Non-Linear Structural Analysis Fundamentals

What Is It?

A structural problem becomes non-linear when the relationship between applied load and structural response is no longer proportional. In a linear analysis, doubling the load doubles the displacement, stress and strain. In a non-linear analysis, this proportionality does not hold — the response depends on the current state of deformation, material behaviour and contact conditions. Non-linear analysis is required when the structural response itself changes the problem being solved.

Why It Matters

Many engineering structures cannot be adequately assessed with linear analysis alone. Thin panels that develop membrane action, components that yield under load, joints that open or close, structures that buckle, and any impact or transient event all involve non-linear behaviour. Applying linear analysis to these problems can produce results that are wildly wrong — not just slightly conservative or unconservative, but qualitatively misleading. Understanding when non-linearity matters and how it is represented is essential for credible structural analysis.

Non-linear analysis is not one method. It is a family of methods addressing geometric, material and contact non-linearity — often simultaneously. The term "non-linear" describes the physics, not a single solver setting.

Linear vs Non-Linear Response

In linear static analysis, several assumptions hold: stiffness is constant, deformations are small, material response is linear elastic, and contact state does not change. Under these assumptions, the load-displacement relationship is proportional — F ∝ u. Non-linearity breaks one or more of these assumptions. When stiffness changes with deformation, when material yields, when contact opens or closes, or when geometry changes significantly, the simple proportionality no longer applies.

The Equilibrium Concept

In a non-linear analysis, equilibrium is not satisfied by a single matrix solve. The internal resisting forces must balance the external applied forces at the current deformation state. The difference between external and internal forces is the residual — the out-of-balance force. When the residual is zero, equilibrium is satisfied. The solver iterates to drive the residual to zero at each load increment.

Non-linear equilibrium:

R(u)  =  F_ext  −  F_int(u)  =  0

where:
R(u)    =  residual (out-of-balance) force vector
F_ext   =  external applied load vector
F_int   =  internal resisting force vector (function of displacement u)
u       =  displacement vector

F_int includes: elastic forces, plastic forces, contact forces, etc.

Incremental Loading

Because the relationship between load and response is not proportional, the load cannot be applied in a single step. Instead, it is applied in increments — small steps from zero to the final load. Within each increment, the solver iterates to find the equilibrium state. The incremental approach allows the solver to track the changing stiffness, material state and contact conditions as the load increases.

Load: 0 → ΔF₁ → ΔF₂ → ΔF₃ → ... → Final load

At each increment: iterate to equilibrium
  Iteration 1: estimate displacement → compute F_int → check residual
  Iteration 2: correct displacement → update F_int → check residual
  ...until residual < tolerance

Tangent Stiffness

The tangent stiffness is the derivative of the internal force with respect to displacement — it describes how the internal force changes as the structure deforms. In a linear analysis, the tangent stiffness is constant (it is the usual stiffness matrix K). In a non-linear analysis, the tangent stiffness changes with deformation, material state and contact. It is updated at each iteration to reflect the current state. The tangent stiffness is used in the Newton–Raphson iteration to predict the displacement correction that will drive the residual toward zero.

When Non-Linearity Is Required

• Stiffness changes with deformation — membrane action, stress stiffening, large rotation
• Material yields or damages — plasticity, creep, progressive failure
• Contact conditions change — joints opening, surfaces sliding, support separation
• Buckling or post-buckling — instability with geometric and material non-linearity
• Impact or transient events — rapid loading, wave propagation, inertia effects
• Very large deformation — crash, forming, large displacement problems

When Linear Analysis Remains Sufficient

Not every structure requires non-linear analysis. If the deformation is small, the material remains well below yield, contact conditions do not change, and the load-displacement relationship is approximately proportional, linear analysis is adequate and is preferred for its simplicity and speed. The engineer must assess whether the assumptions of linear analysis are valid for the specific problem before choosing to include non-linearity. Including non-linearity when it is not needed adds computational cost and complexity without engineering value.

COMMON MISTAKE: Turning on non-linear options without understanding whether the physics genuinely requires them. Non-linearity should be included because the physics demands it, not because the solver can do it.

Non-Linear Analysis Is Not Automatically More Accurate

A non-linear analysis is not automatically more accurate than a linear one. A non-linear model with the wrong material data, incorrect contact definition, insufficient mesh refinement or an inappropriate solver can produce results that are less accurate than a well-judged linear analysis. Non-linearity adds the ability to represent physics that linear analysis cannot — but it also adds more places where errors can be introduced. The engineer must understand the physics, choose the right non-linear options and verify the results.

Key Takeaways

• Non-linearity means the load-response relationship is no longer proportional
• Three main sources: geometric, material and contact non-linearity — often combined
• Equilibrium is found iteratively through increments, not a single matrix solve
• Non-linear analysis is required when the response changes the problem — not by default
• Non-linear does not automatically mean more accurate — it means more physics, more assumptions and more verification