That Day, the PM Came to Me and Asked: This Machine is 20 Years Old, Should We Replace It?
In the morning meeting, the PM suddenly posed a question: "Hey, our old 200mm PVD machine, should we replace it? Look at its Cpk report, it's been dropping recently; last month it was only 1.08, and DPMO is alarmingly high at 6210. The boss has been asking if its ROI is no longer viable." To be honest, hearing such words made my heart skip a beat. This kind of equipment aging management, to put it plainly, is the ultimate test of "when to retire the old gear and bring in new blood." This is not a small sum; it often involves tens of millions or even hundreds of millions. A wrong decision can lead to cost overruns, severe production shortfalls, and an incredibly displeased boss.
Where's the Problem? "Old" Does Not Equal "Broken"
Indeed, the PM is right; data like Cpk 1.08 and DPMO 6210 are genuinely poor. However, "equipment being old" and "equipment being broken to the point of needing replacement" are two different things. Many times, old equipment just needs a bit more care, like replacing consumables or adjusting parameters, and it might last for several more years. But if you blindly keep repairing it, and end up spending more than buying a new one, then you're truly throwing money away. So, the key is not whether it's "old," but whether it's "still worth repairing" and "how significant its impact is on the production line." In other words, we need to evaluate the overall benefits of "continuing to use the old machine" versus "replacing it with a new one."
How to Actually Do It? Understanding "Opportunity Cost"
Frankly, there's no standard answer for when to replace equipment. But I usually think about it from three perspectives:
- Maintenance Cost Trend: Look at the equipment maintenance records. Are you spending a fortune on repairs every month? If one repair costs 500,000 and only lasts a month, that's 6 million over a year. A new machine might cost 50 million but can be used for 10 years, which works out to 5 million per year. In this scenario, continuing to repair the old machine becomes more expensive. So, the key is whether the slope of the maintenance cost growth is already painfully high.
- Performance Degradation Level: Your old PVD machine's Cpk of 1.08 and DPMO of 6210 indicate low yield and high scrap rates, directly impacting downstream product quality. Calculate the losses caused by these rejects, as well as the production bottlenecks resulting from low yield – these are all opportunity costs. If you replace it with a new machine, where Cpk can stabilize at 1.33 and DPMO is only 3.4, the scrap cost savings over a year might be enough to cover several years of the new machine's depreciation.
- Technological Bottlenecks and Future Expansion Needs: This point is the most crucial. Some old equipment isn't just about whether it can be repaired; it simply cannot keep up with new process requirements. For example, if you now need to perform a 7nm process, but your 200mm PVD machine cannot handle 300mm wafers, or its stability simply doesn't meet 7nm requirements. In such cases, any repairs would be futile. If the company has expansion plans, the capacity limit of the old machine will also be an important consideration for replacement.
So the key is, you must quantify all maintenance costs, quality losses, production losses, and future technological requirements into monetary terms, and then compare them with the new machine's procurement cost, depreciation, and maintenance costs.
The Most Common Pitfall: Numbers Speak, But People Can Also Deceive
The most ridiculous thing I ever encountered was when a junior colleague once, to make the reports look good, transferred an old machine's maintenance costs to another machine's account. As a result, when the boss saw the report, he thought the old machine was super durable and had extremely high cost-effectiveness (CP value), almost deciding to extend its service life! Later, we investigated and found out why a new machine had outrageously high maintenance costs, which is how we caught this blunder. To put it simply, data is important, but the people who interpret the data are even more important. Never be deceived by partial data; look at the whole picture and long-term trends. Additionally, sometimes equipment suppliers will tell you they can upgrade your old machine for only one-third of the cost of a new one. At this point, you need to clearly ask if the performance, lifespan, and maintenance costs after the upgrade can truly meet expectations. Many times, an upgrade is merely a way to buy time, treating the symptoms but not the root cause.
One Thing You Can Do Today
Go back and look at the oldest piece of equipment you have. Pull out its maintenance costs, downtime, and yield performance, and create a trend chart for the past three years.