Study on the Constitutive Modeling of (2.5 vol%TiB + 2.5 vol%TiC)/TC4 Composites under Hot Compression Conditions

• Published in: Materials Vol. 17; no. 3; p. 619
• Main Authors: Qiang, Kehao, Wang, Shisong, Wang, Haowen, Zeng, Zhulin, Qi, Liangzhao
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 27.01.2024
• Subjects: Accuracy; Algorithms; Back propagation networks; Constitutive models; Constitutive relationships; Correlation coefficients; Deep learning; Deformation; Experiments; Heat treating; Hot pressing; Hot working; Mathematical models; Mechanical properties; Metal matrix composites; Neural networks; Phase transitions; Strain rate; Stress-strain curves; Temperature; Titanium alloys; Work hardening; Working conditions; Yield strength; flow stress; titanium matrix composites; constitutive model; hot working
• ISSN: 1996-1944; 1996-1944
• DOI: 10.3390/ma17030619

The hot deformation behavior of titanium matrix composites plays a crucial role in determining the performance of the formed components. Therefore, it is significant to establish an accurate constitutive relationship between material deformation parameters and flow stress. In this study, hot compression experiments were conducted on a (2.5 vol%TiB + 2.5 vol%TiC)/TC4. The experiments were performed under temperatures ranging from 1013.15 to 1133.15 K and strain rates ranging from 0.001 to 0.1 s−1. Based on the stress–strain data obtained from the experiment, the constitutive models were established by using the Arrhenius model and the BP neural network algorithm, respectively. Considering the relationship between strain rate, hot working temperature, and flow stress, a comparative analysis was conducted to evaluate the prediction accuracy of two different constitutive models. The research results indicate that the flow stress of (2.5 vol%TiB + 2.5 vol%TiC)/TC4 increases with decreasing temperature and increasing strain rate, and the stress–strain curve shows obvious work hardening and softening behaviors. Both the Arrhenius model and the BP neural network algorithm are effective in predicting the hot compression flow stress of (2.5 vol%TiB + 2.5 vol%TiC)/TC4, but the average relative error and root mean square error of the BP neural network algorithm are smaller and the correlation coefficient is higher, thus possessing higher accuracy and reliability.