Measure Phase

Module 3

1
 Measure Phase Objectives

❑ Validated project definition

❑ Detailed process map with clearly defined process output measures (Y‟s)

❑ Measurement System Analyses on the Y‟s

❑ Control charts on baseline, process performance

❑ C&E Matrix to determine focus within process

❑ FMEA to assess and prioritize process risk

❑ Graphical interpretation of data

Non-Business Use
2
 Questions to answer in the Measure Phase

❑ What are the process output measures? Has data been collected around this metric
previously?

❑ What is/are the defect definition for this project?

❑ What is a unit? Are there more than one opportunity per unit for a defect? Is the defect,
metric, spec unit, problem statement and customer aligned?

❑ Where will you collect data? What is your data collection plan? How much data did
you collect? Is your ability to measure/detect “good enough?” What is the GRR%?

❑ What sampling strategy and sample size did you use? Was it statistically determined?

❑ Have you found any “quick hit” improvements?

❑ What is the current process Sigma level for this project? What's the best that the process
was “designed” to do? Does the impact opportunity justify continuing this project?
Non-Business Use
3
 Agenda

– Process Mapping & Value Stream Map

– Cause & Effect/ Fishbone Diagram

– Cause & Effect/ X -Y Matrix

– Failure Mode Effect Analysis (FMEA)

– Statistics

– Normal Distributions & Normality

– Graphical Analysis

– Measurement System Analysis

– Process Capability

4
Non-Business Use
Process Mapping & Value Stream Map

5
 What is a process map ? (1/2)

❑ A process map is

▪ A graphical representation of the flow of a process

▪ A visual aid for picturing work processes and shows how inputs and tasks are linked and
highlights the steps required to consistently produce a desired output

▪ A tool allowing us to document, analyze, improve, streamline and redesign business

processes to realize organizational efficiencies

❑ A process map encourages new thinking about how work is done, where it is done, who
performs it, what problems frequently occur and how best to solve them.

❑ A common analogy relates to road maps: you can’t plot a route to get to where you want to
go…unless you know where you are.

Non-Business Use
6
 What is a process map ? (2/2)

❑ Process maps represent what is actually happening, not what you think is happening

❑ Process Maps are used to;

▪ Identify areas for focus of improvement efforts

▪ Identify and eliminate non-value added steps

▪ Combine operations

▪ Assist root cause analysis (RCA)

▪ Baseline for failure mode and effect analysis (FMEA)

▪ Identify potential controllable parameters for design of experiments (DOE)

▪ Determine needed data collection points

Non-Business Use
7
 By mapping processes we can identify
important characteristics
1. Process inputs (X’s)

2. Supplier requirements

3. Process outputs (Y’s)

4. Actual customer needs

5. All value-added and non-value added process tasks and steps

6. Data collection points

▪ Cycle times, Defects, Inventory levels, Cost of poor quality, etc.

7. Decision points

8. Problems that have immediate fixes

9. Process control needs

Non-Business Use
8
 Why does a business need a process maps?

❑ Pictures guide better than words

❑ Process maps facilitate improvements in the process, since it becomes easy to pinpoint the
specific areas that need changes

❑ Decision making becomes fast as it deals with the ‘show me’ aspect and not the ‘tell me’
aspect of the process and the problem areas

❑ The improvements made in the process can easily be tracked using process maps since it
becomes possible to audit

❑ Visual illustration for training would be much more effective than any oral tools

Non-Business Use
9
 Process maps uses in DMAIC

❑ In the Define phase, they are used to record top-level processes

❑ In the Measure phase, they are used to record the lower process levels and
reveal differences in the perceptions of stakeholders

❑ In the Analyze phase, process maps are used to investigate the sources of
variation or excessive cycle time

❑ In the Control phase process maps may be used to record process

adjustments

Non-Business Use
10
 Standard symbols for Process Mapping

RECTANGLE indicates a step. Statements should begin with a verb

PARALLELAGRAM shows input or output data

DIAMOND indicates a decision point

Could be two paths from a decision point (e.g., No and Yes) or multiple paths

ELLIPSE shows the start and end of the process

ARROW shows the connection and direction of flow

CIRCLE WITH A LETTER OR NUMBER INSIDE symbolizes the continuation 1

of a flowchart to another page

There are even more notations for Process Mapping, but these are just the most used
Non-Business Use
11
 Types of process maps (1/2)

Process Flowchart – Sometimes called Linear Flow

Calls
Customer for Take Make Cook Pizza Box Deliver Customer
Hungry Order Order Pizza Pizza Correct Pizza Pizza Eats

This diagram shows the process steps in a sequential flow, generally ordered from an upper left
corner of the map towards the right side

Non-Business Use
12
 Types of process maps (2/2)

Functional or Deployment Flow or Swim Lane

Swim Lane map shows who or which department is responsible for the steps in a process. A
timeline can be added to show how long it takes each group to perform their work

Non-Business Use
13
 There are usually three views of a process

1 2 3
What you THINK it is.. What it ACTUALLY is.. What it SHOULD be..

As a result of your project, you will either create the “what it should be” or at least
Non-Business Use
you will be well on your way to get there 14
 Steps to do a Process Map (1/2)

Step 1: Select the process

▪ Name of the process. This may impacts the scope of the process map

Step 2: Identify the scope of the process steps

▪ Identify the start and end points of the process of interest (define clearly the boundaries)

Step 3: Plan and schedule resources

Step 4: Select Mapping techniques

Step 5: Conducting interviews

Step 6: Map the As-Is process

▪ Document the top-level process steps (consecutive steps in the process as it is today)

▪ Identify the inputs and outputs

o What are the results of each process step (Y’s) & the impacts of each input (x’s) on Y?
Non-Business Use
15
 Steps to do a Process Map (2/2)

Step 6: Map the As-Is process (cont..)

▪ Characterise/Classify the inputs. Inputs can be classified as one of three types

o Controllable (C) : Things you can adjust or control during the process such as Speeds, feeds,
temperatures, pressures….

o Standard Operating Procedures (S): Things you always do (in procedures or common-sense
things) such as Cleaning, safety, etc.

o Noise (N): Things you cannot control or do not want to control (too expensive or difficult) such
as Ambient temperature, humidity, operator...

o Critical Inputs (X): X‟s that have been statistically proven to impact the process Y(s)

Step 7: Analyze, evaluate and Sign-off

Non-Business Use
16
 Cooking Pasta Process Example (1/6)

Step 1: “Pasta” High Level SIPOC Map

Input Cooking
Suppliers Output Customers
Pasta

Restaurant Owner Raw Pasta Cooked Patrons

Grocery Store Pan Pasta Owner
H2O Cook
Cup
Stove
Spoon
Salt
Oil

Non-Business Use
17
 Cooking Pasta Process Example (2/6)

Step 2: Identify All Process Steps with designating steps as value-added or non-value-added

Rework/ Recook

0.5 MINUTE 6 MINUTES 0.25 MINUTE 9 MINUTES No 0.01 DPU

Yes
Add water,
Input salt & oil
Boil Water Add Pasta Cook Pasta Done Output

Raw Pasta Cooked Pasta

Pan
H2O
Cup
Stove
Spoon
Salt
Oil

Add cycle time, defect/yield data and ALL rework

Non-Business Use
18
 Cooking Pasta Process Example (3/6)

Step 3: Determine Outputs For Each Step

Rework/ Recook

0.5 MINUTE 6 MINUTES 0.25 MINUTE 9 MINUTES No 0.01 DPU

Yes
Add water,
Input salt & oil
Boil Water Add Pasta Cook Pasta Done Output

Raw Pasta Cooked Pasta

Pan
H2O
Boiling Water Properly Cooked
Cup Water Remaining
Stove Steam
Spoon
Salt
Oil

Non-Business Use
19
 Cooking Pasta Process Example (4/6)

Step 4a : List Inputs For Each Step

Rework/ Recook

0.5 MINUTE 6 MINUTES 0.25 MINUTE 9 MINUTES No 0.01 DPU

Yes
Add water,
Input salt & oil
Boil Water Add Pasta Cook Pasta Done Output

Raw Pasta Cooked Pasta

Pan Boiling Water
Amt. of Salt Water Remaining
H2O Amt. of Oil Steam Properly Cooked
Cup Lid or Not
Ambient Temp
Stove Time Time
Spoon Minerals in H2O Lid or Not
Pan Material Type of Pasta
Salt Pan Size Amount of Pasta
Oil Pan Condition Ambient Temp
Boiling Instructions Cooking instructions

Non-Business Use
20
 Cooking Pasta Process Example (5/6)

Step 4b : Classify Inputs Variables

Rework/ Recook

0.5 MINUTE 6 MINUTES 0.25 MINUTE 9 MINUTES No 0.01 DPU

Yes
Add water,
Input salt & oil
Boil Water Add Pasta Cook Pasta Done Output

Raw Pasta Cooked Pasta

Pan Boiling Water
Amt. of Salt C Water Remaining
H2O Amt. of Oil C Steam Properly Cooked
Cup Lid or Not C
Ambient Temp N
Stove Time N Time C C – Controllable
Spoon Minerals in H2O N Lid or Not C S - SOP
Pan Material C Type of Pasta C
Salt Pan Size C N - Noise
Amount of Pasta C
Oil Pan Condition C
Ambient Temp N
Boiling Instructions S
Cooking instructions S

Non-Business Use
21
 Cooking Pasta Process Example (6/6)

Step 5 : Add Operating Specifications

Rework/ Recook

0.5 MINUTE 6 MINUTES 0.25 MINUTE 9 MINUTES No 0.01 DPU

Yes
Add water,
Input salt & oil
Boil Water Add Pasta Cook Pasta Done Output

Raw Pasta Cooked Pasta

Pan Boiling Water
Amt. of Salt C Water Remaining
H2O Amt. of Oil C Steam Properly Cooked
Cup Lid or Not C
Ambient Temp N
Stove Time N Time C C – Controllable
Lid or Not C
Spoon Minerals in H2O N
S - SOP
Pan Material C Type of Pasta C
Salt Pan Size C Amount of Pasta C N - Noise
Oil Pan Condition C Ambient Temp N
Boiling Instructions S
Cooking instructions S

Controllable Input Variable Target Upper spec Lower Spec

Time (minutes) 9 10 8
Amount of Pasta (ounces) 16 18 14
Non-Business Use
22
 Tips for Quality Process Map

❑ Involve stakeholders

❑ Seek Multiple opinions

❑ Always “Walk the process, repeatedly”

❑ Ask lots of questions

❑ Maintain process maps & update frequently with dates as you capture additional information

❑ Data is the driver when determining whether more detail is necessary. Choosing wrong level costs
time and effort

❑ Never assume or take for granted

❑ Always verify through discussion and data

Non-Business Use
23
 Value stream mapping (VSM)

A Value Stream includes all elements (both value added and non-value added) that occur to a
given product from its inception through delivery to the customer

The Value Stream Map is a very powerful technique to understand the velocity of process
transactions, queue levels and value added ratios in both manufacturing and non-
manufacturing processes

VSM is a tool used to support the implementation of lean strategies

Non-Business Use
24
 Value stream mapping (VSM) – Notations

Timeline of Value-Added Activities

Vs. Non-Value-Added Activities

Non-Business Use
25
 VSM connects the flows of information &
material through the organization

C/T: Cycle Time

Non-Business Use
26
 VSM Example 1: Ordering Lunch

Food Supplier Customers

Food Suppliers Enter Order Place Order
Replenish Foo d (Lunch Patrons)

6 6
Food Delivery 2x/week: L/T 4 min
L/T 3 min
Monday & Friday mornings 100 orders in 2 hrs
P/T 1 min P/T 1 min
at 9:00am Food Ticket Food Sales $1,500
%C&A 70% %C&A 95%
Produced Lead Time 15-60 min
C/O C/O Takt Time: 72 seconds

Available Time (2 hrs or 7200 Seconds)

Takt Time =
Customer Requirement (100 Orders)

Prepare & Serve the Food

Gather Ingredients Cook Order Package the Order
Assemble Order Order

1 1 2 1 6
L/T 7min L/T 7 min L/T 8 min L/T 10 min L/T 4 min
P/T 2min P/T 5 min P/T 8 min P/T 3 min P/T 1 min TOTAL
%C&A 80% %C&A 92% %C&A 98% %C&A 93% %C&A 92%
C/O C/O 3 min C/O 3 min C/O C/O

Non-Business Use
%C&A: % complete and accurate, C/O : Change over , L/T: Lead time , P/T: Processing Time, : # of People in the process 27
 VSM Example 2: Window Manufacturer

Non-Business Use
28
 VA/ NVA/ ENVA Analysis

Value-Add Total Activity Non Value Add

An activity that transforms

the input, thereby bringing
it closer to the form Non-Value Add Essential Non-Value Add
required by the customer (NVA) (ENVA)
Activities that look like
Maximise Any activity that does
waste but are necessary so
not add value & can be
the subsequent can add
completely avoided
value with minimum waste
Waiting Reviews
Defects/ Rework On Job Training

Eliminate Reduce

Non-Business Use
29
 Process mapping Vs Value stream mapping (VSM)

Process Mapping Value Stream Mapping

Concentrate on a single process Considers the whole value stream
Identifies NVA within a process Identifies NVA between processes
Improvements to processes are Improvements to system are
usually small and very easy to usually significant and difficult to
implement achieve
Enable short-term tactical Enable long-term strategic
planning planning

Non-Business Use
30
Cause & Effect/ Fishbone Diagram

31
 Cause-and-Effect (C-E) Diagram

❑ A Cause-and-Effect Diagram (aka Ishikawa, Fishbone) is a picture diagram showing

possible causes (process inputs) for a given effect (process outputs).

