See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/264005236

Control Charts for Variables with Specified Process Capability Indices

Article · January 2012

CITATIONS READS

26 13,583

2 authors:

Jambulingam Subramani Saminathan Balamurali

Pondicherry University Kalasalingam Academy of Research and Education
128 PUBLICATIONS 1,275 CITATIONS 222 PUBLICATIONS 3,426 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jambulingam Subramani on 18 September 2014.

The user has requested enhancement of the downloaded file.

International Journal of Probability and Statistics 2012, 1(4): 101-110
DOI: 10.5923/j.ijps.20120104.03

Control Charts for Variables with Specified Process

Capability Indices
J. Subramani1,* , S. Balamurali2

1
Department of Statistics, Pondicherry University, R V Nagar, Kalapet, Puducherry, 605 014, India
2
Department of Computer Applications, Kalasalingam University, Krishnankoil, 6426 190, Srivilliputhur,Tamilnadu, India

Abstract Control charts, also known as Shewhart charts in statistical process control are the statistical tools used to
determine whether a manufacturing process is in a state of statistical control or not. If analysis of the control chart indicates
that the process is currently under statistical control (i.e. is stable, with variation only co ming fro m sources common to the
process) then no corrections or changes to process control parameters are needed or desirable. In addition, data fro m the
controlled process can be used to predict the future performance of the process with the help of process capability indices.
That is, the suitability and the performance of the manufacturing process are assessed in two stage processes namely control
charts and process capability indices. Th is paper deals with the techniques namely, capability based control charts which
combines the two stage processes into a single stage process in industrial applicat ions for on line process control. The relative
performance of the capability based control charts for variables is assessed with that of existing usual control charts for
variables, namely X − Charts and R − Charts . It is observed that the proposed method is simp le to apply and does not warrant
any tedious computations both for control charts and for co mputing process capability indices. Fo r the sake o f convenience
we have presented the control charts constants for constructing capability based R − Charts and X − Charts . The proposed
method is also illustrated with the help of a numerical examp le.
Keywords Control Charts, Industrial Applicat ions, Process Capability Index, Quality Characteristics

estimate of the quality characteristics. Further one has to

1. Introduction show that the process is under statistical control before
computing the process capability indices. Nowadays process
The present scenario of total quality management capability analysis is a mandatory requirement and to have a
demands the effective use of statistical tools for analyzing minimu m prescribed value for the process capability indices
quality problems and improving the performance of Cp and C pk to be a supplier for original manufacturing
production process. Measuring, monitoring and maintaining companies, particularly in the field of automobile industries.
several quality characteristics to achieve further Process capability indices (PCI) have proliferated in both use
improvements are required in every production process. and variety during the last three decades. Their widespread
Many of the quality characteristics are measurable in nature and often uncritical use led to imp rovements in quality.
and can be expressed in terms of nu merical measurement by Process capability indices are considered as a practical
measuring them using micro meter, vernier calipers etc. For tool by many advocates of the statistical process control
example, length, diameter, weight, pocket width, distance industry. They are used to determine whether the
and groove diameter of a shaft are some of the quantifiab le manufacturing process is capable of producing units within
quality characteristics encountered during the production of the product or process specifications. A capability index is a
shafts. When dealing with measurable quality characteristics, dimensionless measure based on the process parameters and
it is usually necessary to monitor the behaviour of the mean the process specifications, designed to quantify, in a simp le
value of the quality characteristics and its variab ility. To and easily understood way, the performance of the process.
study the behaviour of the production process and to take a Several process capability indices have been developed
necessary action on the process, control charts for variables among wh ich the basic and widely used indices are C p and
are very much used wh ich requ ires to co llect rando m Cpk. (see Kane[8]). A detailed review of the process
samples fro m the production process and to compute the
capability indices can be seen in Kotz and Johnson[9].
The process capability indices are unit less measures
* Corresponding author:
drjsubramani@yahoo.co.in (J. Subramani) showing the capability of a manufacturing process whether it
Published online at http://journal.sapub.org/ijps is capable of operating within its specifications limits. If the
Copyright © 2012 Scientific & Academic Publishing. All Rights Reserved value of the PCI exceeds one, then the process is considered
102 J. Subramani et al.: Control Charts for Variables with Specified Process Capability Indices

