The electricity bill arrives at the end of the month, finance pays it, and the question 'why so much?' stays unanswered. A single utility meter tells you how much energy you used, but not where, when or for what. Those three questions are exactly what energy monitoring is for.
A monitoring system does not save energy by itself. It shows where energy is being lost; the savings come from the decisions you make based on the data. Below we look at how to build a system whose data is actually used, rather than sitting on a dashboard nobody opens.
Why the monthly bill is not data
A bill compresses a whole month into one number. It will not show you that ventilation runs through the weekend, that the server room cooling works at full output all night, or that equipment switching on at the same time every morning creates a demand peak. These patterns only become visible when consumption is broken down by time and by system.
What to measure: the metering hierarchy
A good system starts with a plan that follows the building's electrical single-line diagram: the main incomer, then the main distribution boards, then the circuits that carry significant load. You do not need to measure everything — you need to measure what you can act on.
- Main incomer — total consumption and peak demand
- Heating, ventilation and air conditioning (HVAC) — usually the largest and most controllable load
- Lighting by floor or zone
- Server room: IT load and cooling metered separately
- Kitchen, socket circuits and other specific consumers
- Tenants or departments, if you need to allocate costs
A simple check: the sum of the sub-meters should come close to the main meter. The difference is 'unaccounted consumption'. If it is large, either something is missing from the plan or a meter is installed incorrectly.
Interval data and base load
The value of monitoring lies in interval data — readings recorded, for example, every 15 minutes. The single most useful metric they give you is the base load: how much the building consumes at night and at weekends, when almost nobody is there.
Imagine, for example, a 60-person office where the night-time load approaches half of the working-hours load. That is not necessarily wrong — the server room and refrigerators run at night too — but it is now a specific question you can answer with data broken down by system.
Comparing fairly: normalization
Absolute numbers mislead: consumption rises in winter, headcount changes, floor area grows. So compare normalized figures. Energy use intensity = annual kWh ÷ floor area in m². For heating and cooling, compare consumption against degree days. For the server room, use PUE = total room consumption ÷ IT equipment consumption — the closer PUE is to 1, the less energy goes into cooling and losses.
From data to decisions: typical findings
- HVAC runs outside the working schedule — correcting the schedule is often the cheapest measure
- Heating and cooling run simultaneously in the same zone — a sign of a control or sensor problem
- Lights are on in empty spaces — a candidate for occupancy sensors and scenes
- Server room cooling is set lower than the equipment actually needs
- Morning demand peaks caused by everything starting at once — stagger the start-up
- Consumption creeping up with no visible change — an early sign of equipment faults
Rank measures by cost and effort. First come the things that are almost free: schedules, setpoints, switch-off rules for nights and weekends. Next come control improvements: sensors, scenes, automation. Replacing equipment is the last step, and only when the data shows the problem lies in the equipment itself rather than in how it is controlled. Following this order also lets you see the effect of each change separately.
Integrating with the building management system
Monitoring is most effective when it is connected to the building management system (BMS) and automation. Meters usually send data over standard protocols such as Modbus. A BMS can then do more than display consumption — it can act on it: send an alert when night-time load exceeds a threshold, or adjust a schedule automatically. This depends on a reliable network; meters and controllers are best placed in their own network segment (VLAN).
Questions to ask the supplier
- What accuracy class are the meters and current transformers?
- At what interval is data stored, and for how long?
- Can data be exported in an open format (e.g. CSV) or via an API?
- What happens to data if the connection drops — is it buffered locally?
- Who updates the meter list and diagram when the electrical installation changes?
Making monitoring part of everyday practice
The hardware is only half the job. The other half is organizational: who looks at the data, and what happens next.
- Appoint an owner — a facility manager or engineer responsible for the metrics
- Define a few key indicators: base load, kWh/m², peak demand, PUE
- Set alerts for significant deviations, not for every event
- Review the data regularly, for example monthly, and record what has changed
- After each measure, compare the result against the baseline period
Want to find out where your building's energy goes? Get in touch — we can help you define a metering plan and the scope of a monitoring system.