Statistical Quality Control: Unit V

Unit V
STATISTICAL QUALITY
CONTROL
B. Thilaka
Applied Mathematics
Statistical Quality Control
• Quality Control – Process control &
acceptance sampling.
• Tolerance Limits
• Acceptance Sampling
• Quality- measurable / countable property
of a product
• Potency of a drug, breaking strength of a
yarn, outside diameter of a ball bearing,
number of imperfections in a piece of
cloth,….
Quality Control
The concept of quality control in manufacturing was
first advanced by Walter Shewhart
of the Bell Telephone Laboratories. He issued a
memorandum on May 16, 1924 that featured a
sketch of a modern control chart.
1931 - published a book on statistical quality
control, "Economic Control of Quality of
Manufactured Product"
set the tone for subsequent applications of
statistical methods to process control.
Quality Control
Two other Bell Labs statisticians, H.F.
Dodge and H.G. Romig spearheaded
efforts in applying statistical theory to
sampling inspection.
The work of these three pioneers
constitutes most of the theory of statistical
quality and control
Quality Control - Statistical
Process Control
Typical process control techniques:

1. Histograms
2. Check Sheets
3. Pareto Charts
4. Cause and Effect Diagrams
5. Defect Concentration Diagrams
6. Scatter Diagrams
7. Control Charts
Statistical Process Control
The underlying concept of statistical
process control is based on a comparison
of what is happening today with what
happened previously.
Historical data is used to compute the

initial control limits. The present data are
compared against these initial limits.
Statistical Process Control
• Points that fall outside of the limits are
investigated and, perhaps, some will
later be discarded.
• If so, the limits would be recomputed
and the process repeated.
• This is referred to as Phase I
(retrospective anlaysis of data).
• Real-time process monitoring, using
the limits from the end of Phase I, is
Phase II (prospective monitoring of
future data).
Statistical Quality Control is the process of

inspecting enough product from given lots
to probabilistically ensure a specified
quality level.
Statistical quality control refers to the use of
statistical methods in the monitoring and
maintaining of the quality of products and
services.
Statistical Process Control is concerned
with achieving, maintaining and further
improving a good process that produces
desirable results reliably and consistently
over time.
Applicable to any process that produces

data over time according to some
underlying statistical model – health care,
manufacturing process.
Why and How do we achieve?
• The production processes are not perfect.
• Hence the output of these processes will not be
perfect – correct and deterministic.
• Successive runs of the same production
process will produce distinct(non-identical)
products (each fingerprint is unique -
variation).
•Alternately, seemingly similar runs of the

production process will vary, by some
degree, and impart the variation into the
some product characteristics.
•Because of these variations in the products,

we need probabilistic models and robust
statistical techniques to analyze quality of
such products.
However well a production process is
controlled, these quality measures will vary
from item to item and hence there will be a
probability distribution associated with the
population of such measurements
If all important sources of variations are

under control in a production process, then
the slight variations among the quality
measurements will not cause any serious
problem.
Such a process should produce the same
distribution of quality measurements no matter
when it is sampled, thus ensuring that this is a
“stable system.”
Variations can be broadly classified into
two types.
Chance Variation & Assignable Variation
Chance(common) Variation – Variation in

the quality of the product which occurs due
to minor but random causes such as slight
changes in temperature, pressure, metal
hardness.
Assignable Variation – Variation that
occurs due to non-random causes like
poor quality raw material, mechanical
faults.
A process is said to be in a state of

statistical control, if the variability present
in the process is confined to chance
variation.
Statistical control is achieved by finding
and eliminating all assignable variations.
Once statistical control is achieved, it is

essential to maintain the same .
Control charts (Shewart charts) are

principally used to detect serious
deviations from a state of statistical control
when or before(if possible) they occur.
Control charts
Control charts
Measurements(variables) Attributes
(Mean and Range) (p, np and c)
Control charts
Control charts for measurements:
Observations are measurements
Variables are the quality characteristics of a
product that are measurable.
Control charts for attributes:

