You are on page 1of 18

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1463-5771.htm

Reconfigurable
Assessment of a reconfigurable manufacturing
manufacturing system system
Durga Prasad and S.C. Jayswal
Department of Mechanical Engineering,
Madan Mohan Malaviya University of Technology, Gorakhpur, India
Received 4 June 2018
Abstract Revised 23 August 2018
Accepted 13 September2018
Purpose – The purpose of this paper is to develop the methodology which can facilitate the concept of
reconfiguration in the manufacturing system.
Design/methodology/approach – Design methodology includes the calculation of similarity matrix,
formation of part family, and selection of part family. ALC algorithm has been used for part family formation
and three criteria have been considered for the selection of part family. These criteria are reconfiguration
effort, under-utilization cost, and floor space cost. AHP has been used to calculate the weights of criteria and
reference ideal method has been used for the selection of alternatives.
Findings – In the manufacturing system, machines should be grouped on the basis of reconfiguration cost.
When the time period is less, light machines and Group 1 machines are added and removed. In the case study,
the concept of reconfiguration is useful for families (A, B, C, D). Machines can be reused by adding/removing
some modules of machines. The concept of reconfiguration becomes more useful when it is implemented with
lean manufacturing. Lean manufacturing techniques Jidoka and Poka-yoke are used to increase the
diagnosability of the system.
Practical implications – Industrial case study has been considered.
Social implications – Market competition is increasing rapidly and it increases the demand and variety of
products, due to which manufacturing enterprises are forced to adapt a manufacturing system which can
adjust its capacity and functionality quickly at low cost. To reconfigure manufacturing system from one
product/product family to another product/product family, changes can be done in hardware and/or software
components in response to sudden changes in the market or in regulatory requirements.
Originality/value – An integrated approach for reconfiguration has been proposed considering the
industrial application. It includes weighted Jaccard function, ALCA, AHP, RIM. The methodology
for calculation of reconfiguration effort, under-utilization cost, and floor space cost has been presented for
industrial case.
Keywords ALCA, Reconfigurable manufacturing system, Reconfiguration effort, Reference ideal method,
Similarity coefficient
Paper type Research paper

1. Introduction
Market competition is increasing rapidly and it increases the demand and variety of
products, due to which manufacturing enterprises are forced to adapt a manufacturing
system which can easily respond to these changes. Reconfigurable manufacturing system
(RMS) belongs to this category and it can quickly adjust to the requirements at low cost. To
reconfigure manufacturing system from one product/product family to another product/
product family, changes can be done in hardware and/or software parts so that it can easily
absorb the qsudden changes occurred in the market or in regulatory requirements (Koren
et al., 1999). RMS consists of the dedicated machines, reconfigurable machines, and CNC
machines. It lies between dedicated and flexible manufacturing system. Dedicated
manufacturing systems have the high capacity with limited functionality while flexible
manufacturing systems (FMSs) have less capacity with high functionality (ElMaraghy,
2005). RMSs have the exact functionality and capacity that is required.
With the development of CNC machines and automatic material handling system, the
Benchmarking: An International
concept FMSs was introduced in the 1980s in order to respond to the uncertainties occurred in Journal
the manufacturing environment so that the manufacturing system should be able to run © Emerald Publishing Limited
1463-5771
smoothly without any disturbance. But FMSs were very costly and complicated, also the DOI 10.1108/BIJ-06-2018-0147
BIJ maintenance of FMS is very difficult. To deal with these problems, a new type of manufacturing
system is introduced by the end of the 1990s named RMS. It consists of an adjustable structure
of machine tools and material handling systems. By changing the hardware and software parts
in the machine as well as system, its flexibility is customized and it can be changed whenever a
need occurs (Prasad and Jayswal, 2017b, c, 2018c; Maganha et al., 2018).
Reconfigurable machine tools (RMTs) are also known as modular machines. In these
machines, some parts of the machines (also known as modules) can be hanged (added/
removed/adjusted) to change the functions or capacity of the machines (Goyal et al., 2013b).
The concept of modular machines is not new and it has been used for many years (Rogers
and Bottaci, 1997), but it helped in the development of the concept of RMS. Many definitions
of modularity have been presented (Shaik et al., 2015). The main advantages of using the
concept of modular manufacturing is that it provides opportunity of short-term as well as
long-term goals of the industry, it enables the integration of machine tools, information flow,
processes, etc. It increases the reuse of the machine equipment (Rogers and Bottaci, 1997).
In the present work, reconfigurable machines have been considered in which parts, such as
fixtures and tools, are changed to manufacture another product. It increases the utilization
of the machine but reconfiguration cost occurs.
Initially, when the demand of products was low, process type systems were used.
But when the demand of certain products increased, separate cells were designed for those
products. Initially, cells were designed for single product but as the variety of the products
increased, industries started to design the product for a group. Researchers were focusing
for the group the products so that a manufacturing cell should be designed but after the
development of the concept of reconfiguration even that part family is grouped in the small
groups/families so that cell can easily be reconfigured between one subgroup to another
subgroup. For example, Figure 1 shows the part family formation without the concept of
RMS and with the concept of RMS. In this figure, manufacturing cell is designed for part
families A, B, and C. While in RMS, part family is designed for family A1 then it is
reconfigured to A2 and A3 whenever a need occurs.

Part family A Part family B Part family C

Cell 1 Cell 2 Cell 3


Part family formation without the concept of reconfiguration

Part family A1 Part family A2

Part family A3
Figure 1.
Part family formation
without the concept of
RMS and after the Cell 1
concept of RMS
Part family formation after the concept of reconfiguration
For designing a manufacturing system, question arises that whether it should be designed Reconfigurable
for all the products or for a part family and then reconfigured for another part family? What manufacturing
should be the level of reconfiguration? Should all the machines be reconfigured or some
selected machines should be reconfigured? Research work has been conducted to find the
system
answers of these questions. Case study of Continental Automotive Components (India) Pvt.
Ltd has been used to find the answers.

2. Literature review
Various researchers have worked on RMS. In this section, the literature review related to
similarity index, ALCA, and cost methodology used in work, have been discussed.

