Lean Six Sigma Green Belt

Certification Course
Measurement System Analysis
 Learning Objectives

By the end of this lesson, you will be able to:

Interpret measurement system analysis

List the measurement system characteristics and system concepts

Outline the measurement system properties

Describe G R&R study

 Problems at Java House Coffee

Java House Coffee customers had been complaining about the coffee
temperature. However, investigation showed that the temperature was
within the customers’ desired range.

 The investigation team discovered that the device used to take and read
coffee temperature was malfunctioning. A new device showed coffee
temperature was colder than what they had previously thought.
Calculate and Interpret Measurement System Capability using GR&R
 Measurement System Analysis

Analysis Intended Results

Measurement System Analysis is a technique that identifies measurement error or variation and its
sources to reduce variation.
Measurement System Analysis

Obtain information

Establish criteria

Compare measurement methods

Form basis for evaluating a measurement method

 Measurement System Analysis

Observed value = True value ± Measurement error

 With Parallax Error

 No Parallax Error

 With Parallax Error

 Measurement System Analysis

True variability = Process variability + Measurement variability

 Measurement System Properties

System
Characteristics

Accuracy Precision

μtotal = μprocess ± μmeasurement σ2total = σ2process + σ2measurement

To have stable measurement system, it is important to focus on the accuracy, first by addressing
measurement issues and then getting accurate results.
 Measurement System Properties

Accuracy Precision

Degree of conformity of Degree to which repeated

measured or calculated measurements, under
value to its actual or true unchanged conditions,
value show the same results
 Measurement System Properties

Accurate Precise Not Accurate Precise

Accurate Not Precise Not Accurate Not Precise

Understand Measurement System Characteristics
 Measurement System Characteristics

Measurement System Characteristics

Resolution Accuracy Precision

Bias Repeatability

Linearity Reproducibility

Stability
 Measurement Characteristics: Resolution

Resolution is the smallest reading the measurement scale can detect.

Resolution is the first characteristic that needs to be tested in an MSA.

 Measurement Characteristics: Bias

Bias is the measure of the distance between the measured value and the true or actual value.

Possible causes for bias are need for equipment calibration, improper calibration, or the
wrong gage used for the job.
 Measurement Characteristics: Linearity

Linearity is a measure of consistency of bias over a range of measurement from the

smallest number to a higher number and vice-versa.

Range of the gage

Possible causes for linearity issues could be poor maintenance, worn instrument or
equipment, or application measurement method.
 Measurement Characteristics: Stability

Stability refers to the ability of a measurement system to show the same values,
every time the same item is measured.

Possible causes for instability issues are normal aging and worn instrument.
 Measurement Characteristics: Repeatability

Repeatability is the variation in measurements obtained with one measurement instrument, when it is used
several times by one appraiser while measuring the same characteristic on the same part, repeatedly.

Possible causes for repeatability variation could be appraiser technique, position

on part where measurement is taken, and lack of training.
 Measurement Characteristics: Reproducibility

Reproducibility is variation in the average of measurements made by different operators

using the same gage when measuring identical characteristics of the same part.

Possible causes for reproducibility variation could be difference in

operator training, technique, or experience.
Measurement System Concepts
 Measurement Correlation

Measurement Correlation is the strength of relationship of multiple measurement systems for

use by quantifying accuracy relative to a Gold Standard or Gold Tester.

A Gold Standard is the measurement system used to determine and assign specifications values.
 Precision / Tolerance (P/T)

Precision/Tolerance (P/T) addresses what percent of the tolerance is

taken up by measurement error.

5.15 standard deviation is chosen as an

industry standard and accounts for 99%
of MS variation.

P/T is a ratio that is Tolerance is the difference

usually expressed as a between the upper and lower
percentage. customer specification limit.

A measurement system is
acceptable if P/T is less than 30%.
 GAGE R&R and %GRR

Gage RR or Gage Repeatability and Reproducibility is a

measure of measurement system variation.

