D5: Verifying Corrective Actions with Data

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.

A single good batch is not proof. As a working rule, verify with a minimum of three consecutive production runs, ideally across different shifts and operators, and support the result with a capability study of at least 30 pieces from the post-change process. Compare the before/after capability indices (Cp/Cpk) and defect rate against the target you set in D4. If the process is not statistically capable after the change, the PCA is not verified — return to D4 rather than proceeding to D6.

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.

Containment stays in place until D5 verification is complete and signed off. Removing it early is one of the most common D-step failures: the defect returns, and the team loses the clean data baseline it needs to confirm the fix.

Worked Example: Verifying a Hardness Change

A machined bushing wears prematurely, producing out-of-round bores. The team proposes raising surface hardness from 55 HRC to 60 HRC. Before committing:

- Baseline: 30-piece study shows Cpk = 0.82 on bore roundness; defect rate 4.2%. - Pilot run: Three consecutive lots at 60 HRC, 30 pieces each, across two shifts. - Result: Cpk rises to 1.41; defect rate falls to 0.1%. -