Live Load Distribution Factors in Horizontally Curved Composite Steel I-Girder Bridges: FEM Assessment of AASHTO LRFD Provisions Under HL-93 and Iraqi HB115 Military Loading

Abstract

The American Association of State Highway and Transportation Officials (AASHTO) Load and Resistance Factor Design (LRFD) live-load distribution-factor (DF) equations were calibrated on straight bridges, while their use for horizontally curved I-girder bridges is bounded by the Las/R < 0.06 rad criterion in Article 4.6.1.2.4b of the AASHTO LRFD Bridge Design Specifications, 10th ed. (2024). This study quantifies their accuracy beyond that limit using the finite element method (FEM) in 35 three-dimensional CSiBridge models subjected to numerical consistency checks: three composite plate-girder arrangements (4–6 girders, 9.0 m deck) at central angles of 0–15°, with near-limit, span-transfer, sensitivity, and out-of-range extensions to 25°, under the AASHTO LRFD vehicular design live-load model (HL-93) and the Iraqi Class 100 wheeled military vehicle (HB115; 1150 kN). At the limit, curvature amplification is only 1.8–2.6%. Beyond it, the exterior-moment equations become unconservative almost immediately; FEM demand exceeds AASHTO by 21–29% at 15°, whereas the interior-shear equations remain conservative. A two-part correction factor (CF) of the form CF = R0[1 + (a + a1S/L)(L/R)] is proposed (R2 ≈ 0.97) and predicts the withheld out-of-range cases within 3.3%. Within the tested envelope, exterior-girder amplification depends primarily on L/R; for HB115, its rate is about half that of HL-93. Direct CSiBridge reconstruction of two published 1/10-scale laboratory specimens shows good agreement in global deflection and moderate agreement in strain-based transverse distribution. Because full-scale measurements for the exact 38 m reference configuration were unavailable, this evidence is treated as external experimental benchmarking of the modeling methodology rather than complete validation of the full parametric matrix.