I recently finished Toyota Production System, by Taiichi Ohno, who developed and implemented the system company wide at Toyota. Taiichi Ohno readily admits that he's an engineer, not a speech giver or a writer, so the writing (and translation) is often lacking, but there are real nuggets of insight throughout.
In short, the Toyota system is about eliminating any waste between a customer order coming in and a car going out the door. Cost reduction and improved quality are happy side effects, but focusing on waste is the key.
I think one of the most insightful leaps was to identify the waste in large batch, high volume production. Intuitively, we know that a large machine like a die press has an inherent setup time, and the larger the batch, the better the amortization of that setup cost and the better the economies of scale on the raw materials. So it may be true, that on a per part basis, the cost is cheaper, by Ohno's genius was to look at part manufacturing in the context of a higher system level.
Large batch sizes lead to large amounts of inventory, which is just another way to picture dollar signs sitting around tied up as physical parts. Large inventory levels also mean warehouses to store the inventory, extra people to move the stuff around, higher risk of producing a large batch of defective things, and so on.
The counter intuitive solution is to reduce cost by making only what's needed and only when it's needed. This is call "just-in-time". It's not too much, too soon, or too little too late, it's the right amount, just in time.
The system starts at the very end, with the finished product, i.e. the sales lot sold 5 cars today, so we need to make 5 cars to replace the ones we sold. This would in turn issue a "kanban" card to the previous process, in this instance, say engine assembly. Engine assembly would in turn would issue kanban cards for the needed pistons and gaskets and engine blocks, and those components suppliers would issue kanban cards to their feeder processes, and so on and so forth.
Of course, you can't have massive backwards ripples of kanban cards traveling through the system to get things started at the other end, so Ohno introduced the idea of production leveling, or flow. If the system is geared up to make a steady stream of 5 cars a day, each process can have it's own mini stock of work in progress. It's not going to "sit" in inventory, because it's anticipated to flow through the system. The kanban process can easily handle small variations in flow, say 4 cars instead of 5, and let's the line workers manage themselves to a degree to keep the flow moving.
Taken, collectively, this system dramatically reduced costs and improved quality in car production.
Other important tidbits:
- Ohno's definition of waste also included the human waste of boring jobs. The Toyota system also relies on "autonomation", meaning automation, but with enough sensing to detect process errors and stop the line from making defective parts.
Toyota had it's start in autonomous looms that could stop weaving when a thread broke. One worker could now operate 40 machines simultaneously, and their job went from watching one machine to make sure it was working, a menial task, to actively diagnosing and fixing machines that had automatically stopped, a job that respected their human intelligence.
- Ohno focused on improving efficiency, not by focusing on the numerator of productivity and total parts produced, but by focusing on the denominator, the fixed costs of making the part. For example, if a crankshaft was forged, ground, and honed, requiring three machines and three operators, he would cross train a single operator to run all three machines, meaning he could run crankshafts with a single operator. This allowed them to run smaller, more efficient batches, with fewer people, a competitive advantage in general, but particularly in slow times and for low volume, high mix products.
- Ohno was one of the first to reorganize factory floors to reflect process flows. Instead of a milling department with 30 mills, a lathe department with 30 lathes, a grinding department, and so on, he would have true "lines" with a single mill, lathe, and grinder in the appropriate sequence.
- Remember that setup cost that we talked about earlier? With the focus on small batches, that setup cost clearly stands out for the waste that it is. Toyota was able to reduce the changeover time on some of their presses from hours to minutes.
- Eric Ries, the author of the Lean Startup, leaned on the Toyota production method to apply the ideas to starting a business, especially, to software ones. In short, the idea was to eliminate waste in product development on the path to product market fit. Whereas the SCRUM methodology reduces software development to short batches of development that might last anywhere from 2-6 weeks, Lean Startup advocates for not developing product features any faster than they can be tested with customers. It might slow the engineers and developers down, but from the higher level system context, it's more efficient.
- The five whys. In short, answer ask why 5 times, like a toddler, to get to the root cause of any problem.
- The Ford product system had so much momentum, it was assumed that it was easier to flag errors in process and correct them later, rather than stop the line. The Toyota system has the andon cord, that any worker can pull to stop the entire line to correct the error on the spot. In the short term, it sounds inefficient, but big picture, in the long term, it leads to better root cause detection and correction, and ultimately, higher quality.