Mechanical properties and prediction of fracture parameters of geopolymer/alkali-activated mortar modified with PVA fiber and nano-SiO2. Issue 12 (15th August 2020)

• Record Type: Journal Article
• Authors: Zhang, Peng; Wang, Kexun; Wang, Juan; Guo, Jinjun; Hu, Shaowei; Ling, Yifeng
• Is Part Of: Ceramics international. Volume 46:Issue 12(2020)
• Page Start: 20027
• Page End: 20037
• Publication Date: 2020-08-15
• Subjects: Geopolymer/alkali-activated mortar -- Mechanical properties -- Fracture parameters -- ANFIS; Ceramics -- Periodicals; Céramique industrielle -- Périodiques; Ceramics; Periodicals; Electronic journals; 666
• DOI: 10.1016/j.ceramint.2020.05.074
• Languages: English
• ISSNs: 0272-8842

Abstract

Abstract: Properties of fly ash (FA) and metakaolin (MK) based geopolymer/alkali-activated mortar modified with polyvinyl alcohol (PVA) fiber and nano-SiO2, including workability, compressive strength, flexural performance, elastic modulus and fracture property were tested in this study. PVA fiber content varies from 0 to 1.2%. Nano-SiO2 content is 0 and 1%. Adaptive neuro-fuzzy interfacial systems (ANFIS) method was used to establish the artificial intelligence (AI) model to predict the fracture parameters of geopolymer/alkali-activated mortars. The inputs of ANFIS models include PVA fiber content, nano-SiO2 content, compressive strength, flexural strength, elastic modulus, critical crack mouth opening displacement, crack load and peak load. The outputs of ANFIS model include critical effective crack length, initiation fracture toughness, unstable fracture toughness, and fracture energy. Experiment results showed that PVA fiber addition enhanced the mechanical properties especially the compressive strength and fracture performance, but decreased the workability. 0.8%–1.0% was considered as the optimal content of PVA fiber. Addition of 1% nano-SiO2 shows a slight improvement on both workability and mechanical properties of the mortar no matter how much fiber is added. Based on the ANFIS algorithm and 42 sets of experimental data, the trained models were proved to have high accuracy with root mean square error (RMSE) under 0.15, mean absolute error (MAE) under 0.01, and coefficient of determination (R 2 ) over 0.85. The ANFIS model established in this study combined the fracture properties with the basic mechanical properties of geopolymer/alkali-activated composites, which can provide a new method to assess the fracture performance of geopolymer/alkali-activated mortars modified with PVA fiber and nano-SiO2 in the future.