Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259952477
CITATIONS READS
0 1,435
1 author:
Jacek Mucha
AGH University of Science and Technology in Krakw
88 PUBLICATIONS 123 CITATIONS
SEE PROFILE
All content following this page was uploaded by Jacek Mucha on 30 January 2014.
Extended abstract
Methods applied to estimate the resources of bituminous coals and lignites do not differ
essentially from those used for other mineral raw-materials. However, geology of coal
deposits is commonly (although improperly) regarded as "simple" in comparison with that of
ore bodies due to specific geometry of coal seams. During the early exploration stage of coal
deposits (usually running drilling projects) both the traditional estimation methods of
resources and the classic statistical procedures are applied. On the contrary, the geostatistical
Matherons methods are less common as their usage requires higher number of data (at least
50 units) obtained from thickness measurements or sampling of coal seams at preselected
observation sites.
Among the traditional methods of resources estimation the Boldyrev polygons are
most commonly used, followed by geological cross-section interpolation and isoline
interpolation methods. Their general advantage is relatively high clarity although these
methods do not always guarantee an expected credibility of generated models of minerals
deposits. However, the important constrain of these methods is that we cannot prognose the
accuracy of resources estimations, i.e., the credibility of such estimations cannot be measured
using theoretical evaluation of estimation error. Such drawback of traditional methods can
be avoided by application of classic statistics or modern, more advanced geostatistical
procedures. More detailed exploration of mineral deposits (mostly bituminous coals) with the
mine workings and related acquisition of huge amounts of data makes the usage of
traditional methods rather difficult because construction of geological cross-sections from
irregularly and preferentially distributed observation sites becomes highly laborious. On the
contrary, the credibility of statistical and, particularly, geostatistical methods increases due to
the appearance of stronger autocorrelations of deposit parameters when sampling grids
become more dense.
Among the classic statistical methods two ostensibly similar ones are commonly used:
1
arithmetic mean and mean accumulation. Their popularity results from simple calculation
algorithms. However, both methods differ not only in the algorithms but also in the
applicability, which, unfortunately, is rather rarely verified in practice. Both methods are
based on the concept of independent measurements (observations), which means purely
random variability of deposit parameters (i.e., the lack of autocorrelation of particular
parameters). If such premise is valid the estimated resources or reserves (Q) for particular coal
seam are given by the following formulae:
arithmetic mean method:
where: arithmetic mean of thickness measured at observation sites, arithmetic mean of density
of a raw-material (coal and barren intercalations), F area of a seam or its part (parameter usually regarded as
constant) [m2], arithmetic mean of unit accumulation determined for n observation sites.
The arithmetic mean method provides correct results only when additional premise is valid:
that thickness and density of raw material are independent variables.
For given area of estimation (F) the relative standard errors of coal resources estimation in
seams are calculated from the following formulae:
arithmetic mean method:
where: , relative standard errors of estimations of mean thickness and mean density of a raw-
material [%], vq coefficient of variation of unit accumulation of a raw-material [%], n number of observations
(observation, measurement or sampling sites, or size of a dataset)
Low variability of density (variability coefficient usually below 10%) causes that the
accuracy of resources estimation is only slightly higher than the accuracy of mean thickness
estimation of coal seams. Hence, variability of density can be reasonably ignored and treated
as constant value equal to the arithmetic mean calculated from several to a dozen of
measurements made in the deposit. Current experience indicates that correlation between coal
thickness and density is very rare and, in practice, it does not affect the accuracy of resources
estimations. Therefore, both the arithmetic mean and the mean accumulation methods applied
2
to coal deposits usually provide similar results. Both the lack of correlation between deposit
parameters and the low variability of density of a raw material enable us to regard both
methods as equivalent.
The selection of statistical method of resources estimations is strongly affected by
localization of observation sites, which is described by the terms isotopy heterotopy.
Under isotopic conditions, which means identical localization and equal number of
measurements both estimation methods can be applied: arithmetic mean and mean
accumulation . Under complete or partial heterotopy, which means different localization or
partial overlapping of measurement grids of deposit parameters only arithmetic mean method
can be used.
