Three dimensional geological models in ground engineering: when to use, how to build and review, benefits and potential pitfalls
This paper was originally published in Australian Geomechanics in September 2018 and is reproduced by permission of AGS
ABSTRACT
Three dimensional geological models are increasingly being used to characterise the world of ground engineering. Soil, rock, geological structures such as faults, rock fall zones and slips are often best examined in 3D. In addition, geomorphology of materials above, below or surrounding project sites must be well understood by design and construction teams to optimise both safety and costs. This is especially true for large, complex or unsafe sites or for forensic investigations in both terrestrial and offshore settings. As we shift away from a 2D (long sections, cross sections and design software programs) way of working to a fully 3D system of design it is important to follow suit with site characterisation tools. There are many advantages to thinking and working in 3D as well as some serious pitfalls when using such models.
Models must always make sense geologically and geomorphologically and preferably be reviewed by a geologist/engineer team. The geologist does not necessarily have to be a specialist engineering geologist, they could be a pure geologist but one who can thoroughly and clearly explain all issues and features to the engineering team. Models described herein are geological models first and foremost with an emphasis on being geologically and geomorphologically accurate. The modelling process introduces the engineering aspects once the geology is well understood. These models must always be used with a degree of caution and updated with new information from all sources such as pile excavation records, new drilling, earthworks changes mapping of faces in tunnels and excavations. Those who construct models must always assess confidence levels in the end-product and communicate that level of confidence or areas of ambiguity to all users.
3D Geological models are more suitable for large projects or projects that have potentially complex ground or hazardous site conditions. Examples of suitable types of project include; tunnels, deep excavations, large slopes, landslide remediation, involving soft/hard ground interfaces, areas with structural complexity such as folds or faults, slips and rock fall etc. In addition, forensic investigations may be enhanced by the application of 3D models.
1. INTRODUCTION
Digital geotechnics is developing rapidly in many different applications and forms. Much of the development appears to be heading towards a Virtual Reality (VR) output or end use. This paper explores 3 types of application for 3D geological models including photogrammetry and laser scanning, pure 3D geological and engineering geological models for large infrastructure projects and models in VR using case studies from recent investigations. There are several publications such as Parry et al (2014) (Engineering geological Models: an introduction IAEG commission 25), Stapleton (1982) (subsurface engineering-in a search of a rational approach) and Knill (2003) Core values: the first Hans Cloos Lecture that explore the merits of geological models that discuss issues relating to geological and engineering properties. According to Parry et al (2014), the definition of an engineering geology model is “An approximation of reality created for the purpose of solving a problem”. Both Knill (2003) and Fookes (1997) described engineering geological models as more than purely geological in nature and describe pure geological models as either “Inadequate” or needing to be “More than simply geology that is useful for engineers”. The authors of this paper broadly concur with this conclusion with the proviso that first and foremost the geological correctness of the model is established as the guiding principle. Once the geology is reviewed and assessed as being as close as possible to reality using the available data overprinted with geological doctrines then the last stage in the 3D modelling process can be accomplished. Finally, the introduction of engineering properties into the model can be carried out to complete the process.
The geologist/engineer partnership in this process is critical to achieving success within the project in terms of design and eventual construction. Baynes et al (2005) described two types of model:
1.) The conceptual approach which is not related to 3D space or time and is built up using what might be anticipated in an area by using geological maps, local knowledge and experience.
2.) The Observational approach which is based on the observed and measured distribution of engineering geological units and processes.
Occasionally, geological conceptual models are not used well in the planning phases of investigations and models have to make do with the drilling patterns that are set out without much regard for underlying geology. Most of the models described herein have conceptual and observational elements in that they are built using drilling and mapping data collected first during site investigations. 3D modelling software packages such as Leapfrog are designed by engineers and geologists and rely on algorithms within the software package. The model may require ‘assistance’ from a reviewer (usually a geologist at this stage) using geological and local knowledge to produce correct geological shapes and distributions of soils and rock. Occasionally patterns and shapes that are not possible in nature can be created by modelling software. An example of this experience is where a basalt flow would have to move upslope during deposition prior to cooling and solidification (which is not possible). Above all 3D geological models must evolve and be updated when new information comes to hand. Any change must make sense geologically.
Examples used herein include: one from Christchurch from a site that had severe access and safety issues, selected projects in Melbourne, Auckland and Queensland containing complex geology. In addition, examples of VR applications from projects in NSW and Christchurch are also presented. The VR aspect allows geotechnical specialists to showcase the extent to which 3D imagery and models can be used to aid design, assist with helping all designers, not just the geotechnical specialists to understand their project or site. Importantly, this approach allows the client to see the project at a scale and in a way that is enlightening.
