BORC_PAPER_FINAL

GEMCO
Groote Eylandt Manganese Orebody Study
BORC Conference – Bunbury
13-17 November 2000
Author:Dave Mallon
GEMCO
6 October 2000

GEMCO GROOTE EYLANDTMANGANESE OREBODYSTUDY – BORC Conference,Bunbury,November 2000
'Committed to Safe Production'
1
1 Introduction
The “Groote Eylandt Manganese Orebody Study” was jointly undertaken by GEMCO, SRK and Maptek. The
objective was the construction of a reconcilable resource/reserve model suitable for use as a foundation for the
continued operation of the mine, satisfying all criteria stipulated in the JORC code for reporting resources and
reserves. Modelling commenced in March 1999 and was completed on 29th November 1999.
Previous orebody models were obsolete due to the continuing advance of working faces in many of the operating
quarries. The reverse circulation drilling program of 1997 and a minor open hole precision program in 1998 had
yet to be successfully modelled. In effect, mining was proceeding in areas where ore quality and tonnages could
not be confidently predicted.
The ore deposit on Groote Eylandt is a significantly large world-class ore deposit. The ore horizons are sufficiently
continuous and regular within the mining lease to be mined well into the future and at the current mining rate, there
is at least 20-30 years of resource available.
Groote Eylandt ore is inherently complex with a high degree of variability in respect of key parameters such as;
thickness, Mn, Fe, Si, and P grades, density, ore yield, and lump/fines yield ratio. This may be ascribed to a
complex history of ore forming processes (chemical precipitation, sedimentation, and supergene enrichment) and
ore destructive processes (laterisation, erosion). There is fundamental difficulty in estimating the ore resources,
which has resulted in a complex history of several ore resource estimates.
2 GEOLOGY
2.1 GeologicalSetting
Groote Eylandt is predominantly composed of a stable basement of Proterozoic quartz sandstones and
quartzites. The manganese deposits are part of a blanket of Cretaceous sediments lapping onto the western
margin of the Proterozoic basement sandstones and quartzites of the McArthur Basin.
The manganese orebody is a sedimentary layer and gently undulates beneath the western plains of the island.
It extends over an area of about 50 square kilometres as an almost continuous horizon ranging in thickness up
to 11 metres. Refer to Fig 2.1.1 GEMCO Mine Location”.

2
Figure 2.1.1GEMCO Mine Location”
N
Alyangula
2 1 0 2 4 6 8 10
km
Mined out areas
Mineral Lease
Mineral Lease
Application
Road
Rowel
lHwy
Angurugu
River
Emerald
River
FQ
To Alyangula
CQ
F1Q
GQAQ
EQ
Tailing
Dams
J Deposit
Angurugu
BQ
DQ
Concentrator
Groote Eylandt
N
Groote Eylandt
0 10 20km
North Point
Island
North East Isles
Cape Beatrice
South Point
Winchelsea
Island
Northwest Bay
Gulf of
Carpentaria
Dalumba Bay
Marangula Bay
Mud Cod
Bay
AlyangulaAlyangula
AnguruguAngurugu
UmbakumbaUmbakumba
Sand dunes, ancient and
recent, calcareous in places
Alluvium, laterite, lateritic
soil, ferruginous gravel
Manganese oxides
Massive quartz sandstone,
cross bedded, pebbly in
places
Recent
Recent Mid Cretaceous
Mid Cretaceous
Mid Proterozoic
Port Langdon
Fig 2.1.2 Geological Setting”

