STATISTICAL PROCESS

CONTROL (SPC)

BASIC CONCEPTS
OF

SPC
By
A R SURESH

07/29/09 SPC CM NNCPL 1

 STATISTICAL PROCESS CONTROL (SPC)

CONTENT
1.INTRODUCTION
2.PROCESS CONTROL
3.VARIATION
4.CAUSES
5.STATISITICAL CONTROL
6.TOOLS FOR STATISTICAL CONTROL
7.PROCESS CAPABILITY
8.PROCESS CAPABILITY INDICES
9.CONTROL CHART TYPES
10 CONTROL CHART MEHODOLOGY EVALUATION

07/29/09 SPC CM NNCPL 2

 INTRODUCTION
QUALITY OF CONFORMANCE

Vs

CONFORMANCE TO
SPECIFICATION
07/29/09 SPC CM NNCPL 3
WHAT SPC MEANS?
SPC is not constrained to measuring the Conformance; it is intended
to lead to action on processes, which are operating within the
“specification” to minimize variability. It can be used to analyze a
process or its output, so as to take appropriate actions to achieve and
maintain a state of Statistical Control and to Improve the Process
Capability. It provides objective means of controlling quality in any
transformation process, whether used in the manufacture of articles,
the provision of services or the transfer of Information. Tools of SPC’s
are not calendars / sceneries to hang it on the wall of the industries
where it is made. It should be used as Component part of company-
wide adoption of ‘TOTAL QUALITY’ and act as tool for a never-ending
improvement.

SPC exists because there is, and will always be variation in the
characteristic of materials, articles, services, people present.
Variation has to be understood and assessed in order to be managed.
SPC can be carried out through “control charts”.
07/29/09 SPC CM NNCPL 4
 PROCESS CONTROL
THREE TYPES OF PROCESS CONTROL
• PREVENTION OF DEFECTS
* MISTAKE/ERROR PROOFING
• DETECTION OF CAUSES AND LEAD TO
CORRECTIVE ACTION
* VISUAL CONTROL, SPC
• DETECTION OF DEFECTS
* INSPECTION

07/29/09 SPC CM NNCPL 5

 VARIATION
PRINCIPLES OF SPC
· VARIATION IS INEVITABLE
· VARIATION IS PREDICTABLE
· VARIATION IS MEASURABLE
PREVENTION VERSUS DETECTION
• The Goal of Quality Assurance is Defect Prevention
• Prevention is not possible without Prediction
• Prediction is not possible without the use of Statistical Methods
• Statistical Methods are not possible without Data
• Data is not possible without Detection schemes
• Detection is the Goal of Inspection

This is the linkage between Inspection and Quality Assurance. In

other words, this is the linkage between PREVENTION and
DETECTION.

07/29/09 SPC CM NNCPL 6

 STATISTICAL PROCESS CONTROL

STATISTICAL PROCESS CONTROL IS

· 10 % IS STATISTICS

· 90 % IS PRODUCT & PROCESS

KNOWLEDGE

07/29/09 SPC CM NNCPL 7

 STATISTICS
DETERMINE WHICH SAMPLE IS GOOD ?
SAMPLE –1
15,15,14,15,16,15,15
SAMPLE –2
14,15,20,15,16,10,15
SAMPLE - 3
19,20,15,14,11,10,16

07/29/09 SPC CM NNCPL 8

 DETERMINING WHICH SAMPLE IS GOOD

• SAMPLE –1
15,15,14,15,16,15,15
MEAN: 105 / 7 = 15, RANGE: 16 – 14 = 2

• SAMPLE –2
14,15,20,15,16,10,15
MEAN: 105/7 = 15, RANGE : 20 – 10 = 10

• SAMPLE - 3
19,20,15,14,11,10,16
MEAN: 105 / 7 = 15, RANGE : 20 – 10 = 10

AS WE CANNOT DECIDE WHICH SAMPLE IS GOOD WE

NEED TO CAPTURE AVERAGE, RANGE, STANDARD
DEVIATION TO EVALUATE THE SAMPLES
07/29/09 SPC CM NNCPL 9
 STATISTICS

HENCE WE NEED TO CAPTURE

• AVERAGE
• RANGE
• STANDARD DEVIATION
TO EVALUATE SAMPLES

07/29/09 SPC CM NNCPL 10

 SAMPLE 1
RANGE = 2
MEAN/AVERAGE 105/7 =
No DATA DEVIATION SQUARE OF
15
X (X –AVG) DEVIATION
SUM OF 2
1 15 0 0 SQUARED
2 15 0 0 DEVIATIONS
MEAN SQUARE 2/7
3 14 -1 1
DEVIATION
4 15 0 0
5 16 1 1 ROOT MEAN √2/7
SQUARE = 0.535
6 15 0 0 DEVIATION
7 15 0 0 (STANDARD
DEVIATION σ)

