Drag anchor fluke-soil interaction in clays

M.F. Bransby & M.P. ONeill

Centre for Offshore Foundation Systems, The University of Western Australia, W.A., Australia

ABSTRACT: Finite element analysis of the fluke-soil interaction behaviour of drag anchors in undrained soil
has allowed calculation of plastic yield loci for characterisation of fluke failure states. The yield loci produced
are examined in terms of soil deformation mechanisms and kinematics and are incorporated into a novel
method for drag anchor design.
1 INTRODUCTION
The movement to offshore developments in deeper
water has led to increasing reliance on offshore
structures which are tethered to the seabed rather
than resting upon it. In addition, these are often required to be anchored in soft, normally consolidated
silts and clays.
One common seabed mooring system is the drag
anchor and chain system (Figure 1). The anchor is
lowered to the seabed surface and then installed into
the seabed by dragging the chain laterally. Due to the
shape of the anchor, kinematics govern so that it
must embed during displacement and hence significant holding capacity may be achieved. Most holding capacity is achieved by soil resistance against the
fluke, but the kinematics are controlled by the length
of the shank and the orientation of the shank with
respect to the fluke.
The use of drag anchors in offshore mooring systems requires knowledge about the anchors holding
capacity, embedment depth and drag length required
for mobilisation of the working load. These all vary
for different soil conditions and drag anchor designs.
Historically, drag anchor design has been empirically based (Vryhof Anchors, 1990). More recently,

approaches have been introduced which seek to predict the entire drag trajectory of an anchor from installation to mobilisation of working load capacity
(Neubecker and Randolph, 1996; ONeill and
Randolph, 1997; Thorne, 1998). These approaches
are based on combined kinematic and equilibrium
analyses. They use basic geomechanics principles to
find the geotechnical resistive loads acting on the
anchor in the equilibrium solution and simple assumptions about the anchor movement in the kinematic solution.
The above design approaches use empiricism for
solution (for example, the anchor form factor f,
Neubecker and Randolph, 1996). There appears to
be room for non-empirical prediction methods using
basic soil mechanics.
When drag anchor holding capacity is approached
in soft undrained soils, soil failure consists of local
plastic flow around the fluke and shank. Thus, the
behaviour will be independent of the orientation of
the anchor with respect to the soil surface and so
analysis of the soil around the fluke will lead to insight about the general anchor behaviour.
This paper presents analysis of the soil around
anchor flukes using finite element analysis. The
work allows investigation of the anchor under general loading conditions and the behaviour has been
characterised in terms of a plastic yield envelope.
The findings can be used to validate previous design
methods (e.g. Neubecker and Randolph, 1996) or
alternatively will lead to novel numerical design
methods using the yield envelope approach which
are introduced briefly.
2 PLASTICITY CONCEPTS AND THE YIELD
LOCUS

Figure 1. Drag anchor and chain system

Bransby & ONeill: Drag anchor fluke-soil interaction in clays

Recently, the analysis of offshore foundations such

1

as spudcan and shallow foundations have used plasticity concepts to express the bearing capacity of
foundations under combined vertical (V), horizontal
(H) and moment (M) loading (Murff, 1994; Tan,
1990; Martin, 1994; Bransby and Randolph, 1998).
A plastic yield locus is introduced which expresses the combination of V-H-M loads that result
in foundation failure. This can be given as a mathematical expression, f where f(V,H,M) = 0 at yield, or
shown graphically (Figure 2). Not only may the yield
locus be used to calculate footing capacity under
combinations of loads, but may also form a plastic
potential (g(V,H,M)) for description of plastic vertical, horizontal and rotational displacements of the
footing at failure (Figure 2). Indeed, plasticity theorems (e.g. Chen, 1975) show that the condition of
normality or associated flow will exist for undrained
failure conditions when the soil also remains attached to the footings.
Consider a drag anchor deep in undrained soil
(Figure 3). Loads exerted on the anchor can be expressed in terms of forces parallel (H) and perpendicular (V) to the fluke and moment load (M) about
any one particular reference point on the fluke.
Combinations of these loads (H, V and M) will
cause failure of the anchor with consequent anchor
movements parallel and perpendicular to the fluke
and rotationally about the same reference point.
Because the anchor is deep, soil failure will be
fully constrained and local to the anchor whatever
the direction of the load. Thus, the failure loads H, V
and M will be independent of anchor orientation,
and so the local load and displacement definitions
introduced in Figure 3 may be appropriate generally
for all failure conditions. In addition, the deep condition ensures that there is no soil-anchor detachment. Thus, plastic displacements will be governed
by normality to the failure yield locus, f(H,V,M) and
so determination of the failure locus also allows prediction of anchor displacement directions at failure.
Anchor failure loads (H, V and M) will be a result
of very complicated three-dimensional soil displacements around a complex drag anchor geometry.
To examine this in detail, full 3D analysis would be
required and the detailed 3D geometry of the anchor
would have to be incorporated. This would be extre-

