Analysing high resolution consistent data for structural geology and geotechnical studies

As a geoscientist, gaining access to such information can provide an opportunity to re-asses the understanding of mine scale structural geology and mineral system, with significant implications for assessing geotechnical risk. This session showcases some novel avenues of investigation relevant to the often-neglected overlap between structural geologists and geotechnical engineers.

Overview

[00:00:00.125]
(bright electronic music)

[00:00:10.151]
Hi everyone.

[00:00:10.984]
I am Anthony Reed, a Structural Geologist

[00:00:13.951]
working with Mining One,

[00:00:16.220]
and today I’m presenting analysis of high resolution

[00:00:20.270]
and consistent data for structural geology

[00:00:22.940]
and geotechnical studies.

[00:00:25.777]
I have to put your standard disclaimer in there

[00:00:27.630]
not to make any critical business decisions

[00:00:29.750]
based on this presentation alone.

[00:00:31.090]
But do reach out to me if you’d like

[00:00:34.610]
a similar treatment of your data.

[00:00:38.220]
So a little bit about Mining One,

[00:00:39.257]
we’re one of the largest mining consultancies in the world

[00:00:43.180]
at the moment, and we service, primarily,

[00:00:46.790]
mining companies of course,

[00:00:47.980]
but actually provide services to a number of other

[00:00:52.450]
sectors that are related to mining as well.

[00:00:55.580]
We have offices all over the world

[00:00:58.410]
and we operate in all those orange countries

[00:01:00.610]
that you can see on the screen there.

[00:01:04.490]
Got most of the big players in the mining industry,

[00:01:06.560]
as clients already, but always happy to add some more.

[00:01:10.540]
And we cover everything in the mining space,

[00:01:12.740]
from conceptual stages through to the implementation

[00:01:15.570]
of underground and open cut pits

[00:01:17.660]
all the way through to mine closure.

[00:01:19.110]
And you can see from that list,

[00:01:21.040]
generally everything that you need to run a mine,

[00:01:23.120]
we’re involved with.

[00:01:25.240]
Honing in on the geology space,

[00:01:26.970]
we do cover all stages of geology from early exploration,

[00:01:31.490]
through geoscientific modeling,

[00:01:33.200]
mineral resource modeling and reserve estimation,

[00:01:36.450]
independent valuations for internal or external valuations.

[00:01:42.770]
We can design drill programs,

[00:01:44.540]
and we have competent people qualified

[00:01:46.000]
under JORC and the NI-43-101 codes to sign off.

[00:01:50.200]
We conduct a QA/QC and have independent experts

[00:01:52.950]
qualified under VALMIN.

[00:01:57.260]
In the geotechnical space,

[00:01:58.500]
we cover pretty much everything to do with mine stability,

[00:02:01.550]
so very large geotechnical team.

[00:02:04.580]
And while there’s a large list

[00:02:06.730]
that I’m not going to go through,

[00:02:08.070]
I’d like to focus on down the bottom,

[00:02:10.250]
our partnership with a software package called Cavroc,

[00:02:13.300]
which is essentially a front end plugin

[00:02:16.950]
for the FLAC3D geotechnical modeling software,

[00:02:19.770]
which is mostly command line based.

[00:02:22.120]
It consists of an Octree Mesher

[00:02:25.410]
which brings geotechnical data into the block model space,

[00:02:29.300]
much as you would a resource modeling technique.

[00:02:35.720]
It’s got our IUCM,

[00:02:36.990]
which is a physics engine that uses that data

[00:02:39.170]
to model rock performance.

[00:02:41.060]
And then we’ve got two additional plug-ins,

[00:02:43.469]
StopeX and SlopeX, so one’s open pit

[00:02:46.127]
and one’s for underground,

[00:02:47.610]
which are GUI’s specifically designed for FLAC3D

[00:02:50.600]
to make the entire process far more accessible

[00:02:54.040]
to people on sites,

[00:02:57.010]
and it turns into a very agile and iterative process.

[00:02:59.740]
So that’s the sort of thing

[00:03:00.573]
that we would roll out for mines.

[00:03:04.370]
Now, why do I care about Cavroc?

[00:03:06.790]
So I’m a structural geologist and I’ve been working very

[00:03:09.500]
heavily with geotechnical engineers

[00:03:12.070]
for over a year at Mining One.

[00:03:14.270]
And what a surprise,

[00:03:15.590]
we have significant overlap in our spaces.

[00:03:19.780]
In the rest of the industry that I’ve been involved with,

[00:03:22.190]
structural geologists and geotechnical engineers

[00:03:25.520]
are usually kept at arm’s length,

[00:03:27.330]
but working together we’ve both had our eyes opened

[00:03:29.990]
into each other’s fields.

