Process measurements are critical with respect to product quality and safety. Verifying/calibrating all instruments properly in a timely manner is an important aspect of ensuring that your product is manufactured correctly. I have seen instruments drift because of their natural tendencies and drift that was “self-inflicted.”
Manufacturer accuracy promises are not valid to an auditor checking your records. A technician cannot just say the manufacturer promises it will stay within a certain accuracy for a year or more so you don’t have to verify it. He or she may be using the best, most stable instruments available, but if that instrument was installed in extreme conditions, it could be artificially harming the stability or shortening the lifespan.
What I mean is the problem may be “self-inflicted,” and one would never know unless the data was captured on a regular basis to verify trending. One example I had personally while training a customer at their site involved a high accuracy transmitter that showed far greater drift than documentation acknowledged. The problem was a high-temp steam line venting right on the transmitter. It was simply a poorly chosen install location. The fact that the transmitter was drifting by such a large margin was not the fault of the manufacturer. It was the fact that it was installed in the direct path of a steam vent.
If the technician simply trusted the instrument to be within spec, that would have been incorrect. This example proves there can be instances where technicians may not know the whole picture.
I mentioned “self-inflicted.” The other side of drift is a natural tendency of individual instruments. Some are not stable and some are so stable that it is necessary to look out to the fourth decimal to see the drift. The later looks like a straight line when viewing a graph of the error.
To define drift, let’s first look at accuracy. The instrument manufacturer of my example transmitter defines accuracy as a percent of span. As an example we will say the transmitter is advertised to be accurate to within 0.25% of span, if your span is 200 degrees (0.25% span * 200 degrees = 0 .5 degrees). In this case, the transmitter is accurate to half a degree. In addition to accuracy is stability. Stability can be looked at in terms of different variables such as repeatability, linearity and hysteresis. Alterations to stability factors can be referred to as drift. Drift is one aspect you want to capture when checking your instrumentation and is the reason your transmitter may be showing error beyond the 0.5 degree accuracy guarantee.
It may be a large or minute movement, taking 10 years of drifting before any adjustment is required. Your process may have a requirement of being within a tolerance of 0.5 % to maintain product quality. In this example, let’s say the instrument stability calculates to a drift of 0.1% span a year. If your device accuracy started out at 0.25% span, it would only take three years for your instrument tolerance to exceed your process tolerance if it wasn’t verified or calibrated during that time.
Manufacturer’s promises are not considered proof of an instrument being within tolerance. Environmental factors, such as extreme or caustic conditions, can cause an instrument to drift quicker than expected, and can even be “self-inflicted.” Therefore, all instruments, smart and analog, or whether old or new to the field, should be verified and calibrated on a regular basis.