Plastic Potential in Plasticity Theory

When a material is stressed beyond a certain point, it ceases to behave elastically and undergoes a permanent, irreversible change in shape. This phenomenon, known as plasticity, is fundamental to fields ranging from civil engineering to materials science. While the onset of this permanent deformation is a familiar concept, a deeper question remains: once a material yields, what rules govern the direction and nature of its subsequent flow? Predicting this behavior is crucial for designing safe structures, optimizing manufacturing processes, and understanding geological phenomena.

This article provides a comprehensive overview of the plastic potential, a central concept in modern plasticity theory that answers this very question. The discussion is structured to build understanding from the ground up. In the "Principles and Mechanisms" chapter, we will delve into the theoretical framework, defining the crucial roles of the yield function and the plastic potential and distinguishing between associated and non-associated flow rules. Subsequently, the "Applications and Interdisciplinary Connections" chapter will demonstrate the power of this theory by examining its use in modeling diverse materials like metals and soils, its role in damage mechanics, and its integration into computational simulation. We begin by exploring the elegant geometric and physical principles that form the foundation of this powerful concept.

Imagine stretching a rubber band. It deforms, but when you let go, it snaps back to its original shape. This is ​​elasticity​​, a familiar concept, a temporary change. But what happens when you bend a paperclip? It bends, and it stays bent. You have pushed it beyond its elastic limit into a realm of permanent, irreversible change. This is the world of ​​plasticity​​, and understanding its rules is fundamental to a vast range of engineering, from shaping steel beams to predicting landslides.

After our introduction to the topic, we now dive into the "how" and the "why". How does a material "decide" to deform permanently? And once it does, in which direction does it flow? The answers lie in one of the most elegant constructs in mechanics: the interplay between a ​​yield function​​ and a ​​plastic potential​​.

Let’s picture the state of stress within a material as a point in a multi-dimensional "stress space". Think of it as a landscape. As long as the stress point stays within a certain "safe" region, the material behaves elastically, like a ball rolling on flat ground that always returns to its starting point if nudged. This safe zone is called the ​​elastic domain​​.

The boundary of this domain is the crucial part. It’s a "fence" or a surface that separates purely elastic behavior from the world of plasticity. This boundary is called the ​​yield surface​​, and it is mathematically defined by a ​​yield function​​, typically written as $f(\boldsymbol{\sigma}, \boldsymbol{\kappa}) = 0$. Here, $\boldsymbol{\sigma}$ represents the stress tensor (our location in the stress landscape), and $\boldsymbol{\kappa}$ represents a set of ​​internal variables​​ that describe the material’s history—think of it as the material’s memory of past deformations, which allows it to get stronger (or weaker) through a process called ​​hardening​​ or ​​softening​​.

As long as the stress state is inside the surface, so that $f(\boldsymbol{\sigma}, \boldsymbol{\kappa}) 0$, all is well and elastic. But when the load increases and the stress point reaches the surface, $f=0$, the material is on the brink of yielding. Any attempt to push the stress "outside" this boundary will trigger plastic deformation. This framework, governed by a set of admissibility and consistency conditions, forms the bedrock of modern plasticity theory.

So, our stress state is at the yield surface. We are about to deform plastically. But this deformation isn't just a single number; it's a tensor, describing how the material stretches, shears, and changes volume. In which "direction" in the space of all possible deformations will the material flow?

This is where the concept of the ​​plastic potential​​, $g(\boldsymbol{\sigma})$, comes into play. It’s a scalar function, just like the yield function, that defines another set of surfaces in our stress space. The rule—a truly profound postulate in plasticity—is that the direction of the plastic strain rate, $\dot{\boldsymbol{\varepsilon}}^p$, is always ​​normal​​ (perpendicular) to the surface of the plastic potential at the current stress state.