❑ Fishbones are drawn right to left, with each large bone of the fish drawn out to
include smaller bones that add more detail.

Non-Business Use
32
 When to Use Cause and Effect Diagram

❑ When the problem has multiple possible causes

❑ To identify the possible root causes for an effect

❑ Identify and sort interactions among the factors on an effect

❑ Finally, to initiate appropriate corrective action for existing problems

Non-Business Use
33
 Why use Cause-and-Effect Diagram

❑ It is a basic step for the purpose of studying a problem and determining the root cause

❑ It allows you to study the probable causes of why a process is beginning to have a problem

❑ It helps you identify areas of data collection for further study

Non-Business Use
34
 How to develop a Cause-and-Effect Diagram

Step 1: Identify the problem

Step 2: Determine Effect or Problem
Step 3: Identify major causes contributing to the effect or a problem
Step 4: Identify sub causes
Step 5: Analyze the diagram

Non-Business Use
35
 Cause and Effect Diagram Example 1

Problem: Diameter is out of specification

The team identified the wrong caliper and wrong procedures are the probable causes

Non-Business Use
36
 Cause and Effect Diagram Example 2

Problem: Chemical is not pure

Measurement Manpower Materials

Training on method (S) Raw Materials (C)

Incoming QC (S) C – Controllable
Measurement Insufficient staff (C)
S - SOP
Method (S) Skill Level (S) Multiple Vendors (C) N - Noise
Measurement
Capability (C) Adherence to procedure (S) Specifications (C)

Work order variability (N)

Chemical
Startup inspection (S) Room Humidity (N) Column Capability (C) Purity
Handling (S) RM Supply in Market (N) Nozzle type (C)
Purification Method (S) Shipping Methods (C) Temp controller (C)

Data collection(S)

Methods Mother Nature Equipment

Non-Business Use
37
 Benefits of Cause-and-Effect Diagram

❑ Helps to identify all the probable causes of an effect to identify root causes further

❑ It graphically displays the possible causes related to a problem to discover the root causes

❑ Fosters teamwork and inspires the team to brainstorm until the elimination of the root
cause

❑ Provide users the ability to understand of the factors causing the problem

Non-Business Use
38
Cause & Effect/ X-Y Matrix

39
 What Is a Cause-and-Effect Matrix?

❑ It reveals the correlation between process input variables to the outputs of the customer
during the root cause analysis

❑ A cause-and-effect matrix can be used to examine and document relationships between

input and output variables

❑ Cause and effect matrix also called X-Y diagram, Prioritization Matrix, and Correlation
Matrix

It objectively evaluates the team’s subjective opinions about the Key Process
Input Variables

Non-Business Use
40
 How to Use a Cause and Effect Matrix?
(1/2)
Step 1: Identify the customer requirements, or in other words, understand the voice of customer.
Place those priorities in at the top of the X-Y diagram
Step 2: Assign priority factor for each of the customer outputs. Generally, use 1-10 scale where 1
being the lowest priority and 10 being the highest priority to the customer
Step 3: List all possible key input variables or the improvement factors of the process in each row,
those are the Xs in the X-Y diagram
Step 4: Assess the relationship between key input variables to the customer outputs and rank each
input variables accordingly. Recommended to use geometric progression scale (0,1,3 and 9) where
0 being no impact, 1 – low impact, 3 – medium impact and 9 – Input has strong impact or
correlation on output
Step 5: Cross multiply the customer output priority numbers with correlation rankings and sum
each row at extreme right column
Step 6: Finally, determine the rank based on the highest sum and highlight the critical few
variables. This will help to identify the improvement areas

Non-Business Use
41
 How to Use a Cause and Effect Matrix?
(2/2)

Non-Business Use
42
 Example of an X-Y Diagram

XXX coffee shop franchise located in the Tokyo downtown area has reported falling sales for the
past 6 months. A Six Sigma team conducts a root cause analysis, and also they want to see the
key process inputs that are impacting the process.

Non-Business Use
43
 Benefits of Cause-and-Effect Matrix

❑ It helps to include customer inputs for decision making

❑ Visually depict the correlation between key input variable to the customer outputs

❑ Priority ranking method helps to take the decision based on score rather than
individual opinions

❑ Data collection cost may be reduced by ignoring non-key process inputs

❑ Helps to list all the input variables required for the process

Cause and Effect Matrix

Cause & Effect

Process Map
Diagram
Voice of
Customer

Non-Business Use
44
Failure Mode Effect Analysis (FMEA)

45
 What is FMEA?
Failure Modes Effect Analysis (FMEA) is a structured approach to:

❑ Predict failures and prevent their occurrence in manufacturing and other

functional areas which generate defects

❑ Identify the ways in which a process can fail to meet critical customer
requirements (Y)

❑ Estimate the Severity, Occurrence and Detection (SOD) of defects

❑ Evaluate the current Control Plan for preventing these failures from occurring and
escaping to the customer

❑ Prioritize the actions that should be taken to improve and control the process using
a Risk Priority Number (RPN)

Non-Business Use
46
 Types of FMEAs (1/2)
❑ System FMEA: Performed on a product or service product at the early
concept/design level when various modules all tie together. All the module level
FMEA’s tie together to form a system. As you go lower into a system more failure
modes are considered
▪ Example: Electrical system of a car, consists of the following modules: battery, wiring
harness, lighting control module and alternator/regulator

▪ System FMEA focuses on potential failure modes associated with the modules of a
system caused by design

❑ Design DFMEA: Performed early in the design phase to analyze product fail modes
before they are released to production. The severity rating of a fail mode MUST be
carried into the Process PFMEA

Non-Business Use
47
 Types of FMEAs (2/2)
❑ Process PFMEA: Performed in the early quality planning phase of manufacturing to
analyze fail modes in manufacturing and transactional processes that may escape
to the customer. The failure modes and the potential sources of defects are rated
and corrective action taken based on a Pareto analysis ranking

❑ Equipment FMEA: used to analyze failure modes in the equipment used in a process
to detect or make the part
▪ Example: Test Equipment fail modes to detect open and short circuits

Non-Business Use
48
 Purpose of using FMEA

❑ Improve the quality, reliability and safety of products

❑ Increase customer satisfaction

❑ Reduce product development time and cost

❑ Document and track actions taken to reduce risk and improve the process

❑ Focus on continuous problem prevention not problem solving

Non-Business Use
49
 When To Use FMEA

Detailed Process Cause and Effects

Map Matrix

Initial Control Plans FMEA

Detailed Process Mapping is the foundation of many other

Breakthrough Strategy tools
Non-Business Use
50
 Two Approaches To The FMEA

❑ Two major approaches

▪ Starting with the Cause & Effect Matrix

▪ Starting with FMEA directly from the Process Map

❑ We will explain the approach using the C&E matrix

❑ Both approaches are very similar

Non-Business Use
51
 Starting With C&E Matrix

Identify various X’s

Non-Business Use
Focus on inputs that highly impact a large number of outputs 52
 The FMEA Form
S O D R
Process Actions
Potential Failure Mode Potential Failure Effects E Potential Causes C Current Controls E P
Step/Input Recommended
V C T N

What is the 0 How 0

How can 0
How 0

Effect on the Bad? this be well?

0 0 0 0
Outputs? found?
What
can go 0 0 0 0

wrong What are What can

with the
0
the
0 How 0 0
be done?
Process Causes? Often?
0 0 0 0

What is step?
the
Process
Step

Methodical approach to understanding the process in the absence of

data – depersonalizes the performance issues

Non-Business Use
53
 From Cause To Effect…

Failure Mode
Cause Effect
(Defect)
Material or Key Input (x) External customer
process input or downstream
process step

Controls

Non-Business Use
54
 Risk Priority Numbers, RPN

❑ The risk priority number (RPN) is the product of the rankings for Severity

(SEV), Probability of Occurrence (OCC), Difficulty to Detect (DET)

❑ High RPN’s are flags to take effort to reduce the calculated risk

❑ Regardless of RPN, high severity scores should be given special attention

RPN = SEV x OCC x DET

Effects Causes Controls

Non-Business Use
55
 Rating Definitions

Rating Severity Occurrence Detection

High 10
Hazardous without Very high and Cannot detect or
warning almost inevitable detection with very
low probability
Loss of primary High repeated Remote or low
function failures chance of detection

Loss of secondary Moderate failures Low detection

function probability

Minor defect Occasional failures Moderate detection

probability

No effect Failure unlikely Almost certain

detection
Low 1
Non-Business Use
56
 Detection Scores At Various Levels
Of The Process
Where in the process do we want to implement Controls?

Det = 1 Det = 3 Det = 7 Det = 10

Prevention Detection Detection Detection

Failure Mode
Cause Effect
(Defect)
Material or Process Step External customer
process input or downstream
process step

Non-Business Use
57
 How to Complete A FMEA
❑ For each Process Input

➔ determine the ways in which the Process Step can go wrong (Failure Modes)

➔ For each Failure Mode associated with the inputs, determine Effects

➔ Identify potential Causes of each Failure Mode

➔ List the Current Controls for each Cause

❑ Define Severity, Occurrence and Detection ratings & assign it to each Cause

❑ Calculate RPN, and determine Recommended Actions to reduce High RPN’s

❑ Take appropriate Actions and Document

❑ Recalculate RPN’s
Non-Business Use
58
 Example Of A Completed FMEA
(Part 1)
Shown below is an example for baking a Pizza in which the focus is
the Process Step of Bakin g

Process Step Key Process Input Failure Effects Causes Current RPN

Detection
Severit

Occurrenc
Modes - Controls
What can go

y
wrong?

e
Broken
Bake Pizza Temperature > 500°F Burned Pizza 9 2 None 10 180
Thermcouple
Top Vent too far
9 5 Visual Inspection 6 270
forward
Broken Gas
Gas Burned Pizza 9 Installed wrong 2 Calibration Test 2 36
Valve
Opeartor not
Time > 12 mins. Burned Pizza 9 8 Training 8 576
watching time
Clock not
9 2 None 10 180
calibrated

How do we prioritize what to address first?

Non-Business Use
59
 RPN – Where to focus efforts ?

Would You
Severity Occurrence Detection RPN Recom mend Action?
1 1 1 Ideal situation No Action
1 1 10 Assured mastery No Action
1 10 1 Failure does not reach user No Action
1 10 10 Failure reaches user Yes
10 1 1 Detectable, costly Consider DFMEA
10 1 10 Undetectable, costly Yes
10 10 1 Freq. Failures w/ Major impact Yes
10 10 10 Big trouble! Yes

Not all RPN’s are created equal

- If a Severity number is 9 or 10 we must ensure that our
detection is a 1 or 2
- Consider focusing your initial efforts on those RPN’s with
high Severity numbers
Non-Business Use
60
 Example Of A Completed FMEA
(Part 2)
❑ These recommended actions can be your priority list for the team.

