Design and Optimization of Production Lines

Design and
Optimization of
Production Lines
Edited by
Paolo Renna and Michele Ambrico

Printed Edition of the Special Issue Published in Applied Sciences
www.mdpi.com/journal/applsci
Design and Optimization of
Production Lines
Design and Optimization of
Production Lines
Editors
Paolo Renna
Michele Ambrico
MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin
Editors
Paolo Renna Michele Ambrico
School of Engineering, CRF WCM Research and Innovation,
University of Basilicata Campus Manufacturing Melfi
Italy Italy
Editorial Office
MDPI
St. Alban-Anlage 66
4052 Basel, Switzerland
This is a reprint of articles from the Special Issue published online in the open access journal
Applied Sciences (ISSN 2076-3417) (available at: https://www.mdpi.com/journal/applsci/special
issues/Production Lines).
For citation purposes, cite each article independently as indicated on the article page online and as
indicated below:
LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. Journal Name Year, Volume Number,
Page Range.
ISBN 978-3-03943-961-4 (Hbk)

ISBN 978-3-03943-962-1 (PDF)

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative
Commons Attribution (CC BY) license, which allows users to download, copy and build upon
published articles, as long as the author and publisher are properly credited, which ensures maximum
dissemination and a wider impact of our publications.
The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons
license CC BY-NC-ND.
Contents
About the Editors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Paolo Renna and Michele Ambrico

Special Issue “Design and Optimization of Production Lines”
Reprinted from: Appl. Sci. 2020, 10, 8302, doi:10.3390/app10228302 . . . . . . . . . . . . . . . . . 1
Pyung-Hoi Koo
A New Self-Balancing Assembly Line Based on Collaborative Ant Behavior
Paolo Renna and Sergio Materi

Design Model of Flow Lines to Include Switch-Off Policies Reducing Energy Consumption
Irina Aleksandrova, Eugeniy Koshelev and Elena Koresheva

In-Line Target Production for Laser IFE
Reprinted from: Appl. Sci. 2020, 10, 686, doi:10.3390/app10020686 . . . . . . . . . . . . . . . . . . 37
Marco Antonio Zamora-Antuñano, Jorge Cruz-Salinas, Juvenal Rodrı́guez-Reséndiz, Carlos

Alberto González-Gutiérrez, Néstor Méndez-Lozano, Wilfrido Jacobo Paredes-Garcı́a, José
Antonio Altamirano-Corro and José Alfredo Gaytán-Dı́az
Statistical Analysis and Data Envelopment Analysis to Improve the Efficiency of Manufacturing
Process of Electrical Conductors
Reprinted from: Appl. Sci. 2019, 9, 3965, doi:10.3390/app9193965 . . . . . . . . . . . . . . . . . . . 55
Jure Murovec, Janez Kušar and Tomaž Berlec

Methodology for Searching Representative Elements
Reprinted from: Appl. Sci. 2019, 9, 3482, doi:10.3390/app9173482 . . . . . . . . . . . . . . . . . . . 71
v
About the Editors
Paolo Renna is an Associate Professor at the School of Engineering of the University of
Basilicata. His academic research principally deals with the development of innovative negotiation
and production planning in distributed environments. Several of his contributions have been
presented on designing multi-agent architecture and testing by discrete event simulation in
business-to-business environments. Among these contributions, he is co-author of two research
books about e-marketplaces and production planning in production networks. Moreover, he
has developed coordination approaches in multi-plant production planning environments and
innovative scheduling approaches in flexible and reconfigurable manufacturing systems.
Michele Ambrico is currently a Manufacturing Methodologies Specialist at CRF/WCM Campus

Manufacturing in Melfi. He primarily works on numerical simulation for process/product
optimization. He received his PhD at the Department of Environmental Engineering and Physics
in the Engineering Faculty of the University of Basilicata (Italy), where he also received his degree
in Industrial Engineering. His academic studies are focused on the design of robust cellular
manufacturing systems and the development of simulation models in dynamic environments. He
collaborates with the manufacturing system research group at the School of Engineering (University
of Basilicata) on the development of simulation environments for the design and control of
manufacturing systems.
vii
applied
sciences
Editorial
Special Issue “Design and Optimization of
Production Lines”
Paolo Renna 1, * and Michele Ambrico 2
1 School of Engineering, University of Basilicata, 85100 Potenza, Italy
2 CRF WCM Research and Innovation, Campus Manufacturing Melfi, 85025 Potenza, Italy;
michele.ambrico@crf.it
* Correspondence: paolo.renna@unibas.it

Received: 16 November 2020; Accepted: 19 November 2020; Published: 23 November 2020
The classical models for designing production lines follow the objective of balancing the line so as
to improve the throughput. Nowadays, the development of sustainable processes is a strategic aspect
for the manufacturing industry, and it is a central theme in current innovation projects; the industrial
manufacturers study energy-efficient models because of the costs and environmental impact of energy
consumption. The latest trends of design and optimization models include the management of
reconfigurable machines, switch-off policies, buffer control, and so on, to increase robustness and
modularity of production lines and reduce energy consumption.
The “Fourth Industrial Revolution” (alternatively known as “Industry 4.0”) supports innovative
models for energy consumption and fault tolerance in automated lines, and this drives the changes in
design and optimization models for the production lines.
To meet the objective of sustainable production lines in terms of energy consumption,
peak electricity demand and energy efficiency including Industry 4.0 technologies, new innovative
models are needed to support the design and management of production lines.
This book includes a series of five research studies that reveal new knowledge about the design
and management of sustainable production lines.
The topics covered span many diverse areas associated with the design and management of
production lines, such as: production improvement in uncertain environments [1], design of production
lines to introduce switch-off policies [2], new technologies [3], statistical data analysis [4,5].
In combination, these complementary contributions provide a substantial body of knowledge in
the context of production lines that are currently undergoing an epochal industrial transformation.
Funding: This research received no external funding.

Acknowledgments: This publication was only possible with the invaluable contributions from the authors,
reviewers, and the editorial team of Applied Sciences.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Koo, P.-H. A New Self-Balancing Assembly Line Based on Collaborative Ant Behavior. Appl. Sci. 2020, 10,
6845. [CrossRef]
2. Renna, P.; Materi, S. Design Model of Flow Lines to Include Switch-Off Policies Reducing Energy Consumption.
Appl. Sci. 2020, 10, 1475. [CrossRef]
3. Aleksandrova, I.; Koshelev, E.; Koresheva, E. In-Line Target Production for Laser IFE. Appl. Sci. 2020, 10, 686.
[CrossRef]
Appl. Sci. 2020, 10, 8302; doi:10.3390/app10228302 1 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 8302
4. Zamora-Antuñano, M.A.; Cruz-Salinas, J.; Rodríguez-Reséndiz, J.; González-Gutiérrez, C.A.; Méndez-Lozano, N.;
Paredes-García, W.J.; Altamirano-Corro, J.A.; Gaytán-Díaz, J.A. Statistical Analysis and Data Envelopment
Analysis to Improve the Efficiency of Manufacturing Process of Electrical Conductors. Appl. Sci. 2019, 9, 3965.
[CrossRef]
5. Murovec, J.; Kušar, J.; Tomaž Berlec, T. Methodology for Searching Representative Elements. Appl. Sci. 2019,
9, 3482. [CrossRef]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
2
applied
sciences
Article
A New Self-Balancing Assembly Line Based on
Collaborative Ant Behavior
Pyung-Hoi Koo
Department of Systems Management & Engineering, Pukyong National University, Busan 48513, Korea;
phkoo@pknu.ac.kr; Tel.: +82-51-629-6485

Received: 29 August 2020; Accepted: 28 September 2020; Published: 29 September 2020
Abstract: In most mass-production assembly lines, workers perform a set of tasks repetitively
predefined by assembly line balancing techniques. The static task assignment often leads to low
productivity when the assembly system faces disruptions or uncertainties such as machine breakdown
and uneven worker capabilities. The idea of bucket brigades (BB) has been introduced to address
the static assignment problems where cooperative behavior of ants is applied to flow line control.
This paper examines possible efficiency losses associated with the existing BB-based assembly cell
and presents an improved version for assembly cells under uncertain environments. The new system
attempts to enhance productivity by assigning assembly tasks to workers dynamically and possibly
adding buffers for decoupling consecutive workers. The proposed assembly system is evaluated
through simulation experiments under various manufacturing environments. The experimental
results show that the new system provides higher productivity than the naïve BB-based assembly cell
as well as traditional assembly cells, especially for uncertain assembly environments.
Keywords: assembly systems; assembly cell; line balancing; dynamic task assignment; bucket
brigades; self-balancing; autonomous control
1. Introduction
A common assembly system consists of m workers (or workstations) arranged along a production
line. The components are fed into the line and progressively assembled as they move down from
station to station. At a workstation k (= 1, . . . , m), a worker performs a set of assembly tasks, Sk ,

repeatedly. It takes station time, t(Sk ) = j∈Sk t j to complete all the tasks assigned to workstation k
where tj is the assembly time for task j (=1, . . . , n). The cycle time, c, of the assembly line is determined
by a bottleneck station whose station time is the largest, i.e., c = maxk t(Sk ). The capacity of the
assembly line is 1/c. When an assembly line is operated in a full capacity, station k with t(Sk ) < c has
an idle time of c − t(Sk ) in each production cycle. Line balance efficiency E is defined by E = mc T

where T = t j . The assembly lines are widely employed in various industries for the production of
standardized products.
One of the important decisions in the assembly line involves the work assignment regarding how
much work to assign to each worker. Decisions about assembly task assignments are made through
assembly line balancing (ALB) techniques whose key objective is to obtain a balanced assembly line by
assigning assembly tasks to each worker as evenly as possible. The existing studies on ALB problems
are well summarized in survey papers [1–4]. Most existing studies on ALB problems deal with a static
line-balancing solution with which the workers perform a pre-assigned set of assembly tasks from cycle
to cycle. In general, the ALB decisions are made at the early stage of the assembly line design process
in which assembly time for an individual task is assumed to be constant. However, the real industrial
environment is often different from the situations on which the ALB decisions are based. Manufacturing
systems undergo various types of uncertainties including a machine breakdown, different worker

Appl. Sci. 2020, 10, 6845
skills, rework requirements, and quality problems. With the uncertainties, the assembly time fluctuates
over time so that the assembly lines often fail to achieve the performance as expected [5]. When a
new product is introduced, which frequently happens in today’s fast-moving market, the assembly
systems should be reconfigured. The frequent reconfiguration results in low performance in the
traditional assembly line with a static task assignment. Moreover, it is common that the workers have
different work speeds and each worker cannot perform assembly tasks at a constant speed throughout
a day. When uncertainties are common, it has become more important for production systems to
have the flexibility to respond quickly to uncertainties. Assembly cells have been introduced to solve
these problems of assembly lines (e.g., [6–9]). A long assembly line is reconfigured into several small
assembly cells in each of which a small number of workers (or a single worker in an extreme case)
perform the entire range of assembly tasks for a final product. The assembly cells are now widely
implemented in industries because they provide better manufacturing performance than traditional
assembly lines in terms of productivity, product quality, and customer responsiveness. However,
the assembly cell still has a workload balancing problem, especially under system uncertainty.
To solve the balancing problems associated with static task assignments, a new assembly line
control method has been introduced where the idea of a bucket brigade (BB) is used as control logic in
assembly lines [10–14]. The BB-based assembly cell imitates a cooperative behavior of ants in nature,
as shown in Figure 1a. When some ant colonies find a large amount of food in one place, they move
the food items to their nest in a collaborative way as follows [15]: An ant carries her food item along
the trail. When she encounters a downstream ant (i.e., an ant closer to the nest), she hands over the
food item to the downstream ant and returns back toward the food source to move another food item.
The recipient ant carries the food item further down the trail to the nest until she encounters an unladen
ant (or the nest), and so on. In this control logic, the amount of work that an ant performs depends on
her capability and system status (e.g., a road condition). Stronger ants can perform more tasks while
weaker ants less.
finished
parts assembly line products
assemblers
(a) (b)
Figure 1. Bucket brigades (BBs): (a) cooperative food transport of ants in nature and (b) BB-based
assembly cell.
Figure 1b shows a BB-based assembly cell. The following is how the BB-based cells work [16]:
“Each worker assembles a product towards completion. Every worker except for the last one hands his current job
over to his immediate successor (downstream worker) who is available and goes back upstream to take a new
assembly job from his immediate predecessor (upstream worker). The last worker completes assembling all the
remaining assembly tasks for a product, puts the final product in the output storage, and walks back upstream.
After handing an assembly job over to his immediate successor, the first worker walks back to the beginning
of the line to start a new assembly job.” The BB-based assembly cell is characterized as distributed and
autonomous control. There is a reduced need for central planning and management because bucket
brigades make the flow line self-balancing. The workers do not perform only the jobs previously
assigned to them but perform the jobs that may be different from cycle to cycle. The balancing problems
are mitigated by the self-balancing property. Each worker performs an assembly task with a velocity
commensurate with his own capability. The throughput may be increased because bucket brigades
spontaneously generate the optimal division of work. However, in real-world assembly operations,
some situations undermine the performance of the BB-based cell. Two kinds of efficiency losses
4
Appl. Sci. 2020, 10, 6845
have been identified when the concept of the BB is applied to industrial systems: task hand-offs and
worker blocking [17]. Let us consider a BB-based assembly line in which each assembly task should be
performed by only one operator without any interruption (i.e., the task is not divisible.) In this case,
once a worker starts an assembly task on a product, he should complete the task. Then, it may happen
that a downstream worker cannot take over an assembly task from the upstream worker immediately
after he is available. Then, the downstream worker should wait (hand-off loss) until the predecessor
finishes his current job. Moreover, the assembly time for a task is often not constant but variable over
time. In this case, a situation may happen that an upstream worker cannot start a new assembly task
(blocking loss) when his downstream worker is struggling to finish the same task.
This paper examines performance losses including hand-offs and blocking in the existing BB-based
assembly cell and presents an improved version of the BB-based cell to reduce performance losses.
The proposed system aims to enhance productivity by assigning assembly tasks to workers dynamically
and possibly adding intermediate buffers for decoupling consecutive workers. The assembly systems
under consideration have the following assumptions:
− A single type of product is assembled in the assembly system.

− Multiple elementary assembly tasks are needed to produce a final product.
− Each task is indivisible in assembly. A half-finished assembly task cannot be transferred to the
downstream worker.
− Assembly time for a task may not be deterministic but stochastic with some variabilities.
− The sequence of assembly tasks is predefined on the basis of their precedence constraints.
− The sequence of the worker is fixed (no upstream worker can pass through his downstream
worker).
− Two workers are not allowed to perform the same assembly task at the same time.
This paper is organized as follows. Section 2 reviews the literature on previous studies on
BB-based assembly systems. In Section 3, we present the problems associated with efficiency losses in
BB-based cells. A production capacity model in BB-based cells is introduced by considering hand-off
and blocking delays. Section 4 presents an improved version of the existing BB-based cell, and Section 5
compares the performance of the new assembly cell with that of the other assembly systems through
simulation experiments. Finally, we give concluding remarks in Section 6.
2. Literature Review
Assembly line balancing (ALB) is a practical industrial problem that should be solved in
many different industries. The main task of ALB decisions is to allocate elementary assembly
tasks to workers (or workstations), typically intending to maximize line efficiency. As there are
many different assembly environments, the research works on the ALB problems have extensions
including multi-manned workstations [18,19], multi-product assembly lines [20,21], U-shaped assembly
lines [22,23], asynchronous assembly lines [24], and robotic assembly lines [25]. Most existing studies
deal with ALB problems by using optimization and/or heuristic models. The models provide
static line-balancing solutions with which the workers perform a pre-assigned set of assembly tasks
from cycle to cycle. Even though the optimization/heuristic models are expected to perform well
under static manufacturing environments, high performance may not be realized under dynamic
manufacturing environments.
A BB-based assembly cell has been introduced where assembly tasks are not assigned to a worker
in a static way but can be performed by any worker [16]. Each worker follows a simple control logic,
which is discussed in Figure 1b. The authors of [16] claim that the BB-based system can achieve
the highest productivity when the following three conditions are satisfied: (1) the walk-back time is
ignorable, (2) the task can be handed over immediately when the downstream worker is available,
and (3) each assembler i = 1, . . . , m is characterized by a distinct constant work velocity vi . The work
velocity for each assembler is deterministic. It is proved under the three ideal conditions that if the
5
Appl. Sci. 2020, 10, 6845
workers are sequenced along the flow line from slowest to fastest, the system becomes self-balanced
such that the hand-offs between any two consecutive workers take place at a fixed location, and the

production rate per unit time converges to ni = 1 vi , which is the maximum attainable from the
given workers [16]. The problem is that the ideal conditions are not satisfied in most real-world
production environments.
Reference [26] shows that the BB-based system can be effective even in the presence of stochastic
task times. Reference [27] presents a case study where a traditional tractor assembly line was replaced
by a BB-based assembly line. Here, walk-back times were explicitly considered. It is reported that
the productivity increased by adopting the BB concept more than compensated for the time lost by
walk-back times. Reference [28] examined the effect of labor turnover on the performance of different
assembly methods. A simulation study showed that the BB-based assembly line outperformed a
traditional assembly line with fixed work allocation especially when the labor turnover rate was high.
Reference [10] explores how the BB protocol can be extended to make subassembly lines self-balancing
where all the subassembly lines should be synchronized to produce at the same rate. Reference [29]
examined a two-worker BB system where one worker was faster than the other over some part of the
production line and slower over another part of the line. Reference [30] addresses blocking problems
in a BB-based production system where blocking occurred only when the preceding worker was faster
than the downstream worker. Deterministic task time was assumed. Initial positions of workers were
identified where no blockings occurred. Reference [31] introduces a case study where a BB-based
system was applied to an assembly line of fashion bags, showing the advantages of a BB-based system
in terms of flexibility, low work-in-process (WIP) levels, and the ability to handle uncertainties. In the
warehouse order picking environment, analytical models have been presented to quantify the losses by
hand-off delay [32] and blocking delay [33]. References [14,34] address the walk-back time problems in
a BB-based production system. A new design of bucket brigade was proposed to reduce unproductive
travel with a U-shaped production cell where workers work on one side of the aisle for the forward
direction and they work on the other side for the reverse direction. Reference [14] explicitly considers
hand-off time and walk-back time in a U-shaped BB system. It is argued that the U-shaped BB system
can be more productive than the traditional BB system if the number of workers in the cell is small.
The uncertainties in the manufacturing system are not considered in the model.
As reviewed above, many studies have focused on analytical models dealing with the behavior
of the BB-based system where each worker has a distinct and constant work speed, and the workers
are sequenced from slowest to fastest. Although some research works address practical issues by
presenting improved versions of BB systems by applying BB protocols to U-shaped assembly cells [14]
and sub-assembly cells [10], little work has been done for addressing hand-off and blocking problems
in existing BB-based systems under uncertain environments. This paper aims to fill the research gap of
the existing works. We examine possible efficiency losses associated with hand-off and blocking delays
in the existing BB-based assembly system and present a production capacity model in BB-based cells
with hand-off and blocking delays. An improved BB-based system is introduced for assembly cells
under an uncertain environment that attempts to enhance productivity by assigning assembly tasks
to workers dynamically and possibly adding buffers for decoupling consecutive workers. We also
examine the effect of the sequence of workers in the new BB system.
3. The Efficiency of BB-Based Assembly Cell

We now examine efficiency losses in the BB-based system caused by hand-offs and worker
blocking in this section on assembly lines under uncertainties. Let us first examine the effect of the
uncertain events on the assembly time distribution. The uncertain events in assembly systems include
assembly tool breakdown, defected parts, workers’ job skills, and utility problems, to name a few.
Even though the net assembly task time is deterministic, the uncertain events make effective assembly
times stochastic. Suppose net assembly time for a task is deterministic with ts and an assembly tool
is broken in the middle of an assembly activity. Then, the assembly time is expanded to the net
6
Appl. Sci. 2020, 10, 6845
assembly time plus an independent tool replacement time. Suppose the tool breakage takes place
with probability p and it requires tr minutes on average to fix the problem. The effective task time will
be ts with probability 1-p and ts + tr with probability p. On average, the effective assembly time, τs ,
will be τs = (1 − p)ts + p(ts + tr ) = ts + p·tr . Now, suppose the tool breakage time is stochastic with a
coefficient of variation, vr . The coefficient of variation (CV) is defined as the standard deviation over
the average of a variable. Then, the CV of the effective assembly time, vs , is calculated by using the
model presented in [35] as follows:

t2s + ps v2r t2r + 2tr ts + t2r
vs = −1 (1)
τ2s
To understand the effect of uncertain events on the effective assembly time, let us take a simple
example. Suppose the net assembly time is constant with 3.0 min, and it undergoes a tool breakdown
with a probability of 0.02 following an exponential distribution of average 9.0 min. Then, the average
assembly time is prolonged from 3.0 min to 3.18 min (= 3.0 + 0.02 · 9.0) and the CV of the effective
assembly time is 0.56, calculated from Equation (1). Hence, even with a constant net assembly time,
it is reasonable to assume that the effective assembly time is not deterministic but stochastic in the
system under uncertainties.
Figure 2 shows a situation where a hand-off loss happens. Suppose the upstream worker is
performing an assembly task starting at time ta, whose expected assembly time is (tc − ta) time units.
If the downstream worker becomes available and ready to take over a new assembly job at time tb
from the upstream worker who is performing the assembly task, he should wait until the upstream
worker finishes his current task. Since the current task is expected to be finished at time tc, the expected
waiting time of the downstream worker for the hand-off is (tc − tb) time units.
task finish time

taG tb tcG
(probabilistic)G
Expected hand-off
Upstream workerG Upstream workerG
duration (tc-tb) G
starts new task G Finishes the taskG
Downstream workerG
becomes availableG
Figure 2. Hand-off loss in BB-based assembly cell. The expected hand-off time is tc − tb.
We now estimate the performance of workers when the hand-off times are present. We will adopt
E [ t2 ]
the result of [36] where it is proved that the expected time per each hand-off, ho , is ho = 2E[t] , where t
is a variable representing assembly time. Since E[t2 ] = V(t) + E[t]2 = (cv(t)2 + 1) E[t]2 and E[t] = T/n,
we have
E t2 cv(t)2 + 1 T
ho = = (2)
2E[t] 2n
where V(t) is the variance of t, cv(t) is the CV of t, and T is the total assembly time for a final product.
It is seen from Equation (2) that there is a hand-off time even with no assembly time variability with
cv(t) = 0.
Consider an m-worker assembly line dedicated to a product type where the final product requires
n assembly tasks. When all the workers are fully utilized, then the maximum production rate per unit
7
Appl. Sci. 2020, 10, 6845
time of the assembly line is m/T. In the BB-based cell, each product experiences (m − 1) hand-offs in the
m-worker line. Then, the total expected hand-off time experienced by a final product, h, is as follows:

E t2 (m − 1) cv(t)2 + 1 T
h = (m − 1) = (3)
2E[t] 2n
To complete assembling a product on a BB-based cell, assemblers spend T time units for performing
assembly tasks and h time units being idle. Hence, (T + h) time units are required to complete a product.
Out of (T + h) time units, only T time units are value-added time. The efficiency loss percentage due
h
to hand-off delays, Eloss , is Eloss = T+ h . Let the average efficiency of workers be Eo . Then, Eo can be
obtained by Eo = 1 − Eloss = T+h . Then, the production capacity of m-worker BB-based cell, r, is as
T
follows:
m m m
r = E0 = = (4)
T T + (m − 1) cv(t)2 + 1 T/(2n) T 1 + (m − 1) cv(t)2 + 1 /(2n)
From Equations (3) and (4), it is seen that the production capacity of the BB-based cell is a
function of the number of assembly tasks (n), the number of workers (m), and the CVs of task times.
The production capacity increases as the number of assembly tasks increases and the number of
operators decreases. Figure 3 shows the effects of the number of workers and the number of assembly
tasks on the efficiency loss caused by hand-offs in the case of (a) deterministic task time with cv = 0.0
and (b) stochastic task time with cv = 1.0. When the number of assembly tasks is large and the number
of operators is small, the proportion of the hand-off time in the total assembly time becomes small, thus
decreasing the productivity loss. Note that n/m indicates the average number of assembly tasks handled
by a single worker. If n/m becomes very large, the efficiency loss caused by hand-offs approaches
zero. The results show that assigning a small number of operators to the BB-based cell is preferable.
Comparing Figure 3a,b, we can also find that the system with higher variability of task time leads to
higher hand-off losses.
m=2 m=3 m=4 m=5 m=2 m=3 m=4 m=5

efficiency loss due to hand-offs
efficiency loss due to hand-offs
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
10 15 20 25 30 10 15 20 25 30
number of assembly tasks number of assembly tasks
(a) deterministic task time (cv = 0.0) (b) stochastic task time (cv = 1.0)
Figure 3. Efficiency loss in BB assembly cell with hand-off times under (a) deterministic task time
(cv = 0.0) and (b) stochastic task time (cv = 1.0). It is seen that the hand-off losses are increased by (1)
higher task time variabilities, (2) more workers in a cell, and (3) fewer tasks assigned to a worker.
System efficiency can be also undermined by blocking phenomena in the BB-based cell. When an
upstream worker finishes an assembly task but finds his downstream worker performing the next
task, he cannot start the next assembly task but wait until the downstream worker has finished his
current job. Even though each product requires the same assembly tasks, the blocking may happen
when assembly times are not deterministic. (Note that a major reason for blocking in a warehouse
order picking system is different order contents form order to order [33]). The blocking not only has an
effect on production capacity but also production lead time. Simulation experiments are performed to
examine the blocking time in a three-worker BB-based cell where a single product type requiring 12
8
Appl. Sci. 2020, 10, 6845
assembly tasks is assembled. Figure 4a shows the ratio of blocking time over the total net assembly
time of a product over different assembly time variabilities. It is found in Figure 4a that when the
assembly times are deterministic with cv = 0.0, no blocking occurs. As the assembly time variabilities
increase, blocking time increases (Figure 4a), resulting in a lower production rate (Figure 4b).
20% 600
500
blocking time %
production rate
15%
400
10% 300
200
5%
100
0% 0
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
CV of task time CV of task time
(a) blocking time percentage (b) production rate

Figure 4. Effect of blocking over different assembly time variabilities in terms of (a) blocking time
percentage over total assembly time and (b) production rate.
4. A New Control Method of BB-Based Assembly Cell

