Cracking in prestressed box girders is frequently traced to the inability of beam‐element models to resolve local stress concentrations, yet three‐dimensional solid‐element analysis remains uncommon in routine bridge design because of the prohibitive manual effort involved in model preparation. This paper removes that barrier by presenting an end‐to‐end automated workflow that couples parametric CAD tools (Grasshopper/Tekla) with finite‐element scripting in Abaqus, together with a rigorously validated user‐material subroutine (UMAT) for three‐dimensional concrete creep and shrinkage. The UMAT is first verified against the established commercial software Midas Civil through three independent benchmark tests—uniaxial compression, sustained bending, and incremental loading—with all discrepancies remaining below 4%. The complete framework is then applied to a real‐world extradosed bridge with a 230 m main span. A section‐by‐section comparison with conventional beam analysis reveals three findings that current design practice overlooks: (i) beam models mispredict the sign of web shear stress near cable anchorages and underestimate the maximum principal tensile stress by up to 105%; (ii) vertical prestressing produces a markedly nonuniform compressive field along the segment length, contradicting the constant‐stress assumption embedded in the design codes; and (iii) long‐term creep and shrinkage redistribute web stresses along the segment length, with bidirectional changes in vertical stress that gradually equalize this initial nonuniformity. These results demonstrate that the proposed framework brings high‐fidelity solid‐element verification within reach of everyday bridge design practice.
Lu et al. (Thu,) studied this question.