Rivet Joint Failure — Root Cause & 8D Corrective

Symptoms

• Rivet head pulls through (or into) the workpiece material
• Loose rivet / relative play between joined parts after installation
• Cracking or deformation around the rivet hole
• Joint fails shear or tensile pull test below specification

Affected Components

• Self-pierce (SPR), solid and blind rivets in sheet-metal assemblies
• Aluminum / steel stack-ups in automotive body, chassis, battery trays
• Multi-layer joints with material thickness variation
• Structural and semi-structural joints

Detection Methods

• Visual + gauge: head flushness, hole deformation, cracking
• Mechanical test: shear and tensile pull; record peak load and failure mode (head pull-through vs. rivet fracture)
• Cross-section / macro etch: check rivet flaring, fit-up gap, hole fill
• In-process: squeeze-force trace; rivet length / diameter verification

Possible Causes

• Hole alignment offset between layers (stacked misalignment)
• Rivet length insufficient for total material stack thickness
• Squeeze force too high → material cracking around hole; too low → poor flare
• Material thickness stack variation exceeding rivet tolerance range
• Worn punch / die bushing → oversize hole
• Wrong rivet for the material combination

5-Why Example

Problem: Rivet head pulled through the top layer of a bonded + riveted body panel

1. Why head pulled through? → Top-layer hole 0.5 mm oversize
2. Why oversize? → Punch bushing worn, punch cut an oversize hole
3. Why not replaced? → Punch bushing overdue at PM
4. Why overdue? → PM wear limits too loose / not tracked by cycle count
5. Why loose? → Bushing wear limit never validated against rivet-hole tolerance

TRC: Punch bushing wear limit not validated for rivet-hole tolerance MRC: Tool maintenance limits based on calendar, not wear / cycle count

Fishbone (6M)

• Man: Operator did not flag erratic rivet setting
• Machine: Worn punch bushing; incorrect die alignment
• Method: No hole-size verification before riveting
• Material: Stack-thickness variation; rivet-length margin insufficient
• Measurement: EOL pull test only, no in-process hole check
• Environment: Thermal expansion between stacked dissimilar metals

D3 Containment

• Block parts with oversize holes; 100% hole-size check
• Replace punch bushings on the affected station
• Temporary thicker rivet, rework verified by pull test

D5 Permanent Fix

• Correct punch bushing and die alignment; set hole tolerance to rivet spec
• Validate bushing wear limit from rivet-hole tolerance, not arbitrary calendar
• Select rivet length for the actual max/min stack thickness

D7 Prevention

• Add hole-size (punch/die) verification to PM and change control
• Require stack-thickness measurement at set-up
• Update process FMEA with “worn bushing → oversize hole” failure mode

Worked Example — Quantifying the Gap

A three-layer aluminum/steel stack is riveted with a nominal 6.5 mm rivet. Drawing stack thickness is 4.0 mm nominal, but measured parts range from 3.7 mm to 4.4 mm across the build. The rivet length window for correct flare on this tool is roughly 4.0–4.3 mm of stack, so the upper end of the measured range sits outside the window and the lower end leaves the rivet under-filled.

A 30-piece capability study of hole diameter on the top layer shows a mean of 6.30 mm against a 6.20–6.40 mm tolerance, but individual holes reach 6.55 mm after the punch bushing passes its wear point. Pull tests on those parts fail in head pull-through at roughly 70–80 % of the specified peak load, while parts inside the hole tolerance fail by rivet fracture at or above specification. The failure mode itself is the diagnostic: head pull-through points to hole size or stack thickness, rivet fracture points to rivet strength or squeeze force.

Step-by-Step Verification Checklist

1. Confirm the failure mode on the failed parts — head pull-through, rivet fracture, or loose joint — before touching the process.
2. Measure hole diameter on the top and bottom layers at several points around the circumference; record min/max, not just one reading.
3. Measure stack thickness at the same locations and compare against the rivet length window.
4. Check hole alignment between layers with a pin gauge or cross-section; offset shows as an elliptical or stepped hole.
5. Review the squeeze-force trace for the suspect station and compare peak force and curve shape against a known-good baseline.
6. Inspect the punch and die bushing for wear, and pull the cycle count since the last change.
7. Run a pull test on suspect and conforming parts side by side and record peak load and failure mode for each.
8. Cross-section one part to confirm flare geometry and hole fill before releasing containment.

Common Mistakes and How to Detect Them

• Measuring only one hole per part. A single reading hides the tail of the distribution; sample around the circumference and across several parts.
• Treating the pull test as the only check. End-of-line testing catches escapes but not drift; add in-process hole and stack checks.
• Replacing the bushing without changing the wear limit. The same failure returns at the same interval; verify the new limit is derived from the rivet-hole tolerance.
• Assuming rivet length is correct because the part number matches. Stack thickness variation can push a correct part number outside its working window.
• Ignoring failure mode in the test record. Without it, head pull-through and rivet fracture look identical in a pass/fail column and the wrong cause gets chased.
• Calendar-based PM. Wear tracks cycles, not weeks; compare bushing condition against cycle count to expose the mismatch.

Verifying the Output

A corrective action is only closed when the evidence shows the cause is gone. Re-run the 30-piece capability study on hole diameter and confirm the distribution sits inside tolerance with margin. Repeat pull tests across the full stack-thickness range and confirm every sample meets peak load with a consistent failure mode. Cross-section a sample from the thinnest and thickest stack to confirm flare and hole fill at both extremes. Finally, confirm the updated PM limit and FMEA entry are live in the controlled documents, not just in the 8D file.

Links to Neighbouring D-Steps

This analysis sits between D2 (problem description) and D4 (root cause). A precise D2 — which layer, which hole, which failure mode, how far out of tolerance — is what makes the 5-Why in D4 converge instead of drifting. The Fishbone feeds D4 directly, and the verified root cause then drives D5 and D6. D3 containment must be traceable to the same measurements used in D4, otherwise containment and permanent fix can end up addressing different variables. D7 closes the loop by pushing the validated wear limit and stack-thickness check back into PM, change control and the FMEA.