Prediction of Creep Strain for Self-Compacting Concrete by Artificial Neural Networks

Abstract

Artificial Neural Networks, ANN, technique is a computerized system that is built to simulate the neural networks in the human brain. Throughout the recent couple decades, ANNs had solved with a good degree of success many problems. In the present work, ANN model was developed by SPSS software for estimating creep strain development of self-compacting concrete mixes produced with different types of Portland cement, Type I and Type IL. The independent variables in this model were: age, compressive strength, modulus of elasticity, applied stress, initial strain, water to powder ratio, water to binder ratio, filler to cement ratio, clinker to cement ratio, aggregate size, and slump flow. The used data for model building were local, extracted from the present work. The predictions of the model have been compared to those of an international well-known model, ACI 209 Committee. The comparison revealed the good reliability of the present models in predictions (r = 0.998).