Figure 2. The yield locus and plastic potential function.

Bransby & O Neill: Drag anchor fluke-soil interaction in clays

Figure 3. Loads and displacements at failure for a simplified

drag anchor.

mely time consuming. It is believed that much of the

complex mechanistic behaviour may be understood
by examining a single part of the drag anchor system: the fluke. This will provide a large proportion
of the holding capacity of the anchor and govern
much of the anchor kinematics.
The fluke-soil interaction behaviour is examined
using the finite element analysis below and the results are expressed as plastic yield loci. Methods of
utilising the yield locus in anchor design are introduced and discussed in following sections.
3 FINITE ELEMENT ANALYSIS
The aim of the finite element analyses was to examine soil-fluke interaction behaviour during
undrained, deep failure conditions. In particular, the
aim was to deduce the shape of the H, V, M yield
loci for flukes of different shapes.
The fluke shape was idealised as being infinitely
wide to allow plane strain analysis. Although this is
not necessarily a good approximation to reality, it
will allow detailed analysis of a simpler geometry
which may enable better understanding of the general soil-fluke interaction problem.
Two simple fluke geometries were investigated: a
rectangular fluke and an eccentric wedge. Dimensions were defined as shown in Figure 4. For both
flukes, a length over depth ratio L/d = 7 was
adopted, as this is similar to that of a Vryhof
Stevpris anchor fluke (Vryhof, 1990).
The finite element package CRISP, CRItical State
Program (Britto and Gunn, 1987) was used. In the
analyses, approximately 800 15-noded triangular
elements with 27 degrees of freedom were employed
(cubic strain triangles). Tresca elastic-plastic soil
was used with G/su = 500 where G is the shear

Figure 4. Simplified fluke geometry and definitions.

2

modulus and su is the undrained shear strength. The

soil was modelled as undrained by use of total stress
analysis together with a Poisson s ratio, = 0.49.
Displacement controlled analyses are most suitable for determination of the yield locus (Bransby
and Randolph, 1997). For these present analyses, the
footing was displaced slowly in a certain displacement direction (i.e. v/, h/ = constant) until a
plastic failure load was reached. The full yield locus
was characterised by the final load points after carrying out a range of these displacement probes with
different v/ and h/.
3.1 Rectangular fluke
Results from a series of displacement probes under
various combinations of translation parallel (h) and
normal (v) to the fluke direction are shown in Figure
5. Both parallel and normal load is normalised by
fluke length and undrained shear strength so that the
yield locus is presented in dimensionless H/(Lsu)
V/(Lsu) space. A convex yield locus results with
maximum horizontal load, Hmax/(Lsu) = 4.29 (V and
M = 0) and Vmax/(Lsu) = 11.87 (H and M = 0).
The peak loads Hmax and Vmax can be compared to
solutions with upper bound calculation. Figures 6(a)
and 6(b) show kinematic mechanisms for the upper
bound plasticity calculation of Vmax and Hmax respectively. The upper bound calculations give:
Vmax
tan
d 1

= 4 +
+ 4 + cos
Ls u
2
L2

(1)

H max
d
tan 1

= 4 +
+ 4 + cos
Ls u
L
2 2

(2)

Figure 6. Upper bound mechanisms for calculation of Hmax and

Vmax.

