The 1915 Ҫanakkale Bridge, spanning the Ҫanakkale Straits and located near the western end of
the Sea of Marmara in Turkey between Gelibolu (northeast) and Lapseki (southwest), will be the
world’s longest suspension bridge with a main span of 2,023 m and a total length of 3,563 m (Fig.
1). The proposed bridge is located ~20 km from the North Anatolian Fault, a plate boundary
between the Anatolian and the Eurasian plate, and, as such, has the potential to experience signif-
Figure 1. Project location and bridge layout.
The bridge foundations include two main Towers, herein referred to as European (northwest)
and Anatolian (southeast) towers, for which a hybrid gravity/pile inclusion foundation solution
European Tower Anatolian Tower
Assessment of lateral spreading demands on the 1915 Ҫanakkale
Bridge tower foundation
A. Giannakou, P. Tasiopoulou & J. Chacko
GR8 GEO, Athens, Greece (formerly Fugro)
Daelim Industrial, Co., Ltd, Seoul, Korea
ABSTRACT: The nonlinear, effective-stress, fully coupled dynamic soil-structure interaction
analyses performed to evaluate lateral spreading induced demands on the 1915 Ҫanakkale Bridge
Tower are presented. Calibrated advanced constitutive models were used to simulate the complex
behaviour of loose sands under strong shaking and sloping ground conditions with emphasis also
given to post-seismic deformations. The numerical analyses captured the important liquefaction-
related mechanisms, allowing for a rational and reliable prediction of Tower performance. Re-
sults indicated limited and tolerable tower deformations even for “worst-case” scenarios consid-
like the one adopted for the Izmit Bay Bridge (Steenfelt et al. 2015) and the Rion-Antirion Bridge
(Pecker 2004) has been selected. The foundation consists of a gravity caisson foundation resting
on top of reinforced soil with steel pile inclusions. A gravel bed is placed below the base of the
caisson (with no structural connection between the caisson and the pile heads) acting as a “plastic
hinge” where inelastic deformation and energy dissipation takes place (Pecker 2004).
The Anatolian Tower for the 1915 Çanakkale Bridge is located on the platform about 50 meters
away from the toe of the western slope of a submerged sandbar ridge at water depths between 42
and 46 m. During the geotechnical investigations for the project, loose coarse-grained deposits
were identified beneath and along the western slope, raising the need for assessment of lateral
spreading deformations and their impact to the Anatolian Tower foundation.
Nonlinear, effective stress, fully coupled soil-structure interaction dynamic analyses were per-
formed for the As-Is (i.e. without the bridge structure) and for the design condition including the
site preparation, the foundation elements and the Tower superstructure with the goal to evaluate
the potential for foundation deformations. Calibrated advanced constitutive models were used to
simulate the complex behaviour of loose sands under strong shaking and sloping ground condi-
tions with emphasis also given to the estimation of post-seismic deformations. The ability of the
constitutive model used for the sands to capture lateral spreading displacements was validated
through back analyses of phenomenologically similar centrifuge tests and case histories (Gianna-
kou et al. 2012; Tasiopoulou et al. 2018).
2 SITE CONDITIONS IN THE VICINITY OF ANATOLIAN TOWER LOCATION
The Anatolian Tower for the 1915 Çanakkale Bridge is located on the western slope of a sandbar
ridge at water depths between 42 and 46 m (El. -42 m to -46 m). Offshore boreholes, CPTs and
geophysical surveys were performed as part of an extensive high quality site investigation pro-
gram performed by Fugro to characterize the alignment area site conditions as well as the foun-
dation stratigraphy in an area around the proposed bridge foundations spanning the onshore, near-
shore and offshore environments (Fugro 2018).
