🔢 mathematics

Plasticity as Directional Stationarity: Yielding, Flow, and Hardening from One Functional

This paper unifies the fundamental components of plasticity theory—including yielding, flow, and hardening—into a single scalar functional framework based on directional stationarity over one-sided admissible paths, thereby deriving elastic inequalities, loading-unloading conditions, and both associated and non-associated flow rules while accommodating various hardening mechanisms and rate-dependent evolution.

Original authors: Huilong Ren

Published 2026-08-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Huilong Ren

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

Materials like steel, aluminum, and even the soil beneath our feet share a peculiar trait: they can bend and stretch without snapping, but only up to a point. Once pushed past a certain limit, they undergo a permanent change. Engineers call this plasticity. It is the reason a paperclip stays bent after you twist it, or why a car's crumple zone absorbs the energy of a crash by deforming rather than shattering. For decades, scientists have described this behavior by breaking it down into separate rules. One rule dictates how the material springs back when the force is removed; another defines the exact moment it starts to yield; a third describes how it flows like a thick fluid; and yet another tracks how the material hardens or softens as it deforms. While this step-by-step approach works well for computers, it treats these behaviors as distinct ingredients, missing the deeper connection that ties them all together.

A researcher at Tongji University in Shanghai has proposed a new way to look at this phenomenon, suggesting that all these separate rules actually stem from a single, unified source. Instead of juggling multiple equations, the study shows that the entire behavior of a plastic material can be generated from one simple mathematical object, which the author calls a functional. Think of this functional as a single landscape or terrain. The material's state is like a point on this landscape, and the rules of plasticity are simply the directions a ball would roll if placed on that terrain. By analyzing how this single landscape changes when the material is nudged in specific, allowed directions, the researcher found that the familiar rules of stress, flow, and hardening emerge naturally, without needing to be imposed separately.

The core of this discovery lies in a concept called directional stationarity. In everyday terms, imagine standing on a hillside. If you are at the very bottom of a valley, you are in a stable position; any small step you take will make you go uphill. This is similar to how a material behaves when it is elastic, or springy. It resists change. However, if you are on a flat plateau or a slope, the rules change. The researcher found that the moment a material begins to yield and deform permanently is exactly when the landscape becomes flat in a specific, allowed direction. The material is no longer fighting to stay put; it is finding a path where it can move without increasing its energy cost. This single condition—that the material seeks a flat spot in a specific direction—automatically generates the complex rules engineers use to predict when a bridge will buckle or how a metal sheet will stretch.

One of the most significant findings is how this approach handles materials that do not behave symmetrically. In many real-world scenarios, such as soil under a foundation or certain types of rock, the way a material flows is not perfectly aligned with the force pushing it. Traditional models often struggle to describe this mismatch without adding extra, complicated terms. The new formulation handles this naturally. It allows the direction of flow to be different from the direction of the force, simply by choosing a different path on the landscape. This means the model can describe both standard metals, which flow in a predictable way, and more complex materials like sand or concrete, which might expand or contract as they shear, all within the same framework.

The study also clarifies how materials get stronger or weaker as they deform. As a metal bar is stretched, it often becomes harder to stretch further, a process known as hardening. The research shows that this hardening, along with the shifting of the material's internal structure, is just a change in the shape of that single landscape. The landscape can expand, shrink, or shift its position, and these movements correspond exactly to the material's changing resistance. By treating the material's history and its internal variables as part of this single landscape, the researcher can describe complex behaviors like the way a material remembers past deformations or how it reacts to rapid changes in speed, all derived from the same starting point.

To prove that this abstract idea works in the real world, the researcher applied it to four specific physical problems. First, they modeled a hollow tube being twisted. The model successfully predicted how different layers of the tube would start to deform at different times, creating distinct rings of activity within the material. Second, they looked at a bar with a varying thickness, showing how the internal forces shift smoothly through the material as it stretches. Third, they examined a thick ring under pressure, a classic test for materials that are sensitive to how hard they are squeezed. Finally, they modeled a hole expanding inside a solid block, a scenario crucial for understanding how soil behaves around tunnels or foundations. In every case, the single functional approach produced the correct results, separating the material's strength from its tendency to expand or contract, and showing how these properties interact.

This work does not just offer a new way to calculate; it offers a new way to see. It suggests that the complexity of plastic deformation is an illusion created by looking at the problem in pieces. When viewed as a whole, the behavior of a yielding material is governed by a simple, elegant principle: the material seeks a state of balance in the direction it is allowed to move. This insight could lead to more accurate simulations for designing safer structures, from skyscrapers to aircraft, by ensuring that the underlying physics is treated as a unified whole rather than a collection of separate rules. The researcher has provided the tools to turn this unified view into practical calculations, offering a clearer path to understanding how the solid world around us bends, flows, and holds together.

Technical Summary: Plasticity as Directional Stationarity

Problem Statement
Traditional plasticity theory is conventionally organized through a collection of distinct components: an elastic law, a yield condition, a flow rule, hardening relations, and loading–unloading conditions. While this framework has proven effective in computational applications, it separates relations that share a common energetic and directional origin. The paper addresses the need for a unified variational statement that derives these constitutive relations from a single generating object, thereby unifying the treatment of associated and non-associated flow, mixed hardening, gradient effects, and rate dependence.