We examined efficiency losses caused by hand-offs and blockings in the BB-based cells in the
previous section. To handle efficiency problems in the existing BB-based assembly cell, this paper
presents an improved version. (For comparison, the existing naïve BB-based assembly cell and the
newly proposed BB assembly cell are called NBB and ABB cell, respectively). The idea for the new
model was obtained from the BB-based system in [17], where a BB-based order picking and a zone
picking were integrated to reduce the hand-off and blocking losses in the order picking system. Figure 5
illustrates a three-worker ABB cell to assemble a product type with 12 assembly tasks. The major
difference between NBB cell and ABB cell is twofold: (1) there is a limited set of tasks that each worker
can perform in ABB cell, and (2) there may be buffer space at the end of each worker except the last
worker. Each worker is not allowed to perform assembly tasks out of the assigned set. In Figure 5,
worker 1 is allowed to perform assembly jobs between task A1 and A6, worker 2 between A2 and A9,
and worker 3 between A3 and A12.
B1 B2
Assembly tasks
A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12
worker 1
worker 2
worker 3
Figure 5. Task assignment of workers in the proposed BB assembly (ABB) cell. B1 and B2 indicate
the buffers.
In an ordinary situation, the workers follow a naive control logic of the NBB cell. The new control
logic comes into effect when a worker finishes all the tasks within the assigned tasks but his immediate
downstream worker is not available. For example, if worker 1 (W1) finishes all the tasks up to task
A6, but finds worker 2 (W2) busy, then W1 cannot start performing a new assembly task beyond
A6 but waits until W2 is available to take his job. (Note that in the NBB cells, W1 will attempt to
start A7 assembly tasks.) When W2 is available, W1 hands over his semi-finished assembly to W2
and returns back upstream to the beginning of the assembly line to start the first assembly task for
a new product. If buffer space is allowed at buffer B1, then W1 puts the semi-finished assembly at
9
Appl. Sci. 2020, 10, 6845
buffer B1 after finishing all the tasks assigned to him and returns to the beginning of the assembly line.
When ready for the next assembly job, W2 moves upstream and checks if there is any semi-finished
product in buffer B1. If there exists at least one workpiece at B1, W2 starts the assembly task with it.
Otherwise, as in the NBB cell, W2 goes upstream to take over the assembly job from W1. Worker 3 (W3)
completes assembly jobs by performing task A12 and puts the finished product at output storage and
moves upstream. If there is a semi-finished assembly at buffer B2, W3 starts the assembly task with it.
Otherwise, W3 goes upstream to take a job from W2. Note that if there is a workpiece in a buffer, it
means that the upstream worker does not experience blocking and the downstream worker does not
experience a hand-off time loss. Hence, efficiency losses caused by blocking as well as hand-offs are
expected to be reduced with the new system. With each worker assigned to limited work contents,
they perform assembly tasks far from each other, possibly resulting in less blocking frequencies.
5. Simulation Experiments and Results

Simulation studies were conducted using a simulation package, ARENA, to evaluate the
performance of the proposed assembly method. The experiments were performed with assembly
systems where only one product type (requiring 12 assembly tasks) was assembled. The performance
of the proposed BB-based (ABB) assembly cell was compared with those of the general assembly
cell (GEN cell) and NBB assembly cell. Three workers were sequenced along the assembly line to
perform assembly tasks in a cell. The term “cell” is used because a small number of workers are in
charge of an assembly line. In the GEN cell, each worker performed four pre-assigned assembly tasks.
Ten different simulation scenarios were created with assembly task times randomly generated between
3 and 5 minutes. The dataset is given in Table 1. It is assumed that the work sequence of each task was
predefined by the precedence constraints of assembly tasks. For stochastic assembly times, a Gamma
distribution was assumed because the Gamma distribution is widely used in assembly line analysis
and easy to define assembly time variabilities. Simulation experiments were carried out 10 times for
each scenario in order to have meaningful statistics. Each simulation run covered 30 days. To collect
steady-state data, this study discarded statistics for the first 10 days (4800 min), and data for the next
20 days (1 month, 9600 min) were collected for analysis.
Table 1. Assembly operation data for 10 different scenarios.
Task Scenarios
Number 1 2 3 4 5 6 7 8 9 10
1 4.0 4.8 3.8 3.3 4.4 4.5 3.2 4.7 5.0 4.6
2 3.3 3.2 4.0 5.0 4.0 3.6 4.9 4.4 4.2 4.8
3 4.0 3.1 5.0 3.2 3.7 3.3 4.1 4.1 3.3 3.3
4 3.8 4.8 5.0 3.6 4.0 4.1 3.7 4.5 3.7 5.0
5 3.9 4.0 4.4 3.2 3.3 4.3 3.0 3.8 4.3 4.9
6 4.5 4.7 3.1 4.1 3.6 4.3 3.0 3.1 4.4 4.0
7 5.0 3.5 3.3 4.5 5.0 3.2 4.1 3.9 4.8 3.8
8 4.5 4.9 4.9 3.6 4.3 4.0 3.1 3.1 3.6 3.4
9 3.2 3.2 3.4 4.9 3.6 3.9 3.5 3.6 4.9 4.9
10 3.8 3.3 4.4 3.1 4.3 4.9 3.2 3.2 4.4 4.2
11 4.1 4.9 4.2 4.8 5.0 4.5 4.6 3.1 4.4 3.0
12 3.8 4.6 4.2 5.0 3.7 3.8 4.7 4.6 4.8 3.2 1
1 The values in the cells indicate the expected assembly time in minutes for each task.
10
Appl. Sci. 2020, 10, 6845
5.1. Experimental Analysis with Deterministic Task Times (No Uncertainty)

When the assembly task time was deterministic, the performance of the GEN cell could be obtained
analytically. To understand the performance of the different assembly strategies clearly, this study
takes scenario 1 as an example, whose total assembly time was T = 47.9 min.
GEN cell: In GEN cell, the task times of an assembly cycle for workers 1, 2, and 3 were 15.1, 17.9,
and 14.9 min, respectively. Hence, the bottleneck resource in the GEN cell was worker 2. The system
efficiency of the GEN cell was 0.892 (= 47.9/(3 × 17.9)) and the production rate of the three-worker
GEN cell was 536/month (= 9600 min/17.9 min).
BB-based cell (NBB cell, ABB cell): From Equation (3), the hand-off time for each assembly cycle
was h = E(t) = 3.99. Then, the line efficiency was E0 = T/(T + h) = 47.9/(47.9 + 3.99) = 0.923. Note that
with deterministic task times, blocking is not likely to happen. From Equation (4), the production rate
of the three-worker BB-based cell was 0.0578 units/min, which meant 555 units/month.
It should be noted that in the deterministic task time environment, the efficiency losses in the
GEN cell are due to line imbalance while the efficiency losses in the NBB and ABB cells are caused
by hand-offs. Figure 6 compares the performance of different assembly methods. When workers
are utilized 100%, the production capacity will be 601 units/month (= 9600 × 3/47.9). It is seen that
the balancing losses existed in the GEN cell while hand-off losses were present in the NBB cell and
ABB cell.
Maximum capacity: 601

production rate/month
555
536
Hand-off
loss
Balancing
loss
GEN NBB/ABB
cell cell
Figure 6. Performance of assembly methods under deterministic task time.
5.2. Experimental Analysis with Stochastic Task Times

As discussed previously, in a real-world manufacturing environment, it is common for the effective
assembly times to be stochastic. Reference [37] classified variability levels in manufacturing systems
into low, moderate, and high, on the basis of the magnitude of the CV values: low variation for CVs less
than 0.75, moderate variation for CVs between 0.75 and 1.33, and HV high variation for CVs greater
than 1.33. In this study, the performance of different assembly systems was examined with four CV
values: 0.0, 0.5, 1.0, and 1.5. It was assumed in the ABB assembly cell that the assembly tasks assigned
to the workers were the same as in Figure 5 and that no intermediate buffers existed. (The effect of
in-between buffers will be discussed in the next subsection).
Figure 7 compares the performance of the different assembly systems. Assembly jobs were
transferred to downstream operators when all the workers were ready to start new assembly jobs in
the GEN cell or when the immediate downstream worker was ready to take a new job in the NBB cell
and ABB cell. When there was no variability (i.e., deterministic task time with CV = 0), not much
performance difference could be seen among the three cell control methods. As the variabilities of the
task times increased, the performance became worse in every assembly system. The higher variabilities
of task times aggravated workload imbalance in the GEN cell because the line balancing efficiency
changed due to stochastic task times different from fixed standard time assumed in the line design stage.
The variability of task times also affected the performance of the NBB cell and the ABB cell because
the hand-off and blocking delays were increased with higher task time variabilities (see Equation (3)).
11
Appl. Sci. 2020, 10, 6845
The figure shows that the effect of assembly time variabilities on the performance was less in the NBB
cell and the ABB cell compared to the GEN cell. Among NBB and ABB cells, the ABB cell provided
consistently better performance than the NBB cell under a stochastic task time environment.
GEN Cell NBB Cell ABB Cell

600
production rate/month 550
500
450
400
350
300
0.0 0.5 1.0 1.5
CV (coefficient of variation) of task times
Figure 7. Performance of assembly systems with various assembly time variabilities.
The better performance of the ABB cell over NBB cell was due to reduced hand-off times and
blocking times expected in the ABB cell. Figure 8 shows the hand-off times and blocking times in the
NBB and ABB cells. It is seen that hand-off times and blocking times were consistently less in the ABB
cell than in the NBB cell, and the difference became larger with higher CV values.
1%%&HOO $%%&HOO 1%%&HOO $%%&HOO

hand-off time (min)
hand-off time (min)

&9 RIDVVHPEO\WLPH &9 RIDVVHPEO\WLPH
(a) hand-off time (b) blocking time

Figure 8. Comparison of the performance of naïve BB-based assembly (NBB) cell and ABB cell in terms
of (a) hand-off time and (b) blocking time.
5.3. Performance Comparison of Assembly Systems with Intermediate Buffers

In the previous experiments, this study assumed that there was no in-between buffer. Next, this
study examined a new strategy where there were in-between buffers at the end of each work area for
workers for the GEN cell and ABB cell. Note that a worker was not allowed to perform assembly tasks
out of the assigned set in the ABB cell. When a worker finished the last task of the assigned tasks and
found no downstream worker available, he put the half-finished assembly at the in-between buffer
(if there was an empty space left) located at the end of his work area and went back upstream to take a
new job from his predecessor (or went back to the beginning of the assembly cell for a new product if
he was the first worker of the line). If there was no empty space in the buffer, the worker waited until
an empty space was available. It was expected that the in-between buffers mitigated some adverse
effects of system variabilities. However, if the buffers were too large, it would increase the production
lead time and WIP levels. Therefore, it was necessary to maintain an appropriate buffer size.
12
Appl. Sci. 2020, 10, 6845
Figure 9 compares the production rate and lead time over different buffer sizes. Here, the CVs
of task times were assumed to be 1.0. It is seen that as the buffer size increased, the production rate
increased in all assembly systems except for the NBB cell where a buffer was inherently not allowed.
It is seen that the ABB cell performed better than the other systems in terms of the production rate
when the buffer size was zero or one. It is seen that only a small buffer size (e.g., one) led to a large
increase in throughput rate, with little increase in production lead time in the ABB cell. When the
buffer size was larger than one, the GEN cell provided a good performance in terms of production rate
with the cost of long lead time. Observing Figure 9b, it is clear that the lead time increased sharply in
the GEN cell while the lead time remained almost unchanged in the ABB cell, even with a larger buffer
size. The experimental results indicate that the proposed ABB cell is a promising assembly system
when the lead time, as well as production rate, is an important performance measure and there is very
limited space for intermediate buffers.
GEN Cell NBB Cell ABB Cell GEN Cell NBB Cell ABB Cell
550 120
lead time (minutes)

production rate/month
100
500
80
450
60
400
40
350 20
300 0
0 1 2 3 0 1 2 3
Buffer size Buffer size
(a) production rate (b) lead time

Figure 9. Performance of different assembly systems over varying buffer sizes in terms of (a) production
rate and (b) production lead time.
5.4. Performance Analysis for the System with Different Workers’ Speeds
Most ALB solutions assume that every worker has an identical work speed. However, in real-world
manufacturing systems, experienced and unexperienced workers with different work speeds are
deployed together in an assembly line. In this case, the production capacity is often restricted by
inexperienced and slow workers, resulting in lower productivity than expected. In the BB-based
assembly cells, the workers are not assigned a pre-defined set of tasks but perform assembly tasks as
much as their ability. The dynamic task assignment is expected to lead to a minimum productivity
loss due to different work speeds. This section examines the effect of the different work speeds on the
performance of each assembly system. Figure 10 shows the performance of the GEN, NBB, and ABB
cells. It was assumed that the assembly tasks are performed by three workers with different work
speeds with (a) no buffer and (b) buffer size of one, with an assembly time CV of 1.0. In the figure, speed
notation (0.9, 1.0, 1.1) indicates that the work speeds were 0.9, 1.0, and 1.1 for the first, second, and third
workers, respectively. Since the buffer was not allowed in the NBB cell, there was no experimental
result for the NBB cell in Figure 10b. In every case, the NBB and ABB cells performed better than GEN
cells whether the buffer was allowed or not. It is seen that the ABB cell worked better than the NBB cell
even when the buffer was not allowed. When buffers were allowed, the ABB cell consistently worked
better than the GEN cell. The better performance was realized in the ABB cell no matter where the
slower or faster workers were located. The figure shows that the performance of the GEN cell worked
well when all the workers’ speeds were the same. On the other hand, it is seen that the ABB model
worked well when the workers had different work speeds.
13
Appl. Sci. 2020, 10, 6845

SURGXFWLRQUDWHPRQWK
SURGXFWLRQUDWHPRQWK

*(1&HOO 1%%&HOO $%%&HOO *(1&HOO $%%&HOO
(a) zero-buffer case (b) one-buffer case

Figure 10. Performance of assembly systems over different work speeds with (a) buffer size of zero and
(b) buffer size of one. The notation at the x-axis, (s1, s2, s3) indicates that the work speeds are s1, s2,
and s3 for the first, second, and third workers, respectively.
Figure 10a shows that the ABB cell worked better when faster workers were located at the
beginning of the line in the zero-buffer case. This result is contradictory to the claim in [16] that
BB-based control works best when workers are sequenced from slowest to fastest along the line.
The contradictory result can be explained as follows: since the ABB cell restricts the work contents for
a worker to perform, neighboring workers are less likely to perform consecutive assembly tasks that
may result in blocking. Hence, blocking delays can be reduced in the ABB cell. The hand-off times
depend on the task time of the upstream workers. When the work speed of an upstream worker is fast,
then actual assembly time is reduced, resulting in lower hand-off time. Figure 11 shows hand-off and
blocking times per product in the ABB cell with (a) zero-buffer and (b) buffer size of one. It is seen that
in the ABB cell, hand-off times were much more than blocking times. It is also seen that the hand-off
times could be reduced when faster workers were located at the beginning.

HIILFLHQF\ORVV
HIILFLHQF\ORVV

KDQGRIIWLPH EORFNLQJWLPH KDQGRIIWLPH EORFNLQJWLPH
(a) zero-buffer case (b) one-buffer case

Figure 11. Efficiency losses caused by hand-offs and worker blocking with (a) buffer size of zero and
(b) buffer size of one.
6. Discussion and Conclusions

Once an assembly line is configured on the basis of the assembly line balancing (ALB) decisions,
the system is operated in a static way where workers perform the pre-determined set of assembly tasks
repetitively. The static decision often leads to low performance in dynamic and uncertain assembly
environments. The idea of bucket brigades (BBs) has been applied to assembly systems to respond
to system uncertainties. However, the naïve BB-based assembly cell (NBB cell) suffers from some
efficiency losses including hand-offs and blocking in most assembly flow lines. This paper presented
an improved version (ABB cell) of the NBB cell. The performance of the new assembly cell is compared
14
Appl. Sci. 2020, 10, 6845
with the other assembly systems under various assembly environments. From the numerical and
experimental analysis, we have produced some interesting findings of the new BB-based assembly
system (ABB cell) as follows:
• The key finding of our study is that the ABB cell is robust in that it performs especially well under
dynamic and uncertain assembly environments. As the variabilities increase, both the NBB and
ABB cells provide worse performance. However, the performance of the ABB cells is less affected
by the high variabilities than that of the NBB cells, with fewer hand-off and blocking delays.
• The BB-based cell works well when n/m is large, where n is the number of assembly tasks and m
is the number of workers in an assembly cell. As given in Equation (1), the hand-off losses are
proportional to the n/m ratio. Parallel to this finding is that assigning a small number of workers
to a BB-based cell is preferable.
• Even a minimal buffer size, e.g., buffer size of one, in the ABB cell leads to a big performance
increase in terms of production rate. Moreover, the buffer size has little effect on the production
lead time in the ABB cells, unlike traditional assembly cells. Hence, the ABB cell can be a good
assembly system when the lead time as well as the production rate is a key performance measure.
• Existing research works about the BB-based system assume that the workers are sequenced from
slowest to fastest to have a maximum capacity. Our results are contradictory to this assumption.
In our BB-based system, the workers may be sequenced in the other way, i.e., from fastest to
slowest. It is found that the reverse sequence can reduce hand-off delays and improve performance
in some cases (see Figure 10). This finding is important because we can have the flexibility to
deploy the workers along the line in the new assembly system.
Although the new assembly system has some advantages over traditional assembly cells and naïve
BB-based cells, there are some considerations that should be taken into for the new assembly system to
be successfully implemented. One of the important characteristics in the new assembly system involves
a distributed and autonomous decision making instead of centralized decision making. Each worker
in the production line performs the tasks possibly different from cycle to cycle by following a control
logic without any predefined tasks assigned in advance. Hence, workers should be multi-functional
and willing to work in collaboration with their colleagues. An appropriate training program for
workers should be provided. Moreover, as argued in [38,39], both the benefits and negative aspects of
the dynamic work assignments should be taken into consideration together in selecting appropriate
assignment strategies. Workers in the BB-based system perform their jobs commensurate with their
capabilities. Skilled workers work more while un-skilled workers less. To exert a full capacity of each
worker, an appropriate incentive system may be needed.
This study considers assembly systems producing a single product type. It is believed that our
model can be extended for multi-product assembly systems with minimum adjustment of the control
logic. Walk-back time is ignored in this paper. Hence, the proposed system may be suitable for
assembly systems where light and small components are assembled in a short flow line. For the
assembly systems dealing with big and heavy components, some new ideas may be needed, which is a
possible future work.
Author Contributions: Methodology, Formal Analysis, Investigation, Data Curation, Writing-Original Draft
Preparation, Writing-Review & Editing, P.-H.K. The author have read and agreed to the published version of
the manuscrip.
Funding: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the
Korea government (MSIT) (no. 2019R1F1A1057585).
Conflicts of Interest: The author declares no conflict of interest.
15
Appl. Sci. 2020, 10, 6845
References
1. Hu, S.J.; Ko, J.; Weyand, J.; El Maraghy, H.A.; Lien, T.K.; Koren, Y.; Bley, H.; Chryssolouris, G.; Nasr, N.;
Shpitalni, M. Assembly system design and operations for product variety. CIRP Ann. 2011, 60, 715–733.
[CrossRef]
2. Battaia, O.; Dolgui, A. A taxonomy of line balancing problems and their solution approaches. Int. J. Prod. Econ.
2013, 142, 259–277. [CrossRef]
3. Aguilar, H.; García-Villoria, A.; Pastor, R. A survey of the parallel assembly lines balancing problem.
Comput. Oper. Res. 2020, 124, 105061. [CrossRef]
4. Eghtesadifard, M.; Khalifehb, M.; Khorram, M. A systematic review of research themes and hot topics in
assembly line balancing through the web of science within 1990–2017. Comput. Ind. Eng. 2020, 139, 106182.
[CrossRef]
5. Pereira, J. The robust (minmax regret) assembly line worker assignment and balancing problem.
Comput. Oper. Res. 2018, 93, 27–40. [CrossRef]
6. Johnson, D.J. Converting assembly lines to assembly cells at sheet metal products: Insights on performance
improvements. Int. J. Prod. Res. 2005, 43, 1483–1509. [CrossRef]
7. Ruiz-Torres, A.J.; Mahmood, F. Impact of worker and shop flexibility on assembly cells. Int. J. Prod. Res.
2007, 45, 1369–1388. [CrossRef]
8. Yu, Y.; Tang, J.; Sun, W.; Yin, Y.; Kaku, I. Reducing worker(s) by converting assembly line into a pure cell
system. Int. J. Prod. Econ. 2013, 145, 799–806. [CrossRef]
9. Yu, Y.; Wang, J.; Ma, K.; Sun, W. Seru system balancing: Definition, formulation, and exact solution.
Comput. Ind. Eng. 2018, 122, 318–325. [CrossRef]
10. Bartholdi, J.J.; Eisenstein, D.D.; Lim, Y.F. Bucket brigades on in-tree assembly networks. Eur. J. Oper. Res.
2006, 168, 870–879. [CrossRef]
11. Gurumoorthy, K.S.; Banerjee, A.; Paul, A. Dynamics of 2-worker bucket brigade assembly line with blocking
and instantaneous walk-back. Oper. Res. Lett. 2009, 37, 159–162. [CrossRef]
12. Koo, P.H. Self-balancing assembly line using bucket brigades. In Proceedings of the 37th International
Conference on Computers and Industrial Engineering, Alexandria, Egypt, 20–23 October 2000; pp. 1633–1641.
13. Lim, Y.F. Cellular bucket brigades. Oper. Res. 2011, 59, 1539–1545. [CrossRef]
14. Lim, Y.F. Performance of cellular bucket brigades with hand-off times. Prod. Oper. Manag. 2017, 26, 1915–1923.
[CrossRef]
15. Anderson, C.; Boomsma, J.J.; Bartholdi, J.J. Task partitioning in insect societies: Bucket brigades.
Insectes Sociaux 2002, 49, 1–10. [CrossRef]
16. Bartholdi, J.J.; Eisenstein, D.D. A production line that balances itself. Oper. Res. 1996, 44, 21–34. [CrossRef]
17. Koo, P.H. The use of bucket brigades in zone order picking systems. OR Spectr. 2009, 31, 759–774. [CrossRef]
18. Sahin, M.; Kellegoz, T. A new mixed-integer linear programming formulation and particle swarm optimization
based hybrid heuristic for the problem of resource investment and balancing of the assembly line with
multi-manned workstations. Comput. Ind. Eng. 2019, 133, 107–120. [CrossRef]
19. Dimitriadis, S.G. Assembly line balancing and group working: A heuristic procedure for workers’ groups
operating on the same product and workstation. Comput. Oper. Res. 2006, 33, 2757–2774. [CrossRef]
20. Süer, G.A.; Sinaki, R.Y.; Sadeghi, A. A Hierarchical hybrid heuristic-optimization approach for multi-product
assembly line design problem. Procedia Manuf. 2019, 39, 1067–1075. [CrossRef]
21. Rabbani, M.; Mahmood, S.; Manavizadeh, N. Mixed model U-line balancing type-1 problem: A new approach.
J. Manuf. Syst. 2012, 31, 131–138. [CrossRef]
22. Li, Z.; Kucukkoc, I.; Zhang, Z. Branch, bound and remember algorithm for U-shaped assembly line balancing
problem. Comput. Ind. Eng. 2018, 124, 24–35. [CrossRef]
23. Chiang, W.-C.; Urban, T.L. The stochastic U-line balancing problem: A heuristic procedure. Eur. J. Oper. Res.
2006, 175, 1767–1781. [CrossRef]
24. Lopes, T.C.; Michelsa, A.S.; Sikora, C.G.S.; Molina, R.G.; Magatãoa, L. Balancing and cyclically sequencing
synchronous, asynchronous, and hybrid unpaced assembly lines. Int. J. Prod. Econ. 2018, 203, 216–224.
[CrossRef]
25. Li, Z.; Janardhanan, M.N.; Tang, Q.; Ponnambalam, S.G. Model and metaheuristics for robotic two-sided
assembly line balancing problems with setup times. Swarm Evol. Comput. 2019, 50, 100567. [CrossRef]
16
Appl. Sci. 2020, 10, 6845
26. Bartholdi, J.J.; Eisenstein, D.D.; Foley, R.D. Performance of bucket brigades when work is stochastic. Oper. Res.
2001, 49, 710–719. [CrossRef]
27. Bartholdi, J.J.; Eisenstein, D.D. Using bucket brigades to migrate from craft manufacturing to assembly lines.
Manuf. Serv. Oper. Manag. 2005, 7, 121–129. [CrossRef]
28. Munoz, L.F.; Villalobos, R. Work allocation strategies for serial assembly lines. Int. J. Prod. Res. 2002, 40,
1835–1852. [CrossRef]
29. Armbruster, D.; Ge, E.S. Bucket brigades revisited: Are they always effective? Eur. J. Oper. Res. 2006, 172,
213–229. [CrossRef]
30. Hirotani, D.; Myreshka, K.; Morikawa, K.; Takahashi, K. Analysis and design of self-balancing production
line. Comput. Ind. Eng. 2006, 50, 488–502. [CrossRef]
31. Carlo, F.D.; Borgia, O.; Tucci, M. Bucket brigades to increase productivity in a luxury assembly line. Int. J.
Eng. Bus. Manag. 2013, 5, 28. [CrossRef]
32. Hong, S.; Kim, Y. The effects of loosely coupled hand-off operations on bucket brigade order picking systems.
Ind. Eng. Manag. Syst. 2018, 17, 745–756. [CrossRef]
33. Hong, S.; Johnson, A.L.; Peters, B.A. Quantifying picker blocking in a bucket brigade order picking system.
Int. J. Prod. Econ. 2015, 170 Pt C, 862–873. [CrossRef]
34. Pratama, A.T.; Takahashi, K.; Morikawa, K.; Nagasawa, K.; Hirotani, D. Cellular bucket brigades with worker
collaboration on U-lines with discrete workstations. Ind. Eng. Manag. Syst. 2018, 17, 531–549. [CrossRef]
35. Segal, M.; Whitt, W. A queueing network analyzer for manufacturing. Teletraffic Sci. New Cost-Eff. Syst.
Netw. Serv. 1989, ITC-12, 1146–1152.
36. Hong, S. The effects of picker-oriented operational factors on hand-off delay in a bucket brigade order picking
system. OR Spectr. 2018, 40, 781–808. [CrossRef]
37. Hopp, W.J.; Spearman, M.L. Factory Physics, 2nd ed.; Irwin/McGraw-Hill: Boston, MA, USA, 2000.
38. Schultz, K.L.; McClain, J.O.; Thomas, L.J. Overcoming the dark side of work flexibility. J. Oper. Manag. 2003,
21, 81–92. [CrossRef]
39. Cesani, V.I.; Steudel, H.J. A study of labor assignment flexibility in cellular manufacturing systems.
Comput. Ind. Eng. 2005, 48, 571–5915. [CrossRef]
© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
17
applied
sciences
Article
Design Model of Flow Lines to Include Switch-Off
Policies Reducing Energy Consumption
Paolo Renna * and Sergio Materi
School of Engineering, University of Basilicata, 85100 Potenza, Italy; sergio.materi@unibas.it
* Correspondence: paolo.renna@unibas.it; Tel.: +39-0971-205143

Received: 27 January 2020; Accepted: 18 February 2020; Published: 21 February 2020
Abstract: One of the most promising approaches to reduce the amount of energy consumed in
manufacturing systems is the switch off policy. This policy reduces the energy consumed when the
machines are in the idle state. The main weakness of this policy is the reduction in the production
rate of the manufacturing systems. The works proposed in the literature do not consider the design
of the production lines for the introduction of switch off policies. This work proposes a design model
for production lines that include a targeted imbalance among the workstations to cause designed idle
time. The switch-off policy introduced in such designed production lines allows for a reduction in
the energy consumed with any production rate loss. Simulation tests are conducted to verify the
benefits of switch off policies in production lines designed for its. The simulation results show that
the proposed line design allows for a reduction in energy consumption, with a defined loss in the
throughput. The application of switch-off policies in the proposed flow line leads to a significant
reduction in the energy used in unproductive states controlling the production loss.
Keywords: flow-line design; switch-off policy; energy consumption
1. Introduction and Motivations