2.1 Similarity index


Part families in the manufacturing systems are formed on the basis of similarity of the
operation. For this purpose, similarity coefficient is calculated. Similarity coefficients are
used for the measurement of the groupability in the system. Baroni-Urbani and Buser (1976)
defined some properties for similarity coefficient. These can be discussed as a similarity
coefficient is symmetric, when Smn¼Snm. A similarity coefficient lies between 0 and 1. It is 0
when there is no similarity and if it is 1 then there is complete similarity. The oldest and
best-known similarity coefficient is given by Jaccard (1908). McAuley (1972) used this
coefficient in manufacturing cell formation. Mosier and Taube (1985) modified it to weighted
similarity coefficient. Seifoddini and Wolfe (1986) proposed the similarity coefficient which
was based on the production data by considering production volume, operations sequence,
and processing times. There are some others similarity index developed by Askin and
Zhou (1998), Irani and Huang (2000), Goyal et al. (2013a), etc.

2.2 Linkage algorithm


There are two techniques available in the literature, single linkage cluster algorithm (SLCA)
and average linkage cluster algorithm (ALCA) (Seifoddini and Wolfe, 1986). Initially, these
techniques were used for grouping the machines. McAuley (1972) used a single linkage
clustering algorithm for the formation of manufacturing cell. The method was later used by
other researchers (Seifoddini and Wolfe, 1986; Carrie, 1973; Witte, 1980). Seifoddini and Wolfe
(1986) used ALCA for the formation of machine cells. Galan et al. (2007) used ALCA for group
formation in a RMS. Seifoddini (1989) compared a single linkage clustering algorithm and
ALCA for cell formation. In a single cluster algorithm, the similarity coefficient between two
clusters was defined as the similarity coefficient between two closest members of the two
machine cells. While in ALCA, the similarity coefficient between two machine cells was
defined as the average of pairwise similarity coefficients between all members of the two cells
(Seifoddini, 1989). The results of cluster analysis are represented by dendrogram. It is similar
to the tree diagram which represents the clusters at different similarity levels.

2.3 Reconfiguration effort


When a manufacturing system is reconfigured, it requires some effort and cost. There are
two methodologies available to calculate the cost. The first methodology is the calculation of
reconfiguration effort. Youssef and ElMaraghy (2006) used the term reconfiguration
smoothness (RS). Generally, RS has been calculated by formula:
No: of modules added No: of modules removed
RS ¼ s þð1—sÞ :
Total no: of modules Total no: of modules
It was used for configuration selection in the research of Youssef and ElMaraghy (2007).
Goyal et al. (2013b) used term reconfiguration effort and for the calculation of
BIJ machine reconfigurability. For reconfiguration effort, three terms have been considered;
number of modules added, number of modules removed and number of modules adjusted.
The formula has been shown in Equation (8). Prasad and Jayswal (2017a, c, 2018a, b)
calculated reconfiguration effort for manufacturing system. In both cases, the relative
value is calculated between 0 and 1. In the second method, unit cost is considered
for reconfiguration. Lee (1997) used this concept for the calculation of relocation cost.
Galan et al. (2007) used this concept for the calculation of reconfiguration cost.
Benderbal et al. (2018) outlined an approach to optimize the design of RMS. It consisted of
three objectives. The first objective was to maximize the modularity of the system. The second
objective was to minimize the completion time of the system. The third objective was to
minimize the cost of the system. They used archived multi-objective simulated annealing
method for solution of the problem. Mittal et al. (2017) used an index for the consideration of
reconfigurability considering the different characteristics such as modularity, convertibility,
and diagnosability. Lee et al. (2017) considered factors related to sustainability and developed
a model to simulate self-RMS. Li et al. (2018) presented an approach for limited inventory of
the components that can be configured. Mortensen et al. (2017) defined a framework that can
be used for virtual recommissioning in the RMSs. Bettaieb et al. (2017) presented an approach
for reconfiguration process following a predictive monitoring. Lameche et al. (2017) proposed
an approach based on the design structure matrix to design RMS.
In the literature, it has been found that Jaccard coefficient is the simplest one and easy to use
for industrial application. Since each machine/operation has different importance, therefore in
the present work, weighted Jaccard coefficient has been used for the calculation of similarity
index and average linkage algorithm has been used for part family formation. Part family has
been selected on the basis of three criteria: reconfiguration effort, under-utilization cost and floor
space cost. Case study of industry has been used to validate the methodology.