[00:03:32.260]
So the sort of data that’s particularly useful

[00:03:34.490]
for geotechnical modeling, that I also care about,

[00:03:38.243]
are discrete faults,

[00:03:39.900]
diffuse faults and zones of weakness, alteration zones

[00:03:42.930]
and halos, lithological volumes and boundaries,

[00:03:45.930]
which all geologists are familiar with,

[00:03:47.930]
and any anisotropy that’s in the rock mass, i.e., foliation.

[00:03:54.260]
Back to the title screen,

[00:03:55.430]
what do I mean by high resolution, consistent data sets?

[00:03:58.280]
Well high resolution sort of self explains.

[00:04:01.770]
Consistent, I mean a data set that is collected

[00:04:04.370]
with a well understood method

[00:04:06.530]
that’s very well leveled and very repeatable.

[00:04:09.110]
So things that I would normally consider that fit that bill

[00:04:12.480]
in the geology space are assays from Diamond Drill

[00:04:15.750]
and RC Core for exploration of resource,

[00:04:18.900]
hopefully that data’s in good quality

[00:04:21.030]
if you’ve got an operation.

[00:04:22.850]
More in the geotechnical and mining space,

[00:04:25.660]
we have huge data sets such as blast hole assays,

[00:04:28.600]
which geologists don’t usually get to see,

[00:04:30.710]
geotechnical parameters get logged, such as RQD,

[00:04:33.670]
machine learning derived data, such as color or clustering

[00:04:37.330]
on core or rock chips.

[00:04:39.850]
You can have observations directly from modeled geometry

[00:04:43.050]
of other objects,

[00:04:44.859]
so things like implicitly modeled orebodies

[00:04:47.530]
or other geology surfaces, and of course, photogrammetry,

[00:04:50.760]
which I’m sure everyone is into now,

[00:04:53.230]
and laser scanning for things such as stope pickups.

[00:04:56.270]
They’re all good examples of very high quality datasets.

[00:05:01.370]
I’ll start with the geotechnical data

[00:05:02.727]
and what that looks like.

[00:05:04.527]
So RQD is not something that I’ve played with

[00:05:06.500]
a hell of a lot in the past, as a geologist,

[00:05:08.880]
doing geological demanding,

[00:05:10.860]
but it’s a very consistent data set that remains consistent

[00:05:13.860]
from operation to operation.

[00:05:15.360]
It’s a fairly simple calculation.

[00:05:17.340]
Even when it’s logged manually,

[00:05:19.740]
you do still have to consider that there’s a human factor,

[00:05:22.060]
and occasionally you’ll get some leveling issues

[00:05:24.420]
with different people logging during different shifts.

[00:05:28.030]
But mainly the only problems with the data

[00:05:29.900]
is things like the first five meters of holes

[00:05:32.640]
sometimes have to be excluded

[00:05:33.880]
because you are hitting a rock with a large machine

[00:05:36.670]
and it doesn’t always reflect actual rock quality.

[00:05:40.810]
Things like drilling artifacts can be stuck in the database,

[00:05:43.180]
like Navidril or Wedges that need to be taken out.

[00:05:46.257]
Cover sequences can swamp any deeper weaker signals

[00:05:49.180]
that you might want to consider.

[00:05:51.560]
And occasionally you might want to throw out entire

[00:05:53.580]
campaigns, like the exploration call from the surface.

[00:05:57.480]
It could be something to do with drilling technique

[00:05:59.440]
rather than logging, but you know,

[00:06:01.570]
sometimes you want to make sure

[00:06:02.870]
you have a really consistent data set,

[00:06:04.300]
so that has to be done sometimes.

[00:06:07.450]
And what are we looking for?

[00:06:08.510]
Well the image on the left-hand side

[00:06:11.450]
is a really nice planar fault.

[00:06:13.980]
You’ve got a lovely, bright red signal there,

[00:06:16.220]
and that’s something that can be digitized and wireframed,

[00:06:18.470]
and you’ve got a brilliant fault

[00:06:20.600]
that can head through to your geotechnical modeling.

[00:06:23.410]
And what we would do with this data set is zoom around

[00:06:27.140]
and rotate it until we can find all of those

[00:06:29.870]
very clear planar features and model them all up

[00:06:32.470]
in a fault network.

[00:06:35.210]
Once you’ve done that,

[00:06:36.080]
that data can then be excluded and you can re-level

[00:06:38.330]
the remaining data set to look for weaker signals.

