Sunteți pe pagina 1din 22

See

discussions, stats, and author profiles for this publication at:


https://www.researchgate.net/publication/27650042

Developing an SME based six sigma


strategy

Article in Journal of Manufacturing Technology Management · June 2006


DOI: 10.1108/17410380610662852 · Source: OAI

CITATIONS READS

34 486

2 authors, including:

Andrew J Thomas
Cardiff Metropolitan University
107 PUBLICATIONS 763 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Renewable technology supply chains View project

Value in Higher Education View project

All content following this page was uploaded by Andrew J Thomas on 13 January 2015.

The user has requested enhancement of the downloaded file.


Journal of Manufacturing Technology Management
Developing an SME based six sigma strategy
Andrew Thomas Richard Barton
Article information:
To cite this document:
Andrew Thomas Richard Barton, (2006),"Developing an SME based six sigma strategy", Journal of
Manufacturing Technology Management, Vol. 17 Iss 4 pp. 417 - 434
Permanent link to this document:
http://dx.doi.org/10.1108/17410380610662852
Downloaded on: 13 January 2015, At: 12:14 (PT)
References: this document contains references to 26 other documents.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

To copy this document: permissions@emeraldinsight.com


The fulltext of this document has been downloaded 3117 times since 2006*
Users who downloaded this article also downloaded:
Jiju Antony, Maneesh Kumar, Christian N. Madu, (2005),"Six sigma in small- and medium-sized UK
manufacturing enterprises: Some empirical observations", International Journal of Quality & Reliability
Management, Vol. 22 Iss 8 pp. 860-874 http://dx.doi.org/10.1108/02656710510617265
Sameh Saad, Terrence Perera, Pius Achanga, Esam Shehab, Rajkumar Roy, Geoff Nelder, (2006),"Critical
success factors for lean implementation within SMEs", Journal of Manufacturing Technology Management,
Vol. 17 Iss 4 pp. 460-471 http://dx.doi.org/10.1108/17410380610662889
Andrew Thomas, Richard Barton, Chiamaka Chuke-Okafor, (2008),"Applying lean six sigma in a small
engineering company – a model for change", Journal of Manufacturing Technology Management, Vol. 20
Iss 1 pp. 113-129 http://dx.doi.org/10.1108/17410380910925433

Access to this document was granted through an Emerald subscription provided by 108095 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.

*Related content and download information correct at time of download.


The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1741-038X.htm

Developing an
Developing an SME based six SME based six
sigma strategy sigma strategy
Andrew Thomas and Richard Barton
Manufacturing Engineering Centre, Cardiff University, Cardiff, UK 417
Received October 2005
Abstract Revised November 2005
Purpose – The effective implementation of the six sigma strategy within UK manufacturing Accepted December 2005
industries, in particular SMEs, can be considered to be poor. SMEs cite high costs and complexity of
implementation as being the major limiting factors as to its widespread use. This paper aims to
describe the application of six sigma in an SME and to show how the company applied a cost-effective
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

six sigma methodology to eradicate a critical to quality (CTQ) issue. The paper seeks to conclude by
developing a strategic framework for the widespread use of six sigma in SMEs.
Design/methodology/approach – This paper applies the six sigma strategy in an SME in order to
eradicate a major CTQ issue. It identifies the approach employed, the tools and techniques used and
shows the savings that were made through the structured application of the DMAIC procedure, which
is at the heart of the six sigma approach.
Findings – Through the correct application of experimental design approaches, this paper identifies
the optimum parameter settings that enabled the company to eradicate the CTQ issue and achieve
significant improvements in quality and cost from a modest financial outlay.
Research limitations/implications – The design and implementation of an SME-specific six
sigma strategy and its application to a real engineering problem will enable companies to apply the
techniques and to attain improvements in terms of cost and quality.
Originality/value – The application of the six sigma strategy and the resulting conclusions as to its
effectiveness for industry are the real value of this paper. This paper will be valuable for quality
professionals, design engineers and manufacturing specialists in a wide range of industries.
Keywords Pareto analysis, Taguchi methods, Small to medium-sized enterprises,
Manufacturing industries
Paper type Technical paper

Introduction
Six sigma can be considered both a business strategy and a science that has the aim of
reducing manufacturing and service costs, and creating significant improvements in
customer satisfaction and bottom-line savings through combining statistical and
business process methodologies into an integrated model of process, product and
service improvement. In six sigma, customer focus becomes the top priority. Six sigma
improvements are defined by their impact on customer satisfaction and value (Pande
and Holpp, 2002). From an internal perspective, six sigma provides a way of improving
processes so that the company can more efficiently and predictably produce
world-class products and services. Traditionally a five-phased DMAIC methodology is
applied by six sigma teams that tackle specific problems to reach six sigma levels of
performance (Breyfogle, 1999). These phases are: Journal of Manufacturing Technology
Management
(1) (D)efine. Who are the customers and what are their priorities? Vol. 17 No. 4, 2006
pp. 417-434
(2) (M)easure. How is the process measured and how is it performing? q Emerald Group Publishing Limited
1741-038X
(3) (A)nalyse. What are the most important causes of defects? DOI 10.1108/17410380610662852
JMTM (4) (I)mprove. How do we remove the causes of the defects?
17,4 (5) (C)ontrol. How can we maintain the improvements?

At the heart of the six sigma approach is the application of DOE techniques. These
techniques help to identify key factors and to subsequently adjust these factors in
order to achieve sustainable performance improvements. It is the complexity of these
418 DOE techniques that are often cited by companies as to the reason why they are unable
to employ six sigma. A short overview of the DOE techniques is now covered.

