You are on page 1of 6

Tatay, V., Koltai, T.

: Solving Assembly Line Balancing Models in Excel Environment to Support


Production Management Decisions, microCAD International Scientific Conference, Company
Competitiveness in the 21st Century Section, Miskolc, Magyarország, Miskolci Egyetem Innovációs és
Technológia Transzfer Centruma, 18-20 March, 2010., pp. 165-170.
SOLVING ASSEMBLY LINE BALANCING MODELS IN EXCEL
ENVIRONMENT TO SUPPORT PRODUCTION
MANAGEMENT DECISIONS

Viola Tatay1, Tamás Koltai2


PhD Student, Dr. Habil, Professor of Production and Operations Management
Budapest University of Technology and Economics, Department of Management
and Corporate Economics

INTRODUCTION

Assembly line balancing (ALB) problems occur where several indivisible work
elements (tasks) are to be grouped into (work)stations along a continuous
production line. ALB problems are typical in automobile and bicycle production,
and in the electronic industry (assembling refrigerators, televisions, etc.) [4].
Tasks cannot be allocated to the stations arbitrarily. Cycle time (Tc) constraints,
precedence relations – generally visualized by a precedence graph, zoning
conditions, technological and logical requirements may influence the optimal
assignment. Even considering these restrictions tasks i=1...n with their task time (ti)
can be assigned to stations j=1…m in many different ways. Heuristic methods
trying to approach the optimal solution are discussed in the literature in detail. As a
consequence of the development of computer and information technology, however,
even large ALB problems can be solved by mathematical programming software in
a reasonable time frame.
Research first focused on the simple assembly line balancing problem (SALBP)
with its restrictive characteristics such as deterministic process times, no assignment
restrictions besides the precedence constraints, serial line layout with m stations, etc
[6]. Extended forms of the SALBP exist (like U-shaped lines, parallel stations) that
are referred to general assembly line balancing problems (GALBP). GALBPs may
be closer to practical problems; however their solution procedure in most cases is
based on SALBP algorithms. [2] Depending on the objective different versions of
SALBP models can be formulated. SALBP-1 minimizes the number of stations for
a given the cycle time (Tc), SALBP- 2 minimizes cycle time (Tc) for a given number
(m) of workstations. SALBP-E maximizes line efficiency (E=(Σti)/(mTc)). SALBP-F
is a feasibility problem that seeks for a feasible solution given m and Tc [4], [6].
SALBPs can be solved by mathematical programming models. The first
analytical formulation of ALB was given by Bryton and the first linear
programming problem that might have infeasible solutions because of split tasks
was given by Salveson [1]. Bowman suggested two different integer programming
(IP) models to solve the classical ALB problem, i.e. SALBP-1 [3]. White modified
Bowman’s second IP model and defined 0-1 decision variables for the problem. The
decision variable xij equals to 1 if task i is assigned to station j; and to 0 otherwise
[7]. As ALB models are NP hard problems hard work was made in the past to
reduce computational efforts. One way to reduce computation time is to improve the
objective function and the precedence constraints of the Bowman-Write formula
[1], [5].
Today mathematical programming models for practical ALB problems can be
solved by optimization software in negligible time. In this paper a practical method
using ALB models embedded in Excel is provided. The results can support daily
operative production management decision making.
The paper is structured as follows. The SALBP-1 model minimizing the number
of stations and SALBP-2 model minimizing the cycle time are presented. Then the
application of these two models is illustrated with a sample problem. Finally the
consequences of the results on production management decision making are
discussed.

FORMULATION OF THE SALBP-1 MODEL

The IP model of an SALBP-1 model with more than one last tasks can be
formulated by (1)-(6). The cycle time constraints (2) ensure, that for a station j the
sum of task times is less than or equal to the predefined cycle time. Equalities (3)
make sure that each task is assigned to a station. The precedence relations are
specified by constraints (4). The objective function (1) with the help of constraints
(5) guarantees that the final tasks are assigned to the earliest workstations possible.
The earliest and latest stations for task i can be determined by constraints (6).