3. Case study
In the present work, a case study of Continental Automotive Components (India) Pvt. Ltd
has been considered for the validation of the methodology. Continental Automotive
Components (India) Pvt. Ltd is a part of Continental AG, one of the world’s top five
automotive parts suppliers. In Gurgaon IMT Manesar plant, there are four business units:
clusters – speedometer (mechanical), clusters – speedometer (electrical), sensors and
actuators, and wheel speed sensor (WSS).
In WSS assembly line, four types of WSS models are manufactured. To keep data
confidential, names of model types have been changed to product families A, B, C and D.
Processes which are performed in assembly line are cable cutting, grommet insertion,
stripping, shrinking, wire seal insertion, wire stripping, contact crimping, assembly, jacket
stripping, wire stripping-2, element crimping, molding, pin cutting, height checking, test
(vacuum chamber), insulation testing and marking, adjustment, fastening, clip cutting,
bracket crimping, test (current leakage), functional testing, taping, visual checking, and
packing. For a manufacturing system, maximum productivity is achieved when each person
get independent and clearly defined job description with standard procedures and timing
(Taylor, 2004). Therefore, work study has been carried out in the manufacturing system.
Table I shows the time for operations performed on the assembly line. Continental is using
lean manufacturing to improve the quality of manufacturing system. 5-S has been
implemented in the assembly line. Jidoka and Poka-yoke have been used to reduce the
defected products and overall equipment effectiveness is used as a standard of productivity.
Some points drawn on the basis of work study have been discussed below:
(1) In the present layout, 24 operations are performed. In this assembly line, four types
of product families are manufactured. The names of these product families have
Processing tim e
Reconfigurable
Processes Machine’ s type Machines (in sec) Nk A B C D manufacturing
op 1 – Cable cutting Group 2 M1 1 1 | | | |
system
op 2 – Grommet insertion Group 3 M2 14 1 | | | |
op 3 – Stripping Group 2 M3 4 1 | | | |
op 4 – Shrinking Group 2 M4 16 1 | | | |
op 5 – Wire seal insertion Group 2 M5 16 1 | | | |
op 6 – Wire stripping Group 2 M6 7 1 | | | |
op 7 – Contact crimping Group 2 M7 8 1 | | | |
op 8 – Assembly Group 2 M8 12 1 | | | |
op 9 – Jacket stripping Group 2 M9 5 1 | | | |
op 10 – Wire stripping-2 Group 2 M10 8 1 | | | |
op 11 – Element crimping Group 3 M11 10 1 | | | |
op 12 – Molding Group 3 M12 13 1 | | | |
op 13 – Pin cutting Group 2 M13 9 1 | |
Group 2 M 14 9 1 |
Group 2 M15 9 1 |
op 14 – Height checking Group 2 M16 6 1 | |
Group 2 M17 6 1 |
Group 2 M18 6 1 |
op 15 – Test (vacuum chamber) Group 2 M19 7 1 | | | |
op 16 – Insulation tester and marker Group 3 M20 8 1 | | | |
op 17 – Adjustment Group 1 M21 10 1 |
Group 1 M22 10 1 |
Group 1 M23 10 1 |
Group 1 M24 10 1 |
op 18 – Fastening Group 1 M25 23 2 |
Group 1 M 26 23 2 |
Group 1 M27 23 2 |
Group 1 M28 23 2 |
op 19 – Clip cutting Group 1 M29 8 1 |
Group 1 M30 8 1 |
Group 1 M31 8 1 |
Group 1 M32 8 1 |
op 20 – Bracket crimping Group 1 M33 18 2 |
Group 1 M34 18 2 |
Group 1 M35 18 2 |
op 21 – Test (current leakage) Group 2 M36 7 1 | Table I.
op 22 – Functional testing Group 2 M37 15 1 | | | | Processes and
op 23 – Visual checking and packing Group 1 M38 7 1 | machines used for
op 24 – Taping Group 1 M39 9 1 | | | | product families A, B,
Total 46 26 22 24 24 C and D

been given as A, B, C and D and name of the processes has been given as op 1, op
2, …, op 24. The machines required for these operations are M1, M2, …, M39. These
machines have been divided into three groups: Group 1, Group 2 and Group 3 shown
in Table I. Group 1 machines are very light machines or fixtures used, Group 2
machines are medium machines and Group 3 machines are the heavy machines.
(2) For reconfiguration, some parts of the machines are changed. The parts of machines,
which are changed, can be tools, tool-holding devices, job-holding devices and other
supporting devices. These parts are also known as auxiliary modules. In the assembly
line, machines which are reconfigured are grommet insertion machine (M2), crimping
machine (M11), molding machine (M12), and insulation testing machine (M20).
Machines with auxiliary modules used for different types of products have been
BIJ identified and specific names have been given to these modules shown in Table II. For
example, crimping machine (M11) has two auxiliary modules CT1 and CT2. CT1 is
used for product families A, B and C while CT2 is used for product family D (Table II).
(3) Current leakage test and taping operations are performed only on model A. Bracket
crimping operation is performed on A, C, and D model.
(4) Jidoka has been used for the line. Seven Poka-yoke symbols have been used in the line
( for grommet insertion, housing assembly, crimping, pin cutting and height checking,
insulation testing and current leakage test). Jidoka and Poka-yoke also increases the
diagnosability of the system which is one of the key characteristics of RMS.
(5) Three pin cutting and height checking machines are used for four types of product
families. One for A, C while other two machines are used for B and D.
(6) Single product type flow has been adapted except shrinking, molding operation and
vacuum chamber operations. In the shrinking operation, two workpieces and in
molding operations, four workpieces are operated simultaneously to reduce the
average cycle time while in the vacuum chamber 50 pieces (max) can be kept.

4. Methodology
In the present work, methodology presented is based on industrial situation. The methodology
has been sub-divided in following parts:
(1) calculation of similarity matrix;
(2) grouping of product families; and
(3) selection of the part families.
In the methodology, following notations have been used:
• Nk: number of machines required for kth operation.
• Sij: similarity between a pair of products (i, j).
• REs: reconfiguration effort for schedule s.
• Us: under-utilization cost for schedule s.
• Fs: floor space cost for schedule s.

Machines Machine configurations Auxiliary modules A B C D

M2 M2 1 GIM1 |
M22 GIM2 |
M23 GIM3 |
M24 GIM4 |
M11 M 111 CT1 | | |
M 11
2
CT2 |
M12 M 112 MM1 | |
M 12
2 MM2 |
M 12
3 MM3 |
M20 M 20
1 ITM1 | |
Table II.
Machines with M 20
2 ITM2 |
machine M 320 ITM3 |
configurations Total 4 4 4 4
4.1 Calculation of similarity matrix Reconfigurable
Jaccard similarity coefficient measures the similarity between a pair of products (i, j), and it manufacturing
is defined in terms of the machines that each product has to visit for the operations. This
coefficient (Sij) may be expressed as (Sarker and Islam, 1999):
system
a
Sij ¼ ; 0 oS ij o1; (1)
a þb þc
where a, number of machines that visit both products m and n; b, number of machines that
visit only product i; c, number of machines that visit only product j.
If different importance is given to each machine visited, then the formula can be
changed to:
PK N y x
k k ijk
Sij ¼ PKk¼1 ; (2)
N y ky k ijk
k¼1

where θk, importance given to the kth operation.


.
1 if both products ði; jÞ visits for kth operation
xijk ¼
0 otherwise
. ;
1 if either product ði; jÞ visits for kth operation
yijk ¼
0 otherwise
:
In the present work, similarity coefficient has been considered as the weighted sum of the
similarity coefficient between the machines (S1) and similarity coefficient between the machine
modules (S2). The similarity coefficient (S ) has been calculated by using the following formula:
S ¼ w1 S 1 þ w2 S 2 ; (3)
w1 and w1 are the weights assigned to the similarity between machines and similarity between
modules, respectively, such that w1W w2 and w1+w2 ¼ 1.