Background to DOE techniques


The statistical approach to design of experiments and the analysis of variance
technique was developed in the 1930s (Fisher, 1935). The technique provides a very
powerful and economical method to determining significant factors and factor
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

interactions that affect variability within a product. This classical approach to the
design of experiments was further developed in the late 1980s when Genichi Taguchi
developed his own form of statistical design of experiments. The Taguchi (1987)
method suggested that the design process consisted of three phases namely; system
design, parameter design and tolerance design. The latter two being based on
statistical design of experiments.
Both approaches provide a very powerful means to understanding the process
behaviour and hence the key factors that influence the performance of any system or
process. Whilst the work of both Fisher and Taguchi is relatively well known in
modern manufacturing industry, the work of Dorian Shainin receives less acclaim but
his non-academic statistical technique is gaining greater prominence. The Shainin
DOE technique (Shainin and Shainin, 1988; Antony, 1999; Bohte, 2000) permits the
luxury of considering as many variables as can be identified. The subsequent grouping
of these variables into “families” and identifying the most influential variables based
on statistical significance as opposed to assumptions provides a greater chance of
identifying the major factors of variance (Bhote, 1991). Shainin identifies and
categorises the major factors contributing to variance as Red X, Pink X and Pale Pink
X (Goodman and Wyld, 2001). Red X being the major factor causing variance, Pink X
being the second factor and Pale Pink X being the third.
This paper develops both the Taguchi DOE and Shainin’s components search
technique (CST) by conducting experiments on identifying the influencing variables
that affect the critical to quality (CTQ) issue identified by Orangebox. The DOE
techniques are applied through using the six sigma DMAIC approach and are shown in
the following sections of this paper.

The requirement for an SME based six sigma methodology


There is much debate as to whether formal quality enhancement approaches can be
effectively implemented and subsequently utilised by SMEs. Thomas and Webb (2003)
in their work on analysing quality systems implementation in SMEs highlight the lack
of intellectual and financial capacity within small companies as being the primary
issues that lead to poor systems implementation. They go on to state that the
uniqueness and complexity of SMEs manufacturing operations often hinder the
implantation process. Husband (1997) on studies with Australian SMEs identify
similar problems and go onto identify a series of issues that affect the ability of an SME
to implement formalised quality enhancement techniques. The main issue is one of Developing an
developing a rigorous model that is both suitable to the wide range of SMEs but is not SME based six
so generic that it fails to provide adequate direction and guidance to the company.
In order to do this, a series of objectives need to be identified so as to allow for the sigma strategy
development of a suitable strategic framework for implementing six sigma into SMEs.
These are:
.
A cost-effective approach that aligns itself with the specific SME issues and will 419
have relevance and purpose for all types of SME.
.
A system to allow integration of the model into the strategic and operational
frameworks of SMEs.
.
A model that can adopt specific quality methods and techniques that best suit
SMEs and which directly attack CTQ problems.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

Husband and Mandal (1999) also identify the uniqueness of an SMEs manufacturing
operations as being a limiting factor to quality enhancement implementation
and provide a series of dimensions that are unique to SMEs and suggest that if
these dimensions are not integrated into the model then an SMEs ability to achieve
significant outputs from the application of the model will be compromised. These
dimensions are:
.
Core – products and/or services.
.
Structural – size, location, age, ownership and legal entity/structure.
.
Fundamental – systems, people and measures.
.
Sustainability – leadership and planning, risk and change, and technology and
innovation.
.
Integrative – customers, suppliers and partners.
.
External – competition, stakeholders, government and economy.

Furthermore, Deleryd et al. (1999) identify that SMEs need to make decisions and
improve their processes based on accurate and timely information relating to the
performance of their manufacturing process. To manufacturing companies this is
crucial not least within the design and production areas. This means that a deeper
understanding of the concept of variation, identification of causes of variation and
handling of these causes are important factors within SMEs. It, therefore, follows that
the development of process control theory, experimental design concepts and issues
relating to product reliability cannot solely remain in the domain of the larger
industries in which resources are available to train the workforce to apply these
concepts. These statistical concepts have a major part to play in SMEs and the
application of such principles must come from continued training and development of
the company’s workforce.
The resulting problem shows the lack of application of statistical theory to identify
and solve problems within a manufacturing context. There are several reasons for the
relatively low application of statistical methods in SMEs. Management in small
companies, in general, do not have the sufficient theoretical knowledge to see the
potential of using statistical tools. In many cases, they and their employees, even
become frightened when statistical tools are discussed.
JMTM Small companies also lack resources in the form of time and personnel. Small
17,4 organisations tend to have a lean organisation and, therefore, they find it difficult to
appoint a facilitator or co-ordinator for the implementation process. In addition, they
also have limited resources to provide internal training. Lack of resources in these
aspects leads to a need for a careful analysis of which strategy to use when
implementing statistical methods in order to succeed (Husband, 1997).
420 Having an array of specific tools and techniques available to the SME can allow the
company to develop what can be termed the “quality enhancement” issues relating to
systems and product based quality. These issues are essential to the company’s
continued development and include amongst other things; problem solving,
benchmarking, continuous improvement, etc. These techniques prove to be far more
effective when backed by statistical data and can achieve greater success when
implemented within a systems approach that is designed to suit SMEs.
The primary focus for any SME, therefore, that intends to adopt the six sigma
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

methodology is to undertake the project in the most cost-effective manner and, to


be able to recoup the initial project costs quickly after the completion of the project.
At the heart of this cost-effectiveness is the need to undertake the six sigma project
in-house with the minimum of costly consultancy support. This, therefore, calls for a
simple six sigma approach that utilizes powerful yet simple statistical and DOE tools
to aid in the process.
The following work highlights the tools and techniques that were used by
Orangebox UK in their development of the six sigma programme.

