Smart Site Investigations Save Money!

Martin Th. van Staveren1 and Adriaan J. van Seters2

1 GeoDelft, Stieltjesweg 2, 2628 CK Delft, The Netherlands
m.th.vanstaveren@geodelft.nl
2 Fugro Ingenieursbureau BV, Postbus 63, 2260 AB Leidschendam, The Netherlands
a.vseters@fugro.nl
Tel: +31 15 2693583
Fax: +31 15 2610821

Abstract. More appropriate and well-timed site investigations have a major positive impact
in order to reduce the risk of failure. Design and construction of any infrastructure project
can be optimised within the project specifications by smart site investigations, in order to
save money and time. In this paper this statement is supported by three recent Dutch infra-
structure projects, in which smart site investigations resulted in major cost and or time
savings. A number of generic key success factors are derived from the presented case histo-
ries. These success factors are a risk driven approach of the project, challenging existing
codes of practice by state of the art techniques, true consideration of the impact of geologi-
cal heterogeneity on the project and last but not least, application of hard and soft experi-
ence data. These aspects can be defined as the characteristics of smart site investigations.
They flourish in particular when applied in combination. In conclusion, more attention
towards the strategy and quality of site investigations pays off. It is therefore considered as
the responsibility of the European engineering geological and geotechnical communities to
send this message with clear examples to the owners, designers and constructors of infra-
structure projects. Finally all stake holders of these projects, owners, contractors, engineers,
insurance parties and the European societies as a whole, will benefit from the strength of
smart site investigations.
Keywords: site investigation, costs, degree of risk, cone pressuremeter, liquefaction risk
control.

1 Introduction
The strategy and quality of site investigations for infrastructural projects needs
much more attention in the day-to-day practice. In the Netherlands, and probably
also in most other European countries, there is unfavourable tendency to buy site
investigations on the cheapest price criterion only. The owner provides a shop-
ping list with a number of in-situ and laboratory tests, often with minor or even
no context of the project requirements. The bidder with the lowest price wins the
contract, and apparently, everybody is happy
As figure 1 from Molendijk and Aantjes (2003) demonstrates, the failure costs
in the Dutch infrastructure construction industry are high. In each of the projects
presented in table 1 the subsoil played a major role. Often the lowest cost only site
investigations, without a well-defined strategy and degree of quality, are not the
cheapest solution for all parties involved, at the end of the project.

Robert Hack, Rag Azzam, and Robert Charlier (Eds.): LNES 104, pp. 792800, 2004.


c Springer-Verlag Berlin Heidelberg 2004
 Smart Site Investigations Save Money! 793

Table 1. Planning an cost overruns of a selection of Dutch infrastructre projects.

Project Overrun
Planning Costs

Ramspol inflatable dam + 80 % Unknown

Tramtunnel The Hague +100 % + 90 %
Stormsurge barrier Nieuwe Waterweg + 50 % + 40 %
Betuweroute freight railway unknown + 65 %
Splay railway Amsterdam Utrecht + 65 % + 25 %

2 Why Site Investigations at All?

Anyone involved in design and/or construction of infrastructural projects will

understand that any form of site investigation is required, in order to arrive at an
effective design and construction. Nothing new so far. Probably less well-known
is the fact that about 85 % of all ground related problems are directly related to the
extend and quality of site investigations. Clayton (2001) demonstrates a number of
types of ground-related problems, such as ground geometry not as anticipated and
inadequate planning or execution of the site investigation. The majority of these
problems can at least be reduced by adequate site investigations. Typically costs
of site investigations are less than 1 % of the construction costs (Smith 1996). Site
investigation costs of 0.2 to 0.5 % are typically considered as adequate in the
construction industry (Knill 2003). Groundrelated problems can easily add 5 % to
the cost of construction, while figures as high as 30-50 % are not unheard of. If
real severe unforeseen ground conditions are encountered during construction,
additional costs as high as 100 % of the entire project price may be incurred
(Clayton 2001). It can be even worse. A study by Hoek and Palmieri (1998), in-
volving the study of 71 World Bank hydroelectric projects indicated not only that
costs and schedules were in general 25 % higher than originally estimated, but als
that projects have been stopped and abandoned entirely. For these projects cost
and schedules escalated to several times the original estimates, due to unforeseen
geological conditions. As summarised by Fookes (2000), many similar publica-
tions demonstrate that unforeseen ground conditions are one of the largest drivers
of technical and financial risk in civil engineering projects. The combination of
these figures create an enormous potential to demonstrate the added value of ade-
quate site investigations. What would you decide, if for instance 5 times more site
investigation activities (2.5 % instead of the conventional 0.5 % of the total con-
struction costs) would reduce a 25 % cost overun by a factor 5 to 5 % cost overun
only? Depending on the size of the project, the absolute cost benefit ratio can
easily be 1 : 10 or higher. This nice cost benefit ratio is demonstrated in this
paper by a number of recent cases in the Dutch construction industry. However,
firstly the smart site investigation concept needs an introduction.
794 Martin Th. van Staveren and Adriaan J. van Seters