❑ The FMEA needs to be a living document amended when process changes are made

❑ When the project is complete the R

P
Actions
Resp. Ac ti ons T a k e n
S
E
O
C
D
E
R
P
Recommended
N V C T N

Process Owner is responsible for W h a t are t h e actions

for reducing the
W hose
Re s p o n s i b l e for
W h a t are the
c o m p l e t e d actions
o c c u rra n c e of the t h e re c o mme n d e d t a k e n with the
Ca u s e , or improving a c t i o n ? B e sure recalculated

updating the FMEA

detection? S h o u l d to i nc l ude RPN?
h a v e a c t i o n s o n l y on completion
R P N ' s tha t r e q u i r e it. m o n t h / ye a r
P l a c e a light in line with Jim - 3/10/1
180 t h e t h e rmo c o u p l e t o go 9 2 1 18
off if circuit shorts.

270 0

36 0

Timer w t h b u z z e r will S u e - 3/1/01 Timer w t h buzzer

b e put directly o n the p l a c e d directly on
576 9 2 2 36
oven& opeators will be t h e oven &
trained oerators trained

Non-Business Use
61
 Example Of A Completed FMEA
(Part 2)
❑ These recommended actions can be your priority list for the team.

❑ The FMEA needs to be a living document amended when process changes are made

❑ When the project is complete the R

P
Actions
Resp. Ac ti ons T a k e n
S
E
O
C
D
E
R
P
Recommended
N V C T N

Process Owner is responsible for W h a t are t h e actions

for reducing the
W hose
Re s p o n s i b l e for
W h a t are the
c o m p l e t e d actions
o c c u rra n c e of the t h e re c o mme n d e d t a k e n with the
Ca u s e , or improving a c t i o n ? B e sure recalculated

updating the FMEA

detection? S h o u l d to i nc l ude RPN?
h a v e a c t i o n s o n l y on completion
R P N ' s tha t r e q u i r e it. m o n t h / ye a r
P l a c e a light in line with Jim - 3/10/1
180 t h e t h e rmo c o u p l e t o go 9 2 1 18
off if circuit shorts.

270 0

36 0

Timer w t h b u z z e r will S u e - 3/1/01 Timer w t h buzzer

b e put directly o n the p l a c e d directly on
576 9 2 2 36
oven& opeators will be t h e oven &
trained oerators trained

Non-Business Use
62
 Relationship To Other Tools

❑ Metrics Identification
❑ Process Map
❑ C & E Diagrams
❑ C & E Matrix
Design FMEA (Product or Business Process)
Process FMEA (Product or Business Process)

❑ Multivariate analysis
❑ Identification of vital few inputs
❑ Design reviews
❑ Quality Control Plan

Non-Business Use
63
 (FMEA)-Example of Calculating Risk Priority
Number (RPN)

RPN = Severity x Occurrence x Detection

Severity Scale
Rating Criteria: A Failure Could
10 Injure A Customer Or Employee
9 Be Illegal
8 Render Product Or Service Unfit For Use
7 Cause Extreme Customer Dissatisfaction Occurrence Scale
6 Result In Partial Malfunction Time Period
Rating
5 Cause A Loss Of Performance Which 10 More Than Once Per Day  30%
Is Likely To Result In A Complaint
4 Cause Minor Performance Loss
9 Once Every 3-4 Days  30% Detection Scale
8 Once Per Week  5%
3 Cause A Minor Nuisance But Can Be Rating Definition
Overcome With No Performance Loss 7 Once Per Month  1%
10 Defect Caused By Failure Is Not Detectable
2 Be Unnoticed And Have Only Minor 6 Once Every 3 Months  .03%
Effect On Performance 9 Occasional Units Are Checked For Defect
5 Once Every 6 Months  1 Per 10,000
1 Be Unnoticed And Not Affect The 8 Units Are Systematically Sampled And Inspected
4 Once Per Year  6 Per 100,000
Performance 7 All Units Are Manually Inspected
3 Once Every 1-3 Years  6 Per Million 6 Manual Inspection With Mistake-Proofing Modifications
2 Once Every 3-6 Years  3 Per 10 Million 5 Process Is Monitored (SPC) And Manually Inspected
1 Once Every 6-100 Years  2 Per Billion 4 SPC Is Used With An Immediate Reaction To Out Of
Control Conditions
3 SPC As Above With 100% Inspection Surrounding Out
Of Control Conditions
2 All Units Are Automatically Inspected
1 Defect Is Obvious And Can Be Kept From Affecting
The Customer

Non-Business Use
64
Data & Basic Statistics

65
 What are data types ? (1/2)

Data is a set of values that can be either qualitative or quantitative. It may be a numbers, a
measurement, an observation, or even just a description Qualitative data Quantitative data

Data can be classified into two category

Ordinal Nominal Discrete Continuous

❑ Qualitative data: subjective evaluation of a characteristic which is not directly measurable

or difference between value can not be determined. Such as color, Pass-Fail (also termed
Attribute Data)

❑ Quantitative data consists of measures that take numerical values for which descriptions
such as means, and standard deviations are meaningful.
▪ Discrete data are countable data, for example, the number of defective items

▪ Continuous data, when the parameter (variables) are measurable, are expressed on a continuous
Non-Business Use scale. For example, measuring the length of a part. 66
 What are data types ? (2/2)

Non-Business Use
67
 Understanding the nature of data can
affect the types of statistical tests possible
❑ Nominal : Not much information (just classes with no particular order)

▪ Example : group 1 if you are from city A, and group 2 if you are from city B

Only non-parametric statistics apply ( = , ≠ )

❑ Ordinal: Ranking information (higher or lower numbers have some meaning but not equal
distance)

▪ Example : 1 best company, 100 worst company. 5 very satisfied , 1 very dissatisfied

Only non-parametric statistics apply ( = , ≠ , < , > )

Non-Business Use
68
 Understanding the nature of data can
affect the types of statistical tests possible
❑ Interval : Similar to ordinal, but ensure equal distance between measurement, but not the
multiples

▪ Example : 49 degree Celsius is hotter than 32, the same way 67 is hotter than 50

▪ Example : 30 degree Celsius is NOT 3 times hotter than 10 degree Celsius

Can use parametric statistics ( = , ≠ , < , > , + , - )

❑ Ratio: Has meaning of true zero as origin (lack of measurement). Equal and multiple
properties apply

▪ Example : $4.00 is 2 dollars more than $2.00, the same way $6.00 is more than $4.00

▪ Example : $4.00 is double the purchasing power of $2.00

Can use all statistics ( = , ≠ , < , > , + , - , × , ÷ ) 69

Non-Business Use
 Sample vs Population

Population: a collection of all possible individuals, objects, or measurements of interest

Sample: a portion, or part, of the population of interest

Popu lation Parameters: Arith me tic

de scription s of a popu lation µ ,  , P,  2 , N
Population
Sample Statistics: Arithmetic descriptions of a sample
X-bar , s, p, s2, n
Sample
Sample Sample

It is important to recognize the difference between a sample and a population as we

typically deal with a sample of the what the potential population to make an inference

Non-Business Use
70
 Descriptive vs Inferential Statistics

Population Sample

Inferential: estimate,
Descriptive : summarize forecast or judge the
the population (e.g. population (e.g. average,
total, minimum, maximum,
average, total, minimum,
variance)
maximum, variance)

We will review Descriptive Statistics in this module, and we will review Inferential
Statistics in the Analyze module
Non-Business Use
71
 Descriptive Statistics

❑ Descriptive statistics presents data in helpful mediums like a summary that uses numbers
and graphs

❑ Descriptive statistics is used to describe the traits of the sample or population

❑ Data is summarized tabulated, organized, and presented in the forms of charts and graphs
to summarize the data under consideration for the whole population

Non-Business Use
72
 Parameters and frequency distribution

❑ Parameter (Statistics): a measure to describe population data (e.g. mean or average is a

parameter, variance is parameter)
▪ If we measure a sample, we get a sample statistic, which is used to estimate the parameter

❑ Frequency Distribution: Summarize the data by assigning it to groups or intervals (also

known as classes in statistical language)
▪ Define mutually exclusive and non-overlapping interval

▪ Assign each item to one and only one interval

▪ Display a table of observations count (also known as absolute frequency)

❑ Relative Frequency: the percentage of absolute frequency of the total count

❑ Cumulative Frequency: Summing absolute or relative frequencies from lowest up to the item
Students are expected to understand the probability distribution for both discrete and
Non-Business Use
continuous variables 73
 Parameters and frequency distribution

‫فئات العمر‬ Saudi ‫السعودي‬ Total Saudi Population

Age group MALE ‫ ذكور‬FEMALE ‫ اناث‬Total ‫ جملة‬Relative Frequency Cumulative Frequency
0-4 1,103,151 1,085,223 2,188,374 10.1% 10.1%
05-9 1,165,051 1,097,326 2,262,377 10.4% 20.5%
10-14 1,043,917 1,032,094 2,076,011 9.6% 30.1%
15 – 19 951,818 894,816 1,846,634 8.5% 38.6%
20 – 24 1,074,376 998,140 2,072,516 9.6% 48.2%
25 – 29 1,041,974 1,026,056 2,068,030 9.5% 57.7%
30 – 34 936,104 941,473 1,877,577 8.7% 66.3%
35 – 39 856,123 821,573 1,677,696 7.7% 74.1%
40 – 44 684,143 673,694 1,357,837 6.3% 80.3%
45 – 49 593,073 560,761 1,153,834 5.3% 85.7%
50 – 54 474,238 450,961 925,199 4.3% 89.9%
55 – 59 377,003 350,308 727,311 3.4% 93.3%
60 – 64 258,833 245,608 504,441 2.3% 95.6%
65+ 468,202 484,609 952,811 4.4% 100.0%
‫جملة‬
11,028,006 10,662,642 21,690,648 100.0% 100.0%
Total

Source: General Authority for Statistics (modified) - 2021

Non-Business Use
74
 Some measures used in Descriptive
Statistics
❑ Measures of Location (central tendency)

▪ Mean

▪ Median

▪ Mode

❑ Measures of Variation (dispersion)

▪ Range

▪ Interquartile Range

▪ Standard deviation

▪ Variance

Students should be able to understand these measures and apply them

Non-Business Use
75
 Statistical Notation

Summation An individual value, an observation

A particular (1st) individual value

The Standard Deviation of sample data
The Standard Deviation of population For each, all, individual values
data
The variance of sample data The Mean, average of sample data

The variance of population data The grand Mean, grand average

The range of data
The Mean of population data
The average range of data
A proportion of sample data
Multi-purpose notation, i.e. # of
subgroups, # of classes
A proportion of population data
The absolute value of some term
Sample size
Greater than, less than
Greater than or equal to, less than or Population size
equal to
Non-Business Use
76
 Ranking Measures

❑ A number that indicates a position (location or rank) in ordered set of data

▪ Most used measure is Quantile, which refers to a value at or below which the proportion of the
data lies in a distribution

▪ Quartiles – data is divided into quarters (4 parts)

▪ Quintile – data is divided into fifths (5 parts)

▪ Decile – data is divided into tenths (10 parts)

▪ Percentiles – data is divided into hundredths (100 parts or percents)

❑ Formula for finding quantile measures:

𝒚 n : number of data points sorting from lowest to highest
𝑳𝒚 = (𝒏 + 𝟏) y : target percentile
𝟏𝟎𝟎
L : location of the data point

Students should be able to understand these measures and apply them

Non-Business Use
77
Normal distribution and Normality Test

78
 Normal Distribution

❑ The Normal distribution is used to analyze data when there is an equally likely chance of
being above or below the mean for continuous data

❑ Convert any raw score to a Z-score using the formula:

❑ Refer to a set of Standard Normal Tables to find the proportion between μ and x

❑ Normal distributions have key characteristics that are easy to spot in graphs:
▪ The mean, median and mode are exactly the same

▪ The distribution is symmetric about the mean—half the values fall below the mean and half above
the mean.

▪ The distribution can be described by two values: the mean and the standard deviation.

