Hindawi Publishing Corporation

Shock and Vibration

Volume 2015, Article ID 213159, 11 pages
http://dx.doi.org/10.1155/2015/213159

Research Article
Evolution Balancing of the Small-Sized Wheel
Loader Assembly Line

Rongshen Lai,1 Liang Hou,1 and Yongming Wu2

1
Department of Mechanical and Electrical Engineering, Xiamen University, Xiamen, Fujian 361005, China
2
Key Laboratory of Advanced Manufacturing Technology, Ministry of Education, Guizhou University, Guiyang 550003, China

Correspondence should be addressed to Liang Hou; [email protected]

Received 31 December 2014; Accepted 15 February 2015

Academic Editor: Changjun Zheng

Copyright © 2015 Rongshen Lai et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Assembly line balancing not only directly determines production efficiency but also influences precision and quality of key
assemblies or even the overall performance of final products. Driven by market demand and development of science and technology,
product family must evolve constantly, which necessitates frequent adjustment and rebalancing of product family assembly line
(PFAL). In order to maintain production efficiency, improve assembly quality and precision, and reduce costs for adjustment, the
evolution balancing problem of PFAL for small-sized wheel loader is solved in this paper. Firstly, the evolution balancing model
of PFAL is put forward. Then, with minimizing the number of workstations, the in-station and between-station load indexes
and adjustment costs, and maximizing relevancy between activities as optimization objectives, meanwhile regarding product
platform planning, modularity design and the critical chain technology in concurrent engineering as constraint conditions, the
evolution balancing problem of PFAL is optimized using improved genetic algorithm (IGA). Finally, the whole analysis procedure
is demonstrated by the small-sized wheel loader PFAL case study and the effectiveness of the proposed method is verified.

1. Introduction become the mainstream assembly mode in the industry of

construction machinery. Proposed by Salveson in the 1950s
In most construction machinery manufacturing enterprises, [8], assembly line balancing problem (ALBP) in essence is a
assembly is the last link of production involving many combinatorial optimization problem of rationally assigning
operations and complex processes and has become the core assembly tasks into workstations and meanwhile minimizing
technology. Born in Ford Motor Company, assembly line has differences of operation time between workstations. ALBP
greatly improved labor productivity and opened the prelude can be divided into two categories, both of which are typical
to modern mass production. However, frequent assembly line NP-hard problems [9, 10]: given the cycle time, minimizing
adjustment, such as changes of operating personnel, assembly
the number of workstations and given the number of work-
equipment, and material flow, would decrease assembly pre-
stations, minimizing the cycle time.
cision and quality of final products, which further increases
difficulty of noise and vibration control [1], faults diagnosis Many researchers heretofore have launched in-depth
[2–4], bending, and vibration analysis of structural parts, studies on assembly line balancing problem (ALBP), espe-
also influences the overall performance of such as suspension cially on the mixed-model assembly line balancing problem
systems [5], universal joint in steering handling mechanism (MMALBP) mainly owing to the capability of MMAL of
[6], and rotated shafts [7]. “producing the needed products in needed quantity at needed
With customer demand for personalized and diversi- time” to effectively improve enterprise competitiveness.
fied product models, such as different requirements for Scholl and Becker [11] carried out a comprehensive survey
power, safety, comfortableness, and energy conservation, of simple assembly line balancing problem (SALBP) research
mixed model assembly lines (MMALs) gradually replace with a special emphasis on exact and heuristic solution pro-
the conventional single model assembly lines (SMALs) and cedures of SALBP. Regarding MMALBP, Merengo et al. [12]
2 Shock and Vibration

