The behaviour of a moored oil tanker in the

Port of Leixões, Portugal

-With use of numerical models-

Martin van der Wel

Delft, The Netherlands
May 2011
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 2
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The behaviour of a moored oil tanker in the

Port of Leixões, Portugal
-With use of numerical models-

MSc. Thesis
Martin van der Wel
May 2011

Graduation committee:
Prof. Ir. H. Ligteringen TU Delft CiTG Ports and Waterways
Prof. Dr. Ir. R.H.M. Huijsmans TU Delft 3ME Ship Hydromechanics
Dr. Ir. M. Zijlema TU Delft CiTG Fluid mechanics
Ir. P. Quist TU Delft CiTG Ports and Waterways
Dr. Eng. P. Rosa-Santos FEUP
Dr. Ir. W. van der Molen CSIR
Ir. A. van der Hout Deltares

In cooperation with:

Final 3
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 4
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Preface
This thesis presents a study about the numerical modelling of a moored oil tanker in the port of
Leixões, Portugal. T he research is carried out by a Hydraulic Engineering student from the Faculty of
Civil Engineering and Geosciences of the Technical University of Delft, The Netherlands. The work has
been carried out in cooperation with the Faculty of Engineering University Porto (FEUP), Por tugal and
the Council for Scientific and Industrial Research (CSIR), South-Africa and Deltares, The Netherlands.

Sometimes fellow students asked me w hy I would like to graduate in ports and waterways. To answer
that question it may take some sentences, but the shortest possible answer is because of personal
interest. A port cannot be designed without involvement of other specialism‟s. Subsequently a
graduation student in the field of ports and waterways cannot graduate without having knowledge of
other specialism‟s. Looking back, my graduation pr oject was a journey along the beautiful places of
Delft, Por to and Stellenbosch. But at the same time a journey along the fields o f coastal engineering,
port planning, ship hydromechanics, physical modelling and mainly numerical modelling.

During the journey many people contributed to the progress, while showing me the beautifulness of
their specialism. All these contributions ensured small and sometimes lar ge sidesteps on a path
reaching the goals at the end of the journey. There are however some people whose contribution is
indispensable and therefore I would like to show them my gratitude.
I would like to acknowledge Prof. Ir. H. Ligteringen for providing me this subject and the contacts I
needed that came in at exactly the right times. The discussions and conversations through the
progress definitely helped me reaching goals. From the Technical University of Delft I further would
like to acknowledge Prof. Dr. Ir. R.H.M. Huijsmans, Dr. Ir. M. Zijlema and Ir. P. Quist for their
guidance, suggestions and advises throughout the progress. From FEUP I w ould especially thank Dr. P.
Rosa Santos who showed me the physical model, the protot ype and the surroundings of Porto,
meanwhile finishing his Phd. thesis. Special thanks to Dr. Ir. W. Van der Molen, without his work and
knowledge I w ould not be able to translate the motions of waves into the motions of ships. Also his
help during my weekends of stay in Stellenbosch are highly appreciated (having a flat tire after 30km
was not bad for a weekend). I would like to thank the company of Deltares for offering me the
oppor tunity to carry out additional simulations with their Boussinesq-type wave model. The knowledge
of Ir. A.J. van der Hout was essential to discover the opportunities of the model and placing the
results into a larger perspective. Looking from different perspectives to the same problem was
necessary and the advises from Ir. B. van Vossen, Dr. Ir. M.P.C. De Jong and Ir. O. Weiler helped me
by doing that. In general I would like to thank my family and friends for their endless support, either
during graduation or otherwise.

Martin van der Wel

“Using a complex process based model, like Delft3D or Mike, cannot be used as an excuse for not
understanding the main processes” D.J. Walstra, lecturer Coastal Dynamics II, T U Delft (2010)

“You are the intelligent part in the chain, the model is the calculator” D.J. Walstra again.

Final 5
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 6
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Summary
The Port of Leixões is located in the north of Por tugal. T he operational conditions at berth “A” are
affected by several factors resulting in a dow n-time of the berth of about 20%. In order to describe
the behaviour of the moored vessel physical model tests were performed at the Faculty of Engineering
University of Por to within the research and development program DOLPHIN. Numerical simulations of
wave propagation including non-linear wave interaction and the generation of both sub- and super-
harmonics were not carried out in the DOLPHIN program.

Within this study a sequence of numerical models was applied to describe the moored ship motions
and make a comparison with measurements from the physical model tests . An approach which
combined a Boussinesq-type wave model with a panel model was selected as the appropriate
approach for the present study. A Boussinesq-type wave model was selected, since diffraction around
a breakwater, partial reflection from port structures and non-linear wave pr ocesses are impor tant and
relevant hydrodynamic processes to describe the wave field in the vicinity of the ship.

The output from the Boussinesq-type wave model describes the surface elevations as well as the
velocity field of the waves in the vicinity of the ship. The successive panel model takes the presence of
the ship within the incident wave field into account and calculates the wave forces on the ship. The
calculated total wave forces are a summation of the Froude-Kr ylov, diffraction and second order wave
forces. Time-series of wave forces serve as an input for a ship simulation model, which simulates the
ship motions and the mooring forces, taking into account environmental forces and non-linear
interactions with the mooring system.

The initial approach combined the Boussinesq-type wave model MIKE21 BW, with the panel model
Harberth and the ship simulation model Quaysim. Due to numerical instabilities within MIKE21 BW an
eddy with unreliable velocities was formed after longer period of simulation. This eddy was located in
the vicinity of the ship. The subsequent computations with Harber th resulted in a continuous increase
in second order wave forces on the ship. The first or der wave forces on the ship did not increase
continuously, but via an analysis it is shown that the calculated first or der wave forces on the ship are
not reliable. The finally simulated ship motions in Quaysim are not reliable as well. T he unreliable
wave forces on the ship as well as the unreliable subsquent simulated ship motions are a consequence
of the pre-simulated instabilities within MIKE21 BW.

The cause of the numerical instabilities in MIKE21 BW could not be discovered during this study. Since
longer period of simulations are required to obtain statistical reliable ship motions a switch was made
to an alternative Boussinesq-type wave model called TRITON. Adaptations in the model set-up in
TRITON ensured that numerical instabilities over longer periods of simulation in the simulated velocity
field were avoided/minimized.

The emphasis in simulations with TRITON is on simulating and understanding the low frequency
waves. The focus was on simulating the low frequency waves that caused significant moored ship
responses within the physical model. T he simulated waves in TRIT ON agree well with the measured
waves in the physical model basin. It is shown that a standing wave is measured in the physical model
basin, which caused measured surge responses of the ship. The obtained insight in the generation of
these low frequency waves may be used when performing additional physical and/or numerical model
tests.

It is recommended to simulate the total Port of Leixões with a Boussinesq-type wave model, taking
into account all relevant hydrodynamic processes for moored vessel response. In that case a bound
long wave should already be imposed in the incoming generated waves, since the generation of
primary waves only may lead to an underestimation of the total content of low frequency waves.

The wave simulations, as per formed with TRITON, are expected to be sufficiently accurate to ser ve as
an input for vessel response computations. The recommended additional wave simulations and vessel
response computations are expected to provide additional insight in the causes of excessive ship
motions. New insight in the behaviour of waves as well as vessel response may lead to improvements
of the operational conditions at berth “A” in the Por t of Leixões.

Final 7
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Samenvatting
De haven van Leixões is gelegen in het noorden van Portugal. Door diverse factoren is afmeerplek “A”
gedurende 20% van de tijd niet operationeel. Om het gedrag van de afgemeer de olie tanker te
onderzoeken zijn fysieke model testen uitgevoerd aan de Universiteit van Por to in het kader van een
onderzoeks- en ontwikkelingsprogramma genaamd DOLPHIN. Numerieke golfvoortplanting simulaties,
inclusief niet-lineaire interacties tussen golven en het opwekken van zowel som- als verschilfrequente
golven, ontbraken binnen het kader van het DOLPHIN programma.

Gedurende deze studie zijn diverse aaneengesloten numerieke modellen toegepast om het gedrag van
de afgemeerde olie tanker te beschrijven. De resultaten van de numerieke simulaties wor den
geverifieerd met metingen verricht tijdens proeven met het schaalmodel. Voor deze studie is een
combinatie van een Boussinesq-type golf model met een panelenmodel als meest geschikte aanpak
gekozen. Een Boussinesq- type golf model was geselecteerd, omdat diffractie van golven om een
golfbreker, par tiële reflectie van golven door aanwezige havenconstructies en niet-lineaire interacties
tussen golven beoor deeld waren als belangrij ke en relevante golf processen om de scheepbewegingen
door golfexcitatie accuraat te kunnen beschrijven.

De uitkomsten van het Boussinesq-type golfmodel bestaat uit oppervlakte-uitwijkingen en een

beschrijving van het snelheidsveld door golven om het schip. Het panelenmodel simuleer t de
aanwezigheid van het schip binnen dit inkomende golfveld. De berekende totale golfkrachten op het
schip bestaan uit een sommatie van de Froude- Krylov kracht, diffractiekracht en tweede-orde
golfkrachten. De ver kregen tijdseries van golfkrachten op het schip worden ver volgens gebruikt in een
schip simulatiemodel. Het schip simulatiemodel simuleert de scheepsbewegingen en afmeer krachten,
hierbij rekening houdend met externe excitatiekrachten en niet-lineaire interacties met het
afmeersysteem.

De initiële aanpak combineerde het Boussinesq-type golf model MIKE21 BW, met het panelenmodel
Harberth en het schip simulatie model Quaysim. Gedurende lange perioden van simulatie raakte
MIKE21 BW numeriek instabiel. Ten gevolge van het numeriek instabiel geraken werd er een neer
gevormd in de buurt van het schip met onbetrouw bare snelheden. In de opvolgende berekeningen
met het panelenmodel Harberth is er een continue oplopen van tw eede orde krachten op het schip
geconstateerd. De eerste-or de golfkrachten op het schip liepen niet op, maar uit een analyse is
gebleken dat deze eerste-orde golfkrachten niet betrouwbaar zijn. De uiteindelijk gesimuleerde
scheepsbewegingen met Quaysim zijn dientengevolge ook niet betrouwbaar. Ten gevolge van het
numeriek instabiel raken van MIKE21 BW zijn zowel de berekende golfkrachten op het schip als de
gesimuleerde scheepsbewegingen en afmeerkrachten onbetrouwbaar.

Waardoor MIKE21 BW numeriek instabiel is geraakt kon tij dens deze studie niet worden achter haald.
Omdat simulaties over langere periode benodigd zijn om scheepsbewegingen statisch betrouwbaar te
kunnen beschrijven is een alternatief Boussinesq-type golfmodel genaamd TRITON toegepast.
Veranderingen in de opzet van het model binnen TRITON hebben er toe geleid dat het numeriek
instabiel raken van het model (zelfs na langere periode van simulatie) voorkomen/geminimaliseerd
kon w orden.

De nadruk van de simulaties uitgevoerd met TRITON lag op het simuleren en begrijpen van laag-
frequente golven die excitatie van het afgemeerde schip in het fysieke schaalmodel hebben
veroorzaakt. Het gesimuleer de golfveld in TRITON komt overeen met het gemeten golfveld in het
fysieke schaal model. Het is aangetoond dat er een staande golf is gemeten tij dens proeven met het
schaalmodel. De gemeten schrikbewegingen van het schip worden veroorzaakt door deze staande golf.
Het inzicht ver kregen met de uitgevoerde numerieke simulaties kan worden gebruikt in additionele
proeven met het fysieke schaalmodel.

Final 8
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Het is aanbevolen om het golfveld binnen de gehele haven van Leixões te beschrijven met een
Boussinesq-type golf model. Hierbij dienen alle relevante golfprocessen om het gedrag van de
afgemeerde olie tanker te kunnen beschrijven meegnomen te worden. In dat geval dient een
gebonden lange golf al opgelegd te wor den in de gegenereerde inkomende golven, omdat het alleen
opwekken van primaire golven kan leiden tot een onderschatting van de totale hoeveelheid
laagfrequente golfenergie.

De verwachting is dat het golfveld binnen TRIT O N voldoende nauwkeurig is beschreven om als invoer
te dienen voor het panelenmodel. Verwacht wordt dat de aanbevolen additionele golfsimulaties
alsmede schipsimulaties ver nieuwd inzicht zal geven in de oorzaken van de excessieve
scheepsbewegingen. Het vernieuw de inzicht in het gedrag van golven en de daarmee gepaar d gaande
scheepsbewegingen kan leiden tot verbeteringen in de operationele condities ter plaatse van
afmeerplek “A” in de haven van Lei xões.

Final 9
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Contents
Preface ....................................................................................................................................... 5
Summary ................................................................................................................................... 7
Samenvatting............................................................................................................................. 8
Contents ................................................................................................................................... 10
List of symbols ........................................................................................................................ 12
Definitions ............................................................................................................................... 13
List of figures .......................................................................................................................... 15
List of tables ............................................................................................................................ 17
1 Introduction .................................................................................................................... 18
1.1 Motive of study.......................................................................................................... 18
1.2 Problem definition..................................................................................................... 18
1.3 Objective of study...................................................................................................... 19
1.4 Outline of the thesis ................................................................................................... 19
2 Background information................................................................................................ 20
2.1 Port of Leixões .......................................................................................................... 20
2.1.1 Berth “A”........................................................................................................................................................ 21
2.2 Physical model tests................................................................................................... 22
2.2.1 DOLPHIN project ......................................................................................................................................... 22
2.2.2 Physical model .............................................................................................................................................. 23
2.2.3 Tanker and mooring system characteristics ............................................................................................. 26
2.2.4 Physical model results.................................................................................................................................. 28
2.3 Relevant physics ........................................................................................................ 29
2.3.1 Wave forcing ................................................................................................................................................. 29
2.3.2 Harbour oscillations ..................................................................................................................................... 31
2.3.3 Ship motions in infragravity waves ........................................................................................................... 32
2.3.4 Mooring arrangement of an o il tanker....................................................................................................... 33
3 Physical model analysis .................................................................................................. 34
3.1 Hydrodynamic processes ........................................................................................... 34
3.2 Analysis of measurements ......................................................................................... 35
3.3 Summary of physical model analysis ......................................................................... 38
4 Research methodology ................................................................................................... 40
4.1 Methodology.............................................................................................................. 40
4.2 Description of the Boussinesq-type wave models ....................................................... 41
4.2.1 MIKE21 BW.................................................................................................................................................. 41
4.2.2 TRITON ......................................................................................................................................................... 42
4.2.3 Co mparison MIKE21 BW and TRITON.................................................................................................. 42
4.3 Description of the 3D panel model............................................................................. 44
4.4 Description of the ship simulation model................................................................... 46
5 Numerical simulations with MIKE21 BW, Harberth and Quaysim ......................... 47
5.1 Wave propagation simulations .................................................................................. 47

Final 10
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5.1.1 Model set-up .................................................................................................................................................. 47

5.1.2 Model results ................................................................................................................................................. 48
5.2 Wave force calculations ............................................................................................. 51
5.2.1 Model set-up .................................................................................................................................................. 51
5.2.2 Model results ................................................................................................................................................. 51
5.3 Ship simulations ........................................................................................................ 56
5.3.1 Model set-up .................................................................................................................................................. 56
5.3.2 Model results ................................................................................................................................................. 56
5.4 Discussion of model results ........................................................................................ 57
6 Numerical simulations with TRITON .......................................................................... 60
6.1 Model set-up.............................................................................................................. 60
6.1.1 General settings ............................................................................................................................................. 60
6.1.2 Boundary conditions .................................................................................................................................... 60
6.2 Program of simulations ............................................................................................. 60
6.2.1 Summary of simulat ions .............................................................................................................................. 63
6.3 Explanation of boundary conditions .......................................................................... 63
6.3.1 Partial reflecting boundaries ....................................................................................................................... 63
6.3.2 Generated waves ........................................................................................................................................... 64

7 Comparison physical model results and TRITON ...................................................... 66

7.1 Summary of simulations ............................................................................................ 76
8 Discussion of Boussinesq-type wave model results ...................................................... 77
9 Conclusions and recomme ndations............................................................................... 78
9.1 Conclusions ............................................................................................................... 78
9.2 Recommendations ..................................................................................................... 79
Bibliography............................................................................................................................ 80
Appendices .............................................................................................................................. 85

Final 11
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

List of symbols
Roman symbols
Symbol Unit Designation
A [kg ] Added mass matrix
[ ] Sea floor
B [kg / s] Damping matrix
Bs [ m] Breadth of ship
C [kg / s 2 ] Hydrostatic restoring matrix
Cr [ ] Reflection coefficient
c [m / s] Wave celerity
cg [m / s] Group velocity
[ ] Computational fluid domain
d [ m] Still water depth
ds [ m] Draft of ship
[ ] Free surface
F [N ] Force
Fx , y , z [N ] Surge, sway and heave force
f [ s 1 ] Wave frequency
g [m / s 2 ] Gravitational acceleration
[m2 ] Wetted surface of ship hull
h [ m] Total water depth
K [kg / s 2 ] Impulse response function matrix
LOA [ m] Length over all
LBP [ m] Length between perpendiculars
M [kg ] Mass matrix
M x, y , z [ Nm] Roll, pitch and yaw moment
n [ ] Normal vector
p [N / m ] 2 Pressure
P [m3 / m / s] Depth integrated flux in x-direction
Q [m3 / m / s] Depth integrated flux in y-direction
S [m2 ] Wetted surface of body
t [ s] Time
X [ m] Body motion
u [m / s] Fluid velocity component in x-direction
v [m / s] Fluid velocity component in y-direction
W [m2 ] Wetted surface of fixed structures
w [m / s] Fluid velocity component in z-direction
x [ m] Longitudinal horizontal coordinate
y [ m] Lateral horizontal coordinate
z [ m] Vertical coordinate perpendicular to x-y plane

Final 12
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Greek symbols
Symbol Unit Designation
 [ m] Total displacement
 [ m] Wave surface elevation
 [ ] Waterline of floating body
 [kg / m3 ] Fluid density
 [m2 / s ] Total potential
I [m2 / s ] Incident wave potential
S [m2 / s ] Scattered wave potential
j [m2 / s ] Radiation potential in mode j
 [m2 / s ] Time-varying part of radiation potential
 [m2 / s ] Impulsive part of radiation potential
 [rad ] Rotational displacement

Definitions
COG: Centre of gravity
COB: Centre of buoyancy
COF: Centre of floatation
DWT: Dead weight tonnage

Heave: Translation along z-axis

Surge: Translation along x-axis
Sway: Translation along y-axis
Yaw: Rotation around z-axis
Roll: Rotation around x-axis
Pitch: Rotation around y-axis

Final 13
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 14
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

List of figures
Figure 1-1: Over view of berths and location of the Port of Leixões. ................................................. 18
Figure 2-1: Evolution of the Port of Leixões. Source: Rosa-Santos et al. (2008). .............................. 20
Figure 2-2: Cross section of breakwater at berth “A”. Source: Rosa-Santos et al. (2008). ................. 20
Figure 2-3: Over view of berth “A”. ................................................................................................ 21
Figure 2-4: Asymmetrical mooring lay-out. Source: Rosa-Santos et al. (2008). ................................ 21
Figure 2-5: Over view of the DOLPHIN pr oject. Source: Taveira-Pinto et al. (2008) (adapted). .......... 22
Figure 2-6: Physical model of the first phase. Source: Rosa-Santos et al. (2010) ............................. 23
Figure 2-7: Physical model of the second phase. Source: Rosa-Santos et al. (2010)......................... 23
Figure 2-8: Over view of physical model. Source: Rosa-Santos (2010). ............................................ 25
Figure 2-9: Efficiency of dynamic wave absorption control unit . ..................................................... 25
Figure 2-10: Visualisation of ship described by 676 panels. ............................................................ 26
Figure 2-11: Examples of load elongation curves. .......................................................................... 27
Figure 2-12: Measurements in physical model. .............................................................................. 28
Figure 2-13: Typical wave spectr um. ............................................................................................ 29
Figure 2-14: Bichromatic wave group and bound long wave. .......................................................... 29
Figure 2-15: Generation mechanisms for free infra-gravity waves ................................................... 30
Figure 2-16: Standing waves in closed basins. Source: Rabinovich (2009). ...................................... 30
Figure 2-17: Generation mechanisms for free infra-gravity waves due to discontinuity in bathymetry 31
Figure 2-18: Ship motions in standing waves. ................................................................................ 32
Figure 2-19: Typical mooring arrangement. ................................................................................... 33
Figure 3-1: Hydrodynamic processes in physical model basin. ........................................................ 34
Figure 3-2: Smoothing over wave spectra ..................................................................................... 35
Figure 3-3: Measured wave spectra. ............................................................................................. 36
Figure 3-4: Measured wave spectra. ............................................................................................. 37
Figure 3-5: Distinguished areas for calculation eigen periods. ......................................................... 38
Figure 3-6: Measured wave and ship motion spectra...................................................................... 39
Figure 4-1: Proposed coupling of numerical models to described moored ship behaviour within ports.
Source: Van der Molen 2006A (adapted). ...................................................................................... 40
Figure 4-2: Inclusion of partial boundar y conditions. ...................................................................... 43
Figure 4-3: Bound long wave height in TRITON. ............................................................................. 43
Figure 4-4: Visualization of ship motions in waves. Source: Journée and Massie (2001). .................. 45
Figure 5-1: Applied boundar y conditions in MIKE21 BW. ................................................................ 48
Figure 5-2: Simulated wave spectra with MIKE21 BW. ................................................................... 49
Figure 5-3: Calculated total wave forces ........................................................................................ 53
Figure 5-4: Calculated total wave moments ................................................................................... 53
Figure 5-5: Calculated first order wave forces ................................................................................ 54
Figure 5-6: Calculated first order wave moments ........................................................................... 54
Figure 5-7: Calculated second or der wave forces ........................................................................... 55
Figure 5-8: Calculated second or der wave moments ...................................................................... 55
Figure 5-9: Simulated ship motions ............................................................................................... 56
Figure 5-10: Snapshots and time-series in vicinity of the ship simulated with MIKE21 BW. ............... 58
Figure 5-11: Feedback to coupled numerical model scheme. .......................................................... 59
Figure 6-1: Applied boundar y conditions in TRIT ON. ...................................................................... 60
Figure 6-2: Applied boundar y conditions in TRIT ON for additional simulations. Run 12 to 14. ........... 62
Figure 7-1: Spectrum characteristics run 01 to r un 04. ................................................................... 67
Figure 7-2: Spectrum characteristics run 02, r un 05 and run 06. ..................................................... 68
Figure 7-3: Spectrum characteristics run 02, r un 07 to run 09. ....................................................... 69
Figure 7-4: Schematization tip of breakwater in TRITON. ............................................................... 70
Figure 7-5: Spectrum characteristics run 02, r un 10 and run 11. ..................................................... 71
Figure 7-6: Snapshots of TRIT ON during run 04 and run 011. ........................................................ 72
Figure 7-7: Spectrum characteristics runs 02, 12, 13 and 14. ......................................................... 73
Figure 7-8: Results of Run 12 to Run 14. ...................................................................................... 75

Final 15
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 16
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

List of tables
Table 2-1: Prototype s hip dimensions. .......................................................................................... 26
Table 2-2: Equivalent constant stiffness coefficients of the mooring lines and fenders. .................... 27
Table 2-3: Tested wave conditions. .............................................................................................. 28
Table 2-4: Mooring lines. ............................................................................................................. 33
Table 2-5: Maximum allowable motions amplitudes for a moored oil tanker. Source: PIANC (1995) .. 33
Table 3-1: Calculated eigen periods .............................................................................................. 38
Table 3-2: Calculated eigen frequencies ........................................................................................ 38
Table 5-1: Calculated total wave spectrum characteristics. ............................................................. 50
Table 5-2: Calculated primary wave spectrum characteristics. ........................................................ 50
Table 5-3: Calculated low frequency wave spectrum characteristics. ............................................... 50
Table 6-1: Total simulation duration physical model tests and TRIT ON. .......................................... 60
Table 6-2: Variation of partial reflecting boundaries. ...................................................................... 61
Table 6-3: Variation of primary peak wave period. ......................................................................... 61
Table 6-4: Variation of wave generation. ....................................................................................... 61
Table 6-5: Variation of other boundary conditions. ........................................................................ 62
Table 6-6: Overview of simulations with TRITON. .......................................................................... 63
Table 6-7: Reflection coefficients of a rubble mound breakwater for different wave conditions. ........ 63
Table 6-8: Reflection coefficients of a beach for different wave conditions ...................................... 64
Table 7-1: Qualitative j udgement of sensitivity TRIT ON. ................................................................ 76

Final 17
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

1 Introduction
The behaviour of moored ships in por ts is a topic which is investigated for several deca des. With
increased computational efficiency and coupling of several numerical models it is nowadays possible to
investigate moored ship behaviour in an alternative way than by performing physical model tests only.
Results of both numerical and physical model simulations can be used to gain more insight in system
behaviour and to give recommendations either about designing a port or improving a port.

