Optimizing urban transportation network reliability by analyzing road traffic accidents

Urbanization has led to increased traffic congestion and road traffic accidents (RTAs), significantly impacting public health, urban mobility, and the efficiency of transportation systems. RTAs disrupt road transport networks, reducing their reliability and performance metrics, which are critical for economic and social activities. This study addresses these challenges by integrating statistical analysis and optimization modeling to enhance the reliability of urban transportation networks through targeted interventions. The proposed methodology builds upon the reliability model by Jovanović Dragan et al. (2011), utilizing statistical analysis of historical RTA data to evaluate transport network reliability. This assessment informs of a linear programming (LP) optimization framework designed to allocate intervention budgets effectively. The LP model incorporates road importance, defined by traffic volume, prioritizing investments on high-impact roads to mitigate RTAs and improve overall network performance. The methodology is demonstrated through a case study in Medellín, Colombia, a city facing significant congestion and high RTA rates (average 100 daily). Using geolocated accident data (2017–2019) and vehicle usage metrics, two model variations were tested: one including road importance and another without. Both models yielded efficient solutions using standard optimization solvers with minimal computational time. Findings demonstrate that the model incorporating road importance provides more targeted budget allocations, aligning better with practical priorities by focusing interventions on the busiest and least reliable road segments. This study highlights the value of combining RTA analysis and network reliability perspectives for data-driven strategic transportation planning. The approach offers actionable insights for policymakers and urban planners seeking to reduce accidents and enhance urban mobility through optimized resource allocation. Future research could expand this framework to include other disruption types (e.g., natural disasters) or validate intervention effectiveness through detailed simulation modeling.