analyzed some typical problems of manual, mixed-model or mixed-model assembly line balancing problems from
assembly lines and presented new balancing methodolo- static point of view, and research on dynamic evolution pro-
gies pursuing the common goals of minimizing the rate cess of assembly line is rare, especially the lack of evolution
of incomplete jobs in paced lines and in moving lines or balancing research on modular product family assembly line
the probability of blocking/starvation events (in unpaced which is typical in construction machinery manufacturing
lines) and reducing WIP. Haq et al. [13] dealt with mixed- enterprises. Assembly line evolution would certainly induce
model assembly line balancing to minimize the number of adjustment costs, and product family design techniques such
workstations by incorporating a hybrid genetic algorithm as platform, modularity, and commonality would also affect
approach which used the solution from the modified ranked allocation of assembly tasks and assembly line balancing,
positional method as the initial solution to reduce the search therefore, aiming at the small-sized wheel loader product
space within the global space, thereby reducing search time. family and in order to better guide enterprises to implement
Bukchin and Rabinowitch [14] relaxed the restriction that a product family assembly line planning and innovation we
task that is common to multiple models must be assigned to address the dynamic planning of evolution balancing for
a single station and developed an optimal solution procedure PFAL based on analysis of product family evolution. Firstly,
based on a backtracking branch-and-bound algorithm to relation between external influencing factors of product
minimize the sum of stations and task duplication costs. family evolution and assembly line evolution is selectively
Özcan and Toklu [15] presented a new mathematical model
analyzed. Then, in order to improve assembly efficiency and
minimizing the number of mated-stations as the primary
optimize the assembly line, the precedence graph of assembly
objective and the number of stations as a secondary objective
tasks is converted into AOE-net by specific rules; the critical
for a given cycle time and introduced a simulated annealing
tasks and critical path are identified using critical chain
algorithm for the mixed-model two-sided assembly line
technology. Finally, in this paper minimizing the number
balancing problem. Akpınar et al. [16, 17] proposed a hybrid
of workstations, minimizing in-station and between-station
genetic algorithm sequentially hybridizing Kilbridge and
load indexes, maximizing task correlation, and minimizing
Wester Heuristic, Phase-I of Moodie and Young method and
adjustment costs are set as the optimization objectives;
Ranked Positional Weight Technique to solve mixed model
the design techniques such as product platform and mod-
assembly line balancing problem of type I with objectives
ularity and critical chain technology are introduced into
of minimizing the number of workstations, maximizing
the evolution balancing model of PFAL, and the evolution
the workload smoothness between and within workstations.
balancing problem of PFAL is optimized using improved
Aiming at stochastic environment of MMALs, Xu and Xiao
dual-population genetic algorithm to realize dual objectives
[18] provided min-max related and a-worst scenario based
of improving assembly efficiency and reducing adjustment
robust criteria to formulate corresponding optimization
costs.
models and introduced robust optimization approaches to
balance mixed model assembly lines with uncertain task
times and daily model mix changes by scenario planning. 2. Evolution Balancing Process of
Yagmahan [19] considered operation time variations together PFAL for SWL
with the number of stations, the line efficiency, and the
smooth production to formulate the mixed-model assembly Driven by customer demand, core technology, and market
line balancing problem and proposed a more efficient and competition, product families evolve constantly [22], which
effective multiobjective ant colony optimization (MOACO) is mainly embodied in the replacement of modules or com-
algorithm to solve this problem. Manavizadeh et al. [20] ponents and variation in module partition on microscopic
presented a multiobjective genetic algorithm (MOGA) to level, such as the frame system module in small-sized wheel
solve MMALBP in response to different market demands loader evolving from 915 to 916B and to 918 and 918Te; and
considering cycle time (CT) and the number of stations embodied in the change of product family architecture and
simultaneously and minimizing capital costs of design- configuration difference of product variants on macroscopic
ing multiple assembly lines and carried out a comparison level; as shown in Figure 1. There are two situations in the evo-
between six multiobjective evolutionary algorithms (MOEA) lution process of product family: gradual improvement and
to determine the best method for specific situation. Most abrupt change. The former includes adjustment of produc-
recently, aiming at the balancing problem of product family tion quantity for different product variants and operations
assembly line (PFAL), Hou et al. [21] established a balancing such as addition, removal, modification, and innovation on
model for PFAL and simultaneously proposed an improved variant/unique modules in order to accommodate customer
dual-population genetic algorithm to solve product family requirement differences in function or performance or to
assembly line balancing problem (PFALBP) with minimizing implement continuous improvement in product cost or
the number of stations, minimizing the load indexes between quality, while the latter is mainly due to the emergence of
stations and within each station and maximizing task-related new technology, material, or process which promotes core
degree are used as optimization objectives optimization technology upgrading of product platform or product family.
objectives. Product family evolution propels incessant renewal and
The related research aforementioned is the basis of assem- adjustment of its assembly line. Change of product mix ratio
bly line planning and also the key of assembly line balancing. in different periods of time or operation time variation of
However, most studies are limited to analyzing single-model assembly tasks in new product variants leads the existing
Shock and Vibration 3

Evolution of variant configuration

M1 M1
Module 1

M2 M2
Module 2
···
Power 1 ··· Power 2 Power 3 Purification
···
Upgrading Upgrading and
extension LR4B3-G84X4Q and
ZH4100G2 LR4A3-G84X1Q
purification device
XG916 XG918 XG920T
Evolution of key modules

918

916B
915

918Te

Figure 1: Evolution of small-sized wheel loader.

balancing scheme to be no longer valid; in other words, 3. Evolution Balancing Model of

when new products are introduced or operation time of tasks PFAL for SWL
changes, the tasks should be reassigned to maintain load
balancing of assembly line, which is defined as evolution In this paper, we focus on the evolution balancing process of
balancing problem of assembly line in this paper. Product modular product family assembly line, where the modular
family assembly line planning must be able to rapidly and product family refers to a group/series of products oriented
efficiently respond to the dynamic evolution of product to specific market segmentation derived by adding, removing
family such as adding, removing, and upgrading modules and or replacing the variant or unique modules based on product
changes of various planning elements such as cost and new platform. To demonstrate the concepts of product platform
technology introduction and fully consider the influence of and modularity, a simplified example is shown in Figure 3,
dynamic quantity change of variant products, task addition where the working device is shared by different models and
or removal, operation time variation, and different solutions the extinct accessory modules with standard interface can
of module partition on task assignment and assembly line realize specific functions.
balancing schemes. In all, product family assembly line must Product platform and modularity play an important role
realize balancing in the coevolution process with product in product family development, configuration, production,
family; namely, evolution balancing as shown in Figure 2, and so on. Generally, manufacturing enterprises achieve
share assembly resources between product variants and even product diversification and customization by means of addi-
generational products to the greatest extent and provide high tion, deletion, and/or modification operations onto variant or
quality and low cost individualized products in the shortest unique functional modules and meanwhile try to keep the
possible time and reduce the adjustment costs of assembly product platform constant; hence the product platform and
line [23, 24]. key functional modules possess the characteristic of relative
4 Shock and Vibration

