The PI digital twin places users above seeing streaming data on a trend. We actually want the analytics engines watching that for us.
The architecture and approach is to create then an ecosystem that can accurately detect poor health and underperformance. Each asset slightly unique. The output of the PIDT then can be accurate anomaly records. They are your most valuable data, prepared for analysis. PI Servers (plus optional analytics engines attached to pi) do the rigorous condition monitoring 24/7 so you (operations) can focus on more critical thinking. If there is a new condition detected, PI will tell you. Fewer operators, achieving better results. Humans learning from data and making better data. Nevery wasting their time and resources. Better data sets to feed AI/ML engines. Richer data in deeper context compared to having sensor time series data alone.
To stand up such an accurate and openly evolving monitoring solution we have to be very efficient. Scale is always big in industry. We must always sharpen our focus, always be learning. To be this, we set high standards for our work. For example, Every analytic must be defined in one place and re-used. Improvements made in one template apply to all machines at all plant sites. Users at all sites can see the failures of any site. We can solve problems faster and avoid recurrence with more awareness and storing of our past trials. When we use more PIAF elements, we create more capability.
Your work as a director of such an ecosystem elevates to higher thinking. To the transactional records for poor behavior. To the awareness of how things fail, root causes, failure modes, failure propagation rates, condition detection, forecasting remaining life etc.... Event frames are the transactional scheme in PI you want your users to be reading, studying interacting with, sharing every day . You will always be looking to expand and improve what content belongs in the event frame. This not only sensor data. The raw time series falls more into the background.
This is the 'work by exception' mindset. Operators pivot quickly away from healthy machines.