Checking the Influence of Sudden and Gradual Cooling Regimes on Strength, Near Surface Characteristics and Modulus of Elasticity of Hybrid Fibre Reinforced Blended Concretes Subjected to Sustained Elevated Temperatures Using Artificial Neural Networks

Abstract

Prediction is made on strength, near surface characteristics, and modulus of elasticity of hybrid fibre reinforced blended concretes subjected to sustained elevated temperatures for 3 hours retention period using artificial neural networks (ANN). Temperature ranges from 100°C to 1000°C at an interval of 100°C and after that, specimens are subjected to two cooling regimes, that is, sudden and gradual. These specimens are subjected to tests for compressive strength, split tensile strength, water absorption, sorptivity, and modulus of elasticity. For building ANN models, available 440 experimental results produced with eight different mixture proportions are used. Two major artificial neural networks are used for prediction. One is for all the concrete combinations with sudden cooling [SCR] and other is with gradual cooling [GCR]. The data used in the multilayer feed forward neural network models (architecture, 8-15-1) is designed with eight input parameters covering temperature [T], cement [C], fly ash [FA], GGBFS [GGBFS] Silica Fume [SF], galvanized iron fibre [GIF] polypropylene fibre [PPF], and cooling regime [SCR or GCR]. These five tests are the outputs and they are predicted individually for both the cooling regimes. It shows that neural networks have high potential for predicting the results.