Showing posts with label kanban. Show all posts
Showing posts with label kanban. Show all posts
kanban push-me pull-you
kanban - some quotes
Toyota Production System
kanban push-me pull-you
This is the slide-deck for the presentation I gave at the Lean Kanban UK conference. Many thanks to David for giving me the opportunity and impetus to put some of my thoughts together on this.
Taiichi Ohno's workplace management
When I was a middle school student in the old system, we studied the Chinese classics, and during this class we learned from the Analects of Confucius. In these writings Confucious says, "The wise will mend their ways" and "The wise man should not hesitate to correct themselves."... Confucius was saying that we should change gracefully... I think his words mean that in the end it is not good if you hold onto your ideas too strongly and try stubbornly to justify them.
When we said we would set up a centralized grinding operation, one experienced worker said, "No, we tried that during the war, but it failed. That's why we do it the way we do now." [I said] "I did not see it fail during the war. Show me again how it fails. If I am persuaded by this, I will let you continue doing it the way you do it now."
If you asked me, "What is the most important part of production control?" I would say it is to limit overproduction.
The kanban was a slip that indicated how many pieces they were coming to get, so that if they were going to take ten parts this became a production instruction slip directing the production line to make ten pieces.
When lot sizes are small, you need to do changeovers more frequently.
Stopping the line causes a great loss, so this forces us to think, "How do we keep them from stopping the line?" and this results in more and more quality kaizen.
You can only really tell what is better based on results.
Accounting cannot do any cost reduction... The shop floor reduces inventory. This money goes to the bank... Instead, accounting thinks it just needs to allocate cost savings targets.
There is something called standard work, but standards should be changing constantly. Instead, if you think of the standard as the best you can do, it's all over. The standard is only a baseline for doing further kaizen. It is kaiaku if things get worse than now, and it is kaizen if things get better than now. Standards are set arbitrarily by humans so how can they not change?
You must create a standard for comparison.
Drop a nut once and pick it up. Working at the average time is like trying to catch the nut halfway because letting it drop all the way down takes too long... There is no such thing as average value in this world.
Do not seek to follow in the footsteps of the old masters, seek instead what these masters sought. [Matsu Basho 1644-1694]
Once he asked me how the terms kaizen and kairyo (reform) were differentiated in the West. I said that while kaizen means to make improvements by using brains, kairyo means to make improvements by using money, and that in the West, most managers only think of improvement in terms of money. [Massaki Imai]
Let the flow manage the processes, and not let management manage the flow.
The aim of kanban is to make troubles come to the surface and link them to kaizen activity. I tell people, "Let idle people play rather than do unnecessary work."
The production line that never stops is either excellent or terrible.
Costs exist to be reduced, not to be calculated.
kanban push-me-pull-you
I mentioned how this allows genuine pulling. For example, the Fubarers can signal they're ready to pull a story from the Wibblers by moving an empty red kanban tray into the Wibbling column.
I said that the Wibblers could simply move a done blue story from its yellow kanban into an empty red kanban. Something about that bothered me and I think I now know what it is. Once again it's about visualization. What bothers me is that there is no representation of whether a blue story is done until it's moved into an empty kanban. There is no visual display of whether a blue story actually is "in-progress" and being worked on, or whether its "in-queue" and waiting for an empty red kanban. That feels wrong.
More recently, I blogged about about the common pattern of splitting each column; one for ongoing, one for done, like this. Something about the done column didn't feel quite right either.
So here's what I'm thinking. As well as moving empty-kanbans upstream to signal a pull, you can also move full-kanbans downstream to signal a push. For example, when a Wibbler finishes a blue story they don't leave it in their Wibbling column, waiting for a red kanban pull signal, they move it, still in its yellow kanban, into the Fubaring column, like this:
Ideally, the departing full-yellow-kanban will be just-in-time to meet an arriving empty-red-kanban between the columns...
...and the blue-story will flow from the full-yellow-kanban into the empty-red-kanban:
On the other hand, if the Wibblers are working much faster than the Fubarers then the Fubarer's column will fill up with full-kanban push-requests:
As the Fubarer's column fills up with full-kanban push-requests from upstream it's likely it will also be filling up with empty-kanbans pull-requests from downstream (in green say):
Push-me-pull-you is very visual:
- There's one representation of flow ; thin columns with kanban of one colour only.
