That Day, When the CPK Report Came Out, the Whole Room Fell Silent for Three Seconds
I remember years ago, a batch of new material arrived at our factory, and everyone expected it to perform exceptionally. However, when the yield data came out, the CPK was only 1.08. The conference room immediately went quiet; you could hear a pin drop. At that moment, I thought, "This is terrible. We've been chasing materials relentlessly, only to end up with this dismal result." The boss's face was ashen, and he directly asked, "Do you even know which process step is causing the most severe bottleneck right now?" Honestly, at the time, we really didn't know. We could only rely on manual sampling and wait for reports to identify problems.
Where Was the Problem?
Frankly, in the past, when we practiced Lean Manufacturing, especially the Pull System, it mostly relied on people. Production line workers would look at Kanban cards and pull materials from upstream stations when they saw a shortage. This method was good, but the problem was that when production lines became complex and product varieties increased, human eyes and brains couldn't keep up. For example, you can't expect the team leader to check the utilization rate of every machine every five minutes, can you? Let alone instantly knowing which machine's DPMO suddenly surged to 6210. To put it plainly, the core of the Pull System is "real-time response." But if your reaction speed is half a beat too slow, then Lean is just "half-baked Lean."
How Is It Done in Practice?
So here's the key: IoT can fully maximize this "real-time response" capability. We later implemented sensors, directly installed on machines, to monitor production data in real time.
- Machine Utilization Transparency: With IoT devices, we can constantly see the utilization rate of each machine, clearly knowing which one is down or idle. There's no longer a need to wait for the team leader to complete a round of inspection to discover issues.
- Real-time Work-in-Process (WIP) Tracking: We attached RFID tags to each material tray. When trays enter or leave a station, the system automatically updates the WIP quantity. This way, you can know how much material is accumulated at each station and where bottlenecks might occur.
- Automated Quality Data Collection: Previously, CPK data required manual measurement and input by quality control personnel. Now, many measurement machines can directly upload data to the cloud. If the CPK of a certain batch suddenly drops to 1.08, the system can immediately issue an alert.
In other words, IoT transforms your Pull System from "manual pull" to "automated pull." When the system detects a material shortage at a certain station, it can directly trigger upstream stations to start production; when it finds a machine is down, it can immediately notify maintenance personnel.
The Most Common Pitfalls
When it comes to stumbling blocks, there are many stories. When we first implemented it, the most frequent issue we encountered was data silos. Did you think connecting all machines to the network would solve everything? Wrong! The data format for each machine might be different. Machine A outputs CSV, Machine B outputs XML, and after all that effort, the data still couldn't be integrated. Engineers were swearing just writing conversion programs. Therefore, the key is to plan for unified data standards from the beginning; otherwise, no matter how much data you collect, it will just be a pile of garbage. Also, don't assume that once sensors are installed, they don't need calibration. One time, due to a sensor drifting, the production line continuously false-reported material shortages, causing everyone to scramble needlessly. Only later did we discover that the sensor calibration cycle hadn't been properly set.
One Thing You Can Do Today
Inventory your production line: which critical data points are still being collected manually?