Title:
Optimization of Fiber-Reinforced Concrete Using Data Mining
Author(s):
Emilio Garcia-Taengua - University of Leeds
Publication:
CRC
Volume:
Issue:
Appears on pages(s):
164
Keywords:
DOI:
Date:
12/15/2023
Abstract:
A database of SFRC mixtures and a database of synthetic FRC mixtures have been compiled from papers published over the last two decades and used as the basis for this study. The information compiled on SFRC mixtures constitutes a coherent set of data where the main sources of variability are clear, providing the basis for the development of robust models and conclusions. This has proven more challenging in relation to synthetic FRC mixtures. It has been demonstrated that the mix design variables representing the binder composition and the aggregates content and combined grading explain 49% of the differences in residual flexural strength parameters, as opposed to the 51% explained by the variables directly related to the fibers. This proves that the residual flexural capacity of FRC is heavily influenced by variables other than the fibers and supports the idea that proportioning FRC mixtures mainly by modifying the fiber content is not a good approach. The variability of the residual flexural strength parameters has also been studied, and equations have been obtained to estimate their standard deviation as a function of the mixture proportions, the fiber dimensions, and the test/specimen configuration (3PBT or 4PBT). The mixture proportions and the dimensions of the fibers have a significant effect on the intrinsic variability of the residual flexural strength parameters. The equations obtained, jointly with the model for the residual flexural strength parameters, makes it possible to maximize their value and minimize their variability at the same time. The standard deviation of the residual flexural strength results obtained from testing notched specimens in 3PBT conditions can be up to 49% higher than that of results obtained from unnotched specimens in 4PBT conditions. Generally speaking, the coefficient of variation of the residual flexural strength parameters can typically range from 30% to 50%.