Data Analytics For Advanced Metering Infrastructure A Guidance Note For South Asian Power Utilities
Data Analytics For Advanced Metering Infrastructure A Guidance Note For South Asian Power Utilities
Data Analytics For Advanced Metering Infrastructure A Guidance Note For South Asian Power Utilities
November 2018
Disclaimer
This document has been prepared for the sole purpose of sharing the results of a global AMI survey
and resulting insights related to the AMI strategy for India and South Asia. This does not endorse
individual vendors, products or services in any manner. Therefore, any reference herein to any vendor,
product or services by trade name, trademark, manufacturer or otherwise does not constitute or imply
the endorsement, recommendation or approval thereof.
Copyright © 2018
The International Bank for Reconstruction and Development
THE WORLD BANK GROUP
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.
The material in this publication is copyrighted. However, it may be reproduced in whole or in part and
in any form for educational or non-profit uses, without special permission provided acknowledgment of
the source is made. Requests for permission to reproduce portions for resale or commercial purposes
should be sent to the Manager, Energy and Extractives Global Practice (South Asia) at askeex@
worldbankgroup.org. The World Bank encourages dissemination of its work and will normally give
permission promptly. The Manager would appreciate receiving a copy of or link to the publication that
uses this material for its source sent in care of the address listed.
All images remain the sole property of their source and may not be used for any purpose without
written permission from the source.
Table of Contents
Acknowledgements v
Abbreviations vi
Executive Summary ix
1. Data Analytics in Power Utilities 1
1.1 Overview of Data Analytics 2
1.2 Data Analytics: Scope and Application Points in the Electricity Utility Industry 4
1.3 Meter Data Flow in MBC Process and Analysis System 6
1.4 Core MBC Analytics 8
1.5 Advanced Revenue Cycle Management Analytics 18
1.6 Key Issues/Risks Associated with Data Analytics 22
1.7 Benefits of AMI over AMR for Improved Data Analytics 23
2. Procurement and Deployment of Analytics Systems 28
2.1 Data Analytics System Architecture 28
2.2 Approaches for Procurement and Implementation of Analytics Systems 31
2.3 Technical Specification for Analytics Systems 34
2.4 Functional Requirements of Advanced Analytics System 36
2.5 Sizing and Typical Bill of Quantity for Analytics System 38
References and Links 40
Appendices 41
Appendix A. Description of Key Technology Terms 41
Appendix B. Meter Reading Quality Checks (RQCs) – 2nd Level: Screenshots 43
Appendix C. Billing Quality Checks (BQCs) – 3rd Level: Screenshots 50
Appendix D. Meter Data Exception-Generation Checks: Screenshots 56
Appendix E. MIS Reports: Screenshots 75
Appendix F. AMR Data Analytics – Use Cases: Screenshots 89
Appendix G. Billing and Collection Data Analytics Use Cases – Screenshots 98
Appendix H. Threshold Values for Exception Generation 106
Figures
Figure 1: Stages of Analytics 3
Figure 2: MBC Process Flow Diagram for AMI and Non-AMI Scenario 6
Figure 3: Architectural Diagram of Billing and Data Analysis System 7
Figure 4: Detailed Meter Data Flow Diagram for Exception Generation 8
Figure 5: Sample Exception Generation Check Output 14
Figure 6: Data Analytics System Conceptual Architecture 28
Figure 7: Transition Stages for South Asian Utilities 31
Figure 8: Consolidated Procurement Model 32
Figure 9: Component-wise Procurement Model 33
Figure 10: Software-as-a-Service Procurement Model 34
iv Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Acknowledgements
This report was prepared by a World Bank team comprising Gailius J. Draugelis, Lead Energy Specialist
(co-Team Lead); Rohit Mittal, Senior Energy Specialist (co-Team Lead); Amol Gupta, Energy Specialist
(co-Team Lead); Phillip Matthew Hannam, Energy Economist; and Neetu Sharda, Program Assistant.
The team is grateful to World Bank Group colleagues who peer reviewed the report: Kwawu Mensan
Gaba, Lead Energy Specialist/Global Lead, Power Systems, World Bank; Kelli Joseph, Senior
Energy Specialist, World Bank; and Peter Mockel, Principal Industry Specialist, International Finance
Corporation. The team is also most grateful to Chris Marquardt for his invaluable editing of the final
report. Demetrios Papathanasiou, Practice Manager, Energy and Extractives Global Practice (South
Asia unit), also provided much appreciated guidance and advice for this project.
The report is prepared based on background study undertaken by Deloitte Touche Tohmatsu India LLP
team comprising James Thomson, Principal, US (Project Director), Michael Danziger, Managing Director,
US (Team Leader); Anujesh Dwivedi, Partner, India (Team Leader); Ajay Madwesh, Senior Manager, US;
Pankaj Kumar Goinka, Senior Manager, India; Peter Schmidt, Manager, US; Kyle Webb, Senior Manager,
US; Rohit Deshpande, Specialist, US; Joel Abraham, Consultant, India; which partnered with Tata Power
Delhi Distribution Limited (TPDDL) specifically on aspects of metering, billing and collections analytics
use cases. TPDDL team was represented by Sandeep Dhamija, Deputy General Manager. The Bank
team sincerely appreciates the hard work, dedication and collaborative spirit of the consulting team.
Company surveys that were carried out in the spring of 2018 and technical report preparation were
carried out by Deloitte Touche Tohmatsu under the guidance of the World Bank team. Findings of
the survey and analysis were shared at a workshop in New Delhi on June 8, 2018 with participants
from Bangladesh, Bhutan, India, and Nepal. The report was also informed by an industry stakeholder
consultation in Jaipur, November 2017. The Bank team is deeply thankful to all workshop and consultation
participants for sharing their invaluable insights and comments.
The funding for this report was provided by United Kingdom’s Department for International Development
through the World Bank-managed Trust Fund programs – the South Asia Regional Trade and Integration
Program and the Program for Asia Connectivity and Trade – and the World Bank.
Acknowledgements v
Abbreviations
AMI Advanced Metering Infrastructure
ADMS Advanced Distribution Management System
AMR Automatic Meter Reading
API Application Programming Interface
AT&C losses Aggregate Technical and Commercial Losses
BCS Base Computer Software
BI Business Intelligence
BOM Bill of Materials
BoQ Bill of Quantity
BOOT Build, Own, Operate and Transfer
BOT Build, Own and Transfer
BQC Billing Quality Checks
CAPEX Capital Expenditure
CDW Corporate Data Warehouse
CEA Central Electricity Authority
CMMI Capability Maturity Model Integration
ComEd Commonwealth Edison
COTS Commercial Off-The-Shelf (Software)
CPU Central Processing Unit
CRM Customer Relationship Management
CT Current Transformer
DCU Data Concentrator Unit
DISCOM Distribution Company
DLMS Device Language Message Specification
DMS Distribution Management System
DTR/DT Distribution Transformer
EDA Exploratory Data Analysis
EESL Energy Efficiency Services Limited
ERP Enterprise Resource Planning
ESD Electrostatic Discharge
ETL tool Extract, Transform and Load Tool
Gbps Gigabits Per Second
GHz Gigahertz
GIS Geographic Information System
GPRS/GSM General Packet Radio Service/Global System for Mobile Communications
vi Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
HBA Host Bus Adapter
HDD Hard Disk Drive
HES Head-End System
HHU Hand-Held Units
HT/LT High-Tension/Low-Tension
IBM International Business Machines Corporation
Ib Base Current
IEEE The Institute of Electrical and Electronics Engineers
IoT Internet of Things
IP Internet Protocol
IPDS Integrated Power Development Scheme
IS Indian Standard
IT Information Technology
JDBC Java Database Connectivity
kV kiloVolt
kVARh kiloVolt-Amperes Reactive hours
kWh kiloWatt-hour
lasso Least Absolute Shrinkage and Selection Operator (Regression Analysis Method)
LDAP/AD Lightweight Directory Access Protocol/Active Directory
mA milliAmpere
MAF Manufacturer’s Authorization Form
MBC Metering, Billing and Collection
MDI Maximum Demand Indicator
MDMS Meter Data Management System
MIS Management Information System
MoP Ministry of Power, Government of India
MRD Meter Reading Device
MRI Meter Reading Instrument
mT milliTesla
MTD Month to Date
NIC Network Interface Card
NLP Natural Language Processing
NMS Network Management System
NVM Nonvolatile Memory
OBIEE Oracle Business Intelligence, Enterprise Edition
OAC Office Automation Consultants
OEM Original Equipment Manufacturer
OGC Open Geospatial Consortium
OMS Order Management System
OT Operational Technology
Abbreviations vii
PF Power Factor
PFA Predictive Failure Analysis
PoD Proof of Delivery
PPA Power Purchase Agreement
PSU Public Sector Undertaking
RAPDRP Restructured Accelerated Power Development and Reforms Program
RDBMS Relational Database Management System
RF Radio Frequency
RFP Request for Proposals
RQC Reading Quality Checks
RTC Real-Time Clock
SaaS Software as a Service
SAP BW Sap Business Warehouse (Software)
SAP HANA Sap High Performance Analytic Appliance (Software)
SAP IS-U Sap’s Industry-Specific Solution for the Utilities Industry
SAP PAL Sap Predictive Analysis Library (Part of Sap Hana)
SCADA Supervisory Control and Data Acquisition
SAIDI System Average Interruption Duration Index
SI Systems Integrator
SLA Service Level Agreement
SPSS Statistical Package for the Social Sciences
SQL Structured Query Language
SSD Solid-State Drive
SSH Secure Shell
T Tesla
TB Terabyte
ToD Time of Day
ToU Time of Use
TPDDL Tata Power Delhi Distribution Limited
UDAY Ujwal Discom Assurance Yojana
UoM Unit of Measurement
UPF Unity Power Factor
VAR Volt-Ampere Reactive
WIMS Work Information Management System
XML Extensible Markup Language
YTD Year to Date
viii Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Executive Summary
The power industry in South Asia1 is on the cusp of a transformation driven by technological
advances, decreasing energy intensity, heightened environmental awareness, and evolving customer
expectations. Power distribution utilities in these countries are facing various challenges including high
levels of Aggregate Technical and Commercial (AT&C) losses, increased energy theft, poor customer
services and operational transparency, inefficient load management, and unreliable power supply.
Governments in most South Asian countries are helping utilities by implementing various schemes to
improve their power sectors. In particular, the widespread and successful adoption of smart metering
in advanced economies over the last decade has encouraged South Asian policy makers to take an
increasing interest in smart metering systems in hope that they can address some of the chronic issues.
Now, with high-level policies in place and utilities keen to adopt smart metering, funding requirement
and implementation challenges remain the bottlenecks to mass deployment.
A recent World Bank–funded study, Advanced Metering Infrastructure and Analytics Guidance Study
for South Asian Utilities, carried out in 2018, developed guidance, based on user experience, on
the deployment and operation of Advanced Metering Infrastructure (AMI) and analytics systems by
electricity distribution utilities in India and other South Asian countries. The guidance is intended for
the ready reference of policy makers and utility managers.
The study was divided into two reports. The Survey of International Experience in Advanced Metering
Infrastructure and its Implementation, published separately, covers international best practices
regarding the end-to-end deployment of AMI systems, including such areas as main functions,
procurement options, and the organizational or functional changes needed to implement AMI-enabled
business processes.
The present report, Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South
Asian Power Utilities, is structured as a guidance note to assist utility managers in taking up data
analytics systems to realize the full potential of AMI. The report has two main parts, as follows:
1 The World Bank’s South Asia region comprises Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka. See
https://www.worldbank.org/en/region/sar/overview.
Executive Summary ix
Chapter 1 provides an overview of the use of data analytics by power utilities, including conceptual
architecture for system deployment, meter data flow in Metering, Billing and Collection (MBC)
processes, and data analytics systems. The chapter also covers in detail the core MBC analytics on
data exception checks, meter reading, and billing quality checks. Further, various use cases related
to advanced analytics for revenue management are also elaborated. The discussion of risks and
other issues associated with data analytics systems– including the benefits of AMI over Automatic
Meter Reading (AMR) for improved data analytics – will be particularly insightful to utilities that have
already initiated or are considering an implementation in the near future.
Chapter 2 describes the transition phase utilities go through while adopting data analytics
systems, including an explanation of procurement and implementation models for data analytics
systems. With a focus on the South Asian context, the chapter outlines the basic technical and
functional specifications of analytics systems, sample sizing, and typical Bill of Quantity (BoQ) for
the implementation and qualification requirements of potential system integrators.
The report was prepared by AMI and utility specialists from Deloitte as well as Tata Power Delhi
Distribution Limited (TPDDL), one of the first utilities in India to adopt AMR and AMI technology. Over
the last 15 years, TPDDL has gained expertise in AMR technology and core data analytics, and it has
recently begun implementing AMI and advanced data-analytics systems.
The World Bank conducted a South Asian stakeholder consultation on AMI implementation issues in
India on November 29, 2017, during the International Symposium to Promote Innovation and Research
in Energy Efficiency (INSPIRE-2017). These stakeholders provided validation and guidance on the
scope of the study. The findings of the study were shared on June 8, 2018, at a regional workshop on
AMI titled “International and Regional Approaches to Implementation and Data Analytics in New Delhi.”
Feedback from workshop participants is reflected in the two reports.
Both reports were prepared under the guidance of a team from the World Bank’s Energy and
Extractives Global Practice that was based in the World Bank’s offices in Washington, D.C., United
States, and New Delhi, India.
x Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
1. Data Analytics in
Power Utilities
Although Information Technology (IT) solutions have benefited core power utility operations for
decades, most utilities in South Asian countries have remained weak in using IT to improve operational
efficiency. Nevertheless, a handful of utilities in South Asia – especially the privately owned ones – have
not only deployed best-in-class IT solutions in their revenue management and network operations,
but have also implemented advanced solutions that leverage data to make decision making insightful,
fast and more accurate.
