Int. J. Manufacturing Technology and Management, Vol. 17, No. 4, 2009
Developing and implementing statistical process
control tools in a Jordanian company
R.H. Fouad* and Salman D. Al-Shobaki
Department of Industrial Engineering,
Hashemite University,
P.O. Box 330127,
Zarka 13133, Jordan
E-mail: rhfouad@hu.edu.jo
E-mail: sshobaki@hu.edu.jo
*Corresponding author
Abstract: Statistical Quality Control (SQC) is a branch of Total Quality
Management (TQM) that defines a quality philosophy and set the guiding
principles that represent the foundation for a continuously improving
organisation. Statistical Process Control (SPC) and Acceptance Sampling are
the two major parts of SQC. SPC is an effective tool that aims to get and keep
processes under control and ensure that the product is manufactured as
designed and intended. SPC is comprised of seven tools: Pareto diagram, cause
and effect diagram, check sheets, process flow diagram, scatter diagram,
histogram and control charts. A case study has been carried out to monitor real
life data in a Jordanian manufacturing company that specialises in producing
fertilisers. Pareto diagram, histograms and control charts for variables were
implemented to investigate the major causes of non-conformities and possible
remedies were proposed. The full strength of the seven statistical control tools
was implemented and other companies are encouraged to follow suite and
implement them simultaneously.
Keywords: SQC; statistical quality control; SPC; statistical process control;
TQC; total quality control; basic tools of quality; Pareto diagram; histogram
and control charts for variables.
Reference to this paper should be made as follows: Fouad, R.H. and
Al-Shobaki, S.D. (2009) ‘Development and implementing statistical process
control tools in a Jordanian company’, Int. J. Manufacturing Technology and
Management, Vol. 17, No. 4, pp.337–344.
Biographical notes: R.H. Fouad received his PhD from Bradford University of
Bradford, UK in 1991, in Industrial Technology and Production Management,
studied his MSc at Cranfield University, UK in 1988 in Industrial Engineering
and Operations Management and his BSc at the University of Baghdad, Iraq in
1983 in Mechanical Engineering. Presently he works as an Associate Professor
at the Industrial Engineering Department at the Hashemite University. His
current research interests are operations management, maintenance
management, statistical process control and lean manufacturing.
Salman D. Al-Shobaki is a BSC graduate from Yarmouk University in Jordan,
in 1994, an MSc from Brunel University in London, UK in 1995 and received
his PhD from Imperial College of Science, Technology and Medicine,
University of London, UK in 2000. Presently, he works as an Assistant
Professor at the Department of Industrial Engineering, at Faculty of
Copyright © 2009 Inderscience Enterprises Ltd.
337
338
R.H. Fouad and S.D. Al-Shobaki
Engineering, Hashemite University (HU), Zarka, Jordan. His current research
interests lie in industrial management aspects such as quality control and
management, quality and excellence models, ergonomics, time and motion
study.
1
Introduction
Statistical Process Control (SPC) is receiving increasing attention as a management tool
to observe, assess and compare important characteristics of product to a set standard. The
various procedures in quality control involve considerable use of statistical principles. It
has become clear that an effective quality control programme enhances the quality of
product being produced and increases profits (Besterfield, 2004).
SPC is comprised of seven tools, Pareto chart, Histogram, process flow diagram,
control charts, scatter diagram, check sheets and cause and effect diagram. SPC seeks to
maximise profit by the following ways: improving product quality, improving
productivity, reducing wastage, reducing defects and improving customer value.
A Jordanian manufacturing company was chosen to apply three basic statistical tools of
quality control; Pareto chart, histogram and control charts.
2
Company background
Al-Qawafel Industrial Agriculture Establishment is a Jordanian manufacturing company
that produces Fertilisers. The company has six various production lines. The quality
control department in the company includes two sections: Test Laboratories and
Research and Development (R&D). Suspension fertiliser production line is the pilot
production line was chosen to implement the SPC tools for the purpose of this paper. The
following main tests are applied to fertilisers produced in suspension fertiliser production
line: density test, temperature test, PH concentration test, CL percentage test and
sieve test.
