Evacuations are the quick removal of people from a dangerous area, and are necessary for disaster response. However, traffic congestions due to the surge in demand and inefficient planning often limits the effectiveness of an evacuation. In this study, it is suggested that scheduling the departure time and assisting the destination and path choice of evacuees could optimise the evacuation process. A real-time decision support and pre-planning model is proposed, and a system-optimal dynamic traffic assignment problem is formulated. A simulation-based metaheuristic is used, consisting of an upper-level that is a genetic algorithm engine which optimises the solution, and a lower-level that is a traffic simulator which evaluates the goodness of the solutions generated by the upper-level. The traffic simulator is a macroscopic dynamic traffic assignment model that captures queuing and congestion. The model is tested on the Sioux-Falls network and an expanded, more realistic version of that network, known as Enriched Sioux-Falls network. The effects of varying the number of paths evacuees are allowed to use and the frequency of decisions made is observed. The proposed model showed that scheduling departure time is an effective method to improve evacuation efficiency when compared to a worst case benchmark scenario.