❑ Many statistical inference procedures are based on Normal distributions

Non-Business Use
79
 The 68-95-99.7 Rule (Empirical Rule)

In the Normal distribution with mean µ and

standard deviation σ:

❑ Approximately 68% of the observations fall

within 1 σ of µ

❑ Approximately 95% of the observations fall

within 2σ of µ

❑ Approximately 99.7% of the observations

fall within 3σ of µ

Non-Business Use
80
 The 68-95-99.7 Rule (Empirical Rule)

No matter what the shape of your distribution is, as you travel 3 Standard Deviations from the
Mean, the probability of occurrence beyond that point begins to converge to a very low number

Non-Business Use
81
 Normal Distribution uses

❑ When data is grouped around the mean and there is an roughly equal probability of being
above or below the mean

❑ When the histogram (Cumulative Frequency Distribution) fits a bell curve

❑ When we would like to test population parameters using sample data (inferential statistics)

❑ Normal distribution can be used to approximate many other distributions (e.g., Poisson and
Binomial distribution)

❑ Many statistical tests are designed for normally distributed populations

Non-Business Use
82
 Normality Test (1/2)

❑ While many processes in nature behave according to the Normal Distribution, many
processes in business, particularly in the areas of service and transactions, do not

❑ There are many types of distributions:

❑ There are many statistical tools that assume Normal Distribution properties in their
calculations

❑ So understanding just how “Normal” the data are will impact how we look at the data

Non-Business Use
83
 Normality Test (2/2)

❑ The shape of any Normal curve can be calculated based on the Normal Probability density
function

❑ Tests for Normality basically compare the shape of the calculated curve to the actual
distribution of your data points

❑ For the purposes of this course, we will focus on 2 ways to assess Normality:

▪ The Anderson-Darling test: this test yields a statistical assessment (called a goodness-
of-fit test) of Normality

▪ Normal probability test: this test produces a graph to visual demonstrate just how good
that fit is

Non-Business Use
84
 Goodness-of-Fit (The Anderson-Darling test)

The Anderson-Darling test uses an

empirical density function

❑ Measures the departure of the

actual data from the expected
Normal Distribution

❑ The Anderson-Darling test

assesses how closely actual
frequency at a given value
corresponds to the theoretical
frequency for a Normal
Distribution with the same Mean
and Standard Deviation. 85
Non-Business Use
 Goodness-of-Fit (The Anderson-Darling test)

❑ The Anderson-Darling test

also appears in this output

❑ If the P-value is greater

than .05, there is a
support for assuming
normal data

❑ The P-value concept will

be covered in the Analyze P-value = 0.921
Phase

Non-Business Use
86
 The Normal Probability Plot

❑ The graph shows the probability

density of your data plotted
against the expected density of a
Normal curve

❑ Notice that the y-axis (probability)

does not increase linearly

❑ Normal data will lie on a straight

line (the red line) in this analysis

❑ The graph shows you which values

tend to deviate from the Normal We use the same dataset that’s why the P-value and
the variable names are the same
curve

Non-Business Use
87
 Both the Histogram and the Normality Plot
look very “normal”

Due to a large sample size (500), the Anderson-Darling test is very sensitive and any
slight deviation from Normal will cause the P-value to be very low

This will be addressed in more details in the Analyze Phase

Non-Business Use
88
 Isolating Special Causes from Common
Causes
❑ Special Cause: Variation is caused by known factors that result in a non-random distribution
of output. Also referred to as “Assignable Cause”

❑ Common Cause: Variation caused by unknown factors resulting in a steady but random
distribution of output around the average of the data. It is the variation left over after
Special Cause variation has been removed and typically (not always) follows a Normal
Distribution

❑ If we know that the basic structure of the data should follow a Normal Distribution, but plots
from our data shows otherwise; we can suspect the data contain Special Causes

Non-Business Use
89
Graphical Analysis

90
 What is Graphical Analysis?

❑ Graphical analysis is one of the best ways to analyze the problems in Six Sigma projects, it is
an effective way to visualize data patterns and provides key insights into the data

❑ The purpose of graphical analysis :

o Identify potential relationships between variables

o Identify risk in meeting the critical needs of the

Customer, Business and People

o Provide insight into the nature of the X’s which

may or may not control Y

o Show the results of passive data collection

❑ Many graphical tools are available which can generate graphs quickly and easily

Graphical analysis is the starting point for any problem-solving method

Non-Business Use
91
 What is Graphical Analysis?

❑ Different graphs can reveal different characteristics of your data:

o Central tendency

o Dispersion

o The general shape for the distribution

❑ Conclusions drawn from graphs may require verification through advanced statistical
techniques such as significance testing and experimentation

❑ You need to pick the right graphical tool as there are a lot of different ways to plot your
data

❑ If one graph fails to reveal anything useful, try another one

Non-Business Use
92
 Box-and Whisker Plot

❑ Pictorial representation of continuous data

❑ Box plot shows the Max, Min, median, interquartile range Q1, Q3, and outlier

Non-Business Use
93
 Histogram

❑ Histogram is the graphical representation of a frequency distribution

❑ It is in the form of a rectangle with class interval as bases and the corresponding frequencies
as heights. Particularly, there is no gap between any two successive rectangles

Non-Business Use
94
 Pareto Chart

❑ Pareto Chart is also known as 80-

20 rule

❑ It is a combination of bar chart

and a line chart where the actual
data in descending order using
bar chart and cumulative data in
ascending order using a line
graph

Non-Business Use
95
 Bar Chart

❑ Bar Chart displays the frequency on one axis and the values of the categorical variable on
the other axis

❑ In a bar graph, bars of uniform widths are drawn with various heights where the height of
bars represents the frequency of the corresponding observation.

Non-Business Use
96
 Scatter Diagram

❑ Scatter diagram also known as

correlation chart or XY graph

❑ A scatter diagram plots the

relationship between two continuous
variables with independent variable
in the x-axis and the dependent
variable on the y-axis

Non-Business Use
97
 Line Charts

❑ One of the simplest form of charts

❑ Useful for showing trends in quality, cost or other process performance measures

❑ They represent the data by connecting the data points by straight lines to highlight trends in
the data

❑ A standard or a goal line may also be drawn to verify actual performance against identified
targets

❑ Time series plots, run charts, SPC charts and radar charts are all line charts

Non-Business Use
98
 Run Chart

❑ Run chart is also known as time series plot, it is a line graph of data plotted over time

❑ It helps to identify the pattern of the data across time

❑ The advantage of the run chart is to identify the special cause(s) in the process

Non-Business Use
99
 Time Series Plots

❑ Line charts used to evaluate behavior in data over a time interval

❑ They can be used to determine if a process is stable by visually spotting trends, patterns or
shift in the data

❑ It requires the data to be in the order which actually happened

❑ Different behaviors of the data can be observed such as:

o Upward and downward trends

o Shifts in the mean

o Changes in the amount of variation

o Patterns and cycles

o Anything not random

Non-Business Use
100
 Individual Value Plot

❑ Graphs that are useful to give an overall picture of the individual values that make up a data
set

❑ Often used for comparing distributions that have small number of data.

❑ They give an idea of the distribution shapes

and whether outliers are present

Example – An individual value plot showing the

responses of a particular marketing campaign
that uses multiple advertising methods.

Non-Business Use
101
 Multi-Vari Chart

❑ Variation in the data may come from multiple sources

❑ A Multi-Vari Chart is a graphical tool that allows to visually show where the major variation is
coming from

❑ Multiple variables are plotted together on a single chart

❑ Often used when studying the variation within:

o A subgroup

o Between subgroups

o Over time

Example – A multi-vari chart showing how the

type and composition affect the durability of a
carpet.
Non-Business Use
102
 Dot Plot

❑ A Dotplot is a graphical representation of data using dots plotted on a simple scale

❑ A form of frequency distribution

❑ It is suitable for displaying small to moderate data sets

❑ It is generally used when the data is discrete but can be used for continuous data

❑ It shows where the data are clustered and can help identify outliers

❑ Helpful in comparing distributions in terms of their shape, location and spread

Non-Business Use
103
 Graphical Analysis – Which Graph
Graph Selection:
Count / Attribute Data Continuous Data

Bar Charts Dotplots

Determine the
Comparing shape of the data
Pareto Charts between Histograms*
groups
Pie Charts Individual Value Plots Understanding
the differences
Determine the Boxplots* between groups
Dotplots shape of the data
Understanding Understanding
Time Series Plots Time Series Plots process stability
process stability
Study relationships
Scatter Plot
Study relationships Scatter Plot between variables
between variables
Understanding the
Multi-Vari Charts source of variation
* Larger amount of data

Non-Business Use
104
Measurement Systems Analysis (MSA)

105
 Measurement Systems Analysis (MSA)

❑ Measurement System Analysis (MSA) is

▪ An experimental and mathematical method of determining how much

the variation within the measurement process contributes to overall process
variability

▪ A system used to assess the quality of the measurement system

❑ To ensure a successful improvement project, you will need to complete a MSA to

validate the measurement system.

❑ A comprehensive MSA typically consists of six parts; Instrument Detection Limit,

Method Detection Limit, Accuracy, Linearity, Gage R&R and Long Term Stability.

Non-Business Use
106
 Measurement Systems Analysis (MSA)

❑ A poor measurement system (Metrology) can make

▪ Data meaningless and process improvement impossible

▪ Prevent assessment of process stability and capability, confound Root Cause

Analysis and hamper continuous improvement

❑ Measurement error has a direct impact on assessing the stability and capability of a
process

❑ MSA quantifies the effect of measurement error on the total variation

❑ The sources of this error can be visualized through Sources of Variations and the
elements of Measurement System (Metrology)

Non-Business Use
107
 Sources of Variation

Measured Value

Actual Measurement Measurement Error

Precision Accuracy

Repeatability Reproducibility
Stability Bias Linearity

Resolution
Non-Business Use
108
 Measurement System Elements

Equipment
Hardware
Software

Procedures Environment

Performance
Setup Pe
Calibration Frequency
Calibration Technique
Sample Preparation
Operator Procedure Cleanliness
Data Entry Humidity & Temp
Calculations Vibration
Lighting
Power Source

Non-Business Use
109
 Why MSA is important ?
❑ The main objective of Measurement system analysis is to gauge the validity of the
measuring system and try to minimize the process variation due to the
measurement system

❑ Data accuracy is one of the fundamental requirements in DMAIC process. It helps to

evaluate the effect of the measurement system on the collected data. If the
measurement system data is inaccurate, we will make the decision based on
incorrect data

❑ Monitoring and measuring resources is an important requirement in Quality

management systems such as ISO 9001, AS9100 etc.

Non-Business Use
110
 Accuracy and Precision are influenced by
❑ Accuracy is influenced by resolution, bias, linearity and stability

❑ Precision is influenced by repeatability and reproducibility of the measurement

system

❑ Repeatability is the variation which occurs when the same operator repeatedly
measures the same sample on the same instrument under the same conditions

❑ Reproducibility is the variation which occurs between two or more instruments or

operators measuring the same sample with the same measurement method in a
stable environment

Non-Business Use
111
 Accuracy and Precision in Metrology

❑ Operators are often skeptical of measurement systems, especially those that

provide them with false feedback causing them to “over-steer” their process

❑ This skepticism is well founded since many measurement systems are not capable
of accurately or precisely measuring the process

❑ Accuracy refers to the average of individual measurements compared with the

known, true value.

❑ Precision refers to the grouping of the individual measurements - the tighter the
grouping, the higher the precision.

