You are on page 1of 20

Coal Preparation, 26: 181–199, 2006

Copyright Q Taylor & Francis Group, LLC


ISSN: 0734-9343 print=1545-5831 online
DOI: 10.1080/07349340601104180

COAL QUALITY CONTROL WITH CONTROL


CHARTS

S. ELEVLI
Industrial Engineering Department, Dumlupinar
University, Kutahya, Turkey

Coal properties are inherently variable even within a single coal


seam due to reasons such as coalification history and mining
method. Since nonconsistent coal characteristics negatively affect
the efficiency and production cost of power plants as well as the
profitability of the coal producer, special attention should be paid
to control the variability of coal quality. In this paper, variations in
coal properties delivered to the Tuncbilek Power Plant have been
investigated by means of control charts. The control charts have
mostly indicated that the variations in the properties of coal are sig-
nificantly high. The considerable amount of off-design coal relative
to coal contract specifications has been determined. Finally, the
revenue of the coal producer has been examined by taking into
account the blending operation; the aim of which is to minimize
the variance of coal properties.

Keywords: Blending; Control charts; Lignite; Process capability

INTRODUCTION
The energy efficiency of a power plant is defined as the electrical energy
leaving the plant in a given time interval divided by the fuel energy enter-
ing the plant at the same time interval. Two of the most important factors
that influence the efficiency of power plants are the physical and chemical

Received September 5, 2005; accepted October 6, 2006.


Address correspondence to S. Elevli, Dumlupinar University, Kutahya, 43270, Turkey.
E-mail: selevli@dumlupinar.edu.tr
182 S. ELEVLI

properties of fuel used. These properties should coincide with the design
parameters of the power plant since fuel with off-design characteristics
causes a considerable increase in production costs besides the decrease
in efficiency.
Coal fired power plants make contracts with coal producers to
ensure a coal supply with specific characteristics that are mainly calorific
value, moisture, ash and sulfur content. In these contracts, base values
and limits for these characteristics are described by considering the
design parameters of the power plant. Generally, a price determination
equation on the basis of calorific value is given and a base coal price
is set for a base calorific value. As power plants demand coal with hom-
ogenous characteristics, these contracts occasionally determine the ceil-
ing calorific value. No extra payment is demanded for the coal of which
calorific value is above the ceiling calorific value. Therefore, the coal
having a calorific value below the lower specification limit or above
the ceiling calorific value would have a negative effect on the profitability
of coal producers. That is, both coal producers and power plants do not
desire coal that is out of specification.
Therefore, it is very important to monitor the variations in the char-
acteristics of the coal and to analyze this variability on the base of the
specifications. Currently, Statistical Process Control (SPC) techniques
have been recognized by the mining industry and utilized for this pur-
pose. With the help of Control Charts (CC), which is a tool of SPC, it
can be determined whether coal and other minerals are within the
acceptable limits. Some applications of CC in mining are given by
Ankara and Bilir [1], Aykul et al. [2], and Vapur et al. [3].
In this paper, variations in the characteristics of coal delivered to the
Tuncbilek Power Plant in Turkey have been investigated using control
charts. The variability of calorific value relative to contract specifications
has then been determined. Afterwards, the variation in the revenue of the
coal producer resulting from the premium=penalty application has been
analyzed. Finally, the revenue has been reanalyzed by considering the
blending of coal above the ceiling calorific value and below the lower
specification limit.

Control Charts and Process Capability


The concept of variation states that no two items will be perfectly ident-
ical. Variation is a fact of nature and a fact of industrial life. For example,
COAL QUALITY CONTROL WITH CONTROL CHARTS 183

Figure 1. Process control [7].

there is no such thing as two identical parts on a production line.


The difference between two parts may be very small but they are differ-
ent. The variability of a process is unavoidable and considered to be due
to either assignable causes or random causes [4–5]. An assignable cause
is an influence considered to be significant, unusual, and capable of
being removed from the process. Such causes may be due to operator
errors, defective raw materials, or improperly adjusted machines. A pro-
cess that is operating in the presence of assignable causes is said to be
out of statistical control [6]. Random causes, which are an inherent part
of the process, are smaller and uncontrollable influences that cannot be
removed from the process without fundamental changes in the process
itself. A process that is operating in the presence of random causes is
said to be in statistical control [6]. Figure 1 displays in-control and
out-of-control situations.
The main objective of Statistical Process Control (SPC) is to detect
and to eliminate assignable causes of process variation. SPC is a systema-
tic method for analyzing process data and defined as the application of
statistical methods to the measurement and analysis of variation in any
process [5]. The ultimate goal is continuous process improvement. The
control chart is an effective SPC tool and used to detect assignable
causes of variation in a process.
Figure 2 illustrates the general structure of a control chart. The chart
contains a center line that represents the average value of the quality
characteristic corresponding to the in-control state. Two other horizontal
184 S. ELEVLI

