Professional Documents
Culture Documents
t
ip
ABSTRACT
d cr
The objective of this paper is to document and study the failure of a driven,
te s
di nu
steel pipe pile foundation system supporting an offshore platform. The three-pile
system failed in overturning due to a pull-out failure of the most critically loaded pile in
ye a
a hurricane five years after installation. The calculated tensile capacity of this pile
op M
using the American Petroleum Institute (API) Design Method is close to the estimated
load at failure. This case history is significant because it represents the failure of a
C ted
large-diameter pile in service when loaded by a storm. Also, this case history generally
affirms current design methods and contradicts a widely held perception that offshore
ot p
pile designs are significantly conservative. This case history highlights opportunities to
N ce
1
Senior Engineer, Energo Engineering, 1300 W. Sam Houston Pkwy. S., Suite 100, Houston, TX
Ac
77042; JY.Chen@kbr.com
2
Brunswick-Abernathy Professor, The University of Texas at Austin, 1 University Station C1792,
77042; Frank.Puskar@kbr.com
4
Principal Consultant, Energo Engineering, 1300 W. Sam Houston Pkwy. S., Suite 100, Houston, TX
77042; Sean.Verret@kbr.com
improve design practice by explicitly accounting for pile flexibility when dealing with
strain-softening soils and by considering the capacity of the foundation system as well
INTRODUCTION
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
d cr
Hurricane Ike was the most intense Atlantic hurricane of 2008, with maximum
sustained winds of 230 km/h and a minimum central pressure of 935 mbar (Berg 2009).
te s
di nu
Offshore in the Gulf of Mexico, Ike impacted thousands of platforms used for oil and
gas production and it destroyed at least fifty of them (Energo Engineering 2010). One
ye a
of those platforms, designated as EC368A in the East Cameron area, failed due to an
op M
axial failure in the driven pile foundation, which left the platform leaning after the
hurricane (Fig. 1). Of the more than 200 platform failures in the Gulf of Mexico caused
C ted
by hurricanes from Andrew in 1992 through today (Puskar et al. 1994, Aggarwal et al.
1996a and 1996b, Bea et al. 1999, Puskar et al. 2004, 2006 and 2007, and Gilbert et al.
ot p
failure.
Ac
large-diameter pile in service when loaded by a storm. The experimental databases that
have been used to develop design methods for offshore piles primarily consist of much
smaller test piles that were loaded with static loads within weeks or months of
installation (e.g., Pelletier et al. 1993). This case history is also significant because
there is a widely held perception, due to the lack of known foundation failures until
now, that offshore pile foundation designs are significantly conservative (Aggarwal et
The objective of this paper is to document and study the failure of the EC368A
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
foundation. This paper will describe the EC368A platform configuration, its pile
d cr
foundation design, the loading to which it was subjected during Hurricane Ike, and the
failure mechanism. It will present an analysis of the failure and compare the observed
te s
di nu
performance with predictions. It will conclude with a discussion about what can be
learned from this case history for the design and assessment of driven pile foundations.
ye a
op M
DESCRIPTION OF PLATFORM EC368A
C ted
m of water offshore the coast of Louisiana. It was installed in 2003 and loaded by
ot p
Hurricane Ike in 2008. Figure 2 presents a 3-dimensional (3-D) rendering of the jacket
N ce
platform including the piles. The platform was constructed by placing the jacket on the
Ac
seafloor, driving the steel pipe piles through the jacket legs into the soils, connecting
the piles to the jacket legs with shims at the top of the legs, and then welding the
steel pipe piles. The piles were fabricated using A36 steel. They were driven in sections
and welded in the field during installation. Pile A is a vertical pile that penetrates 80.8
m below the seafloor. Piles B and C are double battered at 1-horizontal to 5-vertical
(1H: 5V) away from Pile A; they both penetrate 67.1 m below the seafloor. The wall
t
ip
Platform EC368A was also equipped with a well conductor north of Pile A.
d cr
This vertical conductor, which contained an oil well, is a 508-mm (20-inch) diameter
steel pipe with a wall thickness of 12.7 mm (0.5 inches). This conductor was attached
te s
di nu
to the jacket via framing that allowed it to carry only horizontal loads. In addition, Pile
at the platform location in 2001. The sampling and testing methods used to drill this
ot p
boring are consistent with the present-day state of practice for the design of offshore
N ce
foundations. The field crew advanced the borehole with an open-hole, wet rotary
Ac
technique, using a drag bit attached to the drill pipe. They added salt water gel and
weight materials as necessary to suspend and remove drill cuttings and to provide
lateral support to the borehole wall. They obtained samples through the open bore of
the drill pipe at 0.9-m intervals to 15.2-m penetration and 3.0-m intervals thereafter to
the bottom of the boring (106.1 m below the seafloor). They used a 64-mm (2.5-inch)
outside diameter (OD) liner sampler to obtain soil samples up to 5.2-m penetration and
pushed a 76-mm (3-inch) OD, thin-walled, Shelby tube sampler to obtain remaining
samples. The 76-mm diameter Shelby-tube samples are considered undisturbed soil
samples in practice.