In India, the use of operational IT by state-owned power utilities has happened on the back of two
schemes promulgated by the Government of India (GoI) – the Restructured Accelerated Power
Development and Reforms Program (RAPDRP) and Integrated Power Development Scheme (IPDS)–
which introduced IT solutions in core utility operations throughout the country.
In November 2015, the GoI launched the Ujwal DISCOM Assurance Yojana (UDAY) program, which
seeks to spur the financial turnaround and revival of power distribution companies (DISCOMs) by
improving their operational and financial efficiency. As part of the program, participating DISCOMs
must install smart meters for all customers consuming more than 200 kilowatt-hours (kWh) per month
by December 2019.
In this context, Energy Efficiency Services Limited (EESL)2 has undertaken the procurement process
for five million smart meters for the states of Haryana and Uttar Pradesh. The key feature of EESL’s
procurement approach is build-own-operate-transfer (BOOT) – that is, a pay-as-you-benefit approach.
This large, centralized procurement initiative has yielded aggressive pricing by bidders, thereby
reducing the cost of smart meters for power utilities. This trend has encouraged several other utilities
in other South Asian countries to go beyond pilots and initiate sizeable smart-meter deployment
programs.
2 EESL is a “super” Energy Service Company (ESCO) that acts as the resource center for capacity building for DISCOMs, Energy Regulatory
Commissions (ERCs), State Development Authorities (SDAs), financial institutions, ESCOs, and so on. Founded in 2010 by the GoI, EESL
is currently implementing the largest energy efficiency portfolio in the world.
The Smart Grid concept and its constituent technologies have added new dimensions to electricity
distribution systems by providing higher-resolution data for enhanced network operations. Advanced
Metering Infrastructure (AMI), which includes smart metering, is considered the key to enabling
active/proactive decision making in utility operations– replacing the existing, largely passive mode of
operations.
By enabling automation at the last node of a utility’s network, smart meters serve as a gateway
between a utility and its customers. It serves as more than just a cash box for the utility, however. In
addition to helping utilities safeguard their revenue interests while helping customers optimize their
consumption, smart meters provide a host of advantages in network operations by boosting utility
managers’ situational awareness of conditions along the “last mile” (that is, the final leg of delivery to
retail end-users, or customers).
Using the “big data” generated by smart meters and systems based on the Internet of Things (IoT),
utilities will have a vast array of opportunities to improve operational and commercial efficiency using
insights gained from data analytics that may not have been previously feasible.
Currently, most utilities in South Asia are focused on addressing growing consumer demand and the
related issues of access, availability, quality and affordability of power supply. They are just beginning
to explore data analytics solutions that will eventually generate tremendous value in terms of efficiency
improvements, enhance customer services and improve financial performance.
To unlock the full value of AMI, each utility will need to implement a robust advanced analytics system
capable of processing data beyond basic meter-to-cash functionality. To do this, it will first need to
determine its business needs and establish a strategy. Once its strategy is aligned with its business
architecture, the utility can decide on the types of analysis it intends to undertake and the benefits that
can be derived.
Data analytics can be broadly grouped into four stages, as shown in Figure 1.
Descriptive analytics helps in understanding what has happened in the past. Examples include
meter ageing analysis; determining the “percentage of average billing” trend across months; trends
of arrears outstanding across months, consumer categories, and so on; network optimization/load
balancing; interconnected feeder analysis; predicting potential defaulting consumers; and theft
analytics.
2 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Figure 1: Stages of Analytics
Difficulty in Implementation
Value Delivered
Diagnostic analytics helps in understanding what has led to success or failure in the past. Examples
include sanctioned load violation analysis, network failure analysis, and theft analytics.
Predictive analytics helps in determining the likelihood of occurrence of an event in future based on
past data patterns. Examples include footfall optimization at payment centers, energy forecasting,
predictive maintenance, predictive fault analysis, and network-asset management.
Prescriptive analytics helps in identifying steps to be taken in future to meet the desired
objectives. Examples include energy optimization and Volt-Ampere Reactive (VAR) control, call
center efficiency, meter reading, and billing complaints analysis.
By its very nature, development of analytics is complex since it aims at integrating structured,
unstructured, and time-series data, then aligning the data with system-generated events and alarms.
Within a utility context, analytics can be defined as the process of converting data from smart grid
sensors and devices by integrating it with a variety of related data sets (including data from operational,
non-operational and external systems) to develop models that predict and/or prescribe the next best
action – thus creating deep situational awareness.
A utility’s external data sources may be based on data that is structured (cellular network data, for
example), unstructured (social media feeds and customer questions, for example), part of a time series
(outside temperature, for example) or generated by an event (lightning, for example). Integrating
the utility’s data and information with these external data sources leads to a deeper understanding
of the causes of issues affecting operational, customer and business performance. Using AMI-type
machine-to-machine communications, consumer-facing web portals, in-home devices, visualization
tools, modelling software, and even spreadsheets results can then give utility managers an enhanced
“situational awareness” – that is, a more thorough and precise understanding of operations.
Analytics built on data from smart meters is valuable in a variety of ways to a utility. The utility landscape
surrounding, and business cases for, adopting AMI analytics vary significantly by region, as well as by
utility ownership structure. This section outlines a broad set of use cases that are now possible using
AMI data. This section also focuses on the “state-of-the-art” as well as the “art of the possible” to show
how utilities can harness AMI data – using both predictive and prescriptive analytical models – not only
to realize greater value, but also to shift to the new utility model that regulators are pushing towards.
The specific drivers for the development of analytical models from AMI vary based on geography,
regulations and underlying demographics. When compared with the cost of implementing AMI itself,
the cost of implementing analytics is quite low and typically has a more immediate payback.
The successes in implementing analytics by industry leading utilities using data from AMI systems have
had a cascading effect in that they have helped other utilities – especially smaller and mid-market
utilities, which tend to be more risk-averse – justify their own accelerated AMI deployments. Globally,
many early adopters of AMI went through multiple cycles of integrations, data management strategies,
data storage, and tools due to non-standard architectures and solutions. The resulting refinement
of architectures and protocols, lessons on effective use of advanced analytics, and development of
successful business cases will be particularly critical in regions like Latin America and Asia, where
operational savings by themselves may not justify the implementation costs of AMI.
Analytics can be useful in taking more informed decisions and actions, whether or not a human is in the
loop. For example, protection and control systems that require sub-second responses will be machine-
automated, whereas dashboards and reports leveraging business data may operate in a workflow that
involves humans making or approving recommended decisions or actions.
Table 1 summarizes the main areas in which data analytics has been applied in the electricity utility
industry.
3 A Head-End System (HES) is hardware and software that receives meter data sent to the utility. Head-end systems may perform a limited
amount of data validation before either making the data available for other systems to request or pushing the data out to other systems.
4 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Application Areas Analytics Process
the feeder’s capacity and reliability can be performed. The ability to analyze voltage sags and
swells at an individual consumption point allows a utility to address capacity or design issues.
Smart meter data is immensely valuable in assessing the load on the transformer (peak and
nominal), which allows the utility to plan transformer upgrades and replacement programs
to minimize outages before they occur. Additionally, capacity and network constraints at the
feeder and lateral levels are assessed to a highly granular degree.
Advanced outage With the smart meter “last gasp” capabilities, along with the logged events, outage analysis
analysis and prediction become significantly more accurate. (“Last gasp” refers to a short transmission
a smart meter makes when it loses power, typically to signal the power loss and often to
specify the time and date of the outage.) When a major weather event is expected, analyzing
the potential extent of the outage, as well as possible nested outages, becomes a highly
effective tool for the utility. This helps maintain a low System Average Interruption Duration
Index (SAIDI) and serves customers by restoring electricity faster following an outage.
Demand segmentation Total demand calculated based on consumption can be analyzed by disaggregating the
demand into multiple cross-sections or segments. The interaction of these segments at
different prices, loads, temperatures, etc. over time improves the ability to predict demand.
Advanced unit In situations where the wholesale markets allow smaller renewable and storage-based
commitment models generation, and where demand aggregators bid into the Capacity and Day-Ahead markets,
the ability to track the performance of these generation sources relative to demand becomes
critical. For example, to address balancing and security constraints while following regulatory
mandates on renewable generation, grid operators use granular demand data to predict
shortfalls ahead of time and avoid buying generation through the costlier real-time and
ancillary markets. To generate near-real-time and predictive demand curves, grid operators
use historical and smart-meter data.
Power quality Utilities have long sought conservative voltage reduction and improved management of
voltage profiles down to the level of feeder lines. With the smart meters capturing voltage
profiles, voltage sags and voltage swells at the consumer level, feeder performance –
including feeder quality and intermittent outages – can be captured with high fidelity. In
instances where the voltages can be managed to the lower limits specified by the reliability
regulator through operating the tap changers at the substation, a utility can optimize
generation.
Technical and non- Distribution feeders can be sectioned to increase greater reliability and feeder reconfiguration
technical loss analysis by using AMI data to measure the technical and non-technical losses more accurately. The
AMI data provides a true measure of the total losses in the distribution feeder and guidance to
target diversion and energy theft. Energy theft models are then used for revenue assurance.
Advanced switching As the distribution feeders are sectioned to increase greater reliability and for feeder
reconfiguration, AMI data can be used to confirm that switching operations have occurred
correctly, as well to validate the network model.
Asset performance Using data from Supervisory Control and Data Acquisition (SCADA) systems, utilities can
analyze the performance of network assets at the substation level, and sometimes at the
feeder level. With AMI data, this analysis can be conducted along the “last mile”.
Customer preferences, As customers become active participants in the energy equation, understanding precisely
customer propensity what drives their behavior becomes critical. AMI data facilitates this. Understanding the
and incentives model consumption patterns of each individual consumer and comparing them to similar consumers
(within the applicable regulatory boundaries) allows for the analysis of the preferences and
patterns of a class of consumers.
There are two scenarios. In the Advanced Metering Infrastructure (AMI) Scenario, the meter data is
directly received at the HES4 if the meter is coupled with a data collector, with the remote meter data
transfer taking place through wired or wireless communication.
In the Non-AMI Scenario the meter reader carries a hand-held Meter-Reading Device (MRD) in the field
to record meter data and run some primary validations on the meter data based on the checks built
into the device. The data is then uploaded to the server.
Figure 2 describes the process of meter reading and billing in both AMI and Non-AMI Scenarios.
In both scenarios, the following further checks are done to ensure that the bill delivered to the
consumers is correct in all respects:
Meter Reading Device (MRD) on-site meter-read check: Here a primary check is performed
to ensure that the meter reader reads the correct meters and provides correct information.
This check is performed only in the non-AMI scenario, where manual meter reading is done
monthly.
Figure 2: MBC Process Flow Diagram for AMI and Non-AMI Scenario
SAP SAP
Planning & Data Downloading in Base Meter Reading Reading Data uploaded
Scheduling MRD from Billing Server through MRD: in Billing Server
1st Level Check
AMI Scenario
Billing Quality Check: Printing & Invoicing Bill Distribution
Non AMI Scenario 3rd Level Check
4 The HES is sometimes also referred as the Meter Data Acquisition System (MDAS) in the case of AMR.
6 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Figure 3: Architectural Diagram of Billing and Data Analysis System
Vendor
Billing Data
1
RS 232/ Billing Data
Vendor 1 Meters Optical Port
Database Server
Full / Analysis Data
Comm. Modems
.rr3/.rb
3 format
Data Analytics
Vendor
APIs Application
2
Optical / Magnetic
RS 232/ Data Storage
Optical Port Devices (CD/ DVD)
Vendor 2 Meters
APIs
RS 232/
Optical Port
Vendor 3 Meters
Reading Quality Check (RQC): This ensures that the meter data captured are accurate prior to
billing in terms of consumption.
Billing Quality Check (BQC): This ensures the quality and correctness of bills in terms of amount.
Further meter data exception-generation checks are performed on the full sets of meter data in
both AMI and Non-AMI scenarios based on the frequency of meter reading and the utility’s business
requirements.
A typical technical architectural diagram showing meter data flow for billing and analytics is shown in
Figure 3.
The diagram depicts meter data flow from smart meters to the billing engine and the data analytics
application servers. A DISCOM may have meters from different vendors deployed at the consumer
premises.
The interoperability of meters from different vendors with HES system is an issue for AMI/AMR
deployment. This is because each of the various vendors uses its own proprietary protocols for meter
data exchange with HES system. Although the Bureau of Indian Standards has formulated the standards
for smart metering (IS 16444), proprietary protocols remain in use for AMR meter data exchange from
different vendors. To resolve this issue, the DISCOMs need to have HES/BCS5 software from respective
meter vendors in their HES server to poll meters using proprietary Application Programming Interfaces
(APIs), as shown above in Figure 3. The meter data received in different formats by the HES/BCS system
will be parsed and converted to standard .xml format, which can then be directed to corresponding
Billing
Billed
Billing Data
Billing Engine BQC
Reads
RQC
• Hardware
NOT OK Problem
Billing Data H OK
Rectification
• Enforcement
E
Manual Checks NOT OK Team
Analysis Data
S NOT OK OK
Smart Meter
General Analysis
Engine (Exception Checks) OK Data Base
Analysis Data
billing and analytics applications. However, this solution has a higher implementation cost. If the
standardization of AMI meter data exchange can be mandated for the vendors through a policy-level
decision or in the tender documents for procurement by the utility, the cost of implementation can be
further reduced.
