The Hidden Ways Manufacturing Can Hurt End-of-Line Processes
At a manufacturing site the focus is often on the production process itself. As a result, up- and down-stream activities seem to receive less attention. If the end-of-line function is in the spotlight, it’s probably because they impacted production.
This departmentalised mindset is detrimental. It neglects the fact that an entire site must function together. It’s bad for culture, degrading the inputs that a team makes to the process; it overlooks improvement opportunities, by not giving due attention to all areas; and worst of all, it can impact the overall output of the plant.
A manufacturing site in Victoria conducted an analysis on the impacts between production and end-of-line processes. It sought to better understand how the areas affect each other and to identify areas for improvement.
Overview of End-of-Line Processes
The process involves several production lines feeding 2 stretch wrappers. The wrappers feed an automated forklift system (AGV, automated guided vehicle), which, depending on the product will store or stage finished pallets. These automated forklifts also handle packaging storage and supply for the plant.
This production week was suitable for review because it was intended to strain the end-of-line processes, allowing flaws to be magnified. It is not necessarily indicative of a standard production week.
Production Distribution by Day
Charts 1 through 3 show what the plant intended to make over the 5 day week. You can see that both Monday and Tuesday had greater than 20% of the volume. Proportionally, those days were overloaded.
Chart 1: Planned production distribution
There is an even more prominent skew on ambient production (which is directly staged, rather than stored) with Monday and Tuesday accounting for 27% and 29% of the volume, respectively.
Chart 2: Planned production of materials to be directly staged
The production of temperature controlled stock (which is stored on site for a time before dispatch) also has Tuesday heavily overloaded compared to the rest of the week.
Chart 3: Planned production of materials to be stored
Impact on end-of-line processes
The AGV system moved more pallets on Tuesday than any other day, as shown by the orange columns in Chart 4.
Chart 4: AGV movements (orange columns) vs average movement duration (blue line), by day
There is also an increase in average movement duration (AMD) from Monday, which then dips on Wednesday before increasing for the remainder of the week.
Increasing AMD means the AGV are taking longer to complete their jobs. This could be:
Caused by traffic congestion which would indicate the AGV are performing ineffectively, or
Formed by the combination of movements completed over a period. As an example, staging is fasting than storing, so if the proportion of ambient pallets is greater than temperature controlled pallets then the AMD would tend lower.
“FASTER MOVEMENTS = LOWER AVERAGE MOVEMENT DURATION (AMD) = GOOD”
In actuality, Monday produced 34% of the week’s ambient production (20% increase from plan). With such a high proportion of fast movements, this explains why Monday had the lowest AMD for the week.
It would be reasonable to expect Tuesday’s AMD to increase as the store vs stage ratio increases. It did, however it should not have increased as much as it did. Wednesday had a similar store vs stage ratio, yet a lower AMD.
We can deduce that the relatively high average movement duration (AMD) on Tuesday was caused by congestion in the AGV system. This was likely the result of the disproportionately high number of pallets forced through the system.
Impact on production
Production-line downtime data indicated that 75% of the reported AGV-related downtime occurred on Tuesday. Given that the plan showed that day to be inordinately overloaded, perhaps this could be seen as self-inflicted.
Chart 5: Production reported AGV-related downtime, by day
Note that some data-points were removed as they related to a machine breakdown outside of normal operation, which occurred on Thursday.
Interestingly, while Tuesday had a high proportion of reported AGV-related downtime, the overall downtime was not much different from any other day. (Note that on Monday one of the measured lines didn’t start until afternoon, and by Friday one of the lines was finishing).
Chart 6: Production reported downtime from all sources, as percentage of week total
Recommendation 1: Improve production distribution by day
Clearly the plan distribution harmed the overall network on Tuesday. Many considerations must go into the planning process, we will take a simplified view but understand that there are further constraints to be considered in practice.
Table 1 shows the number of lines running each day. See that Monday and Tuesday are running many more lines than the tail end of the week. This would explain the higher proportional volume seen on those days.
It would be recommended that the distribution of production lines operating throughout the week be reviewed if optimising end-of-line processes were a priority.
| Mon | Tues | Wed | Thurs | Fri | |
|---|---|---|---|---|---|
| Staged | 3 | 4 | 1 | 1 | 1 |
| Stored | 7 | 6 | 5 | 2 | 4 |
If you are wondering why Monday may not have encountered the same congestion as Tuesday it is likely due to the following:
One of the faster temperature controlled lines did not start until later in the day.
