本論文應用模擬軟體Simulex，以電腦輔助設計軟體AutoCAD所繪製的建築平面圖作為模擬時的建築邊界，並設定該樓層之直通樓梯或緊急逃生梯等為主要避難設施，針對某半導體科技廠1樓至4樓共計四個樓層發生火災或其它緊急狀況時，模擬人員避難行為及預測離開該事故樓層至1樓安全區域所需之時間。本論文以兩種情境進行模擬：(1)情境A的設定為「非火災而需全員疏散之事故」，4個樓層的全體人員透過3個逃生梯避難，移動至1樓逃生梯出口的事件；(2)情境B乃設定「1樓發生火災，導致1樓至4樓的逃生梯B無法使用」時，人員由其它可通行的路徑，前往1樓逃生梯出口避難的情況。經Simulex軟體模擬後得知：情境A中的810人員經由3個逃生梯避難，其總避難時間為4分56.5秒；情境B相同的810人員經由2個逃生梯避難，其總避難時間為5分31.5秒。由模擬結果可知在真實避難狀況中，由於人群對現場不熟悉，容易有盲從行為產生，或由於歸巢性而循進入場所之原路徑避難，並非往最近之出口避難，因此建議Simulex軟體模擬計算所得之避難時間，需乘以1.5倍之安全係數，以更貼近真實的狀況。本論文根據此模擬結果提出後續兩項建議：(1)增加Simulex軟體的模擬涵蓋範圍與條件，以達到模擬結果更接近實際人員避難情形；(2)進行實際人員避難演練，以了解模擬結果與實際情況之差異，以利逃生避難之規劃。;In this study, the software Simulex was used to simulate that when there were fires or other emergencies on the four floors in a semiconductor technology factory, the time required for all the personnel to leave the accident floor to the safe area on the first floor using the emergency exit between each floor or other safety facilities such as a safety ladder, and also the behavior of the people during evacuation. The simulation was performed using the building layout drawn by a computer-aided design software AutoCAD as the architectural boundary, and the straight-through stairs or emergency escape stairs as the main refuge facilities. In addition, two scenarios were included in the simulation: A) There is no fire incident. All personnel from each floor evacuating to the exit on the 1F with equipped three scape ladders. B) Presumed there is a fire incident at 1F that caused safety ladders between the 1st to the 4th floor unavailable. Personnel were evacuated to the exit at 1F through other feasible routes. Results of the simulation showed that: A) 810 members took total of 4 minutes and 56.6 seconds to evacuate using three safety ladders; B) 810 members took total of 5 minutes and 31.5 seconds to evacuate using two safety ladders. It can be seen that in real evacuation situations, people may not be familiar with the environment and take the wrong routes due to blind behavior; others might take the backward route to the entering, instead of using the nearest exit, due to homing ability inherent by human nature. Therefore, the simulated escaping time should be normalized and increased by a factor of 1.5 as the safety coefficient. Based on the simulation results, two suggestions were proposed: 1) improve simulation coverage and conditions of Simulex software to match escaping scenario in the real world; 2) verify the results with real exercise and experiment to understand the differences between the simulated results and real situation for future emergency evacuation planning.