[00:06:40.660]
Now on the right, you can see from the upper left

[00:06:44.470]
to the lower right of that image with all the drill holes,

[00:06:48.170]
some fairly strong striations in that data set.

[00:06:51.720]
Remember this is RQD data,

[00:06:53.120]
so that is showing a rock weakness

[00:06:55.180]
that is lighting up in a plane.

[00:06:58.040]
It turns out that they are, or one of them at least,

[00:07:00.680]
is in the same position as a logged slate horizon

[00:07:04.570]
in this deposit.

[00:07:06.300]
However, you can see there’s a number of striations

[00:07:08.360]
and we really only have a single slate horizon logged,

[00:07:10.640]
so something else is going on.

[00:07:12.920]
So in this case, we did some further investigation

[00:07:16.110]
and found just enough evidence in the rest of the logging,

[00:07:21.070]
looking at a whole myriad of codes

[00:07:22.910]
that could be related to shearing and slates.

[00:07:25.453]
But yes, we do indeed have a lot more slate horizons

[00:07:29.750]
that are heavily sheared in this deposit.

[00:07:31.450]
And so we’ve used the geotechnical data

[00:07:34.000]
and interpreted that as a lithology log

[00:07:36.870]
to build a whole stack of new features.

[00:07:40.820]
In addition to that,

[00:07:42.570]
we found these ramping structures that run between them,

[00:07:45.750]
and after a lot of discussions with the site geologists,

[00:07:49.200]
as well as looking at core photos,

[00:07:51.500]
the entire set of slates has kind of been reclassified

[00:07:54.730]
as a set of a shear zone network,

[00:07:59.670]
where pretty much all of the things that are modeled

[00:08:02.520]
as slates and these ramping structures between them

[00:08:05.490]
are some sort of fault,

[00:08:06.880]
so they’re very, they’re geotechnically important.

[00:08:09.930]
They’re very, very thin,

[00:08:12.880]
but we finally have a way to visualize them and model them.

[00:08:16.260]
Now from a geotechnical modeling’s perspective,

[00:08:20.290]
they’re so thin that in the block model space

[00:08:22.970]
they don’t take up enough volume to even assign blocks

[00:08:28.710]
with their own sort of rock parameters.

[00:08:30.850]
So instead, in this case, we would have treated them

[00:08:33.880]
as discrete faults and fed that through

[00:08:37.090]
into the Cavroc plugin for numerical modeling.

[00:08:42.060]
Blast hole assays are a absolutely brilliant data set,

[00:08:45.570]
and as a geologist, I’ve only recently started asking

[00:08:49.580]
to look at these.

[00:08:51.920]
But while they usually lack in a breadth of assay types,

[00:08:58.770]
they make up for it in the sheer amount of data

[00:09:00.950]
that you get.

[00:09:01.783]
So what we’re looking at here is iron assays

[00:09:04.880]
up high on the pit wall of a deposit.

[00:09:07.980]
Well and truly outside of your normal resource drilling.

[00:09:10.700]
So this is essentially a blank slate up here

[00:09:13.530]
until we get the blast hall assay data.

[00:09:15.970]
In this case, it’s been deemed of critical geotechnical

[00:09:20.030]
value to understand where the dikes are in this deposit.

[00:09:23.100]
And a lot of them haven’t been picked out from the previous

[00:09:25.330]
drilling or mapping efforts because they’re difficult to see

[00:09:27.530]
on the ground, and we don’t fully understand the geometry

[00:09:30.590]
from these sparse resource drill outs.

[00:09:33.260]
But as you can see from the blast holes,

[00:09:34.870]
they just shine.

[00:09:37.370]
Blast holes at any deposit really give you the best insight

[00:09:40.880]
into the shape of your deposit.

[00:09:42.820]
And a lot of how their mineral style is actually working

[00:09:47.380]
for you there as well.

[00:09:52.150]
With some aid of photogrammetry,

[00:09:54.190]
’cause I had it in this case,

[00:09:55.600]
but mainly tracing those features

[00:09:57.220]
we were able to build that into a dike network,

[00:09:58.740]
that again, gets fed into your geotechnical model.

[00:10:03.220]
Now in the previous images we had iron highs,

[00:10:08.000]
but this time we’ve got an economic mineral

[00:10:10.960]
and you can see the dikes are taking that material out,

[00:10:14.850]
and so you’d be tracing the lows in this case.

[00:10:18.260]
And frequently the same feature can track from a signal

[00:10:23.460]
of being anomalously high to anomalously low,

[00:10:26.150]
as it tracks through an orebody.