Figure 2. A typical control chart.

lines, called the Upper Control Limit (UCL) and the Lower Control
Limit (LCL), are also shown on the chart. These limits are chosen so that
there is a high probability that sample values will fall randomly between
these limits if the process is in control. The data from a controlled pro-
cess should have the following characteristics when plotted on a control
chart [8]:

. Most of the plotted points occur near the centerline;


. A few of the points occur near the control limits;
. Only an occasional rare point occurs beyond the control limits;
. The plotted points occur in a random manner with no clustering,
trending, or other departure from a random distribution.

In order to control a production process, it is necessary to identify and to


quantify characteristics that represent the quality of the product. Such
quality characteristics can be either continuous or discrete variables.
The type of control chart used to monitor a process varies depending
on whether the variable of interest is continuous or discrete.
Control charts for continuous variables such as weight, volume, vis-
cosity, and grade are called variables control charts. The X chart is the
COAL QUALITY CONTROL WITH CONTROL CHARTS 185

Table 1. Control chart limits

Chart type Central line Lower control limit Upper control limit

Averages, X X X  A2 R X þ A2 R
Ranges, R R D3 R qffiffiffiffiffiffiffiffiffiffiffi D4 R qffiffiffiffiffiffiffiffiffiffiffi
Percentage nonconforming, p p p  3 pð1pÞ
p þ 3 pð1pÞ
pffiffiffi n pffiffiffi n
Number of nonconformities, c c c  3 c c þ 3 c

most widely used chart for controlling central tendency, while charts
based on either the sample range (R Chart) or the sample standard devi-
ation (S Chart) are used to control process variability. In the case of dis-
crete variables, each unit of product is judged as either conforming or
nonconforming on the basis of whether or not it possesses certain attri-
butes. Control charts for such variables are called attributes control
charts.
Control limit formulas for the most widely used charts are given
in Table 1. These formulas are based on 3 standard deviations (i.e.,
3r) and use a central line equal to the average of the data used in calcu-
lating the control limits. A2, D3, and D4 are factors used in computation
of upper and lower control limits depending on the number of items in the
sample. These factors can be easily found in different sources [4–6].
The ability of the process to comply with externally imposed limits,
which are known as specification limits, is called capability. There is no
connection or relationship between the control limits and specification
limits. The control limits are driven by natural variability of the process,
that is, by the natural tolerance limits of the process [6]. It is customary
to define the upper and lower natural tolerance limits as 3r above and
below the process mean (m). On the other hand, the specification limits
are determined externally by management, design engineers, or custo-
mers. If the natural tolerance limits are inside the specification limits,
then almost all (99.73%) of the individual process measurements are
being produced within the specification limits. In this case, it is said that
the process is capable of meeting specifications. If the natural tolerance
limits are wider than the specification limits, some individual process
measurements are outside of the specification limits. That means the
process is not capable.
Figure 3 illustrates four possible situations that can arise when
the observed=natural variability of a process is compared to design
186 S. ELEVLI

Figure 3. Process capability versus design specifications [9].

specifications. In Figure 3(a), the range of process variation is wider than


the design specification. It is clear that the process is not able to meet
specifications. In Figure 3(b), the process is able to produce according
to specification. However, it will require close monitoring to assure that
it remains in this position. In Figure 3(c), the observed variation is nar-
rower than the specifications. This is the ideal situation from a quality
control point of view, since little inspection or control is necessary.
Finally, in Figure 3(d), the observed variation is the same as the design
specification, but the process is off center. Thus, some nonconforming
product can be expected.
A process in statistical control does not mean that the process is also
capable. Figure 4 displays different capability cases. The left side of
Figure 4 shows that the process is in control but not capable. In the right
side, which satisfies customer expectations, process is both in control
and capable.

