Field calibration validates wind tunnel testing on iconic cable-stayed bridge
The Prince of Wales Bridge (previously Second Severn Crossing) spans the estuarial southern end of the River Severn, which has a tidal range of 14.5 meters (48 feet), the second highest in the world. Opened in 1996, the bridge offers the southernmost crossing between Bristol, England, and South Wales. The six-lane structure has three main sections: viaducts that are roughly 2.1 kilometers (1.3 miles) long on each end with a 945-meter (3,100-foot) cable-stayed bridge, known as Shoots Bridge, in the middle. The main span is 456 meters (1,496 feet) between pylons.
The bridge was constructed to relieve the growing traffic congestion on the first Severn Bridge, a suspension bridge located several kilometers to the north, which frequently must restrict traffic to prevent vehicle rollovers during strong winds that are common in the area. The new bridge included a 3 meter (9.8 foot) high, 50% porous wind screen along both sides of the deck for its entire length across the cable-stayed section. The screens have been very effective in avoiding traffic restrictions during high wind conditions.
Although the wind screens help maintain traffic flow, they also affect the aeroelastic response of a bridge to wind. Our early testing revealed that measures taken to control flutter increased vortex shedding oscillations and vice versa. It was up to us to develop the specific measures that would control the bridge’s aerodynamic response. We began by conducting wind tunnel testing on sectional models of the cable-stayed portion of the bridge. In the end we identified a section that did exhibit limited amplitude vortex shedding oscillations but they were sensitive to damping and turbulence.
With the expected full-scale damping and the estimated turbulence levels at the site, the amplitude of oscillation was expected to be within the design criteria being used. However, once the bridge had been constructed, vortex oscillations were observed with a larger amplitude than expected. Our challenge then became to understand why.
The University of Bristol had instrumented the newly constructed bridge and was able to record in detail the bridge response, wind speed, wind direction and turbulence properties. This gave us the opportunity to identify why the oscillations were larger than expected and also to compare wind tunnel test results with the bridge’s full-scale behavior. This opportunity for field validation of laboratory testing provided us with valuable insights into turbulence effects and sectional wind tunnel model methodology. For example, the full-scale data showed clearly that the damping ratio of the bridge was lower by about a factor of two than had been assumed during design, based on recognized standards and published design guidance that was available at the time.
Our investigations enabled us to identify better analytical methodology for calibrating the wind tunnel simulations to full-scale behavior. When the full-scale data analysis eventually was completed, we identified analytical adjustments that brought the estimated behavior of the bridge into closer alignment with observations.
Having completed this calibration of the as-constructed bridge motion to the sectional wind tunnel study, we developed an aerodynamic solution that eliminated the motions due to vortex shedding without increasing flutter: two vertical baffle plates installed along the central 250 meter length of the main span.
The knowledge we gained from the Prince of Wales Bridge project continues to inform our work today. In addition, this project illustrates both the difficulty and importance of fully accounting for the effects of wind turbulence and damping when assessing the effects of vortex shedding on bridges and other structures.