Solutions To Bottlenecks

Solutions to Bottlenecks

Since I read it and like it, and fit in your first category of readers, can you say a bit more about what has replaced it... which of its recommendations are wrong or out-of-date? Here are the ones I saw:

The "speed of the bottleneck is the speed of the plant"
The "don't run for individual station efficiency"
The "allow yourself to be less efficient at non-bottleneck stations"
The "shorter runs lead to greater profits"
Excess inventory is a liability not an asset.
"Bottlenecks often move - they don't sit still on one machine. Even in a highly repetitive manufacturing process, which are few and far between. Even ones that appear repetitive often aren't as repetitive as you may think."
"If a machine has a backlog of work then people don't just sit and watch it build up. They can sub-contract for instance. Whether they do or do not depends on a large number of factors. The manufacturing process is like the software process - complex and not easily modelled."

In the example of an assembly line, where bottlenecks are approached as problems which must be solved, Let me make the following observations which were derived from a successful example of approaches which were developed over time to effectively solve some of the problems observed above. Let's take them one at a time:

The "speed of the bottleneck is the speed of the plant"

I would rather use throughput than speed in the above. In the simplest form of an assembly line, where there is a single line through which all of the production passes, and each station where a manufacturing action takes place has the same period of time for completion, the design of the process must make sure that specification will allocate no more or less manufacturing action than can be accomplished during the time allocated to each station in the line. This is to ensure that the line moves without halts (halts occur because a manufacturing action at a station required than the interval of time allocated for each station in the line).

In a properly designed system in normal operation, the manufacturing actions take place regularly and on schedule.

Even in such systems, events occur in individual stations which delay the action complete and throughput action. When such delays occur often and not as a result of mechanical breakdowns, a "bottleneck" is identified.

In most such simple systems, it is possible for the downstream processes to continue with the objects which have passed through the bottleneck until the manufacturing action in the last station is complete and the product is passed out of the system. However the upstream processes must halt because while they may have a manufacturing action complete, they cannot perform a throughput action. In this case it is true that "the throughput of the bottleneck is the throughput of the line."

Solutions:

Allocating some of the manufacturing actions of the offending bottleneck station to one or more other stations.
Redesign
- Create two parallel stations with switching capabilities allowing completion by switching the inputs and outputs alternately between the parallel stations (in effect doubling the allocated time for actions.)
- Create two stations (serial) so the former manufacturing actions in the bottleneck station are allocated to separate stations. (in effect halving actions performed by each station)

The "don't run for individual station efficiency" The "allow yourself to be less efficient at non-bottleneck stations"

This indicates an acceptance of defeat and/or lack of ambition. It recognizes a status-quo as inevitable and displays a the lack of vision, desire to optimize, or orientation toward success.

The "shorter runs lead to greater profits" Excess inventory is a liability not an asset.

This indicates an ability to produce more than can be consumed, This occurs when a manufacturing segment is at overcapacity, or is operating more shifts than necessary.

"Bottlenecks often move - they don't sit still on one machine. Even in a highly repetitive manufacturing process, which are few and far between. Even ones that appear repetitive often aren't as repetitive as you may think."

This requires more elaboration. While it is true that breakdowns and supply of the necessary raw manufacturing material may be spread over many stations in a line it does not negate the fact that bottlenecks in an operating system are clearly indicated by their effect on throughput and are analyzable and solveable problems.

"If a machine has a backlog of work then people don't just sit and watch it build up. They can sub-contract for instance. Whether they do or do not depends on a large number of factors. The manufacturing process is like the software process - complex and not easily modelled."

This is a problem that is more related to non-assembly line problems. Where it does apply the internal solution is to expand, or to add shifts. This is however only if the backlog is continuous and likely to last for significant time. If it not likely the short-term resort to sub-contracting or out-manufacturing might be more appropriate. StrategicAlliances? are another possibility.

There is always at least one bottleneck (and one critical path); there may be more than one. When there is a single bottleneck/critical path; other resources may be utilitized, and productivity gains/schedule pull-in may be achieved. On the other hand, when all stations are a bottleneck (or all paths are critical); the ability to achieve further gains becomes limited--as all bottlenecks must be widened.

Bottlenecks may be dealt with in manufacturing by having storage both before and after in which production is stacked or 'lined up in parallel lines with capacity equal to the extra-time you might operate the line in which the bottleneck occurs. The efficient parts might for example be run 16 hours a day with the bottleneck line operating an extra 2 to 8 hours. Another way is to have dual, triple or quadruple stations or lines parallel to one another with a feed-in and feed-out handling the input/output. These can be straight-thru in and out, or buffered in and out. (I have seen and been involved with the programming of PLCs to handle a buffered-in / parallel group of stations where manual rather than automatic processing occurs at less than the automatic rate, and to meet labor contracts on line speed for manual stations/lines. -- DonaldNoyes.Did this stuff.mid 1990s - in a couple mid-american automobile assembly plants or facilities