CASE STUDY
Content of Coal Agreement
The contract between the Electricity Generation Company, Inc., and the
Turkish Coal Enterprise for the year of 2005, which describes the speci-
fication of coal delivered to the Tuncbilek Power Plant from the Western
COAL QUALITY CONTROL WITH CONTROL CHARTS 187

Figure 4. Process capability [7].

Coal Company, is given in Table 2. The price of coal is determined by:


Actual Calorific Value
Coal Price ¼ Price of Base Calorific Value
Base Calorific Value
ð1Þ
The Base Calorific Value, which is 2350 kcal=kg according to Table 2, is
the base value for price determination. The price of the base calorific
value has been set at $35.29 per ton and is fixed for the calorific values
in specification limits. If the calorific value of coal is between
2550 kcal=kg and 2800 kcal=kg or 1600 kcal=kg and 2350 kcal=kg, the
price will be estimated according to Equation (1). If the calorific value
is over 2800 kcal=kg, which is a ceiling calorific value, then calorific value
will be assumed to be as 2800 kcal=kg and Equation (1) will be used. For
the calorific value below 1600 kcal=kg, the coal price will be determined

Table 2. Coal specifications of Tuncbilek Power Plant

Base calorific value (kcal=kg) 2350


Specification limits of calorific value (kcal=kg) 2350  200
Max calorific value (ceiling value) (kcal=kg) 2800
Min calorific value (kcal=kg) 1600
Specification limits of ash content (%) 42(þ %10)
Specification limits of moisture content (%) 24(þ %10)
188 S. ELEVLI

by multiplying the base price by 0.2. This contract structure means that
no coal will be refused. There is no penalty or premium application for
moisture and ash content. In light of above, the coal delivered to the
power plant can be classified as follows:

. Coal having a calorific value more than 2800 kcal=kg;


. Coal having a calorific value between 2550 and 2800 kcal=kg;
. Coal having a calorific value between 2150 and 2550 kcal=kg;
. Coal having a calorific value between 1600 and 2150 kcal=kg;
. Coal having a calorific value lower than 1600 kcal=kg.

Control Chart Formation


In order to analyze the variation in calorific value, moisture content, and
ash content of the coal delivered to the Tuncbilek Power Plant with con-
trol charts, data from 23 months of operation between January 2003 and
June 2005 have been gathered. The total number of data points is 69 since
the calorific value, moisture content, and ash content of coal were
recorded as average values every 10 days. The data arranged as m ¼ 23
(number of sample) and n ¼ 3 (size of sample) are given in the Table 3.
Average and range values have been estimated to determine control limits
for each month and are given in the last line of the tables. Related con-
stants depending on the size of the sample (n) are A2 ¼ 1.023, D3 ¼ 0,
and D4 ¼ 2.574. Control limits for X and R charts for calorific value have
been estimated as an example by using equations in Table 1 as follows:
8
>
> Lower Control Limit ¼ X  A2 R
>
>
>
> ¼ 2476:30  1:023  397:52 ¼ 2069:64
>
<
X Chart Center Line ¼ X ¼ 2476:30
>
>
>
>
>
> Upper Control Limit ¼ X þ A2 R
>
:
¼ 2476:30 þ 1:023  397:52 ¼ 2882:96

8
<Lower Control Limit ¼ D3 R ¼ 0  397:52 ¼ 0
>
R Chart Center Line ¼ R ¼ 397:52
>
:
Upper Control Limit ¼ D4 R ¼ 2:574  397:52 ¼ 1023:22
Modern computer technology has made it easy to implement control
charts in any type of process. Consequently, SPSS 10.0 for Windows
COAL QUALITY CONTROL WITH CONTROL CHARTS 189

