That day, the boss stormed in, face grim, pointing at the report on the large screen: "How can this line's utilization rate be only 65%? Last month it was 80%, wasn't it? What exactly is going on?"
Brother Liang and I exchanged glances, our hearts sinking. This line was our plant's 'cash cow,' and if the utilization rate dropped this much, this month's shipments would surely fall through. The boss continued: "Didn't I tell you to use MTM analysis last time? Where's the report? Or are you giving me work sampling again?"
Now it was awkward. To be honest, MTM (Methods-Time Measurement) and the work sampling method were often used interchangeably in our department. Sometimes, to rush a report, we'd just grab whichever was convenient. Of course, the boss knew the difference, but those of us on the floor often just sought data without considering the true 'analytical value' behind it.
The Core Issue: MTM or Work Sampling?
Essentially, utilization rate analysis is about understanding how much time your machine spends in 'normal production' and how much time it spends 'idle' or 'handling miscellaneous tasks.' It's like your salary: how much is 'base pay,' and how much is 'overtime' or 'bonus.'
MTM analysis, simply put, breaks down your process actions to the finest detail. For example, an operator picking up a wafer, placing it into a machine, closing the door, and pressing start—the time for each action is precisely measured, even to 0.00001 hours. It's built on the foundation of 'standard time,' assuming your actions are the most efficient. Therefore, what MTM calculates is essentially how fast your machine or person *can* operate under 'ideal conditions.'
What about the work sampling method? This one is more grounded. It's essentially 'telling a story from observations.' Throughout the day, we randomly 'peek' at the machines or operators multiple times on-site to see what they are doing. For instance, if you observed 100 times in a day, and 70 times you saw the machine running production, 20 times it was undergoing PM, and 10 times it was waiting for personnel. Then, you can estimate that the utilization rate for this line is approximately 70%.
So the key takeaway is: MTM gives you the efficiency of 'what should be,' while the work sampling method gives you the efficiency of 'what actually is.' When the boss asks about utilization rate, he wants to know what's 'actually' happening, not how good it 'theoretically' can be.
How to Apply Them: Stop the Confusion!
To analyze a drop in utilization rate, you really need to first understand what kind of data you're holding.
- Advantages and Limitations of MTM: MTM is incredibly useful when you need to 'optimize processes' or 'evaluate new equipment.' Suppose you introduce a new machine, and MTM calculates that it can process 100 wafers per hour. This number represents its 'maximum' performance. If it only processes 60 wafers in actual operation, you immediately know there's a gap of 40 wafers, possibly due to slow loading/unloading, long batch changeover times, or abnormal machine downtime. However, MTM requires professional training, is time-consuming and labor-intensive, and isn't something that can be done daily.
- Practicality of the Work Sampling Method: When you need to quickly understand the 'current situation' or 'identify utilization rate bottlenecks,' the work sampling method is the preferred choice.
* Step Two: Plan Sampling. Set the number of samples and timing for each day, ensuring they are 'random.' You can't always observe right after PM finishes, as this would skew the data. For example, sample 20 times a day, with irregular intervals.
* Step Three: On-site Observation. Each time you go, record what the machine is doing.
* Step Four: Data Analysis. If you sampled 500 times, and 325 times you observed the machine in production, then your utilization rate is 325/500 = 65%.
Therefore, to answer the boss 'Why did the utilization rate drop to 65%?', you use the work sampling method to find that 'the machine was not in production for 35% of the time,' and then further break down this 35% into breakdowns, PM, or waiting for personnel. This is far more practical than trying to explain with MTM's 'theoretical value.'
The Most Common Pitfall: Numbers Lie, But So Do People
I once had a colleague who, to make reports look good, 'skillfully' avoided sampling during lunch breaks when using the work sampling method. As a result, the utilization rate soared, and the boss was pleased. But a few days later, production capacity still didn't improve, and the boss stormed in to complain again. Later, it was discovered that he had 'ignored' the machine 'idling' during lunch breaks.
Ultimately, the work sampling method's biggest fear is 'human bias.' If you have preconceptions and only want to see good results, your sampling won't be random enough, and the data will be distorted. Furthermore, if your activity definitions aren't clear enough, for instance, categorizing both 'waiting for material loading' and 'waiting for PM' as 'idle,' then you won't be able to precisely identify bottlenecks.
To give another example, one of our lines had its utilization rate consistently stuck at 70%. Brother Liang used the work sampling method to observe it. He discovered that for as much as 15% of the time, the machine was 'waiting for an engineer to troubleshoot abnormalities.' This figure was quite alarming. If we didn't have this precise data, we might have mistakenly thought it was due to excessively long PM times, leading to wasted effort. Therefore, the more detailed your activity definitions, the better you can uncover problems.
One Thing You Can Do Today
Re-examine the 'activity definitions' for utilization rate analysis within your plant, and make them more detailed!