Deep Construction Pit and Combined Pile-Raft Foundation

for a High-Rise Building at the Red Sea

Alexander Mühl, Michael Brunner, Christian Wawrzyniak

CDM Smith Consult GmbH

A 260m tall high-rise building is to be built for a hotel complex in Jeddah, which is the pilgrims "gateway" on their
pilgrimage to Mecca. The building ground consists of coralline limes stone of various formations. It is has largely
preserved the unique skeletal structure of the coral reef which makes it very erratic. The groundwater level of the
construction pit, which is located directly at the Red Sea, lies 10m above the excavation level. The Retaining and
Shoring System consists of a diaphragm wall and a low-lying grouted sealing blanket. The foundation of the building
is implemented by means of a three-meter thick reinforced concrete slab based on 125 large-diameter bored piles,
which reach a depth of 50m.
The first paragraph of this article is dedicated to the geotechnical planning focusing on the application status of the
FEM. The application and the effects of various soil models on the calibration of soil parameters and the modeling of
piles by means of embedded piles will be discussed. The second section will describe the special production features
regarding the low-lying grouted sealing blanket and the 50m deep piles in the coralline limestone. A special aspect is
the use of micro-silica as additive to the bored pile concrete which is highly resistant to sulfate-bearing water.

1. Introduction
The “Kempinski Hotel & Residences,” a project of the
ASSILA Investment Company, will be located on the
North Corniche of Jeddah about 100 m inland from the
Red Sea. On 1,800 m² of the 9,600 m² sized property a
high-rise building with 65 floors and 259 m height will
be realized. The high-rise building will be founded on a
3 m thick raft slab with 2,900 m² area size and 125
piles.
Figure 2 Coral Rock Drill Core
In the first construction phase, the building pit for the
three basement floors had to be built. The shoring sys-
tem consists on a 400m long and 20m deep, single-
layer anchored diaphragm wall and a low-lying grouted
sealing blanket with a planned thickness of 4m. The
grouted sealing blanket serves to reduce the groundwa-
ter inflow from 9000m³/h to remaining 1,800m³/h,
which was discharged to the Red Sea via 30 deep wells
Figure 1 Kempinski Hotel & Residences, Jeddah, and two discharge lines with a diameter of 800mm.
(www.perkinswill.com)
The foundation soil is composed of coralline limestone
of widely varying strength, as well as silty sand derived
from this formation. The coralline limestone has largely
preserved the unique skeletal structure of the coral reef
which makes it very erratic. Apart from its varying
strength, the coralline limestone is characterized by
irregular cavities, which pose a special challenge to the
foundation of the high-rise building.
 design the analysis had to be revised. As the existing
FE-Model was extremely simplified and the adjust-
ments would have been very time-consuming we decid-
ed to create a completely new model. For this purpose,
we used the calculation program of Plaxis bv. Plaxis
3D-V2012. Soil and structure are modelled by means
of 10-node tetrahedral elements. This type of element
provides a second-order interpolation of displacements.
To obtain the force, the stiffness matrix and the mass
matrix of the tetrahedral elements are integrated with
four integration points. The quality of the 10-node
element is comparable to a 20-node brick element.
The direct three-dimensional continuum finite element
modelling is computationally very expensive. Some
simplifications have been made to enable us to simulate
the complex pile interaction, and the interaction be-
tween soil and structure. Instead of modelling 267
foundation piles by solid elements piles have been
modelled as embedded piles. The embedded pile option
has been developed to describe the interaction of a pile
with its surrounding soil. The interaction at the pile
skin and at the pile foot is described by means of em-
bedded interface elements. The pile is considered as a
beam which can cross the 10-node tetrahedral element
at any place with any arbitrary orientation. Due to the
existence of the beam element three extra nodes are
introduced inside the 10-node tetrahedral element. In
2 Foundation Design - Combine Pile-Raft Foun- order to ensure that a realistic pile bearing capacity as
dation specified can actually be reached, a zone in the soil
volume elements surrounding the embedded pile is
For the structural design a linear static analysis of a 3D-
identified as elastic zone. The size of this zone is de-
structure was performed. In this Model the tower foun-
termined equivalent to the pile's radius. The elastic
dation is considered as linear-elastic by the application
zone makes the embedded pile behave almost like a
of spring bearing supports. The stiffness of the spring
volume pile.
bearings have been defined in result of numerical anal-
yses using a 3D-Finite-Element-Model, in which soil,
A 3D-CAD model representing the structural elements
piles and raft slab are modeled with first order solid
of the basement floor relevant for the load transfer was
elements of hexahedra (brick) shape. To estimate the
created for the implementation of the foundation struc-
influence of cavities on the bearing behavior of the
ture. (Fehler! Verweisquelle konnte nicht gefunden
piled raft an element cluster with lower stiffness has
werden.). The 3D-CAD model was integrated in
been modeled below the base of a group of piles.
PLAXIS and meshed.
The numerical analyses were performed in the early
stage of the design. The Structural Designer applied
lowest and highest pile stiffness, while the position of
the piles was adopted to optimize the distribution of
forces iteratively by running several calculation pro-
cesses, which lead to changes in the size of the raft slab
and the number of piles. The diameter of piles of the
high-rise building was amended in the tender phase
from 1200mm to 1500mm to meet the client's need for
safety. Neither the structural nor the geotechnical calcu-
lation was adjusted.
Difficulties in the production of the piles caused the
company to question the static and geotechnical calcu-
lations. In this context, it needs to be mentioned that
according to the contract (FIDIC) the company carrying
Figure 3 Structural Elements of the Model
out the building work in the next project phase is
obliged to check the static calculations. After that it has
to assume the warranty for all the planning and static
calculations. In order to hand over a complete resilient
Figure 4 Pile Foundation in the Area of the High-Rise
Building
For the analysis the designer of the high-rise building
handed over the data of the loads of the walls on the
ground floor slab (see Figure 5).

