1.0 Introd u c tio n
The U.S. Residential Lighting End-Use Consumption Study aimed to develop reliable estimates of
residential lamp usage and energy consumption at both national and regional levels. Multiple approaches
for pursuing this goal were investigated, exploring tradeoffs in accuracy, flexibility, and required time –
or associated cost. The chosen methodology prioritized flexibility, meaning here the ease of incorporating
new data that might become available in the future. As a result, this effort is best described as the
application of lamp hours-of-use (HOU) models to a newly developed regional estimation framework that
represents the U.S. housing stock. The estimation framework is simply a constructed set of sample
housing units. Each sample housing unit in the estimation framework is described by its household
characteristics (including both housing unit and occupant demographic data) and lighting inventory. This
estimation framework is capable of producing regional and national estimates of lighting usage and
energy consumption for the entire United States, and incorporating new regional data (that meet defined
pre-conditions) for calibrating the HOU models to improve estimation accuracy. This report describes the
development of the estimation framework and the application of the HOU models to the framework,
presents a limited set of lighting estimates, and discusses the accuracy and validity of the presented
estimates. A companion Microsoft Excel spreadsheet contains the full set of estimates produced by this
study, including average number of fixtures, number of lamps, daily HOU per lamp by month, lamp
power, daily energy consumption, and annual energy consumption. The spreadsheet is organized to allow
the estimates to be easily filtered to various levels of aggregation, and by various household and lamp
characteristics and categorical cross-classifications.
Several national and regional studies that occurred between 2008 and 2010 and collected household
characteristics, lighting inventory profiles, and/or lighting end-use metering data were evaluated for use in
developing the HOU models and estimation framework. This study heavily leverages the recent
California Residential Lighting Metering Study (CA RLMS) and U.S. Energy Information Administration
(EIA) Residential Energy Consumption Survey (RECS) datasets. The estimation framework is rooted in
the 2009 RECS housing sample, and the analysis of covariance (ANCOVA) HOU models developed for
the 2008-2009 CA RLMS were used to estimate lighting usage for each lamp type (e.g., incandescent,
compact fluorescent light [CFL], or other type). These and other datasets used in this study are described
in more detail in Section 2.0. Section 3.0 explains the creation of the estimation framework, a combined
dataset containing all the input variables required by the ANCOVA HOU models, and the challenges in
constructing housing unit samples with household characteristics and lighting inventory data that are as
regionally specific as possible. The methods used to apply the models to the estimation framework,
generate lamp-level usage and energy consumption estimates, and aggregate those and other estimates to
various levels are also discussed here. Section 4.0 presents a limited set of lamp usage and related energy
consumption estimates. These estimates were selected to demonstrate the ability of the estimation
framework to generate estimates at regional levels of aggregation and with categorical cross-
classifications. Section 5.0 discusses how the standard error is calculated for all estimates and the data
quality flag in the companion spreadsheet. The section concludes with examples of how the lighting usage
model might be calibrated with end-use data collected in other regions, which represents the primary
objective of potential future updates to the U.S. Residential Lighting End-Use Consumption Study.
All estimates presented in this study are bottom-up, in that they are derived from the lamp and fixture
level within rooms of a housing unit sample, up to the desired level of analysis. Energy consumption is
computed (as the product of lamp power and lamp HOU) at the lowest level and then aggregated up using
1.1