The aggregation of spherical nanoparticles as they propagate through porous media is explored using lattice Boltzmann simulations and tracking of the trajectoties of individual particles. The porous media are modeled as periodic arrays of spheres in different packing configurations. The effects of interparticle interactions on particle aggregation are treated through a single aggregation probability upon particle collision. Fast aggregation is represented by an aggregation probability with values close to one, and slow aggregation is represented by lower aggregation probability. An algorithm that accounts for the different time scales between hydrodynamics and Brownian motion is devised and validated. It is found that primary particle size, initial particle concentration, injection flow rate, and aggregation probability significandy impact the dynamics of the aggregation. Calculations of the transient mean size of the aggregates show that the aggregation rate is high near the entrance of the porous medium, yet dramatically decreases when moving farther downstream. The reason is that the growth of larger aggregates slows the aggregation process. It is also found that the bigger aggregates are formed in the pore space near the pore matrix surface, where particle residence time is long enough to allow the aggregates to grow. An empirical correlation based on measurable parameters is proposed for the prediction of the mean aggregate size. (C) 2018 Elsevier Ltd. All rights reserved.