The term Pull System is often used in Lean manufacturing. In our Lean training courses, we have used a game as a metaphor, trying to push a rope into a hole in a box and out on the other side. Quite a difficult and slow task. The participants feel almost tricked when, as an alternative, we let them pull the rope through the box.
How do we translate this game into the real world of our woodworking plant?
The classic Pull is in connection with a KANBAN system. In its purest form it works well with a narrow product line and a low part variety. The last operation or process takes (pulls) its required components from a KANBAN buffer, stock or supermarket. This allows the operation to work at optimum conditions. All parts are available when you need them. As the consumption of the individual parts reach the trigger point, an optimized production order goes upstream. This could be for one or more production steps upstream. Again, that process will be fed from a KANBAN inventory. These steps can repeat all the way upstream until the last KANBAN in your factory is refilled by a supplier. This system is very prevalent in the automotive industry. The suppliers continue this system in their factory. They will ship exact shipments, well timed, to fill the buffers in the factory. By doing this the suppliers structure their production to the same rhythm and pace as the assembly line. The demand at the last operation (i.e., assembly line) pulls the demand for the entire supply chain.
So how is that different from what most of us are doing?Furniture or cabinet manufacturers have adopted this production scheduling to some extent. The days of big production runs are done. Most manufacturers run on a make-to-order system. They are not producing into big finished goods warehouses. If that is not pull, then what is?
Scheduling issues daily batches of customer orders to be produced one after the other. As the order is completed on a workstation, the work-in-process (WIP) moves to the next workstation. The process is repeated a few times until all the manufacturing and assembly is completed. This model is very typical in our industry. The assumption is, let’s say in a system of daily batches, every day a day’s worth of orders is issued, and a day’s worth of orders is completed. Now what is the difference between Push and Pull in this type of set up?
In the Push scenario orders are completed at each workstation, as much as they can, and each workstation works as efficiently as they can. As the workflow is never absolutely balanced, there are delays along the manufacturing chain. Workers optimize the sequence of jobs to optimize for their workstation (local optimization), which is not always what’s best for the
next one. This imbalance creates the flow to back-up
and buffers need to be built
in to compensate.
So what harm is there in a little back-up?It was always seen as good manufacturing practice to keep all workstations busy and productive. All workers and machines need to work continuously at optimum speed. When they stop working, no value is created! When going Lean, we challenge this statement. It is one of Lean manufacturing’s paradigms, which takes a while to be accepted. We all understand traffic jams. If there is a blockage, do not bring more cars upstream on the road. Traffic control tries to reduce or meter the intake upstream.
When you have imbalance, and practically everyone has some level of imbalance, the flow gets backed up. The backing up in Lean manufacturing terms causes one of the seven deadly wastes in manufacturing. It is over-production!
This over-production is building up WIP. This requires more space. Roller conveyors and carts start getting full. The production slows down as the operators take longer to look for empty carts and to move/park the cart/material. This initially starts out small and is not much of a big deal.
The traditional solutions are:
More WIP- storage required (carts, roller conveyors, floor space…)
Bigger plant
Longer manufacturing lead-time (factory through-put time)
Return to batching for
better efficiency
Hire more people to compensate for the loss of productivity
Although this scenario may be a bit exaggerated, we can see and accept the direction that the push model is leading us.
How do we Pull?The key element of a Pull system is a system of consciously restricting WIP. You start at the end of the process chain. You concentrate on optimizing the last operation. The bottleneck, and especially where the bottleneck (beginning-middle-end) is, has some impact on the approach. The further the bottleneck is downstream the more important the Pull becomes.
Focus on having all parts
available (see my articles on
Fit & Complete)
Prioritize all upstream processes on what is required downstream
Curb upstream production when they start getting ahead
Re-locate available up-stream resources to down-stream and bottleneck processes
When focussing on the last operation, the question may be - What is the last operation? Is it assembly in our cabinet plant, or is it to load the complete order in a truck? The entire value stream does not usually end there, but it is a good start to optimize and start pulling. This, however, requires that there is a natural break. If you are shipping to dealers as you complete, or you have a sufficiently large warehouse, then pulling at the end of the assembly line will work. If, however, your installation operation is backing up and you have no space to put your product, you must solve that bottleneck first, or create a large enough warehouse.
As you pull from the end and do not overproduce, bottlenecks and areas of delay (snags) become very visible. Production management can (and must) react immediately to these delays. As you hold back overproduction, theoretically, you can use the held back resources to keep the flow going at the pinch points. If there is a snag and upstream continues to produce and orders and material keep being pushed into the system, you create inefficiencies not easily seen.
Buffers help but are not the alternatives to Pull.
The more your production is balanced, the less you need buffers. There are two main reasons for buffers:
Inherent imbalance of processes (fast changes in product configuration and product model) makes perfect and consistent balance impossible. This requires some buffers to cover for these imbalances. Also, when you tether two operations too closely together with no buffer at all, the combined capacity of these two machines will be reduced. So, some buffers are required.
Buffers are an insurance policy for imperfection. As with real insurances, what is the risk, how much are you willing to pay, and can you work on reducing the risk (and need less insurance)? This is a critical question which each company needs to answer.
The critical point with buffers and a Pull system is that when the buffer is full, you stop producing more. When a well-defined buffer is full, it is usually better to stop the upstream work cell and bring the upstream resources downstream to help the flow. There is (usually) no benefit to fill the factory beyond that point.
How do you start?It is difficult to change a plant entrenched in a push system to slow down as a solution for their production problems. To get started, a simple solution is to speed up the last operation (i.e. assembly).
Picture a sequence of belt conveyors transporting boxes. If the different belts are all running at exactly the same speed, it will run smoothly. If a belt downstream runs slower, you will have a pile up. A pile up is solved/prevented if a downstream conveyor runs a bit faster than the one upstream. So, by putting extra resources into the last operation you start to pull.
“If we do that, we will run out of work at the assembly line,” is a common comment.
Another effect of Push is that, when buffers get full, workers tend to slow down. We are ahead, so why rush? The reverse is also true; if the workers see that their downstream co-workers (customers) are just about running out of material, it is a natural motivator to speed up. So, let them run the assembly buffer dry.
First, it will surprise you how much the pull-effect will refill the buffers just-in-time, and second, this brings so much visibility to what is going on the shop floor:
Visibility of the real bottlenecks
Understanding how little space is actually required
Seeing where time in factory through-put-time is lost
Understanding the priorities for continuous improvement initiatives
Only by starting to Pull do the improvement opportunities become visible and implementable. It’s one of the initial starting points of becoming a leaner company. That the productivity and capacity increases is just a welcome side-effect.