Spacer grids and mixing vanes exhibit a significant role in the thermal hydraulics of pressurized water reactors (PWRs), especially in the post loss-of-coolant accident regimes. A detailed analysis of the contrasting upstream and downstream turbulent flow features is of great importance to both system codes and computational fluid dynamics (CFD)–Reynolds-averaged Navier–Stokes (RANS) modeling. Further, with the advent of supercomputing resources and machine learning research, a data-driven approach to turbulence modeling is gaining popularity. However, owing to the complexities associated with large-scale, high-fidelity data collection capabilities, the application of machine learning–based turbulence models has been limited to simple geometries. In this work, using a highly scalable CFD code PHASTA, we have collected data from direct numerical simulations of a PWR subchannel with high spatial and temporal resolution. From the collected data we extract key turbulent flow features, including mean velocities and Reynolds stresses that highlight the effects of spacer grids and mixing vanes on downstream turbulence in a typical PWR subchannel. An invariant analysis of the anisotropic stress tensor is also presented, which further elucidates their effect on the nature of turbulence in the immediate upstream and downstream vicinity. The high-resolution data from the simulations are archived and intended for the development of data-driven RANS closure models that are capable of capturing the evolution of anisotropy in PWR subchannels.