A simultaneous calibration and parameter ranking method for building energy models

The existing stock of buildings is a major contributor to energy-related carbon emissions. Significant savings in building energy consumption can be derived through retrofit. Building retrofits are typically guided by analyses through building energy simulation models. Due to the complexity of the physical characteristics of building systems and the lack of field measured data, modellers very often have to work with unknown or unmeasurable parameters either through approximation or with reference to the original design values. Since the values of these parameters usually fail to accurately represent the current conditions of existing buildings, it is important to calibrate these parameters before applying them in a building energy simulation model. In addition, it is also important to rank the input parameters according to their influence on building energy performance when identifying priorities for building retrofit. In this paper, a metamodel-based Bayesian method is proposed to simultaneously calibrate and rank input parameters to building energy simulation models. This proposed method implements both a model calibration procedure and parameter ranking procedure simultaneously when performing an analysis, which is much more efficient than applying these two procedures individually in separate model runs. As a further contribution, we extend the proposed method to one capable of handling large datasets. A case study is developed to demonstrate the accuracy and efficiency of the proposed method. Findings from the case study show that the calibrated parameters are usually different from the initially assumed values. In the context of the chosen existing building in Singapore, most of the considered parameters are key factors influencing building energy performance with cooling plant COP being the most important factor and natural exfiltration rate being the least important factor.