Evaluation on Coal Resource Estimation and Drillhole Spacing Optimization Based on Geostatistical Approach, with a Case Study on Coal Deposit at South Kalimantan, Indonesia. Mohamad Nur HERIAWAN, Rudy Hendrawan NOOR, SYAFRIZAL, and Sudarto NOTOSISWOYO
Evaluation on coal resource estimation and drillhole spacing optimization
1. International Symposium on Earth Science and Technology 2011
Evaluation on Coal Resource Estimation and Drillhole Spacing Optimization
Based on Geostatistical Approach, with a Case Study on Coal Deposit at South
Kalimantan, Indonesia
Mohamad Nur HERIAWAN1
, Rudy Hendrawan NOOR2
, SYAFRIZAL1
, and Sudarto NOTOSISWOYO1
1
Research Group of Earth Resources Exploration, Faculty of Mining and Petroleum Engineering, Bandung Institute
of Technology (ITB), Bandung 40132, Indonesia
2
Magister Program of Mining Engineering, Faculty of Mining and Petroleum Engineering, Bandung Institute of
Technology (ITB), Bandung 40132, Indonesia
Contact e-mail: heriawan@mining.itb.ac.id
Abstract
The evaluation and classification on coal resource and reserve in Indonesia has been guided by SNI 1999 which is
an Indonesian National Standard on coal resource and reserve estimation. However, this standard does not consider
and discuss about any spatial variability on the variable being studied. The coal resource is classified based on the
sample spacing and complexity of geological structures, but not taking into account the coal quantity and quality
parameters. The complexity on geological structures does not always correspond to the complexity on coal seam
geometric and qualities. Estimation variance in geostatistics i.e. generated by Ordinary Kriging method can be used
to assess the spatial variability on coal geometric and qualities based on variogram modelling and estimation
parameters. Therefore geostatistics can be used to evaluate the resource estimation and classification which was
estimated and classified as well using guidance of SNI 1999. Moreover, estimation variance can also be used to
optimize the drillhole spacing on coal deposit by arranging some spacing and configuration then checking the
change on spatial variability of coal variables. The coal resource evaluation takes a case study on two coal seams of
Tanjung Formation where the geological structure is considered to be moderate and of Warukin Formation with
relatively simple geological structure, both are located at South Kalimantan. The drillhole spacing optimization only
takes a coal seam of Warukin Formation. The result showed that estimation variance by Ordinary Kriging was
effective for coal resource evaluation and drillhole spacing optimization.
Key words: coal resources, drillhole spacing, estimation variance
INTRODUCTION
The subject of research is initiated by the procedure on
coal resource estimation and classification in Indonesia
which is based on SNI 1999, an Indonesian National
Standard issued by National Standardization Board in
1999 where coal is estimated and classified based on
the geometry of coal and complexity of geological
structure in the area of interest. The procedure does not
involve any considerations on the spatial variability
either on coal geometric or qualities. The complexity
on geological structure is solely defined based on the
geological knowledge and does not straightly related to
the complexity of the geometric and qualities of coal
deposit. Another problem arise correspond to
determining on the drillhole spacing in coal
exploration as the SNI 1999 requires specific drillhole
spacing to classify the coal resources where it is
related directly to the complexity of geological
structure. For more complex geological structure then
SNI 1999 requires closer drillhole spacing, i.e. up to
100 m for measured resource in highly complex
structure. Accordingly up to now SNI 1999 procedure
only concerns to the continuity of seam structure,
while spatial variability on seam thickness and
qualities have been relatively ignored.
Geostatistical method has capability to measure the
spatial variability of coal geometric and qualities
simultaneously. Estimation variance through kriging
estimation can be used to classify the coal resources
according to the relative error on coal geometric and
qualities estimation by considering a certain
confidence level i.e. 95%. The drilling spacing is also
can be optimized by establishing the estimation
variance for each spacing all together for coal
geometric and qualities although they will produce the
different relative error.
