Anda di halaman 1dari 4

A Model for Assembly Line Balancing Problems

Shwetank Avikal, P. K. Mishra, and Rajeev Jain

Abstract-- The assignment of tasks to work stations in an assembly environment has been the matter of concern to researchers so that the efficiency of the system can be enhanced. An assembly line may be define a flow-oriented production system where the productive units performing the operations, referred to as stations, are aligned in a serial and sequential manner. The proposed work deals with a methodology for the problem of assigning tasks to work stations in assembly lines, where the cycle time is variable. A model has been proposed to assign the tasks to the work station in assembly lines, where variation of cycle time is imposed by the management and then Analysis of Variance bas been applied. Index Terms-- Assembly line, cycle efficiency, heuristics, line balancing.

II. NOMENCLATURE

P CT TT IT
WS LE NT MaxTT MinTT MNF RPW RPW & MTT

= problem = cycle time = task time = idle time = work station = line efficiency = number of task = maximum task time = maximum task time = max number of followers = rank positional weight

I. INTRODUCTION

N assembly line balancing problem is how to streamline the assembly activities, so that the total assembly time required at each workstation is approximately the same. The cycle time of the assembly line is determined by the workstation with maximum total assembly time. Originally, assembly lines were developed for a cost efficient mass production of standardized products, designed to exploit a high specialization of labor and the associated learning effects. Since the times of Henry Ford, product requirements and thereby the requirements of production systems have changed dramatically. In order to respond to diversified customer needs, companies have to allow for an individualization of their products. This individualization forced the management to design the assembly line in minimum time by which the setup time of the line can be minimized. The remainder of the paper is arranged as follows. In Section II, nomenclature is given. Literature review is presented in Section III. Section IV presents the proposed model for assembly line balancing problem solving. The Results of proposed model is presented in Section V. Finally conclusions are given in the Section VI.

MNF & MTT = y yi N T T F S2


=

rank positional weight & maximum task time maximum no of follower and maximum task time
observation

= = = = = =

ith response total number of observations sum of all observation average of all observations fisher ratio
variance

III. LITERATURE REVIEW The problem of assembly line balancing for ideal assignment of tasks to the workstations has been addressed by many researchers for different types of assembly line balancing problems using different types of heuristics and optimizations techniques. Bautista. et. al. [1] proposed a time and space constrained assembly line balancing problem. In this approach they have used Ant algorithm for example given a source of food, an ant colony tends to naturally find out shortest path between the source and its nest via two processes. Firstly, the ant deposit pheromone on the route, and secondly, they normally follows the path on which they find more previous deposited pheromone. Here he proposed different types of line balancing problems line feasibility problem, single objective problems, and multi objectives problems. Sofia. et. al. [2] proposed a genetic algorithm based approach to the mixed model assembly line balancing of type two. In this model he has used genetic algorithm for line balancing. He gradually reduced the cycle time by one unit and tried to solve the problem. If feasible solution is found

Shwetank Avikal, P. K. Mishra, and Rajeev Jain are with the Department of Mechanical Engineering, Motilal Nehru National Institute of Technology, Allahabad-211004, India (e-mail: shwetank@mnnit.ac.in, pkm@mnnit.ac.in, rjain@mnnit.ac.in).

978-1-4673-0455-9/12/$31.00 2012 IEEE

then he again reduced cycle time otherwise he increased cycle time by one unit and solve the problem. He used this model for parallel workstations. Toksari. et. al. [3] proposed simple and U-type assembly line balancing problem with learning effect. He has introduced the learning effect into assembly line balancing problems. In many realistic settings, the produced worker or machine develops continuously by repeated the same or similar activities. Therefore, the production time of product shortens if it is processed later. They show that polynomial solution can be obtained for both simple assembly lines balancing problem and U-type line balancing problem with learning effect. Fleszar et. al. [13] proposed a new heuristic algorithm and new reduction techniques for the type 1 assembly line balancing problem. The new proposed heuristic is based on the well-known Hoffmann heuristic and builds solutions from both sides of the precedence network to choose the best. The reduction techniques aim at augmenting precedences, conjoining tasks and increasing operation times. Boysen. et. al. [6] defined different types of models and explain that which model to use when for assembly line balancing problems. That modal was based on the number of model like single model, multi models and mixed model, and based on line control like paced, unpaced asynchronous & unpaced synchronous and frequency like first time installation & reconfiguration and level of automation like manual lines and automatic lines. Amen [11] presented his work concerned with costoriented assembly line balancing. In his work First a short problem description is presented. After that a classification of existent and new heuristic methods for solving this problem is given. He also proposed a new priority rule called best change of idle cost is proposed. This priority rule differs from the existent priority rules because it is the only one which considers that production cost are the result of both, production time and cost rates. Tetsuo et. al. [4] have presented a management design approach to assembly line system by considering both the cost and lead time under demand fluctuations. First, a design procedure based on the 2-stage design method is presented for the assembly line system with and without stopper in the installation case of production facilities. Next, the strategic relation of economic traffic and lead-time is considered by introducing the production matrix table. Finally, a positioning strategy is discussed for a case in which the lead-time is restricted. Albert. et.al. [7] proposed the technique of balancing assembly line with skilled and unskilled workers. He presented the process of rebalancing the line at motorcycleassembly plant. The company found it necessary to rebalance its line, since it needs to increase production in the spring and summer months. The main characteristics of the problem are: (i) the company hires temporary staff, which needs more time to carry out their tasks than permanent workers; (ii) there must be at least one skilled employee working alongside an unskilled one; and (iii) different task groups are incompatible with each other (clean-hands tasks and dirty-hands tasks). The goal is to minimize the number of temporary workers required, for a cycle time and the team of workers on staff. The problem is modeled as a binary linear program (BLP) and

