19.2 Six Sigma DMAIC Methodology and Quality Function Deployment

Pioneered by Bill Smith and Bob Galvin at Motorola in the mid-1980s and popularized across global manufacturing by Jack Welch at General Electric, Six Sigma is a disciplined, data-driven methodology that combines advanced statistical analytics with structured project management to eliminate defects, drive down process variation, and maximize operational profitability.

---

1. Six Sigma Statistical Foundations and the $1.5\sigma$ Shift

To understand Six Sigma on the FE exam, engineers must distinguish between theoretical short-term capability and empirical long-term performance.

                  Six Sigma Distribution with 1.5σ Shift
                  
                             Centered Mean
                                 (μ_0)
                                   │
                     Shifted Mean  │
                         (μ_s)     │
                           │       │
                           ▼       ▼
                          ┌─┐
                         /   \     │
                        /     \    │
                       /   │   \   │
                      /    │    \  │
                    _/     │     \_│_
          ─────────/───────┼───────\───┬───────────────────►
                  LSL              USL │
                                       │
                  ◄──────── 4.5σ ──────►
                  (Shifted distance to nearest limit)

The Normal Distribution Derivation

Assume a quality characteristic $Y$ is normally distributed with mean $\mu$ and standard deviation $\sigma$, bounded by Upper and Lower Specification Limits ($USL$ and $LSL$):

• Theoretical Centered Process: If the process mean is centered exactly halfway between specification limits placed at $\mu \pm 6\sigma$ (so $USL - \mu = 6\sigma$ and $\mu - LSL = 6\sigma$), the probability of producing a defective part falling outside the specification limits is: $P(\text{Defect}) = 2 \times \left[ 1 - \Phi(6.0) \right] \approx 2 \times (9.866 \times 10^{-10}) = 1.973 \times 10^{-9}$ This corresponds to approximately 2 parts per billion, i.e. 0.002 Defects Per Million Opportunities.
• The Empirical Long-Term $1.5\sigma$ Shift: In real-world industrial environments, processes do not remain permanently centered. Ambient temperature swings, lot-to-lot raw material variations, machine tool wear, hydraulic viscosity changes, and operator shifts cause the process mean to drift over time. Empirical studies across manufacturing operations conducted by Motorola revealed that over the long term, process means exhibit an average drift of approximately $1.5\sigma$.
• The Long-Term Six Sigma Standard: When the mean drifts by $1.5\sigma$ toward a specification limit, the distance from the shifted mean to the nearest specification limit drops from $6\sigma$ to: $Z_{\text{nearest}} = 6.0\sigma - 1.5\sigma = 4.5\sigma$ Evaluating the standard normal tail probability at $Z = 4.5$: $P(Z > 4.5) = 1 - \Phi(4.5) \approx 3.3976 \times 10^{-6} \approx 3.4 \times 10^{-6}$ Multiplying by $10^6$ yields the universal benchmark of Six Sigma quality: 3.4 Defects Per Million Opportunities (DPMO).

Fundamental Defect Metrics

Industrial engineers monitor Six Sigma quality using three core metrics:

• Defects per Unit ($DPU$): The average number of defects observed per unit inspected: $DPU = \frac{D}{U}$ where $D$ is the total count of defects discovered, and $U$ is the total number of units sampled.
• Total Opportunities ($TOP$): The cumulative number of defect opportunities across the entire sample: $TOP = U \times O$ where $O$ is the number of distinct Critical-to-Quality (CTQ) defect opportunities present on a single unit.
• Defects Per Million Opportunities ($DPMO$): $DPMO = \frac{D}{U \times O} \times 10^6 = \frac{DPU}{O} \times 10^6$

---

2. Six Sigma Organizational Governance and Belt Hierarchy

Six Sigma institutionalizes quality improvement by embedding dedicated, trained practitioners across defined organizational tiers:

                  Six Sigma Organizational Belt Architecture
                  
                      ┌────────────────────────┐
                      │  Executive Leadership  │ (Strategic vision & resource sponsor)
                      └───────────┬────────────┘
                                  ▼
                      ┌────────────────────────┐
                      │   Champion / Owner     │ (Business unit head; removes roadblocks)
                      └───────────┬────────────┘
                                  ▼
                      ┌────────────────────────┐
                      │   Master Black Belt    │ (Enterprise methodology expert & coach)
                      └───────────┬────────────┘
                                  ▼
                      ┌────────────────────────┐
                      │       Black Belt       │ (Full-time team leader & statistical analyst)
                      └───────────┬────────────┘
                                  ▼
                      ┌────────────────────────┐
                      │       Green Belt       │ (Part-time departmental project execution)
                      └───────────┬────────────┘
                                  ▼
                      ┌────────────────────────┐
                      │ Yellow / White Belts   │ (Shop-floor contributors & data collectors)
                      └────────────────────────┘

• Executive Leadership / Sponsors: Senior executive officers who establish enterprise strategic quality objectives, allocate capital, and hold business unit leaders accountable for results.
• Champions / Process Owners: Business unit directors who "own" the operating process. Champions define project boundaries, select Black Belts, approve project charters, secure resources, and eliminate organizational roadblocks across functional departments.
• Master Black Belts (MBBs): Elite technical experts who serve as full-time internal consultants, mentors, and instructors. MBBs advise executive leadership, train Black and Green Belts, and ensure statistical rigor across the enterprise portfolio.
• Black Belts (BBs): Dedicated, full-time project leaders who execute high-impact strategic DMAIC projects (typically targeting $100,000+ in annual financial savings). Black Belts mentor Green Belts and possess mastery of statistical techniques including Design of Experiments (DOE), multivariate regression, and non-parametric hypothesis testing.
• Green Belts (GBs): Part-time quality practitioners who lead localized departmental improvement projects while retaining their normal operational job duties (e.g., manufacturing engineers, maintenance supervisors).
• Yellow Belts / White Belts: Operations personnel, technicians, and operators trained in baseline Six Sigma concepts, visual controls, 5S, and data collection protocols who serve as team members on DMAIC projects.

---

3. The DMAIC Roadmap in Depth

DMAIC (Define, Measure, Analyze, Improve, Control) is the closed-loop, five-phase problem-solving architecture applied to stabilize and optimize existing defective processes.

                         The DMAIC Project Pipeline
                         
┌──────────────┐   ┌──────────────┐   ┌──────────────┐   ┌──────────────┐   ┌──────────────┐
│    Define    │──►│   Measure    │──►│   Analyze    │──►│   Improve    │──►│   Control    │
└──────────────┘   └──────────────┘   └──────────────┘   └──────────────┘   └──────────────┘
- Project       - Value Stream    - Root Cause       - Design of Exp.   - SPC Control
  Charter         Mapping           (Fishbone/5Why)    (DOE Factorials)   Charts
- VOC to CTQ    - Gage R&R (MSA)  - Hypothesis Tests - FMEA (RPN)       - Standard SOPs
- SIPOC Macro   - Baseline C_pk     (t, ANOVA, χ²)   - Poka-Yoke Fixture- Control Plan
  Diagram         & Sigma Level   - Regression       - Pilot Trials     - Hand-off

Phase 1: Define

The objective is to establish project scope, identify customer requirements, and formulate the business case.

• Project Charter: The contractual document bounding the project, containing:
  • Problem Statement: Quantifies the historical baseline metric, the performance gap, and the timeline without presuming root cause (e.g., "During Q1-Q3 2026, Line 4 scrap averaged 4.8%, resulting in an annualized loss of $340,000").
  • Business Case: Connects the project to corporate financial and strategic imperatives.
  • Goal Statement: SMART objective (e.g., "Reduce Line 4 scrap from 4.8% to under 1.5% by November 15, 2026, generating $230,000 in net savings").
  • Project Scope: Explicit boundaries identifying what is In-Scope and Out-of-Scope.
• Voice of the Customer (VOC) and CTQ Tree: VOC captures qualitative, subjective customer feedback (e.g., "The hydraulic pump is too noisy"). The Critical to Quality (CTQ) Tree decomposes broad customer needs into measurable, specific engineering parameters with numerical tolerances (e.g., "Acoustic output $\le 68\text{ dBA}$ at 1 meter under full hydraulic load").
• SIPOC Diagram: A high-level process map bounding the system into five macro elements:
  • Suppliers $\to$ Inputs $\to$ Process (4 to 7 macro steps) $\to$ Outputs $\to$ Customers.

