See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237371388

Performance of suction caissons in sand and clay

Article  in  Canadian Geotechnical Journal · January 2011

DOI: 10.1139/t02-030

CITATIONS READS
72 3,854

3 authors, including:

Magued Iskander Sherif El-Gharbawy

New York University, Tandon School of Engineering Abu Dhabi National Oil Company
228 PUBLICATIONS   3,456 CITATIONS    9 PUBLICATIONS   138 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Magued Iskander on 25 February 2017.

The user has requested enhancement of the downloaded file.

Color profile: Generic CMYK printer profile
Composite Default screen

576

Performance of suction caissons in sand and clay

Magued Iskander, Sherif El-Gharbawy, and Roy Olson

Abstract: The use of suction caissons (suction piles) in marine environments has been increasing in the last decade. A
suction caisson is a steel pipe with an open bottom and a closed top that is inserted into the ground by pumping water
out of it. Pumping creates a differential pressure across the caisson’s top that pushes it into place, thus eliminating the
need for pile driving. There are a number of uncertainties in the design of suction caissons. First, the state of stress and
soil conditions adjacent to a suction caisson differs from those around typical driven piles or drilled shafts. Second, the
axial load capacity of suction caissons depends on the rate of loading, hydraulic conductivity, drainage length, as well
as the shearing strength properties of the foundation material. Finally, during pullout, volume change characteristics of
the surrounding soils may change the theoretical suction pressures. A review of the existing knowledge relating to the
design and construction of suction caissons is presented in this paper along with the results of a laboratory study on
model caissons in sand and clay. Test results indicate that the use of suction pressure for installation of caissons is a
viable alternative to conventional methods. Suction was also shown to resist some axial tensile loads.

Key words: suction, pile, bucket, foundation, anchor, capacity.

Résumé : L’utilisation de caissons à succion (pieux à succion) dans des environnements marins a augmenté au cours
des dix dernières années. Un caisson à succion est un tuyau d’acier avec un fond ouvert et le dessus fermé qui est in-
séré dans le sol en extrayant l’eau de l’intérieur du pieu par pompage. Le pompage crée une pression différentielle sur
la tête du caisson et le pousse en place, éliminant ainsi le besoin de battre le pieu. Il y a un certain nombre
d’incertitudes dans la conception de caissons à succion. Premièrement, l’état des contraintes et les conditions du sol ad-
jacent à un caisson à succion diffèrent de celles retrouvées autour de pieux battus ou forés typiques. Deuxièmement, la
capacité portante axiale des caissons à succion dépend de la vitesse de chargement, de la conductivité hydraulique, de
la longueur de drainage, de même que des propriétés de résistance au cisaillement du matériau de fondation. Finale-
ment, durant l’arrrachement, les caractéristiques de changement de volume du sol environnant peuvent changer les pres-
sions théoriques de succion. On présente dans cet article une revue des connaissances existantes se rapportant à la
conception et la construction de caissons à succion de même que les résultats d’une étude en laboratoire sur des modè-
les de caissons dans le sable et l’argile. Les résultats d’essai indiquent que l’utilisation de la pression de succion pour
l’installation de caissons est une alternative viable aux méthodes conventionnelles. On a aussi démontré que la succion
résiste à des chargements axiaux de traction.
Mots clés : succion, pieux, piston à clapet, fondation, ancrage, capacité.

[Traduit par la Rédaction] Iskander et al. 584

Introduction (625 ft) long. The piles penetrated approximately 137 m

(450 ft), with each pile weighing 450 t Lee and Light 1992).
Foundations of offshore structures have undergone a In the last decade, the use of suction pressure to install
steady evolution since the first offshore structure was con- foundations of offshore structures has progressed from a
structed in 1947 (McClelland 1974). The first offshore struc- novelty concept to a viable alternative to pile driving. This
tures were in shallow water, and their foundations were a effort has been driven by the discovery of large oil reserves
natural extension of foundations used on land. Initially, piles in 2000 m (6500 ft) deep waters and the need to install off-
driven with hammers operating above the water level were shore oil platforms at these depths. The cost of traditional
used. With time, structures migrated to deeper waters and steel pile jacket (SPJ) platforms, which resemble a pyrami-
became larger. Piles also became larger and special ham- dal space frame, increases exponentially with depth due to
mers, including ones operated under water were developed. the exponential increase in material and labor costs.
For example, piles used for the Cognac platform in 313 m Bullwinkle, which is the largest SPJ platform, was installed
(1025 ft) of water were 2.15 m (7 ft) in diameter and 190 m in 410 m (1350 ft) of water using 77 000 t of steel (Finn

Received 20 December 2000. Accepted 28 January 2002. Published on the NRC Research Press Web site at http://cgj.nrc.ca on
9 May 2002.
M. Iskander.1 Polytechnic University, Six Metrotech Center, Brooklyn, NY 11201, U.S.A.
S. El-Gharbawy. The Petroleum Projects and Technical Consultations Company (Petrojet), Joseph Tito Street, Haikstep, Cairo, Egypt.
R. Olson. University of Texas at Austin, ECJ 9.227, Austin, TX 78712, U.S.A.
1
Corresponding author (e-mail: [email protected]).