3
“The manganese strata occurs between clay and sand beds which in turn have been deposited over an
undulating terrain of much older quartz sandstones. The sandstones outcrop as a rugged plateau extending
over more than half of Groote Eylandt.” 1 Erosion channels in the basement may have influenced both primary
manganese deposition as sediments and supergene enrichment by laterisation.
“The orebody consists of pisolitic and oolitic manganese oxides. These oxides are thought to have originally
been deposited as a chemical precipitate, forming a tabular sedimentary deposit in wave affected shallowsea-
floor environments during a period of rising and falling sea levels.
Subsequent to deposition, the manganese layer emerged from the sea during a drop in sea level. A period of
weathering followed, covering the manganese with alluvial clays and gravels.
The depositional events were followed by a long period of tropical weathering which extensively modified the
upper parts of the sediment profile. Pisolitic manganese oxides underwent partial to complete remobilisation
and recrystallisation that resulted in the formation of hard cemented pisolite and massive manganese oxides.
The overlying clays and gravels were strongly oxidised and leached to form the iron and alumina rich laterites
that are now excavated off the manganese ore as overburden.” 2
The ore horizon is covered by laterite overburden ranging in thickness up to 25m. In current areas, the
overburden averages 5m thick.
Figure 2.1.2, titled “Geological Setting” shows the geological distribution of manganese ore on Groote Eylandt.
Figure 2.3, titled “Groote Eylandt Lithological Profile” diagrammatically depicts rock types, mining horizon and
the relationship to the stratigraphic model.
2.2 ManganeseOreGenesis
The manganiferous deposits have largely resulted from the chemical precipitation of sedimentary iron, alumina
and manganese minerals. There are two types of mineralisation, primary and secondary.
2.2.1 Primary Sedimentary Features
Whilst there is no clear evidence to indicate whether the primary manganiferous rich sediments are
of organic or inorganic origin, characteristics of both deposits appear to be present within the Groote
Eylandt mineralised resource. Oswald J (1990) has postulated an organic genesis for the
manganiferous pisolites, suggesting the formation may be a phenomenon of coastal waters, river
estuaries, lakes and embayments during a transgression of a stratified metal-enriched sea. An
alternative is to consider an inorganic genesis of the pisolites, in which near shore and beach
conditions (eg. wave action, long shore drift and winnowing) has strongly affected deposition.
The nature and form of these two potential origins are significant in terms of modelling. Those areas
along the original paleo coastline represent the well-sorted sequence dominated by elongated bars
of pisolites and oolites with ribbon like thickening. Whereas in other areas the mineralised strata has
formed a blanket cover of shallow embayed zones and deeply filled channels of the paleo drainage,
with the thickness and form of the layers strongly controlled by the direction and slope of the drainage
system.
Mining has exposed primary sedimentary features that suggest a shallow marine environment. The
ore contains pisoliths and ooliths, which is evidence of wave action influenced accretion. Inverse and
normal graded bedding, cross bedding and ripple marks are evident indicating deposition during a
minor change from marine transgression to regression. In primary ore, manganese pisolites are
loosely embedded in a matrix of white kaolin clay with minor quartz sand. The primary ore is believed
to be most closely represented by loose mangan pisolite and loose mangan oolite.
1
OperationsGuide – GEMCO1991
2
OperationsGuide – GEMCO1991

4
It was the work of Bolton that clarified the geological relationships between primary sedimentary
features and secondary overprinting by manganese and laterite. The relationship between
stratigraphic units, primary deposition and supergene enrichment is directly related to the topography
of the Proterozoic quartzite basement.
2.2.2 Secondary Manganese Enrichment
Lateritisation has mobilised manganese, iron, alumina and silica in the vertical profile. Manganese
oxides through deep tropical weathering, during the Tertiary period have progressively replaced the
kaolin matrix in the loose pisolite. This has caused the development of moderately thick laterite
profiles above the ore. Digenetic and supergene processes have recrystallised the primary
manganese oxide minerals in the loose pisolite (pyrolusite, romanechite, todorokite, vernadite,
lithiophorite, all with kaolinite gangue) to cemented mangan pisolite and massive mangite (dense
cryptomelane and pyrolusite) at the top of the pisolitic ore and at the bottom of the siliceous faces .3
The area has been extensively lateritised. In the wet season leaching of rocks occurs and in dry
season the solution containing the leached ions is drawn to the surface by capillary action to be
evaporated leaving salts to be washed away. Na, K, Ca, Mg are readily depleted and this can result
in a complex vertical and lateral distribution of chemical environments. Specifically a solution
containing these ions under suitable pH conditions can dissolve silica, leaving a clay saprolite.4
3
1997 Exploration& InfillDrilling Report – Cameron & Mallon
4
Runge Mining – October1998 Resource

