0% found this document useful (0 votes)

34 views225 pages

Analyzing PISystem Data

Othi sabak Jkns Kama

Uploaded by

King Lala

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

34 views225 pages

Analyzing PISystem Data

Othi sabak Jkns Kama

Uploaded by

King Lala

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 225

Analyzing PI System Data

Version 2018
How to Use this Workbook
Each Main Heading describes a
high-level valuable learning topic.

Your objectives are skills you

can expect to learn in this
segment.

New concepts are presented as

level 2 headings.

Throughout the class you will be

presented with questions and
challenges to help you learn.

The majority of your time will be

spent learning new skills via
hand-son exercises, either in
small groups or on your own.

Icons help you identify themes,

like exercises, tools, tips, or
documentation references.

User manuals, Learning workbooks, and other materials used in class can be
downloaded from http://techsupport.osisoft.com. Login to an OSIsoft technical
support account is required.

Page i
Analyzing PI System Data

Software Versions Used in this Document

The list below describes the software versions used in this version of the course.

Software Version
PI DataLink 2017 SP2
Microsoft Office 2016
PI ODBC Driver 2016 R2
PI Integrator for Business Analytics Advanced Edition 2018
PI OLEDB Enterprise 2017 R2
Microsoft SQL Server 2014
PI Data Archive 2018
PI Asset Framework 2018
PI Vision 2017 R2 SP1

ii
Contents
1 Welcome ................................................................................................................... 5
1.1 Course Environment ...................................................................................... 5
1.2 Review PI System Architecture .................................................................... 6
1.3 Assets and Tags – The Basic Building Blocks in the PI System .............. 8
1.4 Discussion ....................................................................................................11

2 Business Intelligence ............................................................................................12

2.1 Intro to Power BI ..........................................................................................13
2.2 Directed Activity – Inspect a Sample Power BI Report ............................13

3 Part 1 – Power BI Reports using PI Integrator for BA ........................................19

3.1 Directed Activity – PI AF Hierarchy and Data Set .....................................19

4 PI Integrator for Business Analytics....................................................................21

4.1 Architecture ..................................................................................................21
4.2 PI Integrator Web UI .....................................................................................21
4.3 Directed Activity – Create the Transformer Loading View ......................23

5 Building the Distribution Network Reports .........................................................31

5.1 Preparing and Importing the Tables ..........................................................31
5.2 Building the Report Visuals ........................................................................34

6 Part 2 – Power BI Reports using PI OLEDB Enterprise .....................................87

6.1 Directed Activity – PI AF Hierarchy and Data Set .....................................88

7 PI Analysis Service ................................................................................................91

7.1 Capabilities of the PI Analysis Service ......................................................91
7.2 Expressions ..................................................................................................92
7.3 Rollups ........................................................................................................103

8 Event Frame Generation .....................................................................................110

8.1 What are Event Frames? ...........................................................................111
8.2 Event Frame Generation ...........................................................................120
8.3 Discussion ..................................................................................................126

9 Analyzing Events .................................................................................................127

9.1 Objectives ...................................................................................................127
9.2 PI Event Frames in PI System Explorer ...................................................127
9.3 PI Event Frames in PI DataLink ................................................................133
9.4 PI Event Frames in PI Vision.....................................................................140
Analyzing PI System Data

9.5 Discussion ..................................................................................................147

10 PI OLEDB Enterprise SQL Queries ...................................................................148

10.1 Dissecting the Syntax ...............................................................................148
10.2 PI OLEDB Provider or Enterprise? What’s the difference? ...................150
10.3 Table Aliases ..............................................................................................156
10.4 JOIN Statements ........................................................................................156
10.5 Built-in Functions.......................................................................................163
10.6 Data Tables .................................................................................................164
10.7 Data Transpose Functions & Function Tables .......................................168
10.8 UNION Statements .....................................................................................179
10.9 Saved Views ...............................................................................................183
10.10 Importing PI OLEDB Enterprise data to Power BI ..................................188
10.11 Discussion ..................................................................................................194

11 Building the “Fleet Generation” Report ............................................................195

11.1 Preparing and Importing the Tables ........................................................195
11.2 Augmenting the Data using DAX..............................................................208
11.3 Configuring the Visualizations .................................................................210

12 Final Exercise: Create a Report .........................................................................213

13 Appendix A Substitution Parameters ................................................................216

14 Appendix B Performance Equation Operands and Functions .......................218

15 Appendix C PI SQL Commander Table Relationships .....................................223

4
Analyzing PI System Data

1 Welcome

Welcome to the Analyzing PI System Data Course!

Since you are attending this class, you should have some experience with OSIsoft
Client Tools (PI ProcessBook, PI DataLink, PI WebParts and PI Vision), either using
displays, reports or webpages previously created to analyze your data, or creating
these displays, reports and webpages so that others in your organization have
access to all the powerful data that resides in the Data Archive and data external to
the PI System.
The basic tasks within these tools are presumed to be understood; what you will
experience here can be seen as a factory of ideas, a space for OSIsoft customers to
realize how powerful existing data can be when analyzed with the advanced options
of our tools and additional third party tools, and integrated with non-PI data.
Hope you enjoy!

1.1 Course Environment

The environment for this course is being hosted with Azure. The environment has 3
VM and contain the following:
 PIDC – Domain Controller
 PISRV01 – The server environment
o Microsoft SQL Server 2014
o PI Data Archive 2018
o PI AF Server 2018
o PI AF Client 2018
o PI Vision 2017 R2 SP1
o PI Integrator for Business Analytics Advanced Edition 2018
 PICLIENT01 – This is the primary working environment.
o PI System Explorer 2018
o Microsoft Office Professional Plus 2016 (64-bit)
o Microsoft PowerBI Desktop 2.59.5135.781
o Google Chrome 68.0.3440.106

The userid for each student is pischool\student01, password will be provided by

the instructor.

Page 5
Analyzing PI System Data

1.2 Review PI System Architecture

Objectives:
 Define the components of a PI System
 Draw a diagram of the architecture of a PI System

The PI System Described

The PI System collects, stores, and manages data from your plant or process. You
connect your data sources to one or more PI Interface nodes. The Interface Nodes
get the data from your data sources (control systems, instrumentation, etc) and send
it to the Data Archive. The data is organized and given context using PI Asset
Framework. Users get data from the Data Archive and Asset Framework and work
with it using a variety of client tools, such as PI Vision and PI DataLink.

6
Analyzing PI System Data

1.2.2 Architecture of a Typical PI System

Sometimes the architecture can be very simple. Some customers have as few as
one or two interfaces feeding data to a single Data Archive. Access to data is
through the single Data Archive.

PI AF Server
Asset Analytics
Event Frames
Network Devices Notifications
IT Monitor

Process
HQ PI Data
Controls
Archive
DCS / PLC
OPC

Relational
Database

SQL Server

Text /
Flat Files

PI ProcessBook
PI Manual Logger

PI to PI

PI DataLink

Process Control
Data Sources DMZ Corporate LAN VPN
Network (PCN)

There are often several Data Archives in an organization, aggregating data from
lower levels. Some corporations have Data Archives dedicated to servicing their
clients with restricted company data.

Page 7
Analyzing PI System Data

1.3 Assets and Tags – The Basic Building Blocks in the PI System
Objectives:
 Define an AF Asset with its components element and attributes.
 Define four attribute types: Static (None), PI Point, Formula, and Table
Lookup.
 Define a Data Archive Tag with the attributes Tag Name, Descriptor, and
Point Source.
 Define the different data types that can be stored in Data Archive Tags.

AutoCreate

Tags Assets
Figure 3: Tag Auto Creation

What is an Asset?

The AF Server is a part of the PI System. It contains asset or “metadata” usually

organized according to the assets containing the attributes being monitored. AF can
be helpful to users of the Data Archive who know the assets, but are not familiar with
attribute nomenclature. With assets, data can be located without understanding the
technical details of each piece of equipment. Organized assets help find all of the
attributes associated with a specific piece of equipment.

What is an AF Attribute?

Attributes represent a unique property associated with an asset. The attribute

maybe a constant, a value from an internal AF table, a value from an external
database or a storage point for data in the Data Archive. An AF attribute is simply a
single point of measurement. The point has been the traditional storage method of
data in the Data Archive. The AF Server can automatically generate points as
assets are created.

Some Basic Properties and Why They Are Important to You

AF attributes and Data Archive points have a set of properties that define them.
Some common properties used in client tools are for display or informational
purposes.

8
Analyzing PI System Data

Attribute name
The attribute name is similar in concept to the point description. A detailed name for
the attribute may help the user identify the source of the information.

Figure 6: Attribute Name

Tag name
Unique name is used to create points for storage in the Data Archive. Points for data
attributes storage can be built through AF templates using substitution parameters
for local naming convention or can be searched for on the Data Archive.
Creating points through templates, lends consistency in nomenclature making
searches easier for PI Administrators. For example, which might be easier to locate
in a search?

Point: M03_E1P1_MOTDRV1202_RUNSTAT
Attribute: Machine3 Enclosure 1 Panel 1 Motor Drive 1202 Run Status

Substitution parameters are variables placed in attribute templates for PI point and PI point
array data references representing portions of the AF hierarchy.

For example, %Element% is a substitution parameter that represents the element name. After
you create an element based on that template, you tell AF to create the data reference. When
AF creates the reference, it substitutes the current element name wherever %Element% is
present.

Page 9
Analyzing PI System Data

Descriptor
This is the human-friendly description of the Data Archive Point, similar to the
attribute. The descriptor is often a search criterion since the point name is not
always intuitive. Often the point name is some sort of abbreviated convention and
the descriptor captures the “full name.”
Point source
Points can be related to their interfaces that collect the data by a point attribute
called pointsource. Grouping by point source allows all of points associated with a
particular device to be identified by searching for all points of a certain point source.
This assumes that the user knows the point sources in use and that will not be true in
most situations.
Point type
The PI point attribute that specifies the data type for the values that a point stores.
The possible point types include int16, int32, float16, float32, float64, digital, string,
BLOB, and timestamp.

10
Analyzing PI System Data

1.4 Discussion

This is a discussion designed to maximize learning in a specific topic area.

Your instructor will have questions, and will prompt for communication within
the class. This is an open ended section and the result depends on your
needs.

Objective: The AF data is to be included within our Business Intelligence tools.

What data do we want to see? This can be in the context of your own system, or the
example AF databases available in the class.

Approach
 What do we currently use our AF databases for regrading report generation?
 Within the example database, what else do we want to know? What is
missing?
 Pros and cons of including external asset related data within the AF structure
 What sort of BI tools would we want to use to view this data?

Estimated Completion time 15 minutes.

Page 11
Analyzing PI System Data

2 Business Intelligence

Business intelligence (BI) tools offer solutions to quickly analyze raw, un-normalized,
multidimensional data. In concert with historical values from the Data Archive,
metadata and calculations from Asset Framework, and business intelligence tools,
users can quickly create interactive reports to gain insight on business and
operational processes.

Later on in the course, we will explore the process of preparing the Asset Framework
model to add additional dimensions of information to our AF database. The next step
is extracting desired information (process data, metadata, and event frame data)
from the PI System through PI Data Access tools. This data will be incorporated into
a BI cube and used to develop interactive reports that allow us to “slice and dice” our
data and bring meaning to our multidimensional data cube.

The Distribution Network and Fleet

Generation databases have a
comprehensive amount of information
including a hierarchy of substations,
metadata for each asset. The figure to the
right depicts a data cube that captures
metadata and real-time data of generating
units.

Inclusion of additional attributes

through table lookups and analytics
on existing attributes allow for the
expansion of additional columns (or
dimensions) to the data cube above.
Further, historical data, interpolated
or compressed, add an additional
dimension of information that bring
more meaning in Business
Intelligence reports.

12
Analyzing PI System Data

In the next several chapters in the course, we will be using a pair of AF databases to
expose meaningful data that will help management and engineers make better, more
informed decisions. Specifically, we will add value through the following:

1. Expose the database in a simpler structure for data processing.

2. Develop analytics within PI AF and PI Integrator for Business Analytics
3. Import the data into Microsoft Power BI
4. Draw actionable conclusions from the resulting data sets in our reports

2.1 Intro to Power BI

Power BI is a business analytics service and client provided by Microsoft. It provides interactive
visualizations with self-service business intelligence capabilities, where end users can create
reports and dashboards by themselves, without having to depend on information technology
staff or database administrators.

Some of the benefits of Power BI:

 Less work than Excel for more complex analysis and visuals
 Can solve problems that are simply too large for Excel and PI DataLink
 Cheap – Free download or $9.99 / month per user for Power BI Pro
 Live reporting and centralized web-based dashboards in Office 365
 Slick visuals including 3rd Party Visuals in Microsoft AppSource

2.2 Directed Activity – Inspect a Sample Power BI Report

In this part of the class you will perform a learning activity to explore the
different concepts presented in this chapter or section. You may be invited to
watch what the instructor is doing or perform the same steps at the same time.
You may play a game or hold a quiz. Your instructor will have directions.

Objective:
 Explore a sample Power BI Report.

Page 13
Analyzing PI System Data

Approach
We’ll start by getting a feel for Power BI using a pre-built report. Open C:\Class\Part 1 - PI
Integrator for BA\Solutions\Transformer Loading.pbix

Right now we’re looking at the Report View where the report and visuals are configured. Start
clicking on the visuals and the rest of the report will be filtered to only include the selected items.
This is often referred to as slicing and dicing the data.

Click on the Data Tab to inspect the data set we’ll be working with:

14
Analyzing PI System Data

Note that all the columns are available in the Fields List:

Go back to the Report View, click on the Pole Transformers Table visual, and note the
Visualizations Pane and Fields Pane. These sections are where the bulk of the configuration
takes place. Columns from the data set are dragged and dropped from the Fields Pane onto the
various sections in the Visualizations Pane. We can see that the Secondary Transformer,
Phase, Service Hours, Average of Loading, and PI Vision columns are being displayed in the
table.

Page 15
Analyzing PI System Data

Add the Voltage Average column to the table by selecting the table and doing a drag and drop:

The Voltage Average will be displayed in the table, but by default all the voltages from all the
rows will be summed by transformer. It makes more sense to summarize these as an average,
so click the drop down and change the summary to Average:

16
Analyzing PI System Data

Resize the column directly on the visual similar to Excel:

Average of Voltage Average is a pretty weird header name, so rename it to Average Voltage:

Other Formatting Options are available by clicking the paint roller icon:

Page 17
Analyzing PI System Data

We’ll go over more formatting options in the exercises, so for now just bump up the text size of
the column headers and resize columns and visuals so everything fits:

Your report should now look something like this.

We will build this report from scratch in a future exercise.

18
Analyzing PI System Data

3 Part 1 – Power BI Reports using PI Integrator for BA

This course will be broken down into two main sets of exercises. In Part 1, we’ll use
PI Integrator for BA to publish data from PI and spend a lot of time configuring Power
BI. In Part 2, we’ll make modifications to a PI AF hierarchy and then use PI OLEDB
Enterprise to extract the report data.
In Part 1, we will be working with a data set for a power distribution company, which
includes electrical characteristics for over 1500 single-phase transformers. The
source data comes from a real PI System and will be published in a data-science
ready format using PI Integrator for BA. Once this is done, we'll configure an array of
Power BI visuals and integrate the results with PI Asset Framework and PI Vision.
The transformers we will be analyzing are secondary transformers that deliver power
to homes and businesses, which you may have seen on a pad or pole in your own
neighborhood. There are thousands of them to keep track of, making this a difficult
problem to solve using Excel.
The transformers themselves are not actually instrumented. The power and voltage
characteristics we will analyze have actually been computed by rolling up child
Meters in PI AF.

3.1 Directed Activity – PI AF Hierarchy and Data Set

Objective:
 Better understand the data set used in the following chapters

We will take a few minutes to understand where the data set came from and relate the sample
Power BI report back to the PI System. We are working with a data set for a fictitious power
distribution company. They have built a PI AF Hierarchy for their transformers and meters
serving a number of geographical areas. In this course, we will focus on analyzing the
transformers.

Page 19
Analyzing PI System Data

Open PI System Explorer and head to the Distribution Network AF database. Drill down to a
level with transformers (names starting with PT_) and inspect the available attributes. We will be
using a sub-set of these attributes for all of our analysis, in addition to leveraging the AF
hierarchy.

