Professional Documents
Culture Documents
Mirmd 011 GDFG
Mirmd 011 GDFG
Abstract
In this paper a smoothed particle hydrodynamics (SPH) approach is presented to simulate scouring
due to wave around a marine pipeline on a sloping sea bed. The proposed method is similar to so-
called SPH projection method and consists of three steps. The first two steps play the role of
prediction, while in the third step a Poisson Equation is used for both fluid and sea bed to impose
incompressibility constraint. It is assumed that the sandy sediments around pipeline section which
represent the variable boundary and scour profile, behave as a non-Newtonian fluid. Therefore
scouring below a pipe under waves in a test flume with a piston wave maker and sandy sloping bed is
estimated using Newtonian - non Newtonian fluid flow interaction model. The Bingham Model is
cautiously utilized for simulation of the sea bed behavior. Modeled results with or without viscosity
were compared with analytical, empirical and experimental results and a reasonable agreement was
found. Results show that the developed model is a powerful tool to simulate complex free surface and
Newtonian - non Newtonian fluid interaction problems.
Key Words: Wave scouring, Numerical simulation, Free surface, Marine pipeline, SPH method
1 Introduction
In last few decades the tendency to use offshore platform in deep waters has been dramatically increased.
Submarine pipelines offer an efficient mode of transportation of oil and natural gas continuously from the
production site to onshore terminal facilities. Many foundations of water front structures built on sandy
sediments have suffered extensive damage due to scouring under the pipeline resulting in spanning and
vibration due to vortex induced loading. The pipeline vibration magnifies as the natural frequency of the
pipe and wave frequency are close together resulting in failures of system.
Most of the early investigations on scouring adjacent to the offshore pipelines were through physical
modeling, from which many empirical formulas for the scour depth were proposed (Kjeldsen et al., 1973;
Ibrahim and Nalluri, 1986; Bijker and Leeuwestein, 1984; Chiew, 1991; Moncada and Aguirre, 1999).
Experimental studies were often carried out through small-scale flume tests in laboratories to measure
scour profiles in the vicinity of the pipeline. Physical modeling is a useful tool to understand the
mechanisms of the scouring process. However, the results are affected by limitations of the laboratory
conditions. For example, satisfying the total similarity between model and prototype is, most of the time,
impossible.
Fluid dynamics can be modeled by different numerical methods. However, using common CFD methods
is associated with various difficulties related to some complications as turbulence, long distortions, free
surface and variable boundaries in the model. Numerical models based on the solution of the full
1
M. S.C, Faculty of Marine Technology, Amirkabir University of Technology, 424 Hafez Avenue, Tehran-Iran,
P.O.Box: 15875-4413, E-mail : ahmad.mirmohammadi@gmail.com
2 Assoc. Prof., Faculty of Marine Technology, Amirkabir University of Technology, 424 Hafez Avenue,
Tehran-Iran, P. O. Box: 15875-4413, * Corresponding author, E-mail: ketabdar@aut.ac.ir
Note: The original manuscript of this paper was received in Feb. 2010. The revised version was received in Sept.
2010. Discussion open until Sept. 2012.
International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342 - 331-
Navier–Stokes (N-S) Equations can provide more details of the flow behavior. One of the difficulties in
this procedure is tracking of free surface and variable boundaries due to arbitrary moving boundaries. The
MAC (Harlow and Welch, 1965) and VOF (Hirt and Nichols, 1981) methods are two most flexible and
strong approaches in this regard. The former uses marker particles to define the free surface while the
latter solves a transport equation for the volume fraction of the fluid. They have been successfully applied
to various wave simulations (Chan and Street, 1970; Lin and Liu, 1998). However, in both methods, the
N–S equations are solved on a fixed Eulerian grid. Problems of numerical diffusion in this coordinate
arise due to advection terms in the N–S Equations.
Recently, numerical methods which do not use any grid structures such as Lagrangian or particle
methods have been developed. Among them the smoothed particle hydrodynamics (SPH) and moving
particle semi-implicit (MPS) methods are more strong approaches. In both methods, particles move in
Lagrangian coordinates and advection is directly calculated by particle motion without numerical
diffusion. The SPH method originated in astrophysics for the study of fluid dynamics of interstellar gas
by Lucy (1977) and has later been extended to model a wide range of hydrodynamic problems, for
example, incompressible flows by Monaghan (1992) and Takeda et al. (1994). Other studies have since
contributed to the method and solved various engineering problems including heat transfer (Chaniotis and
Poulikakos, 2002), turbulent flows (Welton, 1998) and interfacial flows problems (Lo and Shao, 2002).