Nowadays, the industrial manufacturers study energy-efficient models because of the costs and
environmental impact of energy consumption [1,2].
The manufacturing activities are characterized by a relevant demand from around 40% with
relevant growth expectations between 2018 and 2050 across all cases. Increases in industrial energy use
from increasing shipments are partially offset by efficiency gains [3].
Studies by the European Association of the Machine Tool Industries [4,5] on the discrete part
manufacturing discussed the importance and environmental impacts of electrical energy consumption.
The main models proposed in the literature, for the design of manufacturing systems, focused
primarily on the performance in regards to productivity, quality, and work in process, etc. Recently,
the models proposed include the energy efficiency issue, but Gahm et al. [6] emphasized how the
scheduling models that include energy-saving may reduce the other goals of the manufacturing
systems. Then, it is more important to propose a model that reduces energy consumption without
reducing the productivity performance of manufacturing companies.
The works proposed regarding machining energy consumption [7,8] identified three main energy
shares: start-up operations (computers and fans, coolant pumps, etc.), runtime operations (tool change,
Jog axis, etc.) and material removal operations (machining). The first and second parts are constant and
independent of the operation, while the third part is variable and depends on the machining operation.
Gutoski et al. [8–10] underlined how the constant independent of the operation is the major part
of the energy consumed by the machining machines and this trend has grown in past years.
The reduction in the energy consumption of these parts can be obtained without studying the
effect on the particular machining operation as: roughness, cutting paths, cutting time, etc. Therefore,

Appl. Sci. 2020, 10, 1475
it is more simply to extend a model that reduces the constant energy to several machining systems and
other production systems.
A promising approach proposed in the literature to reduce the constant energy consumed by
the machining machines is the switch-off policy [11]. Reducing the idle time is a good approach to
realize energy-efficient production [12]. One example is the suppliers for small-parts in aircrafts, where
the idle periods accounted for 16% of the total production time, and about 13% of the total energy
consumption could be saved if the idle machines were switched off [13]. Another study by Weinert and
Mose [14] highlighted how almost 50% to 60% of energy consumption could be reduced by turning the
standby machines into energy-saving states.
The switch-off policies switch off when the machine is in the idle state, which reduces the constant
energy consumed during this state. When there are a determined number of parts waiting for the
operation, the machine switches on, subject to a warm-up period to be operative. The decision of the
switch on is crucial to maintain the productivity level of the production line. Some approaches have
been proposed, by Frigerio and Matta [15,16], with a significant reduction in energy consumption in a
flow lines context. The switch-off models proposed works on a flow line designed following classical
objectives that solve the simple assembly line balancing problem [17,18]. An extensive survey of the
flow line production system balancing was discussed in References [19,20].
Mathematical models to support the decision towards the switch off/on, which works off-line,
have been proposed by Mashaei and Lennartson [21] and Jia et al. [22]. These models did not handle the
uncertainty of the production system and the optimization can lead to high computational complexity
when the number of machines is higher; this reduces the potential application in real industrial
manufacturing systems.
Su et al. [23] and Frigerio and Matta [16] studied upstream, downstream, and mixed policies in
a single machine, and the extension to a production line, assuming stochastic arrivals, and constant
warm-up under to stock control. These models did not consider how the control of the machines may
reduce the production performance of the manufacturing system.
Renna [24] studied a dynamic and adaptive control strategy to switch-off the machines on a
production line under the pull control policy. The policy proposed uses the information of the buffers
and the level of customer satisfaction. The results also underlined the potential application of the
switch-off policies in production lines under pull control.
Duque et al. [25] studied one machine with one buffer with a fuzzy controller that includes the
information about the buffer, the machine state and the production rate required, considering the
warm-up energy. The controller was tested through simulation experiments and it was observed that a
large amount of energy could be saved without affecting the throughput significantly.
Wang et al. [26,27] proposed a method based on a fuzzy controller that includes the information
of the upstream, downstream buffer and the status of the machine. They proposed a set of fuzzy
rules to take the decision to the switch off/on policy. The simulation experiments show that the
proposed approach can be a simple, practical way towards the energy saving operation with accepted
throughput loss.
Marzano et al. [28] proposed a model that controls the machine on-line acquiring data and
estimating the risk of the control policy actions. The model has several limitations because it is tested
for a specific distribution for the estimation of the parameters and the tests are conducted considering
a production line composed of two machines.
The discussion of the past works highlights the following issues:
- The switch-off policies are mainly proposed for a single machine and some works considered a
flow line with the introduction of buffers;
- Few works evaluate the reduction of the performance level due to the introduction of the switch-off
policy, but only energy consumption;
- Works were proposed to consider the possibility of a switch-off policy and were considered from
the design step of the production line.
20
Appl. Sci. 2020, 10, 1475
In response to the limits in the literature, this paper proposed a design model that includes the
switch-off policy by first asking:
RQ1: what is the impact of the design model proposed on the performance of the production line
in terms of energy saving maximizing the production rate?
Some previous works show how the switch-off policy causes production loss, then our second
research question asks:
RQ2: can the constraint of a limited reduction loss improve significantly the energy saving of the
production line obtaining an adequate trade off?
The research proposed concerns about the development of a design model of the flow lines that
includes the possibility of the switch-off already in this design step. Then, the switch-off policy is
introduced in the flow line designed by the proposed model. The proposed design and switch-off
policy is compared to a flow line designed with the classical objectives with the same switch-off policy.
Simulation experiments will be used to answers our two research questions. The simulations
were conducted by considering different levels of production rates in the design model and different
levels of the buffers (upstream and downstream) that control the switch-off policy.
The paper is organized as follows. Section 2 described the design model of the flow line, while
Section 3 introduces the switch-off policy used. Section 4 introduces the reference context investigated
with simulation scenarios, while Section 5 discussed the numerical results. Section 6 provides the
conclusions and a future research path.
2. Flow Line Design Model

The problem deals with the designing of a flow line composed by M stations that manufactures
one product type. The product consists of N operations to process; these operations should be assigned
to the M workstations, following the precedence constrains. The variables of the model are the
assignments of the operations to the workstations of the flow line. The operation assignment binary
variable xij is defined as follows (Equation (1)):
⎧
⎪
⎪ i f the i − th operation is assigned to the j − th station
⎨1
xij = ⎪
⎪ (1)
⎩0 otherwise
The precedence constrains binary variable ensures that the i-th operation must be completed
before the k-th operation, and it is computed as (Equation (2)):
⎧
⎪
⎪ i f the i − th operation preceds the k − th station
⎨1
vik = ⎪
⎪ (2)
⎩0 otherwise
The processing time Tj of the j-th station is the sum of the processing time of the operations
assigned to j-th station, as follows (Equation (3)):
N
Tj = ti xij (3)
i
The cycle time C of the production line is equal to the maximum of the stations’ processing times,
as shown in Equation (4)
C = MAX j T j (4)
The j-th station idle time is defined as follows (Equation (5)):
Td,j = C − T j (5)
21
Appl. Sci. 2020, 10, 1475
The idle time for each cycle is given by Equation (6):

M
TTd = Td, j (6)
j=1
The difference of processing time between nearby stations is called distance dj and it is defined as
follows (Equation (7)):
d j = T j+1 − T j (7)
A positive value of the distance dj forms a couple of workstations when the first has a higher
velocity than the second workstation. Then, the first workstation can fill the downstream buffer and
goes into the off state, reducing energy consumption. Therefore, the flow line consists of a couple of
workstations to facilitate the off state of the first workstation of the couple. Figure 1 shows the concept
of the distance of a couple of workstations. In the following figure, the term WSj has been used to
identify the j-th workstation; the distance between two stations has been obtained as the difference of
the processing time of the second workstation (with higher working time) and the first workstation
of the couple. In Figure 1, the processing time is linked to the stations with, respectively, lower and
higher productivity per couple, as represented in blue and green.
Processing Couple of
time workstations
d1 dn
WS1 WS2 WSn WSn+1
Figure 1. Distance between stations.
To achieve an unbalanced flow line, the objective function is achieved by maximizing the sum of
the distance between stations (Equation (8)):

MAX dj with j = 1, 3, 5, . . . (8)
j
This is subject to the following constrains (Equations (9)–(11))

M
xij = 1 with i = 1, . . . , N (9)
j=1
M M
j xij ≤ j xk j ∀ vik = 1 (10)
j=1 j=1
N
ti xij ≤ C max with j = 1, . . . , M (11)
i=1
Equation (9) ensures that each operation is only assigned to one machine. Equation (10) ensures
that the constraints on the precedence of operations are respected. Finally, Equation (11) ensures
that the station processing times are below the maximum fixed cycle time. By setting the cycle time,
22
Appl. Sci. 2020, 10, 1475
maximizing the objective function achieves an unbalanced line that respects the targeted productivity.
Equation (8) results in an unbalanced flow line with high idle times between sequential stations. In
order to achieve more downtime, and therefore higher energy saving, this research proposes a different
objective function from past literature, where the focus instead is on obtaining the minimum cycle
time [17], the minimum number of stations, and the minimum idle time [18].
Figure 2 shows the framework used for the design of unbalanced flow lines. First, using the
mathematical model, the flow line with maximum productivity and minimum cycle time has been
achieved. The obtained maximum productivity flow line and the minimum cycle time has been
calculated as the bottleneck station processing time. The minimum cycle time has been used as a
parameter to design unbalanced flow lines using the maximization of the sum of distances as the
objective function according Equation (8). Then, increasing and fixing the cycle time, the maximization
of the sum of the distance results in several unbalanced flow lines (in this paper, three unbalanced flow
lines have been obtained). Then, the simulation will be studied if the increment of the cycle time can
lead to important energy reduction.
Minimization of cycle time Precedence constrains Operation processing times
Mathematical Optimization
Line design 1: Max productivity
Minimum
Cycle time
Increment of Increment of Increment of

Maximization Operation
Precedence Minimum minimum of minimum of
of distance processing
constrains Cycle Time of Cycle Time of Cycle Time of
times
'1 '2 '3
Mathematical Optimization
Line design 2 Line design 3 Line design 4
Captions
Parameters Output of mathematical optimization
Objective function
Figure 2. Framework for flow line designs.
3. Switch off Policy
As described in the literature [23], switching off policies based on buffer level information can lead
to significant energy savings without reducing productivity. The proposed policies are upstream (UP),
downstream (DP), and upstream and downstream (UDP). In the upstream policy, the machine switches
off when the upstream buffer is empty and switches on when the upstream buffer level is NU on . The
level of downstream buffer controls the state in the downstream policy. The machine switches off when
the threshold ND off is reached and turns on when the number of pieces in the buffer is equal to ND on .
According to Reference [15], the state of machine sj is defined as follows (Equation (12)):
23
Appl. Sci. 2020, 10, 1475
⎧
⎪
⎪ 1 if out − o f − service
⎪
⎪
⎪
⎪
⎨ 2 if idle
sj = ⎪
⎪ (12)
⎪
⎪ 3 if in start − up
⎪
⎪
⎩ 4 if working
The states 1, 2 and 3 are unproductive, i.e., no pieces are being processed when a machine is in one
of these states. The states 1 and 2 are called inactive states. According to Reference [23], the upstream
and downstream combines the UP and DP policies as follows (Equation (13)):
⎧
⎪
⎪ i f s j = 2 AND (n j = 0 OR n j+1 ≥ NoDf f )
⎨ Switch − O f f
⎪
⎪ (13)
⎩ Switch − On i f s j = 1 AND (n j = Non
U AND n
j+1 ≤ Non )
D
Figure 3 summarizes the states of the generic machine and the transition from one state to another.
Switch-on Warm Part loaded

up
Out of
Idle Working
service
Power
off Switch-off Departure
Figure 3. Machine states.
4. Reference Context and Simulation Scenarios

Using the mathematical model described in Section 2, four production lines with 10 stations and
20 tasks to complete have been designed. The flow line only produces one product type.
The operation processing times and the precedence constraints are described in Table 1 and
have been obtained from the simple assembly line balancing problem dataset (SALBP) according to
Reference [29].
Table 1. Operation processing time [s].
Operations Processing Time (s) Precedence Operation Processing Time (s) Precedence
1 142 - 11 97 7
2 34 - 12 132 8
3 140 - 13 107 10, 11
4 214 - 14 132 12
5 121 - 15 69 12
6 279 1 16 169 13
7 50 2 17 73 13
8 282 4 18 231 13
9 129 5 19 120 15
10 175 6 20 186 14
The basic assumptions for the line design are the following:
24
Appl. Sci. 2020, 10, 1475
The stations can perform every possible operation assigned by the design model;
The precedence constraints are fixed;
No machine failure has been considered;
The operations processing times are deterministic and are the same for each station.
Figure 4 shows the precedence graph. In the precedence graph, each task is represented as a
node, and each direct precedence constraint is illustrated as an arrow that links node i and k if the i-th
operation must precede the k-th operation. For example, in Figure 4, operation 6 must be executed
before operation 10.

Figure 4. Precedence graph.
Four flow lines are obtained in order to get respectively:
1. Minimization of total idle time for each cycle with minimum Cycle Time (MinTTd );
2. Maximization of distance between stations with a 2.5% increment of minimum Cycle Time (2.5%
loss production rate, MaxD_2.5%);
3. Maximization of distance between stations with a 5% increment of minimum Cycle Time (5% loss
production rate, MaxD_5%);
4. Maximization of distance between stations with a 10% increment of minimum Cycle Time (10%
loss production rate, MaxD_10%).
The solution of the mathematical model, considering the previous constrains, gives the following
flow line designs (Table 2); moreover, the station processing times and the station idle times have been
reported in Table 2.
The first production line is considered to be the benchmark of the other production lines, because
it gives a balanced production line with 10 stations and the minimum cycle time.
A discrete event simulation, implemented in Arena, has been used to evaluate the performances
of the four flow line designs and to analyze the application of switch off policies in unbalanced
production lines.
25
Appl. Sci. 2020, 10, 1475
Table 2. Stations processing and idle times [s].
MinTTd MaxD_2.5% MaxD_5% MaxD_10%

Station Station Station Station
Station Idle Station Idle Station Idle Station Idle
Station Processing Processing Processing Processing
Time [s] Time [s] Time [s] Time [s]
Time [s] Time [s] Time [s] Time [s]
1 298 6 248 63 248 70 214 119
2 282 22 311 0 313 5 282 51
3 272 32 282 29 282 36 263 70
4 274 30 298 13 298 20 333 0
5 279 25 274 37 279 39 279 54
6 287 17 306 5 315 3 329 4
7 304 0 279 32 236 82 269 64
8 293 11 304 7 304 14 333 0
9 304 0 276 35 289 29 276 57
10 289 15 304 7 318 0 304 29
TTd 158 228 298 448
Each model has been simulated by considering machines in the “always on” (AO) state and the
UPD switch off policy.
The basic assumptions of the AO model are:
Each station has a buffer;

The buffer capacity is fixed, and equal to K;
The buffer of the first station is always full, that is the raw material is always available;
The power absorbed in each state is equal for all machines.
In addition, for the models with switch off policies, the following conditions apply:
Each station is controlled by a switch off policy;

The control policy parameters (NU on , ND off , ND on ) are the same for stations from 2 to 9;
The first station has only DP policy;
The last station has only UP policy.
As described by Frigerio and Matta [15], the production line machines can be in the following
states:
– Working state: the machine is processing a piece and absorbs the power Pw ;
– Idle state: the machine is ready to work a part, and absorbs the power Pi ;
– Out-of-service (Inactive) state: the machine is not ready to process a part. In this state the machine
absorbs the minimum amount of power Poff ;
– Warmup state: the machine changes its state from Out-of-service in idle or working state,
consuming the power Pwu for the time to complete the warmup twu .
According to [23], the power required by a generic machine in each state is:
– Pw = 12 kW;
– Pi = 5.35 kW;
– Poff = 0.52 kW;
– Pwu = 6 kW for twu = 20 s.
To determine the best switch off control parameters, a full factorial design has been developed.
The factors considered are NU on , ND off , ND on , and three levels for each factor are evaluated as follows:
– NU on = [1, 2, 3];
– ND on = [4, 5, 6];
– ND off = [7, 8, 9];
26
Appl. Sci. 2020, 10, 1475
According to Reference [23], the levels of buffer to switch off or switch on machines respect the
following constraints (Equations (14)–(16)):
NoDf f ,j ≥ Non,
D
j (14)
NoDf f , j ≥ Non,
U
j+1 (15)
D U
Non, j ≥ Non, j+1 (16)
Figure 5 reports on the experiment results for model 1. The design with the lowest inactive time
has a value of 0, the design with the highest inactive time has a value of 1. These results are used to set
these parameters for the simulation tests.
Figure 5. Experiments results for line design MinTTd .
The results of the other experiments for models MaxD_2.5%, MaxD_5%, and MaxD_10% are
reported in Appendix A.
As shown in Figure 2, and in the figures (Figures A1–A3) in Appendix A, for all the four design
models, the set NU on = 3, ND on = 4, ND off = 9, gives the maximum time spent in the idle and inactive
state. Therefore, the simulation scenarios and the evaluation of the performances have been obtained
by considering the best control parameters. According to Reference [23], the discrete event simulation
length is 107 s, and the initial transient is 105 s.
Figure 6 shows the setting of simulation scenarios and performance evaluations. Using
mathematical optimization, the flow line design and the operations assignment to the stations
have been achieved. Using discrete event simulation, the best set of switch off-parameters have been
obtained. Finally, by employing this set, the performances of the four lines have been evaluated
and compared.
27
Appl. Sci. 2020, 10, 1475
Discrete event simulation

Switch-off
control
parameters
(NUon, NDoff,
NDon)
Best Switch off

Flow parameters set Performance
Flow line design
lines evaluation
Power
machines
required
(Pw, Pi,
Poff, Pwu)
Figure 6. Simulation process and performances evaluation.
5. Numerical Results
As shown in the following figure (Figure 7), an increment of the cycle idle time for each station
has been achieved by choosing the maximization of the distance between stations as the objective
function, instead of the minimization of idle time. The maximum increment of TTd has been obtained
in the case where the cycle time has been increased by 10%.

,QFUHPHQWRI&\FOH,GOH7LPH

MaxD_2.5% MaxD_ 5% MaxD_10%
Figure 7. Increment of idle time for each cycle.
28
Appl. Sci. 2020, 10, 1475
The discrete event simulation results, as compared to the model MinTTd , are shown in Figure 8.
The results in the figure only consider the effects of the unbalanced line on the idle time, so the
simulations have been made by considering machines with an always on control policy. It can be
noticed that the higher the idle time, the lower the throughput. Indeed, by increasing the distance,
stations are in the idle state for longer than in a balanced flow line. Increased cycle time allows for
mathematical optimization, which aims to maximize the distance between station process times and
to have greater freedom in assigning operations to the machines. The 10% increase in cycle time
(MaxD_10%) results in a flow line where the idle time has grown by 159%, but productivity has
decreased by only 9%.

Throughput
Idle Time

MaxD_2.5% MaxD_5% MaxD_10%
Idle time Throughput
Figure 8. Results with always on policy.
However, the increase in downtime leads to more energy consumption in a non-productive

state. In order to reduce energy consumption, switch-off policies should be in place. The UDP policy
achieves a significant reduction in energy consumption in non-productive states. In Figure 9, the
energy consumption in unproductive states (Idle, Out-of-service, Warmup) of the four lines (MinTTd ,
MaxD_2.5%, MaxD_5%, MaxD_10%) are compared to a case with machines that are always on and
those applying switch off policies. In all the cases analyzed, adopting a shutdown policy allows for a
significant reduction in energy consumption in non-productive states, ranging from 86% to 89%.
For these reasons, switch-off policies in unbalanced flow lines are necessary and achieve a
reduction in unproductive state energy consumption. If the machines are in always on states, the
unbalancing flow line leads to a reduction in throughput and energy consumption due to more time
being spent in the idle state. Staying for a long time in the idle state is detrimental, since in this state,
the machine is ready to work and then absorbs power without producing. The switch-off policies
warrant energy saving due to the lower energy consumption during warmup.
Designing an unbalanced flow line and controlling the machines state with a switch-off policy can
lead to a reduction in total energy consumption, not only in unproductive states. Figure 10 shows
the increment of energy consumption for an unbalanced flow line design with switch-off policies,
compared to an always on balanced flow line. For this reason, designing a flow line to achieve a high
unbalance under cycle time constraints, and applying a machine shutdown policy, leads to a reduction
in total energy consumption.
29
Appl. Sci. 2020, 10, 1475

5HGXFWLRQRIHQHUJ\FRQVXPSWLRQ

Figure 9. Reduction of energy consumption in unproductive states using switch-off policies.

5HGXFWLRQRIHQHUJ\FRQVXPSWLRQ

MaxD_2.5% MaxD_5% MaxD_10%
Figure 10. Total energy reduction.
Figure 11 shows the total time that stations hold in the inactive state and the number of warmups.
Like the number of pieces in the buffer, the number of times that machines switch on depends on
the position of the bottleneck. Machines turn on fewer times when using model MaxD_5% than model
Max_2.5%, even if the total inactive time is longer, because the last station is the bottleneck. It can be
seen that, if the bottleneck is among the first machines on the line, it can lead to a reduction of the work
in process and storage costs. However, this leads to more switch-ons and higher energy consumption
during warmup.
30
Appl. Sci. 2020, 10, 1475
(
î

(
17597

(
2XWRIVHUYLFHWLPH>V@

1XPEHURI:DUPXSV
î
13136
(
î
(
î
( 6819
4762
(

(
0LQ77G 0D['B 0D['B 0D['B
Number of Warmups Out-of-service time
Figure 11. Warmup and out-of-service time.
Figure 12 reports the switch-off mean times. It can be noticed that the model MaxD_5% gives the
maximum value of the mean inactive time. Thus, this configuration leads to a high inactive time for
each switch off. Instead, in the model MaxD_2.5%, the mean time in the inactive state is the lowest; the
machines turn into off-state since there are only few pieces in the buffer. The UP policy achieves a
reduction in the machine energy consumption removing resource starvation. Indeed, the machine
turns off when the upstream buffer is empty and then does not wait in the idle state using a high
quantity of energy.

Mean Switch-off time [s]

Figure 12. Mean switch off time for the four lines design with UPD policies.
31
Appl. Sci. 2020, 10, 1475
The following figure (Figure 13) shows the mean of the pieces in the buffers of the four models
with the switch off policy. It can be noticed that the number of pieces in the buffers does not depend on
the chosen objective function. Indeed, the number of elements in the buffers depends on the position
of the bottleneck. In fact, in model MaxD_2.5%, as written in Table 2, the bottleneck is the second
machine on the line. For these reasons, if the elements in the buffers are lower than NU on , then the
UPD policy degrades for the machines below the bottleneck in the upstream policy. Future work may
investigate the effect of changing the bottleneck position in an unbalanced flow line in order to obtain
the best compromise between WIP (Work in Process) and the machines switch-off mean times.

