Predicting Compressive Strength of Pervious Concrete Using Artificial Neural Network and Regression

Conventional concrete is amongst the most practicing materials used in construction industry. Now researchers are gradually progressing to nonconventional concrete for sustainable construction. Pervious concrete is sustainable type concrete, it has little or no fines which makes open cell structure that allows water and air to pass through it. The estimation of strength of normal conventional concrete using Rebound Hammer (RN’) and Ultra-Sonic Pulse Velocity (UPV’) tester has been widely discussed and appropriate models also suggested for correlations, but correlation models for pervious concrete are not yet suggested. In this experimental study two types of pervious concretes were prepared. On these samples Destructive Test (DT) and Non Destructive Test (NDT) were performed. Using the concluded results correlation equations were developed using statistical curve fitting method between crushing strength and RN’ and UPV’ derived for pervious concrete. A Neural Network was also developed to estimate compressive strength of pervious concrete using Destructive Test (DT) and Non Destructive Test (NDT) methods, for this dataset were divided into three sets educating data, testing and validation. A mathematical equation was developed for computations of in situ compressive strength with help of weights and biases extracted from ANN. The obtained results from ANN mathematical model was compared with statistical model. The coefficient of determination (R 2 ) is 0.980 while for statistical curve fitting is 0.743. The RMSE, MAPE and VAF of ANN model are 1.889, 8.45 and 93.11 while for statistical curve fitting are 2.094, 11.58 and 74.3. The results indicate that ANN can have better prediction of compressive strength of pervious concrete.