3 The Smart Site Investigation Concept

The smart part of the smart site investigation concept is its simplicity. Smart site
investigations are very project and site specific and risk driven. It is a matter of an
open mind and creative holistic thinking. The smart site investigation concept
means basically that there is no one receipt for site investigation, as well as there
are no general valid rules, guidelines and standards. Every infrastructural project
is unique, therefore every project needs its own specific approach. Sometimes a
limited site investigation with just some CPTs will be sufficient, other times very
sophisticated and multi-phased site investigations are required, to meet an accept-
able level of certainty about the ground conditions. Basically, the type and extent
of a smart site investigation depends on three main factors:
1. the expected geological heterogeneity of the subsoil
2. the boundary conditions of the construction
3. the degree of risk the responsible party accepts to bear
The last factor, the risk factor, should be the main driver in the definition of the
site investigation programme, as the first two factors depend largely on risk accep-
tance. A main and common pitfall is the mindset that a risk driven approach is
inherently difficult and complex, as for instance mentioned by Smith (1996) and
Ho (2000). Risk management does not automatically involve very difficult prob-
abilistic calculations and a jungle of cause and effects diagrammes. Risk manage-
ment starts with common sense and well-structured thinking about the main pro-
ject goals to achieve, the probability of unfavourable events with might obstruct
those goals and the effects of those events on the goals. Based on this type of
analysis, which can be very brief for small projects and extensive for large com-
plicated projects, effective risk remediation measures can be taken. These meas-
ures will almost automatically define the required site investigations and the smart
site investigation concept has been applied! Next the application of this smart site
investigation concept, together with its benefits will be demonstrated by three
studies.

4 Smart Site Investigation Case Studies

4.1 Observational Method Avoids Sheet Pile Wall

The Betuweroute is a 160 km double-track freight railway. It should ensure that

cargo arriving at the port of Rotterdam is transported quickly and safely to the
European hinterland. The section between the cities of Sliedrecht and Gorinchem
is the most difficult part of the new railway track. It runs parallel to sensitive exist-
ing infrastructure, available construction time is short and critical for the total
project and last but not least, local soil conditions are very poor. The alliance of
the owner and a consortium of contractors was challenged to go beyond traditional
design standards, because covering all soft soil uncertainties with a traditional type
of design would cost many millions of euro extra in advance. One example is the 9
 Smart Site Investigations Save Money! 795

km of embankment section, where Betuweroute runs parallel to the existing rail-

way Dordrecht-Geldermalsen and the Motorway A 15. Originally, to minimise
disturbance of the existing rail track, a sheet pile wall was planned in between the
two embankments of the existing and new rail tracks. However, a risk analysis at
the start of the alliance provided interesting new insights. The combination of
the observational method, supported by a smart site investigation, proved to
be a much better cost-effective remediation measure for the deformation risks
(Molendijk and Aantjes 2003). The site investigation involved on-line monitoring
during the construction by open standpipes and vertical inclinometers every 500
m. Settlement plates were installed every 50 m. At geologically relevant cross
sections, were heterogeinity was expected, vibration wire piezometers were used.
A specific aspect of the site charactarisation by monitoring is the on-line avail-
ability of the monitoring results. All monitoring data was sent immediately to a
database, which was accessible via internet for all authorised parties (Molendijk
and van den Berg 2003). The presented observational method, supported by a
smart site characterisation, avoided the requirement of 9 km of sheetpile, which
saved the alliance about 4 million euro. The additional monitoring costs were
about 0.2 million euro, which gives a cost benefit ratio of 1 : 20.