Observations are count data (the
number of defectives in a sample of given data)
Attributes are the quality characteristics of a
product that are identified by their presence or
absence
Control charts
Control chart consists of
a central line corresponding to the average

quality at which the process is to perform
lines corresponding to the upper and lower

control limits – limits are chosen so that
values falling between them can be attributed
to chance, while values falling beyond them
are interpreted as indicating a lack of control.
(warning limits)
Control charts
Plotting the results obtained from samples
taken periodically at frequent intervals can
help determine whether the process is
under control or any assignable variation
has entered the process.
When a sample point falls beyond the

control limits or even when the points fall
within the control limits if a trend or some
other pattern is visible, preventive action is
needed.
Control charts
The control limits are used to assess the

stability of the process and to identify
unusual events (outliers)
Control charts are thus quite useful both

for monitoring if processes get worse and
for testing and verifying improvement
ideas
Control charts
A control chart will tell whether a process

is out of control, unstable, or
unpredictable.
A stable process produces predictable

results consistently.
Processes that are “out of control” need to

be stabilized before they can be improved.
Control charts
Process is in control:
No points outside the control limits
The number of points above and below the
central line are about the same.
Points are seemingly random above and below
the central line.
Most points are near central line, and only a
few are close to control limits.
Basic Assumption: Central Limit Theorem

(Unit II)
Control charts
Process monitoring with control charts is

an important component within an overall
process evaluation and improvement
framework in healthcare.
Control charts
Control chart methods are used
extensively in health care
WHERE? HOW?
To verify improvements in the time

between insulin injections.
Control charts
In cases of epidemiological concerns
such as surgical- site infections,
bacteremia, Clostridium difficile toxin-
positive stool assays, medical intensive-
care-unit (ICU) nosocomial infections, and
needlestick injuries.
Can you relate to something else?
EBOLA
Control charts
In many industries, processes will be
divided into stages and these principles of
SQC are applied to the individual stages
separately.
Unless a product is found to meet the
specified norms, it will not be permitted to
move to the next stage.
This stage-wise SQC scheme ensures that
the desired quality level is maintained at
all levels.
Control charts
Examples???
Food Processing
Wine-making
The right type of grapes,

Crushing and fermentation,
Processing and aging
Control charts for measurements
The x chart :
Determination of the three lines:
If the process mean µ and standard
deviation σ are known and if it is
reasonable to assume that the
measurements are samples from a normal
population (ideal case), then we assert
with probability (1-α) that the mean of a
random sample of size n will fall between
   
  z / 2  and  z / 2  
 n  n
The x chart :
Central line is given by -------------
LCL is given by ------------------
UCL is given by ------------------

The x chart :
It is common industrial practice to use
three-sigma limits obtained by substituting
3 for zα/2. (α=0.0027 – Recall properties of
Normal distribution)
Hence, Central line : µ

Control Limits : µ±Aσ, A=3/√n
If a control chart is being constructed for a
new process, the µ and σ (the population
parameters) will not be known and hence must
be estimated from the sample data.
For establishing control limits, the general rule
of thumb is that at least 30 time points be
sampled before the control limits are
calculated.
For each of the k samples, we compute the
sample mean, the sample variance, and the
range .
The x chart :
To form an unbiased estimator of µ from
k
1
sample means X j we calculate X  k  X . j
j 1
 n 1  2
S `   S j
2
Further j is a better estimator than
 n 
S 2j . The average of the above values is
not an unbiased estimator of σ2.
However, computation of the mean
standard deviation is avoided by using the
range data instead of calculating the
adjusted standard deviation.
The x chart:
1 k
We compute R   Rj.
k j 1
The control chart values for the x chart

are
Central line: X
Control Limits : X  A2 R
The R- chart :
While the x chart monitors the central
value of the process, the R-chart monitors
the variation of the process.
As in the case of the x chart, the three

values of the R- chart in the various cases
are as follows:
The R- chart :
If the underlying population is normal, with
σ known, then
Central line = d 2
UCL = D2
LCL = D1
The R- chart :
If σ is unknown, then it may be estimated
from past data as in the case of the x
chart. Further
Central line = R
UCL = D4 R
LCL = D3 R
Thus, there is a close connection between the use
of control charts and the testing of hypothesis. A
point on the x chart that is out of control
corresponds to a sample for which the null
hypothesis is rejected.
Specifically, control-chart techniques provide

sequential, temporally ordered sets of tests. The
position of the individual points, possible trends,
patterns exhibited by the points representing
successive samples provide an insight into the state
and progress of the process.
Remark: Analyze R-chart before the x chart. (unit

