Introduction to Quality Control (MME- 4306)

Chapter 1

Introduction
Quality:

The term “Quality” has various meanings. The historic meaning of quality is Fitness for use.

Fitness for use: In the ancient market places where commerce was transacted directly between
one man producer (tailor, shoemaker, etc.) and the ultimate user, fitness for use was even
narrower in meaning, since it referred to the degree to which a specific product or service
satisfied the wants of a specific user. As time changed and commerce being transacted through
a distribution chain, there evolved a concept of specification creating two parameters of fitness
for use.

1. Quality of design or Grade

2. Quality of conformance.

Quality of design or Grade:

This is the degree to which a class or category of product possesses satisfaction for people
generally. A difference in specification for the same functional use is a difference in quality of
design, oftenly called as “Grade”. The term “Brand” is often used to describe a producer’s
designation for a particular grade.
Eg: Car type ‘A’ and Car type ‘B’ have same functional use but they may differ in design and
specifications. Therefore, they differ in quality of design.

Quality of conformance:
This is the degree to which a specific product conforms to a design or specification. It relates to
the fidelity with which the product conforms to the design specification.
Eg: Car type ‘A’ which can run and Car type ‘A’ which cannot run have same quality of design but
they differ in quality of conformance.

Quality characteristic:
This is any distinguishing features of a Grade or a product.
Eg: Appearance, dimension, performance, length of life, dependability, reliability, durability,
maintainability, taste, odour, etc.

The Quality function:

This is the name for that area of responsibility in industrial companies through which one can
achieve fitness for use.

1
A Department:
Some industrial departments have names containing the term “Quality”.
Eg: A Quality control department.
Meaning of “Control”:
It means the planned cycle of activities by means of which we achieve an intended goal, objective
or standard.
The other meanings of “Control” are:
1. A device which directs, influences, restrains, commands, verifies or corrects.
2. The act of directing, influencing, etc.
3. A standard of comparison as determined by a statistical test usually referred to as a state of
statistical control.

Meaning of “Quality control”:

It means the entire collection of activities through which we will achieve fitness for use, that is,
carry out the company’s quality function.
It also means:
1. The partial collection of activities related to quality, which are assigned to that company
department which devotes itself primarily to the quality function.
2. Tools, devices or skills through which quality activities are carried out.
3. Name of the department which devotes itself fulltime to quality function.

Statistical Quality Control:

It is one of the tools of scientific management which emphasizes the application of statistical
methods to manufacturing problems. It provides a basis for better understanding of the variables
that exist in quality characteristics and helps directly or indirectly to improve quality or to lower
cost or both. Statistical Quality Control helps to achieve better utilization of raw material and
equipment, less scrap and rework, better inspection, improved producer consumer relation,
better specifications and better process control.

Statistical Quality Control should be viewed as a kit of tools which may influence decisions related
to the functions of specification, production or inspection. Its most effective use generally
requires cooperation among those responsible for three different functions or decisions at a
higher level than any of them. For this reason, the techniques should be understood at a
management level that encompasses all three functions. In 1980’s Dr. Edward Deming has been
most eloquent in impressing upon United States management the importance of these tools.

2
Statistics:

It is the collection, organization, analysis, interpretation and presentation of data. Statistics is

based in large part on the large numbers and mathematical theory of probability. It is in this sense
that the adjective “statistical” is accurately used in the expression Statistical quality control.

Total Quality Control:

It is an effective system for integrating the quality development, quality maintenance and quality
improvement efforts of the various groups in an organization so as to enable marketing,
engineering, production and services at the most economical level which allow for full consumer
satisfaction.
Difference between Inspection and Quality Control:

