Training Module

On
STATISTICAL PROCESS CONTROL-SPC
Refer-SPC AIAG Manual-2th Edition

Neel Auto Private Limited,

Human Resources Department.
Gurgaon
1
Contents
 What is SPC
 Why SPC Required
 Key terms
 Variation
 Common vs Special causes
 Types of Data
 Mean
 Range
 Standard Deviation
 Normal curve-Bell Shaped
 Basic Statistical tools
 What is Control Charts
 Why control Charts
 Types of Control Charts
Variable - X & R Chart
Attribute- P & nP Chart
- C & U chart
 Interpretation for Process Control charts
 Process Capability
 Process Performance
 HMSI SPC Guidelines
2
 What is SPC?
Statistical process control is a collection of tools that
when used together can result in process stability and
variance reduction.


It is a prevention tool
– Inspection = defect detection
– SPC =detect process change
defect prevention

3
 Why SPC required ?
 Initial Belief – Acquiring knowledge is a matter of
chance

 Next Belief – Knowledge is gained through systematic

experimentation

 Current Belief – Knowledge can be acquired &

enhanced through systematic data collection and
analysis
 To gain knowledge about processes on real time basis
 Make economically sound decisions
 Balancing between over control & under control

4
 KEY TERMS
Process
 Set of activities that uses resources & transform inputs into
outputs
Process Control
 Actions taken or checks provided to,
 Prevent the occurrence of causes of nonconformities
 Detect the causes of nonconformities
 Detect the occurrence of failure in the process output
Statistics
 A science used for collection, organization, presentation, analysis
& interpretation of numerical data

5
 KEY TERMS -Continued

Statistical Methods
 Tools & techniques used to collect, organize, present, analyze & interpret data
Statistical Process Control
 A feedback mechanism to detect causes and associated failures by using
statistical techniques on real time basis and leading to corrective /
improvement actions.
Defect (Nonconformity)
 Not acceptable
 Not meeting requirements
 Difference between requirements & expectations
To minimize defects standards / specifications are established
Defective ( Nonconforming)
 Product / service having one or more defects
Variation
 Difference between the similar characteristics of two or more things
Causes of Variation
 Factors those contribute for variation

6
VARIATION

I perceive two kinds of variation:

• Variation from constant causes , the
same causes from hour to hour, lot
to lot, worker to worker- Common
causes
• Variation from a special cause

SPC Developed By
Dr. Walter A. Shewhart
During 1920’s in Bell Lab
7
 Common vs Special cause

Common cause Special cause

• Always present in • Which are not
process normally present in
• GOD only knows process
• Example • Human also Knows
-Slight variation in raw • Example
material - batch of defective material
-Lack of Human perfection in - untrained operator
reading of instruments &
- Faulty set up
setting

8
 Common vs Special cause

Common cause= normal, Inherent

Arrive work between 7:55 to 8:01due to number traffics
lights that you stopped at on your way to work.

Special cause = assignable

Arrived to work today at 8:45 because;
a) flat tire on the way to work
b) Accident on the interstate
c) I met up with an old drinking buddy and
I stayed out later than I should have.

9
If only common cause of variation are present. The voice of the
process is stable & predictable and is said to be under
statistical control

Prediction

Prediction

If special cause of variation are present. The voice of the process is not
stable & predictable and is said to be out of control
10
 Types of Data
• Attribute • Variable
 Anything that can be  Anything that can be
classified measured
 Either/Or  Height
 Pass/ fail  Dollars
 Good/ Bad  Distance
 Go/No Go  Speed
 Discrete  Continues, infinite
 Detection Oriented  Stimulates Prevention

11
 MEAN
The Mean of ‘n’ numbers is the total of the numbers
divided by ’n’
x1  x 2  x3 .....x n
x
n
In Standard Mathematical Notation it is
i n
x  
i 1
xi
n

The symbol “” means ‘sum of’

Sample Value xi (mm)

144

146

154

146

Mean ( X ) = (144+146+154+146) / 4 = 590/4 = 147.5

12
 Range-Measure of Spread or Dispersion
Range : The difference between the largest and the smallest of a
set of numbers. It is designated by a capital “R”

R  X Max  X Min
R  X Hi  X Low

Sample Value xi (mm)

