Final Year Honours Project for the degree of BSc. in Engineering with Management Journal Paper N. Al Helo Method to determine the optimum degree of lean activity applied to an industry using fuzzy logic May 2016 Project Supervisor: Dr. Sibi Chacko School of Engineering and Physical Sciences Heriot-Watt University Dubai Campus
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Final Year Honours Project
for the degree of
BSc. in Engineering with Management
Journal Paper
N. Al Helo
Method to determine the optimum degree of lean activity
3. Literature Review ........................................................................................................................................ 4
Applications of fuzzy logic coupled with lean tools ......................................................................................... 7
4. Overview of Methodology ........................................................................................................................... 8
5. Data collection and Analysis ........................................................................................................................ 8
Brief description of company .......................................................................................................................... 8
Analysis of Data ............................................................................................................................................... 9
Data Analysis of Individual OEE parameters ............................................................................................... 9
6. Development of conceptual models ......................................................................................................... 10
Model 1- Measuring availability .................................................................................................................... 10
Fuzzy system input and output variable(s) ................................................................................................ 10
Different MF shapes .................................................................................................................................. 11
Fuzzy system functionalities ...................................................................................................................... 11
Fuzzy system validation ............................................................................................................................. 14
Model 2-Determine the required downtimes for a specified availability ..................................................... 15
Fuzzy system input and output variable(s) ................................................................................................ 15
Fuzzy system functionalities ...................................................................................................................... 15
Fuzzy system validation ............................................................................................................................. 16
Fuzzy system output .................................................................................................................................. 16
7. Results and Discussion ............................................................................................................................... 16
General recommendation topics ................................................................................................................... 17
Appendix A-Brief descriptions of the various downtime categories ............................................................. 22
Appendix B-Tri, Trap and Gauss MFs [14] ................................................................................................ 22
Appendix C-Data used to define the MF of input and output variables ....................................................... 23
For input variables ..................................................................................................................................... 23
For output variables .................................................................................................................................. 23
Appendix D-How MFs were defined with an example .................................................................................. 24
by using various tools and techniques such as just-
in-time (JIT), Kanban, Kaizen, total productive
maintenance (TPM), single minute exchange of
dies (SMED), value stream mapping (VSM), 5S,
workforce involvement and OEE amongst others.
OEE quantifies the percentage of planned
production time that is truly productive. An OEE
score of 100% signifies a perfect production; where
only good parts are manufactured (100% Quality),
as quick as possible (100% Performance) and with
no down time (100% Availability).
From definition,
𝑂𝐸𝐸 = 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑡𝑦 × 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 × 𝑄𝑢𝑎𝑙𝑖𝑡𝑦
Equation 1-General OEE equation
𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦
=𝑁𝑒𝑡 𝑟𝑢𝑛 𝑡𝑖𝑚𝑒 (minutes)
𝑇𝑜𝑡𝑎𝑙 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑇𝑖𝑚𝑒 − 𝑇𝑜𝑡𝑎𝑙 𝑃𝑙𝑎𝑛𝑛𝑒𝑑 𝑡𝑖𝑚𝑒
Equation 2-General Availability
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 =𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑑 𝐵𝑎𝑡𝑐ℎ𝑒𝑠
𝑇𝑎𝑟𝑔𝑒𝑡 𝐵𝑎𝑡𝑐ℎ𝑒𝑠
Equation 3-Performance equation
𝑄𝑢𝑎𝑙𝑖𝑡𝑦 =𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑑 𝐺𝑜𝑜𝑑 𝐵𝑎𝑡𝑐ℎ𝑒𝑠
𝑇𝑜𝑡𝑎𝑙 𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑑 𝐵𝑎𝑡𝑐ℎ𝑒𝑠
Equation 4-Quality equation
6
OEE factor Implication
Availability Signifies to what extent the process is running during the
planned production time.
Performance Implies to what extent the process is
running at the theoretical maximum
speed.
Quality Determines how many defective parts
there are in the produced batch.
Table 1-This table briefly explains the meaning of each OEE parameters.
From the seven deadly types of waste listed above,
the OEE is powerful at quantifying the proportion
of waste inherent in the production process.
Availability loss incorporates anything that halts
the planned production for an appreciable amount
of time. For example; equipment failure,
changeovers, unplanned breakdowns, and
maintenance. Performance loss can be described
as any factor such as machine wear and usage of
substandard raw material, causing the process to
operate at a lower speed than optimum. The
quality loss includes parts that do not pass quality
inspection standards and require rework or
disposal.