1. Given the values of the sample mean X
and sample range R for 10 samples, each
of size 5. Draw the appropriate mean and
range charts and comment on the state of
control of the process.
Sample No. 1 2 3 4 5 6
Mean 43 49 37 44 45 37
Range 5 6 5 7 7 4
Sample No. 7 8 9 10
Mean 51 46 43 47
Range 8 6 4 6
Solution:
1
Consider X 
N
X i
 43  49  37  ...  47

1
10
= 44.2
1
Also, R   Ri
N

1
5  6  5  ...  6 = 5.8
10
From the table of control chart constants,
for sample size n=5, we have
A2 = 0.577, D3 = 0 and D4 = 2.115.
Hence Control Limits for X chart are:
Central Line : X = 44.2

LCL = X - A2 R= 44.2- 0.577x5.8 =40.85
UCL = X + A2 R = 44.2+0.577x5.8 =47.55
Further, the Control Limits for R chart
are:
CL= R =5.8
LCL = D3 R =0
UCL= D4 R =2.115x5.8=12.27.
We now draw the required two control

charts
State of Control:
Since all the points in the range chart lie
within the control lines, the process is under
control as far as the variability of the sample
values is concerned.
However in the mean chart two points lie above

the upper control line and two points lie below
the lower control line. The process is thus not
under control as far as the average of the
sample values is concerned.
On the whole, the process is out of control.

2. A machine fills boxes with dry cereal. 15
samples of 4 boxes are drawn randomly. The
weights of the sample boxes are shown as
follows. Draw the control charts for the sample
mean and sample range and determine whether
the process is in a state of control.
Sample Number : 1 2 3 4 5
Weights of boxes(X): 10.0 10.3 11.5 11.0 11.3
10.2 10.9 10.7 11.3 11.6
11.3 10.7 11.4 10.7 11.0
12.4 11.7 12.4 11.4 12.1
11.4 11.4 12.1 13.1 12.1
10.7 11.1 12.7 13.1 10.7
11.0 10.3 10.7 12.4 11.5
11.9 12.1 11.1 12.1 11.9
11.8 11.6 12.1 13.1 11.7
11.3 11.4 11.7 12.0 12.1
Solution:
Since the given data is raw data, we first
compute the means and ranges of the given
samples.
Sample 1 2 3 4 5 6 7
No. (i)
X 43.9 43.6 46.0 44.2 46.9 43.8 44.1

Xi 11.0 10.9 11.5 11.1 11.7 11.0 11.0
Ri 2.9 1.4 1.7 0.7 0.8 0.7 1.1
(i) 8 9 10 11 12 13 14 15
X 47.8 49.6 45.6 47.5 47.0 47.0 49.1 46.3

Xi 12.0 12.4 11.4 11.9 11.8 11.8 12.3 11.6
Ri 2.0 2.1 1.4 1.2 0.7 1.0 1.2 1.5
1
Now, X   Xi
N
 11.0  10.9  11.5  ...  11.6

1
15
173.4 = 11.56

15
1
Further, R
N
 Ri
 2.4  1.4  1.7  ...  1.5

1
15
19.9

15
= 1.33
From the table of control chart constants, for

sample size n=4, we have
A2 = 0.729, D3 = 0 and D4 = 2.282.
Hence Control Limits for X chart are:

LCL = X - A2 R = 11.56- 0.729x1.33=10.59
UCL = X + A2 R = 11.56+0.729x1.33=12.53
Further, the Control Limits for R chart are:
CL= R =1.33
LCL = D3 R =0
UCL= D4 R =2.282x1.33=3.04
We now draw the required two control
charts as given below:
State of Control:
Since all the points lie within the upper
and lower control lines in both the mean
chart and the R-chart, the process is under
control.
3. The values of sample mean and sample

standard deviation for 15 samples each of
size 4, drawn from a production process are
given below. Draw the appropriate control
charts for the process average and process
variability. Comment on the state of control.
Sample No. 1 2 3 4 5 6
Mean 15.0 10.0 12.5 13.0 12.5 13.0
S. D. 3.1 2.4 3.6 2.3 5.2 5.4

7 8 9 10 11 12 13 14 15
13.5 11.5 13.5 13.0 14.5 9.5 12.0 10.5 11.5
6.2 4.3 3.4 4.1 3.9 5.1 4.7 3.3 3.3
Solution:
HW
4. The following data gives the coded measurements
of 10 samples each of size 5, drawn from a
production process at intervals of 1 hour. Draw the
appropriate control charts and comment on the state
of control of the process.
Sample 1 2 3 4 5 6
Number(i)
9 10 10 8 7 12
Coded 15 11 13 13 9 15
Measurements 14 13 8 11 10 7
(X) 9 6 12 10 4 16
13 10 7 13 5 10
(i) 7 8 9 10
9 15 10 16
9 15 13 14
(X) 9 10 14 12
13 13 7 14
5 17 11 14
Solution: HW
Control charts for attributes
The control charts seen earlier plots
quality characteristics that can be
measured and expressed numerically. We
measure weight, height, position,
thickness, etc.
If a particular quality characteristic cannot
be represented numerically, or if it is
impractical to do so, then using a quality
characteristic to sort or classify an item
that is inspected into one of two "buckets“
is a good alternate option.
An example of a common quality characteristic

classification would be designating units as
"conforming units" or "nonconforming units".
Another quality characteristic criteria would be

sorting units into "non defective" and "defective"
categories.
Quality characteristics of these types are called

attributes.
We observe that there is a difference between
"nonconforming to an engineering specification"
and "defective" -- a nonconforming unit may still
function fine and be, in fact, not defective at all,
while a part can be defective and not function as
desired.
Examples of quality characteristics that are

attributes
the number of failures in a production run
the proportion of malfunctioning wafers in a lot
Two fundamental control charts for

attributes:
Fraction-defective chart called a “p chart”
Number-of –defects chart called a “c chart”

p-chart is a type of control chart used to
monitor the proportion of nonconforming
units in a sample, where the
nonconforming sample proportion is
defined as the ratio of the number of
nonconforming units to the sample size, n
c-chart is a type of control chart used to

monitor "count"-type data, typically total
number of nonconformities per unit. It is
also occasionally used to monitor the total
number of events occurring in a given unit
of time.
The c-chart differs from the p-chart in that it
accounts for the possibility of more than one
nonconformity per inspection unit, and that
(unlike the p-chart and u-chart) it requires a fixed
sample size.
The p-chart models "pass"/"fail"-type inspection

only, while the c-chart (and u-chart) give the
ability to distinguish between (for example) 2
items which fail inspection because of one fault
each and the same two items failing inspection
with 5 faults each; in the former case, the p-chart
will show two non-conformant items, while the c-
chart will show 10 faults.
Nonconformities may also be tracked by
type or location which can prove helpful
in tracking down assignable causes.
Control charts for attributes(p)
Control Limits for fraction-defective chart :
(Link this to the sampling theory of
proportions – unit 3)
(To what distribution will we associate

this???)
BINOMIAL
(Ultimately, Normal curve approximation)

Control charts for attributes (p)
Hence, if a standard is given, that is, if the
fraction defective should take some pre-
assigned value p , the central line is p and
the three-sigma limits are used for the
control limits.
In other words,
Central line : p
p(1  p)
Control limits: p3
n
If no standard is given(practical situations),
p will have to be estimated from past data.
If k samples are available, di is the number

of defectives in the ith sample, and ni is the
number of observations in the ith sample, p
is estimated as the proportion of
defectives in the combined sample, as
d1  d 2  ...  d k
p
n1  n2  ...  nk
Hence the three control-chart values are
Central Line : p
p (1  p )
UCL : p3
n
p (1  p )
LCL : p 3
n
Note :
1. If p is small, as may often be the case, then,