144

146

154

146

Range ( R ) = (154-144) = 10

13
Measure of Spread or Dispersion-
-Sigma ()
Standard Deviation : The average distance between
the individual numbers and the mean.It is
designated by “s”

s= ( x1
- x) + ( 2
x2
- x) ....(xn
2
- x) 2

n -1
14
 Standard deviation (Sigma ) = Squared deviation from mean
Value xi (mm) Deviation (xi - X) (xi - X) 2
144 - 3.5 12.25
146 - 1.5 2.25
154 +6.5 mm 42.25
146 - 1.5 2.25
Sample Total = 0

Mean ( X ) 147.5 mm Total = 59.00

Average of squared deviation = (xi - X) 2 = 59.0 = 14.75

n 4

=
 (xi -
n
X) 2

 =  14.75 = 3.84 mm

15
 Voice of the Process

-3 s +3 s

Process Width
Design Width

LSL T USL
Voice of the Customer
16
Normal Curve-Bell Shaped

-3s -2s -1s x +s +2s +3s

68.26%
95.48%
99.73%
99.99999975% of the data lies within ± 6 s of the mean

17
 Basic Statistical Tools
Stratification
Run Chart
Check Sheets
Histogram
Cause & Effect Diagram
Pareto
Control Chart
Variable
Attribute

18
 WHAT IS CONTROL CHARTS
• Transformation of a normal distribution curve in the form of 3
parallel lines, where
•The middle line indicates mean and called central line (C.L.)
•The upper line indicates Mean +3 Sigma and called upper control limit
(UCL)
•The lower line indicates Mean -3 Sigma and called lower control limit
(LCL) U.C.L.
+3s

Mean C.L.

-3s
L.C.L.

19
Why CONTROL CHARTS
• Understand Process Variation to
achieve state of Statistical Control
• When Process is in Statistical
Control, its performance is
predictable
• Further reduction of common causes
variation
• Common language for
communicating Process performance
• More precise and accurate technique

20
 Types of Control charts
Variable
 Average and Range ( X-R )

Attribute
 p Chart Proportion of Unit Nonconforming (sample size may not be equal)
 np Chart Proportion of Unit Nonconforming (sample size equal)
 c Chart Number of Nonconformities (sample size equal)
 u Chart Number of Nonconformities (sample size may not be equal)

21
 CONTROL CHARTS (Variable)
Average-Range ( X-R )Chart
• Precondition:
– Measurement must be variable
– Situation must be practically feasible to have at
least 2 measurements in short span.
• Preparatory Steps
– A suitable environment
– Understand the process
– Determine the characteristics
– Ensure Level 1 control -Minimize unneccessary
variations .

22
 CONTROL CHARTS (Variable)
Average-Range ( X-R )Chart
• Preparatory steps
– Select the Subgroup Size
• Rational subgroup: Variability within subgroup should be small
– Select Subgroup Frequency
• Detect change in the Process over span of time.
• For initial study, may be consecutive or a very short interval.
– Select No. of Subgroups (for initial study: to define the control
limits)
• To incorporate Major source of variation (Generally 25 subgroups
or more containing about 100 individual measurements)
– Decide Measurement system: Should qualify MSA

23
 Average-Range ( X-R )Chart
• Data Collection
– On a data collection sheet, called control chart sheet
Date:
 -R Chart
SECTION: PRODUCT: CHARACTERISTICS:

PERSON IN-CHARGE:
X - CHART

Events:
 -C h a r t
R -C h a r t

SAMPLE # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
DATE
TIME
X1
X2
X3
X4
X5

R

24
 AVERAGE-RANGE ( X-R ) CHARTS
 Calculate Average of each Subgroup

X = ( X1 + X2 + … + Xn )/ n TABLE OF CONSTANTS FOR

CONTROL CHARTS
 Calculate Range of each Subgroup
R = Xmax. - Xmin.
Subgroup
 Calculate Process average Size (n)
A2 d2 D3 D4 E2

2 1.880 1.128 - 3.267 2.660

X =(X1 + X2 + … + Xk)/ k
3 1.023 1.693 - 2.574 1.772
 Calculate Average Range 4 0.729 2.059 - 2.282 1.457
5 0.577 2.326 - 2.114 1.290
R = (R1 + R2 + … + Rk )/ k 6 0.483 2.534 - 2.004 1.184
 X1, X2,…., Xn are individual values within the subgroup 7 0.419 2.704 0.076 1.924 1.109
8 0.373 2.847 0.136 1.864 1.054
 n is the Subgroup Sample Size 9 0.337 2.970 0.184 1.816 1.010
 k = No. Trial
of Subgroups
 Calculate Control Limits for Range Chart
UCLR = D4 R
LCLR = D3 R.
 Calculate Trial Control Limits for Average Chart
UCLX = X + A2 R

LCLX = X - A2 R

 D4, D3 and A2 are constant varying as per sample size (n).

