Tunnels and Underground Cities: Engineering and Innovation meet Archaeology,

Architecture and Art – Peila, Viggiani & Celestino (Eds)

© 2019 Taylor & Francis Group, London, ISBN 978-1-138-38865-9

3D geological modelling for the design of complex underground

works

F. Giovacchini, M. Vendramini, L. Soldo, M. Merlo, D. Marchisio, G. Ricci &

A. Eusebio
Geodata Engineering, Turin, Italy

ABSTRACT: Modern planning and design for complex infrastructures provides the concep-
tion of a robust geological conceptual model that accounts for all the involved parameters,
their intrinsic variability and associated risks. Geodata Engineering has undertaken an
innovative approach by systematically implementing 3D geological models to suit this
demand. An implicit modelling engine developed for the Civil Engineering industry is used to
find a unique solution that honors both the provided dataset and the geoscientist interpret-
ation while mimicking natural geological geometries. This paper analyses methods and work-
flows used to integrate 3D implicit modelling and 2.5D explicit modelling. The key
advantages of this approach are examined through relevant examples. Better visualization of
available data brings improved understanding, robust analysis and dynamic interpretation.
The modelling tool of choice not only expedites the first iteration of a 3D geological model,
but it allows a faster updating process and the construction of multiple versions to test several
interpretations. This process leads to improved precision and a reduced geological risk,
becoming a milestone of the Geodata Risk Analysis Based Design (GRBD) approach. In add-
ition, it is explained how 3D geological modelling provides an innovative communication tool
for internal and external collaboration among professionals with a focus on the integration of
3D geological models with engineering designs from BIM.

1 INTRODUCTION

Geologic conditions are the greatest source of unknowns prior to actual construction of
underground excavations, especially for deep and large tunnels in rock. It can, therefore, be
argued that the geological model has a commanding impact on the entire design and risk
evaluation process. It is therefore necessary to conceive a robust Geological Conceptual Refer-
ence Model to minimize risks and costs.
Several authors (Venturini et al., 2001; Knill, 2002; IAEG Commission C25, 2014; Soldo
et al., 2014, Riella A. et al., 2015) have progressively proposed the concept of Geological Ref-
erence Model (GRM) as a framework capable to fulfil these necessities. The GRM is built
progressively, with the contribution of different specialties (e.g. geomorphology, structural
geology, etc.), it is based on natural laws with experts’ judgement. In many projects 3D
models can be very useful in assessing baseline geological and geotechnical reference condi-
tions, hazards, risks and in defining the proper mitigation measures.
The Geological 3D modelling is an industry standard in mining and oil & gas, but it is just
starting to grow in the civil engineering industry.
Subsurface information is intrinsically incomplete since sampling is not as dense as neces-
sary to solve uncertainties and underground conditions are heterogeneous. It is therefore
necessary to use specific software to construct 3D models. This must be able to give ‘geo-
logical sense’ to geometries using custom algorithms. A 3D model is a tool to store and pro-
cess the input data, to understand, define, quantify, visualize, or simulate a certain aspect of

846
geological conditions. It requires interpretation and refinement of the input data as well as a
defined workflow and must be built by a geologist. 3D models allow to see how the layers
interact in 3D which the human brain cannot process from a 2D geological map.
One of the main purposes of a 3D geological model is the conception of a robust geological
model that leads to improved precision and reliability of the reference base model for under-
ground works design.

2 SOFTWARE

3D geological modelling software needs to be capable of handling complex geological infor-

mation, from irregularly spaced data sets (e.g. from boreholes and/or surface exposures).
The use of Geological Modelling tools and methodologies in building the Design Geological
& Geotechnical Model stands among the foundation milestones of the design procedures in
Geodata Engineering.
Recently, the deployment of Leapfrog Works by Seequent has brought interesting out-
comes. The software’s Implicit Modelling Engine is based on a specifically designed algorithm
(FastRBF™, Fast Radial Basis Functions).

