Heat distribution coefficient and cooling load prediction of indoor spaces with attachment ventilation

• A new cooling load prediction method is proposed for attachment-ventilated indoor spaces. • The influence of key parameters on heat distribution and cooling load is analyzed systematically. • A parametric correlation between heat distribution coefficient and Archimedes number is established. • The proposed method achieves higher prediction accuracy than the data-driven regression method and ASHRAE method. Accurate cooling load calculation is essential for air conditioning system design, and is highly influenced by air distribution methods. Attachment ventilation, which targets a specific conditioned zone, creates unique airflow and heat distribution that are not adequately addressed by existing cooling load prediction methods. This study investigated indoor heat distribution characteristics under attachment ventilation and proposed a novel cooling load prediction method based on the heat distribution coefficient. Several influencing parameters were analyzed, including the Archimedes number ( Ar ), dimensionless heat source projection area, intensity and location, and room height. Results show that Ar is the dominant factor, representing the balance between supply air momentum and thermal buoyancy, which significantly affects the heat distribution coefficient ( m ). As Ar increases, m generally decreases as a result of enhanced thermal stratification. Given the identified Ar on m , a parametric correlation between m and Ar was developed . The effects of above parameters were further analyzed in terms of cooling load performance. For office spaces, attachment ventilation can reduce cooling load by 26 % compared to total-room load, with reduction rates ( η c ) often exceeding 20 % when Ar > 0.03, emphasizing the benefits of buoyancy-driven airflow. Based on these findings, a cooling load prediction model for attachment-ventilated spaces was developed and validated. The proposed method outperforms both the data-driven regression and ASHRAE methods. Notably, less than 5 % relative error was observed in 91 % of the predictions, owing to its ability to account for the interaction between wall-attached supply air momentum and heat-source-induced buoyancy.