Figure 4 shows a detailed diagram of meter data flow, along with its application at different steps in
MBC cycle and data analysis to “generate” (i.e., check for and list) exceptions.6
Smart meters can provide billing reads and full data (billing reads, instantaneous data, event data, load
profile, and so on) separately, in line with utility requirements. Billing reads are acquired from meters
each month, and analytics is done on the full data set acquired from meters at custom frequencies set
by the utility for different consumer segments.
After each level of checks, the exception cases are forwarded for manual analysis, which may include
on-site checks. Verified data is forwarded to a database for storage and future reference.
8 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Table 2: List of On-Site Meter Read Checks using MRD
First-level (primary)
Function
Exception Check
Reading check Checks if current meter reading is less than previous reading
Zero consumption Checks if the meter consumption is absolute zero
Drop in consumption Checks if the current consumption is less than the average consumption/threshold value
High MDI*/sanctioned load Compares the current-month MDI reading against the sanctioned load and previous
month’s reading
Meter-unread scenario Mandatory input in each MRD device that captures the reason the meter reading was not
performed – such as a locked house or faulty meter
* MDI = Maximum Demand Indicator.
If these checks yield flags, the meter reader is directed to take a photograph of the meter or to re-read
the meter at the site.
Table 3 lists the RQCs a utility typically performs before generating bills. Sample illustrative screenshots
of RQCs are provided in Appendix B.
Benefits/Use
Function What It Looks for Electrical Parameters Considered
Cases
Low consumption Checks for variations in the Prevent revenue Meter data: kWh consumption
consumption trend of a particular leakage Consumer master data: Type of
consumer (with reference to low institution and sanctioned load with
value thresholds for each category). previous MDIs
High Maximum Checks for variations a consumer’s Prevent revenue Meter data: kW value
Demand Indicator MDI trend with reference to his/her leakage Consumer master data: Type of
(MDI) historical meter reading data and institution and sanctioned load with
category threshold value. previous MDIs
Reading reversed/ Checks for erroneous readings on Prevent revenue Meter data: kWh and kVAh
wrong reading meter energy consumption – such as leakage; verify Consumer master data: Type of
very high or negative consumption. meter data institution and sanctioned load with
previous MDIs
Low power factor Checks for power factor value below Prevent revenue Meter data: Power factor
the threshold defined by utility. leakage Consumer master data: Type of
institution and sanctioned load with
previous MDIs
Contract demand Compares the demand value from Prevent revenue Meter data: MDI (kW)
meter reading data against the leakage Consumer master data: Type of
consumer master database. (There institution and sanctioned load with
is a penalty for higher demand previous MDIs
compared to contract demand.)
Non- Checks to see if: Prevent revenue Meter data: kWh and kW
reprogramming Tariff slab information from meter leakage; verify Consumer master data: Type of
data does not match consumer data institution and sanctioned load with
master data previous MDIs
Meter reset-date info from meter
data does not match consumer
billing master data
Meter is not programmed to store
data in Time-of-Day (ToD) slabs;
HES system after receiving data
identifies consumer to be billed
on ToD
Table 4 provides a typical list of BQCs a utility performs before generating bills. Sample illustrative
screen shots of BQCs are shown in Appendix C.
10 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
1.4.4 Further Checks Using Rule-Based Analysis (further meter data exception
generation checks)
In addition to the aforementioned checks, utilities can use various other exception-generation checks
to identify revenue leakage, theft, and so on. This analysis can be done on a predefined frequency (i.e.,
at various predefined intervals) depending on the category of consumers. The raw meter data received
is subjected to rule-based analysis for identifying anomalies.
These checks are oriented toward either consumers or utilities. Consumer-oriented checks generate
flags when, for example, monthly bill or energy consumption exceeds the average value range; these
checks can be done hourly or daily depending on the resolution of the data received. Utility-oriented
checks includes checks on data for revenue assurance, system-condition monitoring, and so on.
Data analysis rules could be selectively applied to an individual metering node or groups of metering
nodes or to channels common to different metering nodes. A list of exception checks that a utility could
follow based on various data received from meter is provided in Table 5.
Type of Electrical
Consumer
No. Exception Function Benefits/Use Cases Parameters
Categories
Check Considered
1 Consumption Checks that consumption is inline Identification of: Kilowatt- All categories
comparison with historical trend and threshold Tamper condition hours (kWh)
values Meter failure, under-
recording
2 Voltage failure Checks that system voltage System condition check Phase All categories
values are in line with tolerance Identification
of tamper voltage
and threshold values condition
3 Assessed Checks whether the calculated Identification of: kWh All categories
consumption load factor is in the predefined Tamper condition
threshold load-factor range for Situation where
the particular consumer category consumer is not running
load for maximum
hours, causing a loss to
the utility
4 Data corruption Checks the value of maximum Identification of: kW/kWh All categories
demand indicator (MDI)/ Meter failure
consumption data/digits recorded Data downloading
against the threshold values issue
5 Non-Volatile Identifies and records meter Identification of: Meter events All categories
Memory7 (NVM) memory failures and raises “NVM Tamper condition data
failure failure event” flags Meter failure
6 Power failure Checks two threshold values: Identification of: Phase All categories
Number of power failures: The System condition voltages
cumulative total number of Tamper condition
power failures should be less
than the threshold value defined.
Time of power failure: The
cumulative power failure
duration should be less than
threshold value defined.
7 Non-volatile memory (NVM) is a type of computer memory that has the capability to hold saved data even if the power is turned off.
9 CT overload Checks whether the current value Identification of: Phase current All categories
is above the defined threshold Meter damage through
value overloading
Sanctioned load violation
10 Current missing Checks whether the phase Identification of: Phase current All categories
current is zero or less than the Meter failure: under-
specified threshold value recording
Tamper conditions
11 Power Factor Checks whether the PF data Proper asset/network Power factor All categories
(PF) are within the threshold range management
defined for Low and High PF Checks for healthiness
of meter and meter data
recording accuracy
12 Manual reset Checks whether the meter reading Identification of: Meter events All categories
reset duration is other than the Tamper condition data
historical value or predefined Meter failure
value (a flag event is raised
whenever the meter is reset in
between the billing period)
13 Real-Time Clock Checks that the meter’s clock Identification of: Meter events All categories
(RTC) failure accuracy is in line with the HES Meter failure data
clock time.
14 Internal ratio Cross-verifies the CT ratio value Identification of: CT Ratio All categories
at site with the consumer master Tamper condition
data value Non-updating of data
from site
15 Magnet tamper Checks for meter tamper-event Identification of: Magnet Flag/ All categories
flags (the meter records a flag Tamper condition Magnet
event whenever it senses the Tamper
presence of a magnetic field; the
meter starts recording high Imax
values than rated value for longer
duration and this value is checked)
16 Load unbalance Checks whether the phase Identification of: Phase current All categories
current values in three phases Tamper conditions (stringent for
are unequal and whether the Meter failure: under- HT consumers
value of each exceeds the stated recording based
tolerance range Load unbalancing by on utility
consumer requirement)
12 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Type of Electrical
Consumer
No. Exception Function Benefits/Use Cases Parameters
Categories
Check Considered
17 Cover open Checks for any cover-open event Identification of: CoverOpen All categories
flags (the meter records each time Tamper condition Flag/Cover
its cover is opened and flags the Open Tamper
event for notification)
18 Potential missing Checks cases where the current Identification of: Phase All categories
with load value is recorded but with voltage Tamper condition voltage
running value as zero Meter failure: under- Phase current
recording
19 Missing data/data Identifies instances of missing Identification of: Meter billing All categories
download gap data on periodic or on-demand AMI component failure data/full data
polling Tamper conditions
20 ToD (time of day) Checks whether the meter Identification of: Kilowatt- Consumer
consumption has recorded the ToD slab Revenue leakage hours (kWh) category
check consumption for the particular in different having TOD
consumer category ToD slabs rates
21 Meter pulse Checks that the number of pulses Identification of: Physical All categories
overflow check does not exceed the maximum Meter failure verification to
threshold (the meter sets a Pulse check Pulse
Overflow flag when the energy Blinking
consumption in a given interval
exceeds the range of the interval)
22 Test mode check Checks whether data was Meter data validation Meter billing All categories
collected while the meter was in data/full data
test mode
23 Energy sum Checks the total cumulative con- Identification of: kWh All categories
check sumption of incoming data against Tamper conditions
the total consumption of various Meter failure
slots over the same time period Energy Auditing
24 Interval spike Checks incoming data to identify Identification of: kWh All categories
check intervals with evidently high usage Tamper condition
relative to surrounding intervals System fault
25 Unit of Checks to ensure that the Unit Meter data validation All
meter data All categories
Measurement of Measurement (UoM) of the parameters
(UoM) check incoming data matches the UoM
specified in the meter’s profile.
26 Usage of Checks and validates data if Identification of: kWh/kVAh All categories
inactive meters consumption is shown for inactive Meter failure
meter Tamper conditions
27 Negative Checks whether the meter Identification of: kWh/kVAh All categories
consumption reading is lower than the last Meter failure
checks meter reading
28 kVARh check Identifies intervals where the Identification of: Activeload Categories
reactive load (kVARh) is present Tamper conditions (in kWh) and having kVARh-
and the active load (kWh) is not. Meter failure reactive load based billing
(in kVARh)
Current-Missing Check
The data analytics system checks whether the phase current is zero (or less than the specified threshold
value) in the meter data received under each time stamp. The following screenshot shows sample output for
the system-generated exception case:
The aforementioned checks are the basic data-exception checks that can be performed on meter data
in non-AMI and AMI scenarios for all consumer categories. However, the actual number and frequency
of checks will depend on such factors as the number of parameters received, the requirements of the
various consumer categories, and the utility’s business requirements and focus areas.
An illustrative example of frequency of these exception checks for various categories of consumer is
provided in Table 6.
In the AMI scenario, the volume of data varies with the frequency of meter polling and number of meter
parameters recorded. As a result of the high resolution of the data received from meters, the number
of exception checks could be much higher in the AMI scenario than in the non-AMI scenario.
14 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
1.4.4.1 Setting Threshold Values
Analysis of meter data is about estimating the quality of data using rule-based assessments, which
trigger exceptions (or alerts) whenever the monitored metric goes above or below a user-defined
threshold. The utility defines exceptions by specifying the threshold values and data to be included in
the check.
The thresholds are set in the analysis software system based on factors such as the utility’s previous
experiences, regulatory guidance, business requirements, power network operational requirements,
and asset performance requirements. The threshold values are updated from time to time based on
the maturity the utility achieves in each analysis it performs and on changes in any of the business
scenarios.
While the meter has inbuilt algorithms to register events according to the thresholds defined, the
utility has to further develop a list of exceptions to monitor or set an additional filtering layer to extract
meaningful outputs based on its requirements.
The detailed threshold values for key exceptions as of IS 16444 standards is given in the
Appendix H.
Table 7 shows a list of MIS reports that should be generated by a utility, along with recommended
frequencies for better performance management.
16 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Bill Processing Step
No. MIS Report Output after which Report Frequency Report Summary
is Generated
12 Due Date MIS Numbers Bill distribution Daily Area-wise due date report
Reports issued to various head cashiers
at collection centers
Helps in resource planning
13 Demand and Amount and Completion of billing Daily Area- and category-wise
Collection Report percentage cycle report providing amount billed,
variations amount collected (including
subsidies, if any) and collection
efficiency – both month-to-date
(MTD) and year-to-date (YTD)
14 Live and Numbers Not applicable Twice a Area-and category-wise report
Disconnected week on the number of consumers
Arrears Report that have defaulted or
disconnected, including their
arrears amount and arrears
ageing (or days delinquent)
15 Disconnection Numbers Not applicable Twice a Area- and category-wise report
Orders Report week providing the total number of
disconnection orders generated
based on arrears ageing (in
days) and arrear amount
Helps in prioritizing
disconnections
16 Billing Status Percentage Invoicing Monthly Area- and category-wise report
Report providing percentage of total
billing done for various billing
statuses (average/provisional/
actual)
17 Amount Billed Numbers Invoicing Monthly Area-and category-wise report
per Consumer providing total amount billed
for Various Bill per consumer for various billing
Statuses Report statuses (e.g., average billing,
provisional billing)
18 Sundry Adjustment Numbers Invoicing Monthly/ Category-wise report providing
Analysis Report Quarterly the number of consumers with
sundry adjustments in terms
of billing amount and units of
energy consumed
19 Billing Behavior Percentage Invoicing Monthly/ Area- and category-wise report
Report Quarterly providing the number of times
consumers have been billed
using different billing methods
(i.e., provisional billing, average
billing compared to the total
billed generated in a year)
20 Payment Behavior Percentage Completion of billing Monthly/ Area- and category-wise report
Report cycle Quarterly providing the percentage of
consumers paying more than
four, eight, etc. times (or non-
paying) in a year
18 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Table 8: List of Use Cases Based on Smart Meter Data Analytics
Functionality
Analytics Software Type of
No. Use Cases Data Accomplished/
Approach Environment Analytics
Expected
1 Energy Weatherforecasting Performsshort-, Short-term Short-, Predictive
forecasting Historical
data of medium- and long- forecasting medium-
energy procured term forecasting. for day-ahead and long-
purchase of power term power
Long-term procurement
forecasting for management
Power Purchase software
Agreements
(PPAs) and
network planning
Mid-term
forecasting for
billing efficiency
improvement
2 Predictive fault Historical outage Predictive Failure PFA extends Rlanguage Predictive
analysis and details on similar Analysis (PFA) power supply platform Prescriptive
network-asset day basis helps to predict stability and
management Electricity potential network quality by going
consumption details failure probability beyond failure
and network loading Compares the detection to
data from distribution actual network predict problems
transformer (DT)- and asset-loading before they occur
level-and consumer- trend with the Life expectancy
level energy-audit expected loading; of assets can be
modules and feeder- if output deviations improved with
monitoring systems (from system- proper loading
Weather forecast defined conditions) and maintenance
data are detected,
Day of week details exceptions/flags
are generated for
analysis
3 Energy High and low Analyzes the Helps ensure R language Prescriptive
optimization voltage in network network voltage power quality in platform
and volt VAR kVARh reads fluctuation trend the network and Statistical
control and identifies prevents network analytics
points where components from using specific
reactive power unexpectedly software
management breaking down
is required by
switching on
capacitor banks,
compensators, etc.