[00:10:28.030]
Now from sparse resource drilling,

[00:10:30.560]
from one drill hole to another,

[00:10:31.890]
the rock can look completely different and you cannot link

[00:10:34.210]
those features, but in blast holes,

[00:10:36.400]
the drill spacing is so close and the data

[00:10:38.300]
is such high quality that you can see

[00:10:39.880]
that it’s actually the same feature tracking

[00:10:41.600]
all the way through.

[00:10:42.740]
So using this sort of data to reconcile your geology model

[00:10:45.900]
and your understanding of your orebody

[00:10:47.350]
is really quite an amazing opportunity.

[00:10:52.970]
CMS stope scans, so once we have a stope,

[00:10:57.510]
we often send in the CMS to laser scan that stope.

[00:11:01.680]
And in many cases, especially where we have overbreak,

[00:11:05.590]
everyone runs around

[00:11:08.780]
hoping that they can get more information

[00:11:10.630]
about why a stope overbroke.

[00:11:13.280]
So in this case, you can see from upper left

[00:11:17.300]
to lower right, there are a series of structures

[00:11:19.570]
that look like they’re involved in some overbreak

[00:11:23.400]
in this image.

[00:11:24.630]
You’ve got large blocks breaking back

[00:11:26.440]
to a very planar feature.

[00:11:28.510]
Now those were joints or faults that were not picked up

[00:11:31.360]
in the logging, not picked up in the mapping,

[00:11:33.620]
but clearly quite important in this area.

[00:11:35.910]
And so from analysis of this series of CMS scans,

[00:11:40.280]
we’ve identified in the yellow ovals

[00:11:44.220]
areas that we deemed risky,

[00:11:45.910]
even though they’re adjacent to a fallen stope anyway,

[00:11:49.060]
so they would have been thought of as risky,

[00:11:50.640]
but we’ve decided there’s a mechanism there.

[00:11:54.050]
Lo and behold, two months later,

[00:11:55.900]
we’ve got the rest of the data,

[00:11:57.070]
and yes, those zones were of risk,

[00:12:00.000]
they overbroke in the same way,

[00:12:01.720]
and there’s a whole lot more data to suggest that those

[00:12:05.830]
faults were involved in that area as well.

[00:12:09.470]
So, you know, the structures that are important

[00:12:12.570]
to the stope scale, are best visualized in the stope itself.

[00:12:18.700]
So it’s worth your geologists getting in there,

[00:12:20.500]
looking at the scans of the stopes

[00:12:23.560]
after they’ve actually been mined out.

[00:12:26.200]
Often, not an accessible dataset to a geology department.

[00:12:33.300]
So mapping on high resolution photogrammetry,

[00:12:35.670]
everyone’s on the photogrammetry bandwagon,

[00:12:38.320]
but I’m going to look at it again anyway,

[00:12:40.340]
just to show you what we do with it.

[00:12:42.550]
So what we’ve got is a couple of really obvious structures

[00:12:46.840]
sitting in this image.

[00:12:48.530]
You know, we’ve got one on the left-hand side

[00:12:50.370]
that’s colored by our dip direction on the right-hand side,

[00:12:53.400]
it’s just the photographic image.

[00:12:55.490]
And there’s a number of ways to model such a feature

[00:12:58.090]
for use in a geotechnical model.

[00:13:01.450]
So for instance,

[00:13:03.470]
we can see features where blocks

[00:13:06.420]
have fallen out of a pit wall

[00:13:07.780]
and literally just trace it up multiple benches.

[00:13:10.410]
And we can do that with a series of structural disks,

[00:13:13.930]
if we can really pick up the orientation well.

[00:13:17.210]
And if we can’t pick up the orientation,

[00:13:19.240]
then we can use striations,

[00:13:20.740]
so things like polylines or points to tie it together,

[00:13:23.010]
and often there’s a combination of the two.

[00:13:26.320]
Part of the problem that I have

[00:13:27.490]
with some of the more automated data sets that you can get

[00:13:29.850]
from Surrovision or ADAM tech is that the photogrammetry

[00:13:35.030]
meshes themselves, do have some inaccuracies.

[00:13:37.790]
And so if you’re trying to perform structural geology

[00:13:40.970]
on this data, you can’t always rely on an automated pickup

[00:13:44.760]
with flat faces.

[00:13:45.750]
You do need to have your geological brain on.

[00:13:49.240]
Does this make sense?

[00:13:50.260]
Is it a planar feature?

[00:13:51.430]
Does it bend while you’re doing the interpretation?

[00:13:54.510]
And from all that interp, we can make a structural network,

[00:13:56.630]
as you can see on the right-hand side there.