2. WHEN TO USE A THREE DIMENSION GEOLOGICAL MODEL
Whilst many projects can be enhanced using 3D geotechnics, not all are suitable and may not require use of this technology. Below is a guide to assist where and when these applications may be best used or not as the case may be.
2.1 Potential Projects More Suitable For 3D Geological Model Use
1. Where sites contain geological complexity that cannot be adequately represented on 2D sections
2. Specific areas where complexity and risk are high, however geological processes must be recognised and understood
3. On forensic projects where a tight drilling pattern is needed to help identify issues
4. There is sufficient drilling density and other data (seismic, topographical etc) to build a meaningful model
5. The effects of geomorphology and depositional patterns if recent depositional processes need to be well defined
6. Projects where the works methodology will vary with the ground conditions such as dredging in mainly soft ground with potential for the presence of hard rock or dense materials such as gravels
2.2 Potential Projects Less Suitable for 3D Geological Model use
1. Where sites are geologically simple, i.e. contain a single uniform layer
2. Specific areas where complexity and risk are low and ground conditions very well understood
3. When there is insufficient drilling density and other data (seismic, topographical etc) to build a meaningful model
4. In projects that may be faulted or contain slip surface and there is not a good understanding of the geological processes involved
5. Projects where the works are not sensitive to the ground conditions
3. 3D Model Types How to Build and Review
3.1 Photogrammetry and laser scan Models
With the increasing availability of affordable Unmanned Aerial Vehicles (UAV / drones) on the market, there is a huge opportunity to use them for engineering projects. Drones can be used to “visit” places otherwise not easily accessible or inherently unsafe on foot and can in many cases be used to undertake inspections of cliff faces without the need for scaffolding or roped access. The key limitation being that one can only see the rock without being able to touch it, but often this may not be that important. Photogrammetry techniques enable accurate 3D digital models to be developed from the photographs taken from UAVs. The process uses readily available commercial software and when combined with traditional survey or laser scanning data, the models can be just as accurate as traditional survey techniques but with far more detailed coverage, see Figure 1.
Figure 1: Combined photogrammetry and laser scan model (left) transitioning to the laser scan model only (right)
The principal enemy of photogrammetry for use in slope stability applications is vegetation, however with the increasing ability of artificial intelligence, already standard photogrammetry software is including tools to automatically identify isolated objects within the models to aid in “cleaning-up” models. For slopes with limited vegetation however, the quality of the data is nearly indistinguishable from a real image of the site, see Figure 2.
Figure 2: Comparison of a real photograph (A) and a screenshot of the photogrammetry model (B)
Models comprise a point cloud where each point has its own x,y,z coordinate. Recent models have a point every 3mm across a site that is hundreds of metres in size. This level of detail is often unnecessary for some applications, however for engineering geological applications the data can be incredibly valuable. Currently, slope stability models are typically undertaken on an assumed profile, or perhaps a detailed cross section may comprise a surveyed point every 500mm but this is only undertaken by providing the surveyors with a specified line to measure. If a more critical section is discovered later then ideally the survey crew should be sent out to site again to measure the new section. A photogrammetry or laser scan model negates this requirement, providing an accurate model across the entire site and if a new cross section is required then it is a case of simply extracting the data from the 3D model. This technique has been used to extract very accurate cross sections of a cliff on which to then run 2D rockfall analyses. The results of the 2D analyses can then be brought back into the 3D environment. An example of this is presented in Figure 3 below, the rockfall modelling was used to manage health and safety on a seismically active site in Christchurch, New Zealand.
Figure 3: 2D rockfall analyses undertaken on cross sections re-imported back into the 3D environment
3.2 3D Geology Models in Large Infrastructure Projects
3.2.1 City Rail Link (CRL): Auckland, New Zealand
The project comprises twin 3.5km long rail tunnels linking the CBD’s Britomart terminus with the Mt Eden Station.
Located in the heart of the CBD, there is abundant historic ground investigation data in the project corridor. Project-specific geotechnical investigations have largely followed a traditional site investigation philosophy with preliminary investigations (Stage 1), a large main investigation (Stage 2), supplementary investigations (Stages 3 and 4) targeting specific ground risks or design changes and tenderers requests (Stage 5). Figure 4 below depicts the traditional 2D section approach to tunnel investigations and design.