Figure 7 Osterberg Cell (Source: Prof. Dr.-Ing. E.

Vees und Partner Baugrundinstitut)

In order to model soil parameters with the given results

Figure 5 Design Load at the Level of the Bottom Slab from this pile load tests a smaller 3D-FE-Model has
Due to the irregular distribution of the line loads this been set up in PLAXIS. The test pile has been modeled
level was not the appropriate interface level for apply- as embedded pile. The skin resistance of the embedded
ing the loads. We therefore searched an interface, pile element has been generated as function related to
which provided a comprehensive load image and de- the strength properties of the soil in which the pile is
cided in favor of the ground floor. Due to the distribu- located. Using this approach the pile bearing capacity is
tion of pillar and wall loads on this floor (Figure 6) the based on the stress state in the soil. The test piles
load input became significantly easier. The basement showed no or hardly any pile base resistance, which is
floors and the ground floor were taken into account in not exceptional for a heterogeneous coralline lime-
the calculation by modelling them with volume ele- stone. The pile base resistance was therefore set to
ments with a linear-elastic stress strain behavior. zero.
Besides the Mohr-Coulomb model (MC-Model) what
considers a linear-elastic, ideal-plastic material behav-
ior the Hardening Soil Model (HS-Model) and the
Hardening Soil Model with Small Strain Stiffness
(HSsmall-Model) have been studied for the calibration.
The Hardening Soil Model (HS-Model) considers a
hyperbolic stress-strain relationship between the axial
strain and the deviatoric stress in primary triaxial load-
ing. For unloading-reloading stress path a linear elastic
stress-strain relationship is used. The HSsmall-Model
uses the same formulation as the HS-Model, but con-
siders in addition the stiffness of soils at very small
strains with a simple hyperbolic law.
Figure 6 FE Model - Area of High-Rise Building At this point it has to be noted that the PLAXIS code
To determine the load-bearing behavior of the piles and its soil models have been developed to perform
three pile load tests were conducted by means of the calculations of realistic geotechnical problems. In this
Osterberg method. respect PLAXIS can be considered as a geotechnical
simulation tool. The soil models can be regarded as a
qualitative representation of soil behavior whereas the
model parameters are used to quantify the soil charac-
teristics. The simulation of reality remains an approxi-
mation, which implicitly involves that all soil models
do have their limitations. For the here defined task the calculations with unloading due to excavation result in
MC-Model seems to be sufficiently applicable. For large heaves. Calculations with loads applied from the
simplification it can be assumed that the coralline lime- building result in large settlements.
stone does show a linear elastic stress strain behavior. The HSsmall model was favored as a result of the cal-
The shape of the load-displacement curves obtained by culations. When using the MC- and/or HS-Model,
back analysis in a 3D-FE-Model was in good accord- settlements occurred in the area of the raft of the podi-
ance with the load-displacement curves obtained by the um outside the foundation of the high-rise building,
pile tests. For the MC-Model the shaft resistance, which are mainly characterized by heave. This is relat-
which is defined by τf = σn ∙ tan φ + c, is increasing ed to the accumulation of strains in areas of the model