Table 3. Case study data for (a) calorific value, (b) moisture, (c) ash

Number of sample Subgroup 1 Subgroup 2 Subgroup 3 X R

(a) Calorific Value


1 2698 3581 3020 3099.67 883
2 2471 2350 2520 2447.00 170
3 2283 2387 2679 2449.67 396
4 2598 3482 3051 3043.67 884
5 2943 2456 2361 2586.67 582
6 2236 2118 2500 2284.67 382
7 2427 2162 2184 2257.67 265
8 2697 3003 2428 2709.33 575
9 2206 2591 2565 2454.00 385
10 2540 2249 2430 2406.33 291
11 2714 2095 2280 2363.00 619
12 2352 2449 2584 2461.67 232
13 2296 2793 2827 2638.67 531
14 2279 2362 2204 2281.67 158
15 2411 2367 2393 2390.33 44
16 2151 2366 2097 2204.67 269
17 2130 2189 2557 2292.00 427
18 2330 2090 2054 2158.00 276
19 2040 2301 2591 2310.67 551
20 2444 2489 2677 2536.67 233
21 2719 2424 2736 2626.33 312
22 2506 2719 2343 2522.67 376
23 2506 2543 2241 2430.00 302
X ¼ 2476:3 R ¼ 397:52
(b) Moisture Content
1 17.2 18.41 18.24 17.95 1.21
2 16.27 15.58 15.58 15.81 0.69
3 16.11 15.96 16.58 16.22 0.62
4 14.68 18.68 15.7 16.35 4
5 15.06 14.55 13.7 14.44 1.36
6 14.68 14.87 15.07 14.87 0.39
7 14.6 15.47 16.67 15.58 2.07
8 15.43 16.56 16.35 16.11 1.13
9 15.84 16.17 17.51 16.51 1.67
10 14.86 14.93 15.54 15.11 0.68
11 19.42 15.6 15.34 16.79 4.08
12 15.76 15.27 16.5 15.84 1.23
13 17.62 16.09 15.2 16.30 2.42
14 15.87 15.65 15.23 15.58 0.64

(Continued)
190 S. ELEVLI

Table 3. Continued

Number of sample Subgroup 1 Subgroup 2 Subgroup 3 X R

15 15 15.02 14.71 14.91 0.31


16 15.23 15.65 16.21 15.70 0.98
17 15.6 14.78 14.37 14.92 1.23
18 17.37 18.83 16.75 17.65 2.08
19 17.54 15.53 15.25 16.11 2.29
20 20.01 18.18 14.43 17.54 5.58
21 14.51 14.92 14.11 14.51 0.81
22 14.18 13.61 13.42 13.74 0.76
23 13.24 12.34 12.07 12.55 1.17
X ¼ 15:70 R ¼ 1:63
(c) Ash Content
1 44.6 34.64 40.74 39.99 9.96
2 48.9 51.09 48.75 49.58 2.34
3 51.05 50.25 46.15 49.15 4.9
4 48.59 35.04 41.1 41.58 13.55
5 44.53 50.07 51.04 48.55 6.51
6 52.89 53.76 49.51 52.05 4.25
7 50.46 53.9 52.76 52.37 3.44
8 47.08 42.46 50.05 46.53 7.59
9 52.61 47.59 47.18 49.13 5.43
10 49.27 52.4 50.05 50.57 3.13
11 44.29 54.03 53.01 50.44 9.74
12 50.33 49.67 47.32 49.11 3.01
13 50.54 45.06 46.09 47.23 5.48
14 51.52 50.81 52.06 51.46 1.25
15 50.21 51.04 50.75 50.67 0.83
16 52.78 50.09 52.96 51.94 2.87
17 53.25 52.43 48.51 51.40 4.74
18 49.94 51.42 53.45 51.60 3.51
19 53.58 51.42 48.32 51.11 5.26
20 46.8 48.07 47.72 47.53 1.27
21 46.75 49.34 46.47 47.52 2.87
22 49.13 46.81 51.38 49.11 4.57
23 50.51 50.11 53.48 51.37 3.37
X ¼ 49:13 R ¼ 4:78

has been used in this study for drawing the control charts. The X and R
charts for calorific value, moisture content, and ash content are given
in Figures 5–10, respectively. The followings observations can be drawn
from the figures.
COAL QUALITY CONTROL WITH CONTROL CHARTS 191

Figure 5. X Chart for calorific value.

. Figure 5 indicates the process is out of statistical control since there


are two points outside the control limits and the distribution of the
points around the center line is irregular. In addition, six points out-
side the Upper Specification Limit is an indication that the process
is not capable.
. Figure 6 indicates that all the ranges of calorific value lie in the in-con-
trol area. However, due to some points being very close to control
limits, an irregular pattern is observed.
. Figure 7 indicates that there are five points outside the control limits,
which indicates that there is considerable variation in the process
mean. On the other hand, the fact that all values are considerably
lower than specification limit (24%) is an indication that the process
is able to meet specifications.
. Figure 8 indicates that the fluctuations in moisture are excessive and
one point occurs beyond the control limits.
. Figure 9 shows the process is out of control since two points are out-
side the control limits and six consecutive points occur on one side of
192 S. ELEVLI

Figure 6. R Chart for calorific value.

the center line. Additionally, the fact that four points are outside the
specification limits is another problem relating to the process capa-
bility.
. Figure 10 indicates that the process is not stable due to the excessive
fluctuations in ash content and one point being outside the control
limit.

