For many companies, the knock on effects of the “Great Recession” of 2008 are still being felt. The U.S. economy finally appears to be gaining sustainable traction, with the Eurozone recently exiting its worst recession in 40 years. For companies that still have manufacturing facilities in the West and for the people who manage them, what was always a difficult job is becoming even more challenging.
But there appear to be some developments and trends that should give us reason to hope. One of them is the growing phenomenon of “re-shoring,” in which Western manufacturers bring back in-house activities that were previously done overseas. This trend is still in its relative infancy. The number of companies in the US that have brought back some manufacturing element of their operations is thought to be well under 100. But the trend appears to be growing.
One study performed by a Massachusetts Institute of Technology (MIT) professor showed that of 108 US industrial companies, 14 percent had concrete plans to bring some of their manufacturing operations back to the States. Another study by the Boston Consulting Group (BCG) found that in a survey of 106 companies with annual turnover of at least $1 billion, 37 percent either had prepared plans or were seriously contemplating reshoring some of their operations.
With rising wages in China and stagnating wages in the US, these trends should only continue. But what is the best way for an operations manager to take advantage of this “manufacturing renaissance”?
Conventional wisdom holds that the best way to manage a facility is by having balanced production lines. That is to say, all stations should have equal or as close to equal average operating times. We can imagine this as a perfect or ideal state, where raw materials go in and the finished product comes out of the other end with no hold-ups and no variation. But research dating back to the 1960s shows that unbalanced production lines can sometimes outperform their balanced counterparts. This was a great finding, as it could take advantage of the inherent differences in human performance. Add to this the destabilising effects of downtime due to machine failure, and we can see that manned assembly lines in the real world can achieve “notional” balance only, with considerable expenditure of resources. It was found that by having a fast worker in the centre of line, and slow workers at the front and back, output could be increased by 1.37 percent. Not an earth-shattering increase in production, but enough to possibly call into question the logic of expending resources on balancing one’s production facility.
A simulation study was undertaken into the advantages, or disadvantages, of unbalanced lines. The lines studied were manual unpaced production lines. In this type of lines each operator is free to complete some task at their own pace. They then pass the work piece down the line to the next station, where another operator performs another task. In-between stations there is usually some space to store work-in-process (WIP). This allows for more even and regular production, since it reduces the number of occasions when a worker has to stop because they are starved for work or there is no room for them to eject their work piece.
Our study looked at lines having both five and eight stations. In order to make the study more realistic, station breakdowns and repairs, an unavoidable feature of assembly lines were incorporated into the simulations. The lines studied were deliberately unbalanced in two ways; 1) mean worker completion times (MT) differed among the workers and 2) the amount of inter-station buffers were varied. Random breakdown of the stations was programmed to occur during the operation of the line in order to be able to see the interaction between imbalance and downtime due to machine failure and repair.
The above two sources of imbalance were then combined into various patterns of imbalance. That is to say, four different configurations of worker allocation were used:
- 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: /\).
The above four patterns of operator imbalance were then combined with four patterns of buffer imbalance. These were:
- Concentrate the available buffer capacity nearer the end of the line. This policy displays an increasing order of buffer capacity (/).
- Concentrate the available buffer capacity nearer the middle of the line. This policy reflects an inverted bowl shaped buffer size sequence (/\).
- Concentrate the available buffer capacity nearer the beginning of the line. This policy depicts a decreasing order of buffer capacity (\).
- No concentration of buffers: this policy was further broken down into three sub-policies:
- A general allocation (nearly balanced).
- Alternating buffer size between high and low a1ong the line (a zig-zag configuration-/\/\).
- Positioning smaller buffer capacity towards the centre (a bowl-shaped pattern \/).
Three different performance measures were used and analyzed. They were; 1) throughput rate (TR), 2) average worker idle time (IT) and 3) average buffer level (ABL). The goal was to see if performance could be improved by unbalancing a line, and if not, whether the performance of an unbalanced line was dramatically inferior to that of a balanced line.
The results of the investigation showed that in terms of throughput and idle time, there was not generally a difference between the performance of balanced and unbalanced lines suffering breakdown. In other words, balancing the line was not bringing any added advantage to performance. This would call into question the logic of expending time, effort and resources on bringing a production facility into balance, since the improvements obtained, if any, might not justify the cost.
However, when it comes to average buffer levels, the case for running an unbalanced line becomes even stronger. Two highly favourable patterns were identified, namely; 1) a decreasing MT sequence (operators get slower as you go down the line- \), combined with allocating more buffers to the end of the line (/) and 2) a decreasing MT allocation (\), combined with putting more buffers in the center of the line (/\). Significant improvements in these lines undergoing breakdown, as compared to a balanced line, were found to exist from using the above two patterns, with the possibility of an almost 80 percent reduction in average buffer levels for the best case. This represents a huge improvement in operational efficiency.
Since this study used simulation of a limited number of scenarios, managers should note that there is an almost infinite number of possible configurations combining differing average station work times and buffer sizes in the design of a line, and not all of them will lead to better performance. The particular way you unbalance your line is important. Nevertheless, we know that humans do not all work at the same rhythm, and we know that our production lines or elements within them will grind to a halt sometimes, so why not find out how best to design in these facts as efficiently as possible, in order to not only maintain productivity, but also, in some cases, to improve performance?
References
“Reshoring manufacturing,” The Economist, January 19th, 2013. Accessed at: https://www.economist.com/special-report/2013/01/17/coming-home
“Some Firms Opt to Bring Manufacturing Back to U.S.,” by James R. Hagerty, The Wall Street Journal, July 18th, 2012. Accesed at: https://www.wsj.com/articles/SB10001424052702303612804577533232044873766
"The Effect of Some Design Factors on the Efficiency of Production Lines with Variable Element Times," by F.S. Hillier and R.W. Boling, R.W, Journal of Industrial Engineering, Vol.17, No.12, pp. 651-658 (1966).
