Hydrodynamic Forces due to Orbital Stokes 5th Order

Waves on Subsea Pipelines Resting on Porous Seabed

Annelise Karreman
Dr Jeremy Leggoe
School of Mechanical and Chemical Engineering
W/Prof Liang Cheng
School of Civil and Resource Engineering
CEED Client: Woodside Energy Ltd
Abstract
Stability design of subsea pipelines requires an accurate prediction of the hydrodynamic
forces induced by waves and currents. Typically, rectilinear motion is used to simulate
near seabed waves, since the orbital particle paths in wind-generated waves flatten to
ellipsoids with depth. Together with linear Airy wave theory, this simplification predicts
symmetric hydrodynamic forces on forward and reverse half wave cycles. Computational
Fluid Dynamics (CFD) modelling of orbital waves has shown a difference in the overall
pipeline stability of up to 50% between half wave cycles. Further CFD modelling is
proposed to investigate and validate these results, with consideration given to the effects
of higher order wave theory, boundary layer overshoot profile and a porous seabed.

1.

Introduction

Pipeline stability is of significant importance to the success of hydrocarbon developments and

can contribute up to 30% of the capital expenditure for pipeline projects on the North West
Shelf, due to the severe metocean and geotechnical conditions. To avoid lateral instability,
heavy concrete weight coating or costly secondary stabilisation methods (Zeitoun et al. 2009)
are often required. The current design methods use DNV-RP-F109 (DNV 2011). The
applicability of the Recommended Practice is currently being investigated through
experiments conducted in the O-Tube facility under the STABLEPipe JIP. This facility, like
many of the experimental investigations previously conducted, models the near bed wave
velocity as rectilinear. CFD results have indicated an asymmetry in horizontal and vertical
loading due to orbital wave velocities that may require consideration in design (King 2011).
The hydrodynamic load experienced by the pipeline is part of the flow induced forces caused
by the relative motion between the pipe and surrounding water (DetNorskeVeritas 2008).
Morisons equation has traditionally been used to calculate the loads on the pipe. The
submerged weight is then increased to ensure the resistance of the system avoids lateral
displacement. Morisons equation gives the hydrodynamic forces on a stationary cylinder as
!

!!!

F! t = F! t + F! t = C! u t u t +

C! a! t

F! t = F! t = C! Du t
!

CEED Seminar Proceedings 2012

Karreman: Hydrodynamic Forces on Subsea Pipelines

where F! t and F! t are the time dependent horizontal and vertical forces, F! t , F! t and
F! t are the time dependent drag, inertia and lift forces, C! , C! and C! are the constant drag,
inertia and lift coefficients and u t and a! t are the horizontal velocity and acceleration of
fluid particles respectively. Morisons equation doesnt accurately model the forces under
oscillatory flow due to the development of a wake that is swept back over the pipe in
succeeding half wave cycles (Hale, Lammert, & Jacobsen 1989). Use of time-dependent
coefficients and near pipe velocity produces a significant improvement in the calculation of
forces under oscillatory wave loads (Bryndum, Jacobsen, & Tsahalis 1992).
Existing understanding of hydrodynamic loading is based on field measurements, which
traditionally have not been accurate enough to capture the vertical component of orbital
motion and experiments, conducted using the carriage technique, and in u-tube and o-tube
facilities that model only the horizontal velocities. CFD Analysis predicts that an asymmetry
of up to 50% of the overall pipeline stability exists between the forward and reverse half wave
cycles when the vertical component of orbital wave motion is introduced (King 2011). It was
also highlighted that the simplification to rectilinear motion produced conservative results
when an absolute lateral stability design method is used. This suggests improved efficiency
could be obtained by considering orbital motion in design methods.

2.

Computational Fluid Dynamics (CFD) Model

Use of the CFD package ANSYS Fluent 13.0 to solve the Reynold Averaged Navier-Stokes
(RANS) equations, allows for the computation of fluid velocities, pressures, turbulence
parameters and hydrodynamic loading on the pipe structure within the model domain. Based
on the previous mesh verification work, the domain consists of a 200m x 40m fluid region
above a 200m x 2m porous medium seabed region with a 1m diameter pipeline located at the
horizontal centre of the fluid region with its base level with the top of the porous medium.
There is no gap between seabed and pipe and the seabed curves upward toward the pipe to
make a contact angle of 45 reducing the likelihood of scour below the pipeline.
The left and top boundaries of the fluid region have been modelled as velocity inlets, whilst
the right boundary of the fluid region has been set to a pressure outlet. The left, right and
bottom boundaries of the porous region have been set to symmetry and act as a free slip wall.
The pipe has been given the Nikuradse roughness of concrete. The proposed mesh contains
80,128 cells, with cell sizes biased towards the boundary layers at the seabed and around the
circumference of the pipe. The smallest cell height is 0.6mm which is sufficiently small to
capture the boundary layer whilst keeping the cell height larger than the grain diameter in the
porous medium.
The Shear Stress Transport (SST) k- turbulence closure model is adopted in the model as a
result of the previous work comparing the available 2D closure models in ANSYS Fluent
(King 2011).