This research focuses on the application of
geostatistical estimation variance for estimation and
classification of coal resources as well as for drillhole
spacing optimization. A seam of coal deposit from the
two different formations will be used for case study i.e.
Tanjung Formation and Warukin Formation located at
South Kalimantan, Indonesia. The coal resource
estimation and classification will be compared to the
SNI 1999 procedure which considering two different
condition of geological structure, i.e. simple condition
for Warukin Formation and moderate condition for
Tanjung Formation. The similar study in complex
geological structure of coal deposit was addressed in
Heriawan et al. (2010).
2. [Paper Number]
GEOLOGICAL CONDITION
The simple condition of geological structure of
Warukin Formation is located at Mulia area, while
moderate geological structure of Tanjung Formation is
located at West Senakin area. The coal deposit of
Warukin Formation has age of Middle Miocene with
strike direction of South-West – North-East and dip of
25º-45º relatively to South-East. The coal at Mulia
area has relatively lower caloric (sub-bituminous) with
dull brightness and some woody structures. The coal
deposit of Tanjung Formation has age of Eocene with
strike direction of North – South and dip of 5º-14º
relatively to East. The coal deposit at West Senakin
has relatively high caloric (bituminous) with good
brightness. The location and regional geology of both
study area is depicted in Figure 1.
Warukin
Formation
Tanjung
Formation
Figure 1. Location and regional geology of study area
at South Kalimantan, Indonesia (modified from
Sikumbang & Heryanto, 1994; Heryanto et al., 1994).
DATASET AND STATISTICAL DESCRIPTION
A coal seam used for the case study at Mulia and West
Senakin area is Seam EU2 and SM2 respectively,
where they are the main seam in the existing mine. The
statistical distribution of seam thickness and qualities
showed relatively normal distribution as the coefficient
of variation in general was less than 0.5 (Koch and
Link, 1970) for both seams. Due to the different spatial
characteristics on thickness of Seam EU2 at Mulia
area, then the thickness data is separated into three
clusters (Figure 2). The clusters 1, 2, and 3 have
average thickness of 3.62 m, 0.48 m, and 2.53 m
respectively. The histogram of seam thickness of each
cluster and coal qualities after data verification is
shown in Figure 3. The distribution of coal relative
density is a bit positively skewed (Figure 3h). Actually
the information on coal qualities of Seam EU2 is
derived from about 30% of the total number of
drillholes. The geological condition in Mulia area is
considered as simple because there is no presence of
geological structure i.e faults. This fact might indicate
that the simple geological condition does not always
coincide to the simple coal geometry and qualities.
Accordingly the spatial variability on both parameters
must be considered in resource estimation.
Figure 2. The location of 68 drillholes and clustering
on seam thickness data for Seam EU2 of Warukin
Formation.
(a) (b)
(c) (d)
(e) (f)
(g) (h)
Figure 3. Statistical distribution of thickness and
qualities of Seam EU2 of Warukin Formation.
The presence of some fault structures in West Senakin
area is depicted in Figure 4, and then this considered
being moderate geological condition. Due to this then
the estimation domain is separated into five domains
(local blocks) where each of them are bordered by
faults lineament. This assumes that the fault structures
in this area influenced the spatial variability of coal
geometric and quality parameters. The histogram of
seam thickness and qualities of Seam SM2 at West
Senakin area is shown in Figure 5. The distribution of
3. [Paper Number]
coal total moisture shows a bit bimodal with cut-off
about 5.50% (Figure 5b). The information on coal
qualities of Seam SM2 is derived from less than 20%
of the total number of drillholes.
Fault
structure
Figure 4. The location of 111 drillholes and block
clustering due to the presence of fault structures for
Seam SM2 of Tanjung Formation.
(a) (b)
(c) (d)
(e) (f)
Figure 5. Statistical distribution of thickness and
qualities of Seam SM2 of Tanjung Formation.
