When a Design Issue Reaches the Customer
A quality defect that reaches a customer is one of the most stressful events in a mechanical designer’s career. The combination of urgency, blame attribution, and pressure to produce an immediate fix creates conditions where the wrong response is easy to make. Acting too quickly — issuing a design change before understanding the root cause — often creates a second defect. Responding defensively — focusing on what the manufacturing process did wrong rather than what the design might have contributed — prevents the problem from being truly solved.
This article covers how to respond effectively when a design-related quality defect surfaces: how to structure a root cause investigation, how to assess what the design contributed to the failure, and how to develop and implement corrective actions that prevent recurrence.
The First 24 Hours: Containment Before Correction
The immediate priority when a defect is reported is containment — stopping the flow of defective product before understanding why it happened. This is not the time for root cause analysis. It is the time for decisive, temporary action.
Containment Actions
- Identify the population: How many parts or assemblies are potentially affected? What date range of production? Which customers received them?
- Hold in-process and finished goods inventory: Stop shipment of suspect product until it is inspected and dispositioned.
- Issue a field hold if product is already at customers: This is a difficult decision with commercial consequences, but releasing suspect product to avoid short-term disruption while root cause is unknown is rarely the right trade-off.
- Identify a short-term workaround if one exists: Can the product be reworked? Is there an inspection that reliably separates good product from defective product?
Containment buys time for proper analysis. It does not fix the problem. Engineers who skip containment to go directly to root cause often find that by the time the fix is ready, additional defective product has shipped.
Root Cause Analysis: Finding the Design Contribution
Root cause analysis methods range from simple five-why analysis to formal FMEA review to fishbone (Ishikawa) diagrams. The method matters less than the discipline of not stopping at the first apparent cause.
The Five-Why Applied to Design
Consider a field failure: a shaft failed in service by fatigue fracture at a keyway. The five-why investigation might proceed:
- Why did the shaft fail? — Fatigue fracture initiated at the keyway root radius.
- Why did fatigue failure initiate there? — Stress concentration factor at the keyway root radius was higher than analyzed.
- Why was the stress concentration higher than analyzed? — The as-manufactured root radius was smaller than specified on the drawing.
- Why was the as-manufactured radius smaller than specified? — The drawing specified a minimum radius of 1.0 mm, but the manufacturing process consistently produces 0.4 mm with standard keyway broaches.
- Why was a non-producible tolerance specified? — The designer specified the radius based on stress analysis results without verifying that it was achievable with standard tooling.
The root cause is not that the shaft failed. It is not even that manufacturing produced undersized radii. The root cause is a design specification that was not validated against manufacturing capability. The corrective action must address this, not just the radius specification on this one drawing.
Distinguishing Design Defects from Manufacturing Defects
One of the most important and most contested questions in quality response is: was this a design problem or a manufacturing problem? The honest answer is often "both," and a good designer acknowledges the design contribution without deflecting to manufacturing as the primary cause when the design is at fault.
| Failure Characteristic | Suggests Design Contribution | Suggests Manufacturing Contribution |
|---|---|---|
| Failure mode matches design analysis risk | Yes — design analysis identified this as a risk | No — failure mode was not a design risk |
| All parts from same production run affected | No — not a design-specific pattern | Yes — process deviation affected a batch |
| Failure occurs across multiple production runs | Yes — systematic, not batch-specific | No — not a systematic manufacturing pattern |
| Failure rate increases at environmental extremes | Yes — design margin may be insufficient | No — unless process sensitivity to environment |
| Part dimensions within drawing tolerance | Yes — tolerance may be insufficient for function | No — part was within specification |
Developing an Effective Corrective Action
A corrective action that addresses only the immediate defect without preventing recurrence is an incomplete corrective action. Effective corrective actions operate at three levels:
Level 1: Immediate Fix
Correct the specific drawing, design, or calculation that contributed to the defect. This is necessary but not sufficient. Changing the radius specification on one drawing does not prevent the same error on the next similar design.
Level 2: Systemic Fix
Update the design process, checklist, or standard to prevent the same class of error. If the error was specifying a tolerance without verifying manufacturing capability, add a manufacturing capability review step to the design checklist for stress-critical dimensions. This level of corrective action prevents the same mistake from recurring on future designs.
Level 3: Knowledge Capture
Document the failure mode, root cause, and corrective action in a lessons-learned database or design standard. Future designers working on similar components should have access to this knowledge. A design team that does not capture lessons learned will rediscover the same failure modes on future projects.
FAQ
Q: How do I handle pressure from management to issue a quick fix before root cause is fully understood?
Separate the containment action from the corrective action explicitly. You can contain the problem immediately — holding inventory, implementing a temporary inspection, issuing a field advisory — while the root cause investigation continues. Present this distinction clearly: "We have contained the defect. The permanent corrective action will be issued after root cause is confirmed, which I expect will take X days." Issuing a design change before the root cause is understood is a technical risk, and the responsible engineer must communicate that risk clearly.
Q: What is the best way to conduct a root cause analysis for a complex assembly failure?
Start by preserving and examining the failed hardware. Physical evidence — fracture surfaces, wear patterns, deformation — tells you where the failure initiated and often how it progressed. Resist the temptation to disassemble the failed unit too quickly; the as-found condition is data. Supplement physical evidence with production records, dimensional measurements of retained samples, and a review of the design analysis to identify any cases where margins were thin. The combination of physical evidence and analytical review is more reliable than either alone.
Q: Should the designer who created the original drawing be involved in the root cause analysis?
Yes — and not punitively. The designer who created the drawing has context that is often unavailable anywhere else: what trade-offs were made during design, what constraints existed, what the design analysis showed, and what alternatives were considered. Excluding the original designer from the root cause investigation often means re-discovering context that was already documented internally. The investigation should be focused on understanding and preventing the failure, not on establishing blame.
コメント