GEMCO GROOTE EYLANDTMANGANESEOREBODYSTUDY – BORC Conference,Bunbury,November 2000
6
3 Limitations of Previous Models
Due to a complex history of ore forming and ore destructive processes, there is fundamental difficulty in estimating
the ore resources. This has resulted in an assortment of ore resource estimates, MINEX in 1990, MRT in 1993,
BHPE in 1994/5, BHPE/GEMCO hybrid model in 1996 and RUNGE MINING in 1998. This study has been able to
build upon the learnings of these studies.
The limitations of the previous models are:
 Smoothing over heterogenetic stratigraphic ore units contributing to the poor fits of variograms and poor grade
and tonnage reconciliation;
 Inaccuracies in the previous modelling of stratigraphic units;
 Previous models were based on a 40% Mn cut-off grade to identify the ore/waste envelope. This does not
represent the true envelope of that mined. The extraction of ore actually begins at the point where the dense
massive manganese cap is struck often resulting in ore with grades as low as 35%;
 Previous studies cut sample grades, which introduced bias;
 Old RE holes were excluded within 120m of RC drilling in previous work. This study developed procedures to
identify down hole smearing and included drilling on individual merit;
 Previous exclusion of data from areas mined prior to 1990;
 The use of old ‘’look-up tables” of assumed yield have resulted in the overestimation of yields.
4 Drill Hole Inclusion
The model uses 5,597 drill holes for stratigraphic modelling and 5,316 holes with valid assays for grade estimation,
out of the existing 6,314 in the database.
Numerous campaigns of drilling have been conducted, however there are significant differences in quality of the
various drilling techniques and some samples are considered inappropriate for resource estimation.
All drill holes existing in GEMCO’s database have been assessed on individual merit to determine validity of
including all or part of their data.
All drill holes with valid geological logging were used to determine stratigraphic intervals and included: Calweld
bulk sampling; reverse evaluation (open precision or rotatory air blast RE, ST, WB); reverse circulation; and blast
holes. Reverse evaluation and reverse circulation drilling are the only holes used for grade estimation. This data
provides the most comprehensive coverage of the mining leases. They were drilled for resource definition and
have compatible methods of assaying.
5 Data Preparation, Analysis & Validation
Work Instructions govern all the sample preparation and assaying techniques and the lab is audited under the
NATA QA system.
The data validation process undertaken for this model was exhaustive. Hatch & Associates wrote the validation
procedures in conjunction with the mine team. Only when data was validated was it included in the database for
further validation and geological coding.

7
6 Investigation of Down Hole Smearing
Down hole smearing is the name given to the phenomena where particles from the wall of the drill hole continue
to fall out of the wall and contaminate samples taken further down the hole. Open hole percussion drilling
techniques are particularly prone to this problem. If this involves mineralised material and remains undetected,
significant overestimation of deposit size is possible. Normally a distinct “tail” of gradually declining grade of the
contaminating mineral can be seen in an otherwise unmineralised unit below. However if the down hole
contamination is a significant proportion of the recovered sample, then underlying units could be logged as a
mineralised sample.
This study has shown the phenomenon is a major problem in reverse circulation drilling carried out on Groote
Eylandt. Ore quality and thickness in quarry faces were compared to RC holes in close proximity. In a number of
cases, ½-1m of gangue had been logged as ore due to down hole smearing. It quickly became evident that it was
essential to develop methods to identify downhole smearing in the GEMCO drilling database to produce a realistic
resource estimate.
It is important to note that the use of measured yield in this study counteracts the overestimation of product tonnes
resulting from down hole smearing.
7 Bulk Density
The bulk density used in the estimation was investigated. It was found that a more recent study (Mallon 1998) of 227
samples showed significant variability and quite different results to the “look-up” table, based on 98 samples from
Baxter& Krueger(1980). The 1980 and 1998 datasets weremodelledas separatevariables and compared. However,
due tothe vast increaseintonnes usingthenew data, resources and reserves are quoted using themore conservative
Baxter & Krueger data until such time that the bulk densities can be verified.
GEMCO is aware that the measurement of density and the use of look-up tables requires further attention.
8 Yield
The GEMCO concentrator upgrades the manganese ore by crushing, wet screening and heavy media separation.
Two basic types of products are produced:
 Lump, produced by using the heavy media drum separator and has a top size of 75mm.
 Fines, produced by using heavy media cyclones and has a top size of nominally 6.7mm.
Very fine grained material, less than 0.5mm is removed as either sand tails or slimes. This represents
approximately 45% of material processed.
8.1 MeasuredVsLook-upTableYield
Previous studies on Groote Eylandt have used a “look-up table” to assign yield per lithological type. The main
problem with using assumed (look-up table) yields is that product is overestimated in ore hosted within a clay
matrix.
Measured yield (obtained from sample preparation) provides a more accurate estimate as it counteracts the
problem of down hole smearing and eliminates poor reconciliation of previous estimates in clay rich zones.