GEOSTATISTICAL ANALYSIS AND
ESTIMATION
The experimental variogram construction and fitting
model for seam thickness and qualities is performed
prior to estimation. Because of the relatively small
number of drillhole data in each estimation domain,
then variogram is constructed using all data. The
Spherical model is the best fit for all experimental
variogram. The result on variogram fitting for seam
thickness and qualities of Seam EU2 and SM2 is
respectively summarized in Tables 1 and 2.
Table 1. Summary on the variogram parameters of
thickness and qualities of Seam EU2.
Parameters
Thickness
(m)
Total Moisture
(%, adb)
Ash Content
(%, adb)
Total Sulphur
(%, adb)
Caloric (kcal/kg,
adb)
Sill (C) 1.05 2.50 0.96 0.40 48,500
Nugget variance (C0) 0.41 0.09 0.00 0.22 11,000
Range (a), meter 750 600 550 600 850
Nugget ratio (%) 28.08 3.47 0.00 35.48 18.49
Note: The fitting is using Spherical model.
Table 2. Summary on the variogram parameters of
thickness and qualities of Seam SM2.
Parameters
Thickness
(m)
Total Moisture
(%, adb)
Ash Content
(%, adb)
Total Sulphur
(%, adb)
Caloric (kcal/kg,
adb)
Sill (C) 0.03 0.96 3.80 0.0002 67,000
Nugget variance (C0) 0.11 0.13 0.70 0.0009 0.00
Range (a), meter 650 800 650 600 700
Nugget ratio (%) 79.71 11.93 15.56 81.65 0.00
Note: The fitting is using Spherical model.
Based on the variogram parameters, the thickness and
qualities of Seam EU2 at Mulia area has relatively low
spatial variability because their nugget ratio is less
than 50%. While the nugget ratio of thickness and
qualities of Seam SM2 chiefly for total sulphur
indicates very high spatial variability because it is
larger than 75%. Therefore, definition on simple
geological condition of coal deposit at Mulia area
coincides to their coal geometry and qualities, while
for the geological condition of West Senakin area
which is defined to be moderate should be considered
as complex condition related to their coal geometry
and sulphur content.
In term of geostatistical estimation, due to its
capability to perform best unbiased linear estimation,
Ordinary Kriging (OK) method is used to estimate the
seam thickness and qualities for each estimation
domains as mentioned above. Kriging estimation is
always associated with the estimation variance
(kriging variance) where it measures the accuracy
(magnitude of error) of estimation in each estimated
blocks. By assuming that the distribution of error
variance is normal with 95% of confidence level then
the relative error is calculated using:
% error = ± 1.96 σ / z* × 100%
where σ is kriging standard deviation and z* is kriged
value in each estimation block. Based on the percent
relative error as addressed in de Souza et al. (2002)
which was modified from JORC 1999 and Diehl &
David (1982), coal resources could be classified into:
1. Measured resources if the estimation blocks have
relative error ≤ 10%.
2. Indicated resources if the estimation blocks have
relative error 10-20%.
3. Inferred resources if the estimation blocks have
relative error > 20%.
4. [Paper Number]
The coal resources are classified based on the relative
errors of estimated thickness. The coal tonnage for
each resources classification could be calculated by
multiplying the estimated thickness to the estimated
relative density of coal and the block size (100 100
m). The result of coal resources estimation and
classification based on geostatistical approach
followed by the average estimated qualities for Seam
EU2 and SM2 is summarized in Tables 3 and 4
respectively.
Table 3. Coal resources estimation produced by
kriging method for Seam EU2 associated with their
average estimated qualities.
Resources Coal Tonnage Ash Content Total Sulphur Caloric Total Moisture
(Mulia) (Mt) (%, adb) (%, adb) (kcal/kg) (%, adb)
Measured 4.19 5.74 1.41 4,885 34.23
Indicated 18.17 4.75 0.38 4,885 34.23
Inferred 20.18 4.57 0.18 4,885 34.23
Total 42.54 4.76 0.39 4,885 34.23
Table 4. Coal resources estimation produced by
kriging method for Seam SM2 associated with their
average estimated qualities.