4 Predictive Distribution Performs analytics Maintenance Rlanguage Predictive
maintenance transformer (DT) on DT and feeder- of components platform
and feeder meter loading data are predicted to
readings Sets peak value ensure stable
Power outage data for various asset supply
DT temperature and performance Ensures proper
oil-level information factors and loading of assets
Feeder composition analyzes instances to prevent failures
Maintenance history of violation Instead of periodic
Analyzes DT maintenance,
temperature and utility opts for
outage trends condition-based
maintenance
Sample illustrative screenshots of the above-mentioned use cases are provided in Appendix F.
20 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Table 9: List of Use Cases based on Billing and Collection Data Analytics
Functionality
Software Type of
No. Use Case Data Analyzed Analytics Approach Accomplished/
Environment Analytics
Expected
1 Footfall The number of System
studies on Optimized the Statistical Predictive
optimization consumers at number of consumers footfall leading analytics Prescriptive
at payment various centers onturning up to various to better performed
centers various days of the
counters (to which area consumer structured
month that particular consumer satisfaction query
Consumer due belongs to and where and increased language
dates he is making payments) resource (SQL)
Gives an insight on efficiency. application
workload of cashier at Due date
various counters and optimization
statistics on average Improved
footfall in various Crowd
counters at same time Management
of day/week/month.
2 Predicting Consumers’ Performs analytics on Ledto a R language Descriptive
defaulting previous payment assigning a probability reduction in platform Predictive
consumers defaults to number of consumers arrears and SQL
Consumer credit expected to make improved application
ratings defaults in next billing consumers’
Consumer
cycle. payment
payment behavior behavior
3 Consumer Consumer default Segmentation of Took action to Rlanguage Descriptive
segmentation history consumers by number improve upon platform Prescriptive
on default Arrear amount of times they have the reduction of
payments Bill amount
previously defaulted on defaults
payments Prioritization of
cases for action
4 Migration to Consumer mode Identification of target Cost Statistical Diagnostic
digital/online of payment consumers for migrating optimization analytics Prescriptive
payment Payment behavior to online payment Cycle time performed
of consumers modes reduction
5 Consumer Payment behavior Performs analytics on Improved Rlanguage Descriptive
segmentation Number of times various consumers consumer platform Prescriptive
on payment paid and their payment payment
behavior Bill amount
behaviors. behavior and
Segmented based on retention of
their payment behavior. good consumers
6 Theft Previous Performs analytics and Reduction in Rlanguage Descriptive
analytics enforcement identifying areas where AT&C losses platform Diagnostic
records. power theft is high and Predictive
Event reports and assigning probability on
Prescriptive
notification history. theft occurrence.
Load survey data.
DT Energy audit
data.
7 Meter Connection Complaints registered Fewer defective Statistical Descriptive
reading number, meter for meter reading bills analytics Diagnostic
and billing number, previous and billing generated performed Prescriptive
complaints consumption monthly
analysis pattern, complaint
history, consumer
category
Table 10 depicts some of the issues associated with data analytics, along with possible solutions for
each.
Table 10: Key issues and solutions associated with data analytics
22 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
No. Issues Impact Solutions
3 Unavailability of required skill Higher dependency on Identification
of “champion” teams with the
set within the organization external system integrators organization to take up initiatives and run
and Original Equipment the project
Manufacturers (OEMs)
4 Inaccuracy and inconsistency Wrong data will lead to Establishmentof a system to ensure the
in data received wrong interpretations and accuracy and quality of data
less precise outputs
5 Inefficient field-level actions Desired objective of Establishment of a monitoring mechanism to
based on analytics results analytics will not be achieved ensure there is improvement in operations
over time
6 Communication failures where Theft prediction efficiency Facilitatemanual data downloading and
successful polling of meters will not be optimal hardware trouble shooting in the field
cannot be done Load growth estimation will Stringent service-level agreement (SLA)
be inaccurate conditions for timely rectification of cases
7 Mismatch of DT meter and Ithampers proper energy Reading of DT meters and consumer meters
consumer meter reading dates auditing for loss-level at the same date and time, if possible
calculation
8 Consumer meter is faulty and Theft prediction efficiency Estimation of consumption based on
meter cannot be read will not be optimal previous consumption records
Estimation of consumption by analyzing the
feeder- or DT-level metering data along with
other consumer meter points
9 Ad hoc architectures and poor With ad hoc architectures Core-level standardization has to be brought
integration and without proper IT in to successfully integrate systems
infrastructure in place, inter- Utilities should redesign their enterprise
and intra-system integration architecture to allow flow of data for
will be unable to fully leverage operational and meter-to-cash applications
the benefits of the data
10 Change management Traditional organization Effective use of data analytics requires
structures lead to very removing traditional silos and combining
limited applicability historically disparate groups
of modern systems
implemented
11 Incomplete AMR/AMI metering Difficultyin fetching data Implement manual reading of non-AMR
remotely from non-AMR meters using Hand-Held Units (HHUs)
meters or other conventional methods at fixed
Unavailability of full data for intervals for collecting full data
carrying out energy auditing,
load forecasting, outage
management, etc.
Data analytics use cases are
limited to billing
A brief comparison of the functionality of AMR-enabled meters and AMI meters is provided in Table 11.
Meter reading Automated, with a defined schedule Automated, with a defined schedule (meter
(meter pushing the data to HES; usually pushing the data to HES; frequency can be as
once a month) low as 15 minutes)
Meter can be manually polled on need Meter can be manually polled on need basis
basis (Pull mode) (Pull mode)
Optical Port is available for download of Optical port is available for download of meter
meter reading using CMRI reading using CMRI
Remote reading Through external/retrofit modem Built-in
or modular but integrated with meter
Separate power supply for modem housing
No separate power supply for communication
Data availability through Mainly billing data All data including load profile, events, outages
remote reading
Two-way communication No Yes, event notification
Configuration of Limited Allallowed parameters even support remote
parameters in meter firmware update
24 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Functionality Normal Meter (AMR-Enabled) AMI Meters
Data usage Limited, though data is stored in meter All
data is transferred on near real-time basis
to central system, making it available for
analytics
Communication Monthly Configurable;can be configured for as low
frequency as 15 minutes (i.e., data is transmitted every
15 minutes)
Load control No Yes
Security Varies by meter manufacturer and also Varies by meter manufacturer and also based
based on utility requirement mentioned in on utility requirement mentioned in RFP
RFP (meter standards version) (meter standards version)
Administration Limited to metering and billing operations AMI provides data integrations to multiple
and related services utility business systems in addition to
metering and billing systems like OMS, DMS,
NMS,* load forecasting, etc.
* A Distribution Management System (DMS) is a collection of applications designed to monitor and control a distribution network efficiently
and reliably. An Outage Management System (OMS) is an electronic system developed to identify and correct the outages in the electrical
network. A network management system (NMS) is a system designed for monitoring, maintaining, and optimizing a network.
Key features of smart meters include faster network restoration with outage notification from the
meter, remote meter reading, remote disconnection/reconnection, and faster tamper detection. The
real-time data allows utilities to improve in such areas as volt/VAR control, asset management, and
network planning.
Running advanced data analytics on real-time meter data from AMI systems can benefit energy
utilities in the following ways:
Improved customer targeting and segmentation: There are compounding benefits for utilities
to deploy consumer analytics in their operations. Detailed knowledge of consumers can increase
the lifetime consumer value of residential and commercial users. In an increasingly competitive
market, understanding consumer behavior and the needs of customer groups is key to successful
retention of customers. Through targeting specific customers, companies are able to discern
customer use patterns, which allows them to develop and deliver their message more effectively
to customers.
Improved energy leakage analytics and revenue protection: Smart meters already offer
better tampering resistance, but analytics can help advance the detection of power thefts, thus
preventing revenue losses. Analytics checks monthly consumption and energy bills and identifies
any discrepancies that might indicate fraud.
Effectively target energy efficiency programs and demand response analytics: A deeper
understanding of how consumers use energy empowers energy utilities to develop targeted
energy efficiency programs.
Improved billing and payment options: When more information is known about users, companies
are able to customize billing to user segments based on rate class, program participation, interests,
efforts to conserve energy, and so on.
Reduced network maintenance and better assets management: Metering at Distribution
Transformer (DT) and feeder points provides insights into network and asset-loading conditions.
This enables the utility to improve network and asset utilization while reducing the chance of
network failure.
In India, AMR technology has been partially rolled out in several utilities, covering a large proportion of
their high-value industrial and commercial consumers. Meanwhile, over the last few years, Ministry of
Power (MoP) schemes such as IPDS and UDAY have helped accelerate the adoption of smart meters.
More recently, several utilities have proposed advancing to AMI technology.
One of the key roadblocks preventing the adoption of smart metering technology has been the high
cost of smart meters. With the recent EESL aggregated procurement of smart meters, however, the
cost per meter has fallen by more than 50 percent.
India’s AMR metering standards support Time-of Use (ToU) pricing.8 However, various IoT-enabled
services, DSM initiatives, and other initiatives requiring real - time data will require AMI implementation.
8 With time-of-use (TOU) pricing, electricity tariffs rise during peak demand periods and fall during low-demand periods.
26 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
While a full-scale adoption of AMI would be ideal, utilities may follow a phased approach and gradually
enhance their AMI penetration. The following key technical requirements may be useful in a phased
transition to AMI:
Standardization of communication technologies and data-exchange protocols. In other words,
data-exchange protocols that support AMR should be able to support two-way communication in
AMI systems.
Existing AMR head-end software solutions should be scalable so they can adopt AMI in future.
Utilities should deploy smart meters initially, then make relevant technical upgrades where possible
to enable two-way communication and other operational features (such as DSM initiatives, ToU
tariffs, and remote disconnect/reconnect) that will allow them to adopt AMI in future.
Network Interface Cards (NICs) should be easily replaceable in the field so that without replacing
meters, communication infrastructure can be upgraded to support future technologies, such as the
migration from the third to the fourth generation of mobile broadband Internet (3G to 4G).
However, depending on the industry and overall system architecture, the data sources, integration
requirements, and dashboard components may change. Figure 6 shows an illustrative architecture of
a data analytics system for a power distribution utility.
The details of the various components of an analytics system are summarized in Table 12.
Streaming
ENTERPRISE APPLICATIONS
Asset
Operational Work Order Accounting Management Customer
CRM Systems MDMS ERP Systems Management Service Systems
Systems Systems Systems
28 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Table 12: Details of Various Components of Analytics System
Components Details
Data Sources A utility’s data sources are its operational and back-office systems. In data management terms, the
overall information storage system is the “system of record” – and is sometimes referred to as “the
source of truth,” meaning a trusted data source. Due to the nature of these purpose-built systems, they
tend to be siloed – that is the data tend to be stored in different ways without a common organizational
principle. The three main data sources are as follows:
Analytics hub An analytics hub is different from a data warehouse or a simple “big data” repository. It encompasses
a more modern approach in which “in-memory”* technologies are used in conjunction with “big data”
hubs to address Business Intelligence (BI) and real-time analytical computing needs. This architecture
scales vertically and horizontally and uses computing resources efficiently – in a cloud or a traditional
data warehouse. In a real-time or a near-real-time analytics scenario, analytics are deployed on the in-
memory system or database to compute new results even as new data sets become available. Models
are typically developed in languages such as Python, Scala and R to execute statistical and machine-
learning algorithms.
Data Data integration poses several challenges that can be understood in terms of the “four V’s” of big
integration data: variety, volume, velocity and veracity. The variety of data includes structured, unstructured,
time series, events/alarms, and streaming – along with data sources external to the utility, such
as weather data. The volume of data can be significant if a historical time-series data are used as
data use grows exponentially. Since real-time execution of certain analytics is important, building
capabilities to eliminate latency in data integration (i.e., improving velocity) is important. Lastly, knowing
which data are the correct data (i.e., its veracity) is usually a big area of concern and usually requires
careful analysis. Examples of software used for data integration include Informatica, Talend, Oracle
Goldengate, and SAP SLT and Data Services.
Data sciences/ Analytics-model development differs from software development in terms of its approach. It is common
analytics to have an analytics (or data science) team performing the Exploratory Data Analysis (EDA), devising
applications models, and testing models or model “ensembles” (i.e., groups of models). These models are often
tested over time to minimize false positives and negatives before they are moved into production. In
addition, several vendors provide pre-built analytics models as a part of their software applications;
these models have been tested at various utilities and need minimal modifications to be incorporated
into production. The models – which may be descriptive, predictive or prescriptive – are built using
Natural Language Processing (NLP), statistical (open-loop/closed-loop) machine-learning models,
and artificial intelligence models (sometimes in an ensemble), among a variety of other approaches.
Examples of software used include R, SPSS, SAS, Python, Scala, SAP Predictive Analysis Library (PAL),
and Oracle R/OAC/Data Miner/OBIEE.
Table 13 showcases several market-available technology solutions currently in use in various utilities
for analytics systems.
Table 13: Technology Solutions in Use in Various Utilities for Analytics Systems
30 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
2.2 Approaches for Procurement and Implementation of
Analytics Systems
While most South Asian utilities have already implemented software solutions for billing and
customer-related processes, they have not yet implemented the OT systems. As a result, the utilities
will have to undergo the transformation journey depicted in Figure 7 to fully leverage the benefits of
data analytics.
