World Class Manufacturing Solutions
Variation Analysis
Part Details
Gage number
Characteristic Part number
Unit of Part Description Gage
Measurement description
Date Gage Least
Target Value count
Gage R&R
Tolerance value
USL Gage Bias and
Study dates and Uncertainity
Shift (%)
LSL
Any other details
Data Grouping and Sample details
Number of Groups (Number of time
blocks x number of streams)
Number of Samples in Each group (It is
preferable to collect 5 samples
continuously from the process so that
the inherent variations are captured
Checking the consistency of part to part variation (Step 7)
Average Range (R-bar) (Round off to one decimal more Samples D3 D4
than the data) 2 0 3.267
Upper Control Limit (UCL) = D4*R-bar (Round off to the 3 0 2.575
4 0 2.282
same decimal as data) 5 0 2.115
6 0 2.004
Lower Control Limit (LCL) = D3*R-bar
Is the Part to Part variation consistent Yes/No
If the Part to Part variation is not consistent, STOP, do not proceed further. Plan for DOE
Is the Range Chart plotted Yes/No
Are the Stratification levels more than 3 Yes/No
If the stratification level is <=3, then the Part to Part variation is very less and the parameter does not require any
monitoring. STOP. Do not proceed further
Are there 7 consequtive points increasing, decreasing and Yes / No If yes write
one side of the Mean range causes (if
possible)
Rev: 1/Jan 2003 Page: 1/4
World Class Manufacturing Solutions
Variation Analysis
If the range contains patterns as described above, STOP. Plan for DOE
Estimating Part to Part variation (Step 9)
Sigma = R-bar/d2 (Round off to one decimal more than Sample d2
data) 2 1.128
3 1.693
6 * Sigma (Round off to the same decimal as data) 4 2.059
5 2.326
6 2.534
Overall Average (Round off to one decimal more than data)
Construction of Histogram (Step 10)
Number of Groups = Sqrt (Number of data points)
Range = Maximum Value - Minimum Value
Class Interval (C.I) = Range/Number of groups
Class Inverval (C.I) after rounding off to the same decimal
as data
Draw the Histogram here
Does the Overall Average lie in the group having maximum
frequency or in the adjacent groups
Yes/No
Is there a gradual decreasing trend in frequency on both
sides of the group having the maximum frequency
Yes/No
Are there two Modes (Two groups having maximum
frequency) and both the groups are distinctly seperated
Yes/No
Rev: 1/Jan 2003 Page: 2/4
World Class Manufacturing Solutions
Variation Analysis
Based on the above analysis, can you conclude that the
histogram is Normal Yes/No
Note: If the Histogram is non-normal, the estimated rejection which we do in the next step may not be correct. Data
transformation is required to correct non-normality
Rejection Analysis Using Sigma (Only for Normal distributions) (Step 11)
Average Z(USL) = (USL - Average)/Sigma
Sigma Z(LSL) = (LSL - Average)/Sigma
Projected Rejection % above USL (Based
6 * Sigma on the Normal table
Projected Rejection % below LSL (Based
USL on the Normal table)
Cpk = Zusl/3,Zlsl/3
LSL
Rejection Analysis if the results above show rejection on one side higher than the other side
Target (Design) Z(USL) = (USL - Target)/Sigma
Sigma Z(LSL) = (LSL - Target)/Sigma
Projected Rejection % above USL (Based
6 * Sigma on the Normal table
Projected Rejection % below LSL (Based
USL on the Normal table)
LSL Cpk = Zusl/3,Zlsl/3
Six Sigma Analysis (Step 13)
6 Sigma <= 75% of Tolerance
6 Sigma >75% of Tolerance and <=100% of tolerance
6 Sigma > Tolerance
Actions decided based on Six Sigma Analysis
6 Sigma > 75% of Tolerance and <=100%
6 Sigma <= 75% of Tolerance of tolerance 6 Sigma > Tolerance
Actual Part to Actual Part to Part Use Average and Range chart for Use DOE to reduce variation
Part variation <= variation > 50% of monitoring, but it has to be done with 100%
50% of tolerance tolerance inspection
Use Pre-control Use Average and Do DOE to reduce variation or to question 100% Inspection
chart for Range chart for the tolerance for that parameter
monitoring monitoring
Remove 100% inspection if done on
this parameter
Calculation of limits for Average if the decision taken is Average and Range monitoring
Rev: 1/Jan 2003 Page: 3/4
World Class Manufacturing Solutions
Variation Analysis
Sample A2
2 1.880
Average 3 1.023
4 0.729
UCL = Average + A2*R-bar 5 0.577
6 0.483
LCL = Average - A2* R-bar
Select which of the following DOE tools will be used if 6 Sigma > Tolerance
DOE Tools selected for Process Improvement
Multi-Vari Analysis
Concentration Chart
Paired Comparison
Component Search
Product/Process Search
Variable Search
Full Factorial
B Vs C Test
Scatter plot for Process optimization
EVOP (Evolutionary Operations) for Process Optimization
Analysis Completed by Reviewed and Approved by Date
Rev: 1/Jan 2003 Page: 4/4