1.1 Motive of study

Berth “A” in the Port of Leixões, see Figure 1-1, is affected by operational problems. As a
consequence berth “A” is not in operation for 20% of the time. In order to improve the conditions at
the ber th, the behaviour of the moored oil tanker should be described. The Faculty of Engineering
University of Por to (FEUP) was investigating the behaviour of the moored vessel by performing
physical model tests. The behaviour of the moored vessel can be investigated by numerical model
simulations. The results of both the numerical and physical models can be used to give
recommendations how to improve the operational conditions at the berth.

1.2 Problem definition

The Port of Leixões is located in the north of Portugal. The por t has an oil terminal, with three ber ths
(respectively “A”, “B” and “C”), see Figure 1-1. Ber th “A” is located close to the harbour entrance and
is protected by the Leixões North breakwater. Due to its location the berth is exposed to adverse
maritime conditions. The operational conditions at ber th “A” are affected by several factors resulting in
a down-time of the ber th of about 20%. The down- time of the berth leads to extra costs as well as
security risks for the por t authority (Veloso Gomes et al. 2005).
The 20% dow n-time of the ber th does not only include the days that the berth cannot be used due to
adverse maritime conditions, but also the days due to maintenance operations at the jetty or dredging
works in the neighbourhood of the ber th (Rosa-Santos et al. 2008).
Large motions of a moored oil tanker lead to a reduction of the (un)loading efficiency and in adverse
conditions the vessel is not longer allowed to be kept at the berth safely. A research and development
program (DOLPHIN) was initiated by FEUP and the Por t of Leixões to study the behaviour of moored
oil tankers and finally to improve the operational conditions at the berth. The berth “A” operational
conditions are supposed to be influenced by (see Veloso-Gomes et al. 2005):
 Overtopping of the Leixões North breakwater.
 Wave diffraction around the head of the Leixões Nor th breakwater.
 Possible resonance phenomena in the berth “A” area.
 Current transmission through the core of the breakwater.
 Characteristics of existing fenders and mooring system.

Figure 1-1: Overview of berths and location of the Port of Leixões.

Source: Rosa-Santos et al. (2010)

Final 18
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

1.3 Objective of study

The objective of this study is to simulate the moored ship motions at ber th “A” in the Port of Leixões
by applying a sequence of numerical models. The calibrated and validated numerical models provide
more insight into the behaviour of the moored tanker. Both physical and numerical model results may
be used to improve the mooring conditions at berth “A” in the Port of Leixões.

During the study the original objective was adjusted. The focus of this thesis was shifted towards the
understanding of the generated low frequency waves within the physical model. The emp hasis is on
simulating the low frequency waves that caused significant moored ship responses in the physical
model. T he shift was made since discrepancies in the numerical modelling were mainly concer ned
with the wave modelling.

1.4 Outline of the thesis

The thesis is divided in several chapters. In Chapter 1 a general introduction is made. Chapter 2
provides background information about the Port of Leixões, the DOLPHIN project, long waves,
harbour oscillations and mooring of oil tankers in general. In Chapter 3 an analysis is made of the
physical model and the relevant hydrodynamic pr ocesses. Chapter 4 contains the research
methodology and gives a shor t description of the applied numerical models. During the study several
numerical models are applied: MIKE21 BW in combination with Harberth and T RITON. In Chapter 5
the results after applying MIKE21 BW, Harberth and Quaysim will be discussed. Simulations carried
out with T RITON will be explained in Chapter 6. In Chapter 0 the results from the physical model tests
and the TRITON simulations will be compared. The model results obtained from both applied
Boussinesq-type wave models will be discussed in Chapter 8. Final conclusions and recommendations
will be given in Chapter 9.

Final 19
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2 Background information
2.1 Port of Leixões
The Port of Leixões is located in the North of Por tugal, see Figure 1-1, and is one of the most
important ports of the countr y. The por t handles the major types of commodities within maritime
transpor t. The port was established in 1890 when the first docks were completed. Port expansion took
place in the 1930‟s and 1970‟s. Since the late 1990‟s Panamax container vessel can be handled in the
port.
The Port of Leixões has an oil terminal which is un der concession by Petrogal Petróle os de Portugal.
The oil terminal is connected via a pipeline with the Petrogal refinery. The terminal is built nearby a
breakwater, with a crest level at +15 m C.D. and has a total length of 700 m. Along the breakwater
there are three ber ths, respectively berth “A”, “B” and “C” , see Figure 2-1. Berth “A” is located behind
a relatively new breakwater, while berth “B” and “C” are located behind a longer existing breakwater.
The Nor th breakwater protects berth “A” from direct wave action and is constr ucted over an old
submerged breakwater. Permeability conditions of the breakwater are not known accurately. From
observations it appeared that sediment is deposited near the inner breakwater toe, which was not
noticed at the head of the breakwater. The depositing of sediments is supported by ship pilots which
observed flows of sand coming from the breakwater at the stern of the ship, during adverse sea
conditions (see Veloso-Gomes et al. 2005). The breakwater was damaged and repaired several times
during the last decades. In order to protect the head of the breakwater a new submerged breakwater
was installed in fr ont of the older existing breakwater, see Figure 2-1.
Berth “A” can accommodate vessels up to 100,000 dwt carrying cr ude oil and other refined products.
Berth “B” can accommodate vessels up to 27,000 dwt that carry crude oil, refined products, liquefied
gasses and aromatic products. Ber th “C” can accommodate vessels up to 5,000 dw t carrying liquefied
petroleum gasses, refined products and aromatic products.

Figure 2-1: Evolution of the Port of Leixões. Source: Rosa-Santos et al. (2008).

Figure 2-2: Cross section of breakwater at berth “A”. Source: Rosa-Santos et al. (2008).

Final 20
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.1.1 Berth “A”

The jetty str ucture of berth “A” consists of two breasting dolphins (equipped with a pneumatic fender
and double mooring hooks) and a loading platform, see Figure 2- 2 and Figure 2-3. In Figure 2-4 the
most common mooring arrangement for the largest oil tankers that use the be rth is given. In this
arrangement eight mooring legs are used with double mooring lines. The largest oil tankers are
usually moored with steel mooring lines with a synthetic mooring tail (usually nylon).
A detailed analysis of the wave records shows that, in general, ber th “A” is operational when the deep
water wave height is less than 2.5 meter and for wave periods between 7 and 15 seconds. The ber th
is predominantly inoperative for wave height higher than 2.5 meter and for wave periods between 8
and 20 seconds (Veloso-Gomes et al. 2005).
Previous studies have concluded that the downtime of berth “A” was mainly associated with waves
coming from the West and Nor th-West directions. Terminal operators and the ship pilots have stated
that the most problematic sea states were the waves approaching from the West (almost
perpendicular to the North breakwater), since those waves can diffract around the hea d of the
breakwater more easily (Rosa-Santos et al. 2010). Waves from the South-West do not occur very
often and their significant wave height is usually small compared with waves originating from other
directions (Taveira Pinto et al. 2008).

Figure 2-3: Overview of berth “A”.

Figure 2-4: Asymmetrical mooring lay-out. Source: Rosa-Santos et al. (2008).

Final 21
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.2 Physical model tests

2.2.1 DOLPHIN project

The investigation on improving berth “A” has been done within a Research and Development program
called DOLPHIN. The main objective of the DOLPHIN project was to investigate the moored ship
behaviour at berth “A”. An adapted over view of the DOLPHIN program is given in Figure 2-5. Three
different methods are applied to investigate the behaviour of the moored ship:
 Prototype measurements.
 Physical modelling.
 Numerical modelling.

Before the DOLPHIN project star ted the following studies were carried out:
 Analysis of inoperativeness records of berth “A”.
 Physical model tests for the Leixões Nor th breakwater.
 Numerical modelling of wave propagation.
 Analysis of the bathymetry evolution near by berth “A” (alongside the North breakwater).

The analysis of down-time recor ds and the numerical modelling of wave propagation provided wave
conditions which are teste d within the physical model. Results of numerical wave propagation
simulations agreed with ship pilot reports and showed that during high tide, waves of 2 .5 meter
height may be expected in the surrounding area of berth “A” during cer tain (extreme) wave conditions.

In the region of Leixões har bour two ranges of long period waves can often occur: the first one with
periods ranging from 2 to 5 minutes and the second with periods ranging from 15 to 20 minutes.
Simulations with monochromatic long waves showed that the water mass in the neighbour hood of
berth “A” can experience resonance for waves with periods close to both ranges and a standing long
wave node in the region of berth “A” can occur for some of these wave periods (Veloso-Gomes et al.
2005 and Rosa-Santos et al. 2008). Numerical simulations of wave propagation including non-linear
wave interaction and the generation of both sub- and super-harmonics have not been carried out.

Figure 2-5: Overview of the DOLPHIN project. Source: Taveira-Pinto et al. (2008) (adapted).

Final 22
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.2.2 Physical model

Physical model tests have been carried out at the Faculty of Engineering of University Porto within a
test basin with outer dimensions of 28 by 12 meters. Before testing the ship was ballasted and
calibrated and the mooring line and fender characteristics were reproduced. The physical model tests
were carried out according to Froude criteria of similitude at a geometrical scale of 1: 100. The wave
maker consists of multi-elements and has a dynamic wave absorption system. During the first phase
of physical model tests moored ship behaviour was investigated in case of head waves in a simplified
reproduction of the berth arrangement, see Figure 2-6. During the first phase only the berthing
structure was reproduced accor ding to prototype characteristics, since ship behaviour was
investigated in the case of head waves only. Those waves are expected to reach berth “A” after
diffraction around the head of the breakwater. During the second phase a more detailed construction
of the surrounding areas of ber th “A” was represented, see Figure 2-7.

Figure 2-6: Physical model of the first phase. Source: Rosa-Santos et al. (2010)

Figure 2-7: Physical model of the second phase. Source: Rosa-Santos et al. (2010)

Final 23
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

An overview of the second phase of the physical model tests is given in Figure 2-7 and Figure 2-8.
Numerical computations will be carried out for the second phase of the study and compared with
physical model measurements. Some remarks should be made regarding the wave forcing by the
wave maker and the set-up of the physical model with respect to the inclusion of beaches and
breakwaters.

Beach and breakwaters

 The dissipation beach at the end of the physical model (see V in Figure 2-8) is adapted
compared to the first phase of the study to avoid wave reflections from the side walls of the
physical model.
 The beach (see V in Figure 2-8) is cur ved in long-shore direction, but has a uniform pr ofile in
cross-shore direction.
 Both the Nor th and South breakwater structure are represented as accurate as possible.
 A wave guide wall is placed halfway in front of the wave maker (see III in Figure 2-8). The
North breakwater (at VI in Figure 2-8) is made impermeable for the major tests. Wave
penetration through the North breakwater can therefore be neglected.
 A porous (absor bing) beach (see VII and detail Figure 2-8) was designed and installed at the
entrance to the inner harbour basin to reduce reflections. Reflection analysis of this porous
beach was not made.

Wave maker
 During the physical model tests waves were generated according to a standard JONSWAP
spectrum, combined with theoretical set-dow n compensation and dynamic wave absorption.
 The generated waves were long crested waves with out directional spreading.
 The wave maker provides a set-down compensation, in real time, in the form of a second-
order driven signal that is added to the primary waves to ensure that (unwanted) free long
spurious incoming waves are minimized.
 Dynamic wave absorption is used to absorb reflected waves. The efficiency of the absorption
is depending on the frequency, see Figure 2-9. Notice that the defined frequency axis in
Figure 2-9 is on model scale. At a geometrical scale of 1:100 the frequency should be divided
by 10 (according to Froude scaling) to obtain prototype values.
 The efficiency of the dynamic wave absorption for long waves with periods above 50 seconds
is low. Long waves reflected from the beach (V in Figure 2-8) are not absorbed by the wave
maker, but should instead be absorbed along other model boundaries.

Final 24
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

I III
VI
II VII

IV

II

I Wave maker
II Wave gauges
III Wave guide wall
VII
IV Breakwaterhead
V Absorbing beach
VI Ship
VII Porous (absorbing) beach
Figure 2-8: Overview of physical model. Source: Rosa-Santos (2010).

Dynamic Wave Absorption Control Unit

120%

100%

80%
Efficiency (%)

60%

40%

20%

0%
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Frequency (Hz)

Figure 2-9: Efficiency of dynamic wave absorption control unit .

Final 25
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.2.3 Tanker and mooring system characteristics

The ship that regularly moors at ber th “A” is a 105,000 dwt oil tanker. The ship used during physical
model tests was calibrated for maximal loaded conditions and has the dimensions according to Table
2-1, a visualization of the wetted ship hull (described by 676 panels) is given in Figure 2-10.

Designation Symbol Magnitude Unit

Displacement volume  122714 [tons]
Length over all LOA 245.05 [m]
Length between perpendiculars LBP 236.00 [m]
Breadth B 43.00 [m]
Draft d 14.10 [m]
Depth D 21.00 [m]
Transversal metacentric height GMT 5.83 [m]
Longitudinal metacentric height GML 314.14 [m]
Vertical position COG above keel KG 12.46 [m]
Longitudinal position COG in relation to the stern ZCOG,L 128.36 [m]
Transversal position COG in relation to longitudinal central axis ZCOG,T 0.00 [m]
Vertical position COB above keel ZCOB 7.35 [m]
Longitudinal position COB in relation to the stern ZCOB,L 128.41 [m]
Longitudinal position COF in relation to the stern ZCOF,L 119.94 [m]
Waterplane area AW 9368.80 [m2]
Transverse radius of gyration kxx 15.10 [m]
Longitudinal radius of gyration kyy 61.40 [m]
Table 2-1: Prototype ship dimensions.

Bow Stern

Side

3D Overview
Figure 2-10: Visualisation of ship described by 676 panels.

Final 26
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

During the calibration phase of the physical model tests the ship mooring lines as well as the fenders
were simulated. The load elongation curves of the ship moor ing lines were simulated using a
combination of precision springs and taking into account the stiffness of the corresponding force
transducer. The non-linear behaviour of the mooring lines (see Figure 2-11 left) as well as the fenders
(see Figure 2-11 right) was linearized for the model ship. The non-linear mooring lines were linearized
such that the energy absorption capacity of an equivalent linear mooring line was equal. An inelastic
kevlar string and a combination of two precision springs were used to reproduce each one of the eight
double mooring lines of the prototype. Precision coil springs were carefully selected to furnish the
appropriate elasticity to each mooring element (Rosa-Santos et al. 2008 and 2010).

An example of the calibration and linearization of the mooring lines and fenders is included in Figure
2-11, whereas the equivalent constant stiffness coefficients of the mooring lines and fenders are
included in Table 2-2.

Cabo de amarração 3 Defensa 1

1.4 2.4
Cabo de amarração - L=55 m Defensa 1
1.2 Cabo equivalente (igual energia) Defensa equivalente (igual energia)
2.0
Verificação do conjunto - TF3 + Molas C1+K2 Verificação do conjunto - TF9 + Molas E4

1.0
1.6
Força (N)

Força (N)

0.8
1.2
0.6

0.8
0.4

0.4
0.2

0.0 0.0
0 4 8 12 16 20 24 0 2 4 6 8 10 12 14 16 18 20

Alongamento (mm) Deformação (mm)

Load versus elongation mooring line 3 . Load versus elongation fender 1.

Figure 2-11: Examples of load elongation curves.

Mooring line & Approximate length Stiffness

Fender Prototype [m] Model [m] Prototype [kN/m] Model [N/mm]

ML 1 (ASY & SYM) 150 1,50 173,53 0,0169

ML 2 (ASY & SYM) 90 0,90 349,83 0,0341
ML 3 (ASY & SYM) 55 0,55 510,76 0,0498
ML 4 (ASY & SYM) 55 0,55 504,84 0,0493
ML 5 (ASY & SYM) 82 0,82 352,69 0,0344
ML 6 (ASY & SYM 82 0,82 351,99 0,0343
ML 7 (ASY & SYM) 90 0,90 349,42 0,0341
ML 8 (ASY) 120 1,20 317,49 0,0310
ML 8 (SYM) 167 1,67 168,61 0,0165
FD1 (ASY & SYM) -- -- 886,63 0,0865
FD2 (ASY & SYM) -- -- 877,40 0,0856
Table 2-2: Equivalent constant stiffness coefficients of the mooring lines and fenders.

Final 27
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.2.4 Physical model results

Wave conditions
A total number of 15 wave conditions were tested in the physical model, see Table 2-3. Waves were
generated perpendicular to the wave maker paddles without directional spreading using a standard
JONSWAP spectrum with 2nd order wave steering (see also section 2.2.2).

T p [sec] d=16 [m] d=18 [m] d=20 [m]

10 Ѵ Ѵ Ѵ
12 Ѵ Ѵ Ѵ
14 Ѵ Ѵ Ѵ
16 Ѵ Ѵ Ѵ
18 Ѵ Ѵ Ѵ
Table 2-3: Tested wave conditions.

Measurements
During the 2nd phase of the physical model tests the following aspects were measured:
 Surface elevations of waves by 7 resistance wave gauges, without presence of the ship for a
simulation duration of 2~3 hours (in prototype time).
 Ship motions, tracked by an infrared motion capture system ( Qualisys system).
 Mooring lines and fender forces by a total of 8+2 force transducers, see Figure 2-12.

8 6

1234
5 Moored ship measurements
I: Fo rce transducer mooring lines.
II: Fo rce transducer fend ers.
Locations of wave gauges. III: Infrared ship motion cap ture system .

Figure 2-12: Measurements in physical model.

Conclusions
A brief overview of the conclusions from the physical model tests will be presented in this thesis, for
more details one is fur ther referred to Rosa-Santos et al. (2008A, 2008B, 2009 and 2010).
From the physical model tests it was concluded that:
 Modification of the berthing lay-out by relocating a headline had a negligible influence on the
ship motions.
 Increasing the breast lines pretension effectively reduces the moored ship motions, especially
in combination with high friction fenders.
 The tidal level is an important factor to control the behaviour of the moored ship. The ships
added inertia as well as damping increases with decreasing water depth, but the increase of
low frequency wave energy with decreasing water depth leads to a worsening of moored ship
responses.

Relevant measurements from physical model tests can be found in Appendix B.

Final 28
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.3 Relevant physics

2.3.1 Wave forcing

Waves can be classified into several groups. For a moored ship the most relevant waves are short
waves (periods of 5~20 seconds) and infra-gravity waves (periods of 25~300 seconds), see section
2.3.3, section 2.3.4 and Figure 2-13. At the ocean free infra-gravity waves may be generated due to
e.g. seismic activity and storm surges. These waves are for this study not relevant. Relevant is the
forced bound long wave (also referred to as set-down wave) that propagates with the primar y waves
at the scale of wave groups, see Figure 2-14 and e.g. Longuet-Higgins and Stewart (1962).

Wave spectrum
1

0.9

0.8

0.7
Spectral density [m2/Hz]

0.6

0.5

0.4
II
0.3

0.2
I
0.1

0 Primary wave group

0.02 0.04 0.06 0.08 0.1 0.12 0.14
Frequency [Hz] Bound long wave

IG waves Primary waves

I: Eigen period ship e.g. roll
II: Eigen period ship + mooring e.g. surge
Figure 2-13: Typical wave spectrum. Figure 2-14: Bichromatic wave group and bound long wave.

Bound long waves release as free long waves after wave breaking of the primary wave s. Wave
breaking at a shoreline occurs in the surfzone, where the bound long wave reflects as a free wave
from the shoreline. In case the angle of incidence and the shore normal is small, the reflected free
wave may leak fr om the surfzone to deeper water (also referred to as leaky wa ves). In case the angle
of incidence and the shore normal is large the reflected waves may be trapped in the surfzone due to
depth refraction (also referred to as edge waves). T he total motion due to incoming and outgoing
long waves is also referred to as surfbeat, see Figure 2-15. Discontinuities in either bathymetr y or
geometr y may release free long waves as well, see e.g. Bowers (1977). T he amount of bound long
wave ener gy depends on the local wave field and water depth, whereas the amount of free long wave
energy depends on the surrounding shelf and nearby shores, see Herbers et al. (1994).

The main difference between a bound long wave and a free long wave is the propagation speed of
the waves and the shoaling ratio. The wave celerity of both bound long wave and free long waves are
given by Eq. 2-2 and Eq. 2-1 respectively. Bound long waves travel with the speed of wave groups,
whereas free long waves travel with the speed of individual waves according to the linear dispersion
relationship.

Final 29
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

 g
c free   tanh(kd ) Eq. 2-1
k k
  1 2kd  g
cbound    nc free  1   tanh(kd ) Eq. 2-2
k k 2  sinh(2kd )  k

In which:
g = Gravitational acceleration [ms-2]
d = Water depth [m]
k = Wave number [m-1]
 = Wave frequency [s-1]
c = Wave celerity [ms-1]
n = Ratio between group celerity and higher phase celerity [-]

Another impor tant difference between bound long waves and free long waves is the shoaling ratio
when long waves enter shallower areas.
Free long waves shoal accor ding to Green‟s law:   d 0.25 .
Bound long waves shoal, depending on the slope, between d 0.25    d 2.5 .
In which:
 = Wave amplitude [m]

The existence of infra-gravity waves is suppor ted by theories, field observations and experiments. For
more detail one is referred to (amongst others): Baldock et al. (2000 and 2002), Battjes et al. (2004),
Gallagher (1971), Herbers et al. (1994,1995A and 1995B), Huntley et al. (1981), Longuet-Higgins and
Stewart (1962), Munk (1949), Sand (1982), Schäffer (1993), Symonds et al. (1982) and Tucker
(1950).

