A scientific and effective evacuation plan plays an important role in improving the event reaction ability of the urban traffic system. Many unpredictable natural or man-made disasters result in no-notice evacuation events. This type evacuation has a main trait different with tradition hurricane evacuation and no time to evacuate by their cars. Such specific characteristics of the no-notice evacuation contribute to the complexity of the problem. This complexity affects the evacuation zone definition, the evacuation objectives and the solution approach. Firstly, they need to leave the stricken area quickly by foot, than arrive to the evacuation shelter by public transit system. So we want present a new model to solve no-notice evacuation problem. This research presents an two-phase optimization modeling technique to solve this problem: First, all residents or other evacuees (such as the tourists) are evacuated safely from the danger zone to temporary safe stop by foot. Second, the people who will be picked up by public transit, such as buses, taxies or cars (these cars can be pay the bill by the government or voluntary). The simulation result shows that the proposed algorithm and model can be effectively carried out in a no-notice emergency evacuation.