingdom Street Paddington Central London W2 6BD EnglandPhone: +44 (0) 20 7297 7600 Fax: +44 (0) 20 7121 0102
India Operations Headquarters
#5/535 Old Mahabalipuram RoadOkkiyam Pettai, ThoraipakkamChennai, 600 096 IndiaPhone: +91 (0) 44 4209 6000Fax: +91 (0) 44 4209 6060
© Copyright 2017, 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.
Codex 3114
ABOUT COGNIZANT
Cognizant (NASDAQ-100: CTSH) is one of the world’s leading professional services companies, transforming clients’ business, operating and technology models for the digital era. Our unique industry-based, consultative approach helps clients envision, build and run more innova-tive and efficient businesses. Headquartered in the U.S., Cognizant is ranked 205 on the Fortune 500 and is consistently listed among the most admired companies in the world. Learn how Cognizant helps clients lead with digital at www.cognizant.com or follow us @Cognizant.