- There's a different representation of lack of flow ; fat columns filling up (at the bottleneck) with two or three different coloured kanbans.
- If the yellow Wibblers work faster than the downstream red Fubarers, the Wibbler's full yellow kanban get stuck downstream in the Fubarer's column, inviting the Wibblers to help the Fubarers.
- If the red Fubarers work faster than the upstream yellow Wibblers, the Fubarer's empty red kanban get stuck upstream in the Wibbler's column, inviting the Fubarers to help the Wibblers.
- pull does not preclude push; you can have push and pull.
- wip-limits are not limited to pull systems; wip-limits would help in push systems too.
kanban musing
Here's a typical kanban board; three vertical sections, each with a work-in-progress wip-limit, each split into two sections; Ongoing work on the left, Done work on the right. I've been thinking about the Done columns. Specifically, the fact that they are labelled Done. You see, Analysis's Done is Development's Backlog. And Development's Done is Acceptance's Backlog. So none of the Done's are really done. Not done-done. Why are they labelled Done? Why not Backlog? Or donelog? The labelling seems to reflect a push mentality. I wonder whether there is a better way to draw and label them. One that places equal weight on removing items from the queue. One that would help to suggest a more connected process. Perhaps draw it between Development and Acceptance but inside neither?
I've also been pondering the label "wip-limit". In the picture above, Development has a wip-limit of 3. That strikes me as a bit odd. Development's limit of 3 is really a combination of a work-in-progress-limit for its left-half Ongoing column, and a work-not-in-progress-limit for its right-half Done column. Since Development's right-half Done column is also Acceptance's Backlog why does Acceptance not have a say on that part of the limit?
Even the word itself "wip" conjures an image of frenetic activity. A fast moving, whip cracking wip. Can't we have another word for the limit on the work that is not in progress? One that conjures an image of stillness? Any ideas? After all, as Don Reinersten says
making activities more efficient is much less important than eliminating inactivity
smart swarm
As successful foragers return to the nest with seeds, they're met at the nest entrace by foragers waiting in reserve. This contact stimulates the inactive ants to go out. Foragers normally don't come back until they find something. So the faster the foragers return, the faster other ants go out, enabling the colony to tune its work force to the probability of finding food.
Instead of attempting to outsmart the desert environment, the ants, in a sense, were matching its complexity with their own.
Instead of trying to keep fine-tuning a system so it will work better and better, maybe what we really ought to be looking for is a rigourous way of saying, okay, that's good enough. [Deborah Gordon]
If a scout bee was impressed by another scout's dance, she might fly to the box being advertised and conduct her own inspection, which could last as long as an hour. But she would never blindly follow another scout's opinion by dancing for a site she hadn't visited.
J. Scott Turner considers the mound's function as a respiratory system so essential that the termites couldn't live without it. In a sense, he argues, the mound is almost a living part of the colony.
If individuals in a group are prompted to make small changes to a shared structure that inspires others to improve it even further, the structure becomes an active player in the creative process.
Unlike our systems, which are tuned for efficiency, the termites' systems have been tuned for robustness, which they demonstrate by building mounds that are constantly self-healing.
What really made the lights go on was the realization that termites don't pay attention to the environment itself but to changes in the environment.
Not only does this complicated structure represent an indirect collaboration among millions of individuals, it also embodies a kind of ongoing conversation between the colony and the world outside. The mound might look like a structure, but it's better thought of as a process.
We should think of it [the termite mound] as a dynamic system that balances forces both inside and outside its walls to create the right environment for the termites.
When you feel like you belong to something, it gives you so much more freedom and so much more energy that might otherwise be used up in anxiety, to do other things.
On January 12, 2006, several hundred thousand pilgrims had gathered in a dusty tent city at Mina, three miles east of Mecca...
By noon... about a half-million or more pilgrims filled the Jamarat plaza in front of the bridge... The pressure inside the crowd was crushing... More than an hour later, victims were piled up seven layers deep: 363 men and women were dead.
"Those in charge need to remember the root cause of the problem: too many people trying to get through too small a space. The ingress rate at the bridge was 135,000 per hour. The thoughput rate of the pillars was only 100,000 an hour. You can't put a pint into a half-pint jug." [Keith Still]
Subscribe to:
Posts (Atom)