4.2 Smart Liquefaction Risk Control

One of the most densely populated parts of the Netherlands is underlain by the
worst soft soil conditions of the Netherlands. In this Rotterdam The Hague area
a light-rail link will be constructed, RandstadRail. Part of this link will be realised
in a bored tunnel, which will cross the existing and very busy railway Rotterdam -
Utrecht at 14 m depth. During the risk analysis in the design phase, with input of
experience data and historical data, it became clear that the existing Rotterdam
Utrecht railway was constructed on 14 m of very loose anthropogene sand, which
pushed away the originally very soft peat layers. Figure 1 demonstrates a cross
section with soil profile, the proposed tunnel, and the CPT result.
The risk of large soil settlements above the proposed tunnel, affecting the exist-
ing Rotterdam Utrecht railway, was classified as very high. A temporary bridge,
supporting the tracks of the existing railway was proposed as risk mitigation
measure. However, during the risk evaluation it became clear that this measure
was not only very expensive, it would be needed to take the existing railway out of
duty for a couple of weekends as well. In addition, an additional settlement risk
during installation of the temporary bridge was identified. Therefore, another
settlement risk mitigation measure was highly required. In this stage, the benefits
of a smart soil characterisation became clear. The CPT shows cone resistances of
about 3 MPa until a depth of more than 10 m. But how loose would that sand layer
really be? To answer this question a detailed and non-routine additional smart site
investigation was performed. The test programme included 3 in-situ electrical
density measurements, 3 CPTs and a boring. All tests were performed by CPT
equipment and pushed by an angle of inclination of 30 to 45 degrees below the
existing railtrack. Figure 2 demonstrates the performance of the electrical density
test in the field.
796 Martin Th. van Staveren and Adriaan J. van Seters

qc (Mpa)
0 5 10 15
embankement
0

Peat 4
Anthropogene Sand
6

10

12
Gorinchem clay
14

16

Sand 18

20

Fig. 1. Cross section with soil profile and CPT

Fig. 2. Electrical density test below railway.

With special cones the electrical density method measures in-situ and continu-
ously the electrical conductivity of the soil and the groundwater. The results are
calibrated by laboratory tests. The results of the additional site characterisation
demonstrated quite better soil conditions, than could be concluded based on CPTs
 Smart Site Investigations Save Money! 797

only. The temporary bridge support was not needed anymore. Adding extra weight
adjacent to the railway, in combination with a limited lowering of the groundwater
table, was selected as an effective risk mitigation measure (Korff 2003). The risk-
driven and non-routine site investigation saved a significant amount of money on
risk remediation measures, by avoidance of the temporary bridge. The costs of the
additional additional site investigation, including the laboratory tests was much
lower than the savings. The cost-benefit ratio of the smart site investigation for
this case is 1:20.

4.3 Essential Soil Stiffness Investigation for the Hubertus Tunnel

In The Hague plans are developed for the execution of a bored tunnel as part of
the Noordelijke Randweg. The tunnel consists of two 8,65 m inside diameter
tunnels with a bored length of 1600 m each. At both ends a shaft and a cut-and-
cover section are foreseen. The location of the tunnel is near the North Sea coast,
therefore the subsoil conditions consist mainly of sands. The surface level varies
around NAP + 7 m, with at the western side the large (artificially constructed)
Hubertusdune with top at NAP + 23 m. A 1 m thick peat layer is present over
almost the entire tunnel area at NAP 0 m level. The groundwater level is around
NAP 0,0 m. The tunnel itself will be constructed horizontally between NAP 5 m
(top) and NAP 15 m (bottom). At the tunnel entrances the road level will be at
NAP 0 m. The stiffness of the soil is of great importance for dimensioning of the
tunnel lining and therefore largely determines the tunnel cost. This case will focus
on a smart determination of soil stiffness, which was a major topic in the third
phase of the soil investigation in 2003. In this phase the soil investigation con-
sisted of 36 CPTs, 13 boreholes and 31 Cone Pressiometer tests at 7 locations. In
the first two phases already 45 CPTs and a limited number of borings were per-
formed. In the design of the tunnel lining the (un)loading stiffness of the sand is a
key parameter, as the forces in the lining depend highly upon the soil stiffness. In
the first phase of the design, these stiffness values were estimated based upon
literature. In order to maximise the reliability of this crucial parameter during the
third phase of the soils investigation, two methods of determination were applied:
CD-triaxial tests with unloading loop on reconstituted samples and in situ deter-
mination by Cone Pressuremeter testing. In total 23 CD Triaxial tests with unload-
ing phase (triaxial extension test) and 31 Cone Pressuremeter tests were per-
formed. The triaxial tests were multistage consolidated drained tests on isotropi-
cally consolidated samples. The samples were prepared on the in situ density,
using the results of minimum/maximum density determinations and correlations
with the cone resistance qc from literature. As the unloading stiffness proved to be
essential in the design, a direct determination of the parameter was included to
minimise the uncertainty in the value. Thus, an unloading stage was added after
each loading phase, as shown on figure 3. An example of the load/unload curve of
a Cone Pressuremeter test is given on figure 4.
798 Martin Th. van Staveren and Adriaan J. van Seters

Fig. 3. Triaxial test result.

7.2

6.4

5.6

4.8

4.0
Pressure [mpa]

3.2

2.4

1.6

0.8

0.0
0 100 200 300 400 500 600 700 800

Volume [ml]

Fig. 4. Cone Pressuremeter test result.

