Sie sind auf Seite 1von 23

International Journal of Production Research, 2017

Vol. 55, No. 15, 4319–4340, https://doi.org/10.1080/00207543.2016.1249430

A new model to implement Six Sigma in small- and medium-sized enterprises


Taieb Ben Romdhanea , Ahmed Badreddineb∗ and Manel Sansaa
a LISI, Institut National des Sciences Appliquées et de Technologie, Centre Urbain Nord, Tunis, Tunisie; b LARODEC, Institut Supérieur
de Gestion de Tunis, Le Bardo, Tunisie
(Received 22 March 2016; accepted 5 October 2016)

The Six Sigma approach improves the quality of products in order to ensure customers’ satisfaction. This approach has
yielded to interesting results for large enterprises. However, its implementation remains difficult for small- and medium-
sized enterprises (SME). In fact, the use of the same tools is insufficient to achieve the objectives when considering financial
constraints and the lack of data. The regular tools are complex for SMEs which require an adapted model to implement the
approach successfully. In this paper, we propose a new model having the objective to facilitate the integration of Six Sigma
in SMEs by avoiding the use of Black Belts, optimising the implementation costs and period, simplifying the Six Sigma
structure and enhancing the communication between staff and managers. The model includes two imbricated loops: the first
offers immediate improvement actions by estimating the capability and the stability of the process, while the second provides
profound improvement actions using the fuzzy logic system and the analytic hierarchical process (AHP) method. An example
illustrates the application of the proposed model in an SME.
Keywords: Six Sigma; fuzzy logic; AHP; FMEA; statistical process control; SME

1. Introduction
The Six Sigma approach has been recognised as one of the most effective breakthrough improvement methods since it reduces
manufacturing defects. In fact, several enterprises have implemented this approach. General Electric and Delphi Automotive
claimed that the Six Sigma programmes have transformed their organisations (Treicher et al. 2002). In addition, Mahanti
and Antony (2009) conducted a survey among Indian software companies and concluded that the application of Six Sigma
improved the product performance, achieved greater productivity, reduced costs and increased customers’ satisfaction.
Despite the success of Six Sigma, implementing this approach remains reserved primarily for large enterprises because
it requires significant financial resources and the intervention of Black Belt experts. In fact, the hindrances and constraints
in a small- and medium-sized enterprise (SME) are more significant if we consider the lack of adequate financial resources
and manpower, the wrong choice of objectives, the lack of time to achieve the project and the lack of knowledge about Six
Sigma tools (Poznańska 2004). In addition, the SMEs may face difficulties while ensuring an effective implementation as
investing in training and hiring the Six Sigma experts who are vital for any implementation. To overcome these weaknesses,
this paper develops a new implementation model for Six Sigma adapted to SMEs. To this end, we propose to create two
reaction levels. The first is the process level, it warrants the capability and the stability of the process to continuously respect
the requirements of the customers and to reduce the sources of variability. The second is the management level, it determines
thorough improvement actions by combining the fuzzy logic system (Mamdani and Assilian 1975) and the AHP (Saaty 1980)
method. Thus, this proposed model is adapted to SMEs by encapsulating and optimising the Six Sigma selected tools, it aims
at avoiding the use of Blacks Belts, reducing the period and costs, simplifying the Six Sigma structure and enhancing the
communication between staff and managers. The remainder of this paper is organised as follows: Section 2 reviews the Six
Sigma approach. Section 3 presents a brief literature review on the Six Sigma implementation in SMEs. Section 4 describes
the proposed model and details the In-Process and the Out-Process loops. Section 5 details the implementation of the model
in an automotive company. Finally, Section 6 evaluates the efficiency of the proposed model through the experimental study.

2. Brief review of Six Sigma


The fundamental objective of the Six Sigma methodology is to satisfy customers’ requirements while increasing profits and
minimising wasted resources (Snee 1999). The DMAIC approach (i.e. define, measure, analyse, improve and control) is
considered as a guide for the implementation of Six Sigma (Starbird 2002). The five phases of the DMAIC approach are
detailed in Table 1.
∗ Corresponding author. Email: badreddine.ahmed@hotmail.com

© 2016 Informa UK Limited, trading as Taylor & Francis Group


4320 T.B. Romdhane et al.

Table 1. The five DMAIC phases and the appropriate tools for each phase.

The phases Objectives Required activities The tools

Identification of the sig- – Identification of the critical customers SIPOC,b QFD,c Bench Marking, MPMd
nificant long-term quality requirements known as CTQa
Define problems
– Identification of the objectives and
goals set by the organisation
– Selection of the appropriate projects
and team
– Development of an implementation
plan for a better project vision
– Identification of the project metrics
Measurement of the long- Mapping of the process and calculation Control Charts (Dryden et al. 1941), Capability
term problems of the process’s long-term capability indices (Hsiang and Taguchi 1985), Pareto
Measure
Diagrams, Histogram
Identification and verifica- Collection of the necessary data to Cause and Effect diagram, FMEAe , Correlation
tion of the root cause identify and validate the root cause of diagram, Hypothesis Testing
Analyse
variability that may occur in the process
Identification and validation Selection of the long-term improvement 5M,f SCAMPER,g Brain Storming,
Improve
of results actions that addresses the root causes
Poka Yoke (anti-error System)
Development of surveil- Update of the surveillance plan and SPCh techniques, performance indicators, doc-
Control lance plans implementation of control tools to en- umentation of the process
sure the sustainability of improvement
actions
a Critical to quality.
b Supplier input process output customer (Mishra and Kumar 2014).
c Quality function deployment.
d Multiple picking-up method.
e Failure mode effects analysis (USDoD 1980).
f Machine, method, material, manpower, measurement.
g Substitute, combine, adapt, modify, put to other purposes, eliminate, rearrange (Eberle 1971).
h Statistical process control.

The implementation of these phases is ensured by the Black Belts who are considered as the backbone of a Six Sigma
project. In fact, they are technically strong and they provide the appropriate tools package for the implementation. Their main
tasks consist in monitoring, selecting and deploying the appropriate tools and methods. They also ensure all the required
training and select the adequate improvement actions. Each one is expected to achieve the financial goal as a result of the
improvement through a Six Sigma initiative. However, the cost of hiring such experts is significant. In fact, the average
annual salary for a Black Belt is around 101.624 thousand dollars (Voehl et al. 2013). For instance, General Electric spent
1.6 billion dollars on its Six Sigma programme including 50 thousand dollars on each trained worker (General Electric 2002).
Although Six Sigma is applicable for both large and small organisations. The problem is that SMEs do not have adequate
resources to carry out the implementation effectively and they often expose themselves to several limitations (Hernandez
and Turner 2009). In this context, Raghunath and Jayathirtha (2013) presented several barriers for Six Sigma implementation
in SMEs, we can mention in particular the lack of adequate human and financial resources, the lack of leadership from
top executives, poor coaching and training, insufficient organisational alignment and wrong identification of the process
parameters. Moreover, Hoerl et al. (2001) pointed out that SMEs may have difficulties in hiring the Black Belts since they
require significant investments.
In addition, according to (Wurtzel 2008; Chakravorty 2009; Fursule, Bansod, and Fursule 2012), one of the reasons of
Six Sigma programme failure is the lack of a model to effectively guide the implementation. To overcome these problems,
several researches have studied the Six Sigma implementation in SMEs, as detailed below.