Can. Geotech. J. 39: 576–584 (2002) DOI: 10.1139/T02-030 © 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:38 PM
Color profile: Generic CMYK printer profile
Composite Default screen

Iskander et al. 577

1992). Floating structures, such as tension leg platforms thus have greater capacity to resist short-term rather than
(TLP), are becoming, in some cases, the only economically long-term loading.
viable design alternative. For example, the estimated cost of Suction anchors, should not be confused with suction
using an SPJ in 872 m (2860 ft) of water at the site of the caissons. Suction anchors are small inverted cups or hemi-
Auger TLP would have exceeded the estimated value of the spherical shaped anchors installed by pumping water from
oil reserves. Auger, ASCE 1995 Civil Engineering Project within, as with suction caissons (Brown and Nacci 1971;
of the Year, was installed at a cost of $1.1 billion (US) using Sahota and Wilson 1982). The primary distinction between
39 000 t of steel (Shell 2000). the anchors and caissons is that suction anchors are subject
TLP foundation loading consists of static uplift due to to a continuously applied suction and are typically much
buoyancy and dynamic loading due to environmental condi- smaller than suction caissons. Since constant pumping is re-
tions. The foundations are typically installed at great depths quired to maintain their holding capacity, suction anchors
where suction caissons may realize economical advantages are typically used for the temporary mooring of small rigs
over traditional driven piles due to the speed of installation, (Wang et al. 1978; Helfrich et al. 1976).
elimination of the pile driving process, and reduction in ma-
terial costs (Hogervorst 1980). In some cases, suction pres-
Historical background
sures may also contribute to the axial pullout capacity of the
caissons. The use of suction for mooring began almost two decades
The state of stress and soil conditions adjacent to a suc- ago offshore of Norway. At that time, the cost of suction
tion caisson are different from those around a typical driven caissons was in excess of comparative foundation systems. It
pile or drilled shaft; thus typical experience-based design was only after significant experience had been gained with
methods may not work well with suction caissons. It is for permanent structures, such as concrete gravity based struc-
this reason that a laboratory study of model caissons in sand tures (GBSs) and SPJs, that the use of suction re-emerged
and clay was carried out. The results thereof are presented in for the mooring of floating structures.
this paper, along with a review of the knowledge gained In 1989, Gullfaks C was the first structure to employ suc-
from full-scale installation of suction caissons. tion caisson foundations (Tjelta et al. 1990). The platform
was the largest GBS ever built; it was installed in normally
consolidated soft clay in 220 m (720 ft) of water. A soil
Suction caissons drain system was utilized to accelerate consolidation thereby
increasing strength and speeding primary settlement (Tjelta
Suction caissons have also been known as suction piles, et al. 1992). During the design of the platform, two large-
bucket foundations, caisson anchors, and skirt foundations. scale offshore installation trials were conducted using instru-
They can be distinguished from traditional piling in a num- mented steel cylinders that penetrated 22 m (72 ft) into the
ber of ways. Geometrically, suction caissons are larger in di- seabed. The tip penetration resistance in the clay layers was
ameter and shorter in length than traditional piling. Length- approximately equal to the undrained bearing capacity. The
to-diameter (L/D) ratios of typical suction caissons range applied suction decreased wall friction inside the skirt com-
between one and five, compared to a typical L/D ratio of partments in the sand layers and had no influence on wall
30–60 for traditional piling. Diameters of suction caissons friction or tip resistance in the clay layers. (Tjelta et al.
range between 4 and 20 m (12–66 ft), compared with 0.3– 1986). The actual foundation consisted of 16 concrete skirt
3 m (1–10 ft) in traditional piling. For example, Snorre TLPs compartments that were 22 m (72 ft) high, 28 m (92 ft) in
suction caissons are 20 m (66 ft) in diameter and penetrate diameter and soil strength improvement by means of accel-
14.7 m (48 ft) below the seabed (Fines et al. 1991). erated consolidation below the 22 m (72 ft) depth.
Suction caissons are sealed at the top. They are installed Snorre was the first TLP to implement the suction caisson
by lowering the caisson (or caissons in a group) to the sea concept (Fines et al. 1991). Suction caissons were installed
floor, allowing them to settle under their own weight, with in soft clay at a water depth of 320 m (1050 ft). The installa-
internal water draining through a valve in the top plate. tion was preceded by extensive design verification and field
Next, water is pumped out of the caisson using submersible load testing. Comparison between predicted and measured
pumps (Cuckson 1981). Pumping of water from within the pullout loads from the model tests showed an agreement
caisson creates a differential pressure across the sealed top, within 6% (Stove et al. 1992). The actual foundation con-
thus resulting in a net downward hydrostatic force acting on sisted of four concrete foundation templates (CFT). Each
the caisson’s top. The developed hydrostatic force over- CFT consisted of a concrete tricell that was 20 m (65.6 ft)
comes the soil resistance and pushes the caisson to the de- high and 17 m (56 ft) in diameter.
sign depth. Once the design depth is reached, the pumps are Europipe (Draupner E), SPJ, was the first platform to uti-
disconnected and retrieved, the top of the caisson is sealed, lize suction caissons in sand. Suction caissons were installed
and the space between the caisson’s top and the soil plug is in very dense sand at a depth of 70 m (230 ft) (Tjelta 1994).
typically grouted. Pile hammers are not utilized in the instal- Installation was also preceded by a series of field tests on an
lation. instrumented model caisson, due to the limited experience
When a tensile pullout load is applied to a suction cais- with suction caissons in sand. Relationships were developed
son, suction pressure is developed beneath the sealed top, between the weight of the structure, applied suction, and
thus providing resistance to pullout. Given enough time, suc- penetration resistance and also between applied suction and
tion water pressures will dissipate, and prolonged pullout soil heave inside the caisson. (Tjelta 1995; Erbrich and
loading may result in caisson withdrawal. Suction caissons Tjelta 1999). The actual foundation consisted of four suction

