You are on page 1of 2

The Practical Problems of Assembly Line Balancing

A Communication from R. N. Basu, MSc, CEng, MIMechE, MIProdE, SrMem AIIE, Area Manager, Industrial Engineering Division, The Singer Company (UK) Ltd.*

The assembly line balancing problems in simple form have been analysed by a number of researchers, and about ten or so major algorithms have been developed for determining either the exact or an approximate balance1. The problems in simple form attempt to minimise the number of operators for a given cycle time or to minimise the cycle time for a given number of operators (mostly the former) with the assumptions that the element times are deterministic, that an element may be performed at any station, that only one uniform product is made, that the operator performance does not vary during the working period or from station to station, that group size or cycle time has no significance on line performance, and that balancing is independent of learning of operators and the quality level of parts and equipment. In practice, however, this is never so. Although a good deal of time may be saved in line balancing by most of the algorithms when compared with manual balancing, and an optimum or near optimum balance with the above assumptions may be a better starting basis than random or disjointed balance, very often it is found under practical conditions thatstoichastic variables and other extraneous conditions in the line override the merits of a simple and deterministic case. It may be possible to obtain a perfect balance for a problem in simple form, but it is seldom possible to maintain it. Then, one wonders, what is the purpose of all the efforts for a perfect balance, when a workable solution is obtainable by manual methods? A number of authors give lip service to the practical problems, and some have investigated in certain depth one aspect of the problems. The practical problems of line balancing may be briefly described under appropriate headings.

may alter. As a result, at a later date after the implementation of a balanced condition, the time values of each station particularly for the control station, may not be the same as the designed values. The variations of element time and operator performance may result in system losses (due to variation of element time and operator performance at each work station) and balancing losses (due to waiting time caused by differences in operation time between work stations). Moodie & Young4 and Mansoor5 have proposed methods of balancing lines with variable work element times, which, while convenient for their models, seem to be unsuitable for all practical conditions.

Model Mix
The model mix balancing and sequencing problems occur in the situtation where the producer makes more models than he has assembly lines and finds it desirable to keep each model in production rather than to produce batches of each intermittently for inventory. The different models usually require different amounts of assembly work causing an uneven distribution of work along the line. This problem is more predominant in automobile assembly lines. A survey by Kilbridge & Wester6 shows that the American Automobile Industry wasted about 25% of assemblers time through uneven work assignment. Thomopoulos7 has proposed a method of tackling line balancing and sequencing problems for mixed model assembly. There is another situation which may be termed as multi-model assembly, where various models are produced in different lines at any time, and models can be changed over to other lines after producing batches of each. This type of multi-model line can be analysed as a series of single model lines. The assembly of sewing machines, electronic equipment and similar consumer products is usually in the latter group. The training/learning costs at each changeover can be reduced and flexibility of manufacturing facilities in each line can be maintained by maximising the number of identical stations for each model.

Operator Performance
Ignall1 states in his review 'Performance times are usually variable and this has an effect on the line's operation. In fact, if a man at the beginning of the line does not finish a work piece, then no one after him can work on it. If he stays with it after the cycle time is up, then the man after him is under pressure to work faster than usual to gain back the time. If he does not, the problem is passed on to the next man. On the other hand, if a man near the end of the line is late, the consequences are less disturbing, since fewer people follow him'. The performance times of various work stations for any work cycle vary, as regards operator performance because of two factors: (a) Operators do not work at the same pace and (b) Each individual operator varies his pace throughout the working period. The output patterns of operator performance as obtained by Dudley2 and Van Beek3 comply with the hypothesis.

Training/Learning
This is the factor involved in the assembly operators' learning/training to perform their tasks at an acceptable pace. The assignment of elements to work stations and the cycle time should be influenced by the property of learning curve, where the cycle time decreases exponentially with repetition of cycles and subsequently the curve flattens out. Learning cost depends on the length of the task or the cycle time, and the frequency of the changes in model, method and operators. The cost of learning or training may be reflected in two ways (a) loss in production and imbalance due to the slow pace of the inexperienced operator or the new work assignments at assembly stations, and (b) the whole group of assemblers may have to be paid during learning/training period at high guarantee level regardless of production. Kilbridge & Wester8 have put forward a formula for learning cost considering three phases (i) the pace ultimately attained (ii) the initial learning time, and (iii) the recurring learning cost.
Mr. Basu has recently taken up an appointment at A. T. Kearney Ltd., Management Consultants, London, SW1. 369

Variable Work Elements


The element times are determined based on standard or observed conditions. In practice, due to non-uniformity of product quality and specifications, variation in manufacturing facilities, frequency of rework and setting etc. element times cannot be treated as having constant values even for fixed operator performance. Frequently, in course of time by self-adaption of operators (not learning) and modification of methods, element times (and station times for that matter)
The Production Engineer- October 1973

The quantitative aspect of the learning/training, largely depends on extensive empirical results. At least in multimodel assembly the learning period caused by the changeover of models can be greatly reduced by designing as many number of identical or similar work stations as possible for different models.

managers lose interest in line balancing. Assembly operators may be expected to be more co-operative in accepting a better balanced condition and maintaining it under standard condition as much as possible to increase production, when the payment system allows a fixed or less fluctuating allowance for probable non standard elements.

