Prediction model of compressive strength for concrete mixing with waste PET plastic bottle fiber as partial replacement of fine aggregate using artificial neural network

Due to rapid industrialization and urbanization in global, lots of infrastructure developments are taking place. This rapid development led to the acute shortage of construction materials and increased dumping of waste materials. In addition, waste plastic bottles are major reason of solid waste disposal as the bottle consumption have been increasing nowadays. This increment occurred because of plastic bottle is popular for current packaging. The plastic bottle mostly made of Polyethylene Terephthalate (PET) type. Waste PET plastic bottle can be replaced in the concrete mixing as partial replacement of aggregates. In order to reduce the waste production during concrete testing and shorten the time of experimental work, the prediction model method can be developed. Thus, this paper focuses on the prediction model development using Artificial Neural Network (ANN) to predict the compressive strength of concrete mixing with waste PET plastic bottle fiber. Data sample was collected from previous research. Data collection including cement content, water content, water-cement ratio, fine aggregate, coarse aggregate and percentage of waste PET plastic bottle fiber were used in developing prediction model. The percentage used for waste PET plastic bottle fiber was 0% to 20%. The data were divided to 70%, 15% and 15% for data training, data testing and data validation respectively. The data was trained using Levenberg- Marquardt method and tan-sigmoid activation function. The correlation coefficient value for training, testing and validation were 0.98, 0.98 and 0.99 respectively. The result for Mean Absolute Error (MAE) was 1.731% and Root Mean Square Error (RMSE) was 2.210%. The comparison value of compressive strength between predicted and experimental from previous study shows a small difference. Therefore, by using ANN, the compressive strength for 28 days can be predicted efficiently and the natural resources can be reduced. Thus, the application of ANN could help in preserving the environment.