Вы находитесь на странице: 1из 17

3D GROUND MODELLING FOR A DYKE RECONSTRUCTION PROJECT

NORTH RHINE-WESTPHALIA, GERMANY

Myles Lawler, Ph.D.1


Oriol Ciurana, M.Sc.2
Aloys Kisse, Dr.-Ing.3

ABSTRACT

Owing to historic mining activities, the ground level around the cities of Haltern and
Marl in North Rhine-Westphalia (NRW), Germany has been reduced by several metres.
Lowering ground levels has had a negative effect on both water management and on
flood protection, owing to the proximity of Haltern and Marl to the River Lippe, which is
located in the Lower Rhine. New dykes, up to 14 m in height, were commissioned by the
Lippeverband (the Lippe River Authority) as part of the flood protection programme at
Haltern-Lippramsdorf-Marl (HaLiMa). The earthworks were part of the reconstruction of
existing flood barriers, allowing for a wider and shallower flow along this section of the
River Lippe. As part of the design and oversight, CDM Smith were requested by the
Lippeverband to develop a 3D ground model of the construction for both visualisation
purposes and as a means to capture a comprehensive database of the subsurface ground
conditions and earthworks for BIM purposes. The database consisted of some 275 No.
boreholes/coreholes (comprising about 2500 individual soil/rock layer descriptions),
field/laboratory tests, ground investigation and background cartographic data from the
NRW Geologischer Dienst (Geological Office), as well as high-resolution digital terrain
maps, including those from drone surveys. This paper presents a summary of how an
accurate 3D model was developed and shows how the BIM model can be used in
geotechnical practice.

INTRODUCTION

The ground level around the cities of Haltern and Marl in North Rhine-Westphalia
(NRW), Germany have been reduced by several metres owing to ongoing settlement
associated with deep underground mining. Haltern and Marl are situated adjacent to the
River Lippe, a tributary of the Rhine in the Lower Rhine basin; as such, the ground
settlement has adversely affected flood protection barriers and water management in the
region. In order to eliminate the potential for flooding and to meet the requirements of the
Water Resources Act with regard to the continuity of the river basin, upgrading of the
dykes and Lippe were commissioned by the Lippeverband (the Lippe River Authority).

As part of the flood protection programme, a 5.6 km section of river dykes (Lot 3) up to
14 m in height were constructed at Haltern-Lippramsdorf-Marl (HaLiMa). Figure 1
shows the location of the site in NRW. The earthworks, consisting of some 1.4 x 106 m3
of compacted material, were part of the reconstruction of existing flood barriers, allowing

1
CDM Smith Ltd., myles.lawler@cdmsmith.com.
2
CDM Smith Ltd., Dublin, Ireland, oriol.ciurana@cdmsmith.com.
3
CDM Smith Consult Eur. GmbH, Bochum, aloys.kisse@cdmsmith.com.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 1


for a wider and shallower flow along this section of the River Lippe. The plan area of the
site for 3D modelling was 2.563 km2. Figure 2 shows the layout in plan of the existing
and new earthworks, indicating the historical development of the dykes along this section
of the Lippe as well as fill volumes.

Figure 1. Location Plan Map.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 2


Figure 2. Development of Flood Embankments and Proposed Design.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 3


MODEL DEVELOPMENT

Database

Source Data. Geological descriptions were derived from three separate sources and
several site investigation programmes, both historic and recent. Some data (CDM
Spiekermann, 2017) were available in .mdb/.csv format that were mapped to the database
that compiles the data in .ags format. Historic data from the 2010 SI (ICG, 2010) were
inputted manually based on .dwg files generated from GeODin (GeODin, 2017). In
addition, the Geologischer Dienst of NRW provided invaluable open source borehole
data through the Bohrungen in NRW on-line portal in .mdb format (Geologischer Dienst,
2018) that were mapped to the database. Data collection and incorporation within
geotechnical BIM models in general has been described by Morin et al. (2014).

