International Journal of Mathematics and Statistics Invention (IJMSI)

E-ISSN: 2321 4767 P-ISSN: 2321 - 4759

www.ijmsi.org Volume 1 Issue 2 December. 2013 PP-31-40
www.ijmsi.org 31 | P a g e

Some Studies On Normal and Non-Normal Process Capability
Indices

Ajit Goswami
1
& Harendra Narayan Dutta
2
1
Ph.D Scholar, Department of Statistics, Dibrugarh University, Dibrugarh, Assam, India
2
Professor, Department of Statistics, Dibrugarh University, Dibrugarh, Assam, India

ABSTRACT:Nowadays there is no organisation that can operate on the market without constant improvement
of its performance focused on quality of products and services. The challenge in todays competitive market is to
be on the leading edge of producing high quality products at minimum costs. This can be done with a systematic
approach by using Statistical Process Control (SPC) tools which can be used to improve the quality in mass
production process including process and product design prior to manufacturing in quantifying process
variability. It is obvious that customers satisfaction plays a key role for success of any business. So, in
analysing this variability relative to product requirements or specifications, one has to need to eliminate or
greatly reduce this variability by the method of Process Capability Analysis (PCA). The Process Capability
Indices (PCIs) provides a measure of how a process fits within the specification limits.Since 1980s, theories of
PCIs have been developed to analyse the capability of processes. During the past few decades, various PCIs
have been developed both for normal as well as for non-normal process data. In this paper, a brief review of
past works on process capability indices dealing with both normal & non-normal process data will be studied
and some of the PCIs would be discussed for both the cases.
Key Words: PCA, SPC, PCIs, Normal and Non-normal

I. INTRODUCTION:
It is noteworthy that, in todays competitive world market, the manufacturing industry is facing
intensive competition. Therefore, both the cost and quality aspect of the product have become very important
issues among producers. Thus producers thrive on providing economical processes which are also capable of
meeting the customers quality requirements. To achieve the goal for high quality with low cost, Process
Capability Indices (PCI) are developed both for on-line and off-line quality management system. The Process
Capability Analysis (PCA) is a technique that has application in many segments of the product cycle including
product and process design, vendors sourcing, production or manufacturing planning and manufacturer parts
within the tolerance limits and engineering values. Quality engineering experts found that Process Capability
indices are powerful tools of monitoring the process ability for manufacturing a product to meet the particular
specifications. It is note-worthy that certain assumptions are essential to use Process Capability Indices to have
a meaningful interpretation of the Process data.Process capability measures the variability of a process relative
to its specification limit based on three assumptions, namely, (i) the process is itself in control, (ii) target value
and specifications of a quality characteristics are specified, and (iii) the process measures quality characteristics
that follow a normal distribution.

Definition of Process Capability: The capability of a process is the ratio of the distance from the process mean
to the nearest specification limit divided by measure of process variability.
Mathematically, it can be expressed as:
Process capability = |
.
|

\
|
o

o

3
,
3
LSL USL
Min (2.1)
Where and o are the process mean and standard deviation of the measured characteristic of interest. USL
and LSL are the upper and lower specification limits respectively. Usually these values are estimated from the
data collected from the production process. There are several definitions of process capability as put forwarded
by the quality scientist in recent times and more may be offing in future. Here we mention a few definitions for
interested readers in this area.Sinha and Willborn [1] defines that process capability provides a quantified
measures of adequacy and is learned indirectly by quality control done on actual product; while Juran and
Gryana [2] refers that process capability is the minimum variation that a process can achieve. Wasserman et.al
[3] states that -Process capability is the ability of the production process of producing items within specified
tolerances. Pyzdek [4] concluded that Process Capability Analysis is conducted to compare a controlled
process to measurement. It is an analytical study which attempts to estimate future events. Lester, Enrich and
Some Studies On Normal And Non...
www.ijmsi.org 32 | P a g e

Motley [5] defines process capability as the quality that a process can produce with operations a normal
level under practical conditions with companys economic means. Montgomery [6] more specifically defines
Process Capability Analysis as an engineering study to estimate process capability. The estimate of process
capability may be in the form of a probability distribution having a specified shape, center (mean) and spread
(standard deviation). In most recent times, Sagbas [7] defines process capability analysis as the technique
applied in many stages of the product cycle- including process, product design, manufacturing and
manufacturing planning, since it help to determine the ability to manufacture parts within the tolerance limits
and engineering values.

