See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/322708812

A simulation-based approach for plant layout design and production planning

Article in Journal of Ambient Intelligence and Humanized Computing · March 2019

DOI: 10.1007/s12652-018-0687-5

CITATIONS READS

72 3,529

5 authors, including:

Zhinan Zhang Xin Wang

Shanghai Jiao Tong University Shanghai Jiao Tong University
173 PUBLICATIONS 3,115 CITATIONS 12 PUBLICATIONS 574 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Zhinan Zhang on 27 January 2022.

The user has requested enhancement of the downloaded file.

Journal of Ambient Intelligence and Humanized Computing (2019) 10:1217–1230
https://doi.org/10.1007/s12652-018-0687-5

ORIGINAL RESEARCH

A simulation-based approach for plant layout design and production

planning
Zhinan Zhang1 · Xin Wang2 · Xiaohan Wang3 · Fan Cui3 · Hui Cheng3

Received: 7 October 2017 / Accepted: 19 January 2018 / Published online: 25 January 2018
© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract
Facing the new trend of Industry 4.0, manufacturing factories are required to have a more flexible structure to finish produc-
ing customized products within the limited time and at a reasonable cost. Although virtual factory technology is believed
to be helpful for plant layout design and production planning, there still lacks a general framework and algorithms of sim-
ulation-based approach to design an optimized plant layout and the production process. This paper proposes a framework
of simulation-based approach and develops a procedure for the implementation of the proposed framework. The paper also
integrates mathematical algorithms and heuristic methods when applying simulation to balance the operation performance
and the planning cost. An illustrative case demonstrates that the proposed approach can achieve the goal of better plant
layout design and production planning.

Keywords Industrial 4.0 · Plant simulation · Smart manufacturing · Layout planning · Virtual factory

1 Introduction factories are characterized by a more flexible structure to

produce highly customized products in smaller quantities,
Manufacturing planning is a complex engineering problem, at a lower cost, of a higher quality within the required time
which requires a combination of theoretical methods, inter- window. Against such a sweeping trend, it is only possible
net technology and computer-based simulation approaches, when the factories layout and processing flow are rightly
etc. The world is being changed dramatically by Internet of designed and modified quickly.
Things, the Cloud and Industry 4.0 (Molano et al. 2017). A key question about the right designed factory is how
Considering the new manufacturing paradigm, future to ensure the manufacturing system is designed rightly at
the conceptual design phase of a factory. The importance
of conceptual design to new product development cannot
* Zhinan Zhang
[email protected] be overstated. It is widely recognized that roughly 75% of
the cost of producing a product is determined in the end
Xin Wang
[email protected] of conceptual design phase (Ullman 2017). A manufactur-
ing system is a capital intensive engineered system (Suh
Xiaohan Wang
[email protected] 2001), which requires a great deal of modeling and analysis
to ensure designing the manufacturing system rightly. At
Fan Cui
[email protected] the same time, the highly competitive nature of business in
a competitive marketplace is forcing manufacturers to make
Hui Cheng
[email protected] a continuous effort to develop more efficient and effective
approaches and tools to evaluate the impact of design deci-
1
School of Mechanical Engineering, Shanghai Jiao Tong sions on plant layout and production planning.
University, Shanghai 200240, China Against the above background, the concept of “virtual
2
Department of Industrial and Manufacturing Systems factory” emerged and is becoming increasingly popu-
Engineering, The University of Hong Kong, Pokfulam, lar in recent years, which is made possible by the rapid
Hong Kong, China
advancements of information technologies (Lin and Fu
3
Shanghai Spaceflight Manufacture (Group) Co., Ltd, 2001). Terkaj et al. (2015) pointed out that virtual factory
Shanghai 200245, China

