118 LSS BBO3 - Process Capability PDF

Lean Six Sigma Operations
Black Belt Bridge – Week 1

Process Capability
Release 1.0
Objectives
• Hige Level Review of GB Concepts
• Review capability assessment for Single-sided
specifications and non-normal data
• Overview transformation of non-normal data
• Introduce the capability index Cpm
For Ingersoll Rand Internal Use Only

2 Rev. B Printed 3/23/2009
© 2001 by Sigma Breakthrough Technologies, Inc.
Lean Sigma Roadmap - Measure
Steps Tools Outputs
• •SIPOC
SIPOC/ /VSM
VSM
• •Input/Output
Input/Output
Analysis
Understand the
Analysis  Current State
• •C&E
C&EMatrix
Process Matrix
• •Detailed
Process Maps
DetailedProcess
Process
Maps
Maps  Identified and
measured X’s
Measure • •Data
DataCollection
CollectionPlans
Plans (KPIV’s)
• •Data
• Understand the DataIntegrity
IntegrityAudits
Audits  Measurement
Process Develop and Evaluate • •Continuous
ContinuousMSA
MSA
• Develop and
Evaluate
Measurement Systems (Gage
(GageR&R)
R&R)
system
Measurement
Systems
• •Attribute
AttributeMSA
MSA(Kappa
(Kappa verified
Studies)
Studies)
• Measure Current
Process
 Current
Performance capability of
Measure Current
• •Process
Y’s (KPOV’s)
Process Performance ProcessCapability
Capability

Background - Capability Studies
LSL
LSL USL
USL
We continually gather data on our process and ask:

“Is it capable of producing defect free products?”
Gathering and analyzing the data is called....
A Capability Study
Remember This?
Sigma PPM
2 308,537
3 66,807
4 6,210
5 233
6 3.4
Where does it come from?

Capability Roadmap - 30,000 Ft. View
What
WhatType
Typeof
ofData
Data
Do
DoYou
YouHave
Have??
Attribute
AttributeData
Data Continuous
ContinuousData
Data
Collect
CollectData
Data Collect
CollectData
Data
On
OnProcess
Process On
OnProcess
Process
Analyze
AnalyzeData
Data Analyze
AnalyzeData
Data
In
InExcel
Excel/ /Minitab
Minitab In
InMinitab
Minitab
State
StateCapability
Capability State
StateCapability
Capability
Dpu,
Dpu,Sigma,
Sigma,PPM
PPM Dpu,
Dpu,Sigma,
Sigma,PPM
PPM
Control
ControlCharts
Charts Cp,
Cp,Cpk,
Cpk,Pp,
Pp,Ppk
Ppk
Process Capability Ratios
X - LSL USL- X
Cpk = Min( , )
3s 3s
X - LSL USL- X
CpL = CpU =
3s 3s
CCpk accounts for process centering and spread.

pk accounts for process centering and spread.
Note:
Note: the
thessused
usedis
isss(within) .
(within).
Statistical Assumptions Made in
Capability Studies
1. Data comes from a stable process
– If not, work towards getting the process in control
– Don’t despair, you can make big improvements in
processes like these.
2. Has a Normal Distribution
– If not, transform it or use different reference distributions
(we will cover this later in the course)
• If items #1 and #2 aren’t met, results will be misleading.
• Minitab’s capability Sixpack helps you check these
assumptions

Process Indices
• We will discuss indexes about process performance
and process capability
• Process capability is the “Potential” of the process
(Cp and Cpk)
• Process performance is the processes “Real”
performance (Pp and Ppk)

Process Performance Ratios
USL - LSL
Pp =
6s(overall)
X - LSL USL- X
Ppk  Min( , )
3s (overall) 3s (overall)
Pp
Pp&&Ppk
Ppkmeasure
measureprocess
process“performance.”
“performance.”
This
Thisis
ishow
howyour
yourcustomers
customersview
viewthe
theprocess!
process!

Performance vs. Capability
Variability
Std Dev Std Dev
(Within) (Overall)
Accounting for Center &
Spread
Only Cp Pp
Spread Information
Center &
Cpk Ppk
Spread

Process Performance and Capability
• The Ppk can closely approach the Cp when
– The Customer specifications truly reflect customer
requirements
– The process in under statistical control
– The data approximate the normal distribution
• The Cp is like a benchmark or entitlement.
– The sigma for capability is driven primarily by
random error.
• We would like Ppk to be very close to Cp.

