That Day, Chasing the Production Line, I Nearly Tripped in the Cleanroom
I still remember a Friday afternoon a few months ago. I was about to leave work to pick up my child when I was suddenly stopped by a senior team leader. "Ah Liang, take a look at this batch of goods. The client needs it urgently. Shall we work overtime this weekend to push it through?" I thought to myself, "Here we go again! Weren't we so idle for the past two weeks that we were just cleaning machines? And now they want me to come in on the weekend to rush production?" At that moment, I saw a pile of goods jammed in front of machines; some machines were idle, others were overloaded. The entire production line was like a mad person, surging forward then stopping abruptly. I truly felt an impulse to throw my hat on the ground.
You must have encountered this situation too, right? The production line is either so idle that mosquitoes could move in, or so busy that everyone is overwhelmed and running off their feet. This "tidal production" mode is simply physically and mentally exhausting for us frontline engineers.
Where's the Problem? It's That "Instability"!
To put it plainly, the problem lies in poor "production plan stability." Often, the schedules provided by the PMC (Production Management and Control) department simply dump order volumes in without considering machine load capacity, manpower allocation, or even the output of upstream processes, aiming only to fill the "quantity."
What's the result? When large orders come in, the production line gets overwhelmed, everyone rushes frantically, Cpk values are close to 1.08, and quality consequently drops. DPMO skyrockets from the usual 6210 to 12500—this data is very ugly. As soon as order volumes decrease, everyone becomes idle again, machines sit empty, resources are wasted, and you might even see people cleaning machines as maintenance. With such erratic speeds, how can quality be stable? How can delivery times be punctual?
So the key point is, what we need isn't "maximum capacity," but "stable capacity."
How to Implement It in Practice? "Pull" Replaces "Push"
To stabilize production plans, the core concept is to shift from "push production" to "pull production." Previously, PMC scheduling would "push" goods to you, regardless of whether you could accommodate them. Now, we need to switch to "pull," meaning "the downstream station only pulls goods when the upstream station is ready."
- Establish Buffer Zones: Set an appropriate amount of buffer between each key process. This buffer zone is like a reservoir, capable of regulating the flow between upstream and downstream. When upstream capacity is high, surplus products can be stored in the reservoir; when downstream demand is high, the reservoir can provide support. For example, previously, after wafer dicing, products were sent directly to encapsulation without a buffer. Now, we've established a "wafer temporary storage area" after dicing, with a limit of 20 boxes.
- Set Maximum Work-In-Progress (WIP) Limits: Limit the maximum quantity of work-in-progress that can be placed in front of each machine in every process. For example, only a maximum of 5 batches of goods can be placed in front of a machine. Once exceeded, the preceding process must stop production. This might sound strict, but it forces the production line to "pull" goods instead of "pushing" them.
- Visual Management: Clearly display these buffer zones and WIP limits using signboards, indicator lights, or screens. After implementation, when WIP in front of a machine exceeds 5 batches, the signboard will light up red, immediately informing operators and engineers that action is required.
In other words, it's about making the production line like a flexible rubber band, not a rigid iron rod.
The Most Common Pitfall: PMC Tells You "Impossible!"
Frankly, promoting this is not simple. I remember the first time I proposed this concept to PMC, they directly told me, "Ah Liang, you're restricting our capacity! Clients are waiting for goods, and you dare to set a WIP limit?" Their biggest concern at the time was delayed delivery dates.
But honestly, if the plan isn't stable, delivery dates become even harder to control. We later conducted data analysis and found that after implementing these stabilization measures, although capacity might have appeared "reduced" in the short term, the overall on-time delivery rate increased from 85% to 92%, and the quality defect rate also decreased by 15%. Because the process became smoother and quality became more stable. Therefore, the most common pitfall is the "fear of change."
One Thing You Can Do Today
Observe the machines you are responsible for, calculate the WIP quantity of their upstream and downstream processes, and then start thinking about how to make these "quantities" more stable.