High Performance in Data Management --

lettucestewΗλεκτρονική - Συσκευές

21 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

121 εμφανίσεις

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

High Performance in Data Management
--

Moving from Overload to Insights

Jeffrey D. Taft, PhD

Accenture Global Smart Grid Chief Architect

January 2010

2

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Point of View

Smart grid can be viewed as a new means to
support the big three business process
categories of electric utilities:



Power Delivery



reliable delivery of
sustainable,


economical, high quality electric power



Asset Management



asset utilization
optimization and life cycle optimization



Consumer Experience



all aspects of


consumer interaction with the utility


DR,


DG, PHEV, portals, etc


Installed base of
smart technologies
Information volume
Data volumes will
increase 3,000x!


Smart grid data volumes can be
3,000x

what we are used to handling



It is far more than just meter data
-

many new smart devices and data types



Utilities need new tools, architectures, processes to manage smart grid data

The thing that makes a smart grid
smart
is also the thing that causes

the most difficulty:
DATA

3

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Five Architectural Stages of Smart Grid Data
Management

Data Generation

Transport

Persistence

Transformation

Integration

Integrated data
architecture, CIM

Meters, sensors,
devices, substations,
mobile data terminals

Digital communication,
data collection engines

Analytics, visualization

Real time and enterprise
service buses, SOA, ETL

Data

4

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

We Recognize Five Smart Grid Data Classes, Each
With its Own Unique Characteristics

Operational Data

data representing electrical behavior of the grid
.
This includes
:
Voltage and current phasors
,
real and reactive power flows
,
demand response capacity
,
DER capacity and power flows
,
and forecasts for any of the above
.
Non
-
Operational Data

data representing condition or behavior of assets
.
This
includes
:
power quality and reliability data
,
asset stressors
,
utilization
,
and telemetry
from instrumentation not directly associated with grid power delivery
.
Meter Usage Data

including total usage
,
average demand
,
peak demand and time
-
of
-
day or peak demand values
.
Does not include voltages
,
power flows
,
power factor
,
or
PQ data
,
which fall into other categories
,
even though sourced at the meters
.
Event Message Data

asynchronous event messages from smart grid devices
.
This
includes
:
meter loss of voltage
/
voltage restoration messages
,
fault detection event
messages
,
and event outputs from various technical analytics
.
Meta
-
Data

data necessary to organize and interpret any of the above
.
This includes
:
grid connectivity
,
network addresses
,
point lists
,
calibration constants
,
normalizing
factors
,
element naming
,
and network parameters and protocols
.
5

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

The Three Biggest Data Management Challenges
Facing Utilities In Developing Smart Grids



Matching data acquisition infrastructure to required outcomes



Number, kind, and placement of data measurement devices



Communication networks and data collection engines



Data persistence architectures



Case one: minimize data acq infrastructure for given set of outcomes



Case two: maximize benefits from given data acq infrastructure




Learning to apply new tools, standards, and architectures to manage


grid data at scale



New open standards for interoperability



Distributed architectures



New analytics tools




Transforming business processes to take advantage of smart grid


technology

6

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

We Use A Smart Grid Blueprint Process To Lay Out
Smart Grid Data Management Solutions

We use a formal method set for smart grid data management design, with
specific smart grid processes, such as Observability Strategy Development

7

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Analytics Transform Data Into Information


We
Need Technical and BI Analytics

8

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Latency Requirements for Smart Grid Applications
Drive Data System Architectural Considerations

9

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Analytics for Event Streams: Complex Event
Processing



Smart Grid devices and systems increasingly generate


asynchronous event messages



Such event messages tend to come in bursts and floods when


something (usually bad) is happening on the grid



Normal operations also generate event message streams




Processing event streams requires a different approach



Standard approaches use dynamic queries against (more or less) static data



New approach continually runs static queries against dynamic data streams




Two forms: ESP and CEP



ESP


Event Stream Processing


single stream



CEP


Complex Event Processing


multiple streams



Commercial platforms exist for implementation: CEP engines and
development tools; CEP rule bases codify the business process knowledge

10

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Seven Smart Grid Data Management Tips



Recognize smart grid data classes and their characteristics to develop


comprehensive smart grid data management solutions



Consider how any data source can support multiple outcomes via analytics


to get best value from the sensing infrastructure



Consider distributed data and analytics architectures to solve latency and


robustness issues



Look at the entire smart grid problem when planning data management


solutions (not just AMI) to avoid stranded investments or capability


impediment



Design data architecture to match data classes and analytics/applications


characteristics


a giant data warehouse is not the answer



Look to new tools like CEP to handle new classes of data processing


problems



Develop business process transformation plans at the same time as smart


grid designs


11

Copyright © 2010 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accent
ure
.

Questions?

Contact info:

Jeffrey D. Taft, PhD

Global Smart Grid Chief Architect

jeffrey.taft@accenture.com