Mathematically, this is the famous ​​flow rule​​:

$\dot{\boldsymbol{\varepsilon}}^p = \dot{\lambda} \frac{\partial g}{\partial \boldsymbol{\sigma}}$

Here, $\dot{\lambda}$ is the ​​plastic multiplier​​, a positive scalar that tells us how much plastic flow occurs, while the gradient vector $\frac{\partial g}{\partial \boldsymbol{\sigma}}$ tells us the direction of that flow. The gradient of a scalar function at a point is always normal to the level surface passing through that point, a beautiful geometric fact that now has a profound physical meaning.

The simplest and most elegant first guess is to assume that the material flows in a direction normal to the yield surface itself. This means we choose the plastic potential $g$ to be the same as the yield function $f$. This is called an ​​associated flow rule​​. This isn't just a convenient choice; it can be derived from the ​​Maximum Dissipation Principle​​, which states that nature is in a sense "maximally efficient" at dissipating energy during plastic flow. For a given strain rate, the true stress state is the one that maximizes the rate of energy dissipation, and this leads directly to the normality rule for associated plasticity.

This idea of an associated flow rule is stunningly successful for some materials, most notably metals.

For metals, experiments show that yielding is largely unaffected by how much you squeeze them (hydrostatic pressure). Their strength depends on shear. This means their yield function, like the famous ​​von Mises criterion​​, only depends on the deviatoric part of the stress tensor, $\boldsymbol{s}$, which represents pure shear. In principal stress space, the von Mises yield surface is a perfect cylinder aligned with the hydrostatic axis ($\sigma_1=\sigma_2=\sigma_3$).

Now, let's apply our associated flow rule. The normal vector to the surface of a cylinder is always pointing radially outward, perfectly perpendicular to the cylinder's axis. It has no component along the hydrostatic axis. According to the flow rule, this means the plastic strain rate must also have no hydrostatic component. The trace of the plastic strain rate tensor, which represents the rate of volume change, must be zero: $\operatorname{tr}(\dot{\boldsymbol{\varepsilon}}^p) = 0$.

This means the theory predicts that the plastic flow of metals should be ​​isochoric​​, or volume-preserving! When you permanently bend a steel bar, it changes shape, but its total volume remains constant. This is a remarkable prediction, born from pure geometry and a beautiful physical principle, and it matches experimental observations with incredible accuracy.

But what about other materials, like sand, soil, or concrete? Their strength does depend on hydrostatic pressure—the harder you squeeze a pile of sand, the more shear it can resist before shifting. Their yield surfaces are not cylinders but are more like cones or pyramids, such as those described by the ​​Mohr-Coulomb​​ or ​​Drucker-Prager​​ criteria. The surfaces are sloped with respect to the hydrostatic axis.

If we apply the associated flow rule here, the normal vector to this sloped surface is no longer perpendicular to the hydrostatic axis. It now has a component that corresponds to a change in volume. Specifically, it predicts that as you shear the material, it should expand. This phenomenon is called ​​dilatancy​​, and it's real! If you shear a densely packed bag of sand, you can watch it puff up as the grains ride up and over one another. Once again, the theory makes a correct qualitative prediction.

Here, however, our beautiful, simple theory runs into a problem. While associated flow for soils correctly predicts the existence of dilatancy, it gets the amount wrong. Experiments consistently show that the amount of volume expansion predicted by an associated rule, which is rigidly tied to the material's internal friction (the slope of the yield cone), is often far greater than what is actually measured. The theory, in this case, is too predictive, linking two material properties that, in reality, are not so tightly coupled.

This is where the true genius of the plastic potential concept shines. What if the rule that defines the onset of yielding (the yield function $f$) is different from the rule that governs the direction of flow (the plastic potential $g$)?

This is the basis of ​​non-associated flow​​. We maintain the yield function $f$ to accurately model the material's strength. But for the flow rule, we introduce a separate plastic potential $g$, chosen specifically to model the material's deformation characteristics. Now, the plastic strain rate is normal to the surfaces of $g$, not $f$.

