Prediction uncertainty of linear building energy use models with autocorrelated residuals

Autocorrelated residuals from regression models of building energy use pres ent problems when attempting to estimate retrofit energy savings and the un certainty of the savings. This paper discusses the causes of autocorrelatio n in energy use models and proposes a method to deal with autocorrelation. A hybrid of ordinary least squares (OLS) and autoregressive (AR) models is developed to accurately predict energy use and give reasonable uncertainty estimates. Only linear models are considered because both the data and the physical theory for many commercial buildings support this choice (Kissock, 1993). A procedure for model selection is presented and tested on data fro m three commercial buildings participating in the Texas LoanSTAR program. I n every case examined the hybrid OLS-IIR model provided the best estimate o f energy use and the most robust estimate of uncertainty.