Why does one-piece-flow matter?
Dear Gemba Coach,
I am value stream manager in a company that makes consumer goods. We have been working with a sensei for three years and have progressively moved from batch organization to flow and production cells. Recently, during the plant visit, our sensei made a big fuss because there were three parts between two operators and he insisted in one or zero. I am confused at why the big difference between three and one part in the process. Could you shed any light on this?
Great question, which goes to the very core of lean: why do two extra parts matter? Why only one or zero parts in inventory between stations? I have to agree with your sensei that this makes a world of difference. But to spell out this difference we need to go back to the fundamentals of lean. Stripped to its bare bones, lean tries to achieve two things:
- Reduce lead-time in any process (through JIT)
- Involve operators in improving their own workstations to eliminate waste (Kaizen)
The basic thinking is that if you reduce lead-time in a process you will satisfy your customers better, and you will reduce your cost by getting closer and closer to the minimal capacity required to deliver. In doing so, you learn better how to make things. The other key insight is that every person’s time is precious and should only be used for value-added tasks. Because the workers themselves are best positioned to understand their own workstations, they should learn how to see waste in their own working cycles, and so you should work with them to progressively shave the waste out of the process. In doing so, you also develop people to the fullest of their abilities.
The link between these two key intentions is takt time. If we could produce every component of the final product exactly at takt time (daily operating time divided by daily production requirement), which would mean that every component moves at the same takt rhythm, to produce a finished good at the same takt, the stagnation of parts which makes much of the lead-time could be eliminated, and lead-time minimized. However, if you've ever moved from batches to flow you know that this is by no means easy, and in many areas, usually upstream, we're forced by the nature of the process, to produce in batches. Still, in assembly, with the right cell organization it's possible to get pretty close to takt time.
At the level of the workstation, when we watch operator cycles, we can distinguish between value-added work (which enriches the product towards completion) and waste (necessary movements that don't add any value to the product). There are many, many kinds of wastes, but lean practice has codified the main ones in terms of the famous "7 deadly wastes": 1) overproduction, producing too much or too soon; 2) correction, producing defective parts which are scrapped or need repair; 3) waiting, waiting for parts to arrive or for a machine to finish a cycle; 4) conveyance, moving parts or equipment form one place to another; 5) overprocessing, doing more work on the part beyond the strictly necessary to make it; 6) inventory, any more components on hand than the minimum to get the job done; 7) motion, any motion that does not contribute directly to value added.
In any batch process, these seven wastes are fairly obvious to any observer, but not necessarily to the operator. It's hard to explain to an operator, for instance, that carrying a crate of components to her workstation is "waste." This is what they need to do to get the job done, so why do we call this waste? But, as an engineer, it's easy to see all these value-less operations and to try to improve the system to take it out. Eventually, this leads you to create multi-process cells delivered continuously by a small train and so on. This both improves the lead-time and dramatically reduces the waste in the system.
But once you've set up the cells, what next? How do we see the waste if parts are flowing through the line? In fine, this is where lean comes into its own. Operators can now be involved in improving the details of their workstation and standardized work to Kaizen finer waste away.
For instance, I was recently walking through a plant that manufactures auto parts with semi-automated lines. At one of these lines, we could see the operator accumulating bodies although there was a defined space for only one. When we asked him about it, he immediately said "yes, I know we should be working in single-piece flow, but every twenty parts, I need to go and fetch a box of components, so if don't have enough parts in advance, the whole cell will stop." The operator himself can tell you exactly what the problem is.
This is very good plant, and the cell was working at a good rhythm. Operators had all been part of a stable team and knew what they were doing - their movements were both fast and fluid. However, most of them had from three to ten parts on their workstations. We asked them to hold to the single-piece-flow at takt time (actually, at target cycle time which is a bit shorter than takt for various reasons) concept for just half an hour, with the team leader in the cell, and watch the results.
As if we'd wiped a cloth over a grimy window, all the waste that was previously hidden in the cell suddenly appeared. Operators now stopped when they had done their part and could not pick up a new part from upstream until downstream had cleared their own part. What immediately appeared were a huge number of causes of small variations that made the operators stumble, lose time and then hurry up to catch up. Without the in-process inventory, the line kept stopping. There were all sorts of reasons - mostly materials not being brought all the way to the workstation but to racks a few steps away, but also equipment that needed to be fiddled with every other cycle, assemblies which were not as easy as they seemed and so on.
More importantly, every time an operator was stopped, he or she could explain very precisely why, and in many cases, when asked, they had suggestions about what to do about it. As many of these interruptions involved logistics or maintenance, the bulk of these suggestions would be hard to implement by themselves, but they all were great opportunities for Kaizen, either directly within the team or by involving other functions as well.
The difference between single piece flow (asking the operator to put their finished part on the next person's jig) and three-piece flow was clear. With a few parts in the process, the line appeared to work continuously without any major issues - and also without any clear opportunities for Kaizen, when seen from the outside. At zero or one piece (because of the loading-unloading of the automated parts of the line) the line now was both ripe with Kaizen opportunities, and opportunities the operators themselves could see and think about. With one-piece flow, the line moved into the lean world.
I recently had the privilege of spending some time with Masaaki Imaï, who brilliantly introduced Kaizen to the west (I can't recommend his books enough: Kaizen and Gemba Kaizen), and he is insists on three core principles to sustain Kaizen: flow, synchronicity, and leveling - in other words, flowing at takt time, and all the relevant methods to make it so, even when the technical processes are not set up to do so naturally. Indeed, takt is the hinge between lead-time reduction in the total process and waste eradication in every work cycle. As Brazilian sensei Gilberto Kosaka expresses it vividly, lean is about flow-stop-flow-stop: continuous flow and then stop for either a quality issue or an imbalance in the process. Every "stop" is a Kaizen opportunity.
Two extra parts at a station do matter because, in a way, they reflect on the Gemba the two basic, basic aims of lean: improving the flow of value and eliminating the waste. Whenever one leans out some waste out of a process, the flow of value improves. Conversely, whenever one improves the flow and gets it closer to takt time, more waste appears, and more opportunities to Kaizen with the workers themselves. So fight hard to get those three parts down to zero or one! And please let us know how that worked out.