2.1 Basic Principles

There are two methods that are used in geological 3D modelling software; Implicit and Expli-
cit modelling. Explicit Modelling provides that the modeler defines geological structures such
as lithologies, faults, folds and veins by drawing them on regularly spaced sections and joining
them. The data offers constrains for the structures, for example in the form of borehole strati-
graphic data projected onto the cross section, but the process is fundamentally based on draw-
ing and the reasoning is based on 2D elements that are then joined together in a 3D space.
In Implicit Modelling computer algorithms directly create the model from a combination of
measured data and user interpretation. Input data is evaluated against a mathematical func-
tion, by following geologically meaningful rules and parameters that are set up by the user for
a specific geological setting. The interpolant is suitable to construct surfaces using discrete
variables such as lithologies as well as continuous variables such as geotechnical parameters.
The result is a unique solution that honors both the provided dataset and the geoscientist
interpretation while mimicking natural geological geometries.
The advantages of implicit modelling compared to explicit modelling are multiple (Cowan
et al. 2002, Cowan et al., 2004):
• the result of the interpolation is the smoothest surface available that is the most suitable
solution for geological modelling;
• just a limited amount of digitization is necessary;
• evaluating thousands of points takes seconds therefore incorporating a huge amount of
contact points is not a time-consuming process;
• editing is easy and fast, it entails digitization of polylines, points and measurement discs
over existing surfaces or chosen sections with any orientation;
• geometrical constraints in the form of rules as a fault terminating against another are easily
applicable.
• the improved speed of the process compared to explicit modelling leaves more time for the
actual data conditioning and interpretation process.
• the model building and updating process becomes an iterative process;
• updates or production of multiple versions are sped up significantly.
The used interpolant is based on FastRBF™. Radial Basis Functions describe the predicted
value at a point X as a function that depends on the distance from the point where the value is
known. RBFs are a group of global interpolation methods where the interpolant is dependent
on all data points. Fast RBF interpolant uses advanced algorithms to solve the combination
of Radial Basis Functions and evaluate the estimates quickly. The interpolation of surfaces
depends on the distance from the points where values are known.

847
2.2 Input Data and Interpretation
Leapfrog Works can import several forms of input data that are listed below.
• Boreholes and every form of associated data as interval data and point data; this can
include rock type distribution, natural defects, values from in situ or lab tests or calculated
indexes such as Core Recovery, RQD, GSI, RMR, etc;
• Elevation grids;
• Surface data from geological surveys (geological contacts, discontinuity and foliations
trend or orientation);
• Remotely sensed data and ortophotos
• Geophysical data;
• Scanned Cross Sections;
• GIS data as shapefiles;
• CAD data as .dxf o .dwg.

3 KEY POINTS OF THE MODELING APPROACH

When building a geological model, there are a few key rules to keep in mind. It is fundamental
to have a clear final purpose. A good workflow does not involve all the available data, instead
it focuses on modelling the key elements to clarify doubts and concerns that have arisen in the
preliminary evaluations.
Another key element of a successful modelling process is to keep the complexity to the min-
imum to have a fast and agile elaboration process. It is also important to model only over the
area of interest. Radial Basis Functions tend to “flatten out” in the areas where constraints
from data of interpretation are not present making the model less geologically reliable.
A basic version of a geological model in Leapfrog Works can be built based on a small
amount of input data as a geological map and surface elevation detail; this can guide the first
evaluation on the area and help in further investigations planning.

4 3D MODELLING CASE HISTORY IN GEODATA ENGINEERING

In the following paragraphs some GDE case histories are described.

4.1 Example A – 3D Geological Modelling - Hymalaian base Tunnel (India)

The first case history is a long base tunnel (approximately 12 km long), part of the improve-
ment of the service and safety condition of the Manali – Sarchu Road, connecting Lahaul dis-
trict, in Himachal Pradesh, to Ladakh in Jammu and Kashmir, India. The tunnel is located
underneath one of the highest Himalayan passes (el. 4890m a.s.l.), yearly closed for long
periods because of the extreme climatic conditions.

Figure 1. The planning for the next decade provides the construction of several long base tunnels in the
extreme Himalayan environment, requiring an appropriate approach to the identification of the Design
Geological and Geotechnical Reference Model (photo Geodata).