Stage Description
The utility has a low awareness of analytics or its value across the value chain. It has just begin
Nascent stage integrating OT systems into its IT framework, and has standalone reporting capabilities within
each business area.
The utility has conducted an initial assessment of data analytics and has some form of strategic
plan in place. It may have standalone dashboards and reporting capabilities and disparate
Pre-adoption stage
analytics processes within each business area; however, difficult obstacles remain, such as data
integration skill-set shortcomings with regard to tools and human resources.
The utility has strategic and scalable technical capabilities to deliver enterprise-wide analytics
Adoption stage solutions across business areas. There will be a significant shift toward data-driven business
processes and a redefinition of the operating model.
The utility has achieved a complete digital transformation, and all business processes and
Maturity stage
operations are now optimized using advanced analytics.
The utility has generated significant business value from analytics by continuously focusing on
Visionary stage
sharpening its technological capabilities.
Source: TDWI Analytics Maturity Model.
1 2 3
Implementation and Adopting Core Adopting Advanced
Integration of Data analytics related to Data Analytics
Sources MBC Systems
• Implementation of • Adoption of RQC, • Leveraging high
IT and OT systems BQC, Exception resolution “Big
• Integration of OT checks Data” for advanced
with IT systems • MIS Reporting analytics systems
(data Science)
9 See https://www.microstrategy.com/getmedia/9b914607-084f-4869-ae64-e0b3f9e003de/TDWI_Analytics-Maturity-Guide_2014-2015.
pdf, p. 9.
In the consolidated procurement model, the utility hires a System Integrator (SI) to render end-to-end
services to implement the analytics system – including an “as-is” study,10 system design, procurement
of system components, implementation and integration. The utility’s role includes organizing the call for
bids/tenders, monitoring the implementation and making payments to the system integrator. Utilities
with no prior experience in IT implementations typically adopt this “consolidated” model, which is
illustrated in Figure 8.
In India, for example, Kolkata-based utility CESC has adopted this model for deployment of analytics
solutions.
EESL’s “cost-plus” model is very similar to this model. As a third-party nodal agency, EESL is operating
a BOOT (build, own, operate and transfer) model to help utilities take up efficiency improvement
initiatives.
In the component-wise procurement model, the utility first procures the necessary hardware, then
hires a system integrator for software implementation and system integration. This model, illustrated in
Figure 9, is adopted mainly by utilities planning to expand their existing systems.
10 An “as-is” study defines the current state of business processes in an organization. It is often an interim step to improving the current
state by creating a new “to-be” or future-state process.
32 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Figure 9: Component-wise Procurement Model
Analytics System
Hardware Vendors
System
Integrator
The key issue that might arise in this model will be related to interoperability of various system
components that are procured separately. Any utility considering adopting this model should have
sufficient technical expertise and an experienced IT team within the organization.
In India, Delhi-based TPDDL has adopted this model to deploy its analytics solutions.
In the software as a service (SaaS) procurement model, the utility hires a service provider to host the
systems, either on a cloud platform or physically at their host locations, and to provide other services
as needed. As illustrated in Figure 10, the utility’s role is limited to managing the service provider
agreements and monitoring the performance of the systems and services rendered.
In the United States, PECO and ComEd have adopted this model for deploying energy-theft and meter-
operations analytics services. PECO uses Oracle DataRaker, while ComEd uses Operations Optimizer,
which is made by Silver Spring Networks (recently acquired by Itron).
DISCOM Role:
• Releases tender to procure software-as-a-service analytics
platform to implement analytics
• Service provider ensures scalability and service level
agreements are met
• DISCOM manages provider according to agreement and
monitors performance
The pros and cons of each procurement model are discussed in Table 15.
Procurement
Pros Cons
Model
Consolidated Traditional
and mature implementation model Control on the outcome can only be asserted
High probability of success as utility draws contractually
on experience from previous successful Maintenance of system requires training of internal
programs staff
Resources: capability and capacity well Inability to deploy specific resources DISCOM
defined and understood knows and trusts
Component- More control over implementation and Lack of large program experience
wise execution Depends on capacity of internal resources to scale
Flexible and ability to adapt DISCOM responsible for performance, reliability
Change management and organizational and scalability
knowledge DISCOM accountable for delays and overruns
Software as a Short procurement and implementation cycle Data privacy and security concerns
Service (SaaS) Vendor responsible for performance, Control on the outcome can only be asserted
reliability and scalability contractually
Lower total cost of ownership Strong communications network and data
interfaces needed
34 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Table 16: Hardware and Software Specification of Analytics System
Hardware Specifications
The solution should be sufficiently scalable to accommodate future business growth.
The analytical system ideally should have a built-in redundancy feature with no single point of failure.
To protect the utility’s investment, the analytical system should be able to support the coexistence of at least two
generations of hardware in order to avoid hardware re-deployment for future upgrades or expansions.
The solution should allow for data replication between two data centers in asynchronous mode.
The power, cooling and space parameters (Rack Unit and HDD numbers)in the hardware system must be designed to
accommodate future expansion.
The operating system should be bundled with the solution including support.
The solution should be bundled with network switch.
The platform should be capable of the standard archiving process
The platform should be persistent such as for long-term regulatory and analytics data, records that will be used in
time-based analysis; need a clearly defined policy to include on-going integrity checks and other long-term reliability
features, and to address the need for data-in-place upgrades.
Type Features
Analytics Data Integration and Processing
Solution The Extract, Transform and Load (ETL) tool should be integrated with the most commonly used source
systems, such as ERP systems, CRM systems, and RDBMSs.
The ETL tool should allow for the aggregation of all source databases. The data will load into the data
store, and loads can be scheduled.
The solution should allow for real-time data ingestion (i.e., the process of obtaining and importing data
for immediate use or storage in a database) and real- or near-real-time reporting capability from IoT
devices, smart meters and other sources of data.
The solution should have integration tools for transforming data within the analytics system.
2.4.1 Reporting
36 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
2.4.1.2 Mobility Features
In terms of platform support for Microsoft Office applications, the solution should:
Feature seamless integration with Microsoft Office suites
Support Outlook integration
Support Excel integration, with the ability to leverage native Excel charts
Allow printing to PDF as well as exporting data in Excel and comma-separated-value (CSV) formats.
2.4.2 Reliability
High availability (HA) refers to the analytics system’s ability to service users even when confronted
with component failures. The system should ideally have redundancy at both hardware and network-
component levels.
Data protection capabilities let users protect data from any unauthorized access. The system should
offer at-rest encryption of data to ensure data and information security.
2.4.3 Storage
The analytics system should support data storage and applications on a cloud platform or on on-site
physical databases. Cloud-based storage can be a cost-effective solution compared to conventional
on-site physical storage solutions; it can also accelerate technology adoption in power utilities by
reducing the time required to set up data centers. Table 17 compares the various cloud-based and
physical database solutions.
Table 18: Sizing and Typical Bill of Quantity for Analytics System
Parameters Description
Number of consumers 5,00,000
Number of parameters 10
38 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Parameters Description
Processor type OEM-certified processor
Network interfaces 10 gigahertz (GHz), two in number
Operating system Open source (Windows, Linux, Unix)
HBA** connector Two in number, 16 gigabits per second (Gbps)
50 percent of the main setup
Data redundancy
Replication support to be provided by the software vendor
* CPU = central processing unit.
** HBA = host bus adapter.
http://www.tatapower-ddl.com/showcontent.aspx?this=151&f=ABOUT-US&s=Smart-
Grid&t=Overview
http://www.tata.com/article/inside/tpddl-launches-state-of-the-art-smart-grid-lab
http://www.tatapower-ddl.com/Editor_UploadedDocuments/Content/TATA-POWER-DDL%20
Smart-Grid-BROCHURE-Final.pdf
Yi Wang, Qixin Chen, Tao Hong, and Chongqing Kang. 2018. “Review of Smart Meter Data Analytics:
Applications, Methodologies, and Challenges.” IEEE Transactions on Smart Grid. https://arxiv.org/
abs/1802.04117
40 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Appendices
11 See https://openei.org/wiki/Definition:Head-End_System.
Appendices 41
In-Memory In computer science, in-memory processing is an emerging technology for processing data stored
Technologies in database. Older systems are based on disk storage and “relational” databases maintained
using a Relational Database Management System (RDBMS). Because stored data is accessed
much more quickly when placed in random-access memory (RAM) or flash memory, in-memory
processing allows data to be analyzed in real time, enabling faster reporting and decision making.
Internet of Things The IoT is an interconnection of endpoints (devices and other objects) that can be uniquely
(IoT) identified using an Internet Protocol (IP) address. With the IoT, devices can be connected to
the Internet to sense, gather, receive and send data, and communicate with each other and
applications – via IP technologies, platforms and connectivity solutions.
Net Metering Also called Net Energy Metering (NEM), net metering is a billing mechanism that credits solar
energy system owners for the electricity they add to the grid.12
Non-Volatile Non-Volatile Memory (NVM) is a type of computer memory that has the capability to hold saved
Memory data even if the power is turned off.
Smart Grid According to the IEEE, “The Smart Grid is defined as an integration of electricity and
communication, so that the electric network will be ‘always available, live, interactive,
interconnected, and tightly coupled with the communications in a complex energy and information
real-time network.’”13 A smart grid is not a single technology or system, but rather a central system
of systems in which various types of technology are deployed.
42 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Appendix B. Meter Reading Quality Checks (RQCs) – 2nd Level: Screenshots
Low Consumption
This check identifies variations in the consumption trend of a particular consumer, with reference to low-threshold values for each
44 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check identifies variations in the MDI trend of a particular consumer with regard to historical meter reading data and category-wise
threshold values.
Reading Reversed/Wrong Reading
This check identifies erroneous energy consumption readings, such as very high or negative consumption, captured from meters.
46 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check identifies cases where the power factor value is below the threshold defined by the utility.
Zero Consumption
This check verifies that previous-month and current-month consumption readings (in kWh) are the same, so that the resultant per-month
consumption is zero.
This check compares the demand value from the meter data with contractual demand limits in the consumer master database. A penalty
is applied whenever demand exceeds contractual limits.
48 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Non-Reprogramming of Meters
This check verifies that the tariff slab information and reset date from the meter data are in line with the consumer master data.
50 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check performs a trend analysis of current billed demand along with a comparison to threshold value.
Inflated Bill
Inflated Bill
This check identifies cases of high bill amount compared to the threshold value for the particular category.
52 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check identifies cases where a customer’s bill amount for the current month is negative.
High Slab
High Slab
This check compares the slab value with the threshold value for particular category of consumer.
54 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Net Metering Billing Annexure
Net Meter Units Previous Date Current Date Cumulative Peak Normal Off Total Unit- Unit- Unit-Off Units Carried
Details Brought Reading Reading Consumption Peak Peak Normal Peak Forward/Billed
Forward
Draw from 113295 30/11/18 114775 31/12/2018 1479 197 655 627
TPDDL (A)
Net (A-B) 0 470 -28 -88 586 470 -28 -88 588 470
Disclaimer:
1. Excess units if any, will be adjusted against your next month consumption and excess units at the end of the financial year will be refunded thru Cheque.
2. ‘The amount of Government Subsidy & Electricity Tax charged in Bill towards Net metered consumption is subject to revision after receipt of clarification.
This check verifies the correctness of bills for consumers having net metering.14
14,000
12,260.00 kWh
13,000
12,149.00 kWh
12,000 11,356.00 kWh
10,660.00 kWh
11,000
9,831.00 kWh 10,523.00 kWh
10,000 9,124.00 kWh 9,436.00 kWh
8,781.00 kWh
9,000
8,000 6,740.00 kWh
7,000
Consumption
5,000
4,000
3,000
2,000
1,000 13.00 kWh
0
11 days
31 days
31 days
31 days
31 days
31 days
31 days
31 days
Current
30 days
28 days
30 days
30 days
History 1
History 7
History 3
History 4
History 8
History 2
History 5
History 6
History 9
History 11
History 10
Histories
56 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
� Active (I) Total-kWh (CH1) � Reactive (I) kVArh (CH3) � Abs Active - kWh (CH5) � Def. in Active (I)- kWh (CH7)
� Active (E) Total-kWh (CH2) � Reactive (E) kVArh (CH4) � Abs Apparent - kVAh (CH6) � Def. in Apparent - kVAh (CH8)
This check compares the consumer’s current consumption with historical trend and threshold values.
Voltage Failure
Average Voltage
320
300
280
260 Voltage failure in one phase
240
220
200
180
160
140
120
Voltage (in V)
100
80
60
40
20
0
02- May 04- May 06- May 08- May 08- May 12- May 14- May 16- May 18- May 20- May 22- May 24- May 26- May 28- May 30- May 01- Jun
Wed Fri Sun Tue Thu Sat Mon Wed Fri Sun Tue Thu Sat Mon wed Fri
This check verifies that system voltage values are in line with tolerance and threshold values.
60 54.12 kVA
47.32 kVA
50 41.80 kVA 44.38 kVA
40
Maximum Demand
30
16.54 kVA
20 13.42 kVA
10 3.28 kVA
0
Current History 1 History 2 History 3 History 4 History 5 History 6 History 7 History 8 History 9
Histories
� Abs Apparent-IVA (CH4) � Abs Active Total-kW (CH1) � Abs Reactive (Lag)-kVAr (CH2) � Abs Reactive (Lead)-KVAr (CH3)
This check verifies whether the calculated load factor is in the predefined threshold load factor range for the particular consumer
category.
