Transcript UDW - Abb

© ABB Utilities - 1 2002-01; NM-MW-0169/UTUA/NDF
Mats Werner
Utility Data Warehouse
UTUA/NDF
UDW 200
SPIDER Utility Data Warehouse
Data Warehouse for storage of all kinds of data
Presentation in
One-line diagrams
Presentation as Trends &
Historical information
10.25.32
B9
B8
C1
B3
StationA
T1
T2
B1
B2
B4
B5
B6
C2
B7
Data mining and other
analysis tools
C7
Presentation in
Reports & spreadsheets
Vedlikeholdsplaner: Region Syd
Region/Område
Region Syd
Stasjon/LinjeAlle
Status Alle
Type Alle
Spenning300kv
Skala Uker
1996 27 28 29 30 31 32 33
8
15 22 29 5 August
12
Nr Stasjon/Linje
Anl.DelÅrsak Status 1
Juli
96S432-1
Hasle
10 G2 Ombygging
I Gang
96S530-2
Flesaker G1
Årlig Revisjon
Utsatt
96S532-1
Lysebotn F1
MekaniskOK Reg.
96S541-1
Rendalen G3
UtskiftingOK Land
Lukk
Detaljplan
Ny plan
Utskrifter
Hjelp
© ABB Utilities - 2 -
UDW 200
SPIDER
SCADA/EMS/DMS
Data Warehouse
Based on...
other Utility
Information Systems
UDW – Utility Data Warehouse
© ABB Utilities - 3 -

Oracle-based Data Warehouse for Utility data

Able to store the continuous inflow of data from the power process

Able to store information from any other Utility Information System

Makes all this data available for Data Mining.

High User Performance

High Availability

High Redundancy
UDW – Utility Data Warehouse
Both Utility-wide Data Warehouse and on-line Historian!
Decision Support
Historical Analysis
in Office Environment
in Control Room Environment
Real-Time Network Management Systems
Non-real-time Information Systems
AMS
ERP
BMS
© ABB Utilities - 4 -
CIS
…
Oracle
RDB
EMS
DMS
SCADA
Utility Data Warehouse
Utility Data Warehouse – Original Requirements

RDBMS supporting the latest SQL standard


”Ad Hoc” queries, ”Data Mining”, etc.
Continuous operation

Availability > 99.96 %
 Redundancy
 Continuous inflow of data (every 10 sec, 24h/day)



Response times (incl. picture presentation): 3-5 seconds
Extremely large database tables: > 500 million rows
Continuous calculations

sum, max, average etc.
 Complex calculations using MatlabTM

© ABB Utilities - 5 -


Automatic archiving of data that should not be retained on-line
On-line backup
Automatic maintenance of the database

eliminates the need for a DBA at the customer
Relational Analysis

The ability to analyze relations between data is the basis for Decision
Support and Data Mining

Relational Database Design makes this easy
Object
ID
Time
Stamp
Value
Object
ID
Object
Name
Station
123456
12:34:10
123.456
789123
12:34:10
456789
1
123456
TRF1
2.234123
1
789123
12:34:10
213.4324
1
123456
12:34:20
8.456
789123
12:34:20
2.2455
456789
12:34:20
212.143
Station
Station
ID
Name
446
446
City23
Downtown
TRF2
226
226
FOO
WestRing
456789
BAY1
791
…
2
987654
BAY2
446
…
1
…
Quality
ID
…
1
© ABB Utilities - 6 -
…
Example:
”Give me the name and the location of all lines that has been in alarmed state
during the weekend ”
Location
…
UDW - Open but integrated in SPIDER concept

UDW is a completely open system

UDW is also a well integrated product in the SPIDER concept
© ABB Utilities - 7 -
… as shown in the following areas:

Open Commercial Tools and Interfaces

Application Programming Interface, API

Well Integrated in SPIDER

Powerful Database Structure

UDW as an integration platform
UDW - Open Commercial Tools and Interfaces

Based on Oracle

Oracle is an open commercial relational database

© ABB Utilities - 8 -

delivering ”the industry's highest performance, reliability and security for thousands
of popular applications”

All tools applicable for Oracle is also applicable for UDW

Possibility to access UDW data directly in Oracle, using SQL or Oracle tools

New development in Oracle means increased value of UDW
Open interfaces

ODBC/JDBC

Microsoft COM

SQL*Net

UDW API (see below)
UDW Application Programming Interface, API

Simplifies access to UDW data


From C, C++, Java or Microsoft COM-based applications
Simplifies the creation of custom user applications