Introduction to Orangebox
Established in 1985, Orangebox Ltd an SME based in Hengoed, South Wales has
grown to become a world leader in the design and manufacture of office seating and
furniture. This unique position has been achieved due to an unstinting attention to
quality, cost and delivery and through ensuring total customer satisfaction.
The market sector is highly competitive and any possible advantage competitors
can get to obtain increased market share will be mercilessly exploited. Orangebox has
achieved ISO 9000 certification and has implemented company wide lean, TQM and
TPM initiatives to good effect. As an SME the company is unique in the fact that they
have been able to implement these initiatives so well. Teamworking and the ability to
attract highly skilled engineering staff to drive these initiatives forward have been
cited as key issues in this success.
The move towards six sigma was instigated through a need to directly
tackle customer concerns much earlier in the process. The company found that the
TQM process and its associated systems were too slow in responding to such
problems primarily, since they were developed to obtain long-term strategic
direction and focus. Six sigma was, therefore, adopted in order to tackle customer
CTQ issues immediately before relationships suffered and customers were
attracted elsewhere.
The company is highly sensitive to customer quality issues and reacts quickly to
problems they consider to be CTQ. This allows them to keep ahead of the competition,
however, until the introduction of the six sigma strategy, the company attended to
quality problems in an often ad-hoc and unstructured manner. The following section
shows how the company followed the six sigma methodology in an attempt to provide
a structured approach to solving CTQ problems within the company and to achieve Developing an
enhanced customer satisfaction and internal financial benefits. SME based six
The company invested in training a senior engineer in the principles of six sigma to
black belt level. On completion of the training programme, the engineer was given a sigma strategy
small process enhancement team in order to apply the six sigma methodology to highly
sensitive, high value product lines. The following work provides a case study of one
particular project undertaken. 421
DMAIC at Orangebox
The six sigma process concentrates on a simple five phase methodology called DMAIC.
DMAIC is an anagram of the major steps within the methodology namely; define,
measure, analyse, improve, control. The company followed this approach and each
stage is explained in detail in the following section of the paper.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

Define
Orangebox employ field service engineers to undertake basic chair maintenance work
at the client premises. Their primary role is to service their client’s needs that include
basic repair and modifications outlined by the company. The type and frequency of
work undertaken by the field engineers is closely monitored so that any CTQ problems
are highlighted immediately and solved at source.
A Pareto diagram is plotted on a weekly basis and records the number and type of
faults the service engineers rectify during that time. This plot provides advanced
warning to the company of impending problems or trends. Figure 1 shows the Pareto
diagram plotted for the CTQ issue identified.
It was during the monitoring process that Orangebox engineers confirmed a CTQ
issue that threatened their product image and their current and potential client base.
The company was quickly aware that the arm mechanism on their leading product was
working loose over a relatively short period of time. Since, the arm mechanism allowed
the user to adjust the height position of the arm within a pre-defined range, the arm

Pareto Analysis of Year to date quality related customer complaints


50 140

120 120
40
No. of occurances

100
% of total

30 80 80

42 60
20

28 40
28
10
18 20
10 10 12 8 7
5 6
3 4 3 2 2
0 0
Joint/fastener

failure (screw

Delivery
damage

adjustment

Mechanism

prematurely

Wrong spec.

Frame fault

other
G64 2D arm

Upholstery

Figure 1.
Back

parts
failure

Nature of fault
Pareto chart showing CTQ
issue
JMTM fixing has to react to a large amount of repetitive movement as well as to the forces
17,4 exerted in its role as an arm rest. A combination of these was evidently compromising
the joint strength and loosening the single fastener. A similar arrangement is seen in
the back rest fixing, although there are three fasteners used adding further complexity
to the problem.
As a key part of the process, Orangebox planned to use the optimum process
422 conditions identified from the experimentation section of this project as a test condition
to establish the extent of the problem and to estimate the potential for any
improvement in the back fixing.
Orangebox identified the arms were working free after approximately six months of
use which equated conservatively to 60,000 cycles to failure (CTF). The company had
identified the life to service as being 100,000 cycles which was deemed to be well within
the bi-annual service interval provided by the company under their service plan.
Therefore, the problem was defined as:
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

. . . the premature loosening of chair arms through normal cyclical movement by the operator
with the arms working loose at approximately 60 per cent of the expected design life.

Measure
In order for the company to measure the extent of the problem, it was considered
appropriate that simulated accelerated testing would be undertaken at the company
premises on ten chairs drawn at random from the production line. The aim of the test
procedure was to identify the early onset of failure of the arms so small movements
were measured to identify the failure criterion.
A test jig was constructed that simulated the typical movement that would be
experienced under normal operating conditions. The test jig applied a 1 kN vertical
force that created a deflection in the arm of 20 mm. The force was applied for 5 seconds
before being removed for a further 10 seconds. It was expected that the arm would
return to its original set height once the force was removed. A limit switch was
attached to the arm and set so that when the arm deflected to 22 mm or more the switch
would trigger a light to indicate that the deflection was greater than anticipated and
thus indicated the chair had failed. Owing to the need to reduce testing time, the rig
undertook a set of 30,000 cycles of operation. This was deemed to be equivalent to
100,000 CTF as specified in the test procedures in the company. If the switch was
triggered during the 30,000 cycle test, it would thus indicate the cycles to the onset of
failure (CTOF) rather than CTF since the rig was set to trigger at a displacement value
slightly greater than what the engineers would expect a normal chair to have after
30,000 cycles of normal operation.
Accelerated testing was undertaken on each of the ten chairs drawn randomly from
the production line. In eight out of ten cases the chairs failed the test procedure thus
confirming that the faults identified by the customer were real and that a CTQ did
actually exist. The results of the testing are shown in the analysis stage.