more or less linear up to its peak. As the base resistance with small changes of stress, which are avoided in the
is neglected by our definition the pile is yielding at this HSsmall-Model by taking into account a higher stiff-
point. Using the MC-Model results in underestimating ness at small strains.
the pile shaft resistance and the pile, which only sus- The approach of modeling and calculating the founda-
tains due to friction in the 3D-FE-Model, behaves soft- tion is described comprehensively in [L1].
er. While σn is nearly constant for the MC-Model it is The calculation results show, that the piled raft founda-
increasing for the HS- and HSsmall Model. This in- tion provides sufficient capacity under vertical loading
crease is due to the deviatoric and volumetric hardening in the ultimate limit state (ULS). The calculated overall
which these advanced models provide. That means in settlements and differential settlements, both within the
consequence calibrating soil parameters with a pile high-rise area, and between the high-rise and low-rise
load test by the use of the MC-Model results in soil areas are small and are within the limits of the service-
stiffness and shear strength higher than expected in ability limit state (SLS).
reality. In a combined pile-raft foundation in which the To investigate the influence of open cavities as well as
raft transfers a greater proportion of the vertical load or sediment filled cavities element clusters of very low
sections of the raft are less supported by piles, the re- stiffness (0.1 MN/m²) have been modelled at different
sult of the FE calculation could underestimate the set- positions below pile groups of the piled raft foundation
tlement and overestimate the reliability of the load- in the high-rise area. The ground floor areas of the
bearing capacity. Therefore, it must be analyzed if element clusters have sizes of around 100 m² to 200 m².
plausible results can be achieved for the load-bearing The thickness is 5 m. The element clusters are activated
capacity of the raft with the calibrated parameters even in the first calculation phase, whereas clusters beneath
without foundation piles. the 30m long piles are combined with clusters below
the 40m long piles. 13 cluster combinations were ana-
Table 1 Calibrated Material Parameters
lyzed (Figure 8 and Figure 9).
Material Model MC HS HSS
Unit weights gsat/gdry [kN/m³] 19/19 19/19 19/19
Stiffness (Youngs Modulus at depth of 30 m) E [MN/m²] 280 175 80
Oedometer stiffness Eoed [MN/m²] 336 210 96
Shear modulus G [MN/m²] 112 70 32
Cohesion c' [kN/m²] 150 85 85
Internal friction φ' [°] 35 35 35
Dilatancy angle ψ [°] 0 0 0
At rest lat. pressure coefficient K0 [-] 0.426 0.426 0.426
Poisson's ratio n [-] 0.25 0.25 0.25
At rest lat. pressure coefficient for NC K0NC [-] 0.426 0.426
Effective vertical stress at depth of 30 m σ1 [kN/m²] -300 -300
Reference stress for stiffenes pref [kN/m²] 100 100
Power for stress-level dependency of stiffness m [-] 0.5 0.5
Tangent stiffness for primary oedometer loading Eoedref [MN/m²] 152 69
Secant stiffness in standart drained triaxial test E50ref [MN/m²] 152 69 Figure 8 FE-Model with Element Clusters
Unloading-reloading stiffness Eurref [MN/m²] 456 207
Poisson's ratio for unloading-reloading nur [-] 0.2 0.2
The result of the analysis showed that the load-bearing
Dynamic soil stiffness E0ref [MN/m²] 1035 and deformation behavior of the combined pile-raft
Reference shear modulus at very small strains G0ref [MN/m²] 431 foundation is not significantly influenced by cavities in
Shear strain at witch is Gs = 0.722G0 g07 [-] 1E-4 the area of the pile base. The raft can provide redun-
