|
·
ESTIMATION TECHNIQUES
·
Estimation combined several techniques:
o Mineralised layer structures - gridded DTM
surfaces.
o Grades - "un-folding" block modelling for grade interpolation
and regular block modelling for reporting.
·
Mineralised layer structure surface
modelling:
o Software: Modelling and estimation was done in Minex
Genesis software.
o Deposit type: Interpreted as a strongly "layered" VMS
deposit.
o Estimation methods:
§ Geological
structural layer modelling employed computerised gridded
DTM surface interpolation.
§ Interpolated layers were built into an "un-folding" block
model to guide block grade interpolation.
o Appropriateness of estimation methods:
§ Combination of surface interpolation followed by "un-folding"
grade estimation along the layers considered best suited to deposit
type.
§  The
surface DTM method's appropriateness stems from its 3D
computational capability and rigor when applied to thin "layer"
deposits where manual interpretation between relatively widely
spaced drill hole data points cannot match interpolation in
3D.
§ Gridded
surfaces allow simple mathematical operations within and between
surfaces, including preventing over-laps.
§ Bounding
layer surfaces were interpolated from down-hole drill hole layer
intercepts for each layer to give hanging wall (structure roof, SR)
and foot wall (structure floor, SF) boundary surfaces.
§ "Un-folding" block model trending grade estimation as
continuity in the plane of the layers considered to most closely
achieve the strong observed grade trends within and along the
layers.
o Roof/floor/thickness: Here the layer surfaces
computation order was:
§ Overall 23
individual layers were interpreted in drill holes, of which 18 had
enough data to model. Layer list adjacent:
§ Floor
surfaces were interpolated from the layer lower intercept
depths.
§ Thicknesses were interpolated from the down-hole lengths of
the layer intercepts.
§ Roof
surfaces were computed by addition of the thickness surfaces to the
floor surfaces.
o Algorithm:
§  Floor
surface modelling used a trending growth
algorithm to interpolate smooth natural surfaces (as opposed
to straight line methods) as a regular fine mesh. Through
extrapolation this method honours local inflections away from the
reference plane mean orientation. Mesh point interpolations
grow out from data points until all mesh points are
estimated.
§ A long
default scan distance of 1,000 m was used to produce a smooth
regional surface for the method.
o  Algorithm:
§ Thickness
surface modelling used an inverse distance squared
algorithm to keep interpolated values within data
limits (preventing extrapolation).
§ A long
scan distance of 500 m was used to avoid holes in surfaces caused
where distances were great between drill holes.
§  At
the edges surfaces were only expanded 80 m beyond peripheral drill
holes.
o Coordinates and grid mesh size:
§  Surfaces
were interpolated within a coordinate area encompassing all drill
holes with interpreted layer intercepts.
§ The
adjacent figure shows a number of layers in plan view, illustrating
their slightly differing locations.
§ The DTM
mesh point dimensions of all surface gridded DTMs were 10*10 m. This was considered fine enough to
produce smooth surfaces honouring layer intercepts well but not be
greatly finer than the ~50*100-200 m drill hole spacing.
o Orientation:
§ All layer
surfaces were effectively sub-parallel and dipping ~45-50° to the east.
§ The
adjacent figure looks horizontally northwards at a selection of
layer floors.
§ Modelling
could have used an inclined reference plane dipping parallel to the
layers.
§ However to
avoid rotational complications all layers were modelled with
respect to an (assumed) horizontal reference plane at 0
RL.
o Boundary:
§ No
limiting boundary polygons were used.
§ Surfaces
were interpolated to 80 m outside
peripheral drill hole intercepts by layer.
§ This
distance was less than the typical drill hole spacing
(~50-100*100-200 m)
o Stratigraphic model build: After independent
interpolation of each layer's roof and floor the suite of surfaces
was 'built' into a valid model using processes to correct potential
cross-overs between and within lodes. This process resulted
in near zero loss.
o Surface naming:
§ File: Molaoi_20240508.GRD
§ Bounding
surfaces: Layer name + suffix SR and SF.
§ Thickness
surfaces: Layer name + suffix ST.
· Data
population domains:
o Samples and blocks (see below) in layers were uniquely
identified and segregated by domain number for assay analysis and
block grade estimation.
o Domains were set in the drill hole database and in the block
models.
o Domain numbers given above with the layer names (see layer
Table above).
· Drill
hole sample analysis:
o Base
metals (zinc, lead and silver) were the focus of the Project - and
mineralised drill hole intercepts were interpreted as layers based
on grade.
o Brief analysis was performed for the principal base metal zinc
in the layers with most intercepts.
o Brief interpretations showed the
mineralised intercepts to be sharply anomalous against waste
between layers.
