See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323573768

WAVE LOADS ON OFFSHORE STRUCTURES

Conference Paper · March 2018

CITATIONS READS

0 759

2 authors, including:

Peter Tromans
Ocean Wave Engineering Ltd
35 PUBLICATIONS   482 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Peter Tromans on 06 March 2018.

The user has requested enhancement of the downloaded file.

 7th International Conference on “Experiments/Process/System Modeling/Simulation/Optimization”
7th IC-EPSMSO
Athens, 5-8 July, 2017
©LFME

WAVE LOADS ON OFFSHORE STRUCTURES

Peter S Tromans and Luc Vanderschuren
Ocean Wave Engineering Ltd
99 Cumnor Hill, Oxford, OX2 9JR, UK
e-mail: [email protected]

Keywords: Waves, wave loads, offshore, ocean

Abstract. We discuss the estimation of wave loads on steel frame, offshore structures. The conventional design
recipe is based on simple, periodic waves. It may be adequate for the design of new structures, but more realistic
and complex approaches are required for an accurate analysis of wave loads when we re-assess old platforms.
Real, extreme, ocean waves are random, directional and non-linear to the point of breaking. Fully non-linear
calculations of wave kinematics can be employed, but present numerical schemes require significant computer
resources. Moreover, there are considerable difficulties in direct calculation of loads when the wave crest impacts
the deck of a platform. In this case, the best approach is to use random wave tests in a model basin to calibrate an
empirical model. However, such tests are expensive and time consuming. We suggest ideas for making better use
of basin data.

1. INTRODUCTION
Wave loading is a random process. A conservative, deterministic treatment with periodic waves, such as
Stokes, may be adequate for design of steel frame structures supporting decks for offshore oil. In contrast, for re-
assessment and reliability calculations, the most accurate and realistic models required and the probabilistic nature
of the problem has to be confronted. The reliability of an older structure is almost certainly influenced by wave
crests hitting the deck. Here, we highlight the need for fully non-linear, broad-banded models for wave kinematics
for sub-structure loads and the analysis of wave-in-deck load.
The purpose of a platform sub-structure is to keep personnel and deck equipment in safe conditions, out of
the sea water. To achieve this, it must, amongst other things, resist environmental loading from waves, wind and
current. The long-term problem of estimating the return periods of the environmental conditions is critical and
may be tackled in a number of ways such as the response-based method [16]. Here, we focus on the question: given
those environmental parameters, how can we calculate wave kinematics and force on the platform?

1.1 Physical background

Waves load the sub-structure and, if the crest of the wave is high enough, the deck of the platform. The
calculation of wave load depends on several elements. Consider the expression for drag force:

F = ½ ρv2 CdA (1)

The force, F, that we calculate depends on:

1. The geometry of the structure defined here by CdA

2. The force model and its coefficients, such as Cd. The coefficients may be derived from experiments or
calculation. They depend on the flow regime, Reynolds and Keulegan-Carpenter numbers, and the
roughness and shape of the body. The flow regime also determines the method used for the calculation.
Drag is an example (or element) of slender body theory valid when wave length is much greater than
body dimensions. Inviscid forces might also be calculated by diffraction analysis or total force by
computational fluid dynamics
3. The nature of the incoming fluid, water, air, foam or spray, that appears here in the drag equation as ρ
4. The wave and current field around the platform, which appears in this expression as v.

Our skill in these four areas determines the accuracy that we can achieve in calculating wave forces. Physical
and modelling uncertainties in each of them contribute to the total uncertainty in environmental load and are
important in setting load factors for new design and estimating the reliability of all structures.
 Peter S. Tromans and Luc Vanderschuren.
1.2 Practical engineering requirements
For the design of new structures, we require a procedure that is easy to apply and accurate to moderately
conservative in estimating sub-structure load. In general, we have that for sub-structure loading in the API/ISO
wave force recipe and Stokes design wave kinematics. Wave-in-deck loads should not be part of a design
calculation. Apart from design simplicity, there are good economic reasons for setting the deck sufficiently high
that a wave impacting it does not influence the reliability of the system. Overall, for design purposes, it is sufficient
that the wave load model, together with the resistance model and all the explicit and implicit safety factors of the
design process, are straightforward to apply and produce a reliable and cost-effective structure. The present design
wave force recipe first appeared in the 20th edition of API RP2A [13]. More information on its development may be
found in the references [1, 2, 12, 13, 26].
Re-assessment and reliability analysis have different requirements from design. At present, much effort is
going into the re-analysis of old platforms with decks set too low and sub-structures designed to outdated criteria.
For this, accuracy takes priority over ease of application.
The API recipe and its general acceptance provided a huge step forward. The only aspect that might be
criticized, in terms of approaching realism, was the retention of regular, periodic wave kinematics. In the
subsequent adoption of the API Recipe in ISO, some additional possibilities were added: one being the use of
‘representative waves from random seas’ such as NewWave. In its simplest form, a NewWave provides a
deterministic, linear model for the expected time history around a high crest in a random sea [17-19]. The statistics
of departures from the expected history may also be calculated, though, for most practical purposes, they are not
required. More significant than the details of NewWave is that the allowance of ‘representative waves from random
seas’ conveniently opens the door to more realistic wave models, at least for those who have the necessary
computational and modelling resources [15, 20].