Analyse
A brainstorming exercise was carried out by a multi-disciplinary team of engineers
from within the company and led by the black belt six sigma champion (Six Sigma
Qualtec, 2002). The team identified and subsequently classified (Taguchi, 1987) the
factors that could influence the joint integrity and failure characteristics. Figure 2 Developing an
shows the development of cause and effect diagram that highlights the potential major SME based six
contributory factors that limit joint failure. Twenty possible factors were identified and
these factors were subsequently isolated and analysed through further testing for their sigma strategy
contribution to the failure of the arms. A Pareto study was undertaken in an attempt to
identify and prioritise the factors so that a suitable number could be used for the
Taguchi experimentation. However, experimentation was abandoned since it was 423
virtually impossible to identify the key contributing factors from simple
experimentation. Without taking educated guesses as to the factors of real
importance, the team decided to adopt the Shainin CST. The Shainin’s CST was
employed to identify whether the primary cause of premature arm failure lay within
the assembly process itself or within the components used (Bohte, 2000). This allowed
for a first stage filter to be employed that cut down the factors to a manageable
number. There are five key stages in the Shainin CST, the team employed the first
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

three stages in order to obtain a list of important factors, the steps are:
(1) Identify a list of input variables and undertake testing using a suitable sample
size. From this, calculate the median average and then calculate DM : R ratio to
verify whether the variables selected are suitable.
(2) Compute the control limits for the median values obtained from step 1 so that
further analysis can be undertaken to identify if the variables selected are
critical to the experiment.
(3) Separate the critical variables from the non-critical variables in order to identify
the most important factors.

Step 1 of Shainin’s work states that if the ratio of DM : R after first stage testing is less
than 1.25:1 then the team could be confident that the problems lay within the assembly
process, however, if the ratio was higher than 1.25:1, then it could be assumed that the
problems lay within the components used in the construction of the arm (Bohte, 2000).

Men Methods
No / wrong adhesive
Wrong screw used applied

Torque specification Fastener head type


not adhered to No / wrong washer used
Fastener not No lubrication
tightened completely
Cross-threaded Wrong torque specified
fastener Number of threads in
contact Joint
Failure
Thread pitch
Material hardness
Screw diameter
Tool run- Screw shank shears
down speed Thread form
Application environment
Screw length
(heat / vibration / other)
Figure 2.
C þ E diagram of the
Tool torque setting Screw head breaks
off possible factors
influencing joint strength
Machines Materials
JMTM The result of this stage of the Shainin approach identifies the ratio obtained was 14:1
17,4 thus showing that the problem lay firmly in the components used for the arm
mechanism. This allowed the team to quickly focus on the components that could
affect the performance of the arm mechanism.
Steps 2 and 3 enabled the team to identify the key variables, the list of control
variables are shown in Table I and highlight the effectiveness of using a first stage
424 filter in that it achieved a reduction of variables from the original twenty to five. From
here, any interactions could then be added within the orthogonal array (OA) when the
Taguchi technique was employed.
The reason why the Shainin technique was not fully developed was down to the fact
that the team wanted a simple approach to identifying and solving the CTQ and
variability issue simultaneously, the Shainin approach would have meant further
testing for each objective function.
Now that the factors had been identified, a Taguchi DOE experiment was then
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

conducted in order to identify the key factors and the interactions that may exist
between them. (Taguchi, 1987; Phadke, 1989; Dingus, 1989; Quinlan, 1985; Antony and
Kaye, 1995; Montgomery, 1992).

Improve
The DOE exercise. Having selected the parameter design experiment, the next step was to
choose an appropriate OA design for the experiment. The choice of OA design depends on:
.
number of control/noise factors for the experiment;
.
number of levels of each control factor;
.
number of levels of each noise factor;
.
degrees of freedom associated with the control factor effects;
.
degrees of freedom associated with noise factor effects (removed in this case); and
.
cost and time constraints.

Control factors and interactions


Five main control factors at two levels each were considered for the experiment.
Table II shows the control factors and levels used. The classification of factors
(Phadke, 1989) into control and noise factors was achieved via brainstorming and were
subsequently reduced following further analysis. It was decided to concentrate efforts
into reducing the number of noise factors to zero in this experiment in order to achieve
simplicity of analysis.
For the experiment, the interactions A £ B, and A £ E were of interest (Antony and
Kaye, 1995; Six Sigma Qualtec, 2002). There are five degrees of freedom associated
with the control factor effects (i.e. 1 DOF per factor). In addition, two-two-factor
interactions were of interest. Therefore, three degrees of freedom are associated with

Torque setting A
Adhesive type B
Lubrication C
Table I. Washer thickness D
List of control variables Screw type E
the two-factor interactions (i.e. 1 DOF per interaction). The total number of the degrees Developing an
of freedom is equal to the sum of the degrees of freedom for the main and interaction SME based six
effects, and is therefore, seven for the control array. It was decided to use an L8 OA
design (Antony and Kaye, 1995; Six Sigma Qualtec, 2002), as this allows the study of sigma strategy
four main effects and up to three-two-factor interactions. Table III shows the OA
design matrix used for the analysis.
425
Control factors and levels
Control factors were initially tried at two levels. Previous testing of threaded fasteners
identified the need to either increase thread contact surface area or exert a greater
clamping force on the head of the fastener to improve the mechanical strength of the
joint. The second level parameters were, therefore, set to “artificially” create the same
effect. If after analysis, a significant joint strength improvement was not achieved, the
aim would be to look at a greater number of factors and a possible third level. Table II
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

shows the factors chosen for this experiment.

Experimental layout
The arm assemblies were manufactured using standard company operating
procedures, however the test jig was modified slightly. The majority of the arms
deflection was due to other part interfaces, so fixing the chair movement mechanism (a
stiff assembly) meant that only the mechanism/arm joint could be tested. Once a test
had revealed an expected deflection of 5 mm, a new failure deflection was set at 7 mm.
This has become an essential part of the experiment and profoundly affects the
readings if not applied since it reduced virtually all of the expected noise factors to nil.
A benchmark test was carried out on five chairs and a mean CTOF level of 18,250
cycles was identified which was equivalent to 60,225 CTF. A variance of 140.4 was also
calculated for the arms thus indicating a large variation between the test readings. Five

A B AB C D AE E

1 1 1 1 1 1 1
1 1 1 2 2 2 2
1 2 2 1 1 2 2
1 2 2 2 2 1 1
2 1 2 1 2 1 2 Table II.
2 1 2 2 1 2 1 Design matrix for the
2 2 1 1 2 2 1 experiment showing a
2 2 1 2 1 1 2 coded OA

Uncoded level Coded level


Name Abbreviations Units 1st 2nd 1st 2nd

Torque setting A Nm 20 30 1 2
Adhesive type B Type A B 1 2
Lubrication C Type Dry Oil 1 2 Table III.
Washer thickness D mm 1 3 1 2 Factors and levels used in
Screw type E mm 30 35 1 2 analysis
JMTM tests were carried out with the factor levels at each L1 and L2. The analysis of the
17,4 results is shown in the factor identification Table VI and the ANOVA Table VII.

