3D geology models in infrastructure engineering – present and future

Since Aurecon’s first adoption in 2014, Leapfrog has become routinely used to support ground engineering on many projects, particularly for linear infrastructure such as City Rail Link in Auckland.

 

Their uptake of 3D modelling reflects a company-wide emphasis on modelling and visualisation, and on digital integration and interoperability. Seequent Central has become the preferred solution to enable model access, management, and quality assurance for secure model-sharing in the procurement process. While communication of uncertainty remains a challenge and databases have been re-designed, learn how Aurecon have adopted 3D modelling as an integral part of their workflows and increasingly automated engineering design processes for tunnels, bridges and retaining structures.

 

Overview

Video Transcript

[00:00:15.040]
<v ->Thank you for the opportunity to share our 3D ground</v>

[00:00:18.050]
modeling experience with the wider sequent community.

[00:00:22.140]
With my colleagues Anahita and Sam,

[00:00:24.440]
I would like to share what we do,

[00:00:26.890]
how we use 3D ground models to manage ground risk.

[00:00:31.280]
Also share some experiences we’ve had.

[00:00:34.080]
Both good and bad experiences.

[00:00:35.890]
And, and finally talk briefly about where we’re going,

[00:00:39.900]
where we’re focusing our attention in terms of improving

[00:00:42.920]
ground risk management with a 3D model.

[00:00:50.850]
So we’ve been using sequent products,

[00:00:53.950]
geo leapfrog works for about six years

[00:00:58.510]
and mainly in the linear infrastructure project space,

[00:01:02.620]
but also on other projects as well.

[00:01:05.940]
But in linear infrastructure,

[00:01:07.160]
we tend to take one of two roles.

[00:01:10.250]
We’re either on the client side,

[00:01:12.770]
where we do the early part of the

[00:01:14.530]
project planning, concept design, reference design,

[00:01:18.760]
preparing procurement documentation,

[00:01:21.650]
helping with procurement,

[00:01:23.040]
and perhaps having a role going through,

[00:01:25.434]
into construction,

[00:01:27.510]
or we’re focused on the construction side

[00:01:30.050]
where we’re working for the contractor.

[00:01:32.970]
And we are part of the bid team doing tender design,

[00:01:36.370]
detailed design,

[00:01:37.310]
And of course take,

[00:01:38.860]
definitely take that through to construction phase

[00:01:41.310]
supervision and, and hand over back to the owner,

[00:01:49.480]
as well as roles that I want to talk about.

[00:01:51.690]
We’ve got some trends in the industry,

[00:01:53.250]
which I think affect the way

[00:01:54.550]
that we think about ground risk and,

[00:01:57.890]
and the tools that we use to manage ground risk.

[00:02:00.713]
And all of our projects these days seem to be

[00:02:04.772]
what are called accelerated projects,

[00:02:06.910]
which means they just have to be delivered

[00:02:08.940]
a lot faster than the old timeframes.

[00:02:11.170]
And when I say the old timeframes,

[00:02:13.740]
I mean, say five or six years ago,

[00:02:15.410]
we had more time.

[00:02:17.250]
Nowadays, things are being really squeezed

[00:02:20.750]
in terms of program.

[00:02:24.710]
The other thing,

[00:02:25.757]
the thing is that most of our projects

[00:02:26.920]
are in the urban setting

[00:02:28.750]
and that’s got two factors associated.

[00:02:32.280]
That one is, there’s usually plenty of historic data

[00:02:34.800]
or at least some historic data,

[00:02:36.740]
possibly previous development on the same infrastructure,

[00:02:41.670]
possibly neighboring properties and

[00:02:43.680]
developments nearby.

[00:02:45.754]
So there’s that historic data

[00:02:48.140]
but also there’s the responsibility

[00:02:49.930]
to not affect adversely the environment

[00:02:53.190]
that our neighbors are occupying.

[00:02:57.280]
Another factor is that the

[00:03:00.870]
the GI that we do these days is very price-driven.