07/29/09 SPC CM NNCPL 11

 SAMPLE 2
RANGE = 10
MEAN/AVERAGE 105/7 =
No DATA DEVIATION SQUARE OF 15
X (X –AVG) DEVIATION SUM OF 52
1 14 1 1 SQUARED
2 15 0 0 DEVIATIONS
MEAN SQUARE 52/7
3 20 -5 25 DEVIATION = 7.423
4 15 0 0
ROOT MEAN √52/7
5 16 -1 1 SQUARE = 2.726
6 10 5 25 DEVIATION
7 15 0 0 (STANDARD
DEVIATION)

07/29/09 SPC CM NNCPL 12

 SAMPLE 3
RANGE = 10
MEAN/AVERAGE 105/7 =
No DATA DEVIATION SQUARE OF 15
X (X –AVG) DEVIATION SUM OF 84
1 19 -4 16 SQUARED
2 20 -5 25 DEVIATIONS
MEAN SQUARE 84/7
3 15 0 00 DEVIATION = 12
4 14 1 01
ROOT MEAN √12
5 11 4 16
SQUARE = 3.464
6 10 5 25 DEVIATION
7 16 -1 01 (STANDARD
DEVIATION)

07/29/09 SPC CM NNCPL 13

 COMPARATIVE STUDY OF SAMPLES
SAMPLE 1:
MEAN/AVERAGE (X bar) = 15, RANGE (R) = 2,
STANDARD DEVIATION (σ) = 0.535

SAMPLE 2:
MEAN/AVERAGE (X bar) = 15, RANGE (R) = 10,
STANDARD DEVIATION (σ) = 2.726

SAMPLE 3:
MEAN/AVERAGE (X bar) = 15, RANGE (R) = 10,
STANDARD DEVIATION (σ) = 3.464

BY THIS WE CAN INFER THAT SAMPLE ONE IS BEST OF THE

LOT & OF THE 2ND & 3RD SAMPLE 2 IS BETTER AS STANDARD
DEVIATION IS LESS.
07/29/09 SPC CM NNCPL 14
 VARIATION & CAUSES
TYPES OF VARIATION
• RANDOM VARIATION
• NON RANDOM VARIATION

CAUSES OF VARIATION
• COMMON OR CHANCE CAUSES
• SPECIAL OR ASSIGNABLE CAUSES

07/29/09 SPC CM NNCPL 15

UNDERSTANDING THE COMMON AND SPECIAL CAUSES
CONCEPT OF VARIATION
Variation is inevitable. But causes are different. Certain variations
belong to the category of chance or random variations, about
which little may be done, other than to revise the process. This
variation is due to the summation of two causes. They are random
and common causes. The variation due to these causes is small in
a process. Also it is not possible to trace the causes produced by
this random and common causes. When only these random
causes are present in a process then the process is considered to
be “in statistical control”. Top Management has to take action on
these causes.
Causes of variation, which are relatively large in magnitude, and
readily identified are, classified as “assignable” or “special”
causes. When an assignable cause of variation is present, process
variability will be excessive and the process is said to be “out of
statistical control”. This can be identified by the
operator/supervisor and they can take action to eliminate it.
07/29/09 SPC CM NNCPL 16
COMMON CAUSES – refer to the many sources of variation that
consistently act on the process. Common causes within a process
produce a stable and repeatable distribution over time. This is called
“In a State of Statistical Control”

SPECIAL CAUSES – also called Assignable Causes – refer to

many factors causing variation that affect only some of the process
output. They are often intermittent and unpredictable. They result in
non-random patterns of variation.

IMPORTANT NOTE
The changes in the process distribution due to special causes can be
either detrimental or beneficial. When detrimental, they need to be
understood and removed. When beneficial, they should be
understood and made a permanent part of the process.

07/29/09 SPC CM NNCPL 17

 CAUSES
COMMON CAUSES SPL./ASSIGNABLE

FEW IN NOS. PLENTY IN NOS.

VARIATION IS LOW VARIATION IS HIGH

PART OF THE VISITOR TO THE

PROCESS PROCESS
CONSTANT VARIATION FLUCTUATING
VARIATION
PREDICTABLE UNPREDICTABLE

07/29/09 SPC CM NNCPL 18

 CAUSES
COMMON CAUSES SPL./ASSIGNABLE

STATISTICS STATISTICS CANNOT

APPLICABLE APPLY
MANAGEMENT OPERATOR
CONTROLLABLE CONTROLLABLE
e.g Pressure variation, e.g Wrong setting,
Environment variation wrong master
REDUCTION LEAD TO ELIMINATION LEAD TO
IMPROVEMENT MAINTENANCE
07/29/09 SPC CM NNCPL 19
LOCAL ACTIONS
1. Are usually required to eliminate special causes of
variation
2. Can usually be taken by people very close to the
process
• Can correct typically about 15% of process problems

ACTIONS ON THE SYSTEM

1. Are usually required to reduce the variation due to
common causes
2. Almost always require Management action for
correction
3. Are needed to correct typically about 85% of process
problems
07/29/09 SPC CM NNCPL 20
 TOOLS FOR CONTROL

