The Reynolds stress model SSG/LRR- \(\omega \) is validated and analysed for turbulent boundary layer flows with mean-streamline curvature. The computational set-up for convex and concave curvature test cases in almost zero streamwise pressure gradient was designed for selected experiments in the literature. The geometries in the curved region were developed using a simple gradient-descent-based shape optimisation technique. Different RANS models were compared, i.e., linear eddy viscosity models such as SA/SA-RC and SST/SST-RC, and the SSG/LRR- \(\omega \) model. For the latter, different redistribution models (SSG, LRR and the blended SSG/LRR- \(\omega \) ) for the pressure-strain correlation in combination with different turbulent diffusion models (SGDH, GGDH) were investigated for convex and concave curvature test cases. For the convex curvature test case, the SSG/LRR- \(\omega \) gives the best agreement with the experimental data. For the case of concave curvature, SSG/LRR- \(\omega \) gives a better agreement with the experiment than other redistribution models, but open questions arise due to the presence of Taylor-Görtler vortices.