🔬 materials science

Averaging in thermodynamic dislocation theory: general macroscopically uniform stress and strain states

This paper extends thermodynamic dislocation theory to arbitrary macroscopically uniform stress and strain states by deriving an associated $J_2$ flow theory from first principles, which successfully unifies the description of copper's tension, compression, and torsion responses across a wide range of temperatures and strain rates using a single set of material parameters.

Original authors: Khanh Chau Le

Published 2026-09-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Khanh Chau Le

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Metals are not solid, unyielding blocks; under the microscope, they are vast crowds of tiny crystals, each with its own internal structure and orientation. When a metal bar is pulled, squeezed, or twisted, these crystals slide past one another along specific planes, a process driven by defects called dislocations. These dislocations act like traffic jams in the atomic lattice; they can get stuck, or "pinned," by obstacles, and they require energy to break free and keep moving. The way these dislocations move, get stuck, and rearrange themselves determines how hard or soft a metal feels, how it hardens under stress, and how it behaves when heated or stretched at high speeds. Understanding this microscopic dance is crucial for engineers designing everything from aircraft frames to medical implants, yet predicting exactly how a metal will react to a complex mix of forces has long been a difficult puzzle.

For decades, scientists have relied on mathematical rules to describe this behavior, but these rules often required fitting new numbers to every new experiment, making them more like descriptions of what happened than true predictions of what would happen next. A newer approach, known as thermodynamic dislocation theory, attempts to fix this by treating the dislocations not just as mechanical defects, but as a system with its own temperature and disorder, much like a gas. This theory suggests that the movement of dislocations is driven by heat and the energy of the deformation itself. However, this powerful theory had only been successfully applied to simple, one-way stretching or squeezing. It had not yet been shown how to apply it to the complex, multi-directional forces that real-world objects often face, such as the twisting of a drive shaft or the combination of pulling and twisting in a single component.

In a recent study, researchers set out to bridge this gap, extending the theory to handle any uniform state of stress and strain. They began with a simple but profound assumption: that in a typical metal, the tiny crystals are oriented in all directions with equal probability. By averaging the behavior of these randomly oriented crystals, they derived a set of rules that connect the microscopic sliding of dislocations to the macroscopic forces we can measure. They found that when you average the sliding forces across all possible crystal directions, the result is a direct, proportional link to a standard measure of stress used by engineers, known as the von Mises stress. This means that the complex, multi-directional flow of the metal can be described by a single, unified rule that works for tension, compression, and torsion alike, without needing to invent new parameters for each situation.

The researchers then tested this unified theory against a massive collection of real-world data. They gathered experimental results for copper, a common and well-studied metal, covering a wide range of conditions. The data included tests where copper bars were pulled apart at high speeds, compressed in a shock tube, and twisted at various rates, all conducted at temperatures ranging from room temperature up to nearly 1,200 Kelvin. Crucially, they did not just look at the final stress and strain numbers; they worked directly with the raw measurements of torque and twist angle from the twisting tests, avoiding the simplifications that often distort the data in traditional analyses. By fitting their theory to all 142 data points simultaneously, they discovered that a single set of material parameters could accurately describe the behavior of copper in all these different scenarios.

One of the most significant findings was the resolution of a long-standing mystery. Previous studies had noted that copper seemed to behave differently when twisted compared to when it was pulled, leading scientists to believe that the material laws themselves might be different for these two types of loading. The new analysis showed that this discrepancy was not due to a flaw in the theory or a fundamental difference in the metal's laws. Instead, it was caused by the initial state of the dislocations in the specific samples used for the twisting tests. The copper used for the twisting experiments had a higher initial density of dislocations, likely due to its specific manufacturing history, which made it appear harder initially. Once the researchers accounted for this difference in the starting condition, the same set of rules perfectly described both the tension and the torsion data.

The study also clarified the limits of the theory. It works exceptionally well for dynamic loading, where forces are applied quickly, but it does not apply to very slow, quasi-static processes where other mechanisms like recrystallization take over. Furthermore, the theory predicts that at very large amounts of twisting, beyond a certain point, the metal's behavior will change as the internal crystal structure begins to align in a specific direction, a phenomenon known as texture evolution that is not yet included in the current model. Despite these boundaries, the work provides a robust, predictive framework that connects the microscopic world of dislocations to the macroscopic world of engineering. It demonstrates that by understanding the statistical average of how billions of tiny crystals behave, we can create a single, coherent description of how metals deform, harden, and respond to heat and speed, offering a clearer path for designing materials that can withstand the extreme conditions of the modern world.