Monaghan (2005) extended the method and used it to deal with the free surface problems. SPH and
I-SPH have been used by many researchers to investigate different hydrodynamic problems such as the
study of dam breaks, gravity current descending a ramp, solitary wave breaking in mild slopes (Shao and
Gotho, 2005) wave impact on tall structures (Gomez-Gesteira and Dalrymple, 2004), wave overtopping
of a deck (Dalrymple et al., 2001), non-Newtonian flows with free surface (Shao and Lo, 2003) and
incompressible separated flow (Issa et al., 2004). The SPH has also been used to study the impulsive and
landslides water waves (Girolamo et al., 2006; Gallati et al., 2005).
It should be noted that the SPH method is now commonly used in CFD and appears to be promising in
predicting complex free-surface flows. However, increasing flow complexity requires appropriate
approaches for taking into account of turbulent effects, whereas some authors are still working without
any turbulence modeling in the SPH. Developed turbulence models adapted to the SPH method are from
the simplistic point of view of a one-equation model involving mixing length to more sophisticated
models like explicit algebraic Reynolds stress models or large eddy simulation (Issa et al., 2004).
This paper presents a strictly incompressible 2D SPH model to simulate the processes of wave
propagation with variable boundaries in scour around marine pipeline section. It is assumed that the sandy
sediments around marine pipeline section which represent the variable boundary and scour profile,
behave as a non-Newtonian fluid. Therefore scour simulation below marine pipeline under waves is
obtained by Newtonian-non Newtonian fluid flow interaction simulation. In this procedure there exist two
phase flows; water (Newtonian fluid) including free surface tracking and sandy sediment (non Newtonian
fluid).
- 332 - International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342
N
mb
Ah ( r ) | ¦ AbW ( r r c, h) (3)
b 1 Ub
where the summation index b denotes a particle label carrying a mass of mb at the position rb . The
value of A at bth particle is shown by Ab . There are many possible choices of kernel function to satisfy
the following relations:
W (r r c, h)dr c 1
°V³
® (4)
° lim W ( r r c, h) G (r r c)
¯h o 0
In this paper a cubic Spline kernel which has a smaller compact support than quintic kernel (Morris et al.,
1997) is used as follows:
3 2 3 3
°1 2 s 4 s 0 d s 1
°
D °1 3
W ( r , h) u ® (2 s) 1d s 2 (5)
h7 ° 4
st2
°0
°
¯
where s r h , Q is the number of dimensions, and D is the normalization constant with the values:
2/3, 10/7 S , 1/ S in one, two and three dimensions, respectively. Divergence operators need to be
formulated in accordance with the SPH concept. In the current work, the following commonly used forms
are employed for the gradient of a scalar A and the divergence of a vector ui (Colagrossi and Landrini,
2003):
1 §A A ·
Ua
a A ¦ mb ¨¨ U a2 U b2 ¸¸. aWab (6)
b © a b ¹
1 § ui ui ·
Ua
a .uai ¦ mb ¨¨ Ua2 Ua2 ¸¸. aWab (7)
b © a b ¹
where a is the gradient with respect to coordinates of particle a .
A simple way to formulate the Laplacian operator is to envisage it as dot product of the divergence and
gradient operators. This approach proved to be problematic since the second derivative of the kernel is
very sensitive to particle disorder and can easily lead to the pressure instability and decoupling in the
computation due to the co-location of the velocity and pressure. In this paper, the following alternative is
adopted (Cummins and Rudman, 1999):
§1 · 8 Aab rab . aWab
.¨¨ A ¸¸
©U ¹a
¦ mb U Ub
2 2
rab K 2
(8)
b a
where Aab Aa Ab , rab ra rb and K is a small number to be set to 0.1h established to avoid
singularity during computations.
International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342 - 333-
as instructed by Hosseini et al. (2007). In the first step the body forces are considered to account fluid
particles velocity. The computed provisional velocity is then used in the next step to calculate the
divergence of shear stress tensor as:
u~ i uti 't g i 't (11)
In the second step, the divergence of shear stress tensor T fi , is calculated for fluid particles as:
§ 1 wW ij · § P P u~ i ·
T fi ¨
¨ U wx j
¸
¸ ¦ mb ¨¨ U Ua r 2 b Kab2 ¸¸xabi . aWab (12)
© ¹ b © a b ab ¹
where u~abi u~ai u~bi .