States Evolution
balancing
Evolution

Product Product family assembly

family N line N balancing
Product family assembly
... line II balancing ···

Product family assembly

Product line I balancing
family 1

T1 T2 Tn Time

Change of planning environment (labor, material, cost, new technologies, and so on)

Figure 2: Evolution balancing process of PFAL.

which functional modules are divided according to specific

Product platform
criteria and different products are configured through mod-
Working device ule selection and combination to meet diversified market
demand [25]. Modular design supports mass customization
Variant modules with costs close to that of mass production and finds the
best balancing point between product variety, cost, and
performance; in essence, it is the process of product structure
analysis and restructuring. Principle of module division is to
make components in some module with strong correlation,
Accessory 1 Accessory 2 Accessory 3 Accessory 4 Accessory 5 while the components between modules with little correla-
tion. Modular design technique has an important influence
Figure 3: Example of product platform and modularity. on the balancing planning of PFAL, which mainly manifested
that correlation between the assembly tasks for a certain
module is strong; enterprises usually assemble key functional
modules in advance and strictly assign the tasks strongly
higher stability; in other words, the product platform and key
correlated to some specific functional module into the same
functional modules would remain unchanged in the product
station. In this paper, we define task correlation and tend to
family evolution process with much higher probability. We
assign tasks with higher correlation into the same station to
take the relative higher stability of product platform and key
reflect the practical assembly situations.
functional modules into consideration in evolution balancing
Product platform is defined as “a set of subsystems and
planning process with the purpose of realizing assembly line
interfaces developed to form a common structure from which
balancing as well as minimizing adjustment costs.
a stream of derivative products can be efficiently developed
In addition, we pay attention to improving assembly and produced” [26], which is an effective development
efficiency while balancing the assembly line. The critical tasks strategy to improve external diversity and reduce internal
in assembly process have larger space for improvement and differentiation by means of adding, deleting, and/or replacing
restrict/determine the overall assembly efficiency, and the one or several modules based on platform to derive a family
duration of critical tasks and even the tasks themselves are of products and meet customers’ diversified needs. Once the
with relative lower stability compared with product platform product platform is determined, it will generally be relatively
and key functional modules; in other words, the duration of stable in the gradual evolution process of product family.
critical tasks change frequently and/or greatly and the critical Therefore, in the evolution balancing planning process of
tasks themselves are likely to be replaced by some other PFAL, the platform-related assembly tasks should be assigned
tasks. Therefore, we treat critical tasks as flexible planning into fixed stations rather than influenced by variations of
constraints in the evolution balancing optimization model of other tasks, which is conductive to reducing adjustment costs.
PFAL. Modular product family can be represented by a hier-
archical model, as shown in Figure 4, where the functional
3.1. Product Platform/Modularity. Modularity is a prevailing modules are categorized into common modules, optional
design method based on product functional analysis, in modules, and unique modules. Within the same functional
Shock and Vibration 5

Product family
Product variants

Functional modules
Common modules Optional modules Unique modules

Module instances

Figure 4: Hierarchical model of modular product family.

2
a4 = 2
5 a7 For example, the occurrence of event V6 means activities 𝑎5
=3
=4 a1 and 𝑎6 have been completed and activities 𝑎8 and 𝑎9 can be
a1 7 7 0 =2
1
a2 = 6
3
a5
=2 = carried out next, and the duration of activity 𝑎1 is four units
a8
a3 6 a9 = =1 9 of time.
=3 9 a 11 The main research issues of AOE-net include the follow-
4 =2 8
a6 ing: (1) how much time at least is required to complete the
Figure 5: AOE-net. whole project and (2) what are the key activities affecting the
project progress. In an AOE-net, part of the activities can be
performed simultaneously, and the shortest possible time for
completing the whole project is the sum of weighted time in
module, there are several module instances with favorable the longest path from the starting vertex to the end vertex,
interchangeability, and designers can configure different that is, the sum of duration required by the activities on the
product variants to meet diversified customer requirements longest path which is, therefore, defined as the critical path.
by specific selection and combination mechanism. As in Assuming the starting point is V1 , the length of the longest
small-sized wheel loader product family, according to the path from V1 to V𝑖 is called the earliest starting time (EST)
technological capacity in the cooperating company, the brak- of the event V𝑖 which determines the earliest starting time
ing system module is shared by all the variants in the product of activities represented by arcs starting with vertex V𝑖 . Here
family, so it is defined as the common module. In order to we use 𝑒(𝑖) to denote the earliest starting time (EST) and 𝑙(𝑖)
meet different requirements for power, the company pur- to represent the latest starting time (LST) of the activity 𝑎𝑖 ,
chases different power systems such as ZH4100G2, LR4A3- where LST means the latest time to start the activity 𝑎𝑖 on the
G84X1Q, and LR4B3-G84X4Q, as shown in Figure 1, so the premise of not delaying the whole project and the difference
power system module is defined as the variant or optional (𝑙(𝑖)−𝑒(𝑖)) between ES and LS denotes the spare time (namely,
module. time remaining/surplus) of the activity 𝑎𝑖 . When the time
remaining/surplus equals zero, the activity 𝑎𝑖 has no spare
3.2. Critical Chain Technique. AOE-net (Activity on Edge time indicating it is a critical activity. Consequently, the
net) represents the sequence of production activities by critical activities constitute a critical path and it is important
directed edges and is mainly utilized to estimate the comple- for the progress of the whole project to improve efficiency
tion time of the whole project and to search the critical path, of the critical activities, while the early completion of the
by shortening the time of which the efficiency of the entire noncritical activities cannot speed up the progress of the
project would be improved. Therefore, the activities in PFAL project.
could be shown as an AOE-net. In Figure 5, the critical path from V1 to V9 is V1 → V3 →
AOE-net is a weighted directed acyclic graph, where V6 → V8 → V9 with the length 18 which indicates that
the vertices denote events, the arcs, and the corresponding EST is 18, while EST of 𝑎7 is 6 and LST is 16, which implies
weights denote activities and their duration, respectively. An that postponing or starting the activity ahead of time by 10
AOE net including nine events (V1 , V2 , . . . , V9 ) and eleven units of time would not influence the progress of the whole
activities (𝑎1 , 𝑎2 , . . . , 𝑎11 ) is shown in Figure 5. The whole project. Hence it is the key for improving the project progress
process has one and only one starting vertex V1 and end vertex or efficiency to analyze the critical path or activities.
V9 ; that is, the net has only one vertex with zero in-degree and The core of acquiring the critical path is to seek out the
only one vertex with zero out-degree. The occurrence of an critical activity 𝑎𝑖 meeting the requirement of 𝑙(𝑖) = 𝑒(𝑖).
event means the preceding activity has (activities have) been To simplify analysis, we define the earliest and latest starting
completed and the subsequent activity (activities) can begin. time of event V𝑖 as V𝑒(𝑖) and V𝑙(𝑖), respectively. Supposing
6 Shock and Vibration