is obtained between the FE and upper bound calculation of Vmax. However, less good agreement is obtained in prediction of Hmax.
Examination of soil deformation mechanisms calculated at the horizontal and vertical capacities in the
finite element analysis reveal why Vmax is well predicted by upper bound analysis, but Hmax is not (Figure 7). The soil deformation mechanism in the FE
calculation at Vmax is very similar to that of the upper
bound solution, but there is a significant difference
between the FE calculated mechanism and upper
bound mechanism for pure parallel (h) movement.
Further refinement of the methods for Hmax calculation is required.
The yield locus can be deduced in full H, V and
M space when rotational displacements are included
with translation in the displacement probes (Figure
8). The yield locus is symmetrical about the axes and
is convex.

where is defined as in Figures 6(a) and 6(b) and is

varied to minimise Hmax or Vmax.
The predictions of Vmax and Hmax are shown on
Figure 5 for the rectangular footing. Good agreement

Upper bound: V / (L s u ) = 12.10

14

12
h

V / (L su )

10

H
v

Vmax

8
Curve fit (M = 0)

6
4

Upper bound: H / (L s u ) = 5.15

Hmax

0
0

3
H / (L s u )

Figure 5. Rectangular fluke yield locus in V-H space.

Bransby & O Neill: Drag anchor fluke-soil interaction in clays

Figure 7. FE calculated soil displacements at Hmax and Vmax.

1.8

1.8

1.2

1.2

1.2

1.2

0.4

0.8
Curve fit (H = 0)

0.8

6
3
V / (L su)

0.6

Curve fit (H = H1 )

Curve fit (V = 0)

0.4

0
-0.6

0
12

M / (L su)

1.6
M / (L su)

1.6

2
3
H / (L su)
2
3

12

6
3
V / (L su)

0.6

-0.6

-3

3.6
3.6

-3
0

Curve fit (M = 0)

V / (L su)

V / (L su)

1.2
2.4
H / (L su)
1.2
2.4

0
3

Curve fit (V = V1 )

Curve fit (M = M1 )

3
6

12

12

Figure 8. H-V-M yield locus for rectangular fluke.

Figure 10. H-V-M yield locus for wedge fluke.

The moment capacity with V, H = 0 was found to

be Mmax/(L2su) = 1.49. This can be compared to an
upper bound solution using a rotational scoop
mechanism (Figure 9) of Mmax/(L2su) = 1.60 (The
general solution is Mmax/(L2su) = /2(1+(d/L)2).
There is good agreement between the FE and upper
bound solutions because of the similarity of the soil
displacement mechanism (Figure 9).

of the peak H position is due to the changed (and

asymmetric) kinematics of the wedge fluke.

3.2 Wedge shape fluke

The full yield locus in H-V-M space was obtained
for the eccentric wedge shaped fluke (Figure 10).
Unlike for the rectangular fluke, the yield locus is
not symmetric. For example, under pure H loading,
the resultant failure soil displacements will give
positive h and positive v; the anchor will move upwards to the right, partly following the bottom surface of the wedge. Thus, due to normality, the
maximum H load will be sustained with negative V.
The maximum horizontal load is also sustained
with a small negative moment. A detail of the yield
locus in H-M space is shown on Figure 11. The shift

3.3 Curve fitting

The yield loci are shown in full H-V-M space for the
two fluke shapes in Figures 8 and 10. It is more useful to express the yield loci as separate yield functions. Preliminary curve fits are suggested as an offset form of the Murff (1994) equation:
1

q
M M m H H n p
V V1
1
1
1 +
f =
+

M max M1
H max H1
Vmax V1

(3)

where the exponents, q, m, n and p are chosen together with the offsets V1, M1 and H1 after the FE
analyses. The parameters used in the curve fitting are
shown in Table 1, and the fitted curves are shown in
Figures 8 and 10.
Both equations give reasonably good curve fits to

1.8

M
h

1.2
2

M / (L s u )

H1 , M max
Curve fit (V = V1 )

0.6

M=0
0

h/L

Hmax, M 1 ( = 0)
-0.6
0
Figure 9. Upper bound mechanism for calculation of Mmax and
FE calculated soil displacements at Mmax.

Bransby & O Neill: Drag anchor fluke-soil interaction in clays

0.6

1.2

1.8 2.4
H / (L s u )

3.6

Figure 11. H-M yield locus for wedge fluke.

4

Table 1. Yield locus curve fitting parameters.