The field crew extruded most samples from the samplers and a field engineer
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
classified them. The engineer performed unit weight, torvane, pocket penetrometer,
d cr
miniature vane and unconsolidated-undrained (UU) triaxial tests on selected cohesive
soil samples onboard the drilling vessel. The crew also shipped remaining samples to
te s
di nu
the laboratory to perform Atterberg limits and water content measurements as well as
consists of two strata: a very soft, highly plastic, normally consolidated clay to a depth
C ted
PILE DESIGN
Ac
The geotechnical engineer established axial capacities for the driven piles
following the normal industry practice for offshore platforms in the Gulf of Mexico at
the time (API 2000). The maximum side shear for undrained loading at any depth along
a pile, fmax, was calculated from fmax D su where su is the design undrained shear
strength at that depth and D is a dimensionless factor that depends on < the ratio of su
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
to the vertical effective stress, V v' 0 , at that depth (Randolph and Murphy 1985). The
d cr
maximum unit end bearing in compression at the pile tip, qmax, was calculated from
te s
qmax 9su where su is the design undrained shear strength at the tip. The piles were
di nu
assumed to be plugged with the unit end bearing acting over the enclosed cross-section
of the pile tip in accordance with API design guidance for piles in clay, where the
ye a
op M
friction between the soil plug and the pipe pile generally exceeds the end bearing acting
The axial capacity based on the maximum unit side shear and end bearing is
ot p
shown in Figure 5 for a 1,220-mm diameter, vertical pile. This capacity is commonly
N ce
mobilizing the maximum side shear resistance simultaneously along the entire length
Ac
in tension or compression and the maximum end bearing in compression. If the soils
exhibit strain softening, the maximum load that can be resisted by a flexible pile may
be less than the “ultimate” capacity for a rigid pile. The ultimate capacity profiles were
design.
Soil Springs
relationships between unit side shear and relative shear displacement between the pile
wall and soil (t-z curves), end bearing resistance and displacement of the pile tip (Q-z
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
curves), and lateral resistance and pile deflection (p-y curves). These curves were
d cr
developed according to API design guidance. The recommended t-z curve is shown in
Figure 6; this curve accounts for strain softening in the highly plastic clays at this
te s
di nu
location, with a residual-to-peak side shear ratio of 0.8. The recommended p-y curves
were those developed by Matlock (1970) for cyclic loading in soft clay.
ye a
op M
Pile Diameters and Lengths
C ted
The structural engineer selected pile diameters and lengths. First, structural
analyses incorporating the soil springs were performed with different pile diameters
ot p
and wall thicknesses using very long piles. Different combinations of wind, wave,
N ce
current and deck loads were used to check the structural capacity (i.e., unity checks) for
Ac
all steel tubular members and joints including the piles according to API design
guidance. From these structural analyses, the pile diameters and wall thicknesses
A structural analysis was then performed on the final geometry of the structure
in order to select pile lengths. The design axial loads for the piles were determined by
applying the 100-year metocean combination of wind, wave and current loads to the
structure. These design axial loads were increased by a factor of safety of 1.5 to
determine the required axial capacities according to API design guidance. The pile
lengths were selected such that the ultimate axial capacity was equal to the required
axial capacity. To illustrate using Pile C, the design axial load was 11,900 kN in
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
tension, the required axial capacity was 1.5 times the design load or 17,800 kN in
d cr
tension, and the selected pile length was 67.1 m (Fig. 5).
te s
PERFORMANCE OF PILE FOUNDATION IN HURRICANE IKE
di nu
ye a The eye of Hurricane Ike passed within 64 km to the southwest of Platform
op M
EC368A on September 12, 2008. Based on the hurricane hindcast (Oceanweather
2008), the maximum wave height was 22 m and the maximum wind speed was 114
C ted
km/h at Platform EC368A. For reference, the maximum wave height was 15 to 20
percent higher than the design wave height associated with the 100-year metocean
ot p
conditions. Because the base shear on the platform (total horizontal force applied
N ce
above the seafloor) is roughly proportional to the square of wave height, the base shear
Ac
in Hurricane Ike was approximately 30 to 40 percent higher than the design base shear.