Figure 1

Mesh grids in computational domain (left). Mesh grid near pipe

(right).
2

CEED Seminar Proceedings 2012

2.1

Karreman: Hydrodynamic Forces on Subsea Pipelines

Porous Medium

ANSYS Fluent 13.0 allows for the modelling of porous medium as fluids with specified
porosity, and viscous and inertial resistance coefficients. The ANSYS Users Guide (2011,
!
Section 7.2.3) provides that the viscous resistance coefficient of a packed bed, !, and the
inertial resistance coefficient, ! , can be calculated as
!!
!
3.5 1
=

=

!
150 1 !
! !
where ! is the particle diameter and is the void ratio. The soil parameters of the
STABLEPipe JIP O-tube (An et al. 2012, p3) are used for the model.
The ANSYS Users Guide (2011, Section 7.2.3) provides warnings on the treatment of
turbulence in the porous medium and suggests that unless large porosities are used the porous
medium should be treated as a laminar region with zero turbulent viscosity, ! .

2.2

User Defined Function (UDF)

2.2.1

Specified Boundary Condition Profiles

The profiles for the horizontal and vertical component of the velocity and the values for the
Turbulent Kinetic Energy (k) and Specific Dissipation Rate () at the two velocity inlets, are
defined using a User Defined Function (UDF). Considerable focus has been directed at
refining the UDF interpreted into the model in ANSYS Fluent. The main UDF inputs are
wave height (H), wave length (L), wave period (T) and water depth (d). The UDF specifies
the profiles for the horizontal and vertical component of the velocity, and the values for the
Turbulent Kinetic Energy (k) and Specific Dissipation Rate () at the two velocity inlets.
Reversed flow is permitted at the pressure outlet and the same profiles for k and are applied
to this reversed flow.
Increased consistency in the model has been obtained by accurately specifying the k and
profiles in the UDF. The chosen profiles are functions only of the height above the seabed and
are transmitted through an equivalent empty domain without substantial changes occurring.
Specifying these profiles in the UDF replicates the fully developed wave boundary layer.
2.2.2

Stokes 5th Order Wave Theory

A large number of the cases in the test matrix previously analysed (King 2001) lie in the
applicability range of Stokes 5th order wave theory (See Figure 2). Although Airy wave theory
has been found to provide excellent predictions of near-bottom peak water velocities under
waves (Grace & Nicinski 1976, p684), higher order wave theories will provide improved
prediction of seabed kinematics (Zeitoun et al. 2010, p1).
Stokes 5th order wave velocity profiles have been derived and are provided to the ANSYS
Fluent model by interpreting the UDF. Stokes finite amplitude wave theory models the wave
horizontal and vertical velocities as a Fourier series of the form

CEED Seminar Proceedings 2012

u=
!!!

Karreman: Hydrodynamic Forces on Subsea Pipelines

2
x t
! cosh(n y)cos 2n( ) w =
L
L T

! sinh (n
!!!

2
x t
y)sin 2n( )
L
L T

where the ! series of coefficients are calculated for the specific wave height, length, period
and water depth of the wave in accordance with those proposed by Skjelbreia and
Hendrickson (1961). Note that the CFD domain is chosen so as to avoid introducing the freesurface of the wave into the domain.

Figure 2

Wave theory applicability for test matrix (King 2011, p52).

The superposition of sinusoids produces a wave with a steeper crest and shallower trough
than the basic sinusoidal function proposed by Airy wave theory. The horizontal and vertical
velocity profiles are also slightly altered (see Figure 3). The introduction of Stokes 5th Order
wave theory therefore introduces additional asymmetry into the hydrodynamics.

Figure 3

2.2.3

Wave Surface Profile (left), Horizontal Velocity at 1m above seabed

(middle), Vertical Velocity at 1m above seabed (right) for the case 1
test matrix wave.

