76/211/EEC 2.3.3.1 Criteria for Non-Destructive Testing

[Darius]

According with the 76/211/EEC

2.3.3.1.Criteria for non-destructive testing
Number in Criteria __* Acceptance

100 to 500(inclusive) * x_average >=Qn —0.503s

>500 _____________* x_average >=Qn —0.379s

2.3.3.2.Criteria for destructive testing

Number in Criteria ___* sample Acceptance

(=100) ____________* x_average >=Qn —0.640s

Does anybody know where does 0.503, 0.379 and 0.64 come from?

 

[Rick Goodson]

Darius,

I am not familiar with that spec, but have a feeling it has to do with the area under the curve for a distribution. It looks like a comparison of an empirical reading to a calculated number that reflects a point on the curve. What is Qn?

Rick

 

[Darius]

Sorry, I found it

In this formula:
Qn =the nominal quantity of the prepackage,
n =the number of prepackages in the sample for this check,
s =the estimated standard deviation of the actual contents
of the batch,
t (1 —alfa )=0.995 confidence level of a Student distribution with ν
=n —1degree of freedom.

Qn — s/sqrt * t (1 —alfa )

In the table

for non destructive and
Units/batch (100-500] , the samle must be 30
for [500..), the sample must be 50
and for non destructive, Sample size must be 20

So in Excel,
tinv(0.1,30-1)/sqrt(30)= 0.503
tinv(0.1,50-1)/sqrt(50)= 0.379
tinv(0.1,20-1)/sqrt(20)= 0.640