由于过程风险态势高、安全管控难度大，化工装置易发生非计划停车停产、有毒有害物质泄漏以及火灾爆炸等事故，给社会、经济和环境造成了巨大的负面影响。事故致因模型是解决事故情景构建与推演问题的重要工具，但传统模型无法全面捕捉系统要素间的非线性交互关系或者拘泥于系统要素错误的原因分析及其消除措施的有效制定。因此，本工作引入功能共振分析方法(FRAM)，从整体系统的功能特征角度对化工装置事故情景开展推演研究。FRAM通过识别和描述基本功能、评估各功能的性能变化、确定功能共振的可能性以及制定性能变化的响应措施等步骤，能够深入回溯事故发生过程和演化情景，从而揭示化工装置事故的致因机理、挖掘化工装置的薄弱环节。以BP德克萨斯炼油厂爆炸事故为例，研究结果表明，该事故的主要原因是液位计B2、高液位报警器B3、加热炉S2和换热器S3产生了功能共振，从而引发了8种共振影响因素，并导致了12种功能失效连接。与时间与事件序列图(STEP)、事故地图(AcciMap)的分析结果进行了对比，验证了FRAM用于化工装置事故情景推演的可行性、有效性和合理性。

There are more and more accidents in the chemical plants due to high risk state and difficult safety management, such as unplanned shutdown, serious leakage of toxic and harmful chemicals, fire and explosion, and so on. These accidents have a huge negative impact on the society, economy, and environment. Accident causation model is an important technique for accident scenario construction and deduction; however, those traditional models cannot fully capture the dynamic and nonlinear interaction between system elements, or precisely stick to the cause analysis of system element errors and effective development of elimination measures. Therefore, the functional resonance analysis method (FRAM) was introduced to deduce the potential accident scenario of chemical plants from a perspective of systematical function features. There were four major procedures in the FRAM, including the identification and description of basic functions, the performance change assessment of every function, the possibility determination of functional resonance, and the response measure development of performance change. To illustrate its validity, the BP Texas City refinery explosion was selected as a test case in this work. Particularly, each functional performance change and functional failure connection of the raffinate section were clearly illustrated in the functional network diagram. It mainly involved the raffinate splitter tower, heating furnace, heat exchanger, blowdown drum and stack, different kinds of instruments, as well as operators. Results from the above-mentioned accident showed that this method was competent to deeply trace the accident occurrence process and evolution scenario, which could be further applied to reveal the cause mechanism of chemical plant accidents and excavate the weak points of chemical plants. Moreover, the FRAM was also compared with two other traditional methods-sequentially timed events plotting (STEP) and accident map (AcciMap). In summary, the introduced FRAM was verified to be feasible, effective, and reasonable for the accident scenario deduction of chemical plants.