Short Run SPC Presentation - ASQ | Portland Section 0607

By Ed Landauer, C.Q.E., P.E.
November 13, 2012
Short Run SPC
 What it is
 What it isn’t
 SPC is not intended to be a substitute for
inspection and testing
 Programs such as JIT make Short Run SPC
more common
What Constitutes a Short Run?
 A run of 25 parts?
 A run of 100 parts?
 A run of 1000 parts?
 A run of 50,000 parts?
 A single part?
General Categories of Short Runs
 Not enough parts in a single production run to
develop or maintain control limits on the
 Process cycles are so short that production runs
are over before sufficient data can be gathered
 Many different parts are made for many
different customers
 Examples
Example: CNC Machine
 Extremely complex machine
 Has dozens of different tools
 That produce hundreds of different parts
 With thousands of characteristics
Example: CNC Machine- Generic
 Select a tool
 Position a tool
 Rotate a tool
 Move the part
 Remove the metal
Example: CNC Machine
 Nearly all problems encountered after the
initial set-up involve the ability of the
machine to position the tool precisely
 For example, the location of a single hole
provides information on the location of the
tool in both the x and y directions
Sample Set of Data
 Twelve subgroups- 3 observations per subgroup
5.0037 5.0009 4.9973 5.9974 6.0021 5.9981
5.0015 5.0023 5.0029 5.9990 5.9983 6.0018
4.9989 4.9981 4.9985 6.0015 5.9977 6.0023
7.0014 6.9991 7.0003 6.0031 5.9987 5.9987
7.0009 6.9979 6.9975 5.9985 6.0011 6.0021
6.9979 6.9995 7.0021 6.0017 6.0021 6.0001
Method 1: Target Average and
Range Charts
 Subtract nominal (or target) value from
the actual measurements
x* = x - nominal
 Can plot several parts from a given process
on a single chart
 Can plot several features from a single part
on a single chart
Method 1: Restrictions
 Nominal (or target) values must be close to
each other with the same tolerance
 Or the most restrictive tolerance must be
used for all
Method 1: Control Limits
• Calculate averages and ranges of the
deviations from target
• Control limit formulas are the same as for
traditional control charts
Method 2: Standardized Variables
Control Charts
 Good for recurring families of parts
 They are independent of the unit of
 They are scaled so that different
characteristics can be plotted on the same
 They are “true” process control charts
Method 2: Calculations
 Use the normalizing transformations
 Need to use historical averages from
previous data in the calculations
Method 2: Control Limits
 X-bar Chart
 R chart
-A2 and A2
D3 and D4
Short Run Attribute Chart- c-chart
 Job shop welding process
 Produces small quantities of single order
only parts
 Operation always involves joining parts of
similar material and size
 Can plot chart of weld imperfections per 100
inches of weld
Short Run Attribute Chart- p-chart
 Job shop welding process
 Similar operation to previous slide
 Plot chart of the number of weld
imperfections per total welds per 100 inches
of weld
 Only minor change in mechanics to move
from traditional SPC to Short Run SPC
 The biggest difference in the perception of
Statistical Process Control rather than
Statistical Product Control
Thank You!
[email protected]

similar documents