Functions for inserting and reading many data points at once
General browser routines, to access data without knowing the data model
An Oracle-independent general database server service,


Implements Additional functionality

User-friendly access routines for setting and getting a range of time-series
data



When data is stored by event, the data is retrieved as if it was stored cyclically
Implements support for Daylight Savings Time, Seamless Data Retrieval and
Data Sets (as described later in the presentation)
The UDW API is used when:

© ABB Utilities - 9 -
for remote SQL access without SQL*Net
accessing UDW data in the SPIDER environment
 accessing UDW data through the Microsoft COM interface
 accessing UDW data in dedicated programs (including UDW-internal)
UDW - Well Integrated in SPIDER

UDW data can be presented in SPIDER pictures,

on-line diagrams

the report and trend tools used in SPIDER

The object oriented Data Engineering for the UDW is integrated with
SPIDER Data Engineering.

The UDW programs are monitored by SPIDER

Alarms and warnings from UDW are sent to SPIDER alarm list
© ABB Utilities - 10 -

e.g. if the data sampling is not functioning properly due to network or
hardware problems
UDW - Powerful Database Structure



Data is stored in a user-friendly way

Uncompressed in open Oracle tables

Database views to simplify common queries
Data model is temporal

history of each object exist in the database

calculation algorithms related to certain objects are time-stamped

Network relations and calculations are therefore correct even when studying
old data, e.g. data restored from off-line archives
Very good performance regardless of system and database size,

© ABB Utilities - 11 -

By the use of the Oracle partitioning technique
Sampling schemes are easily defined using Object Types and Patterns
UDW – Object Types and Patterns

Sampling schemes defined in Object Types

Each Object Type corresponds to one Oracle Table
Objects to be sampled
Object Types
UDW patterns
sampled Object Types
calculated Object Types
10 sec measurands
1 min average measurands
Measurement pattern 2
indications by event
hourly minimum measurands
Indication pattern
state estimator result
daily maximum measurands
Measurement pattern 1
For each Object Type is defined:
 Sampling
method
cyclically – including sampling frequency
by event – with our without integrated dead band
© ABB Utilities - 12 -
 Retention

An object to be sampled refers to a Pattern

A Pattern refers to a number of Object Types

 If
period
it shall be archived when retention period is over
Enables easy set-up of sampling schemes for sampled objects
UDW as and Integration Platform (example)
Platform for application development and cross-platform integration
Developing a simple
application in MS Excel,
using the UDW COM
interface
Combine the data in the
Utility Data Warehouse
with the functionality of
Excel:
© ABB Utilities - 13 -
• Visual Basic programs
• Excel spreadsheet functions
• Excel Charts
UDW - Data Model

Sampled data is stored in Object Types

A simple Power System model is also provided

including e.g. stations, sub systems, point classes

To enable queries and selections based on relational criteria
e.g. ”show all measured values included in station named AMHERST”


Automatically populated from SPIDER Data Engineering

Temporality is provided
Objects
Stations
10.25.32
B9
B8
Station A
C1
T1
C2
T2
B1
Subsystems
Information and relations
between real-world objects
B2
B7
B3
B4
B5
B6
C7
© ABB Utilities - 14 -
Real-world objects

10 sec measurands
1 hour average
daily maximum
Object Types: values related to
sampled real-world objects
Additional data relations possible to define easily

Since an open relational database technology is used
UDW - Data Sampling

Virtually any type of information can be stored in the UDW


SPIDER SCADA/EMS/DMS Sampling


All types of data in the SPIDER real-time database Avanti can be sampled and
stored in the UDW database.

Cyclically, according to a user-defined sampling rate.

“By Event" (upon change), with or without an integrated dead band
Sampling from other Information Systems

© ABB Utilities - 15 -
The actual sampling is highly dependent on the source system from which the
data is extracted.
There are several ways to store data into UDW:

Direct SQL insert, e.g. through database links (assuming an RDB source)

Direct access through open interfaces (e.g. ODBC, JDBC, Microsoft COM)

Dedicated program that uses the UDW API

Dedicated program that uses the ”bulk data inserter” (as used in SCADA-sampling)

Access through Avanti (e.g. by first inserting data into Avanti)
UDW – By Event sampling, integrated dead band

Sampling data By Event with an integrated dead band

a threshold value in percent of the measurement range is defined

measured value changes are
accumulated between samples
© ABB Utilities - 16 -

if the actual value differs only
slightly from the latest stored,
it will thus eventually be sampled
anyway
UDW - Quality Marking

UDW handles quality information together with all sampled values.

propagates automatically through all calculations

validity of calculation result depends on quality of included values


according to what is defined in Data Engineering
Mapping of quality states is defined in Data Engineering

For SPIDER sampling these mappings are set by default, but are configurable.

