Lecture- 3
APPLICATION OF GEOSTATISTICS IN THE MINING INDUSTRY
1. Estimating the total reserves
The first important step in a feasibility study is to determine the in-situ reserves.
Geostatistics can help the mine planner get accurate estimates of the total tonnage
insitu and the average quality from the available information, and thus help him
decide whether further investment in the project is warranted.
2. Error estimates
No estimation method can ever give exactly the right value since there is
inevitably some error involved. So it is important to know how serious it is.
Decision makers need to know whether the estimated grade is accurate to - 0.1 %
or to 1 %. As well as giving the estimated values, geostatistics also provides a
measure of the accuracy of the estimate in the form of the kriging variance. This is
one of the main advantages of geostatistics over traditional methods of assessing
reserves.
3. Optimal drill-hole (or sample) spacing
The estimation variance (calculated by geostatistics) depends on the variogram
model chosen for the deposit and on the location of the drill-holes but not on the
numerical values of the samples. So, once the variogram has been selected for a
particular deposit, or a particular region, the estimation variance can be found. This
makes it possible to evaluate the estimation variance for a wide variety of possible
drill-hole patterns without actually doing the drilling, and hence to find the sampling
grid that just gives the required accuracy. This can lead to considerable saving on
drilling or sampling budgets.
4. Estimating block reserves
Once a decision has been made to mine a deposit, estimates of the tonnage and
the average grade are needed block by block. Here, a block might represent the
production for a shift, or for a month. In addition to estimating the ore tonnage and
the average grade of mining blocks, geostatistics can provide estimates of quality
variables. For coal, these include ash content, sulphur content and calorific value.
For bauxite, they are the percentage of silica, and of iron oxides, etc…
5. Gridding and contour mapping
Although, most mining companies usually want block estimates of their variables
rather contour maps, geostatistics can be used to estimate the values at the nodes of
a regular grid. After this, a standard contouring package can be used to do the
plotting. This has the advantage of being more accurate than other methods of
evaluating grid node values. Over the past five years, the petroleum industry has
been turning more and more to kriging for this.
6. Simulating a deposit to evaluate a proposed mine plan
Since kriging has been designed to give the minimum variance linear estimates,
the kriged values are “smoother” than any other unbiased linear estimators and also
than the real but unknown values. This mean that if a numerical model of a deposit
is being set up to test various proposed mine plans, the kriged values should not be
fed into this, because they would seriously under-estimate the inherent variability.
In this case, a conditional simulation of the deposit should be used.
7. Estimating the recovery
In many mining operations engineers have to predict the recovery and the
recovered grades when blocks of a specified size are selected for treatment (or
mining) if their average grade is above an economic cutoff. When the sample grid
is about the same size as the selection blocks, their grades can be estimated
individually to a reasonable accuracy. But if the blocks are much smaller than the
grid size as is usually the case at the feasibility stage, it can be misleading trying to
get estimates of each individual block. These are simply not accurate enough. The
best that can be done is to predict the proportion of selection units that will be
recovered, and their average grade. This leads in to nonlinear geostatistics.
