Keith, David Middlebrook (M.S., Civil Engineering)

Thesis directed by Assistant Professor Moncef Krarti

Economic evaluation of building controls requires data on effects on energy use, demand and heating/cooling requirements. For occupancy-sensing lighting controls, economic models should incorporate both average and peak occupancy information, to determine accurately electrical energy savings, electrical demand reduction and changes in heating/cooling energy use.

This thesis describes the data acquisition, organization, analysis, and application of peak occupancy data to DOE-2 building models. A simplified tool for predicting peak occupancy rates based on the number of rooms specified and the associated average occupancy is described and presented.

Twelve months of time-labeled occupancy data were obtained for 99 office and office service rooms, taken every five minutes using the facilities management computer at the Foothills Laboratory of the National Center for Atmospheric Research (NCAR) in Boulder, CO. DOE-2 models which evaluate the effects in private offices of occupancy-sensing controls of lighting show that using peak occupancy data significantly changes the predicted savings. The DOE2 models, for one typical building in four locations, show the difference between using combined peak and average occupancy schedules and using only average occupancy schedules. While energy consumption changes slightly, electrical demand and heating/cooling requirements vary substantially. The sensitivity of DOE2 simulations to the peak schedule is investigated, by comparing the results when each day of the week is the occasion of the peak occupancy. The use of peak occupancy schedules results in temporal and magnitude changes in the peak electrical demand.

Subsets were defined within the 99 room set by two different methods, then used to calculate data triads of set size, average occupancy and peak occupancy values. From this subset data, a prediction tool is developed, with set size and average occupancy as inputs, and monthly peak occupancy for daily and hourly periods as output. The predictions are compared with data from NCAR, with very good agreement for daily and annual averages, and with data from an Energy Center of Wisconsin research project, with close agreement.

The use of peak occupancy data in predicting the potential savings when applying occupancy-sensing lighting controls is promising.