4.2 Formation of part family


ALCA has been used for part family formation. It has been shown in Figure2 (Galan et al., 2007).
In this algorithm, two products are grouped together for which similarity coefficient is the
highest. After grouping the two products, these products are treated as a single product and
new similarity index is calculated. This process is repeated until all the products are grouped.
New similarity matrix can be calculated by using the following equation:
PP
i jSij
S ni0 j0 ¼ ; (4)
ni0 ~ nj0
where ni′, number of products in i′th family, and nj′, number of products in j′th family.
For example, if there are four products A, B, C, and D. Products A and B are grouped,
then similarity coefficient between (A, B) and C is calculated by:
S ðA;CÞ þS ðB;CÞ
S ðA;BÞ;C ¼ :
2
Similarity coefficient between (A, B, C) and D can be calculated as:

S ðA;DÞ þS ðB;DÞ þS ðC;DÞ


S ðA;B;CÞ;D ¼ :
3
BIJ Start

Group together
products for
Max Sij

No If all the products are


Calculate
new Sij (Si'j'*) grouped in same product
family

Yes

Create
dendrogram
Figure 2.
Average linkage
cluster algorithm
Finish

Similarity coefficient between (A, B) and (C, D) can be calculated as:

S ðA;CÞ þS ðA;DÞ þS ðB;CÞ þS ðB;DÞ


S ðA;BÞ;ðC;DÞ ¼ :
4
Part family is formed using ALCA and dendrogram is prepared. Dendrogram presents the
result of cluster analysis in the most convenient way. The abscissa of the dendrogram shows the
parts that are manufactured in the manufacturing system. The ordinate shows the similarity
coefficient in percentage. The junction points of stem mean the resemblance of the products at
that similarity coefficient. The dendrogram shows the part family at the different levels. At each
level, part families are scheduled and for each schedule, cost occurred can be calculated.

4.3 Selection of part family


In a manufacturing system, by using the concept of reconfiguration, machines can be reused
by adding/removing some modules of machines. It also saves the floor space by adding/
removing some machines from the system. Therefore, three criteria have been considered:
reconfiguration effort, under-utilization cost and floor space cost.
In the manufacturing system, reconfiguration cost and under-utilization cost are
different for different machines. But for the simplicity of calculation, machines can be
grouped in “Q” types.
4.3.1 Reconfiguration effort. It is the effort for changing its configuration from one type of
product family to another type of product family. Reconfiguration effort can be at three levels:
market-level reconfiguration effort, system-level reconfiguration effort, and machine-level
reconfiguration effort.
The market-level reconfiguration effort (MKRE) is associated with the activities that are
performed outside the boundaries of the manufacturing system such as financial activities,
shipping activities, bidding activities, logistic activities, etc., that are associated with
purchasing new machines or machine modules, selling old machines or modules and renting
machines or modules. System-level reconfiguration effort (SRE) is associated with the
activities that are performed within the boundaries of the manufacturing system but at a Reconfigurable
level higher than machines. These activities include adding, removing or adjusting the manufacturing
machines in the system, relocating the machines and changing the material flow path.
Machine-level reconfiguration effort (MRE) is associated with the activities that are
system
performed inside the boundaries of the manufacturing system and are all within the limits at
the machine level. These activities include the adding, removing or adjusting machine
modules and adding, removing or adjusting operation clusters. For all the activities
reconfiguration effort is calculated separately by considering machines and/or modules
added, removed or adjusted. Total reconfiguration effort (TRE) can be calculated as the
weighted sum of the all three level reconfiguration efforts:
TRE ¼ c1 ~ MKREþc2 ~ SREþc3 ~ MRE; (5)
where c1, c2, c3 are the weights assigned to the all three types of reconfiguration effort and
c1 þ
c2 c3þ1. ¼
In the present case, only two types activities have been considered: addition/removal of
machines (system level) and addition and removal of modules (machine level) system-level
reconfiguration effort can be calculated as:
No: of machines added No: of machines removed
SREgroup—1=2=3 ¼ a þb
Tota no: of machines Total no: of machines
No: of machines readjusted
þg ; (6)
Total no: of machines
where, α, β, g are constants; α ⩾ β ⩾ g and α+β +g ¼ 1. SRE is the weighted sum of SREgroup—1,
SREgroup — 2, SREgroup — 3⋅⋅⋅
SRE ¼ z1SREgroup—1 þz2SREgroup—2 þz3SREgroup—3; (7)
where, ζ1, ζ2, ζ3 are constants; ζ1 Wζ2 Wζ3 and ζ1+ζ2+ζ3 ¼ 1.
MRE can be calculated as:
No: of modules added No: of modules removed
MRE ¼ a0 þb0
Tota no: of modules Total no: of modules
0 No: of modules readjusted
þg ; (8)
Total no: of modules
where, α′, β′, g′ are constants; α′ ⩾ β′ ⩾ g′ and α′+β′+g′ ¼ 1.
Total no of modules ¼ Total no of modules added
þTotal no of modules removed
þTotal no of modules readjusted: (9)

If, for level schedule s, products are manufactured as A → B → C → D → A, then


reconfiguration effort for schedule s is calculated as:
REs ¼ REA—B þREB—C þ REC—D þRED—A:

In general form, reconfiguration cost for schedule s:


X
REs ¼ REi—j; (10)
s

where ∑s REi−j ¼ summation of all the reconfiguration cost for schedule s.


BIJ 4.3.2 Under-utilization cost. In a multi-product line, a manufacturing system requires the
machines for the all the products. When one type of product is manufactured, the machines
and modules which are required for that product is used and other machines and modules
are not used. Therefore, under-utilization cost is calculated for each family formation and it
is same for all schedules for that family. Under-utilization cost has been divided in the
under-utilization cost of machines and under-utilization cost of modules. Under-utilization
cost of ith machine can be calculated by following formula:

U ðmachineÞi ¼ pi ~ bi ; (11)

pi is the number of products not using ith machine; βi is the under-utilization cost for
ith machine.
Under-utilization cost for modules for ith module can be calculated as:

U ðmoduleÞi ¼ p0i ~ b0i; (12)

where p0i is the number of products not using ith module; b0i is the under-utilization cost for
ith module:
N N0
X X
Us ¼ U ðmachineÞi þ U ðmoduleÞi ; (13)
i¼1 i¼1

where N is total number of the machines; N′ is total number of the modules.


