I've been asked to setup some KPI's using PI data for about 20 bits of equipment. They are all the same type of equipment and will have the same the KPI's. I've been asked for individual values (how long each item has been running at a certain load for a day/week/month, shown as a percentage) and then to create an overall percentage for all the items combined.
I can do this pretty easily in Excel (time filtered expression), but it is wanted in Processbook (and I'm guessing PI Vision as we move forward) so I'm not sure of the best way to do it; AF or performance equations? I can't directly copy the Datalink function as the TimeEQ (and TIMEGE/LE) functions look for a tag not an expression (definitely an improvement list item!) so I think I would have to create a PE for the filter ( speed > 'x' and inlet valve open) and then a 2nd equation to work out how long it was true. Another drawback is that using Datalink I can go back in time with the expression, using AF or a performance equation I think I'm limited to times when the PE used for the filter was available (if I created the PE today, I would only get data from today going forward, nothing historical). Is my thinking sound?
I'm more inclined to go for an AF structure with a template for each item, but I'm concerned about the analysis being done in AF then writing the data back into PI to store the % running values that are being asked for. I've only really used PE's to do calcs before, but the more I look at AF I think it's the better "solution".
Any suggestions about using PI AF over PE's (and vice versa), or if someone knows a better way of generating the data I need, are gladly welcomed.