Professional Documents
Culture Documents
net/publication/283182536
CITATIONS READS
2 279
6 authors, including:
Some of the authors of this publication are also working on these related projects:
Study of Wave-in-deck loading on offshore structures during extreme events View project
All content following this page was uploaded by Yanling Wu on 25 January 2016.
For existing offshore fixed platforms it is often the case that the air gap between the deck
and the sea surface is not adequate and the extreme waves will encroach on the deck
resulting in large wave-in-deck loads. Factors that result in inadequate air gap are seabed
subsidence, sea-level increasing due to climate change and more onerous predictions of
extreme crest heights.
In this paper, a numerical approach based on NewWave theory [Tromans et al.
(1991), Proc. 1st Int. Offshore and Polar Engineering Conf., Vol. 3, Edinburgh, UK,
pp. 64–71] has been developed to represent the extreme wave conditions and implemented
into the framework of an open source software, OpenFOAM, to predict the wave-in-deck
loading. The results have been compared with published FLOW-3D simulations using
Stoke’s 5th order wave theory for a simple box representing the Ekofisk platform deck
in the Norwegian sector of the North Sea [Iwanowski et al. (2002), Proc. 21st Int. Conf.
Offshore Mechanics and Artic Engineering, June 23–28, 2002, Oslo, Norway].
1. Introduction
Wave-in-deck loading can cause local damage as well as toppling of a structure
and there have been many examples of this in the Gulf of Mexico during severe
hurricanes. Fortunately, life safety has not been threatened as these platforms are
de-manned in advance of the hurricane. However, in many regions, the platforms
1640014-1
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
remain manned during extreme events and it is thus important to accurately predict
wave-in-deck loads under these situations.
Presently, analytical or semi-analytical methods (such as Silhouette approaches
[API 2010, ISO19902 (2013)] and Kaplan’s model [DNV-RP-C205 (2014)]) sup-
ported by scaled model tests in laboratory experiments have been the main meth-
ods for calculation of wave-in-deck loads. Due to strong nonlinearity, wave breaking
and the short impact duration of wave-in-deck loading, quantifying the accuracy of
the analytical solutions is challenging. On the other hand, model testing is expen-
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
sive and scale effects may have an influence. As an alternative approach, numerical
simulation based on computational fluid dynamics (CFD) has attracted a lot of
attention recently due to advances in numerical methods and the availability of
high performance computers. CFD is a useful tool to predict impact forces poten-
tially with high accuracy for the full scale deck. Detailed insight into the flow and
Int. J. Comput. Methods Downloaded from www.worldscientific.com
resultant loadings can also be resolved by CFD which is not possible with analytical
methods and is difficult and costly with experimental methods.
To evaluate wave-in-deck impact loads, an extreme wave generator needs to be
employed. The extreme wave occurs as a highly transient event within a multi fre-
quency sea state. Regular waves, such as Stokes’ 5th order, do not represent these
extreme waves accurately and random wave generation is an extremely time consum-
ing process, as the events of interest occur only rarely. NewWave theory [Tromans
et al. (1991)] provides a realistic deterministic description of the largest waves in a
random sea and is an alternative to regular wave theories. It is able to efficiently
generate targeted waves at a prescribed time and location. The representation has
also been studied theoretically by Boccotti (1983) and experimentally and numer-
ically by several investigators, such as Taylor and Haagsma (1994); Baldock et al.
(1996); Borthwick et al. (2006).
In this study, a second-order NewWave model has been employed to generate
input conditions for focused wave CFD simulations using the open source software
OpenFOAM and the wave2FOAM library [OpenFOAM User Guide (2008)]. In the
current OpenFOAM platform, the Navier–Stokes (N–S) equations describe the fluid
flow while the Volume of Fluid method (VOF) captures the movement of the water
free surface.
2. NewWave Theory
The concept of the NewWave formulation is to generate the extreme event from a
specified frequency spectrum by superimposing several relatively small amplitude
waves to form one focused extreme wave at a specified location and specified time.
For the linear NewWave, the amplitude of each wave component ai of frequency fi ,
is defined (see for example [Ning et al. (2009)]) as
S(fi )∆f
ai = A0 N , (1)
i S(fi )∆f
1640014-2
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
where S(f ) is the surface spectral density, ∆f is the frequency step (which depends
on the number of wave components N and bandwidth), and A0 is the target theo-
retical linear wave amplitude of the focused wave. The extreme wave represented by
linear NewWave theory is simply the scaled auto-correlation function corresponding
to a specified spectrum.