Figure 2-15: Generation mechanisms for free infra-gravity waves Figure 2-16: Standing waves in closed
at the shoreline. Source: Rabinovich (2009). basins. Source: Rabinovich (2009).

Final 30
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.3.2 Harbour oscillations

Although the amplitudes of long waves are of the order of centimetres they are important in har bours.
The amplitude of the infragravity waves increases with decreasing water depth in har bours and may
cause harbour oscillations. Because the frequency of infragravity waves is close to the eigen periods
of ship and mooring system, large horizontal moored ship motion responses are expected. Bowers
(1977), showed that differences in wave height in and outside a harbour e ntrance causes an
imbalance in bound long waves. The imbalance in bound long waves causes the generation of free
long waves. Bound long wave ener gy is thus par tly released as free long waves, which may amplify in
case of basin resonance. A visualisation of the generation of free long waves due to a discontinuity in
the bathymetry is given in Figure 2-17.
Alternative analytical solutions of harbour resonance for simplified configurations are made by Mei and
Agnon (1989) and Wu and Liu (1990). When harbour configurations become complex, distinct
numerical or physical models may be used to investigate harbour resonance. Many numerical models
are available to simulate harbour oscillations. Woo and Liu (2004) showed that Boussinesq-type wave
models are able to describe the set-down beneath wave groups and can be used to investigate
harbour resonance under influence of ocean waves.

When the length of the wave equals the basin length a standing wave may develop, see Figure 2-16.
These standing waves can be recognized as local maxima in the infra-gravity wave spectrum at the
eigen frequencies of the basin. This local maximum is more pronounced when measured in an anti-
node of the standing wave. For rectangular basins with an uniform depth the eigen period of a closed
basin can be approximated with the well known Merian‟s formula (Eq. 2-4), whereas Eq. 2-3 should be
applied for open basins (in that case the 0-th mode is the well known Helmholtz mode). For
approximations of eigen periods for other harbour configurations one is referred to Rabinovich (2009).

4L
Open basins: Tn  n  0,1, 2,3... Eq. 2-3
(2n  1) gd
2L
Closed basins: Tn  n  1, 2,3... Eq. 2-4
n gd
In which:
Tn = Eigen period of oscillation [s]
L = Basin length [m]
n = Mode number [-]

I: Incoming bound long wave

II: Shoaled bound long wave
III: Free long wave (out of phase)
IV: Reflected free long wave

Figure 2-17: Generation mechanisms for free infra-gravity waves due to discontinuity in bathymetry

Final 31
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.3.3 Ship motions in infragravity waves

Although the amplitude of long waves in harbours are of the order of centimeters they may cause
large moored ship responses. Damping in the horizontal plane of a moored ship at the eigen
frequency of both ship and mooring system is low. Infra-gravity waves have frequencies close to the
eigen periods of ship and mooring system in the horizontal plane. Thes e infra-gravity waves may
cause harbour oscillations resulting in standing wave patter ns. The resulting ship motions in the
horizontal plane are mainly depending on whether the moored ship is placed in a node or an anti-
node of the standing wave, see Figure 2-18.

Figure 2-18: Ship motions in standing waves.

Ship placed in a node:

 The orbital velocity of the standing wave is horizontally directed, whereas the sur face
elevation of the standing wave is zero.
 As a result of the horizontal velocities drag forces are acting on the ship. The ship responds
by a surge motion.
 Due to the gradient in the surface elevation the ship will make a pitch movement. The
pitching motion may result in surge motions of the ship.

Ship is placed in an anti-node:

 The orbital velocity of the standing wave is ver tically directed, whereas the surface elevation
of the standing wave is maximal.
 The wave forces on the ship are ver tically direc ted.
 The ship responds by a heave motion.

Many harbour configurations are more complex than the illustrated example in Figure 2-18. In order
to describe moored ship behaviour in a harbour configuration it is important to take standing waves
into account.

Final 32
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2.3.4 M ooring arrangement of an oil tanker

A moored oil tanker is subjected to several environmental forces like: wind, currents, tide, seiches,
swell, waves, changes in draft or trim, sur ges from passing ships and i ce. The motions of a moored oil
tanker are characterized by low frequency (periods ~ 1 to 2 minutes) horizontal motions: surge, sway
and yaw and high frequency (periods ~ 5 to 20 seconds) vertical motions: heave, roll and pitch. The
efficiency and security during ber th operations in case of a moored oil tanker are mainly depending on
the ship motions in the horizontal plane (surge, sway and yaw). T hese motions can be effectively
restrained by the mooring arrangement. Motions in the ver tical plane of the oil tanker are less
dependent on the mooring arrangement (in case of vertical motions the restoring forces are due to
buoyancy). A typical mooring arrangement for oil tankers consists of mooring dolphins (for stern- and
headlines) and breasting dolphins (for the breast and spring lines). A typical mooring arrangement is
given in Figure 2- 19 whereas the governing line names are given in Table 2-4.

Line Colour Name

ML 1 Green Stern
ML 2 Green Stern
ML 3 Purple Breast
ML 4 Orange Spring
ML 5 Orange Spring
ML 6 Purple Breast
ML 7 Red Head
ML 8 Red Head
Figure 2-19: Typical mooring arrangement. Table 2-4: Mooring lines.

General recommendations for effective mooring of oil tankers are given by OCIMF (2008). Based on
the recommendations given by OCIMF an asymmetrical as well as a symmetrical mooring
arrangement is investigated during physical model tests. The general recommendations for safe and
efficient mooring of oil tankers will not be given in this thesis, one is referred to the given reference.

In order to guarantee safe wor king conditions when a ship is moored, criteria for the ship motions are
set. These criteria are set to avoid or minimize accidents with line breaking. When the criteria are
exceeded the cargo handling operations should be slowed down or stopped. In extreme conditions the
ship should leave the berth. The most common applie d criteria for oil tankers are given by PIANC
(1995) and can be found in Table 2-5.

Degree of freedom: Heave [m] Surge [m] Sway [m] Roll [0 ] Yaw [0 ]
1.5 2.5 2 4 2

Table 2-5: Maximum allowable motions amplitudes for a moored oil tanker. Source: PIANC (199 5)

Final 33
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

3 Physical model analysis

3.1 Hydrodynamic processes
The main hydrodynamic processes w hich occur in the physical model are:
 Diffraction
 Refraction
 Shoaling
 Reflection
 Non-linear wave interaction

Diffraction of waves takes place at the Leixões North breakwater. Refraction causes waves to bend
towards shallower depth contours (e.g. beaches and breakwaters). Shoaling of waves will mainly
occur at the beach. Reflected waves are expected from the model boundaries e.g. breakwaters,
beaches and side walls. For an explanation of theses basics wave processes one is referred to
Bosboom and Stive (2010) and Holthuijsen (2007).
Non-linear wave interaction will occur in the entire model basin creating both sub- and super-
harmonics. The generated bound long waves from the wave maker may release as free waves from
beaches and breakwaters. Release of free waves may also be caused by discontinuties in the
bathymetry, w hich is particulary relevant for waves that diffract around the head of the breakwater.

I: Diffraction
II: Refraction
III: Shoaling
IV: Reflection

Figure 3-1: Hydrodynamic processes in physical model basin.

Final 34
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

3.2 Analysis of measurements

After obtaining a time-serie of wave sur face elevations a wave spectrum can be obtained by means of
a Fast Fourier Transformation (FFT) analysis, see e.g. Holthuijsen (2007). T he FFT is used to obtain
the amplitude, phase and frequency of wave components in the timeserie. The spectral density per
frequency obtained after FFT is divided over a number of frequency bins. Some smoothing to present
the wave spectra is often used. As a result of smoothing some detail may be lost. In this thesis the
presented wave spectra are based on a smoothing technique, for which the spectral shape and total
wave height is retained, see Van der Molen (2010). By selecting a constant smoothing factor for all
time-series the spectra can be compared directly, see Figure 3-2.

A high level of detail in the low frequency spectrum is useful to recognize possible standing waves
and/or spurious basin resonance. These standing waves and/or basin resonance may affect the ship
motions, see section 2.3. Analysis of the wave spectra for all wave conditions at wave probe 6 and 8
(vicinity of the ship), showed a local peak at a frequency of 0.006Hz, see Figure 3-3 and Figure 3-4.

In the case of small smoothing the spectr um looks “grassy” and the error (defined as the difference
between the expected value and computed value) is relatively large. In order to obtain a more reliable
wave spectrum the spectral resolution decreases. A compromise should be found between spectral
resolution and accepta ble reliability. One method to obtain a compromise between spectral resolution
and acceptable reliability is quasi-ensemble averaging, see also Holthuijsen (2007). Quantifying the
reliability of the wave spectra in terms of confidence inter vals is not made during this study.

For illustration of the errors made in spectral density the following example may be used:
A wave record with a measured duration of 2.5hour has a frequency resolution of ∆f=0.000111Hz.
Averaging the spectral density over larger frequency bins for example δf=0.00125Hz gives an absolute
error in spectral density of:
100%
 8.9%
0.00125 / 0.000111

Smoothing over wave spectra

2.5
No smoothing
Little smoothing
Moderate smoothing
2
Much smoothing

1.5
S [m2/Hz]

0.5

0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
f [Hz]
Figure 3-2: Smoothing over wave spectra

Final 35
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1
5

4
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025

WP 5
5

4
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025
WP 6
5

4 II n= 1 I n=1
d=20m II n= 2
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025

WP 8
5

4
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025
f [Hz]

Figure 3-3: Measured wave spectra. Measured d=16 [m]

With constant smoothing.
Tidal level: d= 16, 18 and 20 [m]. Measured d=18 [m]
Peak period: T p=14 [sec]. Measured d=20 [m]

Final 36
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1
10

8
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025

WP 5
10

8
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025
WP 6
10
II n= 1 I n=1
8 d=20m II n= 2
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025

WP 8
10

8
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025
f [Hz]
Figure 3-4: Measured wave spectra.
Measured d=16 [m]
With constant smoothing.
Tidal level: d = 16, 18 and 20 [m]. Measured d=18 [m]
Peak period: T p=18 [sec]. Measured d=20 [m]

Final 37
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Eigen periods
Within the physical model spurious standing waves can be generated. Basin resonance is expected
when the basin length corresponds to a multiple of the wavelength, see Figure 2-16.
Three possible areas for standing waves are distinguished:
 Area I: Between wave guide wall and side wall = 600 meter.
 Area II: Between side wall and side wall = 1200 meter.
 Area III: From wave maker to absorbent beach ≈ 2000 meter.

The eigen periods for harbour oscillations can be approximated by applying Eq. 2 -4 for the
distinguished areas, see Figure 3-5. The eigen periods are presented in Table 3-1, whereas the
corresponding eigen frequencies are presented in Table 3-2.

Figure 3-5: Distinguished areas for calculation eigen periods.

n I (d=20m) II (d=20m) III (d=20m) n I (d=20m) II (d=20m) III (d=20m)

1 85.7 [sec] 171.3 [sec] 285.6 [sec] 1 0.0117 [Hz] 0.0058 [Hz] 0.0035 [Hz]
2 42.8 [sec] 85.7 [sec] 142.8 [sec] 2 0.0234 [Hz] 0.0117 [Hz] 0.0070 [Hz]
3 28.6 [sec] 57.1 [sec] 95.2 [sec] 3 0.0350 [Hz] 0.0175 [Hz] 0.0105 [Hz]
Table 3-1: Calculated eigen periods Table 3-2: Calculated eigen frequencies

The peak at 0.006Hz is noticed at wave gauge 1 to 4, but not pronounced. Possibly a standing wave is
formed between both side walls of the physical model, since the calculated eigen frequency is close to
the frequency of the found peak (see, Figure 3-3 and Figure 3-4). This standing wave cannot exist at
wave gauge 1 to 4, since the basin length at those gauges is different. At wave probe 1 a peak is
noticed at a frequency of 0.0117 Hz, w hich is possibly a standing wave between the wave guide wall
and the side wall of the basin.
Assuming that a transverse standing wave is measured in the basin this may explain part of the
enhancement of measured wave energy at wave probe 5. At wave probe 6 and 8 the measured low
frequency energy is less compared to wave probe 5, this is probably because these probes are placed
closed to a node. Wave gauge 5 is placed in an anti-node at w hich the amplitude of the standing
wave and hence the spectral density is higher. This presumption is endorsed by the measured ship
motions where both surge and pitch motions do show similar trends, see Figure 3-6. Local peaks at
the same frequencies can be found in the spectra of heave, pitch, spring lines, stern lines and head
lines, see Appendix B.
Other resonant mechanisms in the physical model may well exist. An alter native resonant mechanism
with resemblance in prototype is given in Appendix E. From this study and/or report no conclusion can
be withdrawn which resonant mechanism was dominant to find the corresponding measured surge
responses of the ship.

3.3 Summary of physical model analysis

 From an analysis of the available measured data it is obser ved tha t there is an enhancement
of low frequency energy in the vicinity of the ship at frequencies around 0.006Hz.
 This frequency is close to an eigen period of the physical model basin.
 Possibly a transverse standing wave is measured.
 The existence of a transverse standing wave may explain some resulting ship motions.

Final 38
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Wave spectra Probe 8 vs. Motion spectra Surge

1 60
Wave motion S [m2/Hz]

Ship motion S [m2/Hz]

WP8
0.8 48
Surge
0.6 36

0.4 24

0.2 12

0 0
0 0.005 0.01 0.015 0.02 0.025

Wave spectra Probe 5 vs. Motion spectra Surge

5 60
I n=1
Wave motion S [m2/Hz]

Ship motion S [m2/Hz]

II n= 1 WP5
4 d=20m II n= 2 48
Surge
3 36

2 24

1 12

0 0
0 0.005 0.01 0.015 0.02 0.025

Wave spectra Probe 8 vs. Motion spectra Sway

1 15
Wave motion S [m2/Hz]

Ship motion S [m2/Hz]

WP8
0.8 12
Sway
0.6 9

0.4 6

0.2 3

0 0
0 0.005 0.01 0.015 0.02 0.025

Wave spectra Probe 5 vs. Motion spectra Sway

5 15
Wave motion S [m2/Hz]

Ship motion S [m2/Hz]

WP5
4 12
Sway
3 9

2 6

1 3

0 0
0 0.005 0.01 0.015 0.02 0.025
frequency [Hz]
Figure 3-6: Measured wave and ship motion spectra.
With constant smoothing.
Tidal level: d = 20 [m].
Peak period: T p=14 [sec].
Two different vertical axis.

Final 39
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

4 Research methodology
4.1 Methodology
When a ship is moored at an open jetty in deep water at a non-sloping seabed it is possible to
describe the motions and mooring forces with a ship simulation model and a deep water wave
spectrum only. In case of complex harbour lay-outs with a shoreline close by, coastal processes and
reflections from port structures need to be included. Based on the wor k of Van der Molen (2006B) two
different numerical approaches can be used to describe moored ship behaviour within ports. The
approach which combines a Boussinesq-type wave model with a panel model was selected as the
appropriate approach for the present study, since:
 Diffraction of waves is impor tant.
 Partial reflection of waves from beaches and especially port structures needs to be included.
 Non-linear wave processes need to be included.

By coupling a Boussinesq- type wave model, a panel model and a ship simulation model moored ship
behaviour in a port can be simulated. The proposed flow diagram for moored ship behaviour in ports
is given in Figure 4-1. Summarized, the three successive models are used to:
 The Boussinesq-type wave model takes into account the propagation of ocean waves into the
harbour.
 The panel model takes into account the presence of the ship in the wave field at the ber th
and subsequently calculates the wave forces on the ship.
 The ship simulation model calculates ship motions and mooring forces, taking into account
environmental forces and interactions with the mooring system.

Figure 4-1: Proposed coupling of numerical models to described moored ship behaviour within ports. Source: Van
der Molen 2006A (adapted).

Final 40
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

4.2 Description of the Boussinesq-type wave models

The numerical models used to calculate the propagation of ocean waves into the harbour are the
Boussinesq-type wave models MIKE21 BW and TRITON. In this section general information about the
models is given, for more detail and formulations one is referred to DHI (2005, 2006A and 2006B) in
case of MIKE21 BW and Borsboom et al. (2000) + Deltares (2008) for TRITON.
Making a detailed comparison between both Boussinesq wave models is not the purpose of this thesis
and is therefore not done. A general comparison between both models is given in section 4.2.3.

General princi ples of the applied Boussinesq -type wave models

 Boussinesq-type wave models are phase resolving, they describe individual wave behaviour.
 Within the Boussinesq-type wave model the 3D flow is written in 2D equations.
 The Boussinesq-type wave models are capable of reproducing the most important wave
phenomena like: shoaling, refraction, diffraction, wave breaking, bottom dissipation, moving
shoreline (run-up and run-down), partial reflection, wave transmission, non-linear wave-wave
interactions, frequency spreading and directional spreading.

4.2.1 M IKE21 BW
MIKE21 BW is part of the package MIKE21 from DHI and is successfully applied in several wave
modelling propagation studies for coastal areas. MIKE21 BW solves the time-dependent vertically
integrated Boussinsesq equations of mass and momentum. T he Boussinesq-type equations are solved
using a flux-formulation with impr oved frequency dispersion characteristics (see Madsen et al. 1991,
1992, 1997A and 1997B).

MIKE21 BW is restricted to the following conditions:

 In case of enhanced equations: linear frequency dispersion up to   k p d  3.1 .
 In case of classical Boussinesq equations: linear frequency dispersion up to   k p d  1.4 .
In which:
 = measure of linear dispersion [-]
k = wave number belonging to the peak wave period [m-1]
p

Program structure
For correct modelling of waves in MIKE21 BW at least the following need to be defined:
 Bathymetr y: Up to a predefined minimum water depth.
 Sponge layers: For the absorption of shor t waves along model boundaries e.g. beaches. A
wider sponge layer should be selected if longer waves should be absorbed as well.
 Porosity layers: these layers are applied at breakwaters and walls to model wave transmission
and partial reflections. The porosity value is depending on the wave height and period, water
depth in front of the structure and the reflection coefficient of the str ucture.
 Wave generation: Waves are generated at an internal wave generation line. The wave
generation line serves as an inter nal source for the generation of waves. A sponge layer after
the wave genera tion line is usually applied. T his sponge layer absorbs the outgoing waves, as
well as the radiated waves from the internal wave generation line.

Final 41
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

4.2.2 TRITON
TRITON is a Boussinesq-type wave model from Deltares. TRITON is in development and for that
reason not yet commercia lly available. Although in development, TRITON is able to model wave
propagation up to the following restrictions:
 Linear frequency dispersion up to   kpd  4 .
 Non-linear effects up to   ad 1  0.25 .
  2  0.25 .
Non-linear effects up to
 Wave shoaling up to s  0.5 .
In which:
 = Measure for the non-linearity of waves [-]
a = Wave amplitude [m]
s
= {k p Lbottom, p }1 [-]

kp = Representative wave number [m-1]

Lbottom, p = Typical horizontal length scale over w hich bottom changes take place [m]

Program structure
TRITON uses three modules:
 TriGrid: computational grid and model boundaries.
 TriBath: create depth file such that each computational grid point has a cer tain depth.
 TriGui: defining the boundary conditions and output locations.

Within TRITON the following boundary conditions can be used:

 Closed: in case of closed boundaries or full reflecting walls.
 Partial: in case of partial reflecting walls.
 Outflow: in case of zero reflecting walls.
 Monochromatic: for the generation of monochromatic waves.
 Spectrum: for defining a wave spectr um from which a time-serie is made.
 Time-serie: for defining a time-serie.

4.2.3 Comparison M IKE21 BW and TRITON

Based on the model equations which are solved by the models MIKE21 BW and TRITON, it can be
concluded that both models have the same or der of accuracy and can therefore be applied for the
same range of applications. The main difference between MIKE21BW and TRITON is the inclusion and
treatment of the boundary conditions along the harbour structures and internal wave generation.

MIKE21 BW uses sponge and por osity layers to include physical processes as wave energy dissipation
and partial reflection transmission, see Madsen (1983). Wave generation is done by an inter nal wave
generation line, see Schäffer and Sørenson (2006). TRIT ON does not use sponge or porosity layers,
nor a internal wave generation line backed up by a sponge layer. The model is equipped with a
absorbing boundary procedure, which adapts based on the local wave field, see Borsboom et al. (2000
and 2001). Properties of waves, reflected off str uctures, are based on a typical celerity and directi on
of the outgoing wave. Within TRITON mass and momentum are strictly conserved. Although several
model approximations are made, these physical properties are retained.

The main difference between MIKE21 BW and T RITON is in the applied numerical techniques to solve
the boundar y conditions, on w hich an illustration is given in Figure 4-2. Besides the differences in
techniques used at the model boundaries, different numerical schemes are used to solve the
equations. The (dis)advantages of these schemes for e.g. the robustness of the numerical models are
not investigated.

It should be remar ked that there is no standar d option to include a theoretical bound long wave in the
wave generation in MIKE21 BW. A theoretical bound long wave can however be imposed in the
generated waves within TRITON.

Final 42
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Physical model
Bathymetr y and reflection are determined as accurate
as possible by scaling. T he reflection coefficient of the
breakwater is determined based on physical model
tests.

MIKE21 BW
Bathymetr y defined to hmin. From hmin to the water line
the water depth is kept constant at hmin.
A porosity layer is defined which is build out from the
water line. The defined porosity value is depending on
H, T and Cr. The width of the layer is determined
based on the wave length.

TRITON
Bathymetr y is defined for the full domain, but the user
may define a hmin and apply a par tial model boundary
condition. The partial boundary condition is selected on
Cr only. T he boundary condition procedure calculates
the reflected wave.

Figure 4-2: Inclusion of partial boundary conditions.

Remarks
 In both Boussinesq-type wave models it is not possible to include par tial reflection coefficients
as a function of the wave frequency. Partial reflection should be modelled correctly for waves
with a period corresponding to the peak period of the waves.
 Recent research by De Jong et al. (2009) showed that, in par ticular situations, bound long
wave energy in TRITON is underestimated compared to theoretical values, see Figure 4-3.
Recently improvements are made, see De Jong et al. (2011), but these were not included in
these simulations yet. For long primary waves periods and relatively shallow water conditions,
the bound long waves are however described with sufficient accuracy. The range for which
the bound long waves are sufficiently well described was defined as kd  0.5  1.0 . The
criterion to accurately describe the bound long waves is thus more strict than the criterion for
the primar y waves. For the tested wave conditions in the physical model the bound long
waves are, according to this criterion, described with sufficient accuracy.
 Criteria for accurately describing bound long waves in MIKE21 BW are unknow n. In Madsen
and Sørenson (1993) a comparison between non-linear effects and theoretical values,
according to Stokes theories are given, but only for super-harmonics. It is stated that the
accuracy of modelling super-harmonics is not depending on selecting the enhanced equations
(i.e. irrespective of the B value in case of the enhanced equations). In both Madsen and
Sørenson (1993) and Sørenson et al. (2004) it is stated that the enhanced Boussinesq
equations tend to underestimate energy transfer to super-harmonics, but statements about
energy transfer to sub-harmonics are not given.

Figure 4-3:Bound long wave height in TRITON.

Comparison with theoretical values for different kd values. Source: De Jong et al. (2009)

Final 43
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

4.3 Description of the 3D panel model

The numerical model used to compute the forces on the ship in the irregular and inhomogeneous
wave field is the 3D time-domain panel model Harberth. In this section general information about the
model is given, for more detail and formulations one is referred to Appendix A and Van der Molen
(2006B, 2008 and 2009).