Estimating Process Capability for Calorific Value


The control chart data may be used to describe the capability of the
process. From the X chart, we may estimate the mean calorific value
as X ¼ 2476:3 kcal=kg. The process standard deviation may be esti-
mated as r^ ¼ ðR=d2 Þ ¼ ð397:52=1:693Þ ¼ 234:8. Assuming that the cal-
orific value is a normally distributed random variable with mean 2476.3
and standard deviation 234.8, natural tolerance limits can be estimated
as follows:
COAL QUALITY CONTROL WITH CONTROL CHARTS 193

Figure 7. X Chart for moisture.

Upper Natural Tolerance LimitðUNTLÞ ¼ X þ 3:^


r ¼ 2476:3 þ 3  234:8
¼ 3180:70

Lower Natural Tolerance LimitðLNTLÞ ¼ X  3:^


r ¼ 2476:3  3  234:8
¼ 1771:90

Since the specification limits are 2350  200 kcal=kg, the natural
tolerance limits of the process are outside the lower (LSL) and upper
specification limits (USL). That is, the variation in the process is high
and the general mean is bigger than the nominal value (Figure 11).
Consequently, some out-specification coals are being fed to the power
plants.
By considering coal classification, coal rates for each classification
have been estimated as given below:
194 S. ELEVLI

Figure 8. R Chart for moisture.

. The portion of coal having a calorific value more than 2800 kcal=kg.
   
2800  l 2800  2476:3
P fx > 2800g ¼ P Z > ¼U
r 234:8
¼ Uð1:38Þ ¼ 0:0838

. The portion of coal having a calorific value more than 2550 kcal=kg.
 
USL  l
P fx > USLg ¼ P fx > 2550g ¼ P Z >
r
 
2550  2476:3
¼U ¼ Uð0:31Þ ¼ 0:3783
234:8

. The portion of coal having a calorific value smaller than 2550 kcal=kg.
 
LSL  l
P fx < LSLg ¼ P fx < 2150g ¼ P Z <
r
 
2150  2476:3
¼U ¼ Uð1:39Þ ¼ 0:0823
234:8
COAL QUALITY CONTROL WITH CONTROL CHARTS 195

Figure 9. X Chart for ash.

. The portion of coal having a calorific value smaller than 1600 kcal=kg.

   
1600  l 1600  2476:3
P fx < 1600g ¼ P Z < ¼U
r 234:8
¼ Uð3:73Þ ¼ 0

Depending on the above rates, the yearly revenue (assuming the coal
supply is approximately 1.3 million tons per year) can be estimated as
given in Table 4. As can be seen, the total revenue is $47,718,902. If
calorific value of all coals were between 2350  200 kcal=kg, the total
revenue would be $45,877,000 per year (1; 300; 000  35:29). That is,
although the process is not in statistical control and not capable, the
income of the coal producer is 4.01% high. Despite the fact that the
contracts include a penalty=premium application for different calorific
values, power plants prefer to have a homogenous coal due to efficiency
196 S. ELEVLI

Figure 10. R Chart for ash.

and cost considerations. Therefore, blending is a necessary operation to


homogenize the coal before feeding to power plants.
By blending the coal having calorific value above 2800 kcal=kg with
the coal having calorific value below the 2150 kcal=kg, both the
efficiency of the power plant and the income of coal producer can be
increased. Assuming that the average calorific value of coals having cal-
orific value below 2150 kcal=kg is 1875 kcal=kg and the average calorific

Figure 11. Natural tolerance limits of calorific value.


