Deriving U.s. Household Energy Consumption Profiles from American Time Use Survey Data – a Bootstrap Approach

This paper presents a novel approach to derive U.S. residential building energy load profiles. This approach uses bootstrap sampling method to extract daily activity pattern of occupants of a household from American Time Use Survey (ATUS) data. The characteristics of ATUS data, the relation between time-use and load-demand, and the robustness of this approach are discussed. Virtual experiments were conducted on Energy Plus platform to study the patterns of annual load demand distribution under different household composition and thermal zoning schemes. Simulations of average 24-hr appliance and lighting load profiles were also conducted. The simulated load profiles and those from utility metering studies have good agreement. This novel approach has versatile applications in residential building energy simulation. INTRODUCTION A National Time Use Survey (TUS) is a large scale time use survey administrated by a national government. Each TUS record contains 24 hour period of activities of an individual with this individual’s personal information. TUS records are taken from all walks of life. Scholars generally agree (Robinson and Godbey, 1997) that TUS data are the best available data that represent the time use pattern of a society. In recent years, researchers started to explore the application of national Time Use Survey (TUS) data for simulating schedules in residential building energy consumption calculation. The roulette wheel genetic algorithm (Tanimoto et al., 2008) and Markov Chain Monte Carlo (MCMC) techniques (Richardson et al., 2008) have been applied to TUS data for occupant and load schedule simulation with some success. Two main drawbacks of these approaches come from the methodological constraints that limit the extraction of detailed information embedded in the TUS record and the lack of integration between simulated schedules and commonly used building energy simulation tools. In general, three steps of data transformation are needed for using TUS data for residential building load demand estimation. The first step is to construct a household’s daily activity schedule from TUS data, one-to-many mapping is the key characteristic of this process. i.e., to represent the range of variation of a given household’s daily activity patterns, multiple schedules are simulated from TUS data. The second step is to derive the internal heat gain, lighting and appliance load schedules from a household’s activity schedules. This step involves the interpretation of the spatial and temporal distribution of the occupants’ activities and the corresponding appliance and energy use. The third and final step is to derive heating and cooling load demands from the combined inputs of the TUS derived occupancy, appliance and lighting load schedules, the configuration of the residence and the outdoor environmental conditions. This paper presents a novel approach to simulate occupancy and load schedules from the TUS data in finer details. In constructing the household’s daily activity schedule, the bootstrap method (DeGroot and Schervish, 2002) replaces roulette wheel and MCMC techniques for an individual’s daily activity schedule simulation. Then the individuals’ household demography profiles are matched for household schedule assembly. In appliance load schedule simulation, both spatial and temporal dimensions of the occupants’ activities are referred. Humanphysical integrative household system theory (Hitchcock, 1993) is used to explain the association between the occupants’ activities and appliance load demand. In the calculation of heating and cooling load demands, Energy Plus simulation replaces the self-developed energy consumption estimation method, so the approach will be easier to propagate. Figure 1 illustrates the procedure of the approach. Using 2006 American Time Use Survey (ATUS) data, a series of virtual experiments are conducted to observe the pattern of annual load demand resulted from this novel approach. Results from virtual experiments indicate: 1. patterns of annual energy load demand distribution derived from different batches of randomly sampled ATUS data are highly consistent within each type of household demography, 2.increasing the number of thermal zones has far more significant impact in heating and cooling load demand reduction than increasing building envelope thermal insulation does, 3, the simulated 24-hr appliance and lighting load profiles agree with those from utility metering data. Eleventh International IBPSA Conference Glasgow, Scotland July 27-30, 2009