25
 DATA SHEET FOR X-R CHART
Sample Size X1 X2 X3 X4 X5 SX X R
Sub Group
1 41 32 41 35 20 169 33.8 21
2 21 37 31 25 34 148 29.6 16
3 21 20 21 20 41 123 24.6 21
4 29 29 41 40 38 177 35.4 12
5 28 21 41 36 25 151 30.2 20
6 40 35 21 38 33 167 33.4 19 UCLr = D4 x R = 2.114 x 18.56 = 39.24
7 21 30 24 33 26 134 26.8 12 LCLr =D3 x R = 0 x 18.56 = 0
8 35 41 26 21 38 161 32.2 20
9 27 37 26 20 32 142 28.4 17 UCLx =X + A2 R = 30.34 + 0.577 x 18.56 = 41.05
10 23 41 40 37 35 176 35.2 18 LCLx =X - A2 R = 30.34 - 0.577 x 18.56 = 19.63
11 28 41 24 31 20 144 28.8 21
12 31 25 21 32 22 131 26.2 11
13 22 37 21 41 20 141 28.2 21
14 37 32 24 38 30 161 32.2 14
15 25 40 23 39 21 148 29.6 19
16 21 31 41 20 32 145 29 21
17 38 22 29 40 37 166 33.2 18
18 35 20 35 41 20 151 30.2 21
19 31 20 38 24 41 154 30.8 21
20 20 27 41 40 31 159 31.8 21
21 41 40 21 24 25 151 30.2 20
22 20 31 41 31 28 151 30.2 21
23 29 40 22 32 22 145 29 18
24 28 27 20 32 41 148 29.6 21
25 41 34 24 29 21 149 29.8 20
X R
30.34 18.56

26
 X & R - CHART
45

40 41.05

35 35.4 35.2
33.8 33.4 32.2 33.2 31.8
32.2 30.8 30.2 UCLx
30 29.6 30.2 29.8 X Bar Bar
28.4 28.8 28.2
29.6 29 30.2 30.2 29 29.6
26.8 X Bar
26.2
25 24.6 LCLx

20 19.63

15
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

45
40 39.24
35
30
UCLR
25
20 R Bar
20 21 21 20 19 21 21 21 21 21 21 20 21 21 20
18 19 18 18 R
15 16 17
14 LCLR
12 12 11
10
5
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

27
CONTROL CHARTS (Attribute)

• Select the Subgroup Size

– For p-chart, sufficiently large so that np>5)
• Select Subgroup Frequency
– Detect change in the Process over span of time
• Select No. of Subgroups
– Major source of variation incorporate (25 or more)

28
 P Control Charts
• Calculate the Proportion Defectives

P = nP/n
n = no. of items inspected
nP = no. of Defectives
• Calculate Average Proportion Defectives

P = n1P1 + n2P2 + … + nkPk = P1 + P2 + … + Pk

n1 + n2 +… + nk k
k = No. of Subgroups
• Calculate Upper & Lower Control Limit

UCL = P +3 P(1- P) k

LCL = P -3 P(1- P) k

Process Capability (Attribute) spread = (1-P bar)*100 % of normal curve

NOTE: If the Sample Size variation is 25% of the Average Sample Size take the value
Average n for UCL & LCL calculation

29
 P Control Charts
DATE S AM P LE NO. OF P RO P O RTI O N
S I ZE DEFECTI V ES DEFECTI V ES
01. 04. 2015 62 2 0. 03
02. 04. 2015 62 5 0. 08
03. 04. 2015 62 4 0. 06
04. 04. 2015 62 3 0. 05
05. 04. 2015 62 3 0. 05
06. 04. 2015 62 6 0. 10
07. 04. 2015 62 5 0. 08
08. 04. 2015 62 0 0. 00
09. 04. 2015 62 7 0. 11
10. 04. 2015 62 5 0. 08
11. 04. 2015 62 4 0. 06
12. 04. 2015 62 1 0. 02
13. 04. 2015 62 2 0. 03
14. 04. 2015 62 3 0. 05
15. 04. 2015 62 6 0. 10
16. 04. 2015 62 3 0. 05
17. 04. 2015 62 8 0. 13
18. 04. 2015 62 4 0. 06
19. 04. 2015 62 4 0. 06
20. 04. 2015 62 4 0. 06
21. 04. 2015 62 6 0. 10
22. 04. 2015 62 4 0. 06
23. 04. 2015 62 2 0. 03
24. 04. 2015 62 3 0. 05
25. 04. 2015 62 7 0. 11
T OTAL: 101 1. 629

