Wind Engineers Complete Proof of Concept for New Structural Design Method

Researchers at the University of Maryland (UMD) and University of Florida (UF) have successfully completed proof-of-concept testing for a new design approach that makes it possible to build stronger and lighter structures that are more resilient to wind hazards.

Supported by the National Science Foundation, the fully-automated, cyber-physical approach combines the accuracy of traditional wind tunnel testing with the efficiency of numerical optimization algorithms.

“Cyber-physical systems allow for rapid testing and evaluation of structural design alternatives in a fraction of the time and with greater accuracy than purely numerical approaches,” said Brian Phillips, the assistant professor in UMD’s Department of Civil and Environmental Engineering leading the project with UF’s Forrest Masters. “Cyber-physical experimental methods have been used for decades in earthquake engineering. There are rich opportunities for applying advanced experimentation across other hazards to aid in performance evaluation and optimal design.”

The heart of his approach is automation. In Phillips’ project, a computer-controlled robotic model is placed in a boundary layer wind tunnel. The model iteratively modifies its shape or behavior in response to sensor readings until it converges on the design best suited for a specific need—such as minimizing the use of building materials or reducing pressure loads on the building surface.

These were the targets when the research team, which included CEE alumnus Michael Whiteman M.S. ’17, set out to demonstrate that their approach can yield structural designs optimized for wind hazards.

The study initially focused on the parapet walls of low-rise buildings. Common along rooftops, these short walls mitigate the pressure on building edges and prevent debris from being picked up by the wind. Build them too high, though, and parapets increase the wind’s downward force.

Phillips and the team successfully conducted the automated design of a two-story building with a parapet inside UF’s boundary layer wind tunnel earlier this year. The parapet wall was autonomously raised and lowered in response to sensor readings until the algorithm honed in on the best possible height given the design objective: 2.7 inches at model scale, or 4.0 feet on the actual building.

But the real power of the cyber-physical system, Phillips said, will become more apparent in future stages of the three-year project as the team applies their design approach to more complex structures.

“Because the iterations can be done in real-time, a structural engineer can determine the effect of different architectural or structural features on wind loads in near real time,” he said. “The result is a more collaborative environment for the architect and engineers, a better use of resources, and an objectively optimal design with input from all parties.”

Phase two of the project will also see the incorporation of a UF supercomputer, making it possible to incorporate detailed finite element models into the analysis. The models will enable the researchers to estimate the forces on individual components of a structure as well as the structure as a whole.

Phillips and Masters also plan to leverage hardware upgrades to the boundary layer wind tunnel to recreate gust effects, making the team among the first to experimentally measure how well designs hold up to non-stationary flow.

“Most wind tunnels can only create a single wind profile—the fan runs at a constant rate. You can slowly adjust the velocity, but you can’t add short controlled bursts,” explained Phillips. “But because UF is supplementing their large fan array with hundreds of small lightweight propellers, we will have the unique opportunity to study the effects of non-stationary flow on our designs.”

Published August 23, 2017