Measurement of aerodynamic force partitioning among subcomponents in double-deck girders with varied configurations

With their growing prevalence, double-deck bridges can be classified by the relative width of upper and lower decks into three girder configurations: equal-width (Model A), inverted-trapezoidal (Model B), and lower-deck-cantilevered (Model C). This study combines wind tunnel tests and numerical simulations to quantify the influence of girder configuration, with a focus on measurements of global aerodynamics and load partitioning among subcomponents (upper/lower decks, windward/leeward trusses). Measurements indicate that Model B achieves the lowest drag and smallest absolute lift coefficients under most angles of attack (AoA), yet exhibits the highest moment coefficients. Model A yields peak drag coefficients at AoA = −6° to 2° with intermediate lift and moment values. Model C exhibits drag coefficients comparable to Model A at large AoAs and to Model B at small AoAs, while demonstrating the most pronounced AoA-dependence. Regarding load partitioning among subcomponents, the windward/leeward trusses carry over 50% of mean drag in all girders, with the magnitude sequence of Model C > Model A > Model B. For mean lift and moment, Models A and C distribute generally evenly between upper and lower decks, whereas Model B concentrates them on the upper deck. Furthermore, Dynamic Mode Decomposition (DMD) reveals three dominant vortex modes associated with lower-deck shedding, inter-truss vortex dynamics, and trailing-edge separation. Models A and C exhibit strong inter-deck interference and coherent vortices in the truss gap, resulting in higher drag and greater moment sensitivity, while Model B shows weaker coupling but intensified local separation near the upper deck. These flow characteristics, together with global force and load partitioning results, offer direct guidance for wind-resistant design: Model B is recommended for high-wind regions owing to its lowest drag; however, the concentration of aerodynamic loads on the upper deck necessitates structural reinforcement. Model A supports standardized design through balanced load partitioning between decks, while Model C exhibits relatively high drag and strong sensitivity to AoA, warranting careful aerodynamic mitigation in practical applications.