13
Vol.:(0123456789)
1218 Z. Zhang et al.

research is insightful, especially about the application of 2 Related work

virtual factory concepts to support production planning
(i.e., plant layout planning, production process planning, Considering the benefits of plant simulation as stated
and material flow simulation). To date, multiple com- above, many specialists have got down to seeking effective
mercial plant simulation software has been developed methods to achieve the efficient design of factory layout
and applied to various industrial and commercial fields, and production planning. The literature review focuses on
e.g., application of plant simulation software to evaluate simulation-based approach and commercial software of
material flow for airport planning, to assess the layout of plant simulation.
automobile production line and to identify the bottleneck
of plant logistic system. Numerous researchers have paid
sufficient attention to the plant layout and production plan- 2.1 Simulation‑based approach
ning. It can be concluded that modeling and simulation
are two key components of a virtual factory. However, Traditionally, production planning is treated as an opti-
there is still a lack of frameworks of simulation-based mization problem, which can be solved using various
approach to guide the process of designing the plant layout algorithms. Experts used to commit to improving the
and planning the production to maximize the throughput. arrangement of factories and the manufacturing process
For example, although commercial software is powerful, by adopting a variety of algorithms. For example, Yildiz
it is difficult to master the using method and a general (2013) developed a hybrid optimization method which was
procedure for different software is lacking. Academic based on an artificial bee colony algorithm and Taguchi
approaches are difficult to demonstrate a real sense in vir- method. Dasgupta and Michalewicz (2013) discussed a
tual reality. kind of order-type genetic algorithm to determine the lay-
In addition, along with the application of new genera- out. Kia et al. (2014) proposed another genetic algorithm
tion information (e.g., cloud computing, big data, Internet to solve a multi-floor layout design model in a manufac-
of Things) in manufacturing and industry, smart manufac- turing system. Tao et al. (2008) introduced the grid com-
turing era is coming (Tao and Qi 2017), and a lot of smart puting in the manufacturing simulation, and designed a
production or smart manufacturing model or system are particle swarm optimization and complex network based
proposed and studied, including cloud based manufactur- method for manufacturing resource optimal allocation in
ing (Tao et al. 2017a), Cyber-Physical System based smart shop-floor Tao et al. (2012).
manufacturing (Tao et al. 2017b). To meet the requirement However, it is hard for algorithms to handle dynamic
on smart manufacturing, how to improve the smartness of problems which are closer to reality, provided that the
production line or factory is another challenge. parameters are usually not deterministic. Furthermore,
Therefore, this study aims at enriching virtual factory the convergence speed and efficiency of algorithms are
theoretical foundation for plant layout design and produc- usually slow when the case is complex or there are many
tion planning by taking advantage of a simulation-based scenarios to consider. Thus, algorithms would stand in an
methodology. What is more, most of the previous papers inferior position when compared with heuristic methods
only suggested different algorithms to optimize the factory and simulation in terms of speed and the coincidence with
layout, which could be very time consuming in the real the real status. Rong (2011) applied appling chaos embed-
case. Therefore, this paper will also pay attention to the ded particle swarm algorithms to optimize the automatic
integration of algorithms and heuristic methods to make block section signaling layout design. Lenei et al. (2013)
a tradeoff between the operation performance and plan- adopted the generic algorithm to facilitate the single-row
ning cost. The contribution of this study is to develop a layout design with multiple objectives. Li et al. (2006)
framework for guiding simulation-based plant design and discussed the application of parallel hybrid PSO-GA algo-
evaluation considering algorithms and heuristic methods. rithms to the layout design. There are also some people
The remainder of this paper is organized as follow- focusing on methods other than algorithms to optimize the
ing. Section 2 gives a review of related works. Section 3 design. For example, Yang et al. (2015) expounded a Value
provides the framework scheme for the simulation-based Stream Mapping (VSM) as a tool to identify those low-
approach and procedures of implementing the proposed value activities in a manufacturing process and increase
framework. Section 4 describes the implementation in a the service level of manufacturing, in consideration of five
real industrial case to validate the proposed approach. Sec- factors that are “production unit”, “pacemaker process”,
tion 5 presents a discussion and limitations of the applica- “number of batches”, “production sequence” and “super-
tion of the digital factory approach. Finally, the conclu- market size”. The method was straightforward to follow,
sions and further developments are summarized in Sect. 6. and hence easy to be applied in practice. Nevertheless,

13
A simulation-based approach for plant layout design and production planning 1219

it somehow fails to take other essential factors beyond simulation. Unfortunately, the factory layout design was not
the five factors into account, for instance, the fabrication within the scope of their research. Ducloux (2014) enlarged
process. Furthermore, VSM is not especially effective the scope of simulation for the entire manufacturing pro-
to evaluate, compare, and select the best option among cess to predict the properties of individual components. The
multiple alternative plans with respect to the estimated emphasis hinges on improving the life span prediction of
cost and schedule. Pahl et al. (2007) proposed a system- individual parts, instead of analyzing overall performance of
atic approach of layout drawing. Although 2D drawing the entire production line. Lindskog et al. (2016) proposed a
is effective to make the layout of a floor area visible, it method to improve the design of the manufacturing system
cannot effectively illustrate the deviations in the vertical through the visualization support. In this paper, they got
space. Moreover, it is extremely difficult, if not impossible, an accurate virtual representation of the current area and
to obtain the operational details simply from the layout would combine the representation with some models. Simu-
drawing, making it very difficult to follow the approach to lation was employed to identify the potential risks, while we
guide the implementation. hope to improve the overall design in terms of efficiency.
The substantial potentials of leveraging the simulation- Al-Ahmari et al. (2016) proposed a virtual reality environ-
related technologies and methods to enhance factory design ment, which was intended to train the frontline works who
have been suggested by many previous studies. For instance, operate assembly lines and to evaluate the assembly deci-
Mourtzis et al. (2014) reviewed those major milestones in sions. This environment only addressed a limited number of
the evolution history of simulation technologies, based on operations instead of operating the overall performance of
which, they projected the knowledge gap, future develop- a plant. Yap et al. (2014) put forward a method to facilitate
ment, and potential challenges of simulation in manufactur- the layout planning from the standpoint of virtual reality.
ing. Liu et al. (2013) also reviewed the development and The method emphasized on the human-robot interface and
application of simulation in manufacturing, with respect the layout planning for robot-centered work cells, with the
to, for example, the advantage, disadvantages, and further aim of overcoming the safety issues and the lack of trained
improvement of various simulation technologies. However, personnel.
they stayed on the theoretical level still. Neither have they Based on the above discussion, it is obvious that there
pointed out a systematic method to utilize simulation to is a knowledge gap. Some previous methodologies includ-
reinvent the manufacturing process. Mujber et al. (2004) ing different algorithms did not deal well with the dynamic
pointed out the potentials of employing Virtual Reality to manufacturing process, with the industry becoming more
enhance operation management and manufacturing process, and more highly collaborative, customer-oriented, eco-
based on the augmented virtual interactions between users efficient and knowledge-intensive. Although some people
and machines. Despite the obvious contributions of these interpreted the applications of simulation in the manufactur-
past studies to deepen the theoretical understandings, these ing industry, they left on the theoretical shelf and have not
review-oriented works are lack of details and specifics to elucidated a practical application procedure, which provides
guide the practical execution. a reference to enterprises. Some scholars had attempted to
There were indeed some experts attempting to apply the apply the simulation method to the factory practice, but they
simulation method into practice, but their focuses were lim- usually concentrated on a limited range of aspects, such as
ited to some sectors or different from us. For example, Siano assembly process, risk prevention, life prediction, etc. Few
et al. (2013) discussed the designing method and testing of the previous literature worked on optimizing the overall
method to present the simulation results and verify the effec- performance of a factory by means of simulation. Therefore,
tiveness. The methods are inspiring, but the model is mainly our study might be a significant point of reference to the
designed for electricity supply design and energy savings production activities in the future.
in smart homes, which cannot be applied to manufactur-
ing industry directly. For this shortage, recently, the new 2.2 Commercial plant simulation software
technology named digital twin was introduced in shop-floor,
e.g., Tao et al. (2017c) proposed the concept of digital twin Plant simulation software functions to facilitate managers
shop-floor in 2017, and the key is to realize the modeling and/or engineers to make more informed design decisions
and simulation of physical job-shop, and then realize the based on the visualization of operation or production. Some
integration and fusion between physical and cyber job-shop simulation software is developed by the industrial leading
(Tao and Zhang 2017). users. For example, Siemens, Dassault, and Applied Mate-
On the other hand, some researchers have endeavored rials have all developed their own plant simulation soft-
to apply simulation in practice. Grienitz et al. (2013) pro- ware (e.g., Siemens Tecnomatix Plant Simulation, Dassault
posed a method called GraFem to model and optimize the CATIA V6, and AutoMod) and provided a variety of ser-
manufacturing process, which is supported by the integrated vice packages to manufacturers. For instance, AutoMod can

