Optimal zoning for building zonal model of large-scale indoor space

Zonal model is a powerful numerical modeling approach that can quickly simulate thermal parameter distribution by dividing the indoor space into zones (zoning). However, most previous zoning methods are based on modelers’ experiences or empirical knowledge. Flow field details will be lost if the zoning is too coarse; while the computational load will be high if the zoning is too fine, especially for large-scale spaces. Therefore, this paper proposes an optimal zoning method for large-scale spaces to compromise the model accuracy and computational efficiency. In this method, air velocity and temperature differences before and after the zoning were used as indicators for accuracy. An improved particle swarm optimization method was adopted to obtain optimal Pareto solutions; and the technique for order preference by similarity to an ideal solution (TOPSIS) method was implemented to minimize the number of zones for enhancing computational efficiency. A large high-speed railway station was used as an example to verify the proposed optimal zoning method. The accuracy and computational efficiency of the proposed non-uniform zoning method were compared with a conventional zoning method and a uniform zoning method. Compared with the conventional zoning method, the proposed method reduced the number of zones by 96.2%; while critical temperature distributions and airflow patterns were well preserved. Compared with the uniform zoning method with the same number of zones, the proposed method improved the accuracy by 45.2% at most and 18.5% on average. Applicability analysis was also carried out to show that the proposed zoning method was reliable. [Display omitted] •A novel optimal zoning method for the large-scale space is presented.•Accuracy of optimal zoning results was ensured by improved PSO method.•Used Pareto front and TOPSIS analysis to select optimal zoning results considering efficiency.•This method is effective under different application scenarios for the zonal model.