Note that because the Transformers are not instrumented, Voltages, Voltage Quality, and Wh
Delivered are rolled up from the child meters using Asset Analytics.

Data from this PI AF hierarchy will be published for use in a Power BI report in a later
exercise.

20
Analyzing PI System Data

4 PI Integrator for Business Analytics

Getting the data out of the AF structure and into the client tools requires the use of integration
software such as the PI Integrator for Business Analytics or PI OLEDB Enterprise. This chapter
will discuss the former method of extracting the data.

4.1 Architecture
The PI Integrator for Business Analytics resides on a web server between the client
machines and the source AF server. As such, we are not connecting directly to the
AF Server but instead to a web server that contains a cache of our desired
information. The architecture within our class system however has both the AF
Server and Web Server residing on the PISRV01 machine. The PI Integrator for
Business Analytics site can be accessed via https://pisrv01.pischool.int:777 or from
the desktop. If prompted for credentials, enter your student account, as this has
been given access rights.

Client
PI AF Server Web Server Microsoft Excel
PI AF databases PI Integrator for Business Microsoft Power BI
Analytics Tibco Spotfire
PI SQL DAS Etc.

4.2 PI Integrator Web UI

Views can be created within the PI Integrator portal that is hosted on the Web Server
machine.
A list of previously generated views is present within the portal on the My Views
page, allowing for previewing and maintenance. These existing views can also be
cloned and modified, allowing different views to be created and utilized within BI
client tools.
The following is a breakdown of the My Views page layout, and the different
operations available.

Page 21
Analyzing PI System Data

Note: The information regarding the My Views page layout is available within the PI Integrator
for Business Analytics User Guide.

The My Views page shows details about your views.

1. All the views to which you have access are listed in the table
2. Click to create an Asset View that is based on Elements and Element Templates
3. Click to create an Event View that is based on Event Frames and Event Frame Templates
4. Click to create a Streaming View that is based on Event Frames and Event Frame
Templates
5. To modify a view, select the view in the table and click Modify View.
6. To delete it, click Remove View. Deleting a view removes data from the buffer, therefore
freeing up space. However, this does not free up the available output streams allowed with
your license.
7. For the selected view, the Overview, Log and Security tabs provide the following details
about that view:
· Overview indicates whether the view has been published. This tab also summarizes
information about the view, such the PI AF database it uses, when the view was last run,
and the shape that it uses. If the view is currently being published, the run status bar
indicates progress and you have the option to stop the publishing process.
· Log displays log information. You can adjust the start and end times, and you can filter
the messages to display those of a certain severity, for example, critical errors.
· Security shows who has access to the view, and if you have sufficient privileges, allows
you to change the level of access
8. The red message counter icon at top right show that there are warning and error messages

22
Analyzing PI System Data

recorded by PI Integrator for Business Analytics. Click the icon to open the message list.

9. Click the gear icon at top right to see the version of PI Integrator for Business
Analytics and AF you are using.

4.3 Directed Activity – Create the Transformer Loading View

In this part of the class, you will perform a learning activity to explore the
different concepts presented in this chapter or section. You may be invited to
watch what the instructor is doing or perform the same steps at the same time.
You may play a game or hold a quiz. Your instructor will have directions.

Objective:

 Use the PI Integrator for Business Analytics to create an Asset View, which
will be used in later exercises.
Approach:
Open Google Chrome and Navigate to the PI Integrator for BA Web UI at
https://pisrv01.pischool.int:777
Click Create Asset View and name it Transformer Loading, click Create View:

Click Create a New Shape

Page 23
Analyzing PI System Data

Select Distribution Network as the AF Database, then drill down to PT_XYZ0343.

Drag and drop Alajuela to the Shape Builder

Edit the Filter on Alajuela:

24
Analyzing PI System Data

Clear the Asset Name Checkbox, Change it to filter on the Headquarters template,
click Save:

Drag and drop Avenida Central to the Shape configuration, and change it to filter on
the Substation Template:

Page 25
Analyzing PI System Data

Repeat this pattern for Transformer 1 (Template = Single Transformer), Colegio

Cientifico (Template = Circuit), X Phase (Template = Phase).
Drag and drop PT_XYZ0343 and select Secondary Transformer as the Template,
this time check the box to search derived templates.

The shape configuration should look like this:

26
Analyzing PI System Data

Click PT_XYZ0343 then hold control and multi-select Loading, Maximum KVA, Rated
KVA, Transformer Type, Voltage Average, Voltage Maximum, Voltage Minimum,
Voltage Quality, and Wh Delivered Load. Drag and drop these selections to the
Shape configuration.

There should be over 100 matches in the preview, click Next in the top right corner.

Page 27
Analyzing PI System Data

We now see a preview of the data using the default Time Range and interpolation
mode.

We want to publish Hourly data for the time period 01-Jun-17 00:00:00 to 31-Aug-17
23:00:00. Modify the Start Time and End Time and click Apply:

Click Edit Value Mode and change the time step to 1 hour, then Save Changes:

The TimeStamp column should now reflect changes to the Start, End, and Value
Mode:

28
Analyzing PI System Data

Now we’ll add some additional time columns that will come in handy later when
building the reports. Click Add Colum. Select the Time Column tab. Select Month,
Month Name, Week of the Year, and Hour, then click the arrow to bump them over to
the right:

Click Display 5 Time Columns:

Page 29
Analyzing PI System Data

Now the time ranges and columns have been specified, click Next.

Now we can choose what target to publish to. This depends on the platform used to
support front-end application, but for our purposes we’ll publish to a SQL Server.
Select SQL Server for the Target Configuration, Leave Run Once checked, and click
Publish:

It will take a few minutes to publish the data.

30
Analyzing PI System Data

5 Building the Distribution Network Reports

We will now spend a significant amount of time configuring Microsoft Power BI

reports. The first step is importing the data.

5.1 Preparing and Importing the Tables

Now that the Transformer Loading table has been published, we will import the SQL table into
Power BI.

Directed Activity – Import Data from Microsoft SQL Server.

Objective:
Import the Transformer Loading table.

Approach:
Open Microsoft Power BI

Select Get Data -> SQL Server.

Page 31
Analyzing PI System Data

Enter PISRV01 as the server name and click OK.

If Prompted, Leave “use my current credentials” selected and click Connect:

There may be a warning that the connection is not encrypted, this can be safely ignored, click
OK:

32
Analyzing PI System Data

Expand the PIInt database and Select the Transformer Loading table, click Load

Note that about 3.8 million rows have been imported.

Page 33
Analyzing PI System Data

5.2 Building the Report Visuals

Now that the Transformer Loading table has been imported, the rest of the chapter will be a
walkthrough of configuring various report visuals.

In case there were mistakes or problems with the previous steps, a starter .pbix file
has been created with the raw data set already imported with columns that will match
the exercises exactly.
Open C:\Class\Part 1 - PI Integrator for BA\Starter File - Part 1 Distribution
Network.pbix and use this as a starting point for the remaining exercises.

Directed Activity – Network Load Profile

Objectives:
 Configure a Hierarchy
 Configure a Matrix to show average loading per transformer
 Configure a Clustered Column Chart to show the average energy delivered by
Headquarters
 Configure a Line Chart to show the average energy delivered by hour of the day and by
day of the week
In this exercise, we are going to analyze the loading characteristics of the transformers in our
distribution network. We want to look at the demand curve for various levels of the distribution
network to help forecast the demand seen by various transformers based on the hour of the day
and day of the week.

Loads are expressed as a percentage of the rated value, as well as a raw watt-hours value. The
resulting report will help us understand which transformers are working under the highest loads,
and see how power consumption changes over time. General Steps:

34
Analyzing PI System Data

Approach:

Configuring the Hierarchy

We will now create a hierarchy. In the Fields List, click the ellipses next to Headquarters and
select New hierarchy:

Within the fields list, drag and drop the Substation field on top of the new Headquarters
hierarchy:

Page 35
Analyzing PI System Data

Repeat for Single Transformer, Circuit, Phase, and Secondary Transformer and reorder so
that Substation is below Headquarters in the hierarchy.

Monthly Average Loading - Matrix

We’re going to display the monthly average loading on a per transformer basis and leverage the
Hierarchy we just set up.

Add a Matrix to the canvas by clicking the Matrix icon in the Visualization Pane:

36
Analyzing PI System Data

Drag and drop the Headquarters Hierarchy for the Rows, Month Name for the columns, and
Loading for the Values:

Change the value field to summarize by Average Loading:

Page 37
Analyzing PI System Data

Now we will drill down into the hierarchy. Press the drill down button:

Next, navigate through the layers by clicking on the Headquarters or going right click -> drill
down:

38
Analyzing PI System Data

Expand all the way to the bottom until the Matrix looks like this:

Navigate back up the layers by clicking the Up Arrow on the top left of the visual:

Page 39
Analyzing PI System Data

Drill back down to the lowest level, and turn off Drill Down mode. This will allow us to filter
the rest of the report by clicking on levels and transformers rather than drilling.

To put the Months in the correct order, we will sort the Month Name column in the data set by
the Month column where the months are numbered. Go to the Data View, select Month Name,
open the Modeling Ribbon, and Sort by Column -> Month:

40
Analyzing PI System Data

Go back to the Report View, the Month headers should now display in chronological order.

Next, we will add conditional formatting to highlight transformers with high loading. Conditional
formatting is set from the Values field in the Visualizations Pane. Select the drop down by
Average of Loading and click Conditional Formatting -> Background Color Scales:

Page 41
Analyzing PI System Data

Reverse the minimum and maximum colors so that high numbers are Red. Enable the Diverging
option. Click OK.

Turn off the Subtotals using the formatting options:

42
Analyzing PI System Data

Add a Title to the Matrix using the Formatting Options. Call it Monthly Average Loading, change
the color to black, and bump up the font size:

Bump up the font size of the Column headers, Row headers, and Values:

Page 43
Analyzing PI System Data

The matrix should look something like this:

Watt-hours Delivered by Headquarters – Clustered Column Chart

Next we’ll configure a new visual to show Average watt-hours delivered on a per month basis.
Click some blank space and add a Clustered Column Chart:

44
Analyzing PI System Data

Use Month Name for the Axis, Headquarters for the Legend, and Wh Delivered as the Value:

Summarize the Wh Delivered as an Average:

Note that only 1 headquarters is shown. This is because the visual is being filtered by the
Matrix configuration.

Page 45
Analyzing PI System Data

In the Matrix, drill up to the top level by repeatedly clicking the Drill Up button:

Eventually we’ll see the summary broken down by headquarters and all 5 headquarters will be
shown in the Clustered Column Chart.

Note that Cartago seems to be the HQ with the highest transformer loads. Turn on Drill mode
and drill down and follow the highlighting (drill down wherever the load is highest) to discover
that transformer PT_XYZ0109 is consistently overloaded. Consider that these are averages, not
maximums.

Turn off drill mode and click on PT_XYZ0109 to see the month to month watt-hours delivered
for this particular transformer.

Optionally search for PT_XYZ0109 in PI System Explorer and note how many meters it is
responsible for.

46
Analyzing PI System Data

In Power BI, Click PT_XYZ0109 again to deselect it and remove the filtering.

Select the Clustered Column Chart visual, and adjust the formatting:

 Move the legend to the bottom, remove the legend title, and increase the text size

 Change the Chart Title to “Average Wh Delivered by Headquarters” with black text,
increase the text size
 Change the colors used for the headquarters. You will have to Drill Up in the matrix to
get all the headquarters to show up first.

Page 47
Analyzing PI System Data

Watt-hours Delivered by Weekday – Line Chart

We also want to see how the delivered watt-hours change throughout the day and on different
days of the week. We’ll display this information using a Line Chart.

First some setup. We do not have a day of the week column in the original data set like we did
for the Month, so we will add one using a DAX formula.

Go to the Data View. Select any column, then go right click -> New Column:

48
Analyzing PI System Data

Add the following DAX formula in the formula bar and click the check box or hit enter. Note
that this sets the column name to “Day Name”.

Day Name = FORMAT('Transformer Loading'[TimeStamp], "dddd")

Page 49
Analyzing PI System Data

In order for the days of the week to display in the correct order, we will add another
column for the numerical day of the week using the WEEKDAY() function:

Day = WEEKDAY('Transformer Loading'[TimeStamp])

We can then sort the Day Name column by this new Day column:

50
Analyzing PI System Data

Now it’s time to configure the visual. Click some blank space, and add a Line Chart:

Use Hour as the Axis, Day Name as the Legend, and Average of Wh Delivered as the values
(summarize Wh Delivered as an Average):

Adjust the formatting of the Line Chart:

 Move the Legend to the bottom

 Change the title to “Average Wh by Weekday”, text color black, and increase the text
size of the title to match the other visuals
 Optionally change the trend colors

Page 51
Analyzing PI System Data

Add a report title by inserting a text box:

The end result should look something like this:

Finally experiment with drilling up and down the hierarchy and filtering the report.

52
Analyzing PI System Data

Directed Activity – Transformer Loading Analysis

Objectives:
 Configure a Measure to calculate service hours
 Configure a Group to create bins for different load ranges which can then be used for
highlighting and filtering
 Configure a Stacked Bar Chart to display the service hours spent in each Load Range
by circuit
 Configure a Table to show the top 20 transformers by average Loading
 Configure a Slicer to filter by Month

In this exercise, we will analyze transformer loading characteristics. The goal is to assess the
number of service hours spent in various high load conditions to better understand which
transformers are at risk of failing and also assess whether a given transformer should be
replaced with one that has a higher capacity.

Approach:
Start a new Page by clicking the New Page icon at the bottom of the existing report:

Rename the Page to Transformer Loading.

Service Hours
The first thing to do is configure a Measure to calculate service hours. Each row in the data set
represents 1 hour, so we can simply count the number of rows that have been filtered through
user selection. This should make a bit more sense when it all comes together.

Right click any of the fields from the Fields list and select New measure:

Page 53
Analyzing PI System Data

54
Analyzing PI System Data

Enter the below formula into the configuration box:

The raw text is given below for convenience.

Service Hours = CALCULATE(COUNT('Transformer Loading'[Loading]))

A Note on Measures vs Calculated Columns

From a configuration perspective, Measures and Calculated Columns are configured similarly
so the distinction may not be obvious. Measures and calculated columns both use DAX
expressions. The difference is the context of evaluation. A measure is evaluated on the fly using
a subset of data, whereas a calculated column is pre-calculated at the row level within the table
it belongs to. A simple way to put it is that Measures take into account the filtering that has been
set by the end user of the report (the stuff they’ve clicked on), while calculated columns are
computed row by row and are not influenced by the report filtering.

Loading Groups
Different ranges for Loading will be grouped into bins representing different Load Ranges. It is
normal for transformers to be operating at Loads higher than 100% of their rating, but loads in
the range of 125% and higher are potentially cause for concern. In order to calculate service
hours in the different Load Ranges, a group must be configured in the data set for filtering and
counting by the Service Hours Measure.

Page 55
Analyzing PI System Data

Right click on Loading and select New group.

Change the name to Loading (25%) and set the bin size to 25, then click OK.

56
Analyzing PI System Data

Loading by Circuit – Stacked Bar Chart

Now we can begin to configure the report. Click some empty space and add a Stacked Bar
Chart:

With the Stacked Bar Chart selected, drag and drop Fields from the data set into the field
configuration boxes. Use Circuit for the Axis, Loading (25%) for the Legend, and Service
Hours for the Value:

Next we will apply some formatting and filters to make the data set more manageable. We’ll
change the color scheme and only show Loadings greater than 125%, since loads in the normal
range are not of interest to us.

Page 57
Analyzing PI System Data

In the Visualizations Pane, with Fields Selected, scroll down to the Visual Level Filters and filter
for Loading greater than 125%. Be sure to click Apply Filter:

Next go to the Visualization Options and sort by Service Hours:

58
Analyzing PI System Data

Next change the color scheme. With the Visualization selected, click the Format Icon in the
Visualization Pane and adjust the colors to better convey the severity of the loading levels.

The stacked bar chart should now look something like this:

Page 59
Analyzing PI System Data

Service Hours and Average Load by Transformer – Table

The next visual we will add is a basic table showing the Transformer Name, Phase, Service
Hours, and Average Load. We will then filter the table to show only the top 20 transformers by
average load. This will give us a quick indicator of which Transformers are consistently
overloaded.