❑ The bull’s eye targets in the next slide best illustrate the difference between
accuracy and precision

Non-Business Use
112
 The Center of the Target is the Objective

High Precision High Precision

High Accuracy Low Accuracy

Low Precision
High Accuracy

Low Precision
Low Accuracy
Non-Business Use
113
 Repeatability
❑ The variation in measurements taken by a single person or
instrument on the same item and under the same conditions

❑ Ideally, the results should be identical

❑ Example: Thermometer fluctuates from 72 to 78 degrees

every minute (not repeatable), but actual temperature is
not changing

Non-Business Use
114
 Example: Repeatability

REPEATABLE NOT REPEATABLE

0.0036 0.0046
0.0037 0.0057
0.0035 0.0033
0.0036 0.0039
0.0036 0.0050
0.0037 0.0030
0.0036 0.0036
0.0035 0.0055
GOOD BAD
SAME PART MEASURED OVER AND OVER AGAIN
Non-Business Use
115
 Reproducibility
❑ The variation induced when different operators,
instruments, or laboratories measure the same or replicate
items

❑ Ideally, the average results between instruments or people

should be identical

❑ Example: You think the thermometer shows 56 degrees C,

but your neighbor thinks it shows 58 degrees C

Non-Business Use
116
 Example: Reproducibility

REPRODUCIBLE NOT REPRODUCIBLE

PERSON #1 PERSON #2 PERSON #1 PERSON #2
0.0046 0.0048 0.0043 0.0034
0.0057 0.0050 0.0052 0.0022
0.0032 0.0034 0.0031 0.0021
0.0039 0.0051 0.0033 0.0023
0.0050 0.0037 0.0045 0.0035
0.0030 0.0032 0.0034 0.0024
0.0036 0.0046 0.0039 0.0029
0.0056 0.0044 0.0052 0.0047
AVERAGE AVERAGE AVERAGE AVERAGE
0.0043 0.0043 0.0041 0.0029

GOOD BAD

COMPARE AVERAGES OF SAME PART TO EACH OTHER 117

Non-Business Use
 Resolution
❑ Ability of the measurement system to detect and indicate
small changes – Ideally, the measurement can detect 10 or
more values within likely range

❑ Each increment should be 10% or less of the range of values

to be able to detect a change

❑ Example: Thermometer only displays in increments of 5

degrees (35, 40, 45, etc), unable to get readings between 35
and 40. Prefer to have readings like 35.4 degrees.

Non-Business Use
118
 Example: Resolution

Good RESOLUTION POOR RESOLUTION

0.0036 0.00

0.0037 0.01
0.0035 0.00
0.0036 0.00
0.0036 0.01
0.0037 0.00
0.0036 0.00
0.0035 0.01

SAME PART MEASURED OVER AND OVER AGAIN 119

Non-Business Use
 Bias
❑ How well your measurements compare to a reference,
standard or known value

❑ Ideally, no difference between the measurement and the

reference value

❑ Calibration is often performed to remove bias on a device or

equipment

o Only addresses one source of variation

❑ Example: Thermometer is consistently 2 degrees higher than

actual temperature

Non-Business Use
120
 Example: Bias

BIAS NO BIAS
DEVICE DEVICE
0.0046 0.0043
0.0057 0.0052
0.0038 0.0031
0.0039 0.0033
0.0050 0.0045
0.0042 0.0034
0.0036 0.0039
0.0055 0.0052
AVERAGE STANDARD AVERAGE STANDARD
0.0045 0.0041 0.0041 0.0041

THICKNESS OF PHONE IS KNOWN (REFERENCE) = 0.0041

Non-Business Use
121
 Stability
❑ The change in bias over time (drift)

▪ Ideally, there should be no change in bias over time

▪ Stability issues may increase or decrease the values over

time

❑ Control charts are commonly used to track the stability of a

measurement system over time

❑ Example: Thermometer performs well today, but gets

progressively worse each month
Month 5
Month 4
Month 3
Month 2
Month 1

Non-Business Use
122
 Example: Stability

STABLE NOT STABLE

DEVICE DEVICE

Jan 0.0041 Jan 0.0041

Feb 0.0041 Feb 0.0041
Mar 0.0042 Mar 0.0042
Apr 0.0041 Apr 0.0043
May 0.0041 May 0.0045
Jun 0.0042 Jun 0.0046
Jul 0.0040 Jul 0.0047
Aug 0.0041 Aug 0.0048

THICKNESS OF PHONE IS KNOWN (REFERENCE) = 0.0041

Non-Business Use
123
 Linearity
❑ How accurate your measurements are through the
expected range of measurements in which the device or
instrument is intended to be used

o Ideally, the measurement error will be the same

Linearity = |Slope| * Process Variation
across the range of likely values
% Linearity = |Slope| * 100
o Linearity often shows up as an increase in
measurement error when measuring larger values

❑ Example: Thermometer is very good at low temperatures

(around zero degrees C), but not as good near 100
degrees C or higher

Non-Business Use
124
 Example: Linearity

LINEAR NOT LINEAR

PART DIFFERENCE PART DIFFERENCE
SIZE FROM STANDARD SIZE FROM STANDARD
0.004 0.0001 0.004 0.0001
0.005 0.0000 0.005 0.0000
0.006 0.0002 0.006 0.0002
0.007 0.0001 0.007 0.0004
0.008 0.0001 0.008 0.0005
0.009 0.0001 0.009 0.0009
0.010 0.0000 0.010 0.0010
0.015 0.0001 0.015 0.0012

COMPARE DIFFERENCE FROM STANDARD OVER RANGE OF VALUES

Non-Business Use
125
 Sources of Variation - Summary

Non-Business Use
126
 How to determine data validity
❑ Lots of sources of measurement variation

❑ The most common drivers of measurement variation have

been mentioned:

• Repeatability
Gage R&R study
• Reproducibility

• Resolution

• Bias

• Stability

• Linearity

Non-Business Use
127
 What is Gage R&R
❑ Specialized experiment performed to check likely sources of measurement
variation to determine whether the data is trustworthy

❑ R&R stands for Repeatability and Reproducibility

❑ Gage = Process and devices used for collecting data

❑ Gage R&R depends on type of data

578.94 GOOD
482.02
613.27 BAD

VARIABLE Attribute
Non-Business Use
128
 Notes about MSA design
❑ To prepare for an MSA, you must

▪ Collect samples from the process that span the specification range of the
measurement in question

▪ Include out-of-spec high samples and out-of-spec low samples

▪ Include all Operators in the MSA who routinely measure the product

❑ The number of samples times the number of Operators should be greater than or
equal to fifteen, with three trials for each sample

❑ If this is not practical, increase the number of trials as the table in next slide

Non-Business Use
129
 Notes about MSA design
❑ Measurement System Analysis Design
Samples x Operators Trials
S x O ≥ 15 3
8 ≤ S x O < 15 4
5≤SxO<8 5
SxO<5 6

❑ Code the samples such that the coding gives no indication to the expected
measurement value – this is called blind sample coding

❑ Ask each Operator to measure each sample three times in random sequence.

❑ Ensure that the Operators do not “compare notes”

❑ We will utilize Minitab to analyze MSA for Variable and Attribute Data

Non-Business Use
130
 Notes about MSA design
❑ A Crossed Design is used only in non-destructive testing and assumes that all the parts
can be measured multiple times by either operators or multiple machines
o Gives the ability to separate part-to-part Variation from measurement system Variation

o Assesses Repeatability and Reproducibility

o Assesses the interaction between the operator and the part

❑ A Nested Design is used for destructive testing and also situations where it is not
possible to have all operators or machines measure all the parts multiple times
o Destructive testing assumes that all the parts within a single batch are identical enough
to claim they are the same

o Do not include all possible combinations of factors

o Uses slightly different mathematical model than the Crossed Design

Non-Business Use
131
 MSA (Variable) - Minnesota Polymer Co.
❑ Minnesota Polymer Co. supplies a special grade of silica (POMBLK-15) with
concentration specification by weight is 15 +/- 2%. Silica concentration is
determined by taking a sample of the powdered resin and pressing it into a 4 cm
disk using a 25-ton hydraulic press. The sample disk is then analyzed by x-ray
fluorescence energy dispersive spectroscopy (XRF-EDS) to measure the silica
content. SPC analysis of historical batch silica concentration results have indicated
out-of- control symptoms and poor Cpk.

❑ Fatimah Ali, an industrial engineer, has suggested conducting a measurement

system analysis to find out the contribution of the measurement system to the
process variation.

Non-Business Use
132
 MSA (Variable) - Minnesota Polymer Co.
❑ Fatimah has collected five samples from POMBLK-15 process which span the silica
specification range and included two out-of-specification samples pulled from
quarantine lots. She has asked each operator to randomly analyze three samples

❑ She has sent a portion of each sample to the Company’s R&D Headquarters in Hong
Kong for silica analysis. These results will serve as reference values for each
sample

Sample # Operator 1 Operator 2 Operator 3

Reference1
Trial 1 Trial 2 Trial 3 Trial 1 Trial 2 Trial 3 Trial 1 Trial 2 Trial 3
1 17.3 18.2 17.9 18.2 18.1 18.0 18.0 17.8 17.8 18.2
2 14.0 14.4 14.9 14.8 14.8 14.6 14.8 14.4 14.4 14.5
3 13.3 14.0 13.9 13.8 13.9 14.2 14.0 13.8 13.7 13.8
4 16.7 17.2 17.2 17.4 17.4 17.3 17.5 17.4 17.5 17.5
5 12.0 12.9 12.8 12.5 12.5 12.9 12.8 12.9 12.5 12.6

1
As Reported by Hong Kong R&D Center

Non-Business Use
133
 Create Gage R&R Study Worksheet

Open a new worksheet. Click on Stat ➔ Quality Tools ➔ Gage Study ➔ Create Gage R&R Study Worksheet

Non-Business Use
134
 Create Gage R&R Study Worksheet

Enter the Number of Operators, the Number of Replicates and the Number of Parts in the dialogue box. Click
OK.

Non-Business Use
135
 Create Gage R&R Study Worksheet

The worksheet is modified to include a randomized run order of the samples

Non-Business Use
136
 Create Gage R&R Study Worksheet

Name the adjoining column Silica Conc and transcribe the random sample measurement data to the relevant
cells in the worksheet. 137
Non-Business Use
 Perform Gage R&R Analysis

Click on Stat ➔ Quality Tools ➔ Gage Study ➔ Gage R&R Study (Crossed)

Non-Business Use
138
 Perform Gage R&R Analysis

Select C2 Parts for Part numbers, C3 Operators for Operators and C4 Silica Conc for Measurement data in the dialogue
Non-Business Use box. Click the radio toggle button for ANOVA under Method of Analysis. Click Options 139
 Perform Gage R&R Analysis

Six (6) standard deviations will account for 99.73% of the Measurement System variation. Enter Lower Spec Limit and
Non-Business Use Upper Spec Limit in the dialogue box. Click OK. Click OK. 140
 The output of Gage R&R Analysis

Non-Business Use
141
 Component of Variation
❑ Let us more closely examine the graphical output of the Gage R&R (ANOVA) Report
for Silica Conc.

❑ A good measurement system will have the lion’s share of variation coming from the
product, not the measurement system

❑ Consequently, we would like the bars for repeatability and reproducibility to be

small relative to part-to-part variation

Non-Business Use
142
 MSA Range Chart by Operators
❑ The range chart should be in control i.e., the points in the sample range charts are
within the control limits. If it is not, a repeatability problem is present

❑ If one appraiser is out-of control, the method used differs from the others. If all
appraisers have some out of-control ranges, the measurement system is sensitive
to appraiser technique and needs improvement

Non-Business Use
143
 MSA X-bar Chart by Operators
❑ By contrast, the X-bar SPC chart of should be out of control. This is because we
intentionally picked the samples to cover the spread of the process specification to
indicate that the variability present is due to part to part differences rather than
Operator to Operator differences

❑ Approximately one-half or more of the averages plotted on the X– chart should fall
outside the control limits

❑ If less than half of the averages fall

outside the control limits, then

either the measurement system

lacks adequate effective resolution

or the sample does not represent the expected process variation

Non-Business Use
144
 MSA Silica Concentration by Sample Number
❑ The circles with a cross indicate the mean of the sample data and the solid circles
are individual data points. We want a tight grouping around the mean for each
sample and we want significant variation between the means of different samples.