solved optimally by means of the ILOG CPLEX 9.0 optimizer. The solution provided, namely 12 permanent workers (skilled) and two temporary workers (unskilled), is an improvement on the solution implemented by the business, which involved 12 permanent workers and four temporary workers. This paper presents a model that yield minimum cycle time while using the proposed algorithm. IV. PROPOSED MODEL In proposed model it has been assumed that the variation in cycle time is very small (few units), Change in no of work station can be acceptable, Change in production rate can be acceptable, every work station have a same and constant cycle time. In this proposed model a term used, named variation and it depends upon the assembly line, has been added and it was found out on the bases of market requirements and it should vary within 15 percent of the cycle time because variation up to 15 percent is mostly acceptable for assembly line balancing problems. This variation has been reduced and then added to the cycle time and got a series of the cycle time on which the assembly line can run, with a view that minimum cycle time should not be less than any task time. Here first of all, the problem has been solved and then the efficiency of all the heuristic is calculated for the first cycle time i.e. C1 and then increase the cycle time by one unit and again solve the problem and calculate the efficiency of the all heuristics.

Fig I Proposed assembly line balancing model

The system uses different heuristics that comprises of different priority rules. All the heuristics were used to evolve idle time and cycle time.
Priority Rules and Heuristic:

Optimal solution of the ALB problems having typically been obtained by following approaches. 1. Maximum task time 2. Maximum number of followers 3. Maximum, ranked positional weight 4. Combination of Maximum task time and Maximum number of followers

5. Combination of Maximum task time and Maximum, rank


positional weight
Analysis of variance (ANOVA):

This method was developed by Sir Ronald fisher in the 1930s as a way to interpret the results from agriculture experiments. ANOVA is not a complicated method and also has a lot of mathematical beauty associated with it. ANOVA is a statistical based, objective decision-making tool for detecting any difference in average performance of groups of items tested. The decision rather than using pure judgment, takes variation into account.

Fig II Precedence diagram of simple assembly line

2 Sy =

i =1
kA

( yi T )
N 1

Input Data:

(1)
2

TABLE I TASK AND TASK TIME

2 Sy =

( Ai T )
i =1

Task TT

1 14

2 5

3 6

4 7

5 8

6 5

7 10

8 5

kA 1

(2)

Cycle time = 17 sec No. of theoretical work stations = Sum of task time/ CT = 4
TABLE II VALUE OF CYCLE TIME, IDLE TIME, NO. OF WORKSTATIONS AND LINE EFFICIENCY FOR ALL HEURISTICS

F test for variance comparison: Statistically, there is a tool which provides a decision at some confidence level as to whether these estimates are significantly deferent. This tool is calculated an F-Test, named after Sir Ronald Fisher, A British Statistician, who invented the ANOVA method. The F-Test is simply a ratio is sample variance [16].
F=
2 Sy 1 2 Sy 2

Heuristic MTT MNF RPW MNF & MTT RPW & MTT

CT 17 16 18 18 18

IT 8 4 12 12 12

WS 4 4 4 4 4

LE% 88 93 83 83 83

(3)
It can be seen (tables II) that the optimal idle times by different heuristic are not same and the cycle efficiency is also not same. It is found that maximum task time method yields optimum solution at 17 sec cycle time, maximum no of follower gives optimum solution at 16 sec cycle time, rank positional weight give optimum solution at 18 sec cycle time, combination of maximum task time and maximum no of follower gives optimum solution at 18 sec cycle time and combination of maximum task time and rank positional weight gives optimum solution at 18 sec cycle time. By this we can say that first two heuristics gives different solution both and last three gives same optimal solution. But the solution by maximum no of followers is optimum solution and the assignment schedules of all heuristics are different. The table shows that idle time, number of work stations and cycle efficiency with their cycle time of all heuristic. Ten different problems have been solved and their results are given in TABLE III.