Phase 2: Measure

The objective is to quantify the current process baseline and validate the integrity of the measurement data.

• Detailed Process Mapping: Deployment swimlane flowcharts identifying decision gates, queues, and operator handoffs.
• Measurement System Analysis (MSA) / Gage R&R: Before drawing conclusions from process data, engineers must verify that measurement error does not obscure true part variation. Total observed process variance ($\sigma^2_{\text{total}}$) decomposes into: $\sigma^2_{\text{total}} = \sigma^2_{\text{part-to-part}} + \sigma^2_{\text{measurement}}$ $\sigma^2_{\text{measurement}} = \sigma^2_{\text{repeatability}} + \sigma^2_{\text{reproducibility}}$
  • Repeatability (Equipment Variation, EV): Variation observed when one appraiser measures the same identical part multiple times using the same gage.
  • Reproducibility (Appraiser Variation, AV): Variation observed when different appraisers measure the same part using the same gage.
  • Gage R&R Acceptance Thresholds (%GRR):
    • $%GRR < 10%$: Excellent measurement system; fully acceptable.
    • $10% \le %GRR \le 30%$: Marginally acceptable depending on application criticality and replacement cost.
    • $%GRR > 30%$: Unacceptable; the measurement system must be recalibrated, redesigned, or fixtured before collecting process capability data.
• Baseline Capability Assessment: Calculating $C_p$, $C_{pk}$, $P_p$, $P_{pk}$, and initial $Z$-bench baseline metrics.

Phase 3: Analyze

The objective is to identify, isolate, and statistically verify the vital few root causes ($X$s) that govern process performance ($Y = f(X_1, X_2, \dots, X_k)$).

• Root Cause Identification: Brainstorming potential causes using Ishikawa (Fishbone) diagrams and the 5 Whys technique.
• Statistical Hypothesis Testing:
  • Comparing Two Means: Two-sample independent $t$-test or paired $t$-test.
  • Comparing Three or More Means: One-Way or Two-Way Analysis of Variance (ANOVA) using the $F$-test.
  • Categorical / Attribute Data: Chi-Square ($\chi^2$) test for independence or goodness-of-fit.
• Correlation and Multiple Linear Regression: Modeling the continuous mathematical relationship between process inputs (e.g., cutting temperature, feed rate) and the CTQ output.

Phase 4: Improve

The objective is to formulate, screen, optimize, and pilot breakthrough solutions.

• Design of Experiments (DOE): Full factorial ($2^k$) or fractional factorial ($2^{k-p}$) experimental designs that systematically manipulate input variables to quantify main effects and multi-factor interactions while minimizing experimental trial count.
• Failure Modes and Effects Analysis (FMEA): A structured engineering risk assessment tool that identifies potential failure mechanisms and calculates the Risk Priority Number (RPN): $RPN = \text{Severity } (S) \times \text{Occurrence } (O) \times \text{Detection } (D)$ where $S$, $O$, and $D$ are scored on 1-to-10 integer scales ($1 \le RPN \le 1,000$). High RPN items mandate immediate engineering redesign, error-proofing, or detection enhancement.
• Poka-Yoke Fixturing: Designing mechanical interlocks, asymmetric locating pins, and optical proximity sensors to make human operational error physically impossible.
• Pilot Implementation: Executing a limited trial run to validate projected gains and verify that no unintended side-effects occur.

Phase 5: Control

The objective is to institutionalize improvements, update standards, and ensure sustained capability.