❑ If we do not have variation between samples the MSA has been poorly designed and
we essentially have five samples

of the same thing

Non-Business Use
145
 MSA Silica Concentration by Operator
❑ The shaded boxes represent the interquartile range (Q3-Q1) for each Operator

❑ The interquartile range (IQR) is the preferred measure of spread for data sets
which are not normally distributed

❑ The solid line within the IQR is the median silica concentration of all samples by
Operator

❑ If Operators are performing the

same, we would expect similar

means, medians and IQRs

Non-Business Use
146
 MSA Sample by Operator Interaction
❑ The lines for each Operator should be reasonably parallel to each other
▪ Lines are virtually identical; operators are measuring the parts the same

▪ One line is consistently higher or lower than the others, that operator is measuring parts
consistently higher or lower than the others

▪ Lines are not parallel or crossed, the operators’ ability to measure a part depends on
which part is being measured (an interaction between operator and part)

Non-Business Use
147
 Focus on the analytical Minitab output
❑ We will focus on four Gage R&Rs metrics as defined below

𝜎 2 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
% 𝐶𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 = X 100
𝜎 2 𝑇𝑜𝑡𝑎𝑙

𝜎𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
% 𝑆𝑡𝑢𝑑𝑦 𝑉𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 = X 100
𝜎𝑇𝑜𝑡𝑎𝑙

6𝜎
𝑇𝑤𝑜 − 𝑆𝑖𝑑𝑒𝑑 𝑆𝑝𝑒𝑐 % 𝑃/𝑇 = 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 X
𝑈𝑆𝐿 −𝐿𝑆𝐿
100
3 𝜎𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
𝑂𝑛𝑒 − 𝑆𝑖𝑑𝑒𝑑 𝑆𝑝𝑒𝑐 % 𝑃/𝑇 = X
𝑇𝑂𝐿
100
P/T = Precision to Tolerance Ratio
TOL = Process Mean – LSL for LSL only
TOL = USL – Process Mean for USL only
Non-Business Use
148
 Focus on the analytical Minitab output

𝜎𝑡𝑜𝑡𝑎𝑙
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐷𝑖𝑠𝑡𝑖𝑛𝑐𝑡 𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑖𝑒𝑠 = 𝑡𝑟𝑢𝑛𝑐 (1.41 ∗ )
𝜎𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
TRUNC : removes the fractional part of the number, don’t round

❑ Gage R&R Metrics – Rules of Thumb

Gage R&R Metric Unacceptable Acceptable Excellent

% Contribution > 7.7% 2.0 - 7.7% < 2%

% Study Variation > 28% 14 - 28% < 14%
% P/T Ratio > 30% 8 - 30% < 8%
Number of Distinct Categories <5 5 - 10 > 10
Non-Business Use
149
 Focus on the analytical Minitab output
❑ The highlighted output of the Minitab session
window indicates a % Contribution of the
measurement system of 0.55%. This is in the
excellent region

❑ % Study Variation is 7.39% which is also in the

excellent region

❑ Precision to Tolerance ratio is 25.37% which is in

the acceptable region

❑ Number of distinct categories is 19, well within

the excellent region

❑ Overall, this is a good measurement system

Non-Business Use
150
 Checking for linearity and bias
❑ Now, let us proceed to check for linearity and bias by adding the reference concentrations
as measured by the Hong Kong R&D Center for each of the samples to the worksheet

Return to the active worksheet by clicking on Window ➔ Worksheet 1 *** on

the top menu.

Name the adjoining column Reference Conc and enter the reference sample
concentration values corresponding to each sample (Part) number

Non-Business Use
151
 Checking for linearity and bias

Click on Stat ➔ Quality Tools ➔ Gage Study ➔ Gage Linearity and Bias Study

Non-Business Use
152
 Checking for linearity and bias

Select C2 Parts for Part numbers, C5 Reference Conc for Reference values and C4 Silica Conc for Measurement data in the
Non-Business Use
dialogue box. Click OK. 153
 Checking for linearity and bias
❑ The Minnesota Polymer measurement
system is reading approximately 0.67
wt % Silica higher than Hong Kong

❑ We can’t cannot say either system is

right or wrong from this comparison but
merely that there is a difference
between the two instruments

❑ This difference could have process

capability implications if it is validated

Non-Business Use
154
 MSA (Attribute) - Customer satisfaction
❑ David Raffles Lee has just joined Virtual Cable Co., the leading mobile company as
CEO. During a tour of one of the Customer Service Centers, David witnessed a
couple of heated discussions with the customer service agents. David was shocked
about this and wanted to learn more about customer satisfaction at Virtual Cable.

❑ David formed a team to analyze the measurement of customer satisfaction

❑ The team prepared ten scripts of typical customer complaints with an intended
outcome of pass (customer satisfied with the customer service agent’s response) or
fail (customer dissatisfied with the response)

❑ Twenty “customers” were coached on the scripts, one script for two customers

Non-Business Use
155
 MSA (Attribute) - Customer satisfaction
❑ These customers committed the scripts to memory and presented their service
issue to three different Customer Service Agents at three different Customer
Service Centers

❑ Each customer was issued an account number & profile to be able to rate the
customer’s satisfaction level in the customer feedback database
Operator 1 Operator 2 Operator 3
Script # Reference1
Rep 1 Rep 2 Rep 1 Rep 2 Rep 1 Rep 2
1 F F F F F F F
2 P P P P P P P
3 P P P P P P P
4 P P P P P P P 1
Intended outcome of script from
5 F F F F P F F Customer Satisfaction Team
6 P P P P P P P
7 F F F F F F F
8 F F F F F F F
9 P P F P P F P
10 F F F F F F F
Non-Business Use
156
 Create Attribute Agreement Analysis

Open a new worksheet. Click on Stat ➔ Quality Tools ➔ Create Attribute Agreement Analysis Worksheet

Non-Business Use
157
 Create Attribute Agreement Analysis

Enter the Number of samples, the Number of appraisers and the Number of replicates in the dialogue box. Click OK.

Non-Business Use
158
 Create Attribute Agreement Analysis

The worksheet is modified to include a randomized run order of the scripts (samples).

Non-Business Use
159
 Create Attribute Agreement Analysis

Name the adjoining columns Response and Reference. Transcribe the satisfaction level rating and the reference value
of the script to the appropriate cells.
Non-Business Use
160
 Perform Attribute Agreement Analysis

Click on Stat ➔ Quality Tools ➔ Attribute Agreement Analysis

Non-Business Use
161
 Perform Attribute Agreement Analysis

Select C4 Response for Attribute column, C2 Samples for Samples and C3 Appraisers for Appraisers in the dialogue
box. Select C5 Reference for Known standard/attribute. Click OK.
Non-Business Use
162
 MSA Attribute Agreement (graphical)
Assessment Agreement Date of study:
Reported by:
Name of product:
Misc:
From the graphical
Within Appraisers Appraiser vs Standard results, we can see
1 00 95.0% CI
Percent
1 00 95.0% CI
Percent
that the Customer
Service Agents were
90 90
in agreement with
each other 90% of
the time and were in
80 80
Percent

Percent
agreement with the
expected (standard)
70 70 result 90% of the
time
60 60

1 2 3 1 2 3
Appraiser Appraiser

Non-Business Use
163
 MSA Attribute Agreement (analytical)
❑ From the analytical results, we can see that
the agreement between appraisers was 80%
and the overall agreement vs the standard
values was 80%

❑ The Kappa Value for all appraisers vs the

standard values was 0.90, indicative of
excellent agreement between the appraised
values and reference values

Non-Business Use
164
 In MSA for Attribute data, Kappa value used
❑ The attribute MSA results allow us to determine the percentage overall
agreement, the percentage agreement within appraisers (repeatability), the
percentage agreement between appraisers (reproducibility), the percentage
agreement with reference values (accuracy) and the Kappa Value (index used to
determine how much better the measurement system is than random chance)

Rules of Thumb for Interpreting Kappa Values

Kappa Value Interpretation
-1.0 to 0.6 A g r e e m e n t e x p e c te d a s b y chance
0.6 to 0.7 M a r g i n a l a g r e e m e n t - Si g n i f i c a n t e f fo r t r e q u i r e d t o i m p r o v e M S
0.7 to 0.8 G o o d a g r e e m e n t - So m e i m p r o v e m e n t t o M S i s warranted
0.9 to 1.0 E x c e l l e n t agreement

Non-Business Use
165
 MSA Attribute Agreement (analytical)
❑ Another way of looking at this case is that out of sixty expected outcomes there were
only three miscalls on rating customer satisfaction by the Customer Service Agents
included in this study Script # Reference1
Operator 1 Operator 2 Operator 3
Rep 1 Rep 2 Rep 1 Rep 2 Rep 1 Rep 2
❑ We now have confidence 1 F F F F F F F
2 P P P P P P P
3 P P P P P P P
in the feedback of
4 P P P P P P P
5 F F F F P F F
customer satisfaction 6 P P P P P P P
7 F F F F F F F
measurement system 8 F F F F F F F
9 P P F P P F P
and proceed to identify 10 F F F F F F F

and remedy the underlying root causes of customer dissatisfaction

Non-Business Use
166
 Repeatability & Reproducibility Common
Problems
❑ Repeatability Problems:

▪ Calibrate or replace gage

▪ If only occurring with one operator, re-train

❑ Reproducibility Problems:

▪ Measurement machines

o Similar machines : Ensure all have been calibrated and that the standard
measurement method is being utilized.

o Dissimilar machines : One machine is superior

Non-Business Use
167
 Repeatability & Reproducibility Common
Problems
❑ Reproducibility Problems:

▪ Operators

o Training and skill level of the operators must be assessed.

o Operators should be observed to ensure that standard procedures are

followed.

▪ Operator/machine by part interactions : Understand why the operator/machine

had problems measuring some parts and not others.

o Re-measure the problem parts

o Problem could be a result of gage linearity

o Problem could be fixture problem

o Problem could be poor gage design 168

Non-Business Use
 Signal Averaging

❑ Signal Averaging can be used to reduce Repeatability error when a better gage is not
available

▪ Uses average of repeat measurements

▪ Uses Central Limit theorem to estimate how many repeat measures are necessary

❑ Signal Averaging is a method to reduce

Repeatability error in a poor gage when a

better gage is not available or when a better

gage is not possible

Non-Business Use
169
 Signal Averaging Example

❑ Suppose Study Variation (SV)/Tolerance is 35% represents the biggest problem,

Repeatability, and If SV/Tolerance must be 15% or less to use gage.

❑ Suppose the Standard Deviation for one part measured by one person many times is 9.5

❑ Determine what the new reduced Standard Deviation should be to meet our desire of a
15% gage

❑ We are assuming that 15% will be acceptable for the short term until an appropriate fix
can be implemented

❑ The 9.5 represents our estimate for Standard Deviation of population of Repeatability

Non-Business Use
170
 Signal Averaging Example

❑ Determine sample size:

▪ We now use it in the Central Limit Theorem equation to

estimate the needed number of repeated measures

to do this we will use the Standard Deviation estimated

▪ Using the average of 6 repeated measures will

reduce the Repeatability component of measurement

error to the desired 15% level

This method should be considered temporary!

Non-Business Use
171
Process Capability Analysis

172
 Understanding Process Capability
❑ Process Capability: the inherent ability of a process to meet the expectations of
the customer without any additional efforts.

❑ Provides insight as to whether the process has a :

▪ Centering Issue (relative to specification limits)

▪ Variation Issue

▪ A combination of Centering and Variation

▪ Inappropriate specification limits

❑ Allows for a baseline metric for improvement

Non-Business Use
173
 Capability Analysis
The X’s The Y’s
(Inputs)
Y = f(X) (Process Function) Variation – “Voice of
(Outputs)
the Process”

Frequency
Op i Verified Op i + 1
? Data for
Y1…Yn
X1
Y1 10.16
10.11
10.16 9.87
X2 Off-Line 10.05
10.11 9.99
10.16
9.87 10.11
Analysis Scrap 10.33
10.05 10.12
9.99 10.05
Correction 10.44
10.33 10.43
10.12 10.33

X3 Y2 9.86
10.44 10.21
10.43 10.44
10.01
10.21 9.86
9.80 9.90 10.0 10.1 10.2 10.3 10.4 10.5
10.07
9.86
10.29
10.07 10.15
10.01 10.07
10.36
10.29 10.44
10.15 10.29
10.03
10.44 10.36
X4 10.36
10.33
10.03
10.15
10.33
Yes No Y3 10.15

X5 Correctable
?