HISTOGRAM ----------- BELL SHAPE

CONTROL CHART-- NOT OUT OF CONTROL

NORMAL PLOT ----- STRAIGHT LINE

07/29/09 SPC CM NNCPL 21

 NORMAL DISTRIBUTION

TAKE A SAMPLE OF 100

ASSUME MEAN = 50.00
S. D. = 1.5
48.5 -------------51.5 (+/-1 sigma) 68.26%

47.0--------------53.0 (+/-2 Sigma) 95.44%

45.5--------------54.5 (+/- 3Sigma) 99.73%

07/29/09 SPC CM NNCPL 22

NORMAL DISTRIBUTION
A continuous, symmetrical, bell-shaped frequency distribution
for variable data, that is the basis for the control charts for
variables. When measurements have a normal distributions,
About 68.26% of all individuals lie within plus or minus one
standard deviation unit of the mean,
While about 95.44% lie within plus and minus two standard
deviation units of the mean,
While about 99.73% lie within plus and minus three standard
deviation units of the mean.
These percentages are the basis for control limits and control
chart analysis (since subgroup averages tend to be normally
distributed even if the output as a whole is not), and for many
capability decisions (since the output of many industrial
processes follows the normal distribution).

07/29/09 SPC CM NNCPL 23

 NORMAL DISTRIBUTION

• WHEN AVERAGE, MEDIAN & MODE IS

SAME
• AVERAGE OF ALL VARIABLE DATA
FOLLOWS NORMAL DISTRIBUTION
• ELONGATION,UTS,TWIST,OVALITY,ETC.,
LIKE ONE SIDED DOES NOT FOLLOW
NORMAL DISTRIBUTION

07/29/09 SPC CM NNCPL 24

STATISTICAL CONTROL & PROCESS
CAPABILITY
A PROCESS FREE FROM ASSIGNABLE CAUSES
(PREDICTABLE PROCESS)
PROCESS CAPABILITY IS
A MEASURE OF INHERENT VARIATION
(MANAGEMENT CONTROLLABLE)

PROCESS CAPABILITY
Cp=Potential Process Capability Index
Cp = Tolerance / Total Variation (6 Sigma)

07/29/09 SPC CM NNCPL 25

 PROCESS CAPABILITY INDICES
TV Vs TOL PPM LEVELS
6 SIGMA > TOL Cp = 0.8
125 > 100 HIGHER REJECTION
6 SIGMA = TOL Cp = 1.0
100 = 100 2700 PPM
6 SIGMA < TOL Cp = 1.33
75 < 100 64 PPM
6 SIGMA < TOL Cp = 1.67
60 < 100 4 PPM

07/29/09 SPC CM NNCPL 26

 PROCESS CAPABILITY INDICES

• Cp DOES NOT SPECIFY WHERE THE PROCESS IS

CENTERED.
• HENCE WE NEED TO HAVE ONE MORE INDEX TO
MEASURE ACTUAL PROCESS CAPABILITY
• Cpk = ACTUAL PROCESS CAPABILITY INDEX
• Cpk = Min ( USL – AVG )/ 3 SIGMA,(AVG – LSL)/ 3
SIGMA)
• USE Cp AND Cpk TOGETHER
• Cpk CANNOT EXCEED Cp
• ln METHOD, MIRROR IMAGE TECHNIQUE FOR NON
NORMAL DATA

07/29/09 SPC CM NNCPL 27

 PROCESS CAPABILITY INDICES
Pp & Ppk Cp &Cpk
PROCESS PERFORMANCE PROCESS CAPABILITY
INDEX INDEX

USED DURING INITIAL ON GOING PROCESS

PROCESS STUDY DURING CAPABILITY STUDY
PPAP
CAN BE CAPTURED FOR USED ONLY FOR STABLE
STABLE AND PROCESSES
CHRONICALLY UNSTABLE
PROCESSES

07/29/09 SPC CM NNCPL 28

 PROCESS CAPABILITY INDICES
Pp & Ppk Cp &Cpk
CAPTURES VARIATION CAPTURES VARIATION
DUE TO BOTH COMMON & DUE TO COMMON CAUSES
SPECIAL CAUSES ONLY

SIGMA IS CALCULATED SIGMA IS CALCULATED

USING (n-1) FORMULA USING R bar / d2 FORMULA
USING ALL INDIVIDUAL
READINGS
Ppk > 1.67 Cpk > 1.33

07/29/09 SPC CM NNCPL 29

 CONTROL CHARTS

OBJECTIVES OF CONTROL CHART

• TO DETECT SPECIAL/ASSIGNABLE
CAUSES
• TO MAINTAIN THE ACHIEVED
PROCESS CAPABILITY
• TO IDENTIFY THE OPPORTUNITY FOR
IMPROVEMENT

07/29/09 SPC CM NNCPL 30

 CONTROL CHART STEPS

1. GATHER DATA
2. INITIAL STUDY
3. CALCULATE CONTROL LIMITS
4. INTERPRET FOR PROCESS CONTROL
5. INTERPRET FOR PROCESS CAPABILITY
6. ESTABLISH ONGOING CONTROL CHART
7. MONITOR, REVIEW AND IMPROVE
PROCESS CAPABILITY

