See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/362838118

Implication of Drilling and Sampling Patterns to the Spatial Heterogeneity of

Coal Deposit Based on Geostatistical Approach

Conference Paper · September 2011

CITATIONS READS

0 58

4 authors, including:

Mohamad Nur Heriawan Syafrizal Syafrizal

Bandung Institute of Technology Bandung Institute of Technology
100 PUBLICATIONS 361 CITATIONS 92 PUBLICATIONS 91 CITATIONS

SEE PROFILE SEE PROFILE

Lilik Eko Widodo

Bandung Institute of Technology
73 PUBLICATIONS 90 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mohamad Nur Heriawan on 16 September 2022.

The user has requested enhancement of the downloaded file.

 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

IMPLICATION OF DRILLING AND SAMPLING PATTERNS TO THE SPATIAL

HETEROGENEITY OF COAL DEPOSIT BASED ON GEOSTATISTICAL APPROACH

Mohamad Nur Heriawan1, Syafrizal1, Aditya Mahendra2, and Lilik Eko Widodo1
1
Research Group of Earth Resources Exploration, Faculty of Mining and Petroleum Engineering,
Institut Teknologi Bandung (ITB)
2
Study Program of Mining Engineering, Faculty of Mining and Petroleum Engineering,
Institut Teknologi Bandung (ITB)
Email: [email protected]

ABSTRACT

Drilling pattern and spacing can be correlated to the spatial heterogeneity of coal properties. The optimum
drilling pattern and spacing is directly associated to the number of drillhole and budget required by
mining company. Besides drilling, channel sampling can be conducted much more rapidly based on the
ply by ply sampling, moreover it requires much cheaper budget than drilling. Although the geometrical
support is totally different, drilling sampling is better in term of less subjected by dillution. This study
evaluates the implication of drilling and sampling patterns to the spatial heterogeneity of coal deposit
based on geostatistical approach. The spatial heterogeneity was represented by variogram parameters
principally the nugget ratio and range as well as the estimation error of coal properties by kriging method.
The case study was conducted for a coal seam of Tanjung Formation at South Kalimantan. The existing
drilling and sampling patterns was approximately triangular with much more number of data, and then
they were compared to the square pattern by reducing some data. Obviously the less number of data
would generate the increasing of error estimate. When the increasing of error estimate was not major
anymore, subsequently the drilling and sampling pattern was considered to be optimum.

Keywords: drilling pattern, channel sampling, spatial heterogeneity, geostatistics

INTRODUCTION resources classification. Geostatistical is one

method that has ability to quantify the drillholes
Drilling in coal exploration is essential to spacing based on error variances of regionalized
determine the stratigraphic, coal geometric, variables, i.e. coal thickness, ash content, total
geological structure, wallrock characterization sulphur, inherent moisture, total moisture,
and sampling for coal qualities. The result on volatile matter, and calorific value.
drilling program is used to estimate the coal
resources and reserves. Drillhole spacing is These studies compare the drillholes spacing
usually defined after interpreting the coal and pattern with two different patterns, square
continuity from geological mapping. Statistical and triangular, and then define the most
evaluation on drillhole data can be used to optimum one. Both coal thickness and qualities
define the area of influence and spatial are considered in defining the error variance of
variability of coal thickness and qualities. each pattern using geostatistical approach. A
seam in coal deposit of Tanjung Formation at
Sample derived from the drilling is better and South Kalimantan is used for a case study.
more representative compared to the one from
channel sampling. The problem arised STUDY AREA
frequently in coal exploration is how to define
quantitatively the optimum drillhole spacing Warukin Formation (Tmw) with age of Middle
and pattern. These matters are related to the Meocene is the coal bearing formation in the
proposed budget and confidence level for study site. This formation was deposited in sub-
1
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

