Weight Measurement Uncertainty - F1 10 Kg weight comparison with 10 Kg E2 weight

[ahmadfy2003]

are there any one could help me to know how to calculate the measurement uncertainty for F1 10 Kg weight comparison with 10 Kg E2 weight

[Marc]

Any takers on this?

[skappesser]

I learned how to do this at NIST's Weights and Measures Division.
Do you have the measurement uncertainties as stated on the calibration certificates for the standards and check standards you used?
Do you know your standard uncertainty of your process? (Standard deviation)

If you know these 2 key items I will try to explain in this forum, if not - it would be impossible to arrive at an accurate estimate.

BUZZY

[Benjamin28]

I am going to ressurect this thread as I am working on a similar calculation. We utilize 40, 30, 20, 10, 5, 1, and 0.5lb weights. These weights are utilized in stress rupture, sustained load, and creep testing machines. I need to calculate an uncertainty budget for the weights themselves.

For calibration of these weights we utilize a set of master weights calibrated by an accredited outside laboratory which reports a MU for each weight size.

0.5lb- UNC=3.3 mg k2 coverage
1.0lb - UNC=3.3 mg k2
5.0lb - UNC=8.1 mg k2
20lb - UNC=65 mg k2

Now, if I take 10 measurements on the scale we use for calibration of our working weights against the masters for the 20lb weight I get a std dev of
0.020138lb, I then take this number and divide it by the square root of 10...this should be my type A unc?

So, at this point I would take the unc from our master weight at 20lb and divide it by 2 to return to STD UNC? Then square the type A, B unc values add the squares and take the square root of this number...then multiply this number by 2 to obtain expanded uncertainty.

Does this sound correct for determining the UOM of the 20lb working weights?

There is no way around it, Uncertainty of Measurements is most definately going to give me a headache!

[Hershal]

"Now, if I take 10 measurements on the scale we use for calibration of our working weights against the masters for the 20lb weight I get a std dev of
0.020138lb, I then take this number and divide it by the square root of 10...this should be my type A unc?"

That is correct.

"So, at this point I would take the unc from our master weight at 20lb and divide it by 2 to return to STD UNC? Then square the type A, B unc values add the squares and take the square root of this number...then multiply this number by 2 to obtain expanded uncertainty."

That is also correct.

Don't forget the other Type B contributions which generally will be Rectangular distribution, divided by square root of three (1.732). These contributions will likely include essentricity (repeatability of the specific positioning of weights on the device); if the weights are calibrated in another city (basically you have to ship them away) you may need to include the difference in local acceleration of gravity; bouyancy may not be too much of an issue for you, and likely barometric pressure also may not be; however, you need to consider temperature, that is the temp difference between the calibration temp and the actual usage temp.

Tylenol 3 is good for MU calculations.....Jack Daniels is better.....

Hershal