Subsequently, an intermediate velocity and position for particles are obtained:
~
u~ i u~ i T fi 't (13)
~ ~
xi xti 't u~ i 't (14)
There are no constraints to impose incompressibility in previous steps. Therefore the density of the
moving particles can be variable. In the third step a provisional density can be calculated using the
continuity equation:
aWab aW (ra rb , h)
dW 1 i
drab rab
xa xbj (15)
§ dU~ ·
¨ ¸
© dt ¹ a
Ua ¦
mb ~
Ub
u~ai ubi . aW ra rb , h (16)
b
when two particles reach to each other, their relative velocity u~ai ubi and the gradient of kernel ~
function aW ra rb , h have identical signs. Consequently dU~a / dt will be positive leading to an
increase in U a and vice versa. This provisional density can be used to compute a velocity ui so that the
incompressibility is satisfied. To do so, the pressure gradient term of the momentum equation is combined
with the continuity equation driving a Poisson Equation. This equation creates an exchange between
density and pressure:
§1 · U 0 U~
.¨¨ ~ P ¸¸ (17)
©U ¹ U0 't 2
The above equation can be extended to the following SPH form:
§ U U~ 8mb Pb ~ i
xab . aWab · § 8mb ~ i
xab . aWab ·
¨ 0 ¸ ¨ ¸
Pa a
¦ ¦ (18)
b U a U b
¨ U 't 2 ~ 2 2
rab K 2 ¸ ¨ U~ U 2 r 2 K 2 ¸
© 0 ¹ © b a b ab ¹
The computed pressure is used according to the pressure gradient term of the momentum equation to
calculate the velocity û i as:
§P P ·
uˆai 't ¦ mb ¨¨ ~a2 b2 ¸¸ aWab (19)
b © U a Ub ¹
The velocity of each particle will be obtained as:
~
uti 't u~ i uˆ i (20)
This velocity is smoothed according to the XSPH averaging (Monaghan, 2005):
uai uai T ¦
mb
U ab
u i
b
uai Wab (21)
b
where U ab U a Ub / 2 and 0 d T 1 . Final positions of the particles are calculated using a central
difference scheme in time:
xti xti 't
't i
2
ut uti 't (22)
4 Boundary conditions
In the present study, walls are modeled using three layers of SPH particles with fixed positions. No
- 334 - International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342
special treatment was applied on free surface particles in the computational domain. In fact, using the
SPH method, the free surface is modeled naturally and this is the main advantage of the method.
5 Wave simulation
Wave simulation is performed in a simple test flume with the piston wave maker and impermeable bed
using the SPH model. Figure 1 shows the schematic of this flume.
Fig.1 Schematic of flume test; h is the water level and K is the free surface elevation
If Z is the wave angular velocity, a is surface wave amplitude, k is wave number and h is the
water depth, these values are:
International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342 - 335-
H 0.04m , a 0.02m , O 0.3m , h 0.07m , Z gk tanhkh 13.585
Figure 4 shows the horizontal velocity contours for fluid particles. In this figure bold black parts show
the higher orbits of the velocity.
Results illustrate that there is 10% difference between Hunt’s theory (Eq. 24) and model out puts.
Consider the estimated value for boundary layer thickness and interpolations errors in numerical
simulations; this difference seems to be reasonable.
For the paddle simulation, a moveable rigid wall is allocated at one end of the flume. It is designed to
move periodically with the maximum amplitude of 6 cm and period of 0.45 sec. As the model starts to
run, some turbulence is observed at fluid-sediment intersection. This turbulence particularly occurs in a
layer close to fluid sediment intersection leading to jumping of some sand particles into the water column
(Fig. 9). It is captured even at the other end of the flume prior to reaching the surface wave to this region.
It means that this phenomenon is somehow unreal.
A set of tests is performed to investigate this spurious phenomenon. It is found that for proper simulation
of fluid-sediment interaction, velocity and displacement algorithm in the model must be modified. To
solve the problem, fluid particles and sediment particles are considered separately in the kernel function
calculation and search algorithm. Therefore the model is modified and the test is carried out again. The
new output of the model shows that this modification could efficiently leave out this problem (Fig. 10).
- 338 - International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342
Fig. 10 Spurious turbulence removal at fluid sediment intersection
As shown in Fig. 11, the paddle movement in the first stroke caused a bump of water in front of the
paddle. The generated wave in Fig. 12 then propagates towards pipeline and finally reaches the end of the
flume.
Fig. 12 The propagated wave reaches the pipeline and continue to travel towards flume beach
The particles of sandy sediment start to oscillate with a forward net movement as bed load due to the
passage of the waves overhead. Finally the sediment particles reach to a more stable form as ripples.
Figure 13 clearly shows the formation of these ripples.