[00:03:03.880]
And I say here that that is variable quality.

[00:03:07.370]
What I, really mean is that

[00:03:10.830]
there’s some pretty poor quality work done

[00:03:13.090]
in the name of GI these days.

[00:03:16.480]
So that’s something that we have to address,

[00:03:19.430]
and it’s part of the risk

[00:03:20.750]
that we take into these projects.

[00:03:25.760]
These factors combine into a-

[00:03:30.650]
To cause some trouble

[00:03:32.330]
within in projects or push the risk,

[00:03:36.140]
pushing the ground risk back into the project.

[00:03:39.230]
So previously 5, 10, 15 years ago,

[00:03:42.880]
we used to deal with ground risk

[00:03:46.310]
or discovery of adverse ground conditions

[00:03:48.340]
in the early stages of a project

[00:03:50.813]
where those factors could be brought,

[00:03:53.410]
dealt with in detailed design.

[00:03:56.060]
But these days we tended to be doing

[00:03:57.990]
more and more GI

[00:03:59.540]
into procurement

[00:04:00.500]
and even into the construction phase,

[00:04:02.250]
meaning that any adverse ground discovered

[00:04:05.530]
has to be dealt with pretty late

[00:04:08.560]
when you’re already,

[00:04:10.740]
you’ve already got a fairly well developed design,

[00:04:13.690]
or maybe even under construction.

[00:04:15.830]
And then the cost implications

[00:04:19.010]
of those discoveries are much greater.

[00:04:21.510]
So this pushes the onus onto

[00:04:23.184]
dealing with any discovery of ground risk

[00:04:26.710]
as early as possible.

[00:04:28.040]
And of course, this is where 3D

[00:04:30.633]
and good data handling comes into its own.

[00:04:37.000]
Another factor about the projects

[00:04:38.920]
that we work in is that they are well,

[00:04:42.440]
they’re fairly big

[00:04:43.273]
and fairly complicated in IT terms.

[00:04:45.720]
So our little space in ground engineering,

[00:04:49.210]
which we have over on the side

[00:04:52.060]
is just a small part of a bigger hole.

[00:04:56.870]
And obviously there’s multiple disciplines.

[00:05:01.510]
We may be dealing with joint ventures,

[00:05:03.210]
teams with different IT setups,

[00:05:05.180]
different security,

[00:05:06.450]
and almost certainly dealing with projects

[00:05:09.380]
across several times zones.

[00:05:12.710]
And my view is that

[00:05:14.646]
the real heroes of delivery

[00:05:17.830]
on these big projects

[00:05:19.170]
are the data integration experts

[00:05:22.140]
who look after and federate the models

[00:05:26.010]
that have been developed

[00:05:26.843]
by the different disciplines.

[00:05:29.860]
Now, what does our modeling look like?

[00:05:34.780]
Just sort of a scrapbook snapshot

[00:05:37.645]
of some projects that we’ve worked on.

[00:05:41.457]
The 3D ground models

[00:05:44.930]
are the main part of the story.

[00:05:47.780]
And they do tell the big picture.

[00:05:50.280]
And of course, a big part of the successes,

[00:05:53.860]
the uptake of 3D modeling

[00:05:55.460]
in projects in general has just been the way

[00:05:59.180]
that the models facilitate communication,

[00:06:03.130]
but particularly about

[00:06:04.982]
the big trends in the,

[00:06:06.630]
in the ground story.

[00:06:09.760]
We also continue on every,

[00:06:12.450]
every project I’ve worked on.

[00:06:13.910]
Certainly, we continue to develop sections

[00:06:17.190]
and plans, charts, and texts, of course,

[00:06:21.150]
to inform design around specific structures

[00:06:25.302]
and specific ground risks.

[00:06:26.580]
So there’s a lot of detail

[00:06:28.830]
that goes into these supporting documents

[00:06:33.090]
that perhaps doesn’t get incorporated

[00:06:35.570]
into the 3D ground model,

[00:06:37.310]
or it’s not easy to get it into the 3D ground model.

