Micro-structural and mechanical properties of PVA fiber reinforced engineered cementitious composite incorporating natural and artificial pozzolanic materials under different temperatures

Abstract

• The pozzolans' contents significantly affect the tensile strain capacity of the ECC. • Using fly ash, GGBS and zeolite can improve tensile behavior of the ECC. • The GGBS based ECC has better tensile stress–strain behavior. • At the temperatures above 50 °C, the mechanical properties of the ECC decrease. • ANN and MLR can predict the mechanical properties of the ECC with high accuracy. Engineered cementitious composites (ECC), which have high toughness and tensile ductility, as well as multiple fine cracking, satisfy the high safety and durability criteria in the development of sustainable infrastructures. The effect of different temperatures on the mechanical characteristics of polyvinyl alcohol (PVA) reinforced ECC (PVA-ECC) was examined in this work. A study was conducted to compare the modulus of elasticity, tensile strength, and compressive strength of samples containing various dosages of three available pozzolans: zeolites (ZE), fly ash (FA), and ground granulated blast furnace slag (GGBS), as well as silica fume (SF). Scanning electron microscopy was used to examine the microstructure of PVA-ECC samples subjected to various high temperatures (SEM). For 1 h, all specimens were exposed to temperatures of 150, 100, 50, and 20 °C. The experimental findings revealed that when the temperature rises to 50°, all three mechanical parameters, modulus of elasticity, tensile strength, and compressive strength, increase, then drop. The study of the microstructure of PVA-ECC reveals a basic explanation for the loss of macro mechanical characteristics. Finally, using ANOVA analysis, the relationship between independent parameters such as the age of the sample (AS), temperature (T), and water to pozzolan materials (W/P), and dependent parameters (i.e., compressive strength, tensile strength, and modulus of elasticity) was evaluated, and the multivariate linear regression (MLR) and the artificial neural network (ANN) were shown to be useful for modeling.