II. REVIEW OF PAST WORKS ON PROCESS CAPABILITY INDICES:
Gogoi [8] made an extensive review of the literature on process capability measures developed up to
2003. He had of course mainly studied the past works on PCA dealing with normally distributed product. In
recent times there are several papers on PCA for normally as well as no-normally distributed product. In this
section, we shall only highlight briefly the recent works on PCA for normally distributed product and in next
section we will deal with non-normally distributed product.
3(A) PCI for normally distributed data:
Vannman, and Albing [9], presented a graphical method useful for analysing process capability for a
process having one-sided specification limits. Under the assumption of normality, estimated process capability
plots are used to assess process capability at a given significance level. They suggested that graphical approach
is helpful to determine the variability, the deviation from target, or both which is required to be reduced to
improve the capability. Albing [10] gave emphasis on process capability indices when specification limits are
one-sided. Albing and Vannman [11] discussed the process capability in a situation with non-negative process
data having a skew distribution with a long tail towards large values. Urdhwereshe [12] discussed about
measurement system analysis by comparing ratio of two values, namely Process tolerance and Process variation.
He has also analysed capability indices with a case example.Guevera, et.al [13] discussed process capability
indices for dependent data i.e., when the data are auto-correlated. Though a simulation study they showed that
higher the auto-correlation level, the lower is the capability index value. Jeang, et.al [14] studied process
capability indices for off-line application of production life cycle. The product quality is represented by quality
loss function and production cost is expressed by tolerance cost function. They observed that an economical and
quality of product designed and process planning can be achieved during off-line application of process
capability indices.

Czarski, [15] emphasized for correct determination of process capability indices, identification of
distribution shapes and quality improvement method like Statistical Process Control (SPC) tools to be
understood by the users of capability analysis before using the process capability tools.Ali, et.al [16]
summarised the characteristics of the process capability indices for both real and simulated data. In their paper
they discussed normal and exponential distribution with different values of sample sizes to ascertain sample size
effect on process capability indices.Sagbas, A [7] carried out a process capability analysis in the machining line
of a medium size company that produces machine and spare parts. He suggested that in order to satisfy the
process capability measures, it is necessary to improve the quality level by shifting the process mean to the
target value and reducing the variations in the process.Flaig J.J. [17] proposed a new metric which differs from
the commonly used engineering metrics. The metric consists of two economic capability measures the expected
profit and the variation in profit of the process. This dual economic metric offers a number of significant
advantages over other engineering or economic metric used in Process Capability Analysis. Lundkvist et.al.
[18] discussed about the comparison of decision methods using the process capability index C
pk
when data are
autocorrelated. He has done it through a case study followed by a simulation study and found that two methods
appeared to be better than the others.
3(B) PCI for non- normally distributed data:
Pan & Wu [19] investigated the process capability indices for non-normal data which is divided into
three categories i.e. bilateral specifications; unilateral specification with target value and unilateral specification
without target value. Pearn & Chen [20] introduced C
p
(u,v) which includes the four basic PCIs (C
p
, C
pk
, C
pm