Resources Coal Tonnage Ash Content Total Sulphur Caloric Total Moisture
(West Senakin) (Mt) (%, adb) (%, adb) (kcal/kg) (%, adb)
Measured 4.32 17.96 0.38 6,299 6.11
Indicated 5.21 16.25 0.38 6,299 5.74
Inferred 11.10 15.56 0.38 6,299 5.24
Total 20.63 16.24 0.38 6,299 5.55
The coal resources above were classified based on the
estimation error of seam thickness which is the major
variable considered in resource estimation. The same
methodology could be performed to obtain the coal
tonnage and its classification based on the estimation
error of coal qualities. The coal tonnage for each
category could be different because each coal variables
have the different spatial variability which was defined
from their variogram model. The high variability of
some coal qualities can be reduced by put some infill
drillings in the area where the estimated blocks have
high estimation error i.e. > 20%.
RESOURES ESTIMATION BASED ON SNI 1999
PROCEDURE
Besides using geostatistical approach, a traditional
method based on SNI 1999 procedure was also
performed to estimate and classify the coal resources
for the same area and boundary condition. Polygonal
method is the most appropriate non block model
method used to follow the SNI 1999 procedure. Table
5 shows the spacing between information data required
by SNI 1999 procedure based on the complexity of
geological structure in study area. The spacing
information can also be interpreted as the area of
influence of drillhole to get the coal resource
estimation. The area of influence bordered by polygon
for drillholes at Mulia and West Senakin area are
illustrated in Figures 6 and 7. For area with the
presence of fault structures, then the polygon is ended
to the fault lines.
Table 5. Coal resources classification based on SNI
1999 procedure (BSN, 1999).
Measured Indicated Inferred Hypotetic
Simple x ≤ 500 500 ≤ x ≤ 1,000 500 ≤ x ≤ 1,500 Infinite
Moderate x ≤ 250 250 ≤ x ≤ 500 500 ≤ x ≤ 1,000 Infinite
Complex x ≤ 100 100 ≤ x ≤ 200 200 ≤ x ≤ 400 Infinite
Geological
Condition
Criteria
Resources
Spacing between
information data
(m)
Measured
Indicated
Inferred
Figure 6. Polygonal coal resources estimation for
Seam EU2 with simple geological condition based on
SNI 1999 procedure.
Measured
Indicated
Inferred
Figure 7. Polygonal coal resources estimation for
Seam SM2 with moderate geological condition based
on SNI 1999 procedure.
For Seam EU2 of Warukin Formation at Mulia area
where the geological condition is classified as simple
then the coal resources for measured, indicated and
inferred categories derived from the extension of seam
thickness at each drillholes to the radius of 500 m,
1,000 m, and 1,500 m respectively (Table 6). The
resource estimation was performed for three clusters
based on the different spatial variability of thickness
data. The same procedure was applied for Seam SM2
of Tanjung Formation at West Senakin area where the
5. [Paper Number]
geological condition is considered as moderate then
the extension radius for resource estimation is smaller
than the previous case, i.e. 250 m, 500 m, 1,000 m
respectively for measured, indicated, and inferred
categories (Table 7). The resource estimation was
performed for five blocks based on the presence of
some fault lineaments at this area.
Table 6. Coal resources estimation obtained by
polygonal method based on SNI 1999 classification for
Seam EU2.
Measured Indicated Inferred Total
(Mt) (Mt) (Mt)
Cluster 1 16.16 4.51 1.55 22.22
Cluster 2 0.80 0.49 0.53 1.81
Cluster 3 6.29 5.35 6.93 18.56
Total 23.24 10.35 9.00 42.59
Cluster (Mulia)
Table 7. Coal resources estimation obtained by
polygonal method based on SNI 1999 classification for
Seam SM2.
