A dynamic prediction model for the cooling and heating load in a non-enclosed atrium

Non-enclosed atriums, directly connected to surrounding zones, facilitate uneven airflow and heat transfer—phenomena that can induce significant additional cooling or heating loads, yet remain unaccounted for in current engineering practice. This study proposes a dynamic prediction model to quantify such loads by integrating the established temperature distribution model with the state-space method. The former determines the quasi-steady vertical air temperature distribution in the atrium and adjacent zones under specific thermal boundary conditions, while the latter addresses building thermal processes. This integrated model accurately describes the thermal behaviour of indoor non-uniform temperature fields, enabling dynamic prediction of cooling and heating loads for non-enclosed atriums. Applying this model, we investigated the typical seasonal characteristics of these loads and quantitatively analysed how design parameters influence their intensity and distribution. Key findings reveal that atriums with greater net height and smaller cross-sections amplify extra loads on the top floor in summer and the bottom floor in winter, respectively. Notably, in winter, for a given atrium volume, a lower section aspect ratio effectively mitigates uneven heating load distribution between floors. This study demonstrates the practical value of the dynamic prediction model in engineering and provides a calculation tool to improve energy consumption prediction accuracy for similar non-enclosed spaces.