58 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Data Corruption
11,00,000
9,71,312.00 kVAh
10,00,000
9,00,000
DATA CORRUPTED
8,00,000
7,00,000
6,00,000
5,00,000
Consumption
4,00,000
3,00,000
2,00,000
1,00,000 5,613.00 kVAh 4,427.00 kVAh 5,450.00 kVAh 6,130.00 kVAh 0406.00 kVAh
6,035.00 kVAh
5,185.00 kVAh 7,550.00 kVAh 6,143.00 kVAh
4,666.00 kVAh 6,554.00 kVAh
0
Current History 1 History 2 History 3 History 4 History 5 History 6 History 7 History 8 History 9 History 10 History 11
22 days 31 days 30 days 31 days 28 days 31 days 31 days 30 days 31 days 30 days 31 days 31 days
Histories
� Active(I) - kWh (CH1) � Active(E) Total - kWh (CH3) � Abs Apparent - kVAh (CH5) � Abs Reactive (Lead) -kVArh (CH7)
� Active(E) - kWh (CH2) � Abs Active - kWh (CH4) � Abs Reactive (Lag) - kVArh (CH6) � Active [Del] - NegCur-kWh (CH8)
� Def.in Active (I) - kWh (CH9)
This check looks for data corruption by comparing the values recorded for MDI/consumption data/digits with the threshold values.
60 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Power Failure
Neutral Disturbance
62 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check identifies cases of unbalanced neutral current values or unbalanced voltage values.
Current Reversal
Current Reversal
This check verifies whether the value of the active current is negative or not.
64 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Current Missing
Current Missing
This check identifies cases where the phase current is zero or less than specified threshold value.
0.9
0.8
0.7
0.58 0.57 0.57 0.58 0.57 0.56 0.56 0.55 0.54 0.55 0.55 0.54 0.54
0.6
0.5
Power Factor
0.4
0.3
0.2
0.1
0
Current History 1 History 2 History 3 History 4 History 5 History 6 History 7 History 8 History 9 History 10 History 11 History 12
Histories
66 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check verifies whether the Power Factor (PF) data is within the threshold range defined for low and high PF.
Manual Reset
This check verifies whether the meter reading reset duration is other than the historical value or predefined value.
68 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
This check verifies that the meter clock accuracy is in line with the HES clock time.
Internal Ratio
This check cross-verifies that the on-site CT ratio is consistent with the consumer master data.
120
100
80
60
40
20
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
This check identifies cases of magnetic tampering by analyzing consumption pattern variations.
70 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Load Unbalance
Load Unbalance
This check verifies whether the phase current values in three phases are unequal or vary above the tolerance range in each phase.
38
72 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Potential Missing with Load Running
Potential Missing with Load Running
This check identities cases where the current value is recorded but with voltage value as zero.
Missing Data/Data Download Gap Check Identifies instances of data missing from periodic or on-demand polling.
Time-of-Day (ToD) Consumption Check Verifies whether the ToD slab consumption has been recorded by the meter
for the particular consumer category.
Meter Pulse Overflow Check Checks that the number of pulses does not exceed the maximum threshold.
Test Mode Check Checks whether the data was collected while the meter was in test mode.
Energy Sum Check Calculates the difference between the total cumulative consumption shown
in the incoming data and the total consumption of various slots over the
same time period.
Interval Spike Check Identifies intervals showing high usage relative to surrounding intervals.
Unit of Measurement (UoM) Check Ensures that the Unit of Measurement (UoM) of the incoming data matches
the UoM specified on meter profile.
Usage of Inactive Meters Check Identifies cases of consumption shown for inactive meters.
Negative Consumption Check Verifies whether the meter reading of a consumer is lower than the previous
meter reading.
Reactive Load (kVARh) Check Identifies intervals where reactive load (measured in kVARh) is present and
active load (measured in kWh) is not.
74 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Appendix E. MIS Reports: Screenshots
This is a summary report that is generated after the initial base meter reading (in a non-AMI scenario)
showcasing meter reading status.
REMARK_BASE
INSTALLATION
Reading Date_
DEVICE NO_
MR ID_BASE
Photo (Y/N)_
kVAh_BASE
MRU_BASE
KWh_BASE
SMRD_ID_
PORTION_
kVA_BASE
NO_BASE
kW_BASE
READING
BASE
BASE
BASE
BASE
BASE
527 250 MN08B MN08B004 5000003694 51003314 TL 23 Jul 18 Y
K13 802 KP01D KP01D007 5000004203 41013941 TL 24 Jul 18 Y
S74 828 SN09C SN09C012 5000005971 40275911 ND 23 Jul 18 Y
315 189 SN05C SN05C001 5000011977 41283517 TL 24 Jul 18 Y
254 654 CLO4C CLO4C002 5000012738 40251433 ND 24 Jul 18 Y
254 654 CLO4C CLO4C002 5000012760 43034227 TL 24 Jul 18 Y
831 529 MN08C MN08C002 5000013202 41558813 TL 21 Jul 18 Y
K12 798 KP01C KP01C001 5000013221 42058397 PL 23 Jul 18 Y
738 766 BD02E BD02E004 5000014203 41084490 OB 21 Jul 18 Y
392 707 MP02D MP02D002 5000015065 41072411 TL 24 Jul 18 Y
4 139 CL04D CL04D004 5000017332 72285469 TL 23 Jul 18 Y
This report gives details (including serial numbers) of meters that have failed RQC validations and thus
require follow-up on-site meter readings.
76 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Monitoring Report (Unread Meters)
Shalimar Bagh
Contractor No
Keshavpuram
Shakti Nagar
Model Town
Mangolpuri
Moti Nagar
Pitampura
Civil Lines
Bawana
Narela
Rohini
Badli
Sum:
Sub
GNI 1 1 1 3 1 1 8
HRB 89 119 58 61 138 63 92 113 51 92 61 158 1,095
KCG 4 3 1 1 4 2 2 2 1 20
PP-HCB 17 20 12 21 15 16 17 13 10 36 8 16 201
PP-NOTOD 2 1 2 1 2 8
PP-TOD 1 3 1 2 7
RC1 10 1 7 2 21 4 2 2 2 51
RC2 3 3 1 4 19 6 103 139
RC3 2 1 3 5 5 2 3 23,541 23,562
RC9 1 1,980 3 14 25 1 2,918 7 4 79 5,032
SCB 79 15 7 18 2 8,221 2,410 235 1 1 16 11,005
SHORT TE 1 1
SOLR-HCB 9 2 9 5 3 6 7 8 12 10 6 77
STLT 10 19 7 2 6 1 6 218 2 6 2 15 96
Sum: 10,436 11,335 7,060 5,789 15,768 13,088 11,505 10,858 6,380 10,471 5,532 23,943 13,2165
This report shows the number of meters unread due to meter inaccessibility.
Responsibility Center 1 day 2-5 days 6-7 days 8-15 days >15 days Grand Total
BQC Desk 35 62 5 4 106
JE Authorization Desk 1 1 1 3
Meter Reading Analysis Desk 132 113 7 2 254
Meter Reading Analysis Desk – HRB 3 36 39
Rcg Coordination Desk 2 1 3
Grand Total 173 213 12 6 1 405
This summary report shows the number of bills, disconnection notices distributed (with or without
proof of delivery, or PoD) and undistributed bills.
78 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Reconciliation Report
Not to Remaining
Pending
Consider Unbilled
Total no. of Total Total
consumers Invoiced Active Resolved Resolved Unbilled
First
Cycle Active Inactive TD DT Total
Bill % age % age
Unbilled
H=(C-
(A) (B) (C) (D) (E) (F) (G) (A-H)
(F+G))
RC1 180786 149518 3356 26419 1493 17 520 2819 177967 98.44 1.56
RC2 204103 172759 3315 26469 1560 7 714 2594 201509 98.73 1.27
RC3 227228 187619 3656 34056 1921 30 788 2838 224390 98.75 1.25
RC4 193881 161234 4258 26378 2011 19 698 3541 190340 98.17 1.83
RC5 185259 154329 3523 26133 1274 18 502 3003 182256 98.38 1.62
RC6 188700 158160 3310 25647 1583 24 373 2923 185787 98.46 1.54
RC7 212348 176006 4730 29879 1733 22 491 4217 201831 98.01 1.99
RC8 217974 181426 4652 30344 1552 18 629 4005 213969 98.16 1.84
RC9 41993 2131 3485 36084 293 35 245 3205 38788 92.37 7.63
SCB 170875 130116 4266 33947 2546 13 1117 3136 167739 98.16 1.84
G&I 8580 7273 543 609 155 2 46 495 8085 94.23 5.77
HRB 87665 68363 3179 15173 950 0 478 2701 84964 96.92 3.08
KCG 2453 1872 12 547 22 0 1 11 2442 99.55 0.45
Prepaid 4639 4204 202 104 129 0 22 180 4459 96.12 3.88
Streetlight 5029 4164 278 530 57 0 21 257 4772 94.89 5.11
Grand 1931513 1559174 42768 312295 17279 205 6645 35915 1895598 98.14 1.88
Total
This report yields statistics on meters read and unread as well as consumers billed and invoiced.
This report shows the number of consumers having provisional bills for 60/90 days.
Actual Units
Actual Units
Actual Units
Demand (in
Demand (in
Billed Units
Billed Units
Billed Units
Provisional
Provisional
Ownership
No. of Bills
No. of Bills
Demand
Current
Current
Current
(in MU)
(in MU)
(in MU)
(in MU)
Group
No. of
No. of
(in %)
(in %)
(in %)
Crs.)
Crs.)
Bills
Bills
HCB 815023 5321 126.79 126.2 83.14 1082017 3649 187.59 185.9 119.49 -32.41 -32.12 -30.42
HRB 70138 5 213.62 206.62 224.74 70402 11 221.22 213.12 217.62 -3.43 -3.05 3.27
EXPR 278 0 56.44 54.94 53.94 246 0 57.34 55.86 51.05 -1.57 -1.66 5.67
KCG 1661 5 45.83 44.23 47.11 1525 0 46.74 45.08 46.84 -1.94 -1.87 0.58
SCB 153423 791 14.39 14.34 7.46 187744 374 15.8 15.7 8.32 -8.96 -8.70 -10.42
Others 2 0 0.04 0.03 0.05 -100.00 -100.00 100.00
Sum: 1040523 6122 457.07 446.33 416.39 1341936 4034 528.73 515.69 443.37
This report shows the total units billed, the amount billed for the current month, and the amount billed
for the same month of the previous year, and deviations.
SCG KCG
HRB Consumers HCB Consumers Total (LT) Grand Total
Consumers Consumers
Net Demand
Due
Date
Received
Received
Received
Received
Received
Received
Payment
Payment
Payment
Payment
Payment
Payment
Subsidy)
Demand
Demand
Demand
Demand
Demand
Subsidy
(W/O
(W/O
01 Feb 18 0.71 7.2 4.69 4.29 0.6 0.21 5.99 11.99 0.06 9.19 6.06 21.25
02 Feb 18 0.53 5.24 7.99 4.7 0.09 0.22 8.6 10.17 0.44 0.52 9.05 10.69
03 Feb 18 0.17 2.75 4.59 3.58 0.28 0.2 5.04 6.53 0.33 1.44 5.37 7.97
04 Feb 18 0 0.23 0.03 1.06 0 0.03 0.03 1.32 0 0.03 0.03 1.35
05 Feb 18 0.2 3.47 6.7 4.84 0.21 0.25 7.12 8.61 0 0.89 7.12 9.5
06 Feb 18 0.28 3.02 7.87 4.61 0.15 0.29 8.3 7.92 0 0.29 8.3 8.21
This report shows the daily amount paid by various categories of consumers.
80 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Due Date MIS Reports
25.04.2018
20.04.2018
24.04.2018
Row Labels
23.04.2018
16.04.2018
18.04.2018
21.04.2018
19.04.2018
17.04.2018
Badli 2366 118 42 43 791 8 12824 9 5
Bawana 4403 3063 3893 35 100 8 4118 4396 5
Civil Lines 2253 5484 141 58 2639 174 9971 3276 135
Keshavpuram 80 78 32 24 3270 5 8186 81 1
Mangolpuri 13151 16857 15578 1104 116 16 1068 7866 35
Model Town 191 114 205 23 7657 393 14754 9179 5
Moti Nagar 241 109 232 37 12610 766 24394 4063 9
Narela 4901 3914 5978 77 3135 238 10479 3866 9
Pitampura 965 119 2653 56 5282 12 11880 1495 350
Rohini 705 156 109 69 7732 27 9507 11079 5
This report is issued to head cashiers at collection centers to help them with resource planning.
MTD – Demand vs Collection for Feb-17 Vs Feb-18 (01-Feb-18 to 28-Feb-18), Amount in Rs. crores
Collection Collection
LPSC
Demand Due Date Wise MTD Collection Efficiency Efficiency
Ownership Income
(Gross of LPSC) (Net of LPSC)
Group
17 Feb 18 Feb 17 18
17 Feb 18 Feb Variation 17 Feb 18 Feb 17 Feb 18 Feb
Collection GS Collection GS Feb Feb
HRB 223.56 239.66 7% 207.72 0.24 245.33 0.25 93.02% 102.47% 0.36 0.49 92.86% 102.26%
HCB 123.44 136.59 11% 96.56 26.83 113.89 27.44 99.96% 103.47% 0.68 0.7 99.41% 102.96%
SCB 7.7 8.57 11% 5.8 2.8 6.28 3.11 111.66% 109.54% 0.16 0.13 109.64% 107.98%
Total (LT) 354,71 384.82 8% 310.09 29.87 365.5 30.81 95.84% 102.98 1.19 1.33 95.51% 102.64%
KCG 93.09 88.6 -5% 97.21 - 106.71 - 104.43% 120.4% 0 0.05 104.43% 120.38%
STL 12.17 12.15 0% 12.5 - 16.38 - 102.79% 134.78% - - 102.79% 134.78%
Total KCG 105.25 100.75 -4% 109.71 - 123.09 - 102.24% 122.17% 0 0.05 104.29% 122.11%
Total 459.96 485.58 6% 419.81 29.87 485.59 30.81
Due Date 97.76% 106.96 1.19 1.39 97.50% 106.68%
- - 449.67 519.39
Pending
Collection is net of cheque bounce. Demand & Collection extracted from BIW.
