THE HEATING WAS SCHEDULED OFF. IT HAD NOT BEEN OFF FOR NINE WEEKS.

Posted on: April 7, 2026

The Heating Was Scheduled Off. It Had Not Been Off for Nine Weeks.

Anomaly detection
Cost
FM

A BMS fault after a software update. No alert, no visible failure. The consumption data was the only thing that noticed.

Heating anomaly data story visual

Scenario

Priya manages facilities across a portfolio of six commercial office buildings for a regional property company. Energy costs are a significant overhead and out of hours consumption has always been her primary focus for savings. Scheduled setbacks, where HVAC runs at reduced capacity outside occupied hours, are standard practice across all six sites.

Verifying that those setbacks are actually working is a different matter.

What the data showed

When Monitor Hut was deployed across the portfolio, Priya ran a comparison of daytime versus night time consumption across all six buildings. Five looked broadly as expected. The sixth did not.

One 3,200 m² office building was drawing electrical consumption during the 11pm to 6am window at roughly 68% of its peak daytime load. For a building with no overnight occupancy, no server room, and no scheduled process running during those hours, that number had no legitimate explanation.

Half hourly interval data showed the pattern had been consistent for at least nine weeks. An engineer’s visit identified the cause: a BMS controller fault following a software update had overridden the setback schedule and defaulted the air handling units to a continuous run state. No alert had fired. No one in the building had noticed.

Resolution

The BMS was reconfigured. The following week’s overnight profile dropped to 12% of daytime load, within expected tolerance for the building’s residual loads.

The result
Metric Detail
Overnight consumption vs expected 68% of daytime load
Pattern duration confirmed 9+ weeks
Estimated overnight waste per week ~1,840 kWh
Annualised cost at current rates ~£9,200
Time to identify from go live 6 days
Post fix overnight consumption 12% of daytime load

A BMS fault does not announce itself. Without continuous consumption data, scheduled setbacks are assumed to be working until a bill review three months later suggests otherwise.