Building energy models of existing buildings are unreliable unless calibrated so they correlate well with actual energy use. Calibrating models is costly because it is currently an art that requires significant manual effort by an experienced and skilled professional. An automated methodology could significantly decrease this cost and facilitate greater adoption of energy simulation capabilities into the marketplace. The Autotune project is a novel methodology that leverages supercomputing, large databases of simulations, and machine learning to allow automatic calibration of simulations that match measured experimental data. This paper shares initial results from the automated methodology on commodity hardware applied to the calibration of building energy models (BEM) for EnergyPlus (E+) to provide error rates, as measured by the sum of absolute error, for matching monthly load and electrical data from a highly instrumented and automated ZEBRAlliance research home.
