LECTURE 6

Process and Measurement System Capability

OBJECTIVES
 investigate and analyze process capability using control
charts and histogram
 differentiate between process capability and process
potential
 know how to calculate process capability ratios and
confidence interval on process capability ratios
 learn how to conduct an R& R study
 learn how to estimate the natural limits of a process from
a sample of data

2
PROCESS CAPABILITY
 Process capability refers to the uniformity of the
process
 The variability of critical-to-quality characteristics
in the process is a measure of the uniformity of output
 Two types of variability:
 Natural or inherent variability (instantaneous)
 Variability over time

 Assume that a process involves a quality characteristic

that follows a normal distribution with mean , and
standard deviation, . The upper and lower natural
tolerance limits of the process are

UNTL =  + 3
LNTL =  - 3
3
PROCESS CAPABILITY

Process capability analysis is an engineering study

to estimate process capability
 In a product characterization study, the distribution
of the quality characteristic is estimated ...( mean ,
and standard deviation)
usually measures functional
 Process capability study
parameters or critical-to-quality characteristics on
the product , not the process itself

4
PROCESS CAPABILITY
Major uses of data from a process capability
analysis

 Predicting how well the process will hold the

tolerances
 Assisting product developers/designers in selecting or
modifying a process
 Assisting in establishing an interval between
sampling for process monitoring
 Specifying performance requirements for new
equipment
 Selecting between competing vendors
 Planning the sequence of production processes when
there is an interactive effect of processes on
tolerances 5
REASONS FOR POOR PROCESS
CAPABILITY
 Process may have good potential capability

Some reasons for poor process capability; (a) Poor process centering
(b) Excess process variability

6
PROCESS CAPABILITY

Techniques used in process capability analysis

 Histograms or probability plots

 Control Charts
 Designed Experiments

7
USING A HISTOGRAM OR A
PROBABILITY PLOT
Using a Histogram
 

 The histogram along with the sample mean and

sample standard deviation provides information about
process capability
 The process capability can be estimated
as

 The shape of the histogram can be determined

(such as if it follows a normal distribution)
 Histograms provide immediate, visual
impression of process performance
 Histograms do not provide any information
about the state of statistical control of the
process 8
PROCESS CAPABILITY RATIOS
Use
  and Interpretation of

USL  LSL
Cp 
6σ
where LSL and USL are the lower and upper
specification limits, respectively
The estimate 𝐶^of
 𝑝 is given by

ˆ USL  LSL
Cp 
6ˆ
Where the estimate can be calculated using
the sample standard deviation, s, or

9
USE AND INTERPRETATION OF
CP
One-Sided Specifications
 

USL  
C pu 
3
  LSL
C pl 
3

10
USE AND INTERPRETATION OF CP

 The ppm quantities in T

the table were calculated using
the following important
assumptions:
1. The quality characteristic
has a normal distribution
2. The process is in
statistical control
3. (In case of 2 side spec.)
the process mean is
centered between lower
and upper spec. limits

11
USE AND INTERPRETATION OF
CP
Recommended Minimum Values of the process Capability
Ratio

12
PROCESS CAPABILITY RATIO AN
OFF-CENTER PROCESS
  does not take into account where the process mean is
located relative to the specifications, it is a measure of
potential capability not actual capability
 A process capability ratio that does take into
account centering is
defined as
= min()

  Relationship of and

13
A MEASURE OF ACTUAL
CAPABILITY
   simplyis the one-sided PCR for the specification
limits nearest to the process average
 If the process is centered at the midpoint of the
specifications
 If < the process is off center
 should not be used alone as a measure of process
centering
 depends inversely on  and becomes large as 
approaches zero. (That is, a large value of does not
necessarily reveal anything about the location of the
mean in the interval (LSL, USL)

14
MORE ABOUT PROCESS
CENTERING
  
An improved capability ratio to
measure process centering is
USL  LSL
C pm 
6
where  is the square root of
expected squared deviation from
target: 2
 E  2 2 2
x T     (  T )