Click some blank space on the canvas to deselect any visuals, otherwise you will
accidentally convert the Stacked Bar Chart to a Table.

Create a Table:

Drag and drop the Secondary Transformer, Phase, Service Hours, and Loading Fields into
the Values section:

60
Analyzing PI System Data

Change the Loading Value to summarize by Average:

Change the Visual Level Filters to Show the Top 20 Transformers by Average Loading. Be
sure to drag and drop Loading to the by value field, and be sure to click Apply filter.

Page 61
Analyzing PI System Data

Under By value, change the loading summary to Average, and don’t forget to click Apply Filter
again:

Filtering by Month – Slicer

We’ll now add a basic Slicer to filter by Month. Click some blank space and then add a Slicer:

Drag Month Name to the field list.

62
Analyzing PI System Data

Go into the formatting options and change the orientation to horizontal to change the look of the
Slicer:

The report should now look something like this:

Page 63
Analyzing PI System Data

Sort the table by Average Loading:

Click the bars on the Loading by Circuit chart and the Month slicer buttons and note how
the service hours and transformers for that load range update on the table.

We will save formatting until the end in case we need to save time, but feel free to adjust the
formatting and add a title.

Linking to PI Vision
We have a PI Vision display for Transformers that we can link to from this report. We will utilize
PI Vision URL Parameters to set the same Transformer in the PI Vision display that the user
clicks on in the Power BI report. The URL parameters reference guide can be found in the PI
Live Library.

From within the client virtual machine, Navigate to:

https://pisrv01.pischool.int/PIVision/#/Displays/3/TransformerTrends

Take the above URL and append the following string to it in a text editor, then paste the URL
into Chrome:

?Asset=\\PISRV01\Distribution Network\Secondary Transformers\PT_XYZ0046

Transformer PT_XYZ0046 should be the selected Asset in the TransformerTrends display.

64
Analyzing PI System Data

Note that the ?Asset parameter denotes the path to the Asset in the PI AF hierarchy.

Once that is working, configure a Calculated Column to concatenate the URL with the
Transformer asset path.

Go to the Data Tab:

Right click on the header of any column and select New column:

Page 65
Analyzing PI System Data

For the DAX formula, enter the following and hit enter or click the checkmark:

PI Vision = "https://pisrv01.pischool.int/PIVision/#/Displays/3/TransformerTrends" &

"?Asset=\\PISRV01\Distribution Network\Secondary Transformers\" & 'Transformer
Loading'[Secondary Transformer]

Next scroll all the way to the right and find the PI Vision column, then select it.

Go to the Modeling Ribbon, and change the Data Category to Web URL.

Now go back to the Report Tab and select the Table, then drag and drop the PI Vision field as
one of the table values

66
Analyzing PI System Data

The links are now displayed, and they work, but they are not pretty to look at. Luckily Power BI
has a feature that addresses this.

Go into the Formatting Options, scroll down to the Values section, and turn on the URL icon:

Now the links look much cleaner:

Test the links to confirm that the PI Vision display is launched and the correct transformer is set.

Page 67
Analyzing PI System Data

(Optional) Formatting
Take some time to apply formatting to make the report more visually appealing and easier to
read.

1. Add a Title text box for the report (Home Ribbon -> Insert Text Box)
2. Add titles for the Stacked Bar Chart and Table, change the font color to black and bump
up the font size
3. Adjust the sizes of the header text
4. Resize the columns
5. Move the Legend on the Stacked Bar Chart to the bottom
6. Remove the totals from the Table
7. Remove the header from the Slicer
The end result should look something like this:

Finally test the links and experiment with filtering the report.

68
Analyzing PI System Data

Directed Activity – Circuit Voltage Quality

Objectives:
 Configure a Hierarchy Slicer
 Configure Clustered Column Charts to represent voltage quality distributions
 Configure Multi-row Cards to show some basic statistics
Utilities are required by the Public Utilities Commission to deliver power to customers at a
certain voltage quality. We want to check to see if the various circuits in the network are within
this range.

Voltage Quality is essentially the ratio of the delivered voltage to the nominal voltage. Our
fictitious distribution company aspires to operate at an average quality slightly above 100%.
Recall that we are using Asset Analytics to aggregate the average voltage quality from the child
meters.

We will configure a report to determine whether this is true across all transformers by
plotting the voltage distributions on a per-phase basis. We will use a Hierarchy Slicer
for filtering and display some basic statistics using Multi-row Cards.

Approach:
Start a new sheet and name it Circuit Voltage Quality:

Downloading the Hierarchy Slicer

For this part, there is no need to visit the web site, sign up, or download the file. We have
downloaded the file for use in class so that students do not need to sign up!

The Hierarchy Slicer is a custom visual that can be used to filter reports and mimic the PI AF
hierarchy. This is similar to the PI TreeView from PI WebParts.

Most custom visuals can be found on Microsoft AppSource. We will briefly go through the
procedure of how one would normally obtain a custom visual.

Page 69
Analyzing PI System Data

Search for a custom visual on Google or within AppSource and you’ll arrive at a page like this:

https://appsource.microsoft.com/en-us/product/power-bi-visuals/WA104380820?tab=Overview

At which point you would click Get It Now, sign in using your work or school account, and
download the .pbiviz file.

Importing and Configuring the Hierarchy Slicer

Now it’s time to import the custom visual. Open Power BI, click the ellipses within the
Visualization Pane, and select Import from file:

Navigate to C:\Class\Part 1 - PI Integrator for BA\Power BI Custom Visuals and select the
HierarchySlicer file.

70
Analyzing PI System Data

We should now see the Hierarchy Slicer in the list of available visuals:

Mimic PI AF Hierarchy – Hierarchy Slicer

This exercise requires the Hierarchy Slicer custom visual be imported and assumes the
Hierarchy has been configured.

We will use a Hierarchy Slicer to leverage the existing PI AF hierarchy for filtering. Add a
Hierarchy Slicer:

Page 71
Analyzing PI System Data

Drag and drop the Hierarchy to the visual fields:

Experiment with the Hierarchy Slicer for a bit by drilling down through the levels. Note that
checking a box for a parent will also include the children. This is a great way to visualize how
filtering works in Power BI.

Optionally change the Title of the Hierarchy Slicer to Network in the formatting options and
increase the text size.

Voltage Quality Profiles – Clustered Column Chart

In this part we’ll use a trick to represent a statistical distribution using a combination of a Group
and a Clustered Column Chart.

First configure the Group for voltage qualities, using .1 as the bin size. These bins will form the
x-axis of the chart. Click the ellipses next to Voltage Quality and select New Group:

72
Analyzing PI System Data

Change the name to Voltage Quality (0.1) and the bin size to 0.1 the click OK:

Now add a Clustered Column Chart:

Use the Voltage Quality (0.1) Group as the Axis, Phase as the Legend, and Count of Voltage
Quality (summarize Voltage Quality by Count) as the Value.

The chart is of course not usable in its current form. We will need to apply filtering to show only
the X Phase and remove the outliers from the chart.

Page 73
Analyzing PI System Data

Under Filters in the Visualization Pane, filter to only include the X Phase:

Filter the Voltage Quality (0.1) bins to only include the range of 95 to 105. Be sure to click
Apply Filter:

74
Analyzing PI System Data

Now apply some formatting:

 Change the visual title to “X Phase Distribution”, make the text black, and increase the
text size
 Turn off the legend
 Turn off the Y-axis
 Turn on the X-axis title and increase the text size (expand X-Axis in the formatting
options)
 Change the Data colors to blue
The resulting chart should now look like this:

Now creating the distributions for the Y Phase and Z Phase is easy. Simply copy and paste the
X Phase Distribution Line Chart and change the Filters and Titles:

Now select them all and align left then distribute vertically:

Page 75
Analyzing PI System Data

Voltage Quality Statistics – Multi-row Card

We will now configure multi-row cards showing some voltage quality statistics for the different
phases.

Click some white space and add a Multi-row Card:

76
Analyzing PI System Data

Add Voltage Quality as the first field, summarize as Average and rename the field to
“Average”:

Add Voltage Quality again, this time summarize as Median, and rename the field to “Median”.

Repeat for the Standard Deviation and Variance:

Page 77
Analyzing PI System Data

Filter the card to only include X Phase data, similar to the distribution charts. Drag and drop
the Phase to the filters list and select X Phase:

78
Analyzing PI System Data

Apply some formatting changes to the Multi-row Card:

 Add the title “X Phase Voltage Statistics” to the card using the formatting options.
Change the text color to black and increase the text size.
 Turn on the border
 Turn off Show bar under Card
 Change the category labels to black text
Once you’re happy with the formatting, copy and paste the X Phase Voltage Statistics card and
change the filtering to create cards for the Y Phase and Z Phase, then align them with the
distribution charts. The resulting report should look something like this:

Finally experiment with filtering the report using the hierarchy.

Page 79
Analyzing PI System Data

Directed Activity – Substations on a Map

Objective:
 Configure a Map visual
We want to display the different substations from the Network Load Profile on a map to lend
geographical context to the report. This will help assess which substations may be impacted by
extreme weather patterns and the relative importance of various substations in terms of
delivered power and geographical region.

Approach:

Substations on a Map – Map

We will modify the Report from Exercise 5.2.1 to include a Map of the substations instead of the
transformer matrix.

You may want to start with a copy of the solution file C:\Class\Part 1 - PI Integrator for
BA\Solutions\Network Load Profile.pbix or copy the 5.2.1 report to a new sheet. This is just
to back up your work since we will be deleting the Matrix to free up screen real estate.

You may have noticed that the data set we’ve been working with doesn’t really include
geocoding information, which is the information required to place each substation on a map. We
could potentially use the Substation names and hope for the best, but there will be some
ambiguity because several cities in Latin America and even Europe share these names. To
keep things explicit we will import the Latitudes and Longitudes of all the substations.

80
Analyzing PI System Data

Use Get Data and import data from a Text/CSV file:

Select the C:\Class\ Part 1 - PI Integrator for BA\Substation_Locations.csv file:

Power BI will automatically detect the headers, leave everything as-is and click Load.

Page 81
Analyzing PI System Data

Next, we need to define a relationship between the Transformer Loading data set and the new
Substation_Locations data set. Head to the Relationships View:

In the relationships view you can configure relationships between different tables. In this case
we want to link the two tables using the Substation column from Transformer Loading to lookup
the corresponding Latitude and Longitude from Substation_Locations. If you’re familiar with SQL
queries, configuring relationships between tables in Power BI is similar to choosing an INNER
JOIN column. More information about relationships can be found in the Microsoft Power BI
documentation.

82
Analyzing PI System Data

Expand and reposition the tables so that all fields are visible, then drag and drop Substation
from the Transformer Loading table to the Area field on the Substation_Locations table, then
click the relationship line between them to highlight the related columns:

Page 83
Analyzing PI System Data

Now that we have the Latitudes and Longitudes, Delete the Monthly Average Loading Matrix
and resize the other visuals to make way for a Map, then create a Map:

Use Substation from the Transformer Loading table as the Legend, and Latitude and
Longitude from the Substation_Locations table:

84
Analyzing PI System Data

Configure Latitude and Longitude as Don’t summarize as per the warning on the Map visual:

Resize the Map to fill the left side of the report, and add Average of Wh Delivered as the Size
field:

Page 85
Analyzing PI System Data

Apply some formatting:

 Change the Title of the Map to “Average of Wh Delivered by Substation”, change the text
to black, and increase the text size
 Move the Legend to the bottom
Replace the Legend field in the Clustered Column Chart with Substation:

Apply some more formatting:

 Change the Title of “Average of Wh Delivered by Headquarters” to “Average of Wh

Delivered by Substation”
 Reposition and resize the visuals so that everything looks clean
 Change the text size of the Legends so that all Substations and Weekdays are visible
The end result should look something like this:

Notice that you can now use the Map to filter for different substations.

86
Analyzing PI System Data

6 Part 2 – Power BI Reports using PI OLEDB

Enterprise

An alternative to PI Integrator for BA, which is available with the PI System Access
license, is PI OLEDB Enterprise. PI OLEDB Enterprise has better future data support
and is generally more flexible than PI Integrator for BA. The main drawbacks are the
difficulty of writing SQL queries and reduced throughput. For example, PI OLEDB
Enterprise will have a hard time reliably importing the 3.8 million rows during a report
refresh as was necessary in Part 1.
In Part 2, we will explore the process of preparing the Asset Framework model to
add additional dimensions of information to our AF database. The next step is
extracting desired information (process data, metadata, and event frame data) from
the PI System through PI OLEDB Enterprise. This data will be incorporated into a BI
cube and used to develop interactive reports that allow us to “slice and dice” our data
and bring meaning to our multidimensional data cube.

Page 87
Analyzing PI System Data

6.1 Directed Activity – PI AF Hierarchy and Data Set

Objective:
 Better understand the data set used in the following chapters

Approach:
We will take a few minutes to review the Fleet Generation PI AF Database. We wish
to analyze a number of KPIs for several generating units. Open PI System Explorer
and navigate to the Fleet Generation database.

Browse the hierarchy, which is organized into Region, Station, and Unit.

88
Analyzing PI System Data

Most of the child elements are based on the generic Unit template.

Those in the CENTRAL region are based on the Gas Turbine template, which is
derived from the UNIT template and has additional attributes.

Page 89
Analyzing PI System Data

Gas Turbines have all the attributes from the Gas Turbine template, but also inherit
those from the UNIT Template:

In the following chapters we will augment the AF templates with additional attributes
and KPIs.

90
Analyzing PI System Data

7 PI Analysis Service

PI Asset Framework is a powerful tool to help model the infrastructure of a company,

region, or division. Through PI Asset Framework Formula Data References, you can
create simple, on-the-fly calculations. PI Asset Framework also comes packaged
with the PI Analysis Service, for more advanced analyses. The analytic capabilities
include three analyses types, Expressions, Rollups, and Event Frame Generation,
which allow for calculations to be applied at the template level as well as the ability to
persist the results back to the PI Data Archive.

7.1 Capabilities of the PI Analysis Service

The PI Analysis Service, runs as a service that monitors all analyses and attributes associated
with these analyses.

Expressions:

Expressions allow for multi-lined calculations that utilize mathematical operators and functions,
if-conditions, and PI time-based functions to perform advanced analyses. Expressions, created
for a given asset type (element template), are automatically applied to all elements of that type.

Rollups:

Rollups allow for the calculation of summary statistics (averages, maximums, minimums) of
values from a set of AF attributes. Current statistical values can be written directly to the PI Data
Archive.

Event Frame Generation:

PI Analysis Service allows for the automatic detection of events that occur. These events are
bookmarked and information for any event type can be retrieved for further analysis.

Scheduling:

Expressions and Rollups can be scheduled to run whenever a new event arrives into the PI
Data Archive or calculated on a periodic basis.

Backfilling:
Results from all three types of analyses can be backfilled into the PI System.

Page 91
Analyzing PI System Data

7.2 Expressions

With Expressions, you can implement calculations through a set of built-in functions that take
values of attributes in PI Asset Framework as inputs, and outputs results to other PI AF
attributes. Expressions can be scheduled to run periodically or scheduled to run whenever the
input parameters of the expressions receive a new value.

Multi-line calculation dependency allows for each expression to be written to different output
attributes as well as re-using calculated results in subsequent expressions.

Each set of expressions allows for periodic or event-triggered scheduling.

Example
Function Category
TagAvg, PctGood
Archive Value Statistics
Bod, Hour
Date and Time
And, If
Logical
Abs, Sqr
Math
>, <>, *
Operators
DigState, DigText
PI Data Archive Digital States
TagSpan, TagType
Point Attributes
TimeEq, NextEvent
Search and Retrieval
Rand, Total
Statistical
NoOutput, TagBad
Status
Len, Text
String
A set of built-in performance equation-like syntax allows for access to a range of
functions. The available options include mathematical and logical operators and
functions, date and time functions, PI-specific performance equation functions, and
string manipulation functions.

92
Analyzing PI System Data

It is recommended to configure analyses at the template level.

The following procedure can be used to configure an Expression analysis using a template:
1) In the AF Database Library, create a new analysis template of type Expression.
2) Define expressions for the calculations in the analysis template.
3) Define the scheduling for the analysis template.
4) Define output attribute templates to store results.
5) Create the PI tags used to store the results.
6) Evaluate and preview the data to validate calculations.
7) Backfill the calculation if required.
8) Confirm the backfilled data
9) Backfill the data for other elements sharing the same template.