07/29/09 SPC CM NNCPL 31

 1.GATHER DATA
• ELIMINATE OBVIOUS DEFICIENCIES
• IDENDTIFY THE FACTORS AFFECTING AVG.
& RANGE
• UNDERSTAND PROCESS THROUGH
MASTER CAUSE & WHY WHY ANALYSIS
• PLAN SAMPLE SIZE, FREQ, CONTROL
CHART,NO. OF SUBGROUPS,ETC.,

07/29/09 SPC CM NNCPL 32

 SELECTION OF CHARTS
X BAR & R CHART
• MOST SENSITIVE CHART
MEDIAN & R CHART
• EASY SHOP ACCEPTANCE AS NO CALCULATION OF
AVERAGE
X BAR & S CHART
• SUBGROUP IS > 9,
• EFFECTIVE VARIABILITY INDICATOR
X & MR CHART
• WHERE NOT SUITABLE FOR SUBGROUP SAMPLING
• INSPECTION IS COSTLY/ DESTRUCTIVE IN NATURE
• DIFFICULT TO INTERPRET

07/29/09 SPC CM NNCPL 33

GATHER DATA
• RECORD PROCESS EVENTS WHILE COLLECTION OF DATA
• RECORD AND PLOT READINGS SIMULTANEOUSLY
INTERPRETATION FOR CONTROL
• FOCUS ON RANGE CHART
• ALL 4 CONDITIONS ARE CHECKED AND APPROPRIATE
ANALYSIS MADE AND CAUSES IDENTIFIED FOR ALL 4 OUT OF
CONTROL CONDITIONS
INTERPRETATION FOR CAPABILITY
• CALCULATE SIGMA, 6 SIGMA, Cp,Cpk
• IF Cpk < 1.33 INITIATE CA TO IMPROVE THE VALUE
• INITIATE CA PLAN FOR PROCESSES NOT UNDER STATISTICAL
CONTROL
• IF Cpk >1.33 ESATBLISH ONGOING CONTROL

07/29/09 SPC CM NNCPL 34

ONGOING CONTROL
• FINALISE ONGOING AVGs FOR CONTROL CHART
• ESTABLISH UCL & LCL
• ESTABLISH REACTION PLAN
• TRAIN THE OPERATOR
• ESTABLISH TO CONTROL TO MONITOR ONLINE
CHARTING
• ONLINE MONORING TO ENSURE PROCESS IS
STOPPED IN THE EVENT OF OUT OF CONTROL
• ANALYSED, ACTIONS ARE INITIATED AS PER
REACTION PLAN
• RECORDS OF THE SAME IS MAINTAINED
• PERIODIC REVIEW OF CFT IS ORGANISED TO REVIEW
THE EFFECTIVENESS OF SPC

07/29/09 SPC CM NNCPL 35

EFFECTIVENESS OF SPC
MEASURED BY
• IMPROVED PROCESS KNOWLEDGE
• REVIEW AND REVISION OF UCL/LCL
• REDUCTION IN REJECTION ( ACTUAL Vs ESTIMATED)
• IMPROVED PRODUCTIVITY
• LESS INSPECTION AND ADJUSTMENT
• CUSTOMER SATISFACTION DUE TO LESS VARIATION
ATTRIBUTE CHART
P-CHART
- PROPORTION OF UNIT NONCONFORMING, SAMPLE SIZE NEED NOT
BE EQUAL
np-CHART
- NUMBER OF UNITS NONCONFORMING, SAMPLE SIZE MUST BE EQUAL
C – CHART
- NUMBER OF NONCONFORMITIES, SAMPLE SIZE MUST BE EQUAL
U – CHART
- NUMBER OF NONCONFORMITIES PER UNIT, SAMPLE SIZE NEED NOT
BE EQUAL

07/29/09 SPC CM NNCPL 36

OPERATOR’s ROLE FOR ONE SUBGROUP
PLOTTING

1. CHECK AS PER CONTROL PLAN

2. CALCULATE AVG/RANGE/INDIVIDUALS
3. PLOT THE SAME IN CONTROL CHART
4. CHECK THE PLOTTED POINT IS IN CONTROL ( REFER 4
CONDITIONS IN CONTROL CHART)
5. IF IT IS IN CONTROL, CONTINUE THE PROCESS
6. ELSE, STOP THE PROCESS, TAKE CORRECTIVE ACTION &
DISPOSITION ACTION AS PER REACTION PLAN
7. RECORD THE PROCESS EVENTS INCLUDING OUT OF
CONTROL CONDITIONS

07/29/09 SPC CM NNCPL 37

 EFFECT OF OVER ADJUSTMENT
• THIS IS THE PRACTICE OF TREATING EACH
DEVIATION FROM TARGET IS DUE TO SPECIAL
CAUSES

• TREATING A COMMON CAUSE AS SPECIAL CAUSE &

ADJUSTING THE PROCESS

• IF A STABLE PROCESS IS ADJUSTED ON THE BASIS

OF EACH MEASUREMENT MADE, THEN THE
ADJUSTMENT BECOMES AN ADDITIONAL SOURCE OF
VARIATION

• OVER ADJUSTMENT WILL INCREASE THE VARIATION

07/29/09 SPC CM NNCPL 38
 STATISTICAL PROCESS
CONTROL (SPC)