3
Pareto’s chart
A Pareto chart is simply a frequency distribution (or Histogram) of attribute data
arranged by category (Montgomery, 2005). Pareto chart is a helpful tool for problem
analysis. Problems and their associated frequency or cost are arranged in descending
order according to their relative importance in bar chart form, the chart is a visual
method of identifying which problems are most significant (vital few), that usually
accounts as 80% of the total results and the least significant problems (useful many)
usually accounts as 20% of the total results. The graph has the advantage of providing a
visual impact of those vital few characteristics that need attention. Pareto charts also have
the advantage of limiting the tendency of people to focus on the most recent problems
rather than on the most important ones.
Figure 1 shows a constructed Pareto chart for the main tests performed on fertilisers
at Al-Qawafel Industrial Agriculture. It reveals that the temperature and the PH
Developing and implementing statistical process
339
concentration are the vital few tests and represents around 84% of the total cumulative
percentage. On the other hand, the useful many factors are the density, sieve and CL
percentage and represent around 16% of the total cumulative percentage; moreover, the
main reason of most rework is the temperature.
Figure 1
4
Pareto chart for the tests performed on fertilisers at Al-Qawafel Industrial Agriculture
Histograms
Histogram is a bar chart that shows the number of times (frequency) of each cells
occurred where each cell contains a range of measured data. It is a pictorial display of the
way the data is distributed over the various cells. Histogram analysis clarifies process
capability, conformance to specifications, shape of population and discrepancies and
gaps in data.
Figure 2 shows a Histogram for temperature tests at Al-Qawafel Industrial
Agriculture. It is noted that the measures of central tendency had an average = 24.14, the
median = 12.4 and the mode = 23.6. The measures of dispersion had a range of = 18.0
and a standard deviation = 3.55. Also, the measures of shape had a Skewness
(a3 = 1.17 > 0) indicating that the data is skewed to the right and a Kurtosis
(a4 = 4.24 > 3) indicating that the data is more packed than normal.
Figure 2
Histogram for temperature tests at Al-Qawafel Industrial Agriculture
340
5
R.H. Fouad and S.D. Al-Shobaki
Process capability and tolerances
Process spread will be referred as the Process capability and is equal to 6σ. When design
engineers establish product tolerance without regard to the spread of the process, three
situations are possible (Besterfield, 2004):
Case 1: (Most desirable case): where 6σ < [USL − LSL] [What is USL, LSL].
Case 2: (Satisfactory desirable case): where 6σ = [USL − LSL] .
Case 3: (Undesirable case): where 6σ > [USL − LSL] .
Where USL and LSL are the upper and lower specifications limits.
The process capability is the region between the two boundaries
(X-bar) ±3S = 13.49 − 34.79 . Therefore, process capability is as follows: (13.49 C and
34.79 C). The analysis of data shows that the process capability is greater than the
tolerance, therefore, an undesirable situation exists and a non-conforming product is
produced. This indicates that the company should strive to improve process capability.
Considering Figure 3, the process capability analysis indicates that the range value is
high because the factory temperature limits determined by quality department that has a
wide dispersion. The values of the central tendency measures (average, median and
mode) values are closely together, which means that the distribution is approximately
normal. Also, the process capability is greater than tolerance, which cause high number
of non-conformities as temperature reaches values that are greater than the upper
specification limits.
Figure 3
6
Process capability for controlling temperature tests (see online version for colours)
Control charts
The control chart is a graphical record of the quality of a particular quality characteristic.
In general, the control chart contains a centre line that represents the mean value for the
in-control process. Two other horizontal lines, called the Upper Control Limit (UCL) and
the Lower Control Limit (LCL) are also shown on the chart. The control chart enhances
Developing and implementing statistical process
341
the analysis of the process by showing how the process performs overtime and this will
enable decisions-making concerning future production. It is used to locate any unusual
trends and determine process centring and process variation.
