Updated Lagrangian Formulation

In the fields of engineering and physics, accurately simulating how structures behave under extreme loads—bending, twisting, and deforming significantly—is a critical challenge. This realm of nonlinear mechanics requires a robust framework for tracking motion and forces as an object's shape changes dramatically. The central problem is choosing a consistent "map" or reference frame to describe this deformation. This choice gives rise to two powerful and distinct schools of thought: the Total Lagrangian and the Updated Lagrangian formulations. Understanding the difference between looking back to a fixed past versus constantly updating to the present is key to mastering modern computational analysis.

This article will guide you through this fundamental distinction. First, in the "Principles and Mechanisms" chapter, we will dissect the core ideas behind each formulation, exploring their different mathematical languages for describing stress and strain, how they are implemented computationally, and how they both elegantly capture the crucial concept of geometric stiffness. Following this, the "Applications and Interdisciplinary Connections" chapter will demonstrate why this choice is not merely academic, showcasing how the Updated Lagrangian approach provides an indispensable perspective for tackling complex real-world problems from structural buckling and contact to fracture mechanics and advanced manufacturing processes.

Imagine you are trying to describe the journey of a ship on a vast, ever-changing ocean. You have a fundamental choice to make. Do you describe its position at every moment by referring back to its original port of departure? Or do you describe its next movement based on its current location and the sea conditions right there? This simple question lies at the heart of one of the most powerful distinctions in computational mechanics: the choice between a worldview fixed in the past and one that is constantly updated to the present.

When a body deforms—a bridge under load, a car tire hitting a pothole, a piece of metal being stamped into shape—every particle within it moves. To analyze this complex dance, we need a "map," a coordinate system to track the motion. The two great schools of thought in nonlinear mechanics are born from the choice of this map.

The first approach is the ​​Total Lagrangian (TL) formulation​​. Think of it as the historical method. We plant our flag firmly in the initial, undeformed configuration of the body, which we call the ​​reference configuration​​, $\mathcal{B}_0$. Every single measurement, every calculation, no matter how contorted the body becomes, is referred back to this original, pristine state. It’s like navigating by always giving directions from your childhood home; the starting point is fixed forever. All questions are of the form: "Where is the particle that started at position $\mathbf{X}$ now?"

The second approach is the ​​Updated Lagrangian (UL) formulation​​. This is the "here and now" method. It embraces the change. For each step of our analysis, the "reference" is the body's most recently calculated shape, the ​​current configuration​​, $\mathcal{B}_t$. We plant our flag again, but this time on a moving landscape. All questions are of the form: "Starting from where it is now, where will this particle go next?" After we find the answer, this new location becomes our updated reference for the next step.

This isn't just a philosophical choice; it dictates the very language we must use to describe the physics of deformation.

Just as you wouldn't use a 16th-century map to navigate a modern city, our choice of reference frame demands its own set of "yardsticks" for measuring stretch (strain) and force (stress).

In the Total Lagrangian world, where we are always looking back from the present to the undeformed past, we use ​​material measures​​. A prime example is the ​​Green-Lagrange strain tensor ($\mathbf{E}$)​​. It measures the change in the squared distance between any two nearby particles but describes that change in the coordinate system of the original, undeformed body. It’s a historical record of stretch. The stress that naturally pairs with this strain measure, in an energetic sense, is the ​​Second Piola-Kirchhoff stress tensor ($\mathbf{S}$)​​. It's an abstract stress measure, also defined on the original configuration, that represents the forces of the deformed body transformed back into the language of the past. The beauty of this pairing is its energetic consistency, a concept known as ​​work conjugacy​​. The work done during a small deformation can be expressed perfectly as $\mathbf{S} : \delta\mathbf{E}$.

In the Updated Lagrangian world, we live in the present. Here, we use ​​spatial measures​​. The stress is the one we can physically imagine and measure: the ​​Cauchy stress ($\boldsymbol{\sigma}$)​​. This is the true, physical force per unit of current, deformed area. It’s the stress you'd measure if you could put a tiny sensor on the deforming body. The strain measure that naturally pairs with it is a rate quantity, the ​​rate-of-deformation tensor ($\mathbf{d}$)​​, which describes the instantaneous rate of stretching at the current moment. The work rate is simply and elegantly given by $\boldsymbol{\sigma} : \mathbf{d}$.

