Algorithms, alerts and alarms

An algorithm performs a calculation on real-time or historical (trend) data to generate a result. The result can trigger an alert or an alarm, be displayed on a chart, or can become an input to another calculation.

A single algorithm can run against data collected from an entire building. For example, assume your building has 100 air handling units and an algorithm to monitor their performance. When you add 50 more units, and tag each unit appropriately, without any additional effort on your part the original algorithm applies to all 150 units.

Using a poller, an alert runs an algorithm to monitor point performance. Alerts may trigger alarms, which appear on the normal alarm console. If the alert that triggered an alarm continues to exist, the alarm persists on the alarm console even after it has been acknowledged or force cleared.

The framework provides two uses for algorithms. You can use them to:

Test for an individual condition. The alert running the algorithm can then automatically sound an alarm or trigger a remedy.
Look back at the history of a point and report any events that meet defined criteria. For example, an algorithm can answer the question: Has there been any time in the past when a hot water valve was open more than 90% with a room temperature of three degrees below the setpoint for an hour? Depending on the amount of data (that is, how far back in the past the stored data exist), you may be able to identify a consistent pattern (a trend) that points to a condition requiring attention.
NOTE: Although algorithms run on any platform, if you intend to process large volumes of historical data, consider running algorithms on a Supervisor platform.

Algorithms, alerts and alarms

Related Links