The building
that was fighting
itself.
A six-storey mixed-use commercial building in the Greater Toronto Area — restaurants and retail on the ground floor, five levels of open-sided parking above. Seventeen fan coil units, each with its own condenser. On paper, unremarkable. In the trend logs, a building spending most of its energy undoing its own work.
Twelve compressors
in eight years.
Equipment with a fifteen-year design life should not need replacing twice in under a decade. When it does, the equipment is rarely the actual problem — it is the last thing in the chain to fail, and the only thing anyone gets an invoice for.
The operator raised it as a maintenance complaint. Read as a symptom instead, it pointed directly at short-cycling, poor heat rejection and control logic that never let the machines settle.
Thermostats in the ceiling
Several zone thermostats were mounted in the ceiling plenum rather than the occupied space — reading warmer than the room, so cooling ran long and heating cut out early.
Supply air at 27.5 °C, heating valve shut
A fan coil delivering hot air with its heating valve commanded fully closed. Either the valve is passing, or the sensor is lying. Both cost the same until someone checks.
CO₂ sensors reading, nothing responding
Space CO₂ sensors installed and trending at around 600 ppm, with no corresponding damper movement anywhere in the logs. Demand-controlled ventilation in name only.
Manual overrides locking out control logic. Cooling sequences that fail to engage while zones sit above setpoint. Fans commanded on with status reading off, and no alarm configured to catch it. Individually, each is a work order. Together they describe a building running at constant maximum output with almost no feedback — which is exactly what the utility bills said before anyone opened the BAS.
Five faults,
twenty-one zones.
Every zone was trended and scored against its setpoint. Seventeen of twenty-one were outside a half-degree band. The faults below are not evenly spread — several units carry two or three at once, and those are where the work starts.
Almost none
of this is
capital.
The recommendation was deliberately sequenced into three waves. The first two need nothing but a controls engineer with BAS access and a technician with a wrench. Only the third involves buying anything, and even then it is service work rather than replacement equipment.
Aggregate expectation across the programme: 10–20% of utility consumption within the first billing cycles, with the energy intensity moving back toward benchmark.
Where the
savings sit.
Ranked by what each is expected to return against the load it touches. The largest single item is not equipment at all — it is stopping the building from heating and cooling the same air simultaneously.
Four things
you can check
afterwards.
Recommissioning claims are easy to make and easy to lose. These are the trend-level tests that either pass or do not, two weeks after the work — no interpretation required.
Most savings from controls work are lost within a year — not because the work was wrong, but because a setpoint gets reverted during a service call and nothing is watching. Verification metrics written before the work begins are what make that visible.
Different building.
Same diagnosis.
A 23-storey Calgary tower from 1972 and a six-storey GTA mixed-use building from the early 2000s have almost nothing in common — different climate, different construction, different plant, thirty years apart. Both were found running at constant maximum output because of sensors and sequences rather than equipment at the end of its life.
Published anonymously at the client's request. The building, its owner and its managing agent are not identified, and no address appears on this page. Equipment tags (AC-1 through AC-21) are generic labels from the building automation system and identify nothing. Energy, intensity and percentage figures are reproduced as assessed in the BOMA Phase II Whole Building Optimization report; commercial terms and capital costs have been omitted. Interactive models on this page run on those same figures; inputs you change are your own, not measurements.
Is your building
fighting itself?
Two weeks of BAS trends usually answers it.