the LCL may be a negative quantity. In this case,
the LCL is assumed to be zero and effectively,
only the UCL is used.
2. If the binomial distribution cannot be

approximated to a normal curve(when??? – unit
I), then and only then the UCL may be obtained
directly from the Binomial tables or the Poisson
approximation to the Binomial may be used.
Control charts for attributes (np)
The np-chart is a type of control chart used to
monitor the number of nonconforming units in a
sample.
It is an adaptation of the p-chart and used in

situations where it is easier to interpret process
performance in terms of concrete numbers of
units rather than the somewhat more abstract
proportion.
Control charts for attributes (np)
• The np-chart differs from the p-chart in only the
three following aspects:
• The control limits are np  3 np(1  p) , where n
is the sample size and p is the estimate of the
long-term process mean established during
control-chart setup.
• The number nonconforming (np), rather than the
fraction nonconforming (p), is plotted against the
control limits.
• The sample size, n, is constant.
Control charts for attributes (c)
In many situations, it is necessary to
control the number of defects in a unit of
product, rather than the fraction defective
or the number of defectives.
Example:
number of defects per hundred yards in
carpeting industry
number of defects per roll in newsprint
industry.
Control Limits for no-of-defects chart :
(To what distribution will we associate

this???)
POISSON - WHY???
Hence, the central line for a number-of-

defects chart is the parameter λ of the
Poisson distribution, and the three sigma
control limits is based on the fact that the
standard deviation of this distribution is √λ.
(what will the limits be ?)
If λ is unknown, that is, if no standard

value is given, then its value is estimated
from at least 20 values of c observed from
past data.
If k is the number of units of product available for
estimating λ, and if ci is the number of defectives
in the ith unit, then λ is estimated by c  1 c
k
k

i 1
i
The control chart values are

Central line : c
UCL : c  c
LCL : c  c
Note:
1. np-chart (sample sizes are the same) and p-

chart (sample sizes may be different – in this
case 0.75n  ni  1.25n ,for all i)are used
when p ≥0.05 or np ≥4
2. c-chart is used when c c

≥4 or when
is small compared with the maximum number
of defects in the given data.
5. 15 samples of 200 items each were
drawn from the output of a process. The
number of defective items in the samples
are given below. Prepare a control chart
for the fraction defective and comment on
the state of control.
Sample No. (i) 1 2 3 4
No. of defective (np) 12 15 10 8

(i) 5 6 7 8 9 10
(np) 19 15 17 11 13 20
`(i) 11 12 13 14 15
(np) 10 8 9 5 8
Solution:
Consider ∑np=12+15+10+..+5+8=180
1
 np   np
N
180
 = 12
15
Also,
12
p (since each sample
200
contains 200 items).
The control limits are thus given by
Central line :
p  0.06
LCL : p  3 p(1  p)
n
0.06 x0.094
 0.06  3
200
= 0.01
UCL : p (1  p )
p3
n
0.06 x0.094
 0.06  3
200
=0.11
The fraction defectives (values of p) for the
given samples are:
0.06, 0.075, 0.05, 0.04, 0.095, 0.075,
0.085, 0.055, 0.065, 0.1, 0.05, 0.04, 0.045,
0.025, 0.04.
Since all the sample points lie between the

LCL and UCL lines, the process is under
control.
6. Construct a control chart for defectives for the
following data and comment on the state of
control
Sample No. : 1 2 3 4
No. inspected : 90 65 85 70
No. of defectives : 9 7 3 2
5 6 7 8 9 10
80 60 70 95 90 75
9 5 3 9 6 7
Solution:
Since the size of the sample varies from
sample to sample, we do not construct the
np-chart.
We use the p-chart only if 0.75n  ni  1.25n
for all i.
1
Now, n   ni = (90+65+…+90+75)/10
N = 800/10 = 80
The values of ni are all between 60 and
100 .
Hence, we now draw the p-chart.
Now, p = (Total no. of defectives)/

(Total no. of items inspected)
= 60/800 = 0.075
Hence the control limits are given by