Figure 4: Traditional 2D sections from the early stages of the City Rail Link project showing iterative development of model (as desk-study and early GI data become available), and the implicit 2D ‘language’ of uncertainty: A) Concept Design inverted to match sections B and C; B) Geotechnical Desk Study; C) Geotechnical Interpretative Report
The main lithology for the project will be the relatively benign East Coast Bays Formation (Miocene sedimentary rock). There is, however, significant complexity and geotechnical risk principally at the north and south portal areas due to deep weathering profiles, Pleistocene geomorphology, local Quaternary volcanism and soft Holocene marine sediments. Risks from natural ground are compounded by potential clashes with the existing built environment and limited access to some areas of the site (i.e. with consequent difficulties obtaining additional drilling investigations).
Early work on the ground models for the project were by traditional borehole database with output to CAD or GIS (sections and plans). 3D modelling was initially adopted as a trial (2014) to explore an area of the project with adverse geology interpreted mainly from historical data of varied quality. Leapfrog software was used to construct a 3D model, firstly by using long sections and cross sections placed in 3D within Leapfrog and snapped to the alignment. The model was then constructed using the borehole database and cross checking against the sections. Figure 5 below shows the long section and cross section placement for model checking and refinement.
Figure 5: Using long sections and cross sections to build and refine the CRL project. Complex geology and curved alignments challenge the 3D understanding of potential users.
3D geology modelling has now been fully developed (Figure 6) in three areas of the project where there are specific geotechnical risks and/or design challenges relating to the ground conditions. These areas are at each end of the projects where tunnels or other development is at shallow depth and where ground conditions are most complex, and at an intermediate station where a faulted band of harder rock posed potential groundwater issues (and provided some opportunities for foundations).
Figure 6: Fully developed 3D geological model for the CRL project in Auckland
In each case progress of the models is iterative – the process of modelling lends itself to establishing decision points where value of the model can be assessed and benefit of further development assessed. There has been a cautious stepwise approach to 3D model adoption. The earliest model was internal to the client-side support and used to inform risk and design processes. The latest model was issued (for information only) to tenderers for information with the relevant construction contracts. Models are used to support traditional presentation of ground conditions as sections, and contract baselines have used traditional forms.
3.2.2 Westgate Tunnel Project, Melbourne
The West Gate Tunnel Project comprises 2.8km and 4km tunnels beneath Yarraville in Melbourne. The complex geology encountered within the tunnel alignment includes existing fill embankments, Holocene soft clays, Quaternary alluvium and infilled palaeochannels, high strength basalt of the Newer Volcanics, Tertiary clays and sands of the Brighton Group and Newport Formation, deeply weathered basalt of the Older Volcanics and underlying Tertiary clays and lignite of the Werribee Formation, often appearing as interflow deposits within the Older Volcanics.
During the development of engineering solutions, two key features of the 3D geological model became apparent; the infilled palaeochannel beneath the existing freeway near the inbound south portal and the weathering profiles within the Older Volcanics.
The Stony Creek palaeochannel is infilled with Holocene soft clays and is overlain and partially displaced by predominantly coarse, granular fill (gravels and cobbles) placed around 50 years ago as part of the construction of overpass embankments for the West Gate Freeway. The development of the 3D model for the infilled palaeochannel used a combination of the available borehole information (four sections containing moderately spaced boreholes across the palaeochannel), historic aerial photographs of the former creek and drawings for the construction works as there was, quite reasonably, insufficient investigation information to fully delineate the palaeochannel based on interpolation from boreholes alone. Sections cut at 40m centres along the 400m length of palaeochannel were used to develop the 3D geometry of the palaeochannel around the bends of the former creek alignment (Figure 7).
Figure 7: Development of 3D geological model for Stony Creek palaeochannel
The 3D geological modelling of the weathering within the Older Volcanics presented a different challenge in that the variably weathered material may range from a residual soil to an extremely high strength, fresh basalt over short distances. The uncertainty and ambiguity associated with modelling non-continuous, variable surfaces between sparse borehole information along the tunnel alignment and extrapolation of this away from the tunnel alignment can become difficult to represent in a 3D model.
A beneficial outcome that was observed during the modelling is that the development of the model strongly encouraged 3D thinking based on ‘Geo-logic’ and clearly identified gaps in the available information and understanding.