In general, the sources of variation and out of control points in control charts are
classified as either Assignable causes (special cause) or Chance causes (common causes).
When only chance causes are presenting a process, the process considered to be in a state
of statistical control. When an assignable cause of variation is present the process is
classified as out of control (Montgomery, 2005). Control charts provide information for
quality improvement, to determine the process capability and for decisions concerning
product specifications.
Two types of variable control charts are used extensively when dealing with a quality
characteristic that is variable; the sample average and range control chart and the sample
average and standard deviation control chart. Mean and Range control charts are shown
in Figure 4 and Standard deviation and Range control charts are shown in Figure 5
below.
The first step is to post the preliminary data to the chart along with the control limits
and central line. This has been accomplished and is shown in Figure 4. The next step is to
adopt standard values for the centre lines or, more appropriately stated, the best estimate
of the standard values with the available data. If an analysis of the preliminary data
shows good control, then the control chart can consider, as representative of the process
and these become the standard values. Most processes are not in control when first
analysed. An analysis of Figure 5 shows that there is out of control points on the control
chart that had assignable causes that need to be clarified. The subgroups with assignable
causes are not considered part of the natural variation and are discarded from the date,
and new values of centre line and control limits computed with remaining data. Thus the
centre line and control limits must be revised after discarding the out of control points as
shown in Figures 6 and 7.
Figure 4
Mean and range control chart for temperature tests (see online version for colours)
342
R.H. Fouad and S.D. Al-Shobaki
Figure 5
Standard deviation and range control chart for temperature tests
(see online version for colours)
The interpretation of control charts indicates some process characteristics and
distinguishes between natural and unnatural variation. The unnatural variation results
from assignable causes. From Figures 6 and 7, it was noted that the assignable causes can
be deficiencies in the cooling system, lack of the chemical reactors maintenance and
errors in the sampling process and specification of the correct sample size.
Corrective actions were proposed as to replace the cooling system with a new one,
implementing preventive maintenance, routine tests and inspections and choosing a
representative sample size and frequency for the sampling procedure. As for the natural
variations, which occur due to chance causes, are considered temporary and no corrective
action is taken to modify them.
The first priority is to eliminate all assignable causes then analysing the chance
causes. Through in-depth study and analysis of the process, sources of chance causes
were specified as follows:
1
human errors in calculations of statistical charts or in using test equipments and
measuring equipments were out of calibration
2
poor maintenance plans
3
variation in incoming materials
4
gradual change in temperature and humidity
5
different workers taking samples and using the same chart
6
poor storage conditions.
Developing and implementing statistical process
Figure 6
Revised mean and range control chart for temperature tests
(see online version for colours)
Figure 7
Revised Standard deviation and range control chart for temperature tests
(see online version for colours)
7
343
Conclusion
From this study, several conclusions were developed. The Pareto diagram identifies that
the temperature is the vital view characteristic that need attention. Histogram analysis
shows that the process capability is greater than the tolerance, therefore, an undesirable
344
R.H. Fouad and S.D. Al-Shobaki
situation exists and a non-conforming product is produced. The interpretation of control
charts indicates sources of assignable causes were deficiencies in the cooling system,
lack of the chemical reactors maintenance and sampling process and specifying the
correct sample size. Also the sources of chance causes were human errors
(in calculations of statistical charts or in using test equipments and measuring equipments
were out of calibration), poor maintenance plans, variation in incoming materials,
gradual change in temperature and humidity, different workers taking samples and using
the same chart and poor storage conditions.
Acknowledgement
It is recommended that a further study should be carried out and brainstorming sessions
should be formally held to further analyse the causes of the temperature control problem
using Ishikawa diagram to represent a more meaningful relationship between bad effects
and to take action and correct the problems causes.
References
Besterfield, D.H. (2004) Quality Control, 7th edition, Pearson-Prentice Hall.
Montgomery, D. (2005) Introduction to Statistical Quality Control, 5th edition, John Wiley.