8
The following equation represents the method by which a measured yield percentage is obtained:
100
SplitSampleAwashed-PreofDry Weight
SplitSampleAwashedofDry Weight
=Yield)(AssayedYieldMeasured 
Although it is thought to be a more indicative approach to estimating total yield, the problem of attributing a
lump/fines split is still present. This study must still partially rely on a lithological “look-up table” to assign a
relative ratio of lump and fines to the measured yield. It must be borne in mind that measured yield is based on
RC cuttings which are not identical to ROM and product material in terms of sizing fractions. The processes
involved in drilling and in the concentrator are not identical.
Although lump and fines tonnes have been modelled, the proportioning of lump and fines grade can only be
performed duringreserve calculationstage, as it isdependent onthetonnes and grade withina specific volume.
The fines/lump grade proportioning was calculated using the following equation to perform combined tonnage
and yield weighted average grades:
Using Mn (A assay) as an example;
Lump Tonnes = Lump Yield x SG x Volume
Fine Tonnes = Fine Yield x SG x Volume
and the contained metal for each element.
Contained Lump Mn (A assay) = Mn (A assay) x Lump Yield x SG x Volume
Contained Fine Mn (A assay) = Mn (A assay) x Fine Yield x SG x Volume
It was then a simple procedure of using the Block Reserves Advanced option within Vulcan and summing the
contained metal and yield tonnages for the relevant breakdowns necessary and using the following weighted
average formulae to calculate the yield grades. Using Mn (A assay) as an example;
Mean Lump Mn (A assay) = [Mn (A assay) x Lump Yield x SG x Volume]
[ Lump Yield x SG x Volume]
Mean Fine Mn (A assay) = [Mn (A assay) x Fine Yield x SG x Volume]
[ Fine Yield x SG x Volume]
9 Summary of Data Quality
Table 9.1, titled “JORC Code Data Quality Assessment”, summarises the quality of various parameters and data
in accordance with requirements of the JORC Code, as described in Table 1 of the July 1999 JORC Code.

