PRINCIPLES OF SIX SIGMA

1
PRINCIPLES OF SIX SIGMA

• ISE 412/512
• LECTURE 5-1
• MEASURE (DMAIC)
• THOMAS E. SCOTT, PhD

2
FUNDAMENTALS OF 6s
BELTS YELLOW, GREEN, BLACK, MBB, CHAMPION, CEO
500K and 6 Months
3
6s Problem Solving Methodology
Existing Business Processes
Supplier
Inputs
Processes
Outputs
Customer

• Six Sigma Methodology
• Define
• Measure
• Analyze
• Improve
• Control

DMAIC
Quality Productivity Cost Profitability
Improved Business Performance
4
Six Sigma Premise

• The variation of the process
• (Voice of the Process)
• can use up no more than
• HALF
• of the specification limit
• (Voice of the Customer)

5
Six Sigma Problem Solving Methodology
Existing Business Processes
What is important?
DEFINE
MEASURE
How are we doing?
What is wrong?
ANALYZE
What needs to be done?
IMPROVE
How do we guarantee performance?
CONTROL
IMPROVED BUSINESS PERFORMANCE
6
DMAIC Define

• All about the OUTPUT
• What is important to the customer?
• DEFINE
• Scope and boundary
• Specify Measurable Defects
• Estimate the impact

7
DMAIC Define

• Business case with a project objective
• SIPOC analysis
• Voice of Customer analysis

8
DMAIC Measure

• All about the INPUT
• What things can we control?
• MEASURE
• Map process
• Identify inputs and outputs
• Cause and Effects Matrix
• Preliminary FMEA
• Establish Measurement System Capability
• Establish Process Capability Baseline

9
DMAIC Measure

• Operationally define each CTQ characteristic
• Understand the measurement system for each CTQ
  characteristic
  
• Understand the current capability of each CTQ
  characteristic

10
GAGE RR

• ESTIMATE THE PORTION OF TOTAL VARIATION RELATED
  TO
  
• UNIT-TO-UNIT VARIATION
• R R
• REPEATABILITY
• REPRODUCIBILITY
• OPERATOR-PART INTERACTION
• DIFFERENT PEOPLE MEASURE DIFFERENT UNITS
  DIFFERENT WAYS

11
GAGE RR DATA COLLECTION

• COLLECT THE DATA USING THE FULL RANGE OF
  CONDITIONS EXPERIENCED BY THE MEASUREMENT SYSTEM
  
• IN THE MEASURE PHASE, WE EVALUATE THE MEASUREMENT
  SYSTEM CAPABILITY SO WE CAN BE CERTAIN OF THE
  QUALITY OF THE DATA BEING OBTAINED
  
• ANSWERS THE QUESTION CAN THE DATA BE BELIEVED?

12
GAGE RR

• FOR EXAMPLE
• HAVE TWO INSPECTORS MEASURE FIVE ITEMS, FOUR
  DIFFERENT TIMES
  
• 2 x 5 x 4 40 MEASUREMENTS

13
GAGE RUN STUDY

• GOOD REPEATABILITY LOW VARIATION OF SQUARES
• GOOD WITHIN GROUP REPEATABILITY
• GOOD REPRODUCIBILITY
• GOOD BETWEEN GROUP REPRODUCIBILITY
• CONCLUSION MOST OF THE VARIATION IS DUE TO UNIT
  DIFFERENCES

14
Gage RR Study - ANOVA Method
Two-Way ANOVA Table With Interaction Source
DF SS MS
F P Unit
4 17.6209 4.40522 2005.79 0.000
Inspector 1 0.1061
0.10609 48.31 0.002 Unit Inspector
4 0.0088 0.00220 1.23 0.319
Repeatability 30 0.0536 0.00179
Total 39 17.7894

• UNIT TO UNIT VARIATION MOST OF THE
  VARIATION (SIGNIFICANT) P 0.000
  
• (NOTE P IS THE OBSERVED LEVEL OF
  SIGNIFICANCE)
  
