Other Dimensions: Measurement Uncertainty
August 27, 2007
Editor’s note: This is the third in a three-part series on measurement uncertainty.
In this brief overview of measurement uncertainty, we have arrived at a value for use in an uncertainty budget. There are several ways to use that value and the budget that created it.
First, it is an indicator of how exact a specific measurement process is. The smaller the value, the closer the reading will be to the true size. Remember that every measurement process has uncertainty attached to it; no one is exempt or perfect. Because different processes and equipment produce different values, you can see which is best for your application when compared to the tolerance involved. A typical worst-case scenario would allow the uncertainty to be within 20% to 25% of the tolerance the dimension is supposed to be within. Technical limitations may mean this cannot be achieved, in which case supplier and customer should agree on what percentage both can tolerate.
Another-and probably more controversial-situation arises in applying uncertainty. When the reading of size is on the top limit for example, the uncertainty attached to that reading means that the true size of the part could be over that limit by the uncertainty amount. In Europe they overcome this by insisting that the reading plus or minus the uncertainty shall be within the size limits. This is not too onerous when dealing with most machined parts. However, when it comes to gages, the very small tolerances can make this a difficult state to achieve.
A machinist does not want his tolerance eaten up by the measurement uncertainty under the European system. So there are two choices open to him: reduce the uncertainty with a better measurement method or get a concession from the customer. To avoid problems, I suggest that this matter be discussed with the customer so everyone understands how the uncertainty will be applied.
Another application of measurement uncertainty is for resolving measurement disputes. Because both parties to a dispute will-hopefully-have an uncertainty budget, you start by reviewing values each has rather than ranting on about who is right and who is wrong. If one party doesn’t have a budget, go over the procedure and construct one.
If the two readings are different but the uncertainties overlap, there’s a good chance that is where the true value lies. If this doesn’t happen, then you exchange copies of the uncertainty budgets to see if there are any differences between the parties with respect to the elements listed and how they’ve been processed.
By following this method, you’re not fighting over big right/wrong issues but small line items on the budget. Changes don’t require a lot of face- saving on anyone’s part.
If you are unable to agree on some elements of the budgets, decide on a referee that knows the subject and have him review them. Both parties may have to re-think parts of their budgets.
You could end up with a situation where both budgets generally make sense to the combatants but they are still not within shooting distance of each other. This can occur when elements that have been processed properly had an incorrect starting value.
Some areas to look at in this case would be repeatability that can be resolved by a simple test. The uncertainty attached to the calibration of an instrument or setting master may not be realistic. Part geometry variations may be the source of problems when one person’s instrument contacts a diameter at two points while the other does so using three points.
Noncontact measuring equipment vs. contact measuring equipment can be the reason the measurements are so far apart. This can be due to surface finish and jet sizes when air gages are involved. Spot size and distortion can create problems when laser or optical systems are being compared to contact gages.
The ISO document “Guide to the Expression of Uncertainty in Measurement,” commonly referred to as “the GUM,” is used internationally for computing uncertainty along with similar publications from NCSLI. Both organizations have other documents that will assist you with this subject, as will some ANSI standards that now give examples of budgets.
May your calculator never blow a digit.