Possible to distinguish quality information generated in UDW and quality
information originating from the SPIDER Avanti database.
© ABB Utilities - 17 -
Original quality value
Quality value in UDW
Quality value in UDW
after manual correction in UDW
after calculation (a+b)
Object
Value
Quality
Object
Time
Stamp
Value
Quality
Object
Time
Stamp
Value
Quality
123456
123.46
Invalid
123456
12:34:10
123.46
Invalid
22222
12:34:10
125.46
Invalid
789123
2.000
Corrected
789123
12:34:10
2.000
Corrected in
SPIDER
33333
12:34:10
7.000
OK
456789
213.434
OK
456789
12:34:10
5.00
Corrected in
UDW
UDW - Post Disturbance Review

The Post Disturbance Review, PDR, function allows the user to monitor
the process behavior before, during, and after a disturbance
T rigging
event
(discarded)
T rigging
event
In the DE tool it is specified:
T rigging event
(prolonging)
time
© ABB Utilities - 18 -
period
before
event
Deadzone

which data is to be monitored,

which events that will trigger a PDR

time periods

the number of areas available for
storage of disturbance information
period after
event

A disturbance recording can be manually triggered by an operator

It is possible to playback the recorded disturbance data in a Single Line
diagram.

Upon operator request, data for completed disturbances can be saved on
the off-line medium used for archiving
UDW - Data Archiving

For each Object Type it is defined (in Data Engineering):

how long it shall be retained on-line in the database

if it shall be archived

When retention period has passed, data is automatically exported to the
chosen off-line medium

UDW maintains a library system for the archived data.


Seamless presentation is supported


If restored data and on-line data connects in time
Standard software can be used for the Archiving function.

© ABB Utilities - 19 -
When data shall be restored the user specifies the type of data and for which
time interval the data shall be restored
such as Legato NetworkerTM
UDW - Advanced Data Processing

Powerful and versatile data processing capabilities

Times series calculations:


Snapshot calculations



All power of the Matlab tool is available for utilization
Calculation wizard to ease simple calculation definitions
Complete ROLAP1 aggregation support (configurable)

Data can be automatically aggregated in all dimensions



© ABB Utilities - 20 -
Raw data and processed data are stored and handled the same way,
i.e. both can be used for e.g. data mining
Both during sampling and at manual entry
Quality coding of all values


E.g. summary for e.g. a station or the whole power network
Automatic recalculation of all2 formulas


Any formula/calculation on objects for many points of time
Matlab used as calculation engine


Any formula/calculation on objects for a certain point of time
Custom calculations


Sum, Average, Max, Min
Propagates through calculations
Calculation patterns (“templates”)

For easy connection of sampled objects to a set of pre-defined calculations
and sampling profiles
1ROLAP
= Relational
On Line Analytic
Processing
2Restrictions
apply on
custom calculations to
prevent recursive
deadlocks
UDW - Report and user interaction functions

UDW data is easily accessed from various client tools

There are four main access routes:

Through SPIDER (historical data originating from SPIDER can be presented
in SPIDER trends and spreadsheet reports, and in one-line diagrams).

Direct SQL read (through SQL*Net, Oracle tools, Microsoft Query etc.)

Direct access through open interfaces - ODBC, JDBC, COM, (e.g. Microsoft
Office programs).

Through the API or Java API (dedicated programs accessing UDW data)
Presentation in
Reports & spreadsheets
Presentation in
One-line diagrams
Presentation as Trends &
Historical information
Vedlikeholdsplaner: Region Syd
10.25.32
Station
A
T1
B4
C2
T2
B1
B3
Region/Område
Region Syd TypeAlle
StatusAlle
Stasjon/Linje
Spenning
Skala Uker
Alle
300kv
1996 27 28 29 30 31 32 33
15 22 29 5August
12
Nr Stasjon/Linje
Anl.Del
Årsak Status1 8 Juli
96S432-1
Hasle
10 G2Impinging
I Gang
96S530-2
Flesaker G1 Årlig Revisjon
Utsatt
96S532-1
LysebotnF1 Mekanisk
OK Reg.
96S541-1
RendalenG3 Utskifting
OK Land
B9
B8
C1
B2
B5
B6
Data mining and other
analysis tools
B7
C7
© ABB Utilities - 21 -
Lukk
UDW 200
Database
(Oracle)
Detaljplan Ny plan
Utskrifter
Hjelp
UDW – Client example: SPIDER Time Tagged Trends

The Time Tagged Trend (TTT) function is used for presentation of
historical data in curve diagrams.