 Inspection means checking of material, product or components of product at various stages,

with reference to certain pre-determined factors and detecting and sorting out the faulty or
defective items.
 In inspection activity, the emphasis is placed on the quality of the past production. For
example, if the production schedule calls for manufacturing 1000 shafts with a diameter of
25 ± 0.05 mm the inspector will concern himself only with whether the shafts produced meet
this specification. Those that do not, will be rejected, and will continue until 1000 good units
have been produced.
 Inspection is mainly the responsibility of the inspectors. Inspection is a part of quality control.
Inspection uses precision measuring instruments such as Vernier callipers, micrometers, etc.
and devices such as profile projectors, tool makers microscope, etc.
 Quality control should not be confused with inspection. Quality control is a broad term. It
involves inspection at particular stage but mere inspection does not mean quality control.
 As opposed to the inspection, in quality control activity the emphasis is placed on the
quality of future production. There are various of doing this. For example,
(i) as the shafts are being produced, periodic samples might be taken of the output and the
shafts in each sample inspected. If the quality of the items in particular sample is
satisfactory, production will be allowed to continue. But, if it is not, corrective action will
be immediately taken. This action might involve adjusting the machine, eliminating
defects in raw material, inspecting or replacing the operator, etc.
(ii) care may be taken to provide operating personnel with correct instructions prior to the
production of an item.
 In quality control, what is learned from inspection of a sample of the product is used as a
basis to ascertain whether the quality of the product produced is according to the set norms
of it is necessary to make change in the production process.

3
 Quality control includes a number of activities (including inspection) and involves use of
statistics, control charts, process capability study, acceptance sampling, etc. in order to build
up and regulate the quality of product.
 Quality control stresses that everybody working in an organization is responsible for quality
of products produced.
Some simple Statistical tools:
Many of the techniques developed by mathematical statisticians for the analysis of data may be
used in the control of product quality. The expression Statistical quality control may be used to
cover all uses of statistical techniques for this purpose. However, it often relates to the following
separate but related techniques that constitute the most common working statistical tools in
quality control. These tools are:
1. The Shewhart control chart for measurable quality characteristics. In the technical language
of the subject, these are described as charts for variables or as charts for 𝑋̅ and R (Average
and Range) charts for 𝑋̅ and s (Average and Sample standard deviation).
2. The Shewhart control charts for attributes. These include:
(i) The Shewhat control chart for fraction rejected. In technical language of the subject,
this is described as p chart.
(ii) The Shewhats control chart for number of rejects or defectives. In technical language
of the subject, this is described as np chart.
(iii) The Shewhart control chart for number of non-conformities. In technical language of
the subject, this is described as c chart.
(iv) The Shewhart control chart for number of non-conformities per unit. In technical
language of the subject, this is described as u chart.
3. That portion of sampling theory which deals with the quality protection given by any
specified sampling acceptance procedure (Acceptance sampling plans).
Products and Services:

Human beings acquire products mainly to secure the functional uses which they provide through
their inherent qualities.

Eg: Automobiles, food products, home appliances, etc.

Some products are fully consumed during first usage or during a short period of use.
Eg: Food, soap, etc.

Other products may last for months or years before becoming unfit for use. For such longer life
products, the user need not necessarily buy the product to utilize their quality. He may buy only
the services while ownership of the product rests with someone.

Eg: Education, transport service, telephone service, office space, computer time, etc.

4
The spiral of progress in Quality:

Organization for Quality:

Reliability:

Simply stated, Reliability is quality over the long run. Quality is the condition of the product during
manufacturing or immediately afterward, whereas Reliability is the ability of the product to
perform the intended function over a period of time. A product that “works” for a long period of
time is a reliable one. Since all units of a product will fail at different times, Reliability is a
probability.

5
General Quality control engineering fundamentals:

Patterns of variation:

Variation seems inevitable in nature. Manufacturing processes are no exception to this. No two
pieces are ever made exactly alike, whether one is attempting to control a dimension of a part
that is to go into a precision assembly, the resistance of a resistor, the acidity of a solution, the
weight of the contents in a container or any other quality of the manufactured product, it is
certain that quality will vary. Therefore, it is necessary to have some simple methods of describing
patterns of variations. Statisticians have developed such methods. One method is “Frequency
distribution”. Another method involves the finding of a measure of “Central tendency” of a
distribution (i.e. average) combined with some measure of “Dispersion” or Spread or
Scatteredness of the distribution.

