The ability of wind to damage and even destroy structures has been known to engineers since antiquity. In this century, the well-documented collapse of the Tacoma Narrows Bridge in 1940 served as a pivotal event in clarifying the need for better methods of designing highway structures to resist wind effects. Slightly more the four months after it was opened, the Tacoma Narrows Bridge was literally twisted apart by the aerodynamic forces that developed as a result of the steady winds over the Puget Sound. In spite significant advances in developing successful windresistant design technologies, wind effects remain a hazard to highway structures.

The Federal Highway Administration (FHWA) is currently working to solve some of these problems by developing improved design and retrofit methods and by educating designers in the use of the modern methods. The tools being used in this effort include wind tunnel testing, field measurements, computer simulations, climatological studies, and stochastic analysis.

Problems

Elongated flexible structures with bluff exposed sections, such as bridge decks, cables, columns, and signposts are particularly susceptible to wind loads. These structures are usually designed with plenty of strength to resist the static wind loads of the highest expected wind speeds. However, dynamic fluid-structure interactions can give rise to troublesome aeroelastic phenomena, over a wide range of wind speeds, that defy simple static design methodologies.

These phenomena include flutter, buffeting, vortex-shedding, and wind-rain vibrations.

Flutter occurs when the interaction of a bluff section and the wind create a condition in which small motions of the bridge deck can extract energy from the wind and produce larger motions. This situation is often referred to as negative damping. Flutter produces motions, often in the form of torsional bridge deck oscillations, that grow exponentially. It is believed that flutter caused the collapse of the Tacoma Narrows Bridge. Buffeting is the beating of wind gusts against a structure, causing oscillations. Buffeting rarely causes severe damage to a structure, but it can cause long term fatigue damage and unacceptably large structural motions.

Vortex-shedding occurs when the wake behind a bluff structure causes the formation of large eddies or vortices that shed off alternate sides of the structure. If the frequency of shedding coincides with a natural vibration frequency of the structure, large, steady vibrations of the structure can ensue. These oscillations tend to occur at low wind speeds and can be particularly disconcerting to the public.

Wind-rain vibrations usually appear on the main cables of cable-stayed bridges. These vibrations occur only when it is raining. It is believed that water rivulets on the cables alter the aerodynamic behavior of the cable so that the cable becomes unstable and oscillates.

The occurrence of any one of these wind effects on a highway structure is a big problem that demands attention. Although wind problems often occur only under a rare set of meteorological circumstances, the extended lifetime of most highway structures virtually guarantees that the problems will recur.

The best method of avoiding wind problems is to design the structure properly before it is built. If this is not possible, then retrofit countermeasures are warranted. Some countermeasures can be quite expensive, and to effectively counter the wind effects, it is necessary to have a rational understanding of the phenomena and available design techniques. This requires a coordinated synthesis of field, laboratory (wind tunnel), climatological, computational, and analytical studies.

Wind Tunnel Studies

The wind tunnel is one of the primary tools for investigating wind effects on structures. Through the application of the principles of similitude, small-scale models can be tested to simulate full scale conditions.

The FHWA Vincent Aerodynamics Laboratory at the Turner-Fairbank Highway Research Center in McLean, Va., contains a low-speed, open jet wind tunnel specifically designed to test bridge sections. The tunnel has a speed range of 0 to 15 meters per second (m/s). The test section is 1.5 m by 1.5 m. The wind tunnel is equipped with a variety of force balances, aeroelastic balances, anemometry, data acquisition systems, and turbulence generators.

A major object of study in the Vincent Wind Tunnel has been the Deer Isle-Sedgwick Bridge. This bridge, although not as well known as the Tacoma Narrows, poses many of the same challenges and problems for the wind engineer. The bridge was completed in 1938. The deck has the same plate girder profile as the Tacoma Narrows Bridge, which was subsequently shown to have such poor aerodynamic characteristics. Like the Tacoma Narrows Bridge, the Deer Isle Bridge was built during the Depression and has a relatively light and flexible stiffening structure. Almost immediately after its construction, the Deer Isle Bridge experienced large wind-induced oscillations (in excess of 250 mm). To prevent any severe damage, the bridge was fitted with a variety of diagonal stays and cable restraining blocks. See figure 1.

