Vous êtes sur la page 1sur 8

Frontiers in Offshore Geotechnics: ISFOG 2005 Gourvenec & Cassidy (eds)

2005 Taylor & Francis Group, London, ISBN 0 415 39063 X

Bearing capacity and large penetration of a cylindrical object


at shallow embedment
E.R. Barbosa-Cruz & M.F. Randolph
Centre for Offshore Foundation Systems, The University of Western Australia, Perth, Australia

ABSTRACT: The penetration of a cylindrical object into soft clay starting from a very small embedment
has been investigated using a large deformation finite element approach, the Remeshing and Interpolation
Technique with Small Strains (RITSS). The study has application to pipeline penetration of the seabed, and also
to the interpretation of cylindrical penetrometers such as the T-bar at very shallow depths. The results of the
analyses show the evolution of bearing capacity factor and soil flow mechanisms as the cylindrical object is
penetrated from 0.5% of the diameter up to nearly 5 diameters. Comparisons of results are made for fully
smooth and fully rough interfaces and for homogeneous and non-homogeneous soil profiles.

INTRODUCTION

The penetration resistance of an infinite cylindrical


object at shallow embedment into soft soils is of
importance to offshore pipeline design and the interpretation of T-bar penetration tests close to the surface. Previous studies of pipeline penetration have
made use of bound solutions (Murff et al. 1989, Aubeny
et al. 2005) but these do not account of factors such
as: local heave; variation of shear strength with depth;
and disturbance of the strength profile due to pipe
penetration. For T-bar tests (Stewart & Randolph 1991,
1994), a constant bar factor is generally adopted for
estimating the shear strength from the bearing resistance, regardless of depth. However, at very shallow
depths this factor should be reduced.
At deep penetration, the limiting effective force P
per unit length of cylinder may be related to the shear
strength su of the soil by a bearing factor Nb:
(1)
where D is the diameter of the cylinder. Randolph &
Houlsby (1984) presented a plasticity solution for this
factor, with a lower bound increasing from 9.14 for
fully smooth to 11.92 for fully rough interface conditions. Subsequent upper bound solutions give a very
similar range, from 9.21 to 11.92 (Randolph 2004).
For interpretation of T-bar tests, a bearing factor of
10.5 has been suggested.
Near the surface, however, the bearing factor will
reduce and in the limit at very shallow penetration the

penetration resistance will reduce to the classical


result of 5.14Bsu where B is the contact width between
pipeline and soil.
A numerical study has been undertaken using large
deformation finite element analysis in order to follow
the evolution of the bearing resistance and deformation
mechanism during pipe penetration from the surface
to full burial. The analyses have covered fully smooth
to fully rough interface conditions and both uniform
soil and soil with a significant strength gradient.

2
2.1

NUMERICAL ANALYSIS
Details of the analysis

Deep penetration of objects into the soil from the surface implies very large strains and deformations within
the soil, and thus cannot be simulated using conventional small strain finite element analysis. Hu &
Randolph (1998a) proposed a process that they termed
Remeshing and Interpolation Technique with Small
Strain or RITSS, which falls in the category of
Arbitrary Lagrangian-Eulerian (ALE) finite element
techniques, to analyse this kind of problem. This technique combines a series of small-strain analysis followed by a complete remeshing of the domain and then
interpolation of the stress field and material parameters between the Gauss points of the new and old
meshes. Details of the technique and strategies for the
size of displacement increments and frequency of
remeshing may be found in Hu & Randolph (1998a,
1998b) and Lu et al. (2000, 2001, 2004).

615

The RITSS technique can accommodate remeshing


the domain totally or partially. In the present study the
domain has been remeshed partially, wherever the
nodal displacements exceeded 10% of the maximum
displacement; in the remainder of the domain the
position of the nodal coordinates were updated
according to the displacements. Additionally a minimum zone for remeshing was adopted, bounded
approximately by a circular line centred on the cylinder centre and with a radius equal to 5 times the cylinder radius. Partial remeshing is useful to minimise
any error in the interpolated stress field close to the
domain boundaries, where the mesh is coarse.
Taking advantage of the vertical axis of symmetry
only half of the domain was modelled, using sixnoded triangular elements with three Gauss points;
nodal joint elements were used to model the soilcylinder interface. Four different combinations of
cylinder roughness (fully smooth and fully rough),

Figure 1. Initial finite element mesh (distance scale in


cylinder diameters).
Table 1.

and strength profiles (homogeneous and nonhomogeneous) were modelled.


