19 de diciembre de 2024
Resumen:
Road traffic crashes remain a major public health issue, with vehicle safety measures, particularly restraint systems, playing a crucial role in mitigating injury risk during collisions.
However, the true efficacy of these systems is often difficult to assess, as real-world crash data involves complex variables such as crash conditions, occupant characteristics, and the specific configurations of safety features in different vehicle models. While previous research has explored vehicle safety performance, these studies have not fully included the specific characteristics of the restraint systems across vehicle models. This dissertation aimed to address this gap by developing a comprehensive methodology to assess the impact of restraint systems on injury outcomes, considering the variability inherent in real-world crash conditions.
A method was developed to identify the presence and configuration of restraint system features (such as pre-tensioning and load-limiting devices), enabling an accurate representation of their real-world implementation and facilitating their integration with crash data. The impact of these restraint systems on injury outcomes was assessed using two complementary approaches. The first approach combined real-world crash data with the restraint systems to estimate the injury risk using a logistic regression method. This provided a statistical analysis of the factors contributing to moderate, severe and fatal injuries across various crash scenarios. For the second approach, a vehicle interior sled model, a human model and real-world crash conditions were combined to assess the impact of restraint systems on predicted injury outcomes. These simulations enabled an isolation of crash variables, offering deeper insights into how restraint systems perform across a wide range of real-world scenarios.
This thesis presents several key findings related to vehicle restraint systems and their impact on injury outcomes. A tool was developed to identify the presence and characteristics of pre-tensioner and load-limiting devices in vehicle restraint systems using force-time history measurements. The tool demonstrated excellent performance in identifying the presence of pre-tensioning devices (F1 score: 0.95) and high accuracy for single-stage load-limiting devices (F1 score: 0.90). However, it showed moderate performance for double-stage load-limiting devices (F1 score: 0.77). Detailed restraint system con"gurations for 1,318 vehicles were made publicly available via a GitHub repository, enabling broader community validation and integration with crash test databases. For the first approach to assess the impact of restraint systems on injury outcomes, NASS CDS crash data was augmented with detailed information on pre-tensioning and load-limiting devices. This revealed significant associations between these features and injury outcomes. Load-limiting forces were classiffied into low (below 4.5 kN) and high (above 4.5 kN) to better understand their impact. The presence of these devices was linked to a significant reduction in fatal injuries, with the reduction varying based on load-limiter force level (OR = 0.31 for low-force, OR = 0.42 for high-force devices). Lowforce load-limiting devices were associated with a reduced risk of AIS 3+ whole-body injuries
(OR = 0.70), and both load-limiting force categories were linked to lower AIS 2+ injury risks in the whole-body in high delta-v crashes. However, no significant associations were found between the presence of pre-tensioning and load-limiting devices and injury risks in specific body regions. For the second approach, a method was developed that employed computational models to estimate the impact of restraint systems on predicted injury outcomes.
While the computer modelling-based method did not accurately capture real-world injury distributions, it showed some insights into how variations in crash conditions affect injury outcomes. Significant differences emerged between the Multibody (MB) and Finite Element (FE) models’ predictions, with the morphed FE Human Body Model (HBM) demonstrating greater potential in capturing the impact of restraint systems across a range of crash conditions. The method also highlighted that lower load-limiting forces were significantly linked to a reduction in rib fracture risk across both small and large delta-v values. Though the method demonstrated the ability to assess the effects of restraint system configurations on injury outcomes, it does not yet allow for precise predictions of the number of injuries that could be avoided per 1,000 crashes. Future improvements in injury criteria predictions, particularly for low-severity crashes, may enhance the accuracy of these estimations.
Resumen divulgativo:
Esta tesis evalúa el impacto de los sistemas de retención en lesiones durante choques, utilizando datos de accidentes y simulaciones. Los hallazgos clave muestran una reducción del riesgo de lesiones con pretensores y limitadores de carga. Se desarrolló una herramienta precisa para identificarlos, aunque se requieren mejoras para predicciones más exactas.
Descriptores: Automóviles, Otras
Palabras clave: HBM, human body model, crashworthiness, vehicle, safety, injury, AIS code
Cita:
M. Valdano (2024), Impact of vehicle restraint systems on injury outcomes a combined approach using real-world crash data and computational modelling. Madrid (España).