3. Review on Six Sigma implementation in SMEs


The researches of the Six Sigma approach and its implementation in SMEs can be divided into two categories:
International Journal of Production Research 4321

(1) The first one proposes a set of ideas and factors to improve the Six Sigma implementation. Thus, Wessel and Burcher
(2004) studied the requirements of Six Sigma implementation in a German SME. They concluded that the Six Sigma
approach needs to be adapted for an SME environment. In addition, Rowlands (2004) argued that the Black Belts
training and deployment is not beneficial in the case of SMEs, the implementation needs to use only the SME available
resources and skills. Also, Kumar et al. (2006) highlighted the difficulties relative to the implementation in SMEs
such as the resistance from employees when implementing new business strategies and the difficulty in convincing
top management. The study of Kumar (2007) was carried out in a UK SME, the author pointed out several critical
success factors for the implementation in SMEs. The findings of his study revealed that management involvement
and commitment, the poor training and resource availability were ranked as the highest impeding factors during the
implementation of the Six Sigma project. Moreover, Nonthaleerak and Hendry (2008) insisted on avoiding the use of
Black Belts for SMEs to enhance training and communication between top management and workers. Furthermore,
DeRuntz and Meier (2010) noted that the success of the Six Sigma implementation depends on involving the top
management, adopting clear strategies, motivating employees and training stakeholders. Moosa and Sajid (2010)
claimed that using Black Belts does not lead to efficient results without an adequate training and appropriate strategies.
Fursule, Bansod, and Fursule (2012) concluded that the benefits of implementing Six Sigma are the reduction of scrap
and rework rates, the improvement of quality and the reduction of the manufacturing costs. They insisted that the
implementation depends on the support of top management and staff, the availability of resources and the knowledge
about Six Sigma. In addition, Chakraborty and Chuan (2013) highlighted the importance of the top management
commitment and involvement for the project success and sustainability. Chakraborty and Leyer (2013) showed the
importance of linking Six Sigma to both strategic and operational levels. Finally, Jacobs, Swink, and Linderman
(2015) noted that early Six Sigma adoption in SMEs has more benefits and performance gains than late Six Sigma
adoption.
(2) Using ideas and results from the first category, the second one proposes a set of guidelines and frameworks to improve
the implementation of the Six Sigma project in SMEs. Vandenbrande (2005) simplified the Six Sigma structure in
function of the employees’ number. This structure has also simplified the training programmes using the available
knowledge. Besides, Antony (2006) presented several steps for Six Sigma implementation in SME service companies
through the phases of the DMAIC approach. The author suggested a grid of tools and techniques as guidelines for
the service sector to facilitate the choice of the appropriate tools for each phase. Moreover, Kumar, Antony, and
Tiwari (2011) proposed an implementation framework as a guideline for SMEs. The proposed framework includes
five steps namely: Readiness for Six Sigma, Prepare, Initialise, Institutionalise and sustain. This framework allows
the sustainability of the improvements by enhancing the motivation of employees and sharing the learning across the
company. Finally, Timans et al. (2014) made some improvements to Kumar, Antony, and Tiwari (2011) framework
by reducing the number of phases to three namely: Recognise and Prepare, Initialise and Institutionalise, Sustain.

4. Proposed model to improve the implementation of Six Sigma


The review of the literature, presented above, revealed that only few studies have proposed frameworks/models to implement
Six Sigma effectively in SMEs. Although they have reduced the number of phases (Timans et al. 2014), we remark that they
have ignored the discussion on how SMEs can implement and deploy the proposed frameworks with their limited resources.
Indeed, they are still based on the Black Belts to monitor, choose and deploy the appropriate tools. To overcome these
weaknesses, we propose a new Six Sigma implementation model adapted to SMEs that aims at:
(1) Avoiding the use of Black Belts by relying on the internal expertise of the process. In fact, involving internal resources
and guiding them through a concrete model reinforces the knowledge and the ability of learning.
(2) Optimising the implementation period and costs using appropriate tools and methods. In fact, the increased costs are
essentially due to the length of the project. The minimisation of this period depends on the pertinence of improvement
actions on the process and the management levels.
(3) Simplifying the Six Sigma structure by deploying a guided model which integrates selected, optimised and encap-
sulated tools. These latter reduce significantly the complexity in the measure, analyse and improve phases of the
DMAIC approach.
(4) Enhancing the communication between staff and managers by creating an interface that links and synchronises the
actions and the results of the process level with the improvement mechanism of the management level.
To satisfy these criteria, we propose a new framework to implement Six Sigma in SMEs as illustrated in Figure 1. Beginning
with the requirements of customers and the internal objectives of the organisation, we suggest an imbricated system with two
reactions levels. The first is the process level. The second is the management level. This characteristic allows the continuous
4322 T.B. Romdhane et al.

Definition of the Definition of the Six


internal objectives of Sigma objectives
the organization relative to customers

Improvement
Process actions for
objectives
Improvement
actions Out-
Process
Improvement
actions In-
Process
In-Process loop
Measuring the
capability and the Out-Process loop
stability and decide Measuring the
Measurement to intervene fast on performance of the
of the internal the process In-Process loop
objectives and and react with
the sustainable actions
performance for continuous
of the process improvement

Process Level Management Level

Figure 1. Conceptual framework of the proposed approach.

improvement of communication between the managers and staff. In fact, they need to share the same objectives and deploy
a common and coherent language for a better decision analysis. The model includes two simultaneous improvement loops:
the In-Process and the Out-Process loops. The first guarantees the capability and the stability of the process on the basis
of the customers’ requirements. The second provides profound improvement actions on the basis of the results of the In-
Process loop, of the performance of the process and of the degree of the achievement of the internal objectives. If the customer
requirements are satisfied and the objectives are achieved, the Out-Process is activated to increase them. Thus, the parallelism
between both loops makes a better reactivity while treating problems that cause the process’s inefficiency. This reactivity
minimises the implementation period and costs. In what follows, we will detail the In-Process and the Out-Process loops.

4.1 The In-Process loop


The In-Process loop sets immediate corrective actions from the control charts to guarantee the process capability and stability.
Generally, this phase uses the SPC techniques which require specific skills for the interpretation of data relative to the stability
and capability of the process. The In-Process loop is illustrated in Figure 2.
The first step is the definition of the model’s inputs. The second consists of the surveillance of the process by implementing
the Mean/range control charts. Next, we calculate the capability C P and the stability S P indices. These indices are used as
inputs for the decision matrix to select the appropriate decision Di . The implementation’s steps are detailed below.

