|About this Abstract
||MS&T23: Materials Science & Technology
||Leveraging Integrated Computational Materials Engineering for High-fidelity Physics-based and Machine Learning Models
||Application of Machine Learning Framework in Predicting Creep Response of High Temperature Alloys
||Md Abir Hossain, Jacob T Pellicotte, Calvin M Stewart
|On-Site Speaker (Planned)
||Md Abir Hossain
This study explored the implementation of machine learning (ML) to build high throughput models that predict the creep response of Ni-enriched superalloy Inconel 718. For over a century, creep models were developed by human intuition and empirical knowledge leading to application-, condition-, and/or material-specific models. Machine learning has enabled a widespread improvement in the human ability to develop creep models and makes extrapolations to unprecedented timescales.
In this study, stress-rupture data across four heats of materials at temperature range 500-750 °C are gathered from NIMS. A randomized algorithm is employed to partition the data. A set of 20 different ML algorithms are trained with the training set and 5 best models are chosen. Based on the statistical rationale, the Wide Neural Network (WNN) is found to be the best ML algorithm. The ML framework provides a pathway to design next-generation alloys for elevated temperature applications.