OEE is useful as:
A benchmark to compare operational
performances to industry standards or for
different shifts working on the same asset.
Figure 1-Shows the proportion of true productivity to waste for various benchmarks [9]
Within this project, a low to typical OEE is studied
in an attempt to reach towards the ultimatum, a
world class production.
A baseline to keep track of company
progress in waste elimination. [9]
Establishing a baseline for a process by calculating
OEE provides an objective measure towards
improving manufacturing productivity. However,
using OEE scores to compare divisions across a
company can be problematic since these
comparisons are only truly meaningful if between
the same processes under the same conditions.
OEE is extremely effective at fulfilling the objective
of lean tools in exposing transparent waste
sources, effectively “uncovering the hidden
factory” within a process [10].
Fuzzy logic Fuzzy theory originates from the human inference
process, taking advantage of knowledge without
boundaries [1]. Fuzzy incorporates an alternative
way of thinking in computer language and can be
expressed by linguistic variables mapped as
numerical ranges. Unlike classical logic, not
everything has to be True or not True but rather a
degree of trueness specified [11]. The fuzzy
method, formulated by Zadeh (1965), is a
mathematical theory that allows for ambiguity and
vagueness to be modelled using fuzzy numbers.
This concept was not recognized until Mamdani
applied it in a practical situation to control a steam
engine almost a decade after it was invented [12].
Concepts like fuzzy sets, linguistic variables and if-
then rules are included within fuzzy logic [1]. Since
the 1980s, fuzzy methods have been applied in
many areas such as economics, manufacturing
operations, health sciences, automatic control,
engineering and communication technologies
[12,13]. The use of fuzzy logic has also been seen
applied in customer and domestic products such as
washing machines, microwave ovens, and medical
instrumentation, amongst many others [14]. The
application of fuzzy logic is especially important in
terms of making decisions for lean manufacturing
due to the characteristics of this logic, allowing for
both qualitative and quantitative analysis of
provided data [13].
Fuzzy logic is depicted as a method to transform
input vectors, based on a set of rules, to an output
vector(s).
7
A fuzzy set is a set without a crisp boundary that
contains elements with partial degrees of
membership. A clear illustration can be given by
the example of what season of the year one is
experiencing based on astronomical definitions
and the current climate. According to astronomy,
summer starts when the earth’s orbit points the
North Pole directly towards the sun, hence defining
a crisp boundary of exactly when summer is. In the
northern hemisphere, summer occurs in July and
lasts for one month. However, as we experience
here in Dubai, the climate expresses the heat of
summer from March through September,
regardless of the earth’s orbitals. Therefore, to
define such controversial subjects, fuzzy logic can
be used to allow continuity rather than sharp
boundaries, and the possibility of a assigning a
degree of membership as opposed to either 1 or 0,
as depicted by the Boolean logic [14].
Figure 2-This figure depicts the differences between Boolean and fuzzy logic explained by the example of the four seasons [14]
In more technical terms, linguistic variables are
described by a membership function (MF) µ(x) that
assigns to each number, x, a degree to which the
number satisfies the property.
For example; the property being ‘small’, the
degree of membership describes to what extent x
is small [15].
µ(x) ∈ [0, 1]
Equation 5-Property of the MF in assigning a degree of membership
The value 0 means x has no membership with the
set, while 1 denotes complete membership. Any
degree in the range between 0 and 1 denote a
partial membership to the fuzzy set. A series of if-
then rules are formulated to make the fuzzy
inference system (FIS), which uses input values as
weighted factors to map the final fuzzy output sets.
Once all the rules are inferred, scaled and
combined, a crisp output is obtained by
defuzzification methods [12].
Figure 3-This figure shows a triangular MF and how fuzzy sets and degrees of memberships can be defined [13]
To summarize; fuzzy logic is flexible, tolerant of
imprecise data, can model nonlinear functions, is
based on natural language and can be blended with
other tools; as will be presented in this paper [14].
Applications of fuzzy logic coupled with lean
tools In a paper by Susilawati et al. [13], fuzzy logic was
used to determine the degree of lean activity on
the areas: supplier and customer issues,
manufacturing, R&D and investment. The
proposed fuzzy number based scoring is applied to
aggregate multiple evaluators’ scores and analyse
this vague and subjective data.