CL : p  0.075
p(1  p) 0.075 x0.025
LCL : p 3  0.075  3
n 80
= -0.013
Since LCL cannot be negative, it is assumed
to be zero.
p (1  p) 0.075 x0.025
p 3  0.075  3  0.163
UCL : n 80
The values of pi for the various samples

are 0.100, 0.108, 0.035, 0.029, 0.113,
0.063, 0.043, 0.095, 0.067, 0.093
Since all the sample points lie within the

control limits, the process is under control.
7. In a factory producing spark plugs, the
number of defectives found in the
inspection of 15 lots of 100 each is given
below. Draw the control chart for the
number of defectives and comment on the
state of control.
Sample No. (i) 1 2 3 4

No. of defectives (np) 9 10 12 8
(i) 5 6 7 8 9 10
(np) 6 4 6 3 4 5
(i) 11 12 13 14 15
(np) 4 7 9 3 4
Solution:
Consider ∑np = 5+10+12+….+3+4=90

1
 np   np
N
= 90/15 = 6
and p = 6/100 = 0.06 ( since each sample
contains 100 items)
The control limits for the np-chart are
CL : np =6
LCL : np  3 np(1  p) = 6-3√(6x0.94)=-1.12
Since the LCL cannot be negative, LCL=0.
UCL : np  3 np(1  p) = 6+3√(6x0.94)

= 13.12
Since all the sample points lie between the

upper and lower control limits, the process
is under control.
8. 10 samples each of size 50 were
inspected and the number of defectives in
the inspection were 2,1,1,2,3,5,5,1,2,3.
Draw the appropriate control chart for the
defectives and comment on the state of
control.
Solution : HW
(Hint : The sample sizes are the same.

Either chart, p or np will do)
9. 15 tape recorders were examined for
quality control test. The number of defects in
each tape recorder is recorded below. Draw
the appropriate control chart and comment
on the state of control.
Unit No. (i) 1 2 3 4 5 6

No. of defects (c) 2 4 3 1 1 2
7 8 9 10 11 12 13 14 15
5 3 6 7 3 1 4 2 1
Solution:
Since the number of defects per sample
containing only one item is given, we use
the c-chart.
1
Now, c   ci
N
 2  4  3  ...  2  1
1
15
= 45/15=3
Aside: Observe that even though c <4,
we use only the c-chart at this point of
time.
We now compute the three control limits :
CL : c = 3
LCL : c  3 c = 3-3√3 = -2.20
Since LCL cannot be negative, LCL=0
UCL : c  3 c = 3+3√3 = 8.20
Since all the points lie within the control
limits, the process is under control.
10. A plant produces paper for newsprint

and rolls of paper are inspected for
defects. The results of inspection of 20
rolls of papers are given below. Draw the c
chart and comment on the state of control.
Roll No (i) 1 2 3 4 5
No. of defects (c) 19 10 8 12 15
(i) 6 7 8 9 10 11 12 13
(c) 22 7 13 18 13 16 14 8
14 15 16 17 18 19 20
7 6 4 5 6 8 9
Solution : HW
Tolerance Limits
In quality control, the limiting values between
which measurements must lie if a product is to
be accepted is loosely defined as tolerance
limits.
(Contrast this with confidence limits of unit 3)
(Confidence limits are the lower and upper

boundaries / values of a confidence interval, that
is, the values which define the range of a
confidence interval).
Tolerance Limits
Recall ”A confidence interval gives an

estimated range of values which is likely to
include an unknown population parameter,
the estimated range being calculated from
a given set of sample data.”
If reliable information is available about the

distribution underlying the measurement in
question, the natural tolerance limits are
clearly easily obtained.
Tolerance Limits
In most practical situations, the true values
of µ and σ are not known and therefore,
the tolerance limits are based on the mean
and standard deviation of a random
sample.
While in a normal population, µ±1.96σ are

the tolerance limits that include 95% of the
population, the same cannot be said of for
the limits based on sample mean and
sample standard deviation.
Tolerance Limits
However, it is possible to determine a
constant K so that “one can assert with
(1-α) 100% confidence that the proportion
of the population contained between
x  Ks and x  Ks is at least P”, the
given proportion of the population.
Can you now differentiate between