13
1220 Z. Zhang et al.

help model operations at different levels, from the equipment changes, simulation can deal with more complex situation
level to the more complicated manufacturing, distribution, and approach closer to the reality. In addition to the static
automation and logistics levels. The users of ProModel do and discrete cases, simulation also handles some dynamic
not need to build models themselves and the software is and continuous information. Furthermore, by conducting
easy to learn. Furthermore, the simulation results can be simulation, different plans can be built, evaluated, tested,
generated in a relatively very short time (Promodel 2016). and eventually compared against different metrics. By doing
However, ProModel was only able to show 2D images. Users so, the optimal solution that yields the highest throughput at
could not simulate the 3D arrangement in the plant or get the the lowest cost can be identified. All the work could be com-
operation details accordingly. pleted before the factory is set up or transformed. In contrast,
Although the aforementioned plant simulation software the investment could be considerable given the company has
can provide nice functions for engineers to model and ana- to change the initial layout or modify the production process
lyze plant operations, these software or tools are different after the factory takes shape. Thus, it is of great significance
from each other and there still lacks a general approach for to perform simulation before building or transforming a fac-
engineers (especially the novice engineers) to gain an overall tory or a production line. A meta-model of simulation-based
cognitive understanding of how to perform plant simulation. approach is shown in Fig. 1, which consists of 5 key steps. A
well-constructed simulation model based on the abstraction
of a real physical system and its operating logic is of utmost
3 Proposed framework for plant simulation importance to lean plant simulation.

Section 3 presents the general framework and the procedures 3.2 System architecture of plant simulation
to implement the proposed framework, and thus to realize
the efficient design of plant layout and production line based To obtain a better plant layout and production planning,
on simulation. First, it is necessary to introduce the concept the right input information and data should be efficiently
of lean plant simulation and to develop a systematic archi- transformed into the desired output. Figure 2 illustrates the
tecture of plant simulation. system architecture of how to perform a simulation-based
lean design of layout and production planning. The frame-
3.1 Concept of efficient plant simulation work is composed of five major components: input, output,
design tools, simulation resource repository, and simulation
Simulation facilitates engineers to “see” the manufacturing modeling. Each component consists of different elements
process, which can help imitate the possible manufactur- that are intended to perform different functions. The inputs
ing process to see the outcome, considering the dynamic of the framework include, for example, physical structure of
parameters. According to the German Association of Indus- manufacturing unit, manufacturing process flow, logistics
trial Engineers, simulation refers to imitation of a dynamic planning and optimization, production equipment tooling,
process within a model to arrive at a better result that can and CAD drawings. The outputs of the framework include,
be transferred to the physical system. Unlike some math- for instance, process analysis, and logistics display. The
ematical algorithms, which could only take the parameters core component of the framework is the simulation mod-
as determined figures but fail to consider the arbitrary eling, which can be decomposed into (1) layout modeling,

Fig. 1  A meta-model of simula-

tion based approach

13
A simulation-based approach for plant layout design and production planning 1221

Fig. 2  System architecture of simulation based lean layout and production planning

simulation and optimization; (2) process modeling, simula- systems, and equipment failures can be identified later. The
tion and optimization; and (3) logistics modeling, simulation detailed procedure of building the 3D model is shown in
and optimization. The tools layer shows the tools that can the following.
be used in lean plant simulation. The knowledge of how to Step 1. Analysis of the manufacturing process flow Fig. 3
use the tools is stored in simulation resource repository. The illustrates the procedure for process flow analysis. The first
simulation resource repository allows a company to manage step to build a virtual model is to have some insight into the
its resource related to plant simulation and thus to support process flow of the production line. Experts are required to
design and re-design of their plant layout and production grasp an idea of the characteristics of the production line,
lines. including the stability of the performance, the efficiency of
the production rate, the degree of the automation, the con-
3.3 A structured process efficient plant simulation venience of maintenance and the quality of the products. In
addition, experts are also expected to get more acquainted
Based on the proposed architecture of simulation-based effi-
cient layout and production planning, several procedures are
developed to support plant layout and production planning
design.