For the 3D CAD program to generate—automatically—surfaces based on the soil and


rock layers (there were about Nr. 2500 soil and rock descriptions in the final database),
detailed geotechnical descriptions were simplified; this required geotechnical
interpretation, for example interrogation of the soil log descriptions to ensure continuity
in the ground model at database level. To allow the program to recognise the primary soil
layers, the soils were converted to a stratigraphy by the assignation of a short form
code/identifier (Kurzform) corresponding to Nr. 59 pre-defined layers that the database
would recognise and link to DIN 4023 standard soil symbols. It was found that three
strata were sufficient to both generate the 3D model efficiently without the need for time
consuming manual interpretation of soil layers and to capture the bulk of the data for 3D
visualisation purposes. The iterative process is demonstrated in Figure 3.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 4


Figure 3. Three Levels of Data for use in 3D BIM Model – 1. Full Description, 2.
GeoCode_1/Legend Code, 3. GeoCode_2.

Although the level of simplification shown in Figure 3 allowed very quick 3D modelling,
it is emphasised that this approach was project specific for the following reasons:

1. The ground conditions were straightforward, consisting typically of an upper fill/made


ground (Auffüllung), overlying predominant alluvial sands (Alluvialen), overlying
mudstone/Marl (Mergelstein);

2. Construction was, for the most part, limited to dyke construction rather than heavy
loads/settlement-sensitive foundations, for example.

For more complex geology, or for construction projects that require complex 3D surface
representation of the stratification (e.g. foundations sensitive to small strains), far greater
numbers of ground layers can be incorporated at the expense of time and effort allocated
to manual interpretation at database level to ensure both consistent geology and the
avoidance of attribute clash in the 3D model.

Note that all three levels were captured in the BIM model, whereas for the visualisation,
the GeoCode_2 field (.ags file format) was used. In addition to borehole data, water
levels were recorded as point data within the CAD model and linked to form a continuous
piezometric surface integral to the 3D model.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 5


Within any SI programme, there are inevitable gaps in the geological record; this was
equally true in the case of the river basin in this paper, owing to the highly non-linear
river course and irregular spacing of the boreholes. Accordingly, WMS datasets were
accessed from the ATOM feed of the Geologischer Dienst NRW (Geological Office),
namely 1:50000 soil maps (‘Bodenkarte’ from ground level to 2 m depth) and 1:100000
geology maps (Geologische Karte); these maps were used to define geological
boundaries in plan and to group the upper soil layers as distinct overburden zones.

Figure 4 shows the correlation of the borehole data against the soil cover map at 1:50000
scale; in this example, the soil maps are compared to borehole records that recorded
SAND at the surface; this map demonstrates the improvement in ground model detail that
can be gained by using polylines generated from the geological maps, in the upper 2 m or
so of the ground model, compared to relying solely on borehole coverage.

Borehole survey data were supplemented with an open source Digital Terrain Model
(DTM) at 20 m resolution (derived from laser scans and covering the study area) and two
specially commissioned high resolution Digital Surface Models (DSMs) at 0.1 m spatial
resolution (July 2016 and June 2018); these datasets covered the north side of the river
where the new dyke being constructed and allowed tracking of construction progress.

Figure 4. Use of Background Geological Maps Correlated against Borehole Data to


Improve Ground Modelling.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 6


In addition to the use of geological background templates, it is possible to optimise the
3D model by editing profiles in 2D and back-calculating the 3D model based on the
edited profiles; however, it was found that the river basin, characterised by highly non-
linear topographical features and gaps in sub-surface data, resulted in this being an
inefficient approach. Instead, automatic contour interpolation between boreholes was
used to generate surfaces between boreholes.

The database of sub-surface descriptions was supplemented with Nr. 63 DPH (dynamic
probe) and laboratory tests, the latter including direct shear and soil classification tests, as
well as Nr. 1382 Proctor compaction tests that were used for end-product embankment
compaction testing4. These data were stored and formatted within the 3D model. The
output is discussed in the ‘Use of Model’ section of this paper.