P = 1.629/25 = 0.065
30
 P Control Charts
FRA CTION
= 0.065
DEFECTIV E
UCL = P + 3 SQRT {P(1-P) }/ SQRT (k )
= 0.213
L CL = P - 3 SQRT {P( 1-P) }/ SQRT (k )
= 0.083

P CONTROL CHART
P R O P O R T IO N D EF EC T IV ES

0.25

0.20

0.15

0.10

0.05

0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
0.05

0.10
DATE

PROPORTION DEFECTIVES UCL LCL AVERAGE

31
 np Control Charts
• Calculate the Average Number of Defectives
np = (n1p1 + n2p2 + … + nk) /k
np = No. of Defectives in Sample Size (n)
k = No. of Subgroups
n = Sample Size

• Calculate Upper & Lower Control Limit

UCL = np +3 np (1- np/n)

LCL = np -3 np(1- np/n)

32
 C Control Charts
• Calculate Subgroup Proportion Non Conformity

C = Total no. of defects in a Sample Size (n)

• Calculate Process Average Proportion Non Conformity

C = (C1 + C2 + … + Ck)/ k

k = No. of Subgroups
• Upper Control Limit (UCL) = C +3 C

• Lower Control Limit (LCL) = C -3 C

Process Capability (Attribute) spread = (1-C bar)*100 % of normal curve

33
DAT E
C Control Charts
S AM P L E NO . O F
DAT E S I ZE DEF ECT S
01. 04. 2015 8 8
02. 04. 2015 8 17
03. 04. 2015 8 18
04. 04. 2015 8 15
05. 04. 2015 8 17
06. 04. 2015 8 9
07. 04. 2015 8 19
08. 04. 2015 8 6
09. 04. 2015 8 14
10. 04. 2015 8 17
11. 04. 2015 8 13
12. 04. 2015 8 15
13. 04. 2015 8 16
14. 04. 2015 8 22
15. 04. 2015 8 13
16. 04. 2015 8 10
17. 04. 2015 8 14
18. 04. 2015 8 9
19. 04. 2015 8 19
20. 04. 2015 8 11
21. 04. 2015 8 21
22. 04. 2015 8 23
23. 04. 2015 8 3
24. 04. 2015 8 8
25. 04. 2015 8 12
T O T AL : 349

C = 349 /25 = 13.96

34
 C Control Chart
FRACTION DEFECTIVES = 13.96

UCL = C + 3 SQRT(C)

= 25.169

LCL = C - 3 SQRT(C)

= 2.751

C CONTROL CHART
P R O P O R T IO N D EF EC T S

3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00

DATE

PROPORTION DEFECTS UCL LCL AVERAGE

35
 U Control Charts
 Calculate Subgroup Proportion Non Conformity
U = C/n
n = Variable Sample Size

 Calculate Process Average Proportion Non Conformity

C = C1 + C2 + … + Ck
n1+n2+n3 …. + nk

 Upper Control Limit (UCL) = U +3 U / n

 Lower Control Limit (LCL) = U -3 U / n

NOTE: If the Sample Size variation is 25% of the Average Sample Size take the
value Average n for UCL & LCL calculation

36
INTERPRETATION FOR PROCESS CONTROL CHART

------------------------- ----------------------- -----------------------

--

------------------------- ----------------------- -----------------------

-- Trend (decreasing)
Run Trend (increasing)

----------------------- ----------------------- -----------------------

----------------------- ----------------------- -----------------------

Out of control (no trend)
Cyclic pattern/trend Two universe pattern

37
 INTERPRETATION FOR PROCESS CONTROL CHART

1. Out of Control Limits

Points which are outside the control limit

x Upper Control Limit

Central Line

Lower Control
Limit

Control chart for out-of-control state

38
 INTERPRETATION FOR PROCESS CONTROL CHART
2. Run is the state in which points occur continually on one side of the central line and
the number of points is called the length of the run.
Seven-point length of run is under 6 the following cases are considered to be
abnormal.
a) At least 10 out of 11 consecutive points occur on one side of the central line.
b) At least 12 out of 14 consecutive points occur on one side of the central line.
c) At least 16 out of the 20 consecutive points occur on one side of the central
line.
S even-point length of ru n Ten ou t of 11 consecu tive points occu rring
is abnorm al on one side is abnormal

39
 INTERPRETATION FOR PROCESS CONTROL CHART
3. When the points form a continuous upward or downward curve, this is
said to have a trend.
Seven upw ard points Drastic dow nw ard trend

40
 INTERPRETATION FOR PROCESS CONTROL CHART
4. Considering points which approach the 3 Sigma control limits, if 2
out of 3 points occur outside of the 2 Sigma lines, this case is
considered to be abnormal.

