4. Contact and impact algorithms in PAM-CRASH
Crash energy management depends on how a model transfers load between bodies that were never designed to touch in normal service. In PAM-CRASH, that transfer is handled by the contact interface: a set of algorithms that detect proximity, resist penetration, and convert relative motion into forces that propagate through the structure. If the contact definition is wrong, the energy path is wrong — the barrier may pass through the rail, the spot welds may never see the shear they should, and the occupant load case becomes meaningless. This section covers the three mechanisms that dominate crash models: penalty-based contact, node-to-segment (NTS) projection, and self-contact for folding and self-intersection.
Penalty contact: stiffness, not constraint
PAM-CRASH does not enforce non-penetration as a hard geometric constraint. Instead it uses a penalty method: when a slave node penetrates a master segment by depth d, a restoring force is applied along the segment normal,
F_n = k_pen * d
k_pen = alpha * K_char
K_char = m_eff / (dt^2)where k_pen is the penalty stiffness, alpha is a scale factor (typically 0.1–1.0), and K_char is a characteristic stiffness derived from the effective mass m_eff and the stable time step dt. The practical consequence is that k_pen is tied to the explicit time step. Increase it too far and the contact introduces its own stability limit; leave it too soft and you get visible penetration and a delayed, smeared load rise.
Friction is applied in the tangential direction with a Coulomb cap F_t ≤ μ F_n. For crash, a static friction coefficient around 0.1–0.3 is common for steel-on-steel sliding, while self-contact in folding regions often uses higher values or a stick condition to prevent artificial sliding of already-buckled lobes.
Node-to-segment projection
NTS is the workhorse algorithm for shell-to-shell and shell-to-solid impact. Each slave node is projected onto the closest master segment, and the gap is measured along the segment normal. The projection uses the segment's local coordinates (ξ, η); if the projected point falls outside the segment, the algorithm either clamps to the edge or searches neighbouring segments, depending on the search option.
Two settings dominate NTS behaviour in practice:
- Thickness offset. Shells have no physical thickness in the mesh, so the contact surface must be offset by half the shell thickness. If a 1.6 mm rail is modelled with a mid-surface mesh, the contact gap must account for 0.8 mm on each side. Omitting this offset lets the two mid-surfaces approach to zero gap, producing premature contact and an over-stiff front end.
- Search tolerance / bucket size. PAM-CRASH uses a spatial bucket sort to limit candidate segments. A tolerance smaller than the maximum nodal travel per step causes missed contacts; too large a tolerance wastes CPU and can create spurious contacts between parts that should never touch.
For a typical frontal rail against a rigid barrier, a workable starting point is: slave = rail nodes, master = barrier segments, alpha = 0.5, friction μ = 0.15, thickness offset enabled, and a search tolerance of roughly twice the expected maximum penetration per step. These are starting values, not validated results — they must be checked against the penetration and contact-energy plots.
Self-contact
Self-contact is what makes crash models behave like real structures. When a rail folds, the inner surface of one lobe approaches the inner surface of the next. Without self-contact, those surfaces pass through each other, the fold consumes no additional energy, and the force–displacement curve stays artificially flat. Enabling self-contact on the folding region restores the membrane and bending resistance that real steel provides.
Self-contact is more expensive and more fragile than body-to-body contact because the candidate set is the part against itself. Two rules keep it stable:
- Activate self-contact only on parts that actually fold — typically the crash box, the rail, and the bumper beam. Applying it globally multiplies the search cost and invites false contacts across gaps that are legitimately open.
- Use a slightly larger penalty scale and a small initial gap. A common failure is a node that starts inside its own part's thickness offset, which produces a permanent contact force from step zero and a non-physical preload.
How the three mechanisms combine
The figure below shows the energy path through a frontal impact model and where each algorithm acts. The barrier-to-bumper interface is NTS; the rail-to-rail and rail-to-crossmember interfaces are body-to-body penalty contact; the folding lobes inside the rail are self-contact. Energy enters through the barrier, is transferred through the NTS interface, is absorbed by plastic folding (self-contact), and is redistributed to the rest of the body through the remaining penalty interfaces.
Verification and common failure modes
Contact definitions are verified by three outputs, not by visual inspection alone. First, penetration depth: PAM-CRASH reports maximum penetration per interface; it should stay below roughly 10% of the shell thickness. Second, contact energy: it should be a small fraction of internal energy and should not oscillate. Third, the force–displacement curve: a correct frontal rail shows a sharp rise to a plateau, then a series of fold-related oscillations. A plateau that is too low points to missing self-contact; a curve that is too stiff and short points to an over-large penalty stiffness or a missing thickness offset.
| Symptom | Likely cause | Corrective action |
|---|---|---|
| Node passes through barrier | Penalty stiffness too low or search tolerance too small | Increase alpha to 0.5–1.0; enlarge bucket size |
| Premature contact force at step 0 | Thickness offset missing or initial penetration | Enable shell thickness offset; check initial gap |
| Flat force plateau, low energy absorption | Self-contact not activated on folding part | Add self-contact to crash box and rail |
| Contact energy spike then drop | Node punched through and released | Increase penalty stiffness; refine master segments |
| Excessive CPU in contact search | Self-contact applied globally | Restrict self-contact to folding parts only |
The key engineering judgement is that contact is not a passive bookkeeping step — it is the mechanism that routes energy. Penalty stiffness sets how sharply load rises, NTS projection sets where and when contact begins, and self-contact sets how much energy the folding structure can absorb. Tuning these three together, and verifying them against penetration, contact energy, and the force–displacement curve, is what makes a crash model predictive rather than merely plausible.