Sensor Networks & Metering

A smart building rarely has one sensor of a kind; it has a population of them, deployed to a plan. Two questions govern that plan: coverage (are the things you care about being measured at all?) and density (how finely - one sensor per floor, per zone, per desk?). Density is a genuine trade-off. Too sparse and you get an average that hides the problem - one temperature sensor for a floor tells you nothing about the cold corner by the glass. Too dense and cost, cabling, batteries, network traffic and maintenance balloon for readings that barely differ. Good design matches density to how much the quantity actually varies in space and how granular a decision you need.

Those nodes are organised into a network topology. Some devices sit on a shared wired bus (many meters and BMS points speak Modbus or BACnet MS/TP down a daisy-chained cable). Wireless devices often form a star around a gateway, or a self-healing mesh (as Zigbee and Thread do) where nodes relay for each other to extend range and route around dead spots. Long-range, low-power sensors on LoRaWAN report to a single gateway that can cover a whole site. Module 2 details these protocols; the point here is that deploying many sensors is a design problem - of coverage, density and topology - not just a purchasing one.

Here is the highest-leverage move in the whole module. Most buildings have exactly one utility main meter - a single number for the whole building each month. That number tells you how much but never where or why. Submetering fixes this by installing meters downstream of the main, splitting the total into named loads: HVAC, lighting, tenant A, tenant B, plug loads - and, one level deeper, the chiller, the pumps, the fans.

The figure draws this as a tree: the utility main at the top, branching into systems and tenants, branching again into individual large loads. The effect is transformative. A single bill that rose 12 percent is a mystery; a submetering tree shows the rise is entirely in the chiller, at night, when the building is empty - a stuck valve or a scheduling fault you can now find and fix. Submetering is what makes energy analytics and fault detection (Module 6) possible at all, and it is equally powerful for water, where a submeter on each riser or wet area turns 'the water bill is high' into 'there is a continuous overnight flow on the third-floor branch' - almost always a leak. A rough discipline: submeter every load big enough that you would want to explain it separately. You cannot manage what you cannot see, and the main meter shows you almost nothing.

Every real deployment is a negotiation between wired and wireless, and the figure contrasts them on one floor. Wired sensors and meters sit on a powered, reliable backbone - a Modbus or BACnet bus, or PoE runs. They never need a battery, they are hard to interfere with, and they suit fixed, high-value points: main and sub-meters, BMS field devices, actuators. Their cost is routing the cable, which in an occupied, finished building can be brutal - chasing walls, lifting ceilings, closing rooms.

Wireless flips it. A battery radio node can be placed in minutes almost anywhere, which is why retrofits lean on it heavily. But now coverage and density become a radio problem. Each node has a range; walls, floors and metal attenuate the signal; and the figure shows the real design task - arranging nodes and gateways so their coverage overlaps with no blind spots between them. A mesh helps by letting nodes relay for one another. The honest trade-off: wired gives reliability and power but is costly and disruptive to route; wireless gives speed and flexibility but ties you to battery life, range limits and interference. Most real buildings are a hybrid - a wired backbone for meters, actuators and critical points, wireless for the long tail of light, occasional sensors. Deciding which is which, early, saves enormous pain.

It is worth ending on the instrument most people overlook. A meter is a sensor - one that integrates a flow (of electricity, water, gas, heat) over time - and it is often the single most valuable stream a twin receives, because it maps straight onto money and carbon. Modern smart meters and sub-meters do more than tick a total: they record interval data (typically every 15 or 30 minutes), which reveals the shape of consumption - the overnight base load that should be near zero, the morning start-up spike, the weekend that should be quiet but is not.

Meters expose their data in practical ways you will meet on the job: a pulse output (a contact that closes once per unit consumed), a Modbus or BACnet register you can poll, or a utility data feed. That interval history is the raw material for benchmarking a building against its peers (tools like ENERGY STAR Portfolio Manager do exactly this), for spotting waste, and for verifying that a retrofit actually saved what was promised. The practical instinct to carry forward: before adding exotic sensors, make sure the building is properly metered and submetered. Meters are cheap relative to their value, they rarely lie about totals, and they answer the question owners care about most - where is the money going, and can we prove we cut it?

Put the pieces together on a real pattern. Imagine a mid-size office whose electricity bill has crept up over a year. The single main-meter total confirms the rise but explains nothing - it is one number going the wrong way. Now bring in submetering and interval data. The HVAC sub-meter, plotted at 15-minute resolution, shows a base load - the consumption when the building should be empty, overnight and at weekends - that is not near zero as it should be, but sits stubbornly high. Drill into the chiller sub-meter and that base load traces to the chiller running through the night in an unoccupied building.

That single reading reframes the problem entirely. It is not that the building got busier or the weather harsher; a schedule or a control is wrong - a valve stuck open, an override left on, a time-clock that never turns the plant off. The fix is cheap; finding it without submetering and interval data would have been nearly impossible, because the monthly total hides exactly the overnight shape that gives it away. This is the everyday power of the module: the base load that should be zero, the spike out of hours, the weekend that should be quiet are the signatures a metered building reveals and an unmetered one conceals.

The same reading habit extends to water - a continuous overnight flow on one branch is almost always a leak - and to any load with a predictable rhythm. It is also the raw material for benchmarking a building against its peers and for measurement and verification: proving a retrofit actually delivered the saving it promised. Learn to read the shape of a building from its meters and you hold one of the most immediately useful skills in the field - and the direct on-ramp to the analytics of Module 6.