4.3.3 Floor space cost. It can be calculated by multiplying the total workstations used in
the manufacturing system for the schedule to the unit cost of floor space:
F s ¼ ðTotal number of work stations in scheduleÞ ~ g00 ; (14)
where g00 is the unit cost floor space per work station.
In the problem-integrated approach of AHP and reference ideal method have been used.
Weights of each criterion have been determined by using AHP and ranking of the
alternatives has been obtained by reference ideal method. The structure of the decision
matrix can be expressed as shown in Table III.

5. Reference ideal method (RIM)


RIM is proposed by Cables et al. (2016). The basic principle of this technique is that
ideal solution is not always strictly the maximum value and/or the minimum value,
but rather the ideal solution is a value (or a set of values) that lies somewhere in between
and its another advantage is that it avoids rank reversal problem (Cables et al., 2016).
Rank reversal is the problem that if any new alternative is added, ranking of the
alternative changes. Sometimes, the worst alternative becomes the best (Saaty and Sagir,
2009; Wang and Luo, 2009; García-Cascales and Lamata, 2012). In this problem, criteria

Table III.
Structure of
decision matrix Notes: xij = Performance value of alternative i when it is evaluated in
terms of criterion j; i =1, 2, …, m and j = 1, 2, …, n
used prefer the extreme valves but RIM is beneficial because it avoids the rank reversal Reconfigurable
problem. RIM has following steps: manufacturing
Step 1: define the work context.
In this step, the conditions in the work context are established, and for each criterion the
system
following aspects are defined:
• The range: this is any interval, labels set or simple set of values between which
performance values of each alternative vary.
• The reference ideal: this is an interval, labels set or simple values that represents the
maximum importance or relevance in a given range. The reference ideal can be any
set between the minimum value and the maximum value.
• The weight wj associated to the criterion.
Step 2: obtain the valuation matrix, in correspondence with the defined criteria.
Step 3: normalize the valuation matrix with the reference ideal.
If (A, B) is the range, where A is the lower limit and B is the upper limit of range. (C, D) is
the reference ideal. C is the lower limit and D is the upper limit of the range such that xij∈
(A, B); (C, D) 3 (A, B):
8
1; if xij A ðC; DÞ
>
< ðx Þ
1— d jA—Cj ; if xij A ðA; CÞ; A aC ;
min ij ;½C; D]
yij ¼
>
: 1— dmin ðxij ;½C; D]Þ; if xij A ðB; DÞ; B a D
jD—Bj

where, dmin (xij, [C, D])¼ min(|xij−C|, |xij−D|).


Step 4: calculate the weighted normalized matrix y′, through:
y0ij ¼ yij ~ wj:

Step 5: calculate the variation to the normalized reference ideal I iþ , I —i for each alternative Ai:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi.ffiffiffiffiffiffiffiffiffiffiffiffiffiffiΣffiffiffi2ffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffi.ffiffiffiffiffiffiΣ
ffiffiffi2ffiffi
Xn —
Xn
Iiþ¼ j¼1
y 0—
ij w j ; I i¼ j¼1
y 0
ij :

Taking the above into account, the vector that represents the reference ideal will be the
vector (1, 1, …, 1); also, the reference ideal would be weighted, then the reference ideal
coincides with the vector of weight w.
Step 6: calculate the relative index Ri of each alternative Ai:
I—
i

i ; 0 o Ri o1
I iþ þI —i
Step 7: rank the alternatives Ai in descending order. The alternatives that are at the top
constitute the best solutions.

6. Validation by case study


In case study, system is needed to reconfigure for short period of time (one or two days). In such
systems, when reconfiguration occurs, the heavy machines are not preferred to add/remove.
Relocation of work stations is not preferred. The machines have been divided into three groups:
(1) Group 1 consists of light machines or equipment which can easily be placed in
another place shown in Table I.
BIJ (2) Group 2 consists of heavier machine than Group 1, which are more difficult to add or
remove shown in Table I.
(3) Group 3 consists of the heaviest machines, which are the most difficult to add or
remove shown in Table I.
In this manufacturing system, since heavy machines are not preferred to relocate, machine
module and group 1 machines are added and removed.

6.1 Similarity coefficient


From Table I, it can be seen that for Products A, B, C and D, total machines required
are 26, 22, 24 and 24. For products A and B, the machines which are common are 16
(4: Group 3, 11: Group 2 and 1: Group 1). The total machines which are visited by either A
or B is 32 (4: Group 3, 16: Group 2, and 12: Group 1) (Table I). Importance for Group 1,
Group 2 and Group 3 has been considered as 1, 2, and 3, respectively. From Equation (2) S1
for (A, B):
P 24 N y x
k1 k k ABk
S 1A;B ¼ P24¼ ;
k¼1 N ky ky ABk

N 1y1xAB1 þN 2y2xAB2 þ · · · þN 24y24xAB24


S 1A;B ¼ ;
N 1y1 yAB1 þN 2y2 yAB2 þ · · · þN 24y24 yAB24

xAB, 1 for 4 Group 3 machines, 11 Group 2 machines, and 1 Group 1 machine; yABk, 1 for 4
Group 3 machines, 16 Group 2 machines, and 12 Group 1 machines (Table I). θk for Group
3 products ¼ 3, θk for Group 2 products ¼ 2, and θk for Group 1 products ¼ 1, therefore:

3 ~ 4 þ2 ~ 11 þ1 ~ 1
S 1A;B ¼ ¼ 0: :63
3 ~ 4 þ2 ~ 16 þ1 ~ 12

Similarly, S1 is calculated for (A, C), (A, D), (B, C), (B, D) and (C, D) shown in Table IV. From
Table II, S2 has been calculated. For products A and B, modules which are used for both
products ¼ 2, modules which are used for either A or B ¼ 6 (Table II). Importance given to
each module ¼ 1. Therefore:

1~2
S 2A;B ¼ ¼ 0:33:
1 ~6

Similarly, S2 is calculated for (A, C), (A, D), (B, C), (B, D) and (C, D) shown in Table V.
Similarity coefficient S has been calculated using Equation (3). For w ¼1 0.7 and w¼ 2 0.3;
similarity coefficient S for (A, B) (A, C), (A, D), (B, C), (B, D) and (C, D) has been shown
in Table VI.

