Machine learning application to predict the mechanical properties of glass fiber mortar. (June 2023)

• Record Type: Journal Article
• Authors: Nakkeeran, G.; Krishnaraj, L.; Bahrami, Alireza; Almujibah, Hamad; Panchal, Hitesh; Zahra, Musaddak Maher Abdul
• Is Part Of: Advances in engineering software. Volume 180(2023)
• Publication Date: 2023-06
• Subjects: Hydrated lime -- Fly Ash -- Glass fiber mortar -- RSM -- ANN -- Prediction; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2023.103454
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: Glass Fiber Reinforced Mortar Composites in FA and HL replacement in cement. Response Surface Methodology (RSM) and Artificial Neural Network (ANN). Predicting optimum Glass fiber mortar compressive strength using RSM and ANN. Accuracy of the ANN-based prediction model is higher than that of the RSM model. Abstract: In this study, the mechanical properties of glass fiber mortars have been predicted using machine learning tools, Response Surface Methodology (RSM), and Artificial Neural Network (ANN) approach. This study focused on mortar, in which cement has been partially replaced by 20% fly ash and 20% hydrated lime. In the experiments, the compressive strength of mortars has been determined after curing the mixes for 7 and 28 days. Glass fiber was added to the proportions of 0%, 0.2%, 0.4%, 0.6%, 0.8%, and 1% by weight of cement. The compressive strength of mortar incorporated with glass fiber increases according to an increase in the proportion of glass fiber. Results indicate that the optimal proportion of glass fiber in mortar had been observed to be 0.6%. The predicted compressive strength on day 28 has been modeled using RSM and ANN. The RSM model has been used to predict mechanical properties ( R 2 ≥ 0.7534) accurately. Furthermore, the appropriate R threshold ( R > 0.999) for training, testing, and validation demonstrates that the ANN model has successfully captured the variability in the data. The results show that with the high correlation between the experimental and prediction results, more accuracy has been observed in the ANN model than in the RSM model.