Total 459.96 485.58 6%
Effect of credit/debit JE’s have not been considered in this report.
This report shows the amount billed, amount collected (including subsidies, if any) and collection
efficiency for various due dates.
160 100
142.36 87.07
140 90 79.23
78.19 81.59 78.66 79.82
127.14
80 77.24 78.87 78.98 77.979.06 75.24
120 114.88 126.67 117.17
95.79
109.28 70 75.35
73.48 72.78 74.6275.18 76.51
100 72.44 74.15 72.85 71.76
88.8 88.26 60
86.51 95.71
79.01
80 82.68 50
70.46
74.47 51.46 40
60
30
40
20
20
10
0 0
Apr may Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar April May June July Aug Sep Oct Nov Dec Jan Feb Mar
Arrear LT Segment FY 15–16
Arrear LT Segment FY 16–17 FY 2015–16 FY 2016–17
This report provides data on the number of consumers defaulted/disconnected, their arrears
amounts, and ageing.
This report shows the status of disconnection orders for various categories of consumers.
82 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Billing Status Report
Category/Area Details
5% 6% 89% 0%
Average
This report shows the percentage of total billing done for various billing statuses (average, provisional,
regular, others).
Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Avg 1,542 1,609 1,398 1,381 1,329 1,267 1,258 1,268 1,289 1,265 1,201 1,016
Prov 1,514 1,812 1,425 1,521 1,374 1,460 1,460 1,578 1,452 1,387 1,309 1,167
Reg 3,094 3,747 4,121 4,317 4,410 4,010 4,008 3,798 3,268 2,886 2,672 2,478
Other 4,090 8,956 13,156 1,767
Total 2,796 3,336 3,541 3,603 3,620 3,311 3,309 3,157 2,769 2,486 2,318 2,134
Category/Area Details
Avg 1,543 1,560 1,650 1,107 1,309 1,092 1,327 1,134 1,284 1,140 1,293 1,128
Prov 1,296 1,360 1,387 1,571 1,569 1,540 1,642 1,569 1,568 1,433 1,334 1,214
Reg 2,726 3,098 3,709 3,169 2,875 2,610 2,626 2,564 2,332 2,030 1,988 1,853
Other 578 3,085 3,438 2,574
Total 2,532 2,839 3,366 2,817 2,608 2,353 2,404 2,318 2,155 1,893 1,886 1,749
Avg 3,263 3,205 4,756 3,753 3,053 2,288 2,121 2,241 2,104 2,102 2,130 1,952
Prov 2,210 2,268 3,147 2,323 2,639 2,373 2,253 2,287 2,378 2,236 2,149 1,935
Reg 2,833 3,305 3,784 3,742 3,716 3,707 3,557 3,428 2,784 2,487 2,210 2,203
Other 5,155 93 100 549
Total 2,821 3,234 3,803 3,648 3,601 3,522 3,368 3,270 2,711 2,448 2,202 2,176
Under Billing Over Billing
This report provides the total amount billed per consumer for various billing statuses.
2,000 283
234
1,500
Others
161
1,000 1774 Domestic
1667
Agriculture
500 1040
11 10 6
178 159 26
0 53 67 38
28 25 2
Utility A Utility A Utility B Utility B Utility C Utility C
This report provides the number of consumers with sundry adjustments in terms of amount and units
of energy.
This report shows the number of times consumers of various bill statuses (provisional, average, and so
on) have been billed compared to the total bills generated in a year.
84 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Payment Behaviour Report
Number of times paid
This report breaks down consumer payment behavior in terms of annual payment frequency.
80% 78%
82% 82% 76% 73%
74% 75% 75%
75% 78%
71% 71%
70% 70%
74%
70% 73% 73% 73%
70% 72% 70% 70%
70% 69%
65%
66%
60%
62%
60% 59% 59% 59% 59% 61%
58% 58% 58%
55%
Apr-15 May-15 Jun-15 Jul-15 Aug-15 Sep-15 Oct-15 Nov-15 Dec-15 Jan-16 Feb-16 Mar-16
Category/Area Details
Utility A Utility B Utility C
This report compares the total number of bills generated to the total number of payments received in
each billing cycle.
40
20
0
April May June July August Sept Oct
Arrears - FY 13-14 Arrears - FY 14-15
This report shows the total amount of pending arrears at the end of each billing cycle.
Average Amount
NO Less than Greater Greater Greater Grand Total
Frequency of Late Payment
This report breaks down the number of late-paying consumers by late-payment frequency and
amount, which helps in identifying suspected default consumers.
86 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Meter Status Report
0% 3% 87% 10%
Category/Area Details
Burnt
1% 6% 83% 10% Defective
OK
This report shows the overall status of meter health as of the latest billing cycle.
22% 2%
23% 21% 9% 15%
17% 1%
33% 19% 19% 11% 4%
36% 17% 62% 10% 17%
8% 25%
9%
21% 21%
17%
11% 15%
34% 24% 20% 68%
15% 63%
31%
0-5 Years 6-10 Years 11-15 Years 16-20 Years >20 Years
88 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Apr-14 5000778274 NDP17842 NOT OK OK OK OK OK OK OK OK
Consumer
Apr-14 5000778377 NDP15179 OK OK OK OK OK OK OK OK Meter Number
Apr-14 5000779202 NDP20596 OK OK OK OK OK OK OK OK
This report provides the consumer-wise status of exception checks done on meter data.
Date
Appendix F. AMR Data Analytics – Use Cases: Screenshots
These use cases are used to perform short-, medium- and long-term energy forecasting. The short-
term forecasting takes into consideration weather-related data for day-ahead purchase of power. This
section addresses the following use cases:
Predictive fault analysis and network-asset management
Energy optimization and Volt/VAR control
Predictive maintenance
Network optimization/load balancing
Interconnected feeder analysis
Energization scheme monitoring
480
Line Current
Voltage
Energy
360
240
120
00:15
00:30
00:45
01:00
01:15
01:30
01:45
02.00
02:15
02:30
02:45
03:00
03:15
03:30
03:45
04:00
04:15
04:30
04:45
05:00
05:15
05:30
05:45
06:00
06:15
06:30
06:45
07:00
07:15
07:30
07:45
08:00
08:15
08:30
08:45
09:00
09:15
09:30
09:45
10:00
10:15
10:30
10:45
11:00
11:15
11:30
11:45
12:00
12:15
12:30
12:45
13:00
13:15
13:30
13:45
14:00
14:15
14:30
14:45
15:00
15:15
15:30
15:45
16:00
16:15
16:30
16:45
17:00
17:15
17:30
17:45
18:00
18:15
18:30
18:45
19:00
19:15
19:30
20:00
20:15
20:30
20:45
21:00
21:15
21:30
21:45
22:00
22:15
22:30
22:45
23:00
23:15
23:30
23:45
00:00
kWh kVAh Amp-1 Amp-2 Amp-3 V-1 V-2 V-3
DT loading data is analyzed to ensure optimal loading of the DT to avoid failures and manage the life
of the asset.
Predictive Fault Analysis and Network-Asset Management - DT Unbalancing Analysis
160
80
0
00:15
00:30
00:45
01:00
01:15
01:30
01:45
02:00
02:15
02:30
02:45
03:00
03:15
03:30
03:45
04:00
04:15
04:30
04:45
05:00
05:15
05:30
05:45
06:00
06:00
06:15
06:45
07:00
07:15
07:30
07:45
08:00
08:15
08:30
08:45
09:00
09:15
09:30
09:45
10:00
10:15
10:30
10:45
11:00
11:15
11:30
11:45
12:00
12:15
12:30
12:45
13:00
13:15
13:30
13:45
14:00
14:15
14:30
14:45
15:00
15:15
15:30
15:45
16:00
16:15
16:30
16:45
17:00
17:15
17:30
17:45
18:00
18:15
18:30
18:45
19:00
19:15
19:30
20:00
20:15
20:30
20:45
21:00
21:15
21:30
21:45
22:00
22:15
22:30
22:45
23:00
23:15
23:30
23:45
00:00
SSN NIC Smart Meter Outage Monitor ( Last 24 Hrs Outages & 2 Hrs Restoration )
2017-08-28
RELIANCE JIO INFOCOME KT027145 210722019839 ONGOING 30
18:31:40
2017-08-28
LALIT KUMAR KT027118 210722017021 ONGOING 30
18:31:40
2017-08-28
PRITAM PAL SINGH KT014338 210722010738 ONGOING 30
18:31:40
2017-08-28
LALIT KUMAR JAIN KT027279 210722019064 ONGOING 30
18:31:40
2017-08-28
DHEMENDRA AGARWAL KT014340 210722013699 ONGOING 30
18:31:40
2017-08-28
NEELAM BATALA KT027281 210722014399 ONGOING 30
18:31:40
2017-08-28
TAHIR HUSSAIN KT014566 210722017612 – –
19:05:47
2017-08-28
TAHIR HUSSAIN KT014566 210722017612 – –
18:56:29
2017-08-28 2017-08-28
SUSHIL KUMAR PANDEY AND ROLEY PANDEY KT027185 210722012371 –
18:26:01 18:43:55
2017-08-28
CHAND PRAKASH JAIN KT027098 210722012361 – 18
18:43:54
2017-08-28
STM USHA NARMADEV KT026839 210722012257 – –
18:43:54
2017-08-28 2017-08-28
RAMESH CHAND KT014799 210722012363 18
18:26:17 18:43:54
2017-08-28 2017-08-28
AJEET SINGH KT027199 210722012370 18
18:26:00 18:43:54
Outage event data is analyzed to identify outage locations easily, thereby initiating faster supply
restoration. The data can also be used to identify possible cases of theft at customer locations.
90 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Predictive Fault Analysis and Network - Asset Management - Reliability Indices Analysis
Time Period: SAIDI SAIFI Unauthorized Construction Notices Safety KPI Dashboard PADCI ?
2018-19 34.34 37.87 4,966 5.20
60 75
60.19
46.06 46.06 43.05 53.63
40 50 50.67
32.41 34.34 38.92 37.87
27.08 26.56 28.29 31.84
27.55
20 30.58 25 35.22
24.80 24.77
0 0
2014-15 2015-16 2016-17 2017-18 2018-19 2014-15 2015-16 2016-17 2017-18 2018-19
Actual Target Actual Target
50
10
25
0 0
2015-16 2016-17 2017-18 2014-15 2015-16 2016-17 2017-18 2018-19
Reliability indices values are mapped and analyzed to understand the performance of DISCOM in
supplying reliable power to consumers, predict fault occurrence based on trend analysis, and also
understand network overloading instances.
30 27.81 60 3
20 40 2
20 18.61
10 1
2.77 8.56
0.59 0.05
0 0 0
Category I Category II Distribution Inter.. Distribution Outa.. STS Interruption STS Outage APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR
SAIDI (Hrs.) SAIDI (Hrs.) SAIDI (Hrs.)
6 6 7.5
4.00 6.18
4 3.76 4
5 4.72
2.49 1.92 4.07
2 2 3.43
1.28 0.41 0.63
0.41 2.5
0 0
1.07
-2 -2 0
City Circle Metro Circle Sub urban Ci...Town CircleUrban Circle Badli Rohini Shalimar Bagh 414 503 505 532 Bu Shalimar...
Predictive Maintenance
Under this use case, DT and feeder loading data are analyzed to look for violations above peak value
of various asset-performance factors and to assess DT temperature trends and outage trends. This
ensures proper loading of assets to prevent failures. The utility opts for condition-based maintenance
instead of periodic maintenance.
200
150
100
50
0
Quality Score (%) Average Handling Time (in secs) Average Hold Time (in secs)
Batch without hands on training Batch with hands on training
Here, insights gained from analyzing customer patterns and preferences are used to increase the
effectiveness of self-service channels.
92 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Network Optimization / Load Balancing
Network Optimization/Load Balancing
Network Optimization/Load Balancing - Demand Response
Demand response load profile data is analyzed to ensure the network is optimally loaded and balanced.
94 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
DT loading patterns are analyzed to ensure optimal loading of network assets and plan load shifting.
Interconnected Feeder Analysis
Interconnected Feeder Analysis
Under this use case, the load curves of all interconnected feeders on the same time stamp are analyzed for better load-growth
estimation.
96 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
The energization scheme monitoring is kind of a project monitoring tool which shows the status of energization scheme to various
stakeholders in a single dashboard.
98 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Here, analytics are used to optimize the footfall at payment centers, leading to better consumer satisfaction and increased resource
efficiency.
Predicting Defaulting Consumers
Overall
percentage
DN Trend reduction in
Disconnection Notice Trend DN–16%
50000
Total Count 8,375
42,259
No of consumers not 820 40000 35,450
reaching DN-L1 – Jan to Sep
30000
% Improvement 9.79%
20000
13,016 11,501 14,925 14,318 12,829
11,120
10000
0
Jan-Mar Apr-Jun Jul-Sep Total
2016 2017
Arrear – Disconnection Notice
250
196.83
200
153.11
150
100 70.85
63.79 48.09 62.19 58.05
46.96
50
0
Jan-Mar Apr-Jun Jul-Sep Total
2016 2017
Here, analytics are used to predict defaulting consumers, leading to reductions in arrears and improved
consumers payment behavior.
Analytics are used here to group consumers by default history in order to improve upon the reduction
of defaults.
100 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Here, consumers’ payment behaviour is studied so the utility can work to improve the behavior and retain good consumers.