From both tests stiffness parameters E50 and Eur were derived for all sand layers.
Caution must be taken in the interpretation of the tests. The triaxial tests during
loading were performed at lower strain levels than the CPM tests, whereas the
unloading occurs at lower strains for the CPM tests than for the triaxial tests.
Based on these differences in strain level, it was found that both test methods were
 Smart Site Investigations Save Money! 799

consistent. Finally mean, low characteristic and high characteristic values for the
loading stiffness E50 and unloading stiffness Eur were derived statistically, as
shown on figure 5.

12
Test Results
Belotti, Lower Boundary
10
Belotti, Average
Belotti, Upper Boundary
8
Macht (Test Results)
E/qc 6

0
0 500 1000 1500 2000 2500 3000
qc/sqrt(vert. effect.

Fig. 5. Statistically derived loading stiffness E50 and unloading stiffnes s Eur.

As can be observed from figure 5, the stiffness values derived from the tests are
approximately a factor 2 larger than the well used relation by Bellotti et al. (1989).
This proved very favourable in the design. The mean values were found to be as
follows related to the average cone resistance and the vertical effective stress in
the layer:
0 , 53
E50 q
= 189 c and: Eur = 4 * E50 (1)
qc
v
where: E50 = Youngs modulus at 50 % deviator stress in triaxial test; Eur = Youngs
modulus at unloading/reloading conditions; qc= average cone resistance in the
layer; v = vertical effective stress. By the presented correlation statistically
based stiffness parameters, based on two types of tests, can be derived at every
CPT location for the tunnel. This smart soils investigation was characterised by a
step by step approach, which gradually per phase zoomed into more detail on one
of the most crucial design parameters, i.e. the soils stiffness of the sand layers. The
stiffness was derived from two independent test methods, laboratory and in situ
testing. A large number of tests was carried out, to enable statistical evaluation of
the data, which resulted in a factor two larger stiff ness than would be expected
from literature correlation only. The investment in state of the art testing resulted
therefore clearly in a more economic lining design.
800 Martin Th. van Staveren and Adriaan J. van Seters

5 Conclusions

In total four generic key success factors for the smart site investigations concept
are derived from the briefly presented individual case histories. These success
factors are: 1) a risk driven approach of the project, 2) challenging existing codes
of practice by state of the art approaches and techniques, 3) true consideration of
the impact of geological heterogeneity on the project, and 4) application of hard
and soft experience data. These aspects can be defined as characteristics of smart
site investigations. They flourish in particular when applied in combination and
result in cost : benefit factors of 1 : 20 in the presented cases. In conclusion, more
attention towards the strategy and quality of site investigations pays off. It is there-
fore considered as the responsibility of the European engineering geological and
geotechnical communities, to send this message with clear examples to the own-
ers, designers and constructors of infrastructure projects. Because finally all stake-
holders of these projects, owners, contractors, engineers, insurance parties and the
European societies as a whole, will benefit from the strength of smart site investi-
gations.

References

Bellotti R, Ghiona V, Jamiolkowski M, Robertson PK (1989) Shear strength of sands from

CPT. In: Proceedings 12th International Conference on Soil Mechanics and Foundation
Engineering, Rio de Janeiro.
Clayton CRI (ed) (2001) Managing geotechnical risk: Improving productivity in UK build-
ing and construction. The Institution of Civil Engineers, London, 80 p.
Fookes PG (2000) Total geological history: A model approach to the anticipation, observa-
tion and understanding of site conditions. In Proceedings EngGeo 2000, Sydney, Austra-
lia.
Ho (2000) Quantitative Risk Assesment: application, myths and future direction. In: Pro-
ceedings EngGeo 2000, Sydney, Australia.
Hoek E, Palmieri A (1998) Geotechnical Risks on large civil enginerring projects. In: Pro-
ceedings 8th Congres IAEG 1:79-88.
Knill J (2003) Core values: The first Hans-Cloos lecture. Bulletin of Engineering Geology
and the Environment, vol 62, 1:1-34.
Korff M (2003) Crossing of bored tunnel with existing railway. In Proceedings of 2nd
International Young Geotechnical Engineering Conference, Bucharest, Romania.
Molendijk WO, Aantjes AT (2003) Risk management of earthworks using GeoQ. In: Pro-
ceedings of Piarc XXII World Road Congress, Committee C12, Durban, South Africa.
Molendijk WO, Berg F.P.W. van den (2003) Well-founded risk management for the Betu-
weroute freight railway by the observational method. In: Proceedings of XIIIth Euro-
pean Conference on Soil Mechanics and Geotechnical Engineering, Prague,
Czechoslovakia.
Smith (1996) Allocation of risk The case for manageability. The International Construc-
tion Law Review, 4:549-569.