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:38 PM
Color profile: Generic CMYK printer profile
Composite Default screen

578 Can. Geotech. J. Vol. 39, 2002

caissons fixed to the structure, one in each corner. The cais- Fig. 1. Installation of the suction caisson using maximum vac-
sons were 12 m (40 ft) in diameter with an overall height of uum in sand. Tip, Mid, and Top refer to the locations of pore-
9.45 m (31 ft) and the penetration was set at 6 m (20 ft). presssure measurement inside the caisson.
At the present time, suction caissons have been used to
support other platforms in the North Sea, Africa, and South
America. They have also been used for a variety of subsea
systems such as underwater storage and offloading tanks,
subsea manifolds, and anchor points for pipelines. Suction
caissons also have proven to be a competitive alternative to
driven piles and drag anchors for the temporary mooring of
mobile offshore drilling units (MODUs) and floating, pro-
duction, and storage offshore (FPSOs). A variety of soil
types have been encountered at the installation sites.
Layered soils, especially ones with stiff clays and dense
sands, present more difficult conditions for penetration com-
pared with homogeneous soils. The shape and geometry of
the suction caissons is evolving to conform with specific lo-
calized soil conditions, water depths, installation methods,
and types of loading. In each case, a solid understanding of where iγ w is the seepage force, and all other parameters are
the geotechnical aspects of caisson penetration is required as defined above.
for the successful installation of the foundation. In clays, a total stress approach can be used to calculate
Qso and Qb, as follows:
Pullout capacity of suction caissons H

Drained pullout [5] Qso = ∫ α Su π Do dz

The drained pullout capacity, Q, of a cylindrical caisson 0

can be expressed as π
[6] Qb = Su N c f   Do2
[1] Q = Wc + Qs  4
where Wc is the sum of the submerged weights of the cais- where Su is the undrained shear strength of the soil, α is a
son, templates, and ballast; and Qs is the side shear capacity friction factor, Nc is a bearing capacity factor usually taken
(frictional resistance) on the inside and outside of the cais- as 9, f is a bearing capacity correction coefficient, and all
son wall. Generally Qs can be calculated using an effective other terms are as defined above.
stress approach, as follows:
H Behavior in sands
[2] Qs = π (D o + D i ) ∫ γ z K tan(δ) dz
A testing program was performed to study the generation
0
and dissipation of suction pressures during the installation
where Do and Di are the outside and inside diameters of the and pullout of suction caissons in sand (Pavlicek 1993;
caisson, respectively, H is the tip penetration of the caisson, Jones 1994). Tests were performed using three different
γ is the appropriate effective unit weight of the soil, z is small-scale caissons with diameters ranging between 100
depth, K is the earth pressure coefficient, and δ is the friction and 150 mm (4–6 in.), and L/D ratios of 2:1. The caissons
angle of soil on caisson. Reduced values of γ and K should were installed using a vacuum in fine dense sand having a
probably be used to account for perturbing the soil, particu- saturated unit weight of 20.7 kN/m3 (132 pcf). Loads, dis-
larly in sand. placements, and pore-water pressures inside the model cais-
sons were measured and recorded using a high-speed data
Undrained pullout acquisition system. The results of the testing were reported
The undrained pullout capacity of a cylindrical suction by Iskander et al. (1993) and Jones et al. (1994), and are
caisson can be expressed as summarized in this section along with relevant case histo-
ries.
[3] Q = Qso + Qb + Wc + Ws
where Qso is the side shear on the outside wall of the cais- Installation
The tests shown in Figs. 1–5 were performed using a
son, Qb is the tensile bearing capacity of the foundation soil,
100 mm (4 in.) diameter, 200 mm (8 in.) long caisson (Ta-
and Wc and Ws are the submerged weights of the caisson and
ble 1). Initially, caissons were seated by pushing them for
soil plug, respectively.
approximately 30 mm (1.2 in.) to create a “seal.” In practice,
In sand, an effective stress approach can be used to calcu-
the self-weight of the caisson and ballast pushes the caisson
late Qso as follows:
and provides the seal. Next, the caisson was connected to a
H 550 mm (22 in.) of mercury vacuum source, to model a suc-
[4] Qso = π D o ∫ ( γ z + i γ w) K tan(δ) dz tion installation. Caisson insertion using suction draws water
0 from the sea floor down through the soil outside of the caisson

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:38 PM
Color profile: Generic CMYK printer profile
Composite Default screen

Iskander et al. 579

Fig. 2. Installation of the suction caisson using minimum vac- Fig. 4. Drained pullout of the suction caisson in sand.
uum in sand.

Fig. 5. Undrained pullout of the suction caisson in sand.