Tensile strengh σt [kN/m²] 0 0 0 dancy to the piles, for example, in case a group of piles
Overconsolidation ratio OCR [-] 1 1 1 is defective or weaker, or some pile groups encounter
Interface stiffness ratio Rint [-] 1 1 1 zones of cavities. Under these circumstances, the pres-
ence of the raft allows measure of re-distribution of the
The excavation of the 10m deep construction pit was load from the affected piles to those that are not affect-
another aspect speaking in favor of the HS- and/or ed, and thus reduces the potential influence of pile
HSsmall-Model. The MC-Model does not take into “weakness” or influence of cavities on the foundation
account a stress-depended stiffness. Accordingly the performance.
 classes S3 and C2 according to ACI 318M-08 which is
comparable to the exposure classes XA3 and XS2 ac-
cording to DIN EN 206. The concrete mixtures availa-
LP01 LP02 LP03
ble in Saudi Arabia are produced according to ASTM
specification; however, concrete plants there are not
able to supply concrete with predetermined exposure
LP04 LP05 LP06
classes. The executing company had to choose a recipe
that provides both a high chloride and high sulfate
resistance. According to testing laboratories and sup-
pliers, good results were achieved by the addition of
LP07 LP08 LP09 microsilica in combination with Portland cement along
the Saudi West Coast. Due to the higher tricalcium
aluminate content (C3A) of Portland cement CEM I, it
offers the better chloride resistance [L3] as opposed to
LP10 LP11 LP12
the more sulfate constant CEM V (high content of slag
sand). By the addition of microsilica the sulfate re-
sistance should also be improved. For the pile concrete
LP13 with a compressive strength class of fc=35N/mm² a
Figure 9 3D CAD Model with Cluster Combinations recipe with w/c =0.4 and 10% micro silica was chosen.
Thus, the requirements for exposure class C2 according
3 Design - Retaining and Shoring System to ACI318M-08 were reached. Based on the laboratory
results of the 3rd party, the chloride permeability was
The Retaining and Shoring System was designed by a
determined as "very low" according to the ASTM
subsidiary of the Italian company TREVI Spa., the
C1202 standards, thus a sufficient chloride resistance
ARABIAN SOIL CONTRACTORS. American stand-
were confirmed.
ards form the basis for the design. Static forces were
According to ACI318M-08, the sulfate resistance can
exclusively determined with the help of FE-Models.
be determined in accordance with ASTM C 1012. Here
The coralline limestone was represented by using the
rectangular test blocks with a length of 250 mm are
MC-Model. We used the HSsmall-Model for our com-
stored in a solution of sulfate acid. For the exposure
parative calculations. Additionally, conventional ana-
class S3, the storage time is 18 months.
lytical calculations for the design for safety against
In accordance with ACI318M-08 a maximum elonga-
uplift and hydraulic failure of the base of the construc-
tion for the exposure class S3 is 0.1% within this time
tion pit were also performed. The groundwater flow
frame. After 15 weeks an average increase in length of
and the corresponding pore-water pressure distribution
the sample bodies was determined at 0.012%.
as input values of the analytical calculation were calcu-
In December 2013, after an 18 month storage period,
lated in 2D- and 3D-FE-Models.
an average length increase of 0.047% was determined.
The maximum allowed longitudinal extension was not
reached. The concrete mix used fulfills the require-
ments for exposure class S3.