confidence limits(of unit 3) and tolerance
limits?
Tolerance Limits
Confidence limits are used to estimate a
parameter of a population.
Tolerance limits are used to indicate between

what limits one can find a certain proportion of
the population
Further observe that when n becomes large, the

length of a confidence interval approaches zero
while the tolerance limits approach the
corresponding values of the population.
Acceptance Sampling
Acceptance sampling is an important field
of statistical quality control that was
popularized by Dodge and Romig and
originally applied by the U.S. military to the
testing of bullets during World War II. If
every bullet was tested in advance, no
bullets would be left to ship. If, on the
other hand, none were tested,
malfunctions might occur in the field of
battle, with potentially disastrous results.
Acceptance Sampling
Dodge reasoned that a sample should be
picked at random from the lot, and on the
basis of information that was yielded by
the sample, a decision should be made
regarding the disposition of the lot. In
general, the decision is either to accept or
reject the lot. This process is called Lot
Acceptance Sampling or just Acceptance
Sampling.
Acceptance Sampling
Acceptance sampling is "the middle of the road"
approach between no inspection and 100%
inspection. There are two major classifications of
acceptance plans: by attributes ("go, no-go") and
by variables. The attribute case is the most
common for acceptance sampling.
The main purpose of acceptance sampling is to

decide whether or not the lot is likely to be
acceptable, not to estimate the quality of the lot.
Acceptance Sampling
Acceptance sampling is employed when one or
several of the following hold:
Testing is destructive
The cost of 100% inspection is very high
100% inspection takes too long
This procedure is equivalent to a test of the null

hypothesis that the proportion defective p in the
lot equals some specified value p0 against the
alternative that it equals p1, where p1>p0.
(One tailed test- WHY?? -up for discussion)
Acceptance Sampling
11. The specifications for a certain quality
characteristic are 15.0±6.0 ( in coded
values). 15 samples of 4 readings each
gave the following values for X and R.
Sample No. (i) : 1 2 3

X : 16.1 15.2 14.2
R : 3.0 2.1 5.6
Acceptance Sampling
(i): 4 5 6 7
X : 13.9 15.4 15.7 15.2
R : 2.4 4.1 2.7 2.3
(i): 8 9 10 11
X : 15.0 16.5 14.9 15.3
R : 3.8 5.0 2.9 13.8
(i): 12 13 14 15
X : 17.8 15.9 14.6 15.2
R : 14.2 4.8 5.0 2.2
Acceptance Sampling
Compute the control limits for X and R-
charts using the above data for all the
samples. Hence examine if the process is
in control. If not, remove the doubtful
samples and recompute the values of X
and R . After testing the state of control,
estimate the tolerance limits and find if the
process will meet the required
specifications.
Acceptance Sampling
Solution:
We first compute the following values for
the given data of 15 samples.
1
X  X
N
 16.1  15.2  ....  15.2
1
15
230.9

15
= 15.39
Acceptance Sampling
1
R  R
N
 3.0  2.1  ...  2.2

1
15
1
 x73.9
15
= 4.93
We now compute the control limits for the
X - chart.

Acceptance Sampling
LCL : X - A2 R = 15.39-0.729x4.93 =11.80
UCL : X +A2 R = 15.39+0.729x4.93 =18.98
The control limits of the R-chart are
CL : R = 4.93
LCL : D3 R = 0
UCL : D4 R = 11.25
Acceptance Sampling
From the values calculated above, we see
that though the process is under control
with respect to the average ( X - chart), it is
not under control with respect to the
variability (R- chart), since the R values of
sample numbers 11 and 12 (13.8 and 14.2
respectively) exceed the UCL = 11.25
Hence the process is not under control.

