Mapping Multivariate Influence of Alloying Elements on Creep Behavior for Design of New Martensitic Steels
- PDF / 1,212,615 Bytes
- 15 Pages / 593.972 x 792 pts Page_size
- 59 Downloads / 261 Views
CTION
SIXTY-THREE percent of U.S. electricity generation comes from fossil fuels (coal, natural gas, petroleum, and other gases),[1] of which 48 pct is from coal. Energy security initiatives in 21st century combined with design requirements to lower carbon dioxide (CO2 ) emissions, are pushing the adoption of advanced ultra-supercritical (A-USC) coal power plants with increased steam temperature that promise to increase efficiency.[2] For example, an increase in steam parameters from 566 °C and 24 MPa to 650 °C and 34 MPa would amount to an increase in relative efficiency of 6.5 pct, resulting in a significant decrease in coal use, and hence the reduction of CO2 emissions.[3] Under the proposed A-USC standards,[2] 9 to 12 wt pct Cr martensitic steels are being considered for thick sections such as the main steam pipe
AMIT K. VERMA and JENNIFER L.W. CARTER are with the Department of Materials Science and Engineering, Case Western Reserve University, Cleveland, OH 44106. Contact e-mail: [email protected] JEFFERY A. HAWK is with the National Energy Technology Laboratory, Albany, OR 97321-2198. LAURA S. BRUCKMAN and ROGER H. FRENCH are with the Department of Materials Science and Engineering, Case Western Reserve University and also with the SDLE Research Center, Case Western Research University, Cleveland, OH 44106. VYACHESLAV ROMANOV is with the National Energy Technology Laboratory, Pittsburgh, PA 152360940. Manuscript submitted November 28, 2018. Article published online April 29, 2019 3106—VOLUME 50A, JULY 2019
and header in the boiler which will operate up to 650 °C. Within conventional supercritical coal-fired power plants, martensitic steels already operate up to 610 °C. A 40 °C increase in operating temperature requires development of new martensitic steel alloys with higher creep rupture strengths. Steel alloy discovery today relies on an Edisonian approach with intuitive design choices primarily focused on ppm alloy additions. The 2016 paper by Abe[4] exemplifies this process, where improvement was achieved by alloy additions of V-Nb,[5] later by substituting a part or all of Mo with W,[6] and recently by the addition of Co[7] and B.[8] Initiatives such as Materials Genome Initiative (MGI)[9] and Integrated Computational Materials Engineering (ICME)[10,11] came into prominence in last decade as methodologies to reduce the materials discovery timeline. Aligned with the MGI, this paper presents a data-enabled science approach, focused on leveraging machine learning tools,[12] to improve the creep properties of 9 to 12 wt pct Cr martensitic steels through alloy selection. The aim is to guide the next phase of experiments by gaining insights from previously completed experiments to effectively reduce the time and cost for materials discovery. Supervised learning is a sub-field of machine learning, which can be broadly defined as any computer program that improves its performance at some task through experience.[13] The metal design machine learning problem can be outlined as: task—search the compositional space
Data Loading...
