APCOM 2009 - PROCEEDINGS 95

Modelling and Resource Estimation of a Thin-Layered Lignite Deposit

Ioannis K. Kapageridis
Mining Engineer, PhD CEng CSci MIMMM
Laboratory of Mining Information Technology and GIS Applications
Department of Geotechnology and Environmental Engineering
Technological Educational Institute of Western Macedonia, Greece

Christos J. Kolovos
Mining & Metallurgy Engineer, PhD
Mines Strategic Development Unit, Mines Division
Public Power Corporation of Greece SA

Abstract

The lignite deposits of the Kozani-Ptolemais-Amyntaio basin in western Macedonia (North-West Greece)
present a difficult modelling problem. Each deposit consists of several thin lignite layers ranging from a
few centimetres to a few tens of centimeters in thickness, and thin intercalated sterile layers of marly
limestones, marls, clays, and sands. These layers are practically impossible to model individually due to
fast lateral transition of lignite layers to humus clay and vice versa and the lack of reliable and detailed
stratigraphic correlation, leading to compositing methods being applied prior to any interpolation and
modelling. The complexity of the deposit is further increased by the presence of a number of faults. This
paper presents the modelling and estimation procedures applied during the most recent study of the South
Western Field – a new deposit that is considered for development and lignite production over the next
years and will support the operation of a new power station in the area. Special focus is given to drillhole
database structuring, fault modelling, stratigraphic correlation, mineable lignite intervals compositing, and
resource estimation. The use of specialised mine planning software in all stages of modelling and
estimation is thoroughly explained.

Introduction and Deposit Description

The study presented in this paper concerns the South Western Lignite Field (SW) located between
Ptolemais and Kozani in North-West Greece. The SW Field is one of the “zebra” type lignite deposits of
the Kozani-Ptolemais basin (Kolovos, 2006). The Public Power Corporation of Greece operates a number
of surface lignite mines and power stations in the region. Bucket wheel excavators, belt conveyors and
stackers are used as the main mining equipment. Shovels plus dumpers handle the hard overburden. PPC
considers the development of the SW Field as an expansion to current operations.

The lignite bearing strata in the study area belong to the upper Pliocene. Overburden material belongs to
the Pleistocene and Holocene (Anastopoulos et al., 1972). Over the lignite bearing strata lie a series of
green-gray clay and marl layers - an alternation of mainly sandy clays, calcareous marls and silty clayish
marls. A series of yellow-brown sandy layers follows, consisting of mainly calcareous sands with clay
intercalations and occasionally sandy marls. In this formation, numerous lenticular intercalations of
sandstones and conglomerates-consolidated pebbles exist. Over the yellow-brown layers lies a series of
red-brown clays and conglomerates - an alternation of reddish sandy clays and loose conglomerates with
clay-silica binding matter. The Proastion conglomerates are the most recent system of stream deposits
consisting of loose conglomerates, sands and sandstones.
96 APCOM 2009 - PROCEEDINGS

Software Used in the Study

Two software packages were used in this study. The first is an in-house package called METAL,
programmed by PPC personnel, that allows mineable lignite intervals evaluation, resource and reserves
estimation, hard rock reserves estimation, etc. (Karamalikis, 1992). The second is the well known Vulcan
3D software package built by Maptek Pty Ltd. Vulcan is one of the top general mine planning packages
available, integrating specific coal oriented database and modelling sub-systems called ISIS and GridCalc
(Lee and Kapageridis, 2004). ISIS is an interactive database editor used in conjunction with Vulcan
databases. Grid Calc is a very powerful grid modelling and manipulation tool that provides both an
interactive interface and a sophisticated macro command driven interface. Grid Calc provides facilities to
generate grids from irregularly spaced data points.

Database Management

Raw data from more than 1500 drillholes covering the entire area of lignite deposits in Ptolemais were
exported from METAL into an ISIS database. This data included collar information and lignite quality
analyses (Ash, Lower Heating Value-Kcal, and Relative Humidity) on specific drillhole intervals.
Average drillhole spacing was around 200m. A general description of the material in each interval
(DESC) is included together with a number of coded descriptive fields (R1-R13) regarding parameters
such as colour and texture. Table 1 presents the structure of the drillhole database built in Vulcan with a
short explanation of each field. Only the first two tables (Collar and Raw) were filled directly from the
information exported from METAL, and are discussed here. PPC drillholes have a unique CODE and
PCODE field value corresponding to a numerical and an alphanumerical ID of each drillhole. CODE in
particular is a combination of the X and Y coordinate of each hole allowing quick location of the collar on
a map. The DESC field combined with a number of other descriptive fields (R1-R13) was used to
calculate the COLOUR field that provides a specific colour to raw drillhole intervals for plotting
purposes. An ISIS field calculation script was generated for this calculation.

