Most commercial HVAC systems ventilate for a building that does not exist: the fully occupied, nine-to-five, five-day building their schedules were commissioned for. Hybrid work broke that assumption, and the result is the strangest waste profile in the built environment, floors ventilated hard while empty on a Friday, and meeting rooms under-ventilated while twelve people breathe in them on a Tuesday. Demand-controlled ventilation (DCV) is the fix, and it runs on two measurements this platform happens to specialise in: CO₂ and occupancy.
01 · WHAT DEMAND-CONTROLLED VENTILATION ACTUALLY ISWhat demand-controlled ventilation actually is
Demand-controlled ventilation adjusts outdoor air intake to the number of people actually in a space, instead of a fixed schedule or a worst-case design assumption. The mechanism is not exotic: dampers, variable-speed fans and a building management system that accepts an input signal. What changed recently is the quality of the input signal.
The industry benchmark for ventilation rates is ASHRAE Standard 62.1, which defines acceptable indoor air quality and explicitly recognises DCV as a compliance pathway. The logic is simple: ventilation requirements scale with people. If you know how many people are in a zone right now, you can ventilate for that number, not for the number the architect assumed in 2011.
02 · THE MEETING-ROOM SPIKEThe meeting-room spike
The sharpest version of the problem lives in conference rooms. Outdoor air sits around 420 ppm CO₂. A well-ventilated office holds below 1,000 ppm, the conventional comfort threshold. Put a full meeting in a small room and concentration can pass 2,000 ppm within the hour, into the range where studies associate measurable declines in decision-making performance.
A schedule-driven system cannot see this. It ventilates the room identically whether the meeting is happening or the room is one of the ghost bookings that plague every hybrid office. An occupancy-driven system sees the twelve people arrive and lifts the air supply before the CO₂ curve peaks, then drops it the moment the room empties.
03 · WHY THE OCCUPANCY SIGNAL HAS TO BE ANONYMOUSWhy the occupancy signal has to be anonymous
The obvious way to count people in every room, cameras streaming to a server, is the wrong way, and not only for privacy reasons. It creates a surveillance governance problem that stalls in legal review, and in a workplace it lands in enterprise-agreement territory the moment footage of staff exists anywhere.
The architecture we build avoids the problem structurally: edge computer vision runs the count on the device, raw frames never leave the sensor, and the only output is a number per zone. Ventilation control needs exactly that number and nothing else. A damper does not need to know who is in the room. It needs to know how many lungs are.
A damper does not need to know who is in the room. It needs to know how many lungs are.
04 · WHAT DCV IS WORTHWhat DCV is worth
HVAC accounts for roughly 40 per cent of a typical Australian office building's energy use, per the NABERS energy efficiency guidance, and ventilating unoccupied space is the most avoidable slice of it. The savings mechanics are straightforward:
- Conditioning outdoor air is expensive. Every cubic metre of intake must be heated or cooled to room temperature. Intake matched to occupancy is intake you do not condition for nobody.
- Fan power scales steeply. Fan energy rises roughly with the cube of airflow, so even modest reductions in unneeded airflow compound into real savings.
- The empty-floor problem is chronic. Hybrid attendance means whole zones now sit near-empty on anchor-day off-peaks, exactly where schedule-based ventilation wastes the most.
The same occupancy feed that drives the dampers also lands in the reporting layer, which is where it starts paying twice: the measured alignment between occupancy and HVAC operation is precisely the evidence an energy-performance story needs, whether that is an internal sustainability report or a NABERS submission pack.
05 · WIRING IT UPWiring it up
In practice a DCV loop with anonymous occupancy input is a short chain: the sensor publishes a per-zone count every few seconds over MQTT or HTTPS, the platform holds the zone-to-AHU mapping, and the BMS receives either a setpoint adjustment or the raw count through the integrations layer. Three implementation notes from our scoping work:
- Zone granularity beats building averages. A building-level people count is nearly useless for ventilation; the waste and the under-supply both live at zone level.
- Trend, not snapshot. The useful control signal is the count plus its direction of change. A room at 8 people and filling needs different air than a room at 8 and emptying.
- CO₂ stays in the loop as the auditor. If the count says empty but CO₂ stays high, something is wrong, a miscalibrated zone, a blocked damper, a sensor fault. The disagreement is itself a maintenance alert.
06 · WHERE TO STARTWhere to start
The honest starting point is not a building-wide retrofit. It is one floor with measurable waste: pick the level with the flattest Friday attendance, put an anonymous count on each zone, and run four weeks of occupancy against the existing HVAC schedule. The gap between the two curves is the business case, quantified, before anyone touches a damper. If that is a study you want to run, start with the HVAC use case or talk to us about your floor.