Contact Representation in FE Models

What Is It?

Contact representation is the way the finite element model describes how two separate components interact at their shared interface. In reality, two bolted or pressed components transfer load through the surfaces that touch — the pressure between them, the friction that resists sliding and the gap that opens where they separate. The FE model must capture this surface interaction with a mathematical abstraction. Three representations dominate practice: tied (or bonded) contact, where the surfaces are permanently joined with no relative motion; frictionless contact, where the surfaces can separate and slide freely but cannot interpenetrate; and frictional contact, where the surfaces can separate and slide only when the friction limit is overcome. Each representation embeds a different physical assumption about the interface and produces a different stiffness, a different load path and a different stress distribution. The contact representation is not a meshing detail — it is an engineering idealisation of the joint.

Why It Matters

The contact representation changes the stiffness of the assembly and the way load flows between components. A tied interface is rigid — it transfers both normal pressure and tangential shear across the boundary, behaving like a continuous material. A frictionless interface is much softer — it transfers normal pressure but allows free sliding, so tangential load must find another path. A frictional interface sits between the two — it transfers tangential load up to the friction limit and then allows sliding. Choosing the wrong representation produces a model that is either too stiff (tied where the joint can slip) or too soft (frictionless where the joint is actually gripped). The error is not small: the load path can change entirely, and the peak stress can move from one component to another. The contact representation is one of the most consequential idealisation decisions in any multi-component model.

THE CONTACT REPRESENTATION IS A STIFFNESS ASSUMPTION, NOT A MESHING DETAIL. A tied interface makes the joint rigid in all directions; a frictionless interface lets it slide; a frictional interface lets it slide only beyond the friction limit. Each choice changes the load path and the peak stress location. Choose the representation that matches the physical behaviour of the joint — not the one that is easiest to converge.

Tied (Bonded) Contact

Tied contact — also called bonded, glued or焊接 contact depending on the solver — joins two surfaces so that they cannot separate or slide relative to each other. Mathematically, the tied condition constrains the displacements of the slave surface nodes to match the displacements of the master surface. The constraint can be enforced node-to-node (when the meshes are coincident) or node-to-surface (when the meshes are non-conforming, using projection and interpolation). Tied contact makes the interface behave like a continuous material: normal pressure and tangential shear both transfer across the boundary. It is the simplest and most stable contact representation, and it is the default choice when the joint is genuinely permanent — a weld, an adhesive bond, a brazed joint, a permanently bonded composite layup. The risk is using tied contact where the joint is not actually permanent — a bolted joint that can slip, a press-fit that can separate under load. The tied assumption then over-stiffens the joint and produces an unconservative load distribution.

• Transfers both normal pressure and tangential shear — full stiffness across the interface
• No separation, no sliding — the surfaces are permanently joined
• Node-to-node when meshes are coincident; node-to-surface when non-conforming
• Most stable and cheapest contact formulation — no iteration needed
• Appropriate for welds, adhesive bonds, brazed joints and permanent bonds
• Inappropriate for bolted, riveted or press-fit joints that can separate or slip

Frictionless Contact

Frictionless contact allows the surfaces to separate (gap opening) and to slide freely relative to each other, but prevents interpenetration. The contact condition is enforced only in the normal direction: when the surfaces press together, normal pressure transfers; when they try to separate, the contact releases and the gap opens with zero tension. There is no tangential constraint — the surfaces can slide without resistance. Frictionless contact is the most conservative representation for tangential load: because no shear transfers across the interface, all tangential load must be carried by other means — fasteners, adhesives, friction elsewhere, or the surrounding structure. It is appropriate when the joint is genuinely free to slide (a sliding bearing, a guide surface) or when the analyst wants a conservative bound on the load that the fasteners must carry. The risk is under-predicting the joint stiffness: a frictionless model of a gripped bolted joint will be too soft and will over-predict fastener loads and relative displacement.

Frictional Contact

Frictional contact allows the surfaces to separate and to slide, but sliding is resisted by friction up to a limit. The normal condition is the same as frictionless — no interpenetration, no tension. The tangential condition uses the Coulomb friction law: the tangential force is resisted up to μN, where μ is the coefficient of friction and N is the normal contact force. Below the limit, the surfaces stick (no relative sliding); at the limit, the surfaces slide. Frictional contact is the most physically realistic representation for dry unlubricated interfaces — bolted joints, clamped surfaces, stacked plates. It captures the transition from sticking to sliding and the load redistribution that occurs when a joint slips. The cost is computational: frictional contact is non-linear, requiring iterative solution, and the stick-slip transition can be difficult to converge. The coefficient of friction is also uncertain — it varies with surface finish, lubrication and wear — so the result should be treated with sensitivity analysis for different friction values.

