Occupancy-based HVAC control: run the building on real use, not a schedule from 2009.
Most commercial HVAC runs to a fixed schedule built for the busiest day it will ever see, then conditions empty floors for the other 250 days of the year. Occupancy-based HVAC control replaces that guess with a live, anonymous signal of how the building is actually used, so demand-controlled ventilation and setpoints follow the people instead of the clock.
A schedule is a guess you pay for every hour.
Fixed schedules over-provision by design. They cool the whole floor because three people might come in early, and they ventilate for a full census on a half-empty Friday. Across a commercial portfolio that over-cooling and over-ventilation is one of the largest controllable line items on the energy bill, and it converts straight into avoidable emissions. The money and the carbon are both sitting in the gap between the schedule and reality.
The standards quietly assume it. ASHRAE ventilation rates are written around design occupancy, the theoretical full house, so a building that holds that outdoor-air rate all day is over-ventilating by exactly the margin between design and reality. The same logic inflates energy intensity: the kWh per square metre stays high because the denominator is conditioned space that is mostly empty. Occupancy is the missing variable that turns a static setpoint and a fixed ventilation rate into something that tracks the actual load on the plant, and it is the variable a fixed schedule can never see.
Over-cooling empty space
Conditioning floors and zones nobody is using is the default failure mode of a clock-based system.
Ventilating for a phantom census
Outdoor-air rates set for peak occupancy run all day, even when the real headcount is a fraction of it.
Comfort complaints anyway
Guessing high does not even buy comfort: the busy zones still drift while the empty ones are frozen.
How occupancy-based HVAC control works, zone by zone.
Anonymous counts and density per zone feed thresholds; the thresholds drive actions in your building management system, a setpoint nudge, a ventilation rate, a schedule override. The occupancy signal is camera-free and produced on-device, then delivered over BACnet, Modbus, REST or MQTT. See the full list of BMS and data integrations and the on-device edge AI that produces the counts.
Anonymous counts per zone
On-device sensors report live counts and density. No video, no identities, just numbers per zone.
Thresholds and rules
Define what empty, light and busy mean for each zone. The rules live next to your existing control logic.
BMS action over BACnet
Setpoint, ventilation rate or schedule override, executed by the BMS you already run.
A 28-day model: one schedule change, measured.
This is a worked example modelled on RMIT Building 80, not a live customer deployment. It runs the same anonymous occupancy data against the building's baseline and reconciles the result on a degree-day-normalised basis. Full methodology, baseline and assumptions are in the post on right-sizing HVAC against real occupancy.
Over 28 days, vs baseline
At $0.30 per kWh blended retail
0.68 tCO₂e per MWh, AEMO Victoria
DDH-normalised
One change, not a retrofit
Logged in the BMS portal
Reduce HVAC energy with occupancy: a rough estimate.
Put your own numbers in. This is a back-of-envelope model on the same factors as the worked example, not a quote.
Illustrative only. Assumes $0 per kWh blended retail and 0.68 tCO₂e per MWh (AEMO Victorian grid). Your result depends on your tariff, baseline and how the schedule is changed.
Counts that survive a NABERS submission.
The same anonymous occupancy data that drives the HVAC schedule also underwrites the energy and sustainability story you have to report. It is audit-grade and traceable, which matters when the saving has to hold up in a NABERS pack rather than a slide. Because the occupancy series, the kWh avoided and the NEM grid emissions factor are all logged against the same timeline, the reported tCO₂e carries a baseline an assessor can actually follow, rather than a round number lifted from a vendor deck. See how it feeds sustainability and operations reporting.
Demand-controlled ventilation is only as good as its occupancy signal.
Drive HVAC off a weak signal and you automate a worse guess. The control loop needs occupancy that is accurate, zone-level and anonymous, which is exactly what an on-device optical sensor delivers. See the camera-free occupancy sensor that produces the counts this use case runs on.
Occupancy-driven HVAC, answered plainly.
What is occupancy-based HVAC control?
Occupancy-based HVAC control drives heating, cooling and ventilation from how a building is actually used, measured as live occupancy, rather than from a fixed time schedule. When a zone is empty or lightly used, setpoints relax and ventilation drops; when it fills, the system responds. The result is comfort where people are and far less energy spent conditioning empty space.
How much can occupancy data save on HVAC energy?
Industry figures for right-sizing HVAC against real occupancy typically land between 8 and 18% of a commercial HVAC bill, depending on baseline, climate and how aggressively schedules are tuned. In our 28-day worked example, modelled on RMIT Building 80, a single schedule change avoided about 9,420 kWh and roughly $2,830, a 14% reduction against baseline. Treat it as an illustrative model, not a guarantee.
What is demand-controlled ventilation?
Demand-controlled ventilation (DCV) modulates the outdoor-air ventilation rate to match real demand instead of running flat out for a full building. Traditionally it is driven by CO₂ sensors; pairing it with anonymous occupancy counts gives a faster, zone-level signal, so you ventilate the spaces that are occupied and ease off the ones that are not.
Does occupancy-based HVAC need cameras?
No. Occivar measures occupancy with on-device edge AI: the frame is processed on the sensor and discarded, so no video is stored or transmitted and no faces are recognised. The HVAC system receives anonymous counts and density, never imagery, so you get the control signal without the surveillance liability.
How does the occupancy data reach our BMS?
Anonymous counts and threshold events are exposed over the protocols building systems already speak: BACnet and Modbus for the BMS or BAS, plus REST and MQTT over HTTPS for everything else. The occupancy layer becomes another input your existing control logic can act on.
Will this work with our existing building management system?
In most cases, yes. Because the data is delivered over BACnet, Modbus, REST or MQTT, occupancy-based HVAC control sits on top of the BMS you already run rather than replacing it. You keep your control sequences and add a real occupancy signal to drive them.