Occupant behaviour diagram

+ Stochastic modelling would account for the randomness of human behaviour and enable more accurate predictions of buildings’ energy use.

Despite advances in building design, reducing energy use still relies on how occupants behave. Yet predicting their behaviour is something the industry struggles with and it’s why the actual operating measurement of energy use is often very different from the design value.

It seems people don’t always behave as designers expect, or even in the same way as each other. This is why I believe we need to adopt stochastic energy model simulations, which take account of the randomness of human nature. Without this, we’re likely to fall short of our aspirations for low-carbon buildings. 

Because it’s the occupants who control the operation of a building, they ultimately determine how much energy it uses. So advances in building design and technology will only take us so far if we don’t develop a better understanding of occupant behaviour.

The Florida Solar Energy Center showed that identical houses can vary in their energy use by as much as 300%. If people choose to turn off lights they don’t need, appliances that aren’t in use and air-conditioning or heating that isn’t required then energy consumption is likely to be lower than predicted.

Occupants can also control their own comfort by raising and lowering blinds, opening and closing windows, using their own task lights or fans or even by dressing differently. Even the times that people arrive and leave can affect energy use, because the number of occupants affects the thermal load of an occupied space. 

So any model needs to take account of the fact that people behave individually. In technical terms, it needs to be stochastic rather than deterministic. It needs to use agent-based modelling of individuals to study the effect of a variation in preferences and behaviour.

Of course, reliable data is a crucial prerequisite for the model to be effective. How likely are different individuals to adjust blinds under different glare conditions? If someone is shown their real-time energy use, what is the probability that they will reduce their energy consumption?

With data like this, you could model all sorts of different permutations. You could look at the energy savings from using individual task lights in an office, for example. Or you could investigate shoppers’ decisions to take stairs instead of escalators or lifts in variable indoor climate conditions. 

Stochastic modelling is the logical next step in energy modelling, and a necessary step to provide the foundation for more advanced modelling. To help make it a reality, experts from Arup are contributing to the Annex 66 project of the International Energy Agency’s Energy in Buildings and Community Programme.

As the industry and society shifts towards the big data paradigm, carefully considering and analysing occupant behaviour will prove invaluable in reducing the building industry’s carbon footprint. ?