Pieces in buffer

WIP

0LQ77G 0D['B 0D['B 0D['B
Buffer_WS1 Buffer_WS2 Buffer_WS3 Buffer_WS4

Buffer_WS5 Buffer_WS6 Buffer_WS7 Buffer_WS8
Buffer_WS9 Buffer_WS10 WIP
Figure 13. Pieces in buffer and mean of the work in progress.
6. Conclusions
A design model of flow lines to improve the efficiency of the switch-off policy has been proposed.
The model is based on the introduction of a couple of workstations; the first is more rapid than the
second in order to facilitate the switch-off of the first machine. Moreover, the proposed model allows
for the evaluation of the potential benefit of energy consumption with a determined reduction in the
production rate. In unbalanced flow lines, the higher idle times lead to a reduction in the total energy
required for the production. However, in the idle state, stations absorb energy without working. For
these reasons, applying the switch-off policy in an unbalance flow line is necessary to reduce energy
consumption in the unproductive state.
The design model is used with the discrete event simulation to highlight the energy saving with
the different levels of production loss fixed. In response, our first research question asked: what is the
impact of the design model proposed on the performance of the production line in terms of energy
saving maximizing the production rate?
Using the simulation, we have demonstrated that the model proposed could improve the reduction
of energy consumption of the flow line more than the design model that does not consider the possibility
of introducing the switch-off policy.
However, in response to our second research question: can the constraint of a limited reduction
loss improve significantly the energy saving of the production line obtaining an adequate trade off?
Our results have demonstrated that the model can support the decision about the better trade-off
between the production rate level and energy consumption reduction. Moreover, the results highlighted
32
Appl. Sci. 2020, 10, 1475
the better choice, if the objective is also the reduction in the number of on/off activities that can affect
the maintenance of the machines. The results show that the number of warmups can be reduced by
properly choosing the bottleneck position, respecting the precedence constraints, in order to achieve
further energy saving.
From a managerial viewpoint, the study was motivated by the important issue of energy
consumption of flow lines. Our results suggests that: (i) the design model should be adapted to
introduce the switch-off policy to obtain a higher benefit from the switch-off policies; (ii) it is possible
to evaluate, with the use of the simulation, the effect of a targeted reduction of production rate (for
example in a determined production planning period) to improve the energy consumption reduction;
and (iii) the model proposed can be extended to different flow lines to support the decision making
about the design and potential energy reduction.
A limitation of our study is that the machines were not affected by failures and the processing
times are deterministic. Machine failures have a significant impact on productivity. For this reason,
integrating maintenance during scheduled machine downtime can lead to a reduction in production
losses. Therefore, future works will consider the unbalance flow line design in a pull system that also
considers machine failures and stochastic processing times. Furthermore, different switch-off policies,
for example, that also consider demand fluctuations for shutdowns, and their effect on throughput and
on energy consumption, will be investigated.
Author Contributions: S.M. performed the experiments and wrote the paper, and P.R. designed the experiments
and revised the paper. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding
Nomenclature
C Cycle time
Cmax Maximum fixed cycle time
dj Distance between nearby stations
i = 1,.., N Operation index
j = 1,.., M Station index
Kj j-th buffer capacity
nj Number of parts in j-th buffer
ND off Downstream buffer level to switch off
ND on Downstream buffer level to switch on
NU on Upstream buffer level to switch on
Pi Power in idle state
Poff Power in out-of-service state
Pw Power in working state
Pwu Power in warm up state
sj State of the j-th station
Td,j j-th station idle time
ti i-th operation processing time
Tj j-th station processing time
twu Time in warm-up state
TTd Cycle idle time
vik Precedence constrains binary variable
WSj j-th station
xij Operation assignment binary variable
33
Appl. Sci. 2020, 10, 1475
Appendix A
Figure A1. Experiments results for line design MaxD_2.5%.
Figure A2. Experiments results for line design MaxD_5%.
34
Appl. Sci. 2020, 10, 1475
Figure A3. Experiments results for line design MaxD_10%.
References
1. The Cadmus Group. Regional Electricity Emission Factors Final Report; The Cadmus Group: Boston, MA,
USA, 1998.
2. Deif, A.M. A system model for green manufacturing. J. Adv. Prod. Eng. Manag. 2011, 19, 1553–1559.
[CrossRef]
3. U.S. Energy Information Administration. Annual Energy Outlook 2019. Available online: https://www.eia.
gov/outlooks/aeo/pdf/aeo2019.pdf (accessed on 5 December 2019).
4. Cecimo. Concept Description for CECIMO’s Self-Regulatory Initiative (SRI) for the Sector Specific
Implementation of the Directive 2005/32/EC; Cecimo: Brussels, Belgium, 2009; Available online:
https://www.eup-network.de/fileadmin/user_upload/Produktgruppen/Lots/Working_Documents/Lot_
ENTR_05_machine_tools/draft_self_regulation_machine_tools_2009-10.pdf (accessed on 12 December 2019).
5. Cecimo. The European Machine Tool Sector and the Circular Economy; Cecimo: Brussels, Belgium, 2019; Available
online: https://www.cecimo.eu/wp-content/uploads/2019/05/Circular-Economy-Report.pdf (accessed on 5
December 2019).
6. Gahm, C.; Denz, F.; Dirr, M.; Tuma, A. Energy-Efficient Scheduling in Manufacturing Companies: A Review
and Research Framework. Eur. J. Oper. Res. 2016, 248, 744–757. [CrossRef]
7. Dahmus, J.B.; Gutowski, T.G. An Environmental Analysis of Machining. In Proceedings of the ASME 2004
International Mechanical Engineering Congress and Exposition, Manufacturing Engineering Materials Hand,
Anaheim, CA, USA, 13–19 November 2004; pp. 643–652. [CrossRef]
8. Gutowski, T.G.; Dahmus, J.B.; Thiriez, A. Electrical Energy Requirements for Manufacturing Processes. In
Proceedings of the 13th CIRP International Conference of Life Cycle Engineering, Leuven, Belgium, 31
May–2 June 2006.
9. Gutowski, T.G.; Dahmus, J.B.; Thiriez, A.; Branham, M.; Jones, A. A Thermodynamic Characterization of
Manufacturing Processes. In Proceedings of the 2007 IEEE International Symposium on Electronics and the
Environment, Orlando, FL, USA, 7–10 May 2007; pp. 137–142. [CrossRef]
10. Gutowski, T.G.; Branham, M.S.; Dahmus, J.B.; Jones, A.J.; Thiriez, A.; Sekulic, D.P. Thermodynamic Analysis
of Resources Used in Manufacturing Processes. Environ. Sci. Technol. 2009, 43, 1584–1590. [CrossRef]
[PubMed]
11. Li, W.; Zein, A.; Kara, S.; Herrmann, C. An Investigation into Fixed Energy Consumption of Machine Tools.
In Proceedings of the 18th CIRP International Conference on Life Cycle Engineering—Glocalized Solutions
for Sustainability in Manufacturing, Braunschweig, Germany, 2–4 May 2011; pp. 268–273. [CrossRef]
12. Zhang, X.P.; Cheng, X.M. Energy Consumption, Carbon Emissions, and Economic Growth in China. Ecol.
Econ. 2009, 68, 2706–2712. [CrossRef]
35
Appl. Sci. 2020, 10, 1475
13. Twomey, J.; Yildirim, M.B.; Whitman, L.; Liao, H.; Ahmad, J. Energy Profiles of Manufacturing Equipment
for Reducing Energy Consumption in a Production Setting; Working Paper; Wichita State Univ.: Wichita, KS,
USA, 2018.
14. Weinert, N.; Mose, C. Investigation of Advanced Energy Saving Stand by Strategies for Production Systems.
Procedia CIRP 2014, 15, 90–95. [CrossRef]
15. Frigerio, N.; Matta, A. Energy-Efficient Control Strategies for Machine Tools with Stochastic Arrivals. IEEE
Trans. Autom. Sci. Eng. 2015, 15, 50–61. [CrossRef]
16. Frigerio, N.; Matta, A. Analysis on Energy Efficient Switching of Machine Tool with Stochastic Arrivals and
Buffer Information. IEEE Trans. Autom. Sci. Eng. 2016, 13, 238–246. [CrossRef]
17. Baybars, I. A Survey of Exact Algorithms for the Simple Assembly Line Balancing Problem. Manag. Sci.
1986, 32, 909–932. [CrossRef]
18. Scholl, A. Balancing and Sequencing Assembly Lines, 2nd ed.; Physica-Verlag Heidelberg: Heidelberg, Germany,
1999.
19. Boysen, N.; Fliedner, M.; Scholl, A. A Classification of Assembly Line Balancing Problems. Eur. J. Oper. Res.
2007, 183, 674–693. [CrossRef]
20. Eghtesadifard, M.; Khalifeh, M.; Khorram, M. A Systematic Review of Research Themes and Hot Topics
in Assembly Line Balancing through the Web of Science within 1990–2017. Comput. Ind. Eng. 2020, 139,
106–182. [CrossRef]
21. Mashaei, M.; Lennartson, B. Energy Reduction in a Pallet-Constrained Flow Shop through On-Off Control of
Idle Machines. IEEE Trans. Autom. Sci. Eng. 2013, 10, 45–56. [CrossRef]
22. Jia, Z.; Zhang, L.; Arinez, J.; Xiao, G. Performance Analysis for Serial Production Lines with Bernoulli
Machines and Real-Time Wip Based Machine Switch-On/Off control. Int. J. Prod. Res. 2016, 54, 1–17.
[CrossRef]
23. Su, H.; Frigerio, N.; Matta, A. Energy Saving Opportunities and Value of Information: A Trade-off in a
Production Line. Procedia CIRP 2016, 48, 301–306. [CrossRef]
24. Renna, P. Energy Saving by Switch-Off Policy in a Pull-Controlled Production Line. Sustain. Prod. Consum.
2018, 16, 25–32. [CrossRef]
25. Duque, E.T.; Fei, Z.C.; Wang, J.F.; Li, S.Q.; Li, Y.F. Energy Consumption Control of One Machine Manufacturing
System with Stochastic Arrivals Based on Fuzzy Logic. In Proceedings of the 2018 IEEE International
Conference on Industrial Engineering and Engineering Management (IEEM), Bangkok, Thailand, 16–19
December 2018; pp. 1503–1507. [CrossRef]
26. Wang, J.; Xue, J.; Duque, E.T.; Li, S.; Chang, Q. Fuzzy Decision of Machine Switch On-Off for Energy Efficient
Operation of Manufacturing System. In Proceedings of the 2017 13th IEEE Conference on Automation
Science and Engineering (CASE), Xi’an, China, 20–23 August 2017. [CrossRef]
27. Wang, J.; Fei, Z.; Chang, Q.; Li, S.; Fu, Y. Multi-State Decision of Unreliable Machines for Energy-Efficient
Production Considering Work-in-Process Inventory. Int. J. Adv. Manuf. Technol. 2019, 102, 1009–1021.
[CrossRef]
28. Marzano, L.; Frigerio, N.; Matta, A. Energy Efficient State Control of Machine Tools: A Time-Based Dynamic
Control Policy. In Proceedings of the 2019 IEEE 15th International Conference on Automation Science and
Engineering (CASE), Vancouver, BC, Canada, 22–26 August 2019; pp. 596–601. [CrossRef]
29. Otto, A.; Otto, C.; Scholl, A. Systematic Data Generation and Test Design for Solution Algorithms on the
Example of SALBPGen for Assembly Line Balancing. Eur. J. Oper. Res. 2013, 228, 33–45. [CrossRef]
36
applied
sciences
Article
In-Line Target Production for Laser IFE
Irina Aleksandrova, Eugeniy Koshelev and Elena Koresheva *
P.N. Lebedev Physical Institute of the Russian Academy of Sciences, 119991 Moscow, Russia;
ivaaleks@gmail.com (I.A.); 2814rosti@mail.ru (E.K.)
* Correspondence: elena.koresheva@gmail.com; Tel.: +7-916-159-3641

Received: 25 December 2019; Accepted: 15 January 2020; Published: 18 January 2020
Featured Application: Fuel target supply to high repetition rate laser facilities including inertial
confinement fusion power plant.
Abstract: The paper presents the results of mathematical and experimental modeling of in-line
production of inertial fusion energy (IFE) targets of a reactor-scaled design. The technical approach is
the free-standing target (FST) layering method in line-moving spherical shells. This includes each
step of the fabrication and injection processes in the FST transmission line (FST-TL) considered as a
potential solution of the problem of mass target manufacturing. Finely, we discuss the development
strategy of the FST-TL creation seeking to develop commercial power production based on laser IFE.
Keywords: inertial fusion energy; free-standing target layering; FST transmission line
1. Introduction
In the frame of International Atomic Energy Agency (IAEA) project (IAEA CRP F13016), one of
the specific objectives was to define options for fuel target material choice and mass manufacturing
methods, requirements, development pathways and potential solutions [1]. In this paper we identify
the principle challenges associated with progression from single shot to high repetition rate operation
with regard to manufacturing methods of a cryogenic fuel target (everywhere further—target) for
laser IFE.
A fusion power plant will consume as many as one million targets per day. Therefore, the
free-standing target transmission line (FST-TL) is an integral part of any IFE reactor. Hereupon,
methodologies of the fabrication and injection processes must be applicable to mass-production
layering at low cost.
To meet the above requirements, the LPI has proposed to use line-moving and free-standing
targets to develop a scientific and technological base for repeatable target supply at the laser focus [2–5].
Precisely moving targets co-operate all production steps in the FST-TL that is considered as a potential
solution of mass-production layering and noncontact target delivery in pulsed, repetitively cycled
IFE systems. This approach includes also the development of nano-layering technologies, which
supports the fuel layer survivability under target injection and transport through the reactor chamber
(everywhere further—FST layering method) [6–8].
The target production area includes high-precision technologies for target fabrication, injection
and tracking. These technologies are currently an important research stage in the leading laboratories
of the world [9–18]. Target compression and ignition is regarded as a major challenge to the IFE
community. In addition, to maintain an acceptable tritium inventory, it is needed the development
of reasonably fast layering techniques. An overview of various fuel layering methods is given in
Reference [2,19].
A traditional approach is based on the following concept: targets for laser experiments are filled
by permeation, and a uniform D−T ice layer (molecular composition: 25% of D2 , 50% of DT molecules,

Appl. Sci. 2020, 10, 686
and 25% of T2 ) is formed by “beta layering” method [19–21]. This method involves crystallization
from a single seed crystal in the fixed target. The process takes place in the anisotropic hydrogen
isotopes with extremely slow cooling (q ~ 3·10−5 K/s) and precise temperature control (<100 μK) for
obtaining cryogenic layers like a single crystal. Long-run beta-layering process at very strict isothermal
conditions requires ~24 h for one attempt. But routine practice is between 1 and 4 attempts, or even
6 attempts [20,21] which generally require several days or a week. The beta-layering method can
form a spherical fuel layer in the uniform thermal environment, but the ability of the D−T targets to
withstand acceleration during injection is a key issue in terms of surface perturbation growth in the
anisotropic layers under thermal and mechanical loads. Such layers also do not contribute to avoiding
the Rayleigh-Taylor instabilities during implosion that decreases the odds for achieving ignition [22].
Moreover, this method is not efficient for repetitively cycled schedule of the target fabrication and
injection because the target must be precisely located in the reaction chamber under precise temperature
control. We emphasize that even in the single-shot laser experiments an important role for achieving
ignition (apart from anisotropic layers) can play some other factors, such as effect of the target support
on ice-layer quality, cryogenic shroud retraction, vibration control, target alignment, etc. As noted in
Reference [9,13], the beta-layering targets currently cost thousands of dollars apiece; thereby, over the
next several years, the focus of the target layering should be on the isotropic fuel structure formation
within moving targets to meet the requirements of implosion physics. This places significant onus on
the IFE laboratories in the area of new layering methods development.
At the Lebedev Physical Institute (LPI) the researches focus on the extension of the FST layering
method (cooling rates q = 1 − 50 K/s) on large cryogenic targets to form isotropic ultrafine solid fuel.
Our philosophy in conceptualizing the in-line target production is based on the FST layering method
to generate a dynamical symmetrization of the liquid fuel within moving targets (instead of solid fuel
redistribution inherent in the traditional approach for motionless targets [19]). During FST layering,
two processes are mostly responsible for maintaining a fast fuel layer formation (Figure 1):
• Firstly, the target rotation when it rolls down along the layering channel (LC) (n-fold-spirals at
n = 1, 2, 3) results in a liquid layer symmetrization;
• Secondly, the heat-transport outside the target via the conduction through a small contact area
between the shell wall and the LC wall. The spiral LC is a special insert into the cryostat, and it is
cooled outside by helium. A contact spot moving along the outer surface of the rotating target
results in a liquid layer freezing.
For all these reasons, the FST layering method is a promising candidate for development of the
FST-TL at a high repetitively cycled capability intended for mass manufacturing of low cost IFE targets.
As a result of long-term research effort, the LPI gained a unique experience in the development of
the FST layering module (FST-LM) for target fabrication with an isotropic ultrafine fuel layer inside
polymer (CH) shells of 1-to-2-mm diameter. Below, we discuss the FST-LM development of the
next-generation for targets over 2 mm in diameter.
38
Appl. Sci. 2020, 10, 686
Figure 1. Free-standing target (FST)-layering setup. In (a): general view; in (b): FST-layering
module (LM)—target transport is injection between the basic units: shell container (SC)—layering
channel (LC)—test chamber (TC)); in (c): (vapor (1) − liquid (2)) interface behavior in 1-mm CH -shell
(P = 765 atm of H2 at 300 K); in (d): LC as a part of the insert into the cryostat.
2. Modeling Results
2.1. FST-Layering Method for Classical High Gain Targets

Among the reactor-scale targets is a classical high gain target (CHGT-1) with the following
parameters [23]: it is a 4-mm diameter CH shell with a 45-μm thick wall, the cryogenic layer thickness
is 200 μm. For the CHGT-1, in our previous work we calculated the value of τform for different initial
target temperatures Tin [6] (Table 1).
Table 1. FST layering time for classical high gain target (CHGT-1).
τform
CHGT-1 Design
Tin = 37 K Tin = 27 K
1 CH shell without overcoat 227.5 c 149.0 c
2 Gold-coated CH shell 13.65 c 8.94 c
Another example of a classical high gain target (CHGT-2) was proposed in Reference [24]. The
target consists of four parts: a gold-coated CH shell, a D−T filled CH foam ablator, a layer of pure solid
D−T fuel, and a D−T vapor cavity. We use this CHGT-2 to examine issues affecting the possibility of
its fabrication by the FST layering method taking into account that it has a very thin CH shell (1 μm)
compared with CHGT-1 (45 μm).
The CHGT-2 specifications are given in Table 2, where Di and D0 are the inner and the outer
diameters of each layer, and Δl and ρl are the layer thickness and density, respectively. A thin CH layer
has a gold overcoat (0.03 μm) to withstand the thermal chamber environment.
39
Appl. Sci. 2020, 10, 686
Table 2. CHGT-2 specifications.
Material Di (μm) Δl (μm) D0 (μm) ρl (g/cm3 )

D−T (vapor) 0 3000 3000 0.3·10−3
D−T (solid) 3000 190 3380 0.25
CH (DT)64 3380 261 3902 0.25775
CH layer 3902 1 3904 1.07
The program on CHGT-2 modeling includes the computation of the FST layering time and the LC
specifications, such as: number of spirals, inclination angle of the spiral, total length of the spiral, the
spiral diameter, and number of turns. The obtained results will allow one to calculate the baseline
parameters necessary at designing of the FST-LM as a means of high-gain direct-drive target production.
For a better understanding of the time-integral performance (TIP) criterion under the FST-LM
operation, we emphasize that the LC must have a well-defined geometry to satisfy the condition:
Target is formed in the LC if τform ≤ τres , (1)
where τform is the layering time, and τres is the time of target residence in the LC. Since a liquid sagged
fuel is the initial state before the FST layering (Figure 1c), then for dynamical fuel symmetrization in
a batch of rolling targets the time of liquid phase existence, τliquid , is a key parameter and must be
sufficient for layer symmetrization. This depends on the temperature Tin of the target entry into the
LC and on the temperature distribution along the LC, i.e., the temperature profiling procedure can
play an important role for successful FST layering.
Generally, one can view the target motion in the following rolling conditions:
• Target slides on the LC surface (no rotation: sliding and only sliding or pure S&S-mode);
• Target combines rolling with sliding (rolling with sliding or mixed R&S-mode);
• Target rolls on the LC surface without sliding (rolling and only rolling or pure R&R-mode).
A key problem of our investigations is that it is necessary to realize only the R&R-mode to avoid
the outer shell roughening and to achieve the fuel layer uniformity. Therefore, the TIP criterion can be
written in the following type (τrol is the time of pure target rolling):
τform ≤ τres = τrol . (2)
Thus, determination of the rolling conditions is one of the main problem, which influences
the choice of the FST-LM operation including simplifying the physics design and modifying the
specifications. First of all we should to determine the spiral angles α for realizing the pure target
rolling (R&R-mode).
Consider a cylindrical spiral of radius R0 . In each point of the spiral the tangent makes a constant
angle α with the horizontal. Let a spherical target of radius R and mass m is moving with a velocity
V (t) along the spiral LC. Then the system of equations in accordance with Newton’s second law of
motion has the following form:
dV/dt = gsinα − Ff /m, (3)
(J/R)dω/dt = Ff − kr , (4)
where g is the free fall acceleration, J and ω are the moment of the target inertia and the angular
velocity of its rotation, Ff is the friction force, N is the reaction of support from the LC wall, and kr is
the rolling friction. If the target rolls without sliding the linear and angular velocities are related by
equation of rolling
V = ωR. (5)
40
Appl. Sci. 2020, 10, 686
Excluding the friction force from Equation (3), we obtain the rolling equation:
(1+ζ)dV/dt = gsinα − kr N/m, ζ = J/(mR2 ). (6)
Besides, there is the conditional relation that should be maintained for the accelerated target to be
in pure rolling:
Ff ≤ Fmax = ks N, N = mcosα{g2 + (V4 cos2 α)/R2 }0.5 , (7)
where ks is the sliding friction. The second component under the root is the square of the centripetal
acceleration. Solving together Equations (6) and (7), we obtain the working range Δα, which is given
by two inequalities:
kr = tgαmin < tgα < tgαmax = {(1+ζ)ks − kr }/ζ. (8)
The next step is modeling of the CHGT-2 layering. We used a simulation code based on solving
the Stephen’s problem [25,26]. Using the data in Table 2 and the data on the fuel heat capacity and
conductivity [19], we got the following values:
— Tin = Ts ~ 35 K: τform = 22.45 s for D2 fuel and τform = 28.52 s for D−T fuel,
— Tin = Td ~ 28 K: τform = 12.05 s for D2 fuel and τform = 14.25 s for D−T fuel,
where Ts is the temperature of fuel separation into the liquid and gaseous phases, and Td is the
depressurization temperature at which the excess gas is removed from the SC. Note that depending on
the shell strength Td can be ~ Ts .
The final step is the computation a set of the optimization parameters related to the LC geometry
to maintain the process of the CHGT-2 fabrication by the FST layering method. Using (8), we have
found that the CHGT-2 can be fabricated in the FST-LM by using a double-spiral LC that is confirmed
by the experimental modeling with pilot LC mockups. The obtained results are presented below.
1. Double-spiral LC:
• Specifications (baseline design: Spiral 1 + Spiral 2): angle of each spiral − α = 11.50, radius of
each spiral − 21 mm, height of each spiral − 450 mm, length of each spiral − 2.261 m, total spiral
length − 4.52 m, total time of target rolling − 23.49 s;
• Tin ~ Td → the double-spiral LC specifications are those at which the TIP criterion (2) is valid for
both D2 and D−T fuel;
• Tin ~ Ts → the TIP criterion is valid for D2 fuel;
• Tin ~ Ts → the TIP criterion is not valid for D−T fuel.
Nevertheless, the double-spiral LC can work in the latter case, as well, because the length of Spiral
2 can be extended on ~2 m to meet the TIP criterion (Figure 2). Generally, this approach is workable
for any spiral LC because several interchangeable spirals with different parameters can be joined to a
baseline LC that depends on the experimental goals.
Figure 2. Test mockup of a double-spiral LC with an extra working spiral having a different
inclination angle.
For the CHGT-2 we have considered a three-fold-spiral LC, as well. The obtained results are
presented below.
2. Tree-fold-spiral LC:
41
Appl. Sci. 2020, 10, 686
• Specifications #1 (baseline design: Spiral 1 + Spiral 2 + Spiral 3): angle of each spiral − α = 16.70,
radius of each spiral − 21 mm, height of each spiral − 88 cm, length of each spiral − 3.066 m, total
spiral length − 9.187 m, total time of target rolling − 33.29 s;
• Specifications #2 (Specifications #1 + Spiral 4). Spiral 4: angle − α = 30, radius − 21 mm, height −
10.8 cm, length − 2.070 m, total length of Spiral 3 and Spiral 4 − 5.136 m. Other words, this 3-fold
LC is designed in a special configuration with an extra short Spiral 4 (combined LC).
The modeling has shown that, in ~5 s after a start, the target motion is carried out with a constant
velocity Vmax = 0.3 m/s. The total rolling time for Specifications #1 is 33.29 s. Thus, we can realize the
rolling conditions for the CHGT-2 and satisfy the TIP criterion in the case of 3-fold-spiral LC, even
with a certain time margin.
Note also that the proposed 3-fold-spiral LC (Specifications #1) can have an extra spiral
(Specifications #2) so that two spirals “Spiral 3 + Spiral 4” make a combined LC in which the
Spiral 4 parameters differ considerably from those of Spiral 3. Such combined LC consists of these
spirals assembled one after another: “acceleration Spiral 3 + deceleration Spiral 4” for controlling the
target velocity at the output. For example, in our case, it will take no more than 1.5 s at α = 30 to zero
the target velocity. The obtained results are summarized in Table 3 in which τ2rol and τ3rol are the
rolling times for the double-spiral LC and for the tree-fold-spiral LC, respectively.
Table 3. FST layering for both D2 and D−T.
D2 Fuel
Calculation Experiment
Tin τform τ2rol τ2rol
35.0 K 22.45 s
23.49 s 23.5 s (min)
27.5 K 12.05 s
D−T Fuel
Calculation Experiment
Tin τform τ3rol τ3rol
37.5 K 28.52 s 33.29 s
35 s (min)
28.0 K 14.25 s 34.79 s
A few comments should be made regarding the gold-coated CH shell. In this case for both D2 and
D−T fuel the layering time τform < 0.5 s. This means that in order to have a uniform layer, a temperature
profiling along the LC becomes a necessary condition for increase in time of τrol at temperatures more
than the triple point one. In this case Tin can be ~21 K as the hydrogen isotope vapor pressures near
the triple point determine the minimum operating pressures (do not exceed 0.2 atm [19]) that allows
one to consider an injection filling procedure. Filling of the CH shells with a cryogenic liquid fuel is
suitable for the FST layering method because the fuel in the shell directly before the layering has a
two-phase state “Liquid + Vapor” (Figure 1c).
Thus, in the IFE, the FST layering is a credible pathway to a reliable, consistent, and economical
target supply. Currently, our studies on the creation of FST-LM lie in the field of compatibility of its
work with a noncontact method for delivering targets to the reactor.
2.2. Noncontact Accelerating of a Target

Creation of a delivery system based on noncontact positioning and transport of IFE targets
represents one of the major tasks in the IFE research program. The purpose is to maintain a fuel
layer quality during the target acceleration and injection at the laser focus. The ability of the target
to withstand acceleration during its transport and injection is a key issue in the design of the target
delivery system. The stringent requirements to the delivery process are as follows [13,27]: targets
42
Appl. Sci. 2020, 10, 686
should have a temperature of not higher than 18.3 K; the overloads during its acceleration in the
injector can be a = 500−1000g (but there is a desired acceleration limit a <500g in order to significantly
reduce risks of the target damage due to mechanical overloads arising in the process of its acceleration;
targets must be accelerated to high injection velocities (200−400 m/s) to withstand the reaction chamber
environment. Therefore, the target acceleration is carried out in a special capsule—target carrier or
sabot, which transmits a momentum of motion to the target.
The LPI program includes much development work on creation of different designs of the hybrid
accelerators for IFE target transport with levitation. One of the main directions is an electromagnetic
accelerator (EM-AC) + PMG-System, where PMG is the permanent magnet guideway. The operational
principle is based on a quantum levitation of type-II, high temperature superconductors (HTSC) in the
magnetic field. The target temperature (T = 18.3 K) excludes the use of type-I superconductors as they
have the critical temperature Tc < 10 K, i.e., their heating above 10 K destroys their superconductivity.
At the current research stage, the concept development of a hybrid accelerator «EM-AC + PMG»
is complete and proof-of-principle experiments in the mutually normal magnetic fields are carried
out. The proposed accelerator is a combination of the acceleration system (field coils generating
the traveling magnetic waves) and the levitation system (PMG including a magnetic track). The
experimental setup is shown in Figure 3. The PMG consists of 6 rectangular permanent magnets, and
two in the middle were covered with a ferromagnetic plate. Along the magnetic track (or acceleration
length) the magnetic field is constant (B = 0.33 T onto the permanent magnet surface), which allows
the HTSC-Sabot to move with no energy loss. Normal to the acceleration length, the magnet poles
are aligned anti-parallel to each other N-S-N (Figure 3a) that produces a considerably strong gradient
along the width of the magnetic track. This gradient prevents the motion of the HTSC-Sabots, and they
remain located in the transverse direction (Y-axis in the rectangular coordinate system XYZ).
Figure 3. Noncontact acceleration setup. In (a,b): permanent magnet guideway (PMG)-system

dimensions (as a whole, it lies on a ferromagnetic substrate); in (c): a top view of the mutual
arrangement of the field coil (1) and the PMG-system (2) of two magnets covered with a ferromagnetic
plate; in (d): a starting location of the high temperature superconductors (HTSC)-Sabot (3), just before
the acceleration, with the directions of the levitation force (vector B2 in parallel to the Z-axis) and of the
driving force (vector B1 in parallel to the X-axis), as well as along the acceleration length (X-axis).
In our experiments, the HTSC materials are superconducting ceramics based on YBa2 Cu3 O7−x
(or Y123) [28] and superconducting tapes of the second generation based on GdBa2 Cu3 O7−x (or
Gd123) [29]. These HTSC materials have practically the same values of the critical temperature (Tc )
and critical magnetic field (Bc ), but they differ from each other in terms of their densities (Table 4) and
the magnetic susceptibility (for Y123 (|χ| = 2.5·10−3 [28]) and for Gd123 (|χ| = 4.5·10−4 [29]). Recall that
the phenomenon of superconductivity exists only below the values of Bc and Tc [30].
43
Appl. Sci. 2020, 10, 686
Table 4. Parameters of the HTSC-materials [28,29].
HTSC Type Density (g/cm3 ) Bc at 0 K (T) Tc (K)