Simplified 3D Model

Surfaces. Borehole survey data (Nr. 275 points) were supplemented with open source
DTMs (of 20 m maximum resolution), used mainly for areas at a distance from the
earthworks and specially commissioned high resolution DTMs from drone surveys. The
DTM at 20 m resolution was used as a surface to define the top level of the 3D model and
was found to provide the optimum combination of spatial coverage, topographic detail,
and computation time. The aim was to provide the model with both a good combination
of visualisation (for the existing topographic features) and geological data from boreholes
and geological maps. Although borehole survey data points provide accurate levels where
boreholes are located, these are insufficient to create a continuous representation of the
terrain over the study area; for this reason, high resolution DSMs of the area where the
new dyke was being constructed were commissioned. High resolution surveys were
incorporated into the 3D model to override the 20 m resolution DTM at the location of
the new dykes were being constructed (these DSMs were very high resolution 10 cm cell
size over an area of 256 ha). However, in order to improve computation time and increase
efficiency when generating the 3D model, it was necessary to reduce the DSM resolution
from 10 cm to 5 m. In addition, working with the DSMs adversely affected terrain
elevations; unlike a DTM, a DSM captures the elevation of all types of natural and built
features which may not represent terrain elevations (e.g. tree canopy, buildings, and
powerlines); therefore, the files were checked for inconsistencies that would correspond
to these features.

The top and base elevations of the three-layer model (strata A, B, and C – see Figure 3)
were used to generate the surfaces required to create the 3D solids that defined the 3D
model. As the boreholes (275 No.) were distributed over the site in an irregular pattern,
occasionally in clusters, it was considered that spatial interpolation would be more
efficient than manual editing in the generation of the ground profile between the
boreholes. The interpolation method used in this project was the natural neighbour
algorithm that finds the closest subset of input samples to a query point and applies
weights to them based on proportionate areas to interpolate a value (Sibson, 1981).
4
Einbaukontrolle.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 7


In addition, the borehole data points were supplemented with Nr. 931 data points created
over a 100 m point grid. Levels were then interpolated at these data points using a GIS.
Table 1 summarises the process behind the generation of the surfaces.

Table 1. Generation of Surfaces in the 3D Model.


Stratum Upper/ Description
Lower
A Upper Levels correspond to the DTM. Borehole top elevations were
overridden by surface elevation from the DTM over the study area. In
the case of borehole elevations greater or less than the surface
elevation, the difference in elevation was assumed as ‘cut’ or ‘fill’,
respectively
Lower Surface defined using Stratum A base levels at boreholes (where
applicable) and Stratum B top levels at boreholes where Stratum A
was not present/defined.
B Upper Surface defined using Stratum B top levels at boreholes (where
applicable) and Stratum A base levels at boreholes where Stratum B
was not present/defined. It was assumed that Stratum B was located on
the surface where indicated by Bodenkarte 1:50000 (background
geological map); in this case, the levels from the DTM were taken.
Lower Surface defined using Stratum B base levels at boreholes (where
applicable) and Stratum C top levels at boreholes where Stratum B was
neither present nor defined.
C Upper Surface generated using Stratum C top levels at boreholes (where
applicable) and Stratum B base levels at boreholes where Stratum C
was not present/defined.
Lower Surface set at elevation 0 m, corresponding to the base of the model.

Level of development. The Level of Development, LoD, is defined as the amount of


geometric and other information given to a BIM object; in this paper, the LoD calculation
includes the former, i.e. those data that influence the shape of the 3D model; for example,
a borehole elevation would be included in the calculation but not a soil test result from a
sample in that borehole. The LoD of the 3D model was estimated using the following two
parameters:

1. The number of borehole data points, P, in a grid consisting of 200 m × 200 m squares
contains a combination of borehole points and data points, the latter created by the
interaction of a borehole elevation data point with the DTM;

2. DTM resolution, R.

In order to calculate LoD, first a 200 m square grid was overlayed on a PLAN of the area
in ArcGIS (2016), including the borehole points, p1. As the topography created by the
DTM supersedes the elevations from borehole records, a new data point, p2, is created in
the event of an elevation difference, as indicated by the most up-to-date DTM5.

5
p2 is not assigned equal importance to p1.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 8


For the DTM parameter, R, a value of 1 is given to grid squares defined entirely by the
maximum DTM resolution used in the model, in this case 5 m; lower resolution DTMs
are given a weighting according to the resolution; however, as the DTM by definition
only affects the surface layer, a further reduction factor is applied to R; i.e. where a grid
square was defined by DTMs of different resolution, the relative areas defined by the
different DTMs within the square were calculated in ArcGIS (2016) and a weighting
applied based on the areas, A5 and A20.

