Prediction of mechanical properties in magnesia based refractory materials using ANN
Abstract
Refractory materials are heterogeneous materials having complex microstructures with different constituent's properties. The mechanical properties of these materials change depending on their chemical composition and temperature. Therefore, it is important to select a refractory material, which is suitable for working conditions and is fit to place of use. Artificial neural network (ANN) model is established to investigate the relationship among processing parameters (chemical composition, temperature) and mechanical properties (bending strength, Young's modulus) in magnesia based refractory materials. The mechanical properties of magnesia based refractory materials having four different chemical compositions were investigated using three point bending test at temperatures of 25, 400, 500, 600, 700, 800, 900, 1000 and 1400 °C. The bending strength (σ) and Young's modulus (E) were theoretically calculated by ANN method and theoretical results were compared with experimental values for each temperature. There were insignificant differences between experimental values and ANN results meaning that ANN results can be used instead of experimental values. Thus, mechanical properties of refractory materials having different chemical composition can be predicted by using ANN method regardless of the treatment temperature. © 2009 Elsevier B.V. All rights reserved.
Description
Keywords
Backpropagation , Bending strength , Elastic moduli , Elasticity , Magnesia , Magnesia refractories , Materials properties , Neural networks , Thermal shock , Artificial neural network models , Chemical compositions , Complex microstructures , Experimental values , Heterogeneous materials , Prediction of mechanical properties , Processing parameters , Theoretical result , Three-point bending test , Treatment temperature , Working conditions , Young's Modulus , Mechanical properties