Hydraulic fracturing is an influential revolution in the global energy industry. One of the critical properties in hydraulic fracturing is the fracture conductivity, which measures the capability of a fracture to transmit fluids, and can be calculated as the product of fracture permeability and width. Fractures created in hydraulic fracturing usually have rough rock surfaces due to the influences of natural fractures and/or bedding planes. However, most existing laboratory works on fracture conductivity depended on using fractures between flat rock slabs that led to smooth fracture contact surfaces. In this research, we made fracture conductivity measurements in the laboratory using non-smooth-surface shale slabs. We also conducted numerical studies using a discrete element method/Lattice Boltzmann (DEM/LB) coupled model. We found the fracture conductivity values with ceramic proppant (2 lb/ft2) placed in the fracture space were about 3-8 times higher than those without proppant placement. For the fractures having proppant placement, the gas-measured fracture conductivity values were higher than the water-measured ones due to the Klinkenberg effect. Moreover, the decrease of fracture conductivity with increasing closure pressure was not noticeable when there was no proppant placed in the fracture space. Our experimental results demonstrated that it was preferable to have proppant placed in the fractures, even though the rock surface roughness provided some fracture conductivity via the “self-propping” mechanism. Besides, the effect of solid/liquid interactions, such as rock surface softening, cannot be simply neglected when the shale slabs were in contact with the injection liquid, which caused clay swelling in shale and proppant embedment into rock surfaces. The DEM/LB coupled numerical model was able to measure fracture conductivity as a function of closure pressure for both proppant-supported and “self-propping” fractures.