The offshore geophysical site investigation program was performed along the bridge alignment
and included: i) bathymetric survey to determine the seabed bathymetry and morphology in the
investigated areas: ii) Side Scan Sonar (SSS) survey to identify seabed features and obstructions
in the investigated areas, iii) Sub Bottom Profiler (SBP) survey to obtain a qualitative information
relevant to soil stratigraphy and to identify superficial geohazards, iv) magnetometric survey to
detect existing subsea cables, pipelines, and other metallic obstructions and v) Ultra High Reso-
lution (UHR) survey to identify and map deeper geological and stratigraphic features and phe-
The offshore geotechnical site investigation program at the Anatolian Tower location included
9 seabed Cone Penetration Tests (CPTs), 9 downhole CPTs down to 50 meters below seafloor,
and 9 geotechnical borings down to 110 m below seafloor. In addition, in situ shear wave velocity
measurements were obtained from 2 Seismic CPTs advanced down to 50 meters depth below
seafloor and P-S suspension logging measurements in 2 boreholes advanced down to 110 meters
An integrated approach to site characterization was adopted for the project. In this approach,
the stratigraphy observed in the borings and CPT soundings was compared and integrated with
stratigraphic relationships imaged by the geophysical surveys. All data of the geotechnical explo-
rations, i.e. borings and cone penetration test (CPT) soundings, were captured in a Geographic
Information System (GIS) database to allow synthesis, comparison, analyses and outputs (Fugro
The geotechnical site conditions at the Tower area and at the slope consist of the following
main deposits in descending sequence:
• Active bar deposits: Very loose sands and silts with CPT tip resistances on the order of
1-2 MPa, varying in thickness from 8 m beneath the slope southeast of the Tower to ~1-
2.5 m at the Tower foundation location;
• Medium dense sand (Holocene) deposits: 2-m thick approximately, with tip resistances
ranging between 4 and 10 MPa;
• Stiff overconsolidated clayey (Pleistocene) deposits: 4- to 10-m thick, with undrained
shear strengths ranging between 50kPa and 100kPa;
• Miocene sedimentary rock: Primarily mudstone with UCS values between 0.5MPa and
3MPa encountered at depths of ~8 to 18 m below the seafloor (i.e. below El. -54 m to -
Figure 2 presents an illustration of the idealized stratigraphy at the area of the Anatolian Tower,
along with shear wave velocity (Vs) estimates developed from in situ measurements (i.e. P-S sus-
pension logging and Seismic CPTs).
Figure 2. Idealized stratigraphy and shear wave velocity at Anatolian Tower location.
Estimates of undrained shear strength, developed from correlations with CPT tip resistance
using a cone factor (Nk) of 15 for the Pleistocene clays, varied from about 45 kPa at 4 meters
depth below seafloor to about 125 kPa at 14 m depth below seafloor. To account for strain rate
effects and sample disturbance, the dynamic shear strength of cohesive soils was considered 20
percent higher than the static strength.
Estimated “clean sand” relative density values, Dr, of 30% and 55% from CPT data were used
for the active sand bar deposits and the medium dense sands, respectively.
3 ANATOLIAN TOWER FOUNDATION
As mentioned above, a hybrid solution has been selected for the Tower foundation consisting of
a 21-m-high caisson resting on top of reinforced soil with steel pile inclusions (base of caisson at
El. -45 m). The steel tubular piles are 2.5 m in diameter, 25-mm thick, 22-m long driven in a
triangular pattern below the foundation (Fig. 3). A 3-m thick gravel bed is placed on top of the
pile inclusions and below the base of the caisson. The superstructure consists of a steel Tower
extending to El. +318 m.
Figure 3. Anatolian Tower foundation layout.
Figure 3. Anatolian Tower Foundation Layout
Caisson water filled
-29 -34 -45
composite caisson shaft
Bottom slab of
4 GROUND MOTIONS
Probabilistic Seismic Hazard Analyses (PSHA) were performed to develop design ground mo-
tions for the site (Fugro 2017). A project-specific seismic source model was developed for esti-
mating strong ground motion hazard at the site. The North Anatolian Fault (NAF) was of primary
importance for the seismic hazard at the bridge and was rigorously modeled. A logic tree was
developed to incorporate epistemic uncertainty with respect to earthquake recurrence, slip rate
along NAF, slip partitioning, and fault segmentation in the seismic hazard analyses. A state-of-
the-art methodology was adopted for the development of design ground motions for the bridge
with the goal to reduce conservatism in design. The methodology included using a time-dependent
earthquake recurrence model for probabilistic seismic hazard analyses which considered the oc-
currence of the 1912 earthquake on the Ganos segment, the 1965 earthquake on the Saros segment
and the 2014 earthquake on the Samothraki segment. The Brownian Passage Time-dependent
(BPT) model as well as a Poisson model were incorporated in the logic tree for the NS-NAF with
equal weights. Since the closest distance of the site to the Ganos segment is 20 km, near-source
effects were accounted for by using the recently developed Bayless and Somerville model (Spu-
dich et al. 2013).
PSHA were conducted for the main bridge foundation locations with the goal to develop bed-
rock design acceleration response spectra and time histories for four return periods, 150-years,
475-years, 975-years and 2475-years.