3.3.1 Three‑dimensional modeling of production lines

One important function of the 3D virtual model is to

store a lot of vital information. Some design errors can
be detected during the process of building the 3D models.
What’s more, this method makes it simple to make changes
to the model. Engineers only need to alter the parameters
to modify the whole design. All the relative parts of the
system will automatically change according to the referred
parameters. The software can generate 2D drawings or even
be interconnected to other manage systems of the company
so that various problems in processing technology, logistics Fig. 3  Procedure for the manufacturing process flow analysis

13
1222 Z. Zhang et al.

with the products, including their categories, functions, Step 2. Design and modeling of production line Fig. 4
strengths and shortcomings. shows the procedure for the design and modeling of a pro-
The second step is to model the corresponding relations duction line, which follows the analysis of the manufacturing
between the products and production lines. It is required to process flow. During the process of building models, the
analyze which process flow should be prepared to merchan- corresponding standards that the models have to conform to
dise the products and what devices are demanded. Then, it need to be taken into account. Furthermore, tradeoff should
is possible to build the virtual models of these devices and be made when deciding the final quality of the models, since
attempt to put them in a certain sequence afterward. it takes high time cost and labor cost to build the models of
It is worthy of note that the scope to be studied needs high precision.
to be confirmed so as to avoid invalid work and a waste of Step 3. Model simplification During the process of build-
resources. Engineers are required to pinpoint the bottlenecks ing models, a moderate amount of simplification is regularly
and verify the influence of every segment on the production needed. The first type of models, of which complexity needs
capacity of this plant. to be reduced, are the facilities purchased from the outside
parties, because the engineers have no access to the struc-
ture and operation details of these facilities. Another type is
the facilities of which structures are quite complex, such as
some facilities with multiple degrees of freedom.

3.3.2 Simulation analysis on virtual models

After finishing building 3D models of the facilities, the next

step is to establish the whole digital factory and conduct
the simulation process. The detailed workflow is shown in
Fig. 5. First, the technological flow will be confirmed con-
sidering restrictive conditions and the models built in the
last step. Then, a complete layout is set in the light of the
technological flow. After that, all the involved input param-
eters are set, and the output parameters are defined. The
virtual production process will run and go under simulation
for a period to generate the output results, which provides a
good reference for further analysis. The engineers will come
up with new plans to change the parameters or modify the
Fig. 4  Procedure for the design and modeling of production line

Fig. 5  Procedure for simulation analysis on virtual model

13
A simulation-based approach for plant layout design and production planning 1223

layout accordingly, which will be discussed in Sect. 3.3.3 the performance of the current system and ask “what-if”
in detail. questions accordingly. They will change the parameters of
As mentioned above, the major consideration is the the system and even modify the initial layout to test the pos-
restriction conditions in the first step. The first are the fac- sible effects of those changes on the system. In such way,
tors which have a great effect on the construction address of numerous simulation runs will be performed, and an optimal
this factory, including traffic, labor rates, labor skills, land, setting will be identified.
and closeness to raw materials. These factors may decide
the degree of automation needed, the possible cost, and the 3.3.3 Optimal design of the current plan
feeding strategy. The second restrictive condition is charac-
teristics of demand. For instance, if the annual demand vol- To obtain the optimal design of the whole plan based on
ume is high and the variability is relatively low, the factory the simulation, experts first need to have an idea how they
is suggested to consider improving the degree of automation should understand the output results of the simulation. Then,
and taking advantage of scale economy. What is more, the multiple trials, comparison and analysis from various angles
sophistication of the available equipment is also a signifi- are a must.
cant limiting factor. If some equipment’s capacity is insuf- To find the meaning of the output value, there are two
ficient so that it becomes the bottleneck of the production methods which help use samples to estimate the population
line. Then extra attention is needed to this processing sector. value, i.e. point estimate and interval estimate. Point esti-
Then, according to the limitations and availability of mate is used to calculate the mean and standard deviation
models, the technological flow can be confirmed. The factors of the population, while interval estimate is to give more
to consider consist of rationality, economic efficiency, oper- information about the accuracy of the output parameters
ability, controllability and environmental concerns. Above estimation.
all, the safety of staff and users are the most important. The Then, it is required to settle down the output of current
design of the process flow will directly affect the quality and experiments and conduct more alternative designs. Different
quantity of the end products. plans will possibly have different performance, such as the
The layout is dominated by the process flow. There are average throughput and WIP held in the system. Experts are
four basic types of layout in total, which are fixed-position required to ensure the different performance observed are the
layout, process layout, cellular layout and product layout results of different settings instead of statistical variations so
(Zhang et al. 2015). Each kind of layout has its own advan- that they can reflect actual differences of the physical sys-
tages. The basic layout is mainly determined by the fea- tems. In addition to thinking of the system from the points of
tures of the products and should be modified in accordance engineering capacity, experts from various spheres includ-
with the specific conditions. Fixed-position layout is mainly ing finance, economics and management should help make
for job shop production. The products are typically large some tradeoff from different angles. For example, if the
and heavy so that they are difficult to move. Process layout system gives a high yield every time, it can take advantage
provides a higher flexibility because equipment is arranged of economical scale and avoid some setup cost. However,
according to their functions. In cellular layout, there are the inventory carrying cost is raised consequently. Thus,
many cells, each of which is designed to produce a limited higher throughput does not necessarily mean higher profit.
variety of products. Product layout is for mass production. The comprehensive analysis is crucial to select the optimal
Multiple pieces of devices or workstations are arranged in plan. The procedure for optimal design of current plan is
sequence. The work in process (WIP) are all moved through shown in Fig. 6.
the sequence.
After the initial layout is confirmed, parameters need to
be set. The possible parameters can include arrival rate of
raw materials, processing time on WIP, idle time, the capac-
ity of buffers and the speed of conveyors. It is worth noting
that some parameters are not a deterministic value but a
random number. In this case, some discrete and continuous
probability distribution can be taken use of to simulate these
stochastic variables.
Following the above all preparation work, simulation can
be carried out. The virtual experiment will run for days,
weeks or even months of the system activity in a short
elapsed real time, giving an output representing the result
of the objective functions. Then, those experts will analyze Fig. 6  Procedure for optimal design of current plan

