4. Joint characterization for crash models

In a full-vehicle crash model, joints are where the structure is most likely to be misrepresented. A body-in-white may contain 3,000–5,000 spot welds, several hundred meters of structural adhesive, and dozens of bolted connections in the chassis and closures. Each joint type fails through a different mechanism, is characterized by a different test, and therefore carries a different pedigree burden. Treating all of them as "connectors with a force limit" is the single most common source of non-physical crash results — joints that separate too early, never separate, or transfer load through a stiffness that no physical coupon ever exhibited.

This section separates the four joint families by their characterization logic — what physical quantity governs failure, what test produces that quantity, and what metadata must travel with the resulting parameters so a reviewer can reconstruct where the number came from.

Spot Welds: Nugget Failure Under Combined Loading

A spot weld is a small, stiff, brittle connection. Its crash-relevant behavior is dominated by the weld nugget and the surrounding heat-affected zone, and it fails in one of several modes: interfacial nugget pull-out, button pull-out with sheet tearing, or partial thickness fracture. The mode depends on nugget diameter, sheet thickness ratio, and the loading angle. A weld loaded in pure peel behaves very differently from the same weld loaded in pure shear.

Characterization therefore cannot rely on a single tensile value. The standard practice is a cross-tension and lap-shear test pair at minimum, extended to a mixed-mode suite (typically 0°, 30°, 45°, 60°, 90° relative to the sheet plane) when the joint sits in a load path that will experience rotation during crash. Each specimen yields a force–displacement curve; the model calibrates a connector (often a spot-weld element with a failure criterion such as a force- or energy-based envelope) against those curves.

For the pedigree, the critical fields are not just the peak force. They are:

  • Nugget diameter and sheet stack-up — the exact thicknesses and grades on each side, because a 1.2 mm DP780 to 0.8 mm mild steel weld is not interchangeable with a 1.0/1.0 stack.
  • Welding parameters — current, electrode force, hold time, and whether the schedule was production-representative or a laboratory schedule.
  • Test rate — quasi-static versus intermediate rate (e.g., 0.1–1 m/s). Weld failure force is rate-sensitive; a quasi-static calibration applied to a 30 km/h impact is a documented extrapolation, not a validated result.
  • Failure mode observed — so the model's assumed mode can be checked against the coupon.
A common pedigree gap: the coupon report lists "average peak force 6.8 kN" but omits the stack-up and the loading angle. The connector model then gets calibrated to a number that cannot be traced to any specific joint in the vehicle, and the resulting scatter is misattributed to the crash model rather than to the missing metadata.

Adhesive Bonds: Rate-Dependent, Mode-Dependent, and Sensitive to Bondline

Structural adhesives behave very differently from welds. They are rate-dependent (stiffness and strength rise with loading rate), strongly mode-dependent (mode I peel versus mode II shear), and sensitive to bondline thickness, surface preparation, and cure state. A crash model that represents an adhesive as a rigid tie will over-stiffen the joint and suppress the progressive separation that governs energy absorption in a crash.

Characterization typically uses thick adherend shear tests and peel or double-cantilever tests to extract cohesive-zone parameters: initial stiffness, peak traction, and fracture energy in mode I and mode II. These parameters feed a cohesive element or a connector with a traction–separation law. Because the parameters are rate-dependent, the test matrix must include the rates the crash model will actually see — quasi-static, intermediate, and high rate — and the pedigree must record which rate each parameter set belongs to.

The pedigree fields that matter most for adhesive joints are the ones that are easiest to lose:

  • Adhesive product and batch — a formulation change between batches can shift fracture energy by double-digit percentages.
  • Bondline thickness — measured, not nominal. A 0.2 mm versus 0.5 mm bondline changes both stiffness and failure mode.
  • Surface preparation and cure schedule — grit blast versus wipe, oven cure versus induction cure.
  • Test rate and mode — recorded per parameter, not as a single "adhesive data" label.

Bolted Joints: Preload, Slip, and Bearing

Bolted joints in a crash model are usually fewer in number but higher in consequence — suspension mounts, subframe attachments, seat and restraint anchorages. Their behavior is governed by clamp load, friction at the interface, and the bearing/tear-out strength of the connected plates. Under crash loading, a bolted joint may hold, slip, or fail by bolt shear, plate tear-out, or thread stripping.

Characterization must capture the preload (torque–tension relationship, including friction coefficient scatter), the slip resistance of the interface, and the bearing and tear-out capacity of the plates. Unlike welds and adhesives, bolted joints often have a well-defined design intent — a specified torque and a specified clamp load — so the pedigree must record whether the model used the design preload or a measured as-built preload, and whether the friction coefficient was assumed or tested.

A practical calibration sequence for a bolted joint is: measure torque–tension on the actual fastener and plate stack, test the joint in shear and in tension to capture slip onset and ultimate capacity, then calibrate a connector whose preload, friction, and failure envelope match those measurements. The pedigree entry should state the fastener grade, the torque specification, the measured friction coefficient, and the test configuration.

Why the Pedigree Needs Differ by Joint Type

The four joint families fail through different physics, so the metadata that makes a parameter credible is different for each. A single "joint data" spreadsheet with one set of columns will either lose critical fields or force irrelevant ones. The figure below maps each joint type to its governing failure quantity, its characterization test, and the pedigree fields that must accompany the calibrated parameter.

Joint Type Governing Failure Characterization Test Critical Pedigree Fields Spot Weld nugget + HAZ brittle, mode-dependent Nugget pull-out / button pull-out / partial thickness force–displacement envelope Lap-shear + cross-tension mixed-mode suite 0°–90° loading angle stack-up (grade + t) nugget diameter weld schedule test rate + failure mode Adhesive Bond cohesive zone rate + mode dependent Mode I peel / Mode II shear traction–separation fracture energy G_Ic, G_IIc Thick adherend shear test DCB / peel test quasi-static → high rate product + batch bondline thickness surface prep + cure rate + mode per param Bolted Joint clamp + friction slip / bearing / shear Bolt shear / plate tear-out / slip onset preload + friction envelope Torque–tension shear + tension joint tests design vs as-built preload fastener grade torque spec measured friction coeff. test configuration Pedigree rule of thumb: A joint parameter is credible only if a reviewer can answer: which physical quantity governs failure, which test produced it, at what rate and mode, and on which exact stack-up / product / fastener. Missing any one of these turns a calibrated connector into an unverifiable assumption — and the crash model inherits that uncertainty silently.

Building the Joint Pedigree Record

The practical consequence is that joint data should be organized by joint family, not by vehicle program. A workable structure is a per-joint-type record that carries the governing quantity, the test matrix, the calibrated parameter set, and the traceability fields. For a spot weld, that means the stack-up and weld schedule travel with the force–displacement envelope. For an adhesive, the product batch and bondline thickness travel with the cohesive-zone parameters. For a bolted joint, the torque specification and measured friction coefficient travel with the preload and failure envelope.

When a crash model is reviewed, the first question about any joint should be: can I trace this connector's parameters back to a specific coupon, on a specific stack-up, at a specific rate? If the answer is no for even a fraction of the joints, the model's credibility is bounded by that fraction — regardless of how well the rest of the model was built. The joint pedigree is not administrative overhead; it is the part of the model that determines whether the structure separates where the real structure separates.

A useful audit habit: pick ten joints at random from the crash model, and for each one try to retrieve the coupon report, the stack-up, and the test rate from the pedigree record alone. If more than one or two cannot be retrieved, the joint data is not yet at a reviewable pedigree level.
← Previous 📖 Contents Next →