Page 93
Analyzing PI System Data

Directed Activity – Calculate Utilization for Assets

Objective:
The Utilization is a percentage that represents the amount of electrical power that a
unit produced against its theoretical capacity. Configure, test, run, and validate
analyses to calculate the percent utilization of all generating units.

Approach:
 In PI System Explorer, navigate to the Library in the Fleet Generation
database.
 Under Element Templates, select the UNIT element template.
 Select the Analysis Templates tab to configure the multi-lined expression for
Utilization:
Utilization = Total Hourly Gross Generation / Hourly Capacity
 Specify and configure an attribute template to store the results.
 Schedule the calculation to run periodically every hour.
 Backfill unit GAO01 for the past seven days.

94
Analyzing PI System Data

Approach
From the Unit Template, found in the Library plug-in of the Fleet Generation database,
select the Analysis Templates tab.

Configure a new analysis. Name the analysis Utilization and set the analysis type to
Expression.

Configure the expressions for the hourly total of Gross Generation and Utilization.

HourlyTotal = TagTot('Gross Generation','-1h','') * 24

Utilization = HourlyTotal / 'Hourly Capacity' * 100

Note: The HourlyTotal must be multiplied by 24, as the Performance Equation function TagTot
assumes the units of the input attributes are per day. Conversion factors should not be used
elsewhere with PI Asset Framework, as UOM conversions occur automatically.

Define two new output attribute templates by clicking Map -> New Attribute Template.

Page 95
Analyzing PI System Data

Name them Total Hourly Gross Generation and Utilization, respectively.

The UOMs can be set to MWh and % in the Attribute Templates tab:

Create the PI Tags

After the new attribute template has been configured, switch over to the Element
Hierarchy. The attribute values for the new tags should be “Pt Created.” If not, right-
click on the root Elements object. Select Create or Update Data Reference to
automatically create the PI tags to store the calculated results.

96
Analyzing PI System Data

Switch back to the Unit Template Analysis Templates tab to schedule the Analysis
Template to run periodically at the top of each hour.

Set GAO01 as the Example Element and click on the Evaluate button to validate the
expressions.

Page 97
Analyzing PI System Data

Prior to backfilling data into the PI Data Archive, it is usually a good idea to preview
the results. Right-click Utilization and select Preview Results. Look at the results for
the past 7 days:

98
Analyzing PI System Data

Directed Activity – Backfill Utilization

Objective:
At this point, all the analyses for event frame generation have been set up for all the
units of Fleet Generation. In order to calculate past Utilization values and generate
history for analysis, the calculations must be backfilled.

Approach:
From PI System Explorer, select the Management plugin

Right now, the only Analyses that exist are those we just created, so one can simply select All
or Enabled to view the Utilization Analyses that we want to backfill.

Page 99
Analyzing PI System Data

Normally there would be several types of calculations, so we’d want to filter them by setting up a
search. Create a new search:

Name the search Utilization, then do Add Criteria -> Name and enter the name of the
Analyses and click OK.

100
Analyzing PI System Data

Click the checkbox to select all Utilization Analyses. Then Select the Backfill/Recalculate
operation and set the start time to “*-7d” and the end time to “*”, select “Permanently delete
existing data and recalculate”, then click Queue:

Page 101
Analyzing PI System Data

Exercise – Calculate Generating Efficiency

This solo or group activity is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the activity.

Objective:
Not all of the electricity produced by our generators will make it out to the grid. Some
will be consumed by the internal circuity in the generator itself. The net generation is
defined as the amount of gross generation, or the amount of electricity that a
generator produces, less the electricity required to operate the unit. Calculate the
generating efficiency, or the ratio between the net generation to the gross
generation, expressed as a percentage.

Which unit is performing with the greatest efficiency?

Approach:
 In the PI System Explorer, navigate to the Library in the Fleet Generation
database.
 Under Element Templates, select the UNIT element template.
 Select the Analysis Templates tab to configure the expression for generating
efficiency, named Generating Efficiency.
 Specify and configure an attribute named Generating Efficiency to store the
results with units of %.
 Schedule the calculation to run periodically every hour.
 Evaluate the calculation using example element GAO01 and preview the
results.
 Backfill all Efficiency analyses for the past seven days.

102
Analyzing PI System Data

7.3 Rollups
The second analysis capability of the PI Analysis Service Analytics is known as
rollups. Rollups allow for the calculation of summary statistics for a set of attribute
values.
The types of summary statistics that are allowed are:
 Sum
 Average
 Minimum
 Maximum
 Count
 Median

Examples of rollup calculations include:

 Total mass of all contents in a tank farm
 Total production from all generating units for a particular site
 Maximum temperature of boilers within a building
 Average engine temperature of mining trucks
 Average temperatures for each asset with varying temperature
sensors.
Selecting attributes to rollup
Attributes used in rollup calculations can come from 1) attributes from child elements
relative to the element of interest or 2) the element of interest. One can set search
criteria to specify the specific attributes to rollup. Depending on the source of the
attributes (child elements or current element), the search criteria includes a masking
pattern for the 1) Attribute Name, 2) Attribute Category, 3) Element Category, and 4)
Element Template.

Page 103
Analyzing PI System Data

What is an element Example?

During the configuration of a rollup template analysis, when the source of the
attributes to roll up are from the child elements, PI System Explorer is not aware of
which parent element to retrieve child elements from. As such, when configuring a
roll-up analysis template, you will need to specify an example element. Note that
when configuring a roll-up at the element level, one will not need to select an
example element as the child elements are from the specific, selected element.

Scheduling and backfilling

Similar to Expressions, the rollup analyses can be scheduled to run as new events
come into the PI Data Archive or scheduled to run periodically. The PI Analysis
Service also allow the results from Rollup calculations to be written back to the PI
Data Archive.

The general process to properly configure and backfill an analysis template is:

1) Create a new analysis of type Rollup.

2) Define the source of the attributes to rollup (child element or current element).
3) Select the type(s) of summary statistics to calculate.
4) Define output attributes to store results.
5) Define the scheduling for the analysis.
6) Create the PI tags used to store the results.
7) Evaluate and preview the data to validate calculations.
8) Backfill the calculation.

104
Analyzing PI System Data

Directed Activity – Calculate Average Utilization for Substations

Objective: Management would like to have visibility over the average percent
utilization of all generating units for each substation. Roll up the average utilization to
the substation level.

Approach:

 Open up the Station Element Template from the Fleet Generation Database
Library.
 Add a new analysis called Average Utilization with analysis type of Rollup.
 Select Central\Albertsville as the example element.
 Specify the criteria to select the attributes used for the rollup calculation.
 Select the summary statistic function for the average.
 Specify the output attributes (be sure to create the tags).
 Schedule the calculation to be event-triggered.
 Verify data.
 Backfill for the past 7 days.

Step-by-Step Approach

From PI System Explorer, go to the Library. Then select the Station Element
Template. From the Analysis Templates tab, create a new Analysis called Average
Utilization with Analysis Type Rollup.

Page 105
Analyzing PI System Data

Specify the rollup attributes from child elements and set the example element to be
Central\Albertsville.

Set the attribute name field to Utilization. This mask will automatically select all
Utilization attributes from the child elements of the Albertsville station. However in
the preview only the Utilization from the Sample Child Element will be shown:

Set the scheduling to be event-triggered. Each time the Utilization analysis finishes
calculating each hour, the rollup analysis will run.

106
Analyzing PI System Data

Select Average as the rollup function and create a new Output Attribute called
Average Utilization.

Set the default UOM of this new attribute to % in the Attribute Templates tab:

Page 107
Analyzing PI System Data

In the Analysis Templates tab, Click on the Evaluate button to verify the result of the rollup
function.

Check-in your changes.

From the element hierarchy, verify that the PI tag exists for the attribute.

From the Management pane, backfill your Average Utilization rollup analyses for the past 7
days and verify the data has been backfilled by trending the Average Utilization attributes.

108
Analyzing PI System Data

Exercise – Calculate Total Hourly Gross Generation for Each Station

This solo or group exercise is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the exercise.

Objective:
Management would like to gain more insight into the Total Hourly Gross Generation
at each station. Create a rollup analysis to totalize the Total Hourly Gross
Generation at the station level.

Which station produces the most power?

Approach:
 Open up the Station Element Template from the Fleet Generation Database
Library.
 Add a new analysis called Total Hourly Gross Generation with analysis type
of Rollup.
 Select Central\Albertsville as the example element.
 Specify the criteria to select the attributes used for the rollup calculation.
 Use the Sum function and output Attribute Total Hourly Gross Generation.
 Specify the output attributes (ensure tags are created).
 Set the UOM to MWh.
 Schedule the calculation to be event-triggered.
 Verify data using Evaluate and Preview Results.
 Backfill for the past 7 days and verify.

Page 109
Analyzing PI System Data

8 Event Frame Generation

Events are important process or business time periods that represent something
happening that affects your operations. In the PI System, events are known as event
frames. Thanks to PI Event Frames, you can analyze your PI data in the context of
these events rather than by continuous time periods. Instead of searching by time, PI
Event Frames enables users to easily search the PI System for the events they are
trying to analyze or report on.
With PI Event Frames, the PI System helps you capture, store, find, compare and
analyze the important events and their related data.
PI Event frames represent occurrences in your process that you want to know about,
for example:

 Downtime tracking  Environmental monitoring

excursions

 Process excursions  Product tracking batches

 Equipment startups and  Operator shifts

shut downs

The following table presents some of the features and advantages of PI Event
Frames:

 Reference multiple elements within the

same event.
 Support multiple overlapping events on a
Flexibility
PI AF element.
 Capture any event; a "batch" is just one
type of capturable event.

 Search by time range, type of event or

event frame attribute.
Powerful search  Most common search attributes can be
configures as indexed attributes to speed
up end-user searches

Scalability  PI Event Frames are extremely scalable.

A PI Event Frame is defined by three characteristics:

1. Name.
2. Start time and end time: defines the event’s time range.
3. Context: event attributes and related assets.

110
Analyzing PI System Data

8.1 What are Event Frames?

Creating Event Frames

The Fleet Generation database contains a series of Elements representing the

regions and units associated with each generation plant. In order to keep up with the
power demands, it is important that the plant is up and running. We need to keep
track of the uptime associated with the generation plant.

A ‘Unit Status’ attribute is associated with each generating plant in our hierarchy.
This attribute will be used to monitor the uptime associated with each plant.

Time Range Retrieval Methods

There are three time range retrieval methods, the use of which depends on what data is to be
captured, and how it is to be displayed.

Time Range

This method allows a time range to be supplied by the end user. When any single value query is
made, this period of time is used for calculations. If, however a period of time is supplied from
an application, such as a generated Event Frame or Vision display, then the user specified time
range is discarded and the application time period is used.

Time Range Override

The Time Range Override behaves in the same way as the Time Range method during all
single value queries, as uses the user specified time period. When a period of time is supplied
from an application, the application time range is discarded and the user specified period is
used.

Not Supported

Not Supported does not allow for a time range to be supplied by the end user. As such, an error
is returned by any request for a single value. If a period of time is supplied however, then this
range is adopted by the method for the calculation. The result is then the same from the Time
Range method.

Page 111
Analyzing PI System Data

There are different use cases for the methods, so care must be taken to ensure the correct
method is used.

METHOD SINGLE VALUE APPLICATION SUPPLIED

TIME RANGE User Specified range result Application Specified range
result
TIME RANGE User Specified range result User Specified range result
OVERRIDE
NOT SUPPORTED Error: This attribute requires a Application Specified range
Time Range to calculate a value result
in…
Single timestamp query results (sample element with 1h specifications)

Application supplied time range query results (sample 3h event frame)

112
Analyzing PI System Data

Directed Activity – Create a Temperature Anomaly Event Frame Template

Objective:
The gas turbines in the Fleet Generation database each have two temperature
sensors. Create an Event Frame template with appropriate attributes to help monitor
and analyze potential issues with gas turbines. The event frame should capture the
real-time data specific to gas turbines and the current status and duration of the gas
turbine.

Approach:
 Create an Event Frame template.

Create a template called “Gas Turbine Temperature Anomaly”. Set the Naming
Pattern to %..\ELEMENT% %ELEMENT% %TEMPLATE% %STARTTIME:yyyy-
MM-dd HH:mm:ss%

Select the Attribute Templates tab. Right click in the white space to create an
attribute.

Name the Attribute Unit Status. Select Enumeration Sets => Status as the
value type.

Page 113
Analyzing PI System Data

Select the PI Point Data Reference, then select Settings…

Click the radio button next to Attribute, and enter .\Elements[.]|%Attribute%.

The Event Frame references a PI AF Element. The [.] syntax points to this PI EF
Template’s primary referenced PI AF element within the Elements collection. Set
the By Time Range dropdown option to “Start Time.”

114
Analyzing PI System Data

Note: Substitution parameters cannot be used to make a reference to an attribute from the
Element Template that is not a PI Tag.

Page 115
Analyzing PI System Data

Upon completing the definition, click OK. The Settings will be completed as seen
below:

Create a second attribute to store the Duration of event frame. The Duration
attribute will be populated by the new EventFrame() function in a later
exercise. It’s just a placeholder for now.

Create a third attribute to store the Technology. For the Value Type, select String
and for the Data Reference, select String Builder.

Note: When the event frame attribute’s data reference is set to PI Point, the syntax
.\Elements[.]|Attribute only allows for the reference to PI Point Data Reference attributes.
Element attributes configured as formulas and table lookups cannot be passed to event frames
using a PIPoint Data Reference. Instead, for attributes configured as formulas or table lookups,
select String Builder as the data reference.

116
Analyzing PI System Data

Set the settings for the attribute as .\Elements[.]|%Attribute%:

Continue to create the following additional attributes. Make sure units are
properly set. The fastest way to accomplish this is to copy and paste these
attributes templates from the Gas Turbine element template.

Exhaust Gas Temperature - #1 Probe

Exhaust Gas Temperature - #2 Probe
Gas Fuel Flow
Gas Fuel Pressure
Gas Turbine Speed

Page 117
Analyzing PI System Data

Once these 5 attribute have been pasted into the Gas Turbine Temperature
Anomaly Event Frame Template, select them all and enter
.\Elements[.]|%Attribute%;TimeRangeMethod=StartTime as the configuration
string to set the data references and retrieval method in bulk:

Check in your changes

Note: %attribute% will substitute in the name of the event frame attribute template. This will then
point to the corresponding attribute in the referenced element. You can also select multiple
attributes when making modifications to the attribute configuration.

118
Analyzing PI System Data

Exercise - Create Inactivity Event Frame Template

This solo or group activity is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the activity.

Objective:
Generating units sometimes trip or go down. Management would like to understand
these downtimes, and determine how much demand was not serviced. Event frames
can help capture and bookmark these events for future analysis. Develop an Event
Frame template, called Inactivity using the same Naming Pattern as the previous
exercise, with fields required to track the desired plant information to create reports
for management. Specifically, management would like to know the following:

1. Unit Status – Real-time (copy/paste from previous exercise)

2. Duration – in seconds (copy/paste from previous exercise)
3. Technology – Metadata (copy/paste from previous exercise)
4. Hours Down – in hours (simple formula to convert seconds to hours)
5. Demand – Real-time (PI Point data reference)
6. Operator – Metadata (string builder)
7. Carbon Emissions in g/kWh – Metadata (string builder)
8. Total Demand in MWh – Real-time, Aggregation of Demand

Hints:
 For metadata, use String Builder as the Data Reference.
 For Total Demand, configure the attribute’s source units as MJ / s By Time
as “Time Range”, Relative time as “-1s” and By Time Range as “Total”
 Verify correct event frame template configuration through the creation of a
test event frame.

Page 119
Analyzing PI System Data

8.2 Event Frame Generation

The Event Frames Generation analysis allows for the automated detection and
generation of event frames in the PI AF database based on values from trigger
attributes. The type of events and the types of data captured inside each event are
defined with event frame templates in PI AF.

Some notable features of Event Frame Generation in the PI Analysis Service include the
following:

Generate events: Easily configure event generation and automatically generate your events
from the trigger tags that are already collecting data in the PI Data Archive.

Handle multiple event types: Generate all your different event types, such as downtime,
excursions, batches, and other events, on the same asset with no restrictions on overlapping
events.