Causes of quality variation:

The causes of quality variation are classified into two types:

1. Chance causes
2. Assignable causes
Chance causes:

These are the causes of quality variation due to the operation of forces working at random. These
forces arise from the constant system of chance causes. They represent variability inherent in the
process. These are unavoidable and are termed as “non-significant”. Chance causes merit no
investigation. A process under the influence of chance causes is considered to be stable and
under control (statistical control).

Assignable causes of variation:

These are the causes of quality variation due the operation of forces working consistently in some
particular direction. This leads to non- random variation of the product. The presence of
assignable causes in production process is termed as “significant”. If assignable causes of
variation are allowed then it results in non-conforming or defective product. The process with
assignable causes of variation is termed as unstable or out of control.

Variables and Attributes:

Variables are those quality characteristics which are measurable. When a record is made of an
actual measured quality characteristic, the quality is said to be expressed by variables.

Eg: Diameters in inches, operating temperature in degree oC or Farenheight, Tensile strength in

kg/cm2, weight in Kg, etc.
6
A variable that is capable of any degree of subdivision is referred to as “continuous variable”.

Attributes are those quality characteristic which are classified as either conforming or not
conforming to specifications. When a record shows only the number of articles conforming and
the number of articles failing to conform to any specified requirements, it is said to be record by
attributes. Variables which are discontinuous are called as attributes or discrete variables.

Ex: The number of non-conforming rivets- 5, 10, 9, 6, etc. but not 5.65 non-conforming rivets.

In general, variables are continuous data and are measurable and attribute data are countable.

Frequency distribution:

A frequency distribution is a tabulation of data arranged according to size. When there are a large
number of highly variable data, the frequency distribution can become too large to serve as a
summary of the original data. The data may be grouped into cells to produce a better summary.

The table below shows the grouped frequency distribution for a set of observations:

Cell boundaries Cell midpoints Observed Cumulative

( Weight in Kg) frequency frequency
0.5-1.5 1 1 1
1.5-2.5 2 3 4
2.5-3.5 3 6 10
3.5-4.5 4 10 20
4.5-5.5 5 20 40
5.5-6.5 6 25 65
6.5-7.5 7 15 80
7.5-8.5 8 9 89
8.5-9.5 9 8 97
9.5-10.5 10 2 99
10.5-11.5 11 1 100

7
Definitions related to Frequency distribution:

Grouped frequency distribution: A grouped frequency distribution of a set of observations is an

arrangement which shows the frequency of occurrence of the values of the variable in ordered
classes.

Cell: The interval along the scale of measurement of each ordered class is termed a cell.

Frequency: The frequency for any cell is the number of observations in that cell.

Relative frequency: The relative frequency for any cell is the frequency for that cell divided by
the total number of observations.

The table above shows a set of data or observations for a particular quality characteristic (weight
in Kg) which has been grouped in the form of a frequency distribution. The observations are
grouped in 11 cells. Each cell (e.g. 0.5 -1.5 Kg) has an equal interval or cell width (e.g. 1 Kg) and a
cell midpoint. The frequency of observation in each cell is also shown. Each cell has a lower cell
boundary and an upper cell boundary.

A useful guide in deciding on the number of cells for a set of observations is as follows:

1. Use 5-10 cells when the number of observations is less than 100.
2. Use 8-17 cells when the number of observations is between 100-500.
3. Use 15-20 cells when he number of observations is greater than 500.
Graphic representation of a frequency distribution:

The graphic representation methods of a frequency distribution are as follows:

1. Frequency Histogram
2. Frequency Bar chart
3. Frequency Polygon
1. Frequency Histogram:

A frequency histogram consists of a set of rectangles with bases that represents cell mid points
and with heights (areas) representing frequency within a cell. In this graph, sides (extreme ends)
of the columns represent the lower and upper cell boundaries.

8
2. Frequency Bar chart:
This graph uses bars centred on the midpoint of the cells. The heights of the bars are proportional
to frequencies within the cells.