848
 The bedrock complexity is the result of the juxtaposition of geologic features - formed
during a multi-phase depositional and deformation history - including variable sedimentologi-
cal structures, unconformities, folds, faults and joints, both at regional and local scale. The
state-of-the-art approach entails the geological study and it has been supported by Leapfrog
Works 3D modelling.
A key item for the project is the understanding of the complex relationships and geometry
of some major faults. The fractured rock masses reveal a complex deformation history related
to the interaction of a pervasive ductile and brittle deformation phases that took place in dif-
ferent temperature (T) and pressure (P) conditions.
The rock masses exhibit an intense degree of fracturing, somewhere concentrated,
without large outcrops of fault related rocks (gouges or cataclasites). The complex and
scattered geometry of this faulting-related fracturing, together with the extent of the
areas covered by Quaternary deposits made the field recognition and characterization of
the fault zones difficult. This is particularly relevant for the slope debris fan and
moraines that cover most of the valley downstream the region where literature references
map the presence of thrust planes.
The use of 3D analysis has been of paramount importance providing a powerful
stand-alone environment for data integration, cross-section construction and 3D model
building, making possible the understanding and validation of tectonic geometries other-
wise hidden. The following figures show some examples of the analyses on some relevant
structures.

Figure 2. modelling folded Cambrian Units (purple surface).

Figure 3. the modelled frontal thrust with a double possible interpretation: (1) the pink plane with
southward trend and (2) the red plane south verging; the back thrust is showed by the purple plane (3),
exposed without the DTM.

849
4.2 Example B – 3D Geological Modelling - Topo River Valley (Ecuador)
The geological 3D model presented in this section summarized the results of the geological
and structural studies carried out along the Topo River Valley (Ecuador) to assess the suit-
ability of the valley for hydroelectric power plant design and construction. Because of the
lack of a reliable local geological model, the studies were performed to enquire and represent
the geological layout of one of the most complex area of the country. This location shows
brittle and ductile deformation with scarce outcrops and a widespread quaternary cover.
The geological setting consists of different tectonic units including metamorphic and Meso-
zoic to Tertiary sedimentary geological formations (Hollin, Napo, Tena Formations). Most
common rock types include fine grained phyllite, slate, shale with interbedded meta arenite
and quartz-arenite. Granite intrusions locally cut the whole sedimentary sequence. Rock mass
are strongly deformed, faulted and folded. Regional cleavages, tight to isoclinal folds, major
to local fault planes both intersecting and affecting rock masses formed a composite geological
scenario to investigate and assess the suitability of the area for the project.
Due to the difficulties to get access to the crucial sectors of the valleys (caused by steep
morphology and vegetation cover, presence of private proprieties, distance from the roads,
etc.), the geotechnical investigations were limited to very small areas. For this reason, exten-
sive geological surveys produced detailed geological maps, regularly spaced cross sections and
a longitudinal profile along the Valley axis where the main underground works were foreseen.
Based on the outcomes of the performed ground investigations, the model was built accord-
ing to the following steps:
1. Uploading all the available input data; elevation surface data, geological map and cross
sections images (Figure 4);
2. Digitizing the geological and structural contacts was a crucial step in this workflow because
of the complexity of the local geology and the lack of borehole outcomes in several relevant
areas. The digitized 3D polylines represent therefore the only constraint to build the 3D
surfaces and, consequently, lithology volumes are given in Figure 5;
3. Once the contacts were correctly modelled on the assumed geometries and reciprocal rela-
tionships (stratigraphic pile, deposition, erosion, vein, intrusion) it was possible to get into
the next step and deal with tectonic elements;
4. Faults could therefore be modelled as surfaces whose action will split the volume in fault
blocks that can be managed separately. Each fault block was checked independently, making
sure the correct displacements were shown as well as stratigraphy elements activated;
5. This was followed by modelling the faults as 3D elements to show the expected fault
affected zone and to quantify its volume (Figure 6);
6. All the geological elements were included in the final block model (Figure 7) that shows the
interpolated geology (surface, volume) in different sectors of the valley. Furthermore, it has
been possible to extrude the local geology (3D block volume) along design structural elem-
ents with the aim to assess the expected lithologies and the possible occurring hazards to be
considered. It was therefore possible to give a relatively reliable preliminary estimation of
the volumes by exporting the properties table.
In this case, 3D modelling allowed designers to define the most credible geological
scenario of the area. The model included different geological elements (stratigraphic
units, brittle and ductile tectonics structures, geomorphologic elements, etc.) and their
mutual interactions with internal and external geodynamic processes. Ultimately, the
model highlighted the principal hazards related to unfavorable geological and hydrogeo-
logical conditions.