When this ratio becomes large enough, then the two sample variances are accepted as being unequal at some confidence level. F tables which list the required F ratio to achieve some confidence level are provided in the appendixes. To determine whether an F ratio of two sample variance is statistically large enough, there pieces of information are considered. One, the confidence level necessary: two, the degree of freedom associated with the sample variance in the numerator: and there, the degree of freedom associated with the sample variance in the denominator. Each combination of confidence, numerator degree of freedom, and denominator degree of freedom has an F ratio associated with it [16]. V. RESULTS The present work deals with single model assembly line balancing problems with an objective to propose a efficient heuristic to find out the optimal solution in the terms of minimum idle time with minimum cycle time and minimum number of work stations. The model is illustrated through an example whose precedence diagram is given as:

TABLE III COMPARISON OF LINE EFFICIENCY FOR SAMPLE PROBLEMS

[3]

P1 P2 P3 P4 P5 P6 P7 P8 P9 P 10

MTT % 93.3 85.2 85.2 95.2 96.5 88.2 93.7 92.9 97.9 98.0

MNF% 93.3 85.2 69.7 95.2 86.2 93.8 93.7 86.4 90.4 82.5

RPW % 93.3 85.2 85.2 95.2 92.8 83.3 93.7 92.9 90.4 82.5

MNF & MTT % 93.3 85.2 69.7 95.2 86.2 83.3 93.7 92.9 90.4 82.5

RPW & MTT % 93.3 85.2 85.2 95.2 92.8 83.3 93.7 92.9 90.4 82.5

[4]

[5]

[6]

[7]

[8]

After solving ten sample problems the mean of their efficiency, standard deviation has been calculated and then rank has been assigned to all heuristic.
TABLE IV FINAL RESULTS OF THE PROBLEMS

[9]

[10]

MTT % Mean Efficiency Standard Deviation Rank 92.55 4.823 1

MNF % 86.81 7.434 5

RPW % 88.62 4.834 3

MNF & MTT % 87.18 7.658 4

RPW & MTT % 89.39 4.807 2

[11] [12]

[13]

[14]

Analysis of variance (ANOVA):


2 Sy = 22.49735 1 2 Sy = 37.74149 2

[15]

[16]

M. Duran T., Simple and U-shaped assembly line balancing problem with a learning effect, Applied mathematical modeling, vol.32, P. 29542961, 2007. Testuo Yamada and Masayuki M. A management design approach to assembly line system, International journal of production economics, vol.84, P. 193-204, 2002. Joaquin B and Jaine C., Minimizing work overload in mixed-model assembly line, International journal of production economics, vol.112, P. 177-191, 2007. Nil Baysen and Malte F., Assembly line balancing: which model to use when, 2007, International journal of production economics, vol.111, P. 509-528. Albert C. and Rafael P., Balancing assembly line with skilled and unskilled workers, Omega the International journal of management science, vol.26, P. 1126-1132, 2006. D. D. Sheu and Shiao-Lan P. Assessing manufacturing management performance for notebook computer plant in Taiwan, International journal of production economics, vol.84, P.215-225, 2002. Joaquin B. and Jordi P. A dynamic programming based heuristics for the assembly lone balancing problem, European journal of operation research, vol.194, P. 787-794, 2008. Asar Khan and Andrew J. Day, A knowledge based design methodology for manufacturing assembly lines, Computer and industrial engineering, vol.41, P. 441-467, 2001. Amen, Heuristic methods for cost-oriented assembly line balancing: A survey, Int. J. Production Economics, vol. 68, p. 1-14, 2000. Mendes et.al. Combining heuristic procedures and simulation models for balancing a PC camera assembly line, Computers & Industrial Engineering, vol. 49, p. 413431, 2005. Kfleszar et.al. An enumerative heuristic and reduction methods for the assembly line balancing problem, European Journal of Operational Research, vol. 145, p. 606620, 2003. Yamada T and Matsui M., A management design approach to assembly line balancing systems, International journal of production economics, vol.84, P. 193-204, 2003 J. Rubinovitz and G. Levitin, Genetic algorithm based line balancing, International journal of production economics, vol.41, P. 343-354, 1995. Phillip J. Ross, Taguchi Techniques for Quality Engineering, Edition II, TMH Publishing Company Ltd., New Delhi, 2005.

F test for variance comparison (90% confidence level): F = 0.596091 VI. CONCLUSION This paper presents a model for solving the assembly line balancing problem by varying the cycle time. The results have compared with the reported heuristics and priority rules. It is found that the results obtained by using proposed heuristics are comparable with the variation of cycle time. The performance measures used for comparison are efficiency, idle time of work stations and cycle time. It can be seen (see Table IV) that the maximum task time heuristic gives maximum efficiency and has first rank. The ANOVA and F test also validate the result. Thus, it can be said that heuristic of maximum task time performs better than other heuristic and it yields line efficiency. VII. REFERENCES
[1] Joaquin Bautista and Jordi P., Ant algorithm for a time and space assembly line balancing problem, European journal of operation research, 2006, vol. 177, P. 2016-2032. Ana Sofia Simaria and Pedro m. A genetic algorithm based approach to the mixed model assembly line balancing problem of type-2, 2004, Computer and industrial engineering, vol. 47, P. 391-407.

[2]

Anda mungkin juga menyukai