A digital twin-driven investigation of coupled vertical-torsional soft flutter of a flat closed-box bridge deck section

Soft flutter or limit cycle flutter, a nonlinear self-limiting vibration phenomenon, has been observed in blunt-body cross sections of bridge decks in recent years. This study therefore proposes a digital twin approach to investigate coupled vertical-torsional soft flutter of a flat closed-box bridge deck section and explore its physical mechanisms. In the proposed approach, the coupled vertical-torsional soft flutter of a flat closed-box bridge deck section was first observed through wind tunnel tests utilising a synchronised force-vibration testing system and regarded as a physical entity, and the coupled vertical-torsional vibration responses and self-excited aerodynamic forces of the deck section were recorded. A computational fluid dynamics (CFD) simulation model, accounting for fluid-structure interaction and nonlinearities, was then constructed as a virtual entity of the physical entity. A Gaussian process surrogate model and a phased parameter optimisation strategy were then introduced to establish a self-learning digital twin to achieve high-precision inversion of mechanical damping ratio and stiffness of the deck section. Finally, a parameter-wind speed mapping covering the entire tested wind speed range was constructed, forming a digital twin system capable of predicting soft flutter responses at untested wind speeds. It is demonstrated that the established digital twin accurately reproduces the limit-cycle vibration characteristics of the coupled vertical-torsional soft flutter. The digital twin effectively reveals the intrinsic relationship between aerodynamic energy input, damping dissipation, and vibration stability, and provides an effective pathway for nonlinear soft flutter analysis and response prediction of blunt-body deck sections.