Types of Capability Indexes
• Instantaneous Capability:
– Process Capability over an extremely short period of time
– This should represent the very best perform a process is capable of
over a short time
– This should be a close estimate of process entitlement
– Can be estimated using the “best run” of a short-term or long-term
study
• Short-Term Capability:
– Capability study based on 30-50 data points
– Usually equal to or greater than long-term capability
• Long-Term Capability:
– Capability study based on a large number of data points
– Best estimate of true process capability
– Diagnostics can be made from this data

Goal Setting
Near-Term Goal Move the Ppk to Pp (Center the process)
Move the Pp to Cpk (Reduce

Mid-Term Goal Variation)
Long-Term Goal Move Cpk to Cp (Random Variation)
Six Sigma Process

Cp = 2.00
Ppk = 1.5

Special Considerations:
Single Sided Capability
17-15 Rev. C Printed 3/23/2009

Normal Data, Single Sided
Open project file LS Capability.mpj
Let’s first look at a Histogram and Normal Probability Plot of the
data in column True Position
Histogramof True Position

14
12
Probability Plot of True Position
10 Normal
99
Frequency
Mean 11.98
8 StDev 0.9958
95 N 50
AD 0.228
6 90
P-Value 0.802
80
4 70
Percent
60
50
2 40
30
0 20
9.6 10.8 12.0 13.2 14.4
10
True Position
5
1
9 10 11 12 13 14 15
True Position

Let’s use an USL of 12 (LSL=0, Hard Limit)
Process Capability of True Position
LB USL
P rocess Data Within
LB 0.00000 Ov erall
Target *
USL 12.00000 P otential (Within) C apability
S ample M ean 11.98118 Cp *
S ample N 50 C PL *
S tDev (Within) 0.95525 C PU 0.01
S tDev (O v erall) 1.00087 C pk 0.01
C C pk 0.01
O v erall C apability
Pp *
PPL *
PPU 0.01
P pk 0.01
C pm *
0.0 2.4 4.8 7.2 9.6 12.0 14.4

O bserv ed P erformance E xp. Within P erformance Exp. O v erall P erformance
P P M < LB 0.00 PPM < LB * P P M < LB *
P P M > U S L 500000.00 PPM > U S L 492140.99 P P M > U S L 492499.13
P P M Total 500000.00 PPM Total 492140.99 P P M Total 492499.13
How
Howdo
dowe
wefeel
feelabout
aboutthis
thisprocess’
process’capability?
capability?
• Minitab assumes that the Target is midway between
the Upper and Lower Spec limits
• What if we are targeting a value other than the
midpoint?
• The measure we will get for process capability is the
Cpm
• We want the Cpm to approach 2.00 (or greater)
– This is a measure of centering on the TARGET,
not the middle of the spec
• Let’s take a look….

Adding Cpm

Let’s use the same data with a Target value of 4.00
Process Capability of True Position
LB Target USL
P rocess Data W ithin
LB 0.00000 Ov erall
Target 4.00000
S ample N 50 C PL *
S tD ev (Within) 0.95525 C PU 0.01
S tD ev (O v erall) 1.00087 C pk 0.01
C C pk 2.79
Pp *
PPL *
PPU 0.01
P pk 0.01
C pm 0.33
0.0 2.4 4.8 7.2 9.6 12.0 14.4

O bserv ed P erformance E xp. Within P erformance E xp. O v erall P erformance
P P M < LB 0.00 PPM < LB * P P M < LB *
P P M > U S L 500000.00 PPM > U S L 492140.99 P P M > U S L 492499.13
P P M Total 500000.00 PPM Total 492140.99 P P M Total 492499.13

Now look at the values for Cpk and Cpm when
we shift the mean of the process
Process Capability of True Position Shifted
LB Target USL
P rocess Data W ithin
LB 0.00000 Ov erall
Target 4.00000
S ample N 50 C PL *
C C pk 2.79
Pp *
PPL *
PPU 2.67
P pk 2.67
C pm 2.68
0 2 4 6 8 10 12
P P M < LB 0.00 P P M < LB * P P M < LB *
P P M > U S L 0.00 P P M > U S L 0.00 P P M > U S L 0.00
P P M Total 0.00 P P M Total 0.00 P P M Total 0.00

Special Considerations:
Non-Normal Capability
17-22 Rev. C Printed 3/23/2009

What about Non-Normal Data?
• All of these discussions are great, but what about
Non-normal data?
• We can use the Box-Cox Transformation to
determine a transformation value
– Transformation is like using a different scale
– We must transform the data AND specification
limits for this to be valid
• NOTE: Only uni-modal data should be transformed!