• Statistical Process Control (SPC): Deploying real-time variable control charts ($\bar{X}-R$, $\bar{X}-s$) or attribute control charts ($p$, $np$, $c$, $u$) on shop-floor terminals to detect special cause variation.
• Standard Operating Procedures (SOPs): Documenting validated work steps, tooling offsets, and preventative maintenance schedules.
• Process Control Plan: A master document detailing measurement frequencies, sample sizes, control methods, and specific reaction plans (out-of-control action plans, OCAP) if an alarm triggers.
• Formal Project Handoff: Securing executive sign-off and transferring operational accountability from the Black Belt to the designated Champion / Process Owner.

---

4. Design for Six Sigma (DFSS) and the DMADV Framework

While DMAIC focuses on incremental optimization of existing, operational processes, Design for Six Sigma (DFSS) is deployed to engineer brand-new products, services, or manufacturing processes from scratch.

                  DMAIC vs. DMADV Decision Architecture
                  
                        Operating Process Problem
                                   │
                 Is the process currently operating?
                                  / \ 
                            Yes  /   \  No ──► Deploy DMADV (DFSS)
                                /     \
         Can the process reach   
         target capability (C_pk ≥ 1.5)
         via incremental parameter tuning?
                       / \ 
                 Yes  /   \  No ──► Deploy DMADV (Re-architect system)
                     /     \
                    ▼       ▼
              Deploy DMAIC  Deploy DMADV

DFSS is governed by the DMADV roadmap:

1. Define: Establish customer requirements, product vision, and strategic business goals.
2. Measure: Identify customer CTQs, benchmark competitor capabilities, and assess technical risks.
3. Analyze: Formulate innovative conceptual design architectures, evaluate design alternatives using decision matrices (Pugh Concept Selection), and model preliminary transfer functions.
4. Design: Develop detailed engineering specifications, optimize component tolerances, conduct robust parameter design (Taguchi methods), and simulate performance using finite element analysis (FEA).
5. Verify: Fabricate physical prototypes, conduct pilot production runs, validate Six Sigma capability under operational extremes, and hand off to full-scale manufacturing.

---

5. Quality Function Deployment (QFD) and the House of Quality

Developed by Yoji Akao in 1966 at Mitsubishi's Kobe Shipyard, Quality Function Deployment (QFD) is a structured product development methodology that translates qualitative customer demands into quantitative engineering characteristics.

The foundational graphical tool of QFD is the House of Quality (HoQ):

                      The House of Quality (HoQ)
                      
                              ┌───────────────┐
                              │ 4. Roof:      │ (Interrelationships
                              │  Correlation  │  between HOWs: ++, +, -, --)
                              │    Matrix     │
                              └───┬───────┬───┘
                                  │       │
                     ┌────────────┴───────┴────────────┐
                     │ 2. Technical Characteristics    │ (Direction of
                     │    (HOWs: Engineering Specs)    │  improvement: ▲, ▼, ⊙)
 ┌───────────────────┼─────────────────────────────────┼───────────────────┐
 │ 1. Customer       │ 3. Interrelationship Matrix     │ 5. Competitive    │
 │    Requirements   │                                 │    Assessment     │
 │    (WHATs: Voice  │    ● = Strong (9)               │    (Benchmarking  │
 │     of Customer)  │    ○ = Moderate (3)             │     1 to 5 scale) │
 │    Customer       │    ▲ = Weak (1)                 │                   │
 │    Importance     │                                 │                   │
 │    Weights (w_i)  │                                 │                   │
 └───────────────────┴─────────────────────────────────┴───────────────────┘
                     │ 6. Basement: Technical Targets  │
                     │    - Absolute Weight (W_j)      │
                     │    - Relative Weight (%)        │
                     │    - Engineering Target Values  │
                     └─────────────────────────────────┘