General princi ples 3D panel model Harberth

 The 3D panel model can be applied on arbitrarily shaped bodies, with a zero mean forward
speed.
 The angles of rotation of the ship are assumed to be small (<0.1 radians), such that
linearization and superposition of motions can be applied. Since the angles of rotation are
small the moored ship can be de fined as a linear mass-spring system.
 The resulting motion of the ship in waves may be seen as a superposition of the motion of the
body in still water and the forces on the restrained body in waves, see Figure 4-4.
 The oscillation amplitudes of the fluid and the body should be small relative to the cross -
sectional dimensions of the body.
 The fluid flow around the ship is assumed to be homogenous, incompressible and irrotational
such that it can be described by a velocity potential,  . The gradient of the velocity
  
potential is equal to the flow velocity ( u  ,v  and w  ).
x y z
 Due to the use of potential theor y, effects of flow separation are neglected.
 The potential flow around the body can be calculated based on the principle of Green‟s
second theorem. According to this theorem the pressures in the volume of water around the
ship are transformed in the pressures on the surface of the ship hull. T he applied boundary
conditions on the computational domain, as well as interaction with port str uctures in the
vicinity of the ship, are considered in the formulation of the Green functions, see Appendix A.
 Within Har berth waves are not considered as the sum of regular wave components, but as a
summation of impulsive sources.
 The hydromechanical coefficients (added mass and damping) are calculated in the frequency
domain to minimize numerical instabilities (contrar y to Van der Molen (2009)).
 The first or der free waves do fulfill the linear dispersion relationship in Har berth.

Program structure
Harberth considers the radiation problem (wave due to the moving body in initially still water) and
calculates the wave forces due to the incident and scattered wave on the restrained body, see Figure
4-4. The solution of the radiation problem consists of generated waves due to a moving ship. The
solution of the radiation problem provides the added mass and damping coefficients for each incident
wave direction and wave frequency and is solved in the frequency domain. T he matrices ser ve as an
input in the dynamical simulation with Quaysim. The used retardation functions in Harberth were
verified against measurements of Van Oortmerssen (1976) and the numerical frequency domain panel
model DELFRAC (Pinkster (1995)), see Van der Molen (2006B).
The incident waves to calculate the wave forces on the restrained body are determined with the
Boussinesq wave model. T he Boussinesq wave model cannot take into account the presence of a ship
in the incident wave field, therefore Harber th calculates the scattered and diffracted wave due to the
presence of the ship. The output from the Boussinesq-type wave model (MIKE21 BW) consists of the
two horizontal depth integrated fluxes (P and Q) and the water surface elevation  . Harber th
transforms these data to velocities and pressures at the ship hull using an inverse transformatio n
applied to the variables of the original Boussinesq-type wave model equations. The backward
transformation of the depth-averaged velocity and pressure is required to obtain the distributions over
the water depth. The hull is divided into a lar ge number panels, such that the pressure and velocity at
each panel is assumed to be constant. T he transformation provides therefore mean values of
pressures and velocities at each panel.

Final 44
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Integration of the pressures from the incident waves provides the so called Froude-Krylov force, which
is the wave force due to the incident wave. The Froude-Kr ylov force includes the force due to
nonlinearities in the incident wave field (also referred to as the second order force due to the second
order potential).

Integration of the pressures from the scattered waves provide the diffraction force and the second
order force. Other contributions to the second order force are products of first order quantities and
are obtained based on the calculated incident, scattered and radiated waves and the ship motions.
The total wave force is given by the summation of the Froude-Krylov, diffraction force and the second
order force.

Figure 4-4: Visualization of ship motions in waves. Source: Journée and Massie (2001).
Decomposition into the radiation problem (1) and wave exciting motions (2).

Input Harberth
The main input for the simulations in Har berth consists of the following components:
 Ship characteristics and description of the ship hull.
 Positioning of structure (quay wall or breakwater) relative to the ship-bound coordinate
system.
 Transmission coefficient  t , to include friction and viscous effects between ship and structure,
such that partial reflecting walls in the vicinity of the ship can be included. The transmission
coefficient  t , can be approximated as 1  Cr , in w hich Cr is the reflection coefficient as
defined in the porosity layer in the MIKE21 BW computations.
 Incident wave file (obtained from Boussinesq-type wave model simulation).

Output Harberth
The main output files of Harberth consists of the following components:
 A hydrodynamic file containing the hydrodynamical coefficients.
 A wave force file containing time series of wave forces.

Remark
The program description for Harber th as given within this thesis is based on a coupling with the
Boussinesq-type wave model MIKE21 BW. A coupling between TRITON and Har berth is available and
works according to the same principles as described for the coupling MIKE21 BW and Harber th.
Besides the coupling with Harber th, TRIT ON can also be coupled to DELMULT I, see Wenneker et al.
(2006). DELMULTI calculates time-series of wave forces on a ship and works according to the same
principles as Harberth. The diffraction forces in DELMULT I are determined by solving the diffraction
problem for all frequencies separately, w hereas the diffraction for ces in Harber th are calculated using
the Haskind relations, see Van der Molen (2008).

Final 45
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

4.4 Description of the ship simulation model

The numerical model to compute the ship motions as well as the mooring line and fender forces due
to exter nal forces acting on the ship is the ship simulation model Quaysim. Quaysim is a time domain
simulation program to analyze the dynamic behaviour of moored ships. In case of moored ship
behaviour non-linear restoring forces and wave drift forces fluctuate with respect to time. Linear
superposition cannot be applied in this analysis, contrary to frequency domain calculations, therefore
a time domain simulation should be applied. Quaysim is based on similar formulations as the ship
simulation models Shipmoorings, BAS and Termsim.

General princi ples ship simulation model Quaysim

 The angles of rotation of the ship are assumed to be small (<0.1 radians), such that
linearization and superposition of motions can be applied. Since the angles of rotations are
small the moored ship can be defined as a linear mass-spring system.
 The hydromechanical reaction forces and motions, due to time var ying ship motions, can be
described by using the formulations of Cummins (1962).
 The inertia matrix depends on the mass distribution of the ship. The matrices for added mass,
linearized viscous damping, hydrostatic restoring and impulse response functions are
determined by Har berth and serve as an input for Quaysim. These matrices are the linear
contributions of the hydr odynamic forces due to the moving body.
 The external force may consist of several components, bot h linear and non-linear. The
external force on the ship may be due to waves (including non-linear contributions), currents,
wind and interactions with the mooring system. The exter nal forcing due to waves acting on
the ship are calculated by Harber th and ser ve as an input for Quaysim.
 Viscosity effects due to the ship motions are included in the simulations.

The Cummins equation to describe time-var ying ship motions is given by:
6 

{( M kj  Akj ) X j (t )  Bkj X j (t )  Ckj X j (t )   K ( )kj X j (t   )d }  F k (t )

j 1
Eq. 4-1
0
In which:
M Iner tia matrix
A Added mass matrix
B Linearized viscous damping matrix
C Hydrostatic restoring matrix
K Impulse response functions matrix
X (t ) Body motion

F (t ) External force

j 1~ 6 Degree of freedom of the body

k 1~ 6 Coupled degree of freedom of the body

Input Quaysim
 A hydrodynamic file containing the hydrodynamical coefficients.
 A wave force file containing time series of wave forces.
 Ship characteristics.
 Description of mooring system, including mooring line and fender characteristics.

Output Quaysim
 Statistics of ship motions.
 Statistics of mooring line, fender and environmental forces.
 Time series of ship motions.
 Time series of mooring line, fender and environmental forces.

Final 46
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5 Numerical simulations with MIKE21 BW, Harberth and

Quaysim
5.1 Wave propagation simulations

5.1 .1 M odel set-up

General settings
The first step in MIKE21 BW is to decide whether or not the enhanced equations with improved linear
frequency dispersion characteristics should be solved. The spatial grid size and time step are more
strict when applying the enhanced equations compared to the classical equations and hence the
computational time is much larger. According to the included MIKE21 BW model set- up planner a
spatial cell grid size of Δx=4 meter is appropriate for the pur poses of wave modelling for this s tudy.
The model set-up planner is based on selecting a minimum period for the waves, w hich should be
modelled correctly to a certain defined minimal water depth. Using the linear dispersion relationship
for the selected minimum wave period in the minimum water depth gives the shortest wavelength,
which can be modelled correctly in MIKE21 BW. Based on a rule of thumb of 20 spatial cell grid points
per wave length and applying the linear dispersion relationship a practically smaller spatial grid size
would be found. Based on the following considerations a spatial cell grid spacing of Δx=4 meter was
selected as appropriate, since:
 Only waves with a small wave length at the defined minimal water depth are not modelled
accurately.
 The ship is located at a water depth w hich is much larger than the defined minimal water
depth.
 Waves with a small wave length are not the cause of excessive ship motions.
 The most energetic short waves and long waves are modelled correctly.
 The numerical phase error for the shortest waves is expected to be small.
A time step of Δt=0,1 seconds was selected to ensure numerical stability based on the CFL condition.
The selected time step is a conservative choice which resulted in a lower computational efficiency.

During all simulations both wave breaking and moving shoreline (run-up and run-dow n) were
excluded. Wave breaking and moving shoreline will increase the computational time and are for the
purposes of this study not important. Wave breaking at the beach was avoided due to an applied
sponge layer build out from the minimal water depth. The moving shoreline technique, as proposed
by Madsen et al. (1997B), may introduce some numerical instability problems due to the artificial slot-
technique. By excluding the moving shoreline, numerical instability due to this technique is avoi ded
and computational time decreased.

Boundary conditions
Three boundar y conditions applied in MIKE21 BW are distinguished:
 Sponge layers.
 Porosity layers.
 Wave generation .

Both porosity and sponge layers are created by the applied MIKE21 tools. A large sponge layer was
applied behind the internal wave generation line, w hich should absor b both the radiated short and
long waves from the internal wave generation line. A shorter sponge layer was applied at the beaches,
since these layers should only absorb the short waves and reflect the long waves. Porosity layers were
designed based on expected wave heights, water depths in fr ont of the structure and reflection
coefficients as obtained from physical model tests. The width of the porosity layer (in terms of grid
cells) is based on the wave length for the most energetic waves. Several porosity and sponge layers
were designed to calibrate and validate the model results. Waves were generated at a n internal wave
generation line for which a time-serie was designe d with the supplied MIKE21 tools. The applied
boundary conditions are visualized in Figure 5-1.

Final 47
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Sponge layer Porosity layer Internal wave generation line Closed boundary
Figure 5-1: Applied boundary conditions in MIKE21 BW.

5.1 .2 M odel results

During the first simulations the sensitivity of the model was investigated. These simulations were
short-term simulations with the purpose of defining the porosity and sponge layers such that a good
agreement was obtained between measured and simulated wave spectra.
From the first simulations the following was noticed:
 Shortage of primar y wave energy at especially wave gauge 6 and 8 (in the vicinity of the ship).
 Shortage of low frequency energy at all wave gauges.

Applying alternative sponge layers at the beach did not lead to improvement of results. Decreasing
the effectiveness of the sponge layer (in order to achieve more reflection from the beach), resulted in
an instable model blow-up. In case larger sponge layers are applied at the beach long waves may be
absorbed as well. The recommended s ponge layers for beaches, according to DHI (2006A), were
therefore applied. Applying alternative porosity layers (in terms of length and porosity value) did not
have any noticeable effects on the results. In or der to obtain more primary wave energy at wave
gauge 6 and 8 the head of the breakwater was decreased in size (red striped line in Figure 5-2). In
this way the diffraction losses are less and more primar y wave energy was simulated at wave gauge 6
and 8 in the vicinity of the ship. The simulated spectrum still gave an underestimation compared to
measurements. Since the low frequency waves are more important to obtain ship motions the focus
was shifted to overcome the shortage of wave energy at lower frequencies.

The inter nal generated waves in MIKE21 BW are based on a standard JONSWAP spectrum with a peak
period corresponding to the tested wave condition. During the first simulations a significant shor tage
of simulated low frequency wave energy was noticed at wave gauge 1 to 4. This shortage is due to
the fact that MIKE21 BW does not include the wave set-down compensation, which was used during
the performed physical model tests. T he wave set-dow n compensation simulates the so-called second
order boundar y conditions. The shortage of low frequency ener gy at the internal wave maker in
MIKE21 BW was compensated by an extra time-serie which was accompanied with the time-serie of
the primar y waves. The additional time-serie was based on the measured low frequency spectrum,
but contains free long wave energy (blue dashed striped line in Figure 5-2). The additional time-serie
may be in phase with the pr opagating wave groups (instead of out of phase as with a bound long
wave).

Final 48
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.4 2

0.3 1.5
S [m2/Hz]

0.2 1

0.1 0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.4 2

0.3 1.5
S [m2/Hz]

0.2 1

0.1 0.5

0 0
0.01 0 0.02 0.03 0.04 0
waveprobe 0.05
1 0.1
40 f [Hz] f [Hz]
Figure 5-2: Simulated wave spectra wit h MIKE21 BW. Measured
Tidal level: d=20 [m]. Original breakwaterhead
Peak period: T p=1435
[sec].
Low frequency wave spectra: left. Adapted breakwaterhead
Total wave spectra: right. Including low frequency calibration
30

25
S [m2/Hz]

20
Final 49

15
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The wave spectra as presented in Figure 5-2, are representative for all simulated wave spectra after
simulations with MIKE21 BW. A good agreement is achieved for the primary wave spectra at wave
gauge 1 and 5, w hereas the simulated primar y wave height at wave gauge 6 and 8 is clearly
underestimated. Some improvement in the simulations were obtained by adapting the head of the
breakwater. Improvements due to applying alternative sponge or porosity layers were not noticed
(results not presented in this thesis).
MIKE21 BW underestimates the measured wave energy at lower frequencies, if low frequency
calibration for the performed simulation is absent. T he peaks in the low frequency spectrum of wave
gauge 1 are probably due to the fact that both bound and free long waves are simulated with the
same frequency or as a consequence of calibration. A fairly good agreement in low frequency energy
between measurements and simulations is achieved at wave gauge 5, 6 and 8. At wave gauge 6 both
measured peaks in the low frequency spectrum are simulated. At wave probe 8 the first measured
peak is well simulated, but the second peak is not.

The spectrum characteristics are calculated for the simulation with an adapted head of the breakwater
and including low frequency calibration. Formulas to calculate these characteristics can be found in
Holthuijsen (2007) and will not be represented. A relative error is calculated according to Eq. 5-1. The
spectrum characteristics are calculated for:
 The total wave spectrum including all frequencies, see Table 5-1.
 The primary wave spectrum, only including frequencies >0.04Hz, see Table 5-2.
 The low frequency wave spectrum, only including frequencies <0.04Hz, see Table 5-3.

The simulated spectral periods (see Table 5-1 to Table 5-3) are higher than the measured spectral
periods, although this becomes not clear from the wave spectra as presented in Figure 5-2. A possible
cause is that during the physical model tests more wave ener gy is transferred into super -harmonics,
with higher frequencies. These super-harmonics are not accurately described within MIKE21 BW, since
the spatial grid size is large. Super-harmonics are less relevant for moored ship computations, since
their frequency is not close to the eigen period of ship and mooring system.
Although the wave spectra are not totally satisfactory, the simulations from MIKE21 BW were used for
further processing with Har berth and Quaysim. For further processing with these successive models
long periods of simulation, including a sufficient number of low frequency waves, are necessary to
achieve statistical reliable results. During these long period simulations (of approximately 1.5 hour in
prototype) an eddy was noticed in the computational domain of MIKE21 BW. The size and magnitude
in terms of velocity of this eddy grew in time. A more detailed explanation and the consequences of
this eddy will be explained in section 5.4.
Simulated - Measured
Relative Error  100% Eq. 5-1
Measured

W1 M W1 S Err % W5 M W5 S Err % W6 M W6 S Err % W8 M W8 S Err %

Hm0 [m] 3.09 2.92 -5.4 3.63 2.90 -20.2 0.78 0.51 -34.5 0.59 0.31 -47.8
T m01 [s] 11.39 12.71 +11.7 12.27 12.96 +5.4 10.69 15.13 +41.3 12.61 24.24 +91.9
T m02 [s] 10.50 12.19 +16.1 11.35 12.37 +8.8 9.26 13.28 +43.4 10.39 18.44 +77.3
Table 5-1: Calculated total wave spectrum characteristics.
Based on simulations with MIKE21 BW including low-frequency calibration. Cursive=Simulated.

W1 M W1 S Err % W5 M W5 S Err % W6 M W6 S Err % W8 M W8 S Err %

Hm0 [m] 3.05 2.87 -5.9 3.57 2.83 -20.7 0.74 0.46 -38.3 0.54 0.21 -60.6
T m01 [s] 11.19 12.36 +10.5 11.96 12.48 +4.4 9.84 12.51 +27.1 10.72 13.49 +25.8
T m02 [s] 10.38 11.99 +15.4 11.17 12.09 +8.2 8.82 11.93 +35.3 9.45 12.97 +37.2
Table 5-2: Calculated primary wave spectrum characteristics.
Based on simulations with MIKE21 BW including low -frequency calibration. Cursive=Simulated.

W1 M W1 S Err % W5 M W5 S Err % W6 M W6 S Err % W8 M W8 S Err %

Hm0 [m] 0.47 0.54 +16.1 0.66 0.63 -6.9 0.24 0.23 -5.8 0.25 0.22 -9.7
T m01 [s] 47.24 58.62 +23.8 55.48 55.49 -6.2 62.30 97.55 +53.2 71.95 84.37 +16.2
T m02 [s] 42.18 48.74 +15.4 49.90 47.93 -4.0 52.89 68.80 +30.8 61.79 62.68 +1.0
Table 5-3: Calculated low frequency wave spectrum characteristics.
Based on simulations with MIKE21 BW including low -frequency calibration. Cursive=Simulated.

Final 50
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5.2 Wave force calculations

5.2 .1 M odel set-up

The ship is described by 676 panels (see Figure 2-10) and has the characteristics according to Table
2-1. The reflection coefficient for the scattered waves from the breakwater was defined as 40%. The
retardation functions were obtained from a run with Quaysim in still water taking the non-linear
characteristics of the mooring arrangement into account.

5.2 .2 M odel results

The calculated wave forces by Harberth are presented as time-series of wave forces in Figure 5-3 to
Figure 5- 8. The presented time-series are representative for all other simulations with Harberth based
on MIKE21 BW computations, w hich showed similar trends. The calculation of the wave forces on the
ship showed unexpected results:
 In the calculated total wave forces and moments an upward trend is noticed, see Figure 5-3
and Figure 5-4.
 This upward trend is caused by increasing second order effects, see Figure 5-7 and Figure 5-8.
 The second order forces consists of four contributions. Only the contribution due to the
second order pressure increased significantly, see Figure 5-7 and Figure 5-8.
 The second order forces and moments are finally dominant over the first order wave forces
and moments, see Figure 5-5 to Figure 5-8.

  
2
The 2nd order wave force due to second order pressure is depending on the velocity potential, ,
see Eq. A-11 and section 4.3. In case of only waves the second order pressure wave force should vary
around an equilibrium value. If an increasing current is acting on the body the second error pressure
increases significantly in time. If an increasing current is acting on the body the second order wave
forces due to the relative wave height and the first order rotations and first order iner tia forces
(respectively the third and fourth contribution of the second order wave forces) will not increase, since
these contributions are not depending on the flow potential. The second order wave contribution due
to first order pressure and body motions (second contribution of the second or der wave forces) is
depending on the time derivative of the flow potential and therefore less sensitive to increasing
currents. If a constant current is acting on the body, the time derivative of the flow potential is zero
and hence the second order pressure force constant.

The increasing current acting on the ship is caused by the eddy created in MIKE21 BW. A small
numerical instability in the flux field caused the eddy, w hich in time grew in both size as well as
velocity magnitude. Due to the eddy the fluxes in the vicinity of the ship did not behave according to
an orbital wave motion, but instead as an orbital wave + current motion. It should be noticed that
besides the fluxes grew continuously in time, a spatial gradient in the flux field in the area of interest
is noticed as well. The spatial gradient in the flux field also contributes to excessive wave forces on
the ship due to disturbances in the calculated potential flow, this will be explained in section 5.4

As a consequence of disturbances in the simulated flux field the calculated wave forces on the ship
are not reliable. Within the 3D panel model Harber th the simulated fluxes from MIKE21 BW are
converted into orbital velocities of the waves, which serve as a boundary condition to calculate the
diffraction force due to the presence of the ship. Due to the increasing current in the flux field the 1st
general principle of Harberth no longer holds (see section 4.3). The 1st principle is based on the
assumption that a still ship in a current is equal to a ship sailing in still water.

The 1st general principle yields (see section 4.3):

 The 3D panel model Harberth can be applied on ar bitrarily shaped bodies, with a zero mean
forward speed.
Or equivalently:
 The 3D panel model Harber th can be applied on arbitrarily shaped bodies in waves only.

Final 51
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The calculated first or der wave forces, due to both incident and scattered waves, follow from
integration of the first order pressures over the mean submerged hul l. The first order pressure p is
obtained after applying the Ber noulli equation (see Appendix A):

p  p0  1
  gz      constant Eq. 5-2
 t 2

The first order pressure contains the non-linearity of the waves, but is calculated by assuming
potential flow (second and third term RHS Eq. 5-1). From the calculation of the second order forces it
is noticed that the second order pressure increased continuously. The continuous increase in second
order wave forces is due to spatial and temporal disturbances in the horizontal fluxes, as simulated
with MIKE21 BW. Due to temporal and spatial distur bances a n increasing current is acting on the ship,
  
2
which is most pronounced for terms depending on .

As analysed above the second order forces are not reliable due to the increasing current on the ship.
In the calculation of the first order wave forces, potential flow is used as well (third term RHS Eq. 5-2).
The quadratic pressure related to the second order pressure (thir d term RHS Eq. 5-2) in the
calculation of the first order wave forces on the ship is now dominated by other terms (e.g.
hydr ostatic pressure). The increase in first or der waves is therefore less pronounced compared to the
increase in the second order wave forces. For longer period of calculations an increase in the first
order wave forces is expected as well. Due to both spatial and temporal disturbances in the flux field,
both calculated first or der as well as second order wave forces on the ship are not reliable .