Table 1: Precedence relation between assembly tasks.

Task A1 A2 A3 A4 A5 A6 A7 A8 A9 A10
Predecessor(s) — — — A1, A2, A3 A4 A5 A6 A6 — —
Duration 41 42 40 39 39 48 36 36 28 30
Operators assigned 5 6 4 10 14 2 3 1 1 1
Task A11 A12 A13 A14 A15 A16 A17 A18 A19 A20
Predecessor(s) — A6, A9, A10, A11 A12 A13 A13 A13 A13 A14, A15, A16, A17 A18 A18
Duration 35 14 13 12 15 18 15 5 8 7
Operators assigned 1 1 1 1 1 1 1 1 1 1
Task A21 A22 A23 A24 A25 A26 A27 A28 A29 A30
Predecessor(s) A18 A19, A20, A21 A22 A18, A23 A23 A18 A25, A26 A23 A23 A23
Duration 7 7 11 23 17 4 8 24 22 31
Operators assigned 1 1 1 1 1 1 1 1 1 1

the activity 𝑎𝑖 is represented by the arc between vertex V𝑗 and by adjustment of tools, equipment, and labors, and the
V𝑘 and the time needed is 𝑇𝑗𝑘 , then the critical activity 𝑎𝑖 needs objective of minimizing adjustment costs 𝐼 is represented in
to meet the following requirements: the following:

𝑒 (𝑖) = V𝑒 (𝑗) , 𝐼 = 𝛼ΔNum𝑠 + 𝛽ΔNum𝑝 + 𝛾Δ𝑇total + 𝛿Δ𝑇idle

(1) (2)
𝑙 (𝑖) = V𝑙 (𝑘) − 𝑇𝑗𝑘 . + 𝜀Δ𝑇cycle + 𝜙Δ𝐿 + 𝜑ΔNum𝑤 ,

The algorithm to acquire the critical path or activities is where Num𝑠 , Num𝑝 , 𝑇total , 𝑇idle , 𝑇cycle , 𝐿, and Num𝑤 , respec-
shown as follows: tively, denote total number of stations, total number of
assembly workers, total time of assembly, idle time, cycle
(1) constructing and initializing an AOE-net containing time, logistical cost, and total number of tasks, while 𝛼, 𝛽,
𝑛 vertexes; 𝛾, 𝛿, 𝜀, 𝜙, 𝜑 signify the degree of influence of each index
(2) starting from the vertex V1 and setting V𝑒(1) = 0 on adjustment costs and are determined on the basis of the
and then successively calculating EST V𝑒(𝑖) of all the practical situations of enterprises and expert experiences.
rest vertexes by topology algorithm: if the number of According to (2), the objective of minimizing adjustment
vertexes in the some obtained topological sequence is costs is met only when variations of all indexes mentioned
less than 𝑛, which means a loop exists in the net, then above reach their lowest.
it is impossible to acquire the critical path and the Aiming at the optimization problem for PFAL evolution
algorithm terminates, shifting to step (3), otherwise; balancing, the optimization objective of evolution balancing
is determined by synthesizing the four balancing objectives
(3) starting from the end vertex V𝑛 and setting V𝑙(𝑛) = 𝑍 proposed in [21] and the evolution objective of minimizing
V𝑒(𝑛) and then successively calculating LST V𝑙(𝑖) (1 ≤ adjustment costs 𝐼, as shown in (3). Besides, the influ-
𝑖 ≤ 𝑛 − 1) of all the rest events by topology algorithm; ences of product family design and critical chain techniques
(4) according to V𝑙(𝑖) and V𝑒(𝑖) of each event (vertex), on assembly line evolution balancing are translated into
calculating EST 𝑒(𝑖) and LST 𝑙(𝑖) of activity 𝑎𝑖 , if 𝑒(𝑖) = constraint conditions and introduced into the optimization
𝑙(𝑖), then 𝑎𝑖 is the critical activity; model of PFAL evolution balancing, and the evolution bal-
(5) determining the critical activities and path according ancing problem is optimized by the dual-population genetic
to the results in step (4) and the algorithm terminates. algorithm proposed in [21]:
𝑃min = 𝜇𝑍 + 𝜂𝐼, (3)
3.3. Optimization Objectives. Based on the mathematical
model of product family [21], we study both product family where 𝜇, 𝜂 are coefficients for dimension adjustment.
assembly line balancing and evolution planning comprehen-
sively. The main objective of PFAL evolution planning is to 3.4. Improved Dual-Population Genetic Algorithm. Product
minimize change of assembly line while maximizing product family mixed-model assembly line balancing is a typical NP-
assembly capability, that is, tradeoff between minimization hard problem with many influencing factors and complicated
of adjustment costs and maximization of productivity. We constraints, the number of task combination increases expo-
suppose that the existing assembly line has already balanced nentially with the number of tasks and it is difficult to solve
and needs adjustment owing to the influences of external this type of problem using traditional optimization methods,
factors such as customer requirements, product functions which makes genetic algorithm one of the best choices.
or quantities, materials, and technologies. The adjustment Therefore, in this paper we adopt the dual-population genetic
costs in evolution balancing process include costs induced algorithm proposed in literature [21] to solve the optimization
Shock and Vibration 7

