Abstract:Low carbon and low alloy steel pipes, prevalent in natural gas transmission systems due to their affordability, weldability, and strength, are confronted with significant challenges such as hydrogen embrittlement (HE) when transitioning towards hydrogen energy systems. This necessitates innovative predictive strategies to overcome these hurdles. Most existing research on HE has focused on a limited range of low carbon and low alloy steels under specific experimental conditions and has been constrained by the limitations of experimental facilities. To expand the research scope, our study incorporated seven machine learning (ML) techniques: Random Forest (RF), Decision Tree (DT), Extreme Gradient Boosting (XGBoost), Gradient Boosting (GB), Adaptive Boosting (AdaBoost), Categorical Boosting (CatBoost), and Artificial Neural Networking (ANN). The aim is to predict the HE in terms of the degradation of mechanical properties, specifically the reduction in area. Drawing from tensile test data obtained from 47 distinct low carbon and low alloy steels under pressurized hydrogen gas conditions, we constructed and evaluated a range of ML models, with the aim of identifying the most efficacious one for our study. Our results indicated that the CatBoost ML model offered the best prediction of the reduction in the area of these steels in a hydrogen environment. The CatBoost model provided a low Mean Absolute Error (MAE) of 7.32, Mean Square Error (MSE) of 83.78, Root Mean Square Error (RMSE) of 9.15, and a coefficient of determination (R 2 ) value of 77.62% for the training data and 72.50% for the testing data. Furthermore, the CatBoost model identified hydrogen gas pressure and the steel's ultimate tensile strength as the most influential parameters, contributing 47.4% and 19.2% respectively to the prediction of HE. This research provides valuable insights into the development of advanced materials and infrastructure for efficient hydrogen gas transportation, supporting the broader shift towards a hydrogen-based energy economy.

Multiscale Materials Modelling through Machine Learning: Hydrogen-Steel Interaction during Deformation

Hybrid Data-Driven Deep Learning Framework for Material Mechanical Properties Prediction with the Focus on Dual-Phase Steel Microstructures

Explainable Machine Learning for Hydrogen Diffusion in Metals and Random Binary Alloys

Modeling of necking area reduction of carbon steel in hydrogen environment using machine learning approach

Application of Machine Learning to Predict the Mechanical Properties of High Strength Steel at Elevated Temperature

Exploring nonlinear strengthening in polycrystalline metallic materials by machine learning methods and heterostructure design

The Use of Machine-Learning Techniques in Material Constitutive Modelling for Metal Forming Processes

A machine-learning-enhanced hierarchical multiscale method for bridging from molecular dynamics to continua

Data-driven physics-constrained recurrent neural networks for multiscale damage modeling of metallic alloys with process-induced porosity

A machine learning framework to predict local strain distribution and the evolution of plastic anisotropy & fracture in additively manufactured alloys

Advances in machine learning- and artificial intelligence-assisted material design of steels

Prediction of Ductile Damage in Composite Material Used in Type IV Hydrogen Tanks by Artificial Neural Network and Machine Learning with Finite Element Modeling Approach

A State-of-the-Art Review on Machine Learning-Based Multiscale Modeling, Simulation, Homogenization and Design of Materials

Predictive machine learning approaches for the microstructural behavior of multiphase zirconium alloys

Coupling Physics in Machine Learning to Predict Properties of High-temperatures Alloys

Prediction of grain refinement using multiscale crystal plasticity-integrated dislocation density-based model in multiphase steel alloys

Fusion of experimental and synthetic data for reliable prediction of steel connection behaviour using machine learning

Multi-scale hydrogen embrittlement prediction model of low alloy steel based on multi-dimensional defect reconstruction

Atomistic simulation assisted error-inclusive Bayesian machine learning for probabilistically unraveling the mechanical properties of solidified metals

Deep Learning for Multiscale Damage Analysis via Physics-Informed Recurrent Neural Network