and C
pmk
) and they have generalised C
p
(u,v) as C
Np
(u,v) and C
/
Np
(u,v) for non-normally distributed data and
made a comparison between C
Np
(u,v) and C
/
Np
(u,v) and found that C
Np
(u,v) are superior to C
/
Np
(u,v) in
measuring process capability. Bittanti [21] addresses the problem of defining and computing reliable estimates
for process capability indices (and particularly for Cpk) for nonnormal processes; in particular, a curve-fitting
approach to the estimation problem is taken and the problem of providing confidence intervals for the estimates
of PCIs is considered. Finally, some application examples are also presented.
Some Studies On Normal And Non...
www.ijmsi.org 33 | P a g e
Abbasi, et.al [22] measures Process Capability Indices (PCIs) for bivariate non-normal process using
the bivariate Burr distribution. They have used univariate Burr distribution to improve the accuracy of estimates
of PCIs for univariate non-normal distribution. They also estimated the PCIs of bivariate non-normal
distributions using the bivariate Burr distribution.Casalino & Rotondo [23] proposed the multivariate process
incapability index MCpp (q) for non-normal data. A case study on the comparison between two different cold
extrusion processes has been presented to demonstrate how the proposed index can be applied to real
data.Ahmad et.al [24] reviews the performances of the Clements [25] non-normal percentile method, the Burr
based percentile method and Box-Cox [26] method for non-normal cases. They also conducted a simulation
study using Weibull, Gamma and Lognormal distribution.Albing [27] proposed a class of capability indices that
are useful when the quality characteristic of interest has a skewed, zero-bound distribution with a long tail
towards large values and there is an upper specification with a pre-specified target value, T=0. He has
investigated the process capability indices for Weibull distribution. Czarski [28] discussed the comparative
analysis of capability indices using Clements method and exact method.Lovelace & Swain [29] proposed a
process capability index estimation methodology for Cp and Cpk for the case of non-normal, zero-bound
process data using the delta distribution, a variant of the lognormal distribution. This approach utilizes quantile
estimates derived from a proposed modification of lognormal quality control charts.

Ahmed [30] discussed the estimation of Cp and Cpk , commonly used process capability indices (PCI),
in case of nonnormal data using the characteristics of Weibull distribution. Quantiles are estimated by
probability plotting technique and then control limits are obtained to determine whether the process is in
statistical control or not. Percentage points of the fitted distribution have been used to compute under the
assumption of Weibull distribution. Kantam et.al. [31] studied the point estimates of process capability index as
suggested by Clements through simulation when the underlying distribution is Half Logistic distribution. They
also used Moment estimate, maximum likelihood estimate (MLE) and best linear unbiased estimate (BLUE) .

III. VARIOUS APPROACHES FOR PROCESS CAPABILITY ANALYSIS:
We have already stated that Process Capability Analysis is a technique that has application in many
segments of the product cycle, including product and process design, vendor sourcing, production on
manufacturing planning and manufacturing. There are three primary techniques that can be used to study the
Process Capability Analysis. These are:
(a) Histogram or probability plots
(b) Control Charts
(c) Designed of Experiments.
Jeang et.al [14], observed that in todays competitive World Market, the manufacturing industry is facing
intensive competition. Therefore, both the cost and quality aspect have become very important issues among the
producers. Thus producers thrive on providing economical processes which are also capable of meeting the
customers quality requirements. To achieve the goal for high quality with low cost, Process Capability Indices
(PCI) are developed for on-line as well as off-line quality management system. The PCI value is typically
defined as the ability to carryout the intended goal. The frequently use Process Capability Ratios (PCR) are C
p
,
C
pk
and C
pm
which can be formulated as follows-

(A) On-line Process Capability Indices:
(i) The Process Capability Ratio C
p

This can be expressed as:

o 6
LSL USL
C
p

= (4.1)
where USL and LSL are Upper and Lower specification limits respectively.
In practical application, the Process standard deviation is almost always unknown and must be replaced by an
estimate of . To estimate we typically use either the sample standard deviation s or
2
d
R
(when variables
control charts are used in the capability study).
For one-sided specification we define one-sided PCRs as follows-

o

3

=
USL
C
pu
(upper specification only) ... (4.2)

o

3
LSL
C
pl

= (lower specification only) ... (4.3)
Some Studies On Normal And Non...
www.ijmsi.org 34 | P a g e
It may be noted that during the production process, the process mean can be changed other than its
target value. If the process variance
2
is constant then C
p
value would remain unchanged. This was the major
defect of C
p
owing to the fact that the deviation of the process mean cannot be reflected in the measurement but
only the spread of the process. These are the main reason to develop the C
pk
.