Start Mutation point 1 Mutation point 2

Precedence matrix (topology) Parent 2 3 1 4 5 7 6 9 8 10

Population 1 Population 2
End
Decoding 1 Decoding 2
Saving results
Fitness Fitness
computation computation Offspring 2 3 1 4 5 9 7 6 8 10
Best
Selection chromosome Selection
Figure 8: Process of mutation.
crossover Yes crossover
mutation No No mutation
Termination?
Reserving the Reserving the Product variants: Constraints

Improved genetic algorithm (IGA)

best best Configuration Critical
Exchanging individuals Demand quantity Platform Fixed
chain
between populations Assembly order tasks station
algorithm
Task duration
Figure 6: Process of dual-population genetic algorithm. ... Critical Flexible
tasks station

Comprehensive AOE- Optimization objectives

precedence graph net
Crossover point Minimizing number of stations,
Converting in-station and between-station
load index;
Parent 1 2 3 1 4 5 7 6 9 8 10
Topological maximizing task correlation;
sorting algorithm minimizing adjustment cost
Parent 2 1 3 2 4 5 8 9 6 7 10 Feasible task Task assignment and assembly
sequence line balancing

Figure 9: Evolution planning model of PFAL.

Offspring 1 2 3 1 4 5 7 8 9 6 10
Table 2: Correlation between tasks.
Offspring 2 1 3 2 4 5 8 7 6 9 10 Task Task Correlation Task Task Correlation
A1 A2 0.8 A15 A18 0.6
Figure 7: Process of crossover. A1 A3 0.5 A16 A17 0.7
A1 A4 0.1 A17 A18 0.6
A1 A5 0.1 A19 A20 0.5
problem of product family assembly line. The process of this A2 A3 0.7 A19 A21 0.5
algorithm is shown in Figure 6 and the main characteristics
A2 A4 0.7 A19 A22 0.5
are concluded as follows:
A3 A4 0.8 A19 A23 0.5
A4 A5 0.8 A20 A21 0.5
(1) using real number encoding method, where each gene
A4 A6 0.4 A21 A22 0.8
stands for one task and the encoding length equals the
A5 A6 0.5 A21 A23 0.5
total number of tasks;
A6 A7 0.1 A22 A23 0.7
(2) adopting single-point crossover mode as shown in A6 A12 0.8 A25 A26 0.7
Figure 7 and the mutation operation process as shown A7 A8 0.5 A25 A27 0.7
in Figure 8. A9 A12 0.8 A26 A27 0.4
A10 A12 0.8 A28 A29 0.4
3.5. Evolution Balancing Optimization Process. The evolution
A11 A12 0.8 A28 A30 0.4
planning process of PFAL based on product family design
A14 A15 0.7 A29 A30 0.4
and critical chain techniques is shown in Figure 9 and can be
implemented in the following steps.

Step 1. Based on the comprehensive precedence graph of Step 2. We determine constraints and objectives of the
product family, we determine that the task sequence meets evolution balancing optimization problem. Besides the con-
the precedence order using topological sorting algorithm, straints in the load balancing model, we add two con-
convert the precedence graph into AOE-net, and then seek straints: assigning platform-related tasks to fixed stations and
out the critical tasks and critical path by critical chain assigning critical tasks to flexible stations. The optimization
algorithm. objectives include minimizing number of stations, in-station
8 Shock and Vibration

Hydraulic system Electric system Cab

Power system

Working device

Exterior

Transmission and
Frame system Braking system Air conditioning Driving system torque converter

Figure 10: Modular architecture of small-sized wheel loader.

