Business Moment

Whether you think you can or

whether you think you can't,
you're right!
Henry Ford
Fundamentals of Statistical
Process Control.

SPC
 Statistical Process Control

SPC is a tool for obtaining cost effective

quality, achieved by process monitoring
and forecasting techniques which are
used to predict problems before they
occur. (defect prevention)
 Reliability
Where should you spend your money?
Increased Reliability
- +

$
$
Inspect
Prevention
Rework
Corrective
Rework Insp.
Action

Quality
 Detection vs. Prevention
Manufacture Manufacture

Prevent
$
Inspect & Sort Scrap or SPC
Rework

Feedback

Ship
Ship
Old method
New Method
 Elements of Quality

- Zero Defects (Implied or specified)

- Conformance to Specification
- Reliability (life & uptime)
- Functionality (does what’s intended)
- Price (competitive for performance)
- Delivery (received on promise date)
- Service and Support (incl. Warranty)
 Normal Variation
Variable Data
X Individual Measurements,
.5
.5 the same values stack up.
.5
Variable Data is .4data that
.6 can be measured.
.5
.4 .5 .6
.4 .5 .6
.4 .5 .6
.3 .4 .5 .6 .7
.3 .4 .5 .6 .7
.3 .4 .5 .6 .7
.2 .3 .4 .5 .6 .7 .8

-3 -2 -1 +1 +2 +3

68%
95%
99.7%
Six Sigma
 Sample Quantity

0
-1 -1 0+1
0 -1 0 +1 –1 0+1+2
-1 0+1+2
-1 0 +2 -2 -1 0+1 +2 –1
-1 0 +2 -1 0 +2 3 -2 -1 0 +1 +2 +3 3 -2 -1 0 +1 +2+3 -20+1+2
-1 0 +1 +2
-2 -1 0 +1 +2
3 -1 0 +2 3 -2 -1 0 +1 +2 +3 3 -2 -1 0 +1 +2 +3 3 -2 -1 0 +1 +2 +3
3 -2 -1 0 +1 +2 +3

3 5 20 23 40
x It takes approximately
xxx
xxxxx 30 samples to assure
xxxxxxx the normal bell shape.
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxx
102
 Any Measurement!

Range height 76-60 = 16 in. Range of weight could be 300-60 = 240 lb

Mean (Center) height = 68 in. Mean weight = 180 lbs
 Increment of Measurement
Takes more
Fewer points points to
make bell shape define bell
shape.

.0# .00#

.017
.016
.018
.017
.005
.019
.001
.018
.002
.016
.001

.005 range per block

 Histogram

4.5 to 5.5 5.5 to 6.5 6.5 to 7.5

 Process Control
A process is “In Control” if products vary CONSISTENTLY
within expected limits over TIME. Normal expectations when
displayed by frequency of occurrence will present a curve or
distribution showing a central peak and tapering off smoothly
to tails on either side.
.5
.5
.5
.4 .5 .6
.4 .5 .6
When the normal distribution
.4 .5exits.6
the process is operating
.5 the.6process behavior.
consistently and we can.4predict
.3 .4 .5 .6 .7
.3 .4 .5 .6 .7
.3 .4 .5 .6 .7
.2 .3 .4 .5 .6 .7 .8
 Controlling a Process
1. To control a process means to keep the bell shape normal.
2. Get the Process to a normal distribution. (Remove assignable
causes) You don’t want to control a broken process.
3. Monitor the process property at an interval that can detect
deterioration of the property and allow correction before
rejections.
Example: If the property was effected by a belt that wears over
time, checking once a week may be adequate.
If the property were to use a batch of epoxy in the working range
before it cures to much, checking every 15 minutes would be
appropriate. (The time interval is called “Frequency”)
4. Take corrective action before defects are produced.
Assignable Causes
Easy to identify and correct. (low cost)
 What is poor Quality?
Not meeting customer expectations, implied or specified!
 Customer Dis-satisfaction

A
B

Target
Lower Spec Upper Spec

Reject No Question Perfection – Not Required

A = more variation on target? or B = less variation off target?

