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.


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

https://www.chartwellinc.com/



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.




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




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



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



HeadEndSystem

Social MediaEvents –Twitter,Facebook, etc.

Validate Simulate Assess

Forecast & Actual Data


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


Predictive Asset Failure/Fault Detection Engine (PAFDE)

TextPreprocessing

TextTransformation

PatternDiscovery

Social Events Data



Social Event Processing Engine (SEPE)

Filter Process Identify

AMI + GIS Data



Smart Meter Event Processing Engine (SMEPE)

GeographicalInformation System (GIS)

Figure 5



(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


• 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.



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

at [email protected].

ABOUT THE AUTHORS

mailto:Pankaj.Galdhar%40cognizant.com?subject=

mailto:Pankaj.Galdhar%40cognizant.com?subject=

mailto:Vijin.Varghese%40cognizant.com?subject=

mailto:Vijin.Varghese%40cognizant.com?subject=

mailto:Abhinav.Johnson%40cognizant.com?subject=

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.

http://www.cognizant.com

https://twitter.com/Cognizant

cognizant 20-20 insights · pdf filepredictive analytics to deliver smart power outage ......

Documents