The influence of automotive aerodynamics on spray transport and the implications for ADAS sensor operability

<p dir="ltr">Advanced driver assistance systems (ADAS) are now commonplace, likely foreshadowing the advent of fully autonomous vehicles. These systems rely on suites of external sensors, the performance of which may be hindered by both the soiling of functional surfaces and by spray suspended in their fields of view. Numerous studies have been published describing methods of mitigating the impact of adverse weather on system functionality, with some acknowledging the complex interactions between the spray and the automotive flow field. A robust method of assessing system reliability has yet to be realised, with the validity of on-road tests heavily weather-dependent. Despite this, aerodynamic research has primarily focused on surface contamination and has largely overlooked the impact of the turbulent wake on the suspended spray. A deeper understanding of these interactions is essential for fully autonomous vehicles to gain widespread appeal.</p><p dir="ltr">This work investigates the interactions between automotive aerodynamics, the transport of road spray, and the potential consequences for sensor operability. A novel means of experimentally measuring the structure of road spray is demonstrated at a reduced scale through a comparative study utilising the Windsor Model. The experimental campaign is then replicated numerically using high-fidelity multi-phase computational fluid dynamics. The continuous phase is computed using a detached-eddy-simulation-based approach, with the dispersed phase being computed concurrently in a separate Lagrangian framework. The described numerical method, validated by experiment, is subsequently applied to a more road-representative case. As such, a full-scale version of the Windsor Model is investigated under conditions representative of wet-weather highway driving, and potential applications of the resulting spray data for ADAS development are demonstrated.</p><p dir="ltr">The unsteadiness inherent to spray transport is demonstrated both at a reduced scale and under road-representative conditions. The instantaneous spray field consists of discrete clusters of spray and tendril-like formations, as opposed to a continuous plume. While the high mass loading associated with extremely wet conditions has a significant impact on the flow field close to the vehicle, these structures remain evident downstream. Analysis of the unsteady spray data, replicating the perspective of downstream ADAS sensors, suggests that these structures can both obscure vision and generate deceptive patterns. The resulting signals can be misidentified by object classification algorithms, leading to false detections and potentially erratic system responses.</p><p dir="ltr">This thesis underlines the significance of the turbulent wake on spray dynamics and the need to consider its effects in future studies of automotive multi-phase flows. It also demonstrates how access to detailed, physically-based spray data could revolutionise how ADAS implementations are developed, tested, and refined, potentially addressing one of the primary obstacles to the mainstream adoption of autonomous driving technology.</p>