(ii) The Process Capability Ratio Cpk:
This index can be expressed as:
)
3
,
3
(
o

o
LSL USL
Min C
pk

= (4.4)
) (
, pl pu
C C Min ~ where C
pu
and C
pl
are given in equation (4.2) and (4.3) respectively.
The magnitude of C
pk
relative to C
p
is a direct measure of how off-center the process
is operating. More generally we usually say that C
p
measures potential capability of the
process, whereas C
pk
measures actual capability.

(iii) The Process Capability Ratio C
pm
:
There is still a deficiency which is not measured by the above two P.C.I s namely C
p
or C
pk

since in a process, the mean and variance are generally changed.
To measure such simultaneously changes, C
pm
has been developed. The formula of
C
pm
is given as

t 6
LSL USL
C
pm

= (4.5)
Where is the square root of expected squared deviation from target

2 2
2 2
2 2
) (
) ( ] ) [(
] ) [(
) (
2
1
T
T x E
T X E
LSL USL T
+ =
+ =
=
+ =
o

t
(4.6)
which can finally be expressed as
C
pm
2
1 +
=
p
C
(4.7)

Where

o

T
=
Chan et.al. [32] and Boyels [33] discussed detail about C
pm
and its usefulness in measuring process centering.
The interpretation of C
p
and C
pk
and C
p
relative to C
pm
can be summarized into following ways:
1. If C
p
=C
pk
, then it implies that the process centered at the midpoint of the specification.
2. If C
pk
< C
p
, then it implies that the process is off-center. The user should try to center the process.
3. If C
pk
=0, then the process is exactly equal to the one of the specification limit.
4. If C
p
< 0, then the process mean is outside the specification limits.
5. If C
p
= 1, it implies that the process is centered. For a normally distributed product C
p
= 1 implies that 2700
parts per million (ppm) are non-conforming i.e., fall out rate of 2700 ppm for two sided specifications.
6. If C
p
< 1, it implies that the process is not fully capable.
7. If C
p
> 1, capable of meeting the specification
8. C
pk
< -1 implies that entire process lies outside the specification limit. Usually C
pk
is non-negative.
Similarly,
9. If C
pm
1, implies that process mean lies within the middle third of the specification band. If C
p
= C
pk
=
C
pm
then = Target value, but all the PCI decreases if move away from the target value.



Some Studies On Normal And Non...
www.ijmsi.org 35 | P a g e
IV. PCI FOR NON-NORMAL DATA:
Non-normally distributed processes are not uncommon in practice. Combining this fact with the
misleading results of applying basic PCIs to non-normal processes while treating them as normal distribution
forced academicians and practitioners to investigate the characteristics of process capability indices with non-
normal data.
The situation, where the distribution of the monitored feature of quality does not correspond with normal
distribution can in principle be addressed in the following ways:
a) Transformation of data to a variable coming from normal distribution
b) Use of different theoretical model of distribution

Transformation of data to normally distributed variable
In the case of data transformation, the original data, which do not correspond to normal distribution,
use suitable transformation function to be transformed into data, whose distribution is most closely
approximating normal distribution. The next step will verify the normality of the transformed data using suitable
numerical tests. If the normality of data is confirmed it is followed by the transformation of tolerance limits
using the identified transformation function, and the process capability analysis can be carried out in accordance
with the procedure for normal distribution.

