Investigation on the leak accident prediction of petrochemical installations

摘要

Petrochemical installations are generally of continuous, large-scale, complex and high operation parameters features. As safety problems were hidden in each part, a slight mistake may lead to leak, fire, explosions and toxic pollution. Based on detailed analysis of main factors causing failure of petrochemical installation, a leak accident prediction model of typical petrochemical installation was brought out with the help of LEAK software designed by DNV. The model was then applied to simulate and calculate the leak frequency of a demethanizer and pipelines connected with it. Through the simulation, not only were the reasonable basic leak information of the assessed system including leak frequency of relative hole size distribution, representative leak hole size and mean leak hole size acquired, but also a suggested leak grade method was put forward according to relevant national and industrial standards. Meanwhile, the leak inspection and maintenance programs were improved. This application shows that the leak accident prediction model of typical petrochemical installation is of great feasibility and rationality, which is also of reference value in developing reasonable and well- targeted maintenance programs and optimizing allocation of leak inspection resources.