The structural retrofit, while violating most principles of aesthetic design, prevented serious damage to the Deer Isle Bridge. However, the bridge has experienced several episodes of large wind-induced motions that caused a modest amount of damage. Since the structural retrofits did not alter the fundamentally unstable aerodynamic profile of the bridge deck, it eventually became apparent that an aerodynamic modification would ultimately be the best solution to the problem. 

Figure 2 shows the cross sectional geometry of the decks of the original Tacoma Narrows Bridge, original Deer Isle Bridge, and a fairing-modified version of the Deer Isle Bridge.

The data from a wind tunnel study of the torsional aerodynamic stability of the original and fairing-modified versions of the Deer Isle Bridge appear in Figure 3. The curves show the torsional stability flutter coefficient A *2 as a function of wind speed. Positive values of the A *2 coefficient indicate unstable conditions. It is clear that the fairing modification produces a more stable section.

The Vincent Wind Tunnel is also used to test other highway structures. Most highway structures have bluff cross sections. Since bluff sections cause high drag forces, they have not been studied extensively by the aerospace community. An effort is currently underway to characterize and catalog the aerodynamic behavior of standard bluff sections.
The sections being studied include generic cylinders, polygons, H-shapes, rectangles, trapezoids, as well as common welded signpost box-member shapes. Tapered circular and polygonal shapes are being tested as well. Figure 4 shows the results of a typical test on an octagonal section where the drag force is measured as a function of wind speed at two different angles of attack: where a flat face is perpendicular to the wind and where an edge is pointing into the wind.
The results of these studies will be improved wind design information and standardized data that can be used for intertunnel comparisons and for validating numerical models.

Full-Scale Measurements

The performance and validity of wind resistant design techniques must ultimately be verified by field measurements of full-scale performance. The big difference between field and laboratory measurements is that in the field the engineer has no control over the wind environment. Instead, he/she must wait for the appropriate conditions. The really interesting events may occur only once per decade. The long intervals between events requires a testing strategy that includes the use of automated data acquisition equipment for remote control and event-triggered measurements.

The winds impinging upon and the wind-induced motions of the Deer Isle-Sedgwick Bridge have been monitored in the field since 1981. The field instrumentation consists of 15 accelerometers, six tri-axis anemometers with a relatively high-frequency response, two low-frequency sky-vane anemometers, three thermometers, and various voltage reference channels. In all, 44 channels of data are collected simultaneously in digital form and stored on site. The data acquisition system is queried and controlled via telephone communications with the FHWA facilities in McLean, Va.

An aerodynamic modification to the bridge in the form of triangular-shaped fairings that were added to the full length of the bridge was completed in 1994. The bridge is now being monitored after the fairing modification to determine the effectiveness of the fairings on the aerodynamic performance of the bridge.

A very similar field monitoring project is ongoing at the Luling-Destrehan Bridge, which spans the Mississippi River near New Orleans. The Luling Bridge is one of the first cable-stayed bridges in North America to be built with a streamlined deck in a hurricane-prone setting. As a result, the Luling Bridge is also being monitored for windinduced vibrations. To date, no hurricanes have crossed the bridge site although there have been some close calls.
Therefore, the wind and vibration data measured at the site has been rather uneventful. In general, the bridge has performed well, as predicted in the wind tunnel. However, there have been several occasions where the stay cables have been reported to vibrate. Accelerometers have been deployed on a limited number of cable stays to observe the stay cable vibrations.

One of the biggest challenges in conducting field studies is that an enormous amount of data can be collected very quickly. Analyzing the data and putting it into a useful form requires modern, multi-channel signal processing techniques and data archiving systems.

