Accepted author manuscript, 2.34 MB, PDF document
Available under license: CC BY: Creative Commons Attribution 4.0 International License
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Research on concrete early shrinkage characteristics based on machine learning algorithms for multi-objective optimization
AU - Wang, J.
AU - Liu, H.
AU - Sun, J.
AU - Huang, B.
AU - Wang, Y.
AU - Zhao, H.
AU - Saafi, M.
AU - Wang, X.
PY - 2024/7/15
Y1 - 2024/7/15
N2 - Cracking phenomena in tunnel side wall structures (TSWS) increasingly jeopardize their longevity due to water leakage, reinforcement corrosion, and eventual collapse. The primary contributor, early-age shrinkage (EAS) induced by hydration reactions, significantly undermines structural stability and durability. The integration of expansion agents (EA) and fibers presents a low-cost, efficient strategy to counteract EAS-induced cracking. Despite its promise, limited research on the influencing factors constrains its broader application. This study delves into the impacts of EA content, the CaO–MgO ratio, and fiber reinforcement on flexural strength (FS), compressive strength (CS), and EAS, revealing a complex interplay where EA and CaO content detrimentally affect mechanical properties yet beneficially influence EAS. Results showed that EA and CaO content had negative effects on the mechanical properties, but had positive effect on EAS. Additionally, Random Forest (RF) was developed with hyperparameters refined via the firefly algorithm (FA) based on the experimental data. The validity of the built RF-FA models was verified by substantial correlation coefficients and low root-mean-square errors. Subsequently, a coFA-based firefly algorithm (MOFA) was proposed to optimize tri-objectives of mechanical properties, EAS, and cost simultaneously. The Pareto fronts were obtained effectively for the optimal mixture design. This study contributes to its practical implications, offering a scientifically grounded approach to enhancing TSWS concrete design for improved performance and durability.
AB - Cracking phenomena in tunnel side wall structures (TSWS) increasingly jeopardize their longevity due to water leakage, reinforcement corrosion, and eventual collapse. The primary contributor, early-age shrinkage (EAS) induced by hydration reactions, significantly undermines structural stability and durability. The integration of expansion agents (EA) and fibers presents a low-cost, efficient strategy to counteract EAS-induced cracking. Despite its promise, limited research on the influencing factors constrains its broader application. This study delves into the impacts of EA content, the CaO–MgO ratio, and fiber reinforcement on flexural strength (FS), compressive strength (CS), and EAS, revealing a complex interplay where EA and CaO content detrimentally affect mechanical properties yet beneficially influence EAS. Results showed that EA and CaO content had negative effects on the mechanical properties, but had positive effect on EAS. Additionally, Random Forest (RF) was developed with hyperparameters refined via the firefly algorithm (FA) based on the experimental data. The validity of the built RF-FA models was verified by substantial correlation coefficients and low root-mean-square errors. Subsequently, a coFA-based firefly algorithm (MOFA) was proposed to optimize tri-objectives of mechanical properties, EAS, and cost simultaneously. The Pareto fronts were obtained effectively for the optimal mixture design. This study contributes to its practical implications, offering a scientifically grounded approach to enhancing TSWS concrete design for improved performance and durability.
KW - CaO content
KW - Early age shrinkage
KW - Expansion agent
KW - Machine learning
KW - Mechanical properties
KW - Multi-objective optimization
KW - Cracks
KW - Durability
KW - Expansion
KW - Learning algorithms
KW - Magnesia
KW - Mean square error
KW - Reinforced concrete
KW - Shrinkage
KW - Stability
KW - Early age shrinkages
KW - Firefly algorithms
KW - Machine-learning
KW - Multi-objectives optimization
KW - Random forests
KW - Shrinkage characteristic
KW - Side walls
KW - Wall structure
KW - Compressive strength
U2 - 10.1016/j.jobe.2024.109415
DO - 10.1016/j.jobe.2024.109415
M3 - Journal article
VL - 89
JO - Journal of Building Engineering
JF - Journal of Building Engineering
SN - 2352-7102
M1 - 109415
ER -