Short-term safety analysis and interdependencies of mixed-operation freeways with fully separated express lanes: A copula-based poisson lognormal lindley approach
研究了I-4高速公路快速车道与普通车道事故频率的相互依赖关系，使用Copula模型分析短期安全表现，为管理车道设计和运营提供参考。
The I-4 Ultimate Express Lanes (ELs) are part of a major improvement project on Interstate 4, introducing innovative freeway design that impacts traffic safety and operations. This research examines safety performance and the interdependencies between general-purpose lanes (GPLs) and (ELs). Utilizing crash data from February 2022 to February 2024, short-term crash frequency models are developed at the lane-level, leveraging microscopic traffic detector data and unique geometric design features to estimate annual average weekday crash frequencies and capture temporal safety variations. The study employs Poisson Lognormal Lindley (PLN-L) model to address excessive zeros in crash data, particularly prevalent in ELs, while copula-based framework analyzes the dependency between GPL and EL crash frequencies. Frank copula provided the best fit among five tested copula structures, revealing significant dependency, especially at access points. The analysis incorporates lane-level traffic characteristics, geometric data (e.g., standard and I-4 Ultimate specific segment types (interaction between ELs and GPLs at access points), their lengths, and ramp lengths), and time-period effects. Significant variables for GPL and EL segments include lane-level traffic exposure and other factors. Key findings indicate that the average lane-occupancy in the rightmost lane significantly impacts EL safety, with higher crash rates at merge segments. For GPLs, I-4 Ultimate segments and their shorter lengths (i.e., Long [5000ft >= (length) > 3000ft]) are associated with GPL-related crashes. These findings contribute to developing safer, more efficient managed lane systems and offer guidance for future freeway design improvements. The study gives policymakers and engineers insights into EL safety, design, and operation.