Where T =½(USL+LSL)
15
MORE ABOUT PROCESS
CENTERING
  
can be rewritten another way:
USL  LSL
C pm 
6  2  (  T) 2
Cp

1 2
where T



16
EXAMPLE
  Using the following data:
5.32 5.14 5.97 5.9 5.16
6.2 4.46 5.04 4.17 5.47
5.64 4.71 3.74 5.24 5.02
6.11 5.48 4.99 5.41 4.75
4.45 5.31 5.49 5.26 5.04

 If the process target is set at 5 grams:

 Estimate the
 Estimate

17
 SOLUTIONS
 Sample Mean = 5.18
 Sample s.d. = 0.59

USL  LSL 74

C pm  
6  2  (   T )2 6 0.59   5.18  5 
2 2

 0.814
C pk 0.67
C pkm    0.64
2 2
  T   5.18  5 
1   1  
    0.59 

18
CONFIDENCE INTERVALS AND TESTS
ON PROCESS CAPABILITY RATIOS
 

 is a point estimate for the true , and subject to

variability. A 100(1-) percent confidence
interval on is

 2
1  / 2 , n 1  2
 / 2 , n 1
Ĉ p  C p  Ĉ p
n 1 n 1

19
CONFIDENCE INTERVALS AND TESTS ON
PROCESS CAPABILITY RATIOS

 

 is a point estimate for the true , and subject to

variability. An approximate 100(1-) percent
confidence interval on is

 1 1   1 1 
Ĉ pk 1  Z / 2    C pk  Ĉ pk 1  Z / 2  
 9nĈ pk 2(n  1)   9nĈ pk 2(n  1) 

20
PROCESS CAPABILITY ANALYSIS
USING A CONTROL CHART
  If a process exhibits statistical control, then the
process capability analysis can be conducted
 A process can exhibit statistical control, but may
not be capable
 PCRs can be calculated using the process mean
and process standard deviation estimates
 When dealing with attributes data, defect per unit
(DPU) statistic is used as a measure of capability

* Large samples need tobe used to obtain reliable estimate

21
PROCESS CAPABILITY ANALYSIS
DESIGNED EXPERIMENTS
 Systematic approach to varying the variables
believed to be influential on the process, (Factors that
are necessary for the development of a product)
 Designed experiments can determine the sources of
variability in the process

22
SETTING SPECIFICATIONS ON
DISCRETE COMPONENTS
Components interact with other components
Complex assemblies
Tolerance stack-up problems
Linear combinations
Nonlinear combinations

23
SETTING SPECIFICATIONS ON
DISCRETE COMPONENTS
Linear
  Combination
 when an item is a linear combination of the dimension
of component parts then the dimension of the final
assembly is
y

 where is constant -A- -B- -C-

 If the are normally and independently distributed with
mean
and variance then is normally distributed with mean
and variance
 If and are known for each component, the fraction of
assembled items falling outside the specification can be
determined
24
 TOLERANCES ON DISCRETE
COMPONENTS
 The tolerance for each component is
A±3A
B±3B
C±3C
 The tolerance of L is y
y±3y where
-A- -B- -C-

Note that
3y<3(A+B+C)   2
𝜎 𝑦 =𝜎 2𝐴 +𝜎 2𝐵 +𝜎 2𝐶

25
EXAMPLE
 An assembled product is composed of
three components, the length in cm of the
components have the following means and
standard deviations.
Component Mean Sigma
A 5 .02
B 3.5 .01
C 4 .03

 Give the Natural Tolerance Limits on each of the

components.
 Calculate the Natural Tolerance Limits for the
length (Y = A+B+C) of the assembled product.

