A direct model, using the explicit geometry of stacked products in boxes, was developed and used to study the local and average airflow through stacks of horticultural products. The discrete element method was employed to generate a random stacking of spherical products in the box. A computational fluid dynamics model was then applied to study explicitly the airflow through the air gaps in the box and in the voids between the stacks of different random fillings. The flow resistance was affected by the confinement ratio, product size, porosity, box vent hole ratio, and much less by the random filling. The predicted pressure drop over stacks agreed with experimental correlations for porous media. Air velocity profiles inside the boxes compared well to measurements. The methodology was used to obtain more accurate pressure drop correlation for stacks of vented boxes that can now be used in large scale simulations of cool rooms. (c) 2008 Elsevier Ltd. All rights reserved.