A human approach to improving production line performance – Manufacturing Digital

One of the more popular strategies for manufacturers in the last two decades has been to outsource their production, or transfer their production facilities to low labour-cost countries. That tide is now beginning to turn, however, and according to the Boston Consulting Group, some US industries are bringing back their factories as wages rise abroad and fall at home, coupled with quality and environmental control becoming ever more important to consumers and governments.

So, if the trend is to ‘reshore’, production lines, how best to design production?Where can costs be cut and efficiency gained? How can line managers think best about line design in an environment, where demand fluctuates, machines breakdown and human beings work at different paces and with diverse efficiency? Is there anything new that can be done after the cost-cutting, the streamlining and the mechanisation of the last three decades?

Our research suggest improving the performance of facilities using un-paced production lines in which workers are allowed to work at their own speed, passing on work pieces to the next station, without the aid of an automated moving line.These types of lines usually have some storage space allotted in between the individual stations, called a buffer, for the storage of work-in-process. This allows for smoother production by avoiding the stoppage of work due to an operator not having a work piece available (known as ‘starving’), or not being able to pass on a piece due to there being no room for it after the station (known as ‘blocking’). The importance of these buffers should not be underestimated since they allow workers to process items independently of each other.

For a long time, it has been assumed that a ‘balanced’ line is the best way to improve efficiency,meaning that each work station takes the same amount of time to perform a given task, before being transferred to the next operator. In reality, lines like these are impossible to achieve for several reasons; one being the type of work being done. Sometimes, because of pre-existing technological or precedence restrictions, it might be impossible to break down an individual job into a number of simple tasks in which each one has the same average completion time. Another reason has to do with human nature. Different people have different levels of training, education, skill, ability, and motivation.

Another point, sometimes overlooked, is that the variation in the time taken by an operator to complete a task may be quite considerable. People in general cannot perform a task or a series of tasks differing in complexity and specificity again and again at exactly the same speed over a length of time. Research has shown that the average time to complete a task can vary by up to 66 percent over a day. This variation is measured using the coefficient of variation (CV). Add to this the fact that lines breakdown, or an operator may need to stop unexpectedly, and we see that aiming for a balanced manual line, where a product moves unfailingly from one step to the next with mechanical precision, becomes increasingly unlikely.

In a nutshell, the amount of time it takes for different operators to complete the exact same task will never be equal. Contrary to popular opinion, this is not necessarily a bad thing. Studies have shown that manual lines can be from 25 to 75 percent more efficient than fully automated lines.

In spite of the fact that there are these natural fluctuations, companies make substantial investments in time, money and effort to bring their lines into balance, believing that this is the best way to improve efficiency. But is a balanced line desirable, or even possible? Past studies have shown that sometimes unbalanced lines can actually outperform a perfectly balanced line. But how can this be achieved?

We constructed a simulation investigation into manual production lineshaving both five and eight stations to find out how they performed in terms of throughput, idle time and average buffer level, as well as incorporating station breakdown, so as to mirror real life operating conditions as closely as possible. We assumed that workers fell into one of three classifications according to their mean service time (MT), that is to say an operator could be fast, medium, or slow. Next, four configurations of employee arrangement were looked at: 

• The fastest worker is located at the front of the line, followed by slower and slower workers (an ascending order: /).
• The fastest worker is placed at the end of the line, preceded by progressively slower workers (a descending order: \).
• The fastest worker is stationed in the middle, with workers getting progressively slower as one moves outward in both directions (a bowl shape: \/).
• The slowest worker located in the middle, with workers getting faster as one goes both towards the front and back of the line (an inverted bowl shape: /\). 

Another element that was looked at was the percentage degree of imbalance; that is to say, the difference between the fastest and slowest worker. Three degrees of line imbalance were looked at; two percent (slight), five percent (medium), and 12 percent (high).

We also incorporated worker variability into our study, and four different policies were looked at regarding the placement of workers with different levels of CV:

• Separating the variable workers from one another by steadier workers
• Assigning steadier workers to the line centre, i.e. a bowl arrangement
• Operators having medium variability are allocated to the middle of the line
• The most variable workers are assigned to the centre of the line centre – an inverted bowl arrangement

Regarding the classification of the variability of an individual worker, it was assumed that CV = 0.08 was relatively steady, CV = 0.27 was medium and CV = 0.50 was highly variable.

Finally, inter-station buffers were set at one, two and six units.

The goal of our study was to find out which pattern of worker allocation (if any) would provide the greatest improvement in performance by way of:

1.       An increase in the throughput rate (TR)

2.       A reduction in the overall amount of worker idle time (IT)

3.       A decrease in the average buffer level (ABL)

Some of our results were surprising, and might run counter to the prevailing wisdom concerning production line management. We found that the best pattern resulting in increased production rates and lower average worker idle times was an inverse bowl arrangement of worker MT (/\), combined with a bowl arrangement for CV (\/), that is to say, the slowest and steadiest workers in the middle with faster more variable workers at both ends of the line. The biggest improvement, in certain cases, when compared to a balanced line, was nearly a three percent increase in output and an almost eight percent reduction in average idle time. These improvements might not appear to be significant, but when taken over the expected useful life of a production line, the higher revenue and increased savings could prove to be substantial. 

As for ABL, it was found that the best pattern was a descending order (\) of worker service times (i.e. workers get progressively faster), combined once again with a CV bowl (\/) pattern (steadier workers in the centre, with more variable workers at both ends of the line). It was observed that on certain occasions a reduction in average buffer level of almost 76 percent was possible, a truly vast improvement in performance.

Our results lead us to believe that line managers who choose to benefit from the effects of unbalancing a line are therefore faced with a decision. Would they like to increase output / lower idle times, or would they like to reduce average buffer levels? This decision will most likely be informed by the type of operating environment that they find themselves in. Is product demand high? Is this a technical item in which highly skilled workers, with correspondingly high wages are used, and therefore idle time is of primary concern? In these cases an inverse bowl arrangement of workers’ average processing times (/\), combined with a bowl allocation (\/) of worker variability would most likely provide the best benefits.

On the other hand, managers might find themselves operating in an industry with extremely short product life cycles, or perishable goods. Here, the goal might be lean buffering, and correspondingly, a descending order of workers regarding mean processing time (workers get faster and faster as you move down the line), in conjunction with a bowl allocation of worker CV (steadiest workers in the centre, more variable workers at the front and back) would be the preferred arrangement.

As with all investigations of this type, it should be noted that only a few of the almost unlimited number of possible alternatives for unbalancing a line were examined. Caution should be taken in that if a line is imbalanced in the “wrong” way, adverse performance could be the result, and that is to say, lower production, higher idle time and higher buffer levels might be seen. However, the prospect of being able to allow for the natural rhythms and fluctuations of human work and the inevitable downtime that lines suffer while also improving efficiency, makes both financial and moral sense.

Our magazine is now available on the iPad. Click here to download it