Coulomb friction law (tangential contact condition):

  |τ| ≤ μ · p_n   (stick condition)

  where:
    τ   = tangential (shear) stress at the interface
    p_n = normal contact pressure
    μ   = coefficient of friction

When |τ| < μ·p_n  →  surfaces stick (no sliding)
When |τ| = μ·p_n  →  surfaces slide (tangential force at friction limit)

The friction limit scales with the normal pressure:
  higher clamp load → higher friction capacity → more shear transferred

Comparison of Contact Representations

Contact Stiffness and Penetration

Contact is enforced in most solvers by a penalty or augmented Lagrange method, which introduces a contact stiffness that resists penetration. The penalty method adds a stiff spring between the surfaces when they are in contact — the stiffer the spring, the less the penetration, but the worse the numerical conditioning. A contact stiffness that is too low allows visible interpenetration (non-physical); a contact stiffness that is too high causes convergence difficulties and ill-conditioning. The augmented Lagrange method reduces the penetration by iteratively updating the contact pressure, giving better accuracy with less sensitivity to the penalty value. The analyst should check the penetration in the results — visible penetration means the contact stiffness is too low or the augmented Lagrange iterations are insufficient. The contact stiffness also affects the tangential behaviour in frictional contact: the sticking stiffness determines how much micro-slip occurs before macro-sliding, and a low sticking stiffness can produce artificial compliance in the joint.

COMMON MISTAKE: Setting the contact stiffness too high to eliminate penetration, causing convergence failure or ill-conditioning; or setting it too low, allowing visible interpenetration. Check the penetration in the results and tune the contact stiffness so that penetration is small but the solution converges. Augmented Lagrange methods are less sensitive to the penalty value and are preferred for production work.

Sliding, Gapping and Separation

The contact status — sticking, sliding, open — varies over the interface and over the load history. A bolted joint under increasing shear may start fully stuck, transition to partial sliding at the edges, and eventually slip globally. A flange under bending may open a gap on the tension side while remaining compressed on the compression side. The contact status map is one of the most diagnostic outputs of a contact analysis: it shows where the load is actually transferring and where the joint is losing contact. The analyst should examine the contact status, not just the stress, because the stress distribution depends entirely on which parts of the interface are in contact. A model that assumes full contact when the real joint is partially open will give a completely wrong stress distribution.

Contact status across a bolted joint under shear:

  Clamped region (stick)     Sliding region      Open gap
  ┌─────────────────┐   ┌──────────────┐   ┌──────────────┐
  │ ◄─────────────► │   │ ◄──────────► │   │              │
  │   τ < μ·p_n     │   │  τ = μ·p_n   │   │   p_n = 0    │
  │   full transfer │   │  at limit    │   │  no contact  │
  └─────────────────┘   └──────────────┘   └──────────────┘
         centre              edge             outer edge

  As shear increases: stick region shrinks, sliding grows, gap may open
  Load path shifts from interface friction to fasteners as slip occurs

When to Use Which Representation

The choice of contact representation should be driven by the physical behaviour of the joint and by the engineering question. A joint that is permanently bonded should be tied. A joint that is genuinely free to slide should be frictionless. A joint that grips by friction and may slip under extreme load should be frictional. When the friction coefficient is uncertain, run sensitivity cases — a high-friction case and a low-friction case — and envelope the results. When the joint behaviour is critical to the result, model the fasteners explicitly (as beams or solids) alongside the frictional contact, so that the load transfer between friction and fasteners is captured. When the joint is far from the region of interest, a tied representation may be adequate — the local joint behaviour does not affect the global response, and the cost of frictional contact is not justified.

Idealisation Considerations

IDEALISATION CONSIDERATION: A tied contact representation over-stiffens a bolted joint that can slip, giving an unconservative load distribution. A frictionless representation under-stiffens a gripped joint, giving conservative but potentially excessive fastener loads. When the joint behaviour matters, use frictional contact with explicit fasteners and run friction sensitivity cases. When the joint is remote from the region of interest, tied contact is usually adequate.

Key Takeaways

• Tied contact joins surfaces rigidly — full normal and tangential transfer, no separation or sliding
• Frictionless contact allows free sliding — conservative for tangential load but under-stiffens gripped joints
• Frictional contact uses the Coulomb law — stick until μN, then slide; most realistic for dry interfaces
• Contact stiffness (penalty) controls penetration — too low gives interpenetration, too high causes ill-conditioning
• Contact status (stick/slide/open) is diagnostic — examine it to understand the real load path