3. Frequency Polygon
It represents an effective picture of the frequency distribution. It is constructed by placing series
of straight lines joining small circles which are plotted at cell mid points with a height proportional
to cell frequencies.

9
Cumulative frequency distributions:

Cell boundary Frequency Number of cans having Number of cans having

(Weight in Kg) less than the given greater than the given
weight weight
(Cumulative frequency) (Cumulative frequency)
0.5 0 100
1
1.5 1 99
3
2.5 4 96
6
3.5 10 90
10
4.5 20 80
20
5.5 40 60
25
6.5 65 35
15
7.5 80 20
9
8.5 89 11
8
9.5 97 3
2
10.5 99 1
1
11.5 100 0

Cumulative frequency distribution –Less than ogive curve:

10
Cumulative frequency distribution –Greater than ogive curve:

Measures of Central tendency and Dispersion:

Most commonly used measures of central tendency are:

1. Arithmetic mean (sample mean or average):

The arithmetic mean of a set of n numbers is the sum of the numbers divided by n. Expressed
mathematically as:
𝑛
1
𝑋̅ = ∑ 𝑋𝑗
𝑛
𝑗=1
𝑋̅ = Aarithmetic mean
Xj = X1, X2, … Xn = Specific numbers (observations) in the set
n = Total number of observations in the set (Subgroup or Sample size)
Σ = Mathematical symbol for taking sum
2. Median:
Median is that value of the variate which divides the total frequency exactly into half i.e. half the
values of the variate are greater and half the values are less than the mean value.

If there is an odd number of observations (2m+1) of the variate and they are arranged in
ascending order then the Median is (m+1)th value.
If the number of observations of the variate is even (2m) then the Median is taken to be the
average of the middle values when arranged in ascending order.
𝑚th value + (𝑚 + 1)th value
Median =
2

11
3. Mode:
Mode is that value of the variate which occurs with the greatest frequency or it is the midpoint of the cell
with highest frequency or it is the value corresponding to the highest point in a frequency histogram.
Measures of Dispersion:

Measures of dispersion or scatter or spread are:

1. Range (R):
It is the difference between the largest and the smallest set of numbers. It is denoted by R.

𝑅 = 𝑋𝑚𝑎𝑥 − 𝑋𝑚𝑖𝑛
Where,
Xmax = Largest number in the set
Xmin = Smallest number in the set

2. Sample standard deviation (s):

The Sample standard deviation of a set of numbers from their arithmetic mean is given by

∑ 𝑋𝑗2 − 𝑛 𝑋̅ 2
𝑠= √
𝑛−1
Where,
𝑋̅ = Aarithmetic mean (Sample mean)
n = Subgroup or Sample size
The above formula is used to calculate the standard deviation of a sample taken from the
universe or population or parent distribution.
The Sample variance = s2

3. Standard deviation (𝝈):

It is defined as the square root of the average of the square deviations from the mean. It is a
measure of Dispersion (scatter or spread) in the universe or population. It is given by:
∑𝑀
𝑗=1(𝑋𝑗 − 𝜇)
2
𝜎= √
𝑀

Where,
𝜎 = Standard deviation
1 𝑀
𝜇 = Population mean or Universe average = ∑ 𝑋𝐽
𝑀 𝐽=1
M = Population size

12
4. Root Mean Square deviation (𝝈𝑹𝑴𝑺 ):
The Root Mean Square (RMS) deviation of a set of numbers from their arithmetic mean is given
by:
2
√∑(𝑋𝑗 − 𝑋̅)
𝜎𝑅𝑀𝑆 =
𝑛
Where,
𝑋̅ = Aarithmetic mean (Sample mean)
n = Subgroup or Sample size
Calculation of Sample average (Sample mean) and Sample standard deviation of grouped frequency
distribution – Long Method:

Consider the following frequency distribution:

Cell mid point 1 2 3 4 5 6 7 8 9
(Weight in Kg)
X
Cell frequency 6 8 12 13 20 19 13 11 8
f