Non-Normal Data
• Open project file LS Capability.mpj
• Let’s look at ‘True Position 2’ data which is from a
Chi-square distribution
• This type of distribution is typical of a single sided
spec item where there is a hard limit
Histogram of True Position 2 Probability Plot of True Position 2
18 Normal
99
16 Mean 1.761
StDev 1.499
95 N 50
14
AD 1.610
90
P-Value <0.005
12
80
Frequency
10 70
Percent
60
8 50
40
30
6
20
4 10
2 5
0 1
0 2 4 6 8 0.0 2.5 5.0 7.5
True Position 2 True Position 2

Capability Study
If we perform a capability study with the non-normal
data, we can still get results …. but are they correct?
Process Capability of True Position 2
LB Target USL
P rocess D ata Within
LB 0.00000 Overall
Target 4.00000
S ample N 50 C PL *
C C pk 2.04
Pp *
PPL *
PPU 2.26
P pk 2.26
C pm 0.98
-2 0 2 4 6 8 10 12
P P M < LB 0.00 P P M < LB * P P M < LB *
P P M > U S L 0.00 P P M > U S L 0.00 P P M > U S L 0.00

The Box-Cox Transformation
• When you use the Box-Cox Transformation, Minitab estimates a
l (lambda) value
– Minitab presents the l value which minimizes the s of a
standardized transformed variable
– The resulting transformation is Yl when l ¹ 0 and LogeY
when l = 0
• This method searches through many types of transformations


• The output is a graph which displays:
– the best estimate of l for the transformation
– two closely competing values of l
– a 95% confidence interval for the true value of l
Box-Cox Plot of True Position 2

Lower C L Upper CL
4.5 Lambda
(using 95.0% confidence)
4.0 Estimate 0.400345
Lower C L 0.114732
3.5 Upper C L 0.676266
Best Value 0.500000

3.0
StDev
2.5
2.0
1.5
Limit
1.0
-1 0 1 2 3
Lambda

So What?
Now we can use the  value in a capability analysis.

Capability Study
Process Capability of True Position 2
Using Box-Cox Transformation With Lambda = 0.449
LB* Target* U S L*
P rocess D ata transformed data Within
LB 0.00000 O v erall
Target 4.00000
S ample N 50 C PL *
S tDev (Within) 1.30452 C P U 1.30
C C pk 0.83
A fter Transformation
LB* 0.00000
Target* 1.86348 Pp *
U S L* 3.05177 PPL *
S ample M ean* 1.17527 PPU 1.28
S tDev (Within)* 0.47987 P pk 1.28
S tDev (O v erall)* 0.48786 C pm 0.47
0.0 0.6 1.2 1.8 2.4 3.0

P P M < LB 0.00 P P M < LB* * P P M < LB* *
P P M > U S L 0.00 P P M > U S L* 46.06 P P M > U S L* 59.93

Study Comparisons
• Non-transformed data results:
– Cpk = 2.62
– Ppk = 2.26
• Transformed data results:
– Cpk = 1.30
– Ppk = 1.28
• So which best describes the process ?

Hints:
Capability Studies
• Always perform a Capability Study.
• Think of a Capability Study is a thermometer on your
process.
• If your process is out of control and not normal, don’t
trust the Capability Indices.
• If your process is out of control and not normal, depend
on the graphs/pictures to tell you about the process.

Hints:
Things to Know Before the Capability Study
• Is the measurement system adequate?
• For a Key Process Output Variable is:
– Nominal Best,
– Larger Better
– Smaller Better?
• Is the concern for
– Process Centering
– Process Variation
– or both?
• What is the baseline for the process variable?
– Mean
– Sigma
• Is the Output currently in statistical control?
• Is the Output affected by time?
• Are there multiple responses you’re concerned with?
– What’s the priority?