2. REAL WAVES
Real waves are random; their spectra are broad-banded with energy spread over frequency and direction; and
they are non-linear with complex interactions between frequency components. We shall consider each of these
properties below.
2.1 Broad-banded spectrum
The random and broad-banded nature of real waves introduces ambiguities in specifying a periodic design
wave. The regular wave has a unique, deterministic relationship between wave height, crest elevation and the
associated surface elevation and kinematic fields: real waves do not. Moreover, a real wave, of the same wave
height as a Stokes wave, will, on average, have a much higher crest. We use height to specify the wave amplitude:
yet, crest elevation is more closely related to drag force on a column [3]. The ‘associated period’ of the regular
design wave is the time between two successive down-crossings. Regular waves have a fixed down-crossing wave
period: for real waves this is a random quantity. And, there is no simple expression to obtain even the expected
value of down-crossing period for extreme, random waves. Compared with regular waves, a broad-banded
spectrum such as JONSWAP tends to produce horizontal water particle velocities that are relatively more severe
at the top and bottom of the water column and less severe at an intermediate level. Thus, the bias in mud-line shear
may be different from that in overturning moment. These effects are modified and most often amplified by non-
linearity. Indeed, where bias in the API recipe can be estimated from structural monitoring exercises, it supports
this theoretical observation [2, 21].
2.2 Directional spreading
The directional spreading of wave energy is well established. In a wind sea, low frequency and high frequency
components are spread much wider than the spectral peak. The effect is well defined in the Ewans models for fetch
limited seas and for swell [4]; his directional distribution for wind-sea is shown in Figure 2. Thus, while the standard
deviation of spreading direction might be less than 20° at the spectral peak, the average for the whole spectrum
may exceed 30°.
Spreading reduces horizontal velocity in the main wave direction. The API/ISO Recipe accounts for it with
the directional spreading factor. The spreading factor approximates the velocity reduction ratio beneath the wave
crest, but works less well in the rest of the wave field. It becomes too small as we move away from the crest in the
mean wave direction and too large as we move normal to mean wave direction. Van Weert and Harland suggest
that, for a typical 8-leg sub-structure, this can lead to over prediction of force of order 10% for broadsides wave
attack and a smaller under prediction for end-on [5].
 Peter S. Tromans and Luc Vanderschuren.

+90

Direction, deg
0
ω/ωp

-90
1 2
Figure 2. Directionality in the wave spectrum
(Density of spots indicates density of energy)

2.3 Non-linearity
Regular design wave models are non-linear as exhibited through the bound wave components seen in the
Stokes expansion. However, there are important non-linear processes that are specific to broad-banded, directional
seas: four wave interaction or quartet resonance. In random seas, large crests originate from small linear
components, free waves, coming into phase at a focal point. As the components approach focus, resonant quartets
swap energy. Energy moves away from the spectral peak to lower and, in particular, to higher frequencies.
Simultaneously, energy moves from the more spread directions toward the central direction of the spectrum. This
changes the distribution of velocities over depth and with distance from the peak of the wave crest. The overall
consequence in the water column is to reduce horizontal velocities around still water line and below, while
increasing them high in the wave crest. The effect is illustrated in Figure 4. The shifting of energy may also amplify
the crest elevation, especially if the directional spreading is narrow. The increase in horizontal velocity of fluid at
the top of the crest advances breaking. In turn, breaking will ultimately limit crest elevation and wave height.
Either way, if the top of the crest hits a deck, the amplified velocity will greatly increase load.

