D5 — How Do You Know Your Fix Actually Works? D5 PCA Verification Guide

Part of the 8D methodology (D0–D8).

Discipline 5 (D5) is the "before-the-fact" validation phase. Before you commit to a permanent change in tooling, software, or design, you must prove that the proposed solution actually works and doesn't introduce "Side Effects."

How Many Batches Prove a Fix Works? The team should evaluate multiple options. The ideal PCA is: - Fail-Safe (Poka-Yoke): Makes it physically impossible for the error to occur. - Robust: Works across all shifts, operators, and environmental conditions. - Cost-Effective: Balanced against the risk of the defect.

What’s a Side-Effect Analysis? - DOE (Design of Experiments): Testing the new process parameters. - Pilot Runs: Producing a small batch under "Gold Standard" conditions. - Simulation: Using software to predict the fatigue or stress performance of a new design.

Should You Remove Containment Before Verification? Every change has a consequence. If you increase the hardness of a material to prevent wear (PCA), does it make the part too brittle for low-temperature environments? D5 must explicitly document these "What-If" scenarios.

How Many Samples Are Enough?

For a defect at 3%, you need at least 100 samples for 95% confidence. Formula: n = ln(1 - confidence) / ln(1 - defect rate). With 20 samples, a 3% defect has a 46% chance of being missed. Always present your sample size calculation.

What If the PCA Fails?

A failed D5 is not a dead end. Return to D4 and refine the hypothesis. Document failed attempts with test data. Budget for at least one failed D5 iteration in your 8D timeline.

Real-World D5: Gauge R&R

A manufacturer had 35% GRR in their measurement system. After improving fixtures and training operators, GRR dropped to 8%. Only then did they validate the fix. Without fixing measurement first, any validation is confounded by measurement noise.