Cell mid point Cell frequency Total weight Squared Sum of

(Weight in Kg) f fX weight squares for cell
X X2 f X2
1 6 6 1 6

2 8 16 4 32

3 12 36 9 108

4 13 52 16 208

5 20 100 25 500

6 19 114 36 684

7 13 91 49 637

8 11 88 64 704

9 8 72 81 648

Totals n = ∑ 𝑓 = 110 ∑ 𝑓𝑋 = 575 ∑ 𝑓𝑋 2 = 3527

13
 ∑ 𝑓𝑋 ∑ 𝑓𝑋 575
Sample average = Sample mean = 𝑋̅ = = ∑𝑓
= = 5.227 Kg
𝑛 110
∑ 𝑓𝑋 2 −𝑛 𝑋̅ 2 3527−110(5.227)2
Sample standard deviation = 𝑠 = √ =√ = 2.1876 Kg
𝑛−1 110−1

Calculation of Sample average (Sample mean) and Sample standard deviation of grouped frequency
distribution – Short Method:

Cell midpoint Cell Deviation in Squared Sum of squares

(Weight in Kg) frequency cells from weight for cell
X f Assumed origin fd fd2
d
1 6 -4 -24 94

2 8 -3 -24 72

3 12 -2 -24 48

4 13 -1 -13 13

5 20 0 0 0

6 19 +1 +19 19

7 13 +2 +26 52

8 11 +3 +33 99

9 8 +4 +32 128

Totals n=∑𝑓 = ∑ 𝑓𝑑 = 25 ∑ 𝑓𝑑 2 = 527

110

Consider the middle value of the quality characteristic (Weight) say, X = 5 Kg as the Assumed
origin.
Therefore,
Assumed origin = 5 Kg
Cell width = 1 Kg

Sample mean = 𝑋̅ (in original units from true origin) = Assumed origin + 𝑋̅ (in cells from assumed
origin )* (cell width)

∑ 𝑓𝑑
𝑋̅ (in original units from true origin) = Assumed origin + ( 𝑛 ) * (cell width)

14
 25
= 5 + (110) ∗ (1) = 5.227 Kg

Sample standard deviation = s (in original units) = s (in cell units) * (cell width)

∑ 𝑓𝑑 2 − (∑ 𝑓𝑑)2 /𝑛
= (√ 𝑛−1
) ∗ (Cell width)

(25)2
527−
= (√ 110−1
110
) ∗ (1) = 2.186 × 1 = 2.186 Kg

Sampling statistics and Universe parameters:

In manufacturing, sampling techniques are used to have knowledge about pattern of variation of
the production process under study. The sample is drawn from the production process. The
unknown pattern of variation from which the known sample has been drawn is called as universe,
parent distribution, population, sample space or description space. The true but unknown
numerical values that describe the universe are called as Parameters.

In order to draw conclusions about an unknown universe it is necessary to depend on numerical

values derived from samples drawn from that universe. These numerical values are – Sample
mean, Median, Mode, Sample standard deviation, RMS deviation, Range and Sample variance.
These values summarize the information contained in the sample data. Each of these is referred
as Statistic of the sample and they may be used to estimate the corresponding “parameter” of
the unknown universe. Parametric values of arithmetic mean and standard deviation for a given
universe is denoted by µ (mu) and σ (sigma) respectively. These values are also denoted as 𝑋̅ ′ (X
bar prime) and 𝜎 ′ (sigma prime) respectively.

Oftenly, parameters are estimated from one or more samples drawn from an unknown universe.
Thus, 𝑋̅ (Sample mean or average) of a sample of size n may be used to estimate the parametric
mean µ and s (Sample standard deviation) can be used to estimate parametric standard deviation
σ. To differentiate the estimator from parametric value a “caret” or “hat”() symbol is used. i.e.
𝜇̂ = 𝑋̅ and 𝜎̂ = 𝑠. The parametric symbols may be applied to a known distribution or when
analysing data from all items in a production lot.

The Normal Curve:

For any frequency distribution if we know its measure of central tendency i.e. value of arithmetic
mean µ and measure of dispersion i.e. value of standard deviation σ, then it would be possible
to say just what proportion of the measurements fall within any specified limits (say, µ ± 2σ). For
this to be possible, the equation of a frequency curve that describes the pattern of variation of
the particular quality characteristic should be fully defined.

