Session: 06-06-03 Pipe/Soil Interactions

Paper Number: 134243

134243 - Strain Demand Prediction Model of Buried Pipeline Subjected to Tectonic Fault Displacement Via Deep Learning Models 

Abstract: 

The geological hazard such as normal fault, reverse fault, strike-slip fault is one of the primary threats to the integrity and the safe operation of the current buried in-service pipeline in China, which triggers the fault-induced ground movement often leading to the large-scale deformation and serious tensile and compressive failure on pipeline which will cause serious economic losses and social disruptions, so it is of great significance to carry out the precise pipeline mechanical response for effective assessments of safety status prediction. Therefore, the accurate numerical simulation model of pipeline crossing different geological hazards area was established based on the nonlinear finite element method software ABAQUS with the consideration of the elastoplastic property of high-grade X80 pipe steel. With the combination of ABAQUS and Python co-simulation, the pipeline under different working condition such as different geological hazard types, pipeline geometric dimensioning including pipeline diameter and wall thickness, soil surface movement, pipeline inner pressure and soil type was formulated and the maximum strain of pipeline database for machine learning and deep learning models’ training and validation dataset was formed according to the database above. Base on this database, the deep learning models including the Extreme Gradient Boosting (XGBoost), Auto machine learning (Auto ML), Categorical Boosting (CatBoost) and Residual Neural Network (ResNet) model were trained and employed to effectively acquire the accurate prediction of the maximum strain of pipeline crossing different geological hazard area under various engineering scenarios. This comprehensive analysis also delved into different influencing factors that could influence the pipeline maximum strain. The results show that the finite element - machine learning/deep learning conjoint analysis method has the significant advantages of high prediction accuracy and low time cost. This research serves as a valuable reference and useful guide for the safety state analysis of pipeline under different geological hazard and various engineering scenarios.

Presenting Author: Hao Wang China University of Petroleum, Beijing

Presenting Author Biography: Dr. Wang has a solid knowledge of solid mechanics, metal thermoplastic deformation and fracture theory, proficient in 3D finite element simulation and analysis methods, and proficient in using a variety of finite element analysis software (ABAQUS, ANSYS, MOOSE). Proficient in various experimental methods for microstructure characterization of materials (focused ion beam, scanning electron microscope, transmission electron microscope, electron backscattering diffraction, X-ray diffraction, atomic force probe). I have a solid theoretical knowledge of statistics and am skilled in using various software for data statistics and analysis (JMP, MATLAB, Python). Familiar with multiple programming languages (Python,C++,MATLAB,Fortran), experience in team software development and maintenance.

Authors:

Mengkai Fu China University of Petroleum, Beijing
Hong Zhang China University of Petroleum, Beijing
Dong Zhang China University of Petroleum, Beijing
Jia Shao China University of Petroleum, Beijing
Luyao Bai PipeChina Institute of Science and Technology
Pengchao Chen PipeChina Institute of Science and Technology
Hao Wang China University of Petroleum, Beijing
Xiaoben Liu China University of Petroleum, Beijing