Standardize using event frame templates and populate event attributes: Different event
types have different attributes and information that are important for analysis. Standardize your
events using event frame templates, and use the PI Analysis Service to automatically populate
event’s attributes with data from the PI Data Archive and PI Asset Framework.

Backfill events: PI Analysis Service enables you to define your history backfill time window,
then it backfills the events from previous time periods automatically.

Using PI AF element attributes as event triggers or event attribute values: Trigger

conditions for event frames can be linked to element attributes.

Configure using PI AF element templates: Apply the configuration of event frame detection
and generation to PI AF element templates. The same event detection automatically applies to
newly created assets of the same asset type. There is no need to configure the event frame
generation again.

Root Cause: Event frames are great for capturing events that have occurred.
However, often times, the time period prior to the event provides more information on
the cause of the event. PI Analysis Service allows for root cause analysis and will
capture a fixed time period (default five minutes) before the event start time for
further analysis. This will be recorded as a Child Event Frame.

Time True: The trigger condition for event frames could potentially be noisy. PI
Analysis Service allows for the specification of a minimum time true period before an
event frame will generate.

120
Analyzing PI System Data

Directed Activity – Gas Temperature Anomalies

Objective:
Each gas turbine has multiple temperature sensors. If any temperature reading
deviates more than 20% from the average, then servicing is required. Use the Gas
Turbine Temperature Anomaly Event Frame Temperature to help define these types
of events.

Approach:
From the Fleet Generation Library, select the Gas Turbine Element Template and
select the Analysis Templates tab. Create a new analysis template called Gas
Turbine Temperature Anomaly, Set the example element to GAO01, and set the
Event Frame Template to Gas Turbine Temperature Anomaly.

Page 121
Analyzing PI System Data

Add two new variables called AvgTemp and DeltaTemp.

Set the expressions to:

Avg('Exhaust Gas Temperature - #1 Probe','Exhaust Gas Temperature - #2 Probe')

'Exhaust Gas Temperature - #1 Probe' - 'Exhaust Gas Temperature - #2 Probe'

Define the StartTrigger as:

IF (AvgTemp-Abs(DeltaTemp/2))/AvgTemp > 0.2 THEN TRUE ELSE FALSE

122
Analyzing PI System Data

Add an Output Expression

Enter the expression

EventFrame("Duration")
Map the output to the Duration attribute

Page 123
Analyzing PI System Data

Set the scheduling to Event-Triggered and triggering to Any Input.

Evaluate and preview the results to confirm there are no syntax errors.

From the Analyses plug-in, backfill event frames for the past seven days for all Gas Turbine
Temperature Anomaly analysis templates.

124
Analyzing PI System Data

Exercise - Detect Inactive Units

This solo or group activity is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the activity.

Objective:
Engineering would like to perform a deeper analysis into events over the past week
in which the generating units are inactive. Configure the event frame generation to
automatically capture new events and detect historical events.

How many inactive events have been occurring?

Approach:
 Open up the UNIT Element Template from the Fleet Generation Database
Library.
 Add a new analysis called Inactive Units with analysis type of Event Frame
Generation.
 Specify the event frame template: Inactivity.
 Define the trigger condition to automatically detect inactive events.
 Add an Output Expression using the EventFrame(“Duration”) function.
 Verify data.
 Backfill for the past seven days.

Page 125
Analyzing PI System Data

8.3 Discussion

This is a discussion designed to maximize learning in a specific topic area.

Your instructor will have questions, and will prompt for communication within
the class. This is an open-ended section and the result depends on your
needs.

Objective: Brainstorm some real world uses for event frames at your own company.
Event frames can be used to capture duration and summary information for events
such as process excursions or downtime, but how would this be implemented in your
workplace?

Approach
 What kinds of events are of interest in your own process?
 Can you think of reliable trigger conditions?
 Do you have all the required data to identify these events?

Estimated Completion time 10 minutes.

126
Analyzing PI System Data

9 Analyzing Events

9.1 Objectives
PI Event Frames are stored in PI AF databases. These event frames can be viewed,
filtered, analyzed using PI tools such as PI System Explorer, PI Vision, and PI
DataLink.

9.2 PI Event Frames in PI System Explorer

The easiest way to view PI Event Frames is through PI System Explorer. From the
Event Frames Pane, you can perform searches against all the event frames within
an AF database. You can filter based on specific referenced elements, specific time
ranges, and much more.

From the properties of an Event Frame Search, you can specify the following search
parameters for the time of the event frame, and the properties of the event frame:

Search type: Specify how to perform an event frame search. Find all event frames
that are entirely between a start and end time? Starting or ending between a start
and end time?

Search start: Specify the start time for event frame search.

Search end: Specify the end time for event frame search.

Include descendants: Search for all child event frames in addition to parent event
frames.

Page 127
Analyzing PI System Data

Event Frame Name: Filter based on the name of an event frame. Can use
wildcards.

Element Name: Filter based on the name of the referenced element. Can use
wildcards.

Template: Filter based on the event frame type.

Additional Criteria: Ability to filter based on duration, attribute value, event frame
search root, and specify how many results to return.

The resulting search query is combined into a string within the search field. This
allows for direct manipulation of the data fields without using the menu options.

128
Analyzing PI System Data

The default search results bring back fields detailing the duration, start time, end
time, description, category, template, and a Gantt chart. Any of these fields can be
hidden by using the settings cog on the top right corner of the search results.
Additionally, values from the event frame attributes can be pulled back into the
search results through this same option list.

Page 129
Analyzing PI System Data

Directed Activity – Search for Inactive Events for GAO01

Objective:
Find all Inactive events for the unit GAO01 and GAO02 over the past 24 hours.
Examine the technologies that are involved in these inactive events.

Approach:
Click on the event frame plug-in. Right-click on Event Frame Search 1 and select Properties.

130
Analyzing PI System Data

From the Event Frame Search screen, specify the search start to “*-1d”, end to “*”, and uncheck
the “All Descendants” checkbox. For the Element Name textbox, specify GAO0? and set the
Template to Inactivity.

The search will return several inactive event frames. Select all of them and click on OK.

Click on the gear icon to the right of the fields, and remove the description and category
fields. Then click on “Select Attributes.”

Select the Technology attribute from the Select Attributes wizard.

Examine the Technology that is leading to the downtime for these Inactive Units.

Page 131
Analyzing PI System Data

Exercise – Search for Recent Temperature Anomalies

Objective:
Find all temperature anomaly events for the gas turbines over the past 48 hours that
last for more than one hour. Add columns for Fuel Gas Pressure and for each of the
two gas temperature sensors.

Which unit has the highest starting Gas Fuel Pressure during a temperature anomaly,
and when was it?

Approach:
Perform an event frame search and format results for the desired attributes.

132
Analyzing PI System Data

9.3 PI Event Frames in PI DataLink

PI DataLink allows you to retrieve current, historical, and calculated data back into
Microsoft Excel. In addition to these capabilities, PI DataLink also allows for the
retrieval of event frames back into Excel for further analysis.

There are two retrieval methods for Event Frames inside of PI DataLink:

Explore: Find Event Frames that meet the specified criteria and display them in a
hierarchical format, which is useful to analyze events sharing the same EF template.

Compare: Find Event Frames that meet the specified criteria and compare their
attributes in a flat format. This allows a flat list of events with attributes relating to
child events all within a single row.

For either the Compare or Explore Events, you can specify parameters to search for
specific event frames. You can specify the following:

Page 133
Analyzing PI System Data

Database: AF Database to search against.

Event Name: Search pattern to search for specifically named event frames.

Search Start: Search for all event frames that occurred after this time.

Search End: Search for all event frames that occurred before this time.

Event Template: Search for specific types of events.

Element Template: Search based off of the type of referenced element.

Element Name: Search pattern for the name of the event frame.

More search options: Search based on attribute values, duration, and category.

Number of child event levels: Only for “Explore Events” and allows for the
hierarchical display of events.

134
Analyzing PI System Data

Searching for event frames can be based off multiple attributes.

When searching with Explore Events, the results can be displayed hierarchically
based on the relationships between child and parent event frames.

To return more than 1000 event frames in the search preview, go to Settings in the
ribbon. Change the setting to 10,000 Event Frames.

Page 135
Analyzing PI System Data

136
Analyzing PI System Data

Directed Activity – How many temperature deviations occurred?

Objective:
Temperature deviations could potentially mean damaged machinery. Engineering is
interested in analyzing the Natural Gas units. Find out how many instances of
temperature deviations occurred for gas turbines that lasted for more than 30 minutes.

Approach:
From the PI DataLink tab in of Excel, select cell A1 and click Compare in the ribbon.

Specify the Database as \\PISRV1\Fleet Generation, Event name as “*”, Search start
as “*-1d”, and Event template as “Gas Turbine Temperature Anomaly.”

Page 137
Analyzing PI System Data

From More Search Options, set the minimum duration to 30 minutes.

Select the columns that you would like to display:

138
Analyzing PI System Data

Exercise – Analyzing Inactivity

Objective:
Inactivity events can be costly as the generating units are not generating any power.
Analyze with PI DataLink the total number of Inactivity events as well as the total
amount of time the units were in an Inactive state for the 24 hours.

Which generating unit had the most downtime events? Which generating unit had the
largest total downtime?

Approach:
Use PI DataLink to search for PI Event Frames and specify which attributes to return. Use Excel
to aggregate the events.

Page 139
Analyzing PI System Data

9.4 PI Event Frames in PI Vision

PI Vision enables you to view and analyze your PI data during the time range of a
particular event. For example, you may want to examine the performance of an asset
during an operator shift or compare the data for several assets during a downtime
period.

To view events, open the Events tab on the left side. Here you will find events related to your
process, the color to the left of each event indicates its severity. By default, the time range of the
display and the context of the symbols in the display determine what events are shown in the
Events list in PI Vision. To discover additional events, modify the time range or choose Edit
Search Criteria. When you edit the search criteria, there are a number of filtering options to find
the Event Frames you are looking for.

140
Analyzing PI System Data

You can select an event to find its Data Items (event attributes) and its start and end time.

By right clicking on an event, you can choose Apply Time Range apply the event’s
time range to the display.

Page 141
Analyzing PI System Data

Directed Activity – Inactivity Events in PI Vision

Objective:
Visualize Inactivity Events using PI Vision.

Approach:
Create a new PI Vision display. Drill down to asset GAO01 in the Fleet Generation database

Trend the Exhaust Gas Temperature Probes for the past 24 hours.

Click on Events in the top left and check “Automatically refresh the list”. By default, this will load
Event Frames for Assets on the display (in this case Turbine GAO01).

142
Analyzing PI System Data

Right-click on one of the Inactivity Events and select Apply Time Range. The time range will
be applied to the temperature trends.

Right-click on one of the Events and select Compare Similar Events by Type. Trends of the
Event Frame trigger attributes for the selected Event Frame and 10 recent event frames will be
shown.

Page 143
Analyzing PI System Data

Edit Search Criteria to compare 100 Inactivity Events for All Turbines:

144
Analyzing PI System Data

Other attributes from the Event Frames can be trended, but instead we will trend attributes that
are not included in the Event Frame but are included in the Asset. In the Attributes Pane, drill
into the turbine:

Then drag/drop the Fuel Gas Flow onto the trend area to add new trends

Page 145
Analyzing PI System Data

Use the scroll wheel on the right to scroll down and see the new trends

146
Analyzing PI System Data

9.5 Discussion

This is a discussion designed to maximize learning in a specific topic area.

Your instructor will have questions, and will prompt for communication within
the class. This is an open ended section and the result depends on your
needs.

Objective: Event frames can be difficult to grasp at first. Let’s repeat the discussion
from the previous chapter now that you’ve seen some examples. Brainstorm some
real world uses for event frames at your own company. Event frames can be used to
capture duration and summary information for events such as process excursions or
downtime, but how would this be implemented in your workplace?

Approach
 What kinds of events are of interest in your own process?
 Can you think of reliable trigger conditions?
 Do you have all the required data to identify these events?

Estimated Completion time 10 minutes.

Page 147
Analyzing PI System Data

10 PI OLEDB Enterprise SQL Queries

SQL stands for Structured Query Language. SQL is an American National Standards
Institute (ANSI) definition for the language used to communicate with relational
database systems. It is used by virtually all relational databases in the world today.
(Even the PI Data Archive has a SQL Subsystem that can act as a translator to make
it “look” like a relational database). SQL Commands are often called “SQL
Statements.” They can be executed interactively or as stored procedures.
The good part is that it is a standard and that every relational database you
encounter will understand it. There is no need to learn many languages. However,
there is a down side. Most databases have unique extensions and/or syntaxes that
are unique to those systems.
To give a simple example, when passing dates into Access you use pound signs (#)
for surrounding dates. On the other hand, in SQL Server you need to use
apostrophes (').
Access: [...] WHERE dtColumn >= #2001-11-05#
SQL Server: [...] WHERE dtColumn >= '20011105'
A SQL result set is a set of rows from a database, as well as meta-information about
the query such as the column names, the data types and sizes of each column.
Depending on the database system, the number of rows in the result set may or may
not be known. Usually, this number is not known up front because the result set is
built on-the-fly.
This flexibility allows for complex queries to be constructed and saved to return a
very specific subset of information from the AF Database that would be either too
cumbersome or impossible through the likes of PI System Explorer or PI Datalink.

Trivia: The result is stored in a result table, called the result-set. This table is held in memory.

This is often referred to in code as rs.

10.1 Dissecting the Syntax

A common SQL syntax starting command is SELECT which is used to query the database. The
data retrieved from the statement is based on the criteria specified in the SELECT statement.

Following the SELECT command identifies the columns to be selected from the tables(s).

SELECT * - retrieves all the columns from the table being referenced.

SELECT column1, column2, column3 – retrieves 3 columns of the table being

referenced.

148
Analyzing PI System Data

The FROM command identifies the first (or perhaps only) table being queried.

SELECT * FROM tablename – retrieves all the columns from tablename.

SELECT column1, column2, column3 FROM tablename – retrieves all data for the 3
columns of tablename.

The WHERE command contains criteria to filter the data being retrieved.

The conditional operators include:

equal (=)
greater than (>)
less than (<)
greater than or equal (>=)
less than or equal (<=)
not equal to (<>)
LIKE (which is a pattern matching operator)

Note: If the conditional clause is set to compare to text, the text value is encased in single
quotes (‘text’).

SELECT * from tablename WHERE column1 = 5

Retrieves only rows where column1 has a value equal to the number 5.

AND and OR statements

 AND indicates both statements must be TRUE for the row to be returned
when the query is executed.

o SELECT column1, column2, column3 from tablename WHERE

column1 = 5 and column2 = ‘junk’

o Retrieve only rows where column1 has a value equal to the number 5
and column2 value equals junk.

 OR returns data rows if either condition is met

o SELECT column1, column2, column3 from tablename WHERE

column1 = 5 or column2 = ‘junk’
o Retrieve only rows where column1 has a value equal to the number 5
or column2 value equals junk.

Page 149
Analyzing PI System Data

The LIKE operator is used to search for a specific pattern in a column. In conjunction with the
LIKE operator a wildcard of % is used for comparison. The % can represent a single character
or multiple characters. Another wildcard is the underscore (_) which can be used to represent a
single character.

SELECT * from tablename WHERE column2 LIKE ‘%unk’

Retrieves rows from tablename where column2 values end with the letters ‘unk’

SELECT * from tablename WHERE column2 LIKE ‘%un%’

Retrieves rows from tablename where column2 values contain the letters ‘un’

SELECT * from tablename where column2 like ‘_un_’

Retrieves rows from the tablename where column 2 values only contains 4
characters and the middle two characters are un.

SELECT * from tablename WHERE column2 LIKE ‘j%’

Retrieves rows from tablename where column2 values start with the letter ‘j’

To work with column/table names which have special characters, such as a space, use square
brackets:

If you wish to SELECT a column called Product Orders, enclose it in square brackets: [Product
Orders]

If you’re referring to a table whose full path is Fleet Generation, Region, Station, Unit, that must
be written as [Fleet Generation].[SouthEast].[Brick Canyon].[PLT02]

Any name may be wrapped in square brackets, so when in doubt as to what constitutes a
special character, wrap the name in square brackets.

