|
Abstract:
|
We wish to estimate the end-use energy consumption for U.S. residential space conditioning (i.e., air conditioning and space heating), which is markedly seasonal. Essentially all extant estimates can be traced back to the Residential Energy Consumption Survey (RECS), a nationwide survey conducted periodically by the U.S. Energy Information Administration. This is a tribute to the value of RECS but also an indication of how difficult the estimates are to validate externally. Here we pursue an idea for an "external-to-RECS" estimate of space conditioning consumption: we seek to synthesize monthly residential energy consumption data, which is the total consumption from all residential end uses, and monthly weather data, which should only relate to the space conditioning components. Though the idea is intuitive, the results are sensitive to decisions made throughout the process: for the weather data, we work with population-weighted degree days data, which are sensitive to the base temperatures used in their calculation; monthly consumption data is the product of reporting from utilities, which means the data likely contain unknown time lags. We use simple linear regression within a framework of brute-force optimization to find the "optimal" base temperatures and time lags to use, and hence yield our "optimal" estimates for the consumption due to space conditioning. In the present study, we restrict our analysis to cover only natural gas consumption and space heating estimation. We find the method produces estimates that are quite reasonable; we even find a surprising result implying that some historical RECS results may have been obtained in a potentially inconsistent manner.
|