The soil was represented as a simple elastic, perfectly plastic material with Tresca failure criterion.
Poissons ratio was taken as 0.49, and the modulus
ratio, E/su, was taken as 500. A shear strength profile
of su  5  1.5 z/D kPa was adopted for the nonhomogeneous soil analyses, where z/D is the depth
normalised by the cylinder diameter. For the homogeneous soil the shear strength was taken as su  9 kPa.
Initial geostatic stresses were calculated based on an
effective unit weight for the soil of D  6 kPa (typical for submerged soft soil with a pipe diameter of
about 1 m) and K0 of unity. The quantities su and 
have been non-dimensionalised with respect to D in
order to allow application of the results for D values
close to 1 m, with appropriate adjustment to .
The cylinder was modelled as a cylindrical boundary that was advanced into the soil. The size of the
finite element mesh was chosen to ensure that the
boundaries were far from the plastic zone. The horizontal and vertical extends of the domain were taken
as 15 times the cylinder diameter.
The large penetration analyses were initiated from
a point where the cylinder was very slightly embedded (by 0.5% of the diameter). Figure 1 shows an
example of an initial mesh used in the analysis. As a
check, analyses were performed from an initial preembedment of half a diameter, to confirm that the
response resistance curves from the two different preembedments would merge at deep penetration.
The details of the analysis presented here are summarised in Table 1. In the case of the slightly preembedded fully smooth cylinder penetrating into
non-homogeneous soil, two different analyses were
performed to evaluate the effect of the minimum
mesh size on the results. The combined effect of the
minimum mesh size and remeshing interval was
assessed for the same cylinder and soil conditions,
but for pre-embedment by half a diameter.

Details of finite element analyses.

Analysis

Initial
embedment (m)

Minimum
mesh size (m)

Displacement
increment (m)

Remeshing
interval (m)

Cylinder
roughness

Soil strength
profile*

c1
c2
c3
c4
c5
c6
c7
c8
c9
c10

0.005
0.005
0.005
0.005
0.005
0.5
0.5
0.5
0.5
0.5

0.05
0.01
0.01
0.01
0.01
0.01
0.05
0.05
0.05
0.05

0.00005
0.00005
0.00005
0.00005
0.00005
0.00005
0.00005
0.00005
0.00005
0.00005

0.001
0.001
0.001
0.001
0.001
0.001
0.005
0.005
0.005
0.005

Smooth
Smooth
Rough
Smooth
Rough
Smooth
Smooth
Rough
Smooth
Rough

Non homogeneous
Non homogeneous
Non homogeneous
Homogeneous
Homogeneous
Non homogeneous
Non homogeneous
Non homogeneous
Homogeneous
Homogeneous

* Strength profile: Non homogeneous, su  5  1.5 z/D kPa; Homogeneous, su  9 kPa; (E  500su; v  0.49 for both).

616

It should be noted that this type of analysis is


computationally intensive. Using a computer with a
2.81 GHz Intel Pentium Processor and 2.0 GB of
RAM the analysis times were between 1 to 30 days,
mainly depending on the density of the mesh and the
roughness of the cylinder. For the 10 analyses presented here a total computing time of 111 days were
required. It was found necessary to use a very fine
mesh and frequent remeshing, particularly during the
initial penetration of the cylinder, where the cylindersoil contact region expanded rapidly.
The need for the finer mesh and smaller remeshing
interval at shallow embedment is shown in Figure 2,
which compares analyses c1 and c2 up to an embedment of 0.2D. The two analyses converge for displacements greater than 0.1D, but the coarser mesh,
c1, shows significant jumps at certain remeshing
stages at shallow embedment.
2.2

Processing of results

The computed loads are reported as a bearing capacity factor, Nc, calculated in broad terms as the cylinder load (per unit length) divided by the contact width
(maximum of 1 cylinder diameter) and then by a characteristic soil strength. Details of this processing of
the data are given below.
As the cylinder penetrates the soil, the theoretical
contact width is given by (Figure 3)