Summary
• Reviewed High Level GB Concepts
• Reviewed capability assessment for Single-Sided
specifications and non-normal data
• Provided an overview for transformation of non-
normal data
• Introduced the capability index Cpm

Connecting the Dots
• To understand our process we must “Begin with the
Y” and thoroughly understand current performance or
capability.
• Prior to determining process capability, we must
assess our measurement system using Measurement
System Analysis.
• Capability can be measured differently depending
whether our data is Attribute or Continuous

Appendix:
Connecting Capability and
Measurement
17-36 Rev. C Printed 3/23/2009

Measurement Error Effect
on Capability Indices - Cp
• We know that
USL  LSL
CpObs  where sObs  s2Act  sMS
2
6sObs
• Therefore: USL  LSL

CpObs 
6 s 2Act  sMS
2
What
Whathappens
happensto
toCCppwhen
whenthe
the
Measurement
MeasurementSystem
SystemError
Errorincreases?
increases?
Measurement Error Effect
on Capability Indices - Cpk
• To include the effects of process centering, we know
 USL  X Obs X Obs  LSL  Be

Be careful
careful of
of the
the
CpkObs  MIN  or  direction
direction of
of the
the bias
bias
 3sObs 3sObs  (the
(the sign
sign of
of the
the XXMS
MS)
)
• Where  and 
sObs  sAct
2
 sMS
2
X Obs  X Act  X MS
• Therefore,
  
 USL  X Act  X MS 
X Act  X MS  LSL 
CpkObs  Min  or 
 3 s 2
 s 2
3 s 2
 s 2 
 Act MS Act MS 
What
Whathappens
happensto
toCCpkpkwhen
whenthe
theMeasurement
MeasurementSystem
SystemError
Errorincreases?
increases?
%R&R Guidelines
• Once %R&R has been calculated, you must determine which to
target first:
– Process variation, or
– Measurement systems variation
• Ask the question: is your measurement system capable for the
process improvement work for which you’ll be using it?
• You can assume that the measurement system is capable for
use in process improvement activities when the following
conditions exist for Observed Process Capability (Cpo and
%R&R):
– If Cpo = 1.0 and %R&R < 70%
– If 1.0 < Cpo < 1.5 and %R&R < 40%
– If Cpo = 1.5 and %R&R < 30%
• Note: See detailed graph to follow

%R&R Effect on Capability
8.00
%R&R
7.00
0%
6.00
10%
5.00 20%
30%
CpAct
4.00
40%
3.00 50%
60%
2.00 70%
1.00
0.00
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
1.30
1.40
Source: Larry B.
Barrentine:
Concepts for R&R
CpObs Studies

%R&R vs. Capability
Which might need the most attention -
Measurement System or Process Improvement?
Process %R&R Cpo Which do we work on?

1 10% 0.5 Process
2 40% 1.0 Either one
3 60% 1.5 Measurement system
4 70% 5.5 Measurement system*
Process 2: May Benefit more by addressing Measurement

system than Process improvements at this point.
*Process 4: Would improving %R&R really be worth the effort ?

Appendix:
Z Transformations
17-42 Rev. C Printed 3/23/2009

Process Capability - The Strategy
Centering – Put The Process On Target
Spread – Reduce The Variation
Defects Defects
LSL USL

Z-Transform
• General form: (x - x)
Z=

• This transform produces a value from a distribution where the
Mean = 0 and s = 1
• The value indicates how far the number is from the mean in
units of standard deviations
– For example, if Z = 2, that would say that the value in
question is 2 standard deviations greater than the Mean
• By using this method, we can calculate the proportion of product
that is out-of-spec based on that product’s Output Mean and s

Estimating % Non-conforming
• For estimating yield for a process, we will substitute the
Lower Spec Limit (LSL) and the Upper Spec Limit
(USL) for x
(x - x )
Z =

(SL- x)
Z=

The
Theresult
resultwill
willtell
tellus
ushow
howfarfarthe
theprocess
processmean
meanisisfrom
fromthe
the
specification
specificationlimit
limitininunits
unitsof
ofsigma.
sigma.
Standard Normal Distribution
One Standard
Deviation (

This is a standard Normal distribution where:

Mean = 0
Standard Deviation = 1
Z Score
SL    SL
z= ———

2.5 
= ————

= 2.5


What’s the Area Under the Tail ?
 SL
2.5
We get the area
under the tail from a
Z table ...or we can
use Minitab!