Final 52
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5000

Fx [kN]
0

-5000
0 1000 2000 3000 4000 5000 6000 7000
4
x 10
1
Fy [kN]

0
Ftot
-1
0 1000 2000 3000 4000 5000 6000 7000
4
x 10
2
Fz [kN]

-2
0 1000 2000 3000 4000 5000 6000 7000
t [s]
Figure 5-3: Calculated total wave forces

4
x 10
5
Mx [kNm]

-5
0 1000 2000 3000 4000 5000 6000 7000
6
x 10
1
My [kNm]

0
Mtot
-1
0 1000 2000 3000 4000 5000 6000 7000
5
x 10
5
Mz [kNm]

-5
0 1000 2000 3000 4000 5000 6000 7000
t [s]
Figure 5-4: Calculated total wave moments

Final 53
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2000

F(1)
x
[kN] 0

-2000
0 1000 2000 3000 4000 5000 6000 7000

5000
F(1)
tot
[kN]

0 FFK
F(1)
y

Fd
-5000
0 1000 2000 3000 4000 5000 6000 7000
4
x 10
2
[kN]

0
F(1)
z

-2
0 1000 2000 3000 4000 5000 6000 7000
t [s]
Ftot(1) =Total first order wave force FFK =Froude Krylov force Fd =Diffraction force
Figure 5-5: Calculated first order wave forces

4
x 10
5
[kNm]

0
M(1)
x

-5
0 1000 2000 3000 4000 5000 6000 7000
6
x 10
1
M(1)
[kNm]

tot
0 MFK
M(1)
y

Md
-1
0 1000 2000 3000 4000 5000 6000 7000
5
x 10
2
[kNm]

0
M(1)
z

-2
0 1000 2000 3000 4000 5000 6000 7000
t [s]
(1)
M tot =Total first order wave moment M FK =Froude Krylov moment M d =Diffraction moment
Figure 5-6: Calculated first order wave moments

Final 54
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

2000
F(2)
tot
[kN] 1000
F(2)
I
F(2)

0
x

F(2)
II
-1000
0 1000 2000 3000 4000 5000 6000 F(2)
7000
III
5000 F(2)
IV
[kN]

0
F(2)
y

-5000
0 1000 2000 3000 4000 5000 6000 7000
2000
[kN]

0
F(2)

-2000
z

-4000
0 1000 2000 3000 4000 5000 6000 7000
t [s]
Ftot(2) =Total second order wave force FII(2) =Second order pressure force
Figure 5-7: Calculated second order wave forces

4
x 10
1
M(2)
[kN]

tot
0 M(2)
M(2)

I
x

-1 M(2)
II
0 1000 2000 3000 4000 5000 6000 7000
4
M(2)
III
x 10
5 M(2)
IV
[kN]

0
M(2)
y

-5
0 1000 2000 3000 4000 5000 6000 7000
5
x 10
2
[kN]

0
M(2)
z

-2
0 1000 2000 3000 4000 5000 6000 7000
t [s]

M tot(2) =Total second order wave moments M II(2) =Second order pressure moments
Figure 5-8: Calculated second order wave moments

Final 55
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5.3 Ship simulations

5.3 .1 M odel set-up

The mooring arrangement is described taking into account the mooring lay-out (see Figure 2-4) and
the characteristics of both mooring lines and fenders (see Table 2-2). The ship motions were
simulated applying only the calculated first order wave forces, since it was noticed that the second
order wave forces increased significantly (w hich was thought to be unrealistic). The realistic oscillatory
part of the second order wave forces was such small that it could be neglected.

5.3 .2 M odel results

A representative example of the obtained ship motions simulations is given in Figure 5-9. The ship
motions are obtained after a run through all three successive numerical models. In the ship motions
simulations only the calculated first order wave forces are applied. From the analysis made in section
5.2.2 it is already remar ked that the first or der wave forces are not reliable. The obtained ship
motions are therefore not reliable, but caused by a succession of errors due to increasing horizontal
fluxes in MIKE21 BW. It is also possible that there is too much wave reflection from the breakwater
and/or beach in MIKE21 BW. Due to high wave reflection the ship makes a sway motion, releases
from the fenders and finally starts to surge. Since it is released from the fenders there is no friction to
resist the surge motion of the ship.

5
surge [m]

-5

10
sway [m]

-10

0.5
heave [m]

-0.5

20
roll [deg]

10

0.2
pitch [deg]

-0.2

5
yaw [deg]

-5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
t [h]
Figure 5-9: Simulated ship motions

Final 56
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5.4 Discussion of model results

During the performed simulations with MIKE21 BW a circulation cell of fluxes is formed within the
computational domain at deep water. Physically a circulation cell of fluxes is an eddy, w hich may exist
in the physical model as well. The fluxes of the simulated circulation cell grew in time, with respect to
both size as well as magnitude. Before the end of the abandoned simulation duration the total
velocity field was dominated by the eddy.

Why the circulation cell grew in time in both size as well as velocity magnitude is not understood. The
most obvious reason is that a small numerical instability occurred in a certain grid point, after which
surrounding grid points became instable as well. When the model finally blows -up almost the full
computational domain showed instable fluxes. The final blow-up of the model takes place at the
internal wave generation line, which should generate unidirectional waves. At the beginning of the
simulation the fluxes of the generated waves are indeed unidirectional. During the simulation the
fluxes most close to the wave guiding wall at the inter nal wave generation line star ted to show some
directional spreading. Most likely an oblique reflected wave is not well absorbed by the sponge layer
after the inter nal wave generation line. Before the imposed simulation duration the fluxes at the wave
generation line increased significant in size and more directional spreading was noticed. At the end of
the computation a negative water depth (at 20m water depth) is calculated. The calculated negative
water depth is not caused by an inconsistency in the generation of waves, but due to a numerical
instability in the computation of the flux field.

As a consequence of an unstable internal wave generation line, some high frequency waves may be
generated. According to DHI (2006A) the high frequency “noise” can be avoided by decreasing the
selected time step to ensure numerical stability at the inter nal wave generation line. T he selected
time-step, as described in section 5.1.1, was already decreased several times without any effect on
the obtained results. It should be noticed that the generation of high frequency “noise” was only
noticed for the simulations over longer period of time (i.e. couple of hours). It is remarkable that high
frequency “noise” is only noticed outside the designed porosity layer, w hereas the grid points inside
the porosity layer do not show instabilities in the computation of the flux field. The porosity layer
works effective for waves that propagate along the boundary, but information from waves
perpendicular to the area of interest is lacking.

The screenshots and time-series in Figure 5-10 are a visualization of the fluxes and surface elevations
in the area of interest (vicinity of the ship). It should be remar ked that along other boundaries
instabilities were noticed as well. A numerical instability in the fluxes is clearly noticed, which grew in
time in both size and magnitude. The location of the instability, in the area of interest, is located at
the separation between a porosity layer and the rest of the computational domain. During the
simulation surrounding grid points became unstable as well. The numerical instability finally resulted
in excessive fluxes in the computational grid just outside the porosity layer (see plotted time -series in
Figure 5-10). It should be remar ked that the ship is located in the middle of the plotted areas, which
confirms that the calculated wave forces on the ship are obtained due to a numerical instability in the
simulated fluxes in MIKE21 BW. Two impor tant aspects can be defined that explain the excessive
wave forces on the ship and the resulting unreliable simulated ship motions:
 A spatial gradient in the fluxes at the location of the ship.
 An increase of the simulated fluxes throughout the simulation.

Many attempts were made to make the model stable. These simulations included e.g. the application
of time-extrapolation factors, a smaller time step, alternative sponge and/or porosity layers,
smoothing of bathymetry and an increasing minimal water depth. These attempts w here however not
successful, while as a result of adapting parameters in MIKE21 BW the results may not represent true
physics. Most noticeable was that simulations with alternative seed numbers could result in longer
simulation durations. An alternative seed number is nothing else than an alternative realization of the
same spectrum by defining randomly different phases for the primar y waves. T he random seed
number should not have influence on the final results. If the seed number does have influence on the
final model results, the simulation duration is too short to obtain a wave-spectrum independent of the
incoming time-serie. The questions where and why the model became unstable were not solved
during this study.

Final 57
 The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

t=16Flux
t=16 [min] [min] P [m 3/s/m]
[m 3/s/m]
P Flux t=83Flux
t=83 [min] [min] 3
Flux
P [m P [m 3t=16
/s/m] /s/m][min]
t=16Flux [m3/s/m]
[min]QFlux Q [m3/s/m] t=83 [min]
t=83 Flux [m3Q
[min]QFlux [m3/s/m]
/s/m]

I II

-5 -5 0 0 5 5 -5 -5 0 0 5 -5 5 -5 0 0 5 5 -5 -5 0 0 5 5
Simulated fluxes in- and outside the porosity layer (respectively area I and II).
t=16t=16
[min] Elevation
[min] [m] [m]
Elevation t=83t=83
[min] Elevation
[min] [m] [m]
Elevation Surface elevation
0.5
Elv [m] II: Outside
0 I: Inside

-0.5
0 1000 2000 3000 4000 5000 6000 7000
Flux P
Flux P [m3/s/m]

10

-10
0 1000 2000 3000 4000 5000 6000 7000
Flux Q
Flux Q [m3/s/m]

10

-10
0 1000 2000 3000 4000 5000 6000 7000
-0.4-0.4 -0.2-0.2 0 0 0.2 0.2 0.4 0.4 -0.4-0.4 -0.2-0.2 0 0 0.2 0.2 0.4 0.4 time [sec]
Simulated surface elevations. Simulated time-series in- and outside porosity layer.
Figure 5-10: Snapshots and time-series in vicinity of the ship simulated with MIKE21 BW.

Other studies performed with MIKE21BW found instabilities within the computational domain as well.
Apart from numerical instabilities no convincing reasons were found which could explain instabilities
and/or model blow up. Two examples of vulnerability of numerical instabilities in MIKE21 BW will be
given here:
Brandi Mortensen (2006) showed that the rip current, w hich was simulated originally by Sorensen et
al. (1998), finally became asymmetrical after reaching a steady state and apar t from suggesting that
this was due to small numerical instabilities, no explanations were given (both bathymetry and
computational domain were symmetrical). Brandi Mor tensen (2006) found after a longer period of
simulation an excessive increase of wave energy within a bay. This finally resulted in a blow-up of the
model in deeper water. Extra attempts to obtain longer simulations periods failed.
Mahmoodian (2009), concluded that the application of sponge layers may not w ork optimal if the
direction of the wave is not perpendicular to the direction of the sponge layer. In the case of oblique
incident waves on a sponge layer unwanted reflections can be generated. Increasing the width of the
sponge layer did not lead to improvements of results in case of oblique incident waves. In order to
avoid the unwanted reflections the sponge layer was removed, which yields that full reflection is
imposed along lateral boundaries.

Final 58
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Although the details of both studies are not known, corresponding experiences are found with
simulations in MIKE21 BW during this study. A build-up of fluxes was noticed in the computational
domain, for which the cause could not be discovered. A non-optimal working sponge layer subjected
to oblique incident fluxes may be the reason w hy at the end of the simulation the fluxes most close to
the wave guide wall increased significantly in magnitude and showed directional spreading.

Although the input signal in MIKE21 BW could be even more optimized by the inclusion of a
theoretical defined bound long wave signal, a fair comparison between the simulated and me asured
spectra is not possible. After a long period of simulation, the numerical model becomes unstable and
finally stops. T he numerical simulated duration is much shor ter than the duration of the physical
model tests (about 1.5 hour numerical simulations compared to 3 hours of physical model tests). To
compare the numerical simulations and the physical model tests in a fair way the simulation duration
should be equal. If the simulation time is too shor t the width of the frequency bins is too large to
obtain a good representatio n of the distribution of energy density (especially for the low frequency
waves). T he statistical variability of the simulated low frequency spectr um is relatively high to make a
fair comparison between spectrum characteristics. T he number of waves necessary to obtain a
representative spectrum differs in literature. As a rule of thumb the number of waves should be
persisted as at least Nw aves≈400. With a test duration of 3 hours and a representative long wave
period of 60 seconds the number of waves becomes Nw aves≈180. For the simulated period within
MIKE21 BW the number of waves is only Nw aves≈90.

A switch to an alter native Boussinesq-type wave model was made due to the following reasons:
 As a consequence of numerical instabilities unreliable horizontal fluxes were obtained.
 The cause of increasing horizontal flues in the computational domain of MIKE21 BW could not
be discovered during this s tudy.
 Since the horizontal fluxes from MIKE21 BW were not reliable, the successive computations
with Harberth and Quaysim are not reliable as well, see section 5.2.2 and 5.3.2.
 Simulations over longer periods of time could not be made in MIKE21 BW, since the cause of
the instabilities was not discovered during this study.
 Long period of simulations are required to obtain statistical reliable ship motions.

Remark
In both Harber th and Quaysim numerical errors can be made as well. These errors may be as a
consequence of several factors: irregular frequencies, insufficient low frequency damping in the
retardation functions, other numerical errors etc.. Since already incorrect horizontal fluxes in MIKE21
BW were simulated a sequence of mistakes followed, see Figure 5-11.

Figure 5-11: Feedback to coupled numerical model scheme.

Final 59
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

6 Numerical simulations with TRITON

6.1 Model set-up

6.1 .1 General settings

The spatial grid cell size in TRITON was selected as Δx=4meter. The considerations to select this
spatial grid cell size are the same as described in section 5.1.1. A time step of Δt=0.1 seconds was
selected to ensure numerical stability based on the CFL condition. The selected time step is chosen
conservatively. As a consequence the total simulation duration in terms of total computational time
can be optimized by selecting a larger time step that still fulfills the CFL condition and for which the
waves are described with sufficient accuracy.

6.1 .2 Boundary conditions

The first step in the model set-up of TRITON is to determine which boundary conditions should be
applied. For the model set-up four types of boundar y conditions can be distinguished:
 Wave generation line.
 Closed boundary conditions.
 Partial reflecting boundary condition for beaches.
 Partial reflecting boundary condition for breakwaters.

The used model set-up is visualized in Figure 6-1. Notice that the location of the wave generation line
is not equal to the location of the wave paddles in the physical model. T he wave generation line is
placed at the location of wave gauge 1, which will be explained in section 6.3.2.

Partial reflecting
boundary Breakwater
Partial reflecting
boundary Beach
Closed boundary

Wave generation line

Figure 6-1: Applied boundary conditions in TRITON.

6.2 Program of simulations

A general over view of the performed simulations is given in this section. For more detail about the
applied boundaries conditions in TRITON, one is referred to section 6.1.2 and section 6.3. All
simulations in TRITON are performed for a tidal level of d=20 meters. The numerical simulation
duration in TRITON is equal to the duration of the physical model tests, see Table 6-1. Wave breaking
and moving shoreline (r un-up and run-down) are excluded during all simulatio ns.

Tp [sec] Total simulation duration

10 106 minutes
14 149 minutes
18 193 minutes
Table 6-1: Total simulation duration physical model tests and TRITON.

Final 60
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Partial boundary conditions breakwaters and beaches

During the first four simulations in TRITON the partial reflecting boundaries along the beaches and
breakwaters are varied according to Table 6-2. T hese simulations are performed to investigate the
sensitivity of the obtained primary wave spectra from the simulations for different applied reflection
coefficients. An explanation of the selected reflection coefficients will be given in section 6.3.1.

Run-id Tp [sec] Generated waves Cr Breakwater Cr Beach

01 14 Measured surface elevations 30% 2%
02 14 Measured surface elevations 40% 2%
03 14 Measured surface elevations 50% 2%
04 14 Measured surface elevations 40% 10%
Table 6-2: Variation of partial reflecting boundaries.

Wave conditions with different peak periods

Two alternative wave conditions are simulated in TRITON. One wave condition with a peak period of
T p=10 seconds and a second wave condition with a peak period of T p=18 seconds, see Table 6-3.
These runs are used to verify that other wave conditions in TRIT ON can be simulated accurately as
well.

Run-id Tp [sec] Generated waves Cr Breakwater Cr Beach

05 10 Measured surface elevations 40% 2%
06 18 Measured surface elevations 40% 2%
Table 6-3: Variation of primary peak wave period.

Wave generating boundary

Within TRITON it is possible to directly apply the measured surface elevations from the physical model
tests at the wave generation boundary. The measured surface elevations at the first wave probe in
front of the wave maker of the physical model were used as an incoming wave boundary for runs 01
to 06. In runs 07 to 09 different techniques to generate incoming wa ves are applied according to
Table 6-4. The applied techniques will be explained in more detail in section 6.3.2.

Run-id Tp [sec] Generated waves Cr Breakwater Cr Beach

07 14 First order wave generation 40% 2%
08 14 Second order wave generation method 1 40% 2%
09 14 Second order wave generation method 2 40% 2%
Table 6-4: Variation of wave generation.

Additional simulations
Three additional simulations are carried out with TRITON. These additional simulations are carried out
for the following reasons:

Run 10: The breakwaterhead initially simulated in TRITON is based on the adapted
breakwaterhead which was used performing simulations with MIKE21 BW. This
breakwaterhead is shorter than the breakwater head applied in the physical model. As
a consequence too much high frequency energy in the area of interest may be
simulated.

Run 11: Low frequency waves that diffract behind the Nor th breakwater (IV in Figure 2-8),
may reflect and cause standing wave patter ns between the physical model side walls.
These standing wave patter ns are due to the finite physical model basin sizes. Side
wall I (see Figure 6-1) is schematized as a partial reflecting boundar y with a low
reflection coefficient to investigate the effect of the side wall on possibly standing
waves.

Final 61
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Run 12: The par tial reflecting boundaries applied at side wall II (see Figure 6-1) have a low
reflection coefficient of 2% in Run 01 to Run 11. Small partial reflection fr om this
boundary is expected for the high frequency waves, but waves with low frequencies
may reflect full from the boundar y. As a consequence, the energy of especially the
long waves in the numerical model may be underestimated.

Run 13: The porous absorbent beach (detail Figure 2-8) is removed and a closed boundary
condition is applied at the entrance to the por t basin and side wall II. The expectation
was that the short waves will break on the porous beach installed at the entrance of
the harbour basin (see detail Figure 2-8) and small wave reflection from the porous
beach was expected. The entrance to the harbour basin was therefore schematized as
a partial reflecting boundar y with a low reflection coefficient (Run 01 to Run 11). The
hypotheses is that the porous beach is probably effective in absor bing high frequency
waves, but is less effective in case of low frequency waves.

Run 14: A basin is added and the absorbent beach installed at the entrance is removed. The
boundaries of the basin are full reflecting, see Figure 6-2. The entrance of the basin is
equal to the physical model basin, but the basin itself is a smaller reproduction of the
basin within the physical model.

Remark
Measurements of velocities and/or surface elevations in the harbour basin are not carried out. The
hypotheses cannot be verified by physical model measurements.

Run-id Tp [sec] Generated waves Remark:

10 14 Measured surface elevations Adapted breakwaterhead
11 14 Measured surface elevations Partial reflecting BC side wall I
12 14 Measured surface elevations Full reflecting BC beaches side wall II
13 14 Measured surface elevations Closed boundary at basin entrance
14 14 Measured surface elevations Basin added with full reflecting boundaries
Table 6-5: Variation of other boundary conditions.

Run 12 Run 13 Run 14

Wave generation Partial reflecting beach Partial reflecting breakwater Closed
Figure 6-2: Applied boundary conditions in TRITON for additional simulations. Run 12 to 14.

Final 62
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

6.2 .1 Summary of simulations

An overview of the simulations carried out with TRITON is given in Table 6-6.

Run-id Tp [sec] Incoming wave Cr Br Wat Cr Beach Remark:

01 14 Measured 30% 2%
02 14 Measured 40% 2%
03 14 Measured 50% 2%
04 14 Measured 40% 10%
05 10 Measured 40% 2%
06 18 Measured 40% 2%
07 14 1st order BC 40% 2%
08 14 2nd order BC 1 40% 2%
09 14 2nd order BC 2 40% 2%
10 14 Measured 40% 2% Adapted breakwaterhead
11 14 Measured 40% 2% Partial reflecting BC SW I
12 14 Measured 40% 2% Full reflecting BC SWII
13 14 Measured 40% 2% Closed basin entrance
14 14 Measured 40% 2% Open basin entrance
Table 6-6: Overview of simulations with TRITON.

6.3 Explanation of boundary conditions

6.3 .1 Partial reflecting boundaries

The partial reflecting boundaries in T RITON are depending on the reflection coefficient and the peak
period of the primary waves. In order to determine the reflection coefficients for the rubble mound
breakwater and the beach two different equations are presented.
The reflection coefficient of a rubble mound breakwater can be approximated using the empirical
relationship of Muttray et al. (2006):
1
 2 
CR  1,3  3d   Eq. 6-1
 L0 
In which:
CR = Reflection coefficient [-]
d = Still water depth [m]
L0 = Deep water wave length [m]

Tp [sec] L0 [m] d [m] CR [-]

10 156 16~20 0.30~0.27
14 306 16~20 0.44~0.40
18 506 16~20 0.53~0.49
Table 6-7: Reflection coefficients of a rubble mound breakwater for different wave condit ions.

For a beach the reflection coefficient mainly depends on the beach slope and the type of breaking.
The slope of the beach during the physical model tests is approximated and assumed to be uniform
alongshore. The reflection coefficient of a beach is, according to Battjes (1974), given by:
tan 
CR  0,1 2 with:  Eq. 6-2
H / L0
In which:
tan  = Beach slope angle [-]
 = Iribarren number [-]
H = Wave height [m]

Final 63
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Tp [sec] L0 [m] H [m] tan  [-]  [-] CR [-]

10 156 3 1/25=0,04 0.29 0.008
14 306 3 0,04 0.40 0.016
18 506 3 0,04 0.52 0.027
Table 6-8: Reflection coefficients of a beach for different wave conditions

The calculated reflection coefficients agree well with the range of reflection coefficients which were
obtained by breakwaters tests at FEUP (0.25<C r <0.45), see Table 6-7. The sensitivity of the reflection
coefficients for partial reflecting breakwaters within TRIT ON is tested in run 01 to run 03, see section
6.2. Minimal wave reflection is expected from the absorbent beach at the end of the physical model
opposing the wave generators (see V in Figure 2-8). The applied reflection coefficient from the beach
in TRITON is set to 2% for runs 01 to 03, but was increased in run 04 to 10% in order to investigate
the effect of more wave reflection from the beach, see section 6.2.

6.3 .2 Generated waves

The measured surface elevations at the first wave probe of the physical model tests serve directly
input for the generation of waves in TRITON. T RIT O N calculates the corresponding velocities of the
waves. Both surface elevations and velocities at the incoming wave boundary are controlled by the
adaptive boundar y procedure. It should be noticed that a linear interpolation technique is used to
define the generated waves for TRITON. Application of an interpolation technique was required due to
the fact that the sample period of the surface elevations during the physical model tests is not equal
to the time-step defined in TRITON. T he sample frequency during the physical model tests was 10/24
Hz (prototype value), w hereas the generated waves in TRITON are defined with a sample frequency
of 1/10 Hz. Effects due to application of the applied interpolation technique are expected to be
minimal and non-appreciable for the results.

Directly applying the measured surface elevations as an input for TRITON has disadvantages. The
measured surface elevations at the wave probes in front of the wave maker in the physical model
contains the generated waves, but also reflected waves (and possibly re-reflected waves). Free
reflected waves can be generated by several basin effects, see Voogt et al. (2005):
 Reflections due to the finite size of the basin.
 Mismatch in velocities at the wave maker compared with ocean waves, if no second order
correction is made at the wave maker.
 Release of set-down waves after shoaling resulting in free waves.

Dissipation of reflected free long waves in the physical model basin is minimal since the efficiency of
the dynamic wave absorption unit of the wave maker is minimal for low frequency waves, see Figure
2-9. As a consequence the measured free (re-) reflected long waves are included in the incoming
signal of TRITON. The outgoing waves measured at wave gauge 1 are now generated in TRITON as
well.

In run 07 a so-called 1st order boundary condition is imposed at the wave generation line in TRITON.
This incoming wave signal is based on the measured surface elevations, but waves with periods
higher than 25 seconds are excluded. A bound long wave at the incoming wave boundar y associated
with the primary waves is not imposed, but will be generated within the computational domain of
TRITON in combination with possible spurious free waves.

