Combustion or explosion accident resulting from accidental hydrocarbon release poses a severe threat to the offshore platform's operational safety. Much attention has been paid to the risk of an accident occurring over a long period, while the real-time risk that escalates from a primary accident to a serious one was ignored. In this study, a real-time risk assessment model is presented for risk analysis of release accidents, which may escalate into a combustion or explosion. The proposed model takes advantage of Fault Tree-Event Tree (FT-ET) to describe the accident scenario, and Bayesian network (BN) to obtain the initial probability of each consequence and describe the dependencies among safety barriers. Besides, Computational Fluid Dynamics (CFD) is applied to handle the relationship between gas dispersion and time-dependent risk. Ignition probability model that considering potential ignition sources, gas cloud, and time series are also integrated into this framework to explain the likelihood of accident evolution. A case of release accidents on a production platform is used to test the availability and effectiveness of the proposed methodology, which can be adopted for facilities layout optimization and ignition sources control.

因意外释放碳氢化合物而引起的燃烧或爆炸事故严重威胁着海上平台的操作安全。长期以来发生事故的风险已引起了很多关注，而从主要事故升级为严重事故的实时风险却被忽略了。在这项研究中，提出了实时风险评估模型，用于释放事故的风险分析，该事故可能升级为燃烧或爆炸。所提出的模型利用故障树-事件树（FT-ET）来描述事故场景，并利用贝叶斯网络（BN）来获取每种后果的初始概率并描述安全障碍之间的依存关系。此外，应用计算流体动力学（CFD）来处理气体扩散与时间相关风险之间的关系。考虑到潜在点火源，气体云和时间序列的点火概率模型也被集成到该框架中，以解释事故演变的可能性。在生产平台上发生释放事故的情况用于测试所提出方法的有效性和有效性，该方法可用于设施布局优化和点火源控制。