Engineered cementitious composites (ECC) are widely used in masonry structure reinforcement, bridge deck connection and structural repair due to the strain hardening and multi-crack cracking characteristics, which can overcome some shortcomings as poor ductility and weak crack control ability from ordinary concrete. The existing mechanism analysis for concrete fracture mechanics is mainly based on the micro- and macro-experimental results. Although a research can obtain experimental data, the experimental process and result analysis are time-consuming and labor-intensive, and there is a huge deviation between the experimental results and the intended target. Computer simulation provides an effective research tool for accurately analyzing the working mechanism of cement-based composites. Computer simulation can accurately predict and deeply analyze the experiment results due to the advantages of high efficiency, visualization and low cost, which helps to better understand the achievement criteria of special properties for cement-based composites. The related research on the nano-, meso-, and macro-scopic cracking characteristics during the tensile process of ECC by computer simulation technologies in different scales becomes popular. However, there are a few systematic work for the characteristics, differences and correlations of different scale simulation methods on ECC.

In the nano-scale simulation of ECC, the molecular dynamics (MD) can analyze the influence mechanism of unmodified fiber and modified fiber on the bonding performance at fiber-matrix interface during ECC deformation from the molecular or atomic level. MD can be used to establish the fiber/C-S-H interface molecular model, which can visually display the fiber-matrix interface bonding, and reveal that the interface bonding between unmodified/modified fibers, and C-S-H gel can be achieved via forming polymer, electrostatic interaction, chemical bond, hydrogen bond and the van der Waals force.

In the mesoscopic scale simulation of ECC, the numerical (finite element) simulation can analyze the influence of macroscopic and mesoscopic factors such as fiber, matrix and their interface on the tensile strain hardening behavior of ECC. Firstly, the mesoscopic scale model-the lattice discrete particle model can analyze the influences fiber type and dosage. Secondly, crack propagation mode I can simulate the influence of matrix characteristics such as matrix internal pore structure and matrix fracture toughness. A numerical (finite element) model is established to simulate the single fiber pullout process to analyze the influence of fiber–matrix interface parameters on the tensile properties of ECC. The ECC tensile finite element model is proposed to analyze the influence of sample shape and size on the tensile properties of ECC.

In the macroscale simulation of ECC, there are a few studies on the multi-crack cracking characteristics of ECC from peridynamic (PD) simulation. The fiber-matrix interaction using PD is simulated to investigate the discontinuous cracking process of ECC, or predict the maximum tensile strain of fiber reinforced cementitious composites, indicating the feasibility of PD simulation in this aspect. In addition, this review introduces how to establish an artificial neural network (ANN) prediction model to achieve concrete mix ratio design, predict the correlation between relevant parameters and mechanical properties, and extend the ANN model to the prediction of ECC micro-mechanical properties indicators.

This review represents the research status of four types of simulation technologies according to nano-, meso-, and macro-scale in tensile cracking behavior of ECC. In the nanoscale simulation of ECC, MD can be used to analyze the bonding performance of fiber-matrix interface by virtue of its visualization advantage for molecular/atomic interaction. Using MD model in the molecular or atomic level for investigation of the interface interaction between unmodified/modified fibers and C-S-H gel is enhanced via forming hydrogen bonds, the van der Waals forces, etc., or filling nanopores and bridging nanocracks with nanomaterials, thus improving the tensile properties of ECC. However, the MD simulation on the fiber-matrix interface performance after ECC cracking and the crack propagation behavior is lack. For the mesoscopic scale simulation of ECC, the numerical (finite element) model based on meso-mechanics shows that the uneven dispersion of fibers can reduce the ultimate tensile strain and ultimate tensile strength of ECC, and show that the tensile strain and tensile strength of ECC decrease with the increase of sample width and thickness. However, there is a lack on numerical (finite element) simulation for the tensile strain of ECC from the perspective of the close packing of mortar matrix particles. In the macroscale simulation of ECC, PD can be used to solve the discontinuous cracking problem of micro-/macro-mechanics of cement-based materials. However, the stress transfer between fiber and matrix during ECC multi-cracks cracking and the prediction of crack propagation path after loading are rarely involved. The ANN model can predict the basic mechanical property indexes of ECC and the stress-strain curve of FRC, and optimize the concrete mix proportion. When it is combined with a genetic algorithm to predict the compressive strength, slump and interfacial bond strength of concrete, the prediction accuracy becomes greater. However, the ANN prediction model on ECC micro-mechanical indicators is not reported. In addition, it is also crucial for the improvement of the prediction accuracy and efficiency of computer models to explore how to achieve two-path intelligent optimization of input parameters and prediction targets.