Geoscience Reference
In-Depth Information
Fig.
11.10
, the supergene enriched unit is being compared.
Note how the December model is less smoothed (more high
and low grades) than the February model. Indeed, one of the
objectives of the update was to better control smoothing.
Whenever possible, a detailed decomposition of the main
factors that are thought to affect the comparison must be
made. All differences must be explained in terms of data den-
sity and values, geologic modeling, or other relevant factors.
Another option is to obtain alternative block models for
comparison, using a different estimation or simulation tech-
nique. Simple models such as a Nearest Neighbor model can
be used to check the resource model. Although the models
are not comparable block by block because of the different
methodology used to obtain the grades, they may provide a
general indication that the resource model is reasonable.
Care must be taken to understand and state, where ap-
propriate, the characteristics of each model obtained, and
the differences expected to be encountered due to their
respective properties. Additionally, a clear definition of an
acceptable match is required.
concept goes beyond common industry practice because
reconciliation is typically seen as a material movement and
material balances accounting tool. If used as an optimization
tool, the basic data used to analyze the performance of the
predictive has to be sufficiently accurate and precise.
The main purpose of any reconciliation program in a pro-
ducing mine is to properly account for all material mined, both
ore and waste. But it can also be used to assess the accuracy
of the resource and reserves models, therefore allowing for a
more accurate valuation of the mining property at all times.
These objectives are interrelated, and they all share some
basic requirements if the results are to be meaningful. The
most important requirement is, naturally, reliable data. This
is not trivial, since many operations do not sample mill head
grades and tonnages adequately. Automatic sampling de-
vices placed on the input stream to the processing facility
may be expensive, but are invariably well worth the expense.
Unfortunately, this is often not realized until well into the
operation's life, if at all.
Sometimes there are other issues associated with sam-
pling some of the processing streams. For example, run-of-
mine (ROM) material cannot be sampled because the rocks,
as blasted, are loaded directly into leach pads. Without fur-
ther size reduction, sampling of ROM material is impracti-
cal. In general, leach operations that stack coarse material
do not lend themselves to reliable sampling of head grades,
and they must rely on blast hole information (pre-blast) to
provide an estimate of grade loaded to heaps. Reliable head
tonnages may not be available also, unless a careful program
of truck weighing is implemented or each truck is equipped
with a weightometer. The weightometer should be calibrat-
ed on a regular basis. Some operations, for simplicity, use
truck factors, derived from long-term averages of material
delivered to the mill. The use of truck factors is unreliable
and non-specific to the area or period being mined, and thus
should be avoided.
At some operations blast hole data may not be reliable;
the operation may not sample blast holes; or may not sample
all
the blast holes available. In underground mines, sampling
grades from production stopes is difficult at best, often relying
on grab or muck samples to inform stope grades. Bulk ton-
nage mining methods, such as sub-level or block caving, may
sample sufficient tonnage of ore at the stopes' draw points, but
this is not always done. These problems should be assessed
to evaluate whether implementing a detailed reconciliation
program would result in reliable information, suitable for
long- and short-term model calibration. If certain changes or
additions are required, such as a new sampler for head grades,
it is generally feasible to perform a cost-benefit analysis that
would allow management to make informed decisions.
In addition to having reliable raw data, it is necessary that
the operation's top management be committed and involved
to facilitate the necessary coordination among the geology,
11.6
Reconciliations
Reconciliation of production information with the predictive
models used is critical to evaluating their effectiveness and
may allow for optimization of the resource modeling process
(Rossi and Camacho
1999
; Schofield
2001
; Parker
2012
).
Whether mining open pit or underground, mine-to-mill rec-
onciliations can be one of management's better tools to per-
form proper accounting and evaluate models.
Any reconciliation program should be based on a clear set
of criteria and objectives. It should also be executed through
a stepwise, logical approach. There are a number of assump-
tions and key requisites for reconciliations to be effective,
and they are not without pitfalls. There are benefits and costs
associated with the up-keeping of the information and safe-
guards should be used to avoid collecting and using mislead-
ing information.
Reconciliation procedures must be simple, robust, and
specifically adapted to the operation. The reconciliation
data should be reliable, and the procedures should include if
possible the full production stream (model, mine, process-
ing facilities, and final product comparisons); therefore, the
process may involve several predictive models (long-term
and short-term block models), different open pit and under-
ground mines, stockpiling, and multiple processing streams.
11.6.1
Reconciling against Past Production
Production reconciliation can be considered an optimiza-
tion tool (Rossi and Camacho
1999
; Schofield
2001
). This
