Improve: Solutions, DOE, and Implementation
Improve: Solutions, DOE, and Implementation
Improve generates and tests changes that address verified causes, then implements them without creating new failure modes. The exam tests solution selection, DOE principles, piloting, mistake-proofing, lean tools, and change management.
Core rules and concepts
- Solution selection: use weighted criteria (impact, cost, risk, cycle time) and verify the solution addresses the verified root cause.
- DOE: randomization, replication, and blocking; full factorial for a few factors, fractional factorial for screening many; interactions mean you cannot optimize one factor at a time.
- Center points test for curvature; follow screening with optimization such as response surface methods when needed.
- Lean tools: 5S, standard work, flow, pull, kaizen, and poka-yoke. Mistake-proofing prevents errors at the source.
- Pilot: define success criteria, run in a representative area, compare against baseline using the same measurement system, and update the FMEA.
Worked example
A 2^3 full factorial with three replicates tests temperature, pressure, and speed. The pressure main effect is largest, but the temperature x speed interaction is significant. Reasoning: because the interaction is significant, the effect of temperature depends on speed; setting each factor at its own best level from one-factor-at-a-time trials can miss the optimum. Run a confirmatory combination at the predicted optimum and verify with a confidence interval.
Traps that cost marks
Running DOE before verifying root causes; forgetting randomization or replication; confounding a fractional design without understanding the alias structure; piloting in a non-representative area; ignoring interactions; implementing without the process owner; claiming savings that finance has not verified; leaving the FMEA unchanged.