to be satisfactory or capable. However th is approach ignores process capability indices. We are also presenting the control
the fact that these PCIs are rando m variables with charts constants for constructing capability based control
distributions. Confidence limits play a major ro le for the charts. The proposed method is also illustrated with a
correct interpretation of PCIs. Recently, techniques and numerical examp le.
tables were developed to construct confidence limits for the
process capability indices. These techniques are based on the
assumption that the underlying process is normal. Moreover, 2. Control Charts and Process
since many processes can frequently have skewed or heavily Capability Analysis
tailed distributions. In practice, the interval estimation
2.1. Usual Method
technique that is free fro m the assumption of distribution is
desirable. The various other properties and the behaviour of Set the
m anufactur ing pr ocess
process capability indices at different type of distributions
were studied by many authors. For example, the 95%
confidence limits for the process capability indices Cp and Im plem ent contr ol char ts
for var iables
Cpk were constructed by Chou, Owen and Bo rrego[5]. As
their limits on Cpk can produce 97% or 98% lower confidence
Is
limits (instead of 95%) making them conservative, an R-Char t No
approximation presented by Bissell[4] is reco mmended. Under Contr ol?
Franklin and Wasserman[6] presented an init ial study of the Yes
properties of these three bootstrap confidence intervals for Is
X-bar -Char t No
Cpk. Franklin and Wasserman[7] constructed the confidence Under Contr ol?
limits for some basic capability indices and examined their
Yes
behaviors when the underlying process is normal, skewed or
Com pute Pr ocess capability
heavily tailed distributions. Balamurali and Kalyanasundara Indices Cp and Cpk
m[3] constructed bootstrap confidence limits for the
capability indices Cp , Cpk and Cpm based on lognormal and Is No
chi-squared distributions. Balamu rali[2] considered the Cp Value OK?
above capability indices under short run production
Yes
processes and constructed the confidence limits.
Is
However the problems one has to face are the following: Cpk Value OK?
No
There is a considerable time taken for establishing and
Yes
monitoring the process through control charts for variables;
Establish the
inadequate technical personnel to correctly interpret the m anufactur ing pr ocess
outcome of the control charts; unnecessary inventory and
delay due to the computation of process capability indices Continue the m anufactur ing
and to take a decision on the process based on the resulting
values of process capability indices. Hence it is felt to have Yes Do you No
an alternative method to achieve the benefit of both control want to im pr ove
the pr ocess?
charts for variables and the process capability analysis. As a
result we have proposed new type of control charts for
Figure 1. Present M anufacturing System
variables based on the required process capability indices C p
For the sake of convenience to the readers, we have
and C pk , called as “control charts for variables with presented the control charts for variables for checking
specified value for process capability indices”. Further we whether the process is under statistical control or not. That is,
have presented the control charts constants to construct the to take a decision whether the process can be allowed further
proposed control charts and are explained with the help of a without making any adjustment or to take corrective actions
numerical examp le. For a more detailed discussion on the if any, to bring back the process under statistical control.
control charts for variab les and process capability analysis, Once the process has been brought under the state of
the readers are referred to Montgomery[10], Spiring[11], statistical control by both R − Chart and X − Chart then
Sarkar and Pal[1], Subraman i[12, 13], Subramani and
Balamurali[14] and the references cited therein. In this paper, the process capability indices C p and C pk can be co mputed.
we propose a capability based control chart for on line If the values of the process capability indices are at the
process control. The relat ive performance o f this control satisfactory level, the process can be allowed further without
charts for variables is assessed with that of existing usual making any change or adjustment in the process. Otherwise
control charts for variables. It is observed that the proposed suitable adjustments have to be made so that the process shall
method is simple to apply and does not warrant any tedious satisfy both the requirements of control charts and the
computations both for control charts and for computing process capability indices. The present manufacturing
 International Journal of Probability and Statistics 2012, 1(4): 101-110 103

