USE OM - JANURY 2017
12
a.
If all descriptors result in a rating greater or equal to 95%
of the certified rating, the original rating shall remain and
no first test failure statistic shall be considered.
b.
If a descriptor falls below 95% then the product shall be
considered non compliant and the product shall be re-
rated to the rating from the above procedure.
4.
At any time in the above procedure the Participant may concede
to a re-rate and testing shall cease and the rating shall be
revised to the most recent multi sample rating determined
above.
3.18.3.1.1
Statistical Analysis Procedure
. AHRI shall perform a
statistical analysis based on the number of samples tested.
Mean (
x
)
𝑥 =
∑
𝑥
𝑖
𝑛
𝑖=1
𝑛
Standard Deviation (
s)
𝑠 = √
∑
(𝑥
𝑖
−𝑥)
2
𝑛
𝑖=1
𝑛−1
Standard Error (
s
x
)
𝑠
𝑥
=
𝑠
√𝑛
Lower Confidence Limit (LCL)
𝐿𝐶𝐿 = 𝑥 − 𝑇 ∗ 𝑠
𝑥
Statistical Rating
=
𝐿𝐶𝐿
0.95
For 2 Test Samples,
T
= 3.078
For 3 Test Samples,
T
= 1.886
For 4 Test Samples,
T
= 1.638
The
T
value used shall be based on the number of samples tested
with a 90% confidence limit. To derive at the statistical rating, LCL is
divided by 0.95. The lower of either the mean or statistical rating is
used to compare against the certified rating. If the analysis shows that
the product rates above 95%
of the Participant’s rating, the rating shall
pass, if it does not, then the rating fails. The statistical analysis
calculations are taken from the Code of Federal Regulations (CFR)
Title 10, Part 429, Subpart B, §429.16 Central air conditioners and
heat pumps.
3.18.4
Re-Rate Following
Test Additional
Sample.
Following the decision to Test Additional
Samples, if the Participant concedes to re-rate their rating or it has been determined that the rating
is non-compliant through the statistical analysis, the model combination will be re-rated based on
the DOE Rating Procedures, 10 CFR Part 429, Subpart B, §429.16. Furthermore, all models within
the failed sample’s BMG will be re-rated the same proportionately.
3.18.4.1
Re-Rate
Following First Additional Test “Second Sample”.
If the
Participant concedes to re-
rate their rating after the first additional test or “second sample”,
the model combination will be re-rated to the second sample test results.