• INSPECTOR TO INSPECTOR VARIATION SMALL
  BUT SIGNIFICANT P 0.002
  
• INSPECTOR TO UNIT VARIATION SMALL AND NOT
  SIGNIFICANT P 0.319
  
• BECAUSE OF THIS, NO NEED TO CONSIDER INTERACTION

15
UNIT TO UNIT VARIATION
INSP CROSS UNIT VARIATION
INSP TO INSP VARIATION
16
Two-Way ANOVA Table Without Interaction
(WITHOUT INSPECTOR CROSS UNIT VARIATION)
Source DF SS MS
F P
Unit 4 17.6209
4.40522 2400.86 0.000 Inspector
1 0.1061 0.10609 57.82
0.000 Repeatability 34 0.0624 0.0
0183 Total 39 17.7894
PART TO PART AND UNIT VARIATIONS ARE SIGNIFICANT
17
GAGE RR ASSESSMENT

• Gage RR (THE STUDY OF VARIATION)
• Contribution
• Source VarComp (of VarComp)
• Total Gage RR 0.007048 1.26
• Repeatability 0.001835 0.33
• Reproducibility 0.005213 0.94
• Inspector 0.005213 0.94
• Part-To-Part 0.550423 98.74
• Total Variation 0.557471 100.00
• Study Var
  Study Var
  
• Source StdDev (SD) (6  SD)
  (SV)
  
• Total Gage RR 0.083950 0.50370
  11.24
  
• Repeatability 0.042835 0.25701
  5.74
  
• Reproducibility 0.072199 0.43320
  9.67
  
• Inspector 0.072199 0.43320
  9.67
  
• Part-To-Part 0.741905 4.45143
  99.37
  

• A GOOD MEASUREMENT SYSTEM HAS A MAXIMUM GAGE RR
  OF 10
  
• GAGE RR OF 10 TO 30 ARE MARGINAL SYSTEMS
• THEREFORE, THIS MEASUREMENT SYSTEM IS MARGINAL

TOTAL EFFECT 11.24
18
GAGE RR CONCLUSIONS

• UNIT TO UNIT VARIATION IS A LARGE PORTION OF
  TOTAL VARIATION
  
• THIS IS GOOD

19
UNIT TO UNIT VARIATION IS A LARGE PORTION OF
TOTAL VARIATION
SMALL VARIATION IN UNIT MEASUREMENT (A GOOD THING)
INSPECTORS ASSIGN THE SAME MEASURE TO THE SAME
UNIT (REPRODUCIBILITY)
EASY TO SEE THAT MOST VARIATION IS IN MEASUREMENT
(A GOOD THING)
SMALL RANGE IN REPEATED MEASUREMENT MEANS NARROW
CONTROL LIMITS (A GOOD THING)
20
PAPER ORGANIZERS INTERNATIONAL

• OPERATIONAL DEFINITION OF BENDS
• BEND THE METALLIC SECURING DEVICES (MSDs)
• COLLECT DATA FOR
• WHERE THE MSD WAS OBTAINED
• WHO THE INSPECTOR WAS
• HOW MANY BENDS TO FAILURE

21
GAGE RR RUN CHART
22
YIELD

• BASED ON BEND TESTS (TABLE 16.18)
• DURABILITY
• 6 GOOD OUT OF 16 TRYS (YIELD .375)
• CRITERIA WAS AT LEAST 4 BENDS
• FUNCTIONALITY
• 6 BOXES OUT OF 16 ATTEMPTS (YIELD .375)
• CRITERIA WAS NO MORE THAN FIVE FAILURES PER BOX

23
INDIVIDUAL AND MOVING RANGE CHART
WHOA! WHATS THIS?
24
DOT PLOT

• NOTICE THAT THE DATA IS NOT NORMALLY DISTRIBUTED!
  YIPES!! THEREFORE
  
• SHOULD NOT USE I MR CHART AT THIS TIME
• LOOKS LIKE POISSON, SO USE C-CHART

25
C-CHART (POISSON)
YIPES! WHATS THIS? LOOKS LIKE MAYBE THE FIRST
HOUR THERE WAS A PROBLEM. PERHAPS SLOWER BENDING.
26
C-CHART
LOOKS GOOD!
27
POISSON DISTRIBUTED
28
PROCESS PERFORMANCE FOR CTQs

• CTQs
• DURABILITY
• FUNCTIONALITY

100 FOLD IMPROVEMENT CONSISTENT WITH GOALS STATED
IN THE DEFINE PHASE