4.3 Example C – Geotechnical modelling applied to cavern design

This example deals with the design of a hydroelectric power plant in a complex sedimentary
sequence of Carboniferous age. The structural framework shows a complex brittle deform-
ation process where joints and faults associations from a compressive regime have been found

850
Figure 4. Input data for the 3D geological model represented by a geological map, regularly spaced
cross sections and structural measurements. The accuracy of the model depends on the quantity, quality
and accuracy of the input data. A detailed geological survey is required for a reliable 2D Baseline model
(maps and sections).

Figure 5. Geological contacts modelled onto a folded structure that shows the effect of deformation
(fold and fault displacement). The definition of complex tectonic structures needs to rely on an adequate
amount of field data to conceive the most likely geometry and rock types distribution especially at depth.
Each measurement represents a “natural” constrain for surfaces (geological contacts) to be modelled.

in the rock mass. The focus is on the caverns design and construction when specific requests
are formulated by the engineering teams to obtain input for their numerical simulations. The
main target of the model was to provide a 3D geotechnical characterization of the caverns
location.

851
Figure 6. Volumetric model of fault shear zones surveyed in the areas. For each fault system, the main
features such as orientation, type of movement and expected thickness were defined. Fault zone excava-
tion is one of the most critical hazards to be faced in tunnel construction; fault zones were modelled as
discrete rock mass volumes displacing rock units.

Figure 7. Final geological block model. It includes different volumes displaced by tectonic structures.
Geological sections and 3D orientation of tectonic structures can be easily visualized within the study
area. On the right-hand corner, a detail showing the expected geology and fault thickness along the
tunnel alignment. (A) Fault shear zones are displayed in red, the rock formation in brown.

The 3D numeric model of this area was based on RMR (Rock Mass Rating, Bieniawski,
1989) index data calculated over the boreholes and excavated tunnels. The workflow in this
instance included the study of different aspects within the area:
1. A global model of area was made by imposing the effect of anisotropies (stratigraphy, fault
zones, master joints, etc.) on the RMR numeric model (Figure 8) that generate the RMR
classes distribution. It is important to highlight how the envelope surfaces are considered
reliable only in the proximity of the factual data (boreholes, face mapping).
2. A discrete model of fault planes (Figure 9) was made and then refined over the cavern
volume, this allowed to study their geometries and used as an input for wedge stability
analysis.
3. Faults zones were also modelled as 3D elements (volumes) defining their influence zone
(core zone and damage zones). To these horizons RMR class IV was assigned.
4. The global RMR model was refined over the cavern volumes taking into the effect of the
fault zones (Figure 10).

852
Figure 8. RMR Global model showing the effect of anisotropies in Bieniawsky’s classes distribution. In
the left-hand corner, it is shown the caverns location.

Figure 9. Discrete model of fault planes affecting the caverns excavation. Fault planes were modelled
according to face mapping outcomes.

Figure 10. Final RMR local model (caverns) that account for the fault influence volumes.

5 ADVANTAGES OF 3D GEOLOGICAL MODELLING IN UNDERGROUND

DESIGN

Geodata Engineering has undertaken a systematic approach toward geological 3D modelling

which is now deployed in all the new ongoing projects. Since the earliest projects in which it
was applied, the modelling has quickly shown several advantages, listed in the below
paragraphs.

853
5.1 Iterative process
The design of a geological conceptual model is inherently an iterative process as the model is built
with a little amount of data at the start and as soon as new data comes in it is improved in time.
It is also a dynamic object because the design of the infrastructure is modified due to multiple
factors as geological conditions and other issues. It is therefore fundamental to be able to update
the model in a fast and efficient way to provide a prompt response to the designer’s needs.