What causes defects? 6 M’s
Man (associated operator error)
Machine (improper, misadjusted or broken)
Material (wrong or defective)
Measurement (wrong, bad or misread instrument)
Method (improper, unclear or incomplete)
Mother Nature (environment – temp – moisture …)
 Good or Bad?
LCL UCL
Bad (Out of Spec.)

LCL UCL

6

6
Customer Specification

The Bell shape has nothing to do with good/bad only Normal.

 When do you have rejects?

When Specification Limits are imposed.

(Customer or Company Spec.)
 Capability
CP
LSL USL

LSL USL
LCL
xxx UCL

xxxxx
xxxxxxx
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
Not
Capable Specification
Limits Control
6 limits
Capable
 Precise and Accurate

Accurate and not Precise

Not Precise and Not Accurate

Precise and Accurate

Precise and Not Accurate
Capability Improvement
Variation Reduction
 Capability
CPk
Normal
USL
US
Shift L

xxx xxx
xxxxxxxxxx
xxxxxxx
xxxxxxx
xxxxxxxxx
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxxx
xxxxxxxxxxxxx
xxxxxxxxxxxxxx
xxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
x
xxxxxxxxxxxxxxxxx
Avg.
Cpk=min Avg-USL/3/2
Or LSL-Avg/ 3/2
Six Sigma
Not
Capable Capable
Variation of the Mean
Two Distributions Working Together
(Movement: right, left or up, down or clockwise, counterclockwise …)
(mixed materials, different lots of material, different operators…)
(If parts tighter tolerance parts were selected out it would look like this.)

xxx xxx
xxxxx xxxxx
xxxxxxxxxxxxxx
xxxxxxxxx
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxxxx
xxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx

Bimodal
Six Sigma
Control Limits, the six sigma points
LCL UCL

- 3 +3

Mean
(Average)

6 sigma
Making a time related distribution.
Display the measurements in the order taken.

Each point is an average from the samples measured (5 smpls)

 Add Control Lines to charts

UCL

+3
(Average)
6 sigma

Mean

- 3
LCL
 Xbar and Range Charts
Xbar = X = The average of measured samples (5)
Range = R = Highest reading – Lowest reading in sample (5)
Plotting Procedure
1. Take five measurements / data points, noting any decimal points required.
2. If you are using the current date and time skip step 3.
3. Enter your earlier date and time in military format 10/18/03 14:15 in the Override Date/Time cell
4. Enter your data into the cells provided. Sample 1 2 3 4 5 (Yellow cells)
(Note: Click on the yellow Cell under the number 1 to start entering data, use the Tab key to move to the next cell.)
5. On the last entry use the down arrow to enter and proceed to the Initials cell (otherwise click the Initials cell). Enter your Initials,
Press Tab.
6. Press the "plot button" in the lower right corner and observe that your data point has been plotted.
7. If there appears to be an error in the data, you can click on the defective data and reenter it. Press Tab, then click the Replot Button.
8. Once the point is correctly plotted, Observe if any Run Rules have been violated. If not, enter "none" in the action cell, then click the
"Tab then Action Button." (lower right corner). Otherwise, take and log the corrective action taken.
9. Observe that the data cells are empty, indicating a successful transaction. The lower scroll pane (two rows at the bottom of screen)
can be used to observe the calculations and action entered. (Just click the down scroll arrow until the last row is displayed. (80 rows
maximum can be entered.)