When the platform was re-visited after Ike, it was found to be leaning about 4
degrees to the northwest (Fig. 1). Underwater inspections showed that Pile C was
pulled upward more than 1 m. There was no evidence of damage to the jacket
axial pull-out failure of Pile C, was identified as the cause of failure for Platform
EC368A.
t
ip
Dynamics 2005) was used to predict the response of Platform EC368A to the
d cr
maximum loads imposed by Hurricane Ike. This finite element method based structural
te s
di nu
Assessment is normally performed when the loads on a platform are increased (e.g.,
becomes excessive; this analysis is called a pushover analysis. The pushover analysis
C ted
utilizes the non-linear t-z, Q-z and p-y curves to model the interaction between the piles
and soils.
ot p
N ce
In the pushover analysis, the magnitude and direction of the maximum wind,
Ac
wave and current loads estimated for Hurricane Ike were imposed on the platform,
resulting in an expected base shear of 5,340 kN acting at 290 degrees clockwise from
the true north (Fig. 2). Then, the environmental load was increased until collapse
occurred at a base shear of 5,560 kN, which is within five percent of the expected base
shear in Hurricane Ike. The pushover analysis predicted that Pile C would pull out and
Piles A and B would yield in bending, resulting in a rigid-body rotation of the platform.
In order to gain further insight into the performance of the pile system, a
t
ip
(Murff and Wesselink 1986, Tang and Gilbert 1993 and Chen et al. 2009) was used to
d cr
analyze the pile system. This model, schematically shown in Figure 7, assumes a
plastic collapse mechanism, where all elements of resistance are characterized as rigid
te s
di nu
and perfectly plastic. This model incorporates structural capacities of piles and well
conductors and axial and lateral capacities of soils acting on the piles and conductors.
ye a
The structure supported by the piles is assumed to be perfectly rigid and infinitely
op M
strong so that it can distribute loads as necessary to develop a full foundation collapse
mechanism. Well conductors are modeled as piles that are connected to the structure
C ted
with rollers to constrain them horizontally. Each pile or well in the system collapses
when one hinge forms at the mudline and a second hinge forms at some depth below
ot p
the mudline. Collapse of the entire system occurs when double hinges form in each of
N ce
that incorporates a combination of base translation and rotation to give the minimum
system capacity. Comparisons between upper-bound solutions and results from more
rigorous pushover analyses indicate that the upper-bound solution overestimates the
base shear capacity by approximately 10 percent (Murff and Wesselink 1986 and Chen
et al. 2009). Therefore, results from the system collapse analyses are reduced by 10
Input to the pile system collapse analysis includes the axial and lateral soil
resistances acting on each pile and conductor at failure. For axial soil resistance, strain
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
softening in side shear limits the maximum soil resistance that can be developed along
d cr
a flexible pile. A finite difference (“t-z”) analysis was performed for each pile in
compression and tension to account for soil-pile interaction and establish the axial soil
te s
di nu
resistance at failure. Results for Pile C in tension are shown in Figure 8; the maximum
axial capacity for this pile is about 90 percent of the “ultimate” axial capacity for a rigid
ye a
pile. The effects of pile weight and reverse end bearing on tensile capacity were not
op M
included in the t-z analysis because they represent a negligible contribution to the total
axial capacity of Pile C. For lateral soil resistance, the static ultimate resistance was
C ted
assumed because the piles at failure will push into undisturbed soil that has not been
degraded by previous loading cycles (Murff et al. 1993 and Jeanjean 2009).
ot p
N ce
The capacity of this pile system based on the upper-bound collapse solution is
Ac
and base shear located inside the interaction curve, the pile system is expected to be
stable. Three distinct regions of pile system behavior are shown in Figure 9. At small
overturning moments (or moment arms), pile system capacity is dominated by shear
where the lateral capacities of the piles and conductors govern the system capacity. As
the moment arm increases, the base shear capacity also increases due to pile batters and
large moment arms, pile system capacity is dominated by overturning where axial
Results from the finite element method analysis, including the estimated load
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
applied to the pile system at the peak of Hurricane Ike and the pushover capacity, are
d cr
also shown in Figure 9. The maximum load in Ike is approximately equal to the
te s
di nu
overturning failure of the pile system.
ye a
In order to assess the contribution of individual piles to the capacity of the pile
op M
system, a sensitivity analysis was performed by varying the capacity of each pile (Fig.
10). The maximum axial and lateral soil resistance mobilized at failure along the length
C ted
of each pile was increased and decreased by 30 percent to represent possible variations
in soil resistance due to uncertainties in soil conditions and pile driving effects. The
ot p
measured and predicted capacities for individual piles (e.g., Tang and Gilbert 1993).