4.1.1 Implementation of control charts


The control charts detect the different states of the process via statistical parameters (Mean values, Standard deviations
and Ranges). They anticipate failures. The well known are the Shewhart charts (Dryden et al. 1941). The dispersion of
measures between or beyond the control and warning limits provides information regarding the capability and the stability
of the process. Generally, these charts are used if the SME produces large-batch products. An example of a control chart is
illustrated in Figure 3(a). However, in the case of small-batch products, we can achieve a 100% inspection and a process
monitoring using natural limits rather than the limits of tolerance (see Figure 3(b)). Indeed, this method has the advantage
of confirming the quality of products already manufactured and detecting more quickly the variability of the manufacturing
process (Stamatis 2002).

4.1.2 Estimation of the stability index S P


The stability denoted S P is estimated by analysing the control charts (see Figure 3(a)). We present in Table 2 the different
situations of instability and the corresponding decisions based on Nelson (1984).
International Journal of Production Research 4323

Requirements

Control Process
charts

Measuring the Estimating the stability


capability index Cp index Sp

Determination of
decisions Di (DS:
Analysing the situation of the
Decision Shutdown, DA:
process
matrix Adjustments, DN: Non
interventions required)

Need for
intervention

Figure 2. The flow chart of the In-Process loop.

4.1.3 Calculation of the capability index C P


The capability index provides an indication about the process performance respecting the control limits relative to the
customers’ requirements. Generally, the capability index denoted C P is the ratio of the dispersion to the range of tolerance
(Hsiang and Taguchi 1985; Kane 1986):
U C L − LC L
CP = (1)

where UCL (resp. LCL) is the upper control limit (resp. lower control limit). Once the capability index is computed, each
interval of values represents a state of capability (Tsai and Chen 2006). The set of the possible values is divided into five
states (see Table 3) with the recommended actions relative to each state.

4.1.4 Analysis of the process situation and determination of decisions Di


The capability and the stability indices are considered as inputs for the decision matrix to detect accurate and immedi-
ate decisions. However, if we combine the capability and stability criteria, only four possible situations can be defined
(see Table 4).
In some cases, this may result in unsuitable decisions for instant adjustment or process shutdown. To overcome this
problem, we extend the combination matrix using additional states (i.e. Capable/Almost Capable/Incapable; Stable/Almost
Stable/Unstable) as detailed in Table 5. On the basis of the proposed matrix, we can select one of these three decisions:
shutdowns (D S ), adjustments (D A ) or non-interventions required (D N ).

4.2 The Out-Process loop


The Out-Process loop generates pertinent actions on the process and the management levels to reduce the frequency of
disruptions. A precise quantification and conjunction of the frequency parameters reduce significantly the period and the
costs of a Six Sigma project. To this end, this loop offers three levels of surveillance. The first level calculates the frequency
of the In-Process loop decisions which are the number of interventions on the process while the second one measures the
results’ efficiency in terms of quality, time and cost. At this stage, we compute the frequency of disruptions (FD ) using
the fuzzy logic system. Next, we calculate the severity index, denoted S, using the overall equipment effectiveness (OEE)
(Nakajima 1988). The combination of FD and S allows us to initiate improvement actions selected through the deployment
of the AHP method. Thus, this treatment level represents an implicit implementation of FMEA method that we called FMEA
ONLINE given its real-time nature compared to a classical FMEA that requires prior resources and studies for its completion
such as Pareto diagrams and Fault tree analysis. Through this mechanism, we considerably simplify the implementation
4324 T.B. Romdhane et al.

(a) A Action zone Upper control limit

Variable
B Warning zone
Upper warning limit

C Stability zone

Centreline: CL
CL: Centreline
Stability zone : Standard deviation
C
Lower warning limit

B Warning zone
Lower control limit
A Action zone Time

(b)
Upper natural limit CL + 3
Variable

Centreline: CL
CL: Centreline
: Standard deviation

Upper natural limit CL - 3

Time

Figure 3. Example of a control chart.

of FMEA tool, we guarantee a substantial time saving and relevant improvement actions that are based on significant data
calculated in real time. The third level aims at achieving the internal objectives. It consists of dashboards and management
key performance indicators. Figure 4 illustrates the Out-Process loop which operates as follows.
(1) Calculating the frequencies of occurrences relative to the decisions of:
• adjustments (D A ) denoted by FD A .
• shutdowns (D S ) denoted by FDS .
(2) Aggregating FD A and FDS to estimate the overall frequency FD using the fuzzy logic system,
(3) Computing the severity S using the OEE index.
(4) Selecting a set of actions (A j ) to ensure profound improvement of the process using the AHP Method.

4.2.1 Calculation of FD A and FDS and estimation of the overall decision frequency FD
The estimation of FD is mainly based on FD A and FDS which are calculated on the basis of the decisions generated from the
In-Process loop. However, the combination between FD A and FDS is generally pervaded with uncertainty. To overcome this
problem, we suggest to use the fuzzy logic system. The first step is to convert the input values (i.e. FD A and FDS ) and the
output (i.e. FD ) to fuzzy sets (Zadeh 1965). To this end, we transform each frequency to lexical parameters of low, medium,
high and very high. The fuzzy sets relative to the inputs and the output are defined by trapezoidal and triangular membership
functions, they are known for their simplicity and concise notation (Yang and Hsieh 2009). Their lexical expressions are
defined in Equations (2) and (3).


⎨0 if x ≤ a and x ≥ c
μMedium,High (x) = b−ax−a
if a ≤ x ≤ b (2)

⎩ c−x
c−b if b ≤ x ≤ c
International Journal of Production Research 4325

Table 2. Judgement from samples distribution.

The process’s situation Decision

All the points are placed between the control limits and on each side of the centreline:
Non interventions required

Process under control

The last point has crossed the control limit: Adjust process
Upper limit case: short-term capability deteriorates. We must find its origin and intervene
Lower limit case: short-term capability improves. We must block the measuring system

Point off limits

Seven or more consecutive points are on the same side of the centreline: adjust process
Higher tendency case: short-term capability deteriorates. We must find its origin and intervene
Lower tendency case: short-term capability improves. We must find its origin to maintain this
improvement

Higher/lower tendency

Seven successive points are either on regular increase or decrease: Adjust process
Increasing series case: short-term capability deteriorates. We must find its origin and intervene
Decreasing series case: short-term capability improves. We must find its origin to maintain this
improvement

Increasing/decreasing tendency

The last point is near the control chart: Check by taking another sample
If the point goes back to the middle third: non-interventions required
If the point is near or off the limits: Adjust process
Upper limit case: Supervise the capability. If many points are near the upper limit, the capability
deteriorates. We must find its origin and intervene

Single point near limits

Table 3. Interpretation rules of the capability index C P .

CP Interpretation Recommended actions

C P > 1.67 Highly capable No concerns


1.33 < C P ≤ 1.67 Capable Comfortable situation
1.00 < C P ≤ 1.33 Almost capable Possibility of non-conforming products
0.67 < C P ≤ 1.00 Incapable Existence of non-conforming products
C P ≤ 0.67 Chaotic Shutdown, analysis of causes and corrective actions



⎪ 0 if x ≤ a  and x ≥ d 

⎨ x−a 
b −a  if a  ≤ x ≤ b
μLow,VeryHigh (x) = (3)


⎪ 1 if b ≤ x ≤ c
⎩ d  −x
d  −c if c ≤ x ≤ d 
4326 T.B. Romdhane et al.