Table 1: Database structure in ISIS

TABLE FIELDS
CODE PCODE X Y Z
COLLAR
hole code hole name collar coordinates
FROM TO DESC R1-R13 LENGTH ASH KCAL RH COLOUR
RAW lower
bottom material descr. material
top offset thickness ash heating humidity
offset type fields colour
value
FROM TO DESC LENGTH ASH KCAL RH SR SG
COMPO lower
bottom composite stripping specific
top offset thickness ash heating humidity
offset type ratio gravity
value
DEPTH CODE
CORREL bottom
horizon
offset
FROM TO THICK ZONE
HARD bottom overburden
top offset length
offset zone

Stratigraphic Correlation and Structural Modelling

Stratigraphic correlation of the SW Field deposit was performed by the Institute of Geology and Mineral
Exploration (IGME) of Greece in order to study the tectonics of the deposit. Correlation was based on a
 APCOM 2009 - PROCEEDINGS 97

number of drillhole section plots prepared in Vulcan, and the original drillhole logs. Plotting was
performed using Vulcan’s Batch Plotting utility which allowed quick generation of very complex plots
combining information from multiple databases and tables.

Lignite seams were not modelled individually due to their excessive number and fast lateral transition of
lignite layers to humus clay and vice versa, which makes practically impossible the development of a
reliable correlation mechanism. Lignite seams were grouped in two main bands, the lower and the upper
band. Three characteristic horizons that could be correlated were used for determining faults. The
characteristic sand horizon is a very specific horizon of the Kozani-Ptolemais basin. It is light grey and
10-20cm thick, covering the entire basin. It consists mainly of feldspars (plagioclase), biotite, quartz and
in smaller amounts crystals of hornblende, zircon, apatite and rutile. The Neritina horizon is another
characteristic horizon of the basin that is found between the lower and upper lignite band. It consists of
light coloured marls, ranging from a few centimetres to 9m thick, which contain fossils of gastropod
Theodoxus (Calvertia) Macedonicus, mostly known from its older name Neritina. Final marl lies below
the deepest lignite seam. It is a white-yellowish calcareous mud, 0.2-3m thick, with 92.1% CaCΟ3 and
5.6% MgCO3. The combined identification of the characteristic sand and final marl in a drillhole is a safe
indication of where the lignite bearing strata end, and drilling should stop.

The roof and floor of each lignite band was also correlated and modelled. The section plots were
produced in two directions, along and perpendicular to the long axis of the deposit. Drillhole columns
were split in two parts - the left coloured and patterned according to the COLOUR field of the RAW
table. The right part was coloured according to two of the R1-R13 fields that provide the actual material
colour. The two corresponding colours were mixed into a single composite colour using special fill
patterns. Not all drillholes had these two colour fields defined leading to some intervals showing white
(Figure 1). The following horizons were marked on the sections and were used to fill the CORREL table
of the ISIS database:

Table 2: Correlated and modelled surfaces and horizons

Surface - Horizon Code in Database Colour in Plots

Old mine dump areas MD Cyan
Recent dump areas DND Gray
Würm sediments DV Light blue
Proastio formations DP Red
Red-brown series DR Brown
Yellow-brown series DY Yellow
Green series DG Green
First Geological Roof GO1 Black
First Geological Floor GD1 Black
Second Geological Roof GO2 Black
Neritina horizon N Dark blue
Characteristic sand horizon F Light green
Second Geological Floor GD2 Black
Final Marl Roof OM Black
Base roof BG Dark cyan

Correlation and the study of faults were limited to an area extending around 500m outside of the SW
Field. The surfaces listed in Table 2 were modelled in GridCalc originally as grid models and then
converted to triangulation models for better visualisation and resource calculations.