Y123 ρ = 4.33 >45 T 91
Gd123 ρ = 3.25 >45 T 92
The experimental setup (Figure 3) operates based on the following effects: the levitation of the
HTSC-Sabot is due to the Meissner effect, and the stability of levitation over the PMG system is ensured
by vortex pinning [30]. The experiments were made in the mutually normal magnetic fields: the first is
B1 (from the field coil to move the HTSC-Sabot) directed along the acceleration length, and the second
is B2 (from the permanent magnets to counteract the gravity) directed normally to the acceleration
length (Figure 3c). The PMG-system was optimized in order to achieve simultaneously large levitation
forces and stability of the levitation after small perturbations (Figure 3a–c). Figure 3d shows a starting
location of the HTSC-Sabot (3) just before the acceleration.
A key feature of the proposed PMG-system is that it has a simple arrangement. It is assembled
from several individual magnets that allows it to be extended to any required acceleration length and
create multi-stage accelerators. We start with a simple case. Figure 4 demonstrates the work of a
one-stage accelerator.
Figure 4. One-stage accelerator: magnetic acceleration of the levitating HTSC-Sabot from Gd123. In
(a): results of proof-of-principle (POP) experiments: HTSC-sabot (≤30 mm) levitates over the linear
PMG-system (~30 cm); in (b,c): the HTSC-Sabot looks like it is attached to a magnetic track.
The HTSC-Sabot obtains a velocity of 1 m/s (Figure 4a) and keeps it over the whole acceleration
length (22.5 cm, see Figure 3c). The vertical and horizontal levitating drift is not observed (Figure 4b,4c).
The levitation height is ~3 mm. A set of control experiments has shown that the HTSCs can be
successfully used to maintain a friction-free motion of the HTSC-Sabot over the PMG. Besides, they
can provide a required stability of the levitation height over the whole acceleration length due
to a flux pinning effect [30] inherent in the interaction between a Type-II superconductor and a
permanent magnet.
Technologically, the obtained results allow a convenient spacing of the multiple coils (also
called a multiple-stages accelerator) and lead to realizing very high velocities of the HTSC-Sabot. In
Reference [5], we have reported that, using the MgB2 -superconducting coils [31] as a driving body, it
is possible to achieve the required velocity Vinj = 200 m/s without exceeding the acceleration limits
(a = 400g < 500g), as well as the injector requirement of ~5 m in length. It operates at a very low
temperature (~18 K) allowing no heat energy to be passed into the target from the accelerating medium.
For future IFE power plants, the injection velocities of Vinj > 200 m/s can be easily achieved because
the combination of “EM-AC + PMG-System” successfully works as a multiple-stages accelerator.
Therefore, our next step is the development of a multiple-stages accelerator. Below we propose a
first version of such accelerator (Figure 5). Superconducting sabot includes several HTSC components
(Figure 5a). It comprises not only the accelerating HTSC-coils but also the HTSC-plates for providing
its stable levitation along the magnetic track (Figure 5b). During acceleration, the diameter of the
44
Appl. Sci. 2020, 10, 686
barrel exceeds the diameter of the HTSC-Sabot (Figure 5c), allowing it to protect the target against the
thermal damage and to reduce the risks from wedging of the sabot in the injector guiding tube.
Figure 5. Multiple-stages accelerator for noncontact target transport: (a): HTSC-Sabot, which includes
1—polymer matrix, 2—HTSC-coils (MgB2 -driving body), 3—HTSC-plates, 4—target, 5—shield or
cover to protect the injected target from a head wind of a residual gas in the reaction chamber; (b):
levitation of an HTSC-Sabot mockup ( made from the Gd123-tapes) over the linear PMG-system at 80 K
(magnetic field is equal to 0.33 T on the permanent magnet surface); (c): schematic of the multiple-stages
accelerator, which includes 6—HTSC-Sabot, 7—coil, 8—guiding bore, 9—magnetic track, 10—key to
run a traveling magnetic wave, 11—capacitor, 12, 13—optocoupler.
Note also that the sequence of field coils loaded on the corresponding capacitance is a line with
lumped parameters. From the point of view of the relative position of the accelerating traveling
pulse and magnetic dipole (in our case HTSC-Sabot), steady is only the case when an impulse pushes
(rather than pulls) a magnetic dipole. This means that the phase (longitudinal) stability is on the
front slope of the traveling impulse. Such a technique is called the principle of auto phasing [32] and
corresponds to the case of using superconducting materials in the manufacture of the target carriers as
they HTSC-Sabot are pushed out from the area of a stronger magnetic field [30].
3. FST-TL Key Elements and Their Functional Description

The free-standing targets are an indispensable requirement for any IFE power plant for continuous
target layering and their repeatable injection to the reaction chamber. In other words, targets must
remain un-mounted in each production step, and in the IFE experiments, they will be fed directly from
the FST-LM to the assembly module with HTSC-Sabots.
3.1. FST Layering Module

Our previous results, as well as the expert analysis carried out in the frame of IAEA CRP 13016
(under the IAEA contract No. 20344 [27]), have shown that the proposed FST-LM (see Figure 1) can
meets the requirements for CHGT-2 fabrication.
The construction of the FST-LM is shown in Figure 6.
45
Appl. Sci. 2020, 10, 686
Figure 6. The FST-LM construction: (A) Protection-sluice module, which serves to accommodate
the shell container (SC); (B) Layering channel (LC); (C) Test chamber (TC). 1—cryostat housing,
2—nitrogen vessel, 3—thermal shield, 4—helium vessel, 5—well, 6—internal cavities of the insert,
7—liquid-nitrogen transmission line, 8—PSM, 9—place of the SC disposal in PSM, 10—LC, 11—target
collector, 12—insert into the helium cryostat well, 13—cryostat neck, 14—traction valve, 15—He-gas
output, 16—heat exchanger, and 17—tube space.
The protective-sluice module (Figure 6, item A) serves to accommodate the SC. The LC is made in
the form of a tubular spiral (Figure 6, item B), and the outer surface of which is carefully polished and
covered with a multilayer screen made of a mylar film to further reduce a degree of the surface blackness.
3.2. FST-Projectile Assembly Module

Before the acceleration and injection stages, the free-standing targets must be mounted in a special
module in the form “HTSC-Sabot + Target + Cover” (everywhere further—FST-Projectile). Each step
of the assembly process will be operated at significant rates. Several remarks should be made here.
In the FST-Projectile, the HTSC-Sabot plays a major role allowing for: (a) to transmit effectively an
acceleration pulse to the target, and (b) to protect the target from damage during the acceleration
process. An important characteristic of its design is the shape of a target nest. Our study has shown
that the shape of the target nest allows one to significantly increase the upper limit of the allowable
mechanical overloads on the target and to minimize the injector dimensions. In optimal case the sabot
nest should have a conical shape with the angle of 870 [4]. A chart of the HTSC-Sabot with the target
placed in the conical nest and the cover is shown in Figure 5a.
The FST-projectile assembly module is shown in Figure 7.
46
Appl. Sci. 2020, 10, 686
Figure 7. Repeatable assembly of the FST-projectile for safe noncontact delivery of the free-standing
targets at the laser focus: 1—FST-LM including a double-spiral LC; 7—target accumulator and system for
target transport under gravity to the rotating drum; 3—HTSC-Sabot feeder; 2—a set of the HTSC-Sabots;
3—HTSC-Sabot feeder; 4—extruder of solid D2 cylinder (protective cover production); 5—heater to
cut the cover from D2 cylinder; 6—electromagnetic pusher; 7—rotating drum; 8—field coil for the
FST-projectile transport to the starting point of the acceleration system.
3.3. Noncontact Delivery System

Our near-term goal is to reduce the acceleration length (~5 m) of the linear accelerator by using
a circular PMG of 0.5 m in outer diameter (OD) in the HTSC magnetic levitation (maglev) transport
system. This HTSC-maglev is intended for the high target velocity applications, target trajectory
correction, and creation of a precise injector.
First, experiments with an acceleration length ~31.5 cm (corresponding to one full turn of a median
circle) were made to demonstrate this approach on laboratory scale tests. Figure 8 shows the process
of HTSC-Sabot moving at T = 80 K over the circular PMG-system under an electromagnetic driving
pulse. The HTSC-Sabot is the same as in the experiments with the linear accelerator (Figure 5b). It is a
hollow parallelepiped with dimensions: width 4 mm, height 4 mm, and length 24 mm, which made
from the Gd123-tapes with a thickness of 0.3 mm. The HTSC-Sabot mass is 0.2 g. The PMG-system is a
neodymium magnet with OD = 10 cm and thickness 5 mm. The internal diameter (ID) is 5 cm. This
disc magnet is placed in a ferromagnetic housing open from above (housing thickness 1 mm). Such a
configuration allows achieving the desired magnetic field profile over the PMG-system (Figure 8a).
Figure 8. Laboratory scale tests with the PMG-system of a circular type (a–f—consecutive video frames
of the HTSC-Sabot counterclock-wise moving).
47
Appl. Sci. 2020, 10, 686
3.4. Physical Layout of the FST-TL

Currently, the main problems for target fabrication and injection are as follows:
• There are considered various target designs for direct drive IFE (e.g., compare CHGT-1 and
CHGT-2);
• For each target design, low cost methods of high rate target fabrication are needed;
• Targets must survive mechanical and thermal loads during injection;
• Noncontact options for target acceleration and repeatable injection are also needed.
A schematic of the FST-TL of repeatable operation satisfying the listed above requirements is
shown in Figure 9.
Figure 9. Schematic of the FST- transmission line (TL) operation: 1—SC, 2—FST-LM, 3—extruder
for protective cover preparation, 4—FST-Projectile assembly module, 5—noncontact accelerator,
6—FST-Projectile separation module, 7—system for on-line target tracking and characterization, and
8—reaction chamber wall
Note that a scenario of the FST-TL operation in a batch mode has been shown on a reduced scale
for the targets under 2 mm in diameter:
• Fuel filling of a batch of 5-to-25 free-standing targets at one time (Pmax = 1000 atm at 300 K)
(Figure 1c);
• Fuel layering within the free-standing and line-moving targets using a single LC (Figure 1d), and
the cryogenic layer thickness was up to 100 μm (Figure 1b);
• Levitating HTSC-Sabot acceleration in the mutually normal magnetic fields (Figures 4 and 5b);
• Target injection into the test chamber at a rate of 0.1 Hz at 300 K (Figure 10) and at T = 4.2 K
(Figure 11a);
• Target tracking and characterization using the Fourier holography methods.
The last point is not a topic of this report. Therefore we only note the results that are important
for a general understanding of the problem of techniques integration into a target production line
capable of producing the required amount of the cryogenic targets. As shown in Reference [3], the
Fourier holography of the image recognition is a promising way for on-line characterization and
tracking of the flying targets. In such a scheme the recognition signal is maximal in case of an exact
match between the real and the etalon images; the operation rate is several μs. Achievements in the
Fourier target characterization embodied in creation of a simulation code developed as special software
“HOLOGRAM”. A set of computer experiments have shown that this approach allows to achieve
a high accuracy of 0.7 μm in the following operations: (1) recognition of the target imperfections in
48
Appl. Sci. 2020, 10, 686
both low- & high- harmonics, (2) quality control of both a single target and a target batch, and (3)
simultaneous control of the injected target quality, its velocity and trajectory.
Figure 10. On-holder injection experiments at 300 K (1-6 is the sequence of video frames of the
experiment): (a) free target positioning onto the TC bottom; (b) target injection to a special cylindrical
cavity imitating a hohlraum-like unit; (c) target injection to a circular PMG-holder.
Figure 11. (a–c): A set of the injection experiments at cryogenic temperatures.
4. FST-TL in the Single-Shot Laser Experiments

The FST-TL system shown in Figure 9 can also be used in the laser experiments on a single-shot
target irradiation. In this case, it is necessary to capture the injected target on a holder precisely located
in advance at the given point. We conducted a series of experiments on the target injection using
some different holders: tripod-holder, vacuum suspension on a glass capillary, hohlraum-like unit,
and PMG-holder.
In our experiments we have used the gravity injector mounted at the end of the FST-LM. This
injector covers a fairly small solid angle (about 150 ) which is no greater than the solid angle of any
diagnostic means in the reaction chamber. The target transport by injection is an excellent springboard
for overcoming a departure between two concepts: mounted and free-standing targets. Figure 10a
shows the relative position of the “injector—holder” couple for a multi-beam laser facility, as well
as the results of experimental modeling of the free-standing target assembly with the tripod-holder
49
Appl. Sci. 2020, 10, 686
prearranged at a given point of the test chamber. A final adjustment of the target position relative
to the position of the laser focus is carried out using piezoelectric sensors mounted in the tripod
base. In Figure 10a, the following designations are accepted: 1—is the retractable cryogenic shroud
which must be removed just before the moment of the target irradiation by a laser, 2—output flange;
3—piezoelectric supports made of quartz; 4—nozzle for the target injection from the FST-LM; 5—shell
with fuel layer; 6—threads forming a tripod; 7—thread tensioner. The tripod-holder does not require
the use of glue for mounting the target, which increases the reliability of operation at cryogenic
temperatures and provides a minimum contact area between the target and the threads. The alignment
accuracy is +20 μm. The principle of the target alignment is based on the inverse piezoelectric effect
which realized by using three rectangular micro-supports made of quartz piezoelectric elements
located as shown in Figure 10a (item 3). At the output flange Figure 10a (item 2) there are electrical
contacts (not shown in the figure) through which a signal is applied to change the dimensions of
these piezoelectric elements. They are small in size and able to work at cryogenic temperatures with
a minimal heat loss. The target characterization is based on a backlight shadowgraphy which is a
widely used imaging technique. It can provide high contrast object images for a better understanding
of an object’s shape or size [33]. Specialized software is developed to automatically determine the
coordinates of the target and generate commands for correcting its position in real time. Figure 10b
shows the details of the injection process on the tripod-holder (1, 2, 3 are the frames before, during, and
after the shell injection). The experimental conditions are as follows: CH shell of 2 mm in diameter,
the chamber wall temperature is T = 300 K, the pressure inside it is 1 Torr, the holder is the tripod
made from three glass fibers (10 μm in diameter). Figure 10c shows the injection of a 300-μm diameter
glass shell under gravity from the nozzle of the FST-LM onto the top of a 90-μm diameter capillary
prearranged at a given point of the test chamber (here, the target is trapped by the vacuum mount
mechanism using this capillary): 1—schematic of the start shell position, 2–5—step-by-step experiment
with target injection and its trapping on the top of the capillary, 6—schematic of the final shell position,
7—nozzle, 8—capillary, 9—shell, and 10—restrictive ring used to reliably position the target on the
capillary (this ring is removed from the target irradiation area before the laser shot).
Some other options for the target injection and positioning are shown in Figure 11. A unique
feature of these experiments is the target injection at cryogenic temperatures. Experimentally, injection
can be carrying out in various ways: (1) directly to the TC with a free target positioning onto the TC
bottom (Figure 11a), (2) to a special cylindrical cavity imitating a hohlraum-like unit (Figure 11b), and
(3) to a circular PMG-holder, which is a promising way for non-contact manipulation and positioning
of the cryogenic targets (Figure 11c). In the latter case, for an active target guidance we used an
Y123-coated CH shell of 2 mm in diameter (Y123-layer thickness ~10 μm) and a disc magnet with OD
= 15 mm, ID = 6 mm, and thickness 5 mm. A ferromagnetic insert is located in the central hole of the
magnet. The magnetic field on the periphery of the disc magnet is about 0.3 T.
In the IFE research, these results attract a significant interest due to their potentials for development
of the HTSC-maglev suspension technologies and can be used for enhancement of the operating
efficiency of the injection process in the single-shot laser experiments.
5. Conclusions
In the direct-drive approach to IFE, the free-standing cryogenic targets of high gain design are
required for fueling a laser energy plant. The targets must be delivered to the chamber center with
a high accuracy and a high repetition rate. Because fusion reactions must occur approximately ten
times per second, the FST-TL becomes an integral part of any fusion reactor. The FST-TL consists of a
target production facility, target acceleration, and injection and tracking systems. Currently, target
production is in transition from one-of-a-kind to one million of high quality ignition targets per day.
Creation of the target delivery system based on noncontact positioning and transport of the cryogenic
targets represents one of the major tasks, as well. Therefore, in the frame of IAEA CRP F13016 [1,27], of
50
Appl. Sci. 2020, 10, 686
particular concern is development of a conceptual design of FST-TL including a mass-manufacturing

of reactor-scaled targets and their noncontact delivery at the laser focus.
Achieving controlled thermonuclear ignition is one of the goals for IFE, and a target with the
isotropic fuel structure is a critical aspect of successfully reaching that goal. This indicates that currently
target fabrication technologies for IFE laser experiments are a challenge. A promising candidate is the
FST layering method—fast fuel layering inside moving free-standing targets—which is well suited
to economical mass production of IFE targets. The cost of the targets produced by the FST layering
method may be several orders of magnitude lower than for the traditional target fabrication capabilities
and technologies.
In this paper, high-gain direct-drive cryogenic targets using foam to support the D−T fuel were
analyzed. These are the shells of ~4 mm in diameter with a wall from compact and porous polymers.
The layer thickness is ~200 μm for pure solid fuel and ~250 μm for in-foam solid fuel. A credible path
to solve the issue is the FST layering method developed at the LPI. This method is unique and has no
analogs in the world. The computation showed that the layering time for such targets was τform <
23 s for D2 and τform < 30 s for D−T fuel, and they can be successfully fabricated by the FST-layering
method inside the n-fold LCs at n = 2, 3. So, fast fuel layering is a necessary condition for the producing
targets in the massive numbers and for obtaining fuel as isotropic ultrafine layers. Namely, isotropic
fuel structure supports the layer quality survivability under the target acceleration, injection, and
transport through the reactor chamber. For on-line characterization and tracking of the injected targets
it is planned to apply the method of coherent optics based on the Fourier holography.
Additionally, throughout this paper, the principle of innovative HTSC-maglev transport systems
are explored for the safe, stable and noncontact target delivery at the laser focus. We address this
question both by experimentally demonstrating such system performance and by calculating their
parameters. The HTSC levitation is based on the interaction between the magnet and the HTSC. Due to
unique characteristics, the HTSC materials have demonstrated tremendous potential for the noncontact
target transport.
Any maglev scheme has to address three basic properties: levitation, guidance, and acceleration.
The characteristics of the permanent magnets composing the guideway in the PMG-system are very
important in terms of levitation force and stability. We found that not always complicated magnet
arrangements bring significant improvements with respect to some simpler arrangements that also
provide large levitation force. In our study we use two simple magnet arrangements: linear and circular.
In the experiments with one-stage linear accelerator, the HTSC-Sabot from the second generation
tapes Gd123 gains a speed up to ~1 m/s which remains the same over a 22.5-cm magnetic track
(movement time is 0.22 s). The calculations have shown that, using the driving body from MgB2
superconducting cables as an HTSC-Sabot component, allows reaching the injection velocities of
200 m/s at 5-m-acceleration length without exceeding the established restrictions related to the target
acceleration (a = 400 g < 500 g). These velocities can be obtained by using a set of the field coils the
number of which is more than 100 pieces (multiple-stages accelerator). Besides, our experiments with
strongly pinned superconductors (HTSC-sabot) and permanent magnets in the PMG-system display
high stability of the levitation and acceleration processes. This is directly related to the safe operation
and design of the whole delivery system.
Recent results obtained at the LPI may help improving the actual design of HTSC-maglev
accelerators. Significant reduction of the accelerator dimensions and the number of the field coils
can be obtained in a circular accelerator, in which only several field coils are arranged in a circle in
the PMG-system. In such geometry, the HTSC-sabot will move along a practically constant circular
orbit in pulsed, repetitively cycled mode and will gradually gain the required velocity. The LPI is
currently making major efforts to create such accelerator to reach the injection velocities in the range of
200–400 m/s with a constant PMG dimension and a constant number of the field coils. This is a novel
way to enhance the performance of target accelerators through HTSC guiding technology.
51
Appl. Sci. 2020, 10, 686
Thus, a unique scientific, engineering and technological base developed at the LPI allows creating
a FST-TL prototype for mass targets production with the ultrafine fuel structure and their noncontact
delivery at the laser focus. Such targets for application to high repetition rate laser facilities allow one
to test the reactor-scaled technologies and to run a pioneering research of laser direct drive using, for
the first time, the isotropic hydrogen fuel in the target compression experiments, including a possibility
to apply the FST-TL for single-shot laser experiments for achieving the laboratory-based ignition.
Author Contributions: Conceptualization, I.A. and E.K. (Elena Koresheva); software, I.A.; investigation, E.K.
(Eugeniy Koshelev) and E.K. (Elena Koresheva); writing—original draft preparation, I.A.; writing—review and
editing, E.K. (Elena Koresheva); project administration, E.K. (Elena Koresheva). All authors have read and agreed
to the published version of the manuscript.
Funding: This research was funded by the International Atomic Energy Agency (https://www.iaea.org) under
Research Contract No. 20344 and by the Russian Government in the frame of the State Task Program.
References
1. Pathways to Energy from Inertial Fusion: Materials beyond Ignition. In Proceedings of the 3rd IAEA
Research Coordination Meeting of the Coordinated Research Project (F13016), Vienna, Austria, 21−23 January
2019; Available online: https://www.iaea.org/events/evt1805417 (accessed on 23 January 2019).
2. Aleksandrova, I.; Koresheva, E. Review on high repetition rate and mass production of the cryogenic targets
for laser IFE. High Power Laser Sci. Eng. 2017, 5, e11. [CrossRef]
3. Aleksandrova, I.; Koresheva, E.; Koshelev, E.; Krokhin, O.; Nikitenko, A.; Osipov, I. Cryogenic hydrogen
fuel for controlled inertial confinement fusion (concept of cryogenic target factory based on FST-layering
method). Phys. At. Nucl. 2017, 80, 1227–1248. [CrossRef]
4. Aleksandrova, I.; Koresheva, E.; Koshelev, E. Multiple target protection methods for target delivery at the
focus of high repetition rate laser facilities. Probl. At. Sci. Technol. Ser. Thermonucl. Fusion 2018, 41, 73–85.
5. Aleksandrova, I.; Akunets, A.; Koresheva, E.; Koshelev, E. Contactless HTSC sabot acceleration by mutually
normal magnetic fields generated in a PMG system with magnetic travelling wave. In Proceedings of the
XLV International Conference on Plasma Physics and Controlled Thermonuclear Fusion, Moscow, Russia,
2–5 April 2018.
6. Aleksandrova, I.; Koresheva, E.; Osipov, I.; Timasheva, T.; Tolokonnikov, S.; Panina, L.; Belolipetskiy, A.;
Yaguzinskiy, L. Ultrafine fuel layers for application to ICF/IFE targets. Fusion Sci. Technol. 2013, 63, 106–119.
[CrossRef]
7. Aleksandrova, I.; Bazdenkov, S.; Chtcherbakov, V. Rapid fuel layering inside moving free-standing ICF
targets: Physical model and simulation code development. Laser Part. Beams 2002, 20, 13–21. [CrossRef]
8. Aleksandrova, I.; Bazdenkov, S.; Chtcherbakov, V.; Gromov, A.; Koresheva, E.; Koshelev, E.; Osipov, I.;
Yaguzinskiy, L. An efficient method of fuel ice formation in moving free standing ICF/IFE targets. J. Phys. D
Appl. Phys. 2004, 37, 1163–1179. [CrossRef]
9. Goodin, D.; Aleksander, N.B.; Brown, L.C.; Frey, D.; Gallix, R.; Gibson, C.R.; Maxwell, J.L.; Nobile, A.J.;
Olson, C.L.; Petzoldt, R.W.; et al. A cost-effective target supply for inertial fusion energy. Nucl. Fusion 2004,
44, S254–S265. [CrossRef]
10. Canaud, B.; Fortin, X.; Garaude, F.; Meyer, C.; Philippe, F. Progress in direct-drive fusion studies for the Laser
Mégajoule. Laser Part. Beams 2004, 22, 109–114. [CrossRef]
11. Frey, D.; Alexander, N.; Bozek, A.; Goodin, D.; Stemke, R.; Drake, T.; Bitner, D. Mass production methods for
fabrication of inertial fusion targets. Fusion Sci. Technol. 2007, 51, 786–790. [CrossRef]
12. Bousquet, J.; Hung, J.; Goodin, D.; Alexander, N. Advancement in glow discharge polymer coatings for mass
production. Fusion Sci. Technol. 2009, 55, 446–449. [CrossRef]
13. National Research Council of the National Academies. Inertial Fusion Energy Technologies. In An Assessment
of the Prospects for Inertial Fusion Energy; The National Academies Press: Washington, DC, USA, 2013;
Chapter 3; pp. 89–105. ISBN 978-0-309-27081-6.
52
Appl. Sci. 2020, 10, 686
14. Mori, Y.; Nishimura, Y.; Ishii, K.; Hanayama, R.; Kitagawa, Y.; Sekine, T.; Takeuchi, Y.; Satoh, N.; Kurita, T.;
Kato, Y.; et al. 1-Hz Bead-Pellet Injection System for Fusion Reaction Engaged by a Laser HAMA Using
Ultra-Intense Counter Beams. Fusion Sci. Technol. 2019, 75, 36–48. [CrossRef]
15. Mori, Y.; Iwamoto, A.; Ishii, K.; Hanayama, R.; Okihara, S.; Kitagawa, Y.; Nishimura, Y.; Komeda, O.; Hioki, T.;
Motohiro, T.; et al. Spherical shell pellet injection system for repetitive laser engagement. Nucl. Fusion 2019,
59, 096022. [CrossRef]
16. Ren, L.; Shao, P.; Zhao, D.; Zhou, Y.; Cai, Z.; Hua, N.; Jiao, Z.; Xia, L.; Qiao, Z.; Wu, R.; et al. Target Alignment
in the Shen-Guang II Upgrade Laser Facility. High Power Laser Sci. Eng. 2018, 6, e10. [CrossRef]
17. Tsuji, R. Trajectory adjusting system using a magnetic lens for a Pb-coated superconducting IFE target. Fusion
Eng. Des. 2006, 81, 2877–2885. [CrossRef]
18. Kubo, T.; Karino, T.; Kato, H.; Kawata, S. Fuel Pellet Alignment in Heavy-Ion Inertial Fusion Reactor.
IEEE Trans. Plasma Sci. 2019, 47, 2–8. [CrossRef]
19. Kucheyev, S.O.; Hamza, A.V. Condensed hydrogen for thermonuclear fusion. J. Appl. Phys. 2010, 108, 091101.
[CrossRef]
20. Koziozemski, B.J.; Mapoles, E.R.; Sater, J.D.; Chernov, A.A.; Moody, J.D.; Lugten, J.B.; Johnson, M.A.
Deuterium-Tritium Fuel Layer Formation for the National Ignition Facility. Fusion Sci. Technol. 2011, 59,
14–25. [CrossRef]
21. Mapoles, E. Production of hydrogen ice layers for NIF targets. In Proceedings of the 7th International
Conference on Inertial Fusion Science and Applications, Bordeaux, France, 12–16 September 2011.
22. Bringa, E.M.; Caro, A.; Victoria, M.; Park, N. Atomistic modeling of wave propagation in nanocristals.
J. Miner. Met. Mater. Soc. 2005, 57, 67–70. [CrossRef]
23. Nakai, S.; Miley, G. Physics of High Power Laser and Matter Interactions; Word Scientific Publishing: Singapore,
1992; Volume 1, pp. 87–90.
24. Bodner, S.; Colombant, D.; Schmitt, A.; Gardner, J.; Lehmberg, R.; Obenschain, S. High Gain Target Design
for Laser Fusion. Phys. Plasmas 2000, 7, 2298–2301. [CrossRef]
25. Tikhonov, A.N.; Samarsky, A.A. Equations of Mathematical Physics; Nauka: Moscow, Russia, 1977.
26. Samarsky, A.A.; Mikhailov, A.P. Mathematical Modeling: Ideas, Methods, Examples; Nauka: Moscow,
Russia, 2002.
27. Koresheva, E. FST transmission line for mass manufacturing of IFE targets. In Proceedings of the 1st IAEA
RCM of the CRP Pathways to Energy from Inertial Fusion: Materials beyond Ignition, Vienna, Austria, 16−19
February 2011; IAEA: Vienna, Austria, 2016.
28. Golovashkin, A.I.; Ivanenko, O.M.; Leitus, G.I.; Mitsen, K.V.; Karpinskii, O.G.; Shamrai, V.F. Anomalous
behavior of the structural parameters of the ceramic YBa2Cu3O7 near the superconducting transition. J. Exp.
Theor. Phys. (JETP) Lett. 1987, 46, 410–412.
29. Lee, S.; Petrykin, V.; Molodyk, A.; Samoilenkov, S.; Kaul, A.; Vavilov, A.; Vysotsky, V.; Fetisov, S. Development
and production of second generation high Tc superconducting tapes at SuperOx and first tests of model
cables. Supercond. Sci. Technol. 2014, 27, 044022. [CrossRef]
30. Landau, L.; Lifshitz, E. Theoretical Physics. Electrodynamics of Continuous Media, 2nd ed.; Nauka: Moscow,
Russia, 1982.
31. Goldacker, W.; Schlachter, S.; Zimmer, S.; Reineret, H. High transport currents in mechanically reinforced
MgB2 wires. Supercond. Sci. Technol. 2001, 14, 783–786. [CrossRef]
32. Dolya, S. Acceleration of magnetic dipoles by a sequence of current-carrying turns. Tech. Phys. 2014, 59,
1694–1697. [CrossRef]
33. Settles, G.; Hargather, M. A review of recent developments in schlieren and shadowgraph techniques. Meas.
Sci. Technol. 2017, 28, 042001. [CrossRef]
53
applied
sciences
Article
Statistical Analysis and Data Envelopment Analysis
to Improve the Efficiency of Manufacturing Process of
Electrical Conductors
Marco Antonio Zamora-Antuñano 1,† , Jorge Cruz-Salinas 2 , Juvenal Rodríguez-Reséndiz 3, *,† ,
Carlos Alberto González-Gutiérrez 1 , Néstor Méndez-Lozano 1 , Wilfrido Jacobo Paredes-García 3 ,
José Antonio Altamirano-Corro 1 and José Alfredo Gaytán-Díaz 4
1 Engineering Department, Universidad del Valle de Mexico, 76230 Querétaro, Mexico;
marco.zamora@uvmnet.edu (M.A.Z.-A.); calberto.gonzalez@uvmnet.edu (C.A.G.-G.);
nestor.mendez@uvmnet.edu (N.M.-L.); jaaltami@gmail.com (J.A.A.-C.)
2 Centro de Investigación en Ciencia Aplicada y Tecnología Avanzada, IPN, 76090 Querétaro, Mexico;
j.cruz@inteligenciacolectivakm.com
3 Facultad de Ingeniería, Universidad Autónoma de Querétaro, 76010 Querétaro, Mexico;
wparedes17@alumnos.uaq.mx
4 Engineering Department, Universidad Politécnica de Santa Rosa Jáuregui, 76220 Querétaro, Mexico;
agaytan@upsrj.edu.mx
* Correspondence: juvenal@uaq.edu.mx; Tel.: +52-442-192-12-00
† These authors contributed equally to this work.