Table 4. Combination matrix: capability/stability.

States of the process Description

(1) Represents the comfortable situation, 100% of products conform to


requirements
(2) Represents the limit situation. We must reduce the dispersion or
review the limits
(3) Represents the situation at the edge of chaos. The special causes must
be defined and the process requires the stability state
(4) Represents the chaotic situation. The profound improvements must
be ensured to stabilise the process

Table 5. The proposed improved combination matrix.

Capabilities spotted on the previous control charts


Capable process Almost capable process Incapable process

Decisions on the process Di

Stable process Under control D N : non-interventions D N /D A : capability D A /D S : capability


required improvement improvement
Almost stable process Out of control but all the D A /D S : identify and eliminate special causes
points are between the
control limits
Unstable process Out of control and one or D A /D S : identify and eliminate special causes
many points are beyond
the control limits

where a and c (resp. a  and d  ) locate the feet of the triangle (resp. trapezoid) and b locates the peak (resp. b and d  locate
the peaks). The following step is the identification of the rules and the inference mechanism. These rules are if-then true
rules which are combined by an inference mechanism to map the input values into the output variable. The rules are in the
form: If FD A is Low and FDS is High, then FD is High. Then as inference method, we propose to apply the max–min method
(Zimmermann 1996). The two main steps of this method are:
• Computing the minimum between the discrete value of the membership function relative to FD A and the function
relative to FDS for each rule.
• Computing the maximum of all rules to obtain the resulting membership function of the output variable.
After the inference, we should defuzzify the output to crisp value using the centroid method expressed by:

xμ (x)dx
FD =  F D (4)
μ F D (x)dx
where x = [0, 1] is the universe of discourse and μ F D is the global fuzzy set.

4.2.2 Calculation of the severity S


During the manufacturing cycle, many incidents may occur such as equipment shutdown, reworks, scraps and delivery delays.
Thus, our idea is to calculate the severity S using the OEE to obtain concise information regarding the process variability.
This indicator depends on the availability index (A V ), the performance rate index (PR) and the quality rate index (QR):

O E E = P R × AV × Q R (5)

In order to compute the OEE, we should follow these steps (Singh et al. 2013):
(1) Input the value of Total Shift Time
International Journal of Production Research 4327

Requirements
and objectives

Process

In-Process

FDA Estimation of FD
the overall
FDS
frequency
Fuzzy FMEA
PR Logic ONLINE
Calculating Selection of
Av the OEE as improvement
S
severity index actions Aj
QR

O1
Measuring internal objectives AHP
Ok method

Figure 4. The flow chart of the Out-Process loop.

(2) Input the value of Planned Down Time.


(3) Compute the Loading Time = Total Shift Time − Planned Down Time
(4) Calculate the Operating Time = Loading Time − (All Down Time + All Stop Time)
(5) Calculate A V as follows:
Operating time
AV = (6)
Loading time

(6) Compute the Actual Cycle Time, the Theoretical Cycle Time, the Actual Processing Time and the Total Amount
Produced.
(7) Calculate the Operating Speed Rate = Theoretical Cycle Time/Actual Cycle Time, and the Net Operating Rate =
Actual Processing Time/Operating Time
(8) Calculate PR as follows:

P R = Net operating Rate × Operating Speed Rate (7)

(9) Calculate the value of Defect Amount


(10) Calculate QR using:
Total Amount Produced − Defect Amount
QR = (8)
Total Amount Produced
Once all the outputs are calculated (i.e. FD , S), we propose to select profound actions to improve the efficiency of the
process.

4.2.3 Selection of improvement actions Ai


At this stage, the organisation must consider its internal and required objectives as well as FD and S using a multi-criteria
method which is the most suitable for such problem. Many multi-criteria methods are defined in the literature. However,
the AHP method is commonly used because it decomposes the problem into a multi-level hierarchical structure. The first
level corresponds to the goal, the intermediate level to the different criteria and their sub-criteria, and the last one to the
alternatives. For each level except the first, we set decision matrices for the criteria and the alternatives based on a pair-wise
comparison. The hierarchical structure is illustrated in Figure 5. The criteria and the alternatives are set by the organisation.
First, we set the decision matrix relative to the criteria by defining the degree of importance Fi j between the n criteria
using Saaty’s scale of measurement (Saaty 1980) and on the basis of the internal expertise of the organisation. Second, we
calculate the weight of each criterion denoted WCi using Equation (9). Then, we should ensure the coherence of the matrix
by calculating its consistency ratio (CR) using Equation (10).
4328 T.B. Romdhane et al.

Selection of the
improvement actions

O1 O2 Ok FD S

Alternative 1 Alternative 2 Alternative 3 Alternative m

Figure 5. The proposed hierarchical structure of the AHP method.

n F
n i j
j=1 Fi j
WCi = i=1
(9)
n
CI
CR = (10)
RI
λmax − n
CI = (11)
n−1
where λmax is the maximum eigenvalue of the decision matrix and RI is the random index defined by Saaty (1980) according
to the number of criteria. If CR is lower than 10%, the decision matrix is incoherent and we must review the judgements.
The judgement of the coherence through the CR index minimises implicitly the intervention of the Black Belts due to the
involvement of internal expertise in defining the judgement matrices. Then, we proceed with the same steps to calculate the
weight relative to each alternatives A j regarding each criterion Ci denoted W A j /Ci .
Next, we compute the global scores relative to each alternative W A j as follows:


m
WA j = W A j /Ci × Wci (12)
j=1

where m is the number of alternatives. The last step is to select the best alternative AAHP (i.e. max j (W A j )).

4.3 Compliance of the proposed model with the classical DMAIC approach
In this subsection, we propose a conceptual pre-validation of the model, which involves a comparative analysis with regard
to the DMAIC approach while illustrating the proposed methodology and the used tools (see Table 6). In fact, we can remark
that the phases of DMAIC approach are respected. Besides, the proposed tools and methods are similar to those used in a
classical Six Sigma. We can mention in particular, the SPC techniques and the control charts. Our contribution is marked
with its simplification, synchronisation and by implementing an intelligent system, which combines the fuzzy logic system
and the AHP method for the analysis and the support for the appropriate decision.

5. Case study
The case study is an SME of automobile components located in Tunisia. The company was created in 1999 employing
approximately thirty people. The company manufactures different wires to command airbags for vehicles. The customer
requirements are strict on terms of quality, deadlines and costs. This company’s objective is to implement a Six Sigma
project. To this end, the organisation has agreed to implement our model. Thus, the monitoring team has been selected
as follows: the top manager, the production managers and the manufacturing staff. The technical drawing of the wires
representing the dimensions is illustrated in Figure 6. The manufacturing block of this process is illustrated in Figure 7. The
critical to quality (CTQs) points identified in Table 7 are the dimensions required by the customer (see Figure 6).
The internal objectives set by the organisation are:
• O1 : Increase of profits.
• O2 : Satisfaction of customers.
• O3 : Diversification of product range.
International Journal of Production Research 4329

Table 6. The conceptual pre-validation of the proposed model.