The linear surface displacement η (1) and horizontal and vertical velocities u(1)
and w(1) are given by:
N
η (1) =
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
ai cos[ki (x − x0 ) − ωi (t − t0 )],
i=1
N
ai gki cosh ki (z + h)
u(1) = cos[ki (x − x0 ) − ωi (t − t0 )], (2)
i=1
ωi cosh(ki h)
Int. J. Comput. Methods Downloaded from www.worldscientific.com
N
ai gki sinh ki (z + h)
w(1) = sin[ki (x − x0 ) − ωi (t − t0 )],
i=1
ωi cosh(ki h)
where z is the vertical coordinate measured upwards from the Mean Water Level
(MWL), x0 , t0 are the predefined focal location and focal time, respectively, g is the
gravitational acceleration, h is the water depth, ki = ωi2 /g tanh(ki h) is the wave
number and ωi = 2πfi is the frequency.
The method can be extended to second-order (see for example [Hu et al. (2011)])
with the free surface elevation and velocity components obtained by superimposi-
tion as:
η = η (1) + η (2) ,
w = w(1) + w(2) ,
where η (2) , u(2) and w(2) correspond to the second-order wave elevation and veloc-
ities, respectively. Details of the second-order terms in Eq. (2) can be found in Hu
et al. (2011) and Westphalen et al. (2012).
Various spectra may be used to represent the sea states. The JONSWAP fre-
quency spectrum S(f ) is frequently employed (e.g., [Gao et al. (2012)]) and is used
herein:
2
/2λ2 ]
S(f ) = βJ Hs2 Tp−4 f −5 exp[−1.25(Tp f )−4 ]γαexp[−(Tp f −1) ,
0.06238(1.094 − 0.01915 ln γα ) 0.07 f ≤ fp (4)
βJ ∼
= ; λ= ,
0.230 + 0.0336γα − 0.185(1.9 + γα )−1 0.09 f > fp
where Hs is the significant wave height and Tp and fp are the peak wave period and
frequency, respectively. The peak enhancement factor γα was taken as 3.3. Note that
for the NewWave formulation, the value of Hs is not relevant since the normalized
spectrum is used (see Eq. (1)).
1640014-3
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
eled as a simple box being 30 m wide (normal to the wave propagation direction),
with wave inundation at 2 m and 4 m. In this present study, the calculations were
conducted for an inundation of 4 m and the Stoke’s wave solution is compared with
the NewWave implementation in OpenFOAM.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
For a Stokes 5th order wave, the crest height depends on the wave parameters H,
T and water depth d. The parameters for the target focused NewWave for both crest
height and wave height matching of the Stoke’s wave are summarized in Table 1.
Based on the 2nd order NewWave time profile, a conversion factor of 1/0.93 is used
to obtain Tp in NewWave from the Stoke’s wave period Ts .
From numerical experiments, to get accurate results within a reasonable time
frame, the pre-defined focus position and time (x0 , t0 ) should neither be too small
nor too large. If too small, it is very difficult to get the characteristics of the desired
target focused wave. If too large, the computational cost will increase dramatically.
Through numerical tests, it was found that x0 = 1.5λp and t0 = 1.5Tp gives a
balance between accuracy and efficiency.
With respect to mean sea level (MSL), the NewWave CFD simulation results in
a more asymmetric wave shape than the Stoke’s 5th wave with higher crests and
shallower and wider troughs. This means a NewWave simulation can never match
both the Stoke’s wave crest and wave height. Therefore, in order to compare wave-
in-deck loads, two different approaches are adopted: crest height matching and wave
height matching. The deck height is adjusted to maintain the same inundation level
of 4 m — see Fig. 1.
Stoke’s 5th
Parameter wave NewWave
1640014-4
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Fig. 1. Deck with wave elevation, wave height and wave impact height (wave inundation).
Int. J. Comput. Methods Downloaded from www.worldscientific.com
A0 x0 t0 A1 x1 t1
1640014-5
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
(a)
(b)
Fig. 2. Influence of mesh size for 14.3 m crest matching. (a) Elevation and (b) Wave-in-deck loading.
almost identical but the coarse mesh produces poor results. Therefore, NewWave
needs quite a fine mesh resolution near the free surface. In the present study, ∆x ≈
λP /640, ∆y ≈ H/180 is used in the simulations.
1640014-6
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Fig. 3. Comparison of wave shape for wave height matching (24.3 m) NewWave and 5th order
Stoke’s wave.
the 5th order Stoke’s wave — see Fig. 3 — and the asymmetric profile is indicative
of wave close to breaking.
Comparison of crest velocity for crest height and wave height matching
Table 3 compares the crest velocity at the actual focal point for present CFD tools
for crest height matching and wave height matching.
From Table 3, it is found that the crest velocity for NewWave (for both crest
height matching and wave height matching) is larger than that of the 5th order
Stoke’s wave. For wave height matching, the crest velocity for NewWave is more
than twice that of the Stoke’s wave.