13
1224 Z. Zhang et al.

3.3.4 Integration of algorithms and heuristic methods these three machines. While, all parts will go through K,
L and M after being produced in any of A, B and C. These
It is noteworthy that when applying simulation to the lay- 6 machines will occupy the sites 1, 2, 3, 4, 5, and 6, all of
out plan, the designers should make a wise decision when which are 1 × 1 sq m. The expected quantities of material
choosing mathematical algorithms and heuristic methods. flow between the production machines and the polishing
It is true that mathematical algorithms can better lead to the machines, and that amongst the polishing machines have
optimized solution, but it usually takes a long time, espe- been estimated as shown in the flow-between charts. Assum-
cially when the scenario is complex. In contrast, heuristic ing all the material flow are rectilinear movements between
methods can help obtain a good result much more quickly, centroids, the experts are required to obtain an assignment
which is acceptable and perform better than most solutions. of facilities to minimize the total cost, or quantity-distance
When applying the simulation, our system will first make product Table 1.
a forecast on the planning time needed by the implementa- The first method is to apply the mathematical algorithm
tion of the mathematical algorithm. Given it is within the to calculate the cost of every specific assignment to obtain
required time window, the mathematical algorithm will the global optimized solution. The model representing the
be applied. Otherwise, heuristic methods will be chosen total quantity-distance product should be
instead. An example on plant layout optimization will be
6 j=6
illustrated here to show how to choose from algorithms and 1 ∑∑
heuristics. We will use deterministic numbers here to facili-
TC(𝐚) = w d(a(i), a(j)) (1)
2 i=1 j=1 ij
tate our introduction, while for stochastic values, the simula-
tion methods can be applied. Where TC is the total cost, i.e. the quantity-distance prod-
Assuming that there are six manufacturing facilities in uct and a is the assignment vector of the 6 machines Table 2.
total, which are Production Machine A, B and C and Polish- There should be 6!/2 = 360 kinds of solutions to assign
ing Machine K, L and M. Since A, B and C are producing these facilities to the sites. Two example cases will be intro-
different parts of the products, the WIP will not flow among duced in detail.

Table 1  The layout of the sites (top), and the flow between the facili- Case 1 For A->1, B->2, C->3, K->4, L ->5,
ties (bottom) M->6, or assignment vector = (1,2,3,4,5,6). Then,
the total quantity-distance product for this lay-
1 3 5
out = 0(1) + 0(1) + 7(2) + 10(2) + 5(3) + 0(2) + 9(1) + 6(3)
2 4 6 + 6(2) + 4(1) + 8(1) + 9(2) + 4(2) + 3(1) + 6(1) = 135.
A B C K L M
K 7 9 4 K - 4 3 Case 2 For A->1, B->2, C->3, K->4, L ->6,
L 10 6 8 L - 6 M->5, or assignment vector = (1,2,3,4,6,5). Then,
M 5 6 9 M -

Table 2  Material flow matrix and distance matrix

A B C K L M

A – – – 7 10 5
B – – – 9 6 6
C – – – 4 8 9
K – – – – 4 3
L – – – – – 6
M – – – – – –
1 2 3 4 5 6

1 – 1 1 2 2 3
2 – – 2 1 3 2
3 – – – 1 1 2
4 – – – – 2 1
5 – – – – – 1
6 – – – – – –

13
A simulation-based approach for plant layout design and production planning 1225

the total quantity-distance product for this lay- only 6 times of calculation for the second heuristic method.
out = 0(1) + 0(1) + 7(2) + 10(3) + 5(2) + 0(2) + 9(1) + 6(2) Obviously, more calculation time and storage space are
+ 6(3) + 4(1) + 8(2) + 9(1) + 4(1) + 3(2) + 6(1) = 138. needed for the algorithm.
In this example, the deterministic weight is used for con-
Similarly, we can obtain the total costs for the remaining venience of calculation, while the weight can be a stochastic
358 assignments, and the minimum result is 135. value within a certain range following some distribution, in
The second method is the heuristic one to save time. The which case simulation is more applicable. Therefore, the
whole sites are divided into two parts, i.e. the production experts can set a planning time window first. Given the time
session and the polishing session. The production machines needed for algorithms is within the time limit, algorithm is
are allocated in Site 1, 2 and 3, while the polishing machines suggested to obtain the global optimum. Otherwise, if the
are assigned to Site 4, 5 and 6. We will consider the flow time goes beyond the time limit, heuristic method is recom-
within each session individually to confirm the allocation mended to save planning time.
in this department, while ignoring the flow across different
sessions. First, we consider the allocation within the pro-
duction session. Since there is no flow amongst the pro-
4 Application case
duction machines, we just assign them randomly. We can
just assign A to 1, B to 2 and C to 3. Then, we consider the
This section presents the implementation of the proposed
assignment within the polishing session. Assuming a is the
simulation-based approach, and briefly introduces an
assignment vector of the new departments, there should be
application case of the layout design and optimization of
3! = 6 kinds of assignments, which are a = (4,5,6), (4,6,5),
a workshop.
(5,4,6), (5,6,4), (6,4,5) and (6,5,4).