Product B Product C Product D

Table IV. Product A 0.63 0.72 0.60


S1 for product Product B 0.66 0.66
A, B and C Product C 0.64
6.2 Formation of part family Reconfigurable
ALCA, used for part family formation, has been shown in Figure 2. In initial matrix shown manufacturing
in Table VI, max Sij ¼ 0.61, among products A and C, thus they are grouped in a family.
Sij are recalculated and shown in Table VII:
system
. Σ
S ððA;CÞ;BÞ ¼ S ðA;BÞ þS ðB;CÞ =2 ¼ ð0:54 þ 0:51Þ=2 ¼ 0:53;

. Σ
S ððA;CÞ;DÞ ¼ S ðA;DÞ þS ðC;DÞ =2 ¼ ð0:42 þ0:45Þ=2 ¼ 0:44;

S ðB;DÞ ¼ 0:46:
The new maximum value of the matrix is 0.53 corresponding to products A, C and B. These
products are grouped and similarity coefficient between (A, C, B) and D has been calculated
and shown in Table VIII:
. Σ
S ððA;C;BÞ;DÞ ¼ S ðA;DÞ þ S ðC;DÞ þS ðB;DÞ =3 ¼ ð0:42 þ 0:45 þ0:46Þ=3 ¼ 0:44:
The resulting dendrogram is shown in Figure 3. It shows four levels of part family, each
for different part family. It shows that products can be grouped as (A, B, C, D) (AC, B, D),
(ACB, D) and (ACBD).
For effective working of RMS, one part family is produced in the manufacturing system
and then the system is reconfigured for another part family. Here, there are four levels of
part family, thus four families can be produced. For each family, products are scheduled.
For the problem considered, the schedules will be as described below.

Product B Product C Product D

Product A 0.33 0.33 0 Table V.


Product B 0.14 0 S2 for products
Product C 0 A, B and C

Product B Product C Product D Table VI.


Similarity coefficient
Product A 0.54 0.61 0.42 (S ) for products
Product B 0.51 0.46 A, B, C and D
Product C 0.45 (initial matrix)

Product B Product D
Table VII.
Product (A, C) 0.53 0.44 Sub-matrix for
Product B 0.46 similarity coefficient

Product D Table VIII.


Final matrix for
Product (A, C, B) 0.44 similarity coefficient
BIJ For, part family level 1, there will be six schedules A → B → C → D → A,
A → B → D → C → A, A → C → B → D → A, A → C → D → B → A, A → D → B → C → A,
and A → D → C → B → A. For part family level 2, there will be two schedules, and for
part family level 3 and part family level 4, there will be only one schedule shown in Table X.
For each schedule at each level, the summation of reconfiguration cost, under-utilization
cost and floor space cost have been calculated and for minimum cost, the schedule is
selected. It will provide the optimum solution.

6.3 Selection of product family


For calculations of reconfiguration effort, α′ ¼0.6, β′ ¼0.4, g′ ¼ 0, α¼ 0.5, β ¼ 0.5, g¼ 0,
c1 ¼0, c2 0.4,
¼ c3 0.6.¼ For the calculation of under-utilization cost of machines for each
part family, unit cost has been assumed and shown in Table IX. Unit floor space cost is
considered 1 for all the workstations. Calculated values of reconfiguration effort, under-
utilization cost and floor space cost have been shown in Table X.

7. Results and discussion


Table X shows criteria, reconfiguration effort, under-utilization cost, and floor space cost
for all levels of part families formation. It becomes a multi-criteria decision-making

Products
A C B D
100%
L =1
61%
L =2
53%
Figure 3. L =3
Dendrogram 44%
L =4
0%

Modules/machine not in use Under-utilization cost

Table IX. A module is not used 1


Under-utilization cost A machine of Group 1 is not in use 4
for the machine/ A machine of Group 2 is not in use 7
module A machine of Group 3 is not in use 12

S. No. Schedule Reconfiguration effort Under-utilization cost Floor space cost

1 A→B→C→D→A 0.82 441 120


2 A→B→D→C→A 0.78 441 120
3 A→C→B→D→A 0.82 441 120
4 A→C→D→B→A 0.78 441 120
5 A→D→B→C→A 0.82 441 120
6 A→D→C→B→A 0.82 441 120
Table X. 7 AC → B → D → AC 0.63 537 142
Reconfiguration effort, 8 AC → D → B → AC 0.64 537 142
under-utilization cost, 9 ACB → D → ACB 0.46 633 176
and floor space cost 10 ACBD 0.00 825 216
problem, which has three criteria and ten alternatives. Salient points related to results are Reconfigurable
as following: manufacturing
(1) Since each machine is different, therefore, it has different reconfiguration effort, system
under-utilization and floor space cost. But for simplicity of calculations, machines
have been grouped into three types: Group 1, Group 2, and Group 3.
(2) Level of reconfiguration depends on the time period of configurations. If it is less,
heavy machines are not reconfigured. In the case study, there are 24 Group 1
machines (≈52 percent) which can be easily added or removed. Thus, in any
industry it is not necessary to use the concept of reconfiguration for all the machines,
but it can be used selectively.
(3) Weighted Jaccard similarity index has been used for the calculation of similarity.
Jaccard similarity index most used similarity index because of its simplicity.
Weights provide better results. Table VI provides the similarity coefficient between
part families A, B, C, and D. While Tables VII and VIII provides the similarity
between AC, B, and D, and ACB and D.
Literature shows that ALC algorithm gives better results.
(4) Reconfiguration effort has been calculated using relative weights. Whether it is for
system level or machine level. It provides a better understanding that how much
importance has been given to which type of effort.
(5) Weights of criteria have been calculated by the AHP method. For this purpose,
under-utilization has been given six times importance than reconfiguration effort
and three times to floor space cost. Floor space cost has been given two times
importance than reconfiguration cost. Comparison matrix has been formed. Weights
have been calculated by using R program. Weights have been calculated as 0.1111,
0.6667, and 0.2222.
(6) Ranking of the alternatives has been calculated using the reference ideal method.
Ranking has been calculated manually and with R program. It can be seen that
schedule A → B → D → C → A has first rank.
(7) The question arises the system should be designed for the all the products (ABCD) or
the concept of reconfiguration should be used. The answer is that in present conditions,
the concept of reconfiguration is useful but for families (A, B, C, D). But if floor space
cost or under-utilization cost changes then it can be designed for other families.