Parameter
Rectangular fluke
Wedge fluke
Hmax/(L su)
4.29
3.34
Vmax/(L su)
11.87
11.53
Mmax/(L2 su)
1.49
1.60
H1/(L su)
0
0
V1/(L su)
0
-1.25
M1/(L2 su)
0
-0.57
m
1.26
2.37
n
3.72
2.14
p
1.09
0.93
q
3.16
3.41

the full yield locus. However, for the purposes of

anchor design it is the load conditions close to the
peak H load which are most important. More complex curve fitting to allow for the detailed shape of
the yield locus in this region may be required.
4 KINEMATIC ANCHOR ANALYSIS
The yield loci curve fits produced by the finite element analyses for the fluke-soil interaction can be
used in the kinematic analysis of drag anchors. The
approach is similar to that by Neubecker and
Randolph (1996), but plasticity concepts are used to
determine the force on the fluke in the equilibrium
solution and the associated plastic anchor displacements are used to determine the anchor kinematics.
First, the anchor is wished into place near the
soil surface. Next, equilibrium of the anchor is considered which requires finding the fluke load that
will cause yield. The plastic normal at this load point
will then dictate the plastic displacement of the
fluke. This governs the direction of the next anchor
movement and the anchor is stepped to the next position where the procedure is repeated. Hence, the
entire embedment trajectory of the anchor can be
calculated together with the chain force at each instant. This is explained in slightly more detail below.
4.1 Geometrical simplification
The soil stratigraphy is considered as a single layer
of soil with undrained shear strength, suo at the soil

surface and a shear strength gradient k.

The drag anchor is simplified as shown on Figure
12. The fluke is wide enough such that it can be idealised as being plane strain. The shank is treated as a
plate of length sl and of width sw. The chain of diameter b is connected at the anchor padeye. The
fluke-shank angle, fs, is also defined in Figure 12.
At any instant, the padeye is at a depth, zp below
the soil surface and the top face of the fluke is inclined at angle to the horizontal. The chain is inclined at an angle a to the horizontal at the padeye,
where a is governed by the padeye depth, zp, the
chain width, b, the force on the chain, Ta and the lateral capacity factor of the chain, Nc using the equation of Neubecker and Randolph (1996):

2bN s z + 0.5kz 2
c uo p
p
a =
Ta

(4)

4.2 Equilibrium solution

The forces acting on the anchor are shown on Figure
12. For the purpose of this analysis, the anchor is
assumed to be weightless. The shank drag force, Hs
acts at the midpoint of the shank and parallel to the
direction of fluke travel. It is calculated as the product of the projected shank area perpendicular to the
direction of travel, a bearing capacity factor, Nc = 9,
and the undrained shear strength at the shank midpoint. The unknown chain force, Ta is resisted by the
shank drag force, Hs and the fluke drag (H, V, M).
Given that f(H, V, M) = 0 as the fluke is at failure,
there is enough information to calculate all the force
components.
4.3 Normality displacements
If it is assumed that elastic displacements are negligible in comparison to plastic displacements, the
fluke movement direction will be given by the normal to the plastic yield locus at the load state calculated by the equilibrium solution. Thus, /(h/L) =
(f/(M/L))/(f/H) and v/h = (f/V)/(f/H).
Assuming that the anchor fluke moves a distance
x in the direction parallel to the top fluke face in a
displacement increment, then h = x, v =
((f/V)/(f/H))x and = (f/(M/L))/(f/H)
x/L.
4.4 Re-calculation of anchor position

Figure 12. Simplification of the anchor-chain geometry.

Bransby & O Neill: Drag anchor fluke-soil interaction in clays

The anchor position is recalculated after choice of

x using the displacement equations above. The
chain angle is recalculated for the next equilibrium
step and the direction of the last anchor displacement
is recorded so that the direction of the shank drag
force, Hs is known. The procedure is repeated from
5

10

10

20

30

Padeye depth

40

50
0

100

200
300
Draglength (m)

Padeye depth (m)

Padeye load (MN)

Padeye load

400

Figure 13. Results from kinematic drag anchor analysis.

