Description
The purpose of this KbA is to describe the accuracy of virtual meters. In general use, virtual meters will be at least 80% accurate compared to a dedicated BACnet power meter. For getting general trends of which equipment is using the most power 80% is enough accuracy to get started.
Accuracy
The VM accuracy is related to multiple factors but mostly to TL’s “quality” we are getting from the BMS.
The accuracy of calculated energy consumption is dependent on many factors. Increase accuracy by:
- monitoring Amperage and not status (off/on)
- faster TL frequency. COV with a small increment on amperage is preferred. 15 minute samples are very crude and will not be accurate.
- Voltage variance will typically cause 2-3% error.
Example 1: Pool Circulation pump at constant load
The first (pool circulation) typically switched 2 to max 8 times per day and running at constant load – achieved average accuracy (between Modbus power meter and the VM calculation) was @90% (typically 83% to 96%). The pump operation monitored by the current sensor, AIC calibrated correctly, and COV TL. Voltage was value in AV at average measurement achieved by 24h monitoring. The result fluctuation was mostly caused by the Voltage difference between real value and average value in AV.
Example 1:VSD pump with differential pressure control
The VSD pump differential pressure control modulating almost continuously but had a well-tuned PI loop – achieved average accuracy (between VSD BACnet power consumption value and the VM calculation) was @81% (typically 72% to 91%). The pump operation was monitored by speed signal, 5 min Polled TL, and calculation compensated based on affinity laws. The result fluctuation was mostly caused by speed modulation and 5 min TL sampling.
Using 15 min TL samples was useless as the accuracy was very low.