Percent Gage R and R is a measurement system

capability index.
 Sources of Variation

Long-Term Process
Variation

Short-Term Process
Actual Process Variation
Variation
Observed Process
Variation Variation within a sample
Measurement
Variation
Variation due to operators Reproducibility

Variation due to gage Repeatability

 GRR Study: Components

Difference leads to
Reproducibility

1 2 1 2 1 2
Trial 3 4 3 4 3 4
Reading #1
5 6 5 6 5 6

Difference Six Parts /

leads to Conditions
Reproducibility

1 2 1 2 1 2
Trial 3 4 3 4 3 4
Reading #2
5 6 5 6 5 6

Operator A Operator B Operator C

 GRR Study: Guidelines

GRR studies should be performed over the range of expected

observations

Actual equipment should be used

Written procedures or approved practices should be followed

Measurement variability should be presented as is

Measurement variability should be separated into causal

components, prioritized, and targeted for action
 GRR Study: Variable and Attribute Gage R&R

Variable Gage R&R Attribute Gage R&R

Analyzes measurement systems with the help Analyzes measurement systems using
of variable or continuous data attribute or discrete data

Measurement system typically Measurement system typically utilizes

involves a physical gage. manual or automated
counting/monitoring.
The result of this is quantification of
the percentage of variation The result of this is quantification of
contributed by the measurement the proportion of defective
system. measurements in DPMO, %
Agreement, or Sigma Level.
 Variable Gage R&R: Data Collection

The number of operators are usually 3.

The number of units to measure is usually 10.

General sampling techniques are used to represent the population.

The number of trials for each operator is 2 to 3.

The gage is checked for calibration and resolution.

The units are measured by the first operator in random order, and the same order is
followed by the other operators.

Each trial is repeated.

 Variable Gage R&R: Four Criteria

0% 10% 20% 30% 40%  100%

% Study Variation

% Tolerance

0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%  100%
% Contribution

1 2 3 4 5 6 7 8 9 10 
% Distinct Categories

It is important to assess these values to determine whether each value meets the requirement
to ensure that the measurement system can reliably measure the process.
 Variable Gage R&R

Average and Range Method ANOVA Method

• ANOVA is considered the best method for

• For years, the most used method for Gage
analyzing GRR studies due to the following
R&R was the Average and Range Method
reasons:
o It separates equipment and operator
• Many people refer to the AIAG’s
variation and provides insight on the
Measurement Systems Analysis Manual
combined effect of the two.
when doing Gage R&R studies
o It uses variance instead of standard
deviation as a measure of variation
o AIAG stands for Automotive Industry
and therefore gives a better estimate
Action Group
of the measurement system
variation.
• Calculates the % of the total variation, using
standard deviation, that is due to the Gage
• Allows you to estimate the variance of the
R&R
components
Variable Gage R&R: Continuous Gage R&R Example
Variable Gage R&R: Continuous Gage R&R Example
Variable Gage R&R: Continuous Gage R&R Example
 Variable Gage R&R: Continuous Gage R&R Example

ndc = 2 ndc = 3
 Variable Gage R&R: Interpretation

Check the value of %GRR.

If %GRR < 30, Gage Variation is acceptable, and thus the gage is acceptable. If %GRR > 30, the gage is not
acceptable.

Check EV first.
If EV = 0, the MS is reliable and the variation in the gage is contributed by different operators. If AV = 0, the MS
is precise.

If EV = 0, resolve AV by providing operators with training.

 Attribute Data: Data Collection in GR&R

Select between 20 to 30 samples that represent full range of variation

50:50 mix of good/bad parts is recommended

Have master appraiser categorize each of the test sample as good or bad

Take two or three inspectors and have them assess each test sample

Randomize samples and repeat assessment

 Attribute Data: Example

Test Samples Master/Expert Operator1_1 Operator1_2 Operator2_1 Operator2_2 Operator3_1 Operator3_2

1 okay okay okay okay okay okay okay
2 defective defective defective defective defective defective defective
3 okay okay defective okay okay okay
4 okay okay okay okay okay okay okay
5 okay okay defective okay okay okay okay
6 okay okay okay okay okay okay okay
7 okay okay okay okay okay okay okay
22 defective defective defective defective defective defective defective
23 defective defective defective defective defective defective defective
24 okay okay okay okay okay okay okay
25 okay defective defective defective defective defective defective
26 defective defective defective defective defective defective defective
27 okay okay okay okay okay okay okay
28 okay okay okay okay okay okay okay
29 okay defective defective okay okay okay defective
30 okay defective okay okay okay okay okay
 Attribute Data: Example

