Doctor of Philosophy (PhD)


Department of Civil & Environmental Engineering

Document Type



With expansion of the economy, more and more highway networks extend to coastal areas and mountain valley areas. Vehicles will be exposed to strong crosswinds when driven on these highway roads, especially in hurricane season and in winter in these two different topographic areas. Strong crosswinds threaten the safety of transportation infrastructure and passing vehicles in forms of vehicle accidents that usually result in traffic blockage and driver injury, posing negative effects on economic growth. This dissertation aimed to evaluate the vehicle safety when running through crosswinds in consideration of driver behaviors.

Firstly, the aerodynamic characteristics of road vehicles were identified using computational fluid dynamic method. Aerodynamic coefficients of a high-side lorry running in crosswinds using both traditional resultant-wind velocity method and relative-motion approach were compared. In addition, the aerodynamic coefficients of multiple types of vehicles were investigated. The curves of aerodynamic coefficients for different vehicle types against wind yaw angles were obtained.

Secondly, an experimental investigation on the vehicle performance and driver behavior was conducted by taking advantage of the LSU’s driving simulator. This study revealed the repeatability of driver behavior and the effect of crosswind speeds on the vehicle performance and drivers’ behavior through a statistical analysis. More scenarios were considered, such as driving in windy-rainy conditions. A regression model of the steering wheel angle turned by drivers was obtained.

Finally, safety assessment of vehicles was performed based on an improved wind-vehicle-bridge coupled system and considering driver’s behavior using a series of driver behavior models. For different types of road vehicles, rigid frame vehicle model and flexible frame vehicle model were developed. Accident criteria of lateral side slip, rotational deviation, and rollover were considered. To investigate the influence of driver models, four driver models were considered in different integration methods. Results between cases from different driver models were compared.

Committee Chair

Cai, Steve C.S.