Case 1 For K->4, L->5, M->6 or assignment vector = 4.1 Background

(4,5,6), the total quantity-distance product within the pol-
ishing machines for this layout = 4(2) + 3(1) + 6(1) = 17. As shown in Fig. 7, an enterprise plans to adopt the work-
shop layout to distribute 12 sets of lathes, AGV, three-
Similarly, the other 5 cases can be calculated as well, dimensional warehouse and industrial robots to achieve the
and the optimal layout for polishing machines are a = (4, intelligent manufacturing mode under the framework of
6, 5) and the total cost within the polishing department is industrial 4.0 with the support of Internet of Things and
= 4(1) + 3(2) + 6(1) = 16. Then, we added the cost incurred cloud computing technologies (Groover 2007; Tao et al.
by the flow amongst the production machines and the polish- 2014). At present, there are only two-dimensional layout
ing machines to obtain the total cost, and the result is 138. plans of workshop equipment, but the accurate 3D models
It is true that the first algorithm obtains a better result, of the existing equipment are lacking, which leads to the
i.e. a smaller total cost. However, it can be easily seen that difficulties in selecting and planning of the robots, AGV
360 times of calculation are needed for the algorithms, while and warehouses.

Fig. 7  The original layout of the

workshop

13
1226 Z. Zhang et al.

Therefore, there are almost no reliable guidelines for 4.2.2 The layout of the industrial robots and AGV
adjusting the layout of the workshop so that enterprise is not
able to correctly evaluate the manufacturing system layout The distance between the machine tools is only 1.73 m, so
and equipment configuration before the actual production. the smaller models of industrial robots are selected. Accord-
Due to the lack of three-dimensional dynamic simulation ing to the survey, three industrial robots of two series of
results, enterprises cannot make quantitative evaluation of FANUC can meet the demand. Taking the weight of the
the discrete manufacturing system. For example, it is not workpiece into account, FANUC M-16iB industrial robots
sure whether the systems which have been put into use have are adopted. To load and unload for two machines, an indus-
an adequate capacity to meet the production requirement trial robot is installed between two processing devices. It’s
in the face of new technology, new products, changes in needed to install a total of six industrial robots. The layout
production planning emergency and equipment maintenance is shown in Fig. 9.
status.
Thus, it is urgent to build a virtual working environ- 4.2.3 The path plans of AGV
ment of the whole workshop which can work as an effective
inspection tool for the new planning design. Based on the Forklift AGV path plans are shown in Fig. 10. There is a
above background, the proposed method is used to simulate total of nine stops, among which stop points cp1 to cp6 are
and optimize the layout of the workshop and the dynamic docking with the industrial robots. CpIn is the stop point
performance of the manufacturing system. docking with the shelves. CpWait is the waiting area where
AGV takes goods from the shelf, and cpOut is the stop point
4.2 The layout planning of warehouses during the downtime of forklift AGV. The operating path
of the forklift AGV is a bidirectional reversible path with a
4.2.1 The layout of the lathes turning radius of 1 m.

In the layout shown in Fig. 8, the width of the inter-device 4.2.4 The 3D plan of the whole warehouse
channel and the relative spacing between the two lathes is so
small that AGV and industrial robots cannot go through to The 3D digital model of the workshop is shown in Fig. 11,
complete the work. Considering movement space required including 12 machine tools, 6 FANUC industrial robots, a
by the forklift AGV and industrial robots, the adjustment forklift AGV, and a shelf. Figure 12 shows the relative posi-
of the location of the machine is needed. After calculation, tion of the industrial robot and the machining tools.
the forklift AGV flexible steering channel width should
not be less than 2.8 m. On this basis, the margin of 0.2 m 4.3 The simulation evaluation of production lines
can be added. The interval between the two sets of verti-
cal machines is adjusted to not less than 3 m. The interval The layout of the machines, the AGV path planning and
between the relatively-installed two machines is adjusted to the layout of the forklift truck in the workshop are shown
not less than 1.73 m, as shown in Fig. 8. in Figs. 9, 10, 11 and 12. On this basis, professionals can

Fig. 8  The layout of the work-

shop after adjustment

13
A simulation-based approach for plant layout design and production planning 1227

Fig. 9  The 3D layout of the

devices and industrial robots

Fig. 10  the path planning of AGV

Fig. 11  3D digital model of the

workshop

simulate the operation of the workshop when they are given 4.3.1 Production efficiency
the production process and production planning.
The simulation runs for 12 h, and the output of the cam var-
ies with time is shown in Fig. 13. The output is regarded as
one set when completing every pallet with 6 cams in total.
Thus, the actual cam output is 6 times the output.

