Prediction of Compressive Strength of GGBFS Concrete Subjected to High Temperatures using Artificial Neural Network

Granulated blast furnace slag is widely recycled to reduce the cost of construction as well as to reduce the carbon print on earth. Studies have reported use of GGBFS (Ground Granulated Blast Furnace Slag) as replacement of cement and its ability to improve the long term strength, durability, sulphate attack resistance etc. However, very limited studies have been reported on the compressive strength of GGBFS concrete subjected to high temperatures. In this work, the variation in compressive strength of GGBFS concrete subjected to high temperatures prior to loading is studied. For selecting the proper mix design, trial mixes were prepared and the one with optimum mix proportion was identified.. Experimental results show that addition of around 20% of slag content gives much higher strength than the plain concrete. Also at higher temperatures, the difference in the ultimate strength of GGBFS samples is comparatively less. Using experimental data, ANN models were developed based on LM algorithm and was trained, validated and tested. Input parameters for ANN include the slag content, cement content and temperatures the samples are subjected to. Compressive strength is the output variable of this model. The results of the ANN modeling indicate good performance in the prediction of the compressive strength of the samples subjected to furnace temperatures.