Discrete vortex method simulations of aerodynamic admittance in bridge aerodynamics

The meshless and remeshed Discrete Vortex Method (DVM) has been widely used in academia and by the industry to model two-dimensional flow around bluff bodies. The implementation “DVMFLOW” [1] is used by the bridge design company COWI to determine and visualise the flow field around bridge sections, and to determine aerodynamic forces and the corresponding flutter limit. A simulation of the three-dimensional bridge responseto turbulent wind is carried out by quasi steady theory by modelling the bridge girder as a line like structure [2], applying the aerodynamic load coefficients found from the current version of DVMFLOW in a strip wise fashion. Neglecting the aerodynamic admittance, i.e. the correlation of the instantaneous lift force to the turbulent fluctuations in the vertical velocities, leads to higher response to high frequency atmospheric turbulence than would be obtained from wind tunnel tests. In the present work we have extended the laminar oncoming flow in DVMFLOW to a turbulent one, modelled by seeding the upstream flow with vortex particles synthesized from prescribed atmospheric turbulence velocity spectra [3] . The discrete spectrum is sampled from the continuous spectrum subject to a lower cutoff imposed by the discretisation of the simulation time, to an upper limit that is chosen sufficiently high for the discrete spectrum to reproduce the standard deviation of the continuous spectrum. A time series of velocities is generated at two discrete rows of points perpendicular to the flow, forming quadratic cells over which circulation is integrated and associated with particles centered in the corresponding cells. The cell size is equal the distance traveled by a particle in the free stream during a finite number of time steps. The velocity is sampled downstream of the release area and the measured velocity spectra are found in good agreement with the target spectra. The aerodynamic admittance of the structure is measured by sampling vertical velocities immediately upstream of the structure and the lift forces on the structure. The method is validated against the analytic solution for the admittance of a turbulent flow past a flat plate [4] and two types of bridge girder sections. A fair agreement is observed for sufficiently low turbulence intensities and sufficient spatial and temporal resolutions.