15
The pattern of variation of a particular measured quality characteristic is called as Frequency
curve or Reference distribution. Many such frequency curves pertaining to distributions of
measured qualities of manufactured product and many other frequency distributions found in
nature roughly corresponds to the “Normal curve”. µ and σ are sufficient to describe this Normal
curve. This curve is also known as the Normal curve of error, the probability curve, the Gaussian
curve, the Laplacian curve or Normal distribution curve. It has bell shaped symmetrical form as
shown in the figure.

This curve extends from -∞ to +∞. The proportion (probability) under the curve is unity (1.0). For
a Normal curve the mean, median and mode are equal. Most of the area under this curve is
included within µ ± 3σ limits. The limits in connection with the Normal curve are as follows:

Limits % of total area within limits

µ ± 0.6745σ 50.00
µ ± 1σ 68.26
µ ± 2σ 95.45
µ ± 3σ 99.73

This means that in those distributions which roughly approximates the Normal curve, about 2/3
(68.26%) of occurrences fall within one standard deviation on either side of the average, 95.45%
of occurrences fall within 2 standard deviations and practically all fall within 3 standard
deviations.
Mathematically the Normal curve can be defined as:

𝑧 𝑧2
1
𝐹(𝑧) = ∫ 𝑒2 𝑑𝑧
√2𝜋 −∞
Where,

𝑋−𝜇
𝑧 = Normalized statistic =
𝜎

16
 Problems:
1. The mean value of the modulus of rupture of a large number of test specimens has been
found to be 5600 lb/in2 (PSI).
(i) If the standard deviation is 840 lb/in2 and the distribution is approximately normal, for
what percentage of specimens the modulus of rupture will fall between 5000-6200
Ib/in2?
(ii) For what percentage will it be above 4000 Ib/in2?
(iii) For what percentage will it be below 3500 Ib/in2?
Solution:

(i) Given,
𝜇 = 5600 lb/in2
𝜎 = 840 lb/in2
Let 𝑋1 = 5000 lb/in2
𝑋1 −𝜇 5000−5600
The Normalized statistic = 𝑧1 = = = −0.71
𝜎 840
Let 𝑋2 = 6200 lb/in2
𝑋2 −𝜇 6200−5600
The Normalized statistic = 𝑧2 = = = +0.71
𝜎 840
From the Normal curve table, area under the Normal curve for 𝑧1 = −0.71 is 0.2389 and the
area under the Normal curve for 𝑧2 = +0.71 is 0.7611
Required area under the Normal curve = (0.7611 - 0.2389) = 0.5222.
52.22% of specimens have their modulus of rupture values falling between 5000 and 6200
lb/in2.

(ii) Let X = 4000 lb/in2

𝑋−𝜇 4000−5600
𝑧= = = -1.90
𝜎 840

And the area under the Normal curve for 𝑧 = -1.90 is 0.0287.
Required area under the Normal curve = 1- 0.0287 = 0.9713
97.13% of specimens will have their modulus of rupture values falling above 4000 lb/in2.

17
 (iii) Let X = 3500 lb/in2
𝑋−𝜇 3500−5600
𝑧= = = -2.50
𝜎 840

And the area under the Normal curve for 𝑧 = - 2.50 is 0.0062.
0.62% of the specimens will have their modulus of rupture values falling below 3500 lb/in2.

