The US military has a “standard” of sorts for electronics. But what stands out is the section that says that 60% of the life expectancy is determined by environmental factors alone! That one single statement all but invalidates they entire standard in my opinion. Think about it…life expectancy has very little to do with product design or manufacturing.
Also there are extensive guides put together by a consortium of nuclear plants on life expectancy. They have hundreds of pages of pooled data on switchgear breakers. It is some amazing work. Each one has some very typical failures that I also happen to see in the field. You cannot make wide sweeping generic statements about anything. Things are wildly different between the various GE lines, Amptectors, Square D lines, and ABB.
Even with something as “simple” as a power distribution transformer conventional wisdom says that oil filled transformers last much longer than dry but that hasn’t been my experience. IF you do the annual oil testing and follow recommendations and keep it away from corrosion and don’t overload it excessively you can get 40-50 years out if an oil filled transformer. But fail any of this and 20-25 years is not unusual. With dry transformers there are several on board dragline excavators built in the 1970s that are still running fine despite exposure to mining environments in less than ideal service.
Consider for instance induction disk protection relays. Everything about these screams low reliability. Don’t get me wrong calibration drift is awful. Yet many have been in service for 50 years. In the 1990s we saw the rise of analog solid state protection relays. Everything about the design would seem to suggest it was a step up. However within 10 years the failures started to appear and the reliability was far worse than the old induction disk relays. Then microprocessor relays which one would think would be even worse came out yet have much lower failure rates. The only conclusion we can reach here is you can’t necessarily predict reliability based on design like the military design manuals.
The “predictive” crowd would have you believe you can reliably predict failures ahead of time. I would say this is partly true. My company makes good money on both predictive maintenance and remanufacturing. But the concept here is that you can predict and make repairs reliably. That is simply not true in practice. Plenty of things slip through “the net”. And as we all know you reach a point where you are just throwing money away on something that is well beyond useful life. This flies in the face of theoretical infinite life according to the predictive maintenance crowd.
Another major concern is technological life. When equipment goes obsolete, parts availability is a huge problem. For instance you can continue to repair old GE AK breakers, Powerbreaks, and Limitamo starters but the repair parts are also 30 years old, just as old as the equipment. Reliability of “new” is just as bad as the parts in service. In other words it’s a lost cause.
Nowhere is this more apparent than VFDs. The power components have a 10 year product life cycle (availability). The drive manufacturers in turn adjust the expected component life to 100,000 hours (10 years) and obsolete everything in…ten years. It can be shorter. I have a customer with a terrible hydrogen sulfide issue that only gets 36 months and others with nearly perfect HVAC pushing 20+ years. Durham county, NC had a couple early 1990s Westinghouse VFDs that finally failed almost 30 years later. That’s impressive!
So some of those recommendations are based on planned obsolescence. Can you exceed them or fail to meet them? Yes. So it’s a nice long term planning/budgeting target so you can estimate capital expenditures over time but individual failure rates are not nearly as predictable. As I tell customers that manage to maintain equipment well past ifs expected life, it doesn’t “owe you anything”. By that time the accountants should have easily amortized the value to zero. So when you start to see failures, even crazy ones like all the plastic falling apart or alignment issues and goofy corrosion issues, it’s time to do deep diving as far as repairs or just replace. But I would not suggest replacement based on age alone. I’ve seen far too many cases of old equipment that continues to perform at close to double it’s expected life.
Might want to look at Nolan and Heap. So these two guys started studying failure rates and curves. At the time the belief was everything follows the “bathtub curve”. At first you get a lot of infant mortality issues from things like installation errors and factory defects. After that equipment performs at low failure rates until it is worn out. If everything followed that you could just do an “overhaul” or “replace” based on age or say number of cycles of a starter. Nolan and Heap found in practice a half dozen different failure patterns. Worse still most electronics follows a true random failure pattern. This means “overhauls” don’t work. There is no “clock” to reset. Testing and repairs to detect hidden failures becomes your best option, I agree with their conclusions but aging is a definite factor too.