A9 A12 A13 A14

8 11 12 13 14
A10 A15
9
15
A11
10
1 A16
A3 16
4
A2 A17
3 17
A1 A4 A5 A6
2 5 6 7
A7
32 A18
A8
31
A24 A22 A19
24 22 19 18
A30 A23
30 23 A20
A29 20
29
33 A21
A28 21
28
A27 A25
25
A26
27 26

Precedence relationship ··· Critical task

Critical path
Figure 11: AOE-net of related tasks.

and between-station load index and adjustment cost, and international enterprise [21] is restudied to verify the effec-
maximizing task correlation. tiveness and feasibility of the proposed evolution balancing
model of PFAL. The small-sized wheel loader is a typical
Step 3. We assign the tasks to stations using improved modular product family, which concludes frame system
dual-population genetic algorithm and obtain the optimal module, braking system module, driving system module,
assignment results of PFAL evolution balancing problem. cab module, and so on, as shown in Figure 10. Different
modules have one or more module instances, so the company
4. SWL Case Study can configure hundreds of wheel loader models to meet
customer needs. Unavoidably, the functional or performance
4.1. Evolution Balancing of PFAL. In this section, the assem- requirements of customers in different period of time changes
bly line of small-sized wheel loader product family in some abruptly, so the corresponding assembly line needs frequent
Shock and Vibration 9

Table 3: Relevant times in AOE-net.

Task A1 A2 A3 A4 A5 A6 A7 A8 A9 A10
LST 41 41 80 80 119 167 274 274 167 167
EST 41 41 40 80 119 167 203 203 28 30
Spare time 0 0 40 0 0 0 71 71 139 137
Task A11 A12 A13 A14 A15 A16 A17 A18 A19 A20
LST 167 181 194 212 212 212 212 217 225 225
EST 35 181 194 206 209 212 212 217 225 224
Spare time 132 0 0 6 3 0 0 0 0 1
Task A21 A22 A23 A24 A25 A26 A27 A28 A29 A30
LST 225 232 243 274 266 260 274 274 274 274
EST 224 232 243 255 260 221 260 267 267 274
Spare time 1 0 0 19 6 39 14 7 7 0

Table 4: Relevant tasks on the critical chain.

Critical chain 1 A1 A4 A5 A6 A12 A13 A16 A18 A19 A22 A23 A30
Critical chain 2 A2 A4 A5 A6 A12 A13 A16 A18 A19 A22 A23 A30
Critical chain 3 A1 A4 A5 A6 A12 A13 A17 A18 A19 A22 A23 A30
Critical chain 4 A2 A4 A5 A6 A12 A13 A17 A18 A19 A22 A23 A30

Table 5: Relevant tasks for certain modules. Table 6: Planning results of assembly line evolution.
Module Relevant tasks Station Task Balancing indexes
Module A 1, 2, 4, 5 1 1, 9
Module B 6, 10, 12 2 2, 10
Module C 14, 15, 17 3 3, 11
Module D 16, 18, 19, 21 4 4, 5 (1) Nine stations
Module E 20, 22, 23, 26, 28, 29 (2) Between-station load: 21
5 6, 12, 13 (3) In-station load: 5.1
6 7, 14, 15, 17 (4) Correlation of tasks: 4.9
adjustment which induces considerable cost. To lower or 7 8, 16, 18, 19, 21
control the adjustment cost in product family evolution 8 20, 22, 23, 26, 28
process. 9 25, 27, 29, 30
The characteristics and relevant parameters of the assem-
bly line of the small-sized wheel loader product family are
described as follows. station will bring convenience and operability to the
whole assembly line. By investigation and survey in
(1) The product family contains three series of products enterprise and the empirical knowledge of domain
(916, 918, and 920T), which are assembled on the same experts, correlation values between 2 of 30 tasks are
line. The daily work time is 560 units of time and the shown in Table 2 while the task correlation not listed
demand quantities of 916, 918, and 920T are 2, 4, and in the table all equals zero.
1, respectively; thus the cycle time CT equals 80 units
of time (= 560/(2 + 4 + 1)). According to the precedence relation between tasks as
(2) According to the precedence of assembly task, the shown in Table 1 and the transformation rules of AOE-net,
duration of tasks, and the demand quantity of each the AOE-net of the 30 assembly tasks is constructed as shown
variant, the comprehensive precedence graph can be in Figure 11.
constructed [27] and converted into the precedence Using the introduced algorithm of critical chain in
relation table as shown in Table 1, including 30 tasks Section 3.2, the relevant times such as EST, LST, and spare
and the corresponding operation time (duration). time and critical activities whose EST is equal to LST are
acquired as shown in Table 3.
(3) When operation time of a certain task is greater than Four critical chains are found out and listed in Table 4,
CT in the assembly process, increasing the number by shortening the duration of critical tasks on which would
of operators is more conducive to making operation greatly improve assembly efficiency.
time less than CT. Via investigation in cooperative enterprise and modu-
(4) Certain correlation exists between tasks and grouping larity analysis, among all tasks of the 916, 918, and 920T
tasks with good/better correlation into the same assembly line, the tasks involved in loader platform and
10 Shock and Vibration

Table 7: Comparison of planning results.