process can be explained in the flo w chart as shown in Figure 2.2. Proposed Method- Control Charts for Variables with
1. Specified C p Value
The following are the various computational fo rmulae
Suppose that a customer wants to have products produced
involved in constructing the control charts and for the in a process with a specified value for the process capability
computation of process capability indices.
index C p . To achieve this, the manufacturer has to establish
1 n
∑ xi : Mean of ith samp le
Xi =
n i =1
a process under statistical control and then the process
capability index has to be computed. If the computed value
1 N of Cp is more than the specified value given by the
X = ∑ xi : Grand Mean
N i =1 customer, the process can be run without any adjustment.
Otherwise suitable actions have to be taken to meet the
=Ri Max( xi ) − Min( xi ) : Range of ith samp le
required process capability values. The drawback of this
1 k method is that the manufacturer has to keep all the produced
R= ∑ Ri : Mean of Range values
k i =1 items in the inventory until the co mputation of the process
capability index and to take a decision on the process based
UCLx= X + A2 R : Upper Control Limit for X -Chart on the process capability index. It is a time consuming
procedure as well as requires addit ional skilled personnel. To
LCLx= X − A2 R : Lo wer Control Limit for X -Chart avoid this, we have proposed control charts for variables
with specified p rocess capability indices. In the proposed
UCLR = D4 R : Upper Control Limit for R-Chart method, the process is under statistical control means that it
LCLR = D3 R : Lo wer Control Limit for R-Chart satisfies the required process capability index. Further it is
not necessary to compute the process capability indices
USL − LSL separately. The proposed manufacturing process can be
Cp = : Process Capability
6R / d2 explained in the flow chart shown in Figure 2.

USL − X Specify Cp and Cpk

C pu = : Upper Process Capability Index
3R / d 2
X − LSL Implement control charts for
C pl = : Lower Process Capability Index variables with Cp and Cpk
3R / d 2
C pk = Min(C pu , C pl ) : Process Capability Index
Set the
Tolerance = T = USL − LSL manufacturing process
USL + LSL
T arg et = M =
2 Is
LSL: Lower Specification Limit of the Quality R-Chart No
Characteristics Under Control?

USL: Upper Specification Limit of the Quality Yes

Characteristics Is
When the sample size n =5, the values for the control chart X-bar-Chart No
Under Control?
constants obtained from the table are as given below:
D4 = 2.114, D3 = 0, d 2 = 2.326 and A 2 = 0.577 Yes
The Control limits for constructing R − Charts and Establish the
X − Charts are as given below: manufacturing process
Control limits for R − Chart :
Continue the manufacturing
CLR = R
LCLR = D3 R Do you
Yes want to improve No
UCLR = D4 R the process?
Control limits for X − Chart :

CLx = X Figure 2. Present M anufacturing System

LCLx= X − A2 R The following are the derivation of various computational

formulae involved in constructing the proposed control
UCLx= X + A2 R charts with specified process capability index C p .
104 J. Subramani et al.: Control Charts for Variables with Specified Process Capability Indices

USL − LSL D3d 2

Consider C p = Where D3* =
6σ 6
T
⇒ Cp = UCLR = D4 R
6R / d2
D4 d 2T
If C p value is g iven then R value can be obtained fro m ⇒ UCLR =
6C p
d 2T
the above formula as R = In the similar manner one can compute the control limits
6C p
to construct X − Chart with specified C p value as given
T d
D*
⇒R= *
Where D = 2 below.
Cp 6 Center line= CLX = X
Similarly one can obtain the lower control limit (LCL) and
upper control limit (UCL) of the R − Chart as given Lower control limit for the proposed X − Chart is
obtained as given below:
below:
* T LCLx= X − A2 R
Center Line= CLR= R= D T
Cp X − A2 D*
⇒ LCLx =
LCLR = D3 R Cp
D3d 2T T
⇒ LCLR = X − A2*
⇒ LCLx = Where A2* = A2 D*
6C p Cp

