I had a particular problem that could be solved with the automatic recalculation functionality or at least that is what I was thinking
At our power plant we have 3 different shafts: 2 gas turbines and 1 steam turbine.
The gross power production data of both gas turbines is real time data in MW.
The data for the steam turbine is late arriving data as it is fed to the PI system once a day at midnight. It is fed as quarter hourly gross energy production in MWh in a batch of 96 (24h * 4) values.
The goal is to have a parent element (with 3 child elements) that resembles the quarter hourly energy production of our whole (gas turbine 1 + gas turbine 2 + steam turbine) power plant in MWh.
To reach this goal I have setup to following hierarchy:
Each child node of 'Energy gross active' has an attribute that resembles the quarter hourly gross energy production.
The power production of both gas turbines is converted the a quarter hourly energy production with an analysis that is scheduled each quarter hour.
The energy production of the steam turbine is scheduled each 15 minutes as well, although during the ongoing day the most recent value is stable (late arriving data remember). The most recent data is the one from last night. But I wanted to have a value on the same timestamp as the late arriving data (arriving later on today at midnight).
At the parent level 'Energy gross active', I have setup an attribute as PI point data reference that is linked with a rollup analysis:
This roll up analysis is scheduled event triggered. This analysis is triggered each 15 minutes because child nodes have analysis running each 15 minutes.
In the image underneath, you will find the results generated by the analsyis for the ongoing day in columns 2 tot 5. Highlighted in blue (column 5) is the stable value for the steam turbine which is the value at 't'.
I was convinced - after some automatic recalculation testing on smaller scale - that setting the 'automatic recalculation' flag on the roll up analysis on the parent node would start the recalculation when the child node 'Steam turbine' is receiving its late arriving data.
The expected result is shown in the image above from column 6 to 9.
But the expected result is not met by the setup explained above otherwise I would not be posting this
Is there anyone with some more experience with late arriving data in combination with a roll up analysis and automatic recalculation, that could help me? Others are free to post their thoughts as well