9
Table 9.1 JORC Code Data Quality Assessment
Assessment Criteria Confidence Description
Data Density Reliable Drillingdensityvaries over deposit,generallysamples are collectedat0.5m
intervals,and a drillpatternof60 by120m is consideredadequate for
definingmeasuredresources.
SurveyControl and
AccuracyofSample
Location
Generally
Adequate to
Poor
Numerous drill collarlocationerrors werenoted and corrections madefor
this study.
SamplingandAssaying
Technique
Reliable Sample preparation and analysis is undertakenbythe in-houselaboratory,
which maintains NATAstandards.XRF& ASS assayingmethods are
used.Every30th
sampleis riffle splitatthe drill siteandevery20th
assayis
replicated to maintain analytical quality. An independentstudyofanalytical
accuracyandprecisionwas notundertaken.
GeologicalDescription Reliable A comprehensive geologicalloggingprocedurehas been usedatGEMCO
for RE and RC drilling campaigns anda significantdatabase ofdescriptive
data is available.Logs from older CB holes have beenreinterpreted using
this procedure.This informationwas used as the basis for interpreting
stratigraphic layers.
DrillingTechniqueand
Sample Recovery
Adequate to
Poor
Reliable
RE Series,were earlier holes usingopenhole rotaryandpercussion
techniques,and manyholes maynotbe adequatefor reliableresource
estimate. GEMCOhas redrilledtwinRC holes to replace dubious holes.
This studyhas precludedmanyRE holes with suspectdata includingdown
hole smearing.
RC Series,theseare reversecirculationpercussion holes,with good
sample recoveryand areconsideredreliablefor resourcedefinition.
GeologicalInterpretation Reliable This studyincludes theinterpretationofallintersections intoa numberof
stratigraphic layers as the maingeologicalcontrolused inmodelling.
Yield Reliable Previous resourceestimates relied solelyon the application ofan assumed
(look-up)yieldfactorbasedon loggedlithology.In this study,assumed
yield has been modelled for comparison purposes only.
The resource estimatecarried inthis reportis calculatedusingmeasured
yield.Measured yield (obtained from samplepreparation) is thoughtto
provide a moreaccurate estimate as itcounteracts the problem ofdown
hole smearingandeliminates the poorreconciliation ofprevious estimates
in clayrich zones.
Geologicallogging and sample interval is carriedoutin0.5m increments.
The applicationoflook-up yield fails whenthe intersectionofore withinthis
interval is less than0.5m,ie;the entireintervalhas a highyieldapplied.
TonnageFactors
(bulk density)
Adequate Previous resourceestimates have useda look-uptable to allocate an
assumeddensityfactor basedon lithology.The look-up tableis derived
from a studybyKruegerandBaxter in 1980of97 samples.
In 1998,227 samples were analysedwhichindicatethattheoriginalwork
in 1980 underestimates bulk density.As there is the potentialfor a
significantbias in tonnage,a separate estimateofresources was made
usingthe 1998measurements to investigate sensitivityto density
assumptions.
The resource estimateusingthe revised “look-uptable”with1998 densities
shouldbe viewedas an upside case untilitcan bequalifiedbyongoing
reconciliation. Until thattime,theresource estimateusingthe original
assumeddensities shouldbe carried.
Estimation Techniques Reliable
Reliable
Adequate
Shape,the upper andlower boundaries ofeachstratigraphiclayer have
been gridmodelledsympatheticallyto the basementstructure,re-
triangulated to include intersectionpoints,to produce a series ofsurfaces
Grade,has been estimated usingordinarykriging,constrainedwithin
stratigraphic layers andwithstronglyanisotropic searchellipses (flatplate-
like) whichhave beenvariedbygeologicaldomains (largelybased on
basementmorphology).
Tonnage Factors,The tonnagefactors ofbulkdensityand yieldhave
been interpolatedusingordinarykriging.

10
Cut-Offgrades Reliable Cut-offgrades of40% manganese have been usedto constrainthe
resourceandreserve estimateas thebeneficiationprocess is thoughtto
separatewaste from ore atbetween37 and40%productgrade.
10 Stratigraphic Layer Interpretation
Due to the vertical grade variability and the horizontal homogeneous nature of the deposit, a full stratigraphic model
was required to restrict kriging searches within vertical heterogeneous units. This has been a major advance with
much-improved incorporation of stratigraphic layering into the model.
A very real basement control on the ore-hosting lithologies became apparent. The basement disconformity surface
has dictatedthedepositional environment of theoverlying Cretaceous manganesesediments. Therefore, it was logical
to model the basement first, as subsequent deposition emulates the shape of the disconformity surface.
Figure 10.1
3D View ofSSTBasementDisconformity
Outcrop – green,Subcrop– blue,Domains – red
VE = 1:5 View to the NE