Up to 16 curves in each diagram


Automatic scaling of value axis
depending on curve values.

The values can be fetched from
different object types


© ABB Utilities - 22 -
each with its own value axis.
Can have time offsets relative to each other.
Quality coding
UDW – Client example: SPIDER Single Line Diagram

UDW data can be shown in the Single Line Diagram

© ABB Utilities - 23 -

by selecting a historic time and an Object Type or a Data Set.
Possible to start a replay of the data

using tape-recorder like controls: play, advance, rewind, stop

the selected time is automatically changed with a given speed.
UDW – Client example: Microsoft Excel
© ABB Utilities - 24 -


There are several ways to get UDW data into Excel:

via the SPIDER Excel add-in

directly through the
Oracle ODBC driver

copy/paste or export/
import data retrieved
with another tool,
e.g. Oracle™ Browser

use the UDW COMinterface in an Excel
Visual Basic program,
that reads the appropriate information from
UDW
Once the data is in Excel, its built-in functions can be used

i.e. data can be calculated on, charts can be created etc.
UDW – Client example: Microsoft Query
© ABB Utilities - 25 -

Using the Microsoft Query tool it is possible to create custom queries to
the UDW in a user-friendly way, from the Windows environment
© ABB Utilities - 26 -
UDW – Client example: MATLABTM Graphics

Using the integrated MatlabTM server it is possible to define
advanced formulas and calculations

Using a MatlabTM client tool it is also possible to benefit from
the advanced graphical user
interface of MatlabTM, and that
way visualize data and correlations found in UDW data.
UDW - Data Sets

The Data Set function facilitates presentation of UDW data

Groups a set of Object Types


A Data Set may include object groups from different SPIDER
concepts: Measurands, Process values, Indications etc.

© ABB Utilities - 27 -
Can thereafter be referenced as one identity
Practical e.g. when presenting historical data in Single Line diagram
UDW - Seamless Retrieval – Data Set example


Measured values are often sampled with several frequencies, where higher
frequency normally corresponds to a shorter retention period

Each sampling frequency corresponds to one Object Type

For some points of time there may exist data in more than one Object Type
By creating a Data Set of these Object Types, it is possible to

address the sampled values as one entity

get best possible resolution for all points of time
Data that are retrieved when asking for data from
t1 to tn:
10 second

t1-t2 1 hour values

t2-t3 1 minute values

t3-tn 10 second values
1 minute
© ABB Utilities - 28 -
1 hour
t1
t2
t3
tn
UDW - Daylight Saving Time
Calendar time is normal time
with Daylight Savings offset

The UDW has full support for Daylight Savings Time.

It is possible to store both normal time and calendar time in UDW.


It is recommended for most usages to only store normal time and let the UDW
API take care of the DST conversions.
If the times are retrieved through the UDW API,
the UDW takes care of the conversion

both when it comes to individual points of time and
larger time spans.
The UDW API is used when
accessing UDW data in the
SPIDER environment, by the
MS COM interface and when
accessing UDW data in
dedicated programs
Example: If the user wants to get data for a time span that includes a switch between
normal time and Daylight Savings Time, there will be the correct number of times
retrieved (e.g. an extra hour if going from daylight savings to normal time).
© ABB Utilities - 29 -

The calendar time is used in display pictures and reports
UDW - Authority

UDW uses standard Oracle™ authority routines

Users are assigned to a Profile and one or more Roles

Roles provides specific access rights

Profiles are used to set session parameters


Such as the maximum CPU time for a session
Examples of Roles defined in UDW:

system administrator, privileged user, common user, etc.
Number of sessions
Max CPU per session
Connect time during session
etc.
© ABB Utilities - 30 -
Profiles are used to limit the risk for
low priority users or user groups to
impair the performance of the UDW
database by using badly formed
SQL-statements, for instance, a full
table scan.
Profile
User
Database creator
System administrator
Role
UDW - Audit Trail

The Audit Trail function in UDW tracks all changes made in the UDW
database

For all manual updates the following data is stored:


old and new value

the affected time and object identity

the point of time of the modification

an operator identification
The standard Oracle™ audit trail function can also be used


