毒性化學物質洩漏危害範圍模擬之影響因子研究

因應毒性化學物質災害緊急應變的實際需要，行政院環境保護署選用ALOHA（Area Locations of Hazards Atmospheres）模擬軟體納入「毒化災災害防救決策支援系統」，希早期確認下風危害，即時疏散民眾。而ALOHA輸入條件及數據不同，均影響模擬結果，如何於有限時間蒐集關鍵資訊，使模擬結果合理化，為關鍵重點，因此，針對所有影響因子之重要程度，及其間之交互作用實施全面性探討，提供緊急應變人員應蒐集之參數及其優先順序之指引，實有其必要性。本研究以氯氣及氨氣為模擬對象，結合觀音地區歷史氣象資料，採「田口方法」設計實驗數據，透過ALOHA反覆模擬運算，並以類神經分析系統CAFE分析各影響因子之重要性及其相互影響，律定影響因子蒐集之必要性並排定優先順序，提供緊急應變人員於有限時間蒐集關鍵資訊，俾符緊急應變實須。 CAFE應選擇適當訓練次數(Training Cycle)，方能獲致合理結果，研究發現氯之Train Cycle值應設定於6,500，而氨應設定於700。經分析氯氣影響因子之重要性排序為：地表粗糙度>風速>雲覆蓋程度>溫度>洩漏高度>洩漏孔徑>逆溫層高度>相對溼度；而氨氣影響因子之重要性排序為：地表粗糙度>洩漏孔徑>風速>氣溫>溼度>雲覆蓋程度>逆溫層高度>洩漏高度。總之，毒性化學物質因分子量、密度各異，以致影響因子產生不同排序。惟地表粗糙度排序不變，而且影響程度顯著。粗糙度越高，越不利氣體擴散。這是過去相關研究所忽略的議題，但影響甚鉅。另以ALOHA運算不同下風危害距離之最大可運作量，並運用Grapher完成回歸方程式，提供中央主管機關，據以管制毒化物運作量。桃園某化工廠，運作4萬2千公升，遠超過其最大可運作量(1萬2千公升)，潛在風險，實不容忽略。

關鍵字

ALOHA ； CAFE ；危害模擬；氯氣；氨氣；影響因子；地表粗糙度；下風危害；雲覆蓋程度；最大運作量

並列摘要

Based on actual emergency response needs toward toxic chemical disasters, the Environmental Protection Agency (EPA) of Executive Yuan includes Area Locations of Hazards Atmospheres (ALOHA) software into the toxic chemical disasters prevention decision making system so as to provides early downwind hazard warning to evacuate civilians in time. However, different conditions and data input can affect simulation results of ALOHA. It is essential to collect the key information in time and to rationalize the simulation results. Thus, the importance and interaction of parameters should be verified prior to the ALOHA simulation in order to guide the emergency responders to collect the key data in-situ. This study selected the Kuanyin Industrial Park to testify the emergency response to chlorine and ammonia gas leakage disasters. The historical weather data in Kuanyin area were collected and the parameters input were designed by TAGUCHI method. Several scenarios were assumed and the corresponding parameters were fed into the ALOHA system for simulation. The neural network algorithm was performed using the CAFÉ software to prioritize the importance and interactions of the parameters. During the simulations, adequate training cycles were found to be crucial to acquire accurate results when using CAFÉ. According to the results, the suggested training cycles for chlorine and ammonia are 6500 and 700 respectively. Moreover, the importance of parameters for chorine is sequenced as follows: ground roughness > wind speed > level of cloud coverage > temperature > tank opening height > tank opening diameter > altitude of stratosphere> relative humidity, while the order for ammonia is ground roughness > tank opening diameter > wind speed > temperature > relative humidity> level of cloud coverage > altitude of stratosphere > tank opening height. In conclusion, the priority order of parameters for each TIC varies from different molecular mass and densities, while the ground roughness remains at the fixed order and exerts a significant influence on the simulation. The more rugged the ground surface is, the less will the gases diffuse. This result has not been discussed in other related researches but turns out to be essential for TIC disaster simulation. Use ALOHA to calculate each maximum operation amount of different downwind hazard distances, and design a regression equation with Grapher as a basis for the central competent authority (CCA) to control related TIC operations. For example, one factory in Taoyuan operates 42,000 liters of chemicals, which massively exceeds its maximum operation amount (12,000 liters), so the potential risk behind it absolutely cannot be neglected.

並列關鍵字

ALOHA ； CAFE ； hazard simulation ； chlorine ； ammonia ； parameter ； ground roughness ； downwind hazard ； level of cloud coverage ； maximum operation amount