11
Figure 10.2 Cross-sectional view looking north at 41000N of a slice
of the block model coloured by stratigraphy (5x vertical exaggeration).
Figure 10.3 A view looking north-west of multiple slices of the block
model coloured by stratigraphy (5x vertical exaggeration). Refer to
Figure 10.3 for colour legend.
11 Domainal Stationarity and Variography
The selection of the geological domains for the variography were based on a number of arguments; including ore
primary and secondary phase manganese ore thickness’ and the Proterozoic/basement disconformity topography.
Two types of domain were identified, namely aerial and stratigraphic domains.
It became clear that the basement surface was significant in terms of defining the orientation and shape of the
mineralised layers. Specifically, the presence of deep channels and embayments sympathetic to linear jointing in
the basement outcropping to the east, have a major influence in the location of specific ore types and the direction
of anisotropy.
It is now believed that anisotropy is closely related to the secondary enrichment of the ore as well as the original
depositional environment. The new stratigraphic surfaces have modelled the secondary enrichment by modelling
the massive, cemented, loose and siliceous layers separately. Thus, variography could now be modelled within
the separate heterogeneous stratigraphic units.
The objectives of the SRK component of the report were to calculate the variograms, determine kriging
neighbourhood parameters and to carry out an estimation variance study to determine optimal drillhole spacing.
Each aerial domain was divided into 4 stratigraphic horizons and variography was carried out on 6 variables within
each domain or horizon resulting in 60 fitted models. These models were then used to conduct kriging

12
neighbourhood tests. These established that the minimum block size that could be estimated reliably was 120m x
120m x 0.5m. The estimation variance tests show that if a 60m x 60m drill grid is used on a 120m x 120m x 0.5m
block there is 1 – 1.5% error i.r.o. Mn grade.
From a geostatistical point of view, the estimate is about as good as it could be, given the available dataset.
Generally the variography is poor (high nugget effects, poor model fits etc.) but this due to the inherent nature of
the ore and to the data density. Only by greatly improving the sampling density could the variography be improved.
Furthermore it is not possible to use smaller block sizes in the model because of the very high nugget effects.
12 Grade Estimation
Previous estimates created multiple models to cover the broad deposit. In this study a single three-dimensional
block model was created, thereby simplifying the procedure and enabling full visualisation and validation.
As the majority of the sample lengths were 0.5m long straight length compositing of samples was used. The few
samples less than 0.5m were discarded during grade estimation. The few samples greater than 0.5m were left at
their original length.
The parent block size was set at 120m x 120m x 0.5m as determined by SRK, while the sub-block size was set
20m x 20m x 0.5m for mining related reasons. Under advisement from SRK, ordinary kriging was used for all
variables. A multiple pass estimation process was followed and reported.
There were 24 grade variables, 2 specific gravity variables and 2 sets of lump/fines yield variables. Grade variables
consisted of insitu and product grades. The duplicate specific gravity and yield variables were required to compare
old and new density data and the measured yield verses look-up yield techniques. The shear number of variables
added to the complexity and size of the model as well the degree of difficulty in producing reserve calculations.
Figure 12.1 View looking north-east of the sandstone basement. Notice Paleo-basement highs, basins and
valleys (5x vertical exaggeration).

13
Figure 12.2 Block model slice of average Mn_A grade. Notice the depositional position in relation to the
basement topography. Paleo coastlines are represented by well-sorted sequences dominated by elongated bars
of pisolites and oolites with ribbon likegrade profiles around island and headland features. Whereas, in other areas
the mineralised strata has formed a blanket cover of shallow embayed zones and deeply filled channels of the
paleo drainage, with the thickness and form of the layers strongly controlled by the direction and slope of the
drainage system.
13 Resource Categorisation
The JORC Code is quite clear about the level of confidence required for reporting Mineral Resources and Ore
Reserves. “No amount of confidence in the modifying factors for the conversion of a Mineral Resource into an Ore
Reserve can override the upper level of confidence which exists in the Mineral Resource”.5
The base level for confidence is the Mineral Resource. As the Ore Reserve is a subset of the Mineral Resource,
the confidence in the Ore Reserve cannot be more than that in the Mineral Resource. To meet the JORC Code
criteria, GEMCO had to either reduce the minimum thickness of the Reserve or increase the minimum thickness
of Resource.
On occasion, ore is mined on Groote Eylandt in opened pits at less than 1.0m thickness where thicker ore, already
extracted, has thinned onto basement highs. However, new pits would not be opened in areas where ore thickness
is less than 1.0m. Therefore, ore less than 1.0m thickness was not included in the mineral resource. As the JORC
Code states, “portions of the deposit that do not have reasonable prospects of eventual economic extraction must
not be included in a Mineral Resource”6, the Mineral Resource minimum thickness was increased to one metre.
All categories are now qualified with the percentage of cells estimated in the first, second, third and fourth kriging
neighbourhood passes. The kriging neighbourhood parameters developed by SRK during their variography study,
5
The JORC Code - July 1999
6
The JORC Code - July 1999