The DMAIC
approach phases Proposed model Tools and methods
Define Definition of the objectives and the customers’ requirements Brainstorming

Surveillance of the process SPC techniques, Control charts


In-Process
Measurement of the capability and
stability on the basis of the customers’
Measure
requirements
Measurement of the effectiveness of the OEE (Quality, Availability, Performance
process
Out-Process
Estimation of the In-Process interven- The Fuzzy Logic System
tions occurrence (the shutdowns and the
adjustments frequencies of the process)
Estimation of the achievement level of Dashboard and performance indicators
the internal objectives

In-Process Analysis of the capability and stability Interpretation guide for the control charts
Analyse
of the process to achieve the customers’
requirements
Out-Process Analysis of the process situation FMEA using the frequencies of incidents and
their severities

In-Process Immediate improvement decisions to Decision matrix capability/stability


Improve
adjust the process and improve its
capability and stability
Out-Process Improvement of the requirements and AHP method
objectives level

Control Standardisation through the documentation of the process Guides to apply, control and improve

5.1 Implementation of the In-Process loop


First, we have considered the requirement crimping height as the input (see Table 7) and we have implemented the control
charts as illustrated in Table 8. For instance, the UCL and LCL are calculated from the control charts and they are equal,
respectively, to 0.7357 and 0.7043. If the standard deviation σ is equal to 0.0124 then C P is equal to 0.81 using Equation (1)
and the process is incapable according to Table 4. Concerning the S P index, we can remark from the control chart that many
measure points are near the upper control limit. Thus, the process is out of control (see Table 2). Therefore, the process is
Almost Stable (see Table 5).
The evolution of the In-Process loop implementation is observed weekly through six months as shown in Table 9. The
appropriate decisions have been selected from the proposed combination matrix (see Section 4.1.4). The capability and the
stability of the process have been improved after 12 weeks due to the activation of the Out-Process loop after the first week
and the adjustments in the In-Process loop in the remaining 11 weeks. From week 13 to week 24, the In-process loop has
not been activated. However, the process is capable and stable due to the profound improvement actions of the Out-Process
loop.

5.2 Implementation of the Out-Process loop


At this stage, we have calculated FDS and FD A relative to the decisions selected at the end of the In-Process loop as follows:
• FDS = 0.06
• FD A = 0.48
These frequencies are the inputs of the fuzzy logic system. First, the organisation has set the parameters of the membership
functions (see Table 10).
4330 T.B. Romdhane et al.

Figure 6. The technical drawing of the airbags wires.

Input cables and wires materials and components

Incoming
Inspection

1 2
Cutting/Splitting Cutting

Metric Control 4 6
Cut’s Length Splitting/Crimping Mounting
sheaths
Dynamic inspection
Visual inspection 7
Mounting
components
3
Splitting Electrical
inspection.
Metric
5 inspection
Welding 8
Mounting
sheaths

Supplying Incoming inspection

Storage Intermediate Visual


inspection Inspection
Self-control Final inspection 9
operation Short intermediate Packaging
storage

Figure 7. Manufacturing bloc of Airbag wires.

On the basis of these parameters and using Equations (2) and (3), the membership functions relative to the inputs and the
output are illustrated in Figures 8 and 9, respectively.
As shown in Table 11, we have set the fuzzy rule base on the basis of the internal expertise of the process and we have
chosen to apply the four rules highlighted in gray. Next, as presented in Appendix 1, the max–min inference method has
been applied to obtain the global fuzzy set (see Figure 10). Finally, the defuzzification is obtained by calculating the centre
of gravity of the global fuzzy set using Equation (4). Thus, FD is equal to 0.651.
The second step is to calculate the severity S from the OEE indicator. In fact, the total shift time of the company is
480 min, its planned downtime is estimated to 43 min, and the actual processing time is 352 min. The total downtime and
stop time are, respectively, 44 and 15 min. The theoretical cycle time of the process is 4 min. However, the actual cycle time
is 5 min. The company produces 80.000 m per day of which 2000ṁ are rejected. The loading time is equal to 437 min, the
operating time is equal to 378 min, the operating speed rate is equal to 80% and the net operating rate is 96%. Referring to
International Journal of Production Research 4331

Table 7. The identified CTQs.

Designation Dimension (mm)

L1 Total length 2240 + 20


L2 Unsheathed length 173 + 5
L3 Length MQS side 40 + 5
L4 Length MQS side 25 + 5
L5 Crimping height 0.72 ± 0.03

Table 8. Monitoring by the Mean/range control chart.

Mean/range chart Histogram of crimping height

Mean value = 0.7228 σ = 0.0124 The process is Almost Stable and Incapable

Mean value = 0.7194 σ = 0.01162 The process is Almost Stable and Almost Capable

Mean value = 0.7197 σ = 0.010 The process is Almost Stable and Incapable

these inputs, A V is equal to 86% using Equation (6), PR is equal to 77% using Equation (7) and QR is equal to 98% using
Equation (8). Multiplying these indices, the OEE is equal to 65%. On the basis of the OEE value and the internal expertise of
the process, we have proposed a set of intervals to estimate S as shown in Table 12. We can, hence, deduce that S is medium.
The third step is to select the appropriate improvement actions. First, we have defined five criteria (i.e. C1 : Profits (O1 ), C2 :
Customers’ satisfaction (O2 ), C3 : Diversification (O3 ), C4 : FD and C5 : S) and four alternatives (i.e. A1 : Implementation of an
anti-error system POKA YOKE, A2 : Training employees, A3 : Changing material supplier and A4 : Reviewing maintenance
range). Regarding the achievement level of the objectives Oi , the organisation has defined the set of intervals for the
achievement level of the objectives as presented in Table 13.
On the basis of the defined criteria and alternatives, the proposed hierarchical structure is illustrated in Figure 11.
Then, we have proceeded with the pair-wise comparison and we have calculated the weights relative to each criterion
using Equation (9) (see Table 14). For instance, the weight relative to FD is equal to 0.06487.
From this matrix, we have calculated CR using Equation (10). Since we have five criteria, RI is equal to 1.12. λmax =
5.432098 and C I = 0.108025, thus, CR is equal to 0.096451 ≤ 10%. Therefore, the decision matrix is coherent and we have
proceeded with checking the consistency of the alternatives matrices relative to each criterion. Then, we have calculated the
weights of A j relative to Ci and the global scores using Equations (9) and (12) as presented in Table 15.
According to Table 15, A3 : changing material supplier is the priority improvement action followed by A4 , A1 and
A2 . The results of the Out-Process loop are synthesised in Appendix 2. In fact, we can remark that this loop has been
activated six times, each time different improvement actions have been proposed to ensure the sustainability of the process.
The most appropriate actions have been selected on the basis of the global priority vector column calculated by the AHP
method.
4332 T.B. Romdhane et al.

Table 9. Evolution of the In-Process loop in our case study.

