The right tools are needed to do a proper energy modelling job. For Certified Passive Houses, using the Passive House Planning Package (PHPP) is mandatory. It is the only energy modelling tool in the world that has successfully been tested against real-world results on a large number of projects (see here for some examples). Thousands of Certified Passive Houses deliver results very close to the modelling expectations. Other modelling tools are clearly not performing that well, or have never been tested.
Researchers at Leeds Metropolitan Uni put Passive Houses in the UK to the test. They temporarily evicted the occupants, heated the houses to 25°C for a minimum of one week in winter, and measured the heat loss through the building fabric in relation to the inside-outside temperature gradient. Results of these tests, and in situ U-value measurements were pretty close to the expectations based on PHPP modelling. Not so much so for other buildings tested, which -on paper -all exceeded UK building code requirements, and were considered energy-efficient.
Not always does the consumption of Certified Passive Houses in every single year match the modelling exactly, though. Climate variations between the test reference year and the year in question make a difference. When a larger sample of projects with nearly identical modelling results for individual units is monitored, the influence of user behaviour becomes obvious (or sometimes, particular exposure situations, like end-of-row houses). Some people like it considerably warmer than the assumed 20°C, while others like it a bit cooler. However, averaged out, there typically is good agreement with the modelling for the estate. It’s also clear from the monitoring that while some outliers can fail to meet the expected outcome by a large margin – even if they are consuming considerably more energy than the mean, the consumption is still significantly below energy requirements of a typical houses, while maintaining better indoor environmental quality.