Box- Cox data transformation:
George E.P. Box and David Cox have created a transformation that allows approximation of selection
distribution to normal distribution with respect to skewness and kurtosis. It is a very commonly used
transformation, mainly due to its simplicity. Box-Cox transformation follows the relations:

1
ln( )
x
y
X

for 0 and =0 ... (5.1)

Johnson data transformation:
Johnson transformation chooses one of three types of equations, depending on whether the random
variable is bounded (SB), log-normal (SL) or non-bounded (SU). The equations are:
Type SB-bounded random variable:

( )
.ln
( )
x c
y a b
d x
+
= +

for b>0; -c< x <d ... (5.2)

Type SL- log-normal distribution of random variable:
.ln( ) y a b x c = + + for b> 0; -c<x ... (5.3)
Type SU-non-bounded random variable:

( )
. sinh
x c
y a b A
d

= + for b>0; d>0 ... (5.4)

The use of another theoretical model of distribution
Another way of solution of the problem of non-normality of the monitored characteristics of quality is
based on finding another suitable probability distribution to describe the distribution of characteristics.Perhaps
the most widely used method of evaluation of process capability using a different model of distribution is ISO
method. Similarly as in the case of normal distribution, where the values of -3 and +3 correspond to the
quantiles for which the distribution function reaches the values of 0.00135 and 0.99865, another probability
model uses quantiles corresponding to these values of distribution function. The corresponding values of
capability indices can be calculated as:

0.99865 0.00135
0.5 0.5
0.5 0.00135 0.99865 0.5
min ;
p
pk
USL LSL
C
x x
x LSL USL x
C
x x x x

=
`

)
... (5.5)

where:
X
0.00135
-0.135 % quantile of appropriate distribution
X
0.99865
-99.865 % quantile of appropriate distribution
X
0.5
- median of appropriate distribution

Some Studies On Normal And Non...
www.ijmsi.org 36 | P a g e
Clements Percentile PCI method: Clements method is popular among quality practitioners in industry.
Clements proposed that 6 in equation (4.1) by replaced by the length of the interval between the upper and
lower 0.135 percentage points of the distribution of X. Therefore, the denominator in equation (4.1) can be
replaced by (U
p
-L
p
), i.e.

C
p
= (USL-LSL) / (U
p
-L
p
) (5.6)

where U
p
is the upper percentile i.e. 99.865 percentile of observations and L
p
is the lower percentile i.e. 0.135
percentile of observations. Since the Median M is the preferred central value for a skewed distribution, so C
pu

and C
pl
as follows:
C
pu
= (USL-M) / (Up-M) (5.7)
C
pl
= (M-LSL) / (M-Lp) (5.8)
and C
pk
= min {C
pu
, C
pl
) (5.9)

Clements approach uses the standard estimators of skewness and kurtosis that are based on 3
rd
and 4
th
moments
respectively, and may not be reliable for very small sizes.

An Empirical Study of Normal PCA (In a Chemical Industry)
Problem: The specification of formalin product of a chemical factory is set up as 37 0.5, by weight of
Formaldehyde content. If the weight of formaldehyde gas in Formalin is below 36.5%, the customers do not
accept it and if the same is above 37.5%, it is not affordable to manufacturer so far its cost-benefit margin are
concerned. The data for thirty days has been collected by taking 5 samples from the recorded data sheet from the
factory.
Following section 3, we shall apply the primary techniques like Histogram, probability plot and control
charts (X-bar-R) to the formalin data in our next section for studying process capability analysis.
Histogram and Normal Probability Plot:
Probability plotting is a graphical method for determining whether the sample data conform to a hypothetical
distribution (Normal distribution in our case) based on subjective visual examination of the data. Figure 1 below
presents a normal probability plot (Minitab Version) of the observational data. As the plotted points is falling
approximately along a straight line, so it gives no indication that formalin data is non-normal. However, for
severe departure from normality assumption may need to examine other statistical test, such as non-parametric
test which is outside the preview of our analysis. The histogram and capability analysis is depicted in our figure
2.
By comparing the values of C
p
= 0.67 and C
pk
= .38, we found that C
p
< 1 i.e., 0.67 < 1 so the process is not
fully capable and C
pk
< C
p
i.e., 0.38 < 0.67 implies that the process is off-center. Hence, the user should try to
center the process.

