توصيفگر ها :
مديريت پايدار منابع آب , بتن حاوي انواع پساب , طراحي آزمايش تاگوچي , روش گري , مدلسازي رفتار بتن , LSBoos
چكيده انگليسي :
In recent years, water scarcity has emerged as one of the most critical environmental challenges worldwide. Iran is facing this crisis with increasing severity, as evidenced by the declining groundwater levels, drying wetlands, and widespread limitations in the supply of potable and agricultural water. In this context, the concrete industry—being a major water consumer—can potentially exacerbate the crisis due to the continuous urban development and construction demands. Simultaneously, rapid population growth and industrialization have led to increased volumes of municipal and industrial wastewater, much of which is discharged untreated into the environment, causing irreversible damage to water resources, soil, and ecosystems. This study investigates the feasibility of replacing freshwater with treated wastewater in concrete production. The research aims to reduce the concrete industryʹs dependency on freshwater and mitigate the environmental impact of wastewater discharge by utilizing a sustainable and permanent source. Four types of wastewater were examined: municipal wastewater and industrial effluents from stone-cutting, textile, and sugar factories. Key input variables included the wastewater replacement ratio (25%, 50%, 75%, and 100%), type of additive (microsilica, limestone powder, and superplasticizer), water-to-cement ratio (0.35, 0.40, 0.45, and 0.50), and cement content (300, 350, 400, and 450 kg/m³). A Taguchi L16 orthogonal array was employed to minimize the number of experiments while enabling multivariable analysis.
A comprehensive set of tests was conducted to evaluate the physical, mechanical, and durability properties of concrete, including setting time, workability (slump), compressive, tensile, and flexural strength, water permeability, chloride ion penetration, carbonation depth, and electrical conductivity. To select the optimal mix among the various tested combinations, Grey Relational Analysis (GRA) was used as a multi-criteria decision-making method. Furthermore, advanced machine learning models were developed to predict concrete behavior under different wastewater-based mix designs. These models—trained using experimental data from Taguchi-designed mixes—included Linear Regression, Decision Tree, Random Forest, K-Nearest Neighbors (KNN), and the LSBoost (Least Squares Boosting) algorithm. Notably, LSBoost was applied for the first time in this context and demonstrated superior predictive accuracy.
The results confirmed the viability of using treated wastewater in concrete production. Wastewaters—especially from textile industries—enhanced the compressive, tensile, and flexural strengths, with improvements up to 20% compared to conventional mixes. The stone-cutting wastewater significantly reduced permeability and chloride ingress, while the sugar factory effluent improved resistance to carbonation. The combination of microsilica and superplasticizer yielded notable performance enhancements. Among machine learning models, LSBoost outperformed others, achieving an R² of 0.998 and very low RMSE values, indicating high accuracy and minimal error in predicting complex and nonlinear relationships. Overall, the integration of Taguchi experimental design with LSBoost modeling provides a robust framework for optimizing sustainable concrete production using various wastewaters.