Theft Analytics
Theft Analytics - DT Energy Audit
Here, DT energy audit data are analyzed to identify thefts based on load-profile variations in the area considered.
Fiscal Year: AT&C Loss (with LIP) (%) Billing Efficiency (%) Collection Efficiency (%) Energy Input (MUs) Bill Collection (Crs) Total Billed Amount (Crs) ?
FY 18-19 7.93 92.07 100.01 9,630.29 7,973.90 7,973.30
AT&C Loss (with LIP) (%) Billing Efficiency (%) Collection Efficiency (%)
10 100 100.25
91.80
91.90
92.07
92.20
92.15
91.23
9.87 100.03
90.41
91.38
100.00 100.20
9 100
8.88 50 100.01
8.59 99.93
8 8.40 99.75
7.93 99.69 99.79 99.80
7.67
0 99.5
7
8 15 16 -17 17-
18
18-
19 20 15 -17 18 19 20
14-
15 16 -17 017-1 18-
19 20 14- 15- 2016 19- 14- 16
15- 2016 17- 18- 19-
20 15- 2016 2 20 19- 20 20 20 20 20 20 20 20 20 20
20 20
Actual Target Actual Target Actual Target
Energy Input (MUs) Bill Collection (Crs) Total Billed Amount (Crs)
10k 10k 10k
7,381.18 7,973.90
90,402
9,562
90,630
6,856.97
90,040
7.5k 7.5k
8,423
8,610
7,973
7,393
6,450
6,862
6,936
7,917
5k 5k 6,429.88 6,937.95 5k
2.5k 2.5k
512.01
0 0 0
15 18 19 20 15 16 -17 17-
18
18-
19 20 15 16 17 -18 19 20
14- 16 -17 17- 18- 19- 14- 15- 2016 19- 14- 15- 16- 2017 18- 19-
20 15- 2016 20 20 20 20 20 20 20 20 20 20 20 20 20
20
Actual Target Actual Target Actual Target
Here, Aggregate Technical and Commercial (AT&C) loss data and billing and collection efficiency are
analyzed to understand the DISCOM’s performance patterns, allowing the DISCOM to take actions to
improve in these areas. This analysis also provides insights on theft rates.
Month Wise Detail Top Five BUs/Zones for Bottom Five BUs/Zones for
MAR -2017-18 MAR-2017-18
9.5 9.33 7.5 50 46.76 41.66
9.02 5.35 5.46
9 8.82 4.58
8.76 5 3.89
8.49 22.32 20.77 19.93
8.5 8.33 25
2.5 1.17
8.45 8.52
8.40
8 8.30 0
7.87 7.92 1 1 2 0 3
7.5 56 57 52 1+52 +130 0
50 130
1 512 53
3 517 513 514
L
P
N
R
N
C
CT
G
AR
AY
V
JU
SE
FE
AP
DE
JA
JU
NO
AU
M
M
10 10 41.66
7.64 46.76
7.5 6.43 6.78 6.71 7.5
5 5
5.64
2.5 2.5 20.77
0 0
y tro b wn n
Cit Me Su To Ur
ba NRL BWN
512 513 521 533
Close
Here again, AT&C loss data are analyzed to understand overall AT&C loss variation patterns, allowing
the DISCOM to pursue loss-reduction measures.
102 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Theft Analysis
Theft Collection Detail (Crs.) for 2018-19
YTD Top Five Zones Month Wise Detail Top Five Zones for JAN-2018-19
6 3 0.075 0.06
4.18 2.15 0.05
4 2 1.86 1.86 0.05
1.52 1.75 0.04 0.04
1.28 1.22 1.30 1.20
1.08 1.06
2 1 0.86 0.025
0.75 0.60 0.54
0.12 0.00
0 2 0 0.0
1
8
4
V
AY
AR
G
CT
C
52
N
R
B
P
L
52
41
51
51
2
1
8
4
7
O
JU
AU
AP
DE
SE
FE
JU
52
JA
52
41
51
51
M
M
N
Circle Wise Detail for
JAN-2018-19 0.2 District Wise Detail for
JAN-2018-19/METRO
0.06 0.16
0.21 0.15
0.17
0.10
0.28 0.36
0.05
0.03
0.02
0
CITY METRO SUB TOWN URBAN KPM MGP PPR
Close
Identification of theft instance and analysis on collection data for theft rate reduction and utility
performance improvement.
3.05
3 2.75
2.38 2.28 2.29
2.08 2.16 2.21
1.87 1.98
1.81 1.77
2 1.52 1.6
1.84 1.21 1.1 1.2 1.13 1.09
1.78
1 1.52 1.48
1.27
1.02 1.11 1.03 0.94 0.88 0.8
0
April May June July Aug Sep Oct Nov Dec Jan Feb March
Under this use case, the complaints registered for meter reading and bills generated are analyzed over
various time periods.
92.64 92.20 91.70 91.98 92.06 92.30 92.33 92.39 91.96 92.04 92.34 92.07
100
50
0
APR MAY JUN JULY AUG SEP OCT NOV DEC JAN FEB MAR
District Wise Billing Efficiency (%) Zone Wise Billing Efficiency (%)
Circle Wise Billing Efficiency (%)
for FEB-2018-19/Urban for FEB-2018-19/BDL
150 for FEB-2018-19 150 100
95.05
94.37
88.75
100 94.03 93.42 88.97 94.43 92.91 100 91.71 92.74 94.30
50
50 50
0 0 0
City Metro Sub Town Urban SMB BDL RHN 507 516 581
Close
Here, analysis of circle/district/zone billing efficiency is used to reduce AT&C loss at the DISCOM.
Perfect (100% Available) Very Good (95-99% Available) Good (80-94% Available)
Moderate (40-79% Available) Poor (1-39% Available) No Read
104 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Read Obtained Till Now for: 05 Jun 18
16000
98.69% 98.96% 97.45% 94.55% 98.89% 98.16% 97.88% 94.39% 96.02% 98.6% 98.16%
14000
3.2
12000
98.06
10000
Overall Read meter % Overall Nonread Meter %
8000
6000
4000
2000
0
A1 A2 A3 A4 A5 A6 B1 B2 B3 B4 B5
Target Meters Read Success
Single-Phase Meters
Table 19: Threshold Values for Exception Generation for Single-Phase Meters
Persistence
Persistence Time Threshold Value for Threshold Value for
Exceptions Time for
for Occurrences Occurrence of Events Restoration of Events
Restoration
ESD/JAMMER Immediate (record 0 hr. 01 min. Immunity up to 50 kilovolts Removal of ESD/jammer
only 1 event on first 0 sec. (should (kV) with NIC and logging of signal
application and only restore after event >50 kV
1 event for next 1 min.) 1 min. of last
application)
Magnetic Tamper 0 hr. 2 min. 0 sec. 0 hr. 2 min. >0.5 tesla (T) for permanent <0.5 T for permanent
0 sec. magnet magnet
OR OR
DC magnetic induction DC magnetic induction
>0.2 T <0.2 T
OR OR
AC magnetic induction AC magnetic induction
>10 milliTesla (mT) <10 mT
Meter Top Cover Immediate Immediate If meter top cover is opened Not applicable
Open
Neutral Missing 0 hr. 30 min. 0 sec. 0 hr. 2 min. 1. Battery is used for voltage Voltage >190 V
0 sec. reference: Under tamper
condition of neutral missing,
meter will perform the fraud
energy registration above
500 mA assuming Vref
(reference voltage from the
battery) and Unity Power
Factor (UPF).
OR
2. Third CT is used for
voltage reference: Under
tamper condition of
neutral missing, meter
will perform the fraud
energy registration above
1 A assuming Vref (from
third CT) and Unity Power
Factor (UPF).
Testing: For checking
the meter reliability and
accuracy in lab, the meter is
tested for 30 minutes with
voltage sampled at intervals
of 30 seconds and constant
current.
106 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Persistence
Persistence Time Threshold Value for Threshold Value for
Exceptions Time for
for Occurrences Occurrence of Events Restoration of Events
Restoration
Neutral 0 hr. 01 min. 0 sec. 0 hr. 02 min. Voltage >145% of Vref Voltage <145% of Vref
Disturbance 0 sec. OR OR
Current >10% of base current Current <10% of Ib
(Ib) OR
OR Frequency >47 Hz
Frequency <47 Hz OR
OR Frequency <53 Hz
Frequency >53 Hz
OR
DC voltage/signal/pulse/
chopped signal injection
Current 0 hr. 10 min. 0 sec. 0 hr. 02 min. In-Ip ≥20% of Ib In –Ip <20% of Ib
Mismatch 0 sec. AND
In >Ip
Low Voltage 0 hr. 30 min. 0 sec. 0 hr. 02 min. Voltage <70% of Vref Voltage >70% of Vref
Check 0 sec. AND AND
Current >2% Ib Current <2% Ib
Outage (Power Power OFF = 0 hr. Power ON = Actual Voltage off Actual Voltage On
OFF/ON) 05 min. 0 sec. immediate
Overload Overload (If enabled) 0 hr. 30 min. I >120% Imax I<100% Imax
0 sec.
Microwave Immediate (record 0 hr. 01 min. Any higher frequency Removal of device
only 1 event on first 0 sec. (should magnetic waves, micro
application and only restore after waves >10 mT (or mutually
one event for next 1 min. of last decided)
1 min.) application)
Temperature Rise 0 hr. 30 min. 0 sec. 0 hr. 02 min. Temperature >70˚C Temperature <60˚C
0 sec.
Network Immediate Immediate On removal of card On insertion of card
interface Card
(NIC) removal
(communication
port)
EL WC (earth 0 hr. 30 min. 0 sec. Immediate The difference between The difference between
load – whole phase and neutral currents phase and neutral currents
current) >6.25% of Ib <6.25% of Ib
Table 20: Threshold Values for Exception Generation for Three-Phase Meters
Threshold Value
Persistence Time Persistence Time Threshold Value for
Exception for Occurrence of
for Occurrences for Restoration Restoration of Events
Events
ESD/Jammer Immediate (record 0 hr. 01 min. 0 sec. Immunity up to 50 kV Removal of ESD/JAMMER
only 1 event on first (should restore with NIC and logging of signal
application and after 1 min. of last event >50 kV
only one event for application)
next 1 min)
Magnetic Tamper 0 hr. 2 min. 0 sec. 0 hr. 2 min. 0 sec. >0.5 tesla (T) for <0.5 tesla for permanent
permanent magnet magnet
OR OR
DC magnetic induction DC magnetic induction
>0.2 T <0.2 T
OR OR
AC magnetic induction AC magnetic induction
>10 mT (of any <10 mT
frequency)
Meter Top Cover Immediate Immediate If meter top cover is NA
Open opened
Potential Missing 0 hr. 10 min. 0 sec. 0 hr. 2 min. 0 sec. Voltage <70% of Vref Voltage >80% of Vref
AND AND
Current >2% Ibasic Current >2% Ibasic
Voltage Unbalance 0 hr. 30 min. 0 sec. 0 hr. 2 min. 0 sec. 20% or more between Shall be less than 10%
the phases between the phases and
AND current >2% Ibasic
Current >2% Ibasic
CT Open 0 hr. 10 min. 0 sec. 0 hr. 2 min. 0 sec. Ir + Iy + Ib + In ≥10% of Ir + Iy + Ib + In <5% of
Ibasic (vector Sum) Ibasic.
AND (vector Sum)
Phase current <1% of AND
Ibasic with All current Phase current >10% of
positive Ibasic with All current
positive
CT Reversal 0 hr. 30 min. 0 sec. 0 hr. 2 min. 0 sec. Active current negative Active current positive
AND
I >2% Ibasic
108 Data Analytics for Advanced Metering Infrastructure: A Guidance Note for South Asian Power Utilities
Threshold Value
Persistence Time Persistence Time Threshold Value for
Exception for Occurrence of
for Occurrences for Restoration Restoration of Events
Events
Current Unbalance 0 hr. 30 min. 0 sec. 0 hr. 2 min. 0 sec. Current difference Current difference <20%
≥30% between phases between the phases and
and I min 10% of Ibasic I min >5% of Ib
Low Power Factor 0 hr. 30 min. 0 sec. 0 hr. 2 min. 0 sec. I >1% of Ib I >1% of Ib
AND AND
Power Factor ≤ 0.5 in Power Factor ≤0.7 in
any phase respective phase
Neutral Disturbance 0 hr. 01 min. 0 sec. 0 hr. 2 min. 0 sec. Voltage >145% of Vref Voltage <115% of Vref
and Current >10% Ib AND
OR Current >10% Ib
Frequency <47 Hz AND
OR Frequency >47 Hz
Frequency >53 Hz OR
OR Frequency <53 Hz
DC voltage/signal/
pulse/chopped signal
injection
Outage (Power ON/ 0 hr. 02 min. 0 sec. Immediate Actual Voltage off Actual Voltage On
OFF)
Over Voltage 0 hr. 30 min. 0 sec. 0 hr. 2 min. 0 sec. Voltage >130% of Vref Voltage <110% of Vref
Over Current 0hr. 30min. 0 sec. 0 hr. 2 min. 0 sec. > Preset value (default I <100%Ib
value set at 120% Ib)
Microwave Presence Immediate (record 0 hr. 01 min. 0 sec. Any higher frequency Removal of device
only 1 event on first (should restore magnetic waves, micro
application and only after 1 min. of last waves >10 mT (or
one event for next application) mutually decided)
1 min.)
Temperature Rise 0 hr. 30 min. 0 sec. 0 hr. 02 min. 0 sec. Temperature >70˚C Temperature <60˚C
NIC Card Removed Immediate Immediate On removal of card On insertion of card
Phase Sequence Immediate Immediate Change of phase Restoration of phase
Disturbance sequence sequence