International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342 - 339-
Fig. 13 Bed sand ripples generated by surface waves in the flume
When wave reaches to the pipeline, the sea bed profile around the pipe section is changed. This causes
the occurrence of scouring phenomenon. Figure 14 shows sea bed changes due to the passage of waves. It
can be seen that a major ripple under the pipe is developed.
Voropayev et al. (2003) performed a set of experiments on scouring around pipes due to waves.
According to their tests four regimes of scour are reported as:
(i) No scour, (ii) initial scour, (iii) expanded scour and (iii) periodic burial scour.
Governing parameters for representing these regimes are Shields’ number ( T ) and Keulegan-Carpenter
(KC). Their results are summarized in Fig. 15 in which Sh is Shields’ number. Three splines in this figure
represent the bounds of these regimes.
In this section two efficient parameters on scour phenomenon are calculated using the developed model.
In the SPH Methods all particle properties are represented in time domain. The Keulegan-Carpenter
number can be expressed as:
KC UT / D (29)
where U is maximum velocity in a level 5 cm above the bed; T the wave period; and D the pipe diameter.
For the generated waves with a period of 0.45 seconds the mean velocity captured underneath the pipe is
about 0.13 m/sec leading to a KC of about 2.95. Shields’ Number as a function of shear velocity
U
W 0 U 12 can be expressed as:
T U
2 / 2 g ( s 1) d , (30)
For the condition of the flow and particles it was obtained as 0.18 in these tests. This Shields’ Number
and KC represent a point in Fig. 15. This point is located in the third regime area which is expanded scour
zone. According to this pattern ripples are developing because of high T and they cause gradational
scouring around pipeline section. The occurrence of this mechanism is evident in modeled results as well.
- 340 - International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342
Fig. 15 Results of tests on sandy sediment in 4 different regimes (After Voropayev et al., 2002)
8 Conclusion
In this paper, the SPH method is employed to simulate water and sandy sediment interaction known as
scouring using Newtonian-non Newtonian fluid flow interaction. Wave generation and propagation are
also modeled by free surface simulation. The wave generation model is validated using analytical
equation of inviscid wave propagation and damping empirical model. This comparison showes a good
agreement between model and theoretical results.
It should be noted that sediment transport is a complex and long term phenomenon. Therefore it is very
time consuming to model the under pipe scouring up to a stable condition in both numerical and
laboratory flumes. As SPH is a very computationally demanding method, therefore research on
determination of final depth of scour hole is not in the concept of this work in this stage. However final
results of current research show that this model is a powerful tool to simulate the movement of sediment
particles leading to scouring beneath the marine pipeline. The numerical wave flume developed in this
research can be used to simulate variety of coastal zone phenomena including sediment transport.
Acknowledgements
The authors would like to acknowledge that this research work is financially supported by the Research
and development department of National Iranian Oil Company (NIOC).
References
Ataie-Ashtiani B. and Shobeyri G. 2007, Numerical simulation of landslide impulsive waves by incompressible
smoothed particle hydrodynamics. International Journal for Numerical Methods in Fluids, Vol. 56, pp. 209–232.
Bijker E. W. and Leeuwestein W. 1984, Interaction between pipelines and the seabed under the influence of waves
and currents. International Union of Theoretical and Applied Mechanics Conference. pp. 235–242.
Chan K. C. and Street R. L. 1970, A computer study of finite amplitude water wave. Journal of Computational
Physics, Vol. 6, pp. 68–94.
Chaniotis A. K. and Poulikakos D. 2002, Remeshed smoothed particle hydrodynamics for the simulation of viscous
and heat conducting flows. Journal of Computational Physics, Vol. 182, pp. 67–90.
Chiew Y. M. 1991, Prediction of maximum scour depth at submarine pipelines. Journal of Hydraulic Engineering, Vol.
117, No 4, pp. 452–466.
Colagrossi A. and Landrini M. 2003, Numerical simulation of interfacial flows by smoothed particle hydrodynamics.
Journal of Computational Physics, No. 191, pp. 448–475.
Cummins S. J. and Rudman M. 1999, An SPH projection method. Journal of Computational Physics, Vol. 152, pp.
584–607.
Dalrymple R. A., Knio O., Cox D. T. Gesteira M., and Zou S. 2001, Using a lagrangian particle method for deck
overtopping. Proc., Waves Reston, Va. ASCE, pp. 1082–1091.
Gallati M., Braschi G., and Falappi S. 2005, SPH simulations of the waves produced by a block falling into a
reservoir. ll Nuovo Cimento C, Vol. 28, No. 2, pp. 129–140.
