Time granularity not fixed

How to handle facts where the time component is not of fixed granularity? All the textbook examples I can find seem to assume a fact represents a fixed time period.

My scenario is a SCADA application capturing events of varying duration. The measures I want to record may be valid either at a particular time or over some variable period of time. Measured values don't all occur at the same time or with the same frequency. Are there examples of dimensional models for applications like this?

Thanks

SCADA?

Supervisory Control and Data Acquisition
cpni.gov.uk/advice/infosec/business-systems/scada/

Sounds like process control information - such as metered measurement readings?

Take a look at Kimballs article regarding a Human Resources data mart.

I think the concept applies here .... a new fact record is created for the meter whenever a measurement event occurs.That meter reading is effective dated until the next measurement event occurs.

This approach allows for point-in-time queries (for measures that are statuses/balances) as well as time period queries (for measures that are additive.)

Time can also be a metric, right? So add a column to track elapsed time. You can also add a start and ending dimension or alternatively add an elapsed time banding dimension.