Min (Q ) (1)
n

∑t x
i =1
i ij ≤ Tc j = 1,..., m (2)
m

∑x
j =1
ij =1 i = 1,..., n (3)
m

∑ j ⋅ (x
j =1
qj − x pj ) ≥ 0, ( p, q ) ∈ R (4)
m
Q ≥ ∑ ( j ⋅ xkj ) k∈L (5)
j =1

xij = 0 j ≤ LJ i and j ≥ UJ i i = 1,..., n, (6)


where
xij=0,1,
R={(p;q) | task p immediate precedes task q},
L={i | task i does not precede any other task},
Pi={all the proceedings before task i},
Si={ all the successors after task i },
    
  ∑ tk   ∑ tk  
  k∈Pi , if T ⋅  k∈Pi  − ∑ t ≥ t 
  Tc  c
 Tc  k∈Pi
k i

    
    ,
LJ i =  
 t    
  k∑ k   ∑ tk  
  i , if Tc ⋅  i  − ∑ tk < ti
∈P k∈P

  T c   Tc  k∈Pi 
  




 
 
     
  ∑ tl   ∑ tl  
m + 1 −  l∈Si , if Tc ⋅  i  − ∑ tl ≥ ti
l∈S 
  Tc   Tc  l∈Si 
     
     .
UJ i =  
     
  ∑ tl   ∑ tl  
m + 1 −  i  , Tc ⋅  i  − ∑ tl < ti
l∈S l∈S
if 
  Tc   Tc  l∈Si 
     
     

FORMULATION OF THE SALBP-2 MODEL

The IP problem of SALBP-2 can be written as equations (7)-(10). SALBP-2


minimizes the cycle time (7) that is equivalent to maximizing production rate. Cycle
time constraints (8), occurrence conditions (9) and precedence constraints (10) are
the same as equations (2), (3) and (4) of the SALBP-1 model.

Min Tc (7)
n

∑t x
i =1
i ij ≤ Tc j = 1,..., m (8)
m

∑x
j =1
ij =1 i = 1,..., n (9)
m

∑ j ⋅ (x
j =1
qj − x pj ) ≥ 0, ( p, q ) ∈ R (10)

SAMPLE PROBLEM

To illustrate the solution procedure in Excel environment and the application of


the results a simple ALB problem will be solved. Figure 1. shows the precedence
graph of the problem with the processing times and the index of tasks (i) as well.

12s
45s 11s 9s F
A B C i=6
i=1 i=3 i=5 12s
G
i=7 8s 9s
12s J K
H i=10 i=11
50s 15s
D E i=8
i=2 i=4 12s
I
i=9
Figure 1.
The precedence graph of the sample ALB problem
Table 1. summarizes the input data required to solve the SALBP-1 model of the
sample problem. The first table contains the precedence matrix representing the
precedence graph. Number 1 in the row of task A and in the column of task B
means that task A must be finished before task B can be started. The process times
are given in the second table. The last table shows the cycle time which is equal to
the ratio of the available time for production and the quantity to be produced.

Table 1.
The input data of the SALBP-1 model
Tasks A B C D E F G H I J K Task Time (s)
A 0 1 0 0 0 0 0 0 0 0 0 A 45
B 0 0 1 0 0 0 0 0 0 0 0 B 11
C 0 0 0 0 0 1 1 0 0 0 0 C 9 Total available time
25200
D 0 0 0 0 1 0 0 0 0 0 0 D 50 (second)
E 0 0 0 0 0 0 0 1 1 0 0 E 15 Quantity to be produced
500
F 0 0 0 0 0 0 0 0 0 1 0 F 12 (piece)
G 0 0 0 0 0 0 0 0 0 1 0 G 12 Cycle time
50,4
H 0 0 0 0 0 0 0 0 0 1 0 H 12 (second/piece)
I 0 0 0 0 0 0 0 0 0 1 0 I 12
J 0 0 0 0 0 0 0 0 0 0 1 J 8
K 0 0 0 0 0 0 0 0 0 0 0 K 9

Solving the SALBP-1 model for the sample problem with the data of Table 1 the
results summarized in Table 2 can be obtained. It can be seen that at least four
workstations are needed to perform the eleven tasks. The capacity utilization (the
sum of task times assigned to a station divided by the cycle time) of each station
and the line efficiency are indicated as well. For example, task B, E, H and J are
assigned to station 3. The total time of these four tasks is 50 second that results in
99,2% capacity utilization. The highlighted cells with zero value are those stations
which can not be used for the corresponding tasks as a consequence of constraints
(6). The variables belonging to these cells can be excluded from the calculation, and
this way the possible number of variables (IxJ) can be reduced.