3.2.3 Infrastructure project, Brisbane
An investigation for a section of already constructed infrastructure was undertaken in Brisbane two years ago. A closely spaced drilling pattern of 6 boreholes in an area of 30m by 30m was employed with a cross-hole seismic refraction investigation carried out to look at ground conditions between boreholes and 25m below surface and ground level. Boreholes drilled at 70m apart had originally been completed for the detailed design phase of the project. The area of interest was very small and fell between the original borehole locations. Drilling information from the new and old boreholes was incorporated into a data set that was used to construct a 3D model using Leapfrog software. The 3D model allowed the investigation team to interrogate the geology model constructed from borehole data and enhanced by the seismic refraction data from many different angles, views and levels of magnification. An ability to look at this area in detail with a 3D element resulted in a fault with a small throw being identified at the location of the drilling. This fault was unknown and not expected but provided a neat explanation of the ground conditions encountered during the site investigation.
3.3 The use of Virtual Reality and 3D MODELS
VR incorporating 3D geological a model is a powerful way to combine the latest technology into a useful design tool and client visualisation product. Two actual examples of this combination are described below. The first for a project in Christchurch and the second on a roads project in NSW.
3.3.1 Augmented and Virtual Reality used in Christchurch project
3D models clearly have the advantage of enabling design in three dimensions, which is particularly useful for complex projects, however once the model exists, there is the significant additional advantage of being used for consultation with stakeholders. Figure 8 shows an example of a model viewed through the Microsoft HoloLens. This provides a large virtual model that can be “placed” on a table and stakeholders or designers can collaborate, all viewing the same virtual model in three dimensions. The models can also be viewed within a virtual reality environment where they can be marked up for use as site inductions prior to going to site as the example in Figure 8 shows. The ability to have had a virtual site induction prior to going to high risk sites can have a hugely beneficial impact on health and safety, but also can reduce the need for some site visits where the sole purpose is a high-level review of progress.
Figure 8: Viewing 3D models in Mixed Reality (left) and Virtual reality (right)
3.3.2 Virtual Reality used on the Newcastle Inner City Bypass project, NSW
A series of models for Newcastle Inner City Bypass project in NSW were constructed and then combined. Firstly, a 3D fly through and model of the entire project including cuts, embankments, pavements and bridges was constructed for client visualisation. A 3D Leapfrog geology model was then constructed separately for a single cutting area using existing borehole information and exported into an industry standard #D format. At this point the model was brought into the VR model and attributes for each rock unit added. This method allows the user whether that is the designer or the client to walk around the infrastructure. In this case the geology was visible for this cut and could allow subtle design changes to optimise bench location for drainage or batter angle on specific benches where geological features such as coal deposits are known to exist. This is only possible because the user gains an appreciation of all components of the model in 3D. Screen shots from within the VR model are presented below in Figure 9 and Figure 10. These depict the visual output and how a rock unit can be interrogated for geotechnical parameters.
Figure 9: Geology from the 3D model projected onto the design model (note the location of the black coal measure layers)
Figure 10: Interrogation of a particular rock unit for name and typical UCS value
4. BENEFITS AND POTENTIAL PIT FALLS WHEN USING 3D MODELS
4.1 Benefits of 3D Models
4.1.1 Expected benefits of 3D modelling can be quickly realised
1. The software provides an effective work-platform for geologists to explore a wide range of data types (in conjunction with GIS and other data management tools)
2. Although there is an up-front cost to develop an initial model, later model changes are usually quick and the effort to generate sections is reduced (i.e. there are cost savings with parts of the documentation process)
3. Communication with designers is significantly improved and where geometry becomes complex and inputs are varied the improvement in the quality of communication can be dramatic
4. Improved communication with clients regarding ground risk, potential requirements (and benefits) of additional GI
4.1.2 Benefits – largely unanticipated when we adopted 3D – can be
1. Synergy across modelling from multiple disciplines – a considerable effort is required to federate models within a project, but the benefits are significant
2. Unexpected visualisations – ‘lithology painting’ where engineering geology unit symbology is used to colour the outside surface of a modelled structure (existing or in design) is a favourite visualisation and just one of many possible when model interoperability is achieved
3. A wider reach within the project team was achieved, as ground models become available within project models. For example, planning teams quickly appreciated the addition of ground models in the project 3D environment to aid communication with external stakeholders. On CRL, a 3D-illustrated discussion about local streams and volcanoes with local ‘tangata whenua’ (people of the land) representatives was memorable.
4. On many projects there has been some opportunity to further develop ground models with data obtained during construction. This has been particularly instructive regarding communication of model reliability and uncertainty (see below). The potential benefits of a ‘whole of life’ ground model can be seen, if not quite realised at present (“BIM for Dirt” as described by a colleague from the Buildings team).