8
In a study by Mohanraj et al. [16] a framework for
VSM integrated with fuzzy logic was developed and
tested by an automotive components
manufacturer to prioritize improvement plans
according to the results of the VSM. This initiative
was promoted due to the fact that it is infeasible to
implement all proposals concurrently. VSM is an
approach that identifies and eliminates non-value
added (NVA) activities from the value chain,
effectively removing waste. Examples include
walking long routes to unpack inventory or
transporting goods, which does not add any real
value to the product from the customers’
perspective. Vinodh and Balaji [1] state that one of
the contemporary agendas of lean manufacturing
is the quantification of leanness within a firm. To
build on this, they developed a fuzzy based system
that presents leanness assessment for an
organization manufacturing modular switches.
The case studies and academic articles on these
topics together prove that fuzzy logic can be used
with multiple lean tools to record great benefits in
terms of improved operational productivity. Fuzzy
integrated with a lean tool should provide a
framework for an improved lean tool effectiveness
and elimination of inconsistencies with crisp
values, to enable continuous improvement.
However; to the extent of research carried out,
there is no application in which OEE is fuzzified and
processed by fuzzy logic. The fact that OEE, a highly
numerical lean tool, has not been used in
conjunction with fuzzy logic, a mathematical
model, is the motivation for the developed models.
A successful model incorporating both aspects
should without doubt be deemed extremely useful
for any enterprise longing for a powerful tool to
eliminate or reduce waste while suggesting
improvement areas. Using OEE and fuzzy should
provide sufficient justification and quantifiable
evidence, convincing of lean approaches.
4. Overview of Methodology The methodology followed for this thesis is shown
in Figure 4.
Figure 4-This is the sequence of the processes followed to complete this thesis
The project begins with a literature review of lean
manufacturing tools and techniques, fuzzy logic,
practical applications of both concepts and the
novelty of the conceptual models developed. Since
model development requires data, manufacturing
information was collected and analysed from a firm
in the United Arab Emirates in Section 5. The model
inputs, functionalities, and validation are then
explained in Section 6. Results, discussions, and
implications are studied in Section 7. Concluding
points and statements are listed in Section 8,
before the limitations and future research
directions are mentioned in the concluding Section
9.
5. Data collection and Analysis
Brief description of company This thesis aims to help implement and optimise
lean activity using fuzzy logic, in a private paints
manufacturer located in Dubai.
The objective of the study is to increase the
manufacturing capacity and efficiency using OEE
and fuzzy logic.
The following section contains both qualitative and
quantitative analyses, with a focus on the latter, for
the two main manufacturing processes at the firm:
paint processing and paint filling. Data comes from
the processing and filling of water-based paints,
which makes up 75% of produced product.
Due to an increased competition, the organization
is looking at improving productivity and reducing
costs.
Derivation of improvement proposals
Model Computations and results
Development of conceptual fuzzy lean models
Data collection and analyses
Literature review of lean manufacturing and fuzzy logic
9
OEE needs to be raised with the aim of reaching
world-class standards (OEE≈80%) in the near
future.
Analysis of Data Data was collected and analysed through company
visits, documentation analysis and interview
techniques with various inter-department
personnel. Since the whole manufacturing firm is
vast, analyses of data should direct the study;
firstly, whether to focus on processing or filling
lines; secondly, which OEE parameter is the
limiting; and finally, which machine meets the
above criteria and can be analysed to design the
models accordingly.
The OEE for 2014 and 2015 was analysed, and
some remarks comparing processing and filling are
added after the data is reported. The overall
processing and filling averages are as follows for
2014 and 2015 across all machines and lines:
Functional Line
OEE
2014 2015
Processing 78.68% 73.85%
Filling 47.89% 47.46%
Table 2-Shows a comparison of OEE between the processing and filling lines to help decide on which area to focus
Performance enhancement is to be
focused on filling lines instead of
processing as a result of very low-
efficiency percentages and a lot of room
for improvement.
Data Analysis of Individual OEE parameters Breaking down the OEE scores to the three
parameters gives insight into which area shows
maximum losses, thus highlighting where
improvement efforts need to be focused.
After the data was analysed it was determined that
the parameters ‘Availability’ and ‘Performance’
were affecting the OEE adversely due to their low
scores. However, availability is the easier score to
improve with limited resources, and the
performance is simultaneously improved as a
result. ‘Quality’ was very rarely a non-perfect value
of 100%.
As a result of multiple analyses; it was concluded
that out of the various filling machines, machine
151 had the lowest of Availability scores, and
hence is the machine requiring the most effort.
