**Question: **

In the calculation example in Annex-D (p.43) Probability of wear failure (p.48) is calculated as:

Pwear = 5% or lower.

But if we follow Annex B-Normal Gaussian probability (p.39), for:

y= 0.425354

muy = 0.215956

Sigmay = 0.112623 and

x = ((y-muy)/Sigmay) = 1.859273

Evaluation of Q it stated as:

if abs(x)>1.6448

Q=0.05

Since our value for x, 1.859273 is greater than 1.6448, our value for Q is 0.05.

The section also stated as:

if x>0,Ìý

POF= 1-Q.

Since in this case value for x is positive, our probability of failure is POF=1 â€“ .05 = 0.95 or 95% which clearly does not agree with the 5% given on page 48.

Ìý

**AGMA’s Response:**

Your question has uncovered a bug in Annex B of Å·±¦ÌåÓýÖ±²¥925-A03.Ìý In that Annex, |x| > 1.6448 means that the example is on a tail of the Gaussian distribution.Ìý

If x > 0, the example is on the positive tail.Ìý For scuffing, this indicates that the probability of scuffing is greater than 95%.Ìý

If x < 0, the example is on the negative tail of the Gaussian distribution and the probability of scuffing is less than 5.

With wear, the probability of failure is opposite, with a high risk occurring when x > 0 and a low risk when x < 0.

It should be noted that Å·±¦ÌåÓýÖ±²¥925-A03 is currently being revised and as part of this revision, both Annex A and Annex B will be updated to account for the difference between the scuffing and wear risks.Ìý In the meantime, it is recommended that the probability of wear failure be calculated by placing a value of -1 in equation B.1:

The probability of scuffing failure should continue to be evaluated per equation B.1 as it appears in the information sheet.