Figure 4 shows an example matched acceleration time history (0531 motion, Fault Normal
Component) and the corresponding response spectrum for 2475-year return period at the Anato-
Figure 4. Examples of bedrock horizontal acceleration time history and response spectrum for 2475-year
5 NUMERICAL EVALUATIONS OF LATERAL SPREADING DEMANDS
Two dimensional, effective-stress, dynamic analyses were performed using the finite difference
code FLAC (Itasca 2011), which incorporates the ability to model groundwater flow and pore
pressure dissipation, and the full coupling between the deformable porous soil skeleton and the
viscous fluid flowing within the pore space. The numerical analyses were intended to realistically
model the time-dependent, nonlinear behavior associated with liquefaction of sandy loose mate-
rials, as well as nonlinear behavior of non-liquefiable clay-type materials that underly the site.
The intent of these analyses was to provide estimates of lateral spreading deformations in the
vicinity of the Anatolian Tower foundation locations where liquefiable soils and a slope were
present before (As-Is) and after the Tower construction. The numerical model for the conditions
after Tower construction is shown on Figure 5.
The approach described in Tasiopoulou et al. (2018) was adopted and lateral spreading defor-
mations were conducted in two stages: a) during seismic shakings (i.e. co-seismic); and b) after
the end of seismic shaking (i.e. post-seismic). The first stage analyses (co-seismic) involves per-
forming dynamic, nonlinear, effective stress analyses to estimate the liquefied zones and the de-
formations that occur during strong shaking. Analyses for the second stage (post-seismic) are
static analyses under gravity loads and performed for conditions after the end of strong shaking.
The empirical relationship for residual strength as a function of N1,60 proposed by Idriss & Bou-
langer (2008) considering void redistribution effects was used because lower permeability barri-
ers are present within the liquefiable units.
Ground motions were entered at El. -105 m in the numerical model as “outcrop”. Free-field
lateral boundaries were used in the analyses.
Figure 5. Numerical model including Tower foundation and superstructure.
5.2 Soil Constitutive Models and Model Calibration
5.2.1 Liquefiable Soils
The constitutive model UBCSAND (Beaty & Byrne 1998), developed by Professor Peter Byrne
and his team at the University of British-Columbia, was used to model the behavior of liquefiable
granular soils. This model has been calibrated and validated (Giannakou et al. 2011, Giannakou
et al. 2012) and has been used recently in the vulnerability assessment of important structures
(Travasarou et al. 2012, Tasiopoulou et al. 2018). The model parameters are primarily a function
of the stress-normalized, energy-corrected blow-count value of the standard penetration test
(N1,60). The SPT blow counts for the coarse-grained deposits at the Tower area were derived from
CPT tip resistances using correlations proposed by Idriss & Boulanger (2008) between normal-
ized tip resistance, qc1N, relative density, Dr, and stress-normalized, energy-corrected SPT blow
In the absence of site-specific stress-controlled cyclic tests, UBCSAND was calibrated in order
to capture soil triggering according to Idriss & Boulanger (2008) empirical triggering correlation
and strain accumulation behavior for both level (no-bias) and sloping ground (bias) conditions as
described in Giannakou et al. (2011, 2012).
5.2.2 Non-Liquefiable Soils
The Itasca S3 hysteretic model (Itasca 2011) in combination with a Mohr–Coulomb failure crite-
rion was used to simulate the dynamic nonlinear behavior of the fine-grained deposits. In the
absence of site-specific strain-controlled cyclic tests, the model parameters were fit to approxi-
mate target stress-dependent shear modulus reduction curves developed by Darendeli (2001) with
appropriate plasticity index value for each layer (i.e. PI= 40 for clays).
5.3 Modeling of Tower Foundation and Superstructure
The Tower superstructure was modeled as a series of elastic beams and masses with the appro-
priate boundary conditions to account for the cable at the top of the Tower. The fundamental
period of the fixed base Tower model used in the analyses was estimated to be about 2.54 seconds
(bridge longitudinal direction).
The caisson was modeled using elastic solid elements using a high elastic modulus to achieve
near-rigid-body behavior. Frictional interface elements were used between the caisson and the
gravel fill to allow sliding of the caisson. For these analyses, an interface friction angle equal to
45 degrees was assumed.
