Friday was a visit to the Toyota Museum in the morning and the “1 hour tour” of the Tsutsumi assembly plant in the afternoon.
If you ever get to Nagoya, the Toyota Museum is superb and definitely worth a visit. Even if you have no interest at all in lean manufacturing (so why are you reading this??) you get a really good look at over 100 years of technology development in the weaving industry, as well as their automotive history.
Sakichi Toyoda was one of Japan’s greatest inventors. Starting at the end of the 19th century he started incorporating mechanical assist and then automation into weaving looms. Remarkably his inventions were the first significant advance in weaving technology since John Kay invented the flying shuttle in 1733. Most of Sakichi’s principles remain today. There have been engineering advancements, but the basics are all still there. It was patent licensing of his first fully-automated loom with auto shut-off (jidoka) — the Model G in 1924 — that capitalized his start into the automobile business.
But I have to give credit to Gregg, one of my team-mates here, who summed it all up in one profound sentence:
“… all of this started with a son trying to make life easier for his mother.”
Wow. and Yeah. That insight really got to the core of what kaizen is about – a passion for making the work easier, because people’s burden matters.
Tsutsumi Plant Tour
Just to be clear, this is exactly the same tour that any group gets. There is nothing particularly special. The bus is boarded at the gate by the Public Relations girl (they are always young women), and she starts the spiel. We are on the catwalk over the line following a specific route.
So what did I see?
Wow. And that is not just because it was a Toyota plant, been on this tour before. The “Wow” is that they have made a significant change in their material conveyance. This may be old news to you, but I was last in this plant in 2000, so it was new to me. Previously they had line side racks with stocks of parts for the various models and options. The assembler looked at the manifest for the part code, and picked the appropriate parts for that car from the bins and installed them.
Later on I know they placed RFID on the car roofs which tell the various poka-yokes in the work station what the car needs, but the pick method was not fundamentally different. Kanban replenished the parts are they were used. (more about the RFID in a little bit.)
Now they are kitting car-specific collections of parts and sequencing them to the assembly stations. This is significant because I am a big fan of picking kits and delivering them to assembly at takt. There are a lot of possible problems which are mitigated or eliminated when this is done. But I had always conceded that at some point, takt time was so quick that it might not be practical.
I stand corrected. Here is an operation picking and delivering kits to many hundreds of assembly positions, one-by-one, at a takt of just under 60 seconds. Wow. The picking process is, well, superb, I am not sure what else I can say about it here. I am going to assemble my thoughts over the next couple of days.
RFID – the Car as Customer
The other really interesting bit was the use of an RFID box on top of every car. The box has that particular car’s configuration and options coded in it. (I suppose it could be a serial number linked to an option list in a data base too, but knowing some basic tenants of Toyota’s philosophy regarding information flow, I would not be surprised if the data were actually carried on the car.)
As the car moves through the processes, each work station basically asks the car “What are you? What do you need?” and the “car” responds through the RFID. The work stations’ poka-yokes and other configuration dependent things then adjust to help the assembler give the car what it needs.
So why not just put the sequence list in the computer and have each one called up as it goes by?
What happens if (inevitably) some small variation causes the list to not be accurate. There are thousands of things that can cause small changes. The second that computer sequence list is inaccurate, the entire system breaks down. And inaccurate it will be. Anyone who has tried to run their factory on detailed MRP blowdown knows what I am talking about.
No, in this case, each car “pulls” the work it requires, when it requires it. The information in each work station is delivered just-in-time, and not one second earlier. Thus the information is always the latest. Note that this is really not a fundamental change philosophically. The car has always carried its configuration information with it on the paper manifest. What is different here is that the computer system is facilitating better kaizen, but the information flow philosophy has not changed. The information travels with the car, not ahead of it.
What about that picking and kitting process? Well – and maybe one of you Toyota guys out there can answer this for me – and I will update this accordingly – but I would speculate that it too is driven by the RFID tags rather than a production sequence list. It is a very simple matter to know how many takt-times of lead time are required to pick the kit and get it to the appropriate station. (Well, it is simple for Toyota who is so takt-pulse driven, it may not be as simple for the rest of us – a kaizen opportunity here – basic stability.)
If it takes 10 takt times to pick and get a kit to the line, then 10 positions upstream of the delivery point the RFID is queried. “What are you?” That tells the system what is needed in +10 positions, and the pick list is sent to the picking area. The parts are pulled, kanban cards posted for replenishment, and the kit-cart sent on its way.. first in, first out, one-by-one to the assembly line.
No calculated lead-time offset. No sequenced pick list created in the morning. No sequenced pick list that will be wrong 5 minutes after it is printed. Robust, problem-tolerant, and simple.