N◦ of weeks CP SP Di : DS,A,N Comments

W1 0.52: Incapable Unstable Suspending production: changing tools and Out-Process Loop
adjusting measuring instruments.
W5 0.8: Incapable Almost Stable Suspending production: Non-conforming prod- Nothing to report
ucts. Changing tools and adjusting measuring
instruments
W11 0.97: Almost Capable Almost Stable Repetitive adjustments. Non-conforming Nothing to report
products.
W13 1.27: Capable Stable No Adjustments. Improving the capability of the Out-Process loop
process
W17 2.2: Capable Stable No Adjustments. Improved capability Nothing to report
W24 4.04: Highly capable Stable Few adjustments. A significant improvement of Nothing to report
capability

Table 10. The proposed fuzzy sets intervals.

Membership functions
Low Medium High Very high

a = 0 a = 0.0375 a = 0.1125 a  = 0.225


b = 0 b = 0.1125 b = 0.1875 b = 0.2813
FDS 
c = 0.01875 c = 0.1875 c = 0.2625 c = 1
d  = 0.075 d = 1
a = 0 a = 0.125 a = 0.375 a  = 0.625
b = 0 b = 0.3125 b = 0.5625 b = 0.8125
FD A 
c = 0.0625 c = 0.5 c = 0.75 c = 1
d  = 0.25 d = 1
a = 0 a = 0.125 a = 0.375 a  = 0.625
b = 0 b = 0.3125 b = 0.5625 b = 0.8125
FD 
c = 0.0625 c = 0.5 c = 0.75 c = 1
d  = 0.25 d = 1

Table 11. The proposed inference rule base.

FDS
FD Low Medium High Very high

Low Low Medium High Very high


Medium Medium High Very high Very high
FD A
High High Very high Very high Very high
Very high Very high Very high Very high Very high

5.3 Evaluation of the case study


The results of our model’s implementation during 24 weeks are summarised in Figure 12. It details the evolution of:
• The achievement level of the objectives O1 , O2 and O3 (see Figure 12(a)) using three linguistic terms namely, low,
medium and high as detailed in Table 13.
• The level of the severity S (see Figure 12(b)) using three linguistic terms namely, low, medium and high as detailed
in Table 12.
• The overall decision frequency FD (see Figure 12(c)) on the basis of the capability C P and the stability S P .
Indeed, we note the synchronism and the parallelism between the In-Process loop and the Out-Process loop in order to meet
the customers’ requirements and to achieve the objectives. Furthermore, we can remark that the Out-Process loop has been
International Journal of Production Research 4333

FDS

FDA

Figure 8. The proposed membership functions of, respectively, FDS and FD A .

FD

Figure 9. The proposed membership function of FD .

Low Medium High Very High

Figure 10. The global fuzzy set of FD .

activated two times (weeks 2 and 14) to boost the In-Process loop and five times (Weeks 2, 7, 9, 17, 22) to improve the
achievement level of the objectives O1 , O2 and O3 . In fact, during week 2, the improvement action was ‘Studying and
changing the cutting blades types’ (see Appendix 2). This action has given effect in week 7 since FD has decreased from 77
to 8% and the profits O1 and the satisfaction of customers O2 have reached the medium level. Then, due to the instability and
the lack of the process capability knowing that FD A and FDS were low, the In-Process loop became inefficient. Therefore,
4334 T.B. Romdhane et al.

Selection of the appropriate


improvement actions

Profits Customers’ satisfaction Diversification FD S

Implementation of an anti-error Training Changing Reviewing


system POKA YOKE Employees Material supplier maintenance range

Figure 11. The proposed hierarchical structure.

Table 12. Estimation of S from the OEE.

OEE S

O E E > 70% Low


40% < O E E ≤ 70% Medium
0% < O E E ≤ 40% High

Table 13. Estimation of the achievement level of the objectives.

Oi Achievement level

0% < Oi ≤ 50% Low


50% < Oi ≤ 80% Medium
Oi > 80% High

Table 14. The proposed criteria decision matrix.

FD S Profits (O1 ) Customers’ satisfaction (O2 ) Diversification (O3 ) WC i

FD 1 1/ 1/ 1/ 3 0.06487
4 6 7
S 4 1 4 1/ 7 0.305925
2
Profits (O1 ) 6 1/ 1 1/ 7 0.204482
4 2
Customers’ satisfaction (O2 ) 7 2 2 1 8 0.390618
Diversification (O3 ) 1/ 1/ 1/ 1/ 1 0.034104
3 7 7 8

the Out-Process loop has been activated again in week 7 and the priority action was to set up predictive maintenance of the
cutting machine. This action has reduced the unavailability of the crimping machine and has improved the quality of products
which increased O1 and O2 . However, their achievement level remained medium. During week 9, the OEE indicator has
decreased because of disruptions in the production rates which led to a high severity S and no increase in the level of the
profits O1 . To deal with, a Computer-Aided Production Management connected to the crimping machine was implemented
to improve the pace of production. This action allowed the increase in the profits level O1 from medium to high in week 21.
In parallel with the cited actions, the In-Process loop has become more efficient and has increased the stability S P and
the capability C P . Nevertheless, the rate of non-compliant products was unstable. It is in week 13 that the Out-Process loop
allowed to initiate the setting up of a Poka Yoke system (Shingo and Dillon 1989). It reduced the operators’ assembly defects.
By week 17, the process was perfectly stable and capable. The Out-Process loop has focused on improvement actions that
have been materialised by the implementation of a 5S project (Osada 1995).
It is important to note that the profits O1 (resp. the customers’ satisfaction O2 ) was completely satisfied from week 22
(resp. week 13). However, we can remark that the objective O3 remained low during the 24 weeks (see Figure 12(a)). In fact,
it was difficult to invest for the design and the manufacturing tools, because as shown in Table 14 the implementation model
International Journal of Production Research 4335

Figure 12. Implementation timeline of the proposed model.


4336 T.B. Romdhane et al.

Table 15. The weights of the different alternatives.

Alternatives W A j /C1 W A j /C2 W A j /C3 W A j /C4 W A j /C5 Global score

A1 0.111202 0.156739 0.247824 0.32 0.07 0.233368


A2 0.064987 0.049523 0.143852 0.152859 0.443248 0.123607
A3 0.267759 0.337322 0.08555 0.475169 0.089603 0.326724
A4 0.556053 0.456416 0.522774 0.051269 0.400999 0.316301

Table 16. The operational validation of the proposed model.

