Performance and modeling of volcanic ash and limestone powder waste binary blended alkali activated mortar

Abstract

The greenhouse gas emissions and energy consumption from the production of ordinary Portland cement (OPC) have caused the need to have an alternative sustainable binder system that is eco-friendlier. While alkali-activated material (AAM) is rapidly emerging as a potential eco-efficient and economically viable alternative material to OPC, getting the raw materials for AAM from local sources has been a significant challenge. Local sourcing of raw materials from industrial by-products ensures the sustainability of AAM. Furthermore, to enhance the commercialization of AAM, the mechanical and the durability performance of the AAM synthesized from the local materials need to be established. This research seeks to advance state of the art toward using locally sourced waste materials to develop alkali-activated mortar. Two of the identified viable solid wastes locally available in Saudi Arabia are volcanic ash (VA) and limestone powder (LSP). This study aimed to synthesize alkali-activated mortar using VA and LSP and to develop models using artificial intelligence and statistical techniques that can quickly and accurately predict the compressive strength of alkaliactivated volcanic ash and limestone powder. Firstly, experimental work was carried out to synthesize and evaluate the developed mortar's fresh, mechanical, and durability performance. The impact of constituent variables such as binder ratio varied from (0 to 1), sodium hydroxide concentration [4-14 M], silica modulus (0.52 - 1.18), curing temperatures (ambient room temperature and oven cured temperature maintained from 45 to 90 °C at an interval of 15 °C for 24 hrs), Fine aggregate -tobinder ratio (1.4 - 2.2) and alkaline activator to binder ratio (0.45 - 0.55) on the compressive strengths and microstructure of the developed mortar was investigated. The synthesized alkali-activated VA and LSP binders were examined critically to study the impact of the constituent variables on the product formed, bond characteristics, and microstructures using x-ray diffractometer (XRD), scanning electron microscope coupled with energy-dispersive x-ray spectroscopy (SEM+EDX), and Fourier transforms infrared (FTIR) spectroscopy. Secondly, different models capable of estimating one-day, three-day, 14-day, and 28-day compressive strength (CS) were developed using a hybrid genetic algorithm (GA), support vector regression (SVR) algorithm, and stepwise regression algorithm. The optimization of the mix parameters gave the maximum 90-day compressive strength of 31.3 MPa with 60% LSP and 40% VA using 10 M NaOH( a q ), silica modulus of 0.89 and cured at 75 °C for 24 hrs duration. Besides, about 77% of 28-days compressive strength (27 MPa) could be achieved in 24 hrs using heat curing. Samples synthesized with sole 10 M NaOH( a q ) resulted in a binder with a low 28-day compressive strength (15 MPa) compared to combined usage of Na2SiO3 ( a q )/10 M NaOH( a q ) activators. Curing at a low temperature (25 to 45 ) does not favour strength development, whereas higher curing temperature enhances strength development. The findings also revealed that the synergistic effect of VA with LSP emanated from silica and alumina required to form an aluminosilicate framework, which required cation sourced from LSP (Ca2 + ) for charge balancing in the formed skeletal framework. The binder products formed are anorthite (CaAl2 Si2O8 ) and gehlenite (Ca0.Al2O3.SiO2). Microstructural analysis revealed that the rough texture of activated VA initially characterized with high porosity turned to be filled up by the presence of LSP, thereby improving the microstructural density. SEM+EDX indicated that strong alkali (10 M NaOH) enhanced microstructural density compared to that of mild alkali (4 M NaOH( a q )). The binder synthesized with 60% VA and 40% LSP exhibited the highest resistance to sulfate and acidic attack. The developed hybrid GA-SVR models can estimate the compressive strength of mortar for one day, three days, and fourteen days, up to 96.64%, 90.84, and 93.40% degree of accuracy as measured based on the correlation coefficient between the measured and estimated value. The developed cubic with interactions stepwise regression model (V) was characterized with a high correlation coefficient of 97.2% compared to the other four models (I-IV). The developed mortar can be used where repair work is required due to its early strength. The outcomes of this study also contributed to waste valorization, dumpsite land reclamation, low CO2 footprint, energy consumption reduction, reduction in environmental pollution, and the addition of more sustainable alternative binders for structural purposes. Furthermore, the outcomes of this work would provide a quick and efficient way of predicting the compressive strength of environmental friendly binders with minimal experimental stress and errors inherent in the laboratory.