Mechanics-guided data-driven model for temperature-dependent elastoplastic behavior of conventional and additively manufactured metals

• The effect of temperature on AM and conventional metals deteriorates the mechanical behavior in both linear elastic and nonlinear plastic regimes. • The developed finite element analysis framework coupling thermal and mechanical loadings enabled predicting the elastoplastic behaviors for different AM metals at various temperatures that were not experimentally tested. • The proposed exponential decay functions are suitable to accurately capture the degradation in the elasticity modulus as well as the onset of plastic deformation relying on a single constant. • The modified R-O relationship, Eq. [20] has proven excellent analytical predictions capabilities compared with both numerical results and experimental testing from literature to define the elastoplastic behavior of materials accounting for temperature variation. . Reliable prediction of temperature-dependent mechanical degradation in metals is essential for designing components in aerospace, automotive, and power generation applications. Accurately capturing the temperature-dependency through mathematical modeling is critical for predicting the elastoplastic behavior of metals, thereby enabling safer and more efficient design under thermal loading. This study employed a mechanics-guided, data-driven approach to mathematically model the degradation mechanisms of selected conventional and additively manufactured (AM) alloys. Based on mechanics principles and a comprehensive review of existing literature, trends in the deterioration of mechanical properties with increasing temperature were identified. However, due to scarcity of elevated temperatures experimental data, a finite element model (FEM) was developed and validated against experimental results under both ambient and high-temperature conditions. The validated FEM was then employed to generate synthetic datasets comprising approximately 23,500 data points for different alloys divided into training and testing subsets. The training subset, in combination with the Ramberg–Osgood (RO) relationship, were used to identify the form and parameters of degradation functions capturing temperature-induced deterioration. As a result, new predictive mathematical expressions were developed to represent the temperature-dependent elastoplastic behavior of the studied alloys. The testing subset along with additional experimental data were used to validate the generalizability and accuracy of the proposed expressions. Finally, parametric and sensitivity analyses were conducted to assess the influence and relative importance of each input parameter on the predicted strain response. The resulting expressions provide an efficient, predictive, and easy-to-implement tool for engineers and designers to evaluate the high-temperature elastoplastic behavior of metals.