You are on page 1of 176

The Impact of Wave Drift on the

Tow Resistance and Towing Stability


of a Towed Vessel
A wave drift estimation method assessment
and towing stability simulation
for course stable and unstable barges
M.H.G. (Mathieu) Baas
The Impact of Wave Drift on the
Tow Resistance and Towing Stability
of a Towed Vessel
A wave drift estimation method assessment
and towing stability simulation
for course stable and unstable barges

by

M.H.G. (Mathieu) Baas

in partial fulfilment of the requirements for the degree of

Master of Science
in Offshore & Dredging Engineering

at the Delft University of Technology,

Supervisor: Prof. dr. ir. R.H.M. Huijsmans Delft University of Technology


Thesis committee: Dr.-Ing. S. Schreier, Delft University of Technology
Dr. A.A. Kana, Delft University of Technology
Dr. ir. M. Cvetkovic, Delft University of Technology
Ir. G. Hong, Boskalis
Ir. P.H.A. Hendrickx, Boskalis

An electronic version of this thesis is available at http://repository.tudelft.nl/.

Cover picture: Union Bear + Union Princess tow the accommodation jack-up “Haven”. January 2016
Abstract
Long distance ocean towing is a commonly used transportation method in the oil & gas and renewable energy
industries. Barges with cargo, vessels and various offshore structures are towed from and to location. Boskalis
participates in this industry through its subsidiaries Smit and Fairmount Marine. From experience with these
transports, it was found that the preliminary calculations for determining the towing tug size and sailing
velocity do not match with the experiences offshore. Therefore these calculations are investigated.

In a preliminary study comparing the relative magnitude of current, wind and wave drift forces, the wave
drift is identified as a significant contributor to the total tow resistance and as the contribution with the most
uncertainty in its prediction. The literature shows that research into the impact of waves on the dynamic
behaviour and stability properties of the tow operation is lacking. Therefore wave drift is selected for further
research. To investigate this impact, the problem is split into three parts: wave drift estimation for a stationary
vessel, wave drift estimation for a sailing vessel and the analysis of the towing stability.

From the literature, eleven methods are identified for the wave drift estimation of a stationary vessel. These
are compared for the surge, sway and yaw force components. The different methods are computed using
the diffraction program Delfrac, the strip theory solver ShipMo and own implementations in Matlab. The
comparison is based on the theoretical background and computations of the methods. This serves as the
basis for the stationary tow resistance assessment and the wave drift estimation for a sailing vessel, which
uses the results of the stationary case as input. The most applicable methods found with respect to the tow
resistance are the far field method, the near field method and the Faltinsen short wave method.

With vessel velocities included, ten wave drift estimation methods for the sailing vessel are compared. The
results of the comparison serve as the basis for the sailing tow resistance and the wave drift force input for
the towing stability analysis. Specific attention is paid to the wave drift damping, which relates the stationary
wave drift to vessel velocities. It is implemented using Aranha’s formulations and the panel pressure output
of Delfrac. The most applicable methods with respect to the sailing tow resistance are the far field and near
field methods in combination with the wave drift damping. The Faltinsen short wave method is computed
faster but provides a less accurate approximation. With respect to the total tow resistance, the wave drift
contribution is found to make up roughly 25% of the total resistance, irrespective of forward velocity. In order
to include the wave drift into the towing stability analysis, the far field method with wave drift damping is
chosen for the surge and sway forces. For the yaw moment, the near field method with wave drift damping is
found the most suitable.

The dynamic behaviour of the tow operation is examined for two barges: one course stable and one course
unstable barge. Their towing stability properties are assessed without waves present and for head waves
conditions. By comparing these different conditions, the impact of the waves is identified and quantified. It
was found that the head waves do not affect the stability of the course stable barge assuming the towing tug
has sufficient available towing force. For the course unstable barge, the main stabilising effect is identified as
the extra resistance due to the head waves and not the direct wave drift moment. The relative contribution
of the waves with respect to the wind and current shows that the head waves have a significant stabilising
contribution to the towing stability although the specific relative contributions are highly dependent on the
environmental conditions. In the investigated case and for regular waves with ω > 0.65 the waves have a
beneficial effect for the course unstable barge. They either reduce the path width of the fishtailing motion or
dampen these oscillations into a stable towing position. This was verified by simulations in the time domain.
Irregular waves also promote the stability: they stabilise the unstable barge in the entire range of physically
relevant wave periods, irrespective of towline length.

Overall, the contribution of the wave drift to the tow resistance is significant, especially while sailing. Mainly
due to this increased resistance, head waves are beneficial for the towing stability of a course unstable barge,
both in regular and irregular sea states. This increased insight reduces the uncertainty in the tow resistance
prediction and contributes to the understanding of the towing stability properties of unstable barges.

iii
Acknowledgements
Before you start with your graduation thesis, everybody tells you to find a subject you are really enthusiastic
about. If you do not, it will become one of the hardest projects you will ever work on. Luckily, I found a subject
which I enjoyed thoroughly and I believe this was vital for the successful completion of it.

I knew right from the start of the Offshore & Dredging Engineering Master’s degree I wanted to follow the
floating structures track. I am glad I made that choice. I would like to thank my professor René Huijsmans
for his advice during my studies and especially concerning my semester in Bandung, Indonesia. I enjoyed
the hydrodynamics courses and when I had to choose my graduation thesis topic I knew I wanted to study
advanced hydrodynamics. In the beginning, the exact research topic was hard to define but I am proud of the
final result.

I want to thank my daily supervisor Sebastian Schreier for all the hard work and dedication. I learnt a lot from
you about approaching the problem from a academic viewpoint and really working thoroughly. Thank you.

Guoqing Hong, thanks for keeping me grounded and making me think about the practical use of the research.
This really allowed the thesis to become the whole story. I would also like to thank my supervisor Peter Hen-
drickx for his relaxed approach and giving me the space to pursue the thesis I wanted to write. Additionally
I would like to thank my fellow graduate student Peter, for keeping up with me when I wanted to pitch ideas
and started to think out loud. It helped me a lot in the process.

Finally I would like to thank my family and Eefje. Your support during my studies and especially during my
graduation project are greatly appreciated.

M.H.G. (Mathieu) Baas


Delft, July 2017

v
Contents

Abstract iii
Acknowledgements v
List of Figures xi
List of Tables xiii
Nomenclature xv
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem description and research goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Solution approach and research objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Road map and thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Literature Review 9
2.1 Towing stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.1 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.3 Verification equation of motion and stability analysis . . . . . . . . . . . . . . . . . . . 15
2.2 Environmental forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Stationary vessel wave drift forces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.1 3D Potential theory methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.2 2D Strip theory methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.3 Approximate methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4 Impact forward velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Sailing vessel wave drift forces / Added wave resistance. . . . . . . . . . . . . . . . . . . . . . 26
2.5.1 3D Potential methods - Wave drift damping . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5.2 2D Strip theory methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.3 Approximate methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.6 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3 Methodology 33
3.1 Requirements wave drift for tow resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Requirements wave drift for stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Methodology wave drift estimation method comparison . . . . . . . . . . . . . . . . . . . . . 36
3.4 Methodology towing stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 The simulation models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.6 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4 Stationary Vessel Wave Drift Assessment 45


4.1 Verification of the estimation method computations . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.1 Application assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.2 Updated validity boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Estimation comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1 Stationary vessel surge wave drift force . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.2 Stationary vessel sway wave drift force . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.3 Stationary vessel yaw wave drift moment . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Tow resistance assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.4 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

vii
viii Contents

5 Sailing Vessel Wave Drift Assessment 55


5.1 Verification of the estimation method computations . . . . . . . . . . . . . . . . . . . . . . . 56
5.1.1 Application assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1.2 Updated validity boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.1.3 Wave drift damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.1.4 Wave encounter frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.2 Estimation comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2.1 Sailing vessel surge wave drift force . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2.2 Sailing vessel sway wave drift force. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2.3 Sailing vessel yaw wave drift moment . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.3 Tow resistance assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4 Multi-criteria analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.4.1 MCA sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.4.2 Most suitable estimation method(s) for the stability analysis . . . . . . . . . . . . . . . 72
5.5 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

6 Towing Stability Analysis 73


6.1 Stability assessment without waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.1.1 Equilibrium positions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.1.2 Towing stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.1.3 Time domain simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.2 Wave drift force components Barge 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.3 Stability assessment with waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.3.1 Equilibrium positions regular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3.2 Towing stability regular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3.3 Time domain simulation regular waves . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.3.4 Equilibrium positions irregular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.3.5 Towing stability irregular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.4 Assessment impact of the waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.5 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7 Discussion 99
7.1 Stationary vessel wave drift assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.1.1 Stationary wave drift tow resistance contribution . . . . . . . . . . . . . . . . . . . . . 99
7.2 Sailing vessel wave drift assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.2.1 Sailing wave drift tow resistance contribution . . . . . . . . . . . . . . . . . . . . . . . 101
7.2.2 Sailing wave drift towing stability contribution . . . . . . . . . . . . . . . . . . . . . . 103
7.3 Towing stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3.1 Fishtailing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3.2 Damping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3.3 Minimum required wave drift for stability . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.3.4 Slowly-varying part of the wave drift . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.4 Uncertainties in the results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.4.1 Theoretical uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.4.2 Input uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.4.3 Numerical uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.5 Chapter summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

8 Conclusions and Recommendations 115


8.1 Conclusions stationary tow resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.2 Conclusions sailing tow resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.3 Conclusions towing stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.4.1 Wave drift estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.4.2 Towing stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Contents ix

A Background Case Studies 121


A.1 Case study: Egina FPSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
A.2 Case study: SMIT Barge 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
B Vessel Characteristics 125
B.1 Vessel geometry input. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.2 Delfrac mesh convergence study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.3 Center of gravity impact study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
C Implementation and Verification of the Computations 131
C.1 Wave drift for a stationary vessel implementation and verification . . . . . . . . . . . . . . . . 131
C.2 Wave drift for a sailing vessel implementation and verification . . . . . . . . . . . . . . . . . . 135
C.3 Towing stability analysis implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
D Individual Multi-criteria Scores and Sensitivity Analysis 145
Bibliography 151
List of Figures

1.1 Overview of a towing operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1


1.2 Overview tow resistance and relative contributions for the Egina FPSO case . . . . . . . . . . . . 4
1.3 Overview tow resistance and relative contributions for the Smit Barge 2 case . . . . . . . . . . . . 4
1.4 Roadmap / Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1 Available and required towing force curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9


2.2 The geometry of the towing system [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Definitions of stability: stable (top), unstable (middle) and marginally stable (bottom). From
Lee [44] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Comparison TD simulation Bernitsas and Kekrides [6] results and own implementation . . . . . 16
2.5 Comparison Phase Space Bernitsas and Kekrides [6] results and own implementation . . . . . . 16
2.6 Comparison Lee [44] results without skeg and own implementation . . . . . . . . . . . . . . . . . 17
2.7 Comparison Lee [44] results without skeg and own implementation: individual eigenvalues . . 17
2.8 Comparison Lee [44] results with skeg and own implementation . . . . . . . . . . . . . . . . . . . 18
2.9 Calculation relative velocity from Journee et al. [40] . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.10 Wave Drift Force Regions, taken from Grin [30] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.11 Delfrac wave drift surge force contributions Part I to IV with β = 15° . . . . . . . . . . . . . . . . . 23
2.12 Example of strip theory from Journee et al. [40] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.13 Illustration of the geometry for Faltinsen’s and Fujii and Takahashi’s short wave limits from [63] 25
2.14 Flow separation limits for ζa , ω and Uc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.15 The direct Delfrac output vs the pressure computed Delfrac output and the higher order mo-
ment contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.1 The irregular waves sea states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34


3.2 Overview of the stationary vessel wave drift estimation methods . . . . . . . . . . . . . . . . . . . 37
3.3 Overview of the sailing vessel wave drift estimation methods . . . . . . . . . . . . . . . . . . . . . 37
3.4 Visualisation of the estimation comparison process . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Visualisation of the stability analysis methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.6 Geometry of the barges used as test vessels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.1 Overview of the stationary vessel wave drift estimation methods . . . . . . . . . . . . . . . . . . . 45


4.2 Visualisation of the estimation comparison process . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Stationary surge force estimation method comparison for regular and irregular waves . . . . . . 49
4.4 Stationary sway force estimation method comparison for regular and irregular waves . . . . . . 51
4.5 Stationary yaw moment estimation method comparison for regular and irregular waves . . . . . 51
4.6 Stationary DNV approximations comparison for regular and irregular waves . . . . . . . . . . . . 52