Received: 22 July 2019; Accepted: 19 September 2019; Published: 21 September 2019
Abstract: The main focus of this research was to develop an approach using statistical tools and Data
envelopment analysis (DEA) to tackling productivity measurements and benchmarking problems in
electrical conductor manufacturing environment. In the present work, a tooling efficiency study was
carried out with a nozzle used for the manufacture of 23-AWG wires. The efficiency of five types of
tooling, four non-Mexican-manufactured types and one Mexican-manufactured type, were compared.
Analysis of Variance (ANOVA) and the Tukey test were applied. Six factors were considered that
influence of the performance of the tooling during the manufacturing process: productivity, quality,
time, machine, operator, and color of the insulating material, but the research work focuses on the
efficiency of the tooling die-nozzle. The results demonstrated that two die-nozzle models exhibited
the best performance; one of them was the Mexican model, surpassed by a non-Mexican model,
the capability process index Cpk = 1.26 manifested a better performance for the 3DND die-nozzle
according to the statistical analysis and the tests performed. Subsequently, through a super-efficiency
DEA model of inputs-oriented with non-decreasing returns to scale (NDRS). The results obtained
in the statistical analysis were corroborated using this technique, its application combined with
statistical tools represents an innovation for knowledge in manufacturing processes of electrical
conductors. Input data were obtained at a manufacturer of electrical conductors supplier of the
automotive sector in the Querétaro City of Mexico.
Keywords: electrical conductor production; statistical analysis; design of experiments;

data envelopment analysis
1. Introduction
ANOVA is a statistical technique widely used to evaluate the importance of one or more factors
in a product or a process when comparing the means of the response variable at the different levels
of the factors. ANOVA analyses require data from populations that follow an approximately normal

Appl. Sci. 2019, 9, 3965
distribution with equal variances between factor levels. The name “analysis of variance” is based
on the approach in which the procedure uses the variances to determine if the means are different.
The procedure works by comparing the variance between the means of the groups and the variance
within the groups as a way to determine if the groups are all part of a larger population or separate
populations with different characteristics. Presently, it is possible to combine other techniques to
give greater effectiveness. Over time, the Design of Experiments (DOE) has been applied with
great success in all types of processes and products, mainly in the automotive sector. DOE is used
as the first approach for optimization of processes, devices, and manufacture of electrical circuits,
pinpointing the factors involved in failures in manufacture processes [1], and identification of elements
in the manufacture process to improve the reliability [2]. Based on [3], through the DOE, we made
improvements in organic processes. In [4], developed the DOE, improved manufacturing tolerances in
permanent magnet motors. Taking into consideration [5], applied the DOE in a spot welding process to
identify the key parameters of the process, which influence the tensile strength of welded joints. Also,
DOE had been employed successfully en R&R analysis [6], surface response analysis [7] and, algorithm
analysis [8]. There have been many cases of successful application of the DOE in all types of production
processes. The motivation of this study lies in the interest of analyzing the efficiency of tooling, real
monitoring, and measurement processes developed with DOE, and identifying the efficiency of the
tooling for the manufacture of electrical conductors. Through a combination of techniques of DOE
and DEA, it is intended to establish an analysis of the performance of the tooling and validate the
efficiency of the die-nozzle 3DND model of Mexican manufacture versus the die-nozzle models of
foreign manufacture. The main problem in the manufacture of electrical conductors is the wear on
the die-nozzle tooling. It generates frequent changes and causes stoppages in the production lines
and implies a high consumption of the die-nozzle. In addition to this, there is an increase in the
indirect costs of the process, losses generated by the non-conforming product, waste of materials and
contamination. Therefore, the problem in this research work was defined in terms of the durability
of the torque die-nozzle because the accelerated wear of this leads to the loss of productivity due to
downtime, waste, and contamination [9]. It was developed in a company that manufactures electrical
conductors for suppliers of the automotive sector, the efficiency of die-nozzle tooling was analyzed.
1.1. Manufacturing 23-AWG Wire Process and Die-Nozzle

The insulator of the electrical conductors are applied using a plastic extrusion machine whose
schematic cut is displayed in Figure 1. A die-nozzle is inserted in the head. The equipment is
complemented with a hopper feeding polymer granules and dryer, which are not shown.
A A
Granulated Resin
Heating Hopper
Mold
Screw
Die
Refrigeration
SECTION A-A
Figure 1. Schematic cutting of the die-nozzle composite head.
56
Appl. Sci. 2019, 9, 3965
In this work, we applied the DOE and DEA to find an optimal solution to the manufacturing
problem specifically in an electrical conductor manufacturing industry. The DEA provides a framework
for ranking efficient units and facilitates comparison with rankings based on parametric methods [10].
1.2. Data Envelopment Analysis

DEA is an operational research technique focused on measuring and analyzing the efficiency with
which goods and services are produced. Its main applications have been in sectors such as finance,
health, services, education or to measure the effectiveness of organizational indicators [11].
DEA is a relatively new data-oriented approach for evaluating the performance of a set of peer
entities called Decision Making Units (DMUs) which convert multiple inputs into multiple outputs.
The definition of a DMU is generic and flexible. Recent years have seen a great variety of applications of
DEA for use in evaluating the performance of different kinds of entities engaged in different activities
in certain contexts in lots of countries [12].
To differentiate and rank the performance of efficient DMU, [13] proposed the concept of
super-efficiency. DEA is capable of transforming a productive situation in which diverse resources
generate multiple products in a single index of efficiency. This index is identified with the value that
maximizes the quotient between the weighted sum of the outputs and the weighted sum of the inputs
of the entity analyzed, to estimate the level of relative efficiency of a company or production unit with
respect to the rest of the units that are evaluated simultaneously, making the best use of resources.
It provides useful information to improve efficiency by providing tools establishing which cases do
not use resources efficiently and which are the recommendations on the use of resources to improve
the efficiency of the organization [14].
Analyzing the performance of a set of homogeneous items is very important in the manufacturing
processes. Parametric and non-parametric methods are proposed for evaluating the performance of
DMUs. DEA is one of the most famous and applicable parametric techniques in this field. It has
many applications for interpreting the productivity of complex systems manufacturing [15]. In this
project, we made an extensive review covering the literature from 2000 to 2019 and several analyses
about DEA. [16] applied the DEA in the design of production lines. In [17], due to advances in data
collection systems and analysis tools, data mining (DM) applied DEA and Quality Improvement
(QI), in manufacturing. In [18], it is demonstrated how the DEA model might be useful to screen
training data so a subset of examples that satisfy the monotony property can be identified. They
used real-world health care and software engineering data, managerial monotony assumption, and
Artificial Neural Network (ANN) as a forecasting model. [19] used DEA and an analytic hierarchy
process (AHP) to measure the performance of ball valve production. In [20], taking into account
non-operational factors, the DEA was proposed to measure the technical efficiency, scale efficiency
and pure technical efficiency of innovation in Semiconductor industry of China. [21] analyzed the
effect of reducing energy savings on the efficiency of innovation and performed the DEA and Tobit
regression analysis. This was in agreement with [22], who developed for the Chinese Government a
model on the allocation of carbon emission reduction quotas through DEA. [23] established improved
manufacturing methods using DEA. In addition, [24] studied recent advances and trends in predictive
manufacturing systems in big data environments using DEA. Table 1 illustrates the 23-AWG wire.
We intend to analyze the performance of a 3DND die-nozzle model of Mexican manufacturing versus
four foreign-made die-nozzle models, and its impact on the variation in the thickness of the insulation
to determine the capability and performance of the process of each die-nozzle model, with the range
of ±0.06 mm for a base dimension of 0.58 mm.
Table 1. Features of 23-AWG wire.
AWG Diameter (mm) Section mm2 Number of Turns/cm kg/km Resistance (Ω/km) Capacity (A)
23 0.58 0.26 16.0 2.29 56.4 0.73
57
Appl. Sci. 2019, 9, 3965
2. Methodology
The project was carried out in a manufacturing industry supplier of the automotive sector, in the
City of Queretaro, Mexico. The name of the company is not mentioned for reasons of confidentiality.
What follows is a census of data was made in the process of extrusion of insulating material in 23-AWG
wire. The results were compared vs. the manufacturing standards of each die-nozzle models. Figure 2
depicts the representation of die-nozzle, one manufactured domestically and four imported. Figure 3
shows the tooling code. The data processing was computed by using Minitab v18 R
[25].
Figure 2. Representation die-nozzle.
x D N Y
Brand of tooling accomanufacture to the origing country
N: Nozzle
D: Die
No consecutive identifer
Figure 3. Tooling code for industrial protection purposes.
The manufacture of the 23-AWG wire is ordered in production lines depending on the color of the
insulation. The reason for this arrangement is due to the optimization of spaces and the simplification
of machine cleaning when a color change of the insulation is made in the manufacturing process [26].
• Line A where the colors black, red, brown and blue are processed.
• Line B with the colors orange, green, yellow and violet.
• Line C that processes the colors white, gray, beige and pink.
There is a total of five work stations in the company, each station is composed of a machine and
peripheral equipment, each line produces only four different colors of insulation, the company works
four shifts, for the present work the operators of the machines were the same during the whole period
of data census. The production of the drivers is subject to market demand, which is the availability of
tools of different die-nozzle models. The difference between each die- nozzle pair is defined in the
diameter and the material, gibr models of die-nozzle were used, then the manufacturing origin of each
model is specified as follows:
• 1DNL: die-nozzle of german manufacturing, diameter: 0.60 mm, material: DIN 420.
58
Appl. Sci. 2019, 9, 3965
• 2DNV: die-nozzle of USA manufacturing, diameter: 0.58 mm, material: ASTM (AISI) 316Ti.
• 3DND: die-nozzle of mexican manufacturing, diameter: 0.59 mm, material: ASTM (AISI) 420.
• 4DNK: die-nozzle of australian manufacturing, diameter: 0.58 mm, material: BSI 316S.
• 5DNS: die-nozzle of japanese manufacturing, diameter: 0.60 mm, material: JIS (SUS 304).
Figure 4 presents the methodology used to apply the DOE and validation with the DEA Model.

0HWKRGRORJ\XVHGWRDSSO\WKH'2(LQWKHPDQXIDFWXUHRI$:*

5HVSRQVH
8VHRIWKH'($PRGHO
3UREOHP )DFWRUV 'DWDDQDO\VLV 5HVXOWVRIDQDO\VLV
YDULDEOH
x 1XPEHURINLORPHWHUV
x 3URGXFWLYLW\ x $129$OLIHWLPHYVGLHQR]]OH SURGXFHGIRUHDFKGLH
(IILFLHQF\RI x 4XDOLW\ x $129$VFUDSPYVGLHQR]]OH QR]]OHPRGHO
HDFKGLH x 7LPH PRGHOV x $PRXQWRIVFUDS
QR]]OHSDLU (IILFLHQF\RI 7RYDOLGDWHHIILFLHQF\RI
x 0DFKLQH x $129$SURGXFWLRQNPYVGLH JHQHUDWHGGLHQR]]OH
HDFK HDFKGLHQR]]OHSDLUV
GXHWRWKH x 2SHUDWRU GLHQR]]OHSDLU
QR]]OHPRGHOV PRGHOV
RULJLQRI x $PRXQWRIGLHQR]]OH
x &RORU RI WKH x $129$SURGXFWLRQYVRSHUDWRU
PDQXIDFWXULQJ
LQVXODWLQJ x $129$GLHQR]]OHYVPDFKLQH PRGHOVXVHG
PDWHULDO x &DSDFLW\RIHDFKGLHQR]]OHPRGHO x 'XUDWLRQRIHDFKGLH
QR]]OHPRGHO
x 3URFHVVFDSDELOLW\

Figure 4. Explain the steps to apply the proposed methodology, data analysis through ANOVA and
validation of results with DEA model.
The population that is the object is constituted by the production data for 23-AWG wire: machines,
operators, shifts, colors, meters of cable produced, meters of scrap and die-nozzle models that are
used for a 23-AWG wire. By convenience sampling, it was determined that the sample would be a
proportion of the production from January to June 2018 for the analysis of: Production volume and
scrap, variable production vs. die-nozzle quantity, analysis of the production, machine and operator,
Key Performance Indicator (KPI) analysis, determination of process capacity indexes, result of analysis
and calculation of sigma level [27].
The data available for this investigation correspond to the production history from January to
June 2018 (see Table 2), which were registered according to the internal procedures of the company
and registered by the production staff directly at the work stations, the data resources correspond to
the electrical conductor manufacturing records.
Table 2. Collected data from January to June 2018.
A. Average life duration in hours per month for each die-nozzle model
Model January February March April May June
1DNL 58 60 57 63 66 80
2DNV 53 48 61 45 56 50
3DND 62 64 56 73 76 69
4DNK 39 46 44 51 37 43
5DNS 49 36 54 40 47 38
B. Average production per unit in meters per month
for each model of die and nozzle
1DNL 719,368 699,942 769,032 783,215 694,142 793,167
2DNV 879,598 759,975 869,432 779,216 865,774 894,302
3DND 689,050 659,920 679,106 746,627 653,511 696,131
4DNK 798,534 789,231 898,535 899,477 775,281 892,387
5DNS 899,506 799,926 877,924 895,685 864,309 894,033
C. Amount of average scrap per unit in meters, generated each month
for each model of die and nozzle, used in the extrusion process
1DNL 7366 7620 7239 8001 8382 101,607
2DNV 9129 8268 10,507 7751 9646 8613
3DND 6402 6608 5782 7537 7847 7124
4DNK 7493 8838 8454 9798 7109 8661
5DNS 10,161 7466 11,198 8295 9747 7880
59
Appl. Sci. 2019, 9, 3965
Due to the manufacture efficiency of the 23-AWG wire can be measured through three
indicators: its lifetime, the amount of wire manufactured, and the amount of wire’s scrap generated.
These indicators were measured during six months for 5 different manufacturing models in order to
determine if there was a change at manufacture efficiency of the 23-AWG. However, manufacturing
model is not the only factor can affect manufacture efficiency of the 23-AWG wire. Several noisy factors
such as: machine, operator, supervisor, shift and, color can alter the manufacturing efficiency [28].
In this way, the following expression can be used to modeling the changes in the
manufacturing efficiency:
E = δModel + δMachine + δOperator + δShi f t + δColor + ε (1)
where: δi is the effect of the factor into the manufacture efficiency and ∈ N (0, σ2 ). Nevertheless, as a
result of quality control of the company the previous model can be reduced as follows:
E = δ + δModel + ε (2)
where: ∈ N (0, σ2 )
with σ < γ. Therefore, the effect of the model can be explored by using a 1-way
ANOVA. In Table 2 is presented the measures of the three indicators for every model during the
6-months period and, Figure 5 shows the data analysis workflow. Table 3 presents number of km
produced by cable vs. the number of die-nozzles/month, used for each production line.
Table 3. Number of km produced by cable vs. the number of die-nozzles/month, used for each
production line.
Production Line A
January February March April May June
Production quantity in km 44,889 40,454 46,736 45,650 43,647 47,956
Number of die-nozzles/month 151 145 152 143 159 138
Production Line B
Production Line C
2.1. DEA Model

The concept of efficiency is related to the economy of resources, efficiency is often defined as
the relationship between the results obtained (outputs) and the resources used (inputs). Numerically,
an efficiency score can be obtained as a relation between output and input, called technical efficiency.
DEA uses linear programming models to determine weights and calculate the efficiency score for each
DMUs [29,30]. The only requirements are that the DMUs convert inputs into outputs, both similar and
quantifiable. Simultaneously, the multiple inputs and outputs are analyzed in their natural physical
units instead of having to convert them to a common denominator [31].
It is assumed that if a die-nozzle DMU, named die − nozzle1 can produce or generate Y1 output
units with X1 input units, then other die-nozzle must also be able to do the same if they are operated
efficiently. Similarly, if die − nozzle2 can produce Y2 output units with X2 input units, then the other
die-nozzle must also be able to do the same. Die − nozzle1 and die − nozzle2 can be combined to
generate a die-nozzle composed of inputs and outputs of them. This virtual die-nozzle is used as a
standard of performance for the die-nozzle.
60
Appl. Sci. 2019, 9, 3965
Equation (3) describes the mathematical model of DEA. It allowed to validate the results obtained
in the analysis of the performance of the five die-nozzle types. See Section 3.7 Computational results
of the DEA model.
Formulation of DEA Model for Die-Nozzle Pairs
min θ ∗ ,
Subject to
n
∑ λ j xij ≤ θ ∗ xio for i = 1, 2, . . . , m;
j=1,j=0
n
λ j yrj ≤ yro for r = 1, 2, . . . , s; (3)
∑
j=1,j=0
λ j ≥ 0 for j = 0;
n
∑ λ j ≥ 1.
j =0
where: DMU0 represents one of the n DMUs under evaluation, and xio and yro are the ith input
and rth output for DMU0 , respectively. To rate the performance of a particular die-nozzle pair, DEA
forms a weighted average of all the observations. The average represents a composite die-nozzle
pair. The weights are to be determined so that the composite die-nozzle pair is located on the frontier.
It represents best practice. The actual die-nozzle pair at hand is then compared with its best practice.
Denoting the weights to be attached to a particular die-nozzle pair by λ j . The inputs and outputs for
this model are indicated in Table 4. In Table 4, the inputs and outputs of the DMUs are determined,
the meaning of these is as follow:
Table 4. DMUs definitions.
Inputs Outputs
Machine Time Productivity Cpk
• Productivity: meters produced with a specific die-nozzle pair/total meters by period (%).
• Machine: numbers of machine-hours used with a specific die-nozzle pair/total of machine
hours/period (%).
• Time: time in hours using a specific die-nozzle pair/total of hours by period (%).
• Cpk: capability process index.
3. Results 23-AWG Production Process

Based on production of 23-AWG, statistical information was given to the information collected for
the development of the proposed methodology. The analysis was carried out from January–June 2018.
Table 5 presents ANOVA duration vs. die-nozzle model. Table 6 presents the the results of the ANOVA
for the three variables: duration, volume of production, and volume of scrap, it can be seen that the
value Ptest is limited to a precision of only 1 thousandth, so any value less than this amount is presented
as 0.000 [32]. In this case, it can be observed that there is a significant difference in the average duration
of at least two die-nozzle models, with a value of significance α = 0.05 [33]. The results established that
the 3DND and 1DNL models are the longest, so they produce more meters of cable than the others.
Table 5. ANOVA: duration vs. die-nozzle model.
Source DF SS MS F P
Model 4 2865.5 716.4 15.18 0.000
Error 25 1179.5 47.2
Total 29 4045.0
61
Appl. Sci. 2019, 9, 3965
Table 6. ANOVA: scrap (m) vs. die-nozzle models.
Source DF SS MS F P
Models 4 19,119,336 4,779,834 4.11 0.011
Error 25 29,095,572 1,163,823
Total 29 48,214,907
3.1. Results of Volume vs. Scrap

In Table 6 Ptest = 0.011 <(α) 0.05, therefore, it can establish that there is a significant difference in
the average duration of at least two die-nozzle models.
Table 7A shows the Tukey test, which allowed identifying that the 1DNL and 3DND models have
the longest duration and that the 3DND model has an average duration of 2.8 ± 0.4 days. It was also
verified that there is a significant difference in the production of cable and scrap between die-nozzle
models. Moreover, Table 7B displays the Tukey test results for each indicator and presents two different
groups for production: A(5DNS, 3DND, 4DNK and 2DNV) and B(1DNL), where model group A has
higher production that model group B. Finally, Table 7C points to three different groups for scrap:
A(5DNS and, 2DNV), AB(4DNK and, 1DNL) and, B(3DND), where model group B presents the lowest
scrap generated and two different groups of models for lifetime: A (3DND and 1DNL) and B(2DNV,
5DNS and, 4DNK), where model group A has higher lifetime than model group B [34].
Table 7. Tukey Test.
A. Tukey Test for Duration

Model N Mean Grouping
3DBD 6 66.67 A
1DBL 6 64 A
2DBV 6 52.17 B
5DBS 6 44 B
4DBK 6 43.33 B
B. Tukey Test Production
Model N Production Grouping
5DNS 6 871,897.2 A
3DND 6 860,724.2 A
4DNK 6 842,240.8 A
2DNV 6 841,382.8 A
1DNL 6 743,144.3 B
C. Tukey Test for Scrap
N Scrap Grouping
5DNS 6 9,124,500 A
2DNV 6 8,985,667 A
4DNK 6 8,392,167 AB
1DNL 6 7,869,167 AB
3DND 6 6,883,333 B
3.2. Results of Data Analysis

The results of the data analysis of the three production lines with the monthly data of the
period under study demonstrates the behavior of the operating variables included in the performance
evaluation affects the die-nozzle pair, illustrates for this case [35]. Table 2A shows that average
life. Table 2B displays the average km of electrical conductor produced with each die-nozzle model.
Table 2C shows the scrap generated [36]. To validate the employed model, graphical analysis was
62
Appl. Sci. 2019, 9, 3965
run. Figure 5 describes the evidence, which is often an assumption, randomness and, homogeneity of
variance were accurate for each ANOVA analysis.
Figure 5. Graphical analysis of residuals of the ANOVA contrast of the model vs. scrap.
In Figures 5 and 6, no significant outliers that affect the normality assumption were detected.
A homogeneous variance is maintained in all die-nozzle models.
Figure 6. Graphical analysis of the lifetime vs. model.
The results of the statistical analysis applied to the variables of die-nozzles against the production,
for each production line, allowed to identify that there is a difference in the production rate between
the lines [37]. This is attributed to a dye effect since the production is divided between the lines
according to the color of the cable. The dye in the polymers can be obtained using pigments which
are inorganic materials, or using dyes which are organic compounds. On the other hand, the tonality
is a function of the chemical compound used. In production Line A, the blue dye has a formulation
of cobalt aluminate (cobalt oxide and aluminum oxide), which is an abrasive compound, and for the
violet which belongs to the group of the production Line B, its formulation is di oxazine (Tetrachloride
1, 4-Benzoquinone), which is a corrosive compound. As can be seen, the composition of the dye
formula affects the mechanical elements differently [38].
The results in Table 8, Ftest = 28.66, Ftable = 2.76, and Ptest = 0.000 < 0.05 = α, show that there is
a significant difference, with a value of significance α = 0.05 in the average duration of at least two
die-nozzle models.
63
Appl. Sci. 2019, 9, 3965
Table 8. ANOVA:Production volume vs. die-nozzle models.
Source DF SS MS F P
Model 4 9.88E10 2.47E10 28.66 0
Error 25 2.15E10 8.62E8
Total 29 1.20E11
3.3. Results of Analysis Machine, Production and Operator

Finally, in this case, a behavior similar to that of the Production Lines A and B are observed.
The order of origin is not significant. Then, it can be excluded from the model. The coefficients of
determination are close to 100%. So, it can be concluded that the production in Line C increases at a
rate of 416 ± 28 km per die-nozzle unit. Similarly, the regression analysis is performed for the scrap
variable, the results are presented in computational results of the DEA. Figures 5 and 6 are presented
these results, which were compared with the proposed methods in [39,40]. Table 9 exemplified the
results of ANOVA for machine and operator.
Table 9. Results of ANOVA analysis for the machine and operator variables.
Parameters through ANOVA.

