Urban building energy modeling (UBEM) is arousing interest in building energy modeling, which requires a large building dataset as an input. Building use is a critical parameter to infer archetype buildings for UBEM. This paper presented a case study to determine building use for city-scale buildings by integrating the Geographic Information System (GIS) based point-of-interest (POI) and community boundary datasets. A total of 68,966 building footprints, 281,767 POI data, and 3367 community boundaries were collected for Changsha, China. The primary building use was determined when a building was inside a community boundary (i.e., hospital or residential boundary) or the building contained POI data with main attributes (i.e., hotel or office building). Clustering analysis was used to divide buildings into sub-types for better energy performance evaluation. The method successfully identified building uses for 47,428 buildings among 68,966 building footprints, including 34,401 residential buildings, 1039 office buildings, 141 shopping malls, and 932 hotels. A validation process was carried out for 7895 buildings in the downtown area, which showed an overall accuracy rate of 86%. A UBEM case study for 243 office buildings in the downtown area was developed with the information identified from the POI and community boundary datasets. The proposed building use determination method can be easily applied to other cities. We will integrate the historical aerial imagery to determine the year of construction for a large scale of buildings in the future.

Commercial prototype building models were developed by the United States Department of Energy (DOE) to analyze the energy savings of the ASHRAE 90.1 standard. However, in the DOE models, occupant behavior inputs were deterministic and the stochasticity of occupant behavior was not fully characterized. This study evaluated the impacts of stochastic occupant behavior on building energy consumption and energy savings analysis from upgrading the ASHRAE 90.1-2016 to 2019 in sixteen climate zones in the United States (U.S.). Three occupant behavior styles (austerity, normal, and wasteful) were defined to represent the different levels of energy consciousness. The DOE medium office prototype models were used as the baseline (normal behavior style). The occupant behavior functional mock-up unit (obFMU) was used to model the stochastic occupant behavior models (austerity and wasteful). The EnergyPlus medium office prototype models were modified to co-simulate with the obFMU models. The results of 16 climate zones were aggregated by the relative construction volume of each climate zone. The results showed that the weighted national annual source energy use intensity (EUI) of the austerity, normal, and wasteful behavior styles were 203.81 kWh/m², 287.15 kWh/m², and 388.33 kWh/m² for ASHRAE 90.1-2016, and 192.43 kWh/m², 273.48 kWh/m², and 371.28 kWh/m² for ASHRAE 90.1-2019, respectively. Compared to the normal behavior style, the austerity behavior style consumed 29% less source energy, while the wasteful behavior style consumed 36% more source energy. From upgrading the ASHRAE 90.1-2016 to 2019, the energy saving percentages based on the austerity, normal, and wasteful behavior styles were 5.59%, 4.78%, and 4.42%, respectively. The stochastic occupant behavior significantly impacted the building energy consumption, and their impacts on the energy savings analysis of upgrading ASHRAE 90.1 were also not negligible.

Building performance simulation can be used for retrofit analysis. However, it is time-consuming to create building energy models for existing buildings. This paper presented and implemented a rapid building energy modeling method for existing buildings by using prototype models and automatic model calibration for retrofit analysis with uncertainty. A shopping mall building located in Changsha, China, was selected as a case study to demonstrate the rapid modeling method. First, a toolkit named AutoBPS-Param was developed to generate building energy models with parameterized geometry data. A baseline EnergyPlus model was generated based on the building’s basic information, including vintage, climate zone, total floor area, and percentage of each function type. Next, Monte Carlo sampling was applied to generate 1000 combinations for fourteen parameters. One thousand EnergyPlus models were created by modifying the baseline model with each parameter combination. Moreover, the 1000 simulation results were compared with the measured monthly electricity and natural gas usage data to find 29 calibrated solutions. Finally, the 29 calibrated energy models were used to evaluate the energy-saving potential of three energy conservation measures with uncertainty. The retrofit analysis results indicated that the electrical energy saving percentage of chiller replacement ranged from 1.57% to 13.51%, with an average of 8.27%. The energy-saving rate of lighting system replacement ranged from 1.92% to 11.66%, with an average of 6.43%. The energy-saving rate of window replacement ranges from 0.31% to 1.81%, with an average of 0.55%. The results showed that AutoBPS-Param could rapidly create building energy models for existing buildings and can be used for retrofit analysis after model calibration.