if they are used not only when the line stops, but more regularly when the line is still working, and if data is taken with care and somebody takes the time to compare with previous data, then you may be able to spot media degradation which can take place very slowly, before such an issue can stop the machine.
"Monitor the process" is an oft-ignored cry.
Another drbitboy story, so TL;DR
My dad worked for GE Large Steam Turbine, and would on occasion be sent out on a post-mortem for a
water induction event. He said it was funny because the customer always called it "our turbine," except when something went wrong, at which point they would call GE and say "your turbine has a problem."
Background
Steam turbines have extraction heaters, where some of the steam going through the turbine is extracted and sent to a heat exchanger, where it condenses as it pre-heats feedwater that is heading toward the boiler. The steam condensate collects in a sump under the exchanger tubes, and drains to the condenser along with all the other extraction heaters' condensate.
What became apparent during one post-mortem is that sump drain was blocked, probably a closed valve. In that case, the extracted steam continued to condense, but floods first the sump, second the heat exchanger, and third the extraction piping coming out of the turbine itself. At that point, it only took a small pressure swing (the steam/condensate-side extraction heaters are all connected, with higher-pressure heaters upstream) to push some liquid condensate into the turbine shell, and that is the end of that turbine; you're lucky if pieces of turbine blade don't end up flying around like shrapnel.
There are temperature sensors on the feedwater before, between, and after all extraction heaters, and if you plot them all on a single (analog) strip chart recorder, they show up as a series of parallel lines, each line normally higher in temperature than the last, as each heater has a "temperature rise." If a heater is flooded, little to no heat is being transferred and its feedware inlet and outlet temperatures will be the same so the temperature rise will be zero, and the operator can see at a glance, from the resulting two lines on top of each other, that summat is amiss.
However, in this plant, the control room had been somewhat digitized and that strip chart recorder removed. The feedwater temperatures were "on-demand" only, which meant once a shift or day someone would roll a thumbwheel and push a button to select each channel, and record the data on a paper form on a clipboard.
The post-mortem
My dad asked for the most recent form and saw that one extraction heater had no temperature rise. He went back a day and saw the same thing, and the day before that, and stopped after going back a
fortnight or so because it was obvious that heater had been flooded for some time.
Anyway, that is one reason why "monitor the process" was near and dear to his heart.
It's ironic that, on the boiler side of power generation, a long history of "puffing the boiler" events probably going back to the 19th century meant that the boiler operators were rigorously trained to watch their process constantly, while on the turbine side, in the same control room, that mindset has not penetrated to this day: the link above about water induction ca. 2015 says there were over a dozen events that year (there are other modes of water induction to be sure, but I doubt many of them do not show up in the process first).
