Title :
Reliability-based structural integrity assessment of Liquefied Natural Gas tank with hydrogen blistering defects by MCS method
Author :
Hu, Jun ; Yan, Li-wei ; Liu, Fei ; Duan, Quan ; Zhang, Zao-xiao
Author_Institution :
Sch. of Energy & Power Eng., Xi´´an Jiaotong Univ., Xi´´an, China
Abstract :
Hydrogen blistering is one of the serious threats to safe operation of a Liquefied Natural Gas (LNG) tank, therefore safety analysis of hydrogen blistering defects is very important. In order to assess the reliability-based structural integrity of the LNG tank with defects of hydrogen blistering, the following steps were carried out. Firstly, Abaqus code, one of the Finite Element Method (FEM) software, was utilized to calculate 100 J-integral values of crack tip by defining directly. Secondly, the 100 J-integral values of crack tip were used as training data and testing data by Optimized Least Squares Support Vector Machine (OLS-SVM), Least Squares Support Vector Machine (LS-SVM) and Artificial Neural Networks (ANN) to get other 20000 J-integral values of crack tip. Finally, Monte-Carlo Simulation (MCS) was used to assess the reliability-based structural integrity analysis. The results showed that the hydrogen blistering defect with crack will propagate with about 14 percent chance in such a case. It also proved that MCS combined with FEM and SVM was an effective and prospective method for research and application of integrity assessment, which could overcome the data source problem.
Keywords :
Monte Carlo methods; cracks; finite element analysis; least squares approximations; neural nets; petroleum; reliability; structural engineering computing; support vector machines; tanks (containers); Abaqus code; J-integral values; Monte Carlo simulation; artificial neural networks; crack defect; finite element method; hydrogen blistering defects; least squares support vector machine; liquefied natural gas tank; optimized least squares support vector machine; reliability-based structural integrity assessment; Artificial neural networks; Finite element methods; Liquefied natural gas; Materials; Reliability theory; Support vector machines; J-integral; MCS; Structural integrity; Support Vector Machine;
Conference_Titel :
Natural Computation (ICNC), 2010 Sixth International Conference on
Conference_Location :
Yantai, Shandong
Print_ISBN :
978-1-4244-5958-2
DOI :
10.1109/ICNC.2010.5583691