Developing a correlation model between the frequency of road traffic accidents and road geometric design elements

Abstract

Road traffic crashes (RTCs) are a global problem that affects all sectors of society, irrespective of age, gender, or socioeconomic status. Specifically, Wolaita zone in Ethiopia is currently experiencing a high number of road traffic crashes. The study focused on the Halaba-Sodo area, which is a road segment that frequently witnesses traffic crashes within the Wolaita zone. This research aimed to conduct a comprehensive analysis of the road geometric design elements that contribute to RTCs, with a particular focus on crash frequency. The factors contributing to RTCs were categorized into three groups: human factors, vehicular factors, and road and environmental factors. To achieve the research objectives, a total of 532 road crashes that occurred over a six-year period (2011-2016 G.C.) were collected from secondary sources such as traffic police records and the transport office database along the entire Halaba-Sodo road. Additionally, field measurements and observations were conducted to gather the necessary data for crash analysis, in addition to the secondary data used in this study. Once the important data was collected using both primary and secondary data collection techniques, statistical analysis was performed using the principal component analysis method. Based on the analysis, significant variables with strong predictive abilities for model development were selected from the three categorized contributing factors. Subsequently, a mathematical model was developed using multiple nonlinear and partial least squares regression models. The Monte Carlo approach was utilized to determine the interactive effect of individual components or variables on the model, through 10,000 repeated simulations of samples within the probability density functions of the input data. The findings of the study revealed that gradient, sight distance, and horizontal curve were highly correlated with accident frequency, with correlation coefficients of 0.692, -0.529, and 0.426, respectively, among other road geometric factors. It was observed that gradient had a positive relationship with crash frequency, whereas sight distance and horizontal curve had a negative relationship. These findings provide valuable insights for engineers, urban planners, and policymakers in developing strategies to mitigate traffic crashes by implementing informed road design improvements.