Long-term extreme dynamic response of suspension bridges: Part I - Methodology

Classical Davenport's approach for determining peak response under turbulent wind may be non-conservative for very flexible bridges, since the low-frequency response widens the distribution of maxima, so that the expected value is no longer representative of the whole probability distribution; moreover, the sought extreme response is not necessarily associated with a single design mean wind speed and variability of turbulence properties is not accounted for by this approach. These issues are naturally overcome by full long-term extreme analysis, well established in marine structures and increasingly adopted for bridges. On the other hand, parametric effects induced by atmospheric turbulence through low-frequency fluctuations in the angle of attack can modulate bridge self-excited forces. The associated time-variation in aerodynamic stiffness and damping can create transient low-damping states and reduce flutter stability, affecting response maxima and their long-term statistics. These mechanisms are sensitive to turbulence parameters, especially vertical turbulence intensity. Including turbulence-driven parametric effects in a long-term framework is usually computationally prohibitive because it requires many time-domain simulations. Building on a previous work of the authors (Barni and Mannini, 2024), the “average parametric effect” is exploited to embed the dominant parametric mechanism in a frequency-domain long-term calculation. In addition, the proposed long-term formulation incorporates static response (non-zero mean) and the influence of flutter instability directly in the extreme-value estimate, enabling a unified treatment for suspension bridges that, to the authors’ knowledge, has not been previously reported. An application of the presented methodology and discussion of results is provided in Part II.