5.2 Multiple versions

The ability to create and process a 3D geological model in a time-efficient manner helps to
make different scenarios and test them in different model versions. The capability to calibrate
quickly and in progress the most credible geological layout improves the efficiency of the
whole design process. This lends to a robust geological reference model that can be used to
take decision and find design solutions.

5.3 Integration with BIM

The BIM elements can be managed by Leapfrog Works, this significantly helps to investigate
the impact of site geological and geotechnical conditions on the design. Infrastructure 3D
models from BIM can be imported into the 3D geological Model with a limited post process-
ing effort that consists into simplifying the file to make it suitable for Leapfrog Works.

5.4 3D Model as a Communication tool

3D geological models have a great potential as a communication tool. If displayed correctly
they help to visualize complex geological geometries that the human brain cannot easily pro-
cess from a 2D geological map. This means geological reasonings are much more accessible to
nonprofessional figures. Moreover, they allow a much better understanding of the geological
reference model and its associated risks. A common ground is created, and this improves the
collaboration side of the decision-making process.

6 CONCLUSION

The use of geological 3D models is paramount in the approach to Geodata Risk Analysis
Based Design (GRBD) since it has a capability to simultaneously store, elaborate and interro-
gate multiple sources of input data. It can efficiently support designers in the crucial phases
both of design and construction process.
Notwithstanding the initial effort in collecting, processing and storing the input data, the
consequent advantages of getting a reliable 3D model become consistent during the design
process when the interactions among different professionals need a clear and common base of
discussion.
he chosen software is a flexible tool because it is useful not only for preliminary and quick
considerations, but it allows a faster updating process and the construction of multiple ver-
sions to test several interpretations. This can be considered an added value especially in tunnel
or underground works design because the construction of a Reliable Reference Model is char-
acterized by an intrinsic uncertainty level.
In these contexts, especially for deep tunnel design the introduction of new data (borehole
outcomes exploratory galleries and face mapping) in the model, can deeply affect the inter-
action among the different elements changing the final interpretation. Extensive sensitivity
analysis of the different scenarios are often carried out to assess and mitigate the potential
impacts of the hazards. Most of them can be proper identified (distribution of the hazards
along the alignment), quantified (“volumes” definition for quantities estimation) and visual-
ized to define the design solutions.

854
REFERENCES

Bieniawski, Z.T., 1989. Engineering Rock Mass Classification. John Wiley & Son.
Cowan, J., Beatson, R., Fright, W.R., McLennan, T.J., Mitchell, T.J., 2002. Rapid geological modelling.
International Symposium, Kalgoorlie 23–25.
Cowan, E.J., Lane, R.G., Ross, H.J., 2004. Leapfrog’s implicit drawing tool: a new way of drawing geo-
logical objects of any shape rapidly in 3D 3.
IAEG commission 25 S. Parry, F. J. Baynes, M. G. Culshaw, M. Eggers, J. F. Keaton, K. Lentfer, J.
Novotny, D. Paul., 2014. Engineering geological models: an introduction. Bulletin of Engineering Geol-
ogy and the Environment, Volume 73, Issue 3, pp 689–706.
Knill, J. (2002). Core Values: The First Hans Cloos Lecture’, 9th Congress of the International Associ-
ation for Engineering Geology and the Environment, Durban S. Africa, 45p.
Riella A, Vendramini, M., Eusebio A, Soldo L., 2015. The Design Geological and Geotechnical Model
(DGGM) for Long and Deep Tunnels. Springer International Publishing - Engineering Geology for
Society and Territory - Volume 6, pp. 991-994.
Soldo L., Vendramini M., Eusebio A. 2014. The Design Geological and Geotechnical Model (DGGM) for
major infrastructures. The contribution of Tectonics and Structural Geology. Lyon, AFTES Congress
2014.
Venturini, G., Damiano, A., Dematteis, A., Delle Piane, L., Fontan, D., Martinotti, G., Perello, P., 2001.
“L’importanza dell’affidabilità del Modello Geologico di Riferimento negli studi per il tunneling”. Atti
del congresso “Geoitalia 2001 3°Forum Italiano di Scienze della Terra” – FIST, Chieti, 5-8 settembre,
pp. 426-427.

855