[00:14:00.570]
So just a little bit of a case study of what to do

[00:14:03.840]
with some really complex structural environments.

[00:14:07.300]
So this is, we had some failures in a pit,

[00:14:10.750]
and it was deemed very

[00:14:12.260]
important to get a structural geologist out there,

[00:14:15.250]
to do some mapping and find out what we could.

[00:14:19.820]
Only 12 out of the 40 days he was on site

[00:14:22.780]
was he able to even access the pit.

[00:14:24.230]
There was logistical issues with even getting in there.

[00:14:26.930]
The maximum approximate density that you can collect

[00:14:29.440]
when you’re walking around the pit is about one measurement

[00:14:33.260]
every 20 meters, realistically,

[00:14:35.710]
just in terms of time constraints.

[00:14:38.600]
There are heaps of inaccessible areas that we really needed,

[00:14:41.290]
of course, the ones close to various failures

[00:14:45.020]
are the ones that we didn’t get data for.

[00:14:47.750]
The positional accuracy of the GPS wasn’t particularly good,

[00:14:50.700]
and the relevance of individual measurements is questionable

[00:14:54.710]
because you really can’t tell a global context

[00:14:57.650]
when you’re walking out.

[00:14:59.720]
So in order to augment this study,

[00:15:02.250]
we decided to do a drone survey

[00:15:03.630]
and we have covered the entire pit in this,

[00:15:05.470]
but this is one little example of one little corner.

[00:15:08.460]
Some ground control points, just using a consumer drone,

[00:15:11.700]
a simple inclined grid planned with a back and forth

[00:15:15.630]
flight pattern.

[00:15:17.230]
And we processed fit for purpose, high detail products

[00:15:20.900]
within a day using Agisoft Metashape.

[00:15:25.160]
And this is what they look like.

[00:15:27.120]
So one of the joys of doing it all in house

[00:15:29.940]
is that we can tweak the final products

[00:15:32.750]
to be exactly what we need to pick up the features

[00:15:36.420]
that we want.

[00:15:37.810]
And if it’s not quite right,

[00:15:38.910]
we can dial up that detail and reprocess it again.

[00:15:41.420]
Now, if you’re dealing with a open and shut

[00:15:46.710]
request to provide photogrammetry data

[00:15:48.810]
from someone that’s not involved in the structural or

[00:15:51.840]
geotechnical model, often the product that is returned

[00:15:55.880]
is not always appropriate to do that level of mapping.

[00:15:59.520]
The outputs, we configure them to be compatible

[00:16:02.440]
with all the major stakeholders on the site as well,

[00:16:04.280]
so everyone was able to have a look at this data.

[00:16:07.100]
In this particular case that we’re seeing on the screen

[00:16:08.940]
is this 30-meter-high slip along bedding.

[00:16:11.470]
There’s no physical access to this location.

[00:16:13.940]
There’s rubble, there’s overhangs,

[00:16:15.430]
and even if you were able to walk up to it on foot,

[00:16:17.650]
you’re unlikely to get a representative measurement.

[00:16:19.500]
As you can see, that slip surface is bending

[00:16:22.290]
quite considerably, even just this plane of view.

[00:16:26.590]
So instead took the photogrammetry output into a package

[00:16:29.760]
called Cloud Compare and in a combination of picking faces

[00:16:34.580]
in Cloud Compare and manual digitizing in Leapfrog,

[00:16:37.700]
we’ve got a decent dataset together of bedding and joints,

[00:16:42.970]
all digitized on the computer screen.

[00:16:46.100]
Sighting down measurements for Leapfrog digitization,

[00:16:50.360]
which is often a little bit better when you’ve got,

[00:16:53.020]
you know, rough edges, but you can still see a good feature.

[00:16:56.550]
And the context is decided during the digitizing process.

[00:16:59.530]
So I knew I wanted to make a structural model,

[00:17:01.390]
so I digitized with that context in mind,

[00:17:03.310]
rather than just ending up

[00:17:04.420]
with a cloud of unlabeled measurements.

[00:17:09.870]
It took only about two hours to do this entire area.

[00:17:12.370]
So a similar number of measurements

[00:17:14.920]
to our structural geologist friend

[00:17:17.300]
that was there for, that was in the pit for 12 days.

[00:17:20.490]
It’s not automatic or replacing a skilled geologist,

[00:17:22.730]
it requires a skilled geologists to do this work.

[00:17:26.210]
We haven’t quite managed to automate that out of it yet.

[00:17:29.960]
We picked up all the ethanol and the joints of any visible

[00:17:32.630]
outcrop, and it’s easy to refine targets

[00:17:35.240]
for walk-up in the pit, if we’re not sure about something.

