Image analysis for dispersion of fibres and strength assessment of hybrid fiber reinforced SCC using artificial intelligence

Incorporating fibers in concrete improves its ductility, making it a viable option for large construction projects. The mechanical behavior of Fiber Reinforced Concrete is determined by the fibers ability to bridge cracks and their orientation in the fracture plane. Thus, assessing the distribution characteristics of the fibers is essential for predicting the strength of concrete. The dispersion of fibers plays a critical role in both fresh and hardened properties of concrete. This study investigates the influence of adjusting mix proportions on the orientation and dispersion of fibers in Hybrid Fiber Reinforced Self-Compacting Concrete (HFRSCC). The workability, compressive and split tensile strength of HFRSCC were initially tested for various fiber combinations, followed by image-based analysis and distance transform method to quantify fiber dispersion and orientation within the cement matrix. The dispersion factor, orientation factor, and mechanical test results were subsequently utilized to develop three machine learning models including artificial neural network (ANN), Support Vector Regression (SVR) and Linear Regression (LR). These models effectively predicted the compressive and split tensile strengths of HFRSCC, emphasizing the role of fiber distribution and orientation as critical input parameters. The results revealed that combining shorter and longer fibers significantly improved fiber dispersion, workability, and mechanical properties. The ANN model demonstrated high accuracy in strength predictions, highlighting the benefits of optimizing fiber proportions with varying aspect ratios. This study uniquely integrates experimental testing, image-based fiber analysis, and ANN modelling, focusing on fiber aspect ratios rather than only volume fractions.