These two languages, the material and the spatial, seem completely different. But they describe the same physical reality. And there is a "Rosetta Stone" that allows us to translate between them. The power—the rate of work being done—must be the same regardless of our description. This gives rise to a beautiful identity: the work rate calculated in the past, $\mathbf{S} : \dot{\mathbf{E}}$, must equal the work rate calculated in the present, $\boldsymbol{\tau} : \mathbf{d}$ (where $\boldsymbol{\tau}$ is the ​​Kirchhoff stress​​, a close cousin of Cauchy stress). This ensures that no matter which reference frame we choose, the physics remains unified and consistent.

How does a computer, which only understands numbers and algorithms, handle these concepts? In the Finite Element Method (FEM), we chop the body into small pieces called elements. The magic happens through a clever mapping.

Every element, regardless of its final shape, starts its life in the computer's imagination as a perfect, simple "parent" element—a pristine square or cube in a local coordinate system $(\xi,\eta)$. We then provide a set of rules, called ​​shape functions ($N_a$)​​, to map this ideal parent element to its actual shape in the physical world.

Here's where our two worldviews diverge again.

In the ​​Total Lagrangian​​ formulation, we use this map only once: to relate the parent element to the element's shape in the initial, undeformed body. We calculate the necessary derivatives of the shape functions with respect to the material coordinates $\mathbf{X}$ (i.e., $\nabla_{\mathbf{X}}N_a$) at the very beginning and then store them. They never change for the rest of the simulation. The geometry is fixed in time.

In the ​​Updated Lagrangian​​ formulation, the game is different. Since our reference frame is the current shape, the map from the parent element to the physical element must be re-established at every step of the deformation. The element's geometry is constantly changing, so the derivatives of the shape functions with respect to the current spatial coordinates $\mathbf{x}$ (i.e., $\nabla_{\mathbf{x}}N_a$) must be recomputed in every iteration. This might sound more computationally expensive, and it often is, but as we shall see, this constant awareness of the "now" gives the UL formulation tremendous power.

When we solve these nonlinear problems, we are essentially trying to find the equilibrium shape where all forces balance. The Newton-Raphson method, a standard tool for this, is like a highly intelligent guess-and-check. To make an intelligent guess, it needs to know the system's ​​tangent stiffness ($\mathbf{K}_T$)​​—how much the internal forces will change in response to a small nudge in the structure's position.

Intriguingly, this stiffness comes from two sources. The first is obvious: the ​​material stiffness​​. This is just the intrinsic stiffness of the material itself; steel is stiffer than rubber. But the second source is more profound and beautiful: the ​​geometric stiffness​​.

Geometric stiffness is the change in stiffness that arises purely because the body is already under stress and has changed its shape. The classic example is a guitar string. An unstretched string is flimsy. But once you tighten it, it becomes very stiff to a sideways pluck. This added stiffness doesn't come from a change in the steel itself, but from the tension it's under. This is a geometric stiffening effect. Similarly, a tall, slender column under compression becomes less stiff and will eventually buckle—a geometric softening effect.

This phenomenon arises naturally from the mathematics of both of our formulations. When we linearize the virtual work expressions to find the tangent stiffness, the nonlinear terms in the strain definitions (like the $\mathbf{F}^T\mathbf{F}$ in the Green-Lagrange strain) give birth to a new stiffness term. This term is always proportional to the current stress in the structure—the Second Piola-Kirchhoff stress $\mathbf{S}$ in the TL formulation, or the Cauchy stress $\boldsymbol{\sigma}$ in the UL formulation. It is a beautiful example of unity: two very different mathematical paths lead to the same essential physical insight.

If both TL and UL are exact descriptions of the same physics, why do we need both? Because while they are equivalent in theory, they are not equivalent in practice. The choice depends on the story the material and the boundary conditions have to tell.

​​The Case for Total Lagrangian:​​ The TL formulation shines when a material has a strong "memory" of its original state. The best example is ​​hyperelasticity​​ (think rubber or biological soft tissue). The material's strain energy is a function of its total deformation from the beginning. Using the TL formulation with its material strain $\mathbf{E}$ is incredibly elegant. The stress $\mathbf{S}$ comes directly from differentiating the energy function, and this leads to a symmetric tangent stiffness matrix, which is a huge gift for computational efficiency.

​​The Case for Updated Lagrangian:​​ The UL formulation is the hero when the "now" is what governs the physics. This happens in two major scenarios.

1. ​​When the world acts on the current shape.​​ Imagine the wind pressure on a flapping flag or the contact force between a tire and the road. These forces are defined on the current, deforming geometry. A pressure load that always acts normal to the surface is called a ​​follower load​​. In the UL formulation, where the current shape is the reference, these conditions are laughably easy to apply. In the TL formulation, you would face the nightmare of transforming these ever-changing loads back to the fixed, initial configuration. For problems dominated by evolving contact and follower forces, UL is the undisputed champion.