MACHINE-151
% VALUES
OEE Parameter
2014 2015
Availability 65.1 62.97
Performance 73.83 74.46
Quality 100 100
Table 3-Shows the individual OEE parameters of machine 151
Under Availability; Planned and Unplanned
downtimes are considered important to minimize,
if not eliminate, to improve the overall OEE score.
Unplanned downtimes show variations and hence
a possibility of not only reducing the amount of
downtime, but also minimising the fluctuations to
maximise the operational capacity [7]. By having
well-designed plans complete elimination may also
be sought for some categories of unplanned
downtime.
Conversely, planned downtime remains mostly
constant, shifting focus from the reduction of
variability to magnitude reduction.
Exercise/Communication time and Tea/Lunch
break fall under planned downtime. Line
preparation, changeover, size change,
slip/can/pallet waiting, breakdown and spillage
cleaning are the 6 types of unplanned downtime.
Multiple graphical representations in order to
better understand the magnitudes and
fluctuations of each of these downtimes are
attached in the Appendix.
10
The pie chart below shows the proportion each
category impacts the total downtime; and that a
large proportion, approximately 40%, is due to
changeover time while switching between batches.
Frequent changes in filled product results in a
measurable loss in production due to wasted
changeover times [17].
6. Development of conceptual
models The design and simulations are done using MATLAB
R2013a /Simulink software using the inbuilt fuzzy
logic toolbox.
It was deduced that availability is the constraint,
and machine 151 was the machine with the lowest
score on that measure. Moreover, downtime
parameters are known and can be fuzzified by a set
of MFs. The inference system incorporates a set of
if/then rules and was designed to mimic the
availability formulae aforementioned. Two
computational models were designed: the first
helps measure the availability given a set of
downtimes, and the second operates vice versa; by
inputting the desired availability and the required
amount each downtime category should be
reduced to, displayed. This helps in formulating
recommendation plans reducing the availability
losses and improving the OEE score. The objective
of the models is to eliminate time waste induced
by variability.
Procedure 1. We commence by starting up MATLAB and
typing ‘fuzzy’ into the command line to bring
up the FIS GUI.
2. For the first model, we add the 8 downtime
categories as input variables and a single
output as ‘Availability’.
3. Double clicking on any input variable brings up
the ‘variable editor’ in which we choose which
type and how many MFs define the model. The
analysed statistical data can then be input as
parameters to define these MFs.
4. After all the variables are defined, the series of
if-then rules are then formulated in the FIS
editor.
5. The ‘rule viewer’ shows the crisp defuzzified
value of availability given quantities of each
downtime category.
6. For the second model; a single input as
‘Availability’ is added, with 8 output variables
signifying the respective downtimes.
7. The same set of rules defined in step 4 is used
again, after slight modifications, since they
adequately model the problem in hand.
8. The rule viewer this time around shows the
required downtimes to realize a particular
availability.
Model 1- Measuring availability
Fuzzy system input and output variable(s) The first step in system development is the
identifying of input and output variables necessary
to define the envisaged fuzzy model. The output of
a computational model is just as good as the inputs
specified, with improper data guaranteeing
inaccuracies. Therefore, analysis of a considerable
volume of data and shop floor observation was
performed to confirm consistency of the data
before utilisation. The input and output MFs were
defined by the analysed data. Each of the inputs
uses 3 MFs named ‘Low’, ‘Average’ and ‘Long’.
Availability data for machine 151 over 2015 was
aggregated and statistically manipulated to obtain
certain parameters needed to define each MF. The
‘Low’ defines a range of values, from analysis, that
is accepted to show an appropriate value for the
downtime of the respected category. ‘Average’
defines the range of values that occur on average,
and ‘Long’ specifies the range which incorporates
quantities that are deemed longer than average.
Exercise & Communication
Time (min)5%
Tea/Lunch Break Time (min)
23%
Line Preparation Time (min)
11%
Changeover Time (min)45%
Size change Time (min)
2%
Slip/Can/Pallet Waiting Time
(min)4%
Breakdown Time (min)
6%
Spilage Cleaning Time (min)
4%
FILLING MACHINE 151-DOWN TIMES-2015
Figure 5-A pi chart showing the proportion of each downtime category to total downtime
11
Similarly; the output variable, availability, is
defined by the 3 membership functions ‘Poor’,
‘Average’ and ‘Excellent’ characterized in the same
manner as the inputs.