10.2 PI OLEDB Provider or Enterprise? What’s the difference?

PI OLEDB Provider is an OLEDB data provider that provides access to the PI
System. Given the correct security, the PI OLEDB Provider allows read/write access
to the PI System Archive.
PI OLEDB Enterprise is an OLEDB data provider which provides access to the PI
System in a relational view, accessible through SQL queries. The PI OLEDB
Enterprise provider supports read-only access to asset and event data stored in the
PI Asset Framework (AF), such as AF Elements, AF Attributes and PI Event Frames.
Both data providers can be called by using the PI ODBC via a SQL DAS server to
allow remote query execution, minimizing the required software to be installed and
configured on endpoint machines.

150
Analyzing PI System Data

PI SQL Commander

The PI OLEDB Enterprise installation includes a test environment which handles the
OLE connection process and allows the user to execute queries and perform other
tasks. This test environment is PI SQL Commander.
PI SQL Commander is the user interface to assist with creating queries, transpose
functions, and views against PI AF using PI OLEDB Enterprise. This user interface
also provides access to the classic PI OLEDB Provider which builds queries against
the PI Data Archive components without knowledge of PI AF.
PI SQL Commander Lite is an application to navigate a relational view of the PI
System using SQL Queries that are exposed by PI ODBC. This can be used to
create, edit, test and save SQL queries of PI System data. It does not support the
more advanced functions found within PI SQL Commander however, like transpose
value function creation, or view creation. Previously created transpose value
functions or views may be used however.

Page 151
Analyzing PI System Data

Directed Activity – Review Predefined Queries

Objective:

Review predefined queries associated with the tables defined in PI SQL Commander.

Approach:

 Open PI SQL Commander

 Navigate to the Fleet Generation Database/Catalog
 Execute a Predefined Query associated with the Element Hierarchy table.

Launch PI SQL Commander -

Click Start > All Programs > PI System > PI SQL Commander.

In PI SQL Commander, verify that your PI SQL Object Explorer is visible. If it is not, click
View > PI SQL Object Explorer.

Through PI SQL Commander, either a PI AF Server or a PI Data Archive can be

accessed through SQL statements based on the item selected for connection.

From within the PI SQL Commander Object Explorer, connect to your AF Server, in
this example, PISRV1, by right-clicking Select Connect then select Windows
Integrated Security.

152
Analyzing PI System Data

An arrow next to the server icon indicates that the connection is successful:

After connecting to the PI AF server, you will see a catalog list in the PI SQL Object
Explorer. The catalogs listed correspond to each of the PI AF databases you have
configured for this PI AF server. We will be using the Fleet Generation database
throughout this course.

Right-click an object in the catalog that represents a table, view or function, and then
select Execute Predefined Query.

Page 153
Analyzing PI System Data

PI OLEDB Enterprise includes one sample SQL query for each table in the catalogs for
PI AF server.

This is the environment for building and testing PI OLEDB Enterprise queries.
Queries, written in the editor, can be executed with their results shown in the grid.

Upon selecting Execute Predefined Query, a query window will appear with a Select
statement for the ElementHierarchy will be developed.

SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy]
WHERE
Path = N'\' -- root elements

154
Analyzing PI System Data

The above query does not yield all elements, just the elements at the Region level.

(Note: the “N” declares the path string to be Unicode, which permits lots of different characters.
It will be omitted throughout this document since normally we are only dealing with standard
ASCII characters.)

Modify the query to retrieve all the elements and hit Execute.

SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy]
WHERE
Path like N'\%'
OR

SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy]
Returns the same results.

Page 155
Analyzing PI System Data

10.3 Table Aliases

Sometime table name or columns are lengthy or lack clarity. Using an ALIAS can simplify typing
and clarify table field names that are otherwise unclear. The “AS” command defines an ALIAS
for the item prior to the AS with the abbreviation following the command.

SELECT eh.* FROM [Fleet Generation].[Asset].[ElementHierarchy] as eh

In the above statement, eh can be used to identify the table instead of the full [Fleet
Generation].[Asset].[ElementHierarchy] table name. Aliases become more significant when
creating joins.

10.4 JOIN Statements

Rarely does data exist in one place or in one table. Sometimes the results of a query have to
come from a correlation of two or more distinct tables. To JOIN tables, a relationship is required
between the tables and must be identified in the SQL statement.

Within the joining operations, we want a result set than contains assets with useful information
from both tables, like performing a logical AND operation. There should be no gaps where a
match could not be found. This is called an INNER JOIN, and is the default joining operation
used by PI SQL Commander. Therefore, INNER JOIN and JOIN may be used interchangeably.
If we want to include these empty values, then an OUTER JOIN can be used, the equivalent to
the logical OR.

Two key words are used when creating joins between tables. The words JOIN and ON can be
used in the statement to identify the relationship between the tables being used. The key
word ON sets up the relationship of columns in the selected tables so the desired rows are
returned.

SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy] as EH JOIN [Fleet
Generation].[Asset].[ElementAttribute] as EA
ON eh.name = ea.name

Returns no records. Below is a result of the next query, but shows the names in the tables are
not the same.

156
Analyzing PI System Data

Even though both tables have columns called names, they do not identify identical fields.

Note: Columns named the same are not necessarily referring to the same item. For example,
‘id’ is a column that is frequently found in tables representing a unique identifier for the row, but
rarely do they refer to the same item from table to table.

However, the columns named ElementID in both tables are actually the same and return a
listing of all attributes for all elements defined.

SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy] AS EH join [Fleet
Generation].[Asset].[ElementAttribute] AS EA
on eh.elementid = ea.elementid

Page 157
Analyzing PI System Data

Field Aliases:

There’s an unsightly problem with the query: multiple columns are named the same (Name) but
are not the same. For anyone reading these query results, this is not helpful.

The solution is to rename the columns. Just as a table can be aliased, so can a column be
aliased. The keyword AS is used anytime an ALIAS is defined, whether the field is a table or
column name.

SELECT eh.Name AS [Element Name], ea.Name [Attribute Name]

The above statement gives meaningful names to the columns in the respective tables.

158
Analyzing PI System Data

Directed Activity – Element descriptions

Objective:

To extract the elements in the Fleet Generation database along with their descriptions.

Approach:
 Execute the Predefined Query associated with the Element Hierarchy –
modify to obtain all elements.
 Review the fields in the Element table.
 Determine potential relationships between the tables to create joins.
 Modify the Element Hierarchy query to add the appropriate join information to
extract the description of the elements.

Locate the Element Hierarchy within SQL.

 Execute the Predefined Query.
SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy]
WHERE
Path = N'\'
 Modify the query to obtain all elements
SELECT *
FROM [Fleet Generation].[Asset].[ElementHierarchy]

Review the ElementHierarchy table, which gives us the Path and Name (among
other things) of the elements in the hierarchy, but no description:

Review the Element table. Is there a link between the Element and Element
Hierarchy table?

Page 159
Analyzing PI System Data

Each element pointer within the hierarchy (i.e. each row in the ElementHierarchy
table) corresponds to an element object from the overall set of Elements (i.e. a
row in the Element table). Behind the scenes, these objects are linked by GUIDs
(Globally Unique Identifiers). The purpose of a GUID is to give individual objects
an identifier guaranteed to be unique. Meaningless to the human eye, they look
like:
9abd6084-6c74-4645-a7a0-833f6c25de3d

GUIDs (often contained in table columns ending in ID) are how each table relates
each row (element pointer) in ElementHierarchy to each row (element) in
Element:

 Modify the Element Hierarchy table to include the description from the
Element table.
SELECT eh.path, eh.name, e.description
FROM [Fleet Generation].[Asset].[ElementHierarchy] eh INNER Join [Fleet
Generation].[Asset].[Element] E on eh.elementid = e.id

160
Analyzing PI System Data

Note: In the above statement, the tables have ALIASes, but the word “AS” is not in the
statement as it is understood.

The result of the above query yields the name of the element and the description
associated with the element.

Page 161
Analyzing PI System Data

Exercise - Query for Specific Elements

This solo or group activity is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the activity.

Objective:

To extract the elements in the Fleet Generation database that are “Units” (Element Template)
and are located in the North Region. The fields that we want in our result set are the Unit Name,
Path, and Description.

Approach:
 Start with the query from the previous Directed Activity
SELECT eh.path, eh.name, e.description
FROM [Fleet Generation].[Asset].[ElementHierarchy] eh INNER Join [Fleet
Generation].[Asset].[Element] E on eh.elementid = e.id
 Review the fields in the Element Templates table.
 Determine potential relationships between the tables to create joins.
o Hint: The Element table has a field called ElementTemplateID
 Append a WHERE clause to filter based on the Path and Element Template.
 Determine the fields to return and the tables associated to each field.

162
Analyzing PI System Data

10.5 Built-in Functions

PI SQL Commander has some built-in functions specific to the PI System. If you
are familiar with SQL, you may already be familiar with functions. For example,
aggregation functions such as Max() or Avg() return the maximum or average of
a group of rows.

An entire list of built-in functions is available in the user guide for PI OLEDB
Enterprise.

One of the PI functions that will be used in subsequent exercises is

ParentName(). Instead of returning the complete PATH. The ParentName
function of PI OLEDB Enterprise is used to break up the AF element path name
into separate columns of the table. The strings in double quotes are used to
rename the column name in the table to something perhaps better suited for
reporting. Again, the “eh” ALIAS prefix is required to identify the source of the
field.

SELECT
eh.Name [Unit]
, ParentName(eh.Path,0) [Station]
, ParentName(eh.Path,1) [Region]
FROM [Fleet Generation].[Asset].[ElementHierarchy] eh
Where eh.Level=2

Page 163
Analyzing PI System Data

10.6 Data Tables

In the previous sections, we saw the process to query for elements from the Fleet Generation
database through a series of table joins between Asset Framework object tables within PI SQL
Commander. The tables within PI SQL Commander are not limited solely to Elements, Element
Hierarchy, and Element Templates.

Within PI SQL Commander, there are several tables under the [AF Database].[Data] path that
will allow the user to extract real-time and archive values from the PI Data Archive. A query
against these tables will return either Element Attribute data or Event Frame data. In order to
utilize these tables, a query needs to have an INNER JOIN to the ElementAttribute table and a
specific Data table. The ElementAttribute table allows for the mapping between the data and
specific attributes associated with a set of elements.

The tables corresponding to Element Attribute data are listed below:

Table Description

Archive Returns archive / compressed data

Snapshot Returns values in the snapshot (current values)

ft_InterpolateDiscrete Returns interpolated value given timestamp

ft_InterpolateRange Returns interpolated values at evenly distributed

timestamps
ft_Plot Returns minimum data required for trending

Note: Similar tables exist for data from Event Frame attributes. Typically, only the Event Frame
Snapshot table is queried against as each Event Frames contain individual start and end times.

164
Analyzing PI System Data

Directed Activity – Snapshot Values

Objective: Create a query to extract real-time values for the Gross Generation and
Net Generation attributes of all units. The fields in the result set should contain
Element Name, Station Name, Region Name, Attribute Name, Timestamp, and
Value.

Approach:
The creation of this query requires several steps.
1) Run the predefined query of the Element table to obtain all elements.

2) Remove the WHERE clause.

SELECT *
FROM [Fleet Generation].[Asset].[Element] e

3) Apply an INNER JOIN to the Element table.

SELECT *
FROM [Fleet Generation].[Asset].[Element] e
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
4) Apply an INNER JOIN to the ElementTemplate table.
SELECT *
FROM [Fleet Generation].[Asset].[Element] e
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[Asset].[ElementTemplate] et
ON e.ElementTemplateID = et.ID

5) Add a WHERE clause to return only elements that are Units.

SELECT *
FROM [Fleet Generation].[Asset].[Element] e
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID

Page 165
Analyzing PI System Data

INNER JOIN [Fleet Generation].[Asset].[ElementTemplate] et

ON e.ElementTemplateID = et.ID
WHERE et.Name = 'Unit'

6) Apply an INNER JOIN to the ElementAttribute table and modify the WHERE
clause to return only attributes that are either “Gross Generation” or “Net
Generation.”

SELECT *
FROM [Fleet Generation].[Asset].[Element] e
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[Asset].[ElementTemplate] et
ON e.ElementTemplateID = et.ID
INNER JOIN [Fleet Generation].[Asset].[ElementAttribute] ea
ON ea.ElementID = e.ID
WHERE et.Name = 'Unit' and (ea.Name = 'Gross Generation'
OR ea.Name = 'Net Generation')

7) Apply an INNER JOIN to the Data Snapshot table and specify the fields for
the result set.

SELECT
eh.Name [Unit]
, ParentName(eh.Path,0) [Station]
, ParentName(eh.Path,1) [Region]
, ea.Name [Attribute]
, s.Time
, s.Value
FROM [Fleet Generation].[Asset].[Element] e
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[Asset].[ElementTemplate] et
ON e.ElementTemplateID = et.ID
INNER JOIN [Fleet Generation].[Asset].[ElementAttribute] ea
ON ea.ElementID = e.ID
INNER JOIN [Fleet Generation].[Data].[Snapshot] s
ON s.ElementAttributeID = ea.ID
WHERE et.Name = 'Unit' and (ea.Name = 'Gross Generation'
OR ea.Name = 'Net Generation')

166
Analyzing PI System Data

Exercise – Interpolated data

This solo or group exercise is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the exercise.

Objective:
Create a query to extract hourly interpolated data for the Demand attribute of all
UNITs over the past four hours. The fields for the result set should include Element
Name, Attribute Name, Timestamp, and Value.

At what time does the demand tend to be high over all units?

Approach:
The creation of this query requires several steps.
1) Run the predefined query of the ft_InterpolateRange table.

2) Remove the portion of the WHERE clause that filters the results to all root
elements.

3) Modify the Start Time, End Time and TimeStep.

4) Apply an INNER JOIN to the Element table, which is required to join to the
ElementTemplate table

5) Apply an INNER JOIN to the ElementTemplate table, which is required to

filter for the UNIT template (et.Name)

6) Add a WHERE clause to return only elements that are Units.

7) Append to WHERE clause to return only the “Demand” attribute.

8) Restrict the SELECT to return the desired fields.

Page 167
Analyzing PI System Data

10.7 Data Transpose Functions & Function Tables

As seen above, the data comes back in tabular form, but does not lend itself to easy
interpretation.

Below is the query from the previous section that illustrates the difficulty in reviewing the
snapshot data for the attributes for the Elements. Notice that the attributes as returned in rows.

SELECT eh.name, ea.name, s.time, s.value

FROM [Fleet Generation].[Asset].[ElementHierarchy] eh
INNER JOIN [Fleet Generation].[Asset].[ElementAttribute] ea
ON ea.ElementID = eh.ElementID
INNER JOIN [Fleet Generation].[Data].[Snapshot] s
ON s.ElementAttributeID = ea.ID
OPTION (FORCE ORDER, EMBED ERRORS)
A portion of the results from the above query is displayed below.

Data presented in this format is typically difficult to handle for reports and BI Tools. Instead, we
prefer to rotate or transpose the values as:

As the same column headers are used for every row in the table, every element
needs to have the same set of attributes in order to populate these cells. This brings
us back to using AF Templates for all our elements, so we can get these common
properties. Transpose functions are only available for templatized elements!
This is similar to that of the final view created by the PI Integrator for Business
Analytics. PI OLEDB Enterprise can also generate transpositions similar to what is
above. A wizard walks you through the process of creating a transpose function for
any Element Template of your choosing.

168
Analyzing PI System Data

Transpose Function Wizard

For many use cases, attribute values need to be returned in a way so that each
column represents an attribute. This is contrary to a typical relational representation,
where each value of each attribute is normally represented in consecutive rows. To
represent multiple attributes in this "one column per attribute" format, one could join
data tables with itself multiple times, but the resulting query string would be rather
large and complex. To help with this, we provide a way to create custom Table-
Valued Functions (TVFs) and derived function tables, to get "transposed" result sets
of the related data tables.
Under each PI AF database branch, there are four folders, “Assets”, “Data”, “DataT”
and “EventFrame”.

The “Tables” folder under “Data” shows the tables and columns that provide access
to snapshot and historical data from the PI System.

Page 169
Analyzing PI System Data

Under both the “Assets” and “Data” folders there are two additional folders called
“Views” and “Functions”. These folders are initially empty and provide places for you
to organize the views and functions you create.

In general, creating and editing queries and views is a restricted activity. The changes are
contained in the PI AF SQL Database and access will be controlled by the database
administrator.