An equivalent spacing was estimated for the steel inclusions for the 2D plane strain analyses
by converting the triangular pattern to an equivalent square pattern. This resulted in an equivalent
pile spacing of 6.5 meters, for a total of 12 steel pile inclusions under the caisson and 1.25 m of
“unsupported” caisson length from each edge (Fig. 5). Pile elements were used in FLAC for the
modeling of the steel pile inclusions. Each pile was discretized into 1-meter long elements,
extends to El . + 318 m
El . -45 m
El . -105 m
El . -68 m
Sand bar ridge
allowing for the implementation of detailed variations in the ultimate soil resistance versus depth.
The ultimate soil resistance in both vertical and lateral directions was computed based on API
(2000) recommendations for p-y and t-z springs for sands and clays.
The UBCSAND constitutive model was used to model the behavior of the gravel fill. Although
specifications of gravel grain size result in assumed design permeability values on the order of 1
m/s, the exact permeability of the gravel layer after the Tower construction (i.e. relative density
and permeability) are not known. Sensitivity analyses were performed considering three differ-
ent relative densities for the gravel fill: 40% (loose), 55% (medium dense) and 75% (dense). In
addition three permeability values were considered for the gravel fill: 0.1 cm/s (corresponding to
the boundary between clean sand and gravel), 1 cm/s and 10 cm/s. The lower bound of the range
considers the potential for fine-grained material to permeate into the gravel matrix either during
construction or during the 100-year design life of the bridge, for potential segregation of the gravel
during the placement and for potential particle crushing of the gravel during the performance.
Two dimensional dynamic slope stability analyses were performed to evaluate lateral spreading
deformations near the Anatolian Tower foundation where liquefiable sands and sloping ground
5.4.1 As-Is Conditions (Prior to Tower Construction)
Figure 6a shows contours of horizontal displacements at the end of shaking for the As-Is condi-
tions (i.e. no excavation and backfill). Representative results are presented for a 2475-year ground
motion (0531 motion, Fault Normal Component, Fig. 6). As shown on the figure, large horizontal
displacements occur as a result of liquefaction induced lateral spreading. The maximum slope
horizontal displacements are on the order of 9 m at the seabed about 250 m far away from the
Anatolian tower foundation towards Anatolian side and decrease downslope to about 1 to 2 m at
the foundation location below the seabed. This movement is largely associated with the extremely
loose sand bar material as shown in Figure 5. However, the slope movements also extend verti-
cally down to the underlying medium dense sand with displacement on the order of 3 to 5 m that
extend down to about 20 m below the seabed at the foundation location. However since the Tower
is located on the platform about 50 meters away from the toe of the steeper slope, the horizontal
displacements at the Anatolian Tower foundation location are on the order of 2.5 m to 5.5 m, and
extend about 3 m below the seabed.
Figure 6b presents contours of residual horizontal displacements. As shown the slope becomes
unstable post-earthquake and experiences significant additional deformations in terms of horizon-
tal displacements. The maximum post-earthquake slope horizontal displacements are on the order
of more than 20 m at the seabed about 150 m away from the tower foundation and decrease
downslope to about 2.5 to 5.5 m at the foundation location below the seabed. This is consistent
with observations of the morphology in the multibeam bathymetric surveys that show evidence
of materials having moved downslope towards the Tower foundation.
Figure 6. Horizontal displacement contours for the As-Is Condition (a) at the end of shaking and (b) post-
2.5 - 5.5 m
5.4.2 Design Conditions (After Tower Construction)
Prior to Tower construction, liquefiable sands at the Tower foundation location are planned to be
excavated (up to about El. -48 m) and backfilled with gravel (up to about El. -45 m). Steel pipe
inclusions are planned to be driven to tip elevations of ~El. – 67m prior to backfilling.
Figure 7 presents numerical results for gravel fill relative density of Dr = 40% and assumed
permeabilities of 0.1 cm/s and 1 cm/s in the form of contours of maximum excess pore water
pressure ratio, Ru, and absolute residual horizontal displacement and absolute vertical displace-
ment, respectively. For the lower permeability case [see Figure 7(a)] Ru reached values equal to
1 indicative of liquefaction early on during shaking and the gravel fill liquefied both in the free
field and below the entire length of the caisson. In the higher permeability case [see Figure 7(b)],
negative Ru values were obtained, indicating an increase in the vertical effective stress during the
earthquake and the gravel fill liquefied only close to the edges of the caisson and in the free-field.