Different MF shapes The MF shapes of triangular, trapezoidal and
Gaussian were defined by the analysed data. These
3 specific MF types were found to be the most
accurate of the numerous shapes available within
the toolbox [12,15,18]. As will be analysed in the
validation section, changing the MFs of a variable
has a measurable impact on the functionality of a
common system. Gaussian MFs are used to
represent variables that are smooth, non-linear
and deal in probabilities and statistics [13,19].
Proper choice of MFs allow the user to: define
linguistic variables, and enable the system to
realise accurate results [2].
According to Ali et al. [12] triangular MF, a particular case of trapezoidal [15], shows the best performance.
This accuracy is due to its simplicity and being described by a set of linear regions [18].
Fuzzy system functionalities Illustrations and screenshots are used to how the model was designed and functions.
Figure 6-This is FIS editor, which is the heart of the lean toolbox, in which the input and output variables are specified and the rule editor can be accessed
Rule editor After all variables and MFs are defined, rules need
to be created to allow model simulation. This is
where the series of rules are defined using ‘and/or’
connections to tie the inputs to the output.
Numerous if-then rules were tested until 8 were
concluded as optimal. The series of 8 if-then rules
that make up the FIS are designed to obtain
accurate results and mimic human reasoning in
tackling the vagueness of data used.
Knowing that fuzzy logic is a superset of Boolean
logic, at special extremity conditions standard
logical operations can be extended to use with
fuzzy logic.
12
In order to capture the problem best, logical
operations in logic gates and truth tables are used
as a motivation to design and illustrate the rules
used.
In classical reasoning theory, the truth of any
statement is either True (1) or False (0).
Conversely; in fuzzy logic, a multivalued logic, a
degree of truth is described. Using truth tables, the
AND/OR operators of Boolean logic can be
conserved on an extension to fuzzy reasoning by
use of the MIN and MAX operators respectively.
Figure 7-This image shows how the Boolean operators are extended to fuzzy logic.
AND operates if both A and B conditions are satisfied, OR if any of A or B are present and NOT is a
complementary function.
It can hence be deduced that the operation A AND B is similar to min (A, B), A OR B to max (A, B) and NOT A
to 1-A [14,19].
Figure 8-(a) Figure 6-(b)
Figure 8-(c) the images show the different logic gates that were used and their
operations.
13
The 8 rules defined can be represented by the following illustrations which define the rule operations used
according to logic gates.
Legend for logic gate rules 1-8
Letter Name of variable
A Exercise and Communication Time
B Tea/Lunch Break Time
C Line Preparation Time
D Changeover Time
E Size Change Time
F Slip/Can/Pallet Waiting Time
G Breakdown Time
H Spillage Cleaning Time
X Availability
14
Rule viewer Defuzzification using the centroid technique after the operation of rules on the input sets results in a crisp
output value.
Figure 9-shows the resultant defuzzified crisp output obtained from the specified inputs
The inputs can be entered numerically or by sliding the red cursors. The rules work on the input variables to
produce the output, with a defuzzified crisp value seen in the above example as Availability=73.9%. The usage
of this model can hence calculate the availability from a set of downtimes entered.
Fuzzy system validation Verification is a necessary step in any
computational or simulation model to ensure that
the real system is adequately replicated by the
conceptual model [4].
In order to reach accurate validation, the same set
of variables and rules need to be used for each of
the models produced with the different MF types.
To do this, practical data from the company at 3
different dates were input to the model, and the
defuzzified output compared to that calculated
theoretically. The percentage error table below
shows that trapezoidal MF is highly inaccurate for
this application, and was hence dropped. Although
the Triangular MF yields the least inaccuracy,
Gaussian MFs are also highly accurate and better
represent some data types such as the changeover,
waiting and breakdown times due to the
characteristics of the Gaussian curve. Therefore, a
hybrid model consistent of both Triangular and
Gaussian MFs is used.
% Errors for the different MF shapes
Date 10th April
27th July
1st December Average
Gauss 12 5 4 7
Tri 10 4 3 6
Trap 40 36 56 44 Table 4-error table comparing the accuracy of
different MF shapes
To validate the hybrid model, the same procedure
as above is repeated using data from April 23rd,
2015 and 28th arbitrarily. The resultant % error is
only around 2% in comparison to the practical
setting, which is even more accurate than solely
using triangular MF.
15
Model 2-Determine the required downtimes for a specified availability
This particular model is possibly more useful than the primary model since it informs the user of the amount
each downtime needs to become from a current state to reach the desired availability. The results of this
computation can help direct top management towards improvement plans by focusing resources on specific
areas highlighted by the model. Some recommendations of these improvement plans are discussed in the next
section.