Fig. 3. Installation of the suction caisson by pushing in sand.

tion. The pore pressures measured in reduced suction instal-

and up through the soil plug within the caisson. During this lations (Fig. 2) were similar to those measured in full suc-
process, the upward hydraulic gradients (i = 20–40) exceed tion installations (Fig. 1). It was not possible to install model
the critical gradient, resulting in a quick condition and lique- caissons in sand without liquefying the soil, as evidenced by
faction of the soil within the caisson and beneath its tip. The the measured pore pressure during the installation (Fig. 2)
difference in the pressure head measured by the top and bot- and the formation of an excess soil plug. Reduced suction
tom pore-water pressure transducers corresponds to the head installations were slower than full vacuum installations, but
of the fluidized soil. In Fig. 1, the top and bottom transduc- pore pressures were similar (Table 1).
ers registered 3.5 kPa (0.5 psi) difference in head. For the The penetration force during installations (pressure differ-
given spacing of the transducers that head difference corre- ential times the area of caisson plus the submerged weight of
sponds to a fluid weight of 20.9 kN/m3 (133 pcf), which caisson) was 45–50 N (10–11.3 lb) for full suction installa-
compares remarkably well with the measured total unit tions and approximately 34 N (7.7 lb) for reduced vacuum
weight of the soil of 20.7 kN/m3 (132 pcf). The occurrence installations (Table 1). A number of caissons were pushed
of a quick condition in the model tests appears similar to into position to provide a reference for comparison with
that observed in the field by Senpere and Auvergene (1982) caissons inserted using suction. Typical curves of force and
and Cuckson (1981). It also explains piping that was re- measured pore-water pressure are shown in Fig. 3, as a func-
ported by Larsen (1989) and Brown and Nacci (1971) during tion of penetration depth. The force required to push the
the installation of suction caissons and anchors, respectively. caisson to the design depth was nearly one order of magni-
During installation, the level of soil inside the caisson al- tude larger than that required for suction installation, which
ways rose by approximately 50 mm (2 in.) above that in the is consistent with field experience (Tjelta et al. 1986;
tank, and a small conical depression formed surrounding the Senpere and Auvergne 1982). The measured pore-water
caisson. The difference in elevations between the soil inside pressures were just slightly greater than the static values, in-
the caisson and that in the test tank was termed the excess dicating the development of a small positive excess pore-
soil plug. Caisson penetration continued until the caisson top water pressure, due to installation by pushing.
came in contact with the excess soil plug, at which point the Occurrence of a quick condition has two important practi-
pore-water pressures reached their steady state values cal implications. First, the force required for suction installa-
(Fig. 1). tion is substantially smaller than that needed for pushed or
Several attempts to install the same suction caisson with- driven installations. The second is that the occurrence of an
out developing a quick condition were performed by using excess soil plug reduces the effective length of the suction
the minimum suction pressure that caused caisson penetra- caisson. In practice, formation of an excess soil plug may be

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:39 PM
Color profile: Generic CMYK printer profile
Composite Default screen

580 Can. Geotech. J. Vol. 39, 2002

1.02 [21.2]
mitigated using dredging pumps. Senpere and Auvergne

0.78 [16.3]

1.08 [22.5]
(kPa) [psf]
Side shear

0.31 [6.4]
(1982) employed jetting inside the caisson to prevent the
formation of an excess soil plug. However, jetting may re-
duce caisson capacity.
The observed experimental behavior was in agreement with

68 [15.3]
74.7 [16.8]

81 [18.2]
later field tests conducted by NGI and Fugro-McClelland. In
force (N)

these tests, the penetration resistance was greatly reduced due

Suction

to the use of suction (Tjelta 1995). Soil heave (excess soil

[lb]

na

plug) was also observed inside the caisson.

Displacement

Static pullout
at max load
(mm) [in.]

18 [0.71]
4.3 [0.17]
22 [0.88]

17 [0.68]

Rapid pullout tests were performed on the model caissons

after the dissipation of the installation suction pressures. The
caissons were pulled out at a constant displacement rate of
76 mm/s (0.3 in./s). Model caissons were designed such that
they can be pulled out with a closed top, thus representing
149.7 [33.7]
131.7 [29.6]

156.1 [35.1]
40.9 [9.2]

typical undrained behavior of suction caissons in pullout, or

Max load
(N) [lb]

with their top opened to provide a comparison with the

drained pullout capacity of conventional piles. Sealed-top
tests resulted in a creation of a suction pressure beneath the
caisson’s top, which caused the soil plug to be pulled out
Pullout tests

Sealed top
Sealed top

Sealed top

with the caisson. Open-top tests tend to cause the caisson to

Open top
Method

pullout empty, leaving the soil plug behind. Plugs of soil

were also observed to remain inside of suction anchors dur-
ing pullout (Wang et al. 1978 and Helfrich et al. 1976).
The development of suction resulted in a marked increase
plug, height
Excess soil

(mm) [in.]

in the pullout capacity. Sealed-caissons had approximately

Penetration force is equal to the pressure differential multiplied by the area plus the submerged caisson weight.
17 [0.67]
59 [2.33]
52 [2.05]

47 [1.85]

three times the pullout resistance of open-top caissons, as in-

dicated in Figs. 4 and 5. The differential suction pressure is
the main factor contributing to the increase in capacity (Ta-
ble 1). The developed suction in sealed-top caissons depends
rate (mm/min)

on the rate of loading, drainage length, hydraulic conductiv-

Penetration

533 [21.0]
16.5 [0.65]
152 [6.0]
137 [5.4]

ity, and shearing strength properties of the foundation mate-

[in./min]

rial. A comprehensive relationship between the various

parameters affecting the pullout capacity and the rate of
loading of suction caissons is not available.
The displacement at the maximum load of open-top cais-
Penetration
Table 1. Summary of installation and pullout data of caissons in sand.

(mm) [in.]