5.1 Overview of the sailing vessel wave drift estimation methods . . . . . . . . . . . . . . . . . . . . . 55


5.2 Visualisation of the estimation comparison process . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3 The wave frequency dependency of the Wave Drift Damping (WDD) coefficients with β = 15°
for Barge 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4 The frequency ω and incoming angle of attack β dependency of the Wave Drift Damping (WDD)
coefficients for Barge 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.5 The encounter frequency shift of the regular wave QTFs . . . . . . . . . . . . . . . . . . . . . . . . 62
5.6 Sailing surge force estimation method comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 Sailing sway force estimation method comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.8 Sailing yaw moment estimation method comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.9 Sensitivity Analysis MCA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

xi
xii List of Figures

6.1 Visualisation of the stability analysis methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 73


6.2 Overview of the equilibrium positions without waves . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.3 Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, without waves . . . . . . . . 75
6.4 Maximum real eigenvalues l 00 dependency with U t ug = 2 m/s, x p0 = 0.5, without waves . . . . . . 76
6.5 Time Domain simulation of the towing operation, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5 . . . . . . . 77
6.6 Barge 2a Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5 . . . . . . . . . . . . . . . . . . 77
6.7 Barge 2b Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5 . . . . . . . . . . . . . . . . . 77
6.8 The frequency ω and incoming angle of attack β dependency of the Barge 2 Wave Drift Damping
(WDD) coefficients for Barge 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.9 Barge 2 surge wave drift force X w ave for u = 0, 1 and 2m/s . . . . . . . . . . . . . . . . . . . . . . . 80
6.10 Barge 2 sway wave drift force Y w ave for v = −1, 0 and 1m/s. . . . . . . . . . . . . . . . . . . . . . . 81
6.11 Barge 2 yaw wave drift moment N w ave for r = −0.015, 0 and 0.015r ad /s. . . . . . . . . . . . . . . 82
6.12 Overview of the equilibrium positions with waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.13 Barge 2a: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with waves . . . . . 84
6.14 Barge 2a: Parameter ζa and U t ug dependence for equilibrium position 4 . . . . . . . . . . . . . . 84
6.15 Barge 2b: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with waves . . . . . 86
6.16 Barge 2b: Parameter ζa and U t ug dependence for equilibrium position 4 . . . . . . . . . . . . . . 86
6.17 Barge 2a: Positive eigenvalues contour l 00 and ω dependency with U t ug = 2 m/s, x p0 = 0.5, with
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.18 Barge 2a: Position 3 positive eigenvalues contour with U t ug = 2 m/s, x p0 = 0.5, ψeq = −0.5°, with
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.19 Barge 2b: Positive eigenvalues contour l 00 and ω dependency with U t ug = 2 m/s, x p0 = 0.5, with
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.20 Time Domain simulation with waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 0.80 and
0.90 rad/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.21 Time Domain simulation with waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 1.00 and
1.50 rad/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.22 Barge 2a Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 0.80, ω = 1.00
and 1.50 rad/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.23 Barge 2a: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with irregular waves 93
6.24 Barge 2b: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with irregular waves 93
6.25 Yaw angle equilibrium position 4 x p0 dependency with U t ug = 0.5 m/s, with irregular waves . . . 93
6.26 Positive eigenvalues contour position 3 l 00 and T p dependency with U t ug = 2 m/s, x p0 = 0.5, with
irregular waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.27 Barge 2a: Comparison position 4 yaw angle equilibrium position x p0 dependency with U t ug = 2
m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.28 Barge 2a: Comparison position 3 positive eigenvalues with U t ug = 2 m/s, x p0 = 0.5 . . . . . . . . . 96
6.29 Barge 2a: w/o skegs, TD simulation without and with waves . . . . . . . . . . . . . . . . . . . . . . 96

7.1 Wind and Current coefficients for Barge 2a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100


7.2 Relative force contribution stationary vessel - T p dependency . . . . . . . . . . . . . . . . . . . . 100
7.3 Relative force contribution sailing vessel - T p dependency . . . . . . . . . . . . . . . . . . . . . . . 102
7.4 Relative force contributions sailing vessel - Ur el dependency . . . . . . . . . . . . . . . . . . . . . 102
7.5 TD simulations with wind and waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5 . . . . . . . . . . . . . . . . 104
7.6 Forces and moments of the TD simulations with wind and waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5 105
7.7 TD simulations of Barge 2a for fishtailing discussion, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m 107
7.8 TD simulations of Barge 2a for damping properties determination, U t ug = 2 m/s, x p0 = 0.5, l 00 =
2.5, ζa = 1.25 m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.9 Impact of the numerical error in the equilibrium position on the stability contours - Regular
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.10 Impact of the numerical error in the equilibrium position on the stability contours - Irregular
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
List of Tables

2.1 Slow Motion Derivatives from Bernitsas and Kekrides [6] and Lee [44] . . . . . . . . . . . . . . . . 16
2.2 Reflection coefficients for equation (2.33) [14] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.1 The irregular waves sea state parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34


3.2 Lower and upper boundaries for the tow resistance of the wave drift force variables . . . . . . . . 35
3.3 Lower and upper boundaries for the stability analysis of the wave drift force variables . . . . . . 36
3.4 MCA factors for the comparison of estimation methods . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 The Barge 1 properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.6 The Barge 2 properties (from Hong et al. [37]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7 Slow Motion Derivatives derived by Hong [37] and Nam [51] for a Barges 2a and 2b . . . . . . . . 43

4.1 Boundary values L/λ for the stationary vessel wave drift estimation methods . . . . . . . . . . . 47
4.2 Estimation method assumptions relating to the vessel characteristics . . . . . . . . . . . . . . . . 47
4.3 Stationary vessel surge force relative MSE with the FF method as base value . . . . . . . . . . . . 48
4.4 Stationary vessel sway force relative MSE with the FF method as base value . . . . . . . . . . . . 50
4.5 Stationary vessel yaw moment relative MSE with the FF method as base value . . . . . . . . . . . 50

5.1 Boundary values L/λ for the sailing vessel wave drift estimation methods . . . . . . . . . . . . . 57
5.2 Estimation method assumptions relating to the vessel characteristics . . . . . . . . . . . . . . . . 57
5.3 Sailing vessel surge force relative MSE with the FF method as base value . . . . . . . . . . . . . . 64
5.4 Sailing vessel sway force relative MSE with the FF method as base value . . . . . . . . . . . . . . . 66
5.5 Sailing vessel yaw moment relative MSE with the NF method as base value . . . . . . . . . . . . . 69
5.6 Sailing vessel MCA results for the surge force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.7 Sailing vessel MCA results for the sway force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.8 Sailing vessel MCA results for the yaw moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6.1 Updated lower and upper boundaries for the stability analysis of the wave drift force variables . 78

7.1 Damping properties of the two examined time domain simulations . . . . . . . . . . . . . . . . . 108

xiii
Nomenclature
Forces

αd R-function [−]

αU Effect of advance velocity [−]

αwi Incoming wind angle of attack [°]

αw Relative (wind) angle of attack [°]

F̄ Mean wave drift force [N ]

β Wave angle of attack [°]

δ Logarithmic decrement coefficient [−]

η wl Waterline flare angle [°]


δ
δβ Derivative to the incoming wave angle of attack [−]

δ
δω Derivative to the wave frequency [−]

γ JONSWAP peak enhancement factor [−]

γ Towline angle [°]

γeq Equilibrium towline angle [°]

Ŷi MSE base value

λ Wave length [m]

λi Eigenvalue [−]

µ Wave frequency difference [ r ad


s ]

ω Wave frequency [ r ad
s ]

ω0 Original Wave frequency [ r ad


s ]

ωe Wave encounter frequency [ r ad


s ]

ωs0 Natural slewing frequency [ r ad


s ]

ωsd Damped slewing frequency [ r ad


s ]

bi j Mean wave drift damping coefficient

ψ Yaw angle towed vessel [°]

ψeq Equilibrium yaw angle [°]


kg
ρ Sea water density [ m3 ]

kg
ρa Air density [ m3 ]

kg
ρw Sea water density [ m3 ]

xv
xvi List of Tables

θ Waterline angle [°]

~η Contour/Panel normal vector

r~c Contour coordinate vector

ζ Damping ratio [−]

ζa Wave amplitude [m]

a Vessel added mass [kg ]

B Vessels breadth [m]

Bf Bluntness coefficient [−]

c Celerity; Wave phase velocity [ ms ]

c Straight line stability coefficient [−]

CB Block Coefficient [−]

Cc Current force coefficient [−]

Cf v Kashiwagi et al. correction coefficient [−]

Ct s Total ship resistance coefficient [−]

Cw Wind force coefficient [−]

C oG z vertical center of gravity [m]

d (x) Local Draught [m]

EA Towline stiffness [N ]

f x , f y ,℘z Sectional force components

FT R Tow Resistance [N ]

F wd M Motion induced wave drift force [N ]

F wdR reflection induced wave drift force [N ]

F wd x Surge wave drift force [N ]

F wd y Sway wave drift force [N ]

F wd Wave drift force [N ]

Fx Surge force [N ]

Fy Sway force [N ]

Fn Froude number [−]

g Gravitational acceleration [m
s2
]

H Exposed vessel height above water [m]

Hs Significant wave height [m]

I1 Modified Bessel function of the first kind, order 1

I xx X-axis mass moment of inertia [kg · m 2 ]

Iyy Y-axis mass moment of inertia [kg · m 2 ]


List of Tables xvii

I zz Z-axis mass moment of inertia [kg · m 2 ]

k0 Original Wave number [ rm


ad
]

ke Wave encounter number [ rm


ad
]

K1 Modified Bessel function of the second kind, order 1

k xx X-axis radius of gyration [m]

ky y Y-axis radius of gyration [m]

k zz Z-axis radius of gyration [m]

L Towed vessel length (LPP) [m]

l Towline length [m]

l0 Initial towline length [m]

Lb Distance of the bow to 95% of maximum breadth on the waterline [m]

l eq Equilibrium towline length [m]

m Vessel mass [kg ]

m0 zeroth moment of the spectrum [m 2 ]

M ,0 Zeroth order moment (force) [N ]

M ,1 First order moment [N · m]

M ,2 Second order moment [N · m 2 ]

n Number of data entries

Nman Yaw manoeuvring moment [N · m]

NT Yaw towing moment [N · m]

N w ave Yaw wave drift moment [N · m]

N wd z Yaw wave drift moment [N · m]

N wi nd Yaw wind moment [N · m]

Nz Yaw moment [N · m]

P (ω, ω) In-phase component of the Quadratic transfer function QTF [−]

P (ω, ω)∞ QTF at infinite large frequency (short wave asymptotic limit) [−]

p 20 Second order pressure [ mN2 ]

PO Overshoot percentage [−]

Q Out-of-phase component of the Quadratic transfer function QTF [−]

R Reflection Coefficient [−]

r Yaw velocity towed vessel [ r ad


s ]

R() R-function [−]

R0 Still water resistance [N ]

R aw Added wave resistance [N ]


xviii List of Tables

r eq Equilibrium yaw velocity [ r ad


s ]

S ζ (ω) Wave spectrum [m 2 · s]

S F (µ) Low frequency wave drift force spectrum [m 2 · s]

Sb 0 Normalised tow line breaking strength [−]

T Tension force [N ]

T Vessels draft [m]

T Wave period [s]

Tp Peak wave period [s]

TQ (ω, ω + µ) Quadratic transfer function QTF [−]

Tx Spectrum period with x=1,2 [s]

Tcr i t critical towline tension [N ]

Teq Equilibrium towline tension [N ]

Tl Minimal Draught [m]

T p,emp Empirical Peak period [s]

T p l ower lower boundary Physical Peak period [s]

T p upper Upper boundary Physical Peak period [s]

u Surge velocity towed vessel [ ms ]

Uci Incoming current velocity [ ms ]

Uc Relative current velocity [ ms ]

u eq Equilibrium surge velocity [ ms ]

Ur el Relative vessel velocity (vessel velocity - current velocity) [ ms ]

U t ug Tug forward velocity [ ms ]

Uv Vessel velocity [ ms ]

U wi Incoming wind velocity [ ms ]

Uw Relative wind velocity [ ms ]

v Sway velocity towed vessel [ ms ]

v eq Equilibrium sway velocity [ ms ]

x Global x-coordinate towed vessel CoG [m]

xc Contour/Panel x-coordinate [m]

xt Global x-coordinate bow towed vessel CoG [m]

xG Center of gravity shift x-coordinate [m]

X man Surge manoeuvring force [N ]

xp Bow hawser attachment point with respect to the CoG [m]

XT Surge towing force [N ]


List of Tables xix

X w ave Surge wave drift force [N ]

X wi nd Surge wind force [N ]

y Global y-coordinate towed vessel CoG [m]

y 00 Initial sway offset in TD simulation [−]

yc Contour/Panel y-coordinate [m]

yt Global y-coordinate bow towed vessel CoG [m]

Yi MSE compared value

Yman Sway manoeuvring force [N ]

yp Bow hawser attachment point with respect to the CoG [m]

YT Sway towing force [N ]

Y w ave Sway wave drift force [N ]

Y wi nd Sway wind force [N ]

zc Contour/Panel z-coordinate [m]

Bi j Wave drift damping coefficient

F wd (0) Stationary wave drift force [N ]


1
Introduction
This thesis is the graduation project of Mathieu Baas conducted under the guidance of Boskalis. In this first
chapter the background of the field of study is presented. The motivation for the initial study leading up to the
research goal can be found in section 1.1. The precise problem description and research goal are presented
in section 1.2 while the objectives and the solution approach are presented in section 1.3. The road map of
the thesis and the outline of the chapters are presented in section 1.4.

1.1. Background
One of Boskalis’ businesses is marine transport. This business was recently grown with the acquisition of
Fairmount Marine, Smit and Dockwise, three marine transport specialists. Dockwise mostly focuses on the
Heavy Marine Transport, using semi-submersible vessels to perform dry (dock) transportation of other ves-
sels. Smit uses its expertise mostly for heavy lift and salvage operations and Fairmount Marine is focused on
wet transport: long-distance ocean towing of vessels ranging from FPSO’s to semi-submersible drilling rigs.

For long distance towing it is important that the operation is executed safely and, for commercial purposes,
on time. The correct estimation of the tow resistance due to environmental forces is vital for these goals as
well as their impact on the dynamic towing behaviour. For decades this has thus been a topic of interest in
research.

(a) Sherpa Tug Boat from Fairmount, Boskalis (b) Aerial sight of towing in action

Figure 1.1: Overview of a towing operation

In practice, Boskalis has experienced that the preparational calculations for towing operations do not match
with the experiences of the tug captains in charge of the operation offshore. Unfortunately, no data of these
discrepancies is available. Normal procedure during the engineering phase of a tow operation would be to
use preparational calculations to determine the required propulsion power of the towing tug and the transit

1
2 1. Introduction

time of the operation. Based on these estimates, costs are determined. Due to the discrepancies experienced,
the cost estimates can deviate and reduce profits. Therefore, research into these preparational calculations is
justified and commissioned.

Preparational calculations for a tow operation can generally be classified in two categories: tow resistance
and tow stability. The first determines the minimum required towing force for the operation while the second
determines if the towed vessel can be towed stably and thus safely.