Ac
Pile C is the most influential pile in this system for an overturning-dominated failure in
the loading direction of Hurricane Ike (Fig. 10). The capacity of this pile system in Ike
is essentially governed by the axial capacity of Pile C because the overturning capacity
DISCUSSION
load test on a driven, steel pipe pile in service. Both finite element method analysis and
upper-bound collapse analysis predict that Pile C would fail in tension during
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
Hurricane Ike at an estimated axial load of 16,000 kN (Fig. 8). The characteristics of
d cr
the load on the pile, actual versus predicted pile capacity, and suggestions to improve
te s
di nu
Characteristics of Axial Pile Load in Hurricane Ike
ye a
op M
The axial load on Pile C at failure is not known precisely, unlike the load in a controlled
load test. The estimated axial load of 16,000 kN is the median of the maximum load on
C ted
the pile during the 3-hour sea state in Hurricane Ike that produced the largest waves at
Platform EC368A. There are three sources of uncertainty in the estimated maximum
ot p
load: (1) the inherent variability in wave heights during the sea state (Forristall 1978);
N ce
(2) the uncertainty in estimating the properties of the sea state, primarily the wave
Ac
1991); and (3) the uncertainty in estimating base shear on the platform for a given
framework presented by Puskar et al. (1994) and ABS Consulting (2004) produces the
probability distribution for the maximum axial load on Pile C shown in Figure 11.
There is a 70-percent chance that the maximum load was between 12,000 and 22,000
kN and a 90-percent chance it was between 10,000 and 26,000 kN. While these wide
confidence bounds indicate healthy uncertainty in the axial load that caused this pile to
fail, indirect evidence suggests that the hindcast loads are less uncertain. Detailed
analyses comparing predicted loads from the hindcast with structural damage for
hundreds of platforms suggest that the median value is a reasonable indicator of the
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
actual load and performance of the structure (Puskar et al. 1994, ABS Consulting 2004,
d cr
and Energo Engineering 2007 and 2010).
te s
di nu
Pile C experienced one-way cyclic tensile loading from a mean of about zero
during the 3-hour sea state with the largest waves in Hurricane Ike. The maximum
ye a
tensile load is one of about 750 cycles of load applied to Pile C during the 3-hour sea
op M
state. The maximum load would have been sustained for approximately 3 seconds.
Figure 12 shows the expected number of cycles for different levels of loading based on
C ted
the distribution of wave heights during the sea state (Forristall 1978). There were
expected to be two cycles of axial loads exceeding 85 percent and 25 cycles of loads
ot p
The database used to develop the API design method for driven piles in
originally developed by Olson and Dennis (1982) and recently refined by Najjar (2005),
was used by Randolph and Murphy (1985) to develop the D-method for the design of
driven piles in clay. The D-method remains the recommended method in the recent
update to the design guidance (API 2011). The data shown in Figure 13 represent the
axial capacity for static, undrained loading after the dissipation of driving-induced pore
water pressures.
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
Pile C represents one of a small subset of data points corresponding to typical
d cr
pile dimensions and capacities used in offshore structures; it failed at an expected load
more than 50 percent higher than the load for any other data point (Fig. 13). While the
te s
di nu
average ratio between measured and predicted capacity for all data points is close to
one (e.g., Randolph and Murphy 1985 and Najjar 2005), there is a bias to over-predict
ye a
the capacity of longer piles with higher capacities (Fig. 13).
op M
for longer piles (e.g., Kraft et al. 1981b). In particular, the tests described by Sterling
(1971) and Bogard and Matlock (1998) are both in the highly plastic clays similar to
ot p
those at Platform EC368A. Murff (1980) and Randolph (1983) proposed simple,
N ce
non-dimensional indicators for the significance of strain softening that relate the side
Ac
shear stiffness to the axial stiffness of the pile. Randolph’s factor is , where
Ws is the average peak side shear, 'wres is the displacement required to reduce the side
shear from peak to residual, and D, L, A and E are the diameter, length, cross-sectional
area and Young’s modulus, respectively, for the pile. Higher K values indicate greater
potential for strain softening to reduce the maximum side shear mobilized at failure; K
values greater than one indicate significant strain softening. The K value for Pile C in
Platform EC368A is greater than four, and it exceeds one for all load tests in Figure 13
with measured capacities higher than 5,000 kN. The maximum capacity predicted from
the t-z analysis for Pile C is nearly identical to the expected load at failure, while the
“ultimate” capacity for a rigid pile over-predicts the load at failure, similar to the trend
t
ip
d cr
Degradation of side shear during pile driving also contributes to the bias in
over-predicting the capacity for longer piles using the API design method (e.g.,
te s
di nu
Aldridge and Schnaid 1993, Lehane and Jardine 1994, O’Neill 2001, and Randolph
2003). This effect may have been especially significant for the load tests described by
ye a
Pelletier and Doyle (1982) and Gibbs et al. (1993); both load tests involved low
op M
plasticity, silty clays where rapid dissipation of driving-induced pore water pressures
could cause relatively high radial effective stresses during driving. This effect also
C ted
likely contributed to the significantly over-predicted capacity for the longer pile
(re-driven with extension) in the load tests described by Sterling (1971), the notable
ot p
outlier in Figure 13. This longer pile was installed by driving the original pile with an
N ce
extension more than two and half years after the original pile was installed and loaded
Ac
to failure. Jardine et al. (2005) proposed an alternative method to the API method to
account for the degradation of side shear during pile driving and loading. Jardine and
Saldivar (1999) showed that this alternative method better predicted the measured
capacity for the load test described by Bogard and Matlock (1998). Unfortunately, the
soil properties required in this alternative method are not all directly measured in a
typical site investigation in the Gulf of Mexico, and they were not measured for the
Two notable distinctions between Pile C in Platform EC368A and all other data
points in Figure 13 are set-up and cyclic loading. Pile C was loaded to failure more than
5 years after installation, a set-up time more than twice as long as the set-up time for
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
any other data and much longer than most data. Also, Pile C was loaded to failure by
d cr
cyclic and rapid loads from the waves in a hurricane, unlike the static and slower
loading rates used for all other data points in Figure 13. Cyclic loading reduces the
te s
di nu
axial capacity, while a higher loading rate increases the capacity (e.g., Kraft et al.