178 [7.0]
138 [5.4]
145 [5.7]

150 [5.9]

sons was 4 mm (0.157 in.) (Fig. 4), consistent with typical

distance

displacements required to mobilize side shear of driven pipe

piles in sand. The maximum suction developed for closed-
top caissons peaked at displacements of about 20 mm (0.8 in.)
(Fig. 5). Negative pore-water pressure tended to peak
Penetration

50.3 [11.3]

445 [100]
forcea (N)

34.2 [7.7]
44.5 [10]

sharply and dissipate slowly. The development and dissipa-

tion of these pressures was a function of the rate of pullout
b
[lb]

loading, hydraulic conductivity, and the drainage length. The

measured negative pressures support the contention of Al-
differential across

bert et al. (1989) that negative pressures are sufficient to

hold the soil plug inside the caisson during pullout, thus
top (kPa) [psi]

adding the submerged weight of the plug to the dead weight

4.75 [0.69]
4.07 [0.59]

2.83 [0.41]

of the caisson but eliminating internal side shear.

Pressure

In open top tests, the side shear ( fs) developed during

From load cell measurement.

pullout tests was calculated in Table 1 by rearranging

na

eqs. [1] and [2], as follows:

Max vacuum
Max vacuum

Min vacuum

Q − Wc
Installation

[7] fs = (open top)

π (Do + D i )z
Method

Push

where Q is the pullout tensile load, Wc is the submerged

weight of the caisson, z is the caisson penetration, and Do
series

and Di are the outer and inner diameters of the caisson, re-
Test

a
b

spectively. The side shear measured in open top tests where

1
2
3
4

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:39 PM
Color profile: Generic CMYK printer profile
Composite Default screen

Iskander et al. 581

high vacuums were used during installation was 0.31 kPa ety of amplitudes. Cyclic capacity tended to decrease with
(6.4 psf). This value is an average of internal and external the increase in the frequency of the applied load. Caisson
side friction, which could be different. pullout occurred in cases where the peak sinusoidal load ex-
In sealed top tests, the value of the side shear developed ceeded the drained reference load. Pullout occurred in less
during pull ( fs) depends on the magnitude of end bearing, than 5 s when the peak cyclic load exceeded 125% of the
Qb, which cannot be directly measured. Therefore, fs was reference load. Pullout occurred over several hours when
calculated in Table 1, using eq. [3] and assuming that the peak cyclic loads ranged between 100 and 110% of the ref-
tensile end bearing, Qb, was approximately equal to the suc- erence load. When peak cyclic loads were maintained, even
tion force acting on the caisson’s top, as follows: slightly, below the reference load, pullout did not occur in
Q − pa − Wc − Ws agreement with Allersma et al. (2000). For example, one test
[8] fs = (sealed top) conducted with a peak cyclic load corresponding to 90% of
z π Do the reference load was terminated after 160 000 cycles with-
out any pullout.
where p is the pressure differential across the caisson top, a
The cyclic pullout load capacity of suction caissons de-
is the internal area of the caisson, Ws is the submerged
pends on the same parameters as for static load. The small
weights of the soil plug, and all other variables are as de-
size and the fast rate of loading in the model tests obviously
fined above. The side shear measured in sealed top tests
contributed to the lack of cyclic capacity of model suction
where high vacuums were used during installation was
caissons. However, it is probably imprudent, at this time, to
0.78 kPa (16.3 psf), which is 25% lower than that of cais-
depend on suction to resist any cyclic loading in excess of
sons installed by pushing or using a low vacuum (1.05 kPa,
the drained pullout capacity.
22 psf).
The use of Qb = pa can be justified by the fact that the
maximum suction is directly related to the developed tensile Behavior in clays
end bearing (reverse bearing capacity). It is understood that
the magnitude of the developed suction and tensile end bear- The focus of this part of our investigation has been the be-
ing depends entirely on the rate of loading, the caisson ge- havior of suction caisson foundations for tension leg plat-
ometry, and the hydraulic conductivity of the soil. The forms in water depths of 2000–3000 m (6 500 – 10 000 ft) in
maximum developed suction is limited by the cavitation the Gulf of Mexico. At these depths, the foundation material
pressure and the soil’s tensile capacity. In sand, the tensile is predominantly normally consolidated clay. Traditionally,
capacity depends on the development of negative pore pres- suction caissons have been built in the North Sea with 0.5–
sures during pullout loading. 2 L/D ratios, e.g., Gullfaks C, Snorre, and Europipe. How-
The use of Qb = pa was verified by comparing the mea- ever, in Gulf of Mexico clays, capacity is derived mainly
sured skin frictions to their theoretical values. The external from increased soil shear strength with depth. It was there-
side friction ( fs) can be calculated as follows: fore necessary to examine the possibility of extending the
range of L/D ratios of suction caissons beyond that used in
z the North Sea.
[9] fs = ( γ + i γ w) K tan δ
2 A multidirectional displacement-controlled loading device
capable of applying static and cyclic, axial and inclined ten-
fs (series 1) = 0.075 × [(20.7 – 9.81) + (2.1 × sile loading as well as combined vertical and horizontal cy-
9.81)] × 0.8 × tan(25) = 0.85 kPa (sealed top) clic tension loading representative of actual field loading
fs (series 2) = 0.075 × [(20.7 – 9.81) × 0.8 × conditions was developed. An experimental program was de-
tan(25) = 0.30 kPa (open top) signed to test a set of model foundations under simulated
TLP static and dynamic loads (Goldberg 1994; El-Gharbawy
where z is the caisson penetration (caisson length minus 1998). The foundation material was normally consolidated
height of excess soil plug), γ is the effective unit weight of kaolinite clay consolidated using a vacuum. Model caissons
the soil, iγ w is the increase in vertical stress due to the down- with L/D ratios of 2–12 were used to study the feasibility of
ward hydraulic gradient, i is the hydraulic gradient, γ w is the using suction to install caissons with L/D ratios in that range
unit weight of water, K is the coefficient of lateral earth and the effect of increasing length on penetration resistance
pressure taken per API (1992) as 0.8 for nondisplacement and pullout capacity of the caissons. Each of the instru-
piles, and δ is the friction angle between the caisson and soil. mented caissons consisted of a 50–100 mm (2–4 in.) outside
The computed skin frictions for drained and undrained con- diameter tube fitted with a cap. Loads, displacements, and
ditions (0.30 and 0.85 kPa, respectively) are in agreement pore-water pressures inside and in the vicinity of the model
with the measured values (0.31 and 0.78 kPa, respectively). caissons were measured and recorded using a high-speed
The gradient associated with pullout suction was responsible data acquisition system.
for the increase in frictional resistance.
Installation
Cyclic pullout The tests shown in Figs. 6–8 were conducted using a
Tests were performed to determine the cyclic pullout ca- 100 mm (4 in.) diameter, 600 mm (24 in.) long caisson.
pacity of suction caissons. In these tests, the reference pull- Tests began with seating the caisson on the surface of the
out capacity was that corresponding to fully drained pullout, soil and allowing it to settle under its own weight, thereby
i.e., slow pullout with an open caisson top. Sinusoidal cyclic creating the “seal” required for the application of suction. A
loads ranging between 4 and 15 Hz were applied with a vari- pump was then fitted to the top of the caisson and used to