Op1 with Between All Ops versus

Operator1_1 Operator1_2 Within Op1 Test Samples Master/Expert Operator1_1
Standard Ops Match Standard
okay okay 1 1 1 okay okay 1 1
defective defective 1 1 2 defective defective 1 1
okay defective 0 0 3 okay okay 0 0
okay okay 1 1 4 okay okay 1 1
okay defective 0 0 5 okay okay 0 0
okay okay 1 1 6 okay okay 1 1
okay okay 1 1 7 okay okay 1 1
defective defective 1 1 22 defective defective 1 1
defective defective 1 1 23 defective defective 1 1
okay okay 1 1 24 okay okay 1 1
defective defective 1 0 25 okay defective 1 0
defective defective 1 1 26 defective defective 1 1
okay okay 1 1 27 okay okay 1 1
okay okay 1 1 28 okay okay 1 1
defective defective 1 0 29 okay defective 0 0
defective okay 0 0 30 okay defective 0 0
#Matched 27 27 25 24
#Inspected 30 30 30 30
%Agree 90.00% 80.00% 83.33% 80.00%
Attribute Data: Example
 Interpretation of MSA

Operators are not adequately

trained in using the gage

Calibrations on the gage dial are

not clear
Reproducibility Error > Repeatability Error

MSA Result Gage needs maintenance

Repeatability Error > Reproducibility Error

Gage needs redesign to be more
rigid

Gaging location needs

improvement

Ambiguity is present in SOPs

 Key Takeaways

A Measurement System Analysis (MSA) is needed prior to any

data collection activities.

There are two major system characteristics: accuracy and

precision.

An MSA will help determine if the measurement system

characteristics exist.

The measurement system concepts are Measurement

correlation, gage RR and % GRR, and Precision/Tolerance (P/T).

Gage R&R studies can be conducted for variable and attribute

data.

Average and Range Method and ANOVA Method are the two
methods for analyzing GRR Studies.
Knowledge Check
Knowledge
Check A team studies the results from measurement system analysis of plate thickness and
determines that the average values observed is 3.5 cm, with values close together, and
1 the true value is 4.0 cm. How can the measurement system be described?

A. Low Accuracy, High Precision

B. Low Accuracy, Low Precision

C. High Accuracy, High Precision

D. High Accuracy, Low Precision

 Knowledge
Check A team studies the results from measurement system analysis of plate thickness and
determines that the average values observed is 3.5 cm, with values close together, and
1 the true value is 4.0 cm. How can the measurement system be described?

A. Low Accuracy, High Precision

B. Low Accuracy, Low Precision

C. High Accuracy, High Precision

D. High Accuracy, Low Precision

The correct answer is A

The scenario described here is of low accuracy (average observed value does not equal target value) and high
precision (the values are close together).
Knowledge
Check The change or lack of change of a measurement system readings of the same part over
2 time is an example of ________.

A. Linearity

B. Stability

C. Repeatability

D. Reproducibility
 Knowledge
Check The change or lack of change of a measurement system readings of the same part over
2 time is an example of ________.

A. Linearity

B. Stability

C. Repeatability

D. Reproducibility

The correct answer is B

The change or lack of change of a measurement system readings of the same part over time is an example of stability.
Knowledge
Check A team has measured the thickness of a part over and over and has compared the
average observed value to the actual value. The difference between the average
3 observed value and the actual value is called:

A. Linearity

B. Bias

C. Stability

D. Gage R&R
 Knowledge
Check A team has measured the thickness of a part over and over and has compared the
average observed value to the actual value. The difference between the average
3 observed value and the actual value is called:

A. Linearity

B. Bias

C. Stability

D. Gage R&R

The correct answer is B

Bias is the measurement characteristic described as the difference between the average observed value and a
reference value.
Knowledge
Check What is the major difference between the ANOVA and Average and Range method for
4 Gage R&R study?

A. The results are significantly different

B. ANOVA is based on standard deviation and the Average and Range method is based on
variance

C. There is no difference

D. ANOVA is based on variance and the Average and Range method is based on standard
deviation
 Knowledge
Check What is the major difference between the ANOVA and Average and Range method for
4 Gage R&R study?

A. The results are significantly different

B. ANOVA is based on standard deviation and the Average and Range method is based on
variance

C. There is no difference

D. ANOVA is based on variance and the Average and Range method is based on standard
deviation

The correct answer is D

ANOVA is based on variance and the Average and Range method is based on standard deviation which causes a slight
difference in calculations.
Knowledge
Check
To conduct a continuous Gage R&R study, how many sample parts are needed?
5

A. 10

B. 20

C. 25

D. 3
 Knowledge
Check
To conduct a continuous Gage R&R study, how many sample parts are needed?
5

A. 10

B. 20

C. 25

D. 3

The correct answer is A

For a continuous Gage R&R study, ideally 10 parts are recommended that represent the range of output possibilities
from the process.