Prediction of Compressive Strength of High-Volume Fly Ash Concrete Using Artificial Neural Network: Fid-Bau Portal

Prediction of Compressive Strength of High-Volume Fly Ash Concrete Using Artificial Neural Network

Rajeshwari, Ramachandra / Mandal, Sukomal

Abstract Sustainable development has led to use of waste materials for replacements in conventional concrete. This study focuses on concretes made by cement replaced with high volumes of fly ash, which exhibits good long-term mechanical and desirable durability properties. Usage of high volumes of fly ash in concrete reduces the energy demand globally also saving the natural resources which are on the verge of depletion. Desirable high-volume fly ash (HVFA) concretes are experimentally achieved by trials, leading to wastage of materials, time and money. An alternate approach, artificial neural network (ANN) can be used, which has lately gained popularity in the civil engineering field. ANN is a soft computing technique impersonating the human brain characteristics, learning from previous situations and adapting to new surroundings without any constraints. In this study, HVFA concrete compressive strength (CS) data collected from past experimental investigations are used for ANN modeling. A total of 270 datasets has been collected from literature, of which 12 nos. from an experimental study is used for testing purpose. An ANN model is developed with eight input parameters (i.e., cement, fly ash, water–binder ratio, superplasticizer, fine aggregate, coarse aggregate, specimen and fly ash type) to predict the CS of HVFA concrete; hidden layer nodes along with weights and biases are fixed by trial and error to achieve the better performing model. Coefficients of correlation for train and test data are obtained as 97 and 97.9% respectively, which shows that ANN could be used for predicting the HVFA concrete strengths.