Mean
P
e
r
c
e
n
t
37.75 37.50 37.25 37.00 36.75 36.50
99
95
90
80
70
60
50
40
30
20
10
5
1
Mean
0.868
37.22
StDev 0.2489
N 30
AD 0.201
P-Value
Probability Plot of Mean
Normal


Fig 1. Normal Probability plot
Some Studies On Normal And Non...
www.ijmsi.org 37 | P a g e
37.8 37.6 37.4 37.2 37.0 36.8 36.6
LSL USL
Process Data
SampleN 30
StDev (Within) 0.249
StDev (Ov erall) 0.25102
LSL 36.5
Target *
USL 37.5
Sample Mean 37.216
Potential (Within) Capability
CCpk 0.67
Ov erall Capability
Pp 0.66
PPL 0.95
PPU 0.38
Ppk
Cp
0.38
Cpm *
0.67
CPL 0.96
CPU 0.38
Cpk 0.38
Observ ed Performance
PPM<LSL 0.00
PPM>USL 166666.67
PPMTotal 166666.67
Exp. Within Performance
PPM<LSL 2016.93
PPM>USL 127026.07
PPMTotal 129043.00
Exp. Ov erall Performance
PPM<LSL 2169.74
PPM>USL 128946.30
PPMTotal 131116.04
Within
Overall
Process Capability Analysis for Mean


Fig. 2. Histogram of analyzed data
Control Chart:
If the process is under statistical control, then we are able to reliably use our estimates for spread and
location. In order to assess whether or not a process is under statistical control, which are sensitive to large shifts
in the process parameters we may use X-bar-R chart or X-bar-S chart. We shall depict here the X-bar-R chart
(Minitab version) in Figure-3 below which shows that process is under statistical control and thus we can
reliably use our estimates to measure the spread and location of the Formalin data for analysing process
capability.

Sample
S
a
m
p
l
e

M
e
a
n
15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
37.6
37.4
37.2
37.0
36.8
_
_
X=37.2159
UCL=37.6576
LCL=36.7743
Sample
S
a
m
p
l
e

R
a
n
g
e
15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
0.8
0.6
0.4
0.2
0.0
_
R=0.2348
UCL=0.7673
LCL=0
Xbar-R Chart of Mean

Fig.3. X-bar- R Control Chart


Numerical example for Non-normal PCA:
To illustrate the use of above described method for non-normal PCA, 25 subgroups with a subgroup
range of 5 for Gamma and Weibull distribution were generated by computer software. For Gamma distribution
with parameters of =4.75 and =5.24, the tolerance limits have been set as follows: LSL= 5 and USL = 70. The
character of distribution of the monitored feature and its position in relation to the tolerance limits are shown in
figure 4.
The values of C
p
and C
pk
determined using different methods is presented in Table 1. (for Gamma
distribution)
Some Studies On Normal And Non...
www.ijmsi.org 38 | P a g e
63 54 45 36 27 18 9 0
LSL USL
Process Data
SampleN 25
Shape 4.76
Scale 5.24281
LSL 5
Target *
USL 70
Sample Mean 24.9558
Ov erall Capability
Pp 0.93
PPL 0.93
PPU 0.93
Ppk 0.93
Observ ed Performance
PPM<LSL 0
PPM>USL 0
PPMTotal 0
Exp. Ov erall Performance
PPM<LSL 4575.82
PPM>USL 2173.16
PPMTotal 6748.98
Process Capability of mean
Calculations Based on Gamma Distribution Model


Fig.4 Histogram of analyzed data

Table 1: Capability indices calculated by means of different methods

Method Cp Cpk
Box-Cox transformation 2.16 1.72
Johnson transformation 2.05 1.26
ISO method 0.93 0.93
Clements 2.55 2.13

Similarly, For Weibull distribution with parameters of =4 and =1, the tolerance limits have been set as
follows: LSL= 0.2 and USL = 2. The character of distribution of the monitored feature and its position in
relation to the tolerance limits are shown in figure 5.The values of C
p
and C
pk
determined using different
methods is presented in Table 2. (for Weibull distribution)

