“This unit’s not giving us accurate readings.” When I hear that statement about process instrumentation, I have to step back and ask, “Are you sure it’s an accuracy issue?”
The most basic problem is that people tend to group all unexpected reading errors under the heading of “Accuracy”.
So, here’s a quick primer on measurement terminology.
Definition 1: Accuracy
Accuracy refers to how close the measured reading is to the true value of whatever’s being measured. (Does it read what you expect it to?)
Think of shooting darts at a bullseye. Darts that hit the bullseye, or get closest to it, are considered to be accurate. The dart that hits the second ring, a number, or worse, bounces off the wall, is considered to be inaccurate (and is likely to produce a laugh or two from your competition).
Definition 2: Repeatability
Repeatability refers to precision: To what extent do repeated measurements produce the same result. (Does it read the same way every time you measure it?)
Go back to the target. You may throw three darts, aiming at the bullseye. None of them hit the bullseye, but they all hit the second ring of the 20. Your accuracy is off, because you didn’t hit what you aimed for. But your repeatability is high, because all three darts landed in the same area. They’re precise, but not accurate.
Definition 3: Stability
Stability refers to consistency: the amount of measurement change you can expect over a period of time. (Do changes around the measurement degrade the results?)
In instrumentation, it’s the unit’s ability to perform the same way after a year in your process that it performed on the calibration bench, or its ability to hold onto its calibration over a period of time.
In your dart game, your ability to throw the dart and hit your target can be interrupted by, say, someone bumping your arm when you throw. Or, as the night goes on, your arm tires, and your throws aren’t getting where you expect them to be. Or maybe someone opens a door, and a cold draft comes in from the outdoors. All of these things affect the stability of your performance.
More than likely, what you think of as an accuracy problem may, in fact, be a stability issue. Factors that influence stability include pressure, temperature, humidity, vibration, corrosion, electrical noise, and aging. The good news is this: If you start looking at things that can interrupt the stability of your measurement, your “accuracy” issue may be solved along the way.
Bonus material: Control tuning
I’ve been involved in several situations where what was described to me as a measurement accuracy issue was really a poorly tuned control loop. So, the measurement may be accurate, but improper tuning leads to changing values. But that’s another story for a different day… [See how easy it is to tune a control loop here.]
Process instrument technology training: on-site or online
As part of my job at Lesman, I teach our customers about process control technology, including sessions on the basics of loop control and measurement. If you’re in the Midwest, check the Lesman Training Center for the list of Lesman University process control courses that can be tailored to your needs. If you’re outside the Midwest, feel free to check our library of free recorded process technology webinars, for 45-minute lessons on individual technologies or applications.
What unexpected readings or measurement deviations have you chalked up to accuracy, and later found out to be caused by something else?
#1 by Bill Wilcox on January 6, 2012 - 3:57 pm
You accurately present this information precisely and with stable reasoning.
All kidding aside, I find your tips very interesting and informative.
Thank you for all you do.