Measured Indicated Inferred Total
(Mt) (Mt) (Mt)
Block I 2.28 1.50 2.76 6.53
Block II 2.30 0.49 0.62 3.41
Block III 4.76 1.45 1.12 7.33
Block IV 0.68 0.20 0.23 1.10
Block V 0.22 0.35 0.93 1.50
Total 10.23 3.99 5.65 19.88
Block (West
Senakin)
OPTIMIZATION OF DRILLHOLES SPACING
Optimization of drillholes spacing is pursued based on
the kriging estimation variance as discussed previously
for coal resources classification. Seam EU2 at Mulia
area, chiefly at Cluster 1 is used for examining the
methodology, because the existing drillhole data is
relatively numerous and regular. Detail about the
methodology and data set was originated in Heriawan
et al. (2011). This study needs more numerous and
regular data with certain spacing, then some dummy
drillholes were set based on the geological section on
physical seam distribution to get wider seam thickness
data, while coal qualities were obtained from the
nearest neighborhood point. The different drillholes
spacing from 100 m 100 m (closest) to 500 m 500
m (widest) set at Cluster 1 of Mulia area is depicted at
Figure 8. The study examined the change on spatial
variability of seam thickness, ash content, and total
sulphur of Seam EU2 for different drillhole spacing.
Heriawan et al. (2011) showed that the seam thickness
has lower spatial variability compared to ash content
and total sulphur. Of course much closer the drillhole
spacing will give much smaller variability represented
by the smaller estimation variance (relative error).
Figure 9 shows that the decreasing on relative error
against the additional drillhole data with closer spacing
in general give lower slope on the change of spacing
from 200 m 200 m to 100 m 100 m compared to
the change of spacing 300 m 300 m to 200 m 200
m. Therefore in this case, the drillhole spacing of 200
m 200 m is considered to be optimum.
(a) (b)
(c) (d)
(e)
Note:
Variation on drillhole spacing is established within
areaof 1,300 m x 1,000 m along the strike of coal
seam N70oE:
(a) Drillhole spacing of 500 m x 500 m with 9 data,
(b) Drillhole spacing of 400 m x 400 m with 12
data, (c) Drillhole spacing of 300 m x 300 m with
20 data, (d) Drillhole spacing of 200 m x 200 m
with 42 data, (e) Drillhole spacing of 100 m x 100
m with 154 data.
Figure 8. Variation on drillhole spacing with different
hole density at Cluster 1 of Mulia area (modified from
Heriawan et al., 2011).
-50
-40
-30
-20
-10
0
0 50 100 150
Thickness
Ash Content
Total Sulphur
ErrorReduction(%)
Additional Drillholes
Figure 9. The error reduction versus additional
drillhole of Seam EU2 attributes (modified from
Heriawan et al., 2011).
DISCUSSION
This paper showed that coal resources estimation and
classification using geostatistical approach and SNI
1999 procedure has the remarkably different basic
principal. Geostatistical approach using kriging
estimation variance is based on block model estimation
and considering spatial variability of sample attributes
(seam thickness and coal qualities) represented by
variogram model. In fact SNI 1999 procedure did not
require any particular estimation methods. It only
required some specification on the spacing of
information points (drillhole) according to the
6. [Paper Number]
complexity of geological condition. If an area of
interest has more complex geological condition then
the drillhole spacing is required to be much closer up
to 100 m. The most appropriate estimation method
following the situation of such area of influence is
polygonal method. This method is used by assuming
that the seam thickness within a particular area of
influence is homogeneous (no spatial variability).
In the case of when the number of sample is less or the
sample attributes have high spatial variability then the
amount of coal resource with high level of confidence
(measured resource) will be small in term of
geostatistical approach. The problem here is that the
spatial variability of coal attributes did not always
coincide to the complexity of geological condition.