Figure 4. Comparison of kinematics from a Stokes wave and a non-linear NewWave

(Plot based on data of C. Swan)

We should bear in mind that most design conditions and survival (or failure) conditions correspond to very
high, steep waves. Non-linearity will be significant; some degree of breaking will be present or developing. This
contrasts not only with regular wave models, but also with the waves we encounter more frequently in the sea and
are more likely to observe in offshore measurement programmes.
In summary, there are a host of features that distinguish real ocean waves from regular wave theories. Some,
but not all of them, lead, on average, to reduced horizontal velocities and, thereby, lower loads on sub-structures.
All lead to inconsistencies that demand careful interpretation when we attempt to approximate them with a periodic
wave model. However, we can avoid these difficulties and, most often, obtain significantly lower calculated wave
loads on sub-structures by adopting fully non-linear NewWaves for generating our wave kinematics. More
information on the relevant calculation schemes can be found in Bateman, Swan and Taylor [14], Hague and Swan
[15]
and Chalikov and Babanin [25] and references that they list. Some methods using Fourier schemes [24,25] are
 Peter S. Tromans and Luc Vanderschuren.
relatively efficient Fourier scheme. Though they can treat fairly steep cases, they may fail some way before of
wave breaking. Boundary element methods, such as that of Hague and Swan [15], should work further into breaking,
but the computational effort is high.

3. WAVE-IN-DECK
When wave crests impact a deck, large vertical and horizontal forces are generated. The horizontal loads are
generally the most important for structural reliability and are discussed here.
3.1 Wave-in-deck background
The industry’s position with wave-in-deck load is by no means as mature as that for sub-structure load. Several
semi-empirical approaches have been adopted for calculations. They are variously based on drag models with a
single drag coefficient for the entire deck, Morison models for the components of the deck, von Karman
momentum transfer schemes, and rate of destruction of momentum [6-8]. Much work is summarized by Bolt [23].
The methods may include shielding and blockage effects. Some significant factors are:

1. They scale broadly on the momentum flux into the deck from the incoming wave, the classic ρAv2, where
A and v are representative area and velocity terms.
2. They require a significant amount of model testing to calibrate their coefficients for different wave attack
directions and levels of deck inundation.
3. The wave-in-deck load varies rapidly and should be subject to a dynamic transfer function before
application as a quasi-static load in a structural analysis.

The last point is essential, yet often neglected. The dynamics of the structural system modify the load,
amplifying part and filtering off short duration spikes. An indicative time history of horizontal load on a deck is
shown in Figure 5. Wave-in-deck loads will be used in a plastic analysis of the ultimate strength of the structure.
If that analysis is quasi-static, the wave-in-deck load should first be dynamically transformed with a single degree
of freedom model. A single, peak load can then be input into the strength software. If the plastic analysis uses a
the more realistic dynamic scheme, then we must provide a suitable time history, such as that in Figure 5.

20
Force, MN

15

10

0
-2 0 2 4 Time, s 6

-5

Figure 5. Wave-in-deck load history

The momentum flux into the deck depends on its geometry and the water particle kinematics, in particular the
velocity field at the top of the wave crest and the shape of the intersection between the water surface and the deck
volume; aspects of real waves about which we have the greatest uncertainty and difficulty in treating. Calculations
can go only so far. The velocities in the crest change rapidly and over a wide range, perhaps by a factor of two
during breaking, eventually exceeding the phase speed, while the crest elevation changes little from its highest
value. The favoured approach is to use model basin experiments with random, directional waves supported by a
few additional tests with NewWaves. However, the amount of random wave testing required to make sense of the
data is very large. The cost and time demands are high. We need methods that can extend the applicability of wave
basin tests. We propose some ideas based on dimensional analysis and simple models.
3.2 Dimensional analysis
Model testing depends on correct scaling. Omitting surface tension and viscous effects, we are left with the
Froude number and Euler numbers. For waves, the Froude number may be written as
𝜔
𝐹𝑟 = (1)
√𝑔𝑘
 Peter S. Tromans and Luc Vanderschuren.

