Easy as PI: Asset Framework (AF)
“Context is to data what water is to a dolphin,” wrote American author Dan Simmons. Without context, data can only get you so far, and that's why OSIsoft created Asset Framework. Stephen Kwan has been the acting Product Manager for Asset Framework (AF) at OSIsoft since 2010, interacting with customers, collecting feedback, and continually helping to improve the product. I recently sat down with him to chat about all things AF - how it works, what problems it's solving, and how it will continue to evolve.
What is Asset Framework (AF)?
AF is a tool that allows you to model either physical or logical objects in the way that best suits the way you wish to view those assets and the data associated with them. With AF, you can identify the components/elements that make up a process, specify relationships between those objects, and organize them in whichever way most aligns with your business.
For example, let's say you have a pump. It generates multiple data streams - power consumption, vibration, fluid volume, impeller speed, oil temperature and pressure -- based on different parameters. AF allows you to see them all at once via visualization tools, pretty much like how a studio mixing board lets you hear a symphony rather than different instrumental tracks. High energy consumption alone can indicate a lot of things. Combining it with low water flow, however, can narrow down the problem to a worn-out impeller. With the right data streams, an AF model can be its own analytics.
Is AF a Digital Twin?
AF is a tool that starts as a blank canvas - you use it to represent or create a digital twin of your assets.
How long does it take to create an AF model?
That all depends on the level of detail and the level of complexity that you want to put into it. You can build an AF representation of a single asset (like a pump) in just a few minutes. If you're building a representation of your entire plant or a complete production line, that would take significantly longer. The level of complexity and detail is entirely based on what problems you're trying to solve.
How does AF interact with other components of the PI System?
At its core, AF gives you context - is something large, small, blue, red, hot, cold, etc. The rest of the PI Server, such as the PI Data Archive, is providing real-time historical data. By combining AF with the PI Data Archive, you now have data and context together, which gives you a better understanding of how your assets are functioning.
For instance, if you're in PI Vision, you can find the temperature of a given pump. However, when you utilize AF in conjunction with PI Vision, you can find the temperature of the fluid coming out of pump X which is in the northwest corner of plant Y. It makes for much richer information.
OSIsoft visualization tools also take advantage of the context available in AF templates. With AF templates, you can create a single template that represents multiple related assets (say, wind turbines that are all the same), and this allows users to easily switch between assets without having to build out new displays each time.
What are some features of AF?
- You can create relationships between assets
- You can have multiple representations of an asset
- AF supports templates which allow users to create commonalities between different objects
- Easily adopt to changes over time; when updates are made to templates, they are immediately reflected in all instances from the template thereby ensuring your digital twins are up to date.
- AF supports a rich set of units of measure, so that users can easily convert between units, which means less miscalculations due to differing measurement units
- Real-time streaming calculation engine
- Capture important events with related data and context in near real time
- System notifications that can be sent to users/other third-party systems
- AF is accessible via programmatic and other standard means for custom applications
Who are some of the bigger users of AF?
Shell has over 15,000 users monitoring 7.5 million data streams from industrial sensors, running 100,000 real-time calculations per minute.
Devon Energy deploys over 50 smart asset AF templates across its exploration and production operations, which consists of more than 50,000 assets. The energy company uses AF as an “OT Chart of Accounts,” aggregating and normalizing data via data references/abstraction and then using AF for real-time analytics.
MOL, one of the largest oil and gas companies in Central Europe, uses AF to track over 80 billion data points a year. They've deployed 21,000 AF templates across multiple refineries, which has led to an estimated EBITDA earnings increase of $1 billion over a five-year period.
You can also use AF to make larger Digital Twins. BP is making digital twins of all 40 of its offshore platforms by combining technology from OSIsoft and Element Analytics. BP will ultimately model over 1 million data streams into AF models.
What's on the horizon for AF?
Currently, the team is collecting information from customers about how they use the system and what improvements would be most meaningful to them. There have been a number of releases over the last five years, so the current focus is on gathering feedback to determine what customers truly want and need.
Users can input suggestions at https://feedback.osisoft.com/- whether that suggestion is a feature or a change in behavior. These requests are monitored by the product management team daily and are crucial to building future iterations of AF.