Wave Boundary Layer Velocity Profile

Boundary conditions in a wall boundary layer require the horizontal and vertical velocity to
go to zero at the wall and reach the free stream velocity at the top of the boundary layer.
Lamb (1932) derived a continuous first order solution to the velocity profile in a laminar
boundary layer by resolving the velocity as a cis function, where the velocity is given by the
real part of e!(!"!!!) , and introducing the concept of a defect velocity between the free stream
4

CEED Seminar Proceedings 2012

Karreman: Hydrodynamic Forces on Subsea Pipelines

velocity and the boundary layer velocity with the nature of a damped wave propagat[ing]
away from the bed (Nielsen 1999, p16)
x t
x t
u = u! cos 2( ) e!!! cos 2( ) y
L T
L T
where =

!
!!

and is the kinematic viscosity of the fluid. This solution captures both the

decay and phase difference in the velocity profile approaching a wall as seen in Figure 4.

Figure 4

Boundary layer horizontal velocity profile.

The first order Lambs solution is not appropriate for Stokes finite amplitude waves, since the
velocity profile in the boundary layer is not symmetrical with respect to the vertical axis
(Borghei 1982, p178-178). The profile used in the UDF has been derived through the same
method as Lambs solution.
Lambs solution is also inaccurate in describing the velocity profile in transitional or turbulent
flow (Nielsen 1992, p44). An alteration introducing a power law has been incorporated into
the UDF.
The horizontal velocity profile of the UDF is given by
!

=
!!!

2
x t
!
! cosh(n y) cos 2n( )
L
L T

! !
!!!

x t

cos 2n( )
L T
!!

with !! , z!" and p determined by optimising the fit of the function to the developed
boundary layer velocity outputs from Fluent. The vertical velocity boundary layer (Svendsen
2006, p474-479), is not captured in the UDF.

3.

Results and Discussion

Preliminary results have indicated the continuing presence of hydrodynamic loading

asymmetry. The full test matrix has not been run and as such no reliable conclusions may yet
be drawn.

CEED Seminar Proceedings 2012

4.

Karreman: Hydrodynamic Forces on Subsea Pipelines

Conclusions and Future Work

UDFs have been refined for a number of the original test matrix cases and it is proposed that
these will be run through the improved mesh with porous seabed to verify and further
quantify the identified asymmetry parameter. There is still scope for extension of the analysis
to 3D CFD simulation to model the motion in the axial direction due to vortices.

5.

References

An, H, Luo, C, Cheng, L & White, D, 2012, A Re-examination of the Hydrodynamic Forces
acting on Partially-Buried Submarine Pipelines, Proceedings of the thirty-first International
Conference on Offshore Mechanics and Artic Engineering.
ANSYS 2011, ANSYS FLUENT 13.0 User's Guide, ANSYS Inc.
Borghei, BM 1982, Oscillatory boundary layer over fixed rough beds, Ph.D. Thesis,
University of Nottingham.
Bryndum, MB, Jacobsen, V & Tsahalis, DT 1992, 'Hydrodynamic Forces on Pipelines: Model
Tests', Journal of Offshore Mechanics and Arctic Engineering, vol. 114, no. 4, pp. 231-241.
DetNorskeVeritas, 2008, Offshore Standard DNV-OS-F101: Submarine Pipeline Systems,
DNV, Norway.
DNV 2011, On-Bottom Stability Design of Submarine Pipelines, vol. DNV-RP-F109, Det
Norske Veritas, Norway.
Grace, RA & Nicinski, SA 1976, OTC2676: Wave Force Coefficients from Pipeline
Research in the Ocean, Proceedings of the eighth Offshore Technology Conference.
Hale, JR, Lammert, WF & Jacobsen, V 1989, 'OTC6059: Improved Basis for Static Stability
Analysis and Design of Marine Pipelines', Proceedings of the twenty-first Offshore
Technology Conference.
King, L 2011, Hydrodynamic Forces on Subsea Pipes due to Orbital Wave Effects, Honours
Thesis, University of Western Australia.
Lamb, 1932, Hydrodynamics, 6th Ed, Cambridge University Press.
Nielsen, P 1992, Coastal Bottom Boundary Layers and Sediment Transport, Advanced
Series in Ocean Engineering, Vol 4.
Skjebreia, L & Hendrickson, J 1960, Fifth Order Gravity Wave Theory, 7th Proceedings of
the seventh International Conference of Coastal Engineering, Chapter 10, pp. 184-196.
Svendsen I 2006, Introduction to nearshore hydrodynamics, Advanced Series on Ocean
Engineering, Vol 24 World Scientific, Singapore.
Zeitoun, HO, Tornes, K, Cumming, G & Willcocks, J 2009, ' OMAE2009-79893 A Stability
Design Rationale - A Review of Present Design Approaches', Proceedings of the twentyeighth International Conference on Offshore Mechanics and Artic Engineering.
6