Storage rooms were emptied before starting the week, so staging and dispatch from the rooms was not occurring.
“An in sync manufacturing site should be like a game of catch, each ball being successfully received downstream as the process progresses. But when a departmentalised mindset kicks in, catch quickly turns to dodge-ball.”
Production Distribution by Shift
While the plant ran a 24 hour operation, the night-shift operations were significantly lighter. As a general rule, ambient lines did not run night-shift, and fewer temperature controlled lines ran. Day- and afternoon-shifts are operationally similar from a production standpoint.
| Mon | Tues | Wed | Thurs | Fri | |
|---|---|---|---|---|---|
| Lines on D/S – A/S | 10 | 10 | 6 | 3 | 5 |
| Lines on N/S | 6 | 5 | 5 | 2 | 4 |
Impact on end-of-line processes
For such a simple chart, Chart 7 is packed with insight. It shows the AGV movements and average movement duration (AMD) by shift.
First, we see that night-shift produces only 27% of the pallets. This is lower than the 33% you’d expect with a smoother operation, but we can explain this knowing that the ambient lines are not running on night-shift. However, with consideration into which lines are running, night-shift should only be producing 25% of the pallets. So it’s punching above it’s weight, over-performing.
We also see the AMD drop on night-shift. This is counter-intuitive. We’d expect that with the higher ratio of store vs stage movements, that it should increase. This is a testament to the efficiency of night-shift, where half the lines did not run. Also note the increase in AMD for day-shift, clearly a source of inefficiency.
Chart 7: AGV movements (orange columns) vs average movement duration (blue line), by day
Impact on production
Production reported downtime data shows that that day- and afternoon-shift account for a disproportionate amount of downtime, while night-shift is heavily underrepresented.
This is consistent with expectations. During the day- and afternoon-shift, the system is overloaded, while over night-shift it is under capacity.
Chart 8: Production reported AGV-related downtime, by shift
Note that some data-points were removed as they related to a machine breakdown outside of normal operation, which occurred on Thursday.
Recommendation 2: Improve production distribution by shift
The above shows how a lack of production smoothing across shifts harms end-of-line processes, and as a result production itself.
When the plan overloads day- and afternoon shifts, they suffer significantly greater downtime. Whereas night-shift performs very well, working under capacity.
It would be recommended to investigate the overall cost of downtime based on standard rates vs the increased labour cost of running increased night-shift production. It may be determined that by tipping the production balance slightly, downtime could be reduced sufficiently to offset the additional night-shift labour costs.
Spread of Packaging
Packaging movements within the AGV system have been identified as a likely high-impact function. This is due to layout deficiencies, and it’s quite clear when observed in the field. This includes both storage and production staging processes.
Charts 9 and 10 show that, in line with the heavy Tuesday production, the most packaging movements were required on Tuesday day-shift.
Chart 9: Packaging movements by day.
Chart 10: Packaging movements by shift.
Impact on end-of-line processes & production
The previous two sections have demonstrated the negative impact of overloading any production period. These packaging movements fluctuate in line with those heavy periods and exacerbate the negative effect.
Recommendation 3: Improve the spread of packaging movements
While packaging staging must coincide with production, the storage process has some independence of operation.
Daily volumes are likely dictated by production requirements, but shift distribution is within the control of the end-of-line team.
It would be recommended that packaging storage movements be distributed as smoothly as possible across operational hours, targeting times of lesser production such as night-shift.
Fortunately, looking at the shift distribution, this process has already progressed.
Conclusion
We see that the spread of production across days and shifts can have a significant impact on end-of-line processes, leading to overloading and delays. This planning inflicted outcome can in turn adversely impact production.
By understanding the capabilities of each stage in the process a plant can uncover hidden efficiencies. Planning to optimise on all levels can identify previously unrealised improvements, leading to an overall boost in plant performance.
A manufacturing site isn’t a simple operation. There are many interdependent pieces coming together to achieve a goal – and each of those pieces play their own unique and necessary role.
While production runs a number of individual lines, the end-of-line team has the complex task of understanding how these multiple changing outputs will interact.
An in sync manufacturing site should be like a game of catch, each ball being successfully received downstream as the process progresses. But when a departmentalised mindset kicks in, catch quickly turns to dodge-ball.
Note that this week is not indicative of a normal or standard production week, though the insights it provides could be applied to general scenarios and are consistent with general observation. Note also that the AGV movement duration is indicative proportionately only, but the magnitude should not be taken as actual.
