(411c) Data Analytics for Designing Fe-9Cr Steels (Tensile Strength)

Romanov, V. N. - Presenter, National Energy Technology Laboratory (NETL), Research and Innovation Center, Department of Energy
Hawk, J. A., National Energy Technology Laboratory (NETL), Research and Innovation Center, Department of Energy
Motivation for this research comes from the desire to shorten the rigorous and time-consuming alloy qualification (standardization) process, for new fossil energy materials applications. The main consideration for using 9 to 12% Cr martensitic-ferritic steels is their relatively high microstructural stability at the operating temperature over time, since power plants (e.g., steam boiler and turbine applications) have a design lifetime expectation of over 30 years. A materials data analytics (MDA) methodology was developed in this study to evaluate publicly available information on 9% Cr family steel and to handle nonlinear relationships and the sparsity in materials data for this alloy class. Some of the materials research and development activities are proprietary, which makes it particularly difficult to access and compile high-quality information. Data entries in the analyzed data set for over 90 iron-base alloy compositions, several processing parameters, and results of tensile mechanical tests selected for this study were arranged in 47 columns by 2800+ rows. To address non-linearity in the mechanical properties, data analyses were initially carried out in composition-based clusters. A new algorithm was developed for transparent clustering, c-IG. The modified c-IG using Chebyshev norm produced composition clusters matching the industry classification, without dimensionality reduction. Partitioning revealed the biased nature of available alloy datasets, with implied “rules-of-thumb” in design practices. Processing and test temperature effects on the phases and mechanical properties are the major contributors to predicting the test outcome and need to be modeled first, before searching for minor effects of the composition variations. The evolutionary ensemble models based on the clustering without dimensionality reduction provided a transparent link to the domain knowledge base. While the benchmark linear models (SciPy and TensorFlow) explained less than 50% of the variation in the test data due to controlled alloy design parameters, the ensemble modeling explained over 93% of such variation, for the ultimate tensile strength.


This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.


Do you already own this?



AIChE Members $150.00
AIChE Graduate Student Members Free
AIChE Undergraduate Student Members Free
Non-Members $225.00