UBEM's archetypes improvement via data-driven occupant-related schedules randomly distributed and their impact assessment

In Urban Building Energy Models (UBEMs), buildings are usually modelled via archetypes describing occupants' behaviour via fixed schedules. This research (i) creates data-driven schedules for electric use and occupancy from smart meter readings randomly distributed in the model to improve residential archetypes, (ii) assesses the impact of these schedules on UBEMs' energy results at different temporal resolutions and spatial scales. The novel assessment procedure exploits integrated heat maps based on coefficients of variation of the root means square error (CVRMSE). The outcomes show that differences in energy needs, with randomized schedules, range based on temporal and spatial aggregation. Yearly, for the entire neighbourhood, heating and cooling energy needs, and electric uses are estimated-2%, +1%, and +18% compared to the base case. The outputs show that, when simulations are focused on the entire district, fixed schedules can be enough to describe energy patterns. However, if the simulation is focused on small groups of buildings (e.g., 5 or fewer), randomising the schedules can create variability in the model in terms of electric use and occupancy among buildings characterized by the same archetype. The followed methodology can be exploited also with larger databases and eventually verified with also other types of data.

Automated summary

This research improves the Urban Building Energy Models (UBEMs) by using data-driven schedules created from smart meter readings. It assesses the impact of these schedules on energy results at different temporal resolutions and spatial scales. The results show that randomized schedules can significantly affect energy needs.