Pipeline is now a commonly-used transportation mode for hazardous liquid, whereas followed by frequent pipeline leakage accidents. Research on the leakage post-assessment of hazard liquid pipeline has received increasing attention, but the in-situ data are hard to be accurately captured, resulting in a series of uncertainties affecting the accuracy and practicality of the risk assessment. This paper puts forward a novel method, which is able to accurately achieve leakage detection, cause analysis and leakage volume forecast by avoiding deviation of in-situ data, model parameters and the uncertainties caused by method, thereby quickly assessing the impact on the surrounding environment. The Markov Chain Monte Carlo (MCMC) algorithm is employed for repeatedly sampling leakage position and coefficient, furthermore, the transient hydrothermal and the leakage risk assessment model are established for determining the leakage volume and the risk grade. The influence of real-time measurement data and the deviation of the method are taken into consideration through the frequency distribution statistics of numerous sampling data. Based on two real examples, it is verified that the risk assessment method has practical value for the in-situ analysis and emergency treatment.

管道现在是用于危险液体的常用运输方式，而随之而来的是频繁发生的管道泄漏事故。危险液体管道泄漏后评估的研究受到越来越多的关注，但是现场数据难以准确捕获，导致一系列不确定性影响了风险评估的准确性和实用性。提出了一种新颖的方法，该方法能够避免原位数据，模型参数和方法带来的不确定性的偏差，从而准确地实现泄漏检测，原因分析和泄漏量预测，从而快速评估对周围环境的影响。马尔可夫链蒙特卡罗（MCMC）算法用于重复采样泄漏位置和系数，此外，建立了暂态热液和泄漏风险评估模型，用于确定泄漏量和风险等级。通过大量采样数据的频率分布统计，考虑了实时测量数据的影响和方法的偏差。通过两个实例，验证了该风险评估方法对现场分析和应急处理具有实用价值。