Chemical EngineeringChementator briefs
CO2 Separation Conventional DDR-type zeolite membranes are well suited for…

Comment IIOT Power

Predicting aluminum-alloy mechanical properties at high temperatures

By Gerald Ondrey |

Aluminum is used for a number of applications because it is lighter than iron and easy to machine. However, Al is usually alloyed with Cu, Mg or other elements for improved strength. Developing such alloys that maintain their strength at high temperatures (over 100°C) takes a lot of time, because it requires developers’ knowledge-rich experience and performing many analysis and evaluations.

Aiming to solve these problems, Showa Denko K.K. (SDK; Tokyo, Japan; www.sdk.co.jp) has been participating in a project under Japan’s Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program. In this project, SDK, the National Institute for Material Science (NIMS; Tsukuba, Japan) and the University of Tokyo have collaboratively developed a computer system using neural networks — an artificial-intelligence algorithm — to accelerate the development of materials with optimal mechanical properties.

The researchers focused on 2000-series Al alloys (those that contain Cu and Mg), and utilized design data of 410 records of Al alloys listed in public databases, including the Japan Aluminum Assn., and developed neural network models (NNMs) that accurately predict the strength of Al alloys at various temperatures (from room temperature to over 200°C). In addition, the architecture and parameters of the neural network was optimized with Bayesian inference by applying the replica-exchange Monte Carlo method. This NNM can estimate — within 2 seconds — the alloy’s strength under 10,000 different conditions.

An “inverse design tool” was also developed that suggests a set of Al-alloy design conditions that maximizes the probability of satisfying the desired strength at an arbitrary temperature. The models are said to shorten development time for Al alloys to about one-half to one-third of that required with conventional development methods.

Detail of the study were presented last month at a virtual session of the 2021 Materials Research Society’s Fall meeting (December 6–8, 2021; www.mrs.org).

Related Content
Implementing digital transformation
The recently concluded 5th annual Connected Plant Conference (CPC; www.connectedplantconference.com), held in Austin, Tex., illustrated both the real promise of…

Chemical Engineering publishes FREE eletters that bring our original content to our readers in an easily accessible email format about once a week.
Subscribe Now
SICK Solutions for Cleaner Industries - Powerful Transitions
Gain a Digital Line of Sight Across the Whole Lifecycle of the Plant with a Digital Twin
Granulating High Viscous Melts in the Food Industry
How to Select a Pump for Industrial Applications
Temperature Instruments Improve Operations

View More