2. ​​When the material's memory is short.​​ Consider metal ​​plasticity​​. When you bend a paperclip, it stays bent. Its future behavior depends not on its original straight shape, but on its current bent shape and stress state. The laws of plasticity are almost always written as rate equations: the rate of plastic strain depends on the current stress. The UL formulation, built on rates and the current configuration, is the natural home for such models.

This brings us to one last, subtle, and beautiful point. When using these rate laws in the UL framework, a startling problem appears: a simple material time derivative of stress is not objective. That is, if you subject the material to a pure rigid-body rotation (which should induce no stress), the equations will predict a fictitious stress! To cure this, physicists and engineers developed the concept of an ​​objective stress rate​​, such as the ​​Jaumann rate​​. This is a more sophisticated derivative that is "smart" enough to ignore pure rotations, ensuring that our mathematical model respects physical reality. It is a perfect final illustration of the journey: the choice of a reference frame leads to a specific mathematical language, which in turn reveals subtle challenges that demand even more elegant mathematical solutions, all in the service of capturing the true physics of our world.

Now that we have grappled with the mathematical machinery of the updated Lagrangian formulation, you might be asking a fair question: Why bother? We already had a perfectly good way of looking at the world with the total Lagrangian approach, where everything is neatly referred back to its original, pristine state. Why introduce this new, constantly-shifting viewpoint?

The answer, in a word, is perspective. Imagine you are giving a friend directions for a cross-country road trip. You could give them the entire set of instructions before they leave, a complete manifest of every turn from their starting driveway to their final destination. This is the total Lagrangian (TL) approach—utterly complete, with a single, unchanging reference point. But what if the trip involves navigating a chaotic, ever-changing city, or venturing into uncharted territory where roads might not exist as planned? In that case, a better strategy might be to update the directions at every major intersection, using the current location as the new starting point. This is the spirit of the updated Lagrangian (UL) formulation.

In the world of physics and engineering, both methods, if correctly applied, must lead to the same final answer. They are two different but equally valid "accounting systems" for the same physical reality. For a complex material like a plastic that deforms permanently, a careful calculation shows that the final stress and stiffness can be found with either viewpoint, though the book-keeping along the way looks very different. The art and the beauty lie in knowing which perspective is more natural, more efficient, and more insightful for the problem at hand. It is this art of choosing our perspective that reveals the deep connections between seemingly disparate fields, from the buckling of a bridge to the fracture of a jet engine turbine blade.

Let’s start with the everyday world of structural engineering. Consider a flexible ruler or a fishing rod. You can bend it into a dramatic curve—a large rotation—but the material itself is barely stretching. This is the classic "large rotation, small strain" problem. How can we describe this?

A wonderfully intuitive application of the updated Lagrangian philosophy is the co-rotational method. Instead of staying at a fixed observation post, we imagine ourselves shrinking down and "riding along" on a small segment of the bending ruler. From our moving, rotating vantage point, the deformation looks very simple! We only see a tiny amount of stretching or squashing. We can use the simple math of small strains locally, while the complexity of the large rotation is handled by tracking our moving frame of reference. It’s a beautifully efficient way to separate the large, rigid motion from the small, stress-inducing deformation.

Now, let's think about how things fail. A key failure mode is buckling—the sudden loss of stiffness when a structure is compressed. Think about pressing down on an empty soda can. For a while it holds firm, and then suddenly, it crumples. The tangent stiffness of a structure—its resistance to a small extra push—isn't just a property of the material. It has two parts: a material stiffness, which is the inherent rigidity of the stuff it's made of, and a geometric stiffness, which depends on the stress the structure is already under. Pushing on the can creates compressive stress, which contributes a negative term to the geometric stiffness. At the buckling point, this negative geometric stiffness has grown so large that it effectively cancels out the positive material stiffness, and the total stiffness drops to zero. Catastrophe!

Both TL and UL formulations can capture this, as they must. They both account for material and geometric stiffness. However, the details of the formulation matter immensely in practice. A comparison of a standard UL implementation for a post-buckling beam against a standard TL one reveals a subtle trap. Some UL methods, which build up the stress incrementally without a perfect "memory" of the total deformation, can accumulate small errors during large rotations. This is like our driver who gets directions at each turn but has no overarching map; a series of small misjudgments can lead them astray. This can lead to slower computational convergence and less accurate results compared to a well-formulated TL approach that, by its nature, never forgets the original map. This teaches us a profound lesson: while the UL framework is powerful, the specific "hypoelastic" material laws used within it must be chosen with care.

