PI AF Notification Failure: What to Look for?

At the facility where I work, we need to have near-real-time notification of certain events. I recently created an event frame to be triggered when a digital PI point switches from a value of "Normal" to a value of "Alarm". I received word today that such an event had occurred, but no notification went out. When I looked at the event frames, there was nothing showing for the most-recent event. I checked the archive history for the point that should trigger the event, and sure enough it had shown a value of "Alarm" for about five minutes. Without having changed any aspect of the analysis, I performed a Backfill/Recalculate on the analysis to see what might happen, and doing that caused it to capture the event and (belatedly by fifteen minutes) send out the notification.

 

What should I look for to understand why the event frame failed to generate organically? Are there any known issues that might contribute to this behavior?

EDIT: I did discover that the notification trigger on the template, under Options related to child event frames, had the option selected "When the severity is higher than any previously true trigger condition" instead of "When any trigger condition is true". However, this analysis should not trigger child event frames, so I'm not sure that option changes anything.

 

Thanks,

James Aylard

  • It is hard to say why your Event Frame did not originally trigger simply based on everything that you wrote here. Your edit clause about trigger severity would not have affected it though IF you only had one start trigger.

    Generally speaking in the future when you create an Event Frame make sure to utilize the 'Preview Results' functionality before you call it a day. You can do this by right-clicking on the analysis itself and clicking 'Preview Results'. You can do this from both the template and element. You can also preview results without checking in. I cannot stress enough how useful this functionality is. It will show you over whatever time period you want which events would trigger and their corresponding outputs based on your template and event frame attributes. This goes for regular analytics as well. Always preview results!!

    Once you have previewed you can convince yourself if the results make sense or not. If they do, then backfill and check the event frames and make sure they are being generated in accordance with the Preview Results.

    Philip Babb - Industrial Insight - pbabb@industrialinsightinc.com

  • Check that your event scheduling is set to 'Event-Triggered', trigger on 'Any Input', and like you found, 'When any trigger condition is true'.

     

    What type of compression is set for the tag itself? There may have been something funky with the tag going on in the moment that ended up being compressed/normalized so by the time you ran your backfill/recalculate it would've been working with a potentially slightly different value set.

     

    Do you have a 'True For' requirement that must be met before generating the event frame? That could lead to the delay in your notification.

     

    Was this an email notification or SMS? If SMS: how are you delivering your SMS? Using carrier gateway or dedicated service? I've had issues in the past related to carrier gateway reliance leading to some SMS notifications being delayed to different batches of phone numbers in a list. I've recently swapped to using Azure Communication Services and have much better reliability.