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:39 PM
Color profile: Generic CMYK printer profile
Composite Default screen

582 Can. Geotech. J. Vol. 39, 2002

Fig. 6. Installation of the suction caisson in clay. (a) Penetration. Fig. 7. Undrained pullout of the suction caisson in clay.
(b) Pore-water pressure. (Solid lines are measurements inside the
caisson. Dashed lines are measurements outside of the caisson).

Fig. 8. Incremental, drained pullout of the suction caisson in clay.

of the soil fast enough, in a displacement controlled mode,

create a reduced pressure zone inside the caisson. The pres- so as to prevent pore-water pressure dissipation. Typical
sure differential across the top of the caisson worked to push pullout speeds were on the order of 25–50 mm/s (1–2 in./s).
the caisson into the soil to full penetration. As shown in When the caisson was withdrawn under undrained condi-
Fig. 6, caisson installation was composed of two distinct tions, the soil plug came out with the caisson and the outer
phases: self-weight penetration followed by suction installa- surface of the caisson was covered with about 3 mm of clay.
tion. Pore-water pressures inside and around the caisson in- In the drained tests, loading was applied in increments, in a
creased during the first phase due to the weight and volume load-controlled mode, to allow the pullout suction pressures
displacement of the caisson. In the second phase, pressures to dissipate. Under drained conditions, the caisson came out
within the caisson decreased sharply in response to the ap- with no soil adhering to either the inner or the outer sur-
plication of suction, while those outside the caisson were lit- faces.
tle affected or had proceeded to increase with the increased Negative pore-water pressures developed within the cais-
penetration. Model suction caissons with L/D ratios ranging son during undrained pullout (Fig 7) and remained largely
between 2 and 12 were installed successfully, in agreement unchanged during drained pullout (Fig. 8). The increase in
with the theoretical predictions of House et al. (1999). Self- pullout capacity in the undrained condition was easily attrib-
weight penetration accounted for approximately a third to uted to these pressures (El-Gharbawy and Olson 1998). Neg-
half the penetrable length of the caisson, in agreement with ative pressures also resulted in the soil plug being retained
later full-scale installations in 2100 m (7000 ft) water depths within the caisson during undrained pullout. These results
in the Gulf of Mexico (El-Gharbawy et al. 1999). are in agreement with Finn and Byrne (1972), Byrne and
Finn (1978), and Clukey and Morrison (1993). Accordingly,
Static pullout the weight of the plug could be counted upon to resist the
Pore-water pressures induced during the installation pro- fast pullout load, typical of environmental effects.
cess were allowed to dissipate before caissons were ex-
tracted. The capacities of these caissons were similar to Cyclic pullout
those of piles having similar dimensions and pushed into Stress-controlled cyclic tests were conducted by applying
clay (El-Gharbawy and Olson 1999). Undrained pullout re- and maintaining a bias load on the caisson overnight, then
sulted in larger capacities than drained pullout. For example, superimposing vertical cyclic pullout loads with different
tests run on a 100 mm (4 in.) diameter caisson with a L/D peak values. The bias load was chosen as a function (60–75%)
ratio of 6 indicated that undrained pullout had approximately of the long-term capacity (LTC) of the caisson. Increasing
three times the capacity of drained pullout (Figs. 7 and 8). the peak cyclic load resulted in an increase in displacement
Undrained loading involved pulling the model caissons out after a given number of cycles (Fig. 9). A peak cyclic load

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:40 PM
Color profile: Generic CMYK printer profile
Composite Default screen

Iskander et al. 583

Fig. 9. Axial cyclic loading of suction caissons in clay. LTC, caisson wall; and (d) a reverse bearing capacity mechanism
long term capacity. developed due to suction.
(6) The long-term drained capacity of the caisson was
found to represent an upper limit to the peak cyclic load that
could be withstood by the caisson before significant dis-
placement occurred. The number of cycles before failure de-
creased with an increase in the peak cyclic load. However,
suction pressure can be used to resist transient events.