Criteria The case study results Estimated results from a Sensitivity analysis
classical Six Sigma imple-
mentation

Avoiding the No Black Belt hired At least one Black Belt hired Due to the complexity of their
use of Black products and processes, some
Belts SMEs require the intervention
of the Black Belts. However,
their number will be reduced
compared to the classical
approach
06 months 09 months or more
Optimising the Since we could not estimate the costs of a classical Six Sigma implementation The gain depends on the
period and costs for such SME, we were based on the calculation of gains due to the reduction of dimension of the Six Sigma
of implementa- the period (33%), avoiding the use of Black Belts (100%), the reactivity and the project
tion relevance of the improvement actions (50%). The reduction of costs is estimated
to 25%
Simplifying Compliance with the DMAIC approach. A Compliance with the The use of complex tools
the Six Sigma model application guide was standardised. DMAIC approach such as SPC techniques and
structure Supports for the use of control charts. A classical FMEA make the
more precise matrix for the joint evaluation analysis and improve phases
of the process’s stability and capability. The difficult to achieve for an
implementation of a synchronised FMEA SME with limited resources.
ONLINE with the process allowed a satisfactory A minimum internal expertise
efficiency and reactivity despite the limited of the process is required.
resources A software tool is essential
for calculations and decision
support. The culture of the
automobile companies was a
determining factor for the
success of six sigma project
Enhancing the The In-Process loop was activated three times to The number of improve- If the number of objectives
communication improve the process and three times to reach the ment actions is very limited and improvement actions in-
between staff management objectives. The Out-Process loop during six months in a clas- creases significantly, the deci-
and top mana- was activated six times to improve the stability sical Six Sigma implemen- sion becomes more complex
gement and the capability. The internal expertise and tation. The improvement
the effective communication with the managers actions may be ineffective
allowed to identify relevant and appropriate if the process staff is not
actions involved and the objec-
tives are not shared with
managers

has given priority to O1 and O2 (i.e. W O1 = 0.204482, W O2 = 0.390618 and W O3 = 0.034104). It is only in week 22 that
the Out-Process has generated the action of purchasing a configurable machine with more features in order to achieve the
diversification objective O3 .
International Journal of Production Research 4337

6. Evaluation of the proposed model


The implementation of the proposed model in the automotive company has demonstrated its effectiveness. The company was
able to achieve its internal objectives and to increase the satisfaction of the customers due to the performance of the selected
improvement actions deployed in a short period of time. The parallelism between the In-Process and the Out-Process loops
has led to the involvement of staff and managers and to the reduction of the period. In fact, the company has achieved its
implementation within 24 weeks without the assistance of Black Belt experts. The key of this success was the expertise of
the staff combined with appropriate, simplified and encapsulated tools. The costs of the implementation were relative to the
investments required in order to satisfy the customers and to achieve internal objectives.
Thus, an operational validation of the proposed model has been synthesised in Table 16 analysing the results with regard
of the four criteria defined in Section 4 and of estimated results from a classical Six Sigma implementation. A sensitivity
analysis has been also presented to calibrate the proposed model.

7. Conclusions
In this paper, we have proposed a new Six Sigma implementation model in SMEs. Our contribution is observed through
upgrading the Six Sigma approach to a continuous improvement adapted model coherent with the DMAIC approach. This
model has treated the requirements of customers and the internal objectives of the organisation. First, in the Measure phase,
we have extended the combination matrix of capability and stability from four to nine situations to make suitable decisions
as adjustments or process shutdown. Second, we have estimated the FMEA parameters (i.e. FD and S) in the Analyse phase
using the fuzzy logic system for the overall frequency FD and the OEE for the severity S to overcome the problem relative
to the uncertainty of the process failure modes. Then, we have used the customers satisfaction objective and the internal
objectives (i.e. Profits, diversification) combined with the FMEA parameters as criteria of the AHP method in the Improve
phase to facilitate the selection of profound improvement actions. Finally, we have evaluated our proposed model on the basis
of estimated results of a classical Six Sigma implementation and we have presented a sensitivity analysis in order to calibrate
the proposed model. However, this evaluation is only based on estimated results since the Six Sigma implementation is not
standard and depends on the enterprise’s characteristics. The proposed model is easy to implement due to the encapsulation
of tools and methods. In addition, we have proposed the FMEA ONLINE to move from the process level to the management
level. The results of the case study have demonstrated that the organisation has achieved all its objectives and has generated
profits while ensuring its customers satisfaction. As future works, we will apply our model as part of the Lean Six Sigma
approach. Furthermore, together with customers requirements, we will treat objectives related to eliminating waste or MUDA
(Ohno 1988). In the same direction, we will implement another layer that incorporates the aspect of product design using
DFSS (Design For Six Sigma) combined with a risk management system since early design phase.

Disclosure statement
No potential conflict of interest was reported by the authors.

References

Antony, J. 2006. “Six Sigma for Service Processes.” Business Process Management Journal 12 (2): 234–248.
Chakraborty, A., and T. K. Chuan. 2013. “An Empirical Analysis on Six Sigma Implementation in Service Organisations.” International
Journal of Lean Six Sigma 4 (2): 141–170.
Chakraborty,A., and M. Leyer. 2013. “Developing a Six Sigma Framework: Perspectives From Financial Service Companies.” International
Journal of Quality & Reliability Management 30 (3): 256–279.
Chakravorty, S. S. 2009. “Six Sigma programs: An Implementation Model.” International Journal of Production Economics 119 (1): 1–16.
DeRuntz, B., and R. Meier. 2010. “An Evaluative Approach to Successfully Implementing Six Sigma.” Technology Interface Journal
10 (3): 29–38.
Dillon, A. P., and S. Shingo. 1985. A Revolution in Manufacturing: The SMED System, CRC Press.
Dryden, H. L., T. Von Karman, A. A. Kalinske, T. K. Sherwood, S. S. Wilks, W. A. Shewhart, L. E. Simon, and R. Pound. 1941. Fluid
Mechanics And Statistical Methods in Engineering. Philadelphia: University of Pennsylvania Press.
Eberle, R. 1971. Scamper On: Games for Imagination Development. New York: Buffalo.
Fursule, N. V., S. V. Bansod, and S. N. Fursule. 2012. “Understanding the Benefits and Limitations of Six Sigma Methodology.” International
Journal of Scientific and Research Publications 2 (1): 1–9.
General Electric, Company. 2002. GE Investor Relations Annual Reports.
Hernandez, J., and T. Turner. 2009. “Implementing Six Sigma in Challenging Times: A Case Study.” In IFIP International Conference on
Advances in Production Management Systems, 514–521. Springer.
4338 T.B. Romdhane et al.