13
1228 Z. Zhang et al.

efficiency, equipment utilization and other statistical infor-

mation of the small workshop is obtained. In this project, the
utilization rate of vertical machining center is about 90%,
while the utilization rate of industrial robots and forklift
AGV is not high. Especially, the utilization rate of indus-
trial robots is less than 5% due to the long machining time.
Therefore, improving the process and shortening the work-
ing hours can not only improve the production efficiency
of small workshops, but also improve the utilization rate of
industrial robots and forklift AGV. Because there are only
several devices in this application cast, heuristic methods are
not necessary here to save the time.
Fig. 12  machine tools, industrial robots and forklift AGV

5 Discussion and limitations

4.3.2 The average utilization rate of equipment
5.1 Discussion
After 12 h of operation, the average utilization of the equip-
ment is shown in Table 3. It can be obtained from the table The above case study indicates simulation is of help to
that the utilization rate of vertical machining tools is 90% or test the practicality of the layout. Various information can
so, which indicates the production efficiency has been high be integrated into the simulation process. Revision on the
enough. However, the utilization rate of industrial robots and digital models becomes easy to operate and shorter time
AGV forklifts are not high in general. Especially the utiliza- is required. The iterative modification can help the engi-
tion rate of industrial robots was less than 5%, because they neers acquire the superior design before construction and
are idling for a long time after completing the loading and the potential errors might occur less frequently in the actual
unloading tasks, due to the longer machining time of Cam production. The quality of products was also guaranteed.
3. Therefore, improving technology and shortening work- In addition, the possible cost to reform the factory layout
ing hours, can not only improve the production efficiency, afterwards could be reduced substantially as well. However,
but also improve the utilization rate of industrial robots and professional talents and the specialized software are indis-
forklift AGV. pensable to perform the simulation and carry out analysis,
which will lead to higher labor cost and purchasing cost.
4.4 Summary The factory is also required to provide training courses to
its workers so that they get used to reading the 3D models
Based on the simulation-approach, the proposed workshop and working with this system. It is true that the factory has
layout and production line program is evaluated. Through alternative choices to outsource the simulation work to other
running the workshop simulation for 12 h, the production professional companies. However, in this condition, the

Fig. 13  the output of the cams

13
A simulation-based approach for plant layout design and production planning 1229

Table 3  The average utilization rate of devices achieve the objective of lean design of plant layout. Specifi-
Device category Serial number Utilization rate
cally, the paper suggests one way to judge whether math-
ematical algorithms or heuristic methods should be used,
Vertical machining tools Rmachine (1) 0.908 and the necessity has been proved. It can be speculated that
Rmachine (2) 0.907 this approach will have instructional significance to a slew
Rmachine (3) 0.902 of companies when they build or revamp a plant, arrange the
Rmachine (4) 0.902 facilities or staff and polish up the process flow. However,
Rmachine (5) 0.897 the available model library and the optimization method
Rmachine (6) 0.897 are not comprehensive for the time being. It’s suggested the
Rmachine (7) 0.894 future works make efforts to enrich these contents. And the
Rmachine (8) 0.894 way to achieve commercialization and synchronization will
Rmachine (9) 0.894 be discussed in the future work.
Rmachine (10) 0.894
Rmachine (11) 0.890 Acknowledgements This research is partially supported by the Minis-
Rmachine (12) 0.889 try of Industry and Information Technology of the People’s Republic of
China (2016ZXFM03002), the Shanghai Academy of Space Technol-
Industrial robots Rrobot(1) 0.049
ogy-Shanghai Jiao Tong University Joint Research Center of Advanced
Rrobot(2) 0.049 Aerospace Technology (USCAST2016-16) and National Key Technol-
Rrobot(3) 0.049 ogy Support Program (2015BAF18B02). Special thanks to Mr. Weimin
Rrobot(4) 0.048 Ding for the case study.
Rrobot(5) 0.048
Rrobot(6) 0.047
Forklift AGV Ragv 0.172 References
Alahmari A, Abidi MH, Ahmad A, Darmoul S (2016) Development of
a virtual manufacturing assembly simulation system. Adv Mech
factory will face the risk of confidential information leakage. Eng 8(3):1–13
Therefore, a company must make a comprehensive SWOT Automod (2017) http://www.appli​edmat​erial​s.com. Accessed 17 Dec
2017
analysis and compare the advantages and disadvantages to
CATIA V6 (2017) https:​ //www.3ds.com/produc​ ts-servic​ es/catia/​ produ​
make a wise decision. cts/v6/. Accessed 07 Oct 2017
Dasgupta D, Michalewicz Z (2013) Evolutionary algorithms in engi-
5.2 Limitations neering applications. Springer, Berlin
Ducloux R (2014) Improvement of part or tooling life prediction
through simulation of whole manufacturing process. Procedia
Although it has been pointed out that a wise decision Eng 81:504–509
between algorithms and heuristic methods is needed and Grienitz V, Hausicke M, Goerzel S, Dangelmaier W, Laroque C (2013)
an example has been given, this paper has not suggested the Systemic manufacturing process modelling and optimization with
integrated simulation. In: A. Klass (ed) Simulation in Produktion
most commonly-used algorithms or heuristic methods and
Und Logistk 2013, pp 99–108
has not discussed how to select from them. What is more, the Groover MP (2007) Fundamentals of modern manufacturing: materials
theoretical feasibilities have been proved, while the approach processes, and systems. Wiley, Hoboken
to set the time limit is not suggested as well. Therefore, more Kia R, Khaksarhaghani F, Javadian N, Tavakkolimoghaddam R (2014)
Solving a multi-floor layout design model of a dynamic cellular
work needs to be done before commercialization. Besides, manufacturing system by an efficient genetic algorithm. J Manuf
the paper mainly focuses the simulation in the stage of layout Syst 33(1):218–232
design before construction, while does not well discuss the Lenin N, Siva Kumar M, Islam MN, Ravindran D (2013) Multi-objec-
synchronization problem in the system operation process. tive optimization in single-row layout design using a genetic algo-
rithm. Int J Adv Manuf Technol 67(5–8):1777–1790
It should be better if our system can monitor the operation Li GQ, Zhao FQ, Guo C, Teng HF (2006) Parallel hybrid PSO-GA
within the production line in real time and help make adjust- algorithm and its application to layout design. In: International
ment from time to time. conference on natural computation. Springer, Berlin, Heidelberg,
749–758
Lin MH, Fu LC (2001) A virtual factory based approach to on-line
simulation and scheduling for an FMS and a case study. J Intell
6 Conclusions and future work Manuf 12(3):269–279
Lindskog E, Vallhagen J, Berglund J, Johansson B (2016) Improving
In this paper, we proposed a framework of simulation-base lean design of production systems by visualization support. Pro-
cedia CIRP 41:602–607
plant layout design and an operation procedure for plant sim- Liu XX, Liu QY, Wang W (2013) Development of virtual simulation
ulation. Case study showed that companies can refer to this in manufacturing industry. In: Applied mechanics and materials.
methodology and follow the above-mentioned framework to Trans Tech Publ, Switzerland