where 𝜔 is wave frequency and 𝑘 is wave number. Since we are most often interested in random waves, these will
typically be values associated with the peak of the wave spectrum.
The Froude number is the ratio of inertial and gravity forces. Since 𝜔 and 𝑘 are used to make all time and
length quantities dimensionless, it tells us how variables measured in time and length relate between model tests
and full scale, but it does not define the relevance of an experiment. For that, we need geometric parameters such
as wave steepness, 𝑎𝑘, where 𝑎 is the crest elevation, and relative water depth, 𝑘ℎ, where ℎ is the depth sea-bed
to still water level. We might also add some parameters defining the background wave spectrum such as the peak
enhancement factor, γ, and a dimensionless significant wave height, 𝑘𝐻𝑠 .
The platform deck also has dimensionless geometric parameters. The most important is the elevation of the
underside of the deck, 𝑘𝑒. We approximate the deck as rectangular prism with length and breadth, 𝑑1 and 𝑑2 , and
a height great enough that the top of the deck is always above the top of a wave crest. It has a porosity, 𝑑𝑝 , and a
pore scale, 𝑑𝑝𝑠 . We combine the deck dimensions into one vector 𝒅 and the dimensionless 𝑘𝒅. We should also
include, 𝜃, the orientation of the deck relative to the main wave direction.
Thus, the specification and interpretation of wave basin experiments require attention to 𝐹𝑟 , 𝑘ℎ, 𝜃 and 𝑘𝑒.
Since crest elevation must be approximately equal to or a little greater than deck elevation, 𝑘𝑒 also fills the role of
𝑎𝑘 in defining the experimental model.
Having specified the experiment, we have to scale the force. The Euler number is the ratio of pressure and
inertial forces. For wave-in-deck load, we might generalize it to the ratio of horizontal load on the deck to the
incoming momentum flux. The modified Euler number is 𝐶 and the peak wave-in-deck force, 𝐹𝑑 , is given by

𝐹𝑑 = 𝐶𝑤(𝑎 − 𝑒)𝜌(𝑎𝜔)2 (2)

where 𝑤 is a measure of the width of the wave inundation. It may be the width of the deck if it is not too wide.
Thus, 𝑤(𝑎 − 𝑒) is an effective area of inundation.
We have found that equation (2) provides an adequate parametric model for fitting measured wave-in-deck
loads. 𝐶 or 𝐶𝑤 are a dimensionless force, though they take on the role of a constant. They are best estimated by
fitting equation (2) to measured data using quantile-quantile plots. The worst force is plotted against the highest
wave, 2nd against 2nd, and so on, nicely ordering the data and capturing peak values at comparable probabilities;
see for example Bretschneider [24] and Heidemann [2]. The dependence on wave steepness and water depth is
sufficiently weak that 𝐶 and 𝐶𝑤 may be treated as constant for a given platform and wave attack direction. Some
example results are plotted in Figure 6.

Figure 6. Some measured forces plotted against inundation

The points correspond to different water depths, 𝑘ℎ = 2.34 (blue), 2 (red), 1.85 (green).
Data were low pass filtered to remove noise. Here a safe estimate for 𝐶 might be 1.45.

We will assume the platform sub-structure, on which the deck sits, is sufficiently transparent to waves that it
does not influence the load. However, it will introduce a dynamic response into the problem that we can treat by
adding a dynamic amplification in equation (2) to give a peak in dynamically amplified force:

𝐹𝑑𝑑 = 𝐷𝐶𝑤(𝑎 − 𝑒)𝜌(𝑎𝜔)2 (3)

 Peter S. Tromans and Luc Vanderschuren.
Computational fluid dynamics offers great opportunities for calculating forces. But, it still requires the
kinematics of approaching random wave crests, if only to provide a correct upstream boundary condition for the
flow grid around the deck. There is little point in such calculations if the correct boundary conditions are not
available or if the large physical uncertainties and possibly significant modelling uncertainties are unknown.
Indeed, most often we would still use model tests to calibrate the results. For the present, if accurate wave-in-deck
loads are required, basin tests and parametric models provide a workable, though difficult approach.
3.3 Time history
Our experience is that the equations (2) and (3) work very well. However, they relate to peak values and do
not give us time histories. Very often model tests have recorded only 𝐹𝑑 and not the dynamically amplified 𝐹𝑑𝑑 or
if they have, the dynamic transfer function may be for a different platform from the one of interest. To correct for
this, we need to produce a time history of load. The same need applies if the plastic analysis of ultimate strength
is made with a dynamic scheme. Thus, we would like to have a simple scheme for generating time series that can
be calibrated against peak forces obtained from equation (2) and empirical values for 𝐶 or 𝐶𝑤.
We propose a simple model. We make the assumption that horizontal wave-in-deck load is a dominated by
two components. The first is the integral of the flux of horizontal momentum into a rectangular control volume
drawn tightly around the deck. It could be represented as a horizontal drag force on cylinders surrounding the deck.
Note that although this term appears to be inertial, it is ultimately viscous drag in that the incoming momentum is
assumed to be destroyed. The second component is the total von Karmann momentum transfer in the horizontal
direction [27] (or the rate of change of added mass integrated over the deck volume) as the crest engulfs the deck.
This term is inertial and changes sign as the deck emerges from the wave crest. Thus, the time series of horizontal
force becomes