Table 2.
The optimal solution of the SALBP-1 model
Task\station 1 2 3 4 5 6 7 8 9 10 11
A 1 0 0 0 0 0 0 0 0 0 0 Capacity
Station Time (s)
B 0 0 1 0 0 0 0 0 0 0 0 utilization
C 0 0 0 1 0 0 0 0 0 0 0 1 45 0,893
D 0 1 0 0 0 0 0 0 0 0 0 2 50 0,992
E 0 0 1 0 0 0 0 0 0 0 0 3 50 0,992
F 0 0 0 1 0 0 0 0 0 0 0 4 50 0,992
G 0 0 0 1 0 0 0 0 0 0 0
H 0 0 1 0 0 0 0 0 0 0 0
I 0 0 1 0 0 0 0 0 0 0 0 Efficiency of the
0,967
J 0 0 0 1 0 0 0 0 0 0 0 assignment
K 0 0 0 1 0 0 0 0 0 0 0
The SALBP-2 model minimizes the cycle time for a given number of
workstations. In this case the precedence matrix, the process times and the number
of workstations are the required input data. The results of the SALBP-2 model of
the sample problem with six workstations are given in Table 3. It can be seen that
the minimal cycle time is 50 second. The optimal value of the objective function
equals to the process time of task D thus it is the bottleneck of the system. As the
available time is 25200 second, 504 products can be produced.

Table 3.
The optimal solution of the SALBP-2 model
Task\station 1 2 3 4 5 6 7 8 9 10 11
A 0 1 0 0 0 0 0 0 0 0 0 Capacity
Station Time (s)
B 0 0 1 0 0 0 0 0 0 0 0 utilization
C 0 0 1 0 0 0 0 0 0 0 0 1 45 0,90
D 0 0 0 1 0 0 0 0 0 0 0 2 44 0,88
E 0 0 0 0 1 0 0 0 0 0 0 3 50 1,00
F 0 0 1 0 0 0 0 0 0 0 0 4 15 0,30
G 0 0 1 0 0 0 0 0 0 0 0 5 41 0,82
H 0 0 0 0 0 1 0 0 0 0 0
I 0 0 0 0 0 1 0 0 0 0 0 Cycle time (s/piece) 50
J 0 0 0 0 0 1 0 0 0 0 0 Quantity (piece) 504
K 0 0 0 0 0 1 0 0 0 0 0

CONCLUSION

The input and the output data of the SALBP-1 and the SALBP-2 models are stored
in Excel and the models are solved by the LINGO software. The model of a
practical problem can be solved by these tools in a negligible time, therefore, daily
production management decisions related to ALB can be supported.
The minimal number of workstations required to do all of the work elements can
be determined by the SALBP-1. However, in such production environment, where
low volume production of customized products with frequent setups is typical
varying the number of workstations for nearly each product would be time-
consuming and costly. Therefore, in these situations SALBP-2 model are solved.
With given number of workstations the assignment, which maximizes the
production rate can be determined. Moreover the line efficiency can be analyzed
with the help of SALBP-2. Line efficiency can be written as follows:
n n n

∑t ∑t i i ∑t i
E= i =1
= i =1
=Q i =1
, (11)
mTc T mT
m
Q

where T is the total available time. It can be seen that line efficiency (E) can be
expressed as a function of the quantity to be produced (Q). For different number of
stations Q can be determined by an SALBP-2 model. Figure 2 shows the line
efficiency as a function of the quantity to be produced for all possible stations (1-11
stations) in case of our sample problem. Generating this graph for each product the
efficiency and the quantity to be produced can be analyzed.

E
1,2

0,8

0,6

0,4

0,2

0 Q
500 400 300 200 100 0

Figure 2.
Line efficiency as a function of quantity to be produced

This paper provided some basic models and methodology for ALB problems.
The advantage of the suggested computational framework in this paper is that input
data and results are presented in Excel, therefore, no special software is required to
present the results and to use them for further processing. Since computational time
with any advanced LP solver is very short, even daily assembly line related
operations management decisions can be easily supported.

REFERENCES:

[1] BAYBARS, Í.: A Survey of Exact Algorithms for the Simple Assembly
Line Balancing Problem. Management Science, 1986, Vol. 32, No. 8, pp.
909-932
[2] BECKER, C. AND SCHOLL A.: A Survey on Problems and Methods in
Generalized Assembly Line Balancing. European Journal of Operational
Research, 2006, pp 694-715
[3] BOWMAN E. H.: Assembly Line Balancing by Linear Programming.
Operations Research, 1960, Vol. 8, pp. 385-389
[4] BOYSEN, N., FLIEDNER, M., SCHOLL, A.: Assembly Line Balancing:
Which model to use when? International Journal of Production Economics,
2008, Vol. 111, pp. 509-528
[5] PATTERSON, J. H. AND ALBRACHT J. J.: Assembly-Line Balancing:
Zero-One Programming with Fibonacci Search. Operations Research,
1975, Vol. 23, pp. 166-174
[6] SCHOLL, A. AND BECKER, C: State-of-art Exact and Heuristic Solution
Procedures for Simple Assembly Line Balancing. European Journal of
Operational Research, 2006, pp 666-693
[7] WHITE W. W.: Comments on a Paper by Bowman. Operations Research,
1961, Vol. 9, pp. 274-276

You might also like