In both cases, the soil beneath the caisson close to the edges moved outwards horizontally in
opposite directions, on the order of 30-50 cm at the left edge of the caisson and 30-35 cm at the
right edge of the caisson. The maximum estimated settlement at the center of the caisson varies
from ~7 (low permeability) to 10 cm (high permeability), the maximum horizontal displacement
varies from 6 (low permeability) to 15 cm (high permeability) and the residual horizontal dis-
placement varies from 1.5 (low permeability) to 5 cm (high permeability).
Figure 7. Contours of maximum pore pressure ratio (right), residual horizontal (middle) and vertical dis-
placements (left) for gravel fill with low permeability (upper illustrations) and high permeability (lower
Advanced numerical analyses were conducted to assess the lateral spreading-induced demands
on the Anatolian Tower foundation of the for the 1915 Ҫanakkale bridge. Uncertainties related to
the final properties of the gravel fill layer below the Tower caisson foundation were also addressed
through sensitivity analyses of the assumed properties and evaluation of their effects on Tower
~ 70 cm
~ 4.5 cm
~ 40 cm
~ 10 cm of settlement
~ 7 cm of settlement
displacements and rotations. Two-dimensional nonlinear dynamic numerical analyses were con-
ducted to evaluate the Tower performance using advanced soil constitutive models which were
calibrated to capture liquefaction triggering and post-liquefaction accumulation of shear defor-
mations from published laboratory tests and empirical liquefaction triggering curves. The ad-
vanced numerical analyses captured the important liquefaction-related mechanisms, allowing for
a rational and reliable prediction of Tower performance. Results indicated limited Tower defor-
mations even under the “worst-case” scenarios of assumed gravel fill properties.
American Petroleum Institute. 2000. Recommended Practice for Planning, Designing and Constructing
Fixed Offshore Platforms-Working Stress Design, API Recommended Practice 2a-WSD (RP 2A-WSD)
Twenty-First Edition, December.
Beaty, M., & Byrne, P.M. 1998. An Effective Stress Model for Predicting Liquefaction Behavior of Sand,
Proceedings of a Specialty Conference, Geotechnical Earthquake Engineering and Soil Dynamics,
ASCE 766-777, Seattle.
Darendeli, M. 2001. Development of a new family of normalized modulus reduction and material damping
curves. Ph.D. Thesis, Dept. of Civil Eng., Univ. of Texas, Austin.
Fugro 2017. 1915 Çanakkale Bridge Probabilistic Seismic Hazard Report, Çanakkale, Turkey, prepared for
DSLY JV, November
Fugro 2018. 1915 Çanakkale Bridge Site Characterization Report, Çanakkale, Turkey, prepared for DSLY
Giannakou, A., Travasarou, T., & Chacko, J.M. 2012. Numerical Modeling of Liquefaction-Induced Slope
Movements, Proc. of GeoCongress 2012, Oakland.
Giannakou, A., Travasarou, T., Ugalde, J., Chacko, J.M., & Byrne, P. 2011. Calibration Methodology for
Liquefaction Problems Considering Level and Sloping Ground Conditions, 5th International Confer-
ence on Earthquake Geotechnical Engineering, Chile.
Idriss, I.M. & Boulanger, R.W. 2008. Soil Liquefaction during Earthquake. EERI Publication. Monograph
MNO-12, Earthquake Engineering Research Institute, Oakland.
Itasca Consulting Group Inc. 2011. Fast Lagrangian Analysis of Continua (FLAC2D) version 6.0.
Pecker, A. 2004. Design and construction of the Rion Antirion Bridge. Geotechnical Special Publication,
Spudich, P., Bayless, J.R., Baker, J.W., Chiou, B.S.J., Rowshandel, B., Shahi S.K. & Somerville. P. 2013.
Final Report of the NGA-West2 Directivity Working Group, PEER Report 2013/09, May.
Steenfelt, J.S., Foged, B. & Augustesen, A. H. 2015. Izmit Bay Bridge Geotechnical challenges and inno-
vative solutions, International Journal of Bridge Engineering, 3(3):53-68.
Tasiopoulou, P., Giannakou, A., Drosos, V., Georgarakos, P., Chacko, J., de Wit, S. & Zuideveld-Venema,
N. 2018. Numerical evaluation of dynamic levee performance due to induced seismicity. Bulletin of
Earthquake Engineering., doi.org/10.1007/s10518-018-0426-5, 1-16.
Travasarou T., Chacko J., Chen J., Giannakou A., Ilankatharan M., & Ugalde J. 2012 Application of ad-
vanced numerical methods to assess seismic performance for major projects, Second International Con-
ference On Performance-Based Design In Earthquake Geotechnical Engineering, May 28-30,