Tracking of insert is not recommended

© ABB Utilities - 31 -
providing tracking of user actions on a table level.
the insert rate is very high in the system, and auditing insert would degrade
performance.
UDW – Planned Values


The UDW supports storage of planned values

A ”window” of future data points are a opened for storage

Can be used to store e.g. forecasts or schedules
Defined as an Object Type in Data Engineering

© ABB Utilities - 32 -

Handling of retention period, archiving etc. just like other Object Types
A number of parameters are possible to define:

if the values shall be overwritten by actual sampled values or not

if the values shall initially be set to blank or ”invalid”

the size of the ”window” – the time span to be opened in the future

initial time period of the plan, i.e. including historical times
UDW - Data Engineering

The Data Engineering of the UDW is object oriented and dialog driven.

Is totally integrated with the Oracle-based SPIDER data engineering tool.
© ABB Utilities - 33 -

All data engineering logic for UDW is contained within the UDW product.
UDW – Specification of Object Types

For sampled Object Types:

sampling method: by event (including dead band) or cyclically (including
frequency)

retention period, i.e. how long the object shall be retained on-line in the
database

mapping between the source system (SCADA) and UDW quality codes

properties to be sampled from source system (SCADA) to UDW

© ABB Utilities - 34 -

for instance the actual value and updated flag for a measured value
For calculated Object Types:

source Object Type (a sampled Object Type or another calculated Object
Type)

type of calculation (e.g. Average, Sum, Max, Min)

required percentage factor of valid source data to get a valid result

if time stamping shall be in the beginning or end of the calculated period
UDW - Backup and recovery

For safety copy reasons the on-line data in UDW can be backed up, using
standard Oracle™ functionality.

The backup is done in a way so that all data, including calculated values,
are consistent.

The interval between backup activities is configurable

UDW Safety Backup and UDW Archiving are complementary functions!
© ABB Utilities - 36 -
Oracle
RDB
Utility Data Warehouse
UDW on-line system
UDW
Safety
Backup
UDW
Archiving
Archiving of a safety
backup of on-line data
Archiving of data older
than retention period
UDW - Configuration

The UDW can be configured in different ways to provide the required
degree of redundancy and data security.

Data Redundancy

The data in the UDW database is stored redundantly so that if one disk fails,
no data is lost

To get full data redundancy, RAID disks of type 0+1 are used


Server Redundancy

© ABB Utilities - 37 -

striping of data and disk mirroring is applied
The UDW servers are redundant so that if one UDW server becomes
inoperable (e.g. due to hardware errors) another server is taking over.
Different configurations are applicable depending on the requirements on
allowable downtime and loss of data during switch-over.

Redundant Server solutions, with automatic switchover

Emergency Center solutions, with manual switchover

Single Server solutions
UDW - Redundant Server solution

Redundant Server solution

No single point of failure

Automatic switchover when the on-line server malfunctions

Separate server for Safety Backup and Archiving
Backup
Server
DLT Automated Tape Library with duplicate
drives for archiving and restoring
SCSI
SCSI
Backup LAN
SCADA/EMS/DMS

The two UDW servers are connected in a
cluster configuration

Both servers receive data to be stored
from the sampled system.

One of the servers is on-line while the
other is hot standby and
ready to take over the processing.

The on-line server updates the UDW
database with the sampled data and
calculation results
Memory Channel
© ABB Utilities - 38 -
UDW
Server 1
Server-RAID
Server-RAID
Connection
Connection
SCSI
SCSI
UDW
Server 2
RAID Disk
LAN
UDW – Emergency Center solution

Emergency Center solution

No single point of failure

Manual switchover when the on-line server malfunctions

Original UDW rebuilt by restoring a safety backup
SCADA/EMS/DMS
HIS Server, online
RAID Disk
Server-RAID
Connection
SCSI

The standby server is kept up-to-date with
the on-line server, using Oracle Standby

It is possible to restart the standby database
as the primary UDW database with minimal
loss of time and data.
LAN
SCADA/EMS/DMS
HIS Server, standby
Emergency Control Center
© ABB Utilities - 39 -
RAID Disk
W
A
N
Server-RAID
Connection
SCSI
LAN
UDW – Single Server solution

Single Server solution

With either RAID or disk shelf
SCADA/EMS/DMS
Disk or RAID
Shelf
UDW Server

SCSI
© ABB Utilities - 40 -
LAN
If a disk shelf is used instead of RAID, the
single disk controller is located in the UDW
Server