For penetration greater than D/2 (  /2), B


becomes equal to the diameter, D. Note that since
some heave will occur adjacent to the cylinder, the
actual contact width, B, will initially exceed the theoretical contact width. This is shown in Figure 4, with
results from the finite element analyses c1 and c2
compared with the theoretical relationship.
Since the cylinder is modelled as a hollow cylindrical zone penetrating soil with a non-zero effective
unit weight, a buoyancy force will act on the cylinder.
This buoyancy force is calculated from the (theoretical) volume of cylinder below the soil surface, ignoring local heave, multiplied by the effective unit
weight of the soil. This volume may be related to the
angle, , by
(4)
Once the embedment exceeds the cylinder diameter,
the buoyancy force per unit length of cylinder stays at
R2.
The average contact pressure resisted by the soil is
then calculated as the force per unit length of cylinder
(from the finite element results), less the buoyancy
force, divided by either the theoretical contact width, B,

cylinder

(2)
where R is the radius of the cylinder (D/2), and  is
the semi-angle subtended at the cylinder centre by the
theoretical contact width, which is related to the nominal embedment depth, z, by

heave

seabed
B

nominal
embedment,

B

(3)

Figure 3.

Geometry of cylinder penetrating seabed soil.

1.2
20
Minimum element size 0.01 m

16

c1

14

c2

12
10

Minimum element size 0.05 m

8
6
4

Smooth cylinder
Soil properties su = 5 + 1.5 z/D kPa, E = 500 su

Theoretical
c1
c2

1
Contact widths
B/D or B'/D

Load kN/m

18

0.8
0.6

Analysis Minimum
mesh size
mm
c1
50
c2
10

0.4
0.2

0
0

0.05

0.1

0.15

0.2

Step
size
mm
0.05
0.05

Remeshing
every
mm
1.0
1.0

0
0

Embedment, z/D

Figure 2. Comparison of load-displacement responses


from analyses c1 and c2.

Figure 4.

617

0.1

0.2
0.3
Embedment, z/D

0.4

Variation of contact width with embedment.

0.5

Table 2.

7
Nc max = 5.89
Minimum element size 0.05 m

c2

Case

5
Nc

Nc max = 5.13
Minimum element size 0.01 m

Nc computed
contact width

2
Smooth cylinder
Soil properties su = 5 + 1.5 z/D kPa, E = 500 su

1
0
0

0.05

0.1

0.15

0.2

Normalised embedment, z/D

Figure 5.

Bearing capacity factors.

c1

Detailed variation of bearing capacity factor, Nc.

Initial Nc
P/Bsuo
5.48
5.52
6.12
5.15
5.54

c1
c2
c3
c4
c5
c6
c7
c8
c9
c10

Minimum Nc

P/Bsuo
5.89
5.13
5.43
5.02
5.25
5.17
5.14
6.63
4.49
5.86

P/Bsuo
4.80
4.81
6.01
4.50
5.42

P/Bsuo
4.23
4.18
5.24
4.03
4.96

Max.
Nc

Final
Nc

P/Dsuo
9.10
9.01
10.83
9.26
11.99
9.04
9.48
11.49
9.28
12.01

P/Dsuo
8.96
8.97
10.72
9.25
11.97
8.97
9.40
11.46
9.27
12.01

12
Minimum element size 0.05 m
re-meshing every 0.005 m

10

Nc = 9.40
Nc = 8.97

10

8
Nc 6

8
c7

6
4

0
1

Nc computed
contact width, c2

Nc 5

Smooth cylinder
Soil properties su = 5 + 1.5 z/D kPa, E = 500 su
0

c6, Nc = 8.97

c6

Normalised embedment, z/D

Smooth cylinder
Soil properties su = 5 + 1.5 z/D kPa, E = 500 su

Figure 6.

Overall variation of bearing capacity factor, Nc.

or the actual contact width, B. In order to calculate a


bearing capacity factor, Nc, the contact pressure is
then divided by a characteristic soil strength, suo,
taken as that at the widest (theoretical) contact width
of the cylinder with the soil. Thus suo is taken as the
surface soil strength, sum, for embedment depths less
than D/2, and as
(5)
where k is the strength gradient.
The bearing capacity factor Nc for a smooth cylinder may be expected to increase from around 2  
initially to 9.2 (average of upper and lower bound solutions) once the cylinder is fully embedded. Figures 5
and 6 show the need for finer mesh and smaller remeshing intervals for shallow and deep pre-embedment.
Using the finer mesh size of 1% of diameter and the
smaller remeshing interval equal to 0.1% of diameter
from the analysis, Nc varies from 5.13 at shallow
depth to 8.97 after deep penetration.
3

c2, Nc = 8.97

Nc theoretical
contact width, c2

9
Minimum element size 0.01 m
re-meshing every 0.001 m

BEARING CAPACITY FACTOR

The bearing capacity factors, Nc, are summarised in


Table 2. The overall variation of Nc is shown in Figure 7

Normalised embedment, z/D

Figure 7.