Model Station Task Balancing indexes Evolution assumptions Adjustment cost

1 1, 10
2 2, 9
3 3, 11
(1) Number of stations: 9
4 4, 5
Load balancing (2) Between-station load: 18 5 + 6 + 7 + 5 = 23
model 5 6, 12, 13 (3) In-station load: 4.5
6 7, 16, 18, 19, 20 (4) Correlation of tasks: 4.8
7 8, 14, 15, 17
8 21, 22, 23, 24, 26, 28 (1) Deleting task 7 (unique task)
(2) Deleting task 21 (unique task)
9 25, 27, 29, 30
(3) Adding one new task in station 8
1 1, 9 (4) Adding one new task after deleting task 7
2 2, 10
3 3, 11
(1) Number of stations: 9
4 4, 5
Evolution (2) Between-station load: 21 4 + 5 + 6 + 4 = 19
balancing model 5 6, 12, 13 (3) In-station load: 5.1
6 7, 14, 15, 17 (4) Correlation of tasks: 4.9
7 8, 16, 18, 19, 21
8 20, 22, 23, 26, 28
9 25, 27, 29, 30

certain modules are shown in Table 5. Adding the constraints line balancing oriented to product family evolution process,
mentioned above into the improved genetic algorithm [21], we should not only consider loader balancing but also treat
the number of stations, in-station and between-station loads, evolution planning as one more important optimization
and correlation between activities are optimized, meanwhile, objective.
the evolution-related objectives such as minimizing varia-
tions of quantity of operators, cycle time, and material flows
are taken into consideration, and the optimization results are 5. Conclusions
shown in Table 6. (1) Based on discussion of product family evolution process
and its influencing factors, the dynamic evolution process of
4.2. Performance Analysis. To further demonstrate the effec- PFAL (coevolution of product family and its assembly line) is
tiveness of the proposed planning model for PFAL evolution analyzed and illustrated by the case of SWL product family;
in this paper, the results are compared with those of load combined with characteristics of PFAL and its balancing
balancing model [21] as shown in Table 7. research status, the evolution balancing problem of PFAL is
It is observed from the table that, after adding constraints focused on in this paper.
related to product family design and critical chain techniques, (2) Product platform/key functional modules (such as the
the optimized number of stations is still 9, but the between- working device module or platform) change with a very small
station and in-station indexes have higher values, which probability in product family evolution process and it is the
indicates that the optimization result is not superior to that key to raise efficiency of critical tasks (such as assembling
of when only considering load balancing. However, not only accessory module instances) for improvement of the overall
load balancing of PFAL but also its evolution process is assembly efficiency; therefore, influences of the product
evolved in this paper. Through analysis of labor cost variation family design techniques and critical chain technology on
induced by assembly line adjustment based on different bal- task assignment and assembly line balancing are discussed
ancing models (load balancing model or evolution balancing comprehensively and translated into constraint conditions in
model), it is not difficult to find that, if the load balancing the evolution balancing model of PFAL.
model adopted, then the adjustment cost is 23 units of labor (3) The evolution balancing problem of PFAL is optimized
cost, while if we use the evolution balancing model proposed using an intelligent evolutionary algorithm (dual-population
in this paper, the adjustment cost is 19 units of labor cost. GA) and, compared with the traditional load balancing
Owing to individualization of customer requirement and model, results of the SWL case show that, in the constant
shorter lifecycle of seasonal product family, adjustment or evolution process of product family, adjustment costs of PFAL
evolution balancing of PFAL becomes increasingly frequent; reduce significantly using the evolution balancing model
hence the benefits of evolution balancing model are much with other indicators largely unchanged, which generates
more considerable. Therefore, when dealing with assembly considerable economic benefits for enterprises.
Shock and Vibration 11