5. Availability of 3D ground models has enabled a wide range of analysis. Hydrogeological analysis is well supported by import/export to the main groundwater modelling packages, and beyond that opportunities to develop a range of applications to extract geometry and properties into other geotechnical applications exist.
4.2 Disadvanatges of 3D ground Models
Adoption of 3D ground models has not been without problems, with the key issues perhaps related to over-enthusiastic use of the models (without referencing associated sections).
Traditional sections provide an interpretation with clear documentation of investigation support (i.e. investigation strip logs with offset distance) and a well-established implicit language around uncertainty (line-style, question marks, and reliability as diagrams or in supporting text, as shown in the CRL case study in Figure 4). In 3D models, there is a coverage of GI positions, but this can easily be separated from interpretations of geometry and/or material properties (most commonly when exporting a model from one software package to another). The 3D interpretations/interpolations themselves have no integrated indication of reliability, and the visual quality of a preliminary model is indistinguishable from that of a fully developed model.
This loss of, or de-emphasis of, indications of uncertainty have led to inappropriate reliance on a surface that may be speculative or is poorly constrained by existing GI. Designers relatively unfamiliar with the process of ground model development may make requests of modellers that carry significant risk (warning requests – “just give me that surface” or “extend the area of the model to ..“). The relative ease with which models can be re-generated, re-sampled and re-distributed adds risks concerning model currency, model verification and appropriate design sensitivity to variance from the model.
Works in hand to address risks associated with 3D models include:
1. Integration of reliability information within 3D model data structures
2. Classification of models (or features within models) according to level-of-development
3. Control of model development including versioning and QA control
4. A balance of deterministic modelling and stochastic modelling of uncertainty for critical aspects of the model
CONCLUSIONS
Designing and presenting in three dimensions using 3D geological models is becoming common practice. This powerful tool allows designers, stakeholders and clients to see ‘The same thing’ in terms of geological and engineering complexity below the ground prior to design and construction.
The modelling process can also play a part in guiding where data in the form of boreholes, mapping etc is targeted. In order to achieve the best outcome in terms of a 3D geological model usefulness users might consider the following process which is similar to the Baynes et al (2005) discussion and the C25 model proposed by Parry et al (2014). This is essentially a three-staged process with hold points for review and is also potentially iterative by returning to the conceptual model phase (Figure 11). Conceptual (Stage 1) and Observational (Stage 2) stages are where the geologist/engineering geologist, assisted by the engineer is seeking out ‘truth in geology and geomorphology’ or at least the closest possible approximation of the facts. The third stage (Analytical) is where the engineer takes a leading role and is assisted by the geologist/engineering geologist following the introduction of engineering parameters into the model for design and construction purposes.
Figure 11: Suggested Geological Modelling process modified after Parry et al (2014)
There are several different types of model that can be employed today using the above geological modelling process. 3D modelling options include: photogrammetry or laser scanned models of topography for rock fall or slip hazards and 3D geological models for tunnelling, excavations or foundation design. Finally, 3D geological models can now be incorporated infrastructure models and ultimately into VR. Detailed assessment of how geological elements and their engineering parameters might interact with bridge foundations, cut batters and drainage design is now possible on a more granular scale. Caution must be used with these models as there is always an element of ambiguity. Reliability must also be assessed and conveyed to users, designers, stakeholders and clients. This field is a rapidly developing aspect of our industry, requiring some standardisation in the near future.
REFERENCES
- Baynes, F. J. Fookes, P. G. & Kennedy, J F. 2005 The total engineering geology approach applied to railways in the Pilbara, Western Australia. Bulletin of Engineering Geology and the Environment, 64, 67-94
- Fookes, P. G. 1997 Geology for Engineers: the geological model, prediction and performance. Quarterly Journal of Engineering Geology and Hydrogeology, 30, 293-424
Knill, J. L. 2003. Core values: the first Hans Cloos Lecture. Bulleting of Engineering Geology and Environment, 62, 1-34 - Parry, S. Baynes, F. J. Culshaw, M. G. Eggers, J. F. Keaton, K. Lentfer, J. Novotny & D. Paul 2014 Engineering Geological Models – an introduction: IAEG Commission 25 2014 Bulletin of Engineering Geology and the Environment, 73, 689-706
- Stapleton, T. D. 1982 Subsurface engineering – in search of a rational approach. Australian Geomechanics News 4, 26-33