$\dot{\boldsymbol{\varepsilon}}^p = \dot{\lambda} \frac{\partial g}{\partial \boldsymbol{\sigma}} \quad \text{with} \quad g \ne f$

This master stroke decouples strength from flow. For our soil example, we can now choose a yield function $f$ (like Mohr-Coulomb) with a ​​friction angle​​ that perfectly matches the measured strength of the soil. Then, we can choose a different plastic potential $g$ with a ​​dilation angle​​ that perfectly matches the measured volume change. This gives us tremendous flexibility and allows our models to capture the complex behavior of these materials with much higher fidelity.

This powerful framework is even more robust than it first appears.

• ​​What about corners?​​ Some yield surfaces, like the ​​Tresca​​ or ​​Mohr-Coulomb​​ criteria, are not smoothly curved but have sharp edges and corners. At a smooth point, the normal direction is unique. But at a corner, which way does the normal point? The mathematical theory of convexity provides a beautiful answer: the ​​normal cone​​. At a corner, there is not a single normal vector but a whole "fan" of possible normal directions. The associated flow rule simply states that the plastic flow must lie somewhere within this cone, acknowledging that the flow direction may not be unique at these special points.

• ​​Thermodynamics, the Ultimate Arbiter.​​ This freedom to choose a separate plastic potential does not give us unlimited license. Any valid constitutive model must obey the laws of physics, most fundamentally the second law of thermodynamics, which demands that plastic deformation must always dissipate energy, never create it. This condition, $\boldsymbol{\sigma} : \dot{\boldsymbol{\varepsilon}}^p \ge 0$, places important constraints on the possible choices for $f$ and $g$, ensuring our models remain physically realistic.

• ​​Duality: Two Sides of the Same Coin.​​ There is an even deeper, more abstract layer of beauty here. The entire framework of plasticity can be formulated not only in the space of stresses (using a plastic potential $g$) but also in the space of strain rates (using a ​​dissipation potential​​ $\phi$). These two potentials, living in dual mathematical spaces, are elegantly connected through a relationship known as a Legendre-Fenchel transformation. They are two equivalent perspectives on the same physical reality, a profound symmetry that reveals the deep mathematical structure underlying the messy, irreversible process of plastic flow.

From a simple observation about a bent paperclip, we have journeyed through a landscape of stress, discovered the geometric rules that govern material flow, and developed a sophisticated framework that can be tailored to materials as different as steel and sand. The concept of the plastic potential is a testament to the power of combining physical intuition with elegant mathematics to describe the world around us.

In the previous chapter, we developed a rather beautiful piece of mathematical machinery: the idea of a ​​yield surface​​, $f$, which acts as a boundary to the elastic world, and a ​​plastic potential​​, $g$, which provides the "law of motion" for what happens when we cross that boundary. The direction of plastic flow, we saw, is always "normal" to the surface of this potential. When the potential and the yield function are one and the same ($g=f$), we have what is called an associated flow rule. When they are different ($g \neq f$), the rule is non-associated.

This might all seem a bit abstract, a formal game played with gradients and surfaces in a six-dimensional stress space. But the truth is far more exciting. This framework is not just mathematical scaffolding; it is a key that unlocks a profound understanding of how real materials—the steel in a bridge, the soil under a building, the aluminum in an airplane's wing, the rock in a mountain—actually behave. It is in the applications, in the confrontation with the messy reality of the physical world, that the true power and elegance of the plastic potential concept are revealed. Let's take a journey through some of these realms and see this idea at work.

Let's start with the most familiar case: a simple ductile metal, like steel or aluminum. What is the most salient feature of its plastic deformation? If you take a bar of steel and stretch it, it gets thinner, but its volume hardly changes at all. The atoms just slide past one another, rearranging themselves. This property is called ​​plastic incompressibility​​.

