During Hurricanes Ivan in 2004 and Katrina in 2005, at least 11 highway and railroad bridges along the U.S. Gulf Coast were damaged. When the water rose during the storms, wave forces slammed into the bridges’ supporting substructures, and when it rose high enough, the water’s buoyancy had enough power to lift off sections of a bridge’s superstructure and lay them aside like giant Legos.

To build bridges that can withstand the force of hurricane waves, engineers must be able to estimate the effects those waves will have on bridge structures. An OTREC project led by Oregon State University professor Daniel Cox examined the effects of wave loading on highway bridge superstructures.

Cox and co-investigator Solomon C. Yim, also of Oregon State University, conducted experiments in the Large Wave Flume at the O.H. Hinsdale Wave Research Laboratory at Oregon State University. They used a 1:5 scale, reinforced concrete model of a section of the Interstate 10 Bridge over Escambia Bay, Fla, which failed during Hurricane Ivan.

To see more details about the project, “Hurricane Wave Forces on Highway Bridge Superstructure,” click here, or download the final report.

The problem addressed by this project is that, while calculating the force of buoyancy is a straightforward analytical process, accurate estimation of the force of moving water – hydrodynamic force – is far more difficult.

Engineers often use analytical models: mathematical functions which help calculate the forces that will be acting on a structure, and the shape and strength it will need to stand against them. Highway bridge superstructures, however, have complex geometries. Their response to multidirectional forces can be difficult to estimate. When the added complexities of trapped air, turbulence, and the reactive movements of the bridge’s structure are considered, analytical solutions become impractical and empirical models are the only way to go.

Small-scale models have been designed in the past, but there are limitations to working small. Real-world bridges are made of reinforced concrete, as was the one used in this project. Previous models have been fabricated from wood, plastic or metal due to difficulties working with concrete at small scales. The difference in size and material makes the water affect the model differently from the real thing. Moreover, previous bridge models treated the structure as rigid, with no dynamic response to the forces. This also led to inaccurate results.

The model that Cox used was designed to move like a full-size bridge using Froude scaling, so that the velocities acting on the 1:5 scale model had the same ratio as the velocities acting on a full scale. This assures that the model and the full-size version have geometrically similar motions. Cox’s model also incorporated a roller and rail system, which allowed the model bridge section to move back and forth, simulating the way a real bridge would respond to the waves.

Using the flume to generate different sizes of waves, researchers measured wave conditions along with the resulting forces, pressures, and structural responses. The data they collected will help to calibrate future numerical simulations, leading to the potential for bridge designs that can withstand stronger hydrodynamic forces.

In a follow-up project, also funded by OTREC, Cox and co-investigator Tim Maddux of Oregon State University conducted some more experiments using large-scale test models. This time they wanted to study the transfer of loads from the bridge superstructure to the column supports.

Click here to read more about that project, “Hurricane Wave Forces on Highway Bridge Superstructure: Psuedo-dynamic Testing for Bridge Subassembly,” or download the final report.

Using some of the wave-loading data from Cox’s earlier experiments, Cox and Maddux tested full-size prototypes of a bent-column assembly, part of the supporting substructure of a bridge. They had two identical models, which they tested under different wave loading conditions to gather data about how and when the columns sustained damage.

Part of the experiment was to see how the bent columns would handle being retrofitted to attach rigidly to the bridge deck.

None of the bridges that were damaged by the Gulf Coast hurricanes had rigid attachments between the deck and bent columns. A typical connection between the superstructure and the supporting bent cap consisted of bolted bearing plates with shear keys, but in some cases the spans simply rested on the bent caps, with no physical connections to resist uplift or lateral forces. Once the bent cap connections failed, subsequent waves were able to push the bridge deck off of the supporting substructure.

Cox’s research provides guidance on retrofit methods for existing bridges. The methodology developed in these studies may also serve as a foundation for other hazard assessment tests, including tsunami effects.