A theoretical bound long wave can be included in the incoming wave signal of TRITON. In that case a
2nd order boundary condition is imposed at the wave generation line. This is done by adding a
theoretical derived bound long wave signal with the incoming primary waves, in which the latter is the
generated time-serie of run 07. Two methods to derive a theoretical bound long wave are
distinguished:
 A theoretical bound long wave signal based on the conser vation of radiation stress as
proposed by Longuet-Higgins and Stewart (1962)  Run 08.
 A theoretical bound long wave based on interaction between pairs of primary waves in a wave
group, see Herbers et al. (1994)  Run 09.

Final 64
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The first method calculates a bound long wave based on the wave envelope in which a constant
interaction coefficient is applied between the amplitude of the bound long wave and the wave group.
The second method is based on interaction between pairs of primar y waves in wave group each
having their own interaction coefficient. Summation over all the components of pairs of primary waves
gives the total bound long wave. In TRIT ON the incoming time-serie of waves needs to be defined in
surface-elevations. The original expressions of Herbers et al. (1994) are valid for bottom pressures.
Adjustments to rewrite the transfer functions in surface elevations can be found in Van Dongeren et al.
(2003).

Within the second method a threshold need to be defined to determine the interaction pairs of
primary waves. This is done according to the following methodology:
 The measured surface elevations are converted in a wave spectrum via FFT analysis.
 Waves with a period larger than 25 seconds are removed from the wave spectrum.
 A threshold is defined as 95% of the most energetic primary waves.
 For each possible pair of primar y waves in the defined threshold the contribution to the bound
long wave is calculated.
Defining a theoretical bound long wave including more than 95% of the most ener getic primary waves
did not lead to noticeable differences in the defined time-serie. It can therefore be concluded that the
theoretical bound long wave based on the 95% criteria is accurately described.

Remark
Wave splitting, see Voogt et al. (2005) and Waals (2009), was not performed during the physical
model tests. Wave splitting tools decompose measured waves into:
 Incident free waves
 Incident bound waves
 Reflected waves

Wave splitting tools can be used to indentify spurious free waves and reflected waves in the model
basin. Incorporating these waves in the numerical simulations may further improve the agreement
between measurements and simulations.

Final 65
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

7 Comparison physical model results and TRITON

Within this section references are made to figures which can be found in Appendix C: TRITON
simulations. For the location of the different wave gauges one is referred to Figure 2-12.

For the calculation of spectr um characteristics a distinction is made between:

 The full wave spectrum: including all frequencies.
 The primary wave spectrum: only including frequencies > 0.04 Hz.
 The low frequency wave spectrum: only including frequencies < 0.04 Hz.

Partial boundary conditions breakwaters and beaches

In run 01 to run 04 different reflection coefficients are defined for the breakwaters and beaches along
the model boundaries. A good overall agreement between the simulated and measured wave spectra
for the primary waves in terms of spectrum characteristics and spectral shape is obtained, see
Figure 7-1 and Figure C- 1. At wave gauge 5 the simulations give an underesti mation compared to
measurements. This difference may be due to more reflection of the most energetic waves from the
beach. Despite the increase of the reflection coefficient of the beach in run 04 higher energetic waves
at wave gauge 5 were not noticed. A possible reason is that the beach slope in the physical model
basin is steeper than the beach slope a pplied in the numerical simulations. If the beach slope is not
modelled correctly this mainly affects the amplitude of free reflected waves from the beach, since the
amplitude of free reflected waves increases for increasing beach slopes, see Baldock (2000).

From the simulated low frequency wave spectra at wave gauge 1 and 5 it is observed that the
increase in spectral density magnitude is well simulated for frequencies around 0.02Hz (although a
slight underestimation is noticed). The underestimation of primar y wave energy at wave gauge 5 for
the most energetic waves is only noticed for the tested wave conditions of T p =14 and 18 seconds. For
the tested wave condition of T p=10 seconds there is no underestimation of primary wave energy, see
Figure 7-2 and Figure C- 2.

The simulated wave height for the low frequency waves equals the measured low frequency wave
height, but the spectral shape at low frequencies is not satisfactor y. Wave energy is simulated at
lower frequencies compared to measurements. As a consequence a deviation in both T m01;low and
T m02;low compared to measurements is calculated. The largest deviation between measurements and
simulations is noticed for T m01;low at wave probe 6 and 8. This deviation is due to the simulated wave
energy at very low frequencies (<0.0 05Hz), since the calculated spectral period is most sensitive for
wave energy at these very low frequencies. Although the simulated wave height at low frequencies is
in the same or der of magnitude as measurements, wave energy is predicted at different frequencies.

From run 01 to 04 the following can be concluded:

 The reflections coefficients for the partial reflecting boundaries in TRITON are well selected.
 Selecting alternative reflection coefficients for partial reflecting boundaries mainly affects the
simulated primar y wave energy, but has negligible effects on the low frequency waves.
 The reflection coefficients as applied in run 02 gave the best agreement between
measurements and simulations for the most energetic primary waves in the vicinity of the ship
(see wave spectrum at wave gauge 8 in Figure C- 1).

Final 66
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Hm0;full [m] Hm0;primair [m] Hm0;low [m]

4 4

0.5
2 2

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm01;full [s] Tm01;primair [s] Tm01;low [s]
15 15 100
10 10
50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm02;full [s] Tm02;primair [s] Tm02;low [s]
15 15
10 10 50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Wave gauge number
1000
Figure 7-1: Spectrum characteristics run 01 to run 04. Measured
900 Run01 Cr Bw=0.3 b=0.02
Run02 Cr Bw=0.4 b=0.02
800
Run03 Cr Bw=0.5 b=0.02
Run04 Cr Bw=0.4 b=0.10
700

600

500

400

300

200

100

0
0 1 2 3

Final 67
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Wave conditions with different peak periods

Runs for other tested wave periods (T p=10 [sec] and T p=18 [sec] in run 05 and run 06 respectively),
showed that these wave conditions can be well simulated in TRIT ON, see Figure 7-2 and Figure C- 2.
As noticed in the section before the primary wave height at wave gauge 5 is underestimated, but the
primary waves in the vicinity of the ship are well simulated. The expected trend of increasing low
frequency energy with the incoming peak period of the primary waves is clearly visible in Figure C- 2.
The increase of low frequency energy with increasing period was expected, since the surface
elevations of each of the infra- gravity components increases with increasing peak period of incoming
swell (which can be derived based on the conser vation of radiation stress for a flat bottom, see e.g.
Van Noorloos (2003)). In the vicinity of the ship enhancement of low frequency wave energy is
noticed for increasing peak period. It should be remar ked that in the vicinity of the ship the
frequencies corresponding to the enhanced low frequency wave energy peaks are not predicte d well.

Hm0;full [m] Hm0;primair [m] Hm0;low [m]

4 4 1

2 2 0.5

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm01;full [s] Tm01;primair [s] Tm01;low [s]
20 20 100

10 10 50

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm02;full [s] Tm02;primair [s] Tm02;low [s]
20 20 100

10 10 50

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Wave gauge number
1000
Figure 7-2: Spectrum characteristics run 02, run 05 and run 06. Run05 Measured Tp=10 [sec]
900 Run05 Simulated Tp=10 [sec]
Run02 Measured Tp=14 [sec]
800
Run02 Simulated Tp=14 [sec]
Run06 Measured Tp=18 [sec]
700
Run06 Simulated Tp=18 [sec]
600

500

400

300

200
Final 68
100
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Wave generating boundary

In run 07 to 09 the generated waves in TRIT ON are varied, see Figure C- 3 and Figure 7-3. In r un 07
only primar y waves are generated. The incoming bound long wave associated with the primary waves
groups is now generated within the computational domain of TRITON in combination with spurious
free waves. Not imposing a bound long wave already in the incoming time-serie led to a significant
underestimation of the total amount of low frequency energy compared to measurements. The
calculated low frequency wave height (Hm0;low ) is decreased with a factor two.

An improvement in the agreement between measurements and simulations is made by defining a

second or der boundar y condition for the generated waves. Instead of only generating primar y waves
a theoretical bound long waves associated with the primary waves is now imposed in the incoming
time-serie. Although the amount of wave energy at the lower frequencies is increased there is an
underestimation of low frequency wave energy noticed at wave gauge 1 and 5. At wave gauge 6 and
8 the simulated low frequency wave height is slightly underestimated compared to measurements.
Although the simulated low frequency wave height is almost equal to the measured low frequency
wave height, the distribution over the corresponding frequencies is not well predicted. From the wave
spectrum and the calculation of the spectral periods it is observed that there is an accumulation of
wave energy for the ver y low frequencies (<0.005Hz), w hich does not correspond with a peak in the
measured wave spectrum. An explanation for this accumulation of wave energy at these ver y low
frequencies cannot be given. The expected ship response for these energetic ver y low frequency
waves is however low, see Figure 3-6.

Hm0;full [m] Hm0;primair [m] Hm0;low [m]

4 4

0.5
2 2

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm01;full [s] Tm01;primair [s] Tm01;low [s]
15 15 400
10 10
200
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm02;full [s] Tm02;primair [s] Tm02;low [s]
15 15 200
10 10
100
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
1000 Wave gauge number
Measured
Figure 7-3: Spectrum characteristics
900 run 02, run 07 to run 09. Run02 Measured
Run07 (O)1 BC
800
Run08 (O)2 BC1
700 Run09 (O)2 BC2

600

Final 500
69
400
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

A better agreement between simulations and measurements by imposing a bound long wave was
expected, since during physical model tests second order wave steering was used. In case of second
order wave steering in the simulations within TRITON it is remarkable that especially the energy
density in the range 0.02~0.04Hz is less compared to measurements. Since a wave spectrum does
not include directional information only possible explanations can be given for this shortage.

This shortage may be due to:

 Shortcomings in the description of the theoretical bound long wave.
 Possible measured (re-)reflected free waves from the wave maker, which are not absorbed.
 Transverse measured waves standing between the guide wall and the side wall of the physical
model basin. T hese waves are not or partly absorbed, since the absorption of the wave maker
mainly absorbs normal incident waves.
 Scattering of primar y waves from the breakwaterhead, which interact wi th generated incident
waves resulting in radiated free long waves
 Free long waves which are released due to diffraction of the primary waves.
 More energetic waves reflecting from the beach (V in Figure 2- 8).

Since negligible moored ship response for waves with these frequencies ( 0.02~0.04Hz) is measured,
see Figure 3-6, this shor tage will not be investigated further.

Additional simulations
For the wave spectra for run 10 and 11 one is referred to Figure C- 4, whereas the spectrum
characteristics can be found in Figure 7-5. In run 10 the head of the breakwater is enlarged by
replacing the partial boundaries of the head of the breakwater. The adaptation mainly influences the
simulated primary waves at wave probe 6 and 8. The breakwaterhead as modelled in run 02 gave an
small underestimation of the most energetic primary waves at wave probe 6, w hereas in simula tion 10
the underestimation of the most energetic waves is more significant. Adaptation of the
breakwaterhead did not lead to major changes in the low frequency spectrum, but less energy is
transferred to frequencies around 0.02 Hz in the vicinity of the s hip.

The reason why less wave energy is transferred to sub-harmonics is the following:
The breakwaterhead is a discontinuity in the bathymetry at w hich free long waves are released, see
Figure 2-17. By replacing the par tial reflecting boundaries of the breakwaterhead bathymetrical
information is missing. Due to the adaptation of the breakwaterhead primary wave shoaling and the
generation of sub-harmonics at the harbour entrance is taken less into account, see Figure 7-4.

The resulting effects on the waves due to adaptations of the breakwater head are small , but less wave
energy is transferred to the sub-harmonics with frequencies around 0.02 Hz. As stated in the section
above the moored ship response for these wave frequencies is low and therefore less impor tant.

Modelled tip of breakwater run 02 Modelled tip of breakwater run 10

Figure 7-4: Schematization tip of breakwater in TRITON.

Final 70
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

After run 11 some remarkable results are obtained:

 A lower primary wave height at wave gauge 1 and 5 compared with the base run.
 A higher primary wave height at wave gauge 6 and 8 compared with the base run.
 Significant less low frequency energy at wave probe 5 compared to the base run.

The differences in calculated wave heights are small, see Figure 7-5. T he wave spectr um however
significantly changed in spectral shape, see Figure C- 4. Possibly more wave energy is transferred into
super-harmonics. These super-harmonics have high frequencies and are for moored ship responses
less relevant.

The obtained results are influenced by some numerical imperfections. Par tial reflecting boundaries in
TRITON are not designed for waves that propagate along the model boundary. As a result the
generated waves that propagate along the model boundar y are affected. This results in a distor ted
view in the surface elevations as well as the flow field. This distorted view is due to diffraction losses
as well as numerical imperfections, see Figure 7- 6. The boundar y condition at the side wall is open for
all outgoing waves, but since the generated incoming waves are affected as wel l the final results are
not representative.

Hm0;full [m] Hm0;primair [m] Hm0;low [m]

4 4

0.5
2 2

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm01;full [s] Tm01;primair [s] Tm01;low [s]
15 15 100
10 10
50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm02;full [s] Tm02;primair [s] Tm02;low [s]
15 15
10 10 50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Wave gauge number
1000
Figure 7-5: Spectrum characteristics run 02, run 10 and run 11. Measured
900 Run02 Original
Run10 Adapted breakwater head
800
Run11 Partial reflecting side wall I

700

600

500

400
Final 71
300
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Closed Partial reflecting Closed Partial reflecting

Figure 7-6: Snapshots of TRITON during run 04 and run 011.

Remarks
After run 01 to run 11 the following should be remarked:
 There is a difference in wave energy at wave gauge 1 for ver y low frequencies around
0.005Hz.
 In the vicinity of the ship (at wave gauge 6 and 8) the amount of low frequency wave energy
is well predicted, but the distribution of energy over the corresponding frequencies is not well
predicted. TRITON overestimates the amount of low frequency energy at ver y low frequencies
compared to measurements. The measured peaks in the wave spectra are not simulated in
TRITON.

The difference in wave energy at wave probe 1 for the ver y low frequencies around 0.005Hz, may be
a consequence of the following reasons:
 The distance between the beach and wave maker in the physical model differs from the
numerical model since the wave are generated at the location of the first wave probe instead
of the wave maker.
 The generation line in the numerical model reflects wave energy for these ver y low
frequencies. An improvement in T RITON at the wave generation line can be made by
decreasing the reflection of the generation line (in this case ωexplicit instead of Cr).
 Generation of a so called evanescent wave, for w hich the amplitude decays exponentially from
the boundary.

Since the difference in wave energy at very low frequencies (around 0.005Hz) is only noticed at wave
gauge 1 and not at wave gauge 5 it is most likely the latter reason. This evanescent wave is a local
numerical spurious wave with a small amplitude which decays exponentially from the boundar y. The
evanescent wave is thus a local disturbance and will not have any effects on a global scale (locations
of the other wave gauges).

Final 72
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Hm0;full [m] Hm0;primair [m] Hm0;low [m]

4 4

0.5
2 2

0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm01;full [s] Tm01;primair [s] Tm01;low [s]
15 15 100
10 10
50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
Tm02;full [s] Tm02;primair [s] Tm02;low [s]
15 15
10 10 50
5 5
0 0 0
1 5 6 8 1 5 6 8 1 5 6 8
1000 Wave gauge number
Measured
Figure 7-7: Spectrum characteristics
900 runs 02, 12, 13 and 14.
Run02 Original
Run12 Full reflecting BC SW II
800
Run13 Closed basin entrance
Run14 Open basin entrance
700
In run 12 the reflection coefficient of the beach and the porous beach at sidewall II are higher, see
Figure 6-1. Adaptation of 600this reflection coefficient has no effect on the primar y wave spectrum, but
does affect the spectral density around 0.01Hz at wave probes 6 and 8. Taking the changes in
simulated spectral density500into account as well as the combined measured spectra of waves and surge
motion (see Figure 3-6) it suppor ts that the wave direction of the generated sub-harmonics is mainly
parallel relative to the North
400breakwater.

Although the reflection coefficient

300 of the model boundary is increased for the porous absor bent beach
the simulated spectral density decreased. Two possible explanations can be given:
 Physical: a standing200wave pattern is formed and the output location is placed in a node.
 Numerical: spurious free waves due to a deviated reflection coefficient of the model boundary.
100
For waves with a frequency of 0.01Hz all structures will fully reflect and physically the same standing
wave patterns will be generated
0 compared to the original r un (run 02). Since the conditions for both
runs were constant, except for the reflection
0 coefficient of
1 the model boundaries,
2 a numerical
3 reason
must be the underlying reason.

Final 73
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The numerical reason will be explained with an example for which a typical wave period of 14 seconds
is assumed. The performance of the par tial boundar y condition procedure in TRITON is depending on
a typical period (usually chosen as T p), w hich is defined in the input. The model boundary is expecting
waves with the typical wave period and the par tial reflection coefficient is valid for those waves. For
waves with periods close to the typical period (e.g. 10~18 seconds) the reflected wave is correctly
calculated as well. If the difference between the typical period and the wave period is very large (e.g.
typical wave period of 14 seconds versus an incoming wave period of 100 seconds) the amplitude of
the reflected wave may be different than expected based on the reflection coefficient. Although this
effect may be small in absolute terms (amplitude), for the small amplitude waves as considered here
this effect may be relative large (in terms of spectral density, which is quadratic depending on the
wave amplitude). The performance of partial boundary conditions in TRITON is subjected to further
research and outside the scope of this thesis. The difference between run 02 and run 12 is expected
to be so small that it w ould not have any effect on the response of the moored ship.

In simulations 13 and 14 a modification is made in the harbour configuration. In the original

simulations the bathymetr y as well as the boundar y conditions were represented as accurate as
possible. T he configuration did however not include the harbour basin, w hich was reproduced in the
physical model, see Figure 6-1. The expectation was that the short waves will break on the porous
beach installed at the entrance of the harbour basin (see detail Figure 2-8) and small reflection from
the porous beach was expected. The entrance to the harbour basin was therefore initially schematized
as a partial reflecting boundar y with a low reflection coefficient. By increasing the reflection coefficient
of the porous beach contradicting results were obtained, see discussion run 12.

After simulation 13 the peak at a frequency of 0.015Hz at wave probe 6 becomes more pronounced
(see Figure 7-8), which is possibly due to a simulated standing wave (see both measured and
simulated peaks in wave spectra). T he resulting effec t is small and only some specific simulated wave
amplitudes differ, which causes the differences in the simulated spectral density.

In run 14 an arbitrary basin is added. Due to the arbitrary basin an shift towards the frequency of
0.005 Hz is noticed com pared to previous simulations (see also calculated spectral period in Figure
7-7). The first measured peak in the vicinity of the ship becomes more pronounced and is now
simulated (see Figure C- 5). The effect of the added basin becomes more pronounced when less
smoothing is applied to the presented wave spectr um, see Figure 7-8. It should be noticed that the
number of frequency bins for both simulations and measurements are equal and constant smoothing
is applied.

TRITON did not simulate the peak at a wave frequency of 0.006Hz during all simulations, whereas a
peak for these frequencies is measured. After inclusion of a basin in run 14 a peak appears at a
frequency of 0.007Hz in the simulated wave spectrum. An increase in wave ener gy, due to a standing
wave, at the correct corresponding frequencies is indentified and an improvement in the agreement
between measurements and simulations is obtained.

Possibly long wave transmission took place through the designed porous beach installed at the
entrance of the basin (VII in Figure 2-8). Since there is no damping installed in the physical model
basin, the generated free long waves may amplify and as a result a standing wave is formed in the
physical model basin. Although the response amplitude is not correctly scaled, the corresponding
frequencies at which resonant amplification occurred is correctly simulated. T he measured peak at a
frequency of 0.016Hz at wave gauge 6 is still not simulated, but this peak may very well correspond to
the second mode of the standing wave.

The mismatch in amplitude between measurements and simulations at frequencies around 0.005Hz
for runs 01 to 13 is possibly caused by spurious free reflected waves, due to the finite sizes of the
physical model basin. A transverse standing wave was already suggested to be a possible cause for
the found local peaks in the wave spectra, see chapter 3. The assumption was endorsed by the
measured surge motion and wave spectra, see Figure 3-6. The simulated basin in TRITON is smaller
than the harbour basin build in the physical model. Due to space limitations the represented har bour
basin in the physical model is a smaller reproduction of the total harbour basin (i.e. the total har bour
of Leixões).

Final 74
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 5 Low freq.
5
S [m2/Hz]

0
0 0.005 0.01 0.015 0.02 0.025
WP 6 Low freq.

1
S [m2/Hz]

0.5

0
0 0.005 0.01 0.015 0.02 0.025
WP 8 Low freq.

1
S [m2/Hz]

0.5

0
0 0.005 0.01 0.015 0.02 0.025
frequency [Hz]

Figure 7-8: Results of Run 12 to Run 14.

Measured
Adapted harbour configuration and boundary R02 Original
conditions. R12 SW II closed
Less smoothing compared to Figure C- 5. R13 Small basin
R14 Large basin

Final 75
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

7.1 Summary of simulations

After the performed simulations in TRITON the following can be concluded:
 The harbour configuration is impor tant to obtain standing waves with the corresponding low
frequencies.
 Changes in reflection coefficients for partial reflecting boundaries had a minor effect on the
obtained wave spectrum.
 Whether or not imposing a bound long wave in the generation of waves had a major effect on
the obtained low frequency spectrum.
 Increasing the spectral peak period of the primar y waves resulted in an increase of the
amplitudes of the low frequency waves.

Based on the simulations carried out during this study some important factors for correctly modelling
the waves for a port can be qualitatively judged, see Table 7-1.

High frequency waves Low frequency waves

Lay-out of the port ++ ++
Reflection coefficients of port str uctures + +/-
Imposing a bound long wave in generated waves - +
Table 7-1: Qualitative judgement of sensitivity TRITON.

Final 76
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

8 Discussion of Boussinesq-type wave model results

The simulations of MIKE21 BW and TRIT ON cannot be compared directly. This is due to the fact that
alternative model set-ups and alternative numerical techniques are applied during simulations with
both models. The final results of the simulations carried out with MIKE21 BW and the results of
unstable TRITON simulations show similarities. In both cases eddies are obser ved in the
computational domain. The eddies grew in both size as well as velocity magnitude during the
simulations. Finally, the velocities of these eddies became dominant over the wave induced fluxes.
Both velocity and magnitude of the generated eddies are thought to be physically unrealistic, but
purely a consequence of numerical instabilities. Apart from suggesting that a numerical instability is
the cause of excessive eddies within MIKE21BW as well as in TRITON, no evidence can be given.

MIKE21BW can become numerical unstable for various reasons, see DHI (2006A). Why MIKE21 BW
became numerical unstable during this study is unknown. Attempts to stabilize MIKE21 BW in order to
obtain longer periods of simulations did not w ork out. By altering the model set-up in TRITON stable
model results were obtained for long period simulations. The effects of avoiding numerical instable
boundaries are clearly visible after long time of simulation, see Appendix D.

It should be remar ked that the major aspects (e.g. bathymetry, grid spacing and time -step) to set-up
simulations in both Boussinesq-type wave models were kept constant. The main difference between
both models is the inclusion of the boundary conditions (e.g. par tial reflecting structures and wave
generation) and alternative numerical schemes.

