Journal's Title, Vol. , 2018, no.

HIKARI Ltd, www.m-hikari.com

http://dx.doi.org/10.12988/

Control Charts

Barrios Rodriguez Alvaro Junior

ajuniorbarrios@est.uniatlantico.edu.co

De Alba Diaz Herlis Milagro

hdealba@est.uniatlantico.edu.co

Ojeda Ortiz Luis Miguel

lmojeda@est.uniatlantico.edu.co

Vargas Polo Nicolle Valentina

nvvargas@est.uniatlantico.edu.co

Copyright © 2018 Ojeda Ortiz Luis Miguel. Barrios Rodriguez Alvaro Junior. De Alba Diaz
Herlis Milagro. Vargas Polo Nicolle Valentina. This article is distributed under the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.

Abstract
Control techniques are intended to evaluate the statistical processes. For the
following article the lengths of thirty drywall anchors, which were divided into
three subgroups. The main objective is to evaluate the behavior of the process
using Shewhart's univariate control charts for means and ranges, as well as their
respecting control limits to analyze the process. Subsequently, it is determined
whether the process meets the specifications established in the design by means of
the capability index with target value T.
Keywords: control charts, capacity index, control limits, specification limits.

1 Introduction

During World War I the manufacturing system became more complex, involving
more workers, and reporting only one supervisor for the entire production; later
with Taylor, different changes in the manufacturing process were reflected, since
the first quality control inspectors finally appeared. To avoid complaints from
customers, the products were reviewed at the end of the process, identifying
defective ones by a quality control department, due to this problem and with the
aim of reducing the cost of non-quality the statistical control of the process was
developed and applied. This control includes some methods of which the control
chart is highlighted and it is used to monitor the processes when abnormal
variations occur where the means or ranges are outside the control limits, which
gives us the impression that action must be taken to remove this abnormal
variation.
In order to see more clearly how the control chart works and all the monitoring of
the processes or products; in this case, we took as a sample 30 drywall anchors,
which their lengths were measured, and to analyze the behavior of this data we
took as an aid the Shewhart control chart for the range and the averages, as well as
the univariate capacity index.

2 Control charts

A characteristic that is measured on a numerical scale is called a variable, whether

it be dimensions, volume, time, etc. Control charts are widely used to monitor the
mean and variability of variables, in order to avoid or minimize out-of-
specification products and, if any, to look for the causes of those products.
In practice the control limits are estimated from 20 or 25 preliminary samples or
subgroups, the subgroup size is normally 4, 5 or 6. If m subgroups are had the
large average is calculated as follows:
m

X i
X  i 1

m (1)
This variable represents the centerline of the chart of means.
To estimate the  of the process, the ranges of the subgroups can be used, for each
of the subgroups the range is calculated as:
(2) R = xmax – xmin
If R1,...,Rm , are the ranges of the different subgroups, the average range is:
m

R i
R i 1

m (3)

This variable represents the centerline of the range card.

An estimator of  is R / d2, where d2 is tabulated for different values of n, in this
way, if R is the average range of first simples, using:
R

d 2 (4)

The control limits for a means chart are:

(5) UCL = X + A2 R ; (upper control limits)
(6) LCL = X - A2 R ; (Lower control limits)
(7) CL= X ; (center line)

The value of A2 is found in a table of constants.

In the case of ranges, the centre line is R . The estimator for R can be found as
follows:
R
 R  d3 (8)
d2
Therefore, the control limits for a range chart are given as follows:
R d3
d3
UCL = R + 3  R = R + 3 d 2 = R [ 1+ 3 d 2 ]
UCL= D4 R (9); (upper control limits)
CL= R (10); (center line)
R d3
d3
LCL = R - 3  R = R - 3 d 2 = R [ 1- 3 d 2 ]
LCL= D3 R (11); (lower control limits)

The constants A2, d2 D3 and D4 are found in tables. [1]

CASE OF n CONSTANT
For the S chart you have: For the X chart you have:
UCLs = c4 + 3  1  c4 = B6  (12) UCLX =  + A (15)

CLs = c4 (13) CL =  (16)

 LCLs = c4 - 3  1  c4 = B5 (14) LCLX =  - A (17)

The values for the constants are tabulated for different values of n in the constants
table.
If the population standard deviation is not known, it can be estimated using
several m samples with historical data, where the standard deviation is obtained
for each of them and averaged.

1 m
S   Si (18)
m i 1
__
S
 (19)
c4
As the S /c4 statistic is an unbiased estimator of  , the parameters of the chart
will be as follows
S
UCLs = S  3 1  c42 = B4 S (20)
c4
CLs = S (21)
S
LCLs = S  3 1  c42 = B3 S (22)
c4
In the X chart, the limits are:
S
UCLx = X + 3 = X + A3 S (23)
c4 n

CLx = X (24)
S
LCLx = X - 3 = X - A3 S (25)
c4 n
The values for the constants are tabulated for different values of n in the constants
table. [2]
3. Capability index with target value t

The capability index compares the amplitude of variation allowed by the

specifications between the amplitude of variation between the natural tolerance
limits of the process. [3]
𝑈𝑆𝐿−𝐿𝑆𝐿
𝐶𝑝𝑚 = (26),
6(√(𝜎2 +(𝜇−𝑇)2 )

𝑈𝑆𝐿 = 𝜇 + 𝐾𝜎 (27) 𝑦 𝐿𝑆𝐿 = 𝜇 − 𝐾𝜎 (28); 𝑤𝑖𝑡ℎ 𝑇 𝑎𝑠 𝑡𝑎𝑟𝑔𝑒𝑡 𝑣𝑎𝑙𝑢𝑒.

