Fire resistance of concrete: prediction using artificial neural networks

Abstract

A model based on an artificial neural network (ANN) for predicting the loss of strength of concrete under high temperatures (75–1200°C) is presented. The ANN is a recent development and is particularly useful for evaluating non-linear systems with several variables the relationships between which cannot be defined explicitly by mathematical equations. It has been reported in the literature that the fire resistance of concrete is related to several material and environmental factors, but this relationship is difficult to quantify mathematically. The advantage of the ANN is that it can be trained, using experimental data assembled from published research, to recognize the relationship between these influencing factors and the fire resistance of concrete measured by the loss of strength. After training, the ANN is applied to predict the loss of strength of concrete with new material and environmental factors. In the tests conducted the prediction errors between the network outputs and the actual experimental results were less than 15%.