Variable Ftest Ftest (table value) Ptest α
Machine 7.39 3.10 0.002 0.05
Operator 129.74 3.10 0.000 0.05
In Table 10, it is observed in the results Ftest = 129.74 > FTable = 3.10, therefore, in terms of the
value Ptest shows that Ptest = 0.000 <(α) 0.05, there is a significant difference that at least one operator
produces more units/period.
Table 10. ANOVA: production vs. operator.
Source DF SS MS F P
Operator 3 22821.2 7607.1 129.74 0
Error 20 1172.6 58.6
Total 23 23993.8
3.4. Key Performance Indicator (KPI) Analysis

The performance indicators were calculated using the data of Table 3B. Table 11 presents the
concentration of results the performance indicators for each die-nozzle models.
Table 11. Concentration of results of the performance indicators for each die-nozzle models.
Performance Indicators
Model
Productivity Quality Machine Operator Time Color Results
3DND 0.20 0.13 0.09 0.09 0.27 0.04 81.2%
1DNL 0.12 0.08 0.10 0.10 0.13 0.05 58.7%
2DNV 0.22 0.13 0.08 0.08 0.26 0.04 80.6%
5DNS 0.12 0.08 0.10 0.10 0.14 0.05 59.7%
4DNK 0.10 0.07 0.10 0.10 0.13 0.05 55.1%
3.5. Determination of Process Capacity Indexes

The significant values of this analysis are Cp = 1.59, Pp = 1.09, Cpk = 1.26, and PPM = 4726.1
(parts/million) of allowed defects. In Table 12, the results are presented in die-nozzle models, the best
results were from the model 3DND the die-nozzle of Mexican manufacturing [41].
64
Appl. Sci. 2019, 9, 3965
Table 12. Summary of the results of the processing capacity of each die-nozzle model.
Parameters of Process Capability

Model
Standard Deviation Cp Cpk PPM PPM General Pvalue
1DNL 0.025 0.80 0.73 18,931 18,484 0.431
2DNV 0.023 0.86 0.62 32,184 16,086 0.086
3DND 0.001 1.59 1.26 17,540 4726 0.255
4DNK 0.037 0.55 0.52 101,853 69,866 0.736
5DNS 0.025 0.78 0.55 50,796 103,305 0.625
It is essential to emphasize that the natural variability of the process, 6σ, is intrinsic to it and
independent of the tolerances assigned. Therefore, if 6σ is less than the range of tolerances to be met,
some manufactured products are out of tolerance and non-compliant. If this fact is not taken into
account and it is intended to correct based on the readjustment of the process, i.e., modify the centering,
the only thing that is achieved is to increase its variability. It can be thought that the process is out of
control or unstable in the die of non-Mexican manufacturing because of the variations are caused by
somewhat unpredictable behavior. In Tables 13 and 14, it is observed that the 3DND model that shows
the production process is control Cpk > 1, the other models presents variability, Cpk < 1, that does
not necessarily imply obtaining products outside of specification. The 3DND model process control
allows establishing that the objectives of the project are being met through regular monitoring and
measurement of its progress to identify planned variations [42].
Table 13. Summary of sigma level results for each model.
Models Level Z PPM PPM General Cpk

1DNL 2.08 18,931 18,484 0.73
2DNV 1.85 32,184 16,086 0.62
3DND 3.79 17,540 4726 1.26
4DNK 1.27 101,853 69,866 0.52
5DNS 1.64 50,796.14 103,304.56 0.55
Table 14. DEA model dataset.
Inputs Outputs
Models Machine Time (s) Productivity Cpk
3DND 0.09 0.27 0.20 1.26
1DNL 0.01 0.13 0.12 0.73
2DNV 0.08 0.26 0.22 0.62
5DNS 0.1 0.14 0.12 0.55
4DNK 0.1 0.13 0.10 0.52
3.6. Result of Analysis and Calculation of Sigma Level

Sigma level indexes of the process are calculated using dimensional data from the die-nozzle.
Table 13 indicates the results of the Sigma level analysis for each model die-nozzle. With the analysis
of the level of Sigma, the significant results are Z (level of Sigma), PPM and Cpk [36].
It is observed that the 1DNL and 3DND models have the highest levels of sigma, of 2.08 and
3.79 respectively. For these die-nozzle models, the Cpk indexes were 0.73 and 1.26. Those of Cp were
1.59 and 0.80, and the Pp were 1.09 and 0.80 respectively, so these die-nozzle models are the closest
ones to the reference established by the norm. For that reason, they are the best capacity and ability
of the process, in that order. Models 2DNV, 4DNK, and 5DNS resulted in levels lower than what is
specified by the Standard for Electrical Products-Conductors-Wires and cords with PVC insulation
105 ◦ C, for Electronic Uses-Specifications (NMX-J-429-ANCE-1994) [19].
65
Appl. Sci. 2019, 9, 3965
3.7. Computational Results of the DEA Model

To carry out the calculations, the DEA FrontierTM software was used, which is an Excel.
DEA FrontierTM uses the Excel solver as the engine to solve DEA models and was used to solve
the DEA super efficiency-oriented model to inputs with NDRS [29,30].
4. Discussion and Conclusions
4.1. Discussion
Based on the results obtained by statistical analysis, it can be mentioned that the manufacturing
process of 23-AWG wire presented the best performance in terms of the defined variables and results
obtained through the DOE. The Cpk of 1.26 allowed to verify that the process with the 3DND model
is the most stable compared to the models of this foreign-made die. As for the scrap it generates,
the 3DND model was the one with the best performance, as well as the average hours of use, and the
second-best performance in terms of total km produced. Being in this last point the model 2DNV
the one of better performance. It is proven that through the application of the DOE and process
optimization significant results were achieved in continuous improvement for any product or process.
In [37] applied the SPC, DOE and Six Sigma in the manufacture of electrical circuits achieving a
significant decrease in manufacturing defects. Also many success stories in different scenarios or
organizations used the application of statistical tools and techniques improvements are established.
Successful cases: In accordance with [6], proposed rapid modeling and optimization of manufacturing
processes based on DOE. In [7] is described how processes can be optimized efficiently and how DOE
findings may be applied to scale-up. Others [8], used the DOE to improve the manufacturing results in
engines. In [9], through DOE it was able to establish critical parameters associated with the processes
an indefinite number of success stories in the application of the DOE. The JIT, such as philosophy
developed at Toyota and its different methods and systems have served as an example for many
organizations to establish continuous improvement, this is a reference for any. In [38], established the
methodology allowed the identification of variables that influenced the risk. Davim mentioned that
the approach is based on a combination of Techniques of Taguchi and ANOVA [36]. The Statistical
Process Control (SPC) is a solution developed to easily collect and analyze data, allowing performance
monitoring as well as achieving sustainable improvements in quality which in turn allows increasing
the profitability [35]. Based on [36] has proposed a new paradigm in the application of different
techniques and improvement tools that imply: Quality, Cost, and Delivery (QCD), and this is to
combine techniques and tools already tested as SPC and DOE, with other models in our case the DEA.
For [29,30], DEA it is an extremely powerful tool that can assist decision-makers in benchmarking
and analyzing complex operational performance issues in manufacturing organizations as well as
evaluating processes in banking, retail, franchising, health care, public services, and many other
industries. Rau through a combination of SPC and DEA, optimal solutions were introduced in different
processes [37].
4.2. Conclusions
ANOVA, with a level of significance α = 0.05, was performed on the variables: durability,
production, and scrap. These analyses demonstrated that there is a significant difference in the average
duration between die-nozzle models. The Tukey test identified that the 1DNL and 3DND models have
greater durability. The 3DND model has average durability of 2.8 ± 0.4 days. It was also verified that
there is a significant difference between cable and scrap production of all die-nozzle models [29,30].
With the results of the statistical analysis previously described, it is justified that the variables:
durability, production, scrap, the color of the insulator, machine, and operator, are considered relevant
for the calculation of the performance indicators due to the influence of the die-nozzle pairs, the 3DND
model is the best performer in terms of the generated scrap [38,39].
66
Appl. Sci. 2019, 9, 3965
Therefore, the results demonstrated that there is an effect of the model of die-nozzle on the
variables. The performance is calculated individually with the method of the cable manufacturer,
where the result indicates that the models 1DNL = 81.2% and 3DND = 80.2%, are those with the highest
performance indicators [40].
The second phase was carried out applying formal tools of Total Quality Management:
Benchmarking, Statistical Process Control and DOE. It was obtained that the 3DND and 1DNL models
have a σ level of 3.79 and 2.08. Their indicators Cpk = 1.26, 0.73, Cp = 1.59, 0.80, and Pp = 1.09, 0.80
respectively, so these die-nozzle models are the most approach the references established by the norm,
for that reason, they are the best in the capability of the process, in that order [41].
The indicators of performance under the criterion of the analysis of the company evaluate
in percentages the factors considering a higher weight to three of the six: productivity, quality
and time, obtaining like the result the following order of performance of the die-nozzle models
1DND, 4DNK, 2DNV, and 5DNS. The results of the Statistical Process Control and Six Sigma are
based on the evaluation of the variability of the measurements of the finished product and the
scrap volume generated, obtaining, as a result, the following order: 3DND, 1DNL, 2DNV, 4DNK,
and 5DNS, which It is different because different criteria are used. However, it is observed that the
models with the best performance are those that have the highest indicators of processes capability.
Therefore, the Manufacturing Company could be used the method of Statistical Control and DOE to
evaluate the performance of the die-nozzle pairs [42]. Other authors have applied different statistical
techniques and have proposed a routine for evaluated the failure rate and inspection intervals using
the first-order reliability method (FORM + Fracture) to alleviate the computational cost of probabilistic
defect-propagation analysis. The proposed method is one of the first applying reliability concepts
to additive manufacturing (AM) components [1]. It is proven that significant improvements can be
generated through the systematic application of the TQM process [2]. With the use and application
of statistical analysis, reductions in manufacturing, inspection, materials or methods costs can be
achieved [43]. The concept of additive manufacturing [7], will be very useful for the optimization of
processes in all types of components.
With the results obtained using the DEA model of Super Efficiency oriented to inputs with
non-decreasing returns to scale, the same conclusions were obtained as those obtained through
Statistical Process Control and DOE, so that the order of preference of the die-nozzle models is as
follows: 3DND, 1DNL, 2DNV, 4DNK, and 5DNS [32,33]. The methods to optimize the design proposed
in [1] were very useful for following work in the improvement of tooling and manufacturing processes.
By an early control such as that proposed in [2], substantial improvements in the manufacture of
electrical conductors and other products can be achieved. In the present project, it is very useful to
compare the use of ANOVA in [3,4], to improve the proposed method of analysis. Through optimizing
other variables such as those mentioned in [5,6] it is possible to generate substantial improvements
that will be taken into account in future work in the optimization of manufacturing processes. It can
be concluded that DEA model is a method that can be of great help, and although it will not replace
the DOE, if it allows manufacturing companies to perform analysis and process improvements in a
faster way.
In the informed literature, there are no application cases to measure the efficiency of the production
lines in the manufacturing sector of electrical conductors [28]. With this study, the objective established
in the present work was fulfilled.
Author Contributions: Conceptualization, M.A.Z.-A. and J.R.-R.; methodology, J.C.-S.; software, J.A.A.-C.;
validation, C.A.G.-G., N.M.-L., and J.R.-R.; formal analysis, J.A.A.-C., J.C.-S., and M.A.Z.-A; investigation and
visualization, J.A.G.-D., J.A.A.-C. and J.R.-R.; data curation, J.R.G.-M., W.J.P.-G., and M.A.Z.-A.; writing—original
draft preparation; writing—original draft, review, and editing, all the authors.
Funding: This research was partial funded by CONACYT and PRODEP.
67
Appl. Sci. 2019, 9, 3965
Acknowledgments: The authors thank CONACYT for the academic scholarship for CVU No 573233. We appreciate
the support of Centro de Investigación en Ciencia Aplicada y Tecnología Avanzada del Instituto Politécnico
Nacional (CICATA), Unidad Querétaro. Especially for Master Jorge Cruz Salinas, who was the project leader and
main author of the design of the method of validation of die-nozzle tooling, that allowed the project development,
and for his dedication to obtaining a Master’s Degree in Advanced Technology. We also thank National Research
Vice Rectory of UVM-Mexico, for the support provided. The authors recognize in a special way PhD. Ivan
Domínguez López and Adrián Luis García García, for the advice provided during the development of the project.
The authors appreciate Master Rodrigo Pinto Castillo and César Javier Ortíz Echeverri for his support during the
revision of the project.
References
1. Ambühl, S.; Kramer, M.; Sørensen, J. Reliability-based structural optimization of wave energy converters.
Energies 2014, 7, 8178–8200. [CrossRef]
2. Mejia-Parra, D.; Sánchez, J.R.; Ruiz-Salguero, O.; Alonso, M.; Izaguirre, A.; Gil, E.; Palomar, J.; Posada, J.
In-Line Dimensional Inspection of Warm-Die Forged Revolution Workpieces Using 3D Mesh Reconstruction.
3. Nguyen, H.T.; Hsu, Q.C. Surface roughness analysis in the hard milling of JIS SKD61 alloy steel. Appl. Sci.
2016, 6, 172. [CrossRef]
4. Costa, A.; Cappadonna, F.; Fichera, S. A hybrid metaheuristic approach for minimizing the total flow time in
a flow shop sequence dependent group scheduling problem. Algorithms 2014, 7, 376–396. [CrossRef]
5. Ambühl, S.; Kramer, M.; Dalsgaard Sørensen, J. Structural reliability of plain bearings for wave energy
converter applications. Energies 2016, 9, 118. [CrossRef]
6. Arbogast, G.W. A case study: Statistical analysis in a production quality improvement project. J. Qual. Manag.
1997, 2, 267–277. [CrossRef]
7. Coro, A.; Macareno, L.M.; Aguirrebeitia, J.; López de Lacalle, L.N. A Methodology to Evaluate the Reliability
Impact of the Replacement of Welded Components by Additive Manufacturing Spare Parts. Metals 2019, 9,
932. [CrossRef]
8. Umeda, S.; Nakano, M.; Mizuyama, H.; Hibino, N.; Kiritsis, D.; Von Cieminski, G. Advances in Production
Management Systems: Innovative Production Management towards Sustainable Growth: IFIP WG 5.7 International
Conference, APMS 2015; Springer: Berlin/Heidelberg, Germany, 2015; Volume 459, pp. 7–9. [CrossRef]
9. Boning, D.S.; Mozumder, P. DOE/Opt: A system for design of experiments, response surface modeling, and
optimization using process and device simulation. IEEE Trans. Semicond. Manuf. 1994, 7, 233–244. [CrossRef]
10. Weissman, S.A.; Anderson, N.G. Design of experiments (DoE) and process optimization. A review of recent
publications. Org. Process Res. Dev. 2014, 19, 1605–1633. [CrossRef]
11. Khan, M.A.; Husain, I.; Islam, M.R.; Klass, J.T. Design of experiments to address manufacturing
tolerances and process variations influencing cogging torque and back EMF in the mass production of
the permanent-magnet synchronous motors. IEEE Trans. Ind. Appl. 2013, 50, 346–355. [CrossRef]
12. Rowlands, H.; Antony, J. Application of design of experiments to a spot welding process. Assem. Autom.
2003, 23, 273–279. [CrossRef]
13. Andersen, P.; Petersen, N.C. A procedure for ranking efficient units in data envelopment analysis. Manag. Sci.
1993, 39, 1261–1264. [CrossRef]
14. Khezrimotlagh, D.; Chen, Y. Decision Making and Performance Evaluation Using Data Envelopment Analysis;
Springer: Berlin/Heidelberg, Germany, 2018; Volume 269, pp. 217–234. [CrossRef]
15. Koksal, G.; Batmaz, İ.; Testik, M.C. A review of data mining applications for quality improvement in
manufacturing industry. Expert Syst. Appl. 2011, 38, 13448–13467. [CrossRef]
16. Pendharkar, P.C. A data envelopment analysis-based approach for data preprocessing. IEEE Trans. Knowl.
Data Eng. 2005, 17, 1379–1388. [CrossRef]
17. Naghiha, S.; Maddahi, R.; Ebrahimnejad, A. An integrated AHP-DEA methodology for evaluation and
ranking of production methods in industrial environments. Int. J. Ind. Syst. Eng. 2019, 31, 343–359.
[CrossRef]
68
Appl. Sci. 2019, 9, 3965
18. Li, H.; He, H.; Shan, J.; Cai, J. Innovation efficiency of semiconductor industry in China: A new framework
based on generalized three-stage DEA analysis. Socio-Econ. Plan. Sci. 2019, 66, 136–148. [CrossRef]
19. Shin, J.; Kim, C.; Yang, H. Does Reduction of Material and Energy Consumption Affect to Innovation
Efficiency? The Case of Manufacturing Industry in South Korea. Energies 2019, 12, 1178. [CrossRef]
20. Emrouznejad, A.; Yang, G.L.; Amin, G.R. A novel inverse DEA model with application to allocate the
CO2 emissions quota to different regions in Chinese manufacturing industries. J. Oper. Res. Soc. 2019,
70, 1079–1090. [CrossRef]
21. Kamarudin, N.F.; Rani, R.M.; Halim, F.A. Improving the Food Manufacturing System by Using Simulation
and DEA. In Proceedings of the Third International Conference on Computing, Mathematics and Statistics
(iCMS2017); Springer: Singapore, 2019; pp. 555–561. [CrossRef]
22. Liu, G.; Zhou, Z.; Qian, X.; Wu, X.; Pang, W. Multidisciplinary design optimization of a swash-plate axial
piston pump. Appl. Sci. 2016, 6, 399. [CrossRef]
23. Ammons, D.N.; Roenigk, D.J. Benchmarking and interorganizational learning in local government. J. Public
Adm. Res. Theory 2014, 25, 309–335. [CrossRef]
24. Thue, W.A. Electrical Power Cable Engineering; CRC Press: Boca Raton, 2012.
25. MINITAB. Computer Program Manual; Software; MINITAB: State College, PA, USA, 2014.
26. Yang, Y.; Cao, J. Solving quadratic programming problems by delayed projection neural network. IEEE Trans.
Neural Netw. 2006, 17, 1630–1634. [CrossRef] [PubMed]
27. Tai, Y.T.; Pearn, W.L.; Kao, C.-M. Measuring the Manufacturing Yield for Processes with Multiple
Manufacturing Lines. IEEE Trans. Semicond. Manuf. 2012, 25, 284–290. [CrossRef]
28. Xiao, R.; Liu, B.; Shen, J.; Guo, N.; Yan, W.; Chen, Z. Comparisons of energy management methods for
a parallel plug-in hybrid electric vehicle between the convex optimization and dynamic programming.
29. Zhu, J. Super-efficiency and DEA sensitivity analysis. Eur. J. Oper. Res. 2001, 129, 443–455. [CrossRef]
30. Zhu, J. Quantitative models for performance evaluation and benchmarking: data envelopment analysis with
spreadsheets. Int. Ser. Oper. Res. Manag. Sci. 2014, 213, 175–205. [CrossRef]
31. Andreadou, N.; Lucas, A.; Tarantola, S.; Poursanidis, I. Design of Experiments in the Methodology for
Interoperability Testing: Evaluating AMI Message Exchange. Appl. Sci. 2019, 9, 1221. [CrossRef]
32. Han, C.; Sin, I.; Kwon, H.; Park, S. The Role of the Process and Design Variables in Improving the Performance
of Heat Exchanger Tube Expansion. Appl. Sci. 2018, 8, 756. [CrossRef]
33. Lou, P.; Yuan, L.; Hu, J.; Yan, J.; Fu, J. A Comprehensive Assessment Approach to Evaluate the
Trustworthiness of Manufacturing Services in Cloud Manufacturing Environment. IEEE Access 2018,
6, 30819–30828. [CrossRef]
34. Yaste, R.E.; Budrow, R.B. Void-Free Electrical Conductor for Power Cables and Process for Making Same.
US Patent 4,319,074, 1982.
35. Realyvásquez-Vargas, A.; Arredondo-Soto, K.; Carrillo-Gutiérrez, T.; Ravelo, G. Applying the
Plan-Do-Check-Act (PDCA) Cycle to Reduce the Defects in the Manufacturing Industry. A Case Study.
36. Dyck, T.; Bund, A. Influence of the Bead Geometry and the Tin Layer on the Contact Resistance of Copper
Conductors. IEEE Trans. Compon. Packag. Manuf. Technol. 2018, 8, 1863–1868. [CrossRef]
37. Kilickap, E.; Yardimeden, A.; Çelik, Y.H. Mathematical Modelling and Optimization of Cutting Force, Tool
Wear and Surface Roughness by Using Artificial Neural Network and Response Surface Methodology in
Milling of Ti-6242S. Appl. Sci. 2017, 7, 1064. [CrossRef]
38. Sorensen, D.G.H.; Brunoe, T.D.; Nielsen, K. A classification scheme for the processes of the production
system. Procedia CIRP 2018, 72, 609–614. [CrossRef]
39. Yang, C.M.; Lin, K.P.; Chen, K.S. Confidence Interval Based Fuzzy Evaluation Model for an Integrated-Circuit
Packaging Molding Process. Appl. Sci. 2019, 9, 2623. [CrossRef]
40. Godina, R.; Pimentel, C.; Silva, F.; Matias, J.C. Improvement of the Statistical Process Control Certainty in an
Automotive Manufacturing Unit. Procedia Manuf. 2018, 17, 729–736. [CrossRef]
41. Amasaka, K. Applying New JIT, a management technology strategy model at Toyota-Strategic QCD studies
with affiliated and non-affiliated suppliers. Int. J. Prod. Econ. 2002, 80, 135–144. [CrossRef]
69
Appl. Sci. 2019, 9, 3965
42. Rau, H.; Wu, C.H.; Shiang, W.J.; Huang, P.T. A decision support system of statistical process control for
printed circuit boards manufacturing. In Proceedings of the 2010 International Conference on Machine
Learning and Cybernetics, Qingdao, China, 11–14 July 2010; Volume 5, pp. 2454–2458. [CrossRef]
43. Coro, A.; Abasolo, M.; Aguirrebeitia, J.; de Lacalle, L.L. Inspection scheduling based on reliability updating
of gas turbine welded structures. Adv. Mech. Eng. 2019, 11. [CrossRef]
c 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
70
applied
sciences
Article
Methodology for Searching Representative Elements
Jure Murovec, Janez Kušar and Tomaž Berlec *
Faculty of Mechanical Engineering, University of Ljubljana, 1000 Ljubljana, Slovenia
* Correspondence: tomaz.berlec@fs.uni-lj.si

Received: 25 July 2019; Accepted: 20 August 2019; Published: 23 August 2019
Featured Application: The MIRE method was developed for a specific and practical case, where
there is essentially minimal information about the company’s material flow. It is suitable for
smaller companies as it enables rapid analysis of a large quantity of mutually similar data. MIRE
can relatively quickly give us a sufficient approximation of the material flow, which serves as
the basis and the first step towards lean production, consequently increasing the company’s
competitiveness in the global market.
Abstract: Companies have to assure their share on the global market, meet customer demands and
produce customer-tailored products. With time and production line updates, the layout becomes
non-optimal and product diversity only increases this problem. To stay competitive, they need to
increase their productivity and eliminate waste. Due to a variety of products consisting of similar
components and variants thereof, a huge number of various elements are encountered in a production
process, the material flow of which is hardly manageable. Although the elements differ from each
other, their representative elements can be defined. This paper will illustrate a methodology for
searching representative elements (MIRE), which is a combination of the known Pareto’s analysis
(also known as ABC analysis or 20/80 rule) and a calculation of a loading function, that can be
based on any element feature. Results of using the MIRE methodology in a case from an industrial
environment have shown that the analysis can be carried out within a very short time and this
provides for permanent analysis, optimisation and, consequently, permanent improvement in the
material flow through a production process. The methodology is most suitable for smaller companies
as it enables rapid analysis, especially in cases when there is no pre-recorded material flow.
Keywords: representative elements; large data quantity; loading function; simulation; material
flow optimisation
1. Introduction
Lean production as a production strategy is one of important paradigms of the 21st century which
has been introduced in various variants practically in all companies in the world. A lean environment
can minimise waste by elimination, merging or simplification of technologically unnecessary operations
(stock, transport) and unnecessary elements of the work process (auxiliary and extra activities), that
do not give any added value. Analyses [1] have shown that merely 20% of activities in a production
process directly add added value, 20% of activities do not directly give any added value, but are
needed in order to carry out activities that add added value, while the rest (60%) are the activities
which do not add added value and are, therefore, indicated as waste.
If a technological operation is observed as a fundamental element of a lead time for the production
of a component, it is determined that the lead time of an operation consists of: (1) transition time
calculated as completion of the preceding operation until the beginning of performance of the observed
operation; it represents as much as 90% of the time in the structure of the lead time of an operation, and
(2) operation performance time which represents only a 10% share in the structure of the lead time [2].