Girolamo De., Wu P., Liu P. L. F., Panizzo A., Bellotti G., and Di Risio M. 2006, Numerical simulation of three
dimensional tsunamis water waves generated by landslides comparison between physical model results, VOF, SPH
and depth-integrated models. Coastal engineering, Vol. 53, pp. 1516–1528.
International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342 - 341-
Gomez-Gesteira M. and Dalrymple R. A. 2004, Using a three-dimensional smoothed particle hydrodynamics method
for wave impact on a tall structure. Journal of Waterway, Port, Coastal, and Ocean Engineering, Vol. 130, No. 2, pp.
63–69.
Harlow F. H. and Welch J. E. 1965, Numerical calculation of time-dependent viscous incompressible flow of fluid
with free surface. Phys Fluids, Vol. 8, No.12, pp. 322–329.
Hirt C. W. and Nichols B. D. 1981, Volume of fluid (VOF) method for the dynamics of free boundaries. Journal of
Computational Physics, Vol. 39, pp. 201–225.
Hosseini S. M., Manzari M. T., and Hannani S. K. 2007, A fully explicit three step SPH algorithm for simulation of
non-Newtonian fluid flow. International Journal of Numerical Methods for Heat and Fluid Flow, Vol. 17, pp.
715–735.
Hunt J. N. 1952, Viscous damping of waves over an inclined bed in a channel of finite width. La Houille Blanche, Vol.
7, No. 6, pp. 836–842.
Ibrahim A. and Nalluri C. 1986, Scour prediction around marine pipelines. Proceedings of Fifth International
Symposium on Offshore Mechanics and Arctic Engineering, ASME, New York, No. 3, pp. 679–684.
Issa R., Lee E. S., Violeau D., and Laurence D. 2004, Incompressible separated flows simulations with the smoothed
particle hydrodynamics gridless method. International Journal for Numerical Methods in Fluids, Vol. 47, No. 11,
pp. 1101–1106.
Kjeldsen S. P., Gjorsvik C., Bringaker K. G., and Jacobsen J. 1973, Local scour near offshore pipelines. Proceedings
of 2nd International Port and Oceanic Engineering under Arctic Conditions, University of Iceland, pp. 308–331.
Lin P. Z., and Liu P. L. F. 1998, A numerical study of breaking waves in the surf zone. Journal of Fluid Mechanics,
Vol. 64, pp. 359–239.
Lo E. Y. M. and Shao S. 2002, Simulation of near-shore solitary wave mechanics by an incompressible SPH method.
Applied Ocean Research, Vol. 24, Issue 5, pp. 275–286.
Lucy L. B. 1977, Numerical approach to the testing of the fission hypothesis. Astrophysics Journal, Vol. 82, pp.
1013–1020
Monaghan J. J. 1992, Smoothed particle hydrodynamics. Annual Review Astronomical Astrophysics, Vol. 30, pp.
543–74.
Monaghan J. J. 2005, Smoothed particles hydrodynamics. Reports on Progress in Physics, Vol. 68, pp.1703–1759.
Moncada M. A. T. and Aguirre P. J. 1999, Scour below pipeline in river crossings. Journal of Hydraulic Engineering,
ASCE 125, No. 9, pp. 953–958.
Morris J. P., Fox P. J., and Zhu Y. 1997, Modeling low Reynolds number incompressible flows using SPH. Journal of
Computational Physics, Vol. 136, pp. 214–226.
Shao S. D. and Lo E. Y. M. 2003, Incompressible SPH method for simulating Newtonian and non-Newtonian flows
with a free surface. Advances in Water Resources, Vol. 26, pp. 787–800.
Shao S. D. and Gotho H. 2005, Turbulence particle models for tracking free surfaces. Journal of Hydraulic Research,
Vol. 43, No. 3, pp. 276–289.
Shields A. 1936, Anwendung der Aehnlichkeitsmechanik und Turbulenzforchung auf die Geschiebebewegung. Mitt.
Preuss. Versuchsanstalt fur Wasserbau und Schiffbau, No. 26, Berlin.
Takeda H., Miyama S. M., and Sekia M. 1994, Numerical simulation of viscous flow by smoothed particle
hydrodynamics. Progress of Theoretical Physics, Vol. 92, No. 5, pp. 939–960.
Voropayev S. I., Testik F. Y., Fernando H. J. S., and Boyer D. L. 2003, Burial and scour around short cylinder under
progressive shoaling waves. Ocean Engineering, Vol. 30, No. 13, p. 1647.
Welton W. C. 1998, Two-dimensional pdf/sph simulation of compressible turbulent flows. Journal of Computational
Physics, Vol. 139, pp. 410–443.
- 342 - International Journal of Sediment Research, Vol. 26, No. 3, 2011, pp. 331–342