To estimate the minimum required towing force, the total resistance of the tug and towed vessel must be
estimated. This resistance is dependent on the environmental conditions but also on the configuration of
the towing operation. The towing force should be large enough so the tug is able to keep control of the op-
eration in emergency situations. The minimum required force is estimated for a stationary vessel, where
environmental conditions are harsh, but the vessel is not sailing. Consequences of wrong predictions of the
resistance can result in dangerous conditions. Overestimation of the resistance will result in larger tugs being
deployed for the towing operation, resulting in loss of profit. Underestimation can be far more hazardous:
when there is not enough towing force available in emergency situations, the whole tow operation can turn
rogue and in the worst case result in loss of life. The tow resistance in sailing conditions is also vital, determin-
ing the velocity of the tow operation and thus the transit time. A proper prediction will optimise the profits
while a wrong prediction may even cost the project money. This shows that, for both stationary and sailing
vessels, a proper prediction of the tow resistance is necessary.

Conventionally, the tow resistance calculations are performed for a towing configuration where the towed
vessel lies in an optimal position; right behind the towing tug. During operations however, the towed vessel
will not necessarily sail right behind the tug or even in a straight line. This implies the operational config-
uration of the towed vessel can differ with the calculated resistance values and even that possibly unsafe
conditions can occur during the operation. It is thus vital that this is taken into account. The configuration
of the vessel is dependent on the course stability of the towed vessel and is studied under the term towing
stability.

The three environmental entities that influence the resistance and the towing configuration of a towed vessel
are the wind, the current and the waves. The relative influence on these quantities is important to be aware of
when further specifying the research topic. Therefore, both tow resistance and towing stability are subjected
to a preliminary investigation to estimate the relative influences of the relevant environmental entities.

Relative importance of the environmental conditions on the tow resistance


To investigate the influence of every component in the tow resistance, two test cases are examined. The first
test case is the tow operation of the Egina FPSO, which was towed from South Korea to offshore Nigeria. The
second test case is the tow operation of the SMIT Barge 2 with a jacket installed on top. A detailed description
of the cases can be found in appendix A. Here only the relevant results are discussed.

For the first case, model tank test data is compared to the approximate force calculations from literature. The
comparison is done for a sea state typical for sailing conditions. The current velocity is assumed zero since
this can be simulated by adding or subtracting the velocity from the forward velocity of the vessel. Wind and
current force calculations [55] are based on wind tunnel test coefficients. The wave drift force calculations
are based on the short wave approximation of the force [19]. These approximates are often used during the
engineering phase of a project but can sometimes underpredict the wave drift forces when ship motions are
large.

The results of the comparison are found in figure 1.2. Note that only one data set is available due to limited
test data. In sub-figure 1.2a, it can be seen that with a forward velocity of Ur el = 3 m/s the ratio between the
current data and current literature calculation is ∼ 1.4. Between the wave drift forces the ratio is ∼ 3.5. The
wind forces do not deviate because both model test data and literature calculations are based on the same
wind force coefficients.

Looking at sub-figure b it is very clear that both the wind and the wave drift forces cannot be neglected at
any forward velocity. It can be argued that the current force can be neglected at very low forward velocities
Ur el < 0.5 m/s. The wave drift forces are the dominant hydrodynamic force at low forward velocities and
the uncertainty in their prediction is highest. Improving the wave drift prediction will contributes to the
1.1. Background 3

total accuracy of the prediction. Taken into account that the wave drift force estimation method used can
sometimes underpredict the force, this argument is enhanced.

To further investigate the influence the second test case is examined. Unfortunately, no model test data
is available but the literature calculation methods are showcased in a different light. To compare, a semi-
empirical method used by Boskalis during the analysed project is evaluated. The same sea state is examined
as in the first case.

From examining sub-figure 1.3a, it is clear that the current and wind calculations are similar although the
literature calculation for the current is ∼ 1.5 times larger. The big difference between the calculations is due
to the wave drift force. From sub-figure b it is clear that the increase in wave drift forces according to the
literature methods dominates the total towing force. The wave drift force at Ur el = 2 m/s is according to the
literature method ∼ 4 times as large. This is a very significant difference.

From the two case studies, the calculation of the wave drift forces emerges as the topic where the most can
be gained. It is the dominant hydrodynamic force for low forward velocities, the waves have a higher relative
contribution to total force than the current force, and the waves have a larger uncertainty in their prediction.
This makes the wave drift force, especially with forward velocity, the environmental entity where most can be
gained with respect to the tow resistance.

Relative importance of the environmental conditions on the towing stability


The first test case and references from literature cited below are used to investigate the relative influence
of the environmental conditions on the towing stability. Towing stability has been studied since the first
commercial use of transporting goods through (barge) towing. More information on this research done can
be found in chapter 2.

In the tow stability test for the Egina FPSO it is shown that due to the presence of head waves the yaw angle
greatly reduces in comparison with the still water angle. It was shown that the yaw angle in still water had a
range between ψ = −6° ∼ 16° and in waves a range of ψ = −4° ∼ 2°. This is a significant reduction. This effect
can not yet be quantified based on the wave drift force parameters and a exact relation between this yaw angle
and the incoming waves is unknown. Knowledge about the parameter dependence is worth investigating
since it can provide insight into the process that reduces the yaw angle.

In literature, the impact of current on the towing stability has been studied by including it as relative velocities
in the operation. Using manoeuvring forces to describe the hydrodynamic forces this is possible. Wind forces
have been studied as well, and Fitriadhy [24] concluded that they had significant impact on the towing sta-
bility. This was further highlighted by Sinibaldi [66] who concluded that wind forces tend to reduce stability,
especially in head wind and stern wind conditions, although there are scenarios where they can improve the
towing stability. Fang [20] looked at the wave effects on the towing stability and concluded that they cannot
be neglected when examining towing stability. This was also stated by Varyani [73].

Concluding, both the wind and wave effects should be taken into account when doing a towing stability
analysis. However, more focus has been on the effect of wind on the towing stability than on the effect of
waves. As is clear from the towing tank tests, the effect of head waves can counteract the negative effect of
a wind force on the towing operation and it is thus worth investigating what the impact of the wave drift
phenomenon and its driving parameters are on the towing stability.
4 1. Introduction

Egina FPSO: Relative Forces, = -10°, Sea State #2


Egina FPSO: Forces, = -10°, Sea State #2 100
10-3
5 rel. Current Force
Total Force 90 rel. Wind Force
4.5 Current Force rel. Wave Drift Force
Wind Force 80 rel. Data Current Force

Percentage of Total Force [%]


4 Wave Drift Force rel. Data Wind Force
Data Current Force 70 rel. Data Avg Wave Drift Force
3.5 Data Wind Force
Data Avg Wave Drift Force 60
3
50
2.5
40
2

30
1.5

1 20

0.5 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(a) Tow resistance, ψ = −10° (b) Tow resistance percentage, ψ = −10°

Figure 1.2: Overview tow resistance and relative contributions for the Egina FPSO case

SMIT Barge 2: Relative Forces, = 0°, Sea State #2


SMIT Barge 2: Forces, = 0°, Sea State #2 100
10-3
1 rel. Current Force
Total Force 90 rel. Wind Force
0.9 Current Force rel. Wave Drift Force
Wind Force 80 Boskalis rel. Current Force
Percentage of Total Force [%]

0.8 Wave Drift Force Boskalis rel. Wind Force


Boskalis Total Force 70 Boskalis rel. Wave Drift Force
0.7 Boskalis Current Force
Boskalis Wind Force 60
0.6 Boskalis Avg Wave Drift Force
50
0.5
40
0.4

30
0.3

0.2 20

0.1 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(a) Tow resistance Boskalis and literature, ψ = 0° (b) Tow resistance percentage Boskalis and literature, ψ = 0°

Figure 1.3: Overview tow resistance and relative contributions for the Smit Barge 2 case
1.2. Problem description and research goal 5

1.2. Problem description and research goal


From investigating the relative importance of wind, waves, and current forces in the context of the towing
resistance and towing stability the importance of the wave drift force components emerges.

A proper wave drift force estimation is important for the tow resistance calculation since it is significant in
the area that is of interest for tow operations. In addition to that, the prediction of the wave drift force can be
improved most significantly of the three environmental conditions examined.

The impact of the wave drift force on towing stability is not widely studied, but as is clear from literature, it
cannot be ignored in the towing stability analysis. The wave drift can act as a stabilising factor in head to bow
quartering waves, counteracting possible wind effects acting on the towing operation or simply increasing
the stability. The magnitude of this phenomenon is uncertain and the effect of the different parameters is
unknown.

Research into the wave drift force components (surge, sway and yaw) for both the tow resistance and tow-
ing stability is justified. When towing in seaway, the waves produce extra forces and an extra moment on
the towed vessel changing the regular tow behaviour. Insight in the physical phenomenon taking place is
still missing but a full investigation into this physical phenomenon is out of scope for a masters graduation
project. Since analysing the tow resistance is necessary when investigating the towing stability, the focus of
this project is set on identifying the impact of the wave drift forces and moment on both. The impact on
the tow resistance, and the possible improvement of this calculation, together with the impact on the towing
stability, in particular the sway and yaw motions, will yield more insight into the behaviour of a tow operation
offshore. To be able to use the results of this study in further research, the impact of the waves need to be
quantified. Therefore, the goal of this thesis is formulated as:

To identify and quantify the impact of wave drift on the tow resistance and towing stability of a towed
vessel

1.3. Solution approach and research objectives


As the goal of this thesis is to identify the impact of the wave drift force on the tow resistance and the towing
stability, wave drift force estimation methods need to be assessed. Due to the complexities in predicting the
wave drift force, numerous uncertainties are present in the different estimations. Several methods are re-
ported in literature to estimate the wave drift force, with and without vessel velocities. The solution approach
features these different estimation methods.

The state-of-the-art for estimating the wave drift forces can be split into three main categories. These cat-
egories, either based on approximation techniques, potential theory or computational fluid dynamics can
differ significantly in precision, theoretical background and computational complexity. Of these three cate-
gories, the computational fluid dynamics (CFD) is the most precise but also the most complex and computa-
tionally heaviest. In this study, this method is not taken into account since the complexity of implementing
this method does not fit the time table of this thesis.

For the application of these wave drift estimation methods to the tow resistance and towing stability analy-
sis, the approximation techniques and the potential theory methods are examined. This is done for both a
stationary vessel, without current or vessel velocities, and a sailing vessel. In this last case, there are some
recent developments to include velocities directly in the potential theory calculations. However, since these
improved methods are still under development and are not easily accessible in the public domain they are
not taken into account in this study. Still, some attention is paid to these methods during the literature review.
In this thesis, more conventional approaches are used to arrive at results with vessel velocities included.

To identify the impact of the wave drift force, the multiple estimation methods are assessed and compared to
predefined criteria of application. On the basis of these comparisons, the most suitable method for the wave
drift force estimation in respect to the tow resistance and towing stability analysis will be identified. This
choice will be made from existing estimation methods. For both the stationary vessel, used in the prepara-
tional calculations for determining the minimum required tow force, and the sailing vessel, used to determine
the sailing velocity and transit time, these choices are made. To examine the towing stability, the most suit-
able wave drift estimation method for a sailing vessel is included in the stability analysis and its impact is
6 1. Introduction

assessed. Based on this approach the following research objectives have been identified:

Methodology:
1. Define the requirements for the wave drift force for the tow resistance;
2. Define the requirements for the wave drift force for application in a towing stability analysis;
3. Define the methodology for comparing various wave drift force estimation methods;
4. Define the methodology for the towing stability analysis;
5. Draft the simulation model used as the test case in this thesis.

Stationary Vessel Wave Drift Assessment:


6. Verify the computation of the wave drift estimation methods for a stationary vessel;
7. Comparison between these methods based on the defined methodology;
8. Identify the most suitable wave drift estimation method(s) for the tow resistance of a stationary vessel.

Sailing Vessel Wave Drift Assessment:


9. Verify the computation of the wave drift estimation methods for a sailing vessel;
10. Comparison between these methods based on the defined methodology;
11. Identify the most suitable wave drift estimation method(s) for the tow resistance of a sailing vessel;
12. Find the most suitable wave drift estimation method(s) to include in the towing stability analysis.

Towing Stability Analysis:


13. Assessment of the towing stability without the presence of waves;
14. Compute the wave drift force components for the test case model used in the stability analysis;
15. Assessment of the towing stability with waves present;
16. Identify the wave drift influence on the towing stability.

Overall:
17. Identification of the relative impact of wave drift contribution for the stationary and sailing vessel, as
well as for the towing stability.

1.4. Road map and thesis outline


The road map, presented in figure 1.4, defines the path that is taken to complete the goal of this thesis. This
figure also roughly describes the outline of this thesis.

The motivation behind the thesis is projected in the beginning of the road map. The need to improve the
preparational calculations emerges from the discrepancies seen between the calculations and the experience
offshore. Both the tow resistance and the towing stability were assessed in the initial relative importance
study. Based on a case study and literature, the focus is set on the impact of the waves on the tow resistance
and towing stability. To investigate the impact of the wave drift contributions, a methodology is developed.
For the tow resistance, both the stationary and the sailing vessel cases are assessed and the results are put
in perspective by comparing them with the initial relative importance study. The sailing vessel results are
also used as input for the towing stability analysis, were the impact of the waves on the stability of the tow
operation is assessed. These results will also be put in perspective by addressing their relative contributions.

The first chapter, introduction chapter 1, describes the background, motivation and the goal & objectives of
this thesis. Chapter 2 includes the literature review for both the towing stability and the wave drift estimation
methods. In chapter 3, the methodology and the requirements for the three parts of the research are defined.
The simulation model that is used is also introduced. The stationary vessel wave drift assessment is presented
in chapter 4, while the methods for the sailing vessel are assessed in chapter 5. In the towing stability chapter
6, the wave drift estimation method for the towing stability analysis as well as the towing stability assessment
itself is presented. Chapters 7 and 8 will feature the discussion and conclusions & recommendations drawn
from this research.

