Development and Validation of Wind Assessment Software for UAS Bridge Inspections

The increasing age and number of bridges in the United States have led to the development and integration of new technologies to reduce the workload and cost of bridge inspectors. Unmanned Aerial Systems (UAS) have exemplified a technology that can reduce labor hours and equipment costs while improving safety and increasing the quality of bridge inspections. Despite the benefits of using UAS during bridge inspections, UAS face operational challenges such as turbulent wind flows, wind shear, and limited GPS reception. These challenges can discourage bridge inspectors from using the vehicle adequately during a bridge inspection. Therefore, providing UAS operators with information typically gained through extensive flight experience can expedite the learning curve.

The following research was conducted to develop and validate a process capable of predicting hazardous wind locations around a bridge structure. The methodology focused on utilizing a multi-step process across existing technologies. First, the bridge model was created using Computer Assisted Models (CAD) or photogrammetry modeling. Second, the resulting model is processed in a Computational Fluid Dynamics (CFD) application with a Reynolds Averaged Navier Stokes (RANS) solver to simulate wind flows around the bridge structure. Three inlet conditions were processed using a weather station at the bridge, the closest Aviation Weather Observation Station (AWOS), and height-adjusted velocities depending on the bridge height. Finally, the resulting wind models are compared against the wind velocity values collected by a sonic anemometer mounted to a UAS flown around the bridge.

The validation process used three real-world bridges across multiple weather conditions to validate the CFD model predictions. The three bridges used were a pony truss, a simulated girder, and a truss bridge. The pony truss bridge was extensively used as a validation structure where multiple flights occurred around the same structure with variations in wind heading and velocity. The simulated girder and truss bridge were treated as a simulated bridge inspection to validate the wind velocities around the structure. The resulting Root Mean Square Error (RMSE) of the validation measurements compared to the simulations ranged between 0.5 and 1.0 meters per second (15% - 30% error). The results demonstrated that CFD simulation is a valid prediction tool for estimating wind flows around a bridge structure.