· Grade
continuity control 'un-folding' block model:
o  An
'un-folding' 3D block model (MOL1_D/Z.GR3)
was built within the geological layer surface models to provide
domain control within layers and to control grade trending
continuity within and along the layers (the 'Z' direction in a
Minex 'Z-grid' block model). Dimensions tabulated
adjacent:
o Rotation: As the layers were essentially in an
~45°E dipping plane the Z-grid required no
rotating to have its Z axis normal to that plane (see
below).
o  Extent:
The un-folding block model was given a smaller and tighter plan
extent than the default used for interpolating the layers.
The extent covered all 18 layers being modelled.
o Block size: XY block size set at
10 * 40 m to be roughly 25% of the minimum drill hole spacing of ~50 * 1-200 m (some holes 100 * 200 m).
o Block Z size and number:
§ Block
numbers set according to layer average vertical
thicknesses.
§ Aim to
approximate Z block size 0.5 m.
§ Each
mineralised layer given unique domain number.
§ Parameters
tabulated adjacent:
· Grade
block estimation:
o 3D block grades were estimated into
individual grade block models for each element. The block grade
models had the same coordinate parameters as the un-folding model
(see above).
o  Estimation
parameters tabulated:
o Continuity: Data search directions within the layers
were controlled by the un-folding block model, and layer data was
segregated by domain number. A vertical (Z) distance
weighting of 1.5 was used to enhance
continuity in the plane (XY) of the layering.
o Compositing: Down-hole drill hole sample compositing
done to 0.5 m + residuals +50%.
o Algorithm: Inverse distance
squared (ID2) done in a single pass. Interpolation of
grades in two passes (to overcome issues of very localised highly
anomalous grades) was considered but not undertaken because of the
limited numbers of high grade samples in particular. In a 2
pass estimation an initial 1st pass uses all samples whilst a 2nd
pass uses only high grade samples with severely restricted scan
distances to over-write blocks close to the high grades.
o Scan
distance:
§ A scan of
250 m was used to ensure grades were
estimated in all blocks.
§ Distance
was ~25% longer than generally longest N/S distances between drill
holes.
§ In
practice the boundary limit around the layer surfaces (and hence
block model) limited actual scans to <80 m.
o Data
limits:
§ No lower
cut or clip limits were applied or required as the layer intercept
interpretation process had effectively applied a lower 0.2 % zinc
cut-off already.
§  No
upper cut of clip was applied because of 1) the limited number of
anomalous high grades, 2) their short intervals, and 3) the
positive desire to allow the few high grades to register higher
grades in some blocks.
o Estimation stats tabulated:
· Grade
reporting block model:
o  A
normal "orthogonal-shaped" block model (MOL1.G3*, simply called a
block model or a block database) was built from the un-folding
block model. Parameters tabulated adjacent:
o Block sizes/numbers to cover the same coordinates as the
un-folding block model.
o Primary block sizes were based on the un-folding block model
for X and Y. Z assigned a fixed 10 m vertically to be the
same as X for the ~45° east dipping layers. So primary blocks
10 * 40 * 10 m.
o Primary blocks sub-blocked 5 * 2 * 5 to increase resolution
across strike along layer surfaces. So minimum sub-block
potentially 2 * 20 * 2 m.
o Block grades were loaded from the individual grade block
models (see above).
o Other variables, such as grade totals and JORC classification
variables, were computed using SQL macros.
· Grade
block manipulation:
o  Zinc
equivalent (ZnEq) calculated with an SQL from zinc, lead and
silver
o Elements factorised on 3 month average metals prices to
25th July 2024 and historical Molaoi metallurgical
recoveries. Computations tabulated:
· Check
estimates:
o Resource estimate could be partially checked (for 1 layer and
minor parts of 3 others) against historic estimates (non-JORC) made
by IGME in the 1980s during the original exploration; and against
Rockfire's 2022 (JORC estimate); and against SRK's 2023 (non-JORC)
reconciliation of Rockfire's estimate.
o Those estimates of historic layer West Zone
B could approximately be checked against this layer
L3 as they covered similar areas and the
new recent drilling by Hellenic did not increase the area but
represented internal in-fill drilling simply supplying more
grades. The comparison also appears valid as densities were
similar and Rockfire's ZnEq calculation was similar to here as
parameters had not changed dramatically.  Comparisons
are tabulated adjacent:
o Rockfire's comparison to IGME's estimate had 15% more tonnes
at 15% less zinc grade, resulting in virtually the same contained
zinc - a close comparison.
o Similary SRK's comparison to Rockfire had 14% more tonnes at a
similarly lower zinc grade, resulting in contained zinc being 6%
less - also a close comparison.
o GeoRes's comparison to Rockfire and SRK had 13% more tonnes
then Rockfire (and the same as SRK) at slightly lower zinc grades
than both, resulting in 14% less contained zinc then Rockfire and
9% less than SRK.
o GeoRes considers that its L3 estimate reconciles well with Rockfire's and very well with SRK's.
o All
other 17 layers interpreted here were not estimated in the pas and
cannot be checked.
