2. Contact mechanics fundamentals for bolted joints

A bolted joint is not a single structural member; it is a stack of interfaces. The clamp load generated by the tightened bolt must travel from the bolt head, through the clamped plates, and back into the nut. Every one of those interfaces — head-to-plate, plate-to-plate, nut-to-plate — is a contact problem. Before any nonlinear finite element model can predict joint stiffness, preload retention, or slip capacity, the analyst must understand four coupled mechanisms: normal pressure distribution, the stick–slip transition, Coulomb friction, and contact stiffness. This chapter builds that foundation using the conventions and quantities that appear directly in solver input decks.

Normal pressure and the real contact area

When two nominally flat surfaces are pressed together, they touch only at the summits of their roughness peaks. The true contact area is a small fraction of the apparent area, and it grows as normal load increases because more asperities yield and flatten. This is why contact pressure p is not uniform even under a flat bolt head. Classical Hertzian and Greenwood–Williamson models describe the asperity-scale behavior, but for bolted joints the engineering quantity of interest is the macroscopic pressure distribution under the washer face.

For a through-bolted joint with a stiff washer, the pressure under the head is often idealized as uniform over an annulus. In reality it peaks near the bolt hole and decays outward, following an approximately exponential or conical distribution. A widely used simplification for the effective clamped area is the 45° cone frustum:

A_eff = π/4 * [ (d_w + h * tan(45°))^2 - d_h^2 ]

where:
  d_w  = washer or head bearing diameter
  d_h  = bolt hole diameter
  h    = total clamped plate thickness
  A_eff = effective load-bearing contact area

This A_eff is the area over which the joint's compressive stress is assumed to act when estimating joint stiffness. It is an approximation, but it is the one embedded in most analytical joint models (VDI 2230, Shigley). In FEA, the same physics appears as a pressure field that must be resolved by the mesh; if the mesh under the head is too coarse, the peak pressure is smeared and the predicted joint stiffness is wrong.

Stick–slip transition

Friction in a bolted joint is not a single state. Each contact point is either stuck (no relative tangential displacement) or slipping (tangential force at the friction limit). The transition is governed by the Coulomb condition:

|t| < μ * p        →  stick (elastic tangential compliance)
|t| = μ * p        →  slip  (tangential force saturated)

where t is the tangential traction vector, p is the normal contact pressure, and μ is the friction coefficient. The stick–slip boundary is not fixed; it migrates as the load path changes. In a bolted joint under shear, the outer annulus of the contact patch typically slips first while the region near the hole remains stuck. This partial slip produces the characteristic nonlinear load–displacement curve: an initial steep (stuck) regime, a knee, and a shallow (slipping) regime.

For the analyst, the practical consequence is that a linear contact model cannot capture the knee. The solver must either use a penalty or Lagrange multiplier formulation with a Coulomb friction law, and the tangential stiffness must be allowed to degrade as slip propagates. The transition is also path-dependent: once a point slips, it does not automatically re-stick unless the normal pressure increases or the tangential load reverses.

Coulomb friction and its numerical implementation

Coulomb friction is the default friction model in Abaqus, ANSYS, and most implicit solvers. It is defined by a single coefficient μ, but its numerical behavior depends on three choices:

• Penalty vs. Lagrange: Penalty allows small elastic slip before the friction limit is reached; Lagrange enforces zero slip but adds multipliers and can cause convergence difficulty.
• Stiffness: The penalty tangential stiffness k_t should be large enough to approximate stick but small enough to avoid ill-conditioning. A common starting value is k_t = 0.01 * E / h per unit area.
• Regularization: A small amount of elastic slip (e.g., 0.1–1% of the element size) is often introduced to smooth the stick–slip transition and improve convergence.

For bolted joints, μ is rarely a single number. The bolt head, nut, and plate interfaces may have different surface treatments (as-received, phosphated, lubricated), so a single joint may require multiple friction coefficients assigned to different contact pairs. Typical values for as-received steel-on-steel are 0.15–0.25; lubricated interfaces can drop to 0.08–0.12. These values must be treated as calibration parameters, not constants.

Contact stiffness

Contact stiffness is the resistance to normal and tangential deformation at the interface. It has two components:

• Normal stiffness k_n: governs how much the interface closes under pressure. It is nonlinear — stiff at low load as asperities flatten, then progressively softer as the real contact area saturates.
• Tangential stiffness k_t: governs the elastic stick regime. It is proportional to the normal pressure and the friction coefficient.

In a bolted joint, the total axial stiffness is a series combination of bolt stiffness, plate stiffness, and interface stiffness. The interface contribution is often 10–30% of the total, which is why ignoring it leads to overprediction of joint stiffness and underestimation of preload loss. A practical estimate for the normal contact stiffness per unit area is:

k_n ≈ E_eff / (α * h_roughness)

where:
  E_eff       = effective elastic modulus of the interface
  h_roughness = combined RMS roughness of the two surfaces
  α           = empirical factor (typically 1–3)

This is a simplified form; in FEA the stiffness emerges from the penalty formulation and the mesh. The key point is that k_n and k_t must be consistent with the friction model and the mesh size. If the penalty stiffness is too low, the joint will appear too compliant; if too high, the solver may diverge.

Putting it together: a worked example

Consider an M12 bolt (property class 10.9) clamping two 10 mm steel plates. The target preload is 45 kN. The washer face diameter is 24 mm, the hole diameter is 13 mm, and the total clamped thickness is 20 mm. Using the 45° cone frustum:

A_eff = π/4 * [ (24 + 20)^2 - 13^2 ] = π/4 * [ 1936 - 169 ] ≈ 1388 mm²

Average contact pressure:
p_avg = 45000 N / 1388 mm² ≈ 32.4 MPa

If the friction coefficient is 0.15, the slip load per interface is:

F_slip = μ * F_preload = 0.15 * 45 kN = 6.75 kN

This means the joint will begin to slip when the applied shear load reaches approximately 6.75 kN per interface. In a finite element model, the stick–slip transition should be observable as a change in the tangential stiffness at this load level. If the model predicts slip at a significantly different load, the friction coefficient or the normal pressure distribution is wrong.

What to watch for in a simulation

When you build a bolted joint model, the contact mechanics described above manifest as specific numerical symptoms. A few common ones:

• Preload loss: If the contact stiffness is too low, the bolt will appear to lose preload as the joint is loaded. Check the normal contact pressure under the head — it should remain positive and roughly proportional to the applied preload.
• Premature slip: If the joint slips at a load well below μ * F_preload, the normal pressure distribution is likely wrong (too concentrated or too diffuse), or the friction coefficient is too low.
• Convergence failure at the knee: The stick–slip transition is a non-smooth event. If the solver struggles, reduce the friction regularization stiffness or add a small amount of elastic slip.
• Mesh sensitivity: The pressure peak under the bolt head is mesh-dependent. Refine the mesh locally until the peak pressure and the slip load converge.

The next chapter builds on this foundation by introducing the specific contact formulations used in implicit and explicit solvers, and how to choose between them for bolted joint analysis.