[00:06:40.670]
And so there’s a sense that

[00:06:42.552]
most say 80, 90% of the data

[00:06:46.180]
in a ground model can be,

[00:06:48.001]
can be modeled,

[00:06:49.330]
but there’s really important data

[00:06:51.000]
that doesn’t sit well with,

[00:06:54.370]
in 3D model

[00:06:55.760]
or would take an awful lot of work to get it

[00:06:57.470]
into the 3D model.

[00:06:58.637]
And so we use,

[00:06:59.690]
we use other techniques such as 2D sections for that.

[00:07:03.700]
I think one of the big contributions

[00:07:06.130]
that we’ve made to the industry

[00:07:08.970]
is how we’ve used 3D ground models in procurement.

[00:07:13.867]
And I have two examples here

[00:07:15.160]
from the city rail link here in Oakland.

[00:07:19.520]
These three images on the left

[00:07:22.230]
are from our contracts,

[00:07:24.413]
six procurement,

[00:07:27.500]
And in this one we use the 3D models

[00:07:31.340]
for information only,

[00:07:32.610]
and we provide them, provide them as a

[00:07:34.820]
leapfrog view of file.

[00:07:36.360]
And because of the success of that in our next tender

[00:07:40.130]
for CRL, which was contract three,

[00:07:42.840]
the main works,

[00:07:44.660]
we were even more ambitious.

[00:07:46.370]
And in this one we used leapfrog,

[00:07:50.450]
or sequence central for the first time.

[00:07:52.620]
And we delivered each of the final two tenders

[00:07:55.850]
with their own central environment,

[00:07:58.440]
with their own copy of the,

[00:08:01.030]
all the data of course,

[00:08:02.270]
and the model that they could work with

[00:08:04.810]
and our specimen design,

[00:08:07.422]
and then they could either use or modify,

[00:08:10.930]
or just completely restart the modeling

[00:08:14.960]
according to their own wishes.

[00:08:16.450]
Now, the result of that was that

[00:08:19.520]
both of those tendering parties became,

[00:08:22.601]
you know, quite quickly, they, you know,

[00:08:26.322]
they develop their idea,

[00:08:28.030]
their own ideas on the ground risk on the ground conditions.

[00:08:31.610]
And we got good alignment

[00:08:34.384]
on the ground conditions

[00:08:36.020]
and how much risk they pose

[00:08:37.300]
and how each tender was addressing ground risk

[00:08:41.450]
in their tenders.

[00:08:42.590]
And it’s that alignment

[00:08:44.810]
between the scale of the issue to allow

[00:08:47.550]
fair and reasonable pricing,

[00:08:49.180]
which is the main thing.

[00:08:50.390]
We’re not trying to eliminate or

[00:08:54.250]
not trying to totally eliminate ground risk,

[00:08:56.160]
of course.

[00:08:57.166]
And that would be impossible,

[00:08:57.999]
But it’s in a procurement situation.

[00:09:00.243]
It’s trying to get an even playing field

[00:09:06.380]
and in the ISO 19650 sense,

[00:09:10.540]
what we’re doing here is we’re recognizing of course,

[00:09:13.720]
that as we transitioned from the design phase

[00:09:17.890]
to the construct phase,

[00:09:20.120]
there is traditionally quite a drop

[00:09:23.010]
in information or information loss

[00:09:24.770]
and there’s time.

[00:09:26.160]
And we are trying to minimize

[00:09:28.546]
this problem of this handover

[00:09:31.830]
between design and construction.

[00:09:37.629]
Another big area that we’ve moved into

[00:09:40.647]
is automation as another, quite a success story,

[00:09:44.800]
because many of our engineers

[00:09:46.240]
have been very quick to latch

[00:09:47.620]
onto the design advantages

[00:09:49.370]
of 3D ground, of course,

[00:09:52.990]
and automating a lot of drawing development

[00:09:55.871]
but drawing automation

[00:09:57.630]
is just scratching the surface of the benefits.