Based on the wor k in this study it cannot be concluded that either TRITON or MIKE21 BW is a better
numerical model. Both numerical models have their strengths and weaknesses. T he switch from
MIKE21 BW to TRITON was made due to problems within MIKE21 BW. The switch to an alternative
Boussinesq-type wave model was made, since the causes of numerical instabilities after long periods
of simulation in MIKE21 BW could not be discovered by the model-user. Since long period simulations
are required for the final goals e.g. obtaining ship motions, the switch to TRITON was logical and
finally successful for the purposes of this study. From the analysis made in chapter 5 it was concluded
that the obtained ship motions, based on computations with MIKE21 BW, due to numerical instabilities
in the simulated fluxes were not reliable.

Final 77
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

9 Conclusions and recommendations

9.1 Conclusions
The original objective of this study was to simulate the moored ship motions at ber th “A” in the Port
of Leixões by applying a sequence of numerical models:
 In order to predict the wave motions at berth “A” in the Port of Leixões the effects of
diffraction, par tial reflection and non-linear wave interaction are relevant hydrodynamic
processes. To include the relevant hydrodynamic processes an appropriate Boussinesq-type
wave model was selected.
 It is possible to simulate ship motions based on the output of Boussinesq-type wave models
which serve as an input for a panel model. The calculated wave forces fr om the panel model
serve as an input for a ship simulation model.
 To obtain statistical reliable ship motions t he applied Boussinesq-type wave model should be
robust enough to simulate waves over several hours, meanw hile describing the relevant
hydr odynamic processes in an accurate way.

In the initial approach the Boussinesq-type wave model MIKE21 BW was applied, which became
numerical unsta ble. The cause for these numerical instabilities within MIKE21 BW could not be
discovered during this study, but based on the performed simulations the following can be concluded:
 Numerical instabilities may cause unwanted effects expressed in the velocity field of a
Boussinesq-type wave model.
 Numerical instabilities in the velocity field of the Boussinesq-type wave model resulting in
eddies with high velocities magnitudes will cause an increase in calculated second order wave
forces on the ship, if these eddies are located in the vicinity of the ship.
 As a consequence of numerical instabilities in the simulated velocity field the wave forces on
the ship and thus the obtained ship motions are not reliable.

A switch was made during this study to the Boussinesq-type wave model TRITON and the original
objective adjusted. The focus of this study is shifted to simulate the low frequency waves that caused
the ship motions during the physical model tests. From the simulations carried out with the
Boussinesq-type wave model TRITON the following can be concluded:
 Not imposing 2nd order boundary conditions in TRIT ON may lead to an underestimation of low
frequency energy compared to measurements if 2 nd order boundar y are simulated during
physical model tests.
 TRITON is well able to predict possible basin oscillations.
 A measured standing wave in the physical model is the main cause of the measured surge
responses of the moored vessel.
 The wave simulations as performed with TRITON are expected to be sufficiently accura te to
serve as an input for vessel response computations.
 The wave simulations as performed with TRITON may serve as a base for further research.

Final 78
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

9.2 Recommendations
Numerical modelling:
 Modelling of the total Por t of Leixões with a Boussinesq-type wave model (or equivalent),
taking into account all relevant hydrodynamic processes for moored vessel response.
 Further investigation of the moored ship motions based on simulations with the Boussinesq-
type wave model TRITON or equivalent model. The wave forces on the ship should be
calculated with the panel model Harber th or equivalent model.

Physical modelling (in case of additional tests):

 Place wave gauges in the corners of the physical model basin to discover possible spurious
basin resonance.
 Consider the options of wave splitting, since this may provide more insight in the origin of the
generated free waves in the physical model.

DOLPHIN project:
 Perform wave measurements inside and outside the Por t of Leixões to calibrate and validate
both physical and numerical models.
 Perform local ship motion measurements to calibrate and validate both physical and numerical
models.
 Long wave transmission through the permeable core of the Nor th breakwater may result in
sway motions of the moored vessel. It is recommended to take this mechanism into account
when analysing pr ototype measurements.

Final 79
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Bibliography
Baldock, T.E., Huntley, D.A., Bird, P.A.D., O‟Hare, T.O. and Bullock, G.N. (2000) . Breakpoint generated surf beat
induced by bichromatic wave groups. Coastal Engineering, 39, p.213-242.

Baldock, T.E., and Huntley, D.A. (2002). Long-wave forcing by the breaking of random gravity waves on a beach.
Proceedings Royal Society Londen, serie A 458, p.2177-2201.

Battjes, J.A. (1974). Surf similarity. 14th International conference on Coastal Engingeering, ASCE, p.466-480.

Battjes, J.A., Bakkenes, H.J., Janssen, T.T., and Van Dongeren, A. (2004). Shoaling of subharmonic gravity waves.
Journal Geophysical Research, 109, C02009.

Beck, R.F. (1994). Time-domain computations for floating bodies. Applied Ocean Research, 16, p.267-282.

Bingham, H.B. (2000). A hybrid Boussinesq-panel method for predicting the motion of a moored ship. Coastal
Engineering, 40, p. 21-38.

Borsboom, M.J.A., Doorn, N., Groeneweg, J., and Van Gent, M.R.A. (2000). A Boussinesq-type wave model that
conserves both mass and momentum. Proceedings 27 th ICCE Conference, Sydney, p. 148-161.

Borsboom, M.J.A., Groeneweg, J., Doorn, N., and Van Gent, M.R.A. (2001). Flexible boundary condit ions for a
Boussinesq-type wave model. Waves Conference, San Fransisco, United States of America.

Bosboom, J. and Stive, M.J.F. (2010). Coastal dynamics 1. Lecture notes CT4305 Part 1, Delft University of
Technology, The Netherlands.

Bowers, E.C. (1977). Harbour resonance due to set-down beneath wave groups. Journal Fluid Mechanics, 79,
p.71-92.

Brandi Mortensen, S. (2006). Numerical modeling of the surf zone dynamics in Waimea Bay. Msc. Thesis,
Technical University of Denmark, Denmark.

Cummins, W.E. (1962). The impulse response function and ship motions. International Symposium on Ship
Theory, Hamburg, Germany, nr. 8 .

De Bont, J., (2009). Validation of a numerical model for motions of ships moored with MoorMaster TM units, in a
harbour under influence of ocean waves. Msc. Thesis, Delft University of Technology, The Netherlands.

De Girolamo, P, (1996). An experiment on harbour resonance induced by incident regular waves and irregular
short waves. Coastal Engineering, 27, p.47-66.

De Jong, M.P.C., Borsboom, M.J.A. and Dekker J. (2009). Calculation of low -frequency waves in shallow water
and comparison to common practice in diffraction methods. Proceedings of the 28th International Conference on
Ocean, Offshore and Arctic Engineering, Honolulu, Hawaii, USA.

De Jong, M.P.C., Borsboom, M.J.A., De Bont, J.A.M, and Van Vossen, B. (2011). Evaluation of an extended
operational Boussinesq-type wave model for calculating low -frequency waves in intermediate depths. Accepted
for presentation at the ASME 2011 30 th International Conference on Ocean, Offshore and Arctic Engineering
OMAE2011, June 19-24, 2011, Rotterdam, The Netherlands.

Deltares (2008). Triton 1.0: User and technical manual. Delft, The Netherlands.

DHI (2005). Mike21 Boussinesq wave module: Short description. Hø rsholm, Denmark.

DHI (2006A). Mike21 Boussinesq wave module: User guide. Hø rsholm, Denmark.

DHI (2006B). Mike21 Toolbox: User guide. Hø rsholm, Denmark.

Gallagher, B. (1971). Generation of surfbeats by non-linear wave interactions. Journal Fluid Mechanics, 49, p.1-20.

Herbers, T.H.C., Elgar, S., and Guza, R.T. (1994). Infragravity-Frequency (0.005-0.05Hz) Motions on the shelf.
Part I: Forced waves. Journal of Physical Oceanography, 24, p.917-927.

Final 80
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Herbers, T.H.C, Elgar, S., and Guza, R.T. (1995A). Generation and propagation of infragravity waves. Journal
Geophysical Research, 100, p.24863-24872.

Herbers, T.H.C, Elgar, S., Guza, R.T., and O‟Reilly, W.C. (1995B). Infragravity-Frequency (0.005-0.05Hz) Motions
on the shelf. Part II: Free waves. Journal of Physical Oceanography, 25, p.1063-1079.

Holthuijsen, L. (2007). Waves in oceanic and coastal waters. Cambridge University Press.

Huntley, D.A., Guza, R.T., and Thornton, E.B. (1981). Field observations of surf beats Part 1: Progressive edge
waves. Journal Geophysical Research, 86, p.6451-6466.

Journée, J.M.J., and Massie, W.W. (2001). Offshore hydromechanics, First edition, Delft University of Technology,
The Netherlands.

Kofoed-Hansen, H., Kerper, D.R., Sø rensen, O.R., and Kirkegaard, J. (2005). Simulation of long wave agitation in
ports and harbours using a time-domain Boussinesq model. Proceedings of Fifth International Symposium on
Ocean Wave Measurement and Analysis-Wave, Madrid, Spain.

Kofoed-Hansen, H., Sloth, P., Sø rensen, O.R., and Fuchs, J. (2000). Combined numerical and physical modelling
of seiching in exposed new marina. Proceedings of 27th International Coastal Engineering Conference, Sydney,
Australia

Korsmeyer, F.T, and Sclavounos, P.D., (1989). The large -time asymptotic expansion of the impulse response
function for a floating body. Applied Ocean Research, Vol. 11, No. 2.

Longuet-Higgings, M.S. and Stewart, R.W. (1962). Radiation stress and mass transport in gravity waves, wit h
application to „surf beats‟. Journal Fluid Mechanics, 13, p.529-562.

Madsen, P.A., (1983). Wave reflection from a vertical permeable wave absorber. Coastal Engineering, 7, p.381-
396.

Madsen, P.A., Murray, R., and Sø rensen, O.R., (1991). A new form of Boussinesq equations with improved linear
dispersion characteristics. Part 1. Coastal Engineering, 15, p. 371-388.

Madsen, P.A., and Sø rensen, O.R., (1992). A new form of Boussinesq equations with improved linear dispersion
characteristics. Part 2: A slowly-varying bathymetry. Coastal Engineering, 18, p. 183-204.

Madsen, P.A., and Sø rensen, O.R., (1993). Bound waves and triad interactions in shallow water. Ocean
Engineering, 20, p. 359-388.

Madsen, P.A., Sø rensen, O.R., and Schäffer, H.A, (1997). Surf zone dynamics simulated by a Boussinesq type
model. Part I: Model description and cross-shore motion of regular waves. Coastal Engineering, 32, p. 255-287.

Madsen, P.A., Sø rensen, O.R., and Schäffer, H.A, (1997). Surf zone dynamics simulated by a Boussinesq type
model. Part II: Surf beat and swash zone oscillations for wave groups and irregular waves. Coastal Engineering,
32, p. 289-320.

Madsen, P.A., Bingham, H.B., and Schäffer, H.A., (2003). Boussinesq -type formulations for fully nonlinear and
extremely dispersive water waves: derivation and analysis. Proceedings of the Royal Society London, Part A, 489,
p. 1075-1104

Mahmoodian, M. (2009). Impact assessment of a new wave energy converter, Anaconda. Msc. Thesis. University
of Southampton.

Mei, C.C. and Agnon, Y. (1989). Long-period oscillations in a harbour induced by incident short waves. Journal
Fluid Mechanics, 208, p. 595-608.

Munk, W.H. (1949). Surfbeats. Trans. Am. Geophys. Union 30, p.849-854.

Muttray, M., Oumeraci, H. and Ten Oever, E. (2006). Wave reflection and wave run -up at rubble mound
breakwaters. International Conference on Coastal Engineering, San Diego, 2006.

OCIMF (1997). Mooring equipments guidelines (Second edition). Oil Companies International Marine Forum.
London: Whit erby Seamanship International.

Final 81
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

OCIMF (2008). Mooring equipments guidelines (Third edit ion). Oil Companies International Marine Forum. London:
Whit erby Seamanship International.
Peregrine, D.H. (1967). Long waves on a beach. Journal Fluid Mechanics, 27, p. 815-827.

PIANC (1995). Criteria for movements of moored ships in harbours: a practical guide. Supplement to bulletin no.
88. Report of working group PTC II-24.

Pinkster, J.A. (1980). Low frequency second order wave exciting forces on floating structures. PhD. thesis, Delft
University of Technology, The Netherlands.

Pinkster J.A. (1995). Hydrodynamic interaction effects in waves. Proc. 5th Int. Offshore and Polar Engineering,
The Hague, Vol III, p.414-419.

Rabinovich, A.B., (2009). Handbook of coastal and ocean engineering Kim Y, World Scientific.

Reniers, A.J.H.M., Roelvink, J.A., and Thornton, E.B. (2004). Morphodynamic modeling of an embayed beach
under wave group forcing. Journal of Geophysical Research, 109, C01030.

Rosa-Santos, P. Veloso-Gomes, F., Taveira-Pinto, F., Guedes-Soares, C., Fonseca, N., Alfredo Santos, J., Paulo
Moreira, A., Costa, P., and Brogueira-Dias, E. (2008A). Physical model study of the behavior of an oil tanker
moored at a jetty. Proc. on the 2 nd International Conference on the Application of Physical Modeling to Port and
Coastal Protection, Coastlab 2008, Bari, Italy.

Rosa-Santos, P., Veloso-Gomes, F., Taveira-Pinto, F., Brogueira-Dias, E., and Guedes-Lopes, H. (2008B).
Improving operational conditions at Leixões oil terminal, Portugal. COPEDEC VII, Dubai, UAE.

Rosa-Santos, P., Veloso-Gomes, F., Taveira-P into, F., Guedes-Soares, C., Fonseca, N., Paco, A, Paulo Moreira, A.,
Costa, P., Malheiros, P., and Brogueira-Dias, E. (2009) Influence of the use of mooring line pre -tension on the
behavior of a moored oil tanker. International Conference in Ocean Engingeering (ICOE2009), IIT Madras,
Chennai, India.

Rosa-Santos, P., Veloso-Gomes, F., Taveira-Pinto, F., and Brógueira-Dias, E. (2010). Physical Modelling of Leixões
Oil Terminal, Portugal. Port Infrastructure Seminar, Delft, The Netherlands.

Sand S.E., (1982). Long waves in directional seas. Coastal Engineering, 6, p.195-208.

Schäffer, H.A. (1993). Infragravity waves induced by short -wave groups. Journal Fluid Mechanics, 247, p.551-588.

Schäffer, H.A. and Sø rensen, O.R (2006). On the internal wave generation in Boussinesq and mild-slope
equations. Coastal Engineering 53, p. 319-323

Sempere, J. M. (2006). Study of the swell conditions affecting the moored ships in Coega Harbour (Cape Town,
South-Africa) and proposal for improving the infrastructures. Msc. Thesis, Delft University of Technology, The
Netherlands.

Sø rensen, O.R., Schäffer, H.A. and Madsen, P.A. (1998). Surf zone dynamics simu lated by a Boussinesq type
model. III. Wave-induced horizontal nearshore circulations. Coastal Engineering 33, p. 155-176.

Symonds, G., Huntly, D.A., and Bowen, A.J. (1982). Two-dimensional surf-beat: long wave generation by a time -
varying break point. Journal Geophysical Research, 87, p.492-498.

Taveira-Pinto, F., Veloso-Gomes, F., Rosa-Santos, P., Guedes-Soares, C., Fonseca, N., Alfredo Santos, J., Paulo
Moreira, A., Costa, P., and Brogueira-Dias, E. (2008). Analysis of the behavior of moored tankers. International
Conference in Ocean Engingeering (OMAE2008), Estoril, Portugal.

Tucker, M.J. (1950). Surf beats: sea waves of 1 to 5 min. period. Proceedings Royal Society Londen, serie A 202,
p.565-573.

Van Dongeren, A., Reniers, A., Battjes, J.A. and Svendsen, J. (2003). Numerical modeling of infragravity wave
response during DELILAH. Journal Geophysical Research, 108, 3288, doi:10.1029/2002JC001332

Van der Molen, W., Monárdez, P., and Van Dongeren, A., (2006A). Numerical simu lation of long-period waves and
ship motions in Tomakomai Port. Japan Coastal Engineering Journal, 48, p. 59-79.

Final 82
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Van der Molen, W. (2006B). Behaviour of moored ships in harbours. PhD. thesis, Delft University of Technology,
The Netherlands.

Van der Molen, W., and Wenneker, I., (2008). Time-domain calculation of moored ship motions in nonlinear
waves. Coastal Engineering, 55, p. 409-422.

Van der Molen, W., (2009). Mike21-Harberth: Moored ship response modelling in harbours. User guide.

Van der Molen, W. and Moes, H. (2009). General characteristics of South African ports and the safe mooring of
ships. Proceedings of the 28th Southern African Transport Conference p.308-315.

Van der Molen, W., Rossouw, M., Phelp, D., Tulsi, K., and Terblanche, L. (2010). Innovative technologies to
accurately model waves and moored ship motions. CSIR 3rd Biennial Conference 2010. Science Real and
Relevant, CSIR International Convention Centre, Pertoria, South Africa

Van Noorloos, J.C. (2003). Energy transfer between short wave groups and bound long waves on a plane slope.
MSc Thesis, Delft University of Technology, The Netherlands.

Van Oortmerssen, G. (1976). The motions of a moored ship in waves. PhD thesis, Delft University of Technology,
The Netherlands.

Veloso-Gomes, F. Taveira-Pinto, F., Rosa-Santos, P., Brogueira-Dias, E., and Guedes-Lopes, H. (2005). Berthing
characteristics and the behavior of the oil terminal of Leixões Harbour, Portugal. Marine Heritage and Modern
Ports Wit-Press.

Voogt, A., Bunnik, T., and Huijsmans, R.H.M. (2005). Validation of an analytical method to calculate wave
setdown on current. 24th International Conference on Ocean, Offshore and Artic Engineering, 2005, Halkid iki,
Greece.

Waals, O.J. (2009). On the application of advanced wave analysis in shallow water model testing (wave splitting).
28 th International Conference on Ocean, Offshore and Artic Engineering, 2009, Honolulu, Hawaii, U.S.A..

Wenneker, I, Borsboom, M, Pinkster, J.A., Weiler, O. (2006). A Boussinesq -type wave model coupled to a
diffraction model to simulate wave-induced ship motion. 31th PIANC Congress, Estoril, Portugal.

Woo, S.B. and Liu, P.L.F. (2004). Finite -Element Model for Modified Boussinesq Equations. II: Applications to
Nonlinear Harbor Oscillations. Journal of Waterway, Port, Coastal and Ocean Engineering, 130, p. 17 -28.

Wu, J.K., and Liu, P.L.F. (1990). Habour excitations by incident wave groups. Journal Fluid Mechanics, 217,
p. 595-613.

Final 83
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 84
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendices

Appendix A: Harberth formulations

Appendix B: Physical model results

Appendix C: TRITON simulations

Appendix D: Modification of model set-up boundary conditions

in Boussinesq-type wave model TRITON

Appendix E: Alternative resonance mechanism

Final 85
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Final 86
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendix A. Harberth formulations

Harberth considers the ship in a fluid that is incompressible, homogeneous and irrotational. A
definition sketch of the floating body in the computational domain is given in Figure A- 1. For further
description a right-handed coordinate system is used. An explanation of the symbols is given in Table
A- 1.

Symbol Designation
Computational model domain
Sea floor
Fluid surface
Wetted surface of ship hull
 Waterline
W Wetted surface of fixed structures
Figure A- 1: Computational domain Harberth Table A- 1: Harberth symbols

The fluid flow around the floating body is described by mea ns of a velocity potential. T he velocity
potential can be described with:
6
   I   S   j Eq. A- 1
j 1
In which:
 = Total potential [m2/s]

I = Incident wave potential [m2/s]

S = Scattered wave potential [m2/s]
j = Radiation potential in mode j [m2/s]

Radiation problem
The hydrodynamic coefficients can be obtained by solving the radiation problem. T he radi ation
problem is treated for the body making an impulsive movement in initially sill water, according to:

X j (t )   (t ), j  1, 2,...6 Eq. A- 2

The radiation potential is decomposed into a impulsive and a time-var ying par t:

 j ( x, t )   j ( x) (t )   j ( x, t ) Eq. A- 3

In which:
j = Impulsive par t of the radiation potential [m2/s]
j = Time-varying par t of the radiation potential [m2/s]

 (t ) = Dirac delta function (unit impulse function)

The potentials have to fulfill the boundary conditions according to Table A- 2. BC [I] follows from the
assumption of incompressible and irrotational flow. The dynamic boundary condition (BC [II]) applied
at the free surface ensures that the pressure at the free surface equals the atmospherically pressure.
BC [III] follows from the assumption that no water particles may leave the free surface. BC [IV] and
[V] are no leak conditions, such that the velocities at the applied boundaries are zero. According to BC
[VI] no fluid can cross the ship hull boundar y by setting the velocity of the fluid equal to the velocity
of the body boundar y in the normal direction of the body surface. T he radiation conditions (BC [VII]
and [VIII]) ensures that the sources radiate waves, instead of absorbing them and that the potentials
are outgoing at an infinitely large distance from the oscillating body.

Final 87
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Number: Boundary condition: Applied at: Type of boundary condition:

[I] 2  j 2  j 2  j Laplace or continuity condition
2 j
    0
x 2 y 2 z 2
[II]  j 0 z=0 Linearized dynamic free surface
condition
[III] 2  j  j z=0 Linearized kinematic free surface or
 g 0 Cauchy Poisson condition
t 2 z
[IV]  j z = -h Neumann / no leak condition
0 W and
n
[V]  j z = -h Neumann / no leak condition
0 W
n
[VI]  j Neumann condition
 nj
n
[VII]
 
Lim  j  0
R 
R  Dirichlet or radiation condition

[VIII] Lim    0
j R  Dirichlet or radiation condition
R 
Table A- 2: Boundary conditions radiation problem.

The applied Green functions will not be presented, for these functions one is referred to Van der
Molen (2006B). The hydrodynamic forces can be obtained from pressure integration over the wetted
hull. T he wetted hull is divided in such a number of panels, that the velocities and pressure at each
panel is assumed to be constant. Integration over the impulsive par t of the radiation p otential
provides the added mass coefficients, while integration over the time-varying part of the radiation
potential leads to the impulse response functions.

Akj      j nk dS , k  1, 2,...6 Eq. A- 4

 j k
K kj (t )     n dS , k  1, 2,...6 Eq. A- 5
t
In which:
n k = Outward pointing normal vector
dS = Surface element

Incident and scattered wave computation

The hull of the ship is described by panels in such a way that the pressure and velocity is assumed to
be constant on each panel. Backward transformation of the depth-averaged velocity and pressure is
required to find the original incident velocity and pressure distributions over the water depth. The
formulations for both the velocities as well as the pressure are given as power series in Table A- 3.
The power series are given in the obtained fluxes from Mike21 BW, in which d =still water level and
h  d   . Subscripts x,y and t denote partial differentiation to space and time respectively.
The diffraction pr oblem is solved for the scattering of waves around the floating body. Scattering of
the incident wave ar ound other fixed str uctures in the vicinity of the floating body is assumed to be
included in the incident wave potential (which is obtained by the backward transformation as
described before). T he scattered potential have to fulfill the boundary conditions according to Table A-
4. The applied Green functions will not be presented, for these functions one is referred to Van der
Molen (2006B).