2.00 1.75 1.50 1.25 1.00 0.75 0.50 0.25
LSL USL
Process Data
SampleN 25
Shape 4
Scale 0.927951
LSL 0.2
Target *
USL 2
Sample Mean 0.910077
Ov erall Capability
Pp 1.37
PPL 0.97
PPU 1.80
Ppk 0.97
Observ ed Performance
PPM<LSL 0
PPM>USL 0
PPMTotal 0
Exp. Ov erall Performance
PPM<LSL 2155.51
PPM>USL 0.00
PPMTotal 2155.51
Process Capability of mean
Calculations Based on Weibull Distribution Model


Fig 5. Histogram of analysed data

Some Studies On Normal And Non...
www.ijmsi.org 39 | P a g e
Table 2: Capability indices calculated by means of different methods

Method Cp Cpk
Box-Cox transformation 6.48 1.77
Johnson transformation 2.81 2.21
ISO method 1.37 0.97
Clements 4.71 3.91

V. CONCLUSION:
The tabulated results of process capability indices in Table 1 and 2 above clearly indicate that the
process capability analysis of data for non-normal distribution may significantly differ depending on methods of
analysis of process capability.It is also evident that, the obtained indices from the different approaches reflect
that the process that produces the data considered in this study are off-centered because here C
pk
< C
p
for all the
cases except ISO method though they are capable to meet the specification. If the process is off-centered, the
user should try to adjust the process.