𝐹(𝑡) = ∫ 𝜌𝑣𝑛 𝑢. 𝑑𝐴𝑐𝑣 + ∫ 𝜌𝑢𝜂̇ . 𝑑𝐴𝑑 (4)

where 𝑣𝑛 is velocity normal to the control volume, 𝑢 is the horizontal component of velocity, 𝜂̇ is the local rate of
rise of water surface elevation around an element of the deck, 𝑑𝐴𝑐𝑣 is an infinitesimal element of the surface of
the control volume and 𝑑𝐴𝑑 is an element of plan area of the deck.
We have made an example calculation of static and dynamically transformed force time histories using an
approximate wave model. The results are shown in Figure 7. Though the force amplitudes are consistent with
measurements, more work is required for testing the time histories.

Figure 7. Calculated time histories of horizontal wave-in-deck load

Applied force, blue. Dynamically transformed, orange
𝑎𝑘 ≈ 0.34, 𝑘𝑒 ≈ 0.31

4. UNCERTAINTY IN WAVE LOAD – A BRIEF COMMENT

Since the main application of all that we have discussed is in reliability calculations, for completeness, we
must, at least briefly, mention uncertainty in wave loads. Measurements of both sub-structure and wave-in-deck
loads in realistic seas show large variability even for a fixed crest elevation. It arises because, although we may
know the surface elevation at a point, uncertainty remains concerning elevations and kinematics even close by. We
account for the uncertainty by using quantile-quantile methods in calibrating models. This effectively convolutes
the uncertainty into the relationship between the force and point crest elevation. However, we generate our sub-
structure and wave-in-deck load models separately. Are there consequences from correlation between the uncertain
parts of the deck and sub-structure loads?
Positive correlation is correctly incorporated within the quantile-quantile analysis. We would see negative
correlation when the uncertainties lead to errors in wave-in-deck and sub-structure loads that are of opposite sign.
The discussion in section 2.3 indicates that higher velocities at the top of the wave crest are associated with lower
velocities down the water column. Thus, there is a possibility of some negative correlation. It would cause our
 Peter S. Tromans and Luc Vanderschuren.
calibrations to over predict the sum of wave-in-deck and sub-structure loads. We conclude that the approach may
err to the safe side.

5. CLOSING REMARKS
Though for design, we can produce adequate structures using the API recipe, for reliability calculations, we
should go further. Two aspects demand particular attention. The first is the accuracy and relevance of our design
wave kinematics. Regular waves may be convenient and established, but they are not good representations of real
ocean waves. Real waves are random, directional and have quite different non-linear behaviour. Most of these
features serve to reduce the calculated load on the sub-structure. The significant cost of a careful application of a
fully non-linear random wave model can be more than justified.
The non-linear features are equally important for wave-in-deck loads, though they are likely to increase
calculated loads on decks. The resources required for the wave modelling and the complexity of applying it to the
wave-in-deck problem make the use of fully non-linear schemes much more difficult for this problem. The better
option may be to resort to model testing. Even this requires a large effort to determine deck loads accurately.
Model tests often record only peaks in measured loads. We have suggested a scheme that may allow us to
create time histories.

ACKNOWLEDGEMENTS
This paper would not have been possible without the collaboration of numerous co-workers past and present,
notably Chris Swan, Jason McConochie, John Heideman, Kevin Ewans, Jan Willem van de Graaf and Paul Taylor.

