Since invented in 2003, rock-filled concrete (RFC) has gained much attention and has been successfully applied in more and more civil and hydraulic projects in China. This study developed a numerical framework to simulate self-compacting concrete (SCC) flows in the voids among rocks of RFC, which couples the lattice Boltzmann method and discrete element method (DEM). The multiple-relaxation-time scheme is used to simulate self-compacting mortar (SCM) for better stability while the motion of coarse aggregates in SCC is simulated with DEM. The immersed moving boundary method is incorporated to deal with the interactions between coarse aggregates and SCM. After validation, the coupled framework is applied to study SCC flows in a single channel and in porous media with multi-channels. A passing factor was proposed and calculated to describe quantitatively the passing ability of SCC through a single channel. The study found that jamming of SCC occurs when the ratio of the gap width to particle diameter is smaller than 2.0 and the jamming risk increases with solid particles fraction while the passing ability has a weak relation with the pressure gradient. Further, SCC flow is self-tuning in porous media with multi-channels and it is prone to go through larger or wider gaps. Exceeded existence of narrow gaps will significantly increase the jamming risk.