How does our theory capture this? Beautifully, as it turns out. For most metals, the onset of yielding depends on the differences between principal stresses (the shear), not on the average, hydrostatic pressure. Whether you're at sea level or the bottom of the Mariana Trench, the shear stress required to permanently bend a piece of steel is almost the same. This means the yield function, $f$, for a metal is independent of pressure. Famous criteria like the von Mises, or $J_2$, theory embody this, with yield functions that depend only on deviatoric stress invariants like $J_2$.

Now, if we adopt an associated flow rule ($g=f$), the plastic potential is also independent of pressure. What is the consequence? The rule for plastic flow is $\dot{\boldsymbol{\varepsilon}}^p = \dot{\lambda} (\partial g / \partial \boldsymbol{\sigma})$. The volumetric part of this flow is its trace, $\text{tr}(\dot{\boldsymbol{\varepsilon}}^p)$. But if $g$ does not depend on hydrostatic pressure, its gradient with respect to the stress tensor, $\partial g / \partial \boldsymbol{\sigma}$, must be a traceless (deviatoric) tensor. The result is immediate and elegant: $\text{tr}(\dot{\boldsymbol{\varepsilon}}^p) = 0$. Plastic incompressibility is not an extra assumption we have to tack on; it is a direct and natural consequence of a pressure-independent, associated plastic potential!

This "normality rule" gives us testable, quantitative predictions. For any given stress state on the yield surface, we can calculate the gradient and thus predict the exact ratios of plastic strain rates in different directions. For a metal under a state of stress that puts it on its von Mises yield surface, the theory tells us precisely how it will begin to flow, a prediction that matches experiments with remarkable accuracy.

The associated flow rule works wonderfully for metals, so you might ask, why bother with anything else? Why introduce the complication of a separate plastic potential, $g$? The answer is that the world is full of materials that are not as well-behaved as a simple metal. And for these materials, the associated flow rule makes predictions that are spectacularly wrong.

The Ground Beneath Our Feet: Geomechanics

Consider soil, sand, or rock. Unlike metals, these materials are "frictional." Their strength depends enormously on how much they are squeezed. A handful of sand is weak, but the confined sand deep within the earth can bear the weight of mountains. This means their yield surfaces, described by criteria like the Mohr-Coulomb or Drucker-Prager models, are strongly dependent on pressure.

What happens if we naively apply an associated flow rule here? The yield surface is steeply inclined in stress space, representing the high friction. The normal vector to this surface therefore has a large component pointing "outward" towards expansion. An associated flow rule ($\psi = \phi$, where $\psi$ is the dilatancy angle and $\phi$ is the friction angle) thus predicts that when sheared, a dense sand should expand in volume at a tremendous rate. While dense sands do expand slightly when sheared—a phenomenon called dilatancy—the amount predicted by the associated rule is far beyond anything observed in a laboratory.

This is where the genius of the non-associated framework comes in. We realize that the physics of yielding (governed by inter-particle friction, $f$) is distinct from the physics of flowing (governed by particle rearrangement, $g$). We can therefore "decouple" them. We use the experimentally measured yield surface $f$ (with its high friction angle $\phi$) to tell us when the material will fail. But we devise a separate plastic potential $g$, with a much shallower slope corresponding to the small, observed dilatancy angle $\psi$, to tell us how it will flow. By choosing $g \neq f$, we can accurately model both the high strength and the modest volume change of geological materials. Without the concept of a non-associated plastic potential, modern soil mechanics and geotechnical engineering would be impossible.

Shaping Our World: Anisotropic Metals

Another fascinating case arises in manufacturing. A sheet of steel rolled at the mill is not the same in all directions; the grains get elongated. This makes the sheet anisotropic. Its yield strength might be different along the rolling direction versus across it. Critically, its flow behavior is also anisotropic. When you pull a sample in one direction, the ratio of how much it shrinks in width to how much it shrinks in thickness—a quantity called the Lankford coefficient, or $r$-value—is a key parameter. This $r$-value determines whether the sheet can be successfully stamped into a complex part, like a car door, without wrinkling or tearing.