1640014-7
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
(a)
(b)
Fig. 4. Comparison of wave-in-deck load for crest matching NewWave and 5th order Stokes wave.
(a) Crest matching: Horizontal force and (b) Crest matching: Vertical force.
computational approach provides results that are comparable with other available
CFD studies in published literature.
For the crest matching simulation, it was found that the peak of horizontal force
acting on the simple box for NewWave is about 15% higher than that of the Stokes
5th wave but its duration is less (Fig. 4(a)). The higher peak force is due to the
higher crest velocity while the reduced duration is due to the sharper and narrower
1640014-8
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
(a)
(b)
Fig. 5. Comparison of wave-in-deck load for wave height matching NewWave and 5th order Stokes
wave. (a) Wave height matching: Horizontal force and (b) Wave height matching: Vertical force.
wave shape at the crest of NewWave. On the other hand, the area under the curve of
the NewWave simulation (the impulse) is less than that for the Stokes 5th wave. The
effective horizontal (shear) force applied to the supporting jacket structure depends
on the peak force, the duration of the force and the natural period of the jacket.
Therefore, even though the NewWave force applied to the deck is somewhat larger
than the Stokes 5th force, the response (force in the jacket) may be either larger or
1640014-9
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
smaller, depending on the natural frequency of the jacket. This is illustrated below
in Sec. 4.
For the vertical force comparison in Fig. 4(b), there are significant differences
between the crest-matched NewWave and the Stoke’s 5th wave. The upward vertical
force Fz (+) for NewWave is about half that of the Stoke’s 5th wave. This is most
likely due to the different wave shape (evolution) and vertical crest velocity of the
different wave theories but requires further investigation.
Also from Fig. 4(b), it is seen that our 2D results give large and oscillating
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
downward vertical forces Fz (−). These are due to entrapped air and can be reduced
by three-dimensional (3D) simulation which can capture the strong sideways air
flow occurring under the deck, as reported in our previous 5th order Stokes wave
study on wave-in-deck [Chen et al. (2014)]. If we ignore these spurious spikes, the
NewWave and 5th order Stoke’s downward forces are very similar.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
Figure 5 compares the wave-in-deck loading from our CFD tools where the wave
height is matched to a 5th order Stoke’s wave. It can be seen from Fig. 5(a) that the
horizontal peak force for CFD NewWave is much higher (due to the much larger
crest velocity, as shown in Table 3) and the duration of forces are less than the
Stoke’s 5th wave; the area under both curves (the impulse) is similar.
Comparing Figs. 5(b) and 4(b), we see that for crest matching and wave height
matching, NewWave gives very similar vertical force results (ignoring the large
spurious spike). On the other hand, the upward vertical force Fz (+) for NewWave
is less than 1/2 of that from the 5th order Stoke’s wave while the downward vertical
forces are similar.
1640014-10
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
(a)
(b)
Fig. 6. Lateral dynamic response of jacket structure to wave-in-deck loading (Natural period = 2.5 s;
3% critical damping). (a) Crest matching and (b) Wave height matching.
matched case, while considering dynamic effects, due to the shorter duration of the
impulse, the increase is only 22% and the DAF is now only 1.04.
In the application to wave-in-deck loading on offshore structures, wave crests are
typically calculated from Forristall 3D statistics [Forristall (2000)] and therefore,
1640014-11
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
Stoke’s versus NewWave crests are the most representative comparisons to make.
From the results of this limited study, it seems that a CFD analysis using Stoke’s
5th order wave matched to the Forristall crest height, with a period of 0.93 Tp, will
give reasonable results for a jacket integrity assessment provided dynamic response
is considered.
Comparison with API RP2A and ISO 19902
[API RP2A (2010)] and [ISO 19902 (2013)] use the same recipe for horizontal wave-
in-deck loading based on the silhouette of the impacted deck structure:
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
F x = 0.5ρCdAV 2 . (5)
where ρ is the water density, Cd is a force coefficient, A is the vertical wetted area
and V is the horizontal fluid velocity.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
5. Conclusions
The development of an extreme wave generator based on second-order NewWave
theory was implemented into the OpenFOAM CFD software. It was found that
besides the wave particle crest velocity, wave profile is another important factor
which affects the wave-in-deck loading time history.
A wave-in-deck loading comparison between NewWave and a Stoke’s 5th order
wave indicated that if the wave crests are matched, then the horizontal and vertical
downward forces are similar but the vertical upward loading from Stoke’s is twice as
large as the NewWave loading. If wave heights are matched, the NewWave horizontal
forces are about double those of the Stoke’s forces while the vertical positive force
for Stoke’s 5th is about twice that of NewWave. The downward forces, however, for
both cases are very similar.
