Design of Experiment: Steps:
- Apply brainstorming, and Fishbone, flowchart & process mapping diagrams, to discover KPIVs.
- Limit to most 3 important KPIVs (through pareto, QDF & FMEA).
- Use combination of values & out of range values and repeat each experiment 5 times, design your “if then” results early to direct the experiment.
- Limit DOE to 1 factor (2way Anova:2 factors) and Control other KPIVs.
- Apply statistical tests to result: variable vs proportional. Often the experiment process not the result is what will lead to discoveries.
- Tolerance: The accepted level of variation beyond which a quality adjustment should take place. in improve & control steps of DMAIC. For KPIVs: T=3S, and for KPIs: RSS(root sum of squares) method of contributing KPIVs:
Taguchi Methods for Robust Design: Robustness: Insensitivity to variations in the process (Process Capability).
Taguchi Loss Function. Signal-to-Noise Ratio. Taguchi Design of Experiment.
Taguchi Tolerances and Loss Function:
Loss Function: L(x) = k(x – T)². Where T is the process tolerance, normally given as: X̅ ± T.
Expected Loss = k(S² + D²). D² = (X̅ – T)². S is Sigma (Standard Deviation).