Investigating the compression behavior of concrete‐filled double‐skin steel elliptical tubular columns by a fusion of finite element analysis and machine learning

Abstract This study comprehensively examined the behavior and performance of concrete‐filled double‐skin steel elliptical tubular columns (CFDSSETC) subjected to different loading scenarios. CFDSSETC are gaining attention due to their potential to offer enhanced structural efficiency and architectural versatility compared to traditional columns. This research uses non‐linear finite element analysis and machine learning (ML) to assess the load‐carrying capacity of CFDSSETC under axial and eccentric compression. To do this, ABAQUS software and data from previous research were used to generate finite element models (FEMs) for eight columns. By expanding existing parameters, 172 more FEMs were developed in addition to these 8. Parameters such as eccentric loading ratio; area of concrete portion; outer width, outer depth, inner width, inner depth, and yield strength of internal steel tube; yield strength of external steel tube; and concrete strength of standard cylinder are systematically varied to evaluate their influence on the response of CFDSSETC. Additionally, nine ML models were developed to predict the CFDSSETC's load‐bearing capability under axial and eccentric compression utilizing the database that was acquired from FEM. This work provided a design technique for determining the load‐bearing capacity of short CFDSSETC that were subjected axial and eccentric compression. The outcomes revealed that raising the concrete's area, strength, and yield strength of the internal and external steel tubes as well as reducing the internal steel tube's depth or width and load eccentricity enhanced CFDSSETC's load‐carrying capacity. The support vector regressor demonstrated superior predictive performance among the diverse set of regression models considered. The suggested design formula has shown good prediction accuracy, with 99% confidence with the experimental and FEM findings. The findings provide valuable insights into the design and optimization of CFDSSETC for applications in civil engineering structures, contributing to the advancement of sustainable and resilient infrastructure systems.