A conventional building runs its heating and cooling on schedules. The system starts before the first occupants arrive, runs at a fixed setpoint all day and stops in the evening, whether or not the rooms are in use. This is simple and predictable, and it wastes energy in empty meeting rooms, half-full floors and shoulder seasons. Smart HVAC replaces the schedule with feedback. Sensors report what is actually happening in the building, a controller decides what each zone needs, and the equipment responds. The idea is not new, but cheap sensors, standard communication protocols and local computing have made it practical for ordinary buildings. The result depends less on the algorithm than on the quality of the data feeding it.

What does the system measure?

The basic inputs are temperature and humidity in each zone, supplemented by occupancy detection and carbon dioxide concentration, which is a good proxy for how many people are breathing the air in a room. Demand-controlled ventilation uses that signal to bring in outside air in proportion to occupancy, instead of at a constant rate, which saves the energy needed to condition that air. Outdoor temperature, solar load and weather forecasts add context. Equipment itself also reports data, such as fan speeds, valve positions, supply and return temperatures and power draw. Building protocols such as BACnet and Modbus carry much of this information, and gateways translate it into formats that analytics software can use.

Advertisement

Where does edge computing fit?

Comfort control cannot depend on a cloud connection. A local controller, sometimes called an edge device, runs the fast loops that keep each zone at its setpoint, and it keeps doing so if the internet link fails. Higher-level optimization can sit above it, either on the same device or in the cloud. One approach is model predictive control, in which a model of the building's thermal behaviour and the weather forecast are used to plan ahead, for instance pre-cooling a building overnight when conditions are favourable, or letting a well-insulated zone drift slightly before occupants arrive. The edge device also filters and aggregates telemetry so that only useful data is sent upstream, which reduces bandwidth and storage.

Telemetry, faults and data quality

Telemetry does more than feed the controller. Analysed over time, it reveals faults that waste energy while leaving occupants comfortable, such as a damper stuck open, a valve that leaks, or heating and cooling running at the same time in one zone. Automated fault detection can flag these for maintenance. All of this relies on trustworthy data. Temperature sensors drift, are placed in poor locations near windows or vents, or are mislabelled in the building model, and an optimizer that trusts a faulty sensor will make confident bad decisions. Regular calibration, sensible placement and cross-checks between related measurements are unglamorous but decisive parts of a successful project.

Why it matters

Buildings consume a large share of the energy used in many countries, and heating, ventilation and cooling are a substantial part of that. Control that responds to real conditions reduces waste and operating cost without sacrificing comfort, and it generates data that helps maintenance teams find problems early. The conditions for success are accurate sensing, local control that survives network failures and a continuing process of checking that the system does what the model says it does.

A smart building is only as smart as its least trustworthy sensor.