Anatomy of the House of Quality

1. Customer Requirements (WHATs): The left vertical wall lists customer expectations captured via VOC interviews, focus groups, and surveys. Each requirement is assigned a Customer Importance Weight ($w_i$), typically on a 1-to-5 or 1-to-10 scale.
2. Technical Characteristics (HOWs): The second-story ceiling lists measurable engineering parameters under direct engineering design control. Each characteristic includes a direction of optimization:
   • $\uparrow$ (Maximize: e.g., tensile strength, battery life)
   • $\downarrow$ (Minimize: e.g., acoustic noise, curb weight, vibration)
   • $\odot$ (Target: e.g., nominal hydraulic line pressure, shut height)
3. Interrelationship Matrix: The central room links WHATs to HOWs using standard standardized numerical scoring conventions:
   • Strong Relationship: $\text{Weight } R_{ij} = 9$ (often symbolized by a solid circle $\bullet$)
   • Moderate Relationship: $\text{Weight } R_{ij} = 3$ (symbolized by an open circle $\circ$)
   • Weak Relationship: $\text{Weight } R_{ij} = 1$ (symbolized by an open triangle $\triangle$)
   • No Relationship: $\text{Weight } R_{ij} = 0$ (blank cell)
4. Correlation Roof: The triangular roof models technical trade-offs between engineering characteristics. It highlights synergies (positive correlations $+$, where improving Characteristic A also improves Characteristic B) and engineering conflicts (negative correlations $-$, where improving Characteristic A degrades Characteristic B, such as increasing armor plating thickness degrading fuel economy).
5. Competitive Assessment: The right wing benchmarks the firm's current product against primary marketplace competitors across each customer requirement.
6. Basement (Technical Priorities & Targets): Industrial engineers calculate the Absolute Technical Importance Weight ($W_j$) for each engineering characteristic $j$: $W_j = \sum_{i=1}^m w_i \cdot R_{ij}$ where $w_i$ is the importance weight of customer requirement $i$, and $R_{ij}$ is the relationship score between requirement $i$ and engineering characteristic $j$. The Relative Percentage Weight ($\text{RPW}_j$) is: $\text{RPW}_j = \frac{W_j}{\sum_{k=1}^n W_k} \times 100\%$

---

6. Step-by-Step Worked Engineering Calculations

Worked Example 19.2.1: Six Sigma Defect Metrics and House of Quality Prioritization

Problem Statement: Part 1: An automated printed circuit board (PCB) assembly facility manufactures 25,000 control boards for an avionics supplier. Quality engineers define five Critical-to-Quality (CTQ) defect opportunities per board: (1) solder bridging, (2) missing component, (3) reversed component polarity, (4) dry solder joint, and (5) board delamination. Following automated optical inspection and functional electrical testing, inspectors record a total of 45 defects across the entire lot.

1. Calculate Defects per Unit ($DPU$).
2. Compute Total Opportunities ($TOP$).
3. Calculate Defects Per Million Opportunities ($DPMO$).
4. Determine the approximate long-term sigma level of the assembly process.

Part 2: The avionics engineering team builds a House of Quality (HoQ) matrix to prioritize redesign efforts for the flight computer enclosure. Customer requirements, importance weights ($w_i$), and relationship scores ($R_{ij}$) across three engineering characteristics are tabulated below (using the standard 9-3-1 scoring rule):

Compute the absolute technical importance score ($W_j$) and relative percentage weight ($\text{RPW}_j$) for all three engineering characteristics.

Solution:

Part 1: Six Sigma Defect Analytics

• Step 1: Calculate Defects per Unit ($DPU$): $DPU = \frac{D}{U} = \frac{45\text{ defects}}{25,000\text{ units}} = 0.0018\text{ defects/unit}$
• Step 2: Calculate Total Opportunities ($TOP$): $TOP = U \times O = 25,000\text{ units} \times 5\text{ opportunities/unit} = 125,000\text{ opportunities}$
• Step 3: Calculate Defects Per Million Opportunities ($DPMO$): $DPMO = \frac{D}{TOP} \times 10^6 = \frac{45}{125,000} \times 1,000,000 = 0.00036 \times 1,000,000 = 360\text{ DPMO}$
• Step 4: Determine Sigma Level: Referencing standard Six Sigma conversion tables (with the standard $1.5\sigma$ shift):
  • $4.5\sigma \to 1,350\text{ DPMO}$
  • $4.87\sigma \to 360\text{ DPMO}$
  • $5.0\sigma \to 233\text{ DPMO}$ The PCB assembly line operates at approximately $4.87\sigma$, performing well above four-sigma capability ($6,210\text{ DPMO}$) but slightly short of world-class five-sigma capability.

