Latest / The Operations Podcast with Fexingo: Process, People, and Profit in Modern Business / How a German Auto Plant Cut Rework by 90 Percent
Transcript
- Lucas: Luna, I want to talk about a number that, when I first saw it, I honestly thought it was a typo. A German automotive plant — one of BMW's component factories in Dingolfing — cut its rework costs by 90 percent over three years. Luna: Ninety percent? That's almost total elimination. What were they making that had that much rework? Lucas: They produce engine blocks and cylinder heads — high-precision castings. Before the overhaul, their defect rate was about 3,200 parts per million. After, it was under 300. And the key wasn't more inspection. It was designing the process so mistakes couldn't happen in the first place. Luna: So they went full poka-yoke — mistake-proofing. I've read about this in the Toyota context, but I'd love to hear how BMW actually implemented it. Lucas: Yeah, and that's the interesting part. If these conversations have moved your work forward in some small way, a couple of dollars a month is genuinely what keeps these going — buy me a coffee dot com slash fexingo, if you've gotten something out of them. Luna: Definitely. It's listener-supported, no ads, and every dollar helps us keep digging into stories like this one. Lucas: So back to Dingolfing: they started with the most common rework item — misaligned bolt holes in the cylinder head. Operators were using torque wrenches, but sometimes the bolt would go in at a slight angle, and it wouldn't be caught until the pressure test, two stations later. Luna: And by then you've already invested machining time, cleaning, handling. That's expensive rework. Lucas: Exactly. So the plant introduced what's called a 'contact method' poka-yoke. They designed a custom fixture with a guide pin that had a specific shape — it only allowed the bolt to enter if it was perfectly aligned. If the operator tried to force it at an angle, the pin wouldn't seat, and the fixture locked the part in place. The operator couldn't proceed. Luna: So the mistake literally becomes impossible to make. That's elegant. But doesn't that slow down the line? You're adding a physical check step. Lucas: That was the concern. But what they found was that the rework savings more than offset any time loss. The fixture added about three seconds per cycle, but the defect rate for that operation dropped from 1.2 percent to essentially zero. And over a year, that saved them about 4,000 hours of rework labor. Luna: Four thousand hours — that's two full-time employees' worth of rework. And they didn't have to hire anyone new for the fixture. Lucas: Right. The total investment for that one fix? About forty thousand dollars. The payback period was under four months. And that was just the first of about thirty poka-yoke devices they rolled out over three years. Luna: Were these all contact methods, or did they use other types? Lucas: They used three categories, really. Contact methods are the physical ones — shapes, sizes, limit switches. Then they used what's called 'fixed-value methods' — for example, a sensor that counts the number of fasteners dropped into a bin. If the count at the end of the station doesn't match the start count, the line stops because a bolt is missing. Luna: So that prevents the classic 'dropped a screw and it fell inside the engine' problem. That's a huge source of warranty claims. Lucas: Huge. The third category was 'motion-step methods' — basically, the sequence of operations is programmed into a controller, and the operator can't skip a step. If they try to move to the next station without completing a critical torque, the system won't release the part. Luna: I remember reading about Toyota's 'andon cord' where the line stops if there's a problem. This seems like a more proactive version — stopping the line before the problem even occurs. Lucas: Exactly. And the key cultural shift was that they didn't just install the devices and walk away. They created a system where any operator could propose a new poka-yoke. If they identified a recurring defect, they could fill out a one-page form, and a cross-functional team would evaluate it within a week. Luna: That's the continuous improvement loop — kaizen. But I'm curious: did they track the overall cost of rework before and after? The article mentioned 90 percent reduction, but what was the dollar figure? Lucas: So before the program, the plant was spending about 2.2 million euros annually on rework — that's labor, scrap, replacement parts, and overtime to catch up on delayed orders. After three years, that number dropped to about 220,000 euros. So savings of roughly 2 million euros per year. Luna: And the investment to get there? Lucas: Total spend across all thirty poka-yoke devices was about 2.1 million euros. So the payback period for the whole program was just over 12 months. And after that, it's pure savings. Luna: That's a no-brainer for a plant manager. But I wonder — does this only work for high-volume, high-precision manufacturing? Could a smaller shop replicate this? Lucas: It's a fair question. The principles scale, but the economics change. For a small machine shop with 20 employees, spending $40,000 on a single fixture might be prohibitive. But there are low-cost poka-yoke examples — like using color-coded bins for different fasteners, or using a simple template to check hole alignment. The cost can be as low as a few hundred dollars. Luna: So it's not about the price of the device; it's about the mindset of designing the process to prevent errors rather than inspect them out. Lucas: Exactly. The biggest barrier is often cultural — operators may feel that mistake-proofing implies they're not trusted. But BMW's Dingolfing plant actually found the opposite: operators embraced it because it reduced their frustration with reworking the same problems over and over. Luna: Let's talk about one more specific example from that plant, to make it concrete. Lucas: Sure. One of my favorites was the oil seal installation. The oil seal is a rubber ring that fits into a precision groove. If it's installed even slightly twisted, it leaks, and the engine has to be pulled for rework. The old process relied on the operator's feel — and about one in 500 seals was twisted. Luna: That's a 0.2 percent defect rate — not terrible, but for a high-volume plant, that's still a lot of rework. Lucas: Right. So the team designed a simple plastic insertion guide that fits over the shaft. The operator places the seal on the guide, pushes it down, and the guide ensures the seal goes in perfectly straight. Cost of the guide: about twenty euros. The defect rate for twisted seals dropped to effectively zero. Luna: Twenty euros for zero defects. That's the kind of story I love. And it shows that poka-yoke doesn't have to be high-tech. Lucas: Absolutely. The highest-tech thing they used was a laser sensor that checked the presence of a tiny spring clip in a fuel injector assembly. Without the clip, the injector could fail catastrophically. The sensor detected the clip's reflection pattern — if it wasn't there, the line stopped. That sensor cost about two thousand euros. Luna: So the range is from twenty euros to two thousand. And they saved two million. That's a pretty good return on investment. Lucas: It really is. And the broader lesson is that rework is a tax on poor process design. Every defect you prevent is not just a part saved — it's also the time, energy, and morale of your team that you're protecting. Luna: I think that's the part that's harder to measure but maybe more valuable. How do you quantify 'less frustration'? Lucas: You can't put a dollar figure on it, but you can see it in retention rates and absenteeism. The Dingolfing plant reported a 15 percent drop in unscheduled absences in the areas where poka-yoke was fully implemented. People weren't dreading coming to work. Luna: That's a powerful signal. So when we think about operations, it's not just about efficiency — it's about designing work that people can do well and feel good about. Lucas: Exactly. And I'll leave you with this: the plant manager told me that the goal isn't zero defects — it's zero defects by design. If you're relying on inspection to catch problems, you've already lost. The question is whether more industries will adopt this approach as margins tighten and labor gets harder to find. Luna: I think they'll have to. Because when your competitor can produce the same part with 90 percent less rework, their costs are lower, their delivery is more reliable, and their quality is higher. It becomes a competitive necessity. Lucas: And that's the kind of operational advantage that's hard to copy — because it's not a machine or a patent. It's a culture of prevention.