[00:10:01.090]
We’ve had extraordinary experiences

[00:10:03.420]
in value engineering,

[00:10:06.060]
As we’ve-

[00:10:07.200]
We’ve had really good discussions with the engineers

[00:10:10.820]
designing tunnel linings, or retaining walls

[00:10:14.320]
about the ground conditions.

[00:10:18.244]
The certainly they identify in a very detailed way,

[00:10:22.160]
the sensitive sites, sensitive ground conditions.

[00:10:25.930]
And we’ve been able to go back

[00:10:27.680]
and reclassify the materials

[00:10:29.385]
to optimize their designs,

[00:10:33.004]
and also to optimize the location of new GI

[00:10:37.530]
to maximize the value that the client gets

[00:10:42.100]
from the data.

[00:10:43.300]
Minimize the risk.

[00:10:46.930]
And we can quite readily demonstrate

[00:10:49.470]
cost benefits of additional GI

[00:10:51.720]
to clients using this method.

[00:10:55.890]
Now, we have also had problems

[00:10:57.780]
with 3D ground models,

[00:10:59.600]
and these have a bit of

[00:11:01.300]
a common theme around them,

[00:11:03.700]
and it’s one of miscommunication

[00:11:06.650]
and misunderstanding.

[00:11:08.830]
When we load the 3D geology surfaces

[00:11:12.010]
to the federated models environment,

[00:11:15.499]
we don’t load all the supporting documentation,

[00:11:20.770]
which still resides back

[00:11:22.640]
on our sequence central environment.

[00:11:27.420]
And that has been a problem

[00:11:28.790]
in a couple of key cases.

[00:11:31.616]
So for example, on the left here,

[00:11:34.320]
we have a case where

[00:11:37.800]
we’re underpinning historic building

[00:11:40.350]
that sits over a paleo channel

[00:11:43.080]
running from top left to bottom right.

[00:11:47.700]
We understood and described the uncertainty

[00:11:50.520]
around that paleo channel

[00:11:52.440]
and had that information available.

[00:11:55.445]
When the structure engineers are looking at it,

[00:11:59.690]
looking at the 3D surfaces in isolation,

[00:12:01.810]
they’re not getting that information.

[00:12:05.530]
On the right hand side, we’ve got a similar case.

[00:12:08.130]
We’ve got a paleo channel again,

[00:12:09.910]
as it happens this time filled with basalt.

[00:12:14.551]
And we intend to build a shaft

[00:12:17.130]
in through this area.

[00:12:22.130]
Now, in this case,

[00:12:24.150]
we had identified in the boreholes

[00:12:26.610]
and all the sections that

[00:12:28.420]
there were two main textures of the basalt.

[00:12:30.620]
The solid basalt and the rubbery basalt,

[00:12:34.460]
on the surfaces and sides

[00:12:35.900]
of this infill channel.

[00:12:39.932]
And, but that we didn’t model that in 3D,

[00:12:41.910]
we were unable to resolve

[00:12:44.574]
really complicated spacial variation

[00:12:47.820]
between areas of rubbery basalts and solid basalts.

[00:12:51.620]
And so we just modeled it as one extrusive volcanics.

[00:12:57.600]
Now, when the designers looked at

[00:12:59.821]
that they unfortunately

[00:13:02.100]
gathered that it was all solid basalt.

[00:13:04.870]
So look, those are two examples

[00:13:06.290]
of things that can wrong,

[00:13:07.250]
and I hope that helps other people avoid

[00:13:10.700]
that kind of miscommunication.

[00:13:14.780]
That leads me into the final part of my talk,

[00:13:18.690]
which is looking at what we’re doing

[00:13:21.520]
for the future.

[00:13:22.470]
Where are we putting our energy?

[00:13:25.440]
And a big part of what we’re doing at the moment

[00:13:27.770]
is around the analysis

[00:13:30.710]
and communication of model uncertainty.