REFERENCES
[1] Petrauskas, C., Heideman, J.C., and Berek, E.P., 1993, Extreme wave force calculation for the 20th edition of
API RP-2A, 25th Annual Offshore Technology Conf, May 1993, Houston, TX, USA, OTC 7153, p 201.
[2] Heideman, J.C., and Weaver, T.O., 1992, Static wave force procedure for platform design, ASCE Civil
Engineering in the Oceans V, November, 1992.
[3] Tromans, P.S, Hagemeijer, P.M., and Wassink, H.R., 1992, The statistics of the extreme response of offshore
structures, Ocean Engineering, v. 19, 2, p 161.
[4] Ewans, K.C., 1998, Observations of the directional spectrum of fetch-limited waves J. Phys. Oceanogr., 28, p
495.
[5] Van Weert, P.J., and Harland, L.A., 1998, Wave spreading and extreme wave loading, 17th Offshore Mech.
and Arctic Engg. Lisbon, Portugal.
[6] Kaplan, P., 1992, Wave impact forces on offshore structures: re-examination and new interpretation, 24th
Annual Offshore Technology Conf, , Houston, TX, USA.
[7] Finnigan, T.D., and Petrauskas, C., 1997, Wave-in-deck forces, Proc. 6th ISOPE.
[8] Lande, O., and Johannesen, T.B., 2011, CFD analysis of deck impact in irregular waves: wave representation
by transient wave groups, 30th Offshore Mech. and Arctic Engg. Rotterdam, Netherlands.
[9] Jonathan, P., and Taylor, P.H., 1996, Wave force variability on offshore structures, 15th OMAE, Florence,
Italy.
[10] Harland, L.H., Taylor, P.H., and Vugts, J.H., 1998, The extreme force on an offshore structure and its
variability, Applied Ocean Research, 20, 3-14.
[11] Digre, K.A., Puskar, F.J., Aggarwal, R.K., Irick, J.T., Krieger, W.F., and Petrauskas, C., 1995, Modifications
to and applications of the guidelines for assessment of existing platforms contained in section 17 of API RP2A,
OTC paper OTC 7779.
[12] UK Department of Energy, Offshore installations: Guidance on Design, Construction and Certification,
HMSO, 1993.
[13] API RP2A, Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms 20 th
Edition, 1993.
[14] Bateman, W.J.D., Swan, C.,and Taylor, P.H., 2003, On the calculation of the water particle kinematics arising
in a directionally spread wave field, Journal of Computational Physics, Vol:186, pp. 70-92.
[15] Hague, C.H., and Swan, C, 2009, A multiple flux boundary element method applied to the description of
surface water waves, Journal of Computational Physics, pp. 5111-5128.
[16] Tromans, P.S., and Vanderschuren, L., 1995, Response-based design conditions in the North Sea: application
of a new method, 27th Annual Offshore Technology Conf, May 1995, Houston, TX, USA, OTC 7683, pp.
387-397.
[17] Lindgren, G., 1970, Some properties of a normal process near a local maximum, Ann. Math. Stat., 41, pp.
1870-83.
[18] Boccotti, P., 1983, Some new results on the statistical properties of wind waves, Appl. Ocean Res., 5, pp. 134-
140.
 Peter S. Tromans and Luc Vanderschuren.
[19] Tromans, P.S., Anaturk, A., and Hagemeijer, P., 1991, A new model for the kinematics of large ocean waves,
Proc. 1st Int. Conf. Offshore Mech. And Polar Engg. (ISOPE), Edinburgh, UK, 3, pp. 64-71.
[20] Bateman, W., Swan, C., and Taylor, P.H., 2001, On the efficient numerical simulation of directionally spread
water waves, J. Comp. Phys., 174, p. 277-305.
[21] Atkins, N., Lyons, R.A., and Rainey, R.C.T., 1995, Summary of findings of wave load measurements on the
Tern platform, UK Health and Safety Executive Report, OTH 95 467.
[22] Swan, C., and James, R.L., 2001, A simple analytical model for surface water waves on a depth-varying current.
Appl. Ocean Res., 22, pp 331-347.
[23] Bolt, H., 1998, Review of wave in deck load assessment procedures, BOMEL report to the UK Health and
Safety Executive, HSE Offshore Technology Report OTO 97 073.
[24] Bretschneider, C., 1965. On the probability distribution of wave force and an introduction to the correlation
drag coefficient and the correlation inertial coefficient. Coastal Engineering, Proceedings ASCE Specialty
Conference, Santa Barbara, CA.
[25] Chalikov, D., Babanin, A.V., Sanina, E. 2014, Numerical modelling of 3D fully nonlinear potential periodic
waves, Ocean Dynamics, 645, pp 1469-1486.
[26] ISO 19902:2007 Petroleum and natural gas industries – Fixed steel offshore structures, International Standards
Organisation.
[27] Von Karman, T., and Wattendorf, F.L., 1929, The impact on seaplane floats during landing. NACA Technical
Note No. 321.

View publication stats