CASE STUDY WITH VARIOUS CHARTS

OF

SPC
By
A R SURESH

07/29/09 SPC CM NNCPL 39

 HISTOGRAM EXAMPLE
Reaction Yield
A team in one operating unit is interested in improving the
yield from a given batch reaction. Five batches of material
are made per day. The team wants to use a histogram to
provide them with a snapshot in time of yields over the past
month. Data for percent yield from the last 25 days are
given below.

07/29/09 SPC CM NNCPL 40

 Sample Number
Day Number 1 2 3 4 5
1 81.3 80.4 78.6 86.1 81.8
2 74.3 76.4 82.4 77.8 82.5
3 78.7 77.4 79.4 81.6 81.0
4 80.4 81.7 81.4 79.7 80.2
5 79.4 75.6 80.3 80.2 77.4
6 85.0 75.4 73.8 75.8 78.6
7 78.5 86.2 77.1 73.3 76.4
8 81.7 84.0 80.2 78.6 80.9
9 84.5 82.4 78.8 83.2 83.0
10 82.7 80.5 85.9 82.7 84.0
11 78.4 83.1 80.1 78.5 86.6
12 82.9 82.4 78.9 78.2 78.4
13 75.6 80.1 81.1 78.3 80.4
14 78.2 76.4 82.3 81.7 85.1
15 81.8 80.6 79.1 79.3 83.6
16 75.2 82.2 79.6 83.6 81.9
17 78.6 80.1 80.6 79.3 80.4
18 82.3 80.8 79.7 76.5 85.6
19 83.0 83.6 75.2 83.3 81.3
20 77.6 79.1 78.7 80.8 80.2
21 75.0 81.0 82.9 80.0 81.9
22 82.7 78.8 81.2 74.8 81.7
23 76.9 82.5 82.5 81.4 84.4
07/29/09
24 78.1 SPC
82.9 CM NNCPL
73.7 81.5 75.9 41
25 79.9 78.7 81.3 80.0 78.5
Use these data to construct a histogram for reaction
yield. After constructing the histogram, answer the
following questions.

What is the most common value ( the mode ) ?

How much variation is there in reaction yield?

Describe the pattern of variation.

Is there any evidence that something unusual

occurring in the process?
07/29/09 SPC CM NNCPL 42
 X bar – R CHART EXAMPLE
A team in a unit is interested in improving the yield from
a given batch reaction. The team decides to use a control
chart to determine if the process is in a statistical
control. The team also decides that the control chart will
be used n the future to help monitor the process over
time.
Once any Assignable causes are eliminated, the control
chart can be used to monitor attempts at process
improvement. Five batches of product are made each
day. This provides a frequent data plus a method of
rationally sub-grouping the data. The team decides to
use this historical data for the control chart. The data for
% yield from the last 25 days are given below.
07/29/09 SPC CM NNCPL 43
Day Sample1 Sample2 Sample3 Sample4 Sample5 Sum Avg(Xbar) RangeR
1 81.3 80.4 78.6 83.1 81.8 405.2 81.04 4.5
2 74.3 76.4 82.4 77.8 82.5 393.4 78.68 8.2
3 78.7 77.4 79.4 81.6 81.0 398.1 79.62 4.2
4 80.4 81.7 81.4 79.7 80.2 403.4 80.68 2.0
5 79.4 75.6 80.3 80.2 77.4 392.9 78.58 4.7
6 85.0 75.4 73.8 75.8 78.6 388.6 77.72 11.2
7 78.5 86.2 77.1 73.3 76.4 391.5 78.30 12.9
8 81.7 84.0 80.2 78.6 80.9 405.4 81.08 5.4
9 84.5 82.4 78.8 83.2 83.0 411.9 82.38 5.7
10 82.7 80.5 85.9 82.7 84.0 415.8 83.16 5.4
11 78.4 83.1 80.1 78.5 86.6 406.7 81.34 8.2
12 82.9 82.4 78.9 78.2 78.4 400.8 80.16 4.7
13 75.6 80.1 81.1 78.3 80.4 395.5 79.10 5.5
14 78.2 76.4 82.3 81.7 85.1 403.7 80.74 8.7
15 81.8 80.6 79.1 79.3 83.6 404.4 80.88 4.5
16 75.2 82.2 79.6 83.6 81.9 402.5 80.50 8.4
17 78.6 80.1 80.6 79.3 80.4 399.0 79.80 2.0
18 82.3 80.8 79.7 76.5 85.6 404.9 80.98 9.1
19 83.0 83.6 75.2 83.3 81.3 406.4 81.28 8.4
20 77.6 79.1 78.7 80.8 80.2 396.4 79.28 3.2
21 75.0 81.0 82.9 80.0 81.9 400.8 80.16 7.9
22 82.7 78.8 81.2 74.8 81.7 399.2 79.84 7.9
23 76.9 82.5 82.5 81.4 84.4 407.7 81.54 7.5
24 78.1 82.9 73.7 81.5 75.9 392.1 78.42 9.2
25 79.9 78.7 81.3 80.0 78.5 398.4 79.68 2.8
26 Avg of Xbar 80.1976
07/29/09
27 SPC CM
Avg ofNNCPL R Rbar 44
6.488
From table for constants:
(for 5 samples) d2=2.326,
A4=0.577, D4=2.114, D3=0

