Learn
What is HVAC monitoring?
HVAC monitoring is the continuous collection and analysis of operating data from heating, ventilation and air conditioning equipment — temperatures, damper positions, run status, electrical draw — to find equipment operating incorrectly while still running normally enough not to alarm.
Why it exists
HVAC is typically the largest controllable energy load in a commercial building, and it is the system most likely to be quietly broken. Berkeley Lab analysed fault-detection records from more than 60,000 pieces of commercial HVAC equipment across roughly 90 fault types and found 40% of air handling units and 30% of terminal units carrying a reported fault on any given day — in buildings that all had a building management system. Twenty-one distinct AHU fault types appeared on a fifth or more of every unit monitored. (Crowe et al., Science and Technology for the Built Environment*, 2023, DOI 10.1080/23744731.2023.2263324.)*
Not one of those faults necessarily produced a complaint. A stuck economizer holds space temperature. An AHU in simultaneous heating and cooling holds space temperature. A schedule that never took effect after a holiday holds space temperature. The building is comfortable, nobody calls, and it spends money continuously.
The points that actually matter
Monitoring value comes from a small subset of the available points, and knowing which ones separates a useful deployment from an expensive one.
From the BMS, for air-side analysis:
| Point | Why |
|---|---|
| Outside, supply, return and mixed air temperature | Economizer and mixing analysis; ±2°F accuracy required for code-recognised economizer FDD under California Title 24 |
| Damper and valve commands and feedback | The gap between commanded and actual is the fault |
| Fan and pump status, speed and command | Schedule violation, VFD override, hand-off-auto left in hand |
| Supply air and duct static pressure, with setpoint | Reset strategies that were designed and never enabled |
| Zone temperature and setpoint | Simultaneous heating and cooling, unoccupied setback failure |
From electrical submetering: kW per major HVAC circuit, clock-aligned to the service meter — which gives runtime, cycling, staging behaviour and off-hours operation without touching the control system at all.
On how the data is collected: read the FDD-relevant subset over BACnet using change-of-value subscriptions rather than polling. Aggressive polling of a BACnet MS/TP trunk degrades the control network, and it is the fastest way for an energy vendor to be removed from a site.
Worked example
A single 40-ton rooftop unit with an economizer stuck at minimum position.
- Mechanical cooling runs when outside conditions would have allowed free cooling
- Economizer-available hours in a mid-Atlantic climate: roughly 1,200 hours a year
- Cooling load during those hours, at approximately 1.0 kW per ton: 40 kW
- 40 kW × 1,200 h = 48,000 kWh a year
- At $0.12/kWh: $5,760 a year, on one rooftop unit
A building with fourteen such units, and the 40%-fault rate Berkeley Lab measured, is carrying something in the region of five units with a reported fault at any moment. The arithmetic scales unpleasantly, and none of it appears as a complaint.
The honest boundary on what each data source can see
Electrical data alone. MIT's field study identified seven HVAC faults detectable from electrical load alone. It could not see coil fouling, could not separate small loads at a centralised measurement point, and lost startup signatures behind variable-frequency drives. What electrical data does well is the unglamorous set: equipment running outside schedule, loads that never turn off, standby power, staging that never sheds. Those are consistently where the money is, and they need no air-side sensors at all.
Air-side analysis. Economizer, mixing, reset and simultaneous heating-and-cooling faults require temperature points at ±2°F. Those are not electrical points. Where a BMS carries them and is trended correctly, the analysis is available; where it does not, sensors have to be added, and that is a different project with a different cost.
Neither source substitutes for the other. A vendor claiming full HVAC fault detection from submetering alone is over-claiming, and one claiming it from a BMS with no electrical data cannot dollarise anything.
What HVAC monitoring is not
It is not a building management system. A BMS controls the equipment. Monitoring reads it, analyses it and reports. Monitoring is read-only by default, and any write-back capability belongs behind separate written authorisation.
It is not a thermostat or an alarm. Alarms fire when a space goes out of range. The faults that cost the most money never take a space out of range — that is precisely why they persist.
It does not require replacing the BMS. Most useful HVAC monitoring reads an existing BMS over BACnet or Modbus. A proposal that starts with a controls replacement is a controls project with monitoring attached.
It is not the same as remote access. Being able to log into the BMS from a phone is remote access. Monitoring is the continuous automated analysis that happens whether or not anyone logs in — and LBNL's evidence is unambiguous that the logging-in model saves about a third as much.
Common questions
What is the difference between HVAC monitoring and a building management system?
A building management system controls the equipment — it commands dampers, valves, fans and setpoints. HVAC monitoring reads that system's data, analyses it continuously against expected behaviour, and reports faults. Monitoring is read-only by default and does not replace the control system.
What HVAC faults cost the most?
The ones that never cause a comfort complaint: economizers stuck at minimum position, simultaneous heating and cooling, schedules that never took effect, static pressure and temperature resets that were designed and never enabled, and equipment left in hand rather than auto.
Can HVAC faults be found from electrical data alone?
Partially. MIT's field study identified seven HVAC faults detectable from electrical load alone, and could not see coil fouling or separate small loads at a centralised measurement point. Schedule violations, loads that never turn off, standby power and staging that never sheds are all visible electrically, and those are consistently where the money is.
What sensor accuracy does economizer fault detection require?
Code-recognised economizer fault detection, such as under California Title 24, requires outside, supply and return air temperatures accurate to ±2°F. Those are air-side points and cannot be inferred from electrical measurement.
OptimizeOS reads the FDD-relevant subset of a BMS over BACnet using change-of-value subscriptions, alongside circuit-level electrical data — see alerts and fault detection.