The rapid increase in global waste production poses significant environmental challenges, particularly with regard to polymer wastes. These materials, which are largely non-biodegradable, threaten both vegetation and aquatic life. To tackle this pressing issue, researchers have proposed innovative construction methods, such as the repurposing of polymer waste to create eco-friendly bricks.
This study focuses on the development of bricks made from sustainable materials, combining cement, fly ash, manufactured sand (M sand), and polypropylene (PP) fibers—all sourced from waste polymers. The unique aspect of this research lies not only in the intriguing composition of these bricks but also in the employment of advanced machine learning techniques to predict their compressive strength, making the results more accurate and transparent.
Using models like artificial neural networks (ANN), support vector machines (SVM), Random Forest, and AdaBoost, the researchers recorded the compressive strength of the polymer-infused bricks as the output parameter, analyzing variables such as cement, fly ash, M sand, PP waste, and age as inputs. This innovative approach allows the team to gather insights about the effectiveness of each component and how they influence the final product's strength.
One significant innovation presented by the study is the integration of the SHapley Additive exPlanations (SHAP) interpretation method, which sheds light on the influence of different input variables on the predicted outcomes. By employing SHAP, the researchers aimed to address the opaque nature of machine learning models, making them more interpretable and actionable.
The findings revealed impressive results. The ANN model was particularly successful, achieving R2 values of 0.99674 during training and 0.99576 during testing phases. Alongside these findings, the root mean square error (RMSE) stood at 0.0151 for training and 0.01915 for testing, demonstrating the high accuracy of the ANN model for predicting compressive strength.
Importantly, the SHAP analysis highlighted two factors—age and fly ash—as the most influential variables affecting the final compressive strength of the bricks. Such insights not only point to effective practices for brick creation but also signify the importance of properly integrating various materials to optimize their performance.
With over 50 million tons of polymers consumed every year, India faces significant waste management challenges. A concerning 1.5-2% increase per year exacerbates the environmental hazard posed by polymer waste, highlighting the necessity for efficient recycling and innovative strategies to reduce disposal impacts.
Conventional bricks, which predominantly rely on raw materials, cannot match the affordability and sustainability of polymer-infused alternatives. These developments advocate for innovative solutions to construction material challenges, especially considering the increasing demand for shelters and infrastructure across rapidly urbanizing regions of India. Consequently, environmentally friendly bricks represent not only sustainable waste management but also enhanced performance capabilities—cutting costs and offering technical improvements over traditional building materials.
The methodology utilized to assess these bricks involved rigorous laboratory testing. Specifically, samples were prepared using various proportions of raw materials and subjected to different conditions to identify optimal formulations. The compressive strength of the bricks was tested, which is integral for determining their endurance for construction activities.
The results indicated favorable outcomes; the polymer-infused bricks displayed significant performance capabilities compared to conventional red clay bricks during load testing. Notably, bricks with the M2 polymer-to-sand ratio showed outstanding compressive strength, achieving up to 16.85 N/mm2 under loads of 426 kN, indicating their potential as suitable alternatives for construction purposes.
Advancements like machine learning methods, particularly ANN and Random Forest, have proven invaluable for predicting compressive strength. The ANN model emerged as the most accurate, capable of capturing complex relationships within the data and aligning closely with experimental outcomes. The AdaBoost model also provided strong performance, demonstrating its generalizability across varied datasets, whereas the SVM model exhibited less predictive effectiveness.
These novel approaches to incorporate machine learning techniques spotlight the need for transforming how traditional materials are perceived and utilized within the construction industry. Once perceived as waste, polymers could be converted to become part of resource-sustainable solutions for infrastructure problems, effectively shaping future building practices.
Efforts to optimize formulations and manufacturing processes for polymer-based bricks could yield positive societal impacts too, like job creation and community engagement initiatives aimed at ecological restoration. Therefore, this intersection of machine learning and sustainable development could pave pathways toward not only addressing pollution but fostering circular economy principles.
Looking forward, the study's insights advocate for continued research to expand the dataset size and compare machine learning customized models against existing conventional methodologies. With the aim of ensuring the scaling of these promising findings, researchers must design upcoming studies to fully grasp and refine these applications within the building sector.
Through innovation, collaboration, and commitment, the potential to transition from environmental degradation toward sustainable building practices remains achievable. The unique intersection of waste management, modern construction practices, and advanced machine learning applications opens new doors to combat ecological challenges effectively, aligning with global sustainability goals.