Figure 11 Sulfate test according to ASTM C1012, Test

Blocks
Figure 10 Cross-Section of the Revetment

4 Ensuring the durability of Bored Pile Concrete

Due to the chemical environment in the soil, a high
resistance against chloride and sulfate attack was re-
quired [L2]. As described in the tender documents, the
concrete had to ensure the requirements of the exposure
5 Construction
5.1 Pile Construction
After the site clearing, piles with diameters of 60, 80
and 150 cm were to be constructed in approximately
10m above the excavation base. The lengths of the
piles ranged from 14.5m and 40 m, so that drilling
depths of up to 50 m were required. The drilling was
carried out with a Soilmec SR70 and SR100.

Figure 14 Construction of Reinforcement Cages

Particularly difficult were the sediment-filled cavities

and trenches in the coralline limestone that were en-
countered. Despite careful operation and the use of
support liquid, the sediments have been “washed-out”
due to drilling process. To avoid these voids reaching
large sizes and collapsing, they were filled with sand
cement as preventive measure and the drilling was
continued later. In addition to the geologically-related
Figure 12 Pile Construction, Drilling Work difficulties, there were also problems with the cement
delivery which had a negative impact on the work per-
The first few meters of the drillings were supported formance.
with casing. The rest of the drilling was slurry- In addition to the chloride and sulfate resistance tests
supported using polymer suspension. already discussed, the quality assurance also included
compressive strength and fresh concrete tests (slump
and temperature).

Figure 13 Installation of the casing

The reinforcement cages were constructed on site.

Binding wire was mainly used for the fastenings. The Figure 15 Test Cylinders out of Bored Pile Concrete
30m and 40m long cages, with weights of 4.7 to 10.1 Additionally, caliber tests, integrity test and cross-hole-
tons, were installed in segments and then coupled to- sonic-tests as well as six Osterberg tests were per-
gether with clamps. formed.
Figure 16 Bore Piles in the Area of the High-rise foun- Figure 18 Diaphragm Wall during Excavation
dation
5.3 Grouted Sealing Blanket
5.2 Construction of the Diaphragm Wall
The sealing blanket was constructed to reduce the per-
The diaphragm wall with a wall thickness of 70cm and meability of the coralline limestone in order to decrease
a depth of 20m was constructed by the two-phase notably the water inflow from the bottom of the excava-
method: the trench was excavated, to the full depth tion to ≤ 2.000 m³/h. In principal, the sealing blanket
under a polymer-slurry, by clamshell and hydromill. was constructed by permeation grouting, through Multi
Once the element was fully excavated, the reinforce- Packer Sleeved Pipes (MPSP), under controlled pres-
ment cage was installed and the concrete placed by sure. The method consists in dividing the entire hole in
tremie method. The selected type of polymer is re- sections, separated by permanent packers: the grouting
sistant against the salty water environment. The poly- is performed through the steel pipe, equipped with
mer content was about 0.1 - 0.5%. The density of the rubber valves and bag-packers. The grid of MPSP
fresh (sole) polymer mud is almost equal to that of consisted of a total of 3,200 grouting holes in a square-
water, or even slightly lower, because of air entrapment triangular pattern with a final distribution of about one
that can occur during mixing. Therefore, when only hole each 3.12 m².
polymer is used to prepare the excavation slurry, the
mud density cannot be a useful parameter for determin-
ing and controlling the dosage of the polymer. Depend-
ing on the situation, the bentonite content was between
2% and 5% and led to a Marsh viscosity of > 33s / l.