UCLR=D4*R bar =2.114*6.488 = 13.631, LCLR=0,

UCL/LCL X bar=(Xbar+&-A2*R bar)
= 80.1976+&-0.577*6.488
= 83.9412 &76.454,
Sigma = Rbar/d2 = 6.488/2.326 = 2.7893,
for Spec of 80.15+/-11.25,
Cpk=Min of (USL-Xbar)/3Sigma & (Xbar-LSL)/3Sigma
= Min(91.4-80.1976)/3*2.7893 & (80.1976-68.9)/3Sigma
= Min 1.3387 & 1.35
Cpk = 1.3387
07/29/09 SPC CM NNCPL 45
 CONTROL CHART X BAR – R CHART
PLANT DEPT. OPN. DATE OF CL. SPEC. 80.15+/- 11.25 PART No
M/C No. S SIZE/FREQ. CHARCTR. NAME.
X2bar =Mean=80.20 UCL= X2bar+A2*Rbar=83.94 LCL=X2bar-A2*Rbar=76.45 AVERAGE CHART

R bar =MEAN R = 6.488 UCL R = D4*R bar = 13.63 LCL R=D3*R bar=0 RANGE CHART

D/T 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

AD

IN

SUM

AVG

RNG
07/29/09 SPC CM NNCPL 46
SCALE FOR PLOTTING: FOR X BAR CHART 0.80/DIVISION, FOR R CHART 1.40/DIVISION:
 Questions on reaction yield problem.

1. What variation is being examined on the R chart?

2. Is the R chart in out of statistical control? What does this

mean?

3. If the R chart is in statistical control. What is the estimate of

the process standard deviation σ? What is 'σ measuring’?

4. What variation is being examined on the X chart?

5. Is the X chart in or out of statistical control? What does this

mean?

6. What should the team do next, i.e., where should it begin

looking for improving the yield of this reaction process?
07/29/09 SPC CM NNCPL 47
P – CHART (PROPORTION OF NCs CHART) FOR SAMPLE SIZE – CONSTANT
Situations where sample size is same everyday/ever time. For eg. Everyday 500 pcs are considered or each time a
mould of 4 cavities is used getting 4 Pcs every 2 hours.
i – No of days or No of times sample taken (once every day or 4 or several times in different shifts)
ni -- No of samples in a sample size (n1,n2,n3,n4,…………….ni are same)
nPi – No of NCs in each sample size (nP1,nP2,nP3,…………………….nPi)
Pi -- Proportion of NC occurring (nPi/ni) (nP1/n1,nP2/n2,nP3/n3,……….)
Pbar -- Average proportion of NC = (nPi Sum/ni x i Sum) = (nP1+nP2+nP3+………..+nPi)/ni x i
UCLP & LCLP = Pbar + & - [3 x Sq root {Pbar(1-P bar)/ni}]
Select a proper scale in P chart to accommodate all values of Pis with Pbar as CL & plot the Pi values on different
dates.

1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 i Sum
i-date 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 =25
Time

ni Sample 5k 5k 5k 5k 5k 5k 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 ni sum
Size k k k k k k k k k k k k k k k k k k k 12500

nPi No of 12 15 19 13 9 26 1 1 1 1 1 2 1 3 1 1 1 1 2 1 8 1 1 1 1 nPi sum

NCs 8 4 7 8 6 4 1 1 6 0 6 7 0 5 3 2 7 8 405

Pi prpn of 24 30 38 26 18 52 3 2 3 3 3 4 2 6 3 2 3 3 4 3 1 2 2 3 3
NC μ μ μ μ μ μ 6 8 4 6 2 8 2 2 2 0 2 4 0 0 6 6 4 4 6
μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ

P bar = nPi sum/ni sum = 405/12500 = 0.0324, UCLP & LCLP = 0.0324 + & - [3 x Sq root{0.0324 x 0.9676/500}] = 0.0324 + & - 0.0237 = 0.0561 &
0.0087, Process Capability CPk = (1-P bar) = 1-0.0324 = 0.9676 = 96.76 % (Scale 0.0012 per division & 0.006 per 5 for plotting).
P chart analysis for variation: 1) beyond Control Limits – Evidence of instability at that point because of Spl causes ,calling for analysis &
review. 2) Beyond UCL –not good – control is not proper – plotting may be wrong – process Perf worsened – Measurement system & or
operator changed. 3) beyond LCL – good – process Perf improved – to be studied for further implementation - Measurement system & or
operator changed.