Z Table
.00621
or 6210 ppm
Z Table - Tail Area Probability
2.0 0.00
0.02275 0.01
0.02222 0.02
0.02169 0.03
0.02118 0.04
0.02068 0.05
0.02018 0.06
0.01970
2.1 0.01786 0.01743 0.01700 0.01659 0.01618 0.01578 0.01539
2.2 0.01390 0.01355 0.01321 0.01287 0.01255 0.01222 0.01191
2.3 0.01072 0.01044 0.01017 0.00990 0.00964 0.00939 0.00914
2.4 0.00820 0.00798 0.00776 0.00755 0.00734 0.00714 0.00695
2.5 0.00621 0.00604 0.00587 0.00570 0.00554 0.00539 0.00523
2.6 0.00466 0.00453 0.00440 0.00427 0.00415 0.00402 0.00391
2.7 0.00347 0.00336 0.00326 0.00317 0.00307 0.00298 0.00289
2.8 0.00256 0.00248 0.00240 0.00233 0.00226 0.00219 0.00212
2.9 0.00187 0.00181 0.00175 0.00169 0.00164 0.00159 0.00154
3.0 0.00135 0.00131 0.00126 0.00122 0.00118 0.00114 0.00111
3.1 0.00097 0.00094 0.00090 0.00087 0.00084 0.00082 0.00079

Minitab
In Minitab, go to Calc>Probability Distributions>Normal and check
Cumulative Probability.
Fill in the Z-score of interest. Minitab reports back the probability
indicated below:
Normal with mean = 0 and
standard deviation = 1.00000
x P( X <= x)
2.5000 0.9938
1 - .9938 = .0062
or 6200 ppm

Example
• Assume that we have a process producing 1 gallon bottles of liquid
• If our Upper Spec Limit for the bottle is 1.030 gallons, what percentage
defective (or ppm level) will we be operating at if:
– The mean of our process is 0.995 gallons
– The standard deviation is .010 gallons
SL - x
z= ———
s
1.030 - .995
= —————— = 3.5
.010

Z Table
.00023
or 230 ppm Z Table - Tail Area Probability
3.0 0.00
0.00135 0.01
0.00131 0.02
0.00126 0.03
0.00122 0.04
0.00118 0.05
0.00114 0.06
0.00111
3.1 0.00097 0.00094 0.00090 0.00087 0.00084 0.00082 0.00079
3.2 0.00069 0.00066 0.00064 0.00062 0.00060 0.00058 0.00056
3.3 0.00048 0.00047 0.00045 0.00043 0.00042 0.00040 0.00039
3.4 0.00034 0.00032 0.00031 0.00030 0.00029 0.00028 0.00027
3.5 0.00023 0.00022 0.00022 0.00021 0.00020 0.00019 0.00019
3.6 0.00016 0.00015 0.00015 0.00014 0.00014 0.00013 0.00013
3.7 0.00011 0.00010 9.964E-05 9.577E-05 9.204E-05 8.844E-05 8.498E-05
3.8 7.237E-05 6.951E-05 6.675E-05 6.409E-05 6.154E-05 5.908E-05 5.671E-05
3.9 4.812E-05 4.617E-05 4.429E-05 4.249E-05 4.076E-05 3.909E-05 3.749E-05
4.0 3.169E-05 3.037E-05 2.911E-05 2.790E-05 2.674E-05 2.562E-05 2.455E-05
4.1 2.067E-05 1.979E-05 1.895E-05 1.815E-05 1.738E-05 1.663E-05 1.592E-05


Minitab Approach
x P( X <= x)
3.5000 0.9998
Tail Probability = 1 - .9998

Tail Probability = .0002 or
200 ppm
.9998 .0002
Process Capability

By Definition
 
Z-Transform Example
If we look at the following, we see that, for 150 Melt Index
measurements:
Mean = 10