The “DataT” branch of the hierarchy is for working with transpose functions.
 Transpose functions allow you to obtain tables of PI AF information based on
AF element templates.
 This folder comes with the same subfolder structure as “Assets” and “Data”,
but they are initially empty until they are manually created.
 Transpose functions can be create using the wizard discussed in the next
section.

170
Analyzing PI System Data

Directed Activity – Create Dynamic Asset Transpose Functions

Objective: Create all Dynamic Asset Transpose Functions for the Fleet Generation
database to be used in analyzing plant generation data.

Approach:
There are four transpose functions available. Each transpose function
returns a dataset made up of columns for every attribute of an element
template, where each row returns values based on a different time basis.

Function Snapshot or Rows/element returned

Transpose Snapshot Snapshot 1 row per element attribute for

element(s) selected

Transpose Archive Archive 1 row per element archived attribute

value for element(s) selected over a
specific time range

Transpose Interpolated 1 row per element returns interpolated

Interpolate Discrete value based on timestamp

Transpose Interpolated 1 row per element returned for each

Interpolate Range interval for each element attribute
based on time range and interval

Page 171
Analyzing PI System Data

Access the Transpose Function Wizard by right-clicking on the “DataT” folder under the AF
Database catalog you wish to build a transpose function for, and select “New Transpose
Function -> Dynamic Transpose Function (Asset)…”

Note: Dynamic Transpose Functions will automatically reflect additions and deletions of
Attributes from the template, whereas Custom Transpose Functions allow you to explicitly select
a set of Attributes which will remain static.

We will create all of them, hold control and multi-select the top four then click Next:

Note: Versioned Transpose functions consider PI AF versioning. We will not be using these.

172
Analyzing PI System Data

Select the PI AF template(s) you want to create transpose function for. You can select as many
as you want, the wizard can build multiple transpose functions per pass. Here, we will select
“UNIT” and click Next.

If you want to use non-default Transpose Function names or the element template has sub-
attributes, modifications can be made here. In our case just leave the defaults and click Next.

Page 173
Analyzing PI System Data

Leave the defaults for the next window. You have the option to return values using the
VARIANT data type, which may be useful with some reporting tools. We want to create the
function table.

Review the summary and click Next.

Click Execute. Transpose function creation should be successful.

174
Analyzing PI System Data

From the PI SQL Commander hierarchy, the transpose tables and functions created by the wizard
should appear under the “DataT” folder of the “Fleet Generation” PI AF database.

Page 175
Analyzing PI System Data

As for how to use your newly-created Function, examine the snapshot function it
in Object Explorer:

Reality check: If we call the TransposeSnapshot_UNIT function, the same

columns exist in the function as in the original template, such as Effective
Generating Capacity, Generating Efficiency, etc.

A Predefined Query is associated with transpose functions. Execute the query.

176
Analyzing PI System Data

Exercise – Create Dynamic Event Frame Transpose Functions

This solo or group activity is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the activity.

Objective:
Attributes from the generation units and the event frames will be used to analyze
production data from the plants.

Approach:
Use the transpose function wizard to create a Snapshot event frame function
using the Inactivity and Gas Turbine Temperature Anomaly template.
Verify the results of the transpose function through the execution of the pre-defined
query.

Hint: The steps are almost identical to the ones used when creating an Asset
transpose function.

Page 177
Analyzing PI System Data

Create Transpose Snapshot only:

Create for both Gas Turbine Temperature Anomaly and Inactivity.

Leave all defaults from this point on. Next, Next, Next, Next, Execute.

178
Analyzing PI System Data

10.8 UNION Statements

You may have noticed that the Asset Transpose functions we created only return
Assets that use the UNIT template and not those that use the Gas Turbine template,
despite the Gas Turbine template being derived from the UNIT template via template
inheritance. Unfortunately, this is a limitation of PI OLEDB Enterprise. One way to
address this is with UNIONs.
In simple terms, UNIONs take the results of two queries and stack the result sets on
top of each other to form a single result set. One limitation of UNIONs is that the
input result sets must have identical columns, which may require removing and
aliasing columns to match the data sets. This will be demonstrated in the following
exercise.
The syntax is quite simple, place the keyword UNION in between two queries to
union them together. The OPTION statement must be at the very end:

SELECT * FROM Table1 WHERE Condition=’TRUE’

UNION
SELECT * FROM Table2 WHERE Condition=’TRUE’
OPTION (FORCE ORDER, IGNORE ERRORS, EMBED ERRORS)

Directed Activity – UNITs and Gas Turbines in a single result set

Objective: Create a query to display Snapshots for all UNITs, including Gas
Turbines.

Approach:
The creation of this query requires several steps.

Page 179
Analyzing PI System Data

Create Dynamic Asset Transpose Functions for Gas Turbines. This is identical to a
previous exercise except we will select the Gas Turbine Template. Create all non-
versioned Transpose Functions

Select Gas Turbine, then Next, Next, Next, Next, Execute (leave all defaults after
this point)

180
Analyzing PI System Data

Execute the predefined query for ft_TransposeSnapshot_UNIT, make note of the

resulting column set.

Create a new query. The Goal here is to preserve the text of the UNIT query.

Execute the predefined query for ft_TransposeSnapshot_Gas Turbine, note that

there are five additional columns for those attributes unique to Gas Turbines.

Change ts.* to explicitly select only those columns that are common to the UNIT
template for both queries, then UNION them together and remove one of the
OPTION lines (which has to be the last line). Since this is tedious, just copy/paste
the below query.
SELECT eh.Path + eh.Name Element, ts.[Generating Efficiency],
ts.[Utilization],
ts.[Total Hourly Gross Generation], ts.[Carbon Emissions], ts.[Generation
Rate],
ts.[Unit Status], ts.[Technology], ts.[Shift Hours], ts.[Shift], ts.[Operator],
ts.[Net Generation], ts.[Hourly Capacity], ts.[Gross Generation], ts.[Demand]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeSnapshot_Gas
Turbine] ts
ON eh.ElementID = ts.ElementID
WHERE et.Name = N'Gas Turbine'
UNION

Page 181
Analyzing PI System Data

SELECT eh.Path + eh.Name Element, ts.[Generating Efficiency],

182
Analyzing PI System Data

10.9 Saved Views

Often Administrators would prefer to create Views for end-users who are not familiar
with SQL queries. Often Views are queried using a basic SELECT * query to return
all data without any WHERE clause and without selecting individual columns. This
masks the complexity and size of the query (eg. table JOINS and UNIONs of several
tables) but places the burden of maintaining the query on the administrator. In future
exercises we will be using the queries directly, but it is still useful to know how to
create and query views in PI SQL Commander.

Creating dataset views

PI SQL Commander supports the creation of views. Views allow you to name a
stored query and it is this name that appears in the table list when importing data into
BI clients. Views are the easiest way to allow users to select which datasets they
want from PI AF when creating a report, as they do not need to understand the
complexity of the underlying SQL query.

Views are created using SQL syntax, but OLEDB Enterprise can give you a template
to start with. If you’re trying to create a saved query showing information about
assets, consider creating it in the Asset schema (folder). If you have a saved query
showing data values, for organizational purposes, place it in the Data schema. The
image below shows a right-click menu giving the Create View option:

Selecting Create View produces the beginning of a query:

Page 183
Analyzing PI System Data

At this point, it is a matter of naming the View by replacing <view name> and copy
pasting the query by into <query> placeholder.

184
Analyzing PI System Data

Directed Activity – View Creation for Unit Performance

Objective:
Create a view, Unit Performance, using previously created asset interpolated range
transpose functions for frequently changing process data (Transpose Interpolate
Range).
Approach:
Run the Unit Interpolated Range transpose function using the Execute
Predefined Query. Note that no INNER JOIN to the ElementTemplate table is
necessary as the transpose function was created against the UNIT template.

Modify the date range to include data from the past week and a one hour time
slice.

AND tir.StartTime = DATE('t-7d')

AND tir.EndTime = DATE('t')
AND tir.TimeStep = '1h'

Create a view from the transpose function.

Fill in the required information to create the view.

Name the View: Unit Performance

Use modified transpose function for the query

Page 185
Analyzing PI System Data

CREATE VIEW [Fleet Generation].[DataT].[Unit Performance]

AS
SELECT eh.Path + eh.Name Element, tir.*
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeInterpolateRange_UNIT] tir
ON eh.ElementID = tir.ElementID
WHERE et.Name = N'UNIT'
AND tir.StartTime = DATE(N't-7d')
AND tir.EndTime = DATE(N't')
AND tir.TimeStep = N'1h'
OPTION (FORCE ORDER, IGNORE ERRORS, EMBED ERRORS)

Execute the function.

If successful, a successful message will display, otherwise, an error will be

displayed in the lower region of the query section.

Refresh the View section and verify the Unit Performance View is present.

The definition for the view can be seen by selecting the Alter option in the View
folder.

186
Analyzing PI System Data

Execute the predefined query to see how the view is referenced ([Fleet
Generation].[DataT].[Unit Performance]) and confirm the expected data is
returned:

Page 187
Analyzing PI System Data

10.10 Importing PI OLEDB Enterprise data to Power BI

The first thing to do when using a client is to import the data you want to analyze.
Importing data requires connecting to the data source holding the data, specifying
the data you need from the data source (by selecting a database table, view, or
writing a query), and then importing the data into the client tool. The following steps
will describe how to import the complete datasets from the PI OLEDB Enterprise
views defined in the previous sections.

Directed Activity – Importing Data Using PI OLEDB Enterprise

Objective:
Import Transpose function query results into Power BI.
Approach:
Open MS Power BI Desktop
Select “Get Data” and browse to “other.” Select OLE DB and click “Connect”.

188
Analyzing PI System Data

Select Build, then select PI OLE DB Enterprise and click Next

Skip to the All tab, and edit the Integrated Security property to SSPI. This is
required to authenticate the connection using the running user’s credentials.

Page 189
Analyzing PI System Data

Go back to the Connection tab, enter PISRV01 as the Data Source, check ‘Use
Windows NT Integrated Security’, select Fleet Generation as the Initial
Catalog, and then test the connection.

Use Windows NT Integrated Security basically means connect using the user’s
login (pischool\student01) and password, and hence the user’s level of access.
SSPI must have been added in the previous step to use this option. The
alternative is to hard-code a user and password, which may give more access
than a user is entitled to.

Skip the Advanced tab and click OK to complete the connection string.

190
Analyzing PI System Data

Expand Advanced options, here you can enter any SQL statement that has
been tested in PI SQL Commander. In this case enter the Interpolate Range
UNION query and click OK:

SELECT eh.Path + eh.Name Element, ts.[Generating Efficiency],

ts.[Utilization],
ts.[Total Hourly Gross Generation], ts.[Carbon Emissions], ts.[Generation
Rate],
ts.[Unit Status], ts.[Technology], ts.[Shift Hours], ts.[Shift], ts.[Operator],
ts.[Net Generation], ts.[Hourly Capacity], ts.[Gross Generation], ts.[Demand]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeSnapshot_Gas
Turbine] ts
ON eh.ElementID = ts.ElementID
WHERE et.Name = N'Gas Turbine'
UNION
SELECT eh.Path + eh.Name Element, ts.[Generating Efficiency],
ts.[Utilization],
ts.[Total Hourly Gross Generation], ts.[Carbon Emissions], ts.[Generation
Rate],
ts.[Unit Status], ts.[Technology], ts.[Shift Hours], ts.[Shift], ts.[Operator],
ts.[Net Generation], ts.[Hourly Capacity], ts.[Gross Generation], ts.[Demand]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeSnapshot_UNIT] ts
ON eh.ElementID = ts.ElementID
WHERE et.Name = N'UNIT'
OPTION (FORCE ORDER, IGNORE ERRORS, EMBED ERRORS)

Page 191
Analyzing PI System Data

Click the Windows tab and select ‘Use my current credential’, then click
Connect. This should import the query results.

Inspect the preview, then click Load.

192
Analyzing PI System Data

Change the name of the table from Query1 to Unit Specifications

Page 193
Analyzing PI System Data

10.11 Discussion

This is a discussion designed to maximize learning in a specific topic area.

Your instructor will have questions, and will prompt for communication within
the class. This is an open ended section and the result depends on your
needs.

Objective:
Discuss differences between PI Integrator for BA and PI OLEDB Enterprise

Approach
 Which method do you prefer to create views? PI Integrator for Business
Analytics or for PI OLEDB Enterprise?
 Pros and Cons of both systems?
 What format would we like the data to be in for processing by BI clients?
 What should be added to the SQL queries to improve the format?
 Do these queries match what we want in our reports?
 If not, what is lacking?

Estimated Completion time 10 minutes.

194
Analyzing PI System Data

11 Building the “Fleet Generation” Report

We have now done enough preparation work that we can finally start building the
report.

11.1 Preparing and Importing the Tables

In our case, we are going to separate the time-series data from the static data and configure
table relationships to join the data sets together. Ideally, PI AF can be the glue that ties multiple
data sources together and lends context to the data. Technically, we could put all of our static
data into PI AF using table lookups and by assigning attribute values, then designing the
queries such that the result set is a single table. However, in real life not all of the data is always
in PI and several data sources must be joined together. This can of course be done at the query
level, but also in Power BI.

There are ways to avoid doing many of the following steps, but this will help prepare you for real
world reports where constant modifications and fine-tuning must be performed.

Directed Activity – Select Static Data Only from Snapshot Query

Objective:
Modify the existing query to only include static data.

Approach:
 Open the Power BI query editor to inspect a pre-existing query
 Modify the SELECT statement to only include static data columns
 Replace the query in the query editor with a new query

Page 195
Analyzing PI System Data

On the Home tab, Edit Queries

In the Query Editor window, click the Gear to change the input query

Select all the query text and Copy it:

196
Analyzing PI System Data

Paste the query into PI SQL Commander. Execute it as a sanity check.

Page 197
Analyzing PI System Data

Now the actual modifications. Head over to PI SQL Commander and edit the select statement to
only include static attributes: Carbon Emissions, Generation Rate, Hourly Capacity, Operator,
Shift Hours, and Technology.

198
Analyzing PI System Data

The resulting query is then

SELECT eh.Path + eh.Name Element, ts.[Carbon Emissions], ts.[Generation
Rate],
ts.[Technology], ts.[Shift Hours], ts.[Operator], ts.[Hourly Capacity]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e ON et.ID =
e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh ON e.ID =
eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeSnapshot_Gas Turbine] ts
ON eh.ElementID = ts.ElementID
WHERE et.Name = N'Gas Turbine'
UNION
SELECT eh.Path + eh.Name Element, ts.[Carbon Emissions], ts.[Generation
Rate],
ts.[Technology], ts.[Shift Hours], ts.[Operator], ts.[Hourly Capacity]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e ON et.ID =
e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh ON e.ID =
eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeSnapshot_UNIT] ts ON
eh.ElementID = ts.ElementID
WHERE et.Name = N'UNIT'
OPTION (FORCE ORDER, IGNORE ERRORS, EMBED ERRORS)

Paste the above query back into the Power BI query editor and click OK

Page 199
Analyzing PI System Data

Close the query editor

Click Yes to apply changes

It should reload the data (30 rows) successfully.

200
Analyzing PI System Data

Directed Activity – Split the Element Column

Objective:
There is still a problem with the data set. We will need to split the Element column
into the Region, Station, and Unit name. This could of course be done in the query,
but instead we’ll use the Power BI split columns feature.

Approach:
 Open the Power BI query editor and use Split Columns

On the Home tab, Edit Queries

With the Element column selected, select Split Column -> By Delimiter

Page 201
Analyzing PI System Data

Power BI automatically detected that \ is probably the desired delimiter. The defaults
should do exactly what we want here. Just in case though, ensure the following
settings are selected and click OK:

Power BI creates a blank column for the first split, Right-click -> remove it:

Rename Element.2 to Region, Element.3 to Station, and Element.4 to Unit:

202
Analyzing PI System Data

It should now look like the following. Also, note that a number of transformation steps
have been applied. Whenever the report is refreshed, these transformations will be
performed on the raw input:

Page 203
Analyzing PI System Data

Close the query editor, Click Yes to apply changes. Now the data is in a suitable
format.

Directed Activity – Import the Interpolate Range data

Objective:
Start with predefined queries for Units and Gas Turbines, restrict the result set to
exclude static data, and union the results to form a single table.

