Gupta Megha, Thapa Ishwor, Ghani Sufyan, Prakash Satya
Department of Civil Engineering, Sharda University, Greater Noida, India.
Environ Sci Pollut Res Int. 2025 Jul 22. doi: 10.1007/s11356-025-36767-9.
The construction industry's growing concern for sustainable practices has led to the exploration of alternative materials to replace traditional components in concrete production. One such promising material is glass powder, a byproduct of the glass industry. This study proposes a finite element method (FEM) and deep learning (DL)-based model to assess the structural and mechanical properties of concrete when incorporating waste glass powder (WGP) as a partial replacement for traditional cementitious materials. The research begins by characterizing the physical properties of WGP to understand its potential as a cement substitute. Subsequently, a comprehensive FEM model is developed to simulate the behavior of WGP as a partial replacement of cement in concrete under various loading conditions. The FEM and DL model considers factors such as material heterogeneity, interfacial bonding, and the influence of WGP on the microstructure of the composite material. Through systematic analyses, the study aims to investigate the impact of different proportions of WGP on the compressive strength of concrete. Additionally, the FEM simulations will provide insights into the deformation patterns, stress distribution, and failure mechanisms within this sustainable concrete. Furthermore, the DL model will provide the best prediction input parameters for the prediction of the compression strength of concrete with WGP. The proposed model offers a platform for optimizing the mix design by considering the mechanical performance and sustainability aspects of waste glass powder-replaced concrete. Notably, FEA results show that the best compressive strength is achieved when 10% of cement is replaced with WGP; these findings are corroborated by experimental lab results. Reliability is demonstrated by the convolutional neural network (CNN) used in the DL model, which predicts the compressive strength of WGP concrete with an accuracy of more than 92% both in the training and testing stages. Research findings support using WGP in construction for eco-friendly concrete, offering guidelines to reduce environmental impact and promote sustainability in the industry.
建筑行业对可持续发展实践的关注度不断提高，促使人们探索替代材料，以取代混凝土生产中的传统成分。玻璃粉就是这样一种很有前景的材料，它是玻璃行业的副产品。本研究提出了一种基于有限元法（FEM）和深度学习（DL）的模型，用于评估在掺入废玻璃粉（WGP）作为传统胶凝材料的部分替代品时混凝土的结构和力学性能。研究首先对WGP的物理性能进行表征，以了解其作为水泥替代品的潜力。随后，开发了一个综合有限元模型，以模拟WGP在不同加载条件下作为混凝土中水泥的部分替代品的行为。有限元法和深度学习模型考虑了材料不均匀性、界面粘结以及WGP对复合材料微观结构的影响等因素。通过系统分析，该研究旨在探讨不同比例的WGP对混凝土抗压强度的影响。此外，有限元模拟将深入了解这种可持续混凝土的变形模式、应力分布和破坏机制。此外，深度学习模型将为预测含WGP混凝土的抗压强度提供最佳预测输入参数。所提出的模型提供了一个平台，通过考虑废玻璃粉替代混凝土的力学性能和可持续性方面来优化配合比设计。值得注意的是，有限元分析结果表明，当用WGP替代10%的水泥时，可获得最佳抗压强度；实验室内结果证实了这些发现。深度学习模型中使用的卷积神经网络（CNN）证明了其可靠性，该网络在训练和测试阶段预测WGP混凝土抗压强度的准确率均超过92%。研究结果支持在建筑中使用WGP来生产环保混凝土，为减少环境影响和促进该行业的可持续发展提供了指导方针。