Part 2: House of Quality Scoring Calculations

• Step 1: Absolute Importance for Characteristic 1 (Wall Thickness, $W_1$): $W_1 = (w_1 \cdot R_{11}) + (w_2 \cdot R_{21}) + (w_3 \cdot R_{31})$ $W_1 = (5 \times 9) + (4 \times 1) + (3 \times 0) = 45 + 4 + 0 = 49$
• Step 2: Absolute Importance for Characteristic 2 (Heat Sink Area, $W_2$): $W_2 = (w_1 \cdot R_{12}) + (w_2 \cdot R_{22}) + (w_3 \cdot R_{32})$ $W_2 = (5 \times 0) + (4 \times 9) + (3 \times 3) = 0 + 36 + 9 = 45$
• Step 3: Absolute Importance for Characteristic 3 (Internal Fan Airflow, $W_3$): $W_3 = (w_1 \cdot R_{13}) + (w_2 \cdot R_{23}) + (w_3 \cdot R_{33})$ $W_3 = (5 \times 0) + (4 \times 9) + (3 \times 9) = 0 + 36 + 27 = 63$
• Step 4: Calculate Sum of Absolute Weights and Relative Percentages: $W_{\text{sum}} = W_1 + W_2 + W_3 = 49 + 45 + 63 = 157$ $\text{RPW}_1 (\text{Wall Thickness}) = \frac{49}{157} \times 100\% = 31.21\%$ $\text{RPW}_2 (\text{Heat Sink Area}) = \frac{45}{157} \times 100\% = 28.66\%$ $\text{RPW}_3 (\text{Fan Airflow}) = \frac{63}{157} \times 100\% = 40.13\%$
• Engineering Conclusion: Internal Fan Airflow ranks as the highest technical priority (40.13%), heavily influencing both internal operating temperature and acoustic emissions.

---

7. NCEES Reference Handbook Tips & Realistic Exam Traps

• The $1.5\sigma$ Shift Confusion: On the FE exam, questions may ask for the DPMO of a centered Six Sigma process versus a shifted Six Sigma process. Remember: 3.4 DPMO assumes the long-term $1.5\sigma$ shift ($Z = 4.5$). A perfectly centered Six Sigma process ($Z = 6.0$) produces a negligible 0.002 parts per billion (0.002 DPMO).
• DPMO Denominator Trap: Never divide defects by units alone! If 10 defects are found in 1,000 units, the defect rate is not $10/1,000 \times 10^6$. You must multiply the number of units by the number of defect opportunities per unit ($O$): $\text{Correct: } DPMO = \frac{D}{U \times O} \times 10^6$
• DMAIC vs. DMADV Selection: If an exam prompt describes designing a new product, introducing an unprecedented service, or completely re-engineering an obsolete assembly line that cannot meet customer specs through parameter adjustment, the correct answer is DMADV (DFSS). If the prompt describes reducing scrap or cycle time on an existing active production line, choose DMAIC.
• Gage R&R Percentage Criteria: Memorize the three ASQ/AIAG thresholds:
  • $< 10%$ = Acceptable
  • $10% \text{ to } 30%$ = Marginally acceptable
  • $> 30%$ = Unacceptable (data cannot be trusted for process capability studies)
• QFD Relationship Values: The standard convention for House of Quality cell scores is 9 (Strong), 3 (Moderate), and 1 (Weak). Do not assume a linear 3-2-1 scale unless explicitly stated in the exam problem.