L. Goch

1
L. Goch February 2011
SPC MSA Using Minitab
2
Agenda

• Continuous Attribute Data
• IMR Charts
• Xbar R Charts
• Xbar S Charts
• MSAs
• Using Xbar S Charts for MSAs
• Non-Destructive Variable MSAs
• (NOTE Minitab will also Analyze Destructive
  Variable MSAs Attribute MSAs)

3
The Basic Control ChartKey Components
Control Limits are NOT Spec Limits
UCL
Plotted Data
DATA PLOTTED OVER TIME
MONITORED CHARACTERISTIC
Center Line
LCL
UCL Upper Control Limit / LCL
Lower Control Limit
4
Common Causes vs Special Causes
5
Individual-X Moving Range Chart
6
Control ChartsIndividual /Moving Range

• Use
• When only have 1 measurement per time period
• When want to chart Counts.
• When want to chart Percentages (s).
• Do NOT use P-Charts if the values are lt10 or
  subgroup sample size gt100 since it violates basic
  P-Chart assumptions.
• Need a minimum of 10 data points
• Variation
• Short Term Measures relatively rapid changes
  over time (Moving Range chart)
• Long Term Measures relatively gradual changes
  over time(Individuals chart)

Charts are Based on a Subgroup Size of 1
7
I / MR Chart Stat gt Control Charts gt I-MR

• I-MR Chart analyzes individual data over time.

Open worksheet Exh_QC.mtw
8
I / MR Chart Stat gt Control Charts gt Run Chart
Rule 1
9
Minitab ExerciseIndividuals / Moving Range Output
Is there a shift at Observation 25?
10
Minitab ExerciseIndividuals / Moving Range Output
Group Column Added
11
Minitab ExerciseIndividuals / Moving Range Output
Press ltCtlgt e
12
Minitab ExerciseIndividuals / Moving Range Output
13
Xbar R Chart
14
Control ChartsX Bar / R Chart

• Advantages
• Sensitive to changes in data
• Limitation
• Unlike the Individual Chart, the Xbar Chart
  control limits do not represent the overall data
  range. They represent the range of the subgroup
  averages.

15
Control ChartsX-bar / R Control Charts

• GATHERING DATA
• Define rational basis for subgrouping
• Select sample size for the subgroups
• Select sampling frequency
• Need a minimum of 10 subgroups
• VARIATION
• Short Term Measures the variation within the
  Subgroups based on the subgroup range (Range
  chart)
• Long Term Measures the variation between the
  Subgroups based on the subgroup average (X Bar
  chart)

16
Xbar-R Charts Stat gt Control Charts gt Variables
Charts for Subgroups gt Xbar-R
Open worksheet Camshaft.mtw
17
Minitab Exercise - XBar / R Charts Output
What do the Charts tell us? We need to look at
the individual suppliers.
18
Minitab Exercise - XBar / R Charts

• Stack Data
• Stack the 2 Suppliers so the data can be looked
  at together on the same chart

19
Minitab Exercise - XBar / R Charts
20
Xbar S Chart
21
Control ChartsX Bar / S Chart

• Advantages
• Very sensitive to changes in data.
• Also, the center line of the S-Chart provides a
  good estimate for the datas standard deviation.
• Limitation
• The Xbar Chart control limits do not represent
  the overall data range. They represent the range
  of the subgroup averages.

22
Control ChartsX-bar / s Control Charts

• GATHERING DATA
• Define rational basis for subgrouping
• An S-Chart is typically used when the subgroups
  sample size is gt 9
• Select sampling frequency
• Need a minimum of 10 subgroups
• VARIATION
• Short Term Measures the variation within the
  Subgroups based on the subgroup stddev (S chart)
• Long Term Measures the variation between the
  Subgroups based on the subgroup average (X Bar
  chart)

23
Xbar-S Charts Stat gt Control Charts gt Variables
Charts for Subgroups gt Xbar-S
Open worksheet Bloodsugar.mtw
24
Minitab Exercise - XBar / S Charts Output
What do the Charts tell us? We need to look at
the individual people.
25
Minitab Exercise - XBar / S Charts

• Sort Data
• Sort the data by SubjectID Reading

26
Minitab Exercise I-MR Charts
27
Xbar S Charts used for MSAs
28
MSA X Bar / S Chart

• Advantages
• Pictorial look at MSA results
• Measurement error can easily be calculated
• RR and P/T can easily be calculated
• Daily Calibration Data can be used
• Limitation
• Will NOT detect measurement that varies by
  nominal value (e.g. large parts having more or
  less measurement error than small parts)

29
MSA X-bar / S Chart

• GATHERING DATA
• Need Multiple measurements per time period
• Can use 1 or more Parts
• Need a minimum of 10 subgroups
• VARIATION
• S Chart Displays the Measurement Error. A good
  measurement system will be in-control
• Xbar Chart Displays any Drift over time in the
  average measurements. A good measurement system
  will be in-control for each individual Part.

30
MSA X-bar / S Chart
Open worksheet Daily Calibration Data.mtw

• Multiple Measurements per Day on most Days.
• 1 Calibration Part per Facility
• 3 Facilities
• Sorted by Facility Date Time

31
Xbar-S Charts Stat gt Control Charts gt Variables
Charts for Subgroups gt Xbar-S
Open worksheet Daily Calibration Data.mtw
32
MSA X-bar / S Chart output
Need to add the Measurement Error.
33
MSA X-bar / S Chart Results
Facilitly Sbar Multiplier Meas Error P/T
DHM 31.0 2.575 /- 80 80/50 160
DSM 27.5 2.575 /- 71 71/50 142
GSO 8.0 2.575 /- 21 21/50 42

• Comments
• Spec is /- 50.
• P/T target is lt 30
• None of Facilities meet the measurement error
  target
• Daily Calibration Check procedures need
  investigated since producing dramatically
  different results

34
Variables MSA (Non-Destructive Testing)
35
Variables MSA Stat gt Quality Toolsgt Gage Study
gt Gage RR Study (Crossed)

• Minimum Requirements
• Minimum 10 parts 3 at low end of spec, 4
  parts in middle of Spec, 3 parts at high end of
  spec (NOTE some parts should be out of spec)
• 3 Operators (NOTE if only 1 operator measures
  parts, 1 operator may be used for MSA)
• 3 Repeated Measurements

36
Variables MSA Stat gt Quality Toolsgt Gage Study
gt Gage RR Study (Crossed)
37
Variables MSA Stat gt Quality Toolsgt Gage Study
gt Gage RR Study (Crossed)
38
Variables MSA Graph Results
39
Variables MSA Session Window Results
Both the Part the Operator significantly
effect the part measurements.
Measurement Error /- 0.78 (i.e. 1.55721/2
0.78)
40
SPC/MSA - Summary

• The reliability of Control Charts are dependent
  on the selection of Rational Subgroups. Pick the
  wrong subgroup and the results can be very
  misleading.
• The reliability of MSAs are dependent on the
  selection of the actual Parts. Pick Parts with
  too much or too little variation between them and
  the results can be very misleading.