basins Asam-Asam and Barito. The coal deposit where the samples is separated to anticipate
at study site is located at Satui area where it was dillution. Statistical analysis was performed for
deposited at Tanjung Formation with age of each part of drilling and channel of samples (top,
Eocene-Oligocene (FIGURE 1). The strike of middle, and bottom) as seen on TABLE 1. The
coal seam at the west part is N80ºE in average most significant coal qualities considered in this
and at east part is N45ºE in average. The length research is ash content and calorific value. The
of coal seam distribution is about 14.5 km with comparison on the statististical distribution
2 (two) main seam are Seam SM (middle) and between drilling and channeling samples was
Seam SL (lower) with average thickness is 2.7 conducted only for the clean coal (TABLE 2).
m and 5.4 m respectively.
The location for channel sampling was clustered
The coal seam at Tanjung Formation is high in the north part close to the locations of
caloric (bituminous) with bright luster, while at subcrops, while the location of drillhole is more
Warukin Formation the coal seam is relatively widespread regularly. The difference on these
lower caloric (sub-bituminous) with dull luster data supports produces the different data quality.
and some wood structures. The oldest The ash quality of coal showed relatively higher
sedimentary rock in the study site was deposited value in channel samples rather than in drillhole
in Tanjung Formation with unconformity border due to additional of some impurities material
to the Pre-Tertiary basement. Tanjung during the sampling.
Formation consists of fine to coarse grained
quartz sandstones with thickness 0.5-1.5 m. The The total amount of drillhole data is 28 for
rock structure is fined sedimentary and cross- Seam SL2 which is the main coal seam in the
bedding, intercalated by gray shally claystone in study area. The variables considered in the
the top of formation wih thickness 0.3-1.5 m. analysis are seam thickness, inherent moisture,
Intercalation of coal is found in the bottom of ash content, total sulfur, and calorific value.
formation with black color, bright, brittle, and Statistical analysis showed that the data
thickness 0.5-1.5 m. Lenses limestone is also distribution for all variables is quite normal with
found with brownish gray color associated with coefficient of variations lower than 0.5 (Koch &
pieces of mollusca, echinoid, and forraminifera Link, 1970).
i.e. Nummulites Javanus and Heterostegina sp.,
also small foraminifera from Milholidae family ANALYSIS ON SPATIAL STRUCTURE
with age of Eocene which deposited in parallic-
neritic environment. The thickness of Tanjung Spatial structure of regionalized variable is
Formation is 750 m. performed by variogram model. Experimental
variogram is indispensable to characterize the
STATISTICAL ANALYSIS spatial continuity and variability before doing
estimation. The parameter resulted from
Statistical analysis is indispensable to quantity variogram modelling are range (area of
the trend on data distribution. The result on influence, a), nugget or measurement variance
statistical analysis could explain the data (C0), and dispersion variance (C). The most
correlation and trend of distribution so that we frequently used variogram model is Spherical
could define the most appropriate estimation (Matheron) model (David, 1977, Barnes, 1979)
method. The analyzed data was derived from where the mathematical function is simply
sampling on drilling or channeling. polynomial and the variogram will reach a finite
value for infinite h.
Coal samples from drilling have regular
arrangement for each drillhole i.e. 0.1 m on top The mathematical function of Spherical
part, some meters on middle part, and 0.1 m on variogram model is:
bottom part of coal seams. Those top and for h ≤ a
bottom parts are dirty coal and the middle part
for h > a
is clean coal. This technic is named ply by ply
2
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

for h = 0 according to the geometric and mineralization

represented in variogram.
This variogram model shows a linear behavior
near origin that indicates moderate continuity of A linear estimator is written as:
n
Z (V )* =  i  z (xi )
regionalized variable. The Spherical model is
mostly appropriate applied to the ore grade data
i =1
because ussually give the smallest error in
The OK system in block estimation is written
estimation rather than other models.
as:

   (S , S ) +  =  (S , V )
n
Construction of experimental variogram and
j i j i
fitting model was performed for the drillhole j=1
and channel data of seam thickness, inherent n
moisture, ash content, total sulfur, and calorific
value. Omnidirectional variogram was chosen

i =1
i =1
for all variables as the indication of isotropy The OK variance in block estimation is
was unclear. The lag distance was set 200 m expressed as:

 K2 = − (V ,V ) +  +   j  (S j ,V )
and 150 m respectively for drillhole and channel n

data. The parameters derived from variogram

j =1
model of variable from drillhole and channel is
summarized on TABLE 3. The estimation parameters were used in this
study are:
Exploration drilling is the best way to get the • Grid size is 100  100 m
representative samples, but the cost is mostly • Fault zone is not considered as estimation
high. That is why the optimization on drillhole boundary.
spacing is crucial. The presence of fault • The minimum and maximum number of data
structure in the middle part of study area did not used in estimation is 1 and 5 respectively.
affect significantly the qualities of coal deposit. • Searching radius is 500 m for drillhole data
Different drillhole spacing and pattern and 300 m for channel data.
(triangular and square) by data reducing is
compared for their error variances. The RESULTS
sampling distribution and pattern for the
existing and reduced data is depicted in The calculation on error estimate is based on
FIGURES 2 and 3 respectively. estimation values and kriging standard deviation
with confidence level 95% on normal
GEOSTATISTICAL ESTIMATION distribution of error. Then the relative error is
written as:
The most frequently used geostatistical method 1.96   K
for estimation is kriging. Ordinary Kriging Error relative = 
Z (V ) *
(OK) is an estimation method for regionalized
TABLE 4 summarize the values of relative
variables on point, area, or volume using criteria
errors for existing pattern of channel data,
to minimize the estimation variance (Hohn,
drillhole (triangular), and reduced pattern of
1998). The correlation between grade in
drillhole (square) for each variables of Seam
drillhole data and grade in block showed a
SL2.
systematic scattering. Consequently one
drillhole data is not the best value to estimate
The different samples support between drillhole
the block, and then a correction is necessary.
and channel data showed a large different error
Matheron (1963) proposed a correction by
on estimation up to 39%. The large error
weighting the samples value using variogram
estimate on channel data is due to the low
function. Tke kriging method is known as Best
continuity of coal thickness and qualities near
Linear Unbiased Estimator (BLUE), where
the sub-crops. Moreover the channel data is
simply kriging produces the samples weight
3
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

more sensitive to produce dilution rather than the existing drillhole spacing of 400-500 m is
drillhole data. quite wider.
4. The triangular pattern of drillhole is more
The reduced pattern of drillhole that shows the effective up to 30% to reduce the estimation
square pattern produced higher error estimate error on coal qualities rather than square
on coal qualities rather than triangular pattern of pattern, but the reducing of estimation error
existing drillhole up to 30%. The number of of coal thickness is not more than 3%.
data is reduced from 28 in square pattern to be
22 in triangular pattern. The increasing of error ACKNOWLEDGEMENTS
for ash content and inherent moisture was
extremely significant rather than for seam We would thank sincerely to the Department of
thickness and calorific value when the pattern of Mineral Resources, PT. Arutmin Indonesia in
drillhole was changed from triangular to be supporting to utilize the data set and publish this
square. This result was caused by the high paper. We also appreciate the support of
variation on ash content and inherent moisture Research Grant of ITB 2010 for the Research
data rather than seam thickness and calorific Group of Earth Resources Exploration, Faculty
value. Total sulphur showed the reverse of Mining and Petroleum Engineering, Bandung
condition where the relative error decreased Institute of Technology.
after changing the pattern of drillhole from
triangular to be square. This due to the reduced REFERENCES
data was outliers by chance.
1. Barnes, M.., Computer-Assisted Mineral
CONCLUSION Appraisal and Feasibility, SME-AIME,
New York, 167 pp., 1980.
Based on the study on comparing sampling 2. David, M.., Geostatistical Ore Reserve
pattern of coal deposit, we concluded that: Estimation, Elsevier Scientific Publishing
1. The different sampling support between Company Inc., New York, 1977.
drillhole and channel gave extremely 3. Hohn, M. E., Geostatistics and Petroleum
different estimation error up to 39%, where Geology, Kluwer Academic Publishers,
the drillhole data produced much lower error. Netherlands, 1999.
2. The presence of geological structure i.e. fault 4. Koch, G.S.Jr and Link, R.F., Statistical
in the study site did not affect the coal Analysis of Geological Data, Wiley, New
qualities, so all data is combined to be one York, 1970.
population. 5. Matheron, G., Principles of geostatistics,
3. Based on the area of influence from Economic Geology, 58, pp. 1246-1266,
variogram model 150-250 m indicated that 1963.

