Geoscience Reference
In-Depth Information
The San Cristóbal case presented here illustrates some
of the practical aspects of implementing a simulation-based
grade control method (Isaaks
1990
; Aguilar and Rossi
1996
;
Rossi
1999
). The benefits of the method are evaluated based
on production data, mine and mill reconciliation data, and
cash flow analysis. San Cristóbal is a now-exhausted me-
dium-sized open pit gold and silver mine, which processed
about 10,800 metric tons daily of ore grading approximately
1 g/t Au and between 4 g/t and 6 g/t Ag. The mine operated
on 5 m benches, and blast holes were used for breakage and
to obtain grade samples from the pit, spaced at about 4.5 m
and sampled over the entire 5 m bench height. Blast holes
were drilled, sampled, and loaded with explosives on a daily
basis, typically one blast of 300-400 holes per day. The ore,
after being crushed in three stages, was heap-leached, and
Au and Ag were recovered by passing the enriched solu-
tion through six activated carbon columns. Finally, an Au
and Ag doré was produced with approximately 27-30 % Au.
Up until the introduction of a new grade control method, the
mine produced about 65,000 ounces of Au per year.
also determined that the grade control method used was los-
ing significant quantities of gold and processing waste. This
was because of the difficulty in drawing panels containing
homogeneous ore zones. In an effort to remedy the situation,
a conditional simulation method combined with economic
optimization was designed, tested, and implemented.
14.6.2
Maximum Revenue (MR) Grade Control
Method
Conditional simulations provide conditional probability
distributions which, in conjunction with relevant economic
parameters, can be used to minimize losses caused by im-
perfect selection. Imperfect pit selection and misclassifica-
tion will always occur. The main objective of the method is
to minimize economic losses. This optimization is achieved
based on a set of economic parameters and the probability of
occurrence of ore for every node within a blast. The maxi-
mum revenue grade control method (MR) uses loss func-
tions as a basic tool for decision-making. The MR method
requires two basic steps:
1. A set of conditional simulations is obtained, providing
an uncertainty model about the grade at a specific point
within the blast.
2. An economic optimization process using Loss Functions
is implemented. It is designed to obtain the economically
optimal ore/waste selection. The loss function quanti-
fies the economic consequences of each possible deci-
sion, minimizing losses, see Chaps. 12 and 13 and Isaaks
(
1990
).
The simulation model is based on blast holes, and the pro-
cess is run on a daily basis. Quality of blast hole sampling,
sample preparation and assaying had been addressed before
the implementation of the MR method with a detailed sam-
pling heterogeneity study (Pitard
1995
). The procedures and
protocols implemented were deemed to achieve a targeted
15 % fundamental sampling error variance for the blast holes.
The conditional simulations for the San Cristóbal open
pit were built on a 1 m × 1 m × 5 m grid using the Sequential
Indicator method. Shifts in the attitude of the ore controlling
structures required the separation of the data into different
populations. Evaluation of high-grade populations was re-
quired to control extreme grades in the simulation and cor-
rectly reproduce the observed variability (Parker
1991
; Rossi
and Parker
1993
).
Indicator variograms were modeled from blast hole data,
and a number of simulation parameters were optimized. This
included minimum and maximum data value used; maxi-
mum simulated value allowed; maximum number of con-
ditioning data to be used; and anisotropic search distances.
The conditional simulation models were validated against
original data. Figure
14.57
shows four conditional simula-
tions obtained for Level 2,345 m.
14.6.1
Geologic Setting
Gold mineralization is very erratic, with a highly skewed
distribution that makes grade modeling and resource/reserve
prediction difficult at any scale. Gold and silver minerals are
associated to a sub-volcanic intrusion, mainly consisting of
rhyolite, breccia, quartz-feldspar porphyries, and occasional
dacitic dikes. Alteration is typical of porphyry intrusions,
zoning from a central potassic alteration to an intermediate
alteration zone characterized by actinolite and epidotes, and
to an external halo of propyllitic alteration. Superimposed to
these alterations a different sericite-quartz alteration has been
identified, associated to veining and significant shearing.
Mineralization occurs within discontinuous structures,
oriented North to NW-SE, within a dilational zone limited to
the north and south by two shear zones. The structures host-
ing the mineralization are typically 0.1-1.0 m in width. Gold
is present in a quartz-pyrite association. When the structures
intercept more favorable lithologies, such as breccias, gold
mineralization appears also disseminated into the host rock.
The geology is well understood but provides few markers
with no visual indicators for the occurrence of gold and silver
mineralization. The presence of veins does not ensure the oc-
currence of gold, and not all of the gold is strictly confined to
the stockwork-like fractures. The short- and long-term pro-
duction reconciliations obtained from 1991 (when the mine
began operations) until mid-1994 were poor. Estimation
methods for long-term mine planning were originally ordi-
nary kriging estimates, controlled by the use of geologic and
grade envelopes. A significant improvement was achieved
by using Multiple Indicator Kriging instead of Ordinary
kriging, still constrained by a low-grade envelope. But it was