7.1 Managerial implications


RMS is designed for the part family. This paper provides the insight to the managers for
designing of a manufacturing system for a part family. It will help them for decision making
that manufacturing system should be designed for one part family or more than one.
It provides answers to the questions related to the part family formation. These questions
are: when it should be chosen for reconfiguration? How to group the families? Which are the
criteria that will affect the part family formation? How to calculate reconfiguration effort?
How to select the alternatives? The decision will also affect the required number of machines
because if the system is designed for the entire product, more machines will be required and
it will also acquire more floor space. With the increase of product variety, it is not possible to
design a cell for single product, therefore, reconfiguration is required.

8. Conclusions
Since the variety of products is increasing very fast, therefore, it has become important to
use the concept of reconfiguration. In a manufacturing system, by using the concept of
BIJ reconfiguration, machines can be reused by adding/removing some modules of machines.
It also saves the floor space by adding/removing some machines from the system. The
concept of reconfiguration becomes more useful when it is implemented with lean
manufacturing. Lean manufacturing techniques Jidoka and Poka-yoke are used to increase
the diagnosability of the system. Since each machine has different importance, therefore
machines are divided into groups. In this paper, a methodology has been used for
assessment of the reconfiguration in manufacturing system. Weighted Jaccard function has
been used for the calculation of similarity index. Levels of part families have been identified
using the ALC algorithm. Level of part families has been selected for reconfiguration effort,
under-utilization cost, and floor space cost. The methodology for calculation of
reconfiguration effort, under-utilization cost, and floor space cost has been presented.
Weights of the criteria have been calculated using AHP. For the problem considered weights
are 1,111, 0.6667, and 0.2222. Ranking of alternative has been obtained using the reference
ideal method. For the problem considered best choice is; part family formation as (A, B, C, D)
and schedule A → B → D → C → A. If the weight of reconfiguration effort is increased, the
part family formation (ABCD) becomes the best choice.

8.1 Limitations and future work


The problem considered is based on industrial case study. Similarity index has been
calculated within one layout. It is needed to calculate between two or more systems. Work is
needed related to automation and reconfiguration process plan.

