You may think your pull system is Lean, but what are you pulling?
While the theory of the pull signal sounds simple – to not waste effort, do something only when it is necessary – the implementation of the pull mechanism introduces issues that need to be understood and addressed, especially in an environment of increasing variability.
In essence, the Lean pull signal is a live communication mechanism, linking processes together, one that takes or consumes something from the other. Examples of pure pull signals in operation are actually hard to find. We are all consumers of food, so you would have thought that when we become hungry, this would be the pull signal for us to eat. The more work we do and energy we expend, the more we should eat to be replenished. The hypothesis is fine, but in practice most people adhere to set meal times; many of us overeat, and who can resist that late-night chocolate treat, the result of which, at least for me, is anything
We might call this “human error,” but there are also systematic constraints. Consider kanban-driven local point-of-use storage locations. The optimum minimum stock level of a location and the associated replenishment amounts are the operating parameters needed for the system to work effectively. The theory is that, as materials are used from the location, the stock reduces to the minimum level; the kanban request is raised, and the replenishment stock subsequently arrives.
Sounds simple enough, yet certain assumptions and constraints become part of the determination of operating parameters. For example, setting the parameters to make the storage location as Lean as possible starts with the calculation of time taken from when the kanban signal is raised until replenishment stock arrives. If this is for 10 min., then minimum stock level must be at least 10 min. of stock so that material exhaustion is avoided. If two units of material are then taken from the location for consumption each minute, the minimum stock level should be 20. The replenishment quantity then needs to be enough to return the stock to a level above the minimum stock quantity. The absolute minimum for this would be 20. Any more than this and there would be wasted stock in the location for a period of time. This does mean, however, that the trigger for the next kanban replenishment is pretty much immediate.
If all storage locations had the settings calculated this way, the replenishment work would be huge because every location would have to be replenished every time. This does not make sense because it is a lot of work. What happened is that consumption of materials from the storage location has been tightly coupled with the supply of materials from the main warehouse, with only a minimum buffer in between. The value brought by the kanban storage in this case is only the physical offset of material storage away from the actual production process. In our example, then, the kanban storage is pure and super-Lean, but significant issues need to be considered.
What if the speed of consumption of the material were to change? When the product being made in production changes, the bill of materials changes so that the material in our example location may continue to be used, most likely at a different rate, or it may not be used at all. The kanban settings need to be set to cope with the worst-case scenario, which can be variable over time, requiring careful attention and management. This makes the kanban storage less Lean because there are materials left that need not be there, consuming space, and may require moving back to a main warehouse or on to another point-of-use store at some point.
Actually, the build-up of stock in the local store is a good thing for the logistics effort because they are now far less likely to have to replenish every material in every cycle. The logistics operation also needs to be considered in this example application of Lean thinking to not overload resources. A compromise exists between eliminating waste in one way or another. The local point of use store is unable practically to be completely Lean because it is a part of a larger overall process. This is a fundamental principle to be considered in any Lean project. The scope of any Lean project needs to start at the highest level, the base optimization applied with consideration of the higher level needs and probable effects.
The question then becomes whether it is OK to modify the pull logic to enhance the Lean capabilities in an attempt to avoid compromise, based on the look-ahead to the short-term. This could potentially change what the pull system was intended to be because it is designed to work as a live communication tool.
If look-ahead information were used, assumptions and predictions made as to what will happen in the near-term could change, causing the system to fail. In this case, perhaps materials need not be present. The look-ahead function for example may predict that a material is no longer going to be used in the foreseeable future, and so a local store location could be freed by returning material to the main warehouse. This is great for Lean because it reduces waste at the local store.
The risk, though, is that the material suddenly may be needed at the same line again if a change is made to the plan; for example, rework is assigned, and the material is suddenly needed but is no longer there.
A look-ahead also greatly benefits by smoothing the logistics resource. Statistically, many kanban signals can be raised in a short time that could overload Logistics’ ability to satisfy all requests. Being able to predict this situation in advance based on the planned usage of materials means materials can be picked in advance of the peak demand, thereby reducing the peak to a manageable level. If the production plan on which the look-ahead has depended changes, the local store may fail to have the materials needed. The value of the look-ahead is only good to a point, where the risk of variation becomes significant.
Variation affects Lean negatively in two different ways: The variation of demand for materials from the local store makes the store less Lean, and variation reduces the ability to enhance the Lean operation with knowledge of look-ahead conditions. A solution to this is to link the planning logic with the automated kanban management system so that planning is aware of the situation with materials and flows in and out of the local stores, and issues can be avoided.
As the nature of the electronics business changes, with faster advancement of technology products that exhibit the sales patterns of fashionable goods, the number of products and variants that the SMT-based factory is expected to make is increasing, which in turn causes an increasing need for high-mix production. People following the trends of Industry 4.0 say this will increase to the point at which the factory of the future is expected to “build on demand.” Whether this could actually happen directly in a complex process such as PCB SMT is another discussion, but certainly the trend strongly suggests that variation, as well as increasing volatility of demand to the factory. How, then, will factories retain the efficiency and productivity afforded by established Lean projects and principles, which may be compromised by the increase in variation, beyond the critical point?
This is the question to be considered in upcoming columns, how Lean elements can work together as part of a larger Lean machine. This is the Lean factory of the future, coming sooner than we may think. To answer the question in the title of this column then, the pull signal, unfortunately, is neither Lean nor robust enough in our complex environments when considered in isolation.