(4) The research on evolution balancing focuses on the [12] C. Merengo, F. Nava, and A. Pozzetti, “Balancing and sequenc-
issue of assembly line adjustment when the current assembly ing manual mixed-model assembly lines,” International Journal
line has already balanced but was disturbed by external of Production Research, vol. 37, no. 12, pp. 2835–2860, 1999.
dynamic factors with more planning constraints and much [13] A. N. Haq, K. Rengarajan, and J. Jayaprakash, “A hybrid genetic
more difficulty. Therefore, how to fully consider all factors algorithm approach to mixed-model assembly line balancing,”
such as tradeoff between load balancing indicators and The International Journal of Advanced Manufacturing Technol-
adjustment costs, analysis of other adjustment costs besides ogy, vol. 28, no. 3-4, pp. 337–341, 2006.
labor cost, and so on and then to establish an objective [14] Y. Bukchin and I. Rabinowitch, “A branch-and-bound
model more precisely for the PFAL evolution balancing is an based solution approach for the mixed-model assembly
interesting perspective, in which further research is needed. line-balancing problem for minimizing stations and task
duplication costs,” European Journal of Operational Research,
vol. 174, no. 1, pp. 492–508, 2006.
Conflict of Interests [15] U. Özcan and B. Toklu, “Balancing of mixed-model two-sided
assembly lines,” Computers and Industrial Engineering, vol. 57,
The authors declare that there is no conflict of interests no. 1, pp. 217–227, 2009.
regarding the publication of this paper.
[16] S. Akpınar and G. M. Bayhan, “A hybrid genetic algorithm for
mixed model assembly line balancing problem with parallel
Acknowledgments workstations and zoning constraints,” Engineering Applications
of Artificial Intelligence, vol. 24, no. 3, pp. 449–457, 2011.
This work was financially supported by the National Natural [17] S. Akpınar, G. M. Bayhan, and A. Baykasoglu, “Hybridizing
Science Foundation of China (71172055) and the Special ant colony optimization via genetic algorithm for mixed-model
Program for Innovation method of the Ministry of Science assembly line balancing problem with sequence dependent
and Technology, China (Grant no. 2011IM020400). setup times between tasks,” Applied Soft Computing Journal, vol.
13, no. 1, pp. 574–589, 2013.
References [18] W. Xu and T. Xiao, “Strategic robust mixed model assembly line
balancing based on scenario planning,” Tsinghua Science and
[1] M. Azoulay, A. Veprik, V. Babitsky, and N. Halliwell, “Dis- Technology, vol. 16, no. 3, pp. 308–314, 2011.
tributed absorber for noise and vibration control,” Shock and [19] B. Yagmahan, “Mixed-model assembly line balancing using
Vibration, vol. 18, no. 1-2, pp. 181–219, 2011. a multi-objective ant colony optimization approach,” Expert
[2] Y. Lei, J. Lin, and Z. He, “Fault diagnosis of planetary gearboxes Systems with Applications, vol. 38, no. 10, pp. 12453–12461, 2011.
based on multi-sensor information fusion,” Journal of Vibration [20] N. Manavizadeh, M. Rabbani, D. Moshtaghi, and F. Jolai,
and Shock, vol. 29, pp. 216–218, 2010. “Mixed-model assembly line balancing in the make-to-order
[3] J. Xiang, X. Cui, Y. Wang et al., “Optimized stochastic resonance and stochastic environment using multi-objective evolutionary
method for bearing fault diagnosis,” Transactions of the Chinese algorithms,” Expert Systems with Applications, vol. 39, no. 15, pp.
Society of Agricultural Engineering, vol. 30, pp. 50–55, 2014. 12026–12031, 2012.
[4] Y. Lei, T. Wei, D. Kong et al., “Vibration signal simulation and [21] L. Hou, Y.-M. Wu, R.-S. Lai, and C.-T. Tsai, “Product family
fault diagnosis of planetary gearboxes based on transmission assembly line balancing based on an improved genetic algo-
mechanism analysis,” Chinese Journal of Mechanical Engineer- rithm,” The International Journal of Advanced Manufacturing
ing, vol. 50, pp. 61–68, 2014. Technology, vol. 70, no. 9–12, pp. 1775–1786, 2014.
[5] W. Zhou and D. Li, “Improving performance of cantilevered [22] L. Hou, Y.-M. Wu, R.-S. Lai, and S. H. Huang, “Evaluation and
momentum wheel assemblies by soft suspension support,” driver analysis in product family evolution,” The International
Shock and Vibration, vol. 20, no. 4, pp. 737–748, 2013. Journal of Advanced Manufacturing Technology, vol. 69, no. 1–4,
[6] J. Lu, Y. Xu, C. Hu, A. F. Vakakis, and L. A. Bergman, “5- pp. 863–874, 2013.
DOF dynamic model of vehicle shimmy system with clearance [23] F. Kimura and J. Nielsen, “A design method for product family
at universal joint in steering handling mechanism,” Shock and under manufacturing resource constraints,” CIRP Annals—
Vibration, vol. 20, no. 5, pp. 951–961, 2013. Manufacturing Technology, vol. 54, no. 1, pp. 139–142, 2005.
[7] J. Xiang, Z. Jiang, and X. Chen, “A class of wavelet-based [24] A. Bryan, J. Ko, S. J. Hu, and Y. Koren, “Co-evolution of product
Rayleigh-Euler beam element for analyzing rotating shafts,” families and assembly systems,” CIRP Annals—Manufacturing
Shock and Vibration, vol. 18, no. 3, pp. 447–458, 2011. Technology, vol. 56, no. 1, pp. 41–44, 2007.
[8] M. E. Salveson, “The assembly line balancing problem,” Journal [25] L. Hou, R. Tang, and Y. Xu, “Review of theory, key technologies
of Industrial Engineering, vol. 6, pp. 18–25, 1955. and its application of modular product design,” Chinese Journal
[9] J. Bukchin, E. M. Dar-El, and J. Rubinovitz, “Mixed model of Mechanical Engineering, vol. 40, no. 1, pp. 56–61, 2004.
assembly line design in a make-to-order environment,” Com- [26] M. H. Meyer and A. P. Lehnerd, The Power of Product Platform—
puters & Industrial Engineering, vol. 41, no. 4, pp. 405–421, 2001. Building Value and Cost Leadership, The Free Press, New York,
[10] E. Erel, I. Sabuncuoglu, and H. Sekerci, “Stochastic assembly NY, USA, 1997.
line balancing using beam search,” International Journal of [27] Z. Zhang, W. Cheng, B. Zhong, and J. Wang, “Hybrid behavior
Production Research, vol. 43, no. 7, pp. 1411–1426, 2005. ant colony optimization for mixed-model assembly line balanc-
[11] A. Scholl and C. Becker, “State-of-the-art exact and heuristic ing problem,” Journal of Mechanical Engineering, vol. 45, no. 5,
solution procedures for simple assembly line balancing,” Euro- pp. 95–101, 2009.
pean Journal of Operational Research, vol. 168, no. 3, pp. 666–
693, 2006.