26
 SOLUTION
 Component Tolerance Limits:
=Mean +/- 3 Sigma:

Component Mean Sigma 3Sigma LNTL UNTL

A 5 0.02 0.06 4.94 5.06
B 3.5 0.01 0.03 3.47 3.53
C 4 0.03 0.09 3.91 4.09

 Tolerance of Assembled Product:

= L±3L :
L=12.5,
 L2   A2   B2   C2  .0004  .0001  .0009
L=.0374
Tolerance limits for Length=12.5 +/- 3(.0374)
= (12.388,12.612)

27
SETTING SPECIFICATIONS ON
DISCRETE COMPONENTS
 Vv

28
ESTIMATING THE NATURAL
TOLERANCE LIMITS OF A PROCESS

 When X is normally distributed, the Normal Tolerance

Limits of a process are given by:
 ± Z/2

29
TOLERANCE ANALYSIS
  In practice the mean and standard deviation of the
population are not known, but must be estimated
 and , can be obtained by
 Estimates of
and s from the sample data
 The tolerance limits are then determined
by

X  Ks
Where the factor K is a function of the confidence level
desired, the percentage of the population included within
the limits, and the sample size.

30
 EXAMPLE
 A random sample of n=25 observations gives:

X  40.75
S  1.37
 What tolerance limits contain 99% of all values with
95% confidence?

31
EXAMPLE
 From a table of tolerance factors , we find K = 2.972
 Tolerance Limits:

X  Ks  40.75  (2.972)1.37
 (36.67, 44.82)
Interpretation : 99% of all values are contained within
the calculated tolerance limits with 95% confidence.

32
 MEASUREMENT
SYSTEM
ANALYSIS
A review of statistical procedures for use in test
method development and validation
 GAGE AND MEASUREMENT
SYSTEM CAPABILITY STUDIES
 Two portions of total variability:
 product variability which is that variability that is inherent
to the product itself
 gage variability or measurement variability which is the
variability due to measurement error

    gage
2 2 2
Total product

34
MEASUREMENT SYSTEM
ERROR
 Measurement System Error can be classified into two
categories: Accuracy and Precision
 Accuracy describes the difference between the
measurement and the part’s actual value
 Precision describes the variation observed when the same
sample is measured repeatedly with the same device

35
PRECISION VS. ACCURACY
 The accuracy of a measurement system is usually
broken down into three components:
 Bias
 Linearity
 Stability
 The precision, or measurement variation can be broken
down into two categories:
 Repeatability
 Reproducibility

36
ACCURACY
 Bias is the difference between the
observed average of measurements and
the reference value
 Linearity is the difference in bias values
through the expected operating range of
the gage
 Stability (or drift) is the total variation
in the measurement system on the
same samples when measuring a single
characteristic over an extended period
of time
37
PRECISION

Repeatability
 is the variation in measurements obtained within a
single laboratory with one measurement instrument
when used several times by an appraiser while
measuring the identical characteristic on the same
sample.
Reproducibility
 deals with the variation in the average of the
measurements made between different labs,
appraisers, days, etc. when measuring the identical
characteristic on the same sample.

38
ANALYSIS OF MEASUREMENT
SYSTEMS

 For non destructive tests, measures of precision are

typically estimated by a method called Gauge
Repeatability and Reproducibility, Gauge R&R
 Gauge R&R studies can be performed using a number of
different techniques
 Range Method
 Average and Range Method
 Analysis of Variance (ANOVA) Method

39
CONTROL CHARTS AND TABULAR
METHODS
  and R Charts
 The variability seen on the chart can be interpreted
as that due to the ability of the gage to distinguish
between units of the product

 The variability seen on the R chart can be interpreted

as the variability due to operator

40
CONTROL CHARTS AND TABULAR
METHODS
Precision to Tolerance (P/T) Ratio
 An estimate of the standard deviation for
measurement error is
R
ˆ gage 
d2
 The P/T ratio is
6ˆ gage
P/T 
USL  LSL

41
CONTROL CHARTS AND
TABULAR METHODS
 Total variability can be estimated using the sample
variance. An estimate of product variability can be
found using

 2
Total  2
product  2
gage

S   product   gage
2
ˆ 2
ˆ 2

ˆ 2product  S2  ˆ gage
2

42
CONTROL CHARTS AND
TABULAR METHODS
Percentage of Product Characteristic Variability
 A statistic for process variability that
does not depend on the specifications
limits is the percentage of product
characteristic variability:

ˆ gage
 100

ˆ product

43
CONTROL CHARTS AND TABULAR
METHODS
Gauge R&R Studies
 Gauge repeatability and reproducibility
(R&R) studies involve breaking the total
gauge variability into two portions:
 repeatability which is the basic inherent precision of
the gauge
 reproducibility is the variability due to different
operators using the gauge

44
CONTROL CHARTS AND
TABULAR METHODS
Gage R&R Studies
 Gage variability can be broken down as
  𝟐
𝝈 𝒎𝒆𝒂𝒔𝒖𝒓𝒆𝒎𝒆𝒏𝒕 ¿𝒆𝒓𝒓𝒐𝒓 =𝝈 𝟐𝒈𝒂𝒈𝒆 =𝝈 𝟐𝒓𝒆𝒑𝒓𝒐𝒅𝒖𝒄𝒊𝒃𝒊𝒍𝒊𝒕𝒚 + 𝝈 𝟐𝒓𝒆𝒑𝒆𝒂𝒕𝒂𝒃𝒊𝒍𝒊𝒕𝒚
 More than one operator (or different
conditions) would be needed to conduct
the gage R&R study

45
CONTROL CHARTS AND TABULAR
METHODS
Statistics for Gauge R&R Studies (The Tabular
Method)
 Say there are p operators in the study
 The standard deviation due to repeatability can
be found as R
ˆ Re peatability 
d2

R1  R 2    R p
R
p

where
and d2 is based on the # of observations per part per operator

46
CONTROL CHARTS AND TABULAR
METHODS
Statistics for Gauge R&R Studies (the Tabular
Method)
 The standard deviation for reproducibility is given as
  𝑅 ¯¯𝑥
𝜎^ 𝑟𝑒𝑝𝑟𝑜𝑑𝑢𝑐𝑖𝑏𝑖𝑙𝑖𝑡𝑦 =
𝑑2
R x  x max  x min
x max  max( x1 , x 2 , x p )
x min  min( x1 , x 2 , x p )

where d2 is based on the number of operators, p

47
EXAMPLE

In a study to isolate both gauge repeatability and gauge

reproducibility, two operators use the same gauge to
measure ten parts three times each. Estimate gauge
repeatability and reproducibility
operator 1 operator 2
part meas 1 meas 2 meas 3 xbar R meas 1 meas 2 meas 3 xbar R
1 50 49 50 49.67 1 50 48 51 49.67 3
2 52 52 51 51.67 1 51 51 51 51.00 0
3 53 50 50 51.00 3 54 52 51 52.33 3
4 49 51 50 50.00 2 48 50 51 49.67 3
5 48 49 48 48.33 1 48 49 48 48.33 1
6 52 50 50 50.67 2 52 50 50 50.67 2
7 51 51 51 51.00 0 51 50 50 50.33 1
8 52 50 49 50.33 3 53 48 50 50.33 5
9 50 51 50 50.33 1 51 48 49 49.33 3
10 47 46 49 47.33 3 46 47 48 47.00 2

xdoublebar-1 50.03 xdoublebar-2 49.87

Rbar-1 1.70 Rbar-2 2.30
d2 R/d2
R-xdoublebar 0.17 1.128 0.1478 (reproducibility)
Rdoublebar 2.00 1.693 1.1813 (repeatability)

48
MEASUREMENT
DISCRIMINATION
is the capability of the
 Measurement discrimination
measurement systems to reliably measure small
changes of the measured characteristic (also
known as resolution).

 If the round off in measurement exceeds the

actual measurement variation (repeatability),
artificially low estimates of measurement
variability will result

 In this case it may not be an appropriate system

to identify process variation or to quantify sample
characteristic values.

49
ATTRIBUTE INSPECTION
CAPABILITY
 Two kinds of mistakes can be made in attributes
inspections:
 Conforming items mistakenly classified as nonconforming
 Nonconforming items mistakenly classified as conforming

 The rates with which these errors are made are referred
to as the performance characteristics

50