The architect says this (bridge/power plant/building) will stand for (20/30/40) years with proper maintenance. Then, we should outright replace it. We know it'll cost x dollars now, plus y dollars of the life of the item. Sounds good, so we buy in.
At the end of the lifespan, somebody who is not that architect says we can't afford to replace a (still perfectly good) piece of infrastructure. Let's agree that if we (inspect more often/inspect in greater detail/upgrade this piece here), we can get (10/20/30) more years of life out of it. Y'know, I can already hear the original architect screaming "That isn't what I said!".
The original architect necessarily has to be very conservative in his estimates because he has, in your example, 20-40 years of future uncertainty messing up his predictions. He cannot actually know how high the humidity will be, how much the ambient temperature will fluctuate, how much the soil will shift, what sorts of loads the facility will come under, etc., except as some form of probability distribution. And this distribution becomes more uncertain the further into the future he tries to plan it.
After the 20, 30 or 40 years have actually passed however we know all these things, or can find them out, pretty exactly. And if life has fared gentler with the facility than the architect's worst fears accounted for then there may still be decades of useful life left in it. In this case it is perfectly sensible to make a new maintenance plan and life estimate for it, and then take it from there.