Chemical parks are inherently high-risk environments characterized by complex production processes, densely clustered facilities, and large-scale storage of hazardous materials. Any accident may induce catastrophic domino effects, encompassing human casualties and serious economic losses. An effective accident causation analysis framework including root cause and causal chain is essential, which helps to enhance the overall safety performance of chemical parks. In this paper, we introduce an abductive analytical methodology based on the fault tree analysis (FTA). FTA employs Boolean logic to decompose top events into basic events, and quantifies failure probabilities using reliability data. And the novel method, multi-layered fault tree abductive model, hierarchically analyzes the top event, mechanism, protection barrier, state and human-machine-material-protocol-environment-evaluation (HMMPEE). Then a documented chemical accident is analyzed using the multi-layered fault tree abductive model. The comprehensive analysis systematically identifies accident mechanisms and their root causes, which helps to formulate the targeted and evidence-based recommendations for both accident prevention and systemic safety management.