1981a, Poulos 1981, Dunnavant et al. 1990 and Bea et al. 1999). The current state of
ye a
practice for offshore pile design is to assume that these two competing effects balance
op M
each other (e.g., Pelletier et al. 1993); however, some hypothesize that the rate effect is
more significant than the cyclic loading effect in a hurricane (e.g., Bea et al. 1999).
C ted
Based on the median maximum load that caused Pile C to fail, there is no strong
evidence of a significant bias in predicted capacity due to set-up, cyclic loading and
ot p
loading rate. However, the uncertainty in the actual maximum load on Pile C (Fig. 13)
N ce
does not allow for a definitive conclusion about these effects on the axial capacity of
Ac
Pile C.
there has been a widely-held perception that offshore pile designs are significantly
conservative (e.g., Aggarwal et al. 1996b and Bea et al. 1999). However, this case
history, together with other case histories from recent hurricanes (Gilbert et al. 2010),
t
ip
strain-softening effects and pile system effects in design.
d cr
The design guidance for offshore structures (API 2000) intends that strain
te s
di nu
softening in soils be considered: “In some circumstances, i.e., for soils that exhibit
strain-softening behavior and/or where the piles are axially flexible, the actual capacity
ye a
of the pile may be less than that given by Eq. 6.4.1-1 [for a rigid pile]. In these cases, an
op M
explicit consideration of these effects on ultimate axial capacity may be warranted.”
Similar guidance exists in the API geotechnical and foundation design considerations
C ted
(API 2011). However, the current state of practice does not necessarily achieve this
recommendations for t-z curves with strain softening (Fig. 6), they also provided
Ac
ultimate capacity profiles assuming rigid piles (Fig. 5) because the wall thicknesses for
the piles were not yet known. While the structural engineers included the t-z curves in
their structural analysis to establish the design axial loads on the piles, they selected
pile lengths using the ultimate capacity profiles for rigid piles because they did not
analyze the structure with load cases that fully mobilized the axial capacities of the
piles. As a result, the factor of safety for Pile C was lower than intended, at about 1.35
instead of 1.5. Since this pile failed when the expected maximum load in Hurricane Ike
was about 30 to 40 percent higher than the design load, it is possible that this failure
could have been prevented if strain softening had been considered explicitly in design.
We propose two ways to improve the state of practice with respect to strain
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
softening. First, we recommend a better communication between geotechnical and
d cr
structural engineers. There should be an explicit opportunity for the geotechnical
engineer to review the final design of the pile foundations before the platform is built.
te s
di nu
Second, we recommend that a t-z analysis be conducted to select the final pile lengths.
For example, the minimum wall thickness for Pile C was 19 mm compared to the actual
C ted
wall thickness that ranged from 25 to 38 mm (Table 1). While a t-z analysis using the
minimum wall thickness would be conservative, it is more realistic and defensible than
ot p
assuming a rigid pile. Other pile capacity methods that implicitly include the effects of
N ce
strain softening (e.g., Semple and Rigden 1984, Kolk and van der Velde 1996 and
Ac
Jardine et al. 2005) can also be used to check the maximum pile capacity.
platforms (API 2000 and 2011) is silent on pile system effects. Each pile, just like
every tubular member in the structure, is designed individually based on a check of its
design load and capacity. However, the same factor of safety applied to individual piles
may not provide a consistent level of performance between different pile systems (e.g.,
Tang and Gilbert 1993). Platform EC368A represents an extreme case where the
overturning capacity of the pile system was governed by the axial capacity of a single
pile in Hurricane Ike (Fig. 10). As an illustration, the robustness of this three-pile
system is compared in Figure 14 to a six-pile system for another platform in the Gulf of
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
Mexico. While both systems are significantly more robust in shear than in overturning,
d cr
the six-pile system is more robust in overturning to variations in the axial capacity of a
te s
di nu
ye a In order to consider system effects in design, we recommend that guidelines be
developed for design checks on the overall system in addition to individual members.
op M
Pushover analyses that check the ultimate capacity of the structural system including
the piles have not been part of the offshore platform design practice. However, the
C ted
design practice could include a global pushover capacity check to help ensure that the
foundation has adequate capacity and robustness on a system basis. Another possibility
ot p
checks for cases with one (damage) and two (survivability) mooring lines missing.
Ac
While removing an entire pile from a jacket platform is an unreasonable design check,
one that reduces the capacity of a single pile (specifically the most critical pile in the
foundation has adequate system capacity given the uncertainty in the capacity of
individual piles.
CONCLUSIONS
This paper presents a case history of the failure of a pile system supporting an
offshore platform in a hurricane. The fixed jacket platform was built in 2003 and failed
in 2008 during Hurricane Ike. The three-pile foundation system failed in overturning
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
due to a pull-out failure of the most critically loaded pile. This pile is a 1,220-mm
d cr
diameter, open-ended, steel pipe pile that was driven to a depth of 67.1 m below the
te s
di nu
plastic clay.
ye a
The estimated axial load on the pile at failure was 16,000 kN in tension. The
op M
calculated tensile capacity for this pile using the API Design Method is approximately
equal to the estimated axial load at failure, provided that pile flexibility and strain
C ted
softening in side shear are considered. If the pile is assumed to be rigid, the calculated
capacity is ten percent higher than the estimated load at failure. The actual load at
ot p
failure was estimated based on a hindcast analysis of the hurricane and it was likely
N ce
This case history is the first documented case of foundation failure for a jacket
platform in a hurricane, where the foundation system was loaded beyond its design
capacity and failed. It generally affirms current design methods because the failure is
not a surprise. It also contradicts a perception that offshore pile designs are
improve the current design practice. First, axial pile capacity should be checked in
design by explicitly accounting for pile flexibility when dealing with strain-softening
soils. Second, guidelines should be developed for design checks on the overall
t
ip
ACKNOWLEDGMENTS
d cr
The former U.S. Minerals Management Service (MMS) and American
te s
di nu
Petroleum Institute (API) provided financial support through the Offshore Technology
Research Center for research projects, upon which this case study is based. MMS, API
ye a
and W&T Offshore were instrumental in providing access to hindcast data, design
op M
information and damage assessments. Engineering Dynamics, Inc. donated their
Structural Analysis Computer Software™ for use in this work. Aditya Hariharan and
C ted
Matthew Garcia of Energo Engineering were helpful in providing data and analyses of
this platform. Don Murff, Alan Young and Philippe Jeanjean provided valuable review
ot p
and guidance throughout the research projects. Steve Kay reviewed the draft paper and
N ce
performed an independent pile capacity analysis. The views and opinions contained in
Ac
this paper are ours alone and do not necessarily reflect those of our employers,
sponsors or collaborators.
REFERENCES
ABS Consulting (2004), Hurricane Lili’s Impact on Fixed Platforms, Final Report
Aggarwal, R.K., Dolan, D.K. and Cornell, C.A. (1996a), “Development of bias in
Proceedings of Offshore Technology Conference, Paper No. OTC 8077, May 6-9,
Houston, Texas.
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
Aggarwal, R.K., Litton, R.W., Cornell, C.A., Tang, W.H., Chen, J.H. and Murff, J.D.
d cr
(1996b), “Development of platform foundation bias factors using observed
te s
di nu
Technology Conference, Paper No. OTC 8078, May 6-9, Houston, Texas.
Aldridge, T.R. and Schnaid, F. (1993), “Degradation of skin friction for driven piles in
ye a
clay,” Large-scale Pile Tests in Clay, edited by Clarke, J., published by Thomas
op M
Telford Services Ltd.
API (2000), Recommended Practice for Planning, Designing and Constructing Fixed
C ted
Bea, R.G., Jin, Z., Valle, C. and Ramos, R. (1999), “Evaluation of reliability of
Technology Conference, Paper No. OTC 8767, May 4-7, Houston, Texas.
Chen, J.Y, Gilbert, R.B., Murff, J.D., Young, A.G. and Puskar, F.J. (2010), “Structural
t
ip
Chen, J.Y., Materek, B.A., Carpenter, J.F., Gilbert, R.B., Verret, S. and Puskar, F.J.
d cr
(2009), Analysis of Potential Conservatism in Foundation Design for Offshore
Platform Assessment, Final Project Report Prepared for the Minerals Management
te s
di nu
Service, MMS Project Number 612.
Dunnavant, T.W., Clukey, E.C. and Murff, J.D. (1990), “Effects of cyclic loading and
ye a
pile flexibility on axial pile capacities in clay,” Proceedings of Offshore
op M
Technology Conference, Paper No. OTC 6378, May 7-10, Houston, Texas.
Katrina and Rita, Final Report Prepared for Minerals Management Services,
Offshore Structures and Pipelines in Hurricanes Gustav and Ike, Final Report
Ac
Forristall, G.Z. (1978), “On the statistical distribution of wave heights in a storm,”
Forristall, G.Z., Larrabee, R.D. and Mercier, R.S. (1991), “Combined oceanographic
Gibbs, C.E., McAuley, J., Mirza, U.A. and Cox, W.R. (1993), “Reduction of field data
and interpretation of results for axial load tests of two 762 mm diameter pipe piles
Gilbert, R.B., Chen, J.Y., Materek, B., Puskar, F., Verret, S., Carpenter, J., Young, A.
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
and Murff, J.D. (2010), “Comparison of observed and predicted performance for
d cr
jacket pile foundations in hurricanes,” Proceedings of Offshore Technology
te s
di nu
Haring, R.E., Olsen, O.A. and Johansson, P.I. (1979), “Total wave force and moment
ye a vs. design practice,” Proceedings of the Specialty Conference Civil Engineering in
for Driven Piles in Sands and Clays, Thomas Telford Ltd., London.
C ted
Jardine, R.J. and Saldivar, E. (1999), “An alternative interpretation of the West Delta
Jeanjean, P. (2009), “Re-assessment of p-y curves for soft clays from centrifuge testing
Ac
Kraft, L.M., Cox, W.R. and Verner, E.A. (1981a), “Pile load tests: cyclic loads and
Kraft, L.M., Focht, J.A. and Amerasinghe, S.F. (1981b), “Friction capacity of piles
driven into clay,” Journal of the Geotechnical Engineering Division, ASCE, Vol.
Kolk, H.J. and van der Velde, E.A. (1996), “Reliable method to determine friction
t
ip
Lehane, B.M. and Jardine, R.J. (1994), “Displacement-pile behavior in a soft marine
d cr
clay,” Canadian Geotechnical Journal, 31, No. 2, 181-191.
Matlock, H. (1970), “Correlations for design of laterally loaded piles in soft clay,”
te s
di nu
Proceedings of Offshore Technology Conference, Paper No. OTC 1204, April
ye a 22-24, Houston, Texas.
Murff, J.D., Lacasse, S. and Young, A.G. (1993), “Discussion and summary on
C ted
Murff, J.D. and Wesselink, B.D. (1986), “Collapse analysis of pile foundations,”
Ac
Oceanweather (2008), Fast Delivery Hindcast of Hurricane Ike (2008) in the Northern
Olson, R.E. and Dennis, N.D. (1982), Review and Compilation of Pile Test Results,
Axial Pile Capacity, Final Report to American Petroleum Institute, Project PRAC
O’Neill, M.W. (2001), “Side resistance in piles and drilled shafts,” Journal of
t
ip
Pelletier, J.H., Murff, J.D. and Young, A.C. (1993), “Historical development and
d cr
assessment of the current API design methods for axially loaded piles,”
Proceedings of Offshore Technology Conference, Paper No. OTC 7157, May 3-6,
te s
di nu
Houston, Texas.
Pelletier, J.H. and Doyle, E.H. (1982), “Tension capacity in silty clays – beta pile test,”
ye a
Proceedings, 2nd International Conference on Numerical Methods in Offshore
op M
Piling, Austin, Texas.
Poulos, H.G. (1981), “Cyclic axial response of single pile,” Journal of the
C ted
Puskar, F.J., Aggarwal, R.K., Cornell, C.A., Moses, F., Petrauskas, C. (1994), “A
ot p
Puskar, F.J., Ku, A.P. and Sheppard, R.E. (2004), “Hurricane Lili’s impact on fixed
Puskar, F.J., Spong, R.E., Ku, A.P., Gilbert, R.B. and Choi, Y.J. (2006), “Assessment
Puskar, F.J., Verret, S.M. and Roberts, C. (2007), “Fixed platform performance during
Randolph, M.F. (1983), “Design considerations for offshore piles,” Proceedings of the
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
t
ip
Conference on Geotechnical Practice in Offshore Engineering, ASCE, Austin,
d cr
Texas, 422-439.
te s
di nu
Geotechnique, Vol. 53, No. 10, 847-875.
Randolph, M.F. and Murphy, B.S. (1985), “Shaft capacity of driven piles in clay,”
ye a
Proceedings of Offshore Technology Conference, Paper No. OTC 4883, Houston,
op M
Texas.
Semple, R.M. and Rigden, W.J. (1984), “Shaft capacity of driven pipe piles in clay,”
C ted
Sterling, G.H. (1971), Soil-Pile Adhesion at Eugene Island Block 100, and One Shell
N ce
Tang, W.H. and Gilbert, R.B. (1993), “Case study of offshore pile system reliability,”
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
Figure 3. Soil stratigraphy, index properties and submerged unit weight for pile design.
t
Figure 5. Ultimate axial capacity profiles for 1,220-mm diameter vertical pipe pile.
ip
Figure 6. Recommended t-z curve for 1,220-mm diameter pipe piles.
d cr
Figure 7. Schematic of foundation system collapse model.
te s
Figure 8. Results from t-z analysis for Pile C in tension.
di nu
Figure 9. Results from upper-bound collapse analysis for Platform EC368A pile
system in Hurricane Ike loading direction.
ye a
Figure 10. Results from sensitivity analysis for Platform EC368A pile system by
op M
increasing and decreasing the axial and lateral soil resistance on (a) Pile A, (b) Pile B
and (c) Pile C by 30 percent in Hurricane Ike loading direction.
Figure 11. Probability distribution for maximum axial load on Pile C in Hurricane Ike.
C ted
Figure 12. Expected number of cycles for different levels of axial load in tension on
Pile C in Hurricane Ike.
ot p
Figure 13. Comparison of axial capacity of Pile C with other pile load tests used to
develop the current design guidance for axial pile capacity in clays.
N ce
Figure 14. Comparison of pile system robustness for (a) three-pile platform (Platform
EC368A) versus (b) Six-pile platform (after Chen et al. 2010).
Ac
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
Evidence that
platform was
leaning after
Hurricane Ike
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
Leg C
Leg A
Leg B
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
Hurricane Loading
Seafloor Direction on Foundation
True North
Approx. Axis
of Rotation
Conductor
Pile B
Pile A
Pile C
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
40
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
60 Soft to Hard
Olive Gray Clay
with intermittent
80 silt partings,
Liquid
Limit seams and
Plastic pockets
100
Limit
Water Design
Content Profile
120
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
40 Miniature Vane 40 40
Remolded -
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
Miniature Vane
Pile B and C Pile B and C
60 60 60
Tip Elevation Tip Elevation
80 80 80
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
40
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
60 67.1 m
80
100
Axial Capacity in Compression
Axial Capacity in Tension
120
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
0.8
0.6
t / tmax
0.4
0.2
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
0
0 20 40 60 80
z (mm)
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
Rigid structure
Mudline
Plastic hinges
Well
Pile Pile
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
80
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
120
160
Residual Peak Side Shear
Capacity Mobilized
200
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
12000
Maximum
Capacity
Upper-Bound
Combined Mobilized
Collapse on
Shear Failure All Piles
0
0.0E+00 2.0E+05 4.0E+05 6.0E+05 8.0E+05
Overturning Moment (kN-m)
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
12000
Axial and lateral
soil capacities of
Pile A increased
by 30%
Load in Ike
(a) Pile A
0
0.0E+00 2.0E+05 4.0E+05 6.0E+05 8.0E+05
Overturning Moment (kN-m)
12000 12000
Axial and lateral
soil capacities of
Pile B increased
by 30% Maximum
Base Shear (kN)
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
Median of Maximum
Axial Load = 16,000 kN
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
100
10
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
0.1
50% 60% 70% 80% 90% 100%
Fraction of Median Maximum Axial Load
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
20000
pre-1984 Load Tests used to Develop API Method
Post-1984 Load Tests
Pile C in Platform EC368A with 70% Conf. Bounds
Maximum Capacity
Measured Capacity (kN) 15000 Predicted from t-z Analysis
Ultimate
Capacity
Long Beach Load Test
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
15000
(a) Three-pile platform (Platform EC368A)
12000
6000
Downloaded from ascelibrary.org by Monash University on 09/05/13. Copyright ASCE. For personal use only; all rights reserved.
Maximum
Load in Ike
3000 Range of Foundation System Capacity due to
Increasing and Decreasing the Axial and
Lateral Soil Capacities of One Pile by 30 %
0
0.0E+00 2.0E+05 4.0E+05 6.0E+05 8.0E+05
Overturning Moment (kN-m)
30000
(b) Six-pile platform
25000
20000
Base Shear (kN)
15000
Maximum
Load in Katrina
10000
Range of Foundation System Capacity due to
Increasing and Decreasing the Axial and
5000 Lateral Soil Capacities of One Pile by 30 %
0
0.0E+00 6.0E+05 1.2E+06 1.8E+06
Overturning Moment (kN-m)
Accepted Manuscript
Not Copyedited
Journal of Geotechnical and Geoenvironmental Engineering. Submitted December 29, 2011; January 14, 2013;
posted ahead of print January 16, 2013. doi:10.1061/(ASCE)GT.1943-5606.0000894
1,220
3 27.4-30.5 1,220 32 22.9-65.5 25
4 30.5-79.2 25 65.5-67.1 32
5 79.2-80.8 32 Not Applicable
Accepted Manuscript
Not Copyedited