A sophisticated computational approach for flexural strength prediction and mix proportion optimization of steel fiber-reinforced concrete

Document Type

Article

Publication Date

1-1-2026

Abstract

Accurately predicting the load-bearing performance of steel fiber-reinforced concrete (SFRC) and improving its sustainability remains a critical challenge in civil engineering. This study addresses these challenges via a soft computing framework that integrates Bayesian model updating, lifecycle evaluation and multi-objective optimization at the material scale. Based on 195 flexural peak strength tests, Bayesian updating produced predictive formulations incorporating key internal mix parameters. Subsequently, life cycle assessment coupled with a non-dominated sorting evolutionary method was employed to evaluate sustainability indicators and optimize mix designs, balancing mechanical performance, cost and carbon emissions. Principal observations are as follows: (i) The five identified variables strongly affect SFRC's flexural strength. Maximum gravel size and aggregate-to-cement ratio negatively influence strength, whereas cement content, sand ratio and fiber reinforcement coefficient exert positive effects. (ii) Bayesian model updating, with or without prior regression models, outperforms multiple linear regression, improving flexural strength prediction accuracy by 6.2 %. (iii) Increasing aggregate to cement and sand ratios reduces production costs and carbon emissions without compromising strength, while maximum aggregate size has minimal impact. Higher fiber content, however, substantially raises both cost and emissions, necessitating careful mix design. (iv) Optimized mix proportions meet target strength while cutting carbon emissions and costs by 30 %–50 %. Back-calculation indicates steel fibers are unnecessary below 6.0 MPa, with optimal dosages of 1 % for 10 MPa and 2 % for 12 MPa. Overall, the proposed soft computing framework effectively guides SFRC mix design and promotes material sustainability.

Keywords

Bayesian model updating, Flexural strength, Life-cycle assessment, Mix proportion optimization, Steel fiber reinforced concrete

Publication Title

Applied Soft Computing

ISSN

15684946

DOI

10.1016/j.asoc.2025.114181

Volume

186

Share

COinS