PSI - Issue 41

2

Abdelmoumene Guedri et al. / Procedia Structural Integrity 41 (2022) 564–575 Abdelmoum ne Gu dri et al. / Structural Integrity Procedia 00 (2022) 00 –000

565

Nomenclature ARE

Absolute Relative Error Artificial Neural Networks

ANN MSE

Mean Square Error

S 0 S u

Area of the initial section of the calibrated

Area of the minimum section of the test piece after deformation

T Z

Temperature

Reduction of Area [%]

1. Introduction The ductility is the result of a compromise between convenience deformation parameters such as catering and recrystallization, and various damage mechanisms that could be an increase in the local stress, and / or a decrease in local energy (Zhang, et al., 2017). The elaboration and hot processing of metals are usually restricted by the occurrence of defects, generating a ductility fall, which can compromise their further use (Sellars, 1980). These defects take the appearance of superficial or internal cracks. For some materials, it has been noted a temperature range was ductility is low, due to hot brittleness. For industrials, the understanding of the various mechanisms related to hot brittleness, as well as a perfect knowledge of this ductility trough, are required for the high temperature processing of steels, like forging, rolling or pressing (Vedani, et al., 2008). The ductility of materials has been the subject of several studies with diverse and sometimes contradictory results. It results from a compromise between what favors the plastic deformation and accommodates the local constraints, and which on the contrary, makes appear and grow the damage. The damage is manifested by the formation of voids or cavities, and can occur in the early stages of plastic deformation (Pater and Gontarz, 2019). To control the use of metals and alloys it is necessary to study their thermomechanical behavior before subjecting them to deformation treatment on an industrial scale (Liao, 2019). It is characterized by different laboratory tests according to different parameters such as the deformation temperature, the deformation rate and the microstructural nature of the material (Zhang, et al., 2018). It is well known that the hot deformation behavior of this micro-alloyed steel is sensitively dependent on deformation parameters involving strain, strain rate and temperature. Careful modelling of the flow curves of different materials at elevated temperatures is the first step in mathematical simulation of hot deformation processes such as hot forging and hot rolling. Many researches have been conducted to model the flow curves of different materials (Mirzadeh, and Najafizadeh, 2010; Lin, and Chen, 2011) . Recently, the artificial neural network (ANN) supplied a new method to predict flow stress data by learning the non-linear relationships. The artificial neural network (ANN) is an information processing system by imitating the behavior of biological neural systems, i.e. a data-driven black-box model. The ANN model does not need to build intricate mathematical models (Quan, et al., 2017; Allaoui, et al., 2019) . In this study, hot ductility analysis of (C-Mn-S-Al-Nb-V-Ti) micro-alloyed steel as measured by tensile testing was carried out. The flow stress curves were obtained using the ANN method. 2. Material and experimental techniques Our test specimens were taken from a hot-rolled coil. The average grain size of their initial microstructure is equal to 20 μm (Figure 1). This type of material, called micro-alloyed steel, is intended for the production of welded pipelines used for the transport of gas and oil.