Geoscience Reference
In-Depth Information
Original Dup
5.4.4
Evaluation Procedures and Acceptance
Criteria
Re .
l Error =
<
0.10
(5.1)
Original + Dup
2
ˆ
Š
The unit of acceptance/rejection should be a complete batch.
Individual samples should not be sent out for re-assaying.
The numbers of samples in a batch is variable, but 40 is
commonly used. Laboratories tend to process samples by
batches so whatever problem caused the check sample to
fail is likely to have affected the remaining samples in the
batch.
All acceptance/rejection criteria should be enforceable as
part of the agreements with the laboratories, including re-
assaying if the batch is not compliant. The exploration or
mining company should perform, as part of its QA/QC pro-
gram, the acceptance/rejection tests consistently using the
same procedures. Transparency and good relations with the
laboratory are always necessary to ensure a successful QA/
QC program.
The exploration or mining company should not shortcut
the QA/QC program, and should allow sufficient time, bud-
get, and contractual arrangements for a significant number
of re-assays. The program should be implemented on an on-
going basis and not at the end of drilling campaigns or pre-
determined time periods.
The expected accuracy and precision depend on the type
of mineralization being sampled. Sampling gold is particu-
larly difficult especially if there are coarse particles. Certain
base metal deposits may be easier to sample. The acceptance
criteria will change according to the type of mineralization
being sampled. In the case of gold, commonly accepted cri-
teria include:
1. Coarse blanks : 80 % or more of these samples should
return with a value less than or equal to three times the
detection limit. Thus, at least four control samples are
required to make a decision, which implies 8 batches
(one coarse blank in every other batch). Another way to
express it, is to say that 1 in every 5th blank may fail the
criterion.
2. Pulp blanks : 90 % or more of these samples should return
with a value less than or equal to three times the detection
limit. Therefore, 1 in every 10 can be above the accepted
limit.
3. Standards : In all cases, standards have to fall within the
accepted tolerace limits of the certified reference value .
This can be 2 or 3 times the standard deviation or some
intermediate value, depending on the round robin results.
In the case of duplicates, the suggested criteria for pairs of
samples (original-duplicate) with averages equal or to great-
er than 5 times the laboratory's detection limit (DL), the fol-
lowing formulas are suggested:
4. Pulp duplicates : 90 % of the pairs' absolute relative dif-
ferences equal to or smaller than 10 %. The absolute rela-
tive difference for each pair is defined as:
1. Coarse duplicates : Using the same Eq. 5.1, 90 % of the
pairs' absolute relative differences equal to or smaller
than 20 %.
2. Field duplicates : Using the same Eq. 5.1, 90 % of the
pairs' absolute relative differences equal to or smaller
than 25 %.
3. Additionally, for pulp duplicates, if the absolute value of
the difference [Original-Duplicate] (numerator in Eq. 5.1)
is equal to or less than two times the detection limit (DL),
the pair is accepted.
4. For coarse duplicates, if the absolute value of the differ-
ence [Original-Duplicate] (numerator in Eq. 5.1) is equal
to or less than three times the detection limit (DL), the
pair is accepted.
The standards should be plotted to verify the laboratory's
performance over time. The graph typically shows the ex-
pected value, the upper and lower acceptance limits, and the
assay results for each control sample inserted. This allows
for trends to be detected. For example, if the control samples
are consistently above the expected values (but still within
the acceptance limit), there may be a small persistent bias
that the laboratory should be notified of and correct.
In performing these tests, the practical detection limit
(DL) should be considered. The practical DL may be dif-
ferent (and generally higher) than the nominal or theoretical
DL as stated by the laboratory, because it takes into account
the lower precision of analytical methods when working at
or near their nominal DL. A higher DL should not cause a
problem as long as it is much lower than the mineralized or
economic cutoffs.
The sample preparation and the analytical laboratories
should be supervised constantly. Ideally, the responsible
person should make surprise visits to each laboratory on a
regular basis. These informal inspections should result in a
brief report describing the laboratories operating conditions,
cleanliness, orderliness, sample handling procedures, and,
most importantly, the extent to which they are correctly im-
plementing the prescribed sample preparation and analytical
protocols. Photographs should be used to record and docu-
ment each visit.
5.4.5
Statistical and Graphical Control Tools
There are several tools that can be used in analyzing and
describing QA/QC information. Among the alternatives, the
suggested basic tools are:
• Error histograms and basic statistics: these should include
the relative errors defined in the equations above for
 
Search WWH ::




Custom Search