Investigation of geopolymer mortars and concretes based on blast furnace slag, obsidian, and metakaolin with different fiber content

In this study, the mechanical and microstructural properties of blends containing obsidian powder (OB), blast furnace slag (HFS) and metakaolin (MT) were investigated. It is aimed to save labor and time by predicting the obtained data with appropriate machine learning algorithms. The thesis study consists of 4 stages. In the first stage, geopolymer mortars were produced by mixing the 3 different materials (OB, YF and MT) at 25% by weight in 13 different combinations. During the production process, NaOH with 12 molar concentration was used as alkali activator, alkali/binder ratio was 0.45, curing temperature was 75 °C, 90 °C, 105 °C and 120 °C, curing time was 72 hours, mechanical (flexural and compressive strength) and microstructural analyses by XRD, SEM-EDS and FTIR were performed after 7, 28 and 90 days. Second phase of the thesis mechanical (flexural + compressive) strengths and microstructural analysis by SEM-EDS of the fibrous geopolymer mortars produced by substituting steel, polyvinyl alcohol, and basalt fibers at 1,5%, 2,0% and 2,5% of the binder weight to the 5 mixtures with the highest flexural and compressive strengths (OB100YF0MT0, OB0YF100MT0, OB75YF25MT0, OB75YF0MT25 and OB50YF50MT0). In third stage, the data set consisting of flexural and compressive strengths were predicted by algorithms such as linear regression (LR), k nearest neighbor (kNN), deep neural networks (DNN), decision trees (DT) and Random Forest (RF). In addition, the importance weights of the inputs were obtained by feature ranking analysis. In the final stage of the thesis, concrete specimens were prepared under 90 °C curing temperature and 72 hours curing time by substituting steel fiber, polyvinyl alcohol and basalt fibers at the rate of 2% of the binder weight, which was determined as the optimum fiber ratio, in 5 different mixtures with the highest strength obtained in fibrous geopolymer mortars. The mechanical strength (compressive and splitting tensile strength) and microstructural properties of the concrete specimens were analyzed by SEM-EDS at 28 days. The findings of this thesis show that obsidian powder can be used in the production of geopolymer mortar and concrete, the optimum curing temperature is 90 °C, the optimum fiber ratio is generally 2% of the binder weight, metakaolin is the most important input in machine learning studies and the most successful prediction algorithm is Random Forest (R2=0,983).