Once again, we find that an associated flow rule is too restrictive. A single anisotropic yield function, like the Hill's 1948 criterion, can be calibrated to match the directional yield strengths. But if we use that same function as the plastic potential, the predicted $r$-values often don't match experiments. The anisotropy of strength is simply different from the anisotropy of flow.

The solution is the same as for soils: non-associated plasticity. A sophisticated modeling approach uses two different functions (or two sets of parameters for the same function family): one for the yield surface $f$, calibrated to directional strength measurements, and another for the plastic potential $g$, calibrated to directional $r$-value measurements. This decoupling allows for high-fidelity predictions of sheet metal forming operations.

So far, we have talked about changing shape. But plasticity is also the prelude to fracture. In a ductile metal, failure is not an instantaneous event. It begins with the nucleation of tiny voids, often around small impurities. As the material is stretched, these voids grow and eventually link up, leading to a crack.

Look closely at this process: the growth of voids means the average volume of the material is increasing. We have plastic dilatancy, even in a metal! Our beautiful, incompressible $J_2$ plasticity theory can't describe this.

The key is to build a plastic potential that "knows" about the voids. This is the idea behind the celebrated Gurson-Tvergaard-Needleman (GTN) family of models. The GTN potential is a modification of the $J_2$ function, but it includes terms that depend on both the hydrostatic pressure and the current void volume fraction, $f$. Under high tensile pressure (a positive mean stress), the potential surface is "pulled in," making it easier for the material to yield.

Let's look at the flow rule, $\dot{\boldsymbol{\varepsilon}}^p = \dot{\lambda} (\partial g / \partial \boldsymbol{\sigma})$. Because the GTN potential $g$ now depends on pressure, its gradient has a non-zero trace. And just like that, $\text{tr}(\dot{\boldsymbol{\varepsilon}}^p) \neq 0$. The model automatically predicts plastic volume growth, which is physically interpreted as the growth of the voids. The plastic potential has become the engine for modeling the accumulation of damage. It provides a seamless bridge from the continuum-level description of plasticity to the micro-mechanics of ductile fracture.

We live in an age where much of engineering design and scientific discovery is done not in a physical lab but in a computational one. How do we build a bridge that can withstand an earthquake or design a car that keeps its occupants safe in a crash? We simulate it using tools like the Finite Element Method (FEM). These computer programs solve the laws of physics for millions of tiny elements that make up the structure.

At the heart of every one of these elements is a constitutive model—a set of rules that tells the computer how that little piece of material behaves. And for plastic behavior, that model is a direct implementation of the theory we've been discussing. Algorithms like the ​​return mapping​​ scheme are computational workhorses that, at every tiny time step of a simulation, check if the stress has crossed the yield surface $f$. If it has, the algorithm uses the plastic potential $g$ to calculate how much plastic flow occurs and how the stress should be "returned" back to the new, updated yield surface. Our abstract potential function becomes a concrete set of instructions in a computer code that makes billion-dollar engineering decisions.

This connection reveals a final, deep point. The choice between an associated and non-associated flow rule is not just a matter of physical accuracy; it has profound mathematical consequences for the simulations. An associated flow rule leads to a symmetric material stiffness matrix, a property that guarantees stable and efficient numerical solutions. A non-associated rule, while often more realistic, produces a non-symmetric stiffness matrix. This can lead to computational difficulties, slower convergence, and even non-unique solutions in a simulation. This creates a fascinating tension for engineers and scientists: a trade-off between the physical realism we desire and the mathematical robustness we need for our digital tools.

In the end, we see that the plastic potential is a remarkably versatile and unifying concept. It starts as a simple geometric "compass" for plastic flow. Yet, in our journey, we have seen it explain the silent, constant-volume deformation of pure metals, the dilating crunch of the earth, the complex shaping of anisotropic alloys, and even the microscopic birth of a crack. From the design of new materials to the code running on a supercomputer, this single idea provides a coherent and powerful framework for understanding the rich mechanical world in which we live.