The LoD is calculated from Equation (1)6:

×
= (1)

Where , , and f is a reduction


factor; and , where s represents the
number of surfaces described by a data point and St is the total number of surfaces in the
model. Although in this paper two DTMs of different resolution and 5 surfaces were
used, the equations could be extended to any number of DTMs/surfaces.

Finally, the grid square with the highest LoD was found and the remaining squares
normalised with respect to this value to give a confidence value from 0 to 100. Using
Equation (1), a LoD was then assigned to each square in the grid, as shown in Figure 5a.
An example LoD calculation for grid square D10 is given in Figure 5b.

6 Note that P is multiplied by R, as it was considered that an area without any subsurface information
should be assigned a LoD of zero.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 9


Figure 5a. Grid demonstrating Degree of Confidence in the Level of Development, LoD
(grid square area 40000 m2).

Figure 5b. Example LoD Calculation for Grid Square D10

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 10


USE OF MODEL
Background

The considerable lead-in time in developing a comprehensive 3D model compared with


traditional methods of understanding ground conditions (for example, manual
interpretation of 2D cross-sections from abridged borehole data) means that this
development time should be offset against future time savings and added value. This
section describes some of the applications of the model that add value and improve
project efficiency; model output has been divided into 3D visualisation and BIM.

Visualisation

Figure 6a and 6b shows oblique perspectives of the 3D model. The overburden


distribution is controlled by a combination of the geological maps, borehole data, and
digital terrain modelling; the resulting surface overburden distribution is compared to the
overburden distribution from the geological maps (0 m to 2 m depth) in Figure 6a,
indicating good correlation with the model. Figure 6b demonstrates the layering and
DTM/DSM resolution (note the high resolution area in the top right representing the new
embankments).

Figure 6a. 3D Model Perspectives Showing Correlation with 1:50000 Soil Map Overlay
(Polylines).

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 11


Figure 6b. 3D Model Perspectives Showing Soil Layers and DTM/DSM Resolution.

BIM

Figure 7 and Figure 8 shows example cross-sections taken from the BIM; the 2D cross-
section demonstrates the ability of the model to capture and present technical drawings
(in this case to DIN 4023); useful data that can be automatically displayed in a drawing
include soil layers, soil/rock symbols, soil samples/specimen reference numbers, field
and laboratory test results, and interpreted layers between boreholes; these details are
demonstrated in Figure 7 and Figure 8 in the case of 2D and 3D, respectively.

Figure 7. 2D Cross-Section Generation from 3D Model.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 12


Figure 8. Displaying Borehole Data in 3D.

Applications

Cross-sections for use in design and reporting. Templates of soil symbols/hatches were
created in both the database and within Civil 3D. These templates recognise the
GeologyCode_1 field codes in .ags format (see Figure 3) and assign the equivalent DIN
4023 symbol to the soil layer. Any cross-section drawn within the 3D model will
generate a 2D technical drawing that can be used in geotechnical design; this is
considered very useful for geotechnical practitioners in foundation design.

NURBS/3D Faces/2D Polygons. 3D surface areas can be exported as 3D faces (as a .dxf
file) to 3D FE programs; the surfaces can then be extruded to create 3D FE geometries,
automatically from any area of the model; in addition, objects linked to the geometry can
be extracted, for example boundaries, points, and surfaces. An example of a geometry
generated in this way is demonstrated using PLAXIS FE code (PLAXIS, 2018) in Figure
9. In addition, 2D cross-sections can be generated using the automatic soil and water
level interpretation (described in this paper) that can be used directly in 2D analyses, as
shown in Figure 10.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 13


Figure 9. 3D FE Geometry Generated Automatically in PLAXIS 3D.

Figure 10. 2D FE Geometry in PLAXIS 2D Generated from the 3D Model.

Embankment Construction Monitoring (Einbaukontrolle). Importing additional borehole


data can be handled automatically in the model, improving both topographical and
ground layer/piezometric surface resolution. In addition, by incorporating up-to-date
topographical surveys at regular intervals, construction progress can be tracked,
especially if high resolution surveys are used; in this case, accurate assessments of cut/fill
volumes and areas can be made using the model. The use of the model for earthworks
calculations is demonstrated in Table 2; the data in Table 2 are depicted in Figure 11.