3 / n
2 / n

2 / n
3 / n

Approach to the Control Limits (2 out of 3 points)

41
 INTERPRETATION FOR PROCESS CONTROL CHART
5. When most of the points are arranged within the central 1.5 sigma lines
(the bisectors of the central line and each of the control limits), this is due to
inappropriate way of sub grouping. Approach to the central line does not
mean a controlled state, but it means mixing of data with a different
population in sub groups, which make the width of control limits too wide.
At this situation, it is necessary to change the way of sub grouping.

3 / n

1.5 / n
1.5 / n
3 / n

Approach to the Central Line

42
 INTERPRETATION FOR PROCESS CONTROL CHART

6. When the curve repeatedly shows and up-and-down trend for the
almost same interval, this is also abnormal.

Periodicity

43
 INTERPRETATION FOR PROCESS CONTROL CHART

7. Interpretation should be started with R-Chart first

8. For each indication of special cause, analysis should be done to identify the
root cause and action to be taken.

9. For initial control chart,

– Discard all the subgroup showing out of control situation (starting from R-
Chart)
– Re-Calculate Control Limits, plot and analyse for any out of control situation.
– Re discard if any out of control situation again found.
– Continue the above cycle untill all plot indicate a control situation.
– Repeat same exercise with Average chart.
– If more than 50% data are required to be discarded, reject all data and re-collect.
– Once initial control chart indicates control situation, extend control limits for
ongoing control.

44
 PROCESS CAPABILITY
This is carried out to measure the ability of the process to
meet the specified tolerances.

When Calculate Process Capability ?

• All the Assignable Causes are removed
and process operates only under the
Common Causes
 Skilled operator
 Raw Material, Machine and Measuring
Equipment Condition

45
 PROCESS CAPABILITY
• Calculate Process Standard Deviation
 = R/d2
d2 is a constant varying as per sample size (n)

• Calculate Process Capability (Cp)

Cp = (USL - LSL) / 6 USL = Upper Specification Limit
= Tolerance/ 6 LSL = Lower Specification Limit

• Calculate Process Capability (Cpk)

CpkU = (USL -X) / 3
or CpkL = (X - LSL) / 3
Whichever is minimum will be Cpk

46
 PROCESS CAPABILITY
Standard Deviation () = R/d2
= 0.169/2.33 USL = 0.900
= 0.0725 LSL = 0.500
Process Capability
Cp = (USL - LSL) / 6 = (0.900 - 0.500) / 6 x 0.0725
= 0.920
CpkU = (USL - X) / 3 = (0.900 - 0.738) / 3 x 0.0725
= 0.743
CpkL = (X - LSL) / 3 = (0.738 - 0.900) / 3 x 0.0725
= 1.093
Cpk = 0.743
LSL USL

X

1.093 0.743

47
 PROCESS PERFORMANCE
Standard Deviation s =  (xi-X)2
i=1
for n=80
USL = 0.900
n-1
LSL = 0.500
Process Performance
Pp = (USL - LSL) / 6s = (0.900 - 0.500) / 6 x 0.0759
= 0.880
PpkU = (USL - X) / 3s = (0.900 - 0.738) / 3 x 0.0759
= 0.710
PpkL = (X - LSL) / 3s = (0.738 - 0.900) / 3 x 0.0759
= 1.045
Ppk = 0.710

48
 HMSI SPC Guidelines
Criteria for selection of parameters for Cp / Cpk monitoring
For HS / HA / HB Parts Parameters for which Cp ,Cpk Study to be conducted are as
follows :
• Q Parameters (Which ever applicable).
• Last model and T1 trial feed back.
• Previous model parameters resulting in high PPM.
• Ext / Int QICs - contributing Dimensions Which ever applicable) in previous
models Action Plan required if Cpk
• Fitment and Functional Parameters 1< Cpk < 1.33
• Critical dim. List identified by HMSI-PI •Countermeasure Plan
•
100% Insp. Till countermeasure is
Taken.
• < 1 – Process is not accepted
•>1.33
Maker to make Monthly SPC plan
•Action Plan required if Ppk
<1.67
•Countermeasure Plan
•100% Insp. Till countermeasure is
Taken.
>1.67
• Maker to make Monthly SPC plan

49
50