This situation always gives the much amount of
measured coal resource in polygonal method compared
to geostatistical approach. As for example in the case
of Seam EU2 of Warukin Formation where the
geological condition is simple, the measured coal
resource of polygonal method was about 36% larger
than the same resource category based on geostatistical
approach. For Seam SM2 of Tanjung Formation with
moderate geological condition, polygonal method gave
larger different as about 58%. It seems that the more
complex geological condition then the different of
measured coal resource between polygonal and
geostatistical methods was larger, but the different for
indicated and inferred resources was similar. To
produce the close amount for both approach on
measured coal resource, then the relative error of
geostatistical estimation variance should be set up to
21%.
In the case of drillhole spacing optimization at Cluster
1 of Mulia area showed that relative error through
estimation variance was also effective to estimate the
optimum spacing. The different attributes of course
gave the different estimation variance, but the trend
was same that much closer the drillhole spacing then
the relative error was smaller as well. By plotting the
reducing on relative error against additional drillhole
number we showed that drillhole spacing of 200 m
200 m seemed to be optimum.
CONCLUSION
Based on the evaluation of coal resources estimation at
Mulia and West Senakin coal deposit we conclude
that:
1. More complex geological condition implied to the
larger different on measured coal resource between
polygonal method (based on SNI 1999 procedure)
and geostatistical approach.
2. The geostatistical approach based on spatial
variability of coal attributes can be incorporated to
the SNI 1999 procedure to generate the close
amount of coal resource among two approaches,
i.e. by setting the relative error up to 21% for the
measured category.
3. The drillhole spacing optimization using estimation
variance for Seam EU2 of Cluster 1 at Mulia area
showed that seam thickness had less spatial
variability than ash content and total sulphur, and
then 200 m 200 m can be considered as optimum
spacing.
ACKNOWLEDGEMENTS
We would thank sincerely to the Department of
Mineral Resources, PT. Arutmin Indonesia in
supporting to utilize the data set and publish this paper.
We also appreciate the Institute for Research and
Community Service of ITB (LPPM – ITB) for the
support on Research Grant 2010.
REFERENCES
Badan Standardisasi Nasional (BSN), Klasifikasi
Sumberdaya dan Cadangan Batubara (1999).
de Souza, L E, Costa, J F C L, and Koppe, J C,
Uncertainty estimate in resources assessment: A
geostatistical contribution, Natural Resources
Research, 13(1), p.1-15 (2004).
Diehl, P and David, M, Classification of ore
reserve/resources based on geostatistical
methods, CIM Bulletin CIM Bull., v.75, no.838,
p.127-136 (1982).
Heriawan, M.N., Syafrizal, Suliani, C.P., and
Anggayana, K., Comparing Coal Resources
Classification Based on Geostatistical Approach
and Indonesian National Standard (SNI 1999)
Procedure for Complex Geological Structure of
Tanjung Formation, South Kalimantan,
Proceedings of International Symposium on Earth
Science and Technology 2010, Fukuoka, p.
129-132 (2010).
Heriawan, M.N., Noor, R.H., and Syafrizal, Optimasi
Spasi Pemboran Eksplorasi pada Endapan
Batubara dengan Pendekatan Geostatistik, Studi
Kasus Batubara Formasi Warukin Kalimantan
Selatan, Prosiding Temu Profesi Tahunan TPT
XX PERHAPI 2011, Senggigi – Lombok, p.
11-20 (2011).
Heryanto, R., Supriatna, S., Rustandi, E., and
Baharuddin, Geological Map of Sampanahan
Sheet, Kalimantan, Scale 1:250.000, Geological
Research and Development Center, Bandung
(1994).
Koch, G S Jr, and Link, R F, Statistical Analysis of
Geological Data, Wiley, New York (1970).
Sikumbang, N and Heryanto, R, Geological Map of
Banjarmasin Sheet, Scale 1:250.000, Geological
Research and Development Center, Bandung
(1994).