Retrotec USACE - Manual (Page 311)

Appendix F F17
is not limited to the maximum number of data points taken during the test. It
is recommended to take additional data points so in the analysis the “outliers”
can be omitted from the calculation procedure. Outliers are most frequently
caused by wind gusts, changes in wind direction at the time that data pair was
recorded, among other reasons.
One CFM (L/s) calculated from a
⌬P of 75-Pa (cfm @ 0.3 in. w.g., or
L/s @75Pa) will be calculated for both the pressurization and depressuriza-
tion tests. The average of those CFM values will be divided by the enclosure
area given in the project drawings to determine the normalized air leakage
rate. This mean value will be used as the basis for determination if the build-
ing meets or does not meet the air leakage requirement. The value is to be
rounded to the nearest hundredth. Therefore, a value of 0.255 cfm/sq ft does
not meet the requirement of 0.25 cfm/sq ft.
In addition to reporting the normalized air leakage as cfm/sq ft
envelope
@ 0.3
in w.g. or L/s*m 2
envelope
@75-Pa, the agency is also required to report the corre-
lation coeffi cient (r 2 ) and 95% Confi dence Intervals (95%CI) to determine the
accuracy of the data collected and the quality of the relationship between fl ow
and pressure that was established during the test. The 95%CI should be calcu-
lated in strict accordance with the methodology contained in ASTM E779-03
and the r
2 value can be obtained by data analysis of the plotted data.
In general, a narrower 95%CI to the mean value and higher r 2 value indicates
a clear relationship for the building’s air leakage characteristics was established.
For the collected data to be statistically signifi cant, the 95%CI must not exceed
᭙0.02 for mean values of 0.25 or less, which equates to approximately 8 per-
cent. For example, if the calculated mean value is 0.25 and the 95%CI is shown
to be 0.23 to 0.27, the test data is statistically signifi cant. However, if the mean
value is 0.25 and the 95%CI is 0.16 to 0.33, this exceeds 0.02 and indicates that
the data is not statistically signifi cant, and that a clear relationship between fl ow
and pressure was not established during the test; the test must be repeated. In
cases where the 95%CI exceeds
᭙0.02, but the upper limit is 0.25 or less, the
test would be considered a pass in spite of the statistical insignifi cance because
there is a strong likelihood that the building passes the requirement. Likewise,
the r
2 value must be above 0.98 for the data to be statistically signifi cant. Test
data should have correlations above 0.99.
Similarly, the pressure exponent, n, will also provide some insight as to the
accuracy of the test and relative tightness of the building enclosure. Exponent
values less than 0.5 or greater than 1.0 in theory indicate a bad test, but in prac-
tice, tests outside the range of 0.45 to 0.8 would generally indicate an inaccu-
rate test or calculation methodology. The reason comes down to basic fl uid dy-
namics and the characteristics of developing airfl ow through orifi ces, which is
too lengthy to discuss within this protocol. Except for very rare circumstances,
n values should not take on values less than 0.45 or greater than 0.8. If the n
value exceeds these boundaries, the test must be repeated. In general, an n
value closer to 0.5 indicates large holes that are much shorter in length than
they are wide, where an n value above 0.65 indicates the hole characteristics
that are smaller cracks or holes that are much longer than they are wide. Most