Environmental Engineering Reference
In-Depth Information
Report). The EU Report concluded for this assessment type that RDT, 30MS with
the
first liter, 30MS with the second liter, and the average of the two 30 min
samples (MSA) perform the best, while the FF generally underestimates the amount
of lead in the sample (i.e. the ratio is strictly less than 1). However, the ratio COMP
as well as the prediction range for the ratio varied greatly between test areas A to K.
In some areas (see p. 42 of EU Report), the FF performed better than the other
protocols while it did not perform well in other areas. For instance, in area C (see
EU Report, Fig. 19, p. 42) FF is shown to perform better in terms of prediction
range for the ratio than 30MS,
first and second liters, and their average (30MSA)
while in area G it is shown to perform worse than the other protocols and in area A
it is performing just as well as the other protocols in terms of size of the prediction
ranges. For a given test area when prediction ranges are large (small), prediction
ranges tend to be large (small) for all sampling protocols. Only test areas G and H
seem highly variable in that all protocols seem to have a greater variation in average
ratios compared to lead samples from the measuring device (COMP) although all
protocols have a large prediction range. FF appears to be the most accurate sam-
pling protocol among the most highly variable test areas G and H. All other test
areas (other than G and H), have relatively the same prediction ranges for the ratios.
In other words, within a test area, the accuracy of a sampling protocol is relatively
the same for each protocol although we acknowledge that FF does under predict
lead in more instances than other protocols when compared to lead showed by the
measuring device; sampling protocols are likely to under or over predict lead values
simultaneously for a given test area.
The ability of the protocols to detect problem properties was assessed by ana-
lyzing the percentage of
“
positives,
”“
false positives,
”“
false negatives,
”
and
“
A test was considered positive if both the protocol sample and the
COMP sample produced a value greater than the MCL of 10
negatives.
”
g/L, while a false
positive would be considered a case where the protocol indicated a value higher
than the MCL when the COMP shows a value less than the MCL. A false negative
would mean that the protocol is indicating a value less than 10
µ
µ
g/L while the
measuring device (COMP) is showing a value greater than 10
g/L. A negative is
considered a case where both the protocol and the COMP show a value less than
10
µ
g/L. From the Report, the percentage of positives (the ability of the protocol to
detect problem areas) is highest for RDT followed by the average of RDT and FF,
followed by 30MS, and lastly the FF sample taken alone. False positives were also
highest for RDT and least for the average of RDT and FF. Once again the average
of RDT and FF performed well (or at least as well as the others) but the Report
overlooked this in the
µ
final assessment.
One explanation for the failure to identify problem properties was that it was
“
”
(EU Report, p. 47).
While this may be partially true, the failure to identify problem properties may be
due to characteristics in the sampling protocols themselves. The stagnation time in
the protocol in particular is likely to identify more problem properties if it is
increased (although this would not be cost-effective; see Fig. 2 on page 15 of the
Report for relationship between stagnation time and lead) or if stagnation times
likely to be caused by characteristics of the plumbing system
Search WWH ::
Custom Search