T T Where * D4 d 2
D3*
⇒ LCLR = D4*
⇒ UCLR = D4 =
Cp Cp 6

Table 1. Control Charts Constants for specified C p value

n d2 D3 D4 A2 D* D3* D4* A2*

2 1.128 0 3.267 1.880 0.1880 0.0000 0.6142 0.3534
3 1.693 0 2.574 1.023 0.2822 0.0000 0.7263 0.2887
4 2.059 0 2.282 0.729 0.3432 0.0000 0.7831 0.2502
5 2.326 0 2.115 0.557 0.3877 0.0000 0.8199 0.2159
6 2.534 0 2.004 0.483 0.4223 0.0000 0.8464 0.2040
7 2.704 0.076 1.924 0.419 0.4507 0.0343 0.8671 0.1888
8 2.847 0.136 1.864 0.373 0.4745 0.0645 0.8845 0.1770
9 2.970 0.184 1.816 0.337 0.4950 0.0911 0.8989 0.1668
10 3.078 0.223 1.777 0.308 0.5130 0.1144 0.9116 0.1580
11 3.173 0.256 1.744 0.285 0.529 0.135 0.922 0.151
12 3.258 0.283 1.717 0.266 0.543 0.154 0.932 0.144
13 3.336 0.307 1.693 0.249 0.556 0.171 0.941 0.138
14 3.407 0.328 1.672 0.235 0.568 0.186 0.949 0.133
15 3.472 0.347 1.653 0.223 0.579 0.201 0.957 0.129
16 3.532 0.363 1.637 0.212 0.589 0.214 0.964 0.125
17 3.588 0.378 1.622 0.203 0.598 0.226 0.970 0.121
18 3.640 0.391 1.608 0.194 0.607 0.237 0.976 0.118
19 3.689 0.403 1.597 0.187 0.615 0.248 0.982 0.115
20 3.735 0.415 1.585 0.180 0.623 0.258 0.987 0.112
21 3.778 0.425 1.575 0.173 0.630 0.268 0.992 0.109
22 3.819 0.434 1.566 0.167 0.637 0.276 0.997 0.106
23 3.858 0.443 1.557 0.162 0.643 0.285 1.001 0.104
24 3.895 0.451 1.548 0.157 0.649 0.293 1.005 0.102
25 3.931 0.459 1.541 0.153 0.655 0.301 1.010 0.100
 International Journal of Probability and Statistics 2012, 1(4): 101-110 105

One can use the above control limits to construct index C pk . The fo llo wing are the derivation of various
R − Chart with specified C p value. computational formu lae involved in constructing control
Similarly one may obtain the upper control limit for the charts for variables with specified process capability index
proposed X − Chart as C pk
UCLx= X + A2 R
{
Consider C pk = Min C pl , C pu }
T *
⇒ UCLx =
X + A2 D USL − X X − LSL
Cp Where C pu = and C pl =
T
3R / d 2 3R / d 2
X + A2*
⇒ UCLx = Where A2* = A2 D* After a little algebra, one can rewrite the C pk as
Cp
Thus the control limits for the control charts with specified T
− X −M
C p value are as given below: C pk = 2
3R / d 2
Control limits for proposed R − Chart :
T If C pk value is given then R value can be obtained fro m
CLR = D*
Cp the above formula as
T T  T 
T − X − M d2  − X − M   − X −M 
2 d 2
LCLR = D3* = R 2=  =  2  
Cp 3C pk / d 2 3C pk 3 C pk

T T 
UCLR = D4*  − X −M  d
Cp Dk* 
⇒R=
2  Where Dk* = 2
C pk 3
Control limits for proposed X − Chart :
Similarly one can obtain the lower control limit (LCL) and
CLx = X upper control limit (UCL) of the R − Chart as given
T below:
LCLx= X − A2* T 
Cp − X −M 
 2
T CenterLine= CLR= R= Dk*  
UCLx= X + A2* C pk
Cp
LCLR = D3 R
One can use the above control limits to construct the
T 
R − Chart and X − Chart with specified C p value. For  − X −M 
2
the sample size n (2 ≤ n ≤ 25) , we have presented the D3 Dk* 
⇒ LCLR = 
C pk
values of the control charts constants D* , D2* , D3* and A2*
T 
in the Table 1 given belo w.  − X −M 
2
The advantageous of the proposed control charts is that the D3*k 
⇒ LCLR = 
observed range values are not required for co mputing the C pk
control limits. Further if the process is under the state of
D3d 2
statistical control means that it automatically satisfies the D3*k
Where = = D3 Dk* UCLR = D4 R
required conditions regarding the values of p rocess 3
capability index C p . T 
 − X −M 
2
2.3. Proposed Method- Control Charts for Variables with D4 Dk* 
⇒ UCLR = 
C pk
Specified C pk value T 
 − X −M 
2
In the section 2.2, we have introduced a method for D4*k 
⇒ UCLR =  where
constructing control charts for variables with specified value C pk
for the process capability index C p . In the similar manner D4 d 2
D4*k
= = D4 Dk*
we have developed a method for constructing control charts 3
for variab les with specified value for the process capability One can use the above control limits to construct
106 J. Subramani et al.: Control Charts for Variables with Specified Process Capability Indices