Statistical analysis
Having obtained the response values using an L8 OA (Table IV) the following
steps were used to analyse and interpret the results from the analysis. A sample
426 of five tests were conducted at each experimental point, Table IV shows the mean
CTF values obtained from these experiments.
As the resolution of the design was III (i.e. main effects are confounded with
two-factor interactions). For this study, all the main effects and the two-two-factor
interaction effects A £ B, and A £ E, were calculated. Table V illustrates the estimated
main and interaction effects.
Analysis of the mean response was performed to identify the factors and their
interactions that influenced the mean response (~y).
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

The ANOVA table constructed in Table VII shows that the main effects A and B
and the A £ B interaction were judged to be statistically significant. It is worth noting
that if none of the factors or their interaction effects from the analysis had been
statistically significant, then there would have been a need to:
.
select a new factor or set of factors (possibly the factors that were not considered
in the initial experimentation);
.
ensure that the choice of design was appropriate for the given problem under
investigation;
.
ensure that the measuring equipment was accurate enough and also that the
experiment was performed based on the prepared design matrix; and
.
redefine the objectives of the experiment.

Std order Run no A (mm) B (type) C (type) D (mm) E (A/B) CTF ( £ 1000)

1 2 20 A Dry 1 30 18.05
2 5 20 A Oil 3 35 19.40
3 8 20 B Dry 1 35 21.60
4 6 20 B Oil 3 30 22.89
5 7 30 A Dry 3 35 30.30
Table IV. 6 1 30 A Oil 1 30 31.71
Experimental layout used 7 3 30 B Dry 3 30 32.20
for the experiment 8 4 30 B Oil 1 35 33.68

Average response
Cycles to onset of failure
Main effect L1 L2

Torque setting (A) 20.49 31.97


Adhesive type (B) 24.87 27.59
Table V. Lubrication (C) 25.54 26.92
Response table of main Washer thickness (D) 26.26 26.20
effects Screw type (E) 26.21 26.25
For this study, the objective was to improve the CTF and reduce variability in joint Developing an
strength. The “larger (longer) the better” quality characteristic was chosen in order to SME based six
identify the correct parameter settings for the joint.
The final selection of optimal operational factor settings was based on the responses sigma strategy
obtained from the statistical study shown in Tables IV-VIII. Table VI shows the
optimum factor settings and predicted CTOF for the joint setting identified. The results
of the analysis are shown most effectively in Figures 3 and 4, graphical representations 427
of the higher cycle to failure for each of the optimal settings.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

Table VI.
Factor identification and
predicted parameter
settings

Factor DOF SS MS F ratio Percentage of contribution

A 1 1,319.85 1,319.85 124.94 75 99 per cent


B 1 74.39 74.39 7.04 4.2 95 per cent
C 1 19.06 19.06 1.80 0.4
D 1 0.04 0.04 0.04 1.1
E 1 0.01 0.01 0.00 0.0
AB 1 6.23 6.23 0.59 0.0
AE 1 0.00 0.00 0.00 0.0
Error 32 338 10.5
Total 39 1,757.63
F 0.05, 1, F 0.01, 1,
32 ¼ 4.19 32 ¼ 7.6
Note: 95 and 99 per cent indicates factor/interaction is statistically significant at 5 and 1 per cent level Table VII.
of significance, respectively ANOVA of CTF

CTOF (before) CTOF (after) CTF increase (per cent)

CTF 18,050 34,080 53 Table VIII.


Variation (before) Variation (after) Variance Red (per cent) Results of confirmatory
Variance 140.4 66.13 47 runs
JMTM 33.0
L2
17,4 31.0
29.0
L2
27.0 L2
L1 L1 L2
CTF / 1000 L1
L2
25.0 L1
428 23.0
21.0 L1
Figure 3. 19.0
Graph showing the main 17.0
effects plot of each
15.0
individual variable
Variable
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

35
33
31
29
Response

27
25
23
21
19
Figure 4. 17
Graph showing the effects
plot of the two factor 15
interactions B1 B2 E1 E2
Variable

The statistical analysis of the factor significance was again best represented by the
main effects plots shown in Figures 3 and 4. In support of the ANOVA, it is seen that
only factors A and B are statistically significant, but interestingly progressively less
significant from A to E. The author suggests that this is another benefit of the variable
identification and elimination process that was followed by the team. It is also seen in
Figure 4, the interaction response plot, that factor E was proven not to significantly
affect the joint strength, and that the interaction AE was judged to be insignificant.
This is shown graphically when the two variable lines are virtually parallel, whereas
the AB interaction lines are seen to be converging, proving that a setting of B2 will
further improve the response of an A2 setting, although the fact that they do not cross
demonstrates how much more significant factor A is.
The interaction plots are again seen to be a very good way to de-mystify the
statistics for management purposes.
At this point, further analysis into the proposed improvement conditions could be
carried out through signal to noise ratio calculation, and a pooled ANOVA, which
would help demonstrate the new capability, and a measure of robustness, at reducing
variability and identify the most significant factors. However, one key aspect of this
project is the implementation of a six sigma process in an SME in the UK and as such, Developing an
this analysis is unlikely to be of interest to senior management in place of further work SME based six
on the implementation, and could even threaten to weaken the involvement of some
parties. sigma strategy

Analysis of predicted settings


The selection of the optimal parameter settings, by identifying each optimum 429
individual factor level, is obtained by analysing the mean response table, in order
to obtain maximum joint strength. The predicted joint capability was a maximum
with a 2, 2, 2, 1, 2 joint construction. It is seen that the optimum conditions do not
validate each of the proposed improvements.
Having obtained the optimal factor settings as 2, 2, 2, 1, 2, confirmatory runs were
required. Confirmatory runs were conducted primarily to calculate the variance (SD)
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

for this setting as well as to investigate the failure modes of the new joint against those
of the current bonded joints.
Confirmatory runs were performed and showed that the mean CTOF actually realised
using the joints using the 2, 2, 2, 1, 2 parameter settings was very close to the theoretical
response shown in Table VI. Table VIII shows the results of the confirmatory runs.