As shown in Figure 2, the faults in the area follow a main direction NW - SE. This is the direction of the
main faults that generated the large tectonic graben of Florina - Ptolemais. The main fault population is
98 APCOM 2009 - PROCEEDINGS

located at the West and North West limits of the deposit. The throw of the faults is NE and SW, leading to
a compression in the NW limits of the deposit and the generation of faults vertical to the main direction.

The fault lines were combined into polygonal zones to be used in structural modelling of the deposit.
Some of the originally designed fault lines had to be extended for the zones to close. Grid Calc builds a
grid by modelling each zone independently and connecting the adjoining edges. Zones can also overlap in
which case an average value is used as the final result. Each zone must contain at least three data points
that define a plane. Inverse distance (ID1) was used to model all structural surfaces (roofs and floors,
Figure 3) while inverse distance squared (ID2) was used to model thickness. Four samples were used for
each estimation point within a search area of 300m radius.

Figure 1: Typical section showing surface models based on stratigraphic correlation and
fault modelling. Grey areas show the vertical extents of the lignite deposit
 APCOM 2009 - PROCEEDINGS 99

Figure 2: Plan view of faults and fault zones used for structural surfaces modelling. Red
lines are the original faults and black lines are extensions for fault zone generation

Modelling of Lignite Quality Parameters

Three quality parameters were of interest in this study - ash, mean relative humidity, and mean calorific
value of the mineable lignite intervals. A single grid model was generated for each of them using
weighted average values from the composited drillhole intervals. In other words, a single ash, mean
relative humidity, and mean calorific value was derived from each hole by weight averaging its mineable
intervals. Inverse distance squared was used for generating the quality models and the same number of
samples and search radius used for structural modelling. The quality estimates produced were only used
to get a general picture of the resource quality. As lignite seams are not correlated, it was not possible at
this stage to get a more detailed model of qualities and particularly their distribution along the Z axis.
This will be possible once a reserves study is performed and the mineable lignite is modelled separately
for each bench.
100 APCOM 2009 - PROCEEDINGS

Figure 3: 3D view of the SW Field lignite deposit showing faulted models of geological roofs
and floors of the two lignite bands

Modelling of Hard Formations

Hard formations within the overburden zones needed to be modelled for mining equipment selection and
scheduling purposes. Definition of hard formations was based on an extensive search in the original
drillhole logs. Hard horizons and their thickness were identified in the descriptions of drillhole intervals.
The total thickness of these horizons varies from a few centimetres to more than 20m. Hard formation
intervals were stored in the HARD table of the ISIS database. These intervals were also classified as per
overburden zone. Separate models of hard formations within each overburden zone were generated in
GridCalc and their volume and mass was calculated.

Evaluation of Mineable Lignite Thickness

The evaluation of raw lignite drillholes is a process of compositing seams into blocks of technically
“recoverable lignite” and blocks of “barren (waste) material.” The borehole evaluation criteria were
especially developed for the specific kind of deposit, exploited by the specific type and size of (existing)
equipment and aiming to produce lignite quality acceptable by the power plant. In multi-layered lignite
deposits, the blocks of “recoverable lignite” often include thin layers of barren rocks, having a thickness
of some centimeters. Such waste layers are too thin to be selectively excavated by the bucket wheels, and
constitute “pollution” to the original “geological” lignite. The average quality of the resulting blocks of
“recoverable lignite” should meet certain specifications, depending on the requirement of each consumer.
Evaluation of the average quality of blocks of “recoverable lignite” should take into account that, due to
the geometry of the sickle-shaped cut, the bucket wheel will in fact co-excavate adjacent blocks of barren
rocks. That imposes an additional source of “pollution” (or “dilution”). In a similar way, the blocks of
“barren” material often contain thin layers of lignite, which cannot be selectively excavated. These layers
constitute mining losses (Kolovos, 2007). The algorithm described here is implemented in the METAL
package. The evaluation algorithm goes through the following main steps (Karamalikis, 1992):
 APCOM 2009 - PROCEEDINGS 101

