Comparative study on properties of self compacting concrete using artificial neural network and regression analysis

Abstract

Analytical models are developed by using Artificial Neural Network (ANN) and Multiple Regression Analysis (MRA) for predicting the Slump Flow Diameter (SFD), L-box Ratio (LR), V-funnel Flow Time (VFT) and Compressive Strength (CS) at 28 days of Self Compacting Concrete (SCC). In this work, 60 Mix proportions were prepared for different grades of SCC containing 20%, 30% and 40% of fly ash as partial replacement of cement. The fresh and strength properties of SCC which are arrived through experimental investigations are utilized for developing the analytical models. The mix constituents such as Cement (C), Fly ash (F), Fine Aggregate (FA), Coarse Aggregate (CA), Super plasticizer dosage (SP) and Water-Binder ratio (W/B) were fed as input parameter to achieve the four output parameters as targets. Four models were developed using both ANN and MRA and their results were evaluated and compared. ANN models demonstrated more accuracy and had higher correlation.