Abstract
We present an optimization-based methodology for energy simulation models calibration with hourly monitoring data applied to residential buildings. Goal is obtaining reliable model parameters suitable to estimate energy demand savings after the building retrofit. Since there exists more than one parameter combination matching the actual energy consumption, uncertainty is studied by choosing a number of calibrated models. We apply regularization to account for source reliability of the model parameters by penalizing those parameters with low probability. Calibration criteria with respect to indoor air temperature and cumulative energy demand is added in addition to the standard ones for cross-checking the reliability of obtained models. Occupant interaction with HVAC system is analysed based on monitoring data and interviews to generate stochastic yearly profiles and simulation input files for calibration. Method is illustrated on a case study multi-family residential building, located in Madrid, Spain.