Ask anyone who has spent time comparing utility bills against design-stage energy models and they will tell you the same thing: the two rarely agree.
The research bears this out. Studies of commercial and institutional buildings have found measured energy consumption can run up to two-and-a-half times the design prediction and the divergence tends to be widest in exactly the building types we care most about getting right, such as schools, campuses and health care facilities.
This is not new information. The performance gap has been documented for decades. What is new is that it has started to cost people money.
Institutional and public sector clients across Canada are writing verified performance outcomes into their project requirements, whether through the Canada Green Building Council’s Zero Carbon Building Standard, provincial decarbonization programs or municipal green procurement policies. Once a project carries a performance commitment, the distance between the model and the meter stops being an academic curiosity. There is a contract attached to it.
In my experience that gap traces back to three sources:
First: The assumptions going into the model
Every energy model is a stack of assumptions, and the model is only as honest as the assumptions are. Compliance protocols hand us default values for occupancy density, plug loads, operating hours and so on. Those defaults exist for a legitimate reason: they make models reproducible and comparable across projects. But a default is a statistical average of many buildings. It describes none of them in particular.
Take a standard office profile: nine to five, five days a week. Now put a call centre in that building running three shifts. Same architecture, same systems, wildly different energy outcomes. The further the real operating pattern drifts from the default, the wider the gap. For institutional buildings with complex schedules the drift can be enormous.
The fix is unglamorous: sit down with the client and ask how the building will actually run. It happens far less often than it should.
Second: The people in the building
How people use thermostats, blinds, plug loads and lighting controls is one of the largest drivers of variation between predicted and actual performance, and it is hard to simulate.
The standard modelling treatment assumes occupants who behave: arrive on schedule, leave on schedule, never override a control. Anyone who has walked a floorplate knows better. Plug loads run higher than the model assumed. Lighting overrides accumulate. Setpoints creep. None of this is a failing of the occupants; it is simply what people do in buildings.
The answer is not to model a worst case. It is to model a realistic case and then design systems that hold their performance across a sensible range of behaviour.
Third: What happens between design and handover
The third source of the gap has nothing to do with the model at all. A well-designed, well-simulated building will still underperform if the systems are not commissioned properly. Unbalanced air systems, controls sequences that were never set up as designed, sensors out of calibration.
Part of the problem is structural. Commissioning is still bolted onto the end of most projects, disconnected from the design and modelling work. If the commissioning agent never sees the model’s assumptions, then nobody is actually checking whether the building that got built matches the building that got simulated. That disconnect is the single most reliable predictor of a large performance gap.
Buildings that leave site in good shape can drift within their first year of operation: schedules get changed, overrides pile up, maintenance slips. Ongoing commissioning is how you catch that drift before it hardens into the building’s permanent operating condition.
The gap closes at the design stage or not at all
Most of the performance gap is locked in before a shovel touches the ground. That is uncomfortable, but it is the opportunity. The teams getting this right are the ones interrogating their occupancy assumptions, modelling the building the client will actually operate and bringing commissioning into the conversation early enough that the model and the field work share a common baseline.
Projects pursuing ZCB certification are already operating this way, because the standard’s post-occupancy verification requirement makes the gap visible and expensive. Calibrated models, realistic load profiles, early commissioning integration, post-occupancy monitoring: all established practice.
What has been missing is the expectation that closing the gap is part of the engineer’s job. Producing a model that satisfies the permit is the floor. Producing one that honestly describes the building that will be built and operated is the actual work. That expectation has arrived in the Canadian market, and the firms that organize their practice around it now will not be the ones scrambling to catch up later.
Chris Flood is vice-president, Canada, at IES, a global building performance software company. IES works with engineering firms, developers and institutional clients across Canada on energy modeling, code compliance and whole-building performance simulation. Send Industry Perspectives Op-Ed comments and column ideas to [email protected].