REFERENCES:
[1] M.N. Sinha and W.O. Willborn : Quality Control, 2
nd
Edition. Prentice Hall. New Jersey. NJ, USA. 1985
[2] M. Juran, Joseph and M. Gryana, Frank. : Jurans Quality Control Hand Book, Fourth Edition, 1988, Mc Graw-Hill Book
Company.
[3] G.S. Wasserman, H.A. Mohsen, and L.A. Franklin : A programme to calculate bootstrap confidence intervals for process
capability index CPK, Communication in Statistics Simulation and Computation, 1991, 20, (2/3), 397-510.
[4] T. Pyzdek: Process Capability Analysis using Personal Computers. Quality Engineering, 1992, 4(3), 419-440.
[5] D.H. Lester, N.L. Enrich and H.E. Motely : Quality Control for Profit, Industrial Press, New York, NY USA.1997
[6] D.C. Montgomery: Statistical Quality Control- A modern introduction, Six Edition by John Willey & Sons, Inc., 2011
[7] A. Sagbas : Improving the Process Capability of a turning operation by the application of statistical Techniques, Materials and
Technology, 2009, 43 (1), pg. 55-59.
[8] R. Gogoi : Process Capability Analysis in a working chemical manufacturing industry. Unpublished Ph.d Thesis, 2003,
Department of Statistics, Dibrugarh University (Assam).
[9] K. Vannman , and M. Albing : Process Capability Indices for One sided Specification limits and Skew zero-bound Distribution
Luela University of Technology, Department of Mathematics, 2005, Research Report 11.SE-97187 Lulea, Sweden
[10] M. Albing. : Process Capability Analysis with focus on Indices for One-sided specification limits, Department of Mathematics,
2006, Lulea University of Technology, SE-97187 Lulea, SWEDEN.
[11] M. Albing and K. Vannman : Process Capability Indices for One sided Specification limits and Skew zero-bound Distribution
Luela University of Technology, Department of Mathematics, 2006a, Research Report 11.SE-97187 Lulea, Sweden.
[12] H. Udhayasee: Assessing Initial Process Capability Published in Quality World Feb. 2007.
[13] D.R. Guevera, and J.A. Vargas : Comparison of Process Capability Indices under Autocorrelated data. Revista Colombian de
Estdistica, 2007, Vol.30, No.2. pp. 301-316.
[14] A. Jeang, C.P. Chung , H.C. Li and M.H. Sung : Process Capability Index for off-line Application of Product life cycle.
Proceeding of the World Congress on Engineering 2008, Vol.55, WCE 2008, July 2-4, London, U.K.
[15] A. Czarski : Estimation of process capability indices in case of distribution unlike the normal one. Archieves of Materials
Science and Engineering, International Scientific Journal,2008, Published by World Academy of Materials and Manufacturing
Engineering, Vol.34, Issue-1, Pg. 39-42.
[16] S.M. Ali, M.A. Sarwar and A. Sultana: Finding Characteristics of Process Capability Index Cpk with different distribution and
sample size. Journal of Quality and Technology Management, 2008, Vol. IV, Issue 11, Pg. 5-11.
[17] J.J. Flaig J.J.: A new approach to Process Capability Analysis, Quality Engineering, 2009, 9(2), 205-211
[18] Lundkvist: A Comparison of Decision Methods for Cpk when data are autocorrelated, Quality Engineering, 2012, 24: 460-
472.
[19] J.N. Pan & S.L. Wu : Process Capability Analysis for Non-normal Relay Test Data, Microelectron. Reliab., 1996, Vol.37, No.
3, pp. 421-428
[20] W.L. Pearn, & K.S. Chen : Capability Indices for Non-normal distribution with an application in Electrolytic Capacitor
Manufacturing, Microelectron. Reliab., 1997, Vol.37, No. 12, pp. 1853-1858
[21] S. Bittani, M. Lovera & L. Moiraghi : Application of Non-Normal Process Capability Indices to Semiconductor Quality
Control, IEEE TRANSACTION ON SEMICONDUCTOR MANUFACTURING, 1998, Vol. 11, No. 2
[22] B. Abbasi, M. Ahmad , M. Abdollahian & P. Zeephongsekul : Measuring Process Capability for Bivariate Non-Normal Process
Using the Bivariate Burr Distribution, WSEAS TRANSACTIONS on BUSINESS and ECONOMICS, 2007, Issue 5, Vol.4, Pg.70-
77
[23] G. Casalino, & A. Rotondo : Multivariate Process Incapability Index for Non-normal Data: A case Study, DAAAM
International Scientific Book, 2007, pp- 071-086, Chapter 07.
[24] S. Ahmad, M. Abdollahian, and P. Zeephongsekul : Process Capability Estimation for non-normal quality characteristics: a
comparison of clement, Burr and Box-Cox method, ANEIAM, 2008, J.49 (EMAC 2007). Pp. 642.
[25] J.A. Clements: Process Capability Calculations for non-normal distribution. Quality Progress, 1989 22: 95-100.
[26] G.E.P Box and D.R. Cox : An analysis of transformation, J Roy Stat Soc, 1964, B 26: 211-252.
[27] M. Albing: Process Capability Indices for Weibull Distribution and Upper Specification Limits, Quality and Reliability
Engineering International; 2009, 25:317-334.
[28] A. Czarski : Analysis of CLEMENTs Method for Capability Indices Estimation, Metallurgy and Foundry Engineering, 2009,
Vol. 35, No.1
[29] C.R. Lovelace , & J.J. Swain : Process Capability Analysis Methodologies for Zero Bound, Non-Normal Process Data,
Quality Engineering, 2009, 21: 190-202
[30] E. Ahmed : Process Capability Analysis for Non-normal data, Pakistan Business Review.2010
Some Studies On Normal And Non...
www.ijmsi.org 40 | P a g e
[31] R.R.L. Kantam, K. Rosaiah and R. Subba Rao : Estimation of Process Capability Index for Half Logistic Distribution,
International Transactions in Mathematical Sciences and Computer, 2010, Vol. 3, No. 1, pp-61-66
[32] L.K. Chan, S.W. Cheng, and F.A. Spiring : A New Measure of Process Capability: Cpm, Journal of Quality Technology, 1988,
vol. 20(3), pp.160-175.
[33] R.A. Boyles: The Taguchi Capability Index, Journal of Quality Technology, 1991, Vol.23 (2), pp.107-126