A pore network model (PNM) of drying in a gravity-dominated macroporous material has been developed. The pore network geometry used for the simulations is extracted from microcomputed tomography scans of porous asphalt (PA), a macroporous, hydrophobic material. The drying of liquid water in PA is modeled using a cluster-based approach with a two-step drying process i.e. elements go first from being fully saturated to having liquid only in pore and throat corners, and then to becoming completely dry. The PNM simulations are validated with gravimetric experiments performed under controlled conditions and the simulations show good agreement with experiments for most of the drying period. From experiments, it is seen that drying in PA completely skips the constant drying rate period (CDRP) and instead begins with the decreasing drying rate period (DDRP) due to the poor hydraulic connectivity in PA as a result of its large and hydrophobic pore space. The PNM simulations exhibit CDRP initially and then transitions to DDRP after a third of the total drying time. The CDRP at the beginning of the PNM simulations is due to the simplified liquid configuration assumed in the network, and its duration can be minimized to an extent by increasing the number of hydrophobic pores in the network that does not retain any liquid after drainage. Although promising, the first results call for a more accurate representation of both the complex pore space of PA and the physics involved in drying of a macroporous material. (C) 2019 Elsevier Ltd. All rights reserved.