Acknowledgments
We are grateful to the Offshore Technology Research Cen-
ter (National Science Foundation (NSF) Engineering Re-
search Centers program grant no. CDR8721512) for funding
portions of this work. Our research partners Robert Pavlicek,
Casey Jones, Aaron Goldberg, Wade Williams, Jody Kerwin,
and Pawel Zieminski were instrumental in conducting the ex-
equivalent to 85% of the LTC of the 600 mm (24 in.) long
perimental work and provided us with valuable insights and
caisson could be sustained for at least 10 000 cycles with no
suggestions.
significant cap displacement. However, peak cyclic loads
greater than the LTC resulted in the caisson pulling out at
rates that were directly proportional to the magnitude of the
load. For example, a peak cyclic load reaching twice the References
magnitude of the bias load, the 400 mm (16 in.) long caisson Albert, L., Holtz, R., and Magris, E. 1989. Super pile system: a
(L/D = 4) pulled out after 6 cycles, which is in general feasible alternate foundation for tension leg platform in deep
agreement with the Gullfaks C field test results. When the water. Marine Geotechnology, 8: 133–158.
same loading was applied to the 60 mm (24 in.) long caisson Allersma, H.G.B, Kierstein, A., and Maes, D. 2000. Centrifuge
(L/D = 6), the caisson withstood 80 cycles prior to failure. modelling on suction piles under cyclic and long term vertical
The number of cycles before failure increased with decreas- loading. In Proceedings of the 10th International Offshore and
ing peak cyclic load or with increasing caisson length. Polar Engineering Conference, (ISOPE-2000), Seattle, WA, 28
May – 2 June 2000. Vol. 2, pp. 334–341.
Conclusions and recommendations API. 1992. American Petroleum Institute (API) recommended
practice for planning, designing and constructing fixed offshore
Utilizing suction pressures for installation and develop- platforms, RP2A. API Production Div., 211 North Ervay, Suite
ment of pullout capacity of caissons is a viable concept. 1700, Dallas, TX 75201.
Successful model suction caisson installations were per- Brown, G., and Nacci, V. 1971. Performance of hydrostatic an-
formed with ease. The following observations were made chors in granular soils. In Proceedings, Offshore Technology
during the installation and pullout of model suction caissons: Conference, OTC 1472, pp. 533–542.
(1) A quick condition occurs during the installation of Byrne, P., and Finn, W. 1978. Breakout of submerged structures
suction caissons in sand, which influences the behavior in buried to shallow depth. Canadian Geotechnical Journal, 15:
several ways. First, the penetration resistance is reduced. 146–154.
Second, the penetration distance is reduced due to the for- Clukey, E., and Morrison, M. 1993. A centrifuge and analytical
mation of an excess soil plug. Third, suction installation re- study to evaluate suction caissons for TLP applications in the
sults in a reduction in the frictional capacity of the caissons. Gulf of Mexico. In Design and performance of deep founda-
(2) Model suction caissons with L/D ratios ranging be- tions, piles and piers in soil and soft rock, American Society of
Civil Engineers, Geotechnical Special Publication GSP No. 38,
tween 2 and 12 were installed successfully in clay.
pp. 141–156.
(3) Suction, and the associated tensile bearing capacity,
Cuckson, J. 1981. The suction pile finds its place. Offshore Engi-
are the dominant parameters contributing to the short-term
neer, April, pp. 80–81.
(undrained) pullout capacity of sealed top suction caissons
El-Gharbawy, S. 1998. The pullout capacity of suction caisson
in both sand and clay. Additionally, the suctions developed foundations for tension leg platforms. Ph.D. dissertation, The
during pullout were sufficient to retain the soil plug inside University of Texas at Austin, Austin, TX.
the caisson, thereby increasing the dead weight of the cais- El-Gharbawy, S., and Olson, R. 1998. The pullout capacity of suc-
son and eliminating the internal side shear. tion caisson foundations for tension leg platforms. In Proceed-
(4) During pullout, the gradient associated with the suc- ings of the 8th International Offshore and Polar Engineering
tion pressures resulted in a marked increase in the frictional Conference, Montreal, PQ, 24–28 May 1998.
capacity of caissons in sand. El-Gharbawy, S., and Olson, R. 1999. Suction caisson foundations
(5) Suction caissons derive their vertical pullout capacities in the Gulf of Mexico. In Analysis, Design, Construction, and
from one or more of the following components: (a) weight Testing of Deep Foundations, Proceedings of the OTRC ‘99
of the suction caisson, foundation templates, and ballast Conference, Reston, VA. Edited by J. Roesset. American Soci-
loads; (b) submerged weight of the soil plug captured within ety of Civil Engineers, Geotechnical Special Publication No. 88,
the caisson; (c) shearing resistance between the soil and the pp. 281–295.