Various numerical programs will be used for the computations. MATLAB [70] will be used for most simula-
tions and for coupling the data obtained from other programs. Delfrac [16] and the ShipMo module in Safe-
trans [48] will be used for the computation of wave drift estimation methods. 3D modeling program FEMAP
[65] will be used to develop the model input for these programs.
1.4. Road map and thesis outline 7

Figure 1.4: Roadmap / Thesis outline


2
Literature Review
In this chapter the literature review on towing, towing stability and the relevant environmental entities is
discussed. The focus for the environmental forces are the wave drift forces, both without and with vessel
velocity. To sketch a complete overview, the preparational calculations required for a towing operation are
first discussed.

For a towing tug to be able to tow a towed vessel, it needs to have enough available pulling force to be able
to overcome the overall resistance of both the tug and towed vessel. This minimal required towing force
varies with the forward velocity and the position of the towed vessel in relation to the tug. To determine
the minimum required towing force of the towing tug, an emergency scenario is assumed. In this severe sea
state, the tug should still be able to hold the towed vessel in place, thus the overall resistance is equal to the
minimum required towing force.

Available Towing Force vs Total Tow Resistance


600
Available Tug Towing Force
Total Towed Vessel Resistance
500
Towing Force F TR [kN]

400

300

200

100

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Tug sailing velocity U tug [m/s]

Figure 2.1: Available and required towing force curves

Once the minimum required towing force is known and a towing tug is selected, the propulsion character-
istics can be used to determine the towing velocity. For this, the tow resistance must also be determined.
To do this, a less severe sea state is assumed and the total towed vessel resistance curve, as can be seen in
figure 2.1, is computed. As is clear, both the available tug towing force and the total towing resistance are
velocity dependent. Here, the plotted tug towing force has already taken the tug resistance into account and
represents the available towing force to be used for pulling the towed vessel. The required tow force, or the
total towed vessel resistance, is comprised of wind, current and wave drift forces and generally increases with
velocity. Since both curves are known, the operating forward velocity is determined by the intersection of the
two curves. From the tug velocity the tow operation transit times can then be obtained.

Besides the tow resistance part, it is also vital to know if the towed vessel can be towed in a safe configura-

9
10 2. Literature Review

tion and therefore its dynamic behaviour is important. This dynamic behaviour can be modelled and gives
valuable insight in how the tow operation behaves. The stability of the towed vessel can also be assessed us-
ing stability analysis. This field of study is called the towing stability. An overview of the tow operation, its
dynamic behaviour and the towing stability is featured in section 2.1.

To properly examine the tow resistance and towing stability, more insight in the environmental forces work-
ing on the towed vessel is necessary. A general description of these forces is given in section 2.2. As already
discussed in chapter 1, the wave drift forces are the focus of this thesis and are more intensively featured.
Wave drift forces on a stationary vessel are discussed in section 2.3. Since the towing operation includes for-
ward velocity the impact of this velocity is discussed in section 2.4. The wave drift forces with vessel velocities
are then discussed in section 2.5.

2.1. Towing stability


Towing stability has been an issue to account for ever since the first ships were towed in history. Strandhagen
was the first to describe the dynamics of a towing operation in a simplified manner in 1950 [69]. A barge with
a fixed towline length towed with a steady forward velocity was used to simulate the behaviour and analyse
the stability of such a configuration. Abkowitz published a book on the stability of ocean vehicles in 1969 [1]
where the theory of motion and its stability analysis is addressed. He addresses the equations of motion and
the methods to linearise them. The stability analysis is based on examining the critical points, equilibrium
points where the derivatives are zero, by disturbing with infinitesimal impulses and examining if the system
returns to its original position. Inoue et al. [39] first addressed the different parameter influences on the
course stability of a towing system in 1975. Increasing the distance between the bow hawser attachment
point x p and the center of gravity (CoG) of the ship and increasing the length of the towline l 0 both add to the
course stability. Also increasing the towed vessel resistance increases the course stability. They also showed
that with higher radii of gyration, the towing becomes more unstable. These identified parameters are used
in the investigation of the towing stability later on.

Next improvements in the towing stability research were made in the 1980’s. Bernitsas and Kekrides first de-
veloped a model describing the motion of a towed vessel with an elastic towing line [5]. They used slow mo-
tion derivatives to describe the manoeuvring force on a towed barge with respectively one [6] and two tugs
[4] and looked at the different stability configurations (critical points). Wind and waves were not included in
this research. Lee [44] reproduced this research with more focus on the influence of the towline length and
bow hawser attachment point parameters on the towing stability. A related problem, the weathervaning of
a single point moored vessel in a current, is in theory the same phenomenon since the current and forward
velocity are modelled in the same manner. Bernitsas and Papoulias used this analogy to examine single point
mooring system stability but with some form of wind and wave drift forces included [7, 56]. These develop-
ments in the towing stability analyses are the groundwork for the stability analyses performed with the wind
and wave effects.

In 1990, Kolthoff [42] completed a parametric analysis of course stability and found mainly three impor-
tant parameters that influenced the course stability. The earlier found towline length l 0 and longitudinal
bow hawser attachment point coordinate x p were two but also the transverse location of the bow hawser y p
was identified as an important parameter to optimise for course stability. This is especially interesting for
analysing stable towing points not right behind the towing tug.

Garza-Rios and Bernitsas [27] derived analytic stability criteria for the slow motion dynamics of a towing
operation in 1996. They based these criteria on the Routh stability criteria and they are mainly applicable in
the design stage of an operation. These criteria are a very quick check to see if the towed vessel is stable or
not.

Gokce [29] looked into short towline towing in 2002 with both three degrees of freedom (3DOF) models of
tug and towed vessel. The two most important findings were the impact of the towing force; sufficiently large
values will stabilise "any" system, and the impact of the manoeuvring capabilities of the towing tug. If the tug
is not capable of fast yaw control the system might not have stable towing conditions. The last conclusion is
only valid for short towlines l 0 ≤∼ 0.4 · L, where L is the towed vessel length. With longer towlines, the yaw
control velocity of the tug becomes less significant.
2.1. Towing stability 11

Varyani et al. [71, 73] focused more on the emergency towing of stranded vessels, as part of the European
SAFETOW project, and developed a simulation tool to help tug captains during operations. Focus was on the
impact of parameters on course stability while fishtailing instabilities were also studied. Wind and wave drift
forces were implemented in this paper. These implementations were approximations for stationary vessels
but show insight in the method of including them into the stability analysis.

In recent years a renewed interest has arisen for the dynamic simulations of these nonlinear systems. Fitri-
adhy et al. modelled both the towed vessel as well as the tug in 3DOF and compared nonlinear and linear
approaches. Initially only taking current into account in 2011 [23] while later also including the wind force in
2013 [24]. Focus was on the impact of the towline length l 0 and bow hawser attachment point x p on the mo-
tions of the towed vessel. Sinibaldi et al. [67] investigated the dynamic towing model as described previously
in 2013, but also taking a shift of the center of gravity into account and using a manoeuvring model capable
of handling large yaw angles. Wind effects were also taken into account. The wind direction proved to impact
the stability significantly. Sinibaldi and Bulian [66] extended this model in 2014 with a fourth degree of free-
dom, roll, to investigate its influence. Roll oscillations proved to become especially dangerous once stability
was lost and fishtailing occured. These studies have shown the impact of the wind on the towing stability
which is useful when examining the results of the towing stability analyses.

The impact of waves on the towing dynamics and stability have not always been of interest in research.
Varyani et al. included it in their models by fitting the forces on the basis of harmonic signals for generic
tanker designs [72]. Nakayama et al. [50] used time domain simulations in 2012 to simulate the impact of
waves on the vertical motion of the towed vessel and towline. Head waves were assumed in this case and es-
pecially the towline force variations due to the surge, heave and pitch motion were studied. In 2009, Fang and
Ju [20] proposed a full six degrees of freedom (6DOF) model that included manoeuvring and the influences
of wind and waves. Current was not included since it was argued that the influence was not significant for
the towing stability. The wave drift force including forward velocity was not taken into account but the zero
velocity result was taken as an approximate estimation. Their conclusions matched with previously found
research but insisted that wave effects should be taken into account when modelling.

Hong et al. [37] and Nam et al. [51], used a simple rectangular barge with raked bow and stern for towing
tank tests to compare two manoeuvring models: MMG model and the Cross-flow model. They validated
these methods in still water and also performed tests to show the impact of the waves on the towing stabil-
ity. They concluded that the increase in towline tension of the incoming head waves increased the towing
stability. Shigunov and Schellin [64] focused directly on the towline force using only a force balance instead
of the whole equation of motion. They included the wind from CFD computations and the mean wave drift
force computed with 3D potential theory on a stationary vessel in their balance. Five container ships were
compared with different wind and wave angles of attack. The research into the impact of the wave drift forces
on the towing stability already shows that their effect is significant and cannot be neglected. The distinct
parameter influence of the waves has not yet been investigated and this leaves room for this study. Compar-
ing results once the stability analysis is completed might validate insights. In the following subsections the
implementation of the towing stability theory is discussed in more detail.

2.1.1. Equations of motion


The mathematical model to describe the motions of the tug and towed vessel were, as previously mentioned,
generalised by Abkowitz [1]. The geometry of the towing system is presented in figure 2.2.
In the towing system, various parameters are present and part of those can be taken as state variables to be
used in the modelling. These parameters are presented below. Note that the incoming wave angle of attack
is defined as β = −ψ since head waves are assumed and that head waves are defined as β = 0°, contrary to
most literature.

u = Surge velocity towed vessel


v = Sway velocity towed vessel
r = Yaw velocity towed vessel
l = Towline length
γ = Towline angle tug
ψ = Yaw angle towed vessel
12 2. Literature Review

Figure 2.2: The geometry of the towing system [5]

x = x-coordinate CoG towed vessel, relative to tug


y = y-coordinate CoG towed vessel, relative to tug
xt = x-coordinate bow towed vessel, relative to tug
yt = y-coordinate bow towed vessel, relative to tug

The parameters, except the angles, are normalised in as shown in equation (2.1). The other properties are
normalised as follows, mass by a factor 1/2 · ρ · L 3 , mass moments of inertia by a factor 1/2 · ρ · L 5 , forces by a
factor 1/2 · ρ · L 2 ·U t2ug , moments by a factor 1/2 · ρ · L 3 ·U t2ug and time by a factor Ut ug /L.

u v rL 0 l t l 0 x 0 y 0 xt 0 yt
u0 = , v0 = ,r 0 = , ltl = , x = , y = , xt = , yt = (2.1)
U t ug U t ug U t ug L L L L L

The equations of motions describing the motion of the towed vessel in the horizontal plane are presented
in equations (2.2) to (2.4). It is assumed that the center of gravity (CoG) is located in the center of the local
coordinates of the towed vessel.
¡ 0 0
¢ 0 ¡ 0 0
¢ 0 0 0
¡ 0 0 0¢
u , v , r + X T0 + X wi
¡ 0 ¢¢ ¡ 0
nd u, v, r, ψ + X w ave ω, ζa , u, v, r, ψ
¡ ¡ ¢¢
m + a 11 u̇ − m + a 22 v · r = X man (2.2)
¡ 0 0
¢ 0 ¡ 0 0
¢ 0 0 0
¡ 0 0 0¢
u , v , r + YT0 + Y wi
¡ 0 ¢¢ ¡ 0
nd u, v, r, ψ + Y w ave ω, ζa , u, v, r, ψ
¡ ¡ ¢¢
m + a 22 v̇ + m + a 11 u · r = Yman (2.3)
0 0
) · r˙0 = Nman
0
u 0 , v 0 , r 0 + NT0 + N wi
0 0
nd u, v, r, ψ + N w ave ω, ζa , u, v, r, ψ
¡ ¢ ¡ ¡ ¢¢ ¡ ¡ ¢¢
(I zz + a 66 (2.4)

Where m 0 is the mass, I ψ 0


is the moment of inertia and a 0 the added mass. X man 0 0
, Yman 0
and Nman are the
0
manoeuvring forces described by slow motion derivatives, see equations (2.5) to (2.7), and X T , YT0 and NT0
0
are the towline forces transferred from the tug, see equations (2.8) to (2.10). The (optional) forces X wi nd
,
0 0 0 0 0
Y wi nd and N wi nd and X w ave , Y w ave and N w ave are the wind forces and wave drift forces acting on the towing
system.
0
¡ 0 0 0¢
X man u , v , r = −R 00 · u 02 + X v0 v · v 02 + X r0 r · r 02 (2.5)
0
u 0 , v 0 , r 0 = Y v0 · v 0 + Yr0 · r 0 + Y v0 v v · v 03 + Yr0r r · r 03
¡ ¢
Yman (2.6)
0
¡ 0 0 0¢
Nman u , v , r = N v0 · v 0 + Nr0 · r 0 + N v0 v v · v 03 (2.7)

Note that the slow motion derivative approach assumes small drift angles and becomes more inaccurate once
yaw angles increase. Also, the effects of the propeller or rudder are not included in these equations. Here, the
prime indicates the coefficient is normalised, R 0 is the still water resistance, the X terms describe the surge
coupling, the Y terms describe the sway coupling and the N terms describe the yaw coupling.

X T0 = T 0 · cos γ + ψ
¡ ¢
(2.8)
YT0 = −T 0 · sin γ + ψ
¡ ¢
(2.9)
NT0 = −T 0 · x p0 · sin γ + ψ
¡ ¢
(2.10)
2.1. Towing stability 13

l0
µ ¶
T 0 = E A0 · −1 (2.11)
l 00

Here is T 0 the towline tension described by equation (2.11) where E A 0 is the tow line stiffness and l 00 the initial
towline length. Next to the equations of motion, the kinematic equations describing the towing system are
presented in equations (2.12) to (2.14).
q¡ ¡ ¢¢2 ¡ ¡ ¢¢2
l0 = x 0 + x p0 · cos ψ + y 0 + x p0 · sin ψ (2.12)
1³ ¡ ¢´
sin γ = 0 y 0 + x p0 · sin ψ (2.13)
l
r 0 = ψ̇ (2.14)

Four additional equations (2.15) and (2.18) describe the relative coordinates of the towed vessel to the tug.

x t0 = l 0 · cos γ
¡ ¢
(2.15)
y t0 = l 0 · sin γ
¡ ¢
(2.16)
x 0 = x t0 + x p0 · cos ψ
¡ ¢
(2.17)
y 0 = y t0 − x p0 · sin ψ
¡ ¢
(2.18)

2.1.2. Stability analysis


To analyse the towing behaviour of the towing system the stability of the equations of motion must be deter-
mined. These are three responses for an equation of motion as seen in figure 2.3. Marginally or dynamically
stable systems are for tow operations also known under the name fishtailing.

