The impact of temperature gradient and JPCP design features on surface roughness and curvature

Surface roughness and deflections in jointed plain concrete pavements are strongly influenced by temperature-induced slab curling and slab–subgrade interaction. This study investigates the effect of slab curvature on ride quality using the equivalent linear thermal gradient concept and a combined framework of Westergaard analytical solutions, ISLAB finite element method simulations, and artificial neural network surrogate modeling. Slab shape backcalculation from pavement profiles using Westergaard theory is systematically compared with FEM-based predictions. Results show that the Westergaard solution underestimates slab deflections and International Roughness Index while overestimating the effective temperature gradient required to match field-measured curvature, primarily due to simplified boundary assumptions that neglect joint discontinuities, finite slab geometry, and slab lift-off under large thermal gradients. FEM-based post-processing captures these additional mechanisms and yields convincing predictions of slab curvature and roughness under thermally critical conditions. An ANN surrogate trained on FEM-generated data (R² > 0.99) predicts wheel-path slab shapes and enables rapid evaluation of large design spaces. Sensitivity analysis indicates that joint-moment transfer and slab thickness govern curling-induced roughness, while joint spacing and subgrade stiffness exert a secondary influence. The proposed framework quantifies slab curling relative to diurnal IRI variability and supports the selection of JPCP design parameters to improve ride quality.