The impulsive nature of wave-in-deck-loading results in a transient load being
applied to the supporting structure and large dynamic amplification effects are
evident. These depend on the applied peak deck force, its duration and the natural
period of the jacket. It is this dynamically enhanced force that should be used in
the assessment of structural integrity of the jacket structure. From the results of
this limited study, it seems that in the absence of wave breaking, a CFD analysis
using Stoke’s 5th order wave matched to the Forristall crest height, with a period
of 0.93 Tp, will give reasonable results for a jacket integrity assessment provided
dynamic response is considered.
1640014-12
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
The results presented herein are for unidirectional NewWave simulations for
wave in deck loading on a simple box structure. The effect of spreading, which
reduces the crest velocity and the crest length, and more complex deck geometries
including underdeck beams, will be investigated in future research.
Acknowledgments
The work was funded jointly by Lloyd’s Register Global Technology Centre and
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
the Institute of High Performance Computing, Agency for Science, Technology and
Research (A*STAR) of Singapore, under the project title “Study of wave-in-deck
and breaking wave loading on offshore structures during extreme events”. Thanks
to Mr. Anand Bahuguni for his assistance.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
Disclaimer
The views expressed in this paper are those of the authors and do not necessarily
reflect those of their affiliated companies.
References
API Recommended Practice 2A-WSD: Planning, Designing, and Constructing Fixed Off-
shore Platforms — Working Stress Design, 21st Edition (2010).
Baldock, T. E., Swan, C. and Taylor, P. H. [1996] “A laboratory study of nonlinear surface
waves on water,” Philos. Trans. R. Soc. Lond. A 354, 649–676.
Boccotti, P. [1983] “Some new results on statistical properties of wind waves,” Appl. Ocean
Res. 5(3), 134–140.
Borthwick, A. G. L., Hunt, A. C., Feng, T., Taylor, P. H. and Stansby, P. K. [2006] “Flow
kinematics of focused wave groups on a plane beach in the UK coastal research facility,”
Coastal Eng. 53(12), 1033–1044.
Chen, Y., Wu, Y. L., Stewart, G., Gullman-Strand, J. and Lu, X. [2014] “Numerical
simulation of wave in deck loading on offshore structures,” Proc. ASME 33rd Int.
Conf. Ocean, Offshore and Arctic Engineering, OMAE2014-23847, June 8–13, 2014,
San Francisco, California, USA.
Forristall, G. Z. [2000] “Wave crest distributions — observations and second order theory,”
J. Phys. Oceanogr. 30(8), 1931–1943.
Gao, F., Mingham, C. and Causon, D. [2012] “Simulation of extreme wave interaction
with monopile mounts for offshore wind turbines,” Coastal Engineering Proceedings
Structures 1(33), 22 pp.
Hu, Z. Z., Causon, D. M., Mingham, C. G. and Qian, L. [2011] “Numerical simulation
of floating bodies in extreme free surface waves,” Nat. Hazards Earth Syst. Sci. 11,
519–527.
ISO 19902+A1, Petroleum and Natural Gas Industries-Fixed Steel Offshore Structures
(2013).
Iwanowski, B., Grigorian, H. and Scherf, I. [2002] “Subsidence of the Ekofisk platforms:
Wave in deck impact study — various wave models and computational methods, Proc.
21st Int. Conf. Offshore Mechanics and Artic Engineering, OMAE2002-28063, June
23–28, 2002, Oslo, Norway.
1640014-13
2nd Reading
September 11, 2015 7:47 WSPC/0219-8762 196-IJCM 1640014
Y. L. Wu et al.
Ning, D. Z., Zang, J., Liu, S. X., Taylor, R. E., Teng, B. and Taylor, P. H. [2009] “Free sur-
face evolution and wave kinematics for nonlinear uni-directional focused wave groups,”
Ocean Eng. 36, 1226–1243.
OpenFOAM User Guide [2008]. Available at: http://www.openfoam.org/docs/.
Recommended Practice DNV-RP-C205, Environmental conditions and environmental
loads (2014).
Taylor, P. H. and Haagsma, I. J. [1994] “Focusing of steep wave groups on deep water,”
Proc. Int. Symp.: Waves-Physical and Numerical Modelling, Vancouver, Canada,
pp. 862–870.
Tromans, P. S., Anaturk, A. R. and Hagemeijer, A., [1991] “A new model for the kinematics
by UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA on 11/17/15. For personal use only.
of large ocean waves-application as a design wave,” Proc. 1st Int. Offshore and Polar
Engineering Conf., Vol. 3, Edinburgh, UK, pp. 64–71.
Westphalen, J., Greaves, D. M., Williams, C. J. K., Hunt-Raby, A. C. and Zang, J. [2012]
“Focused waves and wave-structure interaction in a numerical wave tank,” Ocean Eng.
45, 9–12.
Int. J. Comput. Methods Downloaded from www.worldscientific.com
1640014-14