07/29/09 SPC CM NNCPL 48

P – CHART (PROPORTION OF NCs CHART) FOR SAMPLE SIZE – NOT CONSTANT
Situations where sample size is not same everyday/ever time. For eg. Everyday different qty is considered or each time a mould of different
No. of cavities is used every 2 hours.
i – No of days or No of times sample taken (once every day or 4 or several times in different shifts)
ni -- No of samples in a sample size (n1,n2,n3,n4,…………….ni is not same), n bar = mean sample size..
nPi – No of NCs in each sample size (nP1,nP2,nP3,…………………….nPi)
Pi -- Proportion of NC occurring (nPi/ni) (nP1/n1,nP2/n2,nP3/n3,……….)
Pbar -- Average proportion of NC = (nPi Sum/ni x i Sum) = (nP1+nP2+nP3+………..+nPi)/ni x i
UCLP & LCLP = Pbar + & - [3 x Sq root {Pbar(1-P bar)/n bar}]
Select a proper scale in P chart to accommodate all values of Pis with Pbar as CL & plot the Pi values on different dates.

1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 i Sum
i-date 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 =25
Time

ni Sample 96 12 80 14 13 99 1 1 1 5 1 1 1 1 1 2 1 9 1 1 3 1 1 1 1 ni sum
Size 8 16 4 01 76 5 2 0 1 4 3 0 7 3 1 3 3 7 0 2 9 4 2 3 1 29858
0 2 8 2 2 6 2 0 9 0 6 3 5 4 2 3 2 5 8
2 8 4 5 6 1 5 0 6 5 8 4 3 5 2 7

nPi No of 8 13 13 16 14 15 1 1 2 1 1 1 1 9 1 9 1 5 1 1 1 1 1 1 2 nPi sum

NCs 3 0 4 8 6 7 9 4 3 5 9 0 7 3 5 1 356

Pi prpn of 8μ 11 16 10 15 1 1 2 3 1 1 1 7 1 4 1 5 1 1 2 1 1 1 1
NC μ μ 11 μ μ 1 0 0 3 2 6 1 μ 2 μ 0 μ 4 5 6 2 1 1 8
μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ μ

n bar =29858/25 = 1194, P bar = nPi sum/ni sum = 356/29858 = 0.012, UCLP & LCLP = 0.012 + & - [3 x Sq root{0.012 x 0.988/1194}] = 0.012 + & - 0.0095 =
0.0215 & 0.0025, Process Capability CPk = (1-P bar)
= 1-0.012 = 0.9880 = 98.80 % (Scale 0.0005 per division & 0.0025 per 5 for plotting).
P chart analysis for variation: 1) beyond Control Limits – Evidence of instability at that point because of Spl causes ,calling for analysis & review. 2) Beyond
UCL –not good – control is not proper – plotting may be wrong – process Perf worsened – Measurement system & or operator changed. 3) beyond LCL –
good – process Perf improved – to be studied for further implementation - Measurement system & or operator changed.

07/29/09 SPC CM NNCPL 49

C – CHART ( NUMBER OF NCs CHART) FOR SAMPLE SIZE IS ONE
Situations where one sample is taken everyday/ever time. For e.g. Everyday 1 pc is taken or each time a mould of 4 or more cavities is used
every 2 hours & each time a piece is checked for all selected characteristics (hence C – chart) & for NCs in them.
i – No of days or No of times sample taken (once every day or 4 or several times in different shifts)
ni -- No of samples in a sample size (n1,n2,n3,n4,…………….ni are same & being 1)
nPi – No of NCs in each sample size (nP1,nP2,nP3,…………………….nPi)
Pi -- Proportion of NC occurring (nPi/ni) (nP1/n1,nP2/n2,nP3/n3,……….) not applicable.
Cbar -- Average proportion of NC = (nPi Sum/ni x i Sum) = (nP1+nP2+nP3+………..+nPi)/ i Sum
UCLC & LCLC = C bar + & - 3 x Sq root C bar
Select a proper scale in C chart to accommodate all values of Pis & plot the Pi values on different dates.

1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 i Sum
i-date 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 =25
Time

ni 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ni sum
Sample 25
Size

nPi No 9 15 11 8 17 11 5 1 1 7 1 1 4 3 7 2 3 3 6 2 7 9 1 5 8 nPi
of NCs 1 3 0 2 sum
189

C bar = nPi sum/i sum = 189/25 = 7.56, UCLC & LCLC = 7.56 + & - 3 x Sq root 7.56 = 7.56 + & - 8.25
= 15.81 & 0 (ignore negative value -0.69& make it 0), Process Capability CPk = (100-C bar) = 100-7.56
= 92.44 % (Scale 0.4 per division & 2.0 per 5 for plotting). As LCL is 0, C bar line (as bold line) can be referred for any action for correction or
improvement.
C chart analysis for variation: 1) beyond Control Limits – Evidence of instability at that point because of Spl causes, calling for analysis &
review. 2) Beyond UCL –not good – control is not proper – plotting may be wrong – process Perf worsened – Measurement system & or
operator changed. 3) Beyond LCL – good – process Perf improved – to be studied for further implementation - Measurement system & or
operator changed.

