That day, when the CPK report came out, the whole room fell silent for three seconds, and then I decided to restart this lesson.
Remember? A few years ago, we shipped a batch of goods to a client, only to receive a flood of complaints about burrs on the product edges. During that meeting, Quality Assurance presented the CPK report – 1.08. The entire room fell silent for three seconds, the air thick with tension. Everyone exchanged glances, thinking, "Isn't that still within standard?" It was only after tracing back that we discovered the problem: our IPQC frequency and methods completely missed the critical pain points.
Where was the problem? It wasn't just about inspecting, but about whether the key points were inspected.
Often, people assume IPQC (In-Process Quality Control) simply means "as long as an inspection is performed," randomly sampling a few parts periodically. But frankly, that's like driving a car and only watching the speedometer, without realizing the tires are nearly flat.
What is the core of IPQC? Simply put, it's "catching problems in their early stages." Think about it: if you wait until all products are manufactured only to discover a critical dimension is off, all those scrapped items upfront are pure wasted money. So the key isn't "how much" you inspect, but "what" you inspect and "how timely" it is.
How to do it in practice? Let data speak, don't just rely on intuition.
1. Identify Your "Critical Process Parameters"
Every process has a few variables that are most temperamental and prone to issues. These typically originate from high RPN (Risk Priority Number) items in your FMEA (Failure Mode and Effects Analysis). In our burr example, it could have been tool wear, cutting speed, or feed rate. These are the focus points for your IPQC. Measuring every single parameter is a waste of time.
2. How to Determine Inspection Frequency?
Here, you cannot rely on intuition. If your process stability is excellent, with a Cpk above 1.67, then the sampling frequency can be lower, for instance, inspecting 1 out of every 500 products. However, if your Cpk is only 1.33 or even lower, and your DPMO (Defects Per Million Opportunities) has already reached 6210, would you still dare to sample only once every 500 units? At this point, you might need to increase the frequency to inspecting every 100 units, or even every 50 units. For unstable processes, you might even need to consider inspecting the first and last pieces of every batch. This was the lesson we learned from that incident: parameters were unstable, but the sampling frequency wasn't adjusted accordingly.
3. Inspection Methods Must Be "Fast and Effective"
Frankly, if your inspection method takes half an hour, by the time you get the results, hundreds of defective products might have already been produced. Therefore, you must choose methods that provide immediate feedback. For example, replacing manual calipers with vision measuring systems, or implementing sensors for real-time monitoring. The more you can shorten the time to detect a problem, the smaller the loss.
The Most Common Pitfall: Blindly Following Without Thinking
I once encountered a new hire who, upon joining, simply mimicked how their seniors performed inspections, copying both the frequency and methods. Consequently, one time, the machine they were responsible for switched to a new material with entirely different characteristics, yet they continued with the old sampling frequency. By the time the issue was discovered at the next station, several trays of defective products had already been produced.
This is a blind spot. IPQC is not static; it requires flexible adjustment based on process changes, material characteristics, and even customer requirements. If you merely follow instructions without critical thought, an incident is only a matter of time.
One Thing You Can Do Today
Re-examine the IPQC items you are responsible for and identify the one with the lowest Cpk.