References
Askin, R.G. and Zhou, M. (1998), “Formation of independent flow-line cells based on operation
requirements and machine capabilities”, IIE Transactions, Vol. 30 No. 4, pp. 319-329.
Baroni-Urbani, C. and Buser, M.W. (1976), “Similarity of binary data”, Systematic Biology, Vol. 25 No. 3,
pp. 251-259.
Benderbal, H.H., Dahane, M. and Benyoucef, L. (2018), “Modularity assessment in reconfigurable
manufacturing system (RMS) design: an archived multi-objective simulated annealing-based
approach”, The International Journal of Advanced Manufacturing Technology, Vol. 94 Nos 1/4,
pp. 729-749.
Bettaieb, C., Telmoudi, A.J., Sava, A. and Nabli, L. (2017), “Reconfigurable manufacturing system:
Overview and proposition of new approach”, IEEE International Conference on Control,
Automation and Diagnosis (ICCAD), pp. 534-539.
Cables, E., Lamata, M. and Verdegay, J. (2016), “Rim-reference ideal method in multicriteria decision
making”, Information Sciences, Vol. 337, pp. 1-10, available at: www.sciencedirect.com/science/
article/pii/S0020025515009007
Carrie, A. (1973), “Numerical taxonomy applied to group technology and plant layout”, International
Journal of Production Research, Vol. 11 No. 4, pp. 399-416.
ElMaraghy, H.A. (2005), “Flexible and reconfigurable manufacturing systems paradigms”,
International Journal of Flexible Manufacturing Systems, Vol. 17 No. 4, pp. 261-276.
Galan, R., Racero, J., Eguia, I. and Canca, D. (2007), “A methodology for facilitating reconfiguration in
manufacturing: the move towards reconfigurable manufacturing systems”, The International
Journal of Advanced Manufacturing Technology, Vol. 33 Nos 3/4, pp. 345-353.
García-Cascales, M.S. and Lamata, M.T. (2012), “On rank reversal and TOPSIS method”,
Mathemematical and Computer Modelling, Vol. 56 No. 5, pp. 123-132.
Goyal, K.K., Jain, P. and Jain, M. (2013a), “A comprehensive approach to operation sequence similarity
based part family formation in the reconfigurable manufacturing system”, International Journal
of Production Research, Vol. 51 No. 6, pp. 1762-1776.
Goyal, K.K., Jain, P.K. and Jain, M. (2013b), “A novel methodology to measure the responsiveness of Reconfigurable
RMTs in reconfigurable manufacturing system”, Journal of Manufacturing Systems, Vol. 32 manufacturing
No. 4, pp. 724-730.
system
Irani, S.A. and Huang, H. (2000), “Custom design of facility layouts for multiproduct facilities using
layout modules”, IEEE Transactions on Robotics and Automation, Vol. 16 No. 3, pp. 259-267.
Jaccard, P. (1908), “Nouvelles Recherches Sur la Distribution Florale”, Bulletin de la Societe Vaudoise des
Sciences Naturelles, Vol. 44, pp. 223-270, doi: 10.5169/seals-268384.
Koren, Y., Heisel, U., Jovane, F., Moriwaki, T., Pritschow, G., Ulsoy, G. and Van Brussel, H. (1999),
“Reconfigurable manufacturing systems”, CIRP Annals – Manufacturing Technology, Vol. 48
No. 2, pp. 527-540.
Lameche, K., Najid, N.M., Castagna, P. and Kouiss, K. (2017), “Modularity in the design of
reconfigurable manufacturing systems”, IFAC-PapersOnLine, Vol. 50 No. 1, pp. 3511-3516.
Lee, G.H. (1997), “Reconfigurability consideration design of components and manufacturing systems”,
The International Journal of Advanced Manufacturing Technology, Vol. 13 No. 5, pp. 376-386.
Lee, S., Ryu, K. and Shin, M. (2017), “The development of simulation model for self-reconfigurable
manufacturing system considering sustainability factors”, Procedia Manufacturing, Vol. 11,
pp. 1085-1092.
Li, X., Bayrak, A.E., Epureanu, B.I. and Koren, Y. (2018), “Real-time teaming of multiple reconfigurable
manufacturing systems”, CIRP Annals, Vol. 67 No. 1, pp. 437-440.
McAuley, J. (1972), “Machine grouping for efficient production”, Production Engineer, Vol. 51 No. 2,
pp. 53-57.
Maganha, I., Silva, C. and Ferreira, L.M.D. (2018), “Understanding reconfigurability of manufacturing
systems: an empirical analysis”, Journal of Manufacturing Systems, Vol. 48, pp. 120-130.
Mittal, K.K., Jain, P.K. and Kumar, D. (2017), “Configuration selection in reconfigurable manufacturing
system based on reconfigurability”, International Journal of Logistics Systems and Management,
Vol. 27 No. 3, pp. 363-379.
Mortensen, S.T., Chrysostomou, D. and Madsen, O. (2017), “A novel framework for virtual
recommissioning in reconfigurable manufacturing systems”, IEEE International Conference on
Emerging Technologies and Factory Automation (ETFA), pp. 1-4.
Mosier, C. and Taube, L. (1985), “Weighted similarity measure heuristics for the group technology
machine clustering problem”, Omega, Vol. 13 No. 6, pp. 577-579.
Prasad, D. and Jayswal, S.C. (2017a), “Case study of a reconfigurable manufacturing industry”, in
Chauhan, A.K. (Ed.), International Conference on Innovations and Developments in Mechanical
Engineering (IDME’17), KNIT, Sultanpur, pp. 32-36.
Prasad, D. and Jayswal, S.C. (2017b), “Design of reconfigurable manufacturing system”, in Jayswal, S.C.
and Prasad, R.B. (Eds), National Conference on Futuristics in Mechanical Engineering
(FME-2016), Madan Mohan Malviya University of Technology, Gorakhpur.
Prasad, D. and Jayswal, S.C. (2017c), “Reconfigurability consideration and scheduling of products
in a manufacturing industry”, International Journal of Production Research, Vol. 56 No. 19,
pp. 6430-6449.
Prasad, D. and Jayswal, S.C. (2018a), “Scheduling in reconfigurable manufacturing system for
uncertainty in decision variables”, Materials Today: Proceedings, Vol. 5 No. 9, pp. 18451-18458.
Prasad, D. and Jayswal, S.C. (2018b), “Scheduling of products for reconfiguration effort in reconfigurable
manufacturing system”, Materials Today: Proceedings, Vol. 5 No. 2, pp. 4167-4174.
Prasad, D. and Jayswal, S.C. (2018c), “A review on flexibility and reconfigurability in manufacturing
system”, in Chattopadhyay, J., Singh, R. and Prakash, O. (Eds), Innovation in Materials Science
and Engineering-Proceedings of ICEMIT, Vol. 2, Amity University, Ranchi, pp. 187-200.
Rogers, G. and Bottaci, L. (1997), “Modular production systems: a new manufacturing paradigm”,
Journal of Intelligent Manufacturing, Vol. 8 No. 2, pp. 147-156.
BIJ Saaty, T.L. and Sagir, M. (2009), “An essay on rank preservation and reversal”, Mathematical and
Computer Modelling, Vol. 49 No. 5, pp. 1230-1243.
Sarker, B.R. and Islam, K.M.S. (1999), “Relative performances of similarity and dissimilarity measures”,
Computers & Industrial Engineering, Vol. 37 No. 4, pp. 769-807.
Seifoddini, H. and Wolfe, P.M. (1986), “Application of the similarity coefficient method in group
technology”, IIE Transactions, Vol. 18 No. 3, pp. 271-277.
Seifoddini, H.K. (1989), “Single linkage versus average linkage clustering in machine cells formation
applications”, Computers & Industrial Engineering, Vol. 16 No. 3, pp. 419-426.
Shaik, A.M., Rao, V.K. and Rao, C.S. (2015), “Development of modular manufacturing systems – a
review”, The International Journal of Advanced Manufacturing Technology, Vol. 76 Nos 5/8,
pp. 789-802.
Taylor, F.W. (2004), Scientific Management, Routledge, London, available at: https://doi.org/10.4324/
9780203498569
Wang, Y.-M. and Luo, Y. (2009), “On rank reversal in decision analysis”, Mathematical and Computer
Modelling, Vol. 49 No. 5, pp. 1221-1229.
Witte, J.D. (1980), “The use of similarity coefficients in production flow analysis”, International Journal
of Production Research, Vol. 18 No. 4, pp. 503-514.
Youssef, A.M. and El-Maraghy, H.A. (2006), “Assessment of manufacturing systems reconfiguration
smoothness”, The International Journal of Advanced Manufacturing Technology, Vol. 30 Nos 1/2,
pp. 174-193.
Youssef, A.M. and ElMaraghy, H.A. (2007), “Optimal configuration selection for reconfigurable
manufacturing systems”, International Journal of Flexible Manufacturing Systems, Vol. 19 No. 2,
pp. 67-106.

Corresponding author
Durga Prasad can be contacted at: dp.mmmut@gmail.com

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com

You might also like