the equilibrium step until the chain force, Ta becomes constant and the trajectory of the anchor flattens.
4.5 Typical results
A typical drag trajectory is shown in Figure 13 for an
anchor in an undrained, normally consolidated clay
with suo = 0 kPa and k = 1 kPa/m. The anchor has
dimensions similar to those of a 32 tonne Vryhof
Stevpris anchor with a 50 fluke-shank angle (Vryhof, 1990). Figure 13 also shows the holding force
on the chain, Ta as the drag progresses.
Both the drag trajectory and holding capacity results are similar to those seen in model anchor tests
(O Neill and Randolph, 1997) with a gradual increase in holding capacity as the embedment depth
increases. A final holding capacity of 8.7 MN
equates to an anchor form factor f of 1.7. This compares to f = 1.4 derived experimentally, suggesting
that the yield locus approach to the analysis of anchors in clays is promising.
5 CONCLUSIONS
Local soil failure around the fluke of drag anchors
has been analysed using finite element analysis to
investigate how drag anchor capacity and kinematics
are affected by the fluke. Fluke-soil behaviour at
failure has been characterised with a plastic yield
envelope in terms of loads parallel and perpendicular
to the anchor fluke and moment load, using the
framework employed recently for shallow foundation analysis.
Yield envelopes are presented for ideal plane
strain flukes of various shapes and upper bound results are presented to verify these findings. The
varying yield envelope shapes are seen to be the consequence of different soil deformation mechanisms
around the anchor governed by the anchor fluke geometry. However, these can be expressed simply in
terms of curve fits in H-V-M load space, and these
yield loci will also form plastic potentials allowing
Bransby & O Neill: Drag anchor fluke-soil interaction in clays

calculation of fluke displacement direction at failure.

The work has lead to the introduction of a new
numerical design method for predicting drag anchor
trajectory and holding capacity using the yield envelope approach. Despite the simplifications inherent in the analysis and the simplified form of the
fluke-soil yield locus curve-fit used, the analysis
method gave results similar to those seen in model
tests of anchors in soft clays. This suggests that the
method has promise with further refinement of the
yield envelope and better approximation to the anchor geometry.
ACKNOWLEDGEMENTS
The work described in this paper forms part of the
activities of the Special Research Centre for Offshore Foundation Systems, established and supported under the Australian Research Council s Research Centres Program. Special thanks are due to
Mr. Nicholas Spadaccini for his contribution to the
work presented in this paper.
REFERENCES
Bransby, M.F., Randolph, M.F., 1997. Shallow foundations
subject to combined loadings, Proc. of the 9th Int. Conf. on
Computer Methods and Advances in Geomechanics,
Wuhan, Vol. 3, pp. 1947-1952.
Bransby, M.F., Randolph, M.F., 1998. Combined loading of
skirted foundations, Geotechnique, Vol. 48, No. 5, pp. 637655.
Britto, A., Gunn, M., 1987. Critical state soil mechanics via
finite element analysis, Chichester, Ellis Horwood Ltd.
Chen, W.F., 1975. Limit Analysis and Soil Plasticity, Elsevier,
New York.
Martin, C.M., 1994. Physical and numerical modelling of offshore foundations under combined loads, PhD Thesis, University of Oxford.
Murff, J.D., 1994. Limit analysis of multi-footing foundation
systems, Proc. of the 8th Int. Conf. on Computer Methods
and Advances in Geomechanics, Morgantown, Vol. 1, pp.
223-244.
Neubecker, S.R., Randolph, M.F., 1996. The performance of
drag anchor and chain systems in cohesive soil, Marine
Georesources and Geotechnology, 14: 77-96.
O Neill, M.P., Randolph, M.F., Neubecker, S.R., 1997. A
novel procedure for testing model drag anchors, Proc. 7th
Int. Offshore and Polar Eng. Conf., ISOPE-97, Honolulu,
Hawaii, 939-945.
Tan, F., 1990. Centrifuge and theoretical modelling of conical
footings on sand, PhD Thesis, The University of Cambridge.
Thorne, C.P, 1998. Penetration and load capacity of marine
drag anchors in soft clay, Journal of Geotech. and Geoenv.
Eng., Vol. 124, No. 10, pp 945-953.
Vryhof Anchors, 1990. Anchor Manual, Krimpen ad Yssel,
The Netherlands.