· By-product recovery:
o Elements other than base metals and germanium were effectively
not considered in this Resource estimate, hence most potential
by-products were not considered.
o However germanium has recently been
assayed for and is a common by-product of some zinc
deposits.
o GeoRes provides a quantity estimate for germanium, and
considers that the significant tonnage and reasonable grade would
position germanium as by-product if it proves amenable to
extraction from residues of zinc beneficiation.
· Deleterious elements:
o No
deleterious elements have been considered or are known
of
· Block
size - sample size relationship:
o Situation:
o Block sizes: Major block sizes were moderate at 10 * 40
* 10 m - and not considered too small for the typical data
spacing. Sub-blocking (5*2*5) reduces those sizes along the
edges of layers in a proportionate way.
o Sample spacing:
o Down-hole sampling was typically of the order of 1
m.
o Drill N/S section spacing was typically 100 m, sporadically
200 m.
o Hole
E/W spacing on section was ~50 to 100 m.
o Data
search distances: Maximum ~250 m.
o Distance relationships:
o Plan
block sizes were considered well-proportioned to drill hole spacing
(20-25% in X and Y).
o Vertical block sizes were considered very well-proportioned to
down-hole sampling intervals (100% bigger in Z).
· Model
- SMU relationship:
o No
specific focus on selective mining units (SMU) occurred.
o However The primary 10*40*10 m block
size, with potential sub-blocking to 2*20*2 m, would be similar in
size to an underground mining SMU - given that the Consultant
considers that mining by underground would be probable.
o IGME's underground mining was done in drives of those
dimensions.
· Correlation between variables:
o No
work on variable correlation was done.
o However it was clear that the base metals were typically
closely corelated, a feature used extensively through the
mineralised layer intercept interpretation.
· Geological interpretation control of estimate:
o Previously described in detail - mineralised intercept
interpretation layer by layer in each drill hole sub-parallel to
volcanic layering.
o In
summary - the block grade estimates were fundamentally controlled by the geological
interpretation of strong layered sub-parallel mineralization.
Mineralised layers were specifically modelled to match layer shapes
and grades estimated in them were confined by domain control and by
the use of 'un-folding' modelling to emphasise layer continuity
along them.
· Grade
cutting/capping use:
o No grade cutting of clipping was
used.
o Justification for this was
o Layer interpretations had effectively already clipped out low
grades (below 300 ppm TREO).
o Highly anomalous grades were relatively uncommon and where
they existed the Consultant considered that they should be
incorporated to realistically allow the known high grade shutes to
be represented. The fact that REEs consist of 15 individual
elements, each individually estimated here before being combined
into totals meant that high values in any one of the elements had
limited impact overall.
o The
Consultant considers that individual anomalously high grades could
potentially be clipped in future estimation, after consideration
hole-by-hole, if they were found to be completely
isolated.
· Estimate validation:
o Block geology validation:
§ Volume
report: Initial check to compare volumes reported within
geological layer model surfaces with volumes reported from the
blocks built from them. Expect almost exact match.
Checks all considered acceptable.
§ Plots: Visual cross-sectional plot comparison of block
boundaries with geological model surface intersections.
Particular focus on validity of the blocks in each layer (possibly
corrupt if the raw surfaces overlapped). Also check of block
domain assignments. Comparisons considered good.
o Block grade estimate validation:
§ Estimate
stats: Initial basic check to compare overall (not on a
lode/domain basis) stats given during the block estimation - input
drill sample stats with output estimated grade stats. Expect
reasonable but not exact match. Particular focus on closeness
of the maximums and the raw averages. Results considered
acceptable.
§ Plots: Methodical visual cross-sectional plot comparison
of colour-coded block grades with annotated drill hole
samples. Comparisons considered acceptable.
·
Estimate reconciliation:
o Estimate reconciliation: Described above under "Check
estimates" for the L3 layer. Not possible for other layers as
they were not previously interpreted.
o The
Consultant's overall view here was that the past 2022 Resource
estimate by Rockfire was completely valid in itself (and as
confirmed by SRK's 2023 review) but only represented a very small
proportion of this Resource.
|
