0% found this document useful (0 votes)
2 views

2020_OSISoft_Slide_Best Practices when Implementing Asset Analytics

The document outlines best practices for implementing asset analytics, emphasizing the importance of initial design principles, planning, and understanding data density and flow. It discusses tradeoffs between streaming and ad-hoc analytics, the necessity of using templates for consistency, and the need for effective change management. Additionally, it highlights common bottlenecks to avoid and the significance of governance in maintaining a robust analytics environment.

Uploaded by

Leo Zhan
Copyright
© © All Rights Reserved
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)
2 views

2020_OSISoft_Slide_Best Practices when Implementing Asset Analytics

The document outlines best practices for implementing asset analytics, emphasizing the importance of initial design principles, planning, and understanding data density and flow. It discusses tradeoffs between streaming and ad-hoc analytics, the necessity of using templates for consistency, and the need for effective change management. Additionally, it highlights common bottlenecks to avoid and the significance of governance in maintaining a robust analytics environment.

Uploaded by

Leo Zhan
Copyright
© © All Rights Reserved
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/ 22

Best Practices when Implementing

Asset Analytics
Stephen Kwan

#PIWorld ©2020 OSIsoft, LLC 1


Agenda

• Initial design principles


• Best practices
• Planning
• Tradeoffs – what to do, what not to do
• Focus areas
• Avoiding problems
• Summary

#PIWorld ©2020 OSIsoft, LLC 2


Initial Design Principles

• Leverage AF for context and event capture


• Easy to use
• Performance Equation function signature
• High performance
• 1 Analysis Service per 1 AF Server
• Auto backfills for restarts

#PIWorld ©2020 OSIsoft, LLC 3


Basic Concept

Inputs Outputs

#PIWorld ©2020 OSIsoft, LLC 4


Workflow

Configure Test Schedule

Backfill/Recalculate

#PIWorld ©2020 OSIsoft, LLC


PI Analysis Service
PI Analysis Service
1 Load analysis
configuration
Calculation Engine
AF
Server 2
Data
Cache Analysis
A

B

Data
Archive 4 Read from 3
Read from
data source data cache

#PIWorld ©2020 OSIsoft, LLC 6


Design Tradeoffs

Optimized for Not suitable for


• Streaming analytics use case • Executing queries across really
• Prioritize real-time calculations large number of attributes
over backfills/recalculations • Extracting large amounts of
• Leverage Data Cache time series data
• Easy configuration • Ad-hoc calculations

#PIWorld ©2020 OSIsoft, LLC 7


Use cases: Streaming vs. Adhoc

Adhoc investigation and Online, scheduled,


analytics development streaming analytics

#PIWorld ©2020 OSIsoft, LLC 8


Know your
Data Density Understand
Process Sensible
and Data Flow –
Governance Algorithms
Bottlenecks
Data Rate

Good Planning

#PIWorld ©2020 OSIsoft, LLC 9


Change Management

• Consider setting up Dev or Test environment


• Allows for experimentation while developing
calculations
• Many engineers could be writing calculations
• Frequent changes can cause churn for the production
system
• Easier to isolate issues

#PIWorld ©2020 OSIsoft, LLC 10


Best Practices: Use Templates
• Provide manageability, consistency and governance
• Use templates for any repetitive work or for future
extensions.
• A modification to the template is applied to all
analyses from that template.
• Searching and filtering in UI is also easier with
templates.
• Higher performance.
#PIWorld ©2020 OSIsoft, LLC 11
Best Practices: Configuration
• Retain tribal knowledge
• Expressions should be written for readability
• Add comments as appropriate
• Use PI Points for outputs – Higher efficiency
and stores history
• Exit when you’re done – don’t continue to run
calculations
• Use Categories for filtering
#PIWorld ©2020 OSIsoft, LLC 12
Understand the Impact of your Analyses
• Some functions such as summary functions
(TagAvg etc.) might require a lot of data to perform
a calculation on the client side
• Use a smaller range or evaluating less often if the range
is larger
• Hit the cache and not I/O to Data Archive
• Do not trigger faster than evaluation rate
• Watch out for dependent analyses
• Plan for out of order data – recalculation
#PIWorld ©2020 OSIsoft, LLC 13
Avoid Common Bottlenecks

• Inputs from slow data


references, e.g. Table Lookup
• Inputs requiring round trip to
Data Archive (I/O bound)
• Too many analyses triggering at
the same time
• Sudden slug of triggering inputs
#PIWorld ©2020 OSIsoft, LLC 14
High Availability

• PI Analysis Service supports failover using


Windows Server Failover Clustering (WSFC)
• Use PI Buffer Subsystem for writing PI Point
outputs

#PIWorld ©2020 OSIsoft, LLC 15


Lastly

• Nothing scales forever


• Use multiple systems if needed
• Optimizing single system has limitations
• Hard to recover from a perturbation

#PIWorld ©2020 OSIsoft, LLC 16


Summary
• Governance – Test + production system
• Sensible data density and data rate
• Streaming vs. adhoc calculations
• Only calculate what’s needed
• Avoid round trips to Data Archive for inputs
• Highly tuned system may be susceptible to
small perturbations

#PIWorld ©2020 OSIsoft, LLC 17


Customer Talks Product Labs
How We Built An Automated
Recommender With Real-Time Data Optimizing Asset Analytics for
Analytics Using PI AF & PI Vision Better Performance

Predictive Maintenance in GTs


Compressors Usage-based, Condition-Based
and Predictive Maintenance using
Atlantic LNG Energy Management System the PI System
(ARTEMIS) based on Pi Asset Framework

See the Light, Easy Wins to


Using PI analytics to create a more
efficient and optimized Wastewater Improve your Asset Framework
Treatment Plant Experience

#PIWorld
18 ©2020 OSIsoft, LLC
Communicate with OSIsoft Product Managers

https://feedback.osisoft.com
If it is not shared on the feedback portal, it didn’t happen!

#PIWorld ©2020 OSIsoft, LLC 19


• Stephen Kwan
[email protected]
• Product Manager
• OSIsoft, LLC

#PIWorld ©2020 OSIsoft, LLC 20


Questions? Please remember to…
Complete Survey!
Navigate to this session in
Please wait for mobile agenda for survey
the microphone
DOWNLOAD
THE MOBILE
State your
APP
name & company

#PIWorld ©2020 OSIsoft, LLC 21


#PIWorld ©2020 OSIsoft, LLC 10

You might also like