Computer Modeling

The simulation of wind effects on highway structures via computer methods is quite complicated due to the need to model the interaction of the structure and the wind. A comprehensive model that simulates all of the structural and aerodynamic details is beyond the capability of present-day supercomputers. Instead, the numerical modeling effort requires the judicious application of the principles of engineering approximation. The photograph on the right shows the results of one of these studies where the two-dimensional flow around the Chesapeake-Delaware Canal Bridge is simulated with a finite difference scheme.

Reliability Modeling of Highway Support Structures

Highway support structures such as signals and luminaries are also subjected to wind effects, in particular the action of fatigue. Since the wind-induced fatigue failure of highway support structures is a rare, but quite dangerous phenomenon, a research program is underway to develop methods of predicting the fatigue response of these structures. The research is aimed at developing rational stochastic-based models.

The main issue of concern is how to model the effects of vortex-shedding, wind gusting, and truck turbulence on the fatigue life of support structures so that more reliable structures can be built. The inherent randomness of wind loads and the relatively large spreads that are observed in the fatigue life of structures under random loadings is being modeled with stochastic methods. The eventual goal of this research is to develop improved guidelines for the wind load provisions of the American Society of Civil Engineers and for the specifications of the American Association of State Highway and Transportation Officials.

Climatological Studies

The rational aerodynamic design of highway structures, particularly long-span bridges, requires the use of estimates of the speed, direction, and duration of the winds which will attack a structure during its lifetime. One of the best methods of establishing the design wind environment is to use historical wind data as a basis.

The United States has an extensive log of wind data from weather stations that are distributed throughout the country. Unfortunately, this data set consists of a patchwork of different formats, none of which are precisely suited to the task of formulating design wind speeds for highway structures. Most of the data is collected at locations that are near to, but not at, the bridge site. This situation is being remedied by examining the climatological data at or near a particular bridge site and adapting the data to the standard design formats.

Information Dissemination

The ultimate success of the research conducted by the Vincent Laboratory and by other wind engineering centers is the extent to which procedures and practices developed as part of the research program are implemented in the field and by the resulting performance of these procedures and practices. A large impediment to adopting new techniques is that practicing engineers are not readily exposed to the information. As a result, the Vincent Laboratory is in the process of developing educational and promotional literature that supplements the usual papers, reports and technical presentations. The educational literature will be in the form of newsletters, technical guides, technical booklets, and case studies on wind engineering practice, particularly as it pertains to highway structures.

References

1. K. Yusuf Billah and Robert H. Scanlan. "Resonance, Tacoma Narrows Bridge Failure and Undergraduate Physics Textbooks," American Journal of Physics , 59(2), Feb. 1991, pp. 118-124.

2. Robert H. Scanlan. State-of-the-Art Methods for Calculating Flutter, Vortex-Induced and Buffeting Response of Bridge Structures, Publication No. FHWA-RD-80-050, Federal Highway Administration, Washington, D.C., 1981.

3. Harold R. Bosch. "Aerodynamic Performance of the Deer Isle-Sedgwick Bridge," Restructuring: America and Beyond, Proc. Structures Congress XIII , M. Sanayei, ed., Vol. 2, pp. 1558-1562, American Society of Civil Engineers, New York, 1995.

4. Okechukwu U. Onyemelukwe. Computer Simulation of Wind Effects on Bridges and Other Bluff Bodies , Ph.D. Dissertation, Dept. of Civil Engineering, University of Pittsburgh, 1993.

Dryver R. Huston is an Associate Professor of Mechanical Engineering at the University of Vermont. Previously he conducted postdoctoral research into bridge aerodynamics in the Department of Civil Engineering at Johns Hopkins University, and he was a Graduate Research Fellow at the FHWA Turner-Fairbanks Research Center (TFHRC) in McLean, Va. He has a bachelor's degree in mechanical engineering from the University of Pennsylvania and a master's degree and doctorate in civil engineering from Princeton University.

Harold R. Bosch is a research structural engineer and the director of the Vincent Aerodynamics Laboratory at TFHRC. For more than 24 years, he has conducted laboratory and field research studies related to structural loads, dynamics, and aerodynamics. He received his bachelor's degree in civil engineering from the University of New Mexico and has pursued advanced studies at both Colorado State University and the University of Maryland.