COAL QUALITY CONTROL WITH CONTROL CHARTS 197

Table 4. Revenue estimation

Coal type Ratio (%) Quantity (ton) Price ($=ton) Revenue ($)

> 2800 kcal=kg 8.38 108,940 42:05 4,580,927


2800 
2350  35:29

2550–2800 kcal=kg 29.45 382,850 40:17 15,379,084


2675 
(37.83  8.38) 2350  35:29

2150–2550 kcal=kg 53:94 701,220 35.29 24,746,053


½100  ð37:83 þ 8:23Þ
1600–2150 kcal=kg 8.23 106,990 28:16 3,012,838
1875 
2350  35:29

TOTAL 100 1,300,000 – 47,718,902

 Actual Calorific Value


Coal Price ¼  Price of Base Calorific Value:
Base Calorific Value

value of coals having calorific value above 2800 kcal=kg is 2925 kcal=kg,
the mean calorific value of blended coal can be estimated as follows:
ð1; 875  106; 990Þ þ ð2925  108; 940Þ
Mean Calorific Value ¼
106; 990 þ 108; 940 ð2Þ
¼ 2405 kcal=kg

Since the mean calorific value is between 2350  200 kcal=kg, the revenue
estimation will be based on the price of the base calorific value. Thus, the
revenue after the blending operation will be $7,620,169 ð35:29
½108; 940 þ 106; 990Þ. Namely, the revenue of the coal producer will be
$26,404 per year more than the revenue before the blending operation.

CONCLUSIONS
In today’s competitive world, there is a continuing need to produce elec-
tricity more effectively. In this scope, the variability of the coal supply
from coal producers has been one of the main concerns in Turkey. As coal
producers should supply in-specification coals based on design para-
meters of the plant for improving or maintaining the efficiency of the
power plants, the variation of some coal characteristics such as calorific
value, moisture content, and ash content should be dealt with carefully.
In this study, control charts and process capability analysis of stat-
istical quality control methods have been utilized to reach at the follow-
ing conclusions:
198 S. ELEVLI

. Calorific values, ash contents, and moisture contents of coals sold to


the Tuncbilek Power Plant vary within short time periods. Addition-
ally, all of the three variables are not statistically stable.
. According to calorific values, only 53.94% of the coal is within the
specification limits. That is, the process is not able to meet specifica-
tions.
. Even though the rate of coal outside the specification limit is very high,
the income of the coal producer for the year is 4.01% higher than the
income under normal conditions due to the premium paid for coals
with higher calorific values.
. About 8.38% of coal having a calorific value over 2800 kcal=kg is sold
with a lower price than its real value due to the limitations in the con-
tract.
. When coal having a calorific value over 2800 kcal=kg and lower than
2150 kcal=kg is blended, the income of coal producer can be increased
by $26,404.
. It is strongly suggested that the coal should be blended to reduce varia-
bility before selling to the power plant so that the efficiency of the
power plant and the income of the coal producer can be increased.

REFERENCES
1. H. Ankara and K. Bilir, Kriblaj tesisinde kalite denetimi, madencilikte
bilgisayar, Uygulamalari Sempozyumu (1995).
2. H. Aykul, H. Akçakoca, G. Ediz, and M. Taksuk, Garp linyitleri İşletmesi
termik santral kömürleri İçin İstatistiksel süreç kontrol analizi, Türkiye 19,
Uluslararasi Madencilik Kongresi (2005).
3. H. Vapur, O. Bayat, and F. Akyol, Eti gümüş A. Ş. Liç prosesinde İstatistiksel
proses kontrolü uygulamasi, Türkiye 19, Uluslararasi Madencilik Kongresi
(2005).
4. H. M. Wadsworth, Handbook of Statistical Methods for Engineers and Scientists,
McGraw-Hill Inc., New York, 1990.
5. J. M. Juran and F. M. Gryna, Quality Planning and Analysis, McGraw-Hill
Book Company, New York, 1993.
6. D. C. Montgomery, Introduction to Statistical Quality Control, John Wiley &
Sons Inc., New York, 1991.
7. Indiana Department of Transportation (INDOT), Certified Aggregate Tech-
nician Manual: Statistical Quality Control for Aggregate Processing, www.in.
gov/dot/div/testing/manuals/aggregate/chapter_06.pdf, 2003.
COAL QUALITY CONTROL WITH CONTROL CHARTS 199

8. H. M. Wadsworth, Statistical Process Control, In Juran’s Quality Handbook


(J. M. Juran and A. B. Godfrey, eds.), McGraw-Hill International Edition,
New York, 2000.
9. J. R. Evans, Applied Production and Operations Management, 4th ed., West
Publishing Company, Minneapolis–St. Paul 1993.

You might also like