Climatic zoning is an important tool in building energy policy and regulations, smaller and more homogeneous climate zones could help enhance the rationality of energy policymaking. In most countries, climate zoning is based on simplified climate models, and it is difficult to accurately represent the energy demand characteristics of buildings. Therefore, this paper proposes a novel climate subdivision method that tackles the complex relations between climate and building energy-efficient, which move from traditional weather-based approaches to a performance-based approach. Integrate the uncertainty analysis method and the building dynamic performance simulation to extract a novel statistically significant index as the basis for classification. The K-means clustering method and discriminant analysis method are used to obtain and verify the clustering results, respectively. Furthermore, the comparison with clustering results based on degree days confirms the superiority of the new method proposed in this paper in identifying building energy demand.

The method is applied to the cold climate zone of China as a showcase. The results show significant differences in building energy demand and the order of key features affecting building performance between each region in the climatic zones. The zone can be further sub-divided into four groups considering the climate factor and building feature uncertainty, and clustering results have been verified by several evaluation indicators.