Technical Summary: Averaging in Thermodynamic Dislocation Theory

Problem Statement
Thermodynamic Dislocation Theory (TDT) has previously been formulated for single slip systems in simple shear and applied to polycrystals under uniaxial tension or compression. However, a general macroscopic formulation for arbitrary stress and strain states was lacking. Specifically, the theory required a rigorous method to define mean resolved shear stress and strain for a randomly oriented polycrystal under general loading, and to determine the direction of macroscopic plastic flow without resorting to computationally expensive crystal-plasticity simulations. Additionally, existing empirical models (e.g., Johnson-Cook) often fail to reconcile data from different stress states (e.g., tension vs. torsion) without ad hoc adjustments, and classical reductions of torsion test data (torque-twist to shear stress-strain) rely on assumptions incompatible with rate-dependent theories.

Methodology
The paper extends the averaging procedure developed by Le et al. (2020) for axially symmetric loading to arbitrary macroscopically uniform stress and strain states. The methodology rests on two primary hypotheses:

Equal Probability Hypothesis: Grain orientations in a polycrystal are distributed with equal probability (Haar measure in $SO(3)$ ).
Linear Partition of Slip: The plastic resolved shear strain rate on any slip system is proportional to the resolved shear stress acting on that system.

Using these hypotheses, the author defines mean resolved shear stress ( $\bar{\tau}$ ) and mean resolved elastic and plastic shear strains ( $\bar{\varepsilon}^e, \dot{\bar{\varepsilon}}^p$ ) as root-mean-square (RMS) averages over all slip-system orientations. Two exact identities for isotropic fourth-order tensors are derived to relate these averages to macroscopic invariants.

The constitutive framework is then constructed by:

Defining the macroscopic plastic flow direction based on the linear partition hypothesis.
Integrating the single-system TDT kinetics (thermally activated dislocation depinning, evolution of dislocation density $\rho$ , and effective disorder temperature $\chi$ ) driven by the macroscopic plastic power.
Formulating the equations of motion in time-rate form to accommodate arbitrary loading paths.
Deriving the torque–twist relation for bars and tubes directly from the theory, avoiding classical reductions that assume rate-independence.

Key Contributions

Derivation of Associated $J_2$ Flow Theory: The paper proves that the equal probability hypothesis combined with the linear partition of slip yields exactly the associated $J_2$ flow theory of classical plasticity. Specifically, the mean resolved shear stress is shown to be proportional to the von Mises equivalent stress ( $\bar{\tau} = \sigma_{eq}/\sqrt{15}$ ), and the resulting flow rule is the Prandtl–Reuss rule ( $\dot{\varepsilon}^p \propto \sigma'$ ). This structure emerges without adjustable parameters.
Generalization of TDT: The theory is generalized from uniaxial states to arbitrary stress states. The geometric factors connecting macroscopic variables to slip-system variables are updated (e.g., replacing $\pi/2$ with $\sqrt{5/3}$ ), but the dimensionless system remains consistent with previous uniaxial formulations.
Direct Torque-Twist Formulation: A boundary value problem is formulated for torsion that solves for torque and twist angle directly. This eliminates the need for classical reductions (like the Nadai construction) which presuppose a single-valued, rate-independent shear stress-strain curve.
Unified Parameter Identification: The author performs a joint identification of material parameters for copper using data from Hopkinson-bar tension, dynamic compression, and torsion tests across a wide range of temperatures (298 K to 1173 K) and strain rates (10 to 2300 s $^{-1}$ ).

Results

Parameter Consistency: A single set of material parameters ( $\tilde{\chi}_0, K_\rho, c_0, c_1$ ), consistent with a previous compression-only identification, successfully describes tension, compression, and torsion data. The root-mean-square error over 142 data points is 5.9%.
Resolution of Tension-Torsion Discrepancy: The paper addresses the tension-torsion discrepancy noted by Johnson and Cook (1983). The results indicate that this discrepancy is not a failure of the material law but a consequence of the initial dislocation state of the torsion specimens. When the torsion specimens are assigned a distinct initial state (specifically a higher initial dislocation density reflecting their specific annealing history), the theory fits both tension and torsion data accurately.
Limits of the Theory: The theory accurately captures behavior up to strains of approximately $\gamma = 3$ . Deviations occur at higher strains (Stage-IV hardening) where texture evolution and cell structure changes become dominant, phenomena outside the scope of the current isotropic model. Additionally, the theory excludes quasi-static regimes where recovery and recrystallization compete with depinning.

Significance
The paper demonstrates that Thermodynamic Dislocation Theory can be applied to general loading paths without crystal-plasticity computations, retaining the predictive power of the theory while recovering the standard structure of classical $J_2$ plasticity. By deriving the torque-twist relation directly and identifying parameters jointly across stress states, the work validates the theory's ability to describe metal plasticity from room temperature to elevated temperatures and over several decades of strain rates. Crucially, it resolves a long-standing experimental inconsistency (tension vs. torsion) by attributing it to specimen history rather than model inadequacy, reinforcing the robustness of the thermodynamic framework. The theory is presented as a closed macroscopic model (eight ordinary differential equations per material point) suitable for implementation in existing finite-element codes.

Technical Summary: Averaging in Thermodynamic Dislocation Theory

More like this