The statistical methods enable us to solve other problems closely related to resources
estimations, although under some specific conditions:
1) interval estimation of unknown, true resources of coal (determination of confidence
interval, which contains true value of resources at the assumed error risk):
where: QD, QG lower and upper limits of confidence interval for true value of coal resources, respectively,
P probability that confidence interval contains true value of resources (usually, we assume P=0.95),
- risk that confidence interval does not contain the true value of resources (usually, we assume =0.05).
2) determination of guaranteed resources (Qg) i.e., a part of estimated resources (Q) for
which probability that true resources value (Q0) will fall below the guaranteed resources
(Qg) is small, for example, =10%=0.1 or =5%=0.05, which can be defined as:
P(Q0<Qg)=
3
estimation accuracy of individual seams weighted by a square of estimated resources value
and is calculated from formula:
where: qi coal accumulation at i observation (sampling) site, wKi - kriging weights determined from the kriging
equation system, which minimize estimation error of mean value of a parameter, N number of samples
4
involved in the kriging procedure.
It must be emphasized that the kriging error is a credible measure of error of coal
accumulation and coal resources estimations only for the whole seam or its significant parts.
However, it is not a good measure for local estimations referred to small parts of a seam
(calculation blocks) because in such cases it is based on a small number of observations. It
results from the fact that kriging error does not comprise the diversification of measured
accumulation values. For such cases several formulae were proposed in order to correct this
shortage of kriging method (Yamamoto 2000).
The geostatistical evaluations of estimation errors as well as errors determined with the classic
statistical methods are applied to interval evaluation of resources, guaranteed resources and to
categorization of resources assessment. Presented formulae, both statistical and geostatistical,
do not include the errors of area determination. Hence, the true errors of resources estimations
using the polygon kriging on calculation blocks of irregular geometry can be somewhat higher
in some particular cases.
Taking into account the categories of resources assessment after JORC Code (2012) for the
fragments of a deposit planned for extraction in the 5 years period, Bertoli et al. (2013)
proposed the following permissible relative errors of resources or reserves estimation
determined with the ordinary kriging or geostatistical simulation:
inferred from 20% to 50%.
indicated from 10% to 20%
measured up to 10%,
5
It must be noticed that the estimation accuracy of coal resources can be significantly
affected by errors resulting from measurements of seam thickness and determination of
density of a raw material. It is especially important for deposits explored with drillings when
core recovery is unsatisfactory (i.e., less than 95%, after JORC Code). Such problem was
encountered during the drilling of deposits in previous decades due to poor drilling
technologies. Insufficient core recovery caused e.g., by washing out of coal fragments by
circulating mud leads not only to the random errors but also to various systematic errors in
estimation of deposit parameters. If the correlation of coal seams in faulted multi-seam
deposits is proper the main problem appears to be the accuracy of coal resources estimation,
which strongly depends not only on the number of observation sites but also on the observed
variability of deposit parameters involved in estimation procedures. If systematic errors of
their measurements are eliminated, i.e., if only the random errors occur, the observed
statistical variance of given parameter is a simple sum of variances: that resulting from natural
(genetic) variability of given parameter (relevant for applied sampling system, particularly, on
size, geometry and orientation of samples) and that of random errors. In geostatistical
description similar division can be made for random component of observed variability.
If the share of random error variance in the observed variance of given parameter is
high the improvement is recommended of the quality of deposit parameter measurements,
which should result in distinct increase of estimation accuracy of both the resources and the
quality parameters of coals.
Literature
Bertoli O., Paul A., Zach C., Dunn D. (2013): Geostatistical drillhole spacing analysis for
coal resource classification in the Bowen Basin, Queensland. International Journal of
Coal Geology 112, 107-113.
The JORC Code (2012 edition) Australasian Code for Reporting of Exploration Re-
sults, Mineral Resources and Ore Reserves prepared by The Australasian Institute of
Mining and Metallurgy (AusIMM), Australian Institute of Geoscientists and Minerals
Council of Australia.
Yamamoto J.K. (2000): An Alternative Measure of the Reliability of Ordinary Kriging
Estimates. Mathematical Geology, Vol. 32, No. 4, 489-509