Calibration for stochastic existing building stock model for energy simulation

Abstract

Building energy model (BEM) is useful in the retrofit stage of buildings because they allow flexible evaluation of various technology combinations or new technologies. However, the assumptions and uncertainties of the input variables produce a difference between the calculated energy and the measured energy. Old buildings or building stocks are more difficult to develop due to more assumptions and uncertainties. This paper describes a stochastic approach and a calibration method which can account different sources of uncertainty to develop an BEMs that can show energy consumption characteristics of existing building stock in Korea. First, described the process of deriving probability distribution and performing simulation using a database of nationwide building and energy information in Korea. Next, Inverse modelling was proposed to obtain parameters which is an auxiliary variable that describe the relationship between building performance and energy usage. The parameters both the outputs to the performance of the building or system and the input variables that determine energy usage at the same time. Therefore, using the parameters as information for calibration can improve the accuracy of the calibrated model, and identify the cause of inconsistencies with energy usage by ability to compare and calibrate the performance of a building or system in detail. Finally, calibration was performed, and the change of output was discussed.