Fuzzy system input and output variable(s) Reverse to model 1; a single input, ‘Availability’, is created alongside the 8 downtime categories as the output
variables.
Figure 10-Overall structure of developed model 2
Fuzzy system functionalities All variables are described by the same set of information utilised in creating model 1. The logic and MFs
defined are also similar to the case of model 1, with some modifications to incorporate backward integration.
Rule viewer
Figure 11-shows the value of the downtimes required to obtain an availability of 70%
16
Fuzzy system validation Testing these particular output values with
formulae, the model is found to be accurate with
only a 2% error. The same holds true with the
outputs generated when an Availability of 80% is
input.
Fuzzy system output Aiming to raise the Availability to 70% from the
Average of 62% gives a system output of the
downtimes required as:
Downtime Category Value (Minutes)
Exercise & Communication
6.5
Tea/Lunch break 30
Line preparation 30
Changeover 65
Size change 5
Waiting time 10
Breakdown time 55
Spillage cleaning 8
Table 5-similar to Figure 13 in a cleaner tabular form
Calculating this using the formula:
The total available time is calculated to be 570
minutes from an effective 9.5-hour shift starting at
Figure 12-surafce graph showing the relationship between availability, changeover and breakdown in 3D
7. Results and Discussion Simulation of the model is a mean by which
uncertainty is reduced and consensus created to
explore alternative strategies which suit the
production line. One of the many positives about
the fuzzy model on MATLAB is that aside its
accuracy, it remains flexible to details of the
organization.
Some general and specific recommendations for
improvement plans were devised to allow machine
availability maximization to 70%, and these are
outlined below. Since lean is a continuous
improvement philosophy, ‘Kaizen’ in Japanese,
improvements should continuously be sought. For
example, instead of increasing the availability to
70% without an aim to further improve; one should
sustain the improvement and work towards
further eliminating losses. This project not only
suggests the area of improvement but also how to
achieve them. Intangible benefits may also be
achieved in employees’ motivation and a positive
attitude towards change, alongside the primary
measurable benefits of reduction in lead time and
increased productivity [2].
17
General recommendation topics
Tackling downtime Downtime is the largest source of productivity
decrease for manufacturing firms, and if solutions
to eliminate them are devised, gains can be
realised rather quickly. It is due to these reasons
that the sole focus of the conceptual models was
downtime and its various categories. It is
important to improve the constraint and focus
resources on ensuring a strong improvement
impact. The constraint was determined to be the
changeover in the filling lines as per data analyses.
Any downtime should be made visual by the
operators, and if the machine is down for an
extended period of time the supervisor should be
informed of the issue. This prevents small issues
from becoming larger ones. This is both a general
and specific recommendation as it was noticed
once during collection of data that the filling nozzle
was not operating optimally. The designated
operators stopped the machine and tried fixing the
problem. On resuming normal operation; the
problem recurred, and the issue was not even
reported to top management [17].
TPM (Total productive maintenance) Focus is shifted from fixing breakdowns to
preventing them. There is a difference between
being reactive (fixing issues as they occur) and
being proactive (planning long-term fixes). If a
company spends time reactively fixing issues then
progress will be limited. Most of the mechanical
equipment has parts that wear out on repetitive
usage (e.g. seals, bearings, and belts) that may
cause breakdowns towards the end of their
lifecycle. A proper TPM plan makes sure that all
these parts are operable, replacing parts in adverse
conditions. TPM motivates operators to take the
initiative to maintain their equipment. [6,17] TPM
also helps in reducing small stoppages, slow
running, and accidents on the shop floor [20].
VSM (Value stream map) Where there is a product being developed for a
customer there is a value stream. This lean tool
tags operations in the value stream as NVA, VA,
and NNVA; in an attempt to identify waste sources.
The VSM helps visualize cycle times, WIP (work in
progress), manpower deployment and information
flow [4]. Non-value adding activities (NVA) are not
useful for the company or the customer. This
category effectively entails the waste which needs
to be eradicated. This may include material
handling, waiting and transport. Necessary but
non-value adding activities (NNVA) are considered
as waste by the customer but are a company
requirement. Examples include operators
unpacking inventory prior to assembling. Value-
adding activities (VA) are those which convert the
input into a useful output and may include
machining materials and joining subassemblies.
