article · Results in Engineering
Circular hollow sections are increasingly utilised in construction due to their favourable mechanical properties and aesthetic appeal. Research into the structural behaviour of cold-rolled circular hollow section beam-columns, manufactured from both normal and high-strength steel, led to the development of a predictive machine learning tool. A validated finite element model generated a dataset of 3,410 numerical simulations across key structural parameters, supplemented by 13 physical test results from existing literature. These data were used to train and validate an artificial neural network model. The resulting design formula offers an accurate method for predicting cross-sectional load-carrying resistance with minimal computational effort, demonstrating superior performance when compared against standard Eurocode 3 design rules.
Accurately determining the load-bearing capacity of steel components is crucial for designing safe, cost-effective buildings. Circular hollow sections are increasingly popular for both strength and architectural appearance. By replacing slow or conservative conventional calculations with a fast, machine learning-derived formula, structural designers can assess member safety and performance with greater precision.
The primary application is within structural engineering design, where the formula could be incorporated into commercial design software or modernised building specifications. Structural engineers and building software developers are the main prospective users. The tool is at an applied and tested stage, validated against extensive numerical and experimental benchmarks and ready for implementation into engineering design procedures.
AI-generated from the published abstract. Always read the original work before citing.
The use of circular hollow sections (CHS) have seen a large increase in usage in recent years mainly because of the distinctive mechanical properties and unique aesthetic appearance. The focus of this paper is the behaviour of cold-rolled CHS beam-columns made from normal and high strength steel, aiming to propose a design formula for predicting the ultimate cross-sectional load carrying capacity, employing machine learning. A finite element model is developed and validated to conduct an extensive parametric study with a total of 3410 numerical models covering a wide range of the most influential parameters. The ANN model is then trained and validated using the data obtained from the developed numerical models as well as 13 test results compiled from various research available in the literature, and accordingly a new design formula is proposed. A comprehensive comparison with the design rules given in EC3 is presented to assess the performance of the ANN model. According to the results and analysis presented in this study, the proposed ANN-based design formula is shown to be an efficient and powerful design tool to predict the cross-sectional resistance of the CHS beam-columns with a high level of accuracy and the least computational costs.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1016/j.rineng.2023.100902
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
