Prediction of Compressive Strength of Self-Compacting Concrete Containing Silica's Using Soft Computing Techniques

Abstract

Self-compacting concrete (SCC) as the name suggest is a special type of concrete which sets under its own weight without use of any external vibrators. It is one of the most practical concrete with high workability and containing slag, natural pozzolana and silica as supplementary cementitious materials. Cement being the major source of CO2 in air, replacing it gives a wide variety of benefits such as reduced carbon dioxide emission, cost benefits, reduced consumption of natural resources, and improved fresh and hardened properties. SCC is used in in many types of construction work therefore prediction of performance of SCC used is required. This paper investigates the capability of utilising artificial neural network, linear progression, support vector machine, random forest and there hybrid bagging techniques in calculating the compressive strength of SCC containing silica as supplementary cementitious material. In this regard a wide range of experimental data has been collected from various published literature and trained and tested for the models using mentioned artificial intelligence technique. The data used for model consists of input and output parameters with cement, sand, coarse aggregate, silica, superplasticizer and water binder ratio as input and 28 days compressive strength as output parameter. With the help of correlation coefficient, mean absolute error and root mean square error the prediction of compressive strength can be assessed. Result shows that bagged artificial neural network outperform every other every other model in the prediction of compressive strength. Artificial neural network-based model performs better than techniques, i.e., linear regression, support vector machine and random forest. Bagging-based random forest model perform better than random forest model. Hence, Bagging improved the performance of model. Further, sensitivity analysis suggests that cementitious material was the most influencing parameter for prediction of compressive strength of SCC.