4
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

TABLE 1: Summary on statistical analysis for ash content and calorific value (CV) in drillhole and
channel data for each part of seam

Drillhole Data
Parameters Top Middle Bottom
Ash CV Ash CV Ash CV
(% adb) (kkal/kg) (% adb) (kkal/kg) (% adb) (kkal/kg)
Number of data 27 27 28 28 28 28
Minimum Value 2.90 5,693 4.70 6,845 3.20 4,591
Maximum Value 25.90 7,498 10.20 7,455 35.00 7,436
Mean 12.33 6,851 6.36 7,245 8.84 6,911
Standard Deviation 5.84 470 1.14 138 7.78 659
Median 11.50 6,903 6.15 7,263 6.35 7,139
Coefficient of Variation 0.47 0.07 0.18 0.02 0.88 0.10
Channel Data
Top Middle Bottom
Parameters Ash CV Ash CV Ash CV
(% adb) (kkal/kg) (% adb) (kkal/kg) (% adb) (kkal/kg)
Number of data 39 38 39 39 38 38
Minimum Value 2.22 3,325 2.02 6,805 2.52 5,044
Maximum Value 51.77 7,677 12.83 7,695 31.92 7,412
Mean 8.85 6,946 6.72 7,201 12.04 6,711
Standard Deviation 9.17 882 1.85 144 6.94 538
Median 6.02 7,263 6.71 7,196 11.17 6,869
Coefficient of Variation 1.04 0.13 0.27 0.02 0.58 0.08

TABLE 2: Summary on statistical analysis for seam thickness and qualities in clean coal (middle part) in
drillhole and channel data

Drillhole Data
Thickness IM Ash TS CV
Parameters
(m) (% adb) (% adb) (% adb) (kkal/kg)
Number of data 28 28 28 28 28
Minimum Value 0.61 2.70 4.70 0.44 6,845
Maximum Value 1.29 5.10 10.20 0.71 7,455
Mean 0.96 4.20 6.36 0.50 7,245
Standard Deviation 0.15 0.55 1.14 0.07 138
Median 1.01 4.40 6.15 0.49 7,263
Coefficient of Variation 0.16 0.13 0.18 0.14 0.02
Channel Data
Thickness IM Ash TS CV
Parameters
(m) (% adb) (% adb) (% adb) (kkal/kg)
Number of data 39 39 39 39 39

5
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

Minimum Value 0.36 2.99 2.02 0.40 6,805

Maximum Value 1.20 5.95 12.83 1.37 7,695
Mean 0.79 4.20 6.72 0.51 7,201
Standard Deviation 0.20 0.73 1.85 0.15 144
Median 0.78 4.21 6.71 0.49 7,196
Coefficient of Variation 0.25 0.17 0.27 0.29 0.02

TABLE 3: Variogram parameters of coal variables from drillhole and channel data

Drillhole Data
Ash CV IM Thickness TS
Parameters
(% adb) (kkal/kg) (% adb) (m) (% adb)
Nugget Variance (C0) 0.8 12,700 0.05 0.008 0.001
Sill (C) 0.43 6,300 0.247 0.019 0.005
Range (a) 650 600 1,000 570 540
Channel Data
Ash CV IM Thickness TS
Parameters
(% adb) (kkal/kg) (% adb) (m) (% adb)
Nugget Variance (C0) 3 17,500 0.45 0.01 0.00
Sill (C) 0.38 3,200 0.08 0.018 0.02
Range (a) 300 400 380 300 1,000

TABLE 4: The average estimation errors for different sampling patterns.

Channel Data
Ash CV IM Tebal TS

Estimation
28.66 2.11 18.60 26.91 23.94
error (%)

Existing Drillhole Data (Triangular)

Ash CV IM Tebal TS

Estimation
17.48 1.94 11.68 19.64 17.99
error (%)

Reduced Drillhole Data (Square)

Ash CV IM Tebal TS
Estimation
25.28 2.47 15.97 20.21 15.41
error (%)

6
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

Tet

Study Area

FIGURE 1: Geology of study area at Tanjung Formation, South Kalimantan. Note: Mgr =
Granite, Tet = Tanjung Formation, Tomb = Berai Formation, Tmw = Warukin Formation

FIGURE 2: Distribution of the existing drillhole and channel data shows relatively triangular pattern.

7
 PROCEEDINGS JCM MAKASSAR 2011
The 36th HAGI and 40th IAGI Annual Convention and Exhibition
Makassar, 26 – 29 September 2011

FIGURE 3: Distribution of the reduced drillhole data shows relatively square pattern

View publication stats