Using Countermeasures and Error-Proofing to Fix Problems

This point has been made earlier in the article, but it bears repeating: the closer you are to one-piece flow, the quicker quality problems will surface to be addressed. This hit home for me personally in a unique opportunity I had in the summer of 1999. General Motors had a program through its joint venture with Toyota, the NUMMI plant, in Fremont, California, in which they sent GM employees for one week of training in TPS. This one week included two days of working on the Toyota assembly line—actually building cars. I was given the opportunity to participate.

I was assigned to a subassembly operation off the main assembly line that made axle assemblies for the Toyota Corolla and the equivalent GM model. In unibody cars, where there is no chassis, there is not a real axle but four independent modules that include the wheel, brakes, and shock absorber. They are built in the same sequence as the cars on the assembly line, put on pallets, and delivered in the order of cars moving down the assembly line. There are about two hours from the time a module is built until it will be attached to the car, so if there is a problem, you have a maximum of two hours to fix it before the main assembly line segment is shut down.

One of the easy “freshman” jobs I had was to attach a cotter pin to hold a ball joint in place. You put in the cotter pin, spread out the ends, and it locked the ball joint in place. This affected braking, so it was a safety item and very important. At one point early in the afternoon, I saw people scrambling around and there were a number of impromptu meetings. I asked the hourly associate next to me what was going on and he explained that a unit had gotten to the assembly line without a cotter pin and it was a big deal. An assembly line worker who installed the subassembly on the car had caught it. The team knew it happened only a couple of hours earlier. I assumed it was my mistake and immediately felt terrible for having missed installing a cotter pin. The team member claimed that it happened while I was on break. Who knows? But his response to my guilt feelings was even more important. He said:

What is important is that the error went through eight people who did not see it. We are supposed to be inspecting the work when it comes to us. And the guy at the end of the line is supposed to check everything. This should never have gotten to the assembly line. Now we as a team are embarrassed because we did not do our jobs.

The other job I did was the final job on the line—a 100% inspection before loading the axles onto the pallet. The inspection included marking with colored felt-tip pens all the points you are supposed to inspect, including the cotter pin. It turns out that the unit with the missing cotter pin was not marked, so the inspector at the end of the line (which could have been me again—I’m not sure) failed to do a complete inspection. But again, what mattered is that the team went through intense problem solving to identify the root cause and put in place a countermeasure—all within two hours of when the problem occurred.

Although this missing cotter pin went undetected through the entire system of inspection, there were a remarkable number of countermeasures that had already been put in place on the axle line to prevent things like this from occurring. In fact, at every workstation there were numerous poka-yoke devices. Poka-yoke refers to mistake-proofing (also error-proofing or fool-proofing). These are creative devices that make it nearly impossible for an operator to make an error. Obviously, there was not a poka-yoke to detect whether the cotter pin was in place. Nonetheless, the level of sophistication on the line was impressive—there were 27 poka-yoke devices on the front axle line alone. Each poka-yoke device also had its own standard form that summarizes the problem addressed, the emergency alarm that will sound, the action to be taken in an emergency, the method and frequency of confirming the error-proof method is operating correctly, and the method for performing a quality check in the event the fool-proof method breaks down. This is the level of detail that Toyota uses to build in quality.

As an example, though they did not have a poka-yoke to check if the cotter pin was in place, they did have a light curtain over the tray of cotter pins. If the light curtain was not broken by the operator reaching through it to pick up a cotter pin, the moving assembly line would stop, an andon light would come on, and an alarm would sound. Another poka-yoke device required that I replace a tool (somewhat like a file, used to expand the cotter pin) back in its holder after each time I used it or the line would stop and an alarm would sound. It sounds a bit bizarre—one step removed from getting electric shocks for any misstep. But it is effective. Of course there are ways around the system, and the workers on the line find them all. But at Toyota there is a discipline about following the standard tasks that workers tend to adhere to.

Standardized work (Toyota Way Principle 6) is itself a countermeasure to quality problems. For example, the particular job I had was designed so it could be accomplished in 44.7 seconds of work and walk time. The takt time (line speed in this case) was 57 seconds per job, so there was plenty of slack time; hence it was a freshman job. Yet even for this simple job there were 28 steps shown on the “standard work chart,” right down to the number of footsteps to take to and from the conveyor. This “standard work chart” was posted at my job site, where there were visuals that also explained potential quality problems. A more detailed version in a notearticle had each of the 28 steps on its own sheet, described in greater detail along with a digital photo of that step being performed correctly. Very little was left to chance. Whenever there is a quality problem, the standard work chart is reviewed to see if something is missing that allowed the error to occur and, if so, the chart is updated accordingly.