R − Chart with specified C pk value. T 

In the similar manner one can compute the control limits  2 − X − M 
UCLx= X + A2*k
to construct X − Chart with specified C pk value as given C pk
below. The above control limits can be fu rther simp lified by
Center line= CLX = X substituting the values of T and M , wh ich lead to the
following two cases:
Lower control limit for the proposed X − Chart is Case 1: When X > M
obtained as given below:
Control limits for R − Chart with specified C pk value
T 
 − X −M 
2 USL − X 
Dk* 
⇒R=  LCL= X − A R
x 2 CLR= R= Dk*  
C pk
 C pk 
T  USL − X 
 − X − M 
* 2 LCLR = D3*k  
⇒ LCLx =
X − A2 Dk
C pk  C pk 
T  USL − X 
 − X −M  UCLR = D4*k  
⇒ LCLx = X − A2*k 
2  Where A2*k = A2 Dk*  C pk 
C pk
Control limits for the proposed X − Chart with
Similarly one may obtain the upper control limit for the
specified C pk value:
proposed X − Chart as
UCLx= X + A2 R CLx = X
T  USL − X 
 − X −M  LCLx= X − A2*k 
2 
X + A2 Dk* 
⇒ UCLx = 
 C pk 
C pk
USL − X 
T  UCLx= X + A2*k  
 − X −M   C pk 
⇒ UCLx = * 2
X + A2 k Where A2*k = A2 Dk*
C pk
Case 2: When X < M
Thus the control limits for the proposed control charts
Control limits for R − Chart with specified C pk value
with specified C pk value are as given below:
 X − LSL 
Control limits for R − Chart with specified C pk value CLR= R= Dk*  
T   C pk 
 − X −M 
2  X − LSL 
CLR= R= Dk*  
LCLR = D3*k 
C pk 
 C pk 
T 
 2 − X − M   X − LSL 
LCLR = D3*k UCLR = D4*k  
C pk  C pk 
T  Control limits for the proposed X − Chart with
 2 − X − M  specified C pk value:
UCLR = D4*k
C pk
CLx = X
Control limits for the proposed X − Chart with
 X − LSL 
specified C pk value: LCLx= X − A2*k  
 C pk 
CLx = X
 X − LSL 
T  UCLx= X + A2*k  
 2 − X − M   C pk 
LCLx= X − A2*k
C pk
 International Journal of Probability and Statistics 2012, 1(4): 101-110 107

Table 2. Control Charts Constants for specified C pk value

*
n d2 D3 D4 A2 Dk* D3k *
D4k *
A2k
2 1.128 0 3.267 1.880 0.376 0.000 1.228 0.707
3 1.693 0 2.574 1.023 0.564 0.000 1.453 0.577
4 2.059 0 2.282 0.729 0.686 0.000 1.566 0.500
5 2.326 0 2.115 0.557 0.775 0.000 1.640 0.432
6 2.534 0 2.004 0.483 0.845 0.000 1.693 0.408
7 2.704 0.076 1.924 0.419 0.901 0.069 1.734 0.378
8 2.847 0.136 1.864 0.373 0.949 0.129 1.769 0.354
9 2.970 0.184 1.816 0.337 0.990 0.182 1.798 0.334
10 3.078 0.223 1.777 0.308 1.026 0.229 1.823 0.316
11 3.173 0.256 1.744 0.285 1.058 0.271 1.845 0.301
12 3.258 0.283 1.717 0.266 1.086 0.307 1.865 0.289
13 3.336 0.307 1.693 0.249 1.112 0.341 1.883 0.277
14 3.407 0.328 1.672 0.235 1.136 0.372 1.899 0.267
15 3.472 0.347 1.653 0.223 1.157 0.402 1.913 0.258
16 3.532 0.363 1.637 0.212 1.177 0.427 1.927 0.250
17 3.588 0.378 1.622 0.203 1.196 0.452 1.940 0.243
18 3.640 0.391 1.608 0.194 1.213 0.474 1.951 0.235
19 3.689 0.403 1.597 0.187 1.230 0.496 1.964 0.230
20 3.735 0.415 1.585 0.180 1.245 0.517 1.973 0.224
21 3.778 0.425 1.575 0.173 1.259 0.535 1.983 0.218
22 3.819 0.434 1.566 0.167 1.273 0.552 1.994 0.213
23 3.858 0.443 1.557 0.162 1.286 0.570 2.002 0.208
24 3.895 0.451 1.548 0.157 1.298 0.586 2.010 0.204
25 3.931 0.459 1.541 0.153 1.310 0.601 2.019 0.200