Requirements – “Voice
Critical X(s): Data - VOP of the Customer”
Any variable(s) 10.16
10.11 9.87 10.16
LSL = 9.96 USL = 10.44
10.05 9.99 10.11
which exerts an 10.33
10.44
10.12
10.43
10.21
10.05
10.33
9.86 10.44
undue influence on 10.07
10.29
10.01
10.15
9.86
10.07
10.44
the important 10.36
10.03
10.33
10.29
10.36

outputs (CTQ’s) of a 10.15

Defects
process Defects

Capability Analysis Numerically

Compares the VOP to the VOC -6 -5 -4 -3 -2 -1 +1 +2 +3 +4 +5 +6

Non-Business Use
9.70 9.80 9.90 10.0 10.1 10.2 10.3 10.4 10.5 10.6
Percent Composition
174
 Process Output Categories
Incapable Off target
LSL
Average
USL LSL Average
USL

Target Target

Capable and
on target
Average
LSL USL

Target
Non-Business Use
175
 Problem Solving Options – Shift the Mean
.
❑ This involves finding the variables that will shift the process over to the target.
This is usually the easiest option
USL
LSL
Shift

Non-Business Use
176
 Problem Solving Options – Reduce Variation
❑ This is typically not so easy to accomplish and occurs often in Six Sigma projects

LSL USL

Non-Business Use
177
 Problem Solving Options – Shift Mean &
Reduce Variation
❑ Combination of shifting the Mean and reducing variation – This is the primary
objective of Six Sigma projects

LSL USL
Shift & Reduce

Non-Business Use
178
 Problem Solving Options – Move the
specification limits
❑ Obviously this implies making them wider, not narrower. Customers usually do not
go for this option but if they do…it’s the easiest!

LSL USL USL

Move Spec

Non-Business Use
179
 Capability as a Statistical Problem
❑ Our Statistical Problem: What is the probability of our process producing a defect ?

Define a Practical
Problem

Create a
Statistical Problem

Correct the
Statistical Problem

Apply the Correction to

the Practical Problem

Non-Business Use
180
 Capability Studies
❑ Are intended to be regular, periodic, estimations of a process’s ability to meet its
requirements. Can be conducted on both Discrete and Continuous Data

❑ Are most meaningful when conducted on stable, predictable processes

❑ Are commonly reported as Sigma Level which is optimal (short term) performance

❑ Require a thorough understanding of

▪ Customer’s or business’s specification limits

▪ Nature of long-term vs. short-term data

▪ Mean and Standard Deviation of the process

▪ assessment of the Normality of the data (Continuous Data only)

▪ Procedure for determining Sigma level

Non-Business Use . 181

 Capability Studies
❑ A stable process is one that is consistent with time

▪ Time Series Plots are one way to check for stability, Control charts is another

▪ One of the purposes of the Measure Phase is to identify the many X’s possible
for the defects seen, gather data and plot it to see if there are any patterns to
identify what to work on first

❑ When performing Capability Analysis, try to get as much data as are possible, back
as far in time as possible, over a reference frame that is generally representative
of your process

Non-Business Use
182
 Seven Cycle Process

01 1) Select Output for 5) Determine Data Type

ep
St Ste Improvement (LT or ST)
p

02
2) Verify Customer
07

Requirements
6) Check data for
Step

normality

7Step
3) Validate Specification

Step 03
Limits

4) Collect Sample 7) Calculate Z-Score,

Data
St e

PPM, Yield, Capability

p0

Cp, Cpk, Pp, Ppk

6

04

e
p

Ste St
p 05

Non-Business Use
183
 Verifying the Specifications
Questions to consider:

❑ What is the source of the specifications?

▪ Customer requirements (VOC)

▪ Business requirements (target, benchmark)

▪ Compliance requirements (regulations)

▪ Design requirements (blueprint, system)

❑ Are they current? Likely to change?

❑ Are they understood and agreed upon?

▪ Operational definitions

▪ Deployed to the work force

Non-Business Use
184
 Data Collection
❑ Capability Studies should include “all” observations (100% sampling) for a specified period

Short-term data: Collected across a narrow inference space

❑ Daily, weekly; for one shift, machine, operator, etc

❑ Is potentially free of special cause variation

❑ Often reflects the optimal performance level

❑ Typically consists of 30 – 50 data points

Long-term data:Is collected across a broader inference space

❑ Monthly, quarterly; across multiple shifts, machines, operators, etc

❑ Subject to both common and special causes of variation

❑ More representative of process performance over a period of time

❑ Typically consists of at least 100 – 200 data points

Non-Business Use
185
 Short-term vs. Long-term data
❑ You must know if the data collected from process outputs is a short-term or a long-term
representation of how well the process performs

❑ Consider a manufacturing process where the manufacturer is filling bottles with a certain
fluid. Assume that the product is built in lots. Each lot is built using a particular vendor of
the bottle, by a particular shift and set of employees and by one of many manufacturing
lines. The next lot could be from a different vendor, employees, line, shift, etc.

❑ Each lot is sampled as it leaves the manufacturing facility on its way to the warehouse

Non-Business Use
186
 Short-term vs. Long-term data
❑ The results are represented by the graphic where you see the performance data on a lot
by lot basis for the amount of fill based on the samples taken. Each lot has its own
variability and average as shown. The variability actually looks reasonable and we notice
that the average from lot to lot is varying as well

❑ What the customer eventually experiences is the amount of fluid in each bottle

❑ It can now be seen and stated that the long-term variability will always be greater than
the short-term variability Lot 1 Lot 5

Fill Quantity
Lot 3

Lot 2

Lot 4
Short-term studies

Long-term study

Non-Business Use
187
 Short-term vs. Long-term data
Process Baseline: The
average, long-term performance
level of a process when all input
variables are unconstrained. Long-term
baseline

Short Term 4
Performance

` 3

2
1
LSL TARGET USL
Non-Business Use
188
 Short-term vs. Long-term data
❑ From previous slide, the “road” appearing graphic actually represents the target (center
line) and the upper and lower spec limits

❑ Process baseline is a term used to describe the output performance of a process

❑ Whenever you hear the word “baseline” it automatically implies long-term performance

❑ Example: imagine you reported that the process performance baseline based on
distribution 3, you would mislead yourself and others that the process had excellent on
target performance. If you used distribution 2, you would be led to believe that the
average performance was near the USL and that most of the output of the process was
above the spec limit

Non-Business Use
` 3 2 189
 Guidelines to know if the data is short or
long-term data?
❑ A somewhat technical interpretation of long-term data is that the process has had the
opportunity to experience most of the sources of variation that can impact it

❑ Short-term data is a “snapshot” of process performance and characterized by :

One shift One line

One batch One employee

One type of service One or only a few suppliers

❑ Long-term data is a “video” of process performance and characterized by

Many shifts Many batches

Many employees Many services and lines

Many suppliers

Non-Business Use
190
 Guidelines to know if the data is short or
long-term data?
❑ Long-term variation is larger than short-term variation

❑ As a general rule, short-term data consist of 20 to 30 data points over a relatively short
period of time and long-term data consist of 100 to 200 data points over an extended
period of time
▪ Do not be misled by the volume of product or service produced as an indicator of long and short-
term performance

▪ Data that represents the performance of a process that produces 100,000 widgets a day for that
day will be short-term performance

▪ Data the represents the performance of a process that produces 20 widgets a day over a 3
month period will be long-term performance

❑ You should now appreciate why, when we report process performance, we need to know
what the data is representative of
Non-Business Use
191
 Shift over time from short to long

Even stable processes will drift and shift over time by as much as 1.5 Standard Deviations on
the average
Long Term
Overall Variation

Short Term
Between Group Variation

Short Term
Within Group Variation

Non-Business Use
192
 The basis for capability calculations

SS total = SS between + SS within

Precision
Shift (short-term capability)
x
x x
x
x x
x x x
x x
x
x x x Time
x x
x x x x
x x x
x
Non-Business Use
193
 Stability

❑ A Stable Process is consistent over time

▪ Time Series Plots and Control Charts are the typical graphs used to determine stability

❑ At this point in the Measure Phase there is no reason to assume the process is stable

Time Series Plot of PC Data

70

For example, looking at the circled

60 region, it has a much tighter
variation. If we could find the X’s
that are causing the instability this
PC Data

50
is the best the process can perform
in the short term
40

30
1 48 96 144 192 240 288 336 384 432 480
Index

Non-Business Use
194
 Measures of Capability

Hope Cp and Pp
▪ What is Possible if your process is perfectly Centered

▪ The Best your process can be

▪ Process Potential (Entitlement)

Reality
Cpk and Ppk
▪ The Reality of your process performance

▪ How the process is actually running

▪ Process Capability relative to specification limits

Mathematically Cpk and Ppk are the same and Cp and Pp are the same
The only difference is the source of the data, Short-term and Long-term, respectively
Non-Business Use
195
 Measures of Capability

Hope
USL - LSL

Six times the sample

Standard Deviation
Reality

Sample Mean

LSL – Lower specification limit

Three times the sample
USL – Upper specification limit Standard Deviation

If we use LSL ➔ Cpl or Ppl Consider the “K” value the

If we use USL ➔ Cpu or Ppu penalty for being off center
Non-Business Use
196
 Assessing Process Capability
❑ Should be assessed, if the process is in statistical control. That is, the natural behavior of
the process after all sources of instability are removed

❑ Basic assumption is that the process output follows a bell- shaped normal curve. If this is
the case, simple calculations based on data from the control chart can be used.

❑ Calculating the Cp gives only approximate results. This is because: there is always some
sampling variation

❑ The Process Capability Index (Cp) is sometimes called the process potential index

❑ Process Potential (Cp) compares allowed variation to the actual variation. That is, it
compares chance causes of process variation with specifications

Non-Business Use
197
 The Cp Index
❑ Should be assessed, if the process is in statistical control. That is, the natural behavior of
the process after all sources of instability are removed

❑ Basic assumption is that the process output follows a bell- shaped normal curve. If this is
the case, simple calculations based on data from the control chart can be used.

❑ Calculating the Cp gives only approximate results. This is because: there is always some
sampling variation

❑ The Process Capability Index (Cp) is sometimes called the process potential index

❑ Process Potential (Cp) compares allowed variation to the actual variation. That is, it
compares chance causes of process variation with specifications

Non-Business Use
198
 The Cp Index Example

USL - LSL
6
Example: Suppose we want to compute the Cp index for bolt dimensions.
Suppose the specification is 10.75  0.25; that is the upper and lower
tolerances are 11.0 and 10.5. Suppose that the standard deviation is .0868.
What is the process capability index (Cp) ?

Cp = 11.0 -10.5 / 6*(.0868) = 0.96

Non-Business Use 199

 The Cp Index
❑ If Cp = 2.0 The process is capable ➔ The specs are twice as wide as the process spread

❑ If Cp = 1.0 Marginal ➔ The specs are the same as the 6 variability

❑ If Cp = .80. Not capable ➔ The specs are only 0.80 or 80% of the 6 variability

❑ Most Companies have established minimum Cp objectives of 1.33 or higher

Non-Business Use
200
 Examples of Capability
❑ Some examples of where capability analysis can be used:

❑ Process that is not centered

❑ Process with large variability

❑ One-sided specifications

❑ Setting/confirming customer specifications

Non-Business Use
201
 Numerical Example on Capability
❑ Assume we have a process with mean = 50, standard deviation = 4, USL = 58 and LSL = 46

❑ We divide this problem into two parts. First the percent out of specification on the high
end (greater than the USL) and then the percent out on the low end (less than the LSL)
USL − X X − LSL
❑ The normal distribution is: Z= ;
S S
Z is the number of standard deviations that the specification is from the mean

❑ Normal probability tables give you the percent of the distribution that would exceed the
specification limit for a given z value

❑ Remember that 68.3% of the data is within ±1S (therefore 31.7% is outside of ±1S)

Non-Business Use
202
 Out of Specification Calculations
USL − X X − LSL
Z= ;
S S
58 − 50 50 − 46
Z= ;
4 4

Z = 2 for the upper specification and 1 for the lower specification

Z = 2 (2.28% out of specification); z=1 is (15.9% out of specification)

The total percent expected to be out of specification would be 18.1%

Non-Business Use
203
 Capability Extended Example
❑ Assume that a process is in statistical control based on an X-bar and R chart with
subgroups of 5. The grand average (or centerline of the X-Bar chart) is calculated as
0.99832, and the average range (or centerline of the R chart) is calculated as 0.02205

❑ From the table of d2 values, we find d2 is 2.326 for subgroups of 5

❑ Thus, using the equation above for calculating the process standard deviation using the
Range chart method:
0.02205
𝜎ො = = 0.00948
2.326

Non-Business Use
204
 Capability Extended Example
❑ The grand average (or centerline of the X-Bar chart) is calculated as 0.99832, and the
average range (or centerline of the R chart) is calculated as 0.02205, 𝜎ො = 0.00948
❑ If the process requirements are a lower specification of 0.980 and an upper specification
of 1.020 (i.e., 1:000 ± 0:020), the Z values are calculated as:

𝑈𝑝𝑝𝑒𝑟 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 − 𝑋 1.020 − 0.99832

𝑍𝑈 = = = 2.3
𝜎ො 0.00948

𝐿𝑜𝑤𝑒𝑟 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 − 𝑋 0.99832 − 0.980

𝑍𝐿 = = = 1.9
𝜎ො 0.00948
Referring to the Z Table, we find that approximately 1.1% will be outside specifications
(based on the Zu value of 2.3) and approximately 2.9% will be outside specifications (based
on the ZL value of 1.9). Adding the percents finds a total reject rate of 4.0%, which can
be equivalently expressed as a DPMO of 40,000 or a yield of 96.0%

Non-Business Use
205
 Check the Normality of each supplier

Check for Normality: Stat > Basic Statistics > Normality

By looking at the “P-values’’ the data look

to be Normal since P is greater than .05

Non-Business Use
206
 Capability Analysis using Minitab
Stat > Quality Tools > Capability Analysis (Normal)

Create a Capability Analysis

for both suppliers,
assume long-term data.

