Title : PREDICTING COMPRESSIVE STRENGTH OF RECYCLED AGGREGATE CONCRETE USING NEURAL NETWORKS, MODEL TREES AND NON-LINEAR REGRESSION
Authors : Neela Deshpande1, Shreenivas Londhe2 and Sushma Kulkarni3
Publication : 23 - 26 October 2013
Volume :
Pages :
Price : 250
Abstract : Demolished concrete waste can be used in concrete as aggregates and the new concrete termed as Recycled Aggregate Concrete (RAC) can be considered as an important factor for the infrastructure development of a country. Higher water absorption, lower density etc. are some properties of RAC which affect the compressive strength of RAC. Thus, predicting compressive strength of RAC requires extensive testing and is a difficult task. The paper aims at predicting the 28-day compressive strength of RAC using data driven techniques such as Artificial Neural Network (ANN), Model tree (MT) and Non-linear regression (NLR). Two separate models were developed using each of the above mentioned techniques with input parameters as per cubic proportions of materials used in concrete and the replacement ratio of recycled aggregates to natural aggregates in concrete. Comparison of the results by the above mentioned data driven techniques shows that ANN performs better than Model tree and NLR equations in both the models. With limited amount of input parameters also, ANN predicted the strength of RAC better than MT and NLR.