One can use the above control limits to construct the σ = R/d = 0.00992

R − Chart and X − Chart with specified C pk value. For Case 1: Usual Method of Control Charts and the
the given sample size n (2 ≤ n ≤ 25) , we have presented the Co mputation of Process Capability Indices
values of the control charts constants Control Limits for R-Chart are obtained as:

Dk* , D2*k , D3*k and A2*k in the Table 2 g iven below. CLR= R= 0.02324
The advantageous of the proposed control charts with LCL
= R D=
3*R 0*0.02324
= 0
Cp C pk UCL D=
= 4 * R 2.115*0.02324
= 0.0492
specified and values, is that the observed range
R
values are not required for co mputing the control limits. Control Limits for X -Chart are obtained as:
Further if the process is under the state of statistical control
means that it automatically satisfies the required conditions CL=
x X= 74.00118
regarding the values of process capability indices C p and LCL x = X − A2 * R
C pk . = 74.00118 − 0.577 * 0.02324 = 73.98777
UCLx = X + A2 * R
3. Numerical Example = 74.00118 − 0.577 * 0.02324 = 74.01459
Consider the data given below in Table 3 is the Example By plotting the control limits, sample ranges and sample
5.1 (Montgomery[10], page 213). The data is pertaining to means, one may get R − Chart and X − Chart as
the manufacturing of Piston Rings for an auto motive engine given in Figure 3 and Figure 4. Fro m the R − Chart, we
produced by a forging process. Twenty five samples, each of observe that all the plotted sample range values are falling
size five have been taken and the inside d iameter is measured. within the control limits. Hence we may conclude that the
The resulting data together with the sample means and variations are under control. Similarly fro m the X − Chart ,
sample range values are given below in Table 3. we observe that all the plotted sample means are falling
Fro m the above values we have obtained the follo wing: inside the control limits. Hence we may conclude that the
X = 74.0011 , R = 0 . 0232 and averages are also under the statistical control.
108 J. Subramani et al.: Control Charts for Variables with Specified Process Capability Indices

Table 3. Data of Inside Diameter of Piston Rings (Spec: 74.000±0.05 mm)

Sample Number x1 x2 x3 x4 x5 Sample Mean Range R
1 74.030 74.002 74.019 73.992 74.008 74.0102 0.038
2 73.995 73.992 74.001 74.011 74.004 74.0006 0.019
3 73.988 74.024 74.021 74.005 74.002 74.0080 0.036
4 74.002 73.996 73.993 74.015 74.009 74.0030 0.022
5 73.992 74.007 74.015 73.989 74.014 74.0034 0.026
6 74.009 73.994 73.997 73.985 73.993 73.9956 0.024
7 73.995 74.006 73.994 74.000 74.005 74.0000 0.012
8 73.985 74.003 73.993 74.015 73.988 73.9968 0.030
9 74.008 73.995 74.009 74.005 74.004 74.0042 0.014
10 73.998 74.000 73.990 74.007 73.995 73.9980 0.017
11 73.994 73.998 73.994 73.995 73.990 73.9942 0.008
12 74.004 74.000 74.007 74.000 73.996 74.0014 0.011
13 73.983 74.002 73.998 73.997 74.012 73.9984 0.029
14 74.006 73.967 73.994 74.000 73.984 73.9902 0.039
15 74.012 74.014 73.998 73.999 74.007 74.0060 0.016
16 74.000 73.984 74.005 73.998 73.996 73.9966 0.021
17 73.994 74.012 73.986 74.005 74.007 74.0008 0.026
18 74.006 74.010 74.018 74.003 74.000 74.0074 0.018
19 73.984 74.002 74.003 74.005 73.997 73.9982 0.021
20 74.000 74.010 74.013 74.020 74.003 74.0092 0.020
21 73.982 74.001 74.015 74.005 73.996 73.9998 0.033
22 74.004 73.999 73.990 74.006 74.009 74.0016 0.019
23 74.010 73.989 73.990 74.009 74.014 74.0024 0.025
24 74.015 74.008 73.993 74.000 74.010 74.0052 0.022
25 73.982 73.984 73.995 74.017 74.013 73.9982 0.035
Average 74.00118 0.02324

The manufacturing capability of a process can normally be

0.06
evaluated in terms of process capability indices Cp and
0.05
C pk . The process capability indices obtained from the above
Range Values