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:41 PM
Color profile: Generic CMYK printer profile
Composite Default screen

584 Can. Geotech. J. Vol. 39, 2002

El-Gharbawy, S., Olson, R., and Scott, S. 1999. Suction anchor in- Larsen, P. 1989. Suction. Anchors as an anchoring system for
stallation for deep Gulf of Mexico applications. In Proceedings, floating offshore constructions. In Proceedings, Offshore Tech-
Offshore Technology Conference, OTC 10992, pp. 747–754. nology Conference, OTC 6029, pp. 535–540.
Erbrich, C., and Tjelta, T. 1999. Installation of bucket foundations McClelland, B. 1974. Design of deep penetrating piles for ocean
and suction caissons in sand – geotechnical performance. In structures. Journal of Geotechnical Engineering Division,
Proceedings, Offshore Technology Conference, OTC 10990, ASCE, 100(GR7): 705–748.
pp. 725–735. Pavlicek, R. 1993. Axial tensile load capacity of suction piles.
Fines, S., Stove, O., and Guldberg, F. 1991. Snorre TLP tethers and M.Sc. thesis, The University of Texas at Austin, Austin, TX.
foundation. In Proceedings, Offshore Technology Conference, 165 pp.
OTC 6623, pp. 587–597. Sahota, B., and Wilson, Q. 1982. The breakout behavior of a suc-
Finn, L. 1992. Offshore structures — past, present and future. In tion anchor embedded in submerged sand. In Proceedings, Off-
Proceedings Civil Engineering in the Oceans V. College Sta- shore Technology Conference, OTC 4175, pp. 117–132.
tion, Texas, 2–5 November. Edited by R. Hudspeth. American Senpere, D., and Auvergne, G. 1982. Suction anchor piles – a
Society of Civil Engineers (ASCE), pp. 218–242. proven alternative to driving or drilling. In Proceedings, Off-
Finn, W., Byrne, P. 1972. The evaluation of the breakout force for shore Technology Conference, OTC 4206, pp. 483–493.
a submerged ocean platform. In Proceedings Offshore Technol- Shell Oil. 2000. Auger tension leg platform fact sheet [online].
ogy Conference, OTC 1604. Available from Shell Exploration and Production Company
Goldberg, A. 1994. Design and construction of a model suction pile. Web Site, http://www.shell.com [cited 22 March 2002].
M.Sc. thesis, The University of Texas at Austin, Austin, TX. Stove, O.J., Bysveen, S., and Christophersen, H. 1992. New foun-
Helfrich, S., Brazil, R., and Richards, A. 1976. Pullout characteris- dation systems for the Snorre Development. In Proceedings,
tics of a suction anchor in sand. In Proceedings, Offshore Tech- Offshore Technology Conference, OTC 6882, pp. 75–83.
nology Conference, OTC 2469, pp. 501–506. Tjelta, T. 1994. Geotechnical aspects of bucket foundations replac-
Hogervorst, J. 1980. Field trials with large diameter suction piles. ing piles for the Europipe 16/11-E jacket. In Proceedings, Off-
In Proceedings, Offshore Technology Conference, OTC 3817, shore Technology Conference, OTC 7397, pp. 73–82.
pp. 217–224. Tjelta, T. 1995. Geotechnical experience from the installation of
House, A., Randolph, M., and Borbas, M. 1999. Limiting aspect the Europipe jacket with bucket foundations. In Proceedings,
ratio for suction caisson installation in clay. In Proceedings of Offshore Technology Conference, OTC 7795, pp. 897–908
the 9th International Offshore and Polar Engineering Conference Tjelta, T., Guttormsen, T., and Hermstad, J. 1986. Large scale pen-
(ISOPE 99), Brest, 30 May – 4 June 1999, Vol. 1, pp. 676–683. etration test at a deep water site. In Proceedings, Offshore Tech-
Iskander, M., Olson, R., and Pavlicek, R. 1993. Behavior of suc- nology Conference, OTC 5103, pp. 201–212.
tion piles in sand. In Design and performance of deep founda- Tjelta, T., Aas, P., Hermstad, J., and Andenaes, E. 1990. The skirt
tions, piles and piers in soil and soft rock, American Society of piled Gullfaks C platform installation. In Proceedings, Offshore
Civil Engineers, Geotechnical Special Publication GSP No. 38, Technology Conference, OTC 6473, pp. 453–462.
pp. 157–171. Tjelta, T.I., Janbu, N., and Grande, L. 1992. Observations on drain-
Jones, W. 1994. Axial capacity of suction piles in sand. M.Sc. the- age control effects on Gullfaks C structure. In Proceedings of
sis. The University of Texas at Austin, Austin, TX. 90 pp. International Conference on Behavior of Offshore Structures
Jones, W., Iskander, M., Olson, R., and Goldberg, A. 1994. Axial (BOSS 92), Vol. 1, pp. 703–720.
capacity of suction piles in sand. In Proceedings of the 7th Inter- Wang, M., Demars, K., and Nacci, V. 1978. Applications of suction
national Conference on the Behavior of Offshore Structures anchors in offshore technology. In Proceedings, Offshore Tech-
(BOSS’94). Edited by C. Chryssostomidis. Pergamon, pp. 63–75. nology Conference, OTC 3203, pp. 1311–1320.
Lee, G., and Light, L. 1992. Fixed offshore platforms – design and
construction considerations. Notes for short course Design of
Fixed Offshore Platforms. University of Texas, Austin, TX.

© 2002 NRC Canada

I:\cgj\Cgj39\Cgj-03\T02-030.vp
Monday, May 06, 2002 1:39:42 PM
View publication stats