[00:17:39.800]
More to the story, three years prior,

[00:17:42.120]
there were some pseudo bedding models created.

[00:17:45.900]
And it certainly looks like from the old measurements

[00:17:49.080]
that there’s a scoop geometry that aligns pretty well

[00:17:51.360]
with where the failure was in the pit.

[00:17:53.920]
So the thought was, well we think we understand

[00:17:57.630]
that there is a failure risk along bedding.

[00:18:00.870]
Can we highlight that risk in great detail

[00:18:03.110]
on the photogrammetry?

[00:18:06.510]
Often in a pit study, I’ll be asked,

[00:18:10.910]
can I please have a dip and dip direction

[00:18:13.440]
of the features that we’re interested in?

[00:18:15.210]
And as you can see from that image there,

[00:18:16.920]
the feature that we’re interested in,

[00:18:18.560]
doesn’t have a single dip and dip direction,

[00:18:21.190]
it meanders all over the place,

[00:18:22.510]
so the answer would be, no,

[00:18:24.150]
we need to find some way a little bit cleverer to do that.

[00:18:29.330]
And hopefully we did.

[00:18:31.300]
We interpolated the unit vectors from the raw bedding data

[00:18:35.230]
that was taken from the photogrammetry and the previous

[00:18:37.820]
studies and mapped them directly

[00:18:40.960]
onto the photogrammetry data set.

[00:18:43.770]
So we’ve found the angular difference calculated

[00:18:48.170]
between the rock face from a photogrammetry

[00:18:50.800]
and the expected bedding,

[00:18:52.560]
and there’s a couple of little issues with needing,

[00:18:56.880]
we need trig functions in Leapfrog

[00:18:58.690]
to make this a bit more accurate,

[00:19:00.280]
and if we can get access to the,

[00:19:02.950]
the form interpolant algorithm

[00:19:05.310]
to sort of directly map values onto points

[00:19:08.570]
that would make this a lot more accurate as well.

[00:19:10.350]
But if you’re close to data,

[00:19:11.760]
this method works pretty well.

[00:19:15.270]
And this is the output.

[00:19:16.210]
So what we’re looking at here is that same survey area,

[00:19:18.600]
where we have very bright yellow,

[00:19:21.260]
is very close to aligned to bedding.

[00:19:24.660]
So faces in the rock that exists during

[00:19:27.110]
a photogrammetry survey that match the bedding orientation.

[00:19:31.040]
And down to the red pixels are five degrees away from that.

[00:19:35.210]
So the brighter, the color,

[00:19:36.530]
the closer to the bedding we were.

[00:19:38.220]
And you can see from that image,

[00:19:39.500]
that in the zone of the failure, towards the left-hand side,

[00:19:42.630]
there’s a huge area of yellow, and over to the right,

[00:19:45.640]
where the bedding veers away from the exposed,

[00:19:49.310]
from the pit plan, from the bench angles,

[00:19:52.810]
you get little planes of failed rock,

[00:19:55.380]
but none of them are really huge.

[00:19:59.010]
So from that data, we turned a couple of hundred

[00:20:01.720]
measurements manually into possibly millions

[00:20:05.560]
and picked up everywhere in the rock face

[00:20:07.640]
that looks to be relevant to bedding failure.

[00:20:11.270]
And from that we can evaluate, you know,

[00:20:13.800]
risk at any given location

[00:20:15.730]
based on what we can see in the pit walls.

[00:20:18.200]
But I think it’s a good case for having

[00:20:19.360]
some augmented reality as well,

[00:20:21.227]
where you could walk around the pit and see

[00:20:24.320]
what’s in front of you.

[00:20:25.200]
What is bedding?

[00:20:26.033]
What is a joint?

[00:20:27.100]
Is it close to your expected angle,

[00:20:28.580]
or is it something completely different?

[00:20:31.840]
Okay, onwards to our implicit vein

[00:20:34.300]
model structural analysis.

[00:20:35.670]
So we’re looking at the analysis

[00:20:37.880]
of an implicitly modeled object.

[00:20:39.780]
In this case, it’s a high grade ore lens.

[00:20:42.850]
It’s a fairly planar feature,

[00:20:45.690]
but there are a number of structures that interact with it.

[00:20:48.520]
We know from looking at the core and underground,

[00:20:52.480]
that there are structures in a number of orientations

[00:20:55.420]
that are important, but their persistence,

[00:20:58.050]
the significance of them and how they interact,

[00:21:01.280]
and yeah, it’s a little bit difficult to work out.