0.04
0.03 values are given below:
0.02 USL − LSL 74.05 − 73.95
=Cp = = 1.6681
0.01 6R / d2 6*0.2324 / 2.326
0.00 USL − X 74.05 − 74.00118
C pu
= = = 1.6287
1 3 5 7 9 11 13 15 17 19 21 23 25 3R / d 2 3*0.2324 / 2.326
X − LSL 74.00118 − 73.95
LCL-R R-Bar UCL-R Range = =C pl = 1.7075
3R / d 2 3*0.2324 / 2.326
Figure 3. R-Chart- Usual Method
=C pk Min
= (C pu , C pl ) Min(1.6287,1.7075)
= 1.6287
If the customer’s requirement of the process capability
74.03 index Cp is 1.5, then one may conclude fro m the above
74.02 capability indices that the process is an efficient one.
Average Values

However If the customer’s requirement of the process

74.01 capability index Cp is 2.0, then one may conclude fro m the
74.00 above capability indices that the process is not an efficient
one and hence the process has to be adjusted so as the
73.99 resulting process capability index Cp is at least 2.0.
73.98 Case 2: Control Charts with specified C p = 1.5
1 3 5 7 9 11 13 15 17 19 21 23 25 In this case we construct the control limits for the given
value of process capability index C p and hence separate
S_ Mean LCL X-Bar UCL
computation of the process capability index is not required.
Further the proposed control charts states that the process is
Figure 4. X-Bar Chart- Usual Method under the statistical control means that the process satisfies
 International Journal of Probability and Statistics 2012, 1(4): 101-110 109

the requirement of the process capability index. Hence one values are falling within the control limits. Hence we may
can construct several control limits for different values of conclude that the variations are under control. Similarly fro m
process capability index Cp . If any plotted points falls the X − Chart , we observe that all the plotted sample
outside the control limits with specified Cp = 1.5 means that means are falling inside the control limits. Hence we may
the process does not satisfy the required process capability conclude that the averages are also under the statistical
index Cp = 1.5. control. Further we conclude that the process capability
Let the given process capability index Cp = 1.5. Then the index for the given p rocess is at least 1.5 and satisfies the
Control limits for R − Chart with specified Cp = 1.5 are customer’s requirements.
obtained as given below:
Case 3: Control Charts with specified Cp =1.5
T *  0.1  In this case we construct the control limits for the given
CLR D=
= . 0.3877=
1.5  0.2585
Cp value of process capability index Cpk and hence separate
computation of the process capability index is not required.
* T  0.1  Further the proposed control charts states that the process is
LCLR D=
= 3 0=
1.5  0
Cp under the statistical control means, the process satisfies the
requirement of the process capability index. Hence one can
* T  0.1  construct several control limits for different values of
UCLR D=
= 4. 0.8199=
 0.0547
Cp 1.5  process capability index. If any p lotted points falls outside
the control limits with specified Cpk=1.5 means that the
Control limits for X − Chart with specified C p = 1.5 process does not satisfy the required process capability index.
Cpk=1.5.
CL=
x X= 74.00118 Let the given process capability index C p = 1.5. Further
T  0.1  X = 74.00118 > M = 74.00 the control limits with specified
X − A2*.
LCLx = 74.00118 − 0.2159   =
= 73.9868
process capability index Cpk=1.5 are obtained as given
Cp 1.5 
below:
T  0.1  Control limits for R − Chart with specified Cpk value
X + A2*.
UCLx = 74.00118 + 0.2159   =
= 74.0156
Cp 1.5  USL − X   74.05 − 74.00118 
CLR= R= Dk*  = 0.775  = 0.025224
 C pk   1.5


* USL − X
  74.05 − 74.00118 
=LCLR D=
3k   0 =  0
 C pk   1.5

* USL − X
  74.05 − 74.00118 
=UCLR D=
4k   1.64  =  0.053377
C  1.5
 pk 
Control limits for the proposed X − Chart with specified
C pk = 1.5 value are obtained as given below:

CL=
x X= 74.00118
USL − X 
Figure 5. R- Chart with Specified Cp=1.5 LCLx= X − A2*k  
 C pk 
 74.05 − 74.00118 
= 74.00118 − 0.432  
 1.5
= 73.98712
USL − X 
UCLx= X + A2*k  
 C pk 
 74.05 − 74.00118 
= 74.00118 + 0.432  
 1.5
= 74.01524
Figure 6. X-Bar Chart with Specified Cp=1.5
By plotting the control limits, sample ranges and sample
By plotting the control limits, sample ranges and sample means, one may get R − Chart and X − Chart with specified
means, one may get R − Chart and X − Chart with Cpk=1.5 as given in Figure 7 and Figure 8. Fro m the R − Chart ,
specified C p = 1.5 as given in Figure 5 and Figure 6. Fro m we observe that all the plotted sample range values are
falling within the control limits. Hence we may conclude that
the R − Chart , we observe that all the plotted sample range the variations are under control. Similarly fro m the X − Chart ,
 110 J. Subramani et al.: Control Charts for Variables with Specified Process Capability Indices

we observe that all the plotted sample means are falling ACKNOWLEDGEMENTS
inside the control limits. Hence we may conclude that the
averages are also under the statistical control. Fu rther we The authors are thankful to the editor and the referees for
conclude that the process capability index Cpk for the given their useful co mments, which have helped to imp rove the
process is at least 1.5 and satisfies the customer’s presentation of the paper.
requirements.

REFERENCES
[1] Ashok Sarkar and Surajit Pal, “Process Control and
Evaluation in the Presence of Systematic Assignable Cause”,
Quality Engineering, Vol.10, pp.383-388, 1997
[2] Balamurali, S, “Bootstrap Confidence Limits for Short Run
Capability Indices”, Quality Engineering, Vol.15, pp.643-648,
2003

[3] Balamurali, S., and Kalyanasundaram, M “BootstrapLower

Figure 7. R-Chart with Specified Cpk=1.5 Confidence limits for the Process Capability Indices Cp, Cpk
and Cpm ”, International Journal of Quality and Reliability
M anagement, Vol. 19, No.8/9, pp.1088-1097, 2002
[4] Bissell, A. F. “How Reliable is Your Capability Index?”,
Applied Statistics, Vol.39, pp.331-340, 1990

[5] Chou, Y., Owen, D.B., and Borrego, .A “Lower Confidence

Limits on Process Capability Indices”, Journal of Quality
Technology, Vol.22, pp.223-229, 1990

[6] Franklin, L.A., and Wasserman, G.S. “Bootstrap Confidence

Interval Estimates of Cpk: An Introduction”,
Communications in Statistics - Simulation and Computation,
Vol.20, pp.231-242, 1991
Figure 8. X-Bar Chart with Specified Cpk=1.5 [7] Franklin, L.A., and Wasserman, G.S. “Bootstrap Lower
Confidence Limits for Process Capability Indices”, Journal of
Quality Technology , Vol.24, No.4, pp.196-210, 1992
4. Conclusions
[8] Kane, V. E., “ Process Capability Indices”, Journal of
In this paper, we have proposed a control chart which is Quality Technology, Vol.18, pp.41-52, Corrigenda pp. 265,
based on the process capability indices Cp and Cpk for on line 1986
process control, which have the benefit of the usual control
[9] Kotz, S., and Johnson, N.L. “Process Capability Indices – A
charts and the process capability ind ices. That is, the
Review, 1992-2000”, Journal of Quality Technology, Vol.34,
proposed control charts combines the two stage controlling No.1, pp.2-19, 2002
mechanis m namely, Control Charts and the process
capability indices into a single stage controlling mechanism [10] M ontgomery, D.C, Introduction to Statistical Quality Control,
to mon itor the process online and to assess the suitability of Fourth Edition, John Wily, New York, 2001
the manufacturing process. It has been shown that the [11] Spring, F.A, “Assessing Process Capability in the Presence of
relative performance of the p roposed control charts for Systematic Assignable Cause”, Journal of Quality
variables can be assessed with that of usual variable control Technology, Vol.23, pp.125-132, 1991.
charts. It has also been shown that the proposed control chart
[12] Subramani, J, “Application of Systematic Sampling in
is simple to apply and does not warrant any tedious
Process Control, Statistics and Applications,” Journal of
computations both for control charts and for computing Society of Statistics, Computer and Applications, (New
process capability indices. Moreover, we have also presented Series), 1, 7-17, 2004
tables for the control charts constants for co mputing the
control limits of the capability based control charts. The [13] Subramani, J, “Process Control in the Presence of Linear
Trend”, M odel Assisted Statistical Applications, 5, 272-281,
proposed method has also been illustrated with a numerical
2010
example. This study can also be extended to develop control
charts based on other process capability indices. [14] Subramani, J, Balamurali S, “On Line Computation of
Process Capability Indices”, International Journal of Statistics
and Applications, Vol2, no.5, 2012, To appear.

View publication stats