These are activities which the customer associates
with value and is willing to pay for. Within this
project, materials move in batches using pallets at
distances of around 40m instead of by continuous
flow. Reinstating, excess transport and ineffective
motion are some of the 7 types of waste [21].
Single minute exchange of dies (SMED) Setup time reduction is a continuously sought
objective [6]. This lean tool helps reduce the time
it takes for changeovers, and due to the fact that
changeovers make up around 40% of the total
downtime; SMED is a vital tool to solve the
problem in hand. This concept was developed and
tested by Japanese industrial engineer Shigeo
Shingo. SMED involves performing as many steps
as possible before the stage that depends on them,
and to do so with a coordinated team performing
multiple steps in parallel. For example, it takes
many people 15 minutes to change one tire, while
it takes a NASCAR pit crew 15 seconds to change
four. Processes that can be performed while
another is running can be moved externally. This
concept is a lot like multitasking. Using the same
NASCAR example, the tools and materials are
prepared prior to the tire changing stage. The next
step is performing those tasks as quick as possible.
For example, using quick release mechanisms
instead of traditional bolts. Reducing downtime
and building a smooth start-up (both improving
OEE) are some of the short-term benefits of SMED.
18
Some points that a particular process has to
possess in order to be a good candidate for this
lean tool is that the changeover should be
relatively long, variable and performed many times
a week to test proposed improvements quickly; all
conditions of which are satisfied by changeovers in
the filling line. SMED can be achieved by either
human or technical improvements, with the first
being faster and less expensive. This can be
brought about by defining roles properly,
involvement in brainstorming sessions and the
creation and following of standardized work
instructions [22].
Proposed improvement plans Proposed improvements should be assessed
against cost and time to implementation. This
makes larger improvements such as remodelling
the shop floor layout less favourable than
identifying and focusing resources on required
areas.
I. Integrating a swipe card or barcode reader
to ensure employees return to the station
on time in the morning and after breaks.
This will help reduce variability in the
planned downtime.
II. Introducing a belt conveyor system
throughout the shop floor for raw
materials and finished products to ensure
continuous flow while cutting down on
changeover times, waiting times and
waste relevant to the ineffective motion of
around 40m by pallet operators.
III. Automatically updated online order
distribution system at each machine to cut
down on slip waiting times, share
important information and allow for the
authorization of processes at the point;
such as direct communication with the QC
lab or PLC operator.
IV. There are off the shelf systems for
automating OEE measurement [9].
V. SMED is useful in tackling waiting and
changeover times; especially for cans, lids,
and pallets by preparing them prior to the
authorization to start filling.
VI. Flashing lights to show when the line is
down and requires extra attention.
This could be automatically activated or
switched on by the operator.
VII. Other important factors to consider may
include training inter-changeable
operators, which is what is done in the
case of area managers that change areas
every couple of weeks.
VIII. Furthermore, transparent information
sharing is vital and operators should not be
hesitant to report occurring issues [7].
Lean manufacturing is a long-term philosophy that
takes time and cost to arrive at the desired results
[21]. On implementing a few of the plans, the
productivity might improve quickly, not as the
result of the plan’s success, but rather the
‘Hawthorne Effect’; where employees work harder
to impress top management personnel
overlooking the process [22]. These changes
should help ensure an optimised material flow,
especially the filling lines. Customer demands can
now be met with even shorter lead times, adding
to marketing benefits [5]. Employees require
adequate training prior to the introduction of
various lean tools. Improvements should be
implemented one at a time to evaluate the
effectiveness of each as a solution.
8. Conclusions Knowing the benefits lean techniques and fuzzy
logic pose on a manufacturing unit, two conceptual
fuzzy lean integrated models were designed and
developed. Data collected from a paints company
was analysed to conclude that the filling lines’ OEE
score was limited by availability, hence adversely
affecting the overall productivity of the shop floor.
The input and output variables were defined by a
hybrid tri/gauss MF, after deduced that trapezoidal
MFs were highly inaccurate. The model output
allowed for general and specific improvement
proposals to be formulated that would help raise
the availability from 60% to 70%, possibly further,
resulting in a more efficient filling line.
19
9. Limitations and future research
direction The reported results are initial, with additional
efforts being to reduce downtimes further in the
future. The improvement plans were merely noted
down in a list; therefore to help with the
prioritization of proposals, a computerized support
system could be deployed [16]. In the future, more
case studies could be carried out across different
departments (processing lines for example) to
assess and improve the practical validity of the
system. Similar to the availability improvement,
the performance parameter also requires
improvement. The model is found to be weak at
extreme values, and can be further enhanced in
the future to solve this. Future research could build
on these proposed models, comparing the
potential strengths and weaknesses.