[00:21:05.870]
I call this a consistent data set

[00:21:07.440]
because we’ve got fairly good drill hole coverage

[00:21:09.950]
for this object, about 12 1/2 meter centers

[00:21:13.820]
across the entire thing, along a strike of 2 1/2 k’s,

[00:21:17.950]
and down dip of about one kilometer.

[00:21:20.150]
And it’s modeled from assays.

[00:21:22.050]
So even though there’s some geology interp,

[00:21:26.360]
it’s still a fairly consistent shape,

[00:21:28.890]
and every contact is very similar to the next contact along

[00:21:34.030]
in terms of how it’s defined.

[00:21:37.360]
So is there a way to get a bit more value

[00:21:41.060]
out of seeing those perturbations in that shape

[00:21:44.560]
than just looking at it?

[00:21:46.330]
And the answer is yes.

[00:21:47.290]
Again, we’ve taken the vertices from that object

[00:21:50.170]
into Cloud Compare

[00:21:52.010]
and looked at the different dip direction

[00:21:54.800]
and the constituent unit vectors that are orthogonal

[00:21:57.750]
to that mesh.

[00:21:59.090]
And in the top left-hand side,

[00:22:02.230]
you can see our dZ unit factor,

[00:22:04.770]
which is actually fairly similar to dX and dip

[00:22:07.780]
in this instance,

[00:22:09.510]
because it’s a north south striking orebody.

[00:22:11.710]
And we can see those horizontal features that were visible

[00:22:15.280]
are now far more continuous.

[00:22:17.710]
And we can also see some features that are ramping up

[00:22:19.810]
towards the left as well, that are fairly clear.

[00:22:23.390]
Beneath that we’re coloring with dY,

[00:22:25.850]
which is analogous to aspect in this case,

[00:22:29.447]
a different value, but a similar pattern.

[00:22:33.900]
And so some features that are running up and down

[00:22:38.250]
the length of this are seen quite prominently

[00:22:41.900]
in the purple and the green bands next to the purple.

[00:22:45.080]
The top right-hand side, we can see thickness,

[00:22:49.040]
which is calculated from the Leapfrog vein algorithm.

[00:22:53.150]
And those horizontal features can be seen causing a extreme

[00:22:57.680]
thinning of the orebody in the same location.

[00:23:01.700]
So not as much with the ones that ramp up

[00:23:03.720]
towards the left-hand side,

[00:23:05.160]
but definitely those horizontal ones are important.

[00:23:08.440]
The vertical ones in that image

[00:23:10.830]
appear to be thickened rather than thinned as well,

[00:23:14.110]
so starting to understand what the structures are doing

[00:23:18.180]
and what their angles are,

[00:23:19.740]
how they interact with the orebody

[00:23:20.950]
and why they’re important.

[00:23:23.160]
In the lower right-hand side, they’ve all just been traced.

[00:23:26.170]
So running around with the digitizing tool

[00:23:29.060]
and picking out everything we can see

[00:23:30.800]
from any of these datasets.

[00:23:32.280]
And of course there was a large amount of mucking around

[00:23:35.500]
with color and contrast in order to really understand

[00:23:37.780]
where everything was.

[00:23:39.830]
And from that, we generated a structural network.

[00:23:42.630]
Now we didn’t necessarily have the exact angles

[00:23:46.042]
for this study.

[00:23:46.960]
They can be tied into underground observations

[00:23:49.870]
when there’s time,

[00:23:50.980]
but what we did get out of this is an understanding

[00:23:54.340]
of where families of structures interact with the orebody,

[00:23:58.150]
where they interact with each other,

[00:23:59.620]
and what would be considered a structurally complex zone

[00:24:02.120]
in this orebody versus one that’s not.

[00:24:04.250]
And that is quite an important outcome.

[00:24:06.240]
So we went from,

[00:24:08.190]
we went from an understanding

[00:24:09.290]
that there are structures everywhere,

[00:24:10.470]
and we have a risk everywhere in this orebody

[00:24:13.210]
to really nailing down some locations of importance.

[00:24:19.040]
Finally, so all of this is innate of,

[00:24:22.023]
fill in a geotechnical numeric model

[00:24:24.220]
so we get the best evaluation of rock quality,

[00:24:28.910]
rock mechanics, and risk we can in open pit or underground.

[00:24:33.200]
Now Leapfrog does have a part to play

[00:24:34.930]
in populating this block model as well,

[00:24:39.410]
because, especially with the edge module,

[00:24:41.920]
we can populate a number of things into a block model space

[00:24:45.630]
that is useful in Cavroc.

