A REVIEW ON SELF-HEALING AND STRUCTURAL PERFORMANCE OF FIBER-REINFORCED CONCRETE: TOWARD AN EXPERIMENTAL AND ANN-BASED PREDICTIVE FRAMEWORK

Abstract

Concrete is one of the most widely used materials in the world in terms of volume. Approximately 25 billion tons of concrete production occurs annually around the globe. Despite having superior compressive strength and multifunctionality, concrete is still prone to cracking due to plastic shrinkage, thermal gradients, freeze/thaw cycles, or continuous loads. Once formed, cracks become channels for corrosive elements such as chlorides, sulfates, or CO2 in the air. This results in faster reinforcing bar corrosion and reduces concrete service life, leading to premature failures, much earlier than the initial calculations suggested. Currently employed measures, such as manual patching, chemical injection of epoxy, cathodic protection, are costly, laborious and sometimes simply impossible to implement in confined spaces or underwater constructions.

Self-healing concrete (SHC) is considered a bio-mimetic material that has the ability to autonomously heal its cracks without any external actions. In the last decade, significant interest has emerged in researching SHC in academia and industry. This literature review narrows down the field of studies by considering SHC healing methods in conjunction with steel fiber reinforcement in FRCCs. Specifically, it analyzes the principles behind autogenous and autonomous healing of concrete composites, mechanisms through which steel fibers contribute to fracture healing, carbonate precipitation using microorganisms and enzymes, and Artificial Neural Network (ANN) prediction of the structural behavior of FRCCs. As a result, four knowledge gaps have been identified within the cross-disciplinary approach outlined above. They include the non-optimized interaction between silica fume and steel fibers in HPC, influence of continuous loads on healing efficiency in the FRCC context, need for an accurate engineering-oriented ANN model predicting healing potential based on mix design and load history, as well as lack of repeated crack-heal cycles testing under harsh environmental conditions.

The proposed research project addresses these gaps with an experimental approach and ANN modeling focusing on self-healing of M50 High Performance Concrete with silica fume in a range of 50-100% replacement. Hooked-end steel fibers are introduced in order to reinforce the matrix with fibers and promote crack development under constant loads. Healing efficiency and residual strength prediction are modeled with the help of feedforward backpropagation artificial neural network.