14
sets a minimum and maximum number of samples for use in grade estimation within a given distance in a specific
search ellipsoid. This technique was introduced to determine confidence in grade estimation within each category.
The Mineral Resource and Ore Reserve classifications have been revised as follows;
Table 13.1 Ore Reserve/Resource Classifications
1999 Ore Reserve/Resource Classifications
Mineral Resource
Measured Resource - Maximum RC drill hole spacing 60m x 120m
- Minimum ore thickness 1m
- Mn >or = 40%
- A% 1st pass, B% 2nd pass, C% 3rd pass, D% 4th pass
Indicated Resource - Maximum RC drill hole spacing 120m x 120m
- Mn >or = 40%
Inferred Resource - Maximum RC drill hole spacing 240m x 240m
- Mn >or = 40%
Ore Reserve
Proven Ore Reserve - Maximum RC drill hole spacing 60m x 120m
- Mn >or = 40%
Probable Ore Reserve - Maximum RC drill hole spacing 120m x 120m
- Mn >or = 40%
Contained Product
Proved - Beneficiated product tonnes within Proven Ore Reserve
Probable - Beneficiated product tonnes within Probable Ore Reserve
Assigned Recoverable Proved Ore - Proved Ore Reserve
14 Reconciliation
A reconciliation exercise was undertaken upon completion of the project. Due to the blending of ROM ore into the
concentrator, it is difficult to trace the ore back to the specific source. Therefore, the reconciliation exercise only
tested the model on a global scale. A period between 30/07/97 and 30/07/99 was selected, as there were clear
mining records for this period. The results for the total grade and tonnes are tabled below.
Actual 2,253,168
relative
error 5,942,885
relative
error 3,221,289
relative
error 49.0
relative
error 3.9
relative
error 6.9
relative
error 0.080
relative
error 54.2
relative
error
1999 Model 2,115,377 -6.1% 6,230,163 4.8% 3,135,573 -2.7% 49.1 0.2% 3.9 0.5% 6.8 -1.8% 0.083 3.1% 50.3 -7.1%
1996 Model 1,635,001 -27.4% 4,918,291 -17.2% 2,882,059 -10.5% 50.0 2.0% 3.7 -5.8% 6.1 -11.0% 0.084 4.3% 58.6 8.1%
Volume
Reconciliation 30/07/1997 - 30/07/1999
TONNES
INTO PCS FROM ROM* P YieldTONNES Mn Fe Si

15
The reconciliation for the total grade and tonnes was excellent with very low variance from actual. However, when
broken down into lump and fines the reconciliation appears less favourable, although still within an acc eptable
margin of error. What this does indicate is that further investigations into determining a split between lump and
fines are necessary.
15 Conclusion
The new resource model has proven to be a very great improvement on previous models. Many problems identified
during previous studies and audits were addressed and overcome. In particular, better delineation and use of the
stratigraphic and domainal boundaries of the resource.
The inherent problem of smoothing variability by using “look-up tables” has been addressed in terms of yield. Due
to the lack of samples, a look-up table was required to assign density. However, with the planned geophysical
Wire Line logging in 2001, this limitation will be eliminated with the use of real measured data. Further work is also
required in the determination of an accurate method of proportioning lump and fines grade. Remaining problems
are either a function of geological fundamentals and data density or are largely artefacts of the RC drilling method.
Despite the apparent simplistic stratigraphic nature of the Groote Eylandt manganese deposit, the onlapping of
the stratigraphic layers require specialistic modelling techniques. The understanding of product tonnages and
grades alsoadd an order of complexity tothe resource modelling process. The experience and knowledge required
to “come up to speed” is considerable. This has been the underlying cause to the shortcoming of previous models.
To quote Dr John Cottle’s “Resource Estimation Models Review” of 1997, “the geological aspect is a most crucial
part of the entire effective project execution process, and will continue to increase in importance with time”.
16 Acknowledgment
The success of this project could not have been achieved without the sterling effort put in by the GEMCO team
and contributing consultants, namely:
Max Bygraves - Mine Scheduling Engineer, GEMCO;
Andrew Hocking - Chief Mining Engineer, GEMCO;
Marc Thuijs - Hatch
Heather McIntyre – MAPTEK;
Murray Rayner – MAPTEK;
Anthony Wesson - SRK Consulting;
John Vann - SRK Consulting;
Oliver Bertoli - SRK Consulting;
Ed Swindell - Samancor.
Volume TONNES
INTO PCS FROM ROM* TONNES MN FE SI AL P Yield
Actual 2,253,168 5,942,885 1,656,672 48.6 3.5 8.0 4.0 0.083 27.9
1999 Model 2,115,377 6,230,163 1,901,270 48.2 4.2 7.4 3.8 0.083 30.5
Volume TONNES
INTO PCS FROM ROM* TONNES MN FE SI AL P Yield
Actual 2,253,168 5,942,885 1,564,617 49.5 4.4 5.8 3.6 0.078 26.3
1999 Model 2,115,377 6,230,163 1,234,303 50.6 3.5 5.8 3.1 0.083 19.8
LUMP**
FINES**
FINAL PRODUCT

