Geometric Nonlinearity
Large Deformation, NLGEOM & Post-Buckling

Good instinct. For beams and shells, a rough rule of thumb is: if tip displacement exceeds about 5% of the span, your linear result is suspect. At 30% of span you're firmly in large-deformation territory. The three physical effects you're missing in linear analysis are: (1) the actual load path changes — a horizontal force on a deflected beam has a different moment arm; (2) membrane stiffening develops as the beam bends into a catenary shape; (3) for very slender members, follower-force effects change the buckling behaviour. All of these require NLGEOM.

Sure. Three major sources: First, large displacements and rotations — a fishing rod, a flexible robotic arm, or the wing of a high-aspect-ratio aircraft. The structure moves so much that the geometry used to compute stiffness must be updated continuously. Second, membrane stiffening — a guitar string or a prestressed cable: in-plane tension develops that stiffens the structure against transverse loads. A flat plate clamped at all edges becomes dramatically stiffer under large lateral load as mid-plane tension builds. Third, snap-through — a shallow arch or a dome can suddenly "snap" from one stable configuration to another if the load exceeds a critical value. This is a classic bistable problem that linear analysis cannot predict at all because the equilibrium path has a negative-stiffness region.

Both are valid formulations for large deformations — they give identical results for the same material model, just computed differently. The Total Lagrangian (TL) formulation always references quantities back to the original undeformed configuration. Stresses are the second Piola-Kirchhoff tensor; strains are Green-Lagrange. The reference geometry never changes, which makes TL convenient when you have material data defined in the undeformed frame. The Updated Lagrangian (UL) formulation updates its reference configuration at each converged increment to the last deformed state. It uses Cauchy (true) stress and logarithmic strain — physically intuitive quantities. UL is generally preferred in modern codes like Abaqus because it handles very large rotations more naturally and interfaces cleanly with plasticity constitutive models.

The Green-Lagrange strain tensor E is the standard choice for the Total Lagrangian formulation:

The nonlinear term on the right is what distinguishes it from the small-strain engineering tensor. For Updated Lagrangian / large-strain problems with plasticity, the logarithmic (true or Hencky) strain \(\varepsilon^{\ln} = \ln(\lambda)\) is preferred because it is additive — elastic and plastic parts can be simply added. This is what Abaqus uses internally for *NLGEOM analyses with metal plasticity.

At each load increment, Newton-Raphson iterates to find the displacement correction \(\Delta u\) that satisfies equilibrium on the deformed geometry:

The consistent tangent stiffness \([K_T]\) includes both the material stiffness and the geometric stiffness (also called the stress stiffness) \([K_\sigma]\) — extra stiffness terms that arise from the coupling between existing stress and displacement gradients. It's \([K_T] = [K_E] + [K_\sigma]\). The geometric stiffness is what makes pre-tensioned cables stiffer in the transverse direction and what can make compressed columns softer (negative contribution leading to buckling).

Newton-Raphson in load control fails when the load-deflection curve has a limit point — where the tangent stiffness becomes zero or negative. At that point you've reached peak load capacity; any further load increase causes collapse. But the structure actually has a post-collapse equilibrium path — it just requires the structure to "snap through" to a new shape at lower load. Standard load control can't follow this path.

The arc-length method (Riks/Crisfield) solves this by treating the load level itself as an additional unknown and constraining the increment to follow a fixed arc length in the combined displacement-load space. This lets the solver trace the complete equilibrium path through limit points and even negative-slope post-buckling paths — essential for shells, shallow arches, bistable mechanisms, and imperfection sensitivity studies.