Approach:
 Execute predefined query for TransposeInterpolateRange
 Modify the SELECT statement to only include static data columns
 Replace the query in the query editor with a new query

Go to PI SQL Commander and execute the predefined query for

ft_TransposeInterpolateRange_Gas Turbine.

Modify the select statement to split the Element column into Region, Station, and
Unit using the ParentPath() function. Recall that we did this in Power BI in a
previous exercise. Execute the Query to ensure there are no syntax errors.
SELECT ParentName(eh.Path,1) as [Region], ParentName(eh.Path,0) as
[Station], eh.Name as [Unit], tir.*
Also replace the tir.* part to exclude ElementID, StartTime, EndTime, and Timestep
columns so that only the Attributes that change over time are included. We will also
exclude those attributes unique to Gas Turbines. Execute the Query to ensure there
are no syntax errors.

204
Analyzing PI System Data

SELECT ParentName(eh.Path,1) as [Region], ParentName(eh.Path,0) as

[Station], eh.Name as [Unit],
tir.Time, tir.[Generating Efficiency], tir.[Utilization],
tir.[Total Hourly Gross Generation], tir.[Shift],
tir.[Net Generation], tir.[Gross Generation], tir.[Demand]
Change the StartTime to T, EndTime to T-7d, and Timestep to 1h
AND tir.StartTime = DATE(N'T-7d')
AND tir.EndTime = DATE(N'T')
AND tir.TimeStep = N'1h'
Copy and paste the query before the OPTION statement, then replace Gas Turbine
with UNIT in the reference to ft_TransposeInterpolateRange_Gas Turbine and the
WHERE clause, and insert a UNION:
SELECT ParentName(eh.Path,1) as [Region], ParentName(eh.Path,0) as
[Station], eh.Name as [Unit],
tir.Time, tir.[Generating Efficiency], tir.[Utilization],
tir.[Total Hourly Gross Generation], tir.[Shift],
tir.[Net Generation], tir.[Gross Generation], tir.[Demand]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet Generation].[DataT].[ft_TransposeInterpolateRange_Gas
Turbine] tir
ON eh.ElementID = tir.ElementID
WHERE et.Name = N'Gas Turbine'
AND tir.StartTime = DATE(N'T-7d')
AND tir.EndTime = DATE(N'T')
AND tir.TimeStep = N'1h'
UNION
SELECT ParentName(eh.Path,1) as [Region], ParentName(eh.Path,0) as
[Station], eh.Name as [Unit],
tir.Time, tir.[Generating Efficiency], tir.[Utilization],
tir.[Total Hourly Gross Generation], tir.[Shift],
tir.[Net Generation], tir.[Gross Generation], tir.[Demand]
FROM [Fleet Generation].[Asset].[ElementTemplate] et
INNER JOIN [Fleet Generation].[Asset].[Element] e
ON et.ID = e.ElementTemplateID
INNER JOIN [Fleet Generation].[Asset].[ElementHierarchy] eh
ON e.ID = eh.ElementID
INNER JOIN [Fleet
Generation].[DataT].[ft_TransposeInterpolateRange_UNIT] tir
ON eh.ElementID = tir.ElementID
WHERE et.Name = N'Unit'

Page 205
Analyzing PI System Data

AND tir.StartTime = DATE(N'T-7d')

AND tir.EndTime = DATE(N'T')
AND tir.TimeStep = N'1h'
OPTION (FORCE ORDER, IGNORE ERRORS, EMBED ERRORS)

Execute the query to make sure it still works. Then head back to Power BI.

In Power BI, do Get Data -> OLE DB. Build the connection string, enter the query
where it says Advanced options, and click OK. Inspect the preview and Load the
data.

We’ve done this before so there isn’t a screenshot for every click this time.

It should import 5070 rows, 30 units x 24 hours x 7 days = 5040, plus 30 rows (1 per
unit) for the start time.

Rename Query1 to Unit Performance

206
Analyzing PI System Data

Directed Activity – Inspect Table Relationship

Objective:
Inspect the automatically created table relationship. Power BI should have detected
two identically named columns exhibiting a one-to-many relationship.

Approach:
 Open the Power BI relationships tab and inspect the existing relationship

In Power BI, Go to the Relationships tab, then move the Unit Performance table to
the right so that the relationship line is clearly visible and click on the line:

We can see that Power BI has already detected the relationship between the two
tables. This can be thought of as a graphical representation of an INNER JOIN
statement. These tables are now joined on the Unit column. For this to work, one of
the tables must only contain unique values in the Unit column (ie. the column can
serve as a key), as is the case here. This is referred to as a one-to-many relationship
in some documentation. Each Unit only appears once in the Unit Specifications
table, whereas each Unit appears many times in the Unit Performance table.
Relationships can be manually defined using a drag and drop interface, or through
Manage Relationships.

Page 207
Analyzing PI System Data

However at this point there is no need.

11.2 Augmenting the Data using DAX

Next we will add a few calculations to the Unit Performance table that will help
assess the total Emissions produced and the total cost of generation. We will also
add columns for the day of the week and sort the Weekday in Sunday -> Saturday
order.

Directed Activity – Calculate the amount of CO2 produced every hour

Objective:
Add a DAX formula Calculate the amount of CO2 produced every hour

Approach:
 Add and additional column to the Unit Performance table with the amount of
carbon emissions produced.

In Power BI, navigate to the Data Tab and select the Unit Performance table.
Right-click any column and add a new column. Enter the following formula:
CO2 = 'Unit Performance'[Total Hourly Gross Generation]*RELATED('Unit
Specifications'[Carbon Emissions]

Note that Total Hourly Gross Generation has units of MWh, and Carbon Emissions has units of
g/kWh. Grams/kWh is the same as Kilograms/MWh, and therefore the result will be in KG.

208
Analyzing PI System Data

Exercise – Calculate the Generation Cost

This solo or group exercise is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the exercise.

Objective: Add the cost calculation column to your Unit Performance table

Do you prefer having the calculations within AF as a formula data reference, or within
the BI client tools? What are some advantages and disadvantages of each?

Approach:
 Add and additional column named Cost to the Unit Performance table with
the dollar cost per hour.
 Take note of the input units. Cost should be in dollars.

Exercise – Add Column for Day of the Week and sort

This solo or group exercise is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the exercise.

Objective: Add the day of the week to your Unit Performance table, also add a
column with the numerical day of the week and sort by this value

Approach:
 Add and additional column named Weekday which shows the day of the
week as a string using the FORMAT() function
 Add another column named Numday which gives the numerical day of the
week using the WEEKDAY() function
 Sort Weekday by Numday

Page 209
Analyzing PI System Data

11.3 Configuring the Visualizations

Now we will add visuals to the report to convey useful information about the
generating units.

Exercise – Build the Report

This solo or group exercise is designed to maximize learning in a specific

topic area. Your instructor will have instructions, and will coach you if you
need assistance during the exercise.

Objective: Build an interactive Report comparing KPIs for different generation

technologies and operators.

210
Analyzing PI System Data

Approach:
 Add a Table showing Average Generating Efficiency and Average Utilization
by Unit

 Add a Pie Chart showing how the C02 emissions from each generation
technology contribute to the whole. Add a Tooltip that shows the Cost when
the user hovers over the Pie Chart

Page 211
Analyzing PI System Data

 Add a Clustered Column Chart showing the Sum of Total Hourly Gross
Generation with Technology as the Legend and Weekday as the Axis

 Add a Clustered Bar Chart showing the Average Hourly Cost with Operator
as the Legend and Technology as the Axis.

 Add Slicers for the Operator and Technology

 Optionally improve the look and feel of the report through the use of
formatting. Bump up the font sizes, adjust column names and titles, etc.

212
Analyzing PI System Data

12 Final Exercise: Create a Report

Objective:
Determine the carbon footprint of each unit and display on a US map. Also create a
report to analyze downtime (Inactivity) events.
Approach:
 Create a new Sheet in the Fleet Generation Report (the imported tables will
be re-used)
 Geospatial information for all units in Fleet Generation is located in
C:\Class\Final Exercise\Unit Coordinates.xlsx. This data will need to be
imported into the data cube.

.
 To get the Inactivity Events, you can either use PI OLEDB Enterprise or PI
Integrator for BA.
o You need a column to form the relationship between the Unit
Specifications table and the Inactivity Event Frames, it’s probably
easiest to join on Unit Name (GAO01, etc).
o Extract Event frames for the last 7 days
o If using PI Integrator for BA to publish the event frames, it’s
probably easiest to add the Unit Name to the Event Frame template.
o If using PI OLEDB Enterprise, start with the
ft_TransposeEventFrameSnapshot_Inactivity predefined query and
modify it as necessary.

 Import the Inactivity events for the last 7 days using whichever method you
prefer.

 Create the table relationships (should happen automatically if all columns are
named Units).
o Between the Unit Specifications table and the longitude/latitude table
o Between the Unit Specifications table and the Inactivity query results

Page 213
Analyzing PI System Data

 Insert a map within the client to display the location of each of the units and the
associated total hourly carbon emissions.

 Insert a table showing the number of downtime events (Inactivity Event Frames) and
average duration of event frames for each unit. Add the Average Utilization to the same
table.

 Configure the report in such a way that the Table relationships are tested. Use data from
multiple tables in the same Visual.

 Customize the display to make it more user friendly for later use and report generation.
Improve the formatting and add slicers.

Hints:

 If using PI Integrator for BA to publish the Inactivity Event frames, the Data Context must
be set to Second or else it will round to the nearest whole hour (which will always be
zero).

214
Analyzing PI System Data

A sample of what the report could look like:

The above report can be found in C:\Class\Final Exercise\Solution

Page 215
Analyzing PI System Data

13 Appendix A Substitution Parameters

Defining the Substitution Parameters

The substitution parameters are listed in the following table. The ones in bold are the commonly
used “Name” substitution parameters for Elements, Attributes, or Event Frames.

Parameter Will be replaced by this object’s name:

%..\Element% The name of the owning element of the element in
which the attribute resides. To retrieve further
ancestors, use the '..\' notations, such as
%..\..\Element%.
%..|Attribute% The name of the owning attribute in which the attribute
resides. To retrieve further ancestors, use the '..|'
notations, such as %..|..|Attribute%.
%@Attribute% The value of the attribute referenced. To retrieve further
ancestors, use the '..|' notations, such as
%@..|..|Attribute%.
%\Element% The name of the root AF Element in which the attribute
resides.
%<Environment Variable>% The matching System Environment Variable’s value. For
example %COMPUTERNAME% is replaced with the
name of the computer on which the Data Reference is
executing.
%Analysis% The name of the analysis if it can be obtained from the
context.
%Attribute% The name of the attribute that holds this data reference.

%AttributeId% The attribute ID that holds this data reference.

%Database% The name of the AF Database in which the attribute
resides.
%Description% The description of the attribute that holds this data
reference.
%Element% The name of the AF Element in which the attribute
resides.
%ElementDescription% The description of the element in which the attribute
resides.
%ElementId% The element ID that holds this data reference.
%EndTime% The local end time if it can be obtained from the time
context.
%Model% The name of the model if it can be obtained from the
context.
%Server% The name of the default PI Data Archive of the AF
Database in which the attribute resides.
%StartTime% The local start time if it can be obtained from the time
context.
%System% The name of the PI System in which the attribute
resides.

216
Analyzing PI System Data

%Time% The local time if it can be obtained from the time

context.
%UtcEndTime% The coordinated universal (UTC) end time if it can be
obtained from the time context.
%UtcStartTime% The coordinated universal (UTC) start time if it can be
obtained from the time context.
%UtcTime% The coordinated universal (UTC) time if it can be
obtained from the time context.
.\ The current reference
[.] The default object of the parent collection. For example
.\Elements[.] |Temperature returns the temperature
attribute from the primary element of the current
reference’s Elements collection.
[@filter=text] The search string in text (e.g. Tank*) matches the given
filter. Supported filters are: @Name, @Index,
@Template, @Category, @ReferenceType,
@Description, @Type, @UOM.
[@Index=#] Returns the result at location # from the collection result.

Page 217
Analyzing PI System Data

14 Appendix B Performance Equation Operands and

Functions

Taken from the PI Data Archive Application User Guide

Operands in Performance Equations

Operand Type Syntax Requirements Examples

Numbers (none) 1342 98.6 .0015 1.2e2

Tagnames In single quotes 'sinusoid' 'ba:level.1'
'ba.phase.1'
PI Time Expressions In single quotes '01-dec-03' '16-jul-94' '*'
Strings In double quotes "string string string"
"sinusoid"
Functions Must be a Performance TagVal('sinusoid')
Equation function TagAvg('sinusoid')
Cos('sinusoid')

Functions Listed By Type

The following tables list all functions by type. This list can also be found in the
PIPC\HELP\PEReference.chm help file.
Math Functions
Name Description

Abs Absolute value

Asin Arc sine
Acos Arc cosine
Atn Arc tangent
Atn2 Arc tangent (two arguments)
Cos Cosine
Cosh Hyperbolic cosine
Exp Exponential
Float Conversion of string to number
Frac Fractional part of number
Int Integer part of number
Log Natural logarithm
Log10 Common logarithm
Poly Evaluate polynomial
Round Round to nearest unit
Sgn Numerical sign
Sin Sine
Sinh Hyperbolic sine
Sqr Square root
Tanh Hyperbolic tangent

218
Analyzing PI System Data

Tan Tangent
Trunc Truncate to next smaller unit

Aggregate Functions
Name Description

Avg Average
Max Maximum
Median Median selector
Min Minimum
PStDev Population standard deviation
SStDev Sample standard deviation
Total Sum

Miscellaneous Functions
Name Description

BadVal See if a value is bad (not a number or

time)
Curve Get value of a curve
DigState Get digital state from a string
IsDST Test whether a time is in local daylight
savings time period
IsSet Test if a PI value is annotated,
substituted, or questionable
StateNo The code number of a digital state
TagBad See if a point has an abnormal state

PI Archive Retrieval
Name Description

NextEvent Time of a point's next Archive event

NextVal Point's next value after a time
PrevEvent Time of a point's previous Archive event
PrevVal Point's previous value before a time
TagVal Point's value at a time

PI Archive Search
Name Description

FindEq Timestamp when point = value

FindGE Timestamp when point >= value

Page 219
Analyzing PI System Data

FindGT Timestamp when point > value

FindLE Timestamp when point <= value
FindLT Timestamp when point < value
FindNE Timestamp when point ~= value
TimeEq Total period when point = value
TimeGE Total period when point >= value
TimeGT Total period when point > value
TimeLE Total period when point <= value
TimeLT Total period when point < value
TimeNE Total period when point ~= value

PI Archive Statistics
Name Description

EventCount Number of Archive events

PctGood Percent of good time in a period
Range Range of minimum to maximum value
StDev Time-weighted standard deviation
TagAvg Time-weighted average
TagMean Event-weighted average
TagMax Maximum value in a period
TagMin Minimum value in a period
TagTot Time integral over a period

Point Attributes
Name Description

TagDesc Get a point's descriptor

TagEU Get a point's engineering unit string
TagExDesc Get a point's extended descriptor
TagName Get a point's name
TagNum Get a point's ID
TagSource Get a point's point source string
TagSpan Get a point's span
TagType Get a point's type character
TagTypVal Get a point's typical value
TagZero Get a point’s zero value

Time Functions
Name Description

Bod Timestamp for beginning of the day for

given time
Bom Timestamp for beginning of the month for
given time

220
Analyzing PI System Data

Bonm Timestamp for first of the next month for

given time
Day Day of the month from a time
DaySec Seconds since midnight from time
Hour Hour from a time
Minute Minute from a times
Month Month from a time
Noon Timestamp for local noon of day of a
times
ParseTime Convert character string to time
Second Second from a times
Weekday Day of the week from a times
Year Year from a time
Yearday Day of the year from a time

String Functions
Name Description

Ascii ASCII character code for a character

Char String for ASCII character code(s)
Compare Wild comparison of two strings

DigText Text for a digital state

Format Formatting of a numerical number
InStr Instance of a sub-string
LCase Conversion of all characters to lower case
Len Length of a string
Left First characters in a string
LTrim Removal of blanks on the left side of a
string
Mid Extraction of a sub-string from a string
Right Last characters in a string
RTrim Removal of blanks on the right side of a
string
Trim Removal of blanks on both sides of a
string
UCase Conversion of all characters to upper
case

String Conversion
Name Description