Figure 2.3: Definitions of stability: stable (top), unstable (middle) and marginally stable (bottom). From Lee [44]

The type of stability is determined for each equilibrium position of the equation of motion. These equilibrium
positions are points where all the derivatives of the state variables are zero. These can be computed either
analytically or numerically. It is necessary to do this per equilibrium position since the nonlinear equation
of motion needs to be linearised to examine the stability using ordinary linear time invariant (LTI) methods.
The equations of motion can be linearised by deriving the Jacobian of the system at the equilibrium position.
To check stability, multiple approaches are used for towing stability:

• Routh-Hurwitz stability criterion;


• Eigenvalue analysis.
14 2. Literature Review

Note that these approaches describe the local stability around the critical points and not the global stability
behaviour of the towing system. The Routh-Hurwitz stability criterion is a check to determine if all the roots
of the system are negative. Physically, this means that the response of the system dampens to zero since the
power of the response is negative. This criterion provides a way to determine if all these eigenvalues of a sys-
tem are negative without solving the system directly. The downside is that it only provides a binary response:
all eigenvalues are negative or not. More information on this criterion can be found in [25]. Eigenvalue anal-
ysis does provide detailed information on the eigenvalues. Eigenvalue analysis computes the eigenvalues of
the (linear) system either analytically or numerically. On the basis of the eigenvalues the different stability
modes can be determined [44].

• If all eigenvalues have negative real parts, the responses due to the initial conditions will decrease to
zero as time approaches infinity and the system is stable.
• If at least one eigenvalue has a positive real part, the responses due to initial conditions will increase in
magnitude as time increases and the system is unstable.
• If the largest value of all real parts of the eigenvalues is zero, the responses will be the undamped oscil-
lations.

If more information about the eigenvalues is available it can give more insight in the nature of the stability at
an equilibrium position [6]. Assume λi and λ j are the dominant components (the other roots have decayed
sufficiently) then:

• a stable node if λi and λ j are real and negative;


• an unstable node if λi and λ j are real and positive;
• a saddle point if λi and λ j are real of opposite sign;
• a stable spiral point if λi and λ j are complex conjugates with negative real parts;
• an unstable spiral point if λi and λ j are complex conjugates with positive real parts; and
• a center if λi and λ j are pure imaginary conjugate roots.

Global stability behaviour indicates the behaviour of the towing system when released with arbitrary initial
conditions and the behaviour it displays there after [6]. How does the towing system converge to its stable
positions and which eigenvalues dominate this process? Time domain simulation is conventionally used to
investigate this behaviour but the presence of local stability is normally sufficient to determine the towing
stability of the towing system.

These local analyses can be repeated for all equilibrium positions and for different sets of parameters to
investigate the impact of those parameters on the stability. Results from earlier studies have shown the course
stability of towed vessels to be a function of tow velocity U t ug , hull characteristics and thus manoeuvring
forces F man , towline length l 0 , separation of center of gravity from bow hawser attachment point x p as well
as critical towline tension Tcr i t . For deep water, towing stability analysis [6] showed the towed vessel is stable
if:

• xp > N v
Yv ;
• Tcr i t = R; T > Tcr i t ;

So if the towline tension becomes smaller than the still water resistance the towing system becomes unstable.
This is logical since this implies that the tug is not able to tow the towed vessel forward, and thus becoming
unstable. Previous research also showed that the following parameters can increase or decrease the stability
of the towed system:

• Increase Stability;
– More separation between bow hawser attachment point and center of gravity;
– Longer towline;
– Higher overall resistance.
• Decrease stability;
– Multiple towed vessels;
– Higher radius of gyration of the towed vessel.
2.1. Towing stability 15

2.1.3. Verification equation of motion and stability analysis


To verify the implementation of the equation of motion and the stability analysis, the results presented by
Bernitsas and Kekrides (B&K) [6] and Lee [44] are used. In those papers a towing barge was simulated with
only manoeuvring and towline forces. The normalised manoeuvring force coefficients are presented in table
2.1 and equations (2.19) till (2.21). In these papers, the towline tension was calculated for respectively wet
nylon and polyester towlines. The relation between the tension and the materials is described by equation
(2.22). Here, the Sb is the normalised breaking strength and p & q are non-dimensional material properties
with respectively for wet nylon and polyester: Sb 0 = 4.8188 or 4.3011, p = 9.78 or 176 and q = 1.93 or 1.86.

0
¡ 0 0 0¢
X man u , v , r = −R 00 · u 02 (2.19)
0
¡ 0 0 0¢ 1 1 0 1 0 1 0
Yman u , v , r = Y v0 · v 0 + Yr0 · r 0 + · Y v0 vr · v 02 r 0 + · Y vr 0 02 03
r · v r + · Y v v v · v + · Yr r r · r
03
(2.20)
2 2 6 6
0
¡ 0 0 0¢ 1 1 1
Nman u , v , r = N v0 · v 0 + Nr0 · r 0 + · N v0 vr · v 02 r 0 + · N vr 0 0 02 0
r · v r + · Nv v v · v
03
(2.21)
2 2 6
¡ l0 ¢q
T 0 = Sb 0 · p 0 − 1 (2.22)
l0

The towing operation is simulated by a time domain integration of the state variables. The B&K time sim-
ulation used a simple Euler integration scheme to do this. No data on the time step or initial conditions is
known. Therefore, to replicate the simulation, Matlab’s in house ode solver ODE45 is used. This is an adaptive
time step Runge-Kutta scheme. As initial conditions (IC), a small initial yaw angle is adopted.

Sub-figures 2.4 a and b show the sway offset and the yaw angle during the time domain simulation. This
simulation is done for an initial tow line length l 00 = 2.5 and the bow hawser attachment point x p0 = 0.505 in
front of the CoG. It can be seen that the barge will sail to the first equilibrium position at a sway offset y of
twice the barge length from the center line behind the towing tug. Here it sails with a yaw angle of ψ ∼ 15°.
With the less stiff tow line made of nylon, small oscillations in the yaw angle occur. Since the original initial
conditions of the Bernitsas results are unknown, the initial reaction of the system is different when compared
to the present result. All and all, the correspondence between the simulations is good. Besides the transient
part the two match. Although the same oscillation pattern is observed, between the yaw angles there is a slight
difference. This difference possibly originates from the different numerical scheme used in the simulation.
Still, the correspondence is satisfactory.

In figure 2.5, the convergence of the simulations is shown. In sub-figure a, b, and c it shows how the variables
u, v and ψ of the B&K paper and the current simulation converge to their final stable values. The convergence
for the surge and sway offset matches quite accurately while the yaw angle convergence differs. This was ex-
pected from the results in figure 2.4 but it is quantified here. In the current simulation, the barge will have a
high frequency oscillation in the yaw angle for a longer period than in the result of B&K. Since both converge
to the same value, this is deemed not an issue. These oscillations are the results of the high frequency varia-
tions of the tow line tension. Since the wet nylon tow line is less stiff, the variations of the tension are larger
and thus the oscillations are more noticeable.

Besides the time domain simulation, Lee [44] continued on the work of B&K and assessed the stability of the
towing operation on the basis of the two parameters l 00 and x p0 . The procedure of this assessment is explained
in section 2.1. In figure 2.6 and 2.8 the results of the assessment are presented. This is done for respectively
the barge without (w/o) and with (w/) skegs.

In sub-figure 2.6a, the critical or equilibrium points of the towing configuration are plotted based on a shifting
bow hawser attachment point x p0 . In this case, an x p0 = 0 means the towline attaches in the CoG. The findings
of Lee and the current assessment match exactly.

In sub-figure 2.6b, the maximal eigenvalues of the towing configuration with regards to the initial tow line
length l 00 are presented. This is based on an x p0 = 0.505 and a nylon tow line. As can be seen, it is not a complete
match. The maximum eigenvalue curve of position 2 matches decently, but the maximum eigenvalues of
position 1 and 3, which are both represented by the yellow line since they are symmetrical positions are
yield the same result, differ for short l 00 . Also, the current assessment does not include unstable peaks at
l 00 = 0.55, l 00 = 0.59 and l 00 = 1. It could not be determined if these singularities are numerical errors or actual
16 2. Literature Review

Table 2.1: Slow Motion Derivatives from Bernitsas and Kekrides [6] and Lee [44]

Parameter w/o skeg w/ skeg Normalisation factor


m0 0.0170 1
/2 · ρ · L 3
0
I zz 0.00188 1
/2 · ρ · L 5
R 00 0.0014 0.00196 1
/2 · ρ · L 2
Y v0 -0.01153 -0.0259 1
/2 · ρ · L 2 ·U t ug
Yr0 0.00238 0.002815 1
/2 · ρ · L 3 ·U t ug
Y v0 v v -0.3005 1
/2 · ρ · L 2 · 1/Ut ug
Yr0r r 0.0 1
/2 · ρ · L 5 · 1/Ut ug
N v0 -0.007285 -0.0026425 1
/2 · ρ · L 3 ·U t ug
Nr0 -0.00128 -0.00265 1
/2 · ρ · L 4 ·U t ug
N v0 v v -0.00966 1
/2 · ρ · L 3 · 1/Ut ug
Y v0 vr 0.01931 1
/2 · ρ · L 3 · 1/Ut ug
Y vr0
r -0.01025 1
/2 · ρ · L 4 · 1/Ut ug
N v0 vr -0.0111 1
/2 · ρ · L 4 · 1/Ut ug
N vr0
r -0.01025 1
/2 · ρ · L 5 · 1/Ut ug
c -0.0000917 0.0000311 -
N v0 /Y v0 0.6318 0.1020 -

Initial towline length l 0 ' = 2.5 Initial towline length l 0 ' = 2.5

2.5 30

25
2
20
[°]
Sway offset y [-]

Yaw angle

1.5 15

10
1

5
Wet nylon towline Wet nylon towline
0.5 Polyesther towline Polyesther towline
0
Bernitsas: wet nylon towline Bernitsas: wet nylon towline
Bernitsas: polyester towline Bernitsas: polyester towline
0 -5
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Time [s] Time [s]

(a) Bernitsas and Kekrides & Simulated results: sway offset y (b) Bernitsas and Kekrides & Simulated results: yaw angle ψ

Figure 2.4: Comparison TD simulation Bernitsas and Kekrides [6] results and own implementation

20
Wet nylon towline Wet nylon towline Wet nylon towline
0.3 Bernitsas: wet nylon towline 0.4 Bernitsas: wet nylon towline Bernitsas: wet nylon towline
15

0.2 0.3
10
/dt

0.2
dy/dt

0.1
dx/dt

5
d

0 0.1 0

-5
-0.1 0

-10
-0.2 -0.1
-4 -3.5 -3 -2.5 -2 -1.5 -1 0 0.5 1 1.5 2 2.5 3 -5 0 5 10 15 20 25 30
Distance x Sway offset y Yaw angle

(a) Bernitsas and Kekrides & Simulated (b) Bernitsas and Kekrides & Simulated (c) Bernitsas and Kekrides & Simulated
result: Phase Space surge result: Phase Space sway result: Phase Space yaw

Figure 2.5: Comparison Phase Space Bernitsas and Kekrides [6] results and own implementation
2.1. Towing stability 17

unstable positions. The progression of the curve does match the Lee result. These results imply that the
second position is never stable, while, with the current results, the first and third positions yield stable towing
positions. In figure 2.7, all the eigenvalues of the three locations are plotted. Sub-figure a shows the values of
position 1 and 3. Sub-figure b shows the values of position 2. It can be seen that the eigenvalues related to l 0
and γ are the governing eigenvalues in both cases.

Yaw angle at equilibrium points


1.2
80 Equilibrium position 1 Position 1
Equilibrium position 2 Position 2
1
60 Equilibrium position 3 Position 3
Lee eq. position 1 Lee eq. position 1

Max real part of eigenvalues


Lee eq. position 2 Lee eq. position 2
40 0.8
Lee eq. position 3 Lee eq. position 3

20
0.6
0
0.4
-20

-40 0.2

-60 Unstable
0
Stable
-80
-0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5 2 2.5

(a) Lee and Simulated results: stable points (b) Lee and Simulated results: stability with varying l 00

Figure 2.6: Comparison Lee [44] results without skeg and own implementation

Eigenvalues position 1 Unstable Eigenvalues position 2


0 1
Stable
-0.1
0.5
-0.2
Max real part of eigenvalues

Max real part of eigenvalues

-0.3 Unstable
0 Stable
-0.4

-0.5 -0.5

-0.6
-1
-0.7 Eigenvalue u Eigenvalue u
Eigenvalue v Eigenvalue v
-0.8 Eigenvalue r Eigenvalue r
Eigenvalue l -1.5 Eigenvalue l
Eigenvalue Eigenvalue
-0.9
Eigenvalue Eigenvalue
-1 -2
0 0.4 0.8 1.2 1.6 2 2.4 0 0.4 0.8 1.2 1.6 2 2.4

(a) Simulated result: eigenvalues position 1 & 3 (b) Simulated result: eigenvalues position 2

Figure 2.7: Comparison Lee [44] results without skeg and own implementation: individual eigenvalues

In sub-figure 2.8a, the equilibrium positions for the barge w/ skegs are presented. If the x p0 > 0.1020, the
vessel only has one equilibrium position. Again this matches between the current results and Lee’s. In sub-
figure b, the max. eigenvalues are assessed and different results between Lee and the current simulation are
obtained. Since there is only one equilibrium position, all three locations have the same eigenvalues. The
shapes are similar, but the magnitude does not completely match, the present result predicts a larger range of
instability. It is clear from these figures, that the vessel can be towed in a stable position if the tow line length
is l 00 > 1. Since x p0 = 0.505, only position 2 exist. It was found that again, the eigenvalues related to l 0 and γ are
the governing eigenvalues.

From the comparisons presented, it is clear that there are some (small) unexplainable discrepancies between
the two computations. They do however, match on the larger scale. The time domain simulations match
18 2. Literature Review

accurately and the stable towing configurations also match completely. The stability assessment based on
the initial tow line length l 00 is very similar, but the current simulation misses some singularities. All and all,
the implementation of the equation of motion is deemed valid, and it can be used for further investigation.
Yaw angle at equilibrium points

80 Equilibrium position 1 0.02


Position 2
Equilibrium position 2 Lee eq. position 2
60 Equilibrium position 3
Lee eq. position 1 0

Max real part of eigenvalues


40 Lee eq. position 2
Lee eq. position 3

20 -0.02

0
-0.04

-20 Stable

-0.06
-40

-60 -0.08

-80
-0.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5 2 2.5

(a) Lee and Simulated results: stable points (b) Lee and Simulated results: stability with varying l 00

Figure 2.8: Comparison Lee [44] results with skeg and own implementation

2.2. Environmental forces


The three main environmental forces acting on tow operation are classified as the wind, current and wave
drift forces. All three will be shortly discussed.

Wind Force
The wind force is the force exerted on a vessel due to the wind interacting with the above water hull and
superstructure. During sailing conditions the longitudinal component of the force is considered the main in-
terest of the resistance calculation although the transverse component cannot be neglected due to its impact
on the sway and yaw of the vessel. During sailing, its important to take the relative wind velocity into account
instead of the absolute wind velocity. This is the wind velocity measured on the sailing vessel. An overview of
the geometry of the relative velocity can be found in figure 2.9.
q
U w = U v2 +U wi2
+ 2 ·U v ·U wi · cos αwi (2.23)
U wi · sin αwi
µ ¶
αw = arctan (2.24)
U v +U wi · cos αwi
Here is U wi the incoming wind velocity, αwi the incoming wind angle of attack, U v the vessel forward velocity,
U w the relative (wind) velocity and αw the relative (wind) angle of attack. Note the addition of cos αwi below
the root in equation (2.23). Its absence in the source is an error and it is corrected here.

OCIMF [55] presented equations (2.25) to (2.27), to calculate the wind forces on offshore vessels in 1994. They
are based on the dynamic pressure build up in front of the vessel due to the wind velocity. The coefficients
determined by the OCIMF for the calculations were originally developed for large tankers (VLCC’s) but can
be adjusted to a suit smaller vessels. Model test coefficients are always preferred above literature coefficients.
Here C w represent the wind coefficients obtained from tests or database, B · H and L · H the respective longi-
tudinal and transverse above water projected areas. For current, this is the below water projected area B · T
and L · T .
1 2
Fw x = C w x ρ w Uw ·B ·H (2.25)
2
1 2
Fw y = C w y ρ w Uw ·L ·H (2.26)
2
1 2
Nw x y = C w x y ρ w Uw · L2 · H (2.27)
2
2.2. Environmental forces 19

Figure 2.9: Calculation relative velocity from Journee et al. [40]

Another option is MARIN’s inhouse developed WINDOS software is a wind tunnel validated program based
on the Building Block Method: simple building blocks with set resistance coefficients make up the vessel. A
more recent study by He et al. [34] improved the Building Block Method with the inclusion of several correc-
tion coefficients. The method computes the force on the basis of the wind velocity and multiple coefficients.
The shape, height and the exposed area of the block are of importance. In the Modified Building Block Method
shielding and corrections for the shape and wind direction for oblique angles are included. This results in
higher accuracy. Currently, more research is being preformed in area of drag and the effects turbulence of the
wind on the vessel.

Computational fluid dynamics (CFD) is a more definitive solution to determine the wind loads of a vessel but
often needs to be validated through wind tunnel tests. These wind tunnel tests are the main alternative to
computational tools; they can be performed to obtain the wind loads for a specific vessel using scale models.
So, the wind force can currently be predicted through three methods, the generalised OCIMF equations, the
Building Block Method or model tests in a wind tunnel.

Current Force
The current force is the force due to incoming currents (tidal or others) that interact with the vessel. It can
have a big impact on moored vessels, but for sailing vessels it is often considered as an adjustment to the ve-
locities; the relative velocities in surge and sway direction. This is regularly considered a quasi-steady velocity
since it only changes slowly over time.

OCIMF [55] presented similar equations to equation (2.25) to (2.27) to calculate the current forces on offshore
tankers. They are very similar to the wind force equations. Here C w is replaced by the current equivalent C c
and the above water area is replaced by the underwater exposed area.

For the head current condition or still water resistance scenario, more research has been done as the total
resistance of a ship is essential in the design of new vessels. It is similar to the equations presented by OCIMF
but includes all the different components of the resistance separately. These different components can be
captured in the total resistance coefficient C t s . This coefficient can be deduced using the Holtrop and Mennen
method [36] developed on experimental data at MARIN.

Manoeuvring forces are essentially also current forces since they are based on relative velocity. In this case
they are not considered quasi-steady and should include the effects of accelerations. Two methods for es-
timating these manoeuvring forces are the MMG model from the Manoeuvring Model Group [76] and the
cross-flow model developed by Wichers [74]. The MMG model describes the forces through slow motion
derivatives, basically a Taylor’s expansion of the forces related to surge, sway and yaw motions. The force is a
summation of coefficients related to the velocity, velocity squared or velocity cubed depending on the order
of the model. With the assumption of small velocities, lower order models can be sufficiently accurate, but
when moving to higher velocities and drift angles, more complex, higher order manoeuvring models should
be used. The cross-flow model is a conceptually simpler model than the MMG model but only accurate under
the assumption of slow velocities. It consists of two components, a quasi-steady relative current component
and a dynamic relative current component induced by the cross-flow over the hull. This cross-flow compo-
20 2. Literature Review

nent is also known under the name of the Munk Moment [17].

From the manoeuvring coefficients, the course stability of a vessel can generally be determined [1]. This
straight line stability of the vessel is determined by the stability coefficient c in equation (2.28), where xG is
the distance between the geometric center of the vessel and the center of gravity. If the stability coefficient
c is positive, the vessel is straight line stable. If it is negative, the vessel does not have straight line stability.
Basically, this coefficient describes the ratio between the disturbing and stabilising moments acting on the
vessel. If the distance between the working point of the disturbing forces and the CoG is larger than the dis-
tance between the working point of the stabilising forces and the CoG, the vessel is unstable. This is also why
skegs at the stern of a vessel have a stabilising effect. They increase the stabilising forces at a large distance
from the CoG, thus increasing the distance between the working point and the CoG and thus the straight line
stability.

c = Y v0 · Nr0 − m 0 xG0 − N v0 · Yr0 − m 0


¡ ¢ ¡ ¢
(2.28)

Hong et al. [37] considered both manoeuvring models in a comparison for a rectangular barge with a raked
bow and aft. They found that using first order components only (thus a linear approach), both the models
produced satisfactory results. The MMG model produced slightly better results for the yaw motion while
the cross-flow model was much easier to implement due to the simpler nature of obtaining the model co-
efficients. The MMG model was later extended to also include non-linear components by Nam et al. [51],
enabling the model to be applicable for a larger range of yaw angles.

Wave Drift Force


The wave drift force (WDF) is a force that results from the second order force contributions of each wave com-
ponent. The interaction between two waves with a small frequency difference then results in wave groups
producing a low and high frequency force acting on the structure in waves. This force has three components:
the non-zero mean wave drift force, the varying, low frequency, force affecting structures with low natural fre-
quencies like moored vessels, and the varying, high frequency, force affecting stiff structures like TLP’s. The
first force is usually the one of interest during towing operations. The phenomenon of a structure with a (for-
ward) velocity adds complex effects such as the current-wave interaction and the shift from wave frequency
to encounter frequency. The longitudinal wave drift force with a structure having a forward velocity is also
studied under the name of "Added Wave Resistance" or R aw .

There are two phenomena related to the interaction with the vessel that describe the wave drift force as a
function of wave frequency. These two phenomena are the motion induced and the reflection induced forces.
The motion induced part, as seen in figure 2.10a, has a peak roughly around high pitch and heave excitations.
This contribution to the force is due to the interaction of the motion of the ship with the motion of the waves;
the relative motion. The largest relative motions occur either around resonance frequencies or large excita-
tions; for example for pitch, when the wavelength is roughly equal to the ship’s length. The peak force occurs
where the relative motion is largest and the height depends on the properties of the ship motions. The reflec-
tion induced part, see figure 2.10b, will occur when the depth of the influence of the waves is smaller than
the vessel draft and the vessel can be approximated as a vertical wall. In this case the waves reflect or diffract
on the side of the vessel resulting in an asymptotic limit of the force. When both parts are considered and
added the total wave drift force graph versus the frequency emerges, see figure 2.10c. Since the source of the
force is of a second order nature, the force can be made wave height independent by dividing by the wave
amplitude squared ζ2a . This is also known as the Quadratic Transfer Function or QTF. More explanation on
this phenomenon can be found in the book published by Journee et al. [40].

The QTF is the wave drift force related data that is desired for a vessel. Various methods for estimating this
function have been developed over the years and can be classified under the following categories:

• Computational Fluid Dynamics (CFD) methods [21];


The most complete method for estimating the force, it takes both the viscous and the potential part of the
fluid flow into account. CFD can be a very sophisticated tool to use but before data is useful it must be exten-
sively verified by model tests. Therefore, and since this method is the most complex and time consuming it is
not further pursued in this thesis.
2.3. Stationary vessel wave drift forces 21

(a) Motion induced wave drift force F wd M (b) Reflection induced wave drift force F wdR

(c) Total wave drift force

Figure 2.10: Wave Drift Force Regions, taken from Grin [30]

• Potential Theory based methods [49, 59];


Very popular method due to relative simpleness. It ignores the viscous part of the fluid flow. Originally only
capable of addressing the wave drift force on a stationary vessel but in recent years advancements towards
the addition of (forward) velocity have been made. It is also classified under the larger 3D methods, together
with CFD, since it takes the entire geometry of the vessel into account.

• Strip Theory based methods [8, 28, 62];


Is also based on potential theory but is considered a 2D method since it does not consider the vessel as a
whole 3D structure but slices it in tiny 2D sections before examining. Developed when 3D methods were
not yet attainable due to computation power but still widely used due to their computational speed and
simplicity.

• Approximation based methods [19];


Either empirical or potential theory based equations that approximate the force in limited conditions. For
example, the asymptotic limit of the reflection induced part of the QTF.

These categories will be later highlighted in the sections 2.3 and 2.5 for respectively a stationary and sailing,
i.e. with vessel velocities, vessel. Attention must be paid when analysing wave drift forces in shallow water
since extra non-linearities can become dominant, for example the set-down effect resulting from the second
order potential of the wave potential function. In this thesis, deep water is assumed and these effects are not
taken into account.

2.3. Stationary vessel wave drift forces


In the following section the state of the art on calculating the stationary wave drift forces is presented. These
methods are examined and clarified to indicate their use with respect to the tow resistance and for imple-
menting them in a towing stability analysis. Firstly, 3D potential theory methods are discussed. 2D strip
theory methods and approximation based methods are discussed after.
22 2. Literature Review

2.3.1. 3D Potential theory methods


In 1960, Maruo [49] developed the first description of the three-dimensional wave drift force by using the
momentum balance on a control surface at infinity. He did this for the surge and sway motion. Newman
[52] added the description for the yaw motion in 1967 on the basis of the same method: The conservation
of momentum method is also known as the far field (FF) formulation. The advantage of this method is that
it is relatively efficient and it has a (numerically) quick computation time. Since the momentum balance is
drafted at the control surface it is not possible to examine two side by side ships separately, since they are
lumped together in the calculation.

In 1980, based on a different principle, Pinkster [59] developed a separate formulation for the wave drift
force. He used direct pressure integration over the wetted surface of the vessel hull to reach a result. This near
field (NF) formulation has the advantage that it is conceptually simpler than the far field formulation, but as
downside that it is more difficult to compute.

A new formulation, based on the advantages of both the far and near field formulations was developed by
Chen in 2006 [10]. This middle field (MF) formulation starts its derivation with direct pressure integration
over the wetted surface of the ship but combines this result with the conservation of momentum through the
use of Gauss’s and Stokes’s theorem. This shows that both previous formulations are theoretically the same.
Since the control surface in the middle field method is not at infinity, applications like two ships next to each
other can be modelled with much more ease. Overall, this formulation is a robust method of calculating the
horizontal forces and the moment around the vertical axis.

Each of the three-dimensional formulations presented calculate the wave p drift force for a given wave fre-
quency and incoming wave angle in the form of a QTF. This QTF, TQ = P 2 +Q 2 , consists of an in-phase P
and out-of-phase Q part with the low frequency part of the square of the incident waves. These parts (matri-
ces) describe the influence of each frequency combination on the wave drift force. Since every combination
has to be considered the computation can be time consuming. In deep water, Newman’s approximation is a
method of calculating the wave drift force while only using the mean wave drift force values [53]. Instead of
using two waves with different frequencies, Newman used two waves with the mean frequency of the previ-
ous two waves to calculate the results. This simplifies the calculation significantly and one only has to take
the first order pressures into account.

To compute the forces using these methods, the vessel is modelled using source panels and the derived equa-
tions are solved numerically in a similar manner as a regular diffraction analysis. A program developed by
the Delft University of Technology, Delfrac [16], is capable of solving for NF and FF method and is used in this
thesis. However, numerous other (commercial) codes are available yielding similar results.

Delfrac is based on the NF approach developed by Pinkster and uses his approximation of the second order
wave drift force using only first order pressures. The wave drift force components comprise from five terms:
Part I is the relative wave height contribution, part II and III the contributions due to the first order pressure,
part IV the contribution of the ship motions and heave inertia force and part V the contribution of the second
order potential. Detailed information on these contributions can be found in the Hydromechanics book by
Journee [40]. Delfrac only uses the first four terms, and figure 2.11 shows these contributions for the surge
force of a simple rectangular barge.

To examine the QTF in irregular waves, it can be combined with a (directional) wave spectrum S ζ (ω) (for
example a JONSWAP spectrum) to obtain a (directional) wave drift force spectrum. Here, the distinction
between the mean force and the slowly varying force spectrum is made. Equation (2.29) shows the integral
to be solved for the mean wave drift force F̄ , while equation (2.30) shows the low frequency force spectrum
S F (µ) as function of difference frequency µ [40]. These transformations are specific for the spectrum used
in the integral and, for the mean force, to obtain a full range of mean forces, the transformation must be
repeated for a range of spectra. A mean wave drift force spectrum is thus dependent on the peak period of
the spectrum used (in the case of a JONSWAP spectrum) and dependent on the wave height-wave frequency
combinations that occur in practice. Here, the sum frequency part of the square of the incident waves is not
2.3. Stationary vessel wave drift forces 23

Part I: Relative wave heigth contribution Part II: Pressure drop due to velocity squared

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1

-1.2 -1.2
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) Part I (b) Part II

Part III: Pressure due to gradient 1st order pressure and motion Part IV: Contribution of Ship Motions and Heave Inertia force

0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1

-1.2 -1.2
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(c) Part III (d) Part IV

Figure 2.11: Delfrac wave drift surge force contributions Part I to IV with β = 15°

taken into account since this is only relevant for stiff structures. For towing operations this part is negligible.

Z ∞
F̄ = 2 S ζ (ω) · P (ω, ω) · δω (2.29)
Z0 ∞
S F (µ) = 8 S ζ (ω) · S ζ (ω + µ) · |TQ (ω, ω + µ)|2 · δω (2.30)
0

2.3.2. 2D Strip theory methods


Methods described as 2D are a simplification from the 3D models. Most of these were developed before 3D
methods became popular but currently they are still widely applied due to their relatively small computa-
tional effort. These were first developed to analyse ship motions on the basis of potential theory but were
extended for different purposes.

Salvesen [62], used near field formulations in 1974 to calculate the wave drift forces. This was done on the
basis of strip theory. With strip theory the vessel is sliced in small sections which are then examined separately
as if they are part of an infinitely long cylinder, see figure 2.12. Downside of the strip theory method is that it is
impossible to catch detailed changes in the third dimension of the hull shape of the vessel. This is usually the
length of the vessel. This is especially worry-some in bow regions where the change of the hull can be rapid
and abrupt. Strip theory also performs better when forward velocity is part of the analysis. In both scenarios
strip theory will deliver non-optimal results and an alternative method is advised. To examine the wave drift
force in irregular waves the same approach as for 3D potential theory methods can be used.
24 2. Literature Review

Figure 2.12: Example of strip theory from Journee et al. [40]

2.3.3. Approximate methods


There have been several attempts to describe the complex behaviour of the wave drift force on the basis
of simplified equations, either empirical relations or estimations based on the asymptotic behaviour of the
wave drift force. Due to the asymptotic behaviour, the latter is hard to include in other methods. All the short
wave methods presented in the wave drift force with velocities included section 2.5, can also be assessed for
a stationary vessel.

Faltinsen [19] first derived this asymptotic short wave (SW) equation based on the work of Maruo [49] in
1980. It is valid for any structure, as long as it has vertical sides at the waterline and the wavelength λ is small
compared to the ship length L. This can also be interpreted as if the waves do not notice the keel of the vessel
and experience it as a vertical wall. Equation (2.31) only offers a description of the surge and sway forces. In
figure 2.13, the shadow zone is the waterline area that is not subjected to the waves. The waterline angle θ
and incoming wave angle of attack β are also defined in the figure.
Z A
1
~wd = ρ w g · ζ2a
F sin2 (θ + β)~
n δl (2.31)
i
2 B
~
n => n x = sin (θ), n y = cos (θ)

Using this equation to describe the behaviour in irregular waves, equation (2.32) is obtained. This is based on
the assumption that in short waves the wave drift force is no longer frequency dependent and that therefore
the statistical properties of a wave spectrum can be used: H s2 = 16 · m 0 = 8 · ζ2a . If equation (2.29) is used
the same conversion is found, as the mean drift force coefficient P ∞ (ω, ω) = 12 ρg . More information on this
conversion can be found in Jouree et al. [40].

1
Z A
~wd =
F ρ w g · H s2 sin2 (θ + β)~
n δl (2.32)
i
16 B

When examining the short wave limit in irregular waves it is important to understand the limitations due to
the wave frequency to wave height relation. A wave height, for example of 5m, will have a wave period in
open ocean between 8 and 12 seconds. Since this implies a larger wavelength that one that corresponds with
the assumption of short waves there is no significant energy in waves with λ ≥ 0.5 · L, the formula loses its
theoretical groundwork [17].

A more direct and simpler method is prescribed by the DNV standards [14]. They propose the similar limit
as Faltinsen’s but this expression is adjusted for different kinds of vessel front faces through an empirical
reflection coefficient R, see equation (2.33). The reflection coefficients are found in table 2.2. Note the factor
1/2 difference with equation (2.31). The same conversion for irregular waves can be used.

F wd = ρ w g · R 2 · ζ2a · B (2.33)

For very wide structures, the empirical description in equation (2.34) was also published by DNV standards
in 1996 [12]. It is limited to the surge force in the cases where: BL > 3, TB > 6 and U = 0.

B
F wd = ρ w g H s2 · (0.52 · L − 13) (2.34)
L
2.4. Impact forward velocity 25

Figure 2.13: Illustration of the geometry for Faltinsen’s and Fujii and Takahashi’s short wave limits from [63]

Table 2.2: Reflection coefficients for equation (2.33) [14]

Vessels bow shape Reflection Coefficient R


Square face 1.00
Condeep base 0.97
Vertical Cylinder 0.88
Barge with Raked Bow 0.67
Barge with Spoon Bow 0.55
Ship Bow 0.45

2.4. Impact forward velocity


When dealing with (forward) velocity in the field of ship hydrodynamics the Froude number is one of the
most important parameters. The Froude number, F n = U/pg L, describes the ratio between the inertial forces
and the gravitational forces. It describes in which regime a flow around the vessel is flowing and is thus used
for test conditions and scaling of model test results. The Froude number effectively is a ratio between pthe
vessels velocity U v and the wave phase velocity c. The ratio can be expanded for deep water: Ucv = F n · 2π.
Fluid flow is defined as sub-critical if Uv /c ≤ 1, critical if Uv /c ≈ 1 and super-critical if Uv /c ≥ 1. If sub-critical, the
vessels velocity is lower than the wave phase velocity, if critical equal and if super-critical larger than the wave
phase velocity.
p
In ship hydrodynamics the constant value of 2π is usually not included in the Froude number but is incor-
porated in the boundary values. Vessels sailing with F n ≤ 0.15 ∼ 0.2 are classified as low speed vessels as the
flow around these vessels is uniform and steady. In those cases, effects like vessel wave making are not yet
present. If the vessel sails with 0.15 ∼ 0.2 ≤ F n ≤ 0.4 ∼ 0.5 they are classified as moderate speed vessels; vessel
wave making resistance becomes significant but the gravitational forces still dominate the behaviour of the
vessel in the water. For vessels sailing with F n ≥ 0.5 the behaviour is dominated by the inertia forces and is
thus called a high speed vessel. In super-critical flow, planing vessels occur. These boundaries are rough;
deviations between vessels occur. These boundaries were also partly derived by Faltinsen [18].

Fn ≤ p0.5 (= 0.2) = Low Speed vessels



0.2 ≤ F n ≤ p1 (= 0.4) = Moderate Speed vessels

0.5 ≤ F n ≤ p3 (= 1.2) = High Speed vessels: Semi-displacement vessels

F n ≥ 1.2 = High Speed vessels: Planing vessels

At F n ≥ 0.35 trim and heel becomes an important issue in the study of the ship resistance. With low speed
vessels, the influence of the forward velocity can be simplified to solely a shift of the wave frequency to the
encounter frequency ωe , see equations (2.35) and (2.36). The wave number k 0 is adjusted in the same way.
Here are u and v the surge and sway velocity components of the vessel.
26 2. Literature Review

ωe = ω0 + k 0 u · cos β − v · sin β
¡ ¡ ¢ ¡ ¢¢
(2.35)
ω20 ω2e
k0 = ke = (2.36)
g g

Other influences of the forward velocity can be the flow separation in the waves due to current. Although the
current is considered as a relative (forward) velocity in this thesis, the physical phenomenon is different from
vessel velocities. The interaction between the current and waves occurs everywhere in the domain while
the interaction between the waves and vessel velocities only occurs around the hull. To take both current
and waves into account the flow separation should not occur. Equation (2.37) describes the ratio between
the current velocity Uc and the velocity of the particles in the wave and if this ratio remains below one, flow
separation due to the current velocity does not occur in the waves [13].

Uc
<1 (2.37)
ω · ζa

Usually, low wave frequency waves have larger wave heights but for each individual case this assumption
must be checked. Figure 2.14 gives an indication in what areas the assumption might get violated. The as-
sumption holds above the plotted lines.

Flow separation limits


5
Uc = 0.5 m/s
4.5
Uc = 1 m/s
4 Uc = 2 m/s
Uc = 4 m/s
3.5

2.5

1.5

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Figure 2.14: Flow separation limits for ζa , ω and Uc

2.5. Sailing vessel wave drift forces / Added wave resistance


Various methods exist that are used to predict the wave drift force for a sailing vessel. A distinction is made
between the longitudinal (surge) wave drift force and the transverse/oblique and torsional components. The
longitudinal wave drift is of importance during seakeeping, sailing in waves, and is better known as the added
wave resistance. The wave drift force with vessel velocities considers all components of the force and all
components of the vessel velocity and is generally more complex.

2.5.1. 3D Potential methods - Wave drift damping


Within the 3D potential methods predicting the wave drift force with vessel velocities, two general approaches
are identified. Predicting the wave drift forces by implementing velocities in the potential theory code and
computing the impact directly or estimating the zero velocity wave drift force and relating that to the veloci-
ties separately. The first category is still largely under development and no widely accepted results have been
published. An overview of that research is given by Bunnik [9]. This thesis focuses on the second category.

The 3D methods presented in section 2.3, the near field and far field method, are the basis for developing the
wave drift force with vessel velocities. The prime method of relating the stationary wave drift force to vessel
velocity or incoming current is through the wave drift damping. The term Wave Drift Damping B (WDD) was
2.5. Sailing vessel wave drift forces / Added wave resistance 27

first coined by Wichers in 1988 [74]. The phenomenon is named a damping because it represents the wave
drift force derivative to the velocity at zero velocity or in other words the rate of change. It is associated with
a velocity and therefore comparable to a damping term. This term can then be used to relate the stationary
wave drift force F~wd (0) with vessel velocity U~v as presented in equations (2.38) and (2.39). The numbering
follows the conventional motion numbering. For example, the coupling term with subscript B 16 , describes a
surge motion damping due to a yaw velocity.

~wd (ω, β, U
F ~v ) = F
~wd (ω, β, 0) + B
~ (ω, β) · U
~v (2.38)
   
B 11 B 12 B 16 u
~
B (ω, β) = B 21

B 22 B 26  ,
 ~
U = v 
 
(2.39)
B 61 B 62 B 66 r

To arrive at proper descriptions of the wave drift damping two approaches can be taken: computing the entire
matrix directly from potential theory or using approximations of the different (coupling) terms.

Finne et al. [22] developed a method to determine the wave drift damping matrix for arbitrary vessel shapes
as a whole. They based the derivation on potential theory with second order potentials and small forward ve-
locity potentials included. This method is considered the most accurate prediction of the wave drift damping
but since it requires second order and forward velocity potentials to be computed it is not widely adopted.

Approximations are generally split between describing the surge and sway terms versus the yaw terms. The
surge, see equation (2.40), and sway, see equation (2.41), terms can be generally approximated by the equa-
tions developed by Aranha and Malenica et al. [2, 47]. Limitations of the approximation are:

• For slender ships it becomes a more crude approximation when looking at cases deviating significantly
from head waves [35, 38];
• When the first-order motions are near resonance the approximation also become more crude [35];

Both limitations affect the applicability of these approximation equations but they still produce reasonable
results for computing the wave drift damping coefficients.

B 11 (ω, β) ¶ F x (ω, β)
   
ω δ δ
µ
B 21 (ω, β) = ω · cos β · − 2 sin β · + 4 cos β ·  F y (ω, β) 
¡ ¢ ¡ ¢ ¡ ¢
(2.40)
   
g δω δβ
B 61 (ω, β) N z (ω, β)

B 12 (ω, β) ¶ F x (ω, β)
   
 ω ¡ ¢ δ ¡ ¢ δ
µ
B 22 (ω, β) = ω · sin β · + 2 cos β · + 4 sin β ·  F y (ω, β) 
¡ ¢ 
(2.41)
 
g δω δβ
B 62 (ω, β) N z (ω, β)

δ δ
The operators δω and δβ are respectively the derivative to the wave frequency and the derivative to the wave
angle of attack. F x , F y and N z are the wave drift surge force, sway force and yaw moment. For the WDD yaw
term approximations it becomes a bit more complex. Aranha and Martins [3] used slender body approxima-
tions in 1997, B /L << 1, to derive equation (2.42) for the yaw terms of the wave drift damping.

B 16 (ω, β) ¶ M x,1 (ω, β)


   
 ω ¡ ¢ δ ¡ ¢ δ
µ
B 26 (ω, β) = ω · sin β · + 2 cos β · + 4 sin β ·  M y,1 (ω, β)
¡ ¢ 
(2.42)
 
g δω δβ
B 66 (ω, β) M y,2 (ω, β)

where M x,1 and M y,1 are the first order moment contributions of surge force and sway force. M y,2 is the
slender body approximation of the second order yaw moment. At the time, they computed them using a strip
theory approach. The pressure is integrated over the breadth contour or strip C of the vessel. The coordinates
of this strip are described by ~ j + zc~
rc = yc ~ k. Again integrating these sectional drift coefficients over the ship’s
length with x-coordinate x = x c allows for the computation of the zeroth, first and second moments. See
28 2. Literature Review

equations (2.43) and (2.44).

f x (x, ω, β) η x (x,~
   
I rc )
 f y (x, ω, β)  = r c ) ·  η y (x,~
p 20 (x,~ r c )  dC for each x = x c (2.43)
   

℘z (x, ω, β) C −y c η x (x,~
rc )
M x, j (ω, β) f x (x, ω, β)
   
ZL/2
j 
 M y, j (ω, β) = x c ·  f y (x, ω, β)  d x for j = 0, 1, 2 (2.44)
  

M z, j (ω, β) −L/2 ℘z (x, ω, β)

Here is p 20 (x,~ r c ) the second order potential pressure, ~η = (η x , η y , η z ) the normal vector and its components
and f x , f y , ℘z the sectional force components. Note that in the original paper, the sectional yaw normal
vector −y c η x (x,~ r c ) was originally published with the error −y c η y (x,~ r c ). The forces and moments used in
equations (2.40) and (2.41) are related to the zeroth, first and second moment as presented in equation (2.47).

F x (ω, β) = M x,0 (ω, β) (2.45)


F y (ω, β) = M y,0 (ω, β) (2.46)
N z (ω, β) = M z,0 (ω, β) + M y,1 (ω, β) (2.47)

When more detailed information on the pressure distribution is known for example, obtained from 3D panel
methods, the pressure on the panels can be directly integrated to obtain the moment terms, see equation
(2.48). When the second order pressure is not available, the first order approximation as presented in section
2.3 can be used to estimate the WDF moments. This was also shown by Grue and Palm in 1992 and 1993[31,
32]. They showed that the second order potential can be neglected when calculating the stationary wave drift
yaw moment. It allows the wave drift yaw moments to be computed based on the first order potential pressure
instead of the second order potential and this simplifies the calculation significantly. When the panel method
is used to compute the yaw moments, the second order yaw moment can be used in equation (2.42) instead
of the sway approximation.

M x, j (ω, β) η x (x, y, z)
   
Z
j
 M y, j (ω, β) = x c · p 20 (x, y, z) · η y (x, y, z) d S for j = 0, 1, 2 (2.48)
   

M z, j (ω, β) S0 η z (x, y, z)

These moments are computed using the pressure output of Delfrac. To verify them, the forces and first order
yaw moment are compared to the components directly computed with Delfrac for the sample case further
explained in section 3.5. If those match, the computation is correct and the second order moments can be
computed without errors. From figure 2.11 in section 2.3, it is clear that the contribution IV, due to the ves-
sel motion-heave inertia moment, is small. Thus, in these calculations, this term is assumed negligible. The
three force components are plotted in figure 2.15 where they are compared to the NF method results directly
obtained from Delfrac. In the sub-figures 2.15 a, b and c, the force components match quite accurately. The

small deviation is most likely due to the absence of the motion contribution. With the method for calculating
the forces and first order moment verified, the second order yaw moment and first order surge and sway con-
tributions can be computed. The result is shown in sub-figure 2.15 d. Here the surge and sway contributions
will result in the first order yaw moment when added to each other. These moments can be used as input for
the WDD yaw terms calculation in equation (2.42).

The mean wave drift damping for irregular waves can be obtained in the same manner as the mean drift force,
see in equation (2.49).
Z ∞
b i j (β) = 2 B i j (ω, β) · S ζ (ω)δω, (i , j = 1, 2, 6) (2.49)
0
2.5. Sailing vessel wave drift forces / Added wave resistance 29

Surge WDF force recomputed, regular waves Sway WDF force recomputed, regular waves
0.05 0.15
Pressure Computed Mean Surge QTF, = 15° Pressure Computed Mean Sway QTF, = 15°
0 Direct Delfrac Mean Surge QTF, = 15° Direct Delfrac Mean Sway QTF, = 15°

-0.05

-0.1 0.1

-0.15

-0.2

-0.25
0.05
-0.3

-0.35

-0.4
0
-0.45
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) The direct Delfrac F x and pressure computed F x,comp com- (b) The direct Delfrac F y and pressure computed F y,comp com-
parison parison

Yaw WDF moment recomputed, regular waves First and Second order WDF moments, regular waves
0.03 0.04
Pressure Computed Mean Yaw QTF, = 15° Mx,1 , = 15°
0.025 Direct Delfrac Mean Yaw QTF, = 15° My,1 , = 15°
0.03
Mz,2 , = 15°
0.02

0.02
0.015

0.01 0.01

0.005
0

-0.01
-0.005

-0.01 -0.02
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(c) The direct Delfrac N z and pressure computed N z,comp com- (d) The first and second order moments
parison

Figure 2.15: The direct Delfrac output vs the pressure computed Delfrac output and the higher order moment contributions

2.5.2. 2D Strip theory methods


In section 2.3, strip theory was already featured. Generally, the method is more applicable to cases with
forward velocity and is therefore suited for examining the wave drift force with forward velocity.

Boese [8] as well as Gerritsma and Beukelman [28], in respectively 1970 and 1972, proposed methods to cal-
culate the added resistance or surge wave drift force with forward velocity. Gerritsma and Beukelman did this
on the basis of a radiation balance, otherwise known as the balance of energy, and Boese derived his results
from integrating the pressure and taking the longitudinal force component over the ship hull. The derivation
and further explanation of both methods can be found in the book Ship Hydromechanics by Journee et al.
[40]. Both formulations focus on the motion induced part of the wave drift force with forward velocity and
are thus less applicable in short waves.

For both methods there are some comments regarding their applicability [48]:

• Gerritsma and Beukelman’s method is more applicable from head waves to beam waves;
• Boese’s method is more applicable from beam waves to stern waves.

For sway forces and moments the method developed by Salvesen in 1974 [62] can be used. All these methods
lose accuracy once complex bow shapes are examined.
30 2. Literature Review

2.5.3. Approximate methods


For a long time there have been attempts to capture the wave added resistance in empirical, semi-empirical,
and simplified equations. In this section a distinction is made between short wave limits and empirical meth-
ods.

Because the bow area of a vessel often has a very peculiar shape, the previously discussed 2D and 3D methods
can have difficulties describing the forces in short waves since the reflection component becomes dominant.
The first, semi-empirical, description for the short waves phenomenon on a sailing ship was suggested by
Fujii and Takahashi [26] in 1975, see equation (2.50). They based there equation on tests performed on a fixed
vertical cylinder.
1
F wd x = ρ w g ζ2a · B · B f (β)αd (1 + αU ) (2.50)
2 µZ ¶
1
Z
sin2 θ − β sin (θ)δl + sin2 θ + β sin (θ)δl
¡ ¢ ¡ ¢
B f (β) = (2.51)
B W LI W LI I
π2 I 12 (kd )
αd = = R(kd ) (the R-function) (2.52)
π2 I 12 (kd ) + K 12 (kd )
p
αU = 5 F n (2.53)

Where I 1 is a modified Bessel function of the first kind of order 1, K 1 is a modified Bessel function of the
second kind of order 1 and the waterline sections are defined in figure 2.13. This equation was later improved
by Kuroda et al. [43], and their modifications can be found in equation (2.54) and (2.55). They reasoned that
the influence of the encounter frequency had an impact on the reflection coefficient and the advance velocity
coefficient had a different fit with regards to the Froude number.

π2 I 12 (k e d )
αd = (2.54)
π2 I 12 (k e d ) + K 12 (k e d )
αU = CU · F n;
£ ¤
CU = max 10.0; −310 · B f (β) + 68 (2.55)

Downsides of the Fujii and Takahashi and Kuroda et al. equations are:
• They originally only focus on the added wave resistance or surge wave drift force;
• Fujii and Takahashi’s equation only gives reasonable results with blunt bodies;
• The equation from Kuroda et al. can overpredict the forces for slender bodies [68].

Kashiwagi et al. [41] proposed an extra correction C f v for the formula of Fujii and Takahashi to account for
the forward velocity effect. The results should be multiplied with the coefficient found in equation (2.56).

(λ/L)10
µ ¶
C f v = 1 + 4 tanh (F n) · exp −0.02 (2.56)
F n3

Faltinsen [19] developed a more theoretical asymptotic method on the basis of pressure integration to cal-
culate the reflection induced resistance in 1980. This was presented at the same time as his equation for a
stationary vessel. His method was later revised by Sakamoto and Baba [61]. The revised equation is presented
in equation (2.58). The sway velocity dependence was not included in the original formulation by Faltin-
sen but is included by editing the velocity dependent part with respect to the angle component in line with
the velocity. This adjusted part is presented in equation (2.59). Note that there is still ongoing research into
the factor 2ω 4ω
g relating the wave drift force with the velocities. Some argue a factor g is necessary but at this
moment theoretical proof is still absent. More information on this issue can be found in appendix B.

Z
~wd =
F F e ·~
nd l (2.57)
wl
µ ¶
1 2 2
¢ 2ω ¡
F e = ρ w g ζ sin θ + β + u cos β − cos (θ) cos θ + β
¡ ¡ ¢ ¡ ¢¢
(Surge velocity) (2.58)
2 g
µ ¶
1 ¢ 2ω ¡ ¡ ¢
F e = ρ w g ζ2 sin2 θ + β + v sin β − sin (θ) sin θ + β
¡ ¡ ¢¢
(Sway velocity) (2.59)
2 g
~
n => n x = sin θ, n y = cos θ (2.60)
2.5. Sailing vessel wave drift forces / Added wave resistance 31

Liu et al. [45, 46] improved the method of Faltinsen with several empirical correction coefficients in 2015 and
2016. This makes the semi-emperical description in equation (2.61) also applicable for finer hull types with
waterline flares and sectional local draughts. It is also corrected to account for quartering waves. The same
adjustment for sway velocity is adopted.

µ ¶
1 ¢ 2ω ¡
F e = ρ w g ζ2 sin2 θ + β + u cos β − cos (θ) cos θ + β
¡ ¡ ¢ ¡ ¢¢
(2.61)
2 g
p ¢
0.87 f (β) 1+4 F n
µ ¶ ¡
1
· ¢ · Cd
cos η wl
¡
CB
π2 I I2 (k e Tl )
Cd = (2.62)
π2 I I2 (k e Tl ) + K I2 (k e Tl )
Tl = min T cos η wl , d (x)
£ ¡ ¢ ¤
(2.63)

Where f β = cos β for 0 ≤ β < π/2 and f β = 0 for π/2 ≤ β < π. η wl is the waterline flare, C B the block
¡ ¢ ¡ ¢ ¡ ¢

coefficient, T the vessels draft and d (x) the local draft. The main downside of Faltinsen’s equation was that
it could theoretically only be applied to blunt body ships but with the recent modifications by Liu et al. this
asymptotic limit is also capable of predicting the force acting on more slender vessel bodies.

Guo and Steen [33] improved the short wave results for a 2D method by utilising the R-function that was first
proposed by Fujii and Takahashi [26]. With the combination of the motion and reflection induced parts a
full added wave resistance spectrum can be obtained. See equation (2.64) for the use of the R-function to
combine the two parts.

F wd = 1 − R 2 (kT ) · F wd M + R 2 (kT ) · F wdR


¡ ¢
(2.64)

Where F wd M is the motion induced part of the force and F wdR the reflection induced part of the force.

As is common in ship hydrodynamics, model tank tests are performed to validate theories and obtain data. In
2014 Grin [30] developed two empirical models at MARIN to predict the wave added resistance: STAWAVE2®
and SPAWAVE®. This resulted in a database with around fifty different vessels from which the parameter
influence was determined through statistical analysis. From these regression models estimations of the wave
added resistance for new vessels can be made.

Söding and Shigunov [68] showed that an empirical equation, see equation (2.65), first proposed by the IMO
can be used, but generally over-predicts results. This is conservative in a resistance calculation but less con-
servative when analysing the towing stability. This must be kept in mind. The equation is deemed valid
between head waves till bow quartering waves.
s
1 B
F wd x = ρ w g ζ2a B (2.65)
2 Lb

Where L b is the distance of the bow to 95% of maximum breadth on the waterline.
32 2. Literature Review

2.6. Chapter summary


The literature related to towing operations has been reviewed in this chapter. First, the towing stability was
addressed. Research goes back to the 50’s but only in recent years the impact of waves has been on the
agenda. Still no systematic approach to the influence of waves has been presented. The equation of motion
and the stability analysis, including the verification process, was presented and yielded a sufficient match
with consulted literature. Stability criteria and what possible parameters have impact on the towing stability
have been identified.

As input for the equation of motion, wind current and wave environmental forces are discussed. Wind can be
described by implementing the OCIMF [55] equations while current can be modelled by the same equations
or by more complex manoeuvring derivatives.

For the waves, first the wave drift force on a stationary vessel was discussed. Attention is paid to the theoret-
ical background of the QTF and multiple wave drift estimation methods are presented. The impact of veloc-
ities and consecutively the wave drift force with velocities shows the different issues involved with forward
velocity and the wave drift estimation methods supposedly capable of addressing these. Various methods like
3D potential theory, 2D strip theory and approximate methods are addressed. Special attention is paid to the
wave drift damping, from which the computation is verified on the basis of the numerical program Delfrac.
3
Methodology
In this chapter the methodology of the thesis is presented. As presented in chapter 1 the goal of this thesis is
to investigate and determine the impact of the wave drift on the tow resistance and the towing stability. The
approach in this investigation is to examine different estimation methods for the wave drift, see chapter 2,
with the purpose of classifying their applicability and use for both the tow resistance as the towing stability.
With these estimation methods, the quantitative influence of the wave drift on the towing stability can be
determined.

Before the estimation methods can be compared, the requirements that the wave drift force must meet for the
tow resistance and the stability analysis must be drafted. Section 3.