Hoerl, R. W., D. C. Montgomery, C. Lawson, and W. E. Molnau. 2001. “Six Sigma Black Belts: What Do They Need to
Know?/Discussion/Response.” Journal of Quality Technology 33 (4): 391.
Hsiang, T., and G. Taguchi. 1985. “A Tutorial on Quality Control and Assurance the Taguchi Methods.” In ASA Annual Meeting. Las Vegas
Nevada, USA.
Jacobs, B. W., M. Swink, and K. Linderman. 2015. “Performance Effects of Early and Late Six Sigma Adoptions.” Journal of Operations
Management 36: 244–257.
Kane, V. E. 1986. “Process Capability Indices.” Journal of Quality Technology 18 (1): 41–52.
Kumar, M. 2007. “Critical Success Factors and Hurdles to Six Sigma Implementation: The Case of a UK Manufacturing SME.” International
Journal of Six Sigma and Competitive Advantage 3 (4): 333–351.
Kumar, M., J. Antony, R. K. Singh, M. K. Tiwari, and D. Perry. 2006. “Implementing the Lean Sigma Framework in An Indian SME: A
Case Study.” Production Planning and Control 17 (4): 407–423.
Kumar, M., J. Antony, and M. K. Tiwari. 2011. “Six Sigma Implementation Framework for SMEs-A Roadmap to Manage and Sustain the
Change.” International Journal of Production Research 49 (18): 5449–5467.
Mahanti, R., and J. Antony. 2009. “Six Sigma in the Indian Software Industry: Some Observations and Results from a Pilot Survey.” The
TQM Journal 21 (6): 549–564.
Mamdani, E. H., and S. Assilian. 1975. “An Experiment in Linguistic Synthesis with a Fuzzy Logic Controller.” International Journal of
Man-Machine Studies 7 (1): 1–13.
Mishra, P., and S. R. Kumar. 2014. “A Hybrid Framework based on SIPOC and Six Sigma DMAIC for Improving Process Dimensions in
Supply Chain Network.” International Journal of Quality & Reliability Management 31 (5): 522–546.
Moosa, K., and A. Sajid. 2010. “Critical Analysis of Six Sigma Implementation.” Total Quality Management 21 (7): 745–759.
Nakajima, S. 1988. Introduction to TPM: Total Productive Maintenance, Vol. 129, Productivity Press, Inc.
Nelson, L. S. 1984. “Technical Aids.” Journal of Quality Technology 16 (4): 238–239.
Nonthaleerak, P., and L. Hendry. 2008. “Exploring the Six Sigma Phenomenon using Multiple Case Study Evidence”.” International
Journal of Operations & Production Management 28 (3): 279–303.
Ohno, T. 1988. Toyota Production System: Beyond Large-scale Production, CRC Press.
Osada, T. 1995. “Sikap kerja 5S = The 5Ss: Five Keys To A Total Quality Environment.”
Poznańska, K. 2004. “Factors of Innovations Increase of Small and Mediums Firms in Poland.” In International Conference on Small and
Medium Sized Enterprises: Strengths-weaknesses-opportunities-threats, 79-90.
Raghunath, A., and R. V. Jayathirtha. 2013. “Barriers for Implementation of Six Sigma by Small and Medium Enterprises.” International
Journal of Advancements in Research & Technology 2 (2): 1–7.
Rowlands, H. 2004. “Implementation Issues of Six Sigma in an SME.” Vol. 16, First International Conference on Six Sigma.
Saaty, T. L. 1980. The Analytic Hierarchy Process. New York: McGraw-Hill.
Shingo, S, and A.P. Dillon. 1989. A Study of the Toyota Production System: From an Industrial Engineering Viewpoint, CRC Press.
Singh, R., D. B. Shah, A. M. Gohil, and M. H. Shah. 2013. “Overall Equipment Effectiveness (OEE) Calculation-automation through
Hardware & Software Development.” Procedia Engineering 51: 579–584.
Snee, R. D. 1999. “Why should Statisticians Pay Attention to Six Sigma?” Quality Progress 32 (9): 100.
Stamatis, DH. 2002. Six Sigma and Beyond: Design for Six Sigma, Vol. 6, CRC Press.
Starbird, D. 2002. “Business Excellence: Six Sigma as a Management System: A DMAIC Approach to Improving Six Sigma Management
Processes.” In ASQ World Conference on Quality and Improvement Proceedings, 47–55. American Society for Quality.
Timans, W., K. Ahaus, R. van Solingen, M. Kumar, and J. Antony. 2014. “Implementation of Continuous Improvement based on Lean Six
Sigma in Small-and Medium-sized Enterprises.” Total Quality Management & Business Excellence:1–16.
Treicher, D., R. Carmichael, A. Kusmanoff, J. Lewis, and G. Berthiez. 2002. “Design for Six Sigma: 15 Lessons Learned.” Quality progress
35 (1): 33–42.
Tsai, C. C., and C. C. Chen. 2006. “Making Decision To Evaluate Process Capability Index C p with Fuzzy Numbers.” The International
Journal of Advanced Manufacturing Technology 30 (3–4): 334–339.
USDoD, United States Department of Defense. 1980. “Procedures for Performing A Failure Mode, Effects and Criticality Analysis.”
Vandenbrande, W. 2005. “Implementing Six Sigma in Small and Medium-sized European Companies.” Vol. 219, In Proceedings of the
International Conference on Quality; Tokyo.
Voehl, F., H. J. Harrington, C. Mignosa, and R. Charron. 2013. The Lean Six Sigma Black Belt Handbook: Tools and Methods for Process
Acceleration:CRC Press.
Wessel, G., and P. Burcher. 2004. “Six Sigma for Small and Medium-Sized Enterprises.” The TQM Magazine 16 (4): 264–272.
Wurtzel, M. 2008 June. Reasons for Six Sigma Deployment Failures, BPMinstitute.
Yang, T., and C. H. Hsieh. 2009. “Six-sigma Project Selection using National Quality Award Criteria and Delphi Fuzzy Multiple Criteria
Decision-making Method.” Expert Systems with Applications 36 (4): 7594–7603.
Zadeh, L. A. 1965. “Fuzzy Sets.” Information and Control 8 (3): 338–353.
Zimmermann, H. J. 1996. “Fuzzy Control.” In Fuzzy Set Theory and Its Applications, 203–240. Springer.
International Journal of Production Research 4339

Appendix 1. Application of the max–min method


4340

Appendix 2. Results and evolution of the Out-Process loop

Priority of actions
Global priority
N◦ of week FDS FD A FD S O1 O2 O3 Comments Actions vector

W2 Medium 7%High 49%Very high 77.3% Low Low Low Low Improve the customers satisfac- Implementation of the POKA YOKE system 0.102288
tion, improve the product qual-
ity and decrease the adjustments
frequency
Implementation of a Quality Plan 0.228999
Predictive maintenance of the cutting machine 0.148151
Studying and changing the cutting blades types 0.520562
W7 Low 4% Low 3% Low 11.9% Low MediumMediumLow Improve the customers satisfac- Implementation of SMED project (Dillon and 0.246831
tion and the availability Shingo 1985)
Enrolling a new employee 0.071641
Improving the wires conditioning system 0.184012
Predictive maintenance of the cutting machine 0.497515
W9 Low 2% Low 2% Low 8.52% High Medium High Low Improve the profits and the Implementation of a Computer-Aided Produc- 0.51529
effectiveness tion Management connected to the crimping
machine
Acquisition of a SPC software connected to 0.139846
measuring instruments
T.B. Romdhane et al.

Implementation of the POKA YOKE system 0.168326


Implementation of KANBAN 0.176539
W14 Medium 9% Low 3% Medium 31.2% MediumMedium High Low Improve the product’s quality Implementation of the POKA YOKE system 0.393244
and the capability
Implementation of KANBAN 0.080874
Improve the requirements of customers 0.254212
Implementation of a Quality Plan 0.27167
W17 Low 3% Low 4% Low 9% High Medium High Low Improve the profits Implementation of a SMED project 0.241544
Acquisition of new measuring instruments 0.104289
Improve sampling plan 0.10334
Implementation of 5S method in the cutting and 0.550827
setting stations
W22 Low 1% Low 2% Low 8.47% high High High Low Improve the diversification Purchasing a configurable machine with more 0.388729
features
Improve the requirements of customers 0.322933
Increase the objectives and the targets 0.215662
Searching new similar customers 0.072676
Copyright of International Journal of Production Research is the property of Taylor & Francis
Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.

Das könnte Ihnen auch gefallen