[00:13:33.820]
But I can also see that

[00:13:35.410]
there’s a potential in a more quantitative method.

[00:13:38.920]
And I particularly really been looking at

[00:13:44.890]
work research work,

[00:13:46.290]
being done by Florian Wellman

[00:13:48.010]
and his coworkers on Monte-Carlo simulations,

[00:13:53.480]
using constrained perturbations of inputs.

[00:13:56.590]
And what we see here

[00:13:58.700]
are a couple of visualizations

[00:14:00.390]
out of that kind of work.

[00:14:02.590]
He has multiple representations

[00:14:04.480]
of strata contacts offset by a fault.

[00:14:09.600]
And here’s another fault

[00:14:11.860]
case showing the variation impossible unit.

[00:14:17.070]
And this, this method is called entropy.

[00:14:21.150]
These methods are really interesting to me

[00:14:23.760]
because they are some kind of hybrid

[00:14:26.660]
between full deterministic methods

[00:14:31.070]
and probabilistic methods.

[00:14:32.830]
And they’re hybrid because the perturbations

[00:14:36.550]
in the input values can be quite

[00:14:40.360]
closely constrained by the user,

[00:14:44.150]
resulting in a range of if you like,

[00:14:49.500]
of reasonable deterministic models.

[00:14:54.270]
The other big effort that we’re putting in

[00:14:57.870]
at the moment relates to our database

[00:15:01.370]
and it’s look-

[00:15:02.690]
we’ve known that we’ve had bottlenecks

[00:15:05.757]
and silos of data in our database

[00:15:08.350]
for a long time.

[00:15:09.840]
And that data handling has been

[00:15:14.650]
very laborious and slows us down

[00:15:17.300]
in a lot of ways.

[00:15:18.133]
And it makes the handling of data

[00:15:20.440]
the big cost in development of 3D models,

[00:15:24.750]
which is,

[00:15:25.841]
is really the tail wagging the dog.

[00:15:27.970]
So what we’ve designed is a more streamlined

[00:15:31.830]
and sensible story

[00:15:33.268]
around some of these issues that I list on the right.

[00:15:38.250]
So just some concluding comments,

[00:15:44.421]
I’ve talked about ground risk

[00:15:46.550]
in engineering projects

[00:15:48.840]
and certainly 3D models play a vital role

[00:15:52.553]
where we’re thrilled with the progress

[00:15:54.420]
that we’ve made using 3D models.

[00:15:57.920]
It’s a lovely toolbox to work with.

[00:16:01.370]
It doesn’t mean that we don’t use our other tools,

[00:16:03.260]
but 3D is a great way to communicate

[00:16:05.920]
with the wider stakeholders

[00:16:09.516]
in the project and colleagues in the project.

[00:16:13.070]
We also are aspiring to share model space

[00:16:18.280]
with digital twins

[00:16:19.680]
or ecosystems of digital twins that are featurive

[00:16:23.690]
of every smart city.

[00:16:25.810]
All cities these days, essentially.

[00:16:29.389]
And we think that to share that space

[00:16:32.710]
of precision modeling,

[00:16:36.748]
we need the uncertainty-enabled 3D ground models.

[00:16:41.300]
We can’t just put in a ground model

[00:16:45.930]
a deterministic ground model without clearly explaining

[00:16:49.547]
how it varies or how it may vary in space.

[00:16:54.560]
So we think these uncertainty-abled 3D ground models will

[00:16:57.090]
give us the right to

[00:16:58.510]
share space with digital twins.

[00:17:01.050]
And at the same time

[00:17:03.440]
we want to provide or have a transparent way

[00:17:09.150]
of accessing the data support.

[00:17:11.330]
So that is our data linked.

[00:17:14.550]
We would expect with city-wide

[00:17:17.780]
or regional or national databases

[00:17:20.520]
of grounded information.

[00:17:23.240]
Thank you very much.