• CH-24-C106 - How to Represent Occupant Behavior Diversity in Building Performance Simulations? Comparison of Occupant Behavior Profiles Selection Methods and their Applications

CH-24-C106 - How to Represent Occupant Behavior Diversity in Building Performance Simulations? Comparison of Occupant Behavior Profiles Selection Methods and their Applications

ASHRAE , 2024

Publisher: ASHRAE

File Format: PDF

$8.00$16.00


When it is time to compare different building design options, building performance simulations (BPS) usually refer to single schedules to represent occupant behaviors. However, a gap between prior-to-construction estimations and the real energy consumption of buildings often remains because of this oversimplification of occupant behavior (OB). The disparities of energy consumption and comfort between similar buildings, and even dwellings illustrate the need for occupant behavior diversity assumption in BPS. To that purpose, probabilistic occupant behavior models have been developed for many actions such as occupancy, electricity, domestic hot water consumption, etc. As a result, single value of energy consumption for a particular building can be replaced by probability distributions that can be generated through Monte Carlo simulations with different OB profiles. Methods are thus needed to select the appropriate number of OB profiles to reduce the computational cost of BPS and reach adequate energy consumption distributions when assessing for occupant behavior diversity. This paper aims to compare different sorting and clustering methods for the selection of OB profiles for BPS. As a proof of concept, a case-study building in Quebec City, Canada, for which extensive monitoring data is available, is used to calibrate an occupant behavior model and generate thousands of OB profiles. Energy and comfort performance for each profile is assessed through simulation, allowing obtaining detailed distributions. Then, the selection of a limited number of profiles to represent the overall diversity is explored. Well-known clustering algorithms, such as k-mean clustering, will be compared to another selection algorithm based on the finding the most representative set of OB profiles. Different performance metrics are used to compare the selection methods.

More ASHRAE Standards PDF

ASCE Manual of Practice No. 101
ASCE 26-97

ASCE 26-97

$29.00 $59.00

ASCE 34-01

ASCE 34-01

$39.00 $79.00

ASCE 33-01

ASCE 33-01

$39.00 $79.00