Geoscience Reference
In-Depth Information
Area Phase 7 (Lince, Fig. 14.5 a-c): the simulation
model predicts that the block model is conservative (i.e.,
there is upside potential) for cutoffs below 0.7 % TCu,
which is the cutoff of interest, while the opposite is true
for 1.0 % TCu cutoff and above. This is true for each clas-
sification category considered, although more so for the
inferred resources. At the 0.5 % TCu cutoff, blocks cat-
egorized as Measured are within +16 %/- 8 % of the pre-
dicted Resource Model grade, while Indicated resources
are within +30 %/-0 % of the expected grade. The Lower
and Upper probability limits for the Inferred resources
are within +12 %/+39 % of the expected resource model
grade. The grade-tonnage curve for the simulated model
was at the time a major concern, because it appeared as
if the Resource Model was predicting much higher grade
material at higher cutoffs. The issue was eventually
resolved through infill drilling, and indeed the high grade
distribution flattened as predicted by the conditional sim-
ulation model.
Area D4 (Open Pit, Fig. 14.6 a-c): the simulation model
predicts for all cutoffs that the block model is conserva-
tive, although clearly not as much for the higher cutoffs.
There is a clear difference in the width of the upper and
lower limits between the measured and the indicated cat-
egories (Fig. 14.6 a, b), and more so with respect to the
inferred resources (Fig. 14.6 c).
Area D1/D2 (Underground, Fig. 14.7 a-c): In the under-
ground areas, the cutoff grade to be considered is the
higher 1.0 % TCu. Therefore, the resource model appears
optimistic, contrary to the open pit areas. At the time of
this study, most of the resources in this area were clas-
sified as indicated, and prior to significant infill drilling
already planned. The probability intervals for these indi-
cated resources at the 1.0 % TCu cutoff is - 17 %/+6 %,
i.e., the actual grade can be up to 17 % lower and 6 %
higher than the predicted grade, according to the simula-
tion model.
Area A1 (Underground, Fig. 14.8 a-c): In this area the
average of the simulation model predicts a lower grade
for all classes (at a 1.0 % TCu). Note that the grades in
this area are generally higher, compared to the other areas
discussed, noting also that most of the resources in this
area are indicated. However, the probability intervals are
+3 %/-12 % of the predicted resource model grade for the
indicated category, which implies a less variable grade
distribution in this area compared to the other three areas.
The amount of information that can be derived from a simu-
lation model is significantly larger than what has been pre-
sented here. As a consequence of a similar analysis to the one
presented here, infill drilling campaigns, mine call factors,
and other risk mitigation measures were taken to ensure that
the predicted ore through the plant is achieved.
In addition, technical personnel and management have a
tool to better understand the consequences of the resource
classification schemes, and their significance. None of this
detailed analysis is possible with traditional resource clas-
sifications, so it is reasonable to back up the traditional re-
source classification with a probabilistic analysis based on a
conditional simulation model.
14.6
Grade Control at the San Cristóbal Mine
The most important task in daily life of an open-pit mining
operation is to select ore and waste. This grade control pro-
cess can be a simple ore/waste decision or a more compli-
cated process because there may be different destinations or
stockpiling and blending requirements.
Perfect selection, that is, making no mistakes in decid-
ing the destination of every ton of material mined out, is
impossible. Sampling errors, estimation errors, limited or
bad information, and operational constraints and mistakes
always result in ore loss and dilution, which in turn lead to
economic losses. In extreme cases, these losses can be seri-
ous enough to compromise the profitability of the operation.
Poor grade control may cause an operation to fail, such as
the São Vicente mine in Brazil's Matto Grosso in the mid-
1990s. Studies on mine failures and not realizing expecta-
tions have been completed be several researchers, for ex-
ample Burmeister ( 1998 ) in Australia, and Knoll ( 1989 ) and
Clow ( 1991 ) for Canadian operations. In many cases, the
failures to meet expectations have been attributed to poor re-
source estimation and lack of grade control. Minimizing ore
loss and dilution is critical to a successful operation, since
every mistake made detracts from the maximum amount of
ore that could, theoretically, be recovered from the pit. The
San Cristóbal mine discussed here was on its way to join the
unpleasant list of failed operations, until improvements in
grade control turned around the mine.
In open-pit mines it is often difficult to define accurately
the position of the dig boundaries prior to loading, particular-
ly where there are few or no visual markers. Commonly used
grade control methods include simple visual observations of
the blast holes grades, some form of blast hole averaging into
panels of arbitrary shape, or polygonal methods. In more re-
cent years, several forms of kriging have gained acceptance,
including ordinary and indicator kriging. Even more recently,
grade control methods based on conditional simulations and
economic optimization have gained some popularity.
Conditional simulation-based methods may be better than
more traditional grade control methods, including kriging,
when (a) ore and waste populations are intermixed, making
it difficult to identify ore pods without leaving ore blast holes
unrecovered; similarly, the recognizable ore pods may have
significant amounts of waste within; (b) no visual markers
are available; even if higher-grade controlling structures are
identified, there is never assurance that they are mineralized;
and (c) grade variability is significant.
 
Search WWH ::




Custom Search