Large-span soil–steel bridges present distinct design challenges, primarily due to the high demands and deformations induced during backfilling. The Canadian Highway Bridge Design Code (CHBDC) currently mandates a 2% deformation-to-rise allowable limit during construction; however, this criterion fails to fully capture the complex relationship between deformation and strength limit states. To address this limitation, the present study proposes a systematic design framework for evaluating construction-stage structural performance of soil–steel bridges. The proposed framework was developed through extensive finite element simulations, conducting a comprehensive parametric analysis on bridges with spans ranging from 7 m to 32.4 m and varying rise-to-span ratios, backfill cover heights, and material properties. The resulting dataset was then used to develop Evolutionary Polynomial Regression models capable of predicting crown deformation, as well as moment–thrust and buckling utilization factors. Results indicate that the stage of maximum structural demand often occurs prior to construction completion, highlighting the importance of monitoring intermediate backfilling stages. For spans exceeding 25 m, crown deformations may surpass the 2% limit while remaining within strength and stability thresholds. Without additional stiffening measures, a 32.4 m span appears to define the practical upper bound for current deeper-corrugation profiles, though notable crown deformations are expected. These findings demonstrate that deformation-based criteria, when decoupled from strength performance, can be excessively conservative for large-span bridges. The proposed predictive equations provide a rational basis for performance-based design, construction monitoring, and inspection of these special structures. • Performance-based framework for large-span soil–steel bridges during construction. • Parametric study covers spans 7–32.4 m, rise-to-span ratios, cover heights, and backfill types. • Integration of finite element simulations with artificial intelligence models. • Predictive equations developed for crown deformation and stability indicators. • Provides practical tools for design, construction monitoring, and inspection.
Elsawwaf et al. (Tue,) studied this question.