Appl. Sci. 2019, 9, 3482
Based on the above, the transition time is by far the most important in the structure of the lead
time of an operation and it consists of a waiting time after a preceding operation, a transport time, and
a waiting time before the observed operation [3].
Each production-based company has a certain arrangement of sectors and workplaces, which can
vary in its optimal layout. A great potential in terms of reducing losses and consequently in raising
competitiveness can be hidden in a sub-optimal layout. It is common that smaller companies add
newly bought machinery to places where space was sufficient, despite the sub-optimal layout.
This finding calls for a need to reorganise the material flow within a production system in a way
to have minimum [4] transition time between operations. This has a direct impact on the cost of a
company’s internal transport [5]. Francis, McGinnis and White [6] maintain, that 20% to 50% of direct
costs can be attributed to the costs of material manipulation [7], wherein innovative organisational
concepts have a decisive impact [8].
The material flow can be defined as a structured and organised movement of a material from
point A to point B through a production system in compliance with a prescribed technological method
under consideration of efficient use of space, savings in energy and human resources [5]. Material
flow is a complex process of each production system and can be optimised without adequate tools
for planning a material flow. Before optimising a material flow, the inventory must first be taken
(the current situation assessed). A material flow usually consists of a huge number of elements (each
with its own technological method) that are not identical but are similar in certain criteria. This is why
families need to be formed on the basis of selection of representative elements. The quantity of data to
be inventoried and analysed can thus be considerably reduced.
Bhosale and Pawar [9] have modified a mathematical model to optimise material flow optimisation
of a flexible manufacturing system with a real coded genetic algorithm. In their article [10], the authors
showed application of a statistical analysis and use of a group technology for the introduction of lean
concepts into production systems in order to reach better leanness and consequently effectiveness
of production systems. The group technology is also used for the analysis of a material flow in the
process of converting a company having mass production and shop-floor configuration of workplaces
into a cellular organisation of workplaces [11].
Due to a diversity of elements the conventional ABC analysis and the multi-criteria ABC analysis,
such as [12,13], are no longer sufficient in our case to inventory the material flow.
Based on a literature review a new model called the methodology for searching representative
elements (MIRE) was designed which assists us in processing huge quantities of data.
The result of the method is an approximation of the real situation, while deviations are checked
by feedback. All five steps of the MIRE are repeated as necessary, until a satisfactory ratio between the
approximation accuracy and time savings is obtained in the processing of data.
2. Methods
2.1. Literature Review

Group technology (GT) is an approach of production management, in which part families are
formed from a plurality of various parts that are manufactured in a treated production system and that
use the same group of machines to be transformed from an input material into the required final shape
and property of the product.
The methods designed to search families of parts and the group of machines they belong to
are treated in the literature as cell formation techniques. Based on a proposal of various authors
Selim et al. [14] proposes the following overview of the cell formation technique:
(1) Descriptive procedures (part family identification, machine group identification, part
families/machine grouping simultaneously);
(2) Cluster analysis;
(3) Graph partitioning;
72
Appl. Sci. 2019, 9, 3482
(4) Artificial intelligence;

(5) Mathematical programming (linear programming, linear and quadratic integer programming,
dynamic programming, goal programming).
Descriptive procedures part families/machine grouping were first represented by Burbidge as a

Production flow analysis [15]. Dekleva et al. [16] extended the Burbridge’s production flow analysis
and called it the extended production flow analysis. The extended material flow analysis included
two steps:
(1) Macroanalysis is analysing the material flow (flow of parts) on a macro level, i.e., between
departments of a company. The goals of these analysis were: identification of extra paths between
the departments of the company and their elimination (usually by changing a technological
method) and determination of a standard material flow between the departments of the company.
(2) Microanalysis is included in a material flow analysis within an individual department. The goals
of this analysis were: simultaneous identification of part families and groups of needed machines
for their manufacturing and formation of production cells for the part families. A cluster analysis
method and an incidence matrix method were proposed for the formation of cells.
Nouri et al. [17] propose, in particular, metaheuristic techniques such as: genetic algorithm,
artificial neural networks, fuzzy set theory, colony systems, tabu search, branch and bound method,
simulated annealing.
Nowadays, the goals of group technology and formation of various shapes of production systems
(cells) need to be connected, especially with the goals of leanness, flexibility and agility of production.
Due to market demands, a large number of different, yet mutually similar parts appear in the production
process. That means that the production program contains a large number of different products that
are similar to each other, as the companies emphasizes the possibility of making a tailored products
for each customer to maintain its competitiveness in global markets. It is quite common in smaller
companies that the record of material flow has never been made. They simply lack the time and the
human resources, while trying to maintain acceptable production and make profit. With time and
production line updates, the layout becomes increasingly non-optimal and the diversity of products
only increases this problem.
So, it is not reasonable to analyse a material flow and a value flow on the entire amount of
available data. Within thousands of pieces, representative parts (hereinafter representative elements)
need to be selected. This is made possible by the methodology for searching representative elements,
MIRE, which can relatively quickly give us a sufficient approximation of the material flow, which
serves as the basis and the first step towards lean production, consequently increasing the company’s
competitiveness in the global market.
The methodology is most suitable for smaller companies that throughout their growth and
development simply did not have the time, resources and competences to create a material flow
inventory because their priorities were mainly focused on delivering the products and keeping up to
date with the latest technological trends.
2.2. Methodology for Searching Representative Elements (MIRE)

Basic data for the implementation of the MIRE can differ, yet they usually refer to one selected
influencing variable, e.g., mass, dimension, volume, size, material value, production time, added value.
It is important that a selected variable is used throughout the methodology from its very beginning.
The MIRE methodology follows five steps represented in Figure 1. First, a time interval for
data gathering and a creating criterion(s) are selected, which are then considered through the entire
analysis process.
73
Appl. Sci. 2019, 9, 3482
Figure 1. The methodology for searching representative elements (MIRE) circular flow diagram.
2.2.1. Step 1: Quantities

In a first step, an influential variable is selected that will serve as the basis for searching
representative elements.
A production of n number of products is anticipated for the selected time interval, each of them
having value x.
The loading function is defined as a product of the variable of one product with their own quantity
and functions as a weight function (Figure 2). The total load in the selected time interval is:

n
n
OI = OI,i = (xi ·yi ) (1)
i=1 i=1
OI —total load
74
Appl. Sci. 2019, 9, 3482
n—number of products
xi —value of the selected variable of the ith product
yi —quantity of the ith product (always 1 before formation of families)
Figure 2. Total load.
2.2.2. Step 2: Formation of Families

The products having a similar sequence of technological operations are grouped into families
(Figure 3), where the selected variable is not yet taken into consideration. The larger the number of
families, the more precisely the material flow will be inventoried.
Gl —indication of family l
K—number of families; where: K ≤ n
p—number of products in a family
Figure 3. Formation of product families.
2.2.3. Step 3: Selection of a Representative Product of the Family

The family joins products with different values x, which have an identical or similar sequence
of technological operations. The variables selected with an identical or similar value are grouped
(Figure 4). The accuracy of grouping impacts the accuracy of material flow inventory. The more groups
that are formed, the more accurate the inventory will be.
So, j groups are obtained and each group has a load OI,i .
OI,i = xI,i ·yI,i (2)
Ol,i —load of the ith group of products having an identical or similar value of the selected variable
within family l
xl,i —value of the selected variable of a product in the ith group of family l
yl,i —number of products in the ith group of family l
75
Appl. Sci. 2019, 9, 3482
j
OI OI,i
xI = = ip=1 (3)
yI yI,i i=1
xl —average value of the selected variable of a product of family l

Ol —total load of family l
yl —number of all products of family l
Figure 4. Grouping within a family.
The obtained xl is used to search for such a product within family l, to which the value of variable
x gets closest. In this way a representative product of family l, Rl , is selected.
The representative product should be assigned a path through the production and a sequence
of treatment processes. This path should be written in a square FROM–TO matrix Pa×a , where a is
the number of workplaces. Column or row a should represent the last workplace in the production
(assembly, warehouse, takeover, etc.).
⎡ ⎤
⎢⎢ I11 ··· I1a ⎥⎥
⎢⎢ ⎥⎥
⎢ ⎥⎥
Pa×a = ⎢⎢⎢ ... ..
.
..
. ⎥⎥ (4)
⎢⎢ ⎥⎥
⎣ ⎦
Ia1 ··· Iaa
The representative product of the Ith family thus has the value of the selected variable xR,l and
path P.
RI = xR,I ·P; xR,I ≈ xI (5)
2.2.4. Step 4: Material Flow

The calculation of representative products of individual families allows us to define their
material flow:
MR,I = yI ·RI = yI ·xR,I ·P (6)
The total material flow of all families in this case is:
MR = MR,A + MR,B + MR,C + · · · + MR,K (7)
2.2.5. Step 5: Checking Material Flow Approximation Accuracy

All representative products have been obtained and used to calculate an approximation of the
total material flow. Since its accuracy is of interest, the equation below is used:
a
i=1 mR(i,a)
=α (8)
O
mR(i,a) —ith element of the ath column in matrix MR .
The closest α comes to value 1, the more accurate the material flow MR inventory is.
I f α = 1, then MR = M (9)
76
Appl. Sci. 2019, 9, 3482
M—actual material flow.

The actual material flow or 100% accuracy of inventory of material flow MR is obtained with the
proviso that the number of families equals the number of all products in the selected time interval
(K = n). In such a case, exactly one product is present within each family and this automatically means:
xR,I = xI and j=1 (10)
As there are no deviations in the values of the selected variable and FROM–TO matrices of
representative products, the material flow is inventoried with 100% accuracy.
The essence of this methodology to reduce the quantity of data to be processed in order to obtain
a satisfactory material flow inventory. Therefore it is not reasonable, that K = n.
To obtain credible results consistency is needed:
• In classifying the products into families based on a criterion of sequence of treatment processes;
• In grouping the products within families based on the criterion (selected variable); and
• In selecting a representative product based on xI .
As described, MIRE helps us process large amounts of data and it is especially useful for cases
where the elements are different and similar at the same time, depending on the criteria (size, weight,
cost, etc.). By creating families and selecting their representative elements, the amount of data, that
needs to be processed is significantly reduced. In this case, due to the variety of elements, the classic
ABC analysis is not sufficient [17]. The result of the process is an approximation of the real material
flow, and the deviations are checked by a feedback loop. All five MIRE steps are repeated as needed,
until a satisfactory ratio between the accuracy of the approximation and the saved time is achieved.
MIRE can also be used by companies that already had their material flow recorded, but need to
quickly check how optimal their layout is. This is suitable after adding or updating the machinery
or presenting a new product into the production line. In cases where the company never had their
material flow recorded, MIRE can significantly shorten the time needed by reducing the quantity of
data to be inventoried and analysed, because we do not need to consider all the material flow data, but
only the elements, based on which we can satisfactorily describe the material flow. Needed input can
be accumulated through positions of assembly drawings or any other source, that contains information
about product parts, technological operations, etc.
3. Case Study
A Slovene company having 80 employees counts on adaptability and the possibility of
manufacturing custom-made products, which provides an opportunity for being competitive on
the global market.
Since the company does not have an Enterprise Resource Planning system (ERP system) or does
rather not use it for production management, there is no material flow traceability. The company has
a material flow inventory for positions of an assembly drawing; however, there is no material flow
overview through the production to the assembly. So, there is no overview, through which processing
phases blanks are transferred when leaving the main warehouse. Since a material flow inventory is
needed to prepare a layout, this was a major problem. There was a huge quantity of data in the form of
launching orders and structural product tallies. This is why a methodology had to be found, with
which an accurate material flow inventory could be obtained in the simplest possible way.
An analysis of production assortments of companies in Slovenia has shown that data on masses
of parts are the easiest to obtain; it was used in this paper as the main influencing variable x.
77
Appl. Sci. 2019, 9, 3482
3.1. Application of the MIRE Methodology

The main advantage of the MIRE methodology is the reduction of the number of products, for
which a path through the production needs to be traced in order to achieve a satisfactory inventory of
the total material flow.
The period from 2012 to 2015 was selected for the analysis. Only the last year could have been
included in the material flow inventory, which would reduce the quantity of data, however, a possibility
of influence of potential fluctuation in orders would be increased. Inclusion of a longer period of time
provides for a more precise annual sales average and more reliable anticipation of production load for
the years to come.
The company produced approximately 3300 products in the past four years.
3.2. Selection of Representative Products

By using the clustering technology, the products were classified pursuant to their production
mode into 96 groups. A two-dimensional ABC analysis was used [18] on criteria of product size and
frequency of production. With the help of the employees, the products were divided into three main
families and designated:
• Product (Table 1),

• Supplier product (Table 2),
• Module (Table 3).
The products were classified into 18 groups based on the quantity of produced products. Table 1
shows the data on the quantities produced in the period from 2012 to 2015.
Table 1. Quantities of the products produced in the period 2012–2015.
Name 2012 2013 2014 2015 Sum

Product 1 1 5 3 6 15
Product 2 0
Product 3 1 1
Product 4 3 3
Product 5 1 3 4 2 10
Product 6 3 6 1 10
Product 7 4 1 1 3 9
Product 8 1 1 2
Product 9 1 1 1 3
Product 10 3 2 5 1 11
Product 11 1 1
Product 12 0
Product 13 2 2
Product 14 0
Product 15 1 1
Product 16 1 1
Product 17 0
Product 18 0
Sum 17 23 16 13 69
78
Appl. Sci. 2019, 9, 3482
For a further analysis, five families were selected from the entire family of products (18 products)
The selection was based on experience and the quantities of products these families cover: Product 1,
Product 5, Product 6, Product 7 and Product 10.
Suppliers supply a large quantity of various products (supplier products—77 families) for the
analysed final product, yet many of them appear rarely. Table 2 shows the 10 most frequent products
from suppliers, which are used for the analysis.
Table 2. Most frequent products from suppliers in the period 2012–2015.
Name Quantity Per Month Per Year

Suppl. product 1 236 4.917 59.00
Suppl. product 2 86 1.792 21.50
Suppl. product 3 87 1.813 21.75
Suppl. product 4 85 1.771 21.25
Suppl. product 5 83 1.729 20.75
Suppl. product 6 37 0.771 9.25
Suppl. product 9 266 5.542 66.50
Suppl. product 10 30 0.625 7.50
Additionally, a module was selected that is integrated into all products (products from suppliers
excluded). The modules differ only in size but are treated separately, because a buyer does sometimes
not need it or orders it separately. As they are all practically only scaled versions of Module 1,
weights were introduced on the basis of size/mass. A representative product of the module family was
determined, which is shown in Table 3.
Table 3. Quantity and type of modules produced in the period 2012–2015.
Module Type Quantity Weight Factor × Weight

Module 1 4 2 8
Module 2 2 4 8
Module 3 9 6 54
Module 4 5 8 40
Module 5 3 12 36
Module 6 3 16 48
Module 7 5 20 100
Module 8 11 24 264
Module 9 9 32 288
Module 10 2 40 80
Module 11 1 64 64
Sum 54 990
Average module weight 18.33333
As a representative product, only one module is selected, the weight of which comes closest to
the average weight of the module. In this case, Module 7 with weight 20 came closest to the average
79
Appl. Sci. 2019, 9, 3482
weight of the module which amounts to 18.3. A weight function method was used to determine all
representative products of the selected families.
4. Simulation
A production line with 356 different product groups was simulated, creating more than
44,000 individual products. The properties of each were: produced quantity, mass, the sequence
of technological operations, size, cost and production time. This represented a common year of
production in a smaller company. Data was written in a spreadsheet, then read and handled with
a code written in MATLAB (Matlab 2018a, The MathWorks Inc., Natick, MA, USA). The presented
simulation shows how we can choose a small fraction of products according to desired variables and
still achieve a sufficient portion of the whole material flow.
A simulation by all steps of the MIRE methodology for the selected production will be shown
below. In a first step, an influencing variable needs to be selected, which is a product mass in our case.
Figure 5 graphically illustrates all products of the production as a function of mass.
Figure 5. Simulation of all products.
Figure 5 shows an entire population of products of the selected production at the selected
influencing variable (mass). The arrangement in Figure 5 is not ordered by size, it is random. There is
huge dissemination of a number of products based on the criterion of product mass since the mass
ranges from several kilos to more than two tons. The illustrated case represents a typical production
assortment of a company that produces custom-made products.
This step is followed by grouping of products within a family (Figure 6) and selection of a
representative of the family that will make the best inventory of the products within a family (Figure 7)
as the representative of this family.
Figure 6 illustrates a distribution of products by mass within a certain family that was formed
based on the sequence of treatment processes. More than 500 products are classified in 25 groups.
If a closer look is taken at each group, an average mass of one product of the entire family can be
determined (Figure 7).
The representatives of groups within a family and their mass are shown in Figure 7. The red line
represents an average mass of a product in a family. A representative of group 13 comes closest to this
value and this is why it is selected as the representative of the family.
80
Appl. Sci. 2019, 9, 3482
Figure 6. Simulation of grouping of products within a family.
Figure 7. Simulation of selection of a representative product within a family.
Then follows simulation of descending distribution of representatives by share under consideration

of the criterion of the mass (Figure 8).
The graph in Figure 8 can be used to determine how many representatives will be selected and
their material flow inventoried to include a satisfactory share of the entire production assortment.
The criteria of size, price and production time are also shown.
An increase in the share of production time and size is noticed with the twelfth heaviest
representative, and it will therefore be the most reasonable to select the 12 heaviest representatives
and define their material flow. In this way, more than 80% share of the mass of all products, almost
60% share of the size of all products, approximately 50% share of the total production time, and
approximately 45% share of the price of all products would be captured.
81
Appl. Sci. 2019, 9, 3482
Figure 8. Simulation of descending distribution of representatives by several criteria.
Figure 9 shows a graph of selection of a number of representatives in a multi-criteria selection.

The important variables are the mass and size of a product in a ratio of 3:2, which serves as an example
of approximation of criteria importance regarding transport and stock. As evident from the figure, the
reasonable number of representatives to be selected could be 6, 11 or 17, as the processing time/share
of the inventory ratio is the most beneficial. This depends on the share of the inventory of all variables,
with which we are satisfied. The larger the number of selected representatives, the higher share of the
production assortment will be included. If the first 11 most important representatives are taken as an
example based on the mass and size (in a 3:2 ratio), an approximately 80% share of the total mass of all
products, an almost 60% share of the size and some 30% share of the price and production time of all
products are inventoried.
Figure 9. Simulation of a descending distribution of representatives based on two criteria with

different weights.
82
Appl. Sci. 2019, 9, 3482
The black line represents an order of the representatives which are distributed from the most
important to the least important one based on both criteria: mass and size in a 3:2 ratio. If these two
criteria would be equally important, the black curve would be a line of symmetry of the red and blue
lines or a centroidal axis of the hatched area, but we see that it comes closer to the curve of the mass.
5. Results
The methodology for searching representative elements provides a satisfactory material flow
inventory. In fact, it proves to be a very efficient way for the processing of a large quantity of
mutually similar data. The simulation shows how MIRE manipulates product information through
each step. The generated data represents a typical production assortment of a company that produces
custom-made products. Figures 8 and 9 describe how the share of selected variable increases together
with the number of representatives of each product family. The increase of the share of other variables
can be seen, too, and an example of one and two criteria simulation is illustrated.
Out case study did not include any pre-recorded material flow, thus only an approximation of
the material flow share could be done with MIRE. Sufficient data had to be obtained manually from
assembly drawings and it would be very time consuming to gather all the information needed to
calculate the entire material flow. A solution was needed quickly, which led to the creation of MIRE
methodology. Table 4 shows that the representative products inventory almost a half of the total
material flow. There may be certain deviation, yet it can be neglected if we bear in mind that only
0.5% of all products were selected for the 44% share of the inventory. The work is reduced by almost a
factor of 100, while the obtained material flow is more than sufficient to design a new layout because it
includes the most frequent, the largest and the most important products of the company.
Table 4. The share of material flow inventory and the share of the number of products.
Number of Number of Material Flow

Period 2012–2015 Time Consumed
Products Families Share
estimation:
All products 3300 96 100%
4–6 months
Representative products
16 16 44% 2 weeks
of selected families
Share 0.5% 17% 44% 8%–16%
MIRE is estimated to speed up the process of creating the material flow inventory up to 10 times.
It can be also be used to check how optimal is the current layout. Instead of weeks, the material flow
can be calculated in hours, saving time and consequently money. Although the output does not cover
100% of the material flow, the final share of it is sufficient as we can choose and adjust and multiple
variables and its weights until they meet our needs and achieve high enough certainty of the result.
6. Conclusions
There are many efficient methods designed to search families of parts. They vary from relatively
simple to very sophisticated, but they all have one thing in common: a large amount of data. In our
case, the company did not have a pre-recorded material flow. There was no ERP system, so we had
to be creative in our search for relevant product information. We gathered the necessary data from
assembly drawings manually. With MIRE methodology we were able to find the sufficient data for the
calculation of an approximation of the total material flow.
The MIRE saves plenty of time in processing a huge quantity of data. It is mostly applicable
for cases, where the elements differ from each other based on some criteria and are similar to each
other based on others. It can be of great help to the companies that do not have a digital observer of
inventory of material flow in the ERP system, because it can give a sufficient approximation of the
83
Appl. Sci. 2019, 9, 3482
actual situation. The MIRE result helped in conceiving a suboptimal production layout that will serve
as a basis for re-organisation of the production of the company.
Having proved that the MIRE is applicable and efficient, it is reasonable to develop it further.
The steps, in which weight functions are used, can be even more precisely defined.
Simulations would help us detect the influence of deviation of a representative product from an
average product, the influence of the number of families which is inversely proportional to the number
of elements in an individual family, and the influence of accuracy of grouping of products within a
family. Input data, such as number of workplaces, number of families, and number of all products
would additionally be changed and their correlation with other factors could be monitored.
Ideally, a mathematical model could create graphs and tables that will be easy to use and from
which it would be easy to deduct what share and reliability of a material flow inventory can be
anticipated based on known input parameters. Simultaneously with simulations, the MIRE would
be used on the data in the companies which have an accurate material flow inventory as an upgrade
to the ERP system. The simulated and real results could be compared for an even more thorough
understanding of MIRE’s advantages. This could be done in a company that has a complete overview
and tracking of its material flow. In that case, many other methods have already been developed,
but all of them require an organised database. The MIRE method was developed for a specific and
practical case, where there is essentially minimal information of the company’s material flow. Therefore,
comparison to other methods is a little difficult and unwarranted. Regarding further development of
MIRE methodology, some programming knowledge could convert the final product into computer
software, into which a person could enter input parameters, select the desired material flow, and
share its accuracy. This would result in instructions that would suggest which families should be
selected and what should the representative products be, to which the FROM–TO matrix would need
to be assigned.
Author Contributions: J.M. and T.B. conceived and designed the experiments; J.M. performed the experiments;
J.M., T.B. and J.K. analyzed the data; J.K. contributed analysis tools; J.M. and T.B. wrote the paper. All authors
have read and approved this paper for submission.
Funding: This research received no external funding.
References
1. Ward, A.; Runcie, E.; Morris, L. Embedding Innovation: Design Thinking for Small Enterprises. J. Bus.
Strategy 2009, 30, 78–84. [CrossRef]
2. Wiendahl, H.P.; Reichardt, J.; Nyhuis, P. Handbool Factory Planning and Design; Springer: Berlin/Heidelberg,
Germany, 2015.
3. Zupan, H.; Herakovič, N.; Starbek, M.; Kušar, J. Hybrid Algorithm Based on Priority Rules for Simulation of
Workshop Production. Int. J. Simul. Model. 2016, 1, 29–41. [CrossRef]
4. Alvarado-Iniesta, A.; Garcia-Alcaraz, J.L.; Rodriguez-Borbon, M.I.; Maldonado, A. Optimisation of the
Material Flow in a Manufacturing Plant by Use of Artificial Bee Colony Algorithm. Expert Syst. Appl. 2013,
40, 4785–4790. [CrossRef]
5. Mayers, F.E.; Stewart, J.R. Motion and Time Study for Lean Manufacturing, 3rd ed.; Prentice Hall: Engelewood
Cliffs, NJ, USA, 2001.
6. Francis, R.L.; McGinnis, L.F.; White, J.A. Facility Layout and Layout: An Analytical Approach; Prentice Hall:
Engelewood Cliffs, NJ, USA, 1992.
7. Singh, S.P.; Sharma, R.R.K. A Review of Different Approaches to the Facility Layout Problems. Int. J. Adv.
Manuf. Technol. 2006, 30, 425–433. [CrossRef]
8. Koren, R.; Prester, J.; Buchmeister, B.; Palčič, I. Do Organisational Innovations Have Impact on Launching
New Products on the Market? Strojniški Vestnik J. Mech. Eng. 2016, 62, 389–397. [CrossRef]
9. Bhosale, K.C.; Pawar, P.J. Material Flow Optimisation of Flexible Manufacturing System Using Real Coded
Genetic Algorithm (RCGA). Mater. Today Proc. 2018, 5, 7160–7167. [CrossRef]
84
Appl. Sci. 2019, 9, 3482
10. Božičković, R.; Radošević, M.; Ćosić, I.; Soković, M.; Rikalović, A. Integration of Simulation and Lean Tools
in Effective Production Systems—Case Study. J. Mech. Eng. 2012, 58, 642–652. [CrossRef]
11. Suzić, N.; Stevanov, B.; Ćosić, I.; Anišić, Z.; Sremčev, N. Customizing Products through Application of Group
Technology. J. Mech. Eng. 2012, 58, 724–731. [CrossRef]
12. Douissa, M.R.; Jabeur, K. A New Model for Multi-Criteria ABC Inventory Classification: PROAFTN Method.
Procedia Comput. Sci. 2016, 96, 550–559. [CrossRef]
13. Liu, J.; Liao, X.; Zhao, W.; Yang, N. A Classification Approach Based on the Outranking Model for Multiple
Criteria ABC Analysis. Omega 2016, 61, 19–34. [CrossRef]
14. Selim, H.M.; Askin, R.G.; Vakharia, A.J. Cell Formation in Group Technology: Review, Evaluation and
Directions for Future Research. Comput. Ind. Eng. 1998, 34, 3–20. [CrossRef]
15. Burbidge, J.L. Production Flow Analysis. Prod. Eng. 1971, 50, 139–152. [CrossRef]
16. Dekleva, J.; Kušar, J.; Menart, D.; Starbek, M.; Zavadlav, E. Extended Production Flow Analysis. Robot.
Comput. Integr. Manuf. 1988, 4, 63–68. [CrossRef]
17. Nouri, H.; Tang, S.H.; Hang Tuah, B.T.; Ariffin, M.K.A.; Samin, R. Metaheuristic Techniques on Cell Formation
in Cellular Manufacturing System. J. Autom. Control. Eng. 2013, 1, 49–54. [CrossRef]
18. Starbek, M.; Petrišič, J.; Kušar, J. Extended ABC analysis. Strojarstvo J. Theory Appl. Mech. Eng. 2000, 42,
103–108.
85
MDPI
St. Alban-Anlage 66
4052 Basel
Switzerland
Tel. +41 61 683 77 34
Fax +41 61 302 89 18
www.mdpi.com
Applied Sciences Editorial Office

E-mail: applsci@mdpi.com
www.mdpi.com/journal/applsci
MDPI
St. Alban-Anlage 66
4052 Basel
Switzerland
Tel: +41 61 683 77 34
Fax: +41 61 302 89 18
www.mdpi.com ISBN 978-3-03943-962-1

Design and Optimization of Production Lines

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Design and Optimization of Production Lines

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Design and Optimization of Production Lines

Uploaded by

Copyright:

Available Formats

Design and

Paolo Renna and Michele Ambrico

ISBN 978-3-03943-961-4 (Hbk)

About the Editors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

Paolo Renna and Michele Ambrico

Paolo Renna and Sergio Materi

Irina Aleksandrova, Eugeniy Koshelev and Elena Koresheva

Marco Antonio Zamora-Antuñano, Jorge Cruz-Salinas, Juvenal Rodrı́guez-Reséndiz, Carlos

Jure Murovec, Janez Kušar and Tomaž Berlec

Michele Ambrico is currently a Manufacturing Methodologies Specialist at CRF/WCM Campus

Funding: This research received no external funding.

Appl. Sci. 2020, 10, 8302; doi:10.3390/app10228302 1 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 6845; doi:10.3390/app10196845 3 www.mdpi.com/journal/applsci

− A single type of product is assembled in the assembly system.

3. The Eﬃciency of BB-Based Assembly Cell

task finish time

m=2 m=3 m=4 m=5 m=2 m=3 m=4 m=5

(a) blocking time percentage (b) production rate

4. A New Control Method of BB-Based Assembly Cell

5. Simulation Experiments and Results

Table 1. Assembly operation data for 10 diﬀerent scenarios.

5.1. Experimental Analysis with Deterministic Task Times (No Uncertainty)

Maximum capacity: 601

5.2. Experimental Analysis with Stochastic Task Times

GEN Cell NBB Cell ABB Cell

Figure 7. Performance of assembly systems with various assembly time variabilities.

1%%&HOO $%%&HOO 1%%&HOO $%%&HOO

(a) hand-off time (b) blocking time

5.3. Performance Comparison of Assembly Systems with Intermediate Buﬀers

lead time (minutes)

(a) production rate (b) lead time

*(1&HOO 1%%&HOO $%%&HOO *(1&HOO $%%&HOO

(a) zero-buffer case (b) one-buffer case

KDQGRIIWLPH EORFNLQJWLPH KDQGRIIWLPH EORFNLQJWLPH

(a) zero-buffer case (b) one-buffer case

6. Discussion and Conclusions

Keywords: ﬂow-line design; switch-oﬀ policy; energy consumption

1. Introduction and Motivations

Appl. Sci. 2020, 10, 1475; doi:10.3390/app10041475 19 www.mdpi.com/journal/applsci

2. Flow Line Design Model

The j-th station idle time is deﬁned as follows (Equation (5)):

The idle time for each cycle is given by Equation (6):

WS1 WS2 WSn WSn+1

Figure 1. Distance between stations.

This is subject to the following constrains (Equations (9)–(11))

Minimization of cycle time Precedence constrains Operation processing times

Line design 1: Max productivity

Increment of Increment of Increment of

Line design 2 Line design 3 Line design 4

Figure 2. Framework for ﬂow line designs.

Switch-on Warm Part loaded

Figure 3. Machine states.

4. Reference Context and Simulation Scenarios

Table 1. Operation processing time [s].

Figure 4. Precedence graph.

Four ﬂow lines are obtained in order to get respectively:

Table 2. Stations processing and idle times [s].

MinTTd MaxD_2.5% MaxD_5% MaxD_10%

 Each station has a buﬀer;

(1&HOO 1%%&HOO $%%&HOO (1&HOO $%%&HOO

KDQGRIIWLPH EORFNLQJWLPH KDQGRIIWLPH EORFNLQJWLPH

Each station has a buﬀer;

Each station is controlled by a switch oﬀ policy;

î

( 6819