This investigation is a synergistic combination of laboratory experimentation and numerical simulation to quantify the practical impact of geometric imperfections in the flow channels of a reverse osmosis (RO) system. To this end, carefully executed experiments are performed to quantify the fluid flow in a system containing feed spacers which are embedded in the RO membrane. In a complementary activity, numerical simulations were performed both for an ideal geometric situation (without embedments) and the actual geometric configuration including the embedments. It was found that the presence of unaccounted embedments affected the pressure drop predictions for the system by 14-19%. When account was taken of the embedments, the simulation results were found to be virtually coincident with the experimental results. This outcome suggests that deviations between experimental and simulation results encountered in the literature might well have been due to geometrical deviations of the type investigated here. The numerical simulation of the feedwater fluid flow was based on the often-used but unverified assumption that the velocity field experiences the geometric periodicity of the feed spacer. This assumption was lent support by results from a non-periodic simulation model and by the excellent agreement between the numerically based predictions and the experimental data.