Methodology
The paper proposes a formulation based on the directional stationarity of a single scalar functional, $\Pi$ , evaluated over a set of one-sided admissible plastic paths.

Scalar Functional: The core of the formulation is a functional defined over the domain $\Omega$ , comprising elastic energy $\phi$ , plastic-state energy $\psi_p$ , and activity resistance $W$ .
$\Pi[u, \varepsilon^p, \lambda] = \int_\Omega [\phi(\varepsilon - \varepsilon^p) + \psi_p(\varepsilon^p) + W(\lambda)] dx - \ell(u)$
Directional Stationarity: Instead of imposing separate yield and flow rules, the paper restricts the first variation of the functional to an admissible plastic tangent cone. The admissible variations are parameterized by non-negative activity increments $\delta\lambda_a \geq 0$ and direction vectors $M_{Ia}$ .
Derivation of Constitutive Laws:
- The first variation yields the stress $\sigma$ , internal-variable forces $X_I$ , and resistance $R$ .
- Restricting this variation to the admissible tangent defines a reduced directional force $F_a$ for each mechanism:
  $F_a = \sum_I X_I \cdot M_{Ia} - R_a$
- The condition of directional stationarity on the non-negative cone leads directly to the Kuhn–Tucker conditions:
  $F_a \leq 0, \quad \delta\lambda_a \geq 0, \quad F_a \delta\lambda_a = 0$
- These conditions simultaneously define the yield boundary ( $F_a=0$ ), the elastic inequality ( $F_a < 0$ ), and the complementarity conditions for loading and unloading.

Key Contributions

Unified Variational Framework: The paper demonstrates that the yield function can be identified with the negative first variation per unit admissible plastic activity. This approach generates stress, internal forces, and resistance derivatives from one functional, while the admissible tangent set defines the irreversible mechanisms.
Associated and Non-Associated Flow:
- Associated Flow: Arises naturally when the admissible tangent is selected as a support direction of the admissible plastic metric (the subdifferential of a support function). In this case, the flow direction is normal to the yield surface.
- Non-Associated Flow: Is accommodated by assigning a tangent direction that is not normal to the resulting yield boundary. The paper provides a local representation for regular non-associated directions via positive rescaling, provided the directional work remains positive.
Self-Similar Associated Families: A self-similarity analysis reveals that positively homogeneous stress gauges generate a broad family of yield surfaces. The paper shows that a single gauge $\rho$ fixes the surface shape and flow direction, while the resistance term $R$ controls the scale (isotropic expansion), and the plastic-state energy $\psi_p$ controls the center (kinematic translation). This separates the roles of geometry, size, and translation.
Generalized Hardening and Gradients: The framework naturally incorporates:
- Mixed Hardening: Isotropic, kinematic, and coupled hardening are derived from distinct derivatives of the resistance $W$ and plastic-state energy $\psi_p$ .
- Gradient Plasticity: By including gradient terms in the resistance functional (e.g., $W(\lambda, \nabla\lambda)$ ), the formulation yields nonlocal resistance terms and natural micro-boundary conditions directly from the first variation.
- Multi-Mechanism Plasticity: Independent activity variables allow for the modeling of multiple independently activated mechanisms (e.g., crystal slip systems or multi-surface models) with latent hardening coupling.
Viscoplastic Extension: The construction is extended to rate-dependent evolution by introducing a viscous potential $V(\dot{\lambda})$ . The rate-independent limit is recovered as the viscosity scale tends to zero, converging to the standard KKT graph.

Results and Analytical Examples
The paper validates the formulation through four closed-form analytical solutions:

Multi-Threshold Torsion: Illustrates the formation of spatially separated active annuli in a circular tube under torsion, demonstrating how multiple thresholds and hardening moduli create distinct activation radii.
Non-Uniform Bar with Kinematic Hardening: Provides an explicit solution for a tapered bar, showing how the backstress field and plastic strain distribute spatially under monotone loading.
Hill-Type Annulus: Recovers a pressure-sensitive limit state for a thick annulus under internal pressure, explicitly separating the pressure-dependent strength boundary (controlled by parameter $\alpha$ ) from the plastic dilatancy (controlled by parameter $\beta$ ).
Elastic–Plastic Cavity Expansion: Solves for the expansion of a cylindrical cavity, demonstrating the decoupling of strength and flow kinematics. The solution shows that while the stress field depends on the yield surface shape ( $\alpha$ ), the displacement field and plastic dilatation depend on the flow direction ( $\beta$ ), allowing for the modeling of non-associated flow.

Significance and Claims
The paper claims that this directional stationarity formulation offers a common variational statement for relations that are traditionally separated. By identifying the yield function with the negative first variation, the framework unifies the derivation of constitutive laws. It explicitly separates the geometric shape of the yield surface from its scale and translation, and it provides a rigorous mechanism for handling non-associated flow without ad-hoc modifications to the yield function.

The author notes that the formulation covers mixed hardening, multiple plastic activities, gradient resistance, and rate dependence within a single structure. The analytical examples serve as complementary checks, confirming that the formulation correctly reproduces known limit states and distinct displacement fields associated with normal and non-normal flow. The paper concludes that the most immediate extensions involve establishing well-posedness for specific non-associated and gradient models, developing consistent finite-element linearizations, and calibrating the model against cyclic and rate-dependent experiments.

More like this