Acceptance Sampling
We now remove the samples numbered11
and 12 from the given data and recompute
the values for the remaining 13 samples
only.
X  16.1  15.2  ....  14.9  15.9  14.6  15.2

1
13
197.8
 = 15.22
13
R
1
3.0  2.1  ...  2.9  5.8  4.0  2.2
13
1
 x 45.9 = 3.53
13
Acceptance Sampling
We now compute the revised control limits
for both the charts.
The control limits for the X - chart are:

CL : X = 15.22
LCL : X - A2 R = 15.22- 0.729x3.53=12.65
UCL : X - A2 R =15.22+0.729x3.53=17.79
The control limits for the R-chart are:

Acceptance Sampling
CL : = 3.53
LCL : D3 R = 0
UCL : D4 R = 11.25
We now observe that the process is under

control with respect to the 13 samples
considered.
We now compute the tolerance limits

X R
Acceptance Sampling
The tolerance limits are given by :
3R
X
d2
3x3.53
 15.22   15.22  5.14
2.059
Thus the required tolerance limits are
(10.08, 20.36).
Since these tolerance limits lie within the

specification limits (9.0,21.0), the process
meets the required specifications.
Acceptance Sampling
12. The specifications for a certain quality
characteristic are (60±24) in coded values.
The table below gives the measurements
obtained in 10 samples. Find the tolerance
limits for the process and test if the
process meets the specifications.
Acceptance Sampling
Sample No. (i) : 1 2 3 4 5

Measurements (X) : 75 48 57 61 55
66 79 55 71 68
50 53 53 66 58
62 61 61 69 62
52 49 72 77 75
70 56 53 63 64
Acceptance Sampling
(i) 6 7 8 9 10
49 74 67 66 62
(X) 98 63 70 65 68
65 62 68 58 66
64 57 56 52 68
66 62 61 58 73
64 64 66 50 68
Acceptance Sampling
Solution : HW
(Hint: In control, does not meet the

specifications – under discussion)

Statistical Quality Control: Unit V

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Statistical Quality Control: Unit V

Uploaded by

Copyright:

Available Formats

Unit V

Typical process control techniques:

Historical data is used to compute the

Statistical Quality Control is the process of

Applicable to any process that produces

•Alternately, seemingly similar runs of the

•Because of these variations in the products,

If all important sources of variations are

Chance Variation & Assignable Variation

Chance(common) Variation – Variation in

A process is said to be in a state of

Once statistical control is achieved, it is

Control charts (Shewart charts) are

Control charts for attributes:

a central line corresponding to the average

lines corresponding to the upper and lower

When a sample point falls beyond the

The control limits are used to assess the

Control charts are thus quite useful both

A control chart will tell whether a process

A stable process produces predictable

Processes that are “out of control” need to

Basic Assumption: Central Limit Theorem

Process monitoring with control charts is

To verify improvements in the time

Can you relate to something else?

The right type of grapes,

LCL is given by ------------------

UCL is given by ------------------

Hence, Central line : µ

The control chart values for the x chart

As in the case of the x chart, the three

Specifically, control-chart techniques provide

Remark: Analyze R-chart before the x chart. (unit

 43  49  37  ...  47

Hence Control Limits for X chart are:

Central Line : X = 44.2

We now draw the required two control

However in the mean chart two points lie above

On the whole, the process is out of control.

X 43.9 43.6 46.0 44.2 46.9 43.8 44.1

X 47.8 49.6 45.6 47.5 47.0 47.0 49.1 46.3

 11.0  10.9  11.5  ...  11.6

 2.4  1.4  1.7  ...  1.5

From the table of control chart constants, for

Central Line : X = 11.56

3. The values of sample mean and sample

Mean 15.0 10.0 12.5 13.0 12.5 13.0

S. D. 3.1 2.4 3.6 2.3 5.2 5.4

An example of a common quality characteristic

Another quality characteristic criteria would be

Quality characteristics of these types are called

Examples of quality characteristics that are

Two fundamental control charts for

Fraction-defective chart called a “p chart”

Number-of –defects chart called a “c chart”

c-chart is a type of control chart used to

The p-chart models "pass"/"fail"-type inspection

(To what distribution will we associate

(Ultimately, Normal curve approximation)

If k samples are available, di is the number

1. If p is small, as may often be the case, then,

2. If the binomial distribution cannot be