Control
Once the confirmatory runs showed a positive response, the design department was
informed of the parameter changes and were instructed to modify the drawings in line
with the changes identified. This final process was critical since it closed the loop and
ensured that the improvements achieved through the six sigma project were delivered
each time a product was made.
The improvements identified were also used to set future tolerances, materials
specifications and dimensional characteristics of future products. Inspection levels
were increased initially in order to ensure that the improvements in the arm integrity
were repeatable, this proved to be the case and after six months, the inspection level
dropped to normal once more.

Developing an SME based six sigma model


Having worked to instigate a process to identify opportunities to improve quality
and reduce variability using statistical tools, further work can now be done to
integrate the DOE process throughout the existing model. The aim being to defeat
the biggest “excuses” cited by SMEs as the reasons six sigma is not feasible: high
costs and complexity of implementation, by extending the scope and effect of the
work, emphasising the “doing” elements and by simplifying the DOE exercise (most
complex aspect) by rationalising the variables using the knowledge within the
company.
In addition, it helps to break down so many of the barriers that stand in the way of
individuals using statistical and/or unfamiliar problem solving methods by acting as a
step-by-step guide. Figure 5 shows the development of the proposed model (Antony
and Kaye, 1995).
The model clearly shows the major stages in the process. It shows initially the
sequential nature of the stages whereby the six sigma elements are used to accurately
input to the DOE stage of the main factors and interactions that have a major impact
JMTM Six Sigma Stage DOE Stage
17,4
Multi-disciplinary team to
identify a potential problem.

430
Measure the problem to identify
requirement and benefit of
solution.

Identify the key process, its Confirm the objective, and


variables and constraints. specific target result of
experiment
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

Map the relevant process and Identification of correct


estimate their capability. orthogonal array and interactions

Identification of problem causes,


brainstorming , C+E etc Conduct experiment

Obtain parameter data from DOE


Undertake first stage filter using system with all interaction data.
Shainin s CST approach to
reduce and prioritise factors.
Carry out ANOVA and SNR on
interactions to identify statistical
significance

Undertake Taguchi DOE on


factors identified from Shainin s Verification tests to test
CST prediction with actual result.

No Yes
Conduct test run to prove
problem & need for improvement

No Yes Do values provide a close match


actual performance
requirements.?.
Does the test run value provide a
close match to prediction?

Solution Found control measures, &


Figure 5. continued monitoring to be evaluated
Integrated six sigma/DOE Review process in light of
success difficulties and each month
model
gaps in information

on the performance of the system being modelled. Integration of various stages of the
model comes by way of ensuring that the objectives of the experiment are agreed upon
in the form of targets such as KPIs, e.g. process output, or business MOPs, e.g. scrap
reduction (Breyfogle, 1999).
The next planned test of the integrated model will be to take a more pro-active Developing an
approach, in a less specific environment. Consider the customer service department for SME based six
example; a key, customer facing, facility that frequently delivers poor quality product
in terms of information presented to departments responsible for meeting customers’ sigma strategy
requirements. Successfully improving quality here will prove that quality
improvement is applicable everywhere in the business, through a systematic
approach which, with guidance, is neither frightening nor mysterious. 431
Managerial aspects of using the integrated model
There are many benefits of such an integrated method, but the feedback-based nature
is particularly important to Orangebox. It means that the many opportunities already
known at Orangebox can be prioritised by a multi-disciplinary team considering their
individual targets, the company’s goals and the effects of the expected improvement.
This in turn means that the iterative nature of the model can be used to achieve
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

optimisation and/or continuous improvement of a system, in an almost continuous


systematic approach, and hence help manage the entire business process of the
company.
As a result of the above, the model also enables the company to help identify the
need to have a multi-disciplinary team working strategically, early in the product
planning stage (Quinlan, 1985; Antony et al., 2003; Kolarik, 1995; De Mast, 2004;
Thomas and Antony, 2003). Especially relevant to Orangebox due to the size of the
company which means that there is a particularly good return on the time spent by the
team; well managed, cost-oriented implementation without disruption and far-reaching
implications on new and existing product specification.
Two issues emerged here:
(1) The introduction of process engineers early in the design stage provided
valuable inputs into how the fastening specification could be changed so that
process build and product strength variation could be reduced.
(2) The introduction of quality and service engineers at this stage made them an
integral part of the design function providing test data to verify the changes in
joint design. This made the quality department more pro-active in the design
function and less reactive as they were traditionally used.

Bottom-line cost savings realised through using the six sigma process
Service costs of “loose arms” to date ¼ 120 £ £350 ¼ £42,000.
Potential cost of failed partial solution ¼ (3 months) 60 £ £350 ¼ £21,000.
Actual cost of project ¼ £4,800.
Actual cost of failures since optimum condition implementation ¼ 3 £ £350 ¼
£1,050.
Realistic saving ¼ £21,000 2 £5,850 ¼ £15,150.