07/29/09 SPC CM NNCPL 50

nP – CHART (NUMBER OF NCs FOR A PARTICULAR CHARACTRISTIC) FOR SAMPLE SIZE – CONSTANT
Situations where sample size is same everyday/ever time. For eg. Everyday 62 pcs are considered or each time a mould of 4 cavities is
used getting 4 Pcs every 2 hours & a particular characteristic say CD 1 is considered for study.
i – No of days or No of times sample taken (once every day or 4 or several times in different shifts)
ni -- No of samples in a sample size (n1,n2,n3,n4,…………….ni are same)
nPi – No of NCs in each sample size (nP1,nP2,nP3,…………………….nPi)
Pi -- Proportion of NC occurring (nPi/ni) (nP1/n1,nP2/n2,nP3/n3,……….) not applicable as only one characteristic is considered.
nP bar -- Average proportion of NC = (nPi Sum/i Sum) = (nP1+nP2+nP3+………..+nPi)/i sum
UCLP & LCLP = nP bar + & - [3 x Sq root {nP bar(1-P bar)}]
P bar – Average No. of Nc in each piece = nP bar/ni
Select a proper scale in P chart to accommodate all values of Pis & plot the Pi values on different dates

1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 i Sum
i-date 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 =25
Time

ni 62 62 62 62 62 62 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 NA
Sample 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
Size

nPi No 2 5 4 3 3 6 5 0 7 5 4 1 2 3 6 3 8 4 4 4 6 4 2 3 7 nPi
of NCs sum
101

nP bar = nPi sum/i sum = 101/25 = 4.04, P bar = nP bar/ni = 4.04/62 = 0.0652, UCLP & LCLP = 4.04 + & - [3 x Sq root{4.04 x 0.9348} = 4.04 + &
- 5083 = 9.87 & 0 (ignore negative value & make it 0), Process Capability CPk = (1-P bar) = 1-0.0652 = 0.9348 = 93.48 % (Scale 0.22 per
division & 1.1 per 5 for plotting).
nP chart analysis for variation: 1) beyond Control Limits – Evidence of instability at that point because of Spl causes ,calling for analysis &
review. 2) Beyond UCL –not good – control is not proper – plotting may be wrong – process Perf worsened – Measurement system & or
operator changed. 3) beyond LCL – good – process Perf improved – to be studied for further implementation - Measurement system & or
operator changed.

07/29/09 SPC CM NNCPL 51

U– CHART (PROPORTION OF NCs CHART) FOR SAMPLE SIZE – NOT CONSTANT
Situations of dock audit, pre dispatch audit where sample size is not same everyday/ever time. For eg. Everyday certain No. of units like
boxes are checked for all NCs.
i – No of days or No of times sample taken
ni -- No of samples in a sample size (n1,n2,n3,n4,…………….ni is not same), n bar = mean sample size..
nPi – No of NCs in each sample size (nP1,nP2,nP3,…………………….nPi)
Ui -- Proportion of NC occurring (nPi/ni) (nP1/n1,nP2/n2,nP3/n3,……….)
U bar -- Average proportion of NC = (nPi Sum/ni Sum) = (nP1+nP2+nP3+………..+nPi)/ni Sum
UCLU & LCLU = U bar + & - [3 x Sq root {U bar/nbar}]
Select a proper scale in U chart to accommodate all values of Pis with U bar as CL & plot the Ui values on different dates.

1 2 3 4 5 6 7 8 9 1 1 1 1 1 i Sum
i-date 0 1 2 3 4 = 14
Time

ni 8 8 9 8 8 7 7 8 8 8 7 8 9 9 ni sum
Sample 112
Size

nPi No 8 17 18 15 23 9 1 6 1 1 1 1 1 2 nPi
of NCs 9 4 7 3 5 6 2 sum
212

Pi prpn 1. 2. 2. 1. 2. 1.3 2. 0. 1. 2. 1. 1. 1. 2.
of NC 0 1 0 9 9 7 8 8 1 9 9 8 4

n bar =112/14 = 8, U bar = nPi sum/ni sum = 212/112 = 1.89, UCLU & LCLU = 1.89 + & - [3 x Sq root{1.89/8}] = 1.89 + & - 1.46 = 3.35 & 0.43,
Process Capability CPk = (100-U bar) = 100 – 1.89 = 98.11 % (Scale 0.08 per division & 0.4 per 5 for plotting).
U chart analysis for variation: 1) beyond Control Limits – Evidence of instability at that point because of Spl causes ,calling for analysis &
review. 2) Beyond UCL –not good – control is not proper – plotting may be wrong – process Perf worsened – Measurement system & or
operator changed. 3) beyond LCL – good – process Perf improved – to be studied for further implementation - Measurement system & or
operator changed.

07/29/09 SPC CM NNCPL 52