The Cpmk index takes into account the centering of the process with respect to the
specifications, in this case it is called Cpmk and is evaluated by taking the
minimum between the Cpu and Cpl.
𝑈𝑆𝐿−𝜇 𝜇−𝐿𝑆𝐿
𝐶𝑝𝑘𝑚 = min[𝐶𝑝𝑢 = , 𝐶𝑝𝑙 = (29) [3]
3(√(𝜎2 +(𝜇−𝑇)2 ) 3(√(𝜎2 +(𝜇−𝑇)2 )

4 Results

The 30 stroke length measurement is taken as an experience to monitor the

process using the range and mean control chart.

X1 X2 X3 ̅
𝑿 R
1 24,15 23,90 24,15 24,07 0,25
2 24,05 24,40 24,20 24,22 0,35
3 24,25 24,90 24,15 24,43 0,75
4 25,00 24,20 24,10 24,43 0,90
5 24,85 24,15 24,20 24,07 1,70
6 24,15 24,70 24,20 24,35 0,55
7 24,25 24,20 24,70 24,38 0,50
8 24,80 24,60 24,25 24,55 0,55
9 24,50 24,65 24,45 24,53 0,20
10 24,30 24,45 24,15 24,30 0,30
Table 1. drywall anchors length.
From the data of table 3, and the formula (3), is had that 𝑅̅ =0,51; with formula
(1), is had that 𝑋̿=24,37.
 Histograma

15

12

frecuencia
9

0
23 23,5 24 24,5 25 25,5
Figure 1.Col_1
histogram

With this information a range chart can be constructed, where 𝑅̅ = 0.51 the center
line, and by means of the formulas (9) y (11) it is determined that LCS= 1.30, and
LCI=0.00, these values contribute to the control chart for the range.
Gráfico de Rangos para Col_1-Col_3

1,5

1,30
1,2

0,9
Rango

0,6
0,51

0,3

0 0,00
0 2 4 6 8 10
Figure 2. Subgrupo
Control chart for R.

Ideally, for a range control chart the data should be below the central limit without
going outside the limits, in this case, the data obstructs between the limits; some
with an upward and downward trend, as well, a decrease in variability is observed.

To determine the control limits for the mean of the data, it is necessary to use
formulas (5), (6), (7). Determines UCL=24.88, LCL=23.85, CL=24.37
 Gráfico X-bar para Col_1-Col_3

25
24,88

24,6
24,37
X-bar 24,2

23,8 23,85

23,4

23
0 2 4 6 8 10
Subgrupo
Figure 3. Control chart for the means.

For the mean control chart it is observed that the data do not go beyond the
control limits, although some data suffer trends, the data are taken as a process
and not as individual points; therefore the mean control chart proves the
hypothesis that the mean of the process is in control.
To find the capacity index Cpm and the specific capacity index Cpkm, it is
necessary to determine the limits of specifications; for this the values of
𝑘, 𝑑2 , 𝜇 𝑎𝑛𝑑 𝜎 are needed. The value of 𝑑2 is found in the constants table; taking
as n=3, the value of the constant k is determined by a table of factors for normal
bilateral tolerance limits taking 95% confidence and 90% power, the value of μ
was calculated previously; taking that 𝜇 = 𝑥̿ , finally the value of σ is determined
by means of the formula (4).
𝑘 = 2.140, 𝑑2 = 1.693, 𝜇 = 24.37 𝑦 𝜎 = 0.30

Then, USL= 25, LSL=24.

there is a probability of compliance with the specifications, taking the normal

distribution and specification limits; P=87.34%.

Finally, the capacity (26) and location (29) indices, with T=25, are given by:

𝐶𝑝𝑚 = 0.24 𝐶𝑝𝑘𝑚 = −0.30

 Capacidad de Proceso para Col_1-Col_3
LIE = 24,0; LSE = 25,0

15 Normal
Media=24,3667
Desv. Est.=0,284766
12
Cp = 0,56
Pp = 0,59

frecuencia
9 Cpk = 0,41
Ppk = 0,43
DPM a Corto Plazo = 126357,37
6 DPM a Largo Plazo = 112014,13

0
23 23,5 24 24,5 25 25,5
Col_1-Col_3

Figurae 4. Process capacity

With a Cpm= 0.24 it implies that the process requires very serious modifications,
on the other hand the decentralization of the process must be corrected.

5 Conclusions
The main objective of statistical control is to detect the presence of assignable
causes and anomalies in the process in order to then make an investigation and
begin to correct these problems. By distinguishing these causes, they indicate
when these problems should be corrected and when departments or different
levels of the organization should be involved in order to attack them.
The control charts are a useful tool to look at the tracking of all data; taking them
as a distribution and not as a set of points which give a broad view of the behavior
and monitoring process.

References
[1] Montgomery, (2004), Statistical quality control 3rd edition.

[2] Feigenbaum Armand V.(1994), Total quality control .3rd edition-CESCA.

[3] Grant Eugene I. (1966). Statistical quality control.3rd edition.Editorial

Continental S.A.

Received: August, 2018