Creep-Rupture Prediction of Inconel 617 Using a Python-Based Machine-Learning Approach

Abstract Creep rupture data is only sometimes readily available at the desired temperature or stress levels, and performing creep tests can be both time-consuming and expensive. Creep-rupture data from various sources are often combined for model calibration and validation. However, such combined data may overlap or exhibit a wide scatter band because of different metadata factors. A small change in chemical composition may affect the creep properties creating a large variation in the rupture data. Advances in data mining techniques make it possible to use machine learning to consider metadata such as the chemical composition of different heats in modeling for improved prediction. In this study, a Python-based machine-learning approach is applied to predict the creep rupture of alloy Inconel 617. Data from five different sources (General Electric Company (GE), Oak Ridge National Laboratory (ORNL), German HTGR, Huntington Alloy, and Korea Atomic Energy Research Institute (KAERI)) which encompasses multiple heats are used. Pearson Correlation Coefficient (PCC) and Spearman Correlation Coefficient (SCC) are employed to identify the dominant chemical elements and operating conditions (stress) influencing creep rupture. Six different regression methods (Random Forest Regression (RF), Linear Regression (LR), K-Nearest Neighbor (KNN), Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Regression (SVR), and Gradient Boosting Regression (GB)) are used for model training. The resulting prediction curve is validated against data not used in calibration. A detailed flow diagram elucidating the methodology is also provided.