2. A certain process with a standard deviation of 0.001 has been statistically controlled at a
mean of 0.036. The product is currently sold to two users A and B. User A’s specifications are
0.038 ± 0.004 and those of user B are 0.036 ± 0.004.
(i) Assuming a Normal distribution of the product what % of product does not meet user
A’s requirements? and user B’s requirements?
(ii) A suggestion is made that the process be adjusted to 𝑋̅ ′ of 0.037. Determine the effect
of this shift in terms of percent nonconforming for each user. Do you consider it
desirable?
Solution:

(i) Given,
μ = 0.036
σ = 0.001
User A’s specifications:
𝑋1 = U = 0.038 + 0.004 = 0.042
𝑋2 = L = 0.038 – 0.004 = 0.034
𝑋1 −𝜇 0.042 −0.036
The Normalized statistic = 𝑧1 = = = +6
𝜎 0.001
𝑋2 −𝜇 0.034−0.036
The Normalized statistic = 𝑧2 = = = −2
𝜎 0.001
The area under the Normal curve for 𝑧1 = +6 is unity.
The area under the Normal curve for 𝑧2 = −2 is 0.0228.
Hence, 2.28% of the product does not meet user A’s requirements (specifications).
User B’s specifications:
𝑋1 = U = 0.036 + 0.004 = 0.040
𝑋2 = L = 0.036 – 0.004 = 0.032
𝑋1 −𝜇 0.040 −0.036
The Normalized statistic = 𝑧1 = = = +4
𝜎 0.001
𝑋2 −𝜇 0.032−0.036
The Normalized statistic = 𝑧2 = = = −4
𝜎 0.001
The area under the Normal curve for 𝑧1 = +4 is unity.
The area under the Normal curve for 𝑧2 = −4 is negligible.

18
Hence, a negligible percentage of the product does not meet user B’s requirements
(specifications).

(ii) When 𝑋̅ ′ = 𝜇 = 0.037.

User A’s specifications:
𝑋1 = U = 0.038 + 0.004 = 0.042
𝑋2 = L = 0.038 – 0.004 = 0.034
𝑋1 −𝜇 0.042 −0.037
The Normalized statistic = 𝑧1 = = = +5
𝜎 0.001
𝑋2 −𝜇 0.034−0.037
The Normalized statistic = 𝑧2 = = = −3
𝜎 0.001
The area under the Normal curve for 𝑧1 = +5 is unity.
The area under the Normal curve for 𝑧2 = −3 is 0.00135.
Hence, 0.135% of the product does not meet user A’s requirements (specifications).

User B’s specifications:

𝑋1 = U = 0.036 + 0.004 = 0.040
𝑋2 = L = 0.036 – 0.004 = 0.032

𝑋1 −𝜇 0.040 −0.037
The Normalized statistic = 𝑧1 = = = +3
𝜎 0.001
𝑋2 −𝜇 0.032−0.037
The Normalized statistic = 𝑧2 = = = −5
𝜎 0.001
The area under the Normal curve for 𝑧1 = +3 is 0.99865.
The area under the Normal curve for 𝑧2 = −5 is negligible.
Required area = 1-0.99865 = 0.00135.
Hence, 0.135% of the product does not meet user B’s requirements (specifications).
Since maximum proportion of the product conform to specifications of both users and the
percentage of nonconforming product is very small for both users, i.e. 0.135%, the change in
process mean to 0.037 is desirable.

3. In the precision grinding of a complicated part it is economical to rework the part than to
scrap it. Therefore, it is decided to establish the rework percentage at 12.5%. Assuming
Normal distribution of the data, a standard deviation of 0.01mm and Upper specification
limit of 25.38mm, determine the process center.

Solution:

The rework at 12.5% denotes the area (proportion) under the Normal curve above the Upper
specification limit (USL or U) of 25.38mm is 0.125. Therefore, the area under the Normal curve
till the USL is 1-0.125 = 0.875.
The z value corresponding to the area of 0.875 is obtained using the Normal curve table on
interpolation between z = +1.15 and z = +1.16 and corresponding probabilities 0.8749 and 0.8770
as shown below:

19
 (1.16−1.15)(0.875−0.8749)
1.15 + = +1.1504
(0.8770−0.8749)

Let X = USL = 25.38 mm

𝑋−𝜇
The Normalized statistic = 𝑧 = 𝜎

25.38 −𝜇
+1.1504 =
0.01

𝜇 = 25.38 – (1.1504 × 0.01) = 25.36mm

Therefore, to have the rework percentage at 12.5% the process should be centred at
𝜇 = 25.36 mm.

20