Monte Carlo housing stock model to predict the energy performance indicators

This study presents a new physics-based model of housing stock energy using Monte Carlo, where inputs are probability distribution functions originated from Energy Performance Certification (EPC) Portuguese database. The overall performance of the model in predicting the energy indicators used in EPC is extremely satisfactory, considering that the inputs required to run the calculations are not always available. The model outputs are validated against EPC data with residual sum of squares (RSS) below 2×10−3, except for cooling energy benchmark with RSS below 4×10−2. The main output of EPC, the distribution among classes, is successfully reproduced by the model, with differences in the number of occurrences below 3.1%. The developed model constitutes a tool that helps on further research on energy policies, namely, studying the impact evaluation of more restrictive thermal quality requirements; evaluating other methodological approaches to calculate energy indicators; analysing policies of building elements retrofitting and bottom-up estimations of housing stock energy consumption.