It’s surprising how many engineers misunderstand this crucial difference.
As metrology industry experts will agree, it’s surprising how many engineers don’t understand the difference between accuracy and precision. To laypeople, accuracy and precision are synonyms, both descriptors of how close an attribute is to an ideal. In the manufacturing world, however, the advent of statistical process control meant that no dimension or attribute is ever considered to be “right” or “wrong.” Attributes that are both accurate and precise file within upper and lower control limits, but it’s possible for a dimension or measured attribute to be highly precise, yet not accurate. Jim Anderton explains the difference between accuracy and precision.
Watch episodes of End of the Line on Engineering TV along with all of our other series.
* * *
Episode Transcript:
“Accuracy.” “Precision.”
To people who don’t work in manufacturing engineering or quality assurance, these two words are synonyms. For decades, advertising has touted everything from Swiss watches and German cars to Japanese audio equipment as being “precision engineered.” But for professionals whose job it is to see that the important attributes of mass-production goods stay inside their customers’ acceptable quality limits, accuracy and precision are not the same thing.
Let me illustrate what I mean with this simple example, using these darts and that dartboard. The attribute we’re measuring is a dart on target on the bull’s-eye. I’ll throw all 16 darts, then let’s take a look.
What we’re looking at here is an example of poor accuracy and poor precision. The darts are all over the board.
Let’s try it again. Now we have accuracy, but poor precision. If you use a coordinate measuring system with the bull’s-eye at the origin, then the mean of all the deviations of distance from the bull’s-eye would be zero. But the individual measurements are still all over the place.
Let’s try again. Now we have very high precision, but very poor accuracy. The darts are consistently hitting the same spot on the board, but they are not on target. This is actually better than it looks, because regardless of whether it’s the diameter of a hole, the fit up of a robotic assembly process or the amount of sugar on a doughnut, a condition that is accurate but not precise suggests that the equipment is not capable, in the quality sense of that word, or that it is operating at or outside its design limits.
I have seen this frequently with robotics and machine tools which are run at 100 percent feeds and speeds. It’s one reason why many manufacturing engineers operate equipment well within the maximum performance limits of production machines, despite the loss in potential productivity.
Accuracy without precision can be a challenging problem to correct, and an expensive one if the capability study shows that a piece of production equipment can’t deliver both. In the hole drilling example, the engineer would look for wear in a bushing or bearing, or maybe a concentricity issue or flex in the cutting head.
High precision with poor accuracy is usually a much simpler problem to solve. A robot, a cutting head or a nozzle that can consistently deliver the same outcome, but one that is off-target, can usually be adjusted to make it both accurate and precise—which is, of course, the goal. High precision with poor accuracy usually means a capable machine that is out of adjustment or has another problem such as fixturing.
Of course, everyone wants both accuracy and precision, =like this Boeing robotic assembly system for 777 production. But if you’re troubleshooting a production problem and have to choose one, I suggest looking at precision first, and accuracy later. And for engineers who have to write reports—and who doesn’t—a really good idea may be to stop using the word precision at all, mainly because it’s entirely possible that the supervisor reading the report doesn’t fully understand the difference between accuracy and precision…. and confusion can result.
I suggest using the word “repeatability” instead of precision in those reports. Accuracy is good, but precision, or “repeatability,” is really what we’re all after.
