煉油石化業製程安全事故分析及探討

Abstract

煉油石化業在本質上為高風險之產業，稍一不慎極易產生火災爆炸事故，對企業本身及社會大眾造成重大影響。因而煉油石化產業非常重視安全衛生，先進國家推行的製程安全管理均具備事故調查的要求，利用事故調查資料提供管理回饋機制，作為重要的工安先期指標，是改善製程設計及安全操作有效的方法。 近期研究顯示大部份事故情境及原因相似且重複，顯見業界並未善用事故經驗學習到教訓，導致事故重複發生。對於發生頻率低但後果嚴重之煉油石化業事故特性，事故資料庫統計分析將更形重要。目前國際上已有許多化學事故資料庫之建立，國內現有勞動部「職業災害統計月報表」係依職業災害事故之特性編列，不敷製程產業如煉油石化業之需求。導致相關事故致災關鍵資訊未能深入探討並儲存於事故資料庫中，不僅無法累積失敗學習經驗，並錯失與同業間寶貴之經驗傳承及分享機會。 本研究參考文獻及實務經驗，架構一套符合煉油石化產業特性之製程安全事故資料庫共計28欄位，藉以強化現有資料庫之完整性及實用價值。依此分類重新審視國內某大型煉油石化公司近12年20件製程安全事故，以頻率分析法統計事故特性，發現事故最常在正常操作狀態下，但人員心態較易產生浮動之午休、換班/下班期間發生；並應特別重視油料泵送過程，由於管路系統或泵浦/壓縮機損壞或未正常運轉下，導致高濃度氫氣之燃料油/循環氣等內容物洩漏，造成自燃火災之可能事故情境，事業單位可參考作為事故預防之方向。 20件案例採用預設邏輯樹模式，架構事故情境及剖析事故導因，藉由全方位事故導因表鑑別，各列舉關鍵及次要導因共計60項，以「四象限分析法」進行統計分析。本研究發現，設備因素為製程安全事故優先應重視之事故導因，包括現場人員設備操作、檢查及維修等直接執行層面，操作期間預知保養等間接管理層面，及技術設計和建構等基本源頭管控層面等，具有顯著之事故發生頻率及重要性，與歐盟 MARS石化事故資料庫具有相同事故導因趨勢，值得事業單位借鏡及作為防範對策之參考。

Crude oil refining processes or refineries are classified as high risk operations because of their potential for major accidents such as fire and explosions. Hence incident investigation has been an integral part of process safety management regulations. It is a regulatory requirement for petrochemical and refining companies to conduct thorough accident investigation after an unfortunate event to identify the root causes and to take corrective actions to prevent the same and similar accidents from happening again. However, recent accidents occurred in Taiwan involving the petrochemical and refining processes revealed the unfortunate fact that the lessons learned mentality or practices is nonexistent. This is primarily due to the fact that Occupational Safety and Health Act of the Ministry of Labor do not specifically differentiate occupational injuries from chemical process or manufacturing process related accidents. Secondly, the mandatory requirement on major accident reporting or the monthly occupational injury report requires no information on the root causes or main contributors of incidents or accidents. This study is aimed to design an accident database suitable for the refining industry. This is accomplished by referencing major accident databases of the United States, Japan, the European Union and 20 process-related accident reports compiled by a domestic petroleum company. Information gathered from these accidents is further classified into 60 major causal factors and minor causal factors. These factors are further analyzed by using British Petroleum’s Comprehensive List of Causes and classified as direct causes, indirect causes and root causes. The Four Quadrant Analysis technique is utilized to identify the major causal factor contributor of the 20 accidents is equipment related such as design, maintenance management such as preventive maintenance, inspection and testing, and operation. The findings are similar to those of the European Union Major Accident Reporting System. The frequency analysis reveals the fact that most of the accidents occurred during lunch break, or shift change. The modes of operations involved in these accidents were pumping, normal operations. Piping and pressurizing systems were the most common equipment encountered in the accidents. And the most frequently occurred accident was fire. It is hoped that the major causal factors and minor causal factors contributing to the accidents of the domestic petroleum company can alleviate the consequences of limited compilation, analysis, dissemination and utilization of accident information within and between companies.