Determination of H_{i}(K) and A_{i}(K) should generally be based on the sectional wind tunnel test, or recently, by numerical simulation on computer. Both of the methods physically apply one of the following procedures, [97,2]: 1. Vibration tests where a fixed girder section is given an initial vertical and torsional displacement. The flutter derivatives are based on the transient behavior that occurs when the section girder is released. 2. A forced oscillation technique that involves forcing the model through a prescribed motion and measuring the aerodynamic forces on the model. The aerodynamic forces may be determined using pressure measurements at a number of pressure taps on the model. 3. Buffeting tests, where the behavior of girder section is analyzed for different wind velocities in the tunnel. The behavior observed in the simulated natural wind is analyzed in accordance with the differential modal equations for vertical and torsional motion.

Flutter derivatives primarily depend on the girder configuration, see Figure 4.2, for the main kind of the streamlined girder and the truss girder. During the preliminary computation of the suspension bridge flutter (Chapter 5), when the final girder is not yet determined, an equivalent flat plate aerodynamics model can be used as a test input with the structural mass, natural frequencies and damping ratios corresponded to the actual bridge. The predicted critical wind velocity of the actual bridge when choosing the streamlined girder will be about 80% to 90% of the value of the flat plate model, [97,2]. 123456789_123456789_123456789_123456789_12345

The buffeting wind loads occur because of the wind turbulence as described in the section about wind. Buffeting wind load acts on the bridge girder cross section can be commonly expressed in terms of a vertical lift force L_{b}, a moment M_{b} and a horizontal force D_{b}, all are per unit span as shown in Figure 4.4. C_{L }, C_{M }, and C_{D} , respectively, the non-dimensional lift, moment and drag coefficient, determined experimentally in the wind tunnel as a function of the girder rotation r_{b} in flow (angle of attack) The buffeting forces given above assume that there are no interaction between the aeroelastic and the buffeting forces. This formulation is only approximate since it bases unsteady, time-dependent results upon a quasi-steady formulation [92,3]. That means the time dependency of buffeting forces only depend on the oncoming turbulence and not definitely upon the signature turbulence (produced by the structure itself inthe flow, even if the incoming flow is perfectly smooth, i.e. u = w = 0). I.e. L_{b} and M_{b} requires the state of the air updating, but does not depend on the turbulence factor due to the girder motion. 123456789_123456789_123456789_123456789_12345