Run Rules
1. Any point that crosses the upper or lower control limit on the XBar Chart or the upper control limit on the R Chart.
2. 7 points in a row on one side of the center line on the Bar Chart.
3. 7 points in a continuing upward or downward direction on the Bar Chart or R - Range (lower) chart (Trend).
4. 7 points in a row at the upper and lower extremes (no points on or near the centerline) Bar Chart.
5. 7 points in a row clustered closely around the center line on the X Bar Chart (Capture improvement or speed up the process).
6. 7 points in a row clustered near bottom of R Chart or trending toward bottom (Capture improvement or speed up the process)
 Run Rules (normal no action required)
Any point that crosses the upper or lower control limit on the
Xbar Chart or the upper control limit on the R chart.
Xbar (upper chart)

Range – R (lower chart)

* Note random points, some near control lines some near center.
 Run Rules (when to take action)
Any point that crosses the upper or lower control limit on the
Xbar Chart or the upper control limit on the R chart.
Xbar (upper chart)

Range – R (lower chart)

* Check for possible bad measurement or data entry error.

 Run Rules (when to take action - Run)
7 points in a row on one side of the center line on the upper chart.
Xbar (upper chart)

Or

*Check material, lot, operator, machine adjustment…

 Run Rules (when to take action - trends)
7 points in a continuing downward or upward direction on X chart
Xbar (upper chart)

Or

* Check machine wear, material, method or operator fatigue…

 Run Rule (when to take action - trends)
7 points in a continuing downward or upward direction on R chart
Range R (lower chart)

Improving – capture or speed up

Or

Getting Worse - correct

* Check machine wear, material, method or operator fatigue…

 Run Rule (clustered points)
“Over controlled” If bell was normal points would be near the edges.
7 points in a row clustered around the center line on the Xbar chart
or clustered around zero (bottom line) on the R Range chart
Xbar (upper chart)

Improved – capture or speed up

Range – R (lower chart)

Improved – capture or speed up

 Run Rules (when to take action – Unusual pattern)
No points near the center (possibly bimodal)
Xbar (upper chart)

*Two lots of material, two operators, selected parts removed…

 Attribute Data
Non-measurable data
Good, Bad; Pass, Fail; True, False; Go, NoGo; Present, Missing…

Charts: P (Percent Defective), NP (Number Defective)

C (Defects per unit), U (Defects per Subgroup)

Simple if sample size is constant and equated easily to 100. Ex:

Number defective from sample 50 multiply by 2, 25x4, 10x10,
5x20, 2x50 EXAMPLE: 5 bad in sample of 25 (5x4) =20%
The software will do the math but keeping a fixed sample size is
preferred.
Fixed sample size will keep the control lines straight and understood.
Plotting Procedure
1. Count out the sample size and note the number defective.
2. If you are using the current date and time skip step 3.
3. Enter your earlier date and time in military format 10/18/03 14:15 in the Override Date/Time cell
4. Enter your data into the cells provided. Sample Size, Qty Defective (Yellow cells)
(Note: Click on the yellow Cell to start entering data, use the Tab key and or arrow keys to move to the next yellow cell.)
5. On the last entry use the arrow keys to enter and procede to the Initials cell (otherwise click the Initials cell). Enter your
Initials, Press Tab.
6. Press the "plot button" in the lower right corner and observe that your data point has been plotted.
7. If there appears to be an error in the data, you can click on the defective data and reenter it. Press Tab, then click the
Replot Button.
8. Once the point is correctly plotted, Observe if any Run Rules have been violated. If not, enter "none" in the action cell,
then click the "Tab then Action Button." (lower right corner). Otherwise, take and log the corrective action taken.
9. Observe that the data cells are empty, indicating a successful transaction. The lower scroll pane (two rows at the bottom of
screen) can be used to observe the calculations and action entered. (Just click the down scroll arrow until the last row is
displayed. (80 rows maximum can be entered.)
Run Rules
1. Any point that crosses the upper or lower control limit on the P Chart
2. 7 points in a row on one side of the center line.
3. 7 points in a continuing upward or downward direction on the P chart (Trend).
4. 7 points in a row at the upper and lower extremes (no points on or near the centerline)
5. 7 points in a row clustered near bottom of R Chart or trending toward bottom (Capture improvement or speed up the
process)
The End