The EPI normalizes energy use relative to building area and
the AAhEPI normalizes building area as well as operational hours.
Sharp [20] used a stepwise regression on building data from the
CBECS 1992 database to identify the strongest determinants of
office energy use intensities from 75 variables. He reported that
the two dominant variables (most important) correspond to the
logarithm of the number of workers per square foot and the category
describing the number of personal computers in the building.
These are followed by the number of operating hours and whether
the building is owner-occupied. The remaining variables of less
importance were the presence of an economizer and a chiller.
Chung and Hui [16] used 14 factors in their benchmarking
study of office buildings in Hong Kong. They categorise these into
age, occupancy factors (floor area; operation schedule; number
of employees), climate factors (degree-day temperatures), people
factors (occupant behaviour and maintenance, temperature set
points) and energy end-use factors (chiller equipment type; air side
distribution type; air side control; water side distribution control;
lighting equipment; lighting control; office equipment).
In their study of benchmarking energy consumption of supermarkets
in Hong Kong, Chung et al. [17] used similar factors—age,
occupancy factors (internal floor area, operational schedule, number
of customers per year), people factors (occupant behaviour and
maintenance, and indoor temperature set point) and energy system
(chiller type, lighting equipment and lighting control). Lee [19] also
identifies occupant density as a significant predictor of energy use
intensity in a study of 47 government office buildings in Taiwan.