In this case, the above may be considered idealistic due to the small material cost of the
improvement. However, the extra time spent considering the controlling factors meant
that cost could always be kept as a key consideration, maximising the influence of this
particular interpretation of the process, and further lending credibility to the six sigma
technique within the company.
JMTM Conclusions
17,4 .
An experimental study was undertaken in order to identify the optimum
parameter settings and predicted CTF values for the joint. The results of the
experimental study enabled a change to be made to the parameter settings
resulting in a 53 per cent increase in CTOF. This equates to approximately
115,000 CTF and well above the design requirement of 100,000 CTF.
432 .
Through the use of ANOVA, it was also possible to tighten the variability of
failure between the chair arms by approximately 47 per cent thus indicating a
more reliable wear out rate and a more reliable service life interval.
.
The use of Shainin’s CST allowed the engineers to focus quickly on the
components that were deemed important to the process. Through a more
rigorous approach than the simple Pareto study, CST also allowed the engineers
to pick up essential product knowledge from the study.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

.
This relatively simple application of Taguchi’s experimental design technique
should allow for increased use of the methodology for tackling many process
quality problems. Likewise, the results can also provide the stimulus for the
wider application of the technique to create process improvements at relatively
lower costs.
.
The application of the six sigma approach to a CTQ problem at Orangebox
achieved savings in excess of £60,000 for an initial outlay of less than £5,000 in
experimental and project costs.
.
The development of a six sigma model developed a culture towards continuous
improvement and the systematic implementation of the approach throughout the
organisation.
.
The application of the Shainin and Taguchi DOE techniques allowed the
company to develop advanced statistical techniques and to become generally
more “technical” in their approach to problem solving (Thomas and Antony,
2003).

References
Antony, J. (1999), “Spotting the key variables using Shainin’s variables search technique”,
Journal of Logistics and Information Management, Vol. 12 No. 4.
Antony, J. and Kaye, M. (1995), “Experimental quality”, Journal of Manufacturing Engineer,
Vol. 74 No. 4, pp. 178-81.
Antony, J., Forturis, M. and Thomas, A. (2003), Using Six Sigma, Manufacturing Engineer,
IEE Publication, Stevenage.
Bhote, K.R. (1991), World Class Quality, American Management Association, New York, NY.
Bohte, F. (2000), World Class Quality: Using Design of Experiments to Make It Happen, Amacom,
New York, NY.
Breyfogle, F.W. (1999), Implementing Six Sigma, Smarter Solutions – Using Statistical Methods,
Wiley, New York, NY.
Deleryd, M., Garvare, R. and Klefsjo, B. (1999), “Experiences of implementing statistical methods
in small enterprises”, The TQM Magazine, Vol. 11 No. 5.
De Mast, J. (2004), “A methodological comparison of three strategies for quality improvement”,
International Journal of Quality & Reliability Management, Vol. 21 No. 2, pp. 198-213.
Dingus, G. (1989), “An application of Taguchi methods in the foundry industry”, paper presented Developing an
at Seventh Symposium on Taguchi Methods, Michegan, pp. 517-32.
Fisher, R.A. (1935), The Design of Experiments, Oliver and Boyd, Edinburgh.
SME based six
Goodman, J. and Wyld, D.C. (2001), The Hunt for the Red X: A Case Study in the Use of Shainin
sigma strategy
Design of Experiment (DOE) in Industrial Honing Operations, Kennedy-Western
University, Cheyenne, WY.
Husband, S.G. (1997), Innovation in Advanced Professional Practice: Doctor of Technology 433
(Report No 2), Faculty of Science and Technology, Deakin University, Geelong.
Husband, S.G. and Mandal, P. (1999), “A conceptual model for quality integrated management in
small and medium size enterprises”, International Journal of Quality & Reliability
Management, Vol. 16 No. 7, pp. 699-713.
Kolarik, W.J. (1995), Creating Quality: Concepts, Systems, Strategies and Tools, McGraw-Hill,
Maidenhead.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

Montgomery, D.C. (1992), “The use of statistical process control and design of experiments in
product and process improvement”, IIE Transactions, Vol. 24 No. 5, pp. 4-17.
Pande, P. and Holpp, L. (2002), What is Six Sigma, McGraw-Hill, New York, NY.
Phadke, M.S. (1989), Quality Engineering Using Robust Design, Prentice-Hall,
Englewood Cliffs, NJ.
Quinlan, J. (1985), “Process improvement by application of Taguchi methods”, paper
presented at Transactions of the Third Symposium on Taguchi Methods, Michegan,
pp. 11-16.
Shainin, D. and Shainin, P. (1988), “Better than Taguchi orthogonal tables”, Quality & Reliability
Engineering International, Vol. 4, pp. 143-9.
Six Sigma Qualtec (2002), The Great Balancing Act, Six Sigma Qualtec, Princeton, NJ, available
at: www.ssqi.com/library/whitepapers
Taguchi, G. (1987), System of Experimental Design,Vol. 1, 2, ASI, Dearborn, MI.
Thomas, A.J. and Antony, J. (2003), “Applying Shainin’s variable search methodology in
aerospace applications”, Journal of Assembly Automation, Vol. 24 No. 2, pp. 184-91.
Thomas, A.J. and Webb, D. (2003), “Quality systems implementation in Welsh small-
to medium-sized enterprises: a global comparison and a model for change”,
Proceedings of the I MECH E, Journal of Engineering Manufacture, Vol. 217 No. 4,
pp. 573-9.

Further reading
Antony, J. and Kaye, M. (1996), “Optimisation of core tube life using experimental design
methodology”, Journal of Quality World, pp. 42-50.
Antony, J., Kaye, M. and Frangou, A. (1998), “A strategic methodology to the use of
advanced statistical quality improvement techniques”, The TQM Magazine, Vol. 10 No. 3,
pp. 169-76.
Taguchi, G. (1989), Introduction to Quality Engineering, UNIPUB, New York, NY.