400
= 0.5 300
LSL USL
Frequency
Let’s say: 200
LSL = 9 100
0
USL = 11 8 9 10 11 12
Population
What
Whatisisthe
theestimated
estimatedpercentage
percentageof
ofthe
the
distribution
distributionthat
thatwill
willbe
beout
outof
ofspec?
spec?
Z-Transform
The task is to determine estimates of the proportion of the
normal curve that is outside both the upper and lower
specification limits. We do that by calculating Z-score for each
spec limit.
(LSL- x) (USL- x)
ZL = ZU =
s s
9 - 10 11 - 10
= =
0.5 0.5
= -2.00 = 2.00
We can now calculate the areas below the lower spec

and above the upper spec using the normal probability
function.
Minitab for Estimating Proportion
Outside Lower Spec
Go to Calc > Probability Distributions >
Normal
Mean and Sigma
Lower Spec
Limit

Minitab Output
Cumulative Distribution Function
Normal with mean = 10.0000 and standard deviation = 0.500000
x P( X <= x)
400
9.0000 0.0228
300
LSL USL
Frequency
LSL 200
100
8 9 10 11 12
Population

Minitab for Estimating Proportion
Outside Upper Spec
Mean and Sigma
Upper Spec
Limit

Minitab Output
Cumulative Distribution Function
Normal with mean = 10.0000 and standard deviation = 0.500000
x P( X <= x)
400
11.0000 0.9772
300
LSL USL
Frequency
LSL 200
100
8 9 10 11 12
Population
1- .9772 = .0228

Example:
Outside the Spec Limits
The fraction outside the spec limits can be determined

by using Minitab functions.
Pr (x≤9) + Pr(x≥11) = Pr(Z≤-2.00) + Pr(Z≥2.00)

= 2.28% + 2.28%  4.56%

Problem
• Suppose we have a process with:
– Mean = 37.5
– Standard Deviation = 7.83
– Specification:
Specification 30 +/- 15
• Determine the total estimated percentage outside
specification for this process
• How many standard deviations is the mean from the
upper spec limit?

The Z-Transform and
Estimating Capability
SL- 
Z 
 T(Target)
=
USL  (Mean)
SL =
LSL
SL - 
Z 
  ST
ST (Short Term)
LT (Long Term)  LT
Revision History
Rev. Name Date Course Changes Name

A 5/2/03 MS Lean Sigma First Draft Kristine Nissen
B 11/19/04 Lean Six Sigma Minitab 14 Updates Ausifa Munshi
C 9/12/08 OPS LSGB GS Switched to 1page GB roadmap Ian Wedgwood
D 1/31/09 LSS GBO Initial IR Release Daryl Holley


118 LSS BBO3 - Process Capability PDF

Uploaded by

Copyright:

Available Formats

118 LSS BBO3 - Process Capability PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

118 LSS BBO3 - Process Capability PDF

Uploaded by

Copyright:

Available Formats

Lean Six Sigma Operations

Black Belt Bridge – Week 1

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

We continually gather data on our process and ask:

Gathering and analyzing the data is called....

Where does it come from?

CCpk accounts for process centering and spread.

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

Near-Term Goal Move the Ppk to Pp (Center the process)

Move the Pp to Cpk (Reduce

Long-Term Goal Move Cpk to Cp (Random Variation)

Six Sigma Process

For Ingersoll Rand Internal Use Only

17-15 Rev. C Printed 3/23/2009

Histogramof True Position

For Ingersoll Rand Internal Use Only

0.0 2.4 4.8 7.2 9.6 12.0 14.4

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

0.0 2.4 4.8 7.2 9.6 12.0 14.4

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

17-22 Rev. C Printed 3/23/2009

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

Box-Cox Plot of True Position 2

Best Value 0.500000

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

0.0 0.6 1.2 1.8 2.4 3.0

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

17-36 Rev. C Printed 3/23/2009

• Therefore: USL  LSL

 USL  X Obs X Obs  LSL  Be

• Note: See detailed graph to follow

For Ingersoll Rand Internal Use Only

For Ingersoll Rand Internal Use Only

Process %R&R Cpo Which do we work on?

Process 2: May Benefit more by addressing Measurement

*Process 4: Would improving %R&R really be worth the effort ?

For Ingersoll Rand Internal Use Only

17-42 Rev. C Printed 3/23/2009

Centering – Put The Process On Target

Spread – Reduce The Variation

For Ingersoll Rand Internal Use Only