This study presents a framework of traffic evacuation microsimulation modeling that accounts for uncertain network disruptions endogenous to traffic operations. While evacuation modeling considers external stresses such as flooding-related network disruptions, the risks inherent to the transport operations, particularly vehicle collisions may also cause disruptions to evacuation traffic flows. This study adopts a combined Bayes theory and Monte Carlo simulation approach to identify collision hotspots and their occurrence over different times of an evacuation day. A traffic evacuation microsimulation model is developed which explicitly incorporates vehicle collision-related disruptions at the hotspots identified by this probabilistic model. The proposed probabilistic approach identifies 128 candidate collision locations within the study area. The probabilities of candidate locations to anticipate a vehicle collision range between 0.21 and 7.0%. Based on the probabilities, the Monte Carlo simulation approach identifies five hotspots for traffic microsimulation modeling of vehicle collisions during the evacuation. The results from the traffic simulation reveal that due to concurrent collision occurrence, evacuation times vary within 23–31 h depending on the time required to remove traffic disruptions from the network. On the other hand, the concurrent collision occurrence at the hotspots increases the complete evacuation time by almost 11 h if the disruption is not removed from the network, an increase of 50%, compared to an evacuation scenario without disruptions. The analysis of simulated queue length reveals that the hotspots’ traffic queues range from 0.28 to 2.06 km depending on their locations in the study area. The study asserts that an evacuation model without the consideration of the network disruptions due to endogenous risks may underestimate the traffic impacts and network clearance time for an evacuation. These results will provide emergency professionals with insights into managing emergency traffic operation subjected to uncertainties.

这项研究提出了一个交通疏散微观模拟模型的框架，该模型解决了交通运营所固有的不确定网络中断。疏散模型考虑了诸如洪水泛滥的网络中断之类的外部压力时，运输操作固有的风险，特别是车辆碰撞也可能导致疏散交通流的中断。这项研究采用贝叶斯理论和蒙特卡洛模拟相结合的方法来识别碰撞热点及其在疏散日不同时间的发生情况。开发了一种交通疏散微观模拟模型，该模型在此概率模型确定的热点处明确纳入了与车辆碰撞相关的破坏。所提出的概率方法确定了研究区域内的128个候选碰撞位置。候选位置预计车辆碰撞范围在0.21和7.0％之间的概率。基于概率，蒙特卡洛模拟方法确定了五个热点，用于疏散过程中车辆碰撞的交通微观模拟建模。流量模拟的结果表明，由于同时发生冲突，疏散时间在23–31 h内变化，具体取决于从网络中消除流量中断所需的时间。另一方面，如果没有从网络中消除干扰，则在热点上同时发生的碰撞发生将整个避难时间增加了将近11小时，与没有干扰的避难情况相比，增加了50％。对模拟队列长度的分析表明，热点的流量队列范围从0.28到2。06 km，取决于他们在研究区域中的位置。该研究断言，疏散模型不考虑由于内生风险引起的网络中断，可能会低估交通影响和疏散的网络清理时间。这些结果将为紧急情况专业人员提供在不确定情况下管理紧急交通运营的见解。