Note the subgroup size for

this example is 5.
LSL=598 USL=602

Non-Business Use
207
 MINITAB™ Example – Supplier 1

599.548 is the process Mean Process Capability of Supplier 1

which`falls short of the target
LSL USL
(600) for Supplier 1. The left P rocess Data Within
tail of the distribution falls LS L 598 Ov erall
Target *
outside the lower specification USL 602 P otential (Within) C apability
Cp 1.19
limits S ample M ean
S ample N
599.115
100 C P L 0.66
S tDev (Within) 0.559239 C P U 1.72
S tDev (O v erall) 0.602583 C pk 0.66
The CPL index for Supplier 1 O v erall C apability
Pp 1.11
is .66 which tells us they need PPL 0.62
reduce the process variation PPU 1.60
P pk 0.62
and work on centering C pm *

597.75 598.50 599.25 600.00 600.75 601.50

O bserv ed P erformance E xp. Within P erformance E xp. O v erall P erformance
P P M < LS L 30000.00 P P M < LS L 23088.05 P P M < LS L 32130.26
PPM > USL 0.00 PPM > USL 0.12 PPM > USL 0.84
P P M Total 30000.00 P P M Total 23088.18 P P M Total 32131.10

Non-Business Use
208
 MINITAB™ Example – Supplier 2

600.06 is the process man Process Capability of Supplier 2

for Supplier 2 and is very
close to the target although LSL USL
both tails of the distribution LS L
P rocess Data
598
Within
Ov erall
fall outside of the Target *
USL 602 P otential (Within) C apability
specification limits S ample M ean 600.061 Cp 0.66
S ample N 100 C P L 0.68
S tDev (Within) 1.00606 C P U 0.64
C pk 0.64
The Cpk index is very S tDev (O v erall) 1.14609
O v erall C apability
similar to Supplier 1 but this Pp 0.58
infers that we need to work PPL 0.60
PPU 0.56
on reducing variation P pk 0.56
C pm *

When making a comparison

between Supplier 1 and 2
597 598 599 600 601 602 603
relative to Cpk vs Ppk we
O bserv ed P erformance E xp. Within P erformance E xp. O v erall P erformance
see that Supplier 2 process P P M < LS L 40000.00 P P M < LS L 20251.30 P P M < LS L 36065.24
is more prone to shifting PPM > USL 60000.00 P P M > U S L 26969.82 P P M > U S L 45337.82
P P M Total 100000.00 P P M Total 47221.11 P P M Total 81403.06
over time.

Non-Business Use
209
 MINITAB™ Example – Z-score comparison

Generate the new capability

graphs for both suppliers
and compare Z values or
sigma levels.

Stat>Quality Tools>Capability Analysis>Normal…>Options…Benchmark Z’s (sigma level)

Non-Business Use
210
 Long-term Z-score – Supplier 1
Process Capability of Supplier 1
The overall long term sigma
level is 1.85 for supplier 1 LSL USL
you should also note that it P rocess Data Within
LS L 598 Ov erall
has the potential to be 1.99 Target *
P otential (Within) C apability
sigma as the process stands USL 602
Z.Bench 1.99
S ample M ean 599.115
in its current state S ample N 100 Z.LS L 1.99
S tDev (Within) 0.559239 Z.U S L 5.16
S tDev (O v erall) 0.602583 C pk 0.66
O v erall C apability
Z.Bench 1.85
Z.LS L 1.85
Z.U S L 4.79
P pk 0.62
C pm *

597.75 598.50 599.25 600.00 600.75 601.50

O bserv ed P erformance E xp. Within P erformance E xp. O v erall P erformance
P P M < LS L 30000.00 P P M < LS L 23088.05 P P M < LS L 32130.26
PPM > USL 0.00 PPM > USL 0.12 PPM > USL 0.84
P P M Total 30000.00 P P M Total 23088.18 P P M Total 32131.10

Non-Business Use
211
 Long-term Z-score – Supplier 2
Process Capability of Supplier 2
The overall long term sigma
level is 1.39 for supplier 2, LSL USL
P rocess Data Within
you should also note that it LS L 598 Ov erall
has the potential to be 1.39 Target *
P otential (Within) C apability
USL 602
sigma as the process stands S ample M ean 600.061 Z.Bench 1.67

in its current state S ample N 100 Z.LS L

Z.U S L
2.05
1.93
S tDev (Within) 1.00606
S tDev (O v erall) 1.14609 C pk 0.64
O v erall C apability
Z.Bench 1.40
Z.LS L 1.80
Z.U S L 1.69
P pk 0.56
C pm *

597 598 599 600 601 602 603

O bserv ed P erformance E xp. Within P erformance E xp. O v erall P erformance
P P M < LS L 40000.00 P P M < LS L 20251.30 P P M < LS L 36065.24
PPM > USL 60000.00 P P M > U S L 26969.82 P P M > U S L 45337.82
P P M Total 100000.00 P P M Total 47221.11 P P M Total 81403.06

Non-Business Use
212
 With short-term data do one of the
following:
Option 1 Option 2

Enter “Subgroup size:” = total Go to “Options”, turn off “Within

number of samples subgroup analysis”

Using data from Column “Bi modal” in the Minitab worksheet “GraphingData.mtw”

Non-Business Use
213
 With short-term data do one of the
following:

Capability indices assume

Normally Distributed data
Always perform a Normality test
before assessing Capability

Non-Business Use
214
 Seven Cycle Process does not work for
Attribute data

01 1) Select Output for 5) Determine Data Type

ep
St Ste Improvement (LT or ST)
p

02
2) Verify Customer
07

Requirements
6) Check data for
Step

normality

7Step
3) Validate Specification

Step 03
Limits

4) Collect Sample 7) Calculate Z-Score,

Data
St e

PPM, Yield, Capability

p0

Cp, Cpk, Pp, Ppk

6

04

e
p

Ste St When we follow the steps in performing a Capability

p 05
Study on Attribute Data we hit a wall at step 6.
Attribute Data is not considered Normal so we will use
a different mathematical method to estimate capability
Non-Business Use
215
 Seven Cycle Process does not work for
Attribute data

01 1) Select Output for 5) Calculate

ep
St Ste Improvement DPU
p

02
2) Verify Customer
07

Requirements
6) Find Z-Score
Step

7Step
3) Validate Specification

Step 03
Limits

4) Collect Sample 7) Calculate Z-Score, PPM,

Data
St e

Yield, Capability
Cp, Cpk, Pp, Ppk
p0
6

04

e
p

Ste St Notice the difference when we come to step 5…

p 05

Non-Business Use
216
 Z Score Z Scores

❑ Z Score is a measure of the distance in Standard Deviations of a sample from the Mean.

❑ Given an average of 50 with a Standard Deviation of 3 what is the proportion beyond the
upper spec limit of 54?

50

54 217
Non-Business Use
 Z Score

In our case we have to

lookup the proportion for the
Z score of 1.33. This means
that approximately 9.1% of
our data falls beyond the
upper spec limit of 54. If we
are interested in determining
parts per million defective
we would simply multiply the
proportion .09176 by one
million. In this case there
are 91,760 parts per million
defective.

Non-Business Use
218
 Z short and Z long
❑ Attribute data is always long-term in the shifted condition since it requires so many
samples to get a good estimate with reasonable confidence.

❑ Short-term Capability is typically reported, so a shifting method will be employed to

estimate short-term Capability.

Sigma Short-Term Long-Term

Level DPMO DPMO
You Want to Estimate : ZST ZLT 1 158655.3 691462.5

Short Term Long Term 2 22750.1 308537.5

Your Data Is : Capability Capability
3 1350.0 66807.2
Short Term Subtract
ZST Capability 1.5
4 31.7 6209.7

5 0.3 232.7
Long Term Add
ZLT Capability 1.5
6 0.0 3.4

Non-Business Use
219
 Z short and Z long
❑ By viewing these formulas you can see there is a relationship between them.

❑ If we divide our Z short-term by 3 we can determine our Cpk and if we divide our Z long-
term by 3 we can determine our Ppk.

Non-Business Use
220
 Attribute Capability Example
❑ A customer service group is interested in estimating the Capability of their call center.

❑ A total of 20,000 calls came in during the month but 2,666 of them “dropped” before they
were answered (the caller hung up).

❑ Results of the call center data set:

Samples = 20,000

Defects = 2,666

Non-Business Use
221
 Attribute Capability Example

1. Calculate DPU
2. Look up DPU value on the Z-Table
3. Find Z-Score
4. Convert Z Score to Cpk, Ppk

Example:
Look up ZLT
ZLT = 1.11
Convert ZLT to ZST = 1.11+1.5 = 2.61

Non-Business Use
222
 Attribute Capability Example

1. Calculate DPU
2. Look up DPU value on the Z-Table
3. Find Z-Score Example:
4. Convert Z Score to Cpk, Ppk Look up ZLT
ZLT = 1.11
Convert ZLT to ZST = 1.11+1.5 = 2.61

2.61
.87

Non-Business Use
223
 Measure Phase Overview - The Goal
The goal of the Measure Phase is to:

❑ Define, explore and classify “X” variables using a variety of tools.

▪ Detailed Process Mapping, Fishbone Diagrams, X-Y Matrixes, FMEA

❑ Demonstrate a working knowledge of Basic Statistics to use as a communication tool and a

basis for inference.

❑ Perform Measurement Capability studies on output variables.

❑ Evaluate stability of process and estimate starting point capability.

Non-Business Use
224
 Measure Phase - The Roadblocks
Look for the potential roadblocks and plan to address them before they become problems:

❑ Team members do not have the time to collect data

❑ Data presented is the best guess by functional managers

❑ Process participants do not participate in the creation of the X-Y Matrix, FMEA and Process
Map

Non-Business Use
225
 Measure Phase Checklist
Measure Questions

❑ Identify Critical X’s and potential failure modes

• Is the “as is” Process Map created?

• Are the decision points identified?

• Where are the data collection points?

• Is there an analysis of the measurement system?

• Where did you get the data?

❑ Identify critical X’s and potential failure modes

• Is there a completed X-Y Matrix?

• Who participated in these activities?

Non-Business Use
226
 Measure Phase Checklist
• Is there a completed FMEA?

• Has the Problem Statement changed?

• Have you identified more COPQ?

❑ Stability Assessment

• is the “Voice of the Process” stable?

• If not, have the special causes been acknowledged?

• Can the good signals be incorporated into the process?

• Can the bad signals be removed from the process?

• How stable can you make the process?

Non-Business Use
227
 Measure Phase Checklist
❑ Capability Assessment

• What is the short-term and long-term Capability of the process?

• What is the problem, one of centering, spread or some combination?

❑ General Questions

• Are there any issues or barriers that prevent you from completing this phase?

• Do you have adequate resources to complete the project?

Non-Business Use
228