About the authors


Andrew Thomas is a Lecturer and Project Manager at the Manufacturing Engineering
Centre, Cardiff University. He entered academia having followed an industrial career in
production engineering. He is a Chartered Engineer and obtained his BE in Mechanical
Engineering, MSc in Manufacturing and a Doctor of Engineering degree in Manufacturing
JMTM Systems. His main research interests include manufacturing systems design and analysis,
manufacturing fitness, design process development. He has developed close links with
17,4 industry conducting numerous large-scale engineering projects in the areas of systems
engineering and has published a series of international journal and conference papers in the
area of manufacturing systems design and analysis. Andrew Thomas is the corresponding
author and can be contacted at: thomasaj3@cf.ac.uk
Richard Barton is a Research Fellow at the Manufacturing Engineering Centre, Cardiff
434 University. He obtained his BE in Mechanical Engineering and MPhil in Lean Manufacturing.
Upon leaving university he pursued an industrial career in production engineering before
returning to academia. His main research interests include manufacturing systems design and
analysis, lean and agile systems development. Over the years he has developed close links with
industry and continues to service industrial needs through providing consultancy services on
behalf of the university.
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com


Or visit our web site for further details: www.emeraldinsight.com/reprints
This article has been cited by:

1. Ravi S. Reosekar, Sanjay D. Pohekar. 2014. Six Sigma methodology: a structured review. International
Journal of Lean Six Sigma 5:4, 392-422. [Abstract] [Full Text] [PDF]
2. Wagner Cezar Lucato, Milton Vieira Júnior, José Carlos da Silva Santos. 2014. Eco-Six Sigma: integration
of environmental variables into the Six Sigma technique. Production Planning & Control 1-12. [CrossRef]
3. Alireza Shokri, David Oglethorpe, Farhad Nabhani. 2014. Evaluating Six Sigma methodology to improve
logistical measures of food distribution SMEs. Journal of Manufacturing Technology Management 25:7,
998-1027. [Abstract] [Full Text] [PDF]
4. Hyunjei Jo, Sang Do Noh, Yongju Cho. 2014. An agile operations management system for green factory.
International Journal of Precision Engineering and Manufacturing-Green Technology 1, 131-143. [CrossRef]
5. Manoj Dora, Dirk Van Goubergen, Maneesh Kumar, Adrienn Molnar, Xavier Gellynck. 2013. Application
of lean practices in small and medium-sized food enterprises. British Food Journal 116:1, 125-141.
[Abstract] [Full Text] [PDF]
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

6. Steven Cox, John Garside, Apostolos Kotsialos, Valentin Vitanov. 2013. Concise process improvement
definition with case studies. International Journal of Quality & Reliability Management 30:9, 970-990.
[Abstract] [Full Text] [PDF]
7. M. Prasanna, Sekar Vinodh. 2013. Lean Six Sigma in SMEs: an exploration through literature review.
Journal of Engineering, Design and Technology 11:3, 224-250. [Abstract] [Full Text] [PDF]
8. Manoj Dora, Maneesh Kumar, Dirk Van Goubergen, Adrienn Molnar, Xavier Gellynck. 2013. Operational
performance and critical success factors of lean manufacturing in European food processing SMEs. Trends
in Food Science & Technology 31, 156-164. [CrossRef]
9. Manoj Dora, Maneesh Kumar, Dirk Van Goubergen, Adrienn Molnar, Xavier Gellynck. 2013. Food quality
management system: Reviewing assessment strategies and a feasibility study for European food small and
medium-sized enterprises. Food Control 31, 607-616. [CrossRef]
10. D.T. Matt, E. Rauch. 2013. Implementation of Lean Production in Small Sized Enterprises. Procedia
CIRP 12, 420-425. [CrossRef]
11. Sunil V. Deshmukh, Ashish Chavan. 2012. Six Sigma and SMEs: a critical review of literature.
International Journal of Lean Six Sigma 3:2, 157-167. [Abstract] [Full Text] [PDF]
12. Mehmet Taner. 2012. A Feasibility Study for Six Sigma Implementation in Turkish Textile SMEs. South
East European Journal of Economics and Business 7. . [CrossRef]
13. Mohamed Gamal Aboelmaged. 2011. Reconstructing Six Sigma barriers in manufacturing and service
organizations. International Journal of Quality & Reliability Management 28:5, 519-541. [Abstract] [Full
Text] [PDF]
14. T.Y. Lee, W.K. Wong, K.W. Yeung. 2011. Developing a readiness self‐assessment model (RSM) for Six
Sigma for China enterprises. International Journal of Quality & Reliability Management 28:2, 169-194.
[Abstract] [Full Text] [PDF]
15. José H. Ablanedo-Rosas, Bahram Alidaee, Juan Carlos Moreno, Javier Urbina. 2010. Quality improvement
supported by the 5S, an empirical case study of Mexican organisations. International Journal of Production
Research 48, 7063-7087. [CrossRef]
16. Rodney McAdam, Shirley‐Ann Hazlett. 2010. An absorptive capacity interpretation of Six Sigma. Journal
of Manufacturing Technology Management 21:5, 624-645. [Abstract] [Full Text] [PDF]
17. Carlos Henrique Mora Júnior, Edmilson Lima. 2010. Programa seis sigma em pequenas e
médias empresas: revisão e recomendações. Revista Eletrônica de Ciências Administrativas 9:10.5329/
RECADM.20100901, 19-34. [CrossRef]
18. Carlos Henrique Mora Júnior, Edmilson Lima. 2010. Programa seis sigma em pequenas e
médias empresas: revisão e recomendações. Revista Eletrônica de Ciência Administrativa 9:10.5329/
RECADM.2010.0901, 19-34. [CrossRef]
19. Mohamed Gamal Aboelmaged. 2010. Six Sigma quality: a structured review and implications for future
research. International Journal of Quality & Reliability Management 27:3, 268-317. [Abstract] [Full Text]
[PDF]
20. Farhad Nabhani, Alireza Shokri. 2009. Reducing the delivery lead time in a food distribution SME through
the implementation of six sigma methodology. Journal of Manufacturing Technology Management 20:7,
957-974. [Abstract] [Full Text] [PDF]
21. Julie Eatock, Dorian Dixon, Terry Young. 2009. An exploratory survey of current practice in the medical
device industry. Journal of Manufacturing Technology Management 20:2, 218-234. [Abstract] [Full Text]
Downloaded by OPEN UNIVERSITY At 12:14 13 January 2015 (PT)

[PDF]

View publication stats

S-ar putea să vă placă și