[00:24:48.700]
So things like lithological instructional domains,

[00:24:52.260]
structures themselves, structural trends,

[00:24:56.120]
and they can be individual per domain,

[00:24:58.150]
or they can be global.

[00:25:00.530]
And we can feed, yeah,

[00:25:01.830]
we can feed the trend data into Cavroc as well.

[00:25:06.230]
For the geotechnical logging, RQD can be interpolated

[00:25:12.510]
fairly nicely in Leapfrog.

[00:25:14.440]
You do have to flip it around to do 100 – RQD

[00:25:17.300]
as the RBF algorithm likes to contour around high values.

[00:25:21.250]
And so I find that visualizing RQD as it’s being

[00:25:25.270]
interpolated is an extremely powerful thing

[00:25:27.960]
that a lot of people don’t do.

[00:25:30.480]
And you can ensure that your RQD interpolation

[00:25:35.270]
is geologically reasonable before it gets sent onto

[00:25:38.980]
a geotechnical numeric model.

[00:25:45.460]
Orientation interpolation, there is actually a workflow

[00:25:49.380]
using edge to get downhole measurements

[00:25:53.060]
through into the block model space

[00:25:55.370]
that involves creating form interpolant surfaces,

[00:25:59.310]
and then using VAT to populate a variable

[00:26:02.286]
interpolation field in edge.

[00:26:05.600]
But you can see from the lower right image,

[00:26:07.260]
the final result of that is an evaluation of angle

[00:26:10.720]
from the input data at every centroid in a block,

[00:26:14.930]
which is a particularly useful thing.

[00:26:19.250]
The final geotech block model,

[00:26:21.100]
in this case from Leapfrog only,

[00:26:23.100]
I was able to populate the domain, foliation orientation,

[00:26:26.470]
foliation intensity, the RQD, distance functions to objects,

[00:26:30.410]
and distance to reliable data in the model.

[00:26:36.020]
And so while it’s not my place to evaluate risk,

[00:26:40.220]
I was able to do little calculations

[00:26:42.210]
like looking inside the orebody for poor quality rock

[00:26:44.810]
that’s highly foliated

[00:26:46.240]
that’s sitting next to some valid data and pointing out,

[00:26:50.690]
areas of particular interests, geotechnically,

[00:26:53.980]
that are worth focusing on.

[00:26:56.490]
And that block model can just be delivered directly

[00:26:58.650]
to our numeric models.

[00:27:02.360]
So some final thoughts,

[00:27:04.530]
there’s a lot of opportunities for geology

[00:27:07.610]
in the geotechnical space.

[00:27:09.960]
Geotechnical engineers often have access to a lot more,

[00:27:13.800]
really high quality data than a geology team

[00:27:16.350]
would normally see.

[00:27:18.150]
And so it’s worth the two teams working together,

[00:27:21.810]
the data of critical mass to do this level of interpretation

[00:27:24.960]
in the structural space,

[00:27:25.960]
and just in the orebody knowledge space

[00:27:29.080]
is usually when a mine is more mature

[00:27:31.070]
than when you’ve got your exploration department going

[00:27:36.470]
great guns on your deposit.

[00:27:40.200]
Interpretation and reconciliation of the geology model

[00:27:43.520]
using these massive data sets at a mine

[00:27:46.130]
is not commonly done from what I can tell.

[00:27:49.930]
Most of the data is restricted to very narrow use cases,

[00:27:53.130]
such as the blast hole databases being for mining block

[00:27:57.220]
models, rather than being seen as a opportunity

[00:28:02.680]
to really understand your orebody again.

[00:28:06.460]
And better geology always feeds into better

[00:28:09.350]
geotechnical modeling, the understanding and the products.

[00:28:12.510]
So it’s really worth having this partnership

[00:28:14.380]
between geologists and geotechnical engineers.

[00:28:18.130]
Sadly, geotechnical numeric modeling is often the only good

[00:28:22.860]
justification for getting this work done.

[00:28:26.170]
They tend to be the ones that have the budgets and the need,

[00:28:29.870]
and the link directly to production,

[00:28:32.880]
rather than a structural geologist, such as myself,

[00:28:37.010]
who would often only be asked to do this

[00:28:40.570]
as part of a more academic study.

[00:28:43.460]
So hopefully what I’ve shown is that

[00:28:47.320]
it actually does provide a level of rigor to geotechnical

[00:28:51.620]
numeric modeling to get some really high quality structural

[00:28:54.550]
geology in there,

[00:28:56.220]
and especially from these datasets that geotechnical

[00:28:58.520]
engineers get to use every day.

[00:29:01.460]
So thank you very much for your time.

[00:29:03.519]
(bright electronic music)