In times of disaster, ensuring the safe evacuation of affected populations is crucial for saving lives and mitigating community risks. This research presents a Dynamic Population Evacuation (DPE) approach, which combines strategic planning and real-time management aided by vehicular communication technology. By addressing the impact of disasters on transportation and communication systems, the DPE framework utilizes dynamic shelter allocation and simulation-based traffic assignment techniques to enhance planning accuracy. It incorporates a trip-based traffic simulator to account for the effects of disasters on transportation networks and a Vehicular Ad Hoc Network (VANET) simulator to manage communication dark zones. To manage dark zones, The proposed framework continuously updates the evacuation plan in real time to manage road blockages and the disabling of VANET resources. Accounting for the temporal evolution of areas with communication dark zones improves evacuation efficiency regarding clearance time. A sensitivity analysis is conducted on the compliance rate of evacuees to instructions provided by the vehicular communication management system. Furthermore, the framework's effectiveness is evaluated by simulating the real test case of the 2018 Camp Fire wildfire in Paradise, California, where roads and communication were severely disrupted. Additionally, by comparing the architectures of cloud computing and fog computing while accounting for blocked roads, the framework shows more than a 12% improvement in network clearance time compared to fog computing and a 19% improvement compared to traditional evacuation planning.