The choice of viewpoint also has dramatic practical consequences in problems involving contact. Imagine simulating the compression of a rubber seal. We need to track the contact pressure. In a total Lagrangian world, the points on the seal's surface are labeled once in the reference configuration. We can attach the history of the contact pressure to these fixed material points like pinning notes to a corkboard. It's straightforward. In the updated Lagrangian world, the surface itself is moving and deforming with each increment. The "corkboard" is made of pliable dough. A pressure value recorded at one step refers to an area that will have changed its size and orientation by the next step. To be consistent, we must perform a careful mapping of these history-dependent quantities from one configuration to the next. Failing to do so introduces errors, like an accountant who forgets to adjust for inflation. For this reason, the history-keeping simplicity of the total Lagrangian formulation makes it a very compelling choice for many contact problems.

While the total Lagrangian view has its elegance, the updated Lagrangian formulation truly comes into its own when the geometry of an object changes so profoundly that the original shape becomes a distant memory.

Consider one of the most critical problems in materials science and safety engineering: fracture. When we model the tip of a crack in a ductile metal, theories based on small strains predict that the stress and strain should become infinite right at the tip. This is a mathematical absurdity. What happens in reality? The material yields, and the crack tip blunts, rounding out and relieving the infinite stress. This process of blunting involves extreme, localized changes in geometry. The original, perfectly sharp crack tip is a poor reference point for what’s happening.

This is a problem that cries out for an updated Lagrangian formulation. By constantly updating our frame of reference to the current, blunted shape of the crack, we can accurately track the evolution of stress and strain in this critical region. This allows engineers to predict the conditions under which a crack will grow and a component will fail. Sophisticated UL simulations, which incorporate advanced models for plasticity and employ adaptive numerical techniques to handle the severe mesh distortion, are the workhorses of modern fracture mechanics. They are the reason we can step onto an airplane with confidence that its engine components have been designed to resist a level of damage far beyond what they will ever experience.

A similar story unfolds in manufacturing. Processes like forging, rolling, and stamping involve taking a simple block of metal and deforming it into a complex shape like a car door or a turbine disk. The deformations are enormous. The UL approach, which follows the material step-by-step as it flows and deforms, is the natural and often only feasible way to simulate these processes, allowing for the optimization of manufacturing techniques and the design of stronger, lighter components.

The power of the Lagrangian viewpoints extends far beyond solid mechanics. It provides a fundamental framework—a computational stage—upon which we can simulate all manner of complex, coupled physical phenomena.

Think of piezoelectric materials, which deform when a voltage is applied and generate a voltage when deformed. They are the heart of countless sensors and actuators. At very high electric fields, these materials behave nonlinearly; their properties change with the field strength. How do we model this? We use the exact same framework. The equations of equilibrium are augmented with Gauss's law for electricity, and the constitutive laws now link stress, strain, electric field, and electric displacement. Whether we use a TL or UL approach, the underlying logic of tracking the system's evolution and linearizing the equations to find the next step remains the same. The framework is general enough to accommodate this rich multiphysics behavior.

This generality extends to other advanced frontiers of simulation. Modern meshless methods, which do away with the rigid connectivity of traditional finite element meshes, often rely on an updated Lagrangian formulation. Why? Because the shape functions that describe the motion are constructed "on the fly" in the current configuration of scattered nodes, a concept that fits hand-in-glove with the UL philosophy. Similarly, problems involving fluid-structure interaction—the flapping of a flag in the wind, the flow of blood through a flexible artery—are often tackled with methods that blend the Lagrangian view (for the structure) and an Eulerian view (for the fluid) in a way that continuously updates the domain, a clear echo of the UL approach.

So, which is better? Total or updated Lagrangian? It is, of course, the wrong question. It is not about better, but about what is more illuminating. The total Lagrangian formulation is the historian's view: grounded, consistent, and powerful when history matters on a fixed landscape. The updated Lagrangian formulation is the explorer's view: adaptive, immediate, and indispensable when the landscape itself is the thing that is changing.

The true understanding comes not from picking a side, but from appreciating the duality. It is the ability to see a physical problem and know whether to approach it with the grand, overarching map of the historian or the nimble, step-by-step compass of the explorer. In this choice, we find the flexibility and power that allows us to model our complex, ever-deforming world.