Search

Highlights: [•] An artificial neural network was utilized to predict endurance limits from carefully selected inputs. [•] The worst-case correlation factor of 0.9 indicated that the neural network has been well trained. [•] Comparison of predicted and experimental data confirmed the accuracy of the model. [ABSTRACT FROM AUTHOR]

Abstract: The characteristic formation of graphite microstructures defines the physical and mechanical properties of cast irons. The graphite morphology of cast iron structures can be classified according to their shape, arrangement and size using as reference the structure micrographs specified in the standards ASTM A 247-67 and DIN EN ISO 945. These specific values do not provide any mathematical correlation between microstructure and mechanical properties. The application of quantitative image analysis in microstructure investigations allows a statistically reproducible evaluation method. Although cast irons were quite often focus of research works in this sector, the graphite morphology classification has been mainly handled devoid of any quantitative correlation to material properties. In this article an interrelationship between the fatigue strength of cast iron materials and a quantitative characteristic of the graphite microstructure was established. Thereby it is necessary to identify the largest graphite precipitates, which can be statistically considered as potential crack initiation sites. In this context three different cast iron materials were investigated: ENGJL-270, EN-GJV-450 and EN-GJS-700. [Copyright &y& Elsevier]

Abstract: The artificial neural network methodology presented in this paper was trained to predict the green shear strength of compacted samples made from iron powder. Iron powders of three different morphologies admixed with three types of lubricants in different amounts were considered. Green compacts were pressed uniaxially in a square floating die. The more or less cubic slugs were sheared to fracture perpendicularly and parallel to the direction of compaction. From the maximum loads at the start of decohesion the green shear strengths were calculated. Compaction parameters together with corresponding shear strength records were used as sets of data for the training process. The performance of the network was verified by putting aside one set of data and testing the network against it. Comparison of the predicted and experimental data confirmed the accuracy of the model. [Copyright &y& Elsevier]