13
1230 Z. Zhang et al.

Molano JIR, Lovelle JMC, Montenegro CE, Granados JJR, Crespo RG Tao F, Guo H, Zhang L, Cheng Y (2012) Modelling of combinable rela-
(2017) Metamodel for integration of Internet of Things, Social tionship-based composition service network and theoretical proof
Networks, the Cloud and Industry 4.0. J Ambient Intell Humaniz of its scale-free characteristics. Enterp Inf Syst 6(4):373–404
Comput. https​://doi.org/10.1007/s1265​2-017-0469-5 Tao F, Zuo Y, Xu LD, Zhang L (2014) IoT-based intelligent perception
Mourtzis D, Doukas M, Bernidaki D (2014) Simulation in manufactur- and access of manufacturing resource toward cloud manufactur-
ing: review and challenges. Procedia CIRP 25:213–229 ing. IEEE Trans Industr Inf 10(2):1547–1557
Mujber TS, Szecsi T, Hashmi MS (2004) Virtual reality applications Tao F, Cheng JF, Cheng Y, Gu SX, Zheng TY, Yang H (2017b)
in manufacturing process simulation. J Mater Process Technol SDMSim: a manufacturing service supply–demand matching
155:1834–1838 simulator under cloud environment. Robot Comput Integr Manuf
Pahl G, Beitz W, Feldhusen J, Grote KH (2007) Engineering design: a 45(6):34–46
systematic approach. Springer, Berlin Tao F, Cheng J, Qi Q (2017c) IIHub: an industrial internet-of-things
Plant Simulation (2017) https​://www.plm.autom​ation​.sieme​ns.com. hub towards smart manufacturing based on cyber-physical system.
Accessed 07 Oct 2017 IEEE Trans Industr Inf. https:​ //doi.org/10.1109/TII.2017.275917​ 8
Promodel (2016) Simulation Conversations with ProModel. 2016. https​ Tao F, Cheng J, Qi Q, Zhang M, Zhang H, Sui F (2017d) Digital
://blog.promo​del.com. Accessed 07 Oct 2017 twin-driven product design, manufacturing and service with big
Promodel (2017) https​://www.promo​del.com/Produ​cts/ProMo​del. data. Int J Adv Manuf Technol. https​://doi.org/10.1007/s0017​
Accessed 07 Oct 2017 0-017-0233-1
Rong H (2011) Design of automatic block section signalling layout of Terkaj W, Tolio T, Urgo M (2015) A virtual factory approach for in situ
appling chaos embedded particle swarm optimization algorithm simulation to support production and maintenance planning. CIRP
based on skew tent map. Inf Control, Autom Robot 132:785–792 Ann Manuf Technol 64(1):451–454
(Lecture Notes in Electrical Engineering book series) Ullman DG (2017) The mechanical design process. McGraw-Hill, New
Siano P, Graditi G, Atrigna M, Piccolo A (2013) Designing and test- York
ing decision support and energy management systems for smart Yang T, Kuo Y, Su C, Hou C (2015) Lean production system design
homes. J Ambient Intell Humaniz Comput 4(6):651–661 for fishing net manufacturing using lean principles and simulation
Suh NP (2001) Axiomatic design: advances and applications. Oxford optimization. J Manuf Syst 34:66–73
University Press, Oxford Yap HJ, Taha Z, Dawal SZ, Chang S (2014) Virtual reality based sup-
Tao F, Qi Q (2017a) New IT driven service-oriented smart manufactur- port system for layout planning and programming of an industrial
ing: framework and characteristics. IEEE Trans Syst Man Cybern robotic work cell. PloS One 9(10):e109692
Syst. https​://doi.org/10.1109/TSMC.2017.27237​64 Yildiz AR (2013) A new hybrid artificial bee colony algorithm for
Tao F, Zhang M (2017e) Digital twin shop-floor: a new shop-floor para- robust optimal design and manufacturing. Appl Soft Comput
digm towards smart manufacturing. IEEE Access 5:20418–20427 13(5):2906–2912
Tao F, Zhao D, Hu Y, Zhou Z (2008) Resource service composition Zhang ZN, Liu G, Jiang ZC, Chen Y (2015) A cloud-based framework
and its optimal-selection based on particle swarm optimization in for lean maintenance, repair, and overhaul of complex equipment.
manufacturing grid system. IEEE Trans Industr Inf 4(4):315–327 J Manuf Sci Eng 137(4):040908

13

View publication stats