How long is long enough? toward an optimal year span for typical meteorological year selection in a changing climate

A typical meteorological year (TMY) is a fundamental boundary condition for building performance simulations (BPSs), and its representativeness governs both the reliability of simulation results and the credibility of design decisions. In the context of global warming, conventional long-span (30-year) TMYs risk blending outdated conditions with recent trends, and cannot reflect warming trends. This study adopted the ERA5 reanalysis from 1980 to 2024 as a database and evaluated seven historical year spans (1, 5, 10, 15, 20, 25, and 30 years) based on the BPS of various prototype buildings across 20 cities in China. TMYs were generated from multiple periods and validated against subsequent decades of actual meteorological years, with the performance measured by the deviation in annual cumulative heating and cooling loads. The analysis showed that the local climatic variation, rather than the climate zone or building type, was the primary factor shaping the optimal span. Shorter spans consistently improve representativeness compared with the traditional 30-year baseline, and a span of approximately 10 years achieves the most robust balance between bias and variability. These findings provide evidence for revising the current practices and support the use of a 10-year TMY in China to deliver more reliable inputs for building energy assessments, efficient designs, and long-term planning.