1. Initial raw seam coding - all seams of a hole are characterised initially as waste if the ASH field
is blank or the values of ASH field and seam thickness are greater than user defined limits. Seams
not satisfying the above are temporarily considered as lignite.
2. Waste seam compositing - the algorithm composites consecutive waste seams derived from the
first pass into “waste” blocks.
3. Compositing lignite down to next parting (Figure 4) - with the simplified picture of the hole
generated after the first two steps, the algorithm goes into the main task of appropriate
compositing seams into recoverable blocks of “lignite” and “waste”. To achieve this, it scans the
characterised lignite and waste seams of the hole until it finds a waste seam with thickness greater
than the parting thickness defined by the user. The algorithm temporarily composites into one
block the seams before the parting (the seven seams of Block A in Figure 4) and performs a
number of calculations. These include the specific gravity in dry form that will be then used to
calculate the mean ash value of the block, the subtraction of losses from the thickness of the first
and last seam of Block A, the mean weighted values for moisture, ash, specific gravity and
minimum calorific value of Block A. The algorithm can also take into account the increase of
thickness due to “dilution”; however, introducing a “dilution” factor to the borehole evaluation
software is not justified for the Greek multi-layered lignite deposits because it has resulted to an
undesired and incontrollable alteration to the geometry of the pit and to the calculation of the
reserves of “recoverable” lignite. (Kolovos, 2007). If the final mean ash of the block and the final
block thickness (after dilution and losses) satisfy the requirements for recoverable lignite, the
algorithm characterises the block as “lignite” and assigns to it the calculated final ash, moisture,
calorific value and specific gravity. If the final mean ash of the block or the final block thickness
does not satisfy the requirements for recoverable lignite then the algorithm removes the last seam
of the block and repeats Step 3 until it finds a part of the block that will satisfy the requirements
for recoverable lignite. This part is characterised as “lignite” and the process continues with all
combinations of the remaining seams in the block.

Figure 4: Example of mineable lignite intervals evaluation of a drillhole

102 APCOM 2009 - PROCEEDINGS

4. Compositing consecutive waste seams - once the recoverable blocks of lignite are constructed, the
program continues with the compositing of any consecutive waste blocks into composited waste
blocks.

There are other steps that follow regarding drillhole totals. The evaluated drillhole intervals were
imported into the COMPO table in ISIS which is used in the next stage of resource estimation.

Resource Estimation

Estimation of lignite resources for the SW Field was based on the structural floor and total thickness
model of the mineable lignite intervals (Figure 5). The total thickness model was adjusted to exclude any
lignite that is already mined or scheduled to be mined by existing operations. Similar models were
developed for overburden (volume above the structural roof of the mineable lignite intervals) and
intermediate intercalations (intervals classified as waste between mineable lignite). Figure 5 shows a
mineable lignite thickness map. It is clear that the higher thickness appears in the deep area of the deposit
where a number of faults bring the lower lignite band much deeper than usual. This is an area where the
mining method will possibly have to differentiate from the common bucket wheel excavators used as the
main excavating equipment in PPC lignite mining operations.

Figure 5: Mineable lignite thickness map of the SW Field (dark colours show areas of
higher thickness)

Conclusions

This paper presented the stratigraphic modelling and resource estimation study of the South Western
lignite field in North-West Greece. Data management and the evaluation of raw drillhole intervals to
recoverable lignite blocks were critical aspects of the study. The data and methodology used in modelling
the geometry and extents of this lignite deposit was thoroughly discussed. The particular problems arising
from the deposit complexity and the given solutions were explained. The study also clarified the
 APCOM 2009 - PROCEEDINGS 103

significance of the highly fragmented, deep area of the deposit as to the contained recoverable lignite and
provided important information for further mine planning and development of the South Western field.

References

ANASTOPOULOS, J.C., KOYKOYZAS, C.N., 1972. Economic geology of the southern part of
Ptolemais Lignite Basin (Macedonia – Greece), Institute for Geology and Subsurface Research, Athens.

KARAMALIKIS, N., 1992. Computer software for the evaluation of lignite deposits, Mineral Wealth
(Oriktos Ploutos – in Greek) 76M: p. 39-50.

KOLOVOS C.J., 2006, Technical Aspects Regarding the Opening of a New Mine with Old Already-
Existing Equipment, Energy Sources, Part B, 1: p. 1–7

KOLOVOS C.J., 2007, Statistical Tests for Prediction of Lignite Quality, Energy Sources, Part A, 29: p.
685–693.

LEE, D., and KAPAGERIDIS, I.K., 2004. New dimensions in mining software, Materials Word, Institute
of Materials, Minerals and Mining, 12:9, p. 31-33.