16
17 Bibliography
COTTLE, J., 1997 Groote Eylandt Manganese Project, Cotlo report, February 1997, 27p.
EDGAR, C., MALLON, D. T., 1997, 1997 Exploration & Infill Drilling Report. GEMCO report, August 1998.
GEMCO, Operations Guide – 1991.
HORROCKS, N., 1964 Physical Geography and Climatology. Second edition, 1964, 370p.
JARMAN E, A., MATTHEWS, W. H., 1970. Geology Made Simple. September 1970, 305p.
JOINT COMMITTEE OF THE AUSTRALASIAN INSTITUTE OF MINING AND METALLURGY, AUSTRALIAN INSTITUTE
OF GEOSCIENTISTS AND MINERALS COUNCIL OF AUSTRALIA. The Australasian Code for Reporting Mineral
Resources and Ore - The Jorc Code. July 1999, 17p.
KHOSROWSHAHI, S., SHAW, W. J., 1993,Geological and Geostatistical Modellingofthe GEMCO Mamnganese Deposits,
Groote Eylandt, NT. Mining and Resource Technology (MRT) report, GEMCO repo9rt G449A and B, June 1993
LEENDERS,R., 1995,Groote Eylandt Manganese Deposit - July 1995 Resources.BHPE reportRE01965,GEMCO report
G475 A and B, July 1995.
MALLON, D. T., 1998. 1998 Density Study. GEMCO report, 29 January 1999.
MALLON, D. T., 1999b.Groote Eylandt Manganese OrebodyStudy - November 1999 Resource Volume 1,Part1: Geology,
Data Preparation, Stratigraphic Modelling & Resource Comparisons. GEMCO Report, 74p.

17
RAYNER, M. J., 1999b. Groote Eylandt Manganese Orebody Study - November 1999 Resources Volume 3,Part 1: Block
modelling, Grade Estimation, Resource and Reserve reporting. Maptek Report, 87p.
SWINDELL, E. P. W., 2000. Technical Audit – GEMCO Mineral Resource Model, May 2000, SAMANCOR Report, 20p.
THOMPSON, H., 1998 Groote Eylandt Manganese Deposit - October 1998 Resource,Volume 1, Runge Mining report no.
3048a, Novembe1998.
THUIJS, M. J., 1996,GEMCO Drill hole System (GDHS) User Guide.BHP Engineering,doc no.B200-8001/MJT,April 1996,
84p.
WESSON, A. J., BERTOLI, O., JACKSON, S. & VANN, J., 1999a.. Groote Eylandt Manganese OrebodyStudy - November
1999 Resources - Volume 2: Kriging neighbourhood and estimation variance study. SRK Consulting Report,ProjectCode
MPG01, 91p.

BORC_PAPER_FINAL

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to BORC_PAPER_FINAL

Similar to BORC_PAPER_FINAL (20)

BORC_PAPER_FINAL