ACKNOWLEDGEMENTS
I would like to show appreciation and thanks to my
academic supervisor, Dr. Sibi Chacko, who has
been extremely open, helpful and cooperative
throughout the entire project.
I would also like to thank my industrial supervisor,
Mr. Shukla, for the opportunity to experience
practical implementation of the theoretical
knowledge gained, and the support and
information passed on from him.
I would also like to extend my thanks to Mr. Titus,
of the industrial company’s management, for
making this whole industrial collaboration
possible.
20
10. References
[1] Vinodh, S., and Balaji, S. R., 2011, “Fuzzy logic based leanness assessment and its decision support system,” Int. J. Prod. Res., 49(13), pp. 4027–4041.
[2] Achanga, P., Shehab, E., Roy, R., and Nelder, G., 2012, “A fuzzy-logic advisory system for lean manufacturing within SMEs,” Int. J. Comput. Integr. Manuf., 25(9), pp. 839–852.
[3] Adullah, F., 2003, “Lean Manufacturing tools and techniques in the process industry with a focus on steel,” University of Pittsburgh.
[4] Xia, W., and Sun, J., 2013, “Simulation guided value stream mapping and lean improvement: A case study of a tubular machining facility,” J. Ind. Eng. Manag., 6(2), pp. 456–476.
[5] Anvari, A., Zulkifli, N., and Yusuff, R. M., 2013, “A dynamic modeling to measure lean performance within lean attributes,” Int. J. Adv. Manuf. Technol., 66(5-8), pp. 663–677.
[6] Dal, V., and Akçagün, E., 2013, “Using Lean Manufacturing Techniques to Improve Production Efficiency in the Ready Wear Industry and a Case Study,” FIBRES Text. East. Eur., 21(100), pp. 16–22.
[7] Vinodh, S., and Chintha, S. K., 2011, “Leanness assessment using multi-grade fuzzy approach.,” Int. J. Prod. Res., 49(2), pp. 431–445.
[8] Dora, M., and Gellynck, X., 2015, “Lean Six Sigma Implementation in a Food Processing SME: A Case Study,” Qual. Reliab. Eng. Int., 31(7), pp. 1151–1159.
[11] Fuller, R., 1998, Fuzzy Reasoning and Fuzzy Optimization.
[12] Ali, O. A. M., Ali, A. Y., and Sumait, B. S., 2015, “Comparison between the Effects of Different Types of Membership Functions on Fuzzy Logic Controller Performance,” Int. J. Emerg. Eng. Res. Technol., 3(3), pp. 76–83.
[13] Susilawati, A., Tan, J., Bell, D., and Sarwar, M., 2015, “Fuzzy logic based method to measure degree of lean activity in manufacturing industry,” J. Manuf. Syst., 34(C), pp. 1–11.
[14] The MathWorks, I., 2016, “Getting Started with Fuzzy Logic Toolbox,” MathWorks [Online]. Available: http://www.mathworks.com/help/fuzzy/getting-started-with-fuzzy-logic-toolbox.html.
[15] Barua, A., Mudunuri, L. S., and Kosheleva, O., 2014, “Why trapezoidal and triangular membership functions work so well: Towards a theoretical explanation,” J. Uncertain Syst., 8(3), pp. 164–168.
[16] Vimal, K. E. K., Mohanraj, R., Sakthivel, M., and Vinodh, S., 2015, “A framework for VSM integrated with Fuzzy QFD,” TQM J., 27(5), pp. 616–632.
[17] Vorne Industries, 2013, “Down_Time.”
[18] Zhao, J., and Bose, B. K., 2002, “Evaluation of Membership Functions for Fuzzy Logic Controlled Induction Motor Drive,” IEEE 2002 28th Annu. Conf. Ind. Electron. Soc., pp. 229–234.
[20] Shahid, S., Chacko, S., and Shukla, C. M., 2013, “Energy Conservation for a Paint Company Using Lean Manufacturing Technique,” Univers. J. Ind. Bus. Manag. 1(3), 1(3), pp. 83–89.
[21] Vamsi, N., Jasti, K., and Sharma, A., 2014, “Lean manufacturing implementation using value stream mapping as a tool,” Int. J. Lean Six Sigma, 5(1), pp. 89–116.