Final 88
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Taylor expansion of imported data: With:

1 P 1 1
u ( z )  u (0)  ( z  d )2 u (2) u (0)   dPxx  d x Px
2 h 6 3
P
u (2)   xx
d
1 Q 1 1
v( z )  v(0)  ( z  h)2 v(2) v (0)   dQyy  d y Qy
2 h 6 3
Qyy
v (2)  
d
1 P Q
w( z )  w(0)  ( z  d ) w(1)  ( z  d )3 w(3) w(0)  d x  d y
6 h h
w  u x  v y
(1) (0) (0)

w(3)  u x(2)  v (2)

y

1 1
p( z )  p (0)  ( z  d )2 p (2) p (0)   g  d 2 p (2)
2 2
p (2)

  tt (1  3BB )  BB g ( xx   yy )
 d
d P dy Q
 x ( t  3BB g x )  ( t  3BB g y )
d 2d d 2d
Table A- 3: Power series of the imported data from the Boussinesq wave model Mike21BW

Number: Boundary condition: Applied at: Type of boundary condition:

[I]  2 S  2 S 
2 S Laplace or continuity condition
2 S
    0
x 2
y 2
z 2
[II]  2 S  S z=0 Linearized kinematic free
 g 0 surface condition
t 2 z
[III]  I  S Neumann condition
 0
n n
[IV]  S z = -h and Neumann condition
0
n W
[V]
R 
 
Lim  S  0 R  Dirichlet or radiation condition

Table A- 4: Boundary conditions incident/scattered wave problem

BC [I] follows from the assumption of incompressible and irrotational flow. BC [II] follows from the
assumption that no water particles may leave the free surface. The normal velocity in the scattered
wave is the opposite of the normal velocity in the incident wave (see BC [III]), such that the boundary
condition V  n  0 at the hull of the ship is fulfilled. According to BC [IV] no fluid can cross the ship
hull boundary by setting the velocity of the fluid equal to the velocity of the body boundar y in the
normal direction of the body surface. The radiation condition (BC [V]) ensures that the source radiate
waves, instead of absorbing them and that the potential is outgoing at an infinitely large distance
from the oscillating body.

Final 89
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The first order wave forces, due to both incident and scattered wave, follow from integration of the
first order pressures over the mean submerged hull. The pressure p is obtained by applying the
Bernoulli equation:
p  p0  1
  gz      constant Eq. A- 6
 t 2

In which:
gz = Hydrostatic pressure [m2/s2]
 = Linear component of dynamic pressure [m2/s2]
t
1 = Quadratic pressure related to the second order forces [m2/s2]
 
2
The Froude-Krylov force can be calculated due to integration of pressures in the undisturbed incident
wave and the diffraction force is found by integration of pressures in the scattered wave. The total
first order wave force follows from summation of the Froude-Krylov force plus the diffraction force.
Second or der wave effects are included in the first order forces, since a nonlinear wave model is used
to compute the incident wave field. Other second order contributions are however not taken into
account. These remaining terms for the second order force and moment include the products of first
order quantities, due to relative wave height (first term RHS Eq. A-11 and Eq. A-12), second order
pressure (second term RHS Eq. A-11 and Eq. A-12), products of first order pressures and first order
body motions (third term RHS Eq. A-11 and Eq. A-12) and the contribution of the product of first
order rotations and first order inertia forces (fourth term RHS Eq. A-11 and Eq. A-12), see also
Pinkster (1980).

The general formulas to calculate the wave forces or moments are given by:
F     pndS Eq. A- 7
S

M     p( x  xG )  ndS Eq. A- 8
S
In which:
p = Fluid pressure [kN/m2]
n = Outward pointing normal vector of surface element [-]
dS = Surface element [m2]
S = Total wetted surface of the body [m2]
x = Coordinate of surface element [m]
xG = Coordinate of the centre of gravity [m]

F = Force [kN]

M = Moment [kNm]

Final 90
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

The first or der wave forces and moments due to the incident and scattered wave are given by:
  S 
F(O1 )     p I   ndS
 t  Eq. A- 9
   S
M(O1 )     p I   ( x  xG )  ndS
 t 
Eq. A- 10
And the second order wave forces and moments are given by:
1 1  
F(O2 )   g  (   3 ) 2 ndl          (O1 )  ndS
2  2 t 
 (O1 )  MX (O1 ) Eq. A- 11
1 1  
M(O2 )   g  (   3 ) 2 ( x  xG )  ndl         (O1 )  ( x  xG )  ndS
2   2 t  Eq. A- 12

 (O )  M (O )
1 1

In which:
  X  ( x  xG )  n = Total displacement [m]

3 = Displacement in ver tical direction [m]

 = Rotational displacement [rad]
dl = Line element of the waterline [m]

Final 91
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendix B. Physical model results

WP 1 Low freq. WP 1 Total
4 60

3
40
S [m2/Hz]

2
20
1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

4 60

3
40
S [m2/Hz]

2
20
1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
2 3

1.5
2
S [m2/Hz]

1
1
0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

2 3

1.5
2
S [m2/Hz]

1
1
0.5

0 0
0
0.020.01
0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]
Figure B- 1: Measured wave spectra.
Measured Tp=10 [sec]
Tidal level: d=20 [m].
Low frequency spectra: left. Measured Tp=14 [sec]
Total wave spectra: right. Measured Tp=18 [sec]

Final 92
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

8 50

40
6
S [m2/Hz]

30
4
20
2
10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

8 50

40
6
S [m2/Hz]

30
4
20
2
10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
2 3

1.5
2
S [m2/Hz]

1
1
0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

2 3

1.5
2
S [m2/Hz]

1
1
0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]

Figure B- 2: Measured wave spectra. Measured Tp=18 [sec] d=20 [m]

Two different wave periods. Measured Tp=18 [sec] d=16 [m]
Two different tidal levels. Measured Tp=10 [sec] d=20 [m]
Measured Tp=10 [sec] d=16 [m]

Final 93
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Surge Low freq. Surge Total

S [m2/Hz] 100 100

50 50

0 0
0 0.01 0.02 0.03 0 0.05 0.1
Sway Low freq. Sway Total

50 50
S [m2/Hz]

0 0
0 0.01 0.02 0.03 0 0.05 0.1
Heave Low freq. Heave Total

1 1
S [m2/Hz]

0.5 0.5

0 0
0 0.01 0.02 0.03 0 0.05 0.1
Roll Low freq. Roll Total
1 4
S [deg2/Hz]

0.5 2

0 0
0 0.01 0.02 0.03 0 0.05 0.1
Pitch Low freq. Pitch Total
0.2
S [deg2/Hz]

0.5
0.1

0 0
0 0.01 0.02 0.03 0 0.05 0.1
Yaw Low freq. Yaw Total
5 5
2
S [deg /Hz]

0 0
0.01 0
0.02 0.03 0 0.05 0.1
f [Hz] f [Hz]
Figure B- 3: Measured ship motions spectra.
Measured Tp=10 [sec]
Tidal level: d=20 [m].
High friction fenders. Measured Tp=14 [sec]
Measured Tp=18 [sec]

Final 94
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

5 6
x 10 ML1 Stern x 10 ML2 Stern
10 2
S [kN2/Hz]

5 1

0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
6 6
x 10 ML3 Breast x 10 ML6 Breast
2 2
S [kN2/Hz]

1 1

0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
6 6
x 10 ML4 Spring x 10 ML5 Spring
2 4
S [kN2/Hz]

1 2

0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
6 6
x 10 ML7 Head x 10 ML8 Head
2 2
S [kN2/Hz]

1 1

0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03

7 7
x 10 Fender 1 x 10 Fender 2
4 4
S [kN2/Hz]

2 2

0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
f [Hz] f [Hz]

Figure B- 4: Measured spectra mooring lines.

Measured Tp=10 [sec]
Tidal level d=20 meter.
High friction fenders. Measured Tp=14 [sec]
Measured Tp=18 [sec]

Final 95
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendix C. TRITON simulations

WP 1 Low freq. WP 1 Total
0.8 30

0.6
20
S [m2/Hz]

0.4
10
0.2

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.4 2

0.3 1.5
S [m2/Hz]

0.2 1

0.1 0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.4 1.5

0.3
1
S [m2/Hz]

0.2
0.5
0.1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]

Figure C- 1: Results of run 01 to run 04.

Measured
Sensitivity different reflection coefficients
partial reflecting boundaries. R01 Sim. Cr Bw=0.3 b=0.02
R02 Sim. Cr Bw=0.4 b=0.02
R03 Sim. Cr Bw=0.5 b=0.02
R04 Sim. Cr Bw=0.4 b=0.10

Final 96
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

3 40

30
2
S [m2/Hz]

20
1
10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

6 40

30
4
S [m2/Hz]

20
2
10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.8 2

0.6 1.5
S [m2/Hz]

0.4 1

0.2 0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.8 1

0.6
S [m2/Hz]

0.4 0.5

0.2

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]

Figure C- 2: Results of run 05 and run 06.

Verification partial reflecting boundaries Tp=10 [sec] Measured
for different wave periods. Tp=10 [sec] Sim. (R05)
Tp=14 [sec] Measured
Tp=14 [sec] Sim. (R02)
Tp=18 [sec] Measured
Tp=18 [sec] Sim. (R06)

Final 97
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

0.8 30

0.6
20
S [m2/Hz]

0.4
10
0.2

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.4 2

0.3 1.5
S [m2/Hz]

0.2 1

0.1 0.5

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.4 1

0.3
S [m2/Hz]

0.2 0.5

0.1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]

Figure C- 3: Results of run 07 to run 09.

Sensitivity generated waves. Measured
R02 Sim. Direct
R07 Sim.(O)1 BC
R08 Sim.(O)2 BC1
R09 Sim.(O)2 BC2

Final 98
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

0.8 30

0.6
20
S [m2/Hz]

0.4
10
0.2

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.4 3

0.3
2
S [m2/Hz]

0.2
1
0.1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.4 1.5

0.3
1
S [m2/Hz]

0.2
0.5
0.1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]

Figure C- 4: Results of run 10 and run11.

Additional simu lations Measured
Adaptation of breakwater head. R02 Original
Partial reflecting side wall I. R10 Adapted BW head
R11 SW I partial reflecting

Final 99
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

WP 1 Low freq. WP 1 Total

1 30

20
S [m2/Hz]

0.5
10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 5 Low freq. WP 5 Total

2 40

1.5 30
S [m2/Hz]

1 20

0.5 10

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1
WP 6 Low freq. WP 6 Total
0.4 3

0.3
2
S [m2/Hz]

0.2
1
0.1

0 0
0 0.01 0.02 0.03 0.04 0 0.05 0.1

WP 8 Low freq. WP 8 Total

0.4 1

0.3
S [m2/Hz]

0.2 0.5

0.1

0 0
0 0.01
0.02 0.03 0.04 0 0.05 0.1
f [Hz] f [Hz]
Figure C- 5: Results of run 12 to run 14.
Additional simu lations. Measured
Different configurations. R02 Original
Full reflecting side wall II. R12 SW II closed
Small basin.
R13 Small basin
Large basin.
R14 Large basin

Final 100
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendix D. Modification of model set-up boundary conditions in

Boussinesq-type wave model TRITON

Within TRITON a boundary condition procedure calculates the reflected wave, see section 4.2.2. From
previous studies it is know n that oblique model boundaries may cause numerical instabilities within
TRITON. These instabilities are due to discretization of the dispersion boundar y condition s along
oblique model boundaries. The effect of an instable model boundary is illustrated in Figure D- 1: and
Figure D- 2, in which the same time-serie was used under the same conditions except for the
descritization of the par tial reflecting boundar y.

A numerical instability in the velocity field is noticed in case of oblique model boundaries, which grew
in time. After longer period of simulation this instability dominates the total velocity field, see Figure
D- 2. By avoiding oblique model boundaries, no numerical instabilities were noticed in the flow field.
In case of a stable flow field, the applied model boundaries were defined parallel along grid lines, see
Figure D- 3. In simulations with applied oblique model boundaries flow velocities above 3 [m/s] along
the model boundaries in TRITON were calculated. T hese small instabilities finally formed a jet along
the model boundaries effecting the total flow field (simulations not presented in this thesis). The
occurrence of such a numerical instability along the model boundaries was already noticed in the case
of an uniform bottom with low energetic waves (simulations not presented in this thesis).

If the flow velocities are calculated incorrect, but far outside the area of interest, the results from the
Boussinesq-type wave model simulations can be used in further processing to calculate the wave
forces on the ship. If incorrect described fluxes and/or velocities are located in the area of interest
(the vicinity of the ship) this may lead to an incorrect description of the flow field around the ship and
hence an incorrect description of the wave forces on the ship.

Due to an alternative model set-up of the model boundaries in TRITON no numerical instabilities in
the flow field are observed. TRITON is robust enough to calculate the flow over a long period of
simulation.

Conclusion
For the simulations carried out with TRITON during this study the following can be concluded:
Generated numerical instabilities along the model boundaries are responsible for the generation of
eddies within the computational domain of TRIT ON. During longer period of simulation these eddies
may increase in size as well as velocity magnitude. The existence of these ver y large eddies with large
velocities amplitudes in the numerical model is physically not possible and is purely due to numerical
instability along the model boundaries.

Numerical instabilities along model boundaries in TRITON can be avoided or minimized. Adaptations in
the set-up of the model boundaries ensured that unwanted eddies were avoided during further
performed simulations within TRITON.

Final 101
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

!!

Without oblique model boundaries. With oblique model boundaries.

Surface elevations above. Velocit y field beneath. Surface elevations above. Velocit y field beneath.
Figure D- 1: Snapshots of TRITON after 42 minutes of prototype simulation.

Final 102
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Without oblique model boundaries. With oblique model boundaries.

Surface elevations above. Velocit y field beneath. Surface elevations above. Velocit y field beneath.
Figure D- 2: Snapshots of TRITON after 139 minutes of prototype simulation.

Use of stair case model boundaries (left) and oblique model boundaries (right) in computational domain TRITON.
Figure D- 3: Discretization of model boundaries in computational domain

Final 103
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Appendix E. Alternative resonance mechanism

The following comments were handed in by Dr. P. Rosa Santos from the Faculty of Engineering
University of Porto. These comments should be taken into account during further research of wave
modelling and/or moored ship response calculations.

Together with the transversal resonance mode mentioned in the dissertation, the following hypothesis should
also be analysed: resonance of the mass of water limited by the north breakwater, the old north breakwater
and the south breakwater. This mass of water may be considered as a triangular edge, with constant water
depth (conditions in the physical model).

For d=20 m water depth and a height of the triangular edge of about 650 m, the following results may be
obtained: a period of nearly 172s (0.0058Hz) for the first resonance mode considering some curvature at the
triangle summit and a period of about 142s (0.0070Hz) if a perfect triangular edge is considered. As the
triangular edge is reasonably wide it is expected that the associated response curve will be smooth,
corresponding to a poorly selective situation. Nevertheless, pure reflective non-resonant situations caused by
the corner between the two breakwaters may occur for a wider range of periods.

This mechanism may explain why energy measured i n WP5 at frequencies about 0.006 Hz is not very high. Long
waves reflected from the absorbing beach and “Prainha” (both are close to the left side wall of the wave basin)
will be subjected to diffraction phenomena, either to Matosinhos Beach and the wave maker (see red lines in
Figure E- 1).

In a previous study numerical simulations, for monochromatic long period waves, were carried out. Harbour
geometry, bathymetry and surroundings were taken into account (i.e. si mulation of real prototype conditions).
For a period of 80 s (0.0125 Hz) and 100 s (0.01 Hz) a nodal line is located nearby the centre of Berth “A”. So
WP6 & WP8 do not measure correctly this standing wave. As the long wave period increases, the node line
approaches the head of the north breakwater. For periods between 140 ~ 170s it should be placed nearly in the
base of the triangle represented in Figure E- 1 and measured by WP6 and WP8.

Praia absorvente perfurada

"Prainha"

TANQUE DE ONDAS 200 11

12
13
14

3 10

(28.0x12.0x1.2 m )
9

S9 3
4
5
6
125

5
S8 43 S6
125

42
182

Parede divisória em acrílico

S7
PRAIA
DISSIPADORA
DO
1023 TANQUE DE ONDAS
CONTROLO

Praia de Matosinhos
SALA

(adaptada)
DE

1100

S1-S4
S5
400
150

150

SISTEMA DE GERAÇÃO DE AGITAÇÃO MARÍTIMA

Figure E- 1: Resonant mechanism between North breakwater, old North breakwater and South breakwater.

Considering L= 350 ~ 500 m, d=20 m and the simplified expression for resonance of a closed basin, the periods
st
obtained for the 1 mode of resonance are between 50s (0.020Hz) and 71s (0.014Hz).
It is difficult to be sure about the r esonant mechanisms that occur (and are dominant) in the physical model.
When planning the second phase of the study, this kind of issues was taken into account. It is remar kable to
confirm that despite limitations and simplifications, resonant mechanisms identified in the model have
correspondence with the prototype.

After diffraction, most of the wave´s energy will be directed to the south br eakwater and then reflected to the
Berth “A” location (see repr esentation bellow). Then that energy will be re-directed to the south breakwater
and Matosinhos beach (effects not too much different from what happens in the prototype).

Final 104
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Only a small amount of the incident (or generated) l ong wave energy will be directly directed to the left side of
the wave tank. Part of this energy will be reflected in ‘Prainha’ (beach that exists in the prototype). The
remaining energy will enter the inner harbour basin (very simplified due to space limi tations), by the gap
between the two breakwaters, but only after being partially dissipated in the perforated beach installed there.
Morais, C. C.; Abecasis, F., 1978. Storm surge effects at Leixões. Laboratório Nacional de Engenharia
Civil, Memória Nº503, Lisboa - this paper was presented at Fourteenth Coastal Engineering
Conference, Copenhagen, 1974.
st nd
This reference reports an accident occurred at Berth “A” with a 137000 dwt oil tanker on the 21 and 22
January 1974. Long period waves of 2 to 4 minutes with amplitudes of up to 50 cm, probably amplified,
occurred. It was hypothesized that the situation was worse for periods near two minutes, because for the
conditions existing at the time, the resonant period of the triangular edge mentioned befor e was o f about 133
s. The oil tanker was unloading. In the worse situation the surge amplitude was of about 10 to 15 m and the
sway amplitude of about 3 to 4 m from the berth. With tug assistance those peak -to-peak amplitudes reduced
to 8 and 2 m, respectively.

Praia absorvente perfurada

"Prainha"
o
çã
ita
ag

TANQUE DE ONDAS 12
13
14
de

11

3 10

(28.0x12.0x1.2 m )
9

7
na

4
zo

1
da
ite
Lim

Parede divisória em acrílico

PRAIA
Limite da zona de expanção DISSIPADORA
P O 75 150 225 300 375 450 525 600 DO
TANQUE DE ONDAS
CONTROLO

Li
m Praia de Matosinhos
ite
SALA

da (adaptada)
DE

zo
na
de
al
L /4 im
en
taç
ão

SISTEMA DE GERAÇÃO DE AGITAÇÃO MARÍTIMA

Figure E- 2: Diffraction in physical model.

Almost all the low frequency energy measured by WP1 to 4 is measured in WP5. WP5 is also close to a beach.
In beaches, bound long waves are released and start to travel as free waves . Beaches may also generate low
frequency energy by a so called “breakpoint forcing mechanism”. According to the location of WP5, and taking
into account the conditions of the previous figure, some reflections from the south breakwater may also be
measured at the location of WP5. The low-frequency spectrums of WP5 show a high concentration of energy in
the range between 0.007 to 0.01 Hz. The lengths associated with those frequencies are between 1000 m and
st
700 m (1 mode of oscillation). The wave tank is 12 m wide (1200 m in prototype dimensions). Considering the
second mode of resonance, the possibility of WP 5 being close to an anti -node reduces considerably.
Most of the low frequency energy travels in the longitudinal direction. Low frequency energy may r each berth
“A” area by diffraction or after reflection on the south breakwater or Matosinhos Beach. In the analysis
presented in the dissertation it is considered that low frequency energy is trapped in the transversal direction.
Low fr equency energy travels, mainly, in the longitudinal direction. Diffraction will occur either in the
propagation to the berth (right > left) or from the berth (left > right).
st
Considering only the 1 mode of resonance, length associated with those critical ‘resonant’ frequenci es are
between 1000 and 700 m, which are difficult to fit transversally in an alignment including the berth “A”. In this
case WP5 & WP8 are not in a node. In practice partial standing wave systems occur. Therefore, WP placed
where the amplitude is minimal will record some energy.
For the triangular edge mentioned before and a d=20 m, the period of the first resonance mode was between
172 s (0.0058Hz) and 142 s (0.0070 Hz). As the triangular edge is reasonably wide, it is expected that the
associated response curve will be smooth, corresponding to a poorly selective situation.

Final 105
The behaviour of a moored oil tanker in the Port of Leix ões, Por tugal
With use of numerical models

Analysis of Figure 3-6 > results of WP8. Three peaks can be seen between 0.005 Hz and 0.01 Hz in the wave
spectra. Consider three partial standing wave systems associated with those three peaks. All may be explained
as a result of the edge mechanism described before (0.0058 Hz and 0.0070 Hz) and therefore can have
correspondence with reality. It is important to mention that the right side wall of the basin may also have some
effec t on that. below pr esents, as example, results of numerical simulations carried out for long period waves.
In the example results are for T=120s.

Figure E- 3: Monochromatic long wave computations for port of Leixões (T=120s).

Three ‘resonant’ situations could be associated with those three peaks. The first one, with a T~160 s (~0.0062
Hz), may correspond to a situation with a mode nearby the head of the breakwater. This standing wave is the
more easily measured by WP8. The second one, with a T~143 s (0.007 Hz) and a node between the head of the
north breakwater and the bow of the ship. A higher response is measured from the ship. Because the node is
more close to WP8 this ‘standing wave’ is not so easily measured. The third one, with a T~125 s (0.008 Hz) and
a nodal line in the space occupied by the ship’s bow. This hypothetical ‘standing wave’ is therefor e more
difficult to measure but leads to the bigger response from the ship. The natural period of oscillation of the
moored oil tanker in the surge direction is about 70 to 80 s (0.013 to 0.14 Hz), what may parti ally explain the
second observed peak in the surge spectra.

Sway motion should mainly be associated with reflections from the south breakwater (or Matosinhos beach),
but also the possible installation of partial standing wave systems between the two break waters. Smaller
lengths lead to higher resonance frequencies. The sway natural period of oscillation is also in the range of 70 to
80 s (0.013 to 0.14 Hz). Differ ences of mooring system stiffness longitudinally and transversally are not very
high. Some sway may also be due to coupling between surge and sway.

From Figure 3-6, it can be seen that the spectral energy density of surge decreases when approaching 0.01Hz
(from the left), however it seems that the relative importance of sway is very high around 0.01 Hz. Sway would
certainly be more easily excited if a transversal standing wave system wer e installed between the two
breakwaters. Only simple reflections (i.e. without resonance) from the south breakwater may significantly
contribute to the obtained res ults. Due to surge and sway coupling, large surge motions may also leads to
significant sway motions.

Long waves could pass the perforated dissipation beach, but suffer energy dissipation when doing that. The
rubble slopes installed inside that small basi n (in all sides except in the right side wall of the wave basin)
provided additional damping.

Final 106