570
Views
25
Downloads
30
Crossref
N/A
WoS
31
Scopus
0
CSCD
A poor depiction of occupant behavior in building performance simulation frequently results in substantial divergences between real and simulated results. The problem may be of particular concern with simulation supporting the renovation of older multi-unit residential buildings, buildings whose occupants use windows for temperature control even during heating season. Here, we investigated the impact of window operation models (as well as other occupant behaviors) on simulated energy performance in university residence halls. Based on environmental monitoring, along with information collected from occupant surveys and wearable devices, we estimated air exchange rates and developed a probabilistic window-operation prediction model. The data were collected in 76 dormitory rooms sampled from a pre-renovated historic building and two similar buildings. We then evaluated the window-operation model’s predictive performance in 15 dormitory rooms in the post-renovated building with new occupants. The results of our predictive model were also compared with previously reported window-operation models. We implemented each window-operation model in a calibrated EnergyPlus building performance model, comparing the results of each simulation to metered hourly steam consumption. The impact of the different window operation models on simulated heating energy use was significant (annual error ranging from 0.2% to 10%). Our model demonstrated the highest capability of predicting window state (accuracy=85.8%) and steam use (NMBE=-0.2%); however, some previously published window-operation models also produced satisfactory performance, implying that such models may be generalizable to some extent. The results suggest that data collected from somewhat ubiquitous indoor environmental quality sensors can glean insights into occupant behavior for building performance simulation. Furthermore, the energy impacts resulting from the variations in occupant behavior studied here were large enough to show that the choice of behavior model can have meaningful implications for real-world applications, such as estimating saving from heating and lighting system upgrades.
A poor depiction of occupant behavior in building performance simulation frequently results in substantial divergences between real and simulated results. The problem may be of particular concern with simulation supporting the renovation of older multi-unit residential buildings, buildings whose occupants use windows for temperature control even during heating season. Here, we investigated the impact of window operation models (as well as other occupant behaviors) on simulated energy performance in university residence halls. Based on environmental monitoring, along with information collected from occupant surveys and wearable devices, we estimated air exchange rates and developed a probabilistic window-operation prediction model. The data were collected in 76 dormitory rooms sampled from a pre-renovated historic building and two similar buildings. We then evaluated the window-operation model’s predictive performance in 15 dormitory rooms in the post-renovated building with new occupants. The results of our predictive model were also compared with previously reported window-operation models. We implemented each window-operation model in a calibrated EnergyPlus building performance model, comparing the results of each simulation to metered hourly steam consumption. The impact of the different window operation models on simulated heating energy use was significant (annual error ranging from 0.2% to 10%). Our model demonstrated the highest capability of predicting window state (accuracy=85.8%) and steam use (NMBE=-0.2%); however, some previously published window-operation models also produced satisfactory performance, implying that such models may be generalizable to some extent. The results suggest that data collected from somewhat ubiquitous indoor environmental quality sensors can glean insights into occupant behavior for building performance simulation. Furthermore, the energy impacts resulting from the variations in occupant behavior studied here were large enough to show that the choice of behavior model can have meaningful implications for real-world applications, such as estimating saving from heating and lighting system upgrades.
This research was partially funded by the Harvard University Climate Change Solutions Fund. Data collection and analysis is sponsored by the EFRI-1038264 award from the National Science Foundation (NSF), Division of Emerging Frontiers in Research and Innovation (EFRI). Dr. Cedeno Laurent was partially supported by the Fundacion Mexico en Harvard and the Mexican Science and Technology Council (CONACYT). Thank you to the fellow participants of IEA-EBC Annex 66 (Definition and Simulation of Occupant Behavior in Buildings) for multiple helpful discussions and Farhang Tahmasebi for advice on EMS coding.