Waiting Time
Waiting time may be defined as the time during which the worker is available but is prevented from working. Waiting time may be analysed under Department not responsible (DNR) and Department responsible (DR) 9 . As regards assembly lines, DNR waiting time is actually individual station idle time due to imbalance, variable element times, and operator performance. DR waiting time is irregular and caused by extraneous conditions viz. shortage of parts, tools and equipment breakdown, excessively bad quality of parts, high reject rates etc. If all the extraneous factors are properly analysed and the significance and probability of occurrence established from actual data, a mathematical model may be constructed for proper feedback and control of a balanced assembly line.

Attitude of Operators
The effectiveness of any system, however efficient, cannot be achieved without good industrial relations. The trade unions use their organised strength to resist any innovation that, in their view, constitutes a threat or a shabby deal. Assembly lines in all mass producing industries all over the world, particularly automobile industries, have always been on the top of the league table in terms of grievances and industrial disputes. The following lines are quoted from Business Week Magazine, July 1964, 'A rash of auto strikes has occurred throughout the industry because of dissatisfaction over standards . . . . The grievances over so-called cycle timethe time allowed for assembly workers to install car components . . . may be harder to resolve by compromise than past production standard disputes'. The problem is still affecting industries. A recent analysis10 has revealed that in a large light engineering company, to cite as an example, about 60% of industrial disputes occurred in assembly lines. The problem of human aspects or operator attitudes is vast and complex. The technique of line balancing will in no way be a total solution of the problem, but can form a sound basis for evaluating the grievances and negotiating in a rational and intelligent manner. Henry Ford II once stated, 'Machines alone do not give us mass production. Mass production is achieved by both machines and men. And while we have gone a long way toward perfecting our mechanical operations, we have not successfully written into our equations whatever complex factors represent Man, the human element'. One should not be led to believe that the assignment of work elements to work stations solves the line balancing problemthis is just a part of the overall problem. The practical problems, as outlined briefly above, are the real crux of the system. The approach of investigation must be to analyse as a total system. 'Knowing whether these problems occur together is important, because analysing them separately is not sufficient if they do. Using the "sum" of the results obtained by analysing each problem separately as the procedure for the combined problem can be a dangerous pastime'1.

Group Size
One objective of line balancing, in theoretical sense, is to minimise the group size or number of work stations for a given cycle time. Cycle time is usually determined by the expected output rate. The required production level may be achieved either by a small number of lines with a large group size and low cycle time or by a large number of lines of a small group size and high cycle time. Therefore, where there is a choice, besides theoretical factors some practical and empirical considerations should be made in deciding a group size or cycle time. It has been found that lower cycle time reduced the learning time, but too much repetition of small work content also causes monotony on the operator and consequently high labour turnover. The smaller group size is better for group effectiveness and control, but higher cycle time usually slows down the pace of work. Often, a compromise is made, without adequate quantitative basis, by partitioning the assembly line in smaller groups without altering cycle time. It is indeed very difficult to set up a generalised rule, but for any particular situation a basis may be formulated to near optimise the group size. The current trend towards job enrichment is an important contributing factor for determining a group size.

Repair Operations
In order to maintain a smooth flow of production on conveyorised asembly lines, it is often a general practice to allow a number of operators 'off the line' to carry out repair operations. Although the frequency and nature of repair operations vary a great deal, it may be possible to measure the work content of repair operations by post analysis. Once a scientific basis is established (for example, a multi-factor system related to the total work content of basic assembly tasks, key mechanisms requiring repair, and frequency of occurences) the requirement of repair operators for a new model can be determined. Any arbitrary method of deciding the number of repairmen may undermine the benefits gained by line balancing.
REFERENCES 1. Ingall, E. J. A Review of Assembly Line Balancing, 'Journal of Industrial Engineering/ Vol. 16 No. 4 (1965). 2. Dudley, N. A. Output Pattern in Repetitive Tasks, Institution of Production Engineers Journal, Vol. 37, pp. 187-191, 257-264, 303-313,382-384 (1958). 3. Van Beek, H. G., Working on Assembly Lines, 'Occupational Psychology,'Vol. 38, pp. 161-172 (1964). 4. Moodie, C. L. 8- Young, H. H., A Heuristic Method of Assembly Line Balancing for Assumptions of Constant or Variable Work Element Times, 'Journal of Industrial Engineering,' Vol. 16, No. 1 (1965). 5. Mansoor, E. M., Assembly Line BalancingA Heuristic Algorithm for Variable Operator Performance Levels. 'Journal of Industrial Engineering,' Vol. 19, No. 12 (1968). 6. Kilbridge, M. D. and Wester, L, The Assembly Line ModelMix Sequencing Problem, Proc. of 3rd International Conference on O.R., Oslo (1963). 7. Thomopoulos, N. T., Line BalancingSequencing for Mixed Model Assembly, 'Management Science,' Vol. 14, No. 2 (1967). 8. Kilbridge, M. D. and Wester, L, An Economic Model for the Division of Labour, 'Management Science,' Vol. 12, No. 6 (1966). 9. BS3138, Glossary of Terms in Work Study, British Standards Institution, London 1969. 10. Roche, A. J., An Analysis of Control and Communication in the Business Firm, Unpublished PhD Thesis, Strathclyde University (1972). The Production Engineer- October 1973

Payment System
The payment system practised in an assembly line influences the effectiveness of line balancing. Very often a policy allowance9 is added to cycle time to provide a satisfactory level of earning and allowed time9 is issued for payment purposes. The decision of policy allowance should be related to balance delay rather than some percentage or fixed increment on cycle time alone. If a straight proportional incentive scheme is properly maintained for assembly line payment, the benefits of line balancing will be fully realised. However, in practice, most frequently non standard elements {viz. DR waiting time, disputes, etc.) affect the system, and
370

You might also like