Figure 19 Grouting Pattern

The grout was a Bentonite-Cement suspension with a
cement/water ratio of C/W = 0.5 to 0.35 and a benton-
ite/water ratio of B/W = 0.03 to 0.05. The composition
of the grout mixes has been defined by preliminary trial
tests, carried out on samples. In addition to the refusal
criteria of pressure and flow rate, a maximum grout
Figure 17 Construction of the Diaphragm Wall
take was set for the primary and secondary holes, which
The construction of the diaphragm wall lent itself very had to be increased at the beginning of the work. The
well under the prevailing ground conditions. During tertiary holes were grouted with no limit of grout take.
excavation of the construction pit, mostly good quality The refusal criteria were led by pressure.
wall surfaces were observed. Analogous to the pile
cages, the reinforcement cages of the diaphragm wall
were also pre-constructed on site.
 in different sizes were mixed on site in order to provide
a suitable filter pack.

Figure 20 Drilling Equipment for Injection Boreholes

The permeability of the grouted sealing blanket was

tested before and after the injection by means of water Figure 22 Installation of the Gravel Filter at a Well
permeability tests in accordance with the Lefranc meth- The extracted groundwater was passed through a ring
od. Here, a defined volume (determined on drilling line to the sedimentation basin.
depth and drilling diameter) is sealed with a pneumatic
packer and subjected to a constant hydrostatic pressure.
The local permeability was calculated on the basis of
the discharged subsoil water quantity over a certain
amount of time. The construction site was divided into
20 equal sized sections, in which one to two tests were
performed before and after the injection.
From the results of water tests, a reduction in permea-
bility of about 2*10-4 m/s to 2*10-5 m/s could be deter-
mined. The remaining water inflow was in total
1.800m³/h. This amount is corresponding to a permea-
bility of 2*10-5 m/s.

Figure 19 Results of a Drill Core after Injection

The pipe bridges commonly used in Germany are not
common in Saudi Arabia. Underground pipes are
standard here. The construction of a discharge line is
therefore very complex. Here, the water was discharged
over 2 pipelines DN800.

Figure 21 Results of a Drill Core after Injection

5.4 Dewatering
In order to control groundwater levels and ensure al-
most dry conditions inside the construction pit the in-
stallation of an array of deep wells around the internal
perimeter of the diaphragm wall shoring system and
additional wells inside the pit area was installed. The in
total 30 wells had a diameter of 600mm and were
equipped with a PVC well casing of a diameter of
Figure 24 Water Drainage in Operation
300mm. Since there are no standardized filter grains
suitable for filter gravel in Saudi Arabia, various grains
Literatur:

[L1] A. Mühl; K. Röder; C. Wawrzyniak: “Anwen-

dung und Grenzen der numerischen Berechnung
in der Geotechnik im Hinblick auf die Schnitt-
stelle zur Hochbaustatik - Planung einer Pfahl-
Platten-Gründung ”, Pfahl-Symposium 2013,
Tagungsband S. 427-446,
[L2] ACI 318-08, Building Code Requirements For
Structural Concrete
[L3] ASTM C1012, Standard Test Method for
Length Change of Hydraulic-Cement Mortars
Exposed to a Sulfate Solution

Authors
Dipl.-Ing. Alexander Mühl
[email protected]
Dipl.-Ing. Michael Brunner
[email protected]
CDM Smith Consult GmbH
Weißenfelser Straße 65H www.cdmsmith.com
04229 Leipzig Tel.: 0341 333 89 580
Dr.-Ing. Christian Wawrzyniak
[email protected]
CDM Smith Consult GmbH
Motorstraße 5 www.cdmsmith.com
70499 Stuttgart Tel.: 0711 83 07 620