Another important Einbaukontrolle application is 3D visualisation of material suitability


testing for end-product compaction. Figure 12 shows laboratory (Proctor) compaction
tests conducted on samples taken from the eight sections (Baubereich) of the flood
embankment (for example, embankment core, Stützkörper Nord I and compensation
layer, Ausgleichschicht Nord I); red is a compaction (Proctor) test failure and green
indicates that the test met the minimum (95% of optimum) requirements7.

7
The end-product material densities were compared to the minimum dry density values required for each
embankment section to generate groups of test results by section. The results were included in the Civil 3D
model as georeferenced points with an associated compaction test Location ID.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 14


Table 2. Cut/Fill Volumes Calculated from 3D Model.
Baubereich Area (ha) Cut volume (m3) Fill volume (m3)
Nord Baubereich 1 5.03 151,015
Nord Baubereich 2 11.91 40,103
Nord Baubereich 3 3.23 70,780
Nord Baubereich 4 6.07 24,504
Nord Baubereich 5 0.90 1,809

Figure 11. Net fill/cut volumes of construction areas (Baubereiche) in PLAN.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 15


Figure 12. End-Product Compaction Test Results (green/red corresponding to pass/fail,
respectively) in PLAN and 3D.

CONCLUSION

3D BIM ground models can improve the visualisation of construction, crystallise large
amounts of disparate site investigation data, and provide a basis for improved design
workflows and construction visualisation. However, a sound understanding of
geotechnical conditions remains as important as ever should a comprehensive and
accurate 3D model be developed. For example, it was found that rational ground models
could not be generated automatically based on raw SI data and a very strong
understanding of the different data types imported to the database was required to ensure
correlation of the model with both known geology and accurate DTMs. Nevertheless, it
was found that great potential exists for 3D ground modelling to improve the efficiency
of geotechnical design, including FE modelling and for tracking earthworks progress.
Future improvements to the HaLiMa model include the export of all data in .ifc format
from the 3D model as property sets to improve interoperability with other BIM programs
and additional topographic and precise level surveys to define features such as roads,
rivers, water features, and embankments; in addition, more complex soil stratigraphies
and embankment structure will be included within the 3D model.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 16


ACKNOWLEDGEMENTS

The authors would like to acknowledge the support of Ilja Prinz and Albano Haruni of
CDM Consult Europe GmbH, Bochum for their invaluable input, including proof reading
of the paper, data procurement, software support, and with developing the LoD section of
the paper. The authors would, in addition, like to acknowledge the Lippeverband for
allowing the HaLiMa – Los 3 data to be published and for their assistance throughout the
project. Finally, Stephan Bach of the Geologischer Dienst NRW is thanked for compiling
the open source borehole data.

REFERENCES

ArcGIS (2016) ArcGIS Desktop 10.5, Esri Inc.

CDM Smith (2017) HaLiMa Deichrückverlegung bohrungung (Field Investigations in the


Construction Areas, in German).

DIN4023_v3 (2012) CADsys Vertriebs- und Entwicklungsgesellschaft mbH.


Schraffuren-DIN4023-V3.

GeODin (2016) 8.3 Release GeODin SITE.

Geologischer Dienst (2018) www.bohrungen.nrw.de (accessed April to July, 2018).

Gary Morin, Steve Hassall, and Roger Chandler. 2014. "Case study – The real life
benefits of Geotechnical Building Information Modelling." In: D.G. Toll et al., Eds.,
Information Technology in Geo-Engineering, IOS Press, 95-102.

ICG (2010) Planungsgemeinschaft HaLiMa, Sondierprofiles (Geotechnical Reports for


Project HaLiMa, in German).

PLAXIS (2018) 2D and 3D FE code.

R. Sibson. 1981. "A Brief Description of Natural Neighbor Interpolation." In: Barnett, V.,
Ed., Interpreting Multivariate Data, John Wiley & Sons, New York, 21-36.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 17

Вам также может понравиться