Bearing capacity factor, Nc for smooth cylinder.

for initial embedments of 0.005D and 0.5D for a fully


smooth cylinder in non-homogeneous soil (analyses
c2 and c6). For the former case, results are presented
in terms of the theoretical contact width, B, and the
computed contact width, B, allowing for soil heave.
Details of the corresponding Nc variation for shallow
embedment are shown in Figure 8.
As commented above, the value of Nc increases
from just over 5 at very shallow embedment to just
over 9 once it becomes fully embedded. The computed asymptotic Nc factor at large embedment is
about 2% below the theoretical lower bound of 9.2,
which is excellent accuracy considering the complexity of the analysis. The curves in Figure 7 show a
slight jump at embedment greater than 2.5D. This
jump is associated with closure of the soil around the
top of the cylinder; a small region of water appears to
be trapped immediately above the cylinder before closure, but the remeshing strategy ignores this and
replaces this zone with soil (see later, Figure 11). The
close agreement of the occurrence of this closure,
with that for an initial embedment of 0.005D occurring just ahead of that for an initial embedment of

618

14
Nc theoretical contact width
Nc max = 5.52

7
6

10

Nc min = 4.81

Nc

5
Nc 4

Nc max = 5.13
Nc computed contact width

c2,Nc = 8.97
c3,Nc = 10.72
c4,Nc = 9.25
c5,Nc = 11.97

12

8
6

Nc min = 4.18

c2,
c3,
c4,
c5,

4
2

2
1

Smooth cylinder
Soil properties su = 5 + 1.5 z/D kPa, E = 500 su

0
0

0.1

0.2

0.3

0.4

Non homogeneous soil, smooth cylinder


Non homogeneous soil, rough cylinder
Homogeneous soil, smooth cylinder
Homogeneous soil, rough cylinder

2
3
4
Normalised embedment, z/D

0.5

Normalised embedment, z/D

Figure 9.

Summary of overall variations in Nc.

Figure 8. Detailed variation of Nc at shallow embedment


for smooth cylinder.
8
su = 5 + 1.5 z/D kPa
6
Nominal Nc

0.5D (with additional soil heave present in the former


analysis) is further confirmation of the robustness of
the analyses.
At shallow embedment (Figure 8) the initial peak in
Nc based on the computed contact width is very close
(within 0.2%) to the theoretical value of 5.14. The
bearing capacity factor then reduces slightly reaching
a local minimum of 4.18 (based on computed contact
width) or 4.81 (theoretical contact width) for embedments of 0.18D to 0.24D. This is due to the smooth
cylinder-soil interface condition adopted, so that the
cylinder surface represents a weakness introduced into
the soil, reducing the bearing capacity below the theoretical value for a flat strip.
The reverse would occur for a fully rough cylinder
placed into non-homogeneous soil (analysis c3),
although in this case soft soil from the surface level
will be dragged down more with the cylinder, so there
is a combined effect of the cylinder geometry and
distortion of the soil stratigraphy (Figure 9). In this
case the Nc based on computed contact width, B, is
5.43 at shallow penetration then reduces slightly to
5.24 and reaches an asymptotic value of 10.72 at deep
penetration.
For the homogeneous strength profile (analyses c4
and c5) the bearing capacity factor Nc of the cylinder
calculated from the computed contact width varies
from 5.02 (fully smooth) to 5.25 (fully rough) at shallow embedment (Figure 9). Then the Nc decreases to
4.03 (fully smooth) and 4.96 (fully rough) and reaches
limiting values at deep displacement between 9.25
(fully smooth) and 11.97 (fully rough). At shallow
embedment the values are very close to the theoretical lower value 5.14 and for deep penetration the calculated Nc values are in agreement with the plastic
solution of 9.14 for a fully smooth cylinder and 11.94
for a fully rough cylinder.

c2
c3

su = 9 kPa

c4
c5

E = 500 su
0
0

0.1

0.2

0.3

0.4

0.5

Normalised embedment, z/D

Figure 10.

Nominal bearing capacity factor, P/Dsuo.

It is also useful to see the variation of the nominal


bearing factor, P/Dsuo (where P is the net cylinder
force, allowing for buoyancy effects), which is plotted
in Figure 10. This factor increases monotonically,
reaching Nc values between 5.00 and 6.64 depending
on the interface roughness and the soil profile once
the pipe embedment is 0.5D.
Examples of the patterns of incremental displacement vectors for the fully smooth cylinder and non
homogeneous soil (analysis c2) are shown in Figure 11
for cylinder embedments of 0.014D, 0.205D, 1.505D,
2.524D (before closure), 2.525D (after closure), and
2.555D. The incremental displacements show the
classical rotational velocity field around the fully
embedded cylinder, extending out to about 2.2 cylinder radii from the centre of the cylinder. The plastic
zones (not shown here) were somewhat larger than
anticipated from upper bound solutions, but this did
not seem to affect the overall load displacement
response of the flow patterns.

619

z/D = 0.014

z/D = 0.205

z/D = 2.525 (just after enforced closure)

z/D = 1.505

z/D = 2.555

z/D = 2.524 (just before closure)

Figure 11. Incremental displacement vectors at increasing penetration of cylinder (analysis c2: smooth cylinder in nonhomogenous soil).

CONCLUSION

This paper has summarised the results of large displacement finite element analysis of a cylindrical
object penetrating soft clay. The work used the RITSS
algorithm developed by Hu & Randolph (1998a). The
aim of the study was to evaluate the gradual evolution

of the bearing factor for a cylindrical object penetrating from very small initial embedment. The analyses
involved a 1 m diameter cylinder displaced into soft
soil, from an initial embedment of 0.005D, to a final
embedment from 3 to 5D.
It was found that the bearing capacity factor (expressed as force normalised by contact width times

620

soil strength at maximum cylinder contact width)


depends on the combination of the interface roughness and soil strength profile. The calculated Nc values are close to those predicted from plasticity theory.
Analyses run from an initial embedment of 0.5D
showed essentially identical resistance to that for the
initial embedment of 0.005D, with the curves merging by a penetration of 1.5D.
The analyses showed that the mechanism gradually
evolved from a classical Prandtl field at shallow
embedments to a rotational mechanism similar to
confined flow of soil around a laterally loaded pile.
Closure of the soil over the top of the cylinder
occurred (for the particular soil properties modelled)
at an embedment from 2.4D to 4.2D.
ACKNOWLEDGMENTS
This work was carried out while the first author was
an International Postgraduate Research Scholar in
Australia. Part of the work was undertaken in collaboration between the Norwegian Geotechnical Institute
(NGI) and the Centre for Offshore Foundation Systems
(COFS), University of Western Australia.

REFERENCES

Hu, Y. & Randolph, M.F. 1998a. A practical numerical


approach for large deformation problem in soil. Int. J.
Num. and Anal. Methods in Geomech., 22(5): 327350.
Hu, Y. & Randolph, M.F. 1998b. Deep penetration of shallow foundations on non-homogeneous soil. Soils and
Foundations, 38(1): 241246.
Lu, Q., Hu, Y. & Randolph, M.F. 2000. FE analysis for T-bar
and ball penetration in cohesive soil. Proc. 10th Int.
Offshore and Polar Engineering Conf. ISOPE 2000,
Seattle, USA, 2: 617623.
Lu, Q., Hu, Y. & Randolph, M.F. 2001. Deep penetration in
soft clay with strength increasing with depth. Proc. 11th
Int. Offshore and Polar Engineering Conf. ISOPE 2001,
Stavanger, Norway, 2: 453458.
Lu, Q., Randolph, M.F., Hu, Y. & Bugarski, I.C. 2004.
A numerical study of cone penetration in clay.
Geotechnique 54(4): 257267.
Murff, J.D., Wagner, D.A. & Randolph, M.F. 1989. Pipe
penetration in cohesive soil. Geotechnique 39(2):
213229.
Randolph, M.F. 2004. Characterisation of soft sediments for
offshore applications, Keynote Lecture, Proc. 2nd Int.
Conf. on Site Characterisation, Porto, 1: 209231.
Randolph, M.F. & Houlsby, G.T. 1984. The limiting pressure
on a circular pile loaded laterally in cohesive soil.
Geotechnique 34(4): 613623.
Stewart, D.P. & Randolph, M.F. 1991. A new site investigation tool for the centrifuge. Proc. Centrifuge 91, Boulder,
Balkema, 531538.
Stewart, D.P. & Randolph, M.F. 1994. T-bar penetration testing in soft clay. J. Geotechnical Engineering, ASCE,
120(12): 22302235.

Aubeny, C.P., Shi, H. & Murff, J.D. 2005. Collapse loads for
a cylinder embedded in trench in cohesive soil. Int. J. of
Geomechanics, ASCE (in press).

621

Vous aimerez peut-être aussi