3. Linear vs. nonlinear buckling

Thin-walled structures — stiffened panels, cylindrical shells, cold-formed sections, and pressure-vessel skirts — rarely fail by material yielding first. They fail by losing stability. The engineering question is not whether buckling matters, but which buckling model gives a defensible allowable load. Two families of analysis dominate practice: linear eigenvalue buckling and nonlinear (geometrically nonlinear, often elastic-plastic) buckling. They answer different questions, and confusing them is one of the most common sources of unconservative shell designs.

What Linear Eigenvalue Buckling Actually Solves

Linear eigenvalue buckling is a generalized eigenvalue problem built on the pre-buckling stiffness matrix K and the geometric (initial stress) stiffness matrix Kg, which is assembled from a reference load state:

(K + λ · K_g) · φ = 0

Here λ is the eigenvalue (the load multiplier), and φ is the corresponding buckling mode shape. The critical load is P_cr = λ · P_ref, where P_ref is the applied reference load. The formulation assumes three things that are almost never simultaneously true for real thin-walled hardware:

• Small pre-buckling deformation. The structure is assumed to deform negligibly before the bifurcation point, so the pre-buckling stress state is computed on the undeformed geometry.
• Perfect geometry and ideal loading. No out-of-flatness, no weld distortion, no load eccentricity, no residual stress.
• Linear elastic material. No yielding, no plasticity, no stiffness degradation prior to bifurcation.

Under these assumptions the eigenvalue problem is cheap and informative: it returns a load factor and a mode shape. The mode shape is genuinely useful — it tells you where the structure wants to buckle and therefore where imperfection sensitivity and stiffener placement matter. The load factor, however, is an upper bound, not a capacity.

Why the Eigenvalue Result Is Not a Capacity

The gap between the eigenvalue load and the real collapse load is not a small correction factor. For thin cylindrical shells under axial compression, classical linear theory predicts a critical stress on the order of E·t / (R·√(3(1−ν²))), while carefully tested laboratory specimens routinely reach only 15–40% of that value. The discrepancy is not experimental error; it is the physics of imperfection sensitivity.

Three mechanisms drive the gap:

1. Imperfection sensitivity. A geometric deviation of amplitude w₀ aligned with the critical mode reduces the limit load sharply. For axially compressed cylinders, the knockdown can be expressed through a factor that decays roughly as 1 / (1 + k·w₀/t) for the classical imperfection-sensitive mode.
2. Mode interaction and post-buckling path. Many thin-walled configurations have an unstable post-buckling path — the load drops immediately after the bifurcation point. The eigenvalue marks the top of a cliff, not a plateau.
3. Plasticity and residual stress. In stocky or locally thick-walled regions, yielding occurs before the elastic bifurcation load is reached, so the elastic eigenvalue is never physically attainable.

What Nonlinear Buckling Analysis Adds

Nonlinear buckling analysis removes the three assumptions above. The standard approach is a geometrically nonlinear static analysis with incremental load stepping, tracking the equilibrium path P(u) where u is the displacement vector. The governing equation at each increment is:

K_t(u) · Δu = ΔP_ext − F_int(u)

where K_t is the tangent stiffness matrix, F_int is the internal force vector, and ΔP_ext is the external load increment. The limit point — the maximum of the equilibrium path — is the nonlinear buckling load. Two ingredients are essential:

• An imperfection. Either a measured geometric deviation, a scaled eigenmode (commonly the first mode with amplitude w₀ = t/10 to t/100 for shells), or a combination of modes. The choice must be documented because it directly sets the result.
• A path-tracking solver. Newton–Raphson with arc-length (Riks) control is standard, because the equilibrium path may pass through a limit point where the tangent stiffness becomes singular and load control fails.

If plasticity matters, the material model is added: an elastic-plastic constitutive law with a yield surface and hardening rule, evaluated at each integration point. The limit load then reflects both geometric softening and material yielding.

Side-by-Side Comparison

How the Two Analyses Work Together

The practical workflow is not "choose one." Eigenvalue analysis is the fast reconnaissance step: it identifies the critical mode, the region of highest sensitivity, and a reference load factor. That mode then seeds the nonlinear analysis as an imperfection shape. The nonlinear run produces the capacity. The ratio between the two — the effective knockdown — is itself a diagnostic: a ratio near 1.0 suggests a mildly imperfection-sensitive structure, while a ratio of 0.3 or lower signals a shell that must be treated with explicit imperfection sensitivity in the design basis.

Reading the Path Diagram

The figure above shows the three load–displacement paths that matter in practice. The blue dashed line is the linear eigenvalue prediction: a straight pre-buckling path terminating at the bifurcation load P_cr. The teal curve is the nonlinear response of a perfect structure; it softens as deformation grows and reaches a limit point below the eigenvalue load. The orange curve is the nonlinear response of an imperfect structure; it departs from the perfect path early, reaches a lower limit load, and then descends — the unstable post-buckling branch. The vertical gap between the eigenvalue point and the imperfect limit load is the knockdown, and it is the single most important quantity that linear analysis cannot provide.

Practical Guidance for the Assessment Workflow

For a thin-walled component, a defensible assessment sequence is:

1. Run a linear eigenvalue analysis to obtain the first few modes and the reference load factor. Use at least 5–10 modes; closely spaced eigenvalues indicate mode interaction and higher imperfection sensitivity.
2. Select the imperfection shape. Use the first eigenmode scaled to an amplitude consistent with fabrication tolerance (e.g., w₀ = t/10 for welded shells, t/100 for machined shells), or a measured deviation if available.
3. Run a geometrically nonlinear analysis with arc-length control. Include plasticity if the yield stress is within roughly 1.5× the expected limit stress.
4. Extract the limit load from the equilibrium path. Verify that the solver detected a true limit point, not a numerical divergence.
5. Compare the nonlinear limit load to the eigenvalue load. Report the knockdown ratio and use it to justify the design allowable.

The essential takeaway is that linear eigenvalue analysis is a screening and mode-identification tool, not a capacity method. Nonlinear analysis is what converts a perfect-structure bifurcation load into a realistic limit load that accounts for imperfections, large deformation, and material nonlinearity. For thin-walled structures, that conversion is not optional — it is the difference between a design that survives fabrication and one that does not.