I have a truck-load of pressure and temperature transducers that I deal with. These transducers are calibrated once a year. We have a couple of major shut-downs through the year. The transducers are calibrated in groups according to their turn.
Sometimes a transducer goes out of calibration and we can plainly see that this is so. It has an effect on the process.
Most of my transducers have back-ups. Both transducers (the Primary and Secondary for any given part of the process) are running but only the Primary is normally used for control.
If the Primary fails, the process automatically switches to the Secondary. At that time, a visually painful remainder is posted on the screen every cycle until I, or one of my compadres, acknowledges the failure and puts the message into a less aggressive mode. The replacement is installed as soon as practical.
If, on the other hand, through simple observation, it is seen that a Primary transducer has gone out of calibration, the Primary can taken off-line and the Secondary will automatically jump in to control that part of the process. (Of course, this is playing the Odds-Game"; the odds being that the Secondary is OK.)
IF, the Primary transducer has not gone too far out of calibration, I attempt to "calibrate" the transducer by software means. This means adjusting the scaling. Hopefully, the transducer has still maintained a certain amount of linearity. Even if it lost a large part of its linearity (at the extremes), I try to take advantage of what linearity there is. Again, HOPEFULLY, the linear part is across the normal (or critical) operating range.
I have to assume that the Secondary, which is now controlling the process, is reasonably calibrated and produces reasonable results.
So, using the Secondary as my "Yard-Stick" I try to "re-calibrate" the Primary using software. Since both are operating together I can follow and compare the response of each. Sometimes that adjustment takes me outside of the normal areas. The numbers you finally end up with depend on the linearity of the transducer.
I then change the operating status of the Secondary to Primary and the Primary to Seconday. The less reliable transducer becomes the back-up until such time that the scheduled calibrations happen again.
Of course, if the new Primary goes bad, then the both of them are scheduled ASAP!
The point of this whole walk-through-the-park is that sometimes, when necessary, software calibrations are done on transducers. And sometimes those calibrations take the numbers outside of the normal range.
It certainly is not the right way to do it, but it can produce reasonable results - depending...!