Road crossings and culverts may adversely impact the stream network connectivity and fish habitats of the catchment. Research into the passage of small‐body‐mass fish in pipe culverts has been relatively limited, compared to the literature for box culverts. For small‐bodied and juvenile fish species, the excessive barrel velocities are often a major hindrance, because of their weak swimming capabilities. In the present study, some physical testing of low‐velocity‐zones was undertaken in standard pipe culvert. The physical modelling was conducted under controlled flow conditions to test comparatively three designs, aiming to minimise the change in energy losses and to maximise low‐velocity zones and secondary circulation conducive to small‐body‐mass fish passage. In the whole pipe culvert experiment (Model 1), both baffle and longitudinal rail boundary treatments provided low‐velocity zones. The baffles however induced a strongly turbulent flow, associated with substantially larger energy dissipation than the reference smooth boundary pipe culvert. The longitudinal rail boundary treatment produced energy losses comparable to the smooth boundary reference configuration. Both boundary treatments were tested comparatively at near‐full‐scale (Model 2) to quantify the low‐velocity‐zone (LVZ) characteristics. The small longitudinal rail (0.06 m × 0.02 m), installed at 30° from the centreline, induced some flow asymmetry, as well as some low‐velocity‐zones on both sides of the rail. Some strong secondary motion was further observed as the combined effect of the flow asymmetry and singularities of the rail corners. The secondary motion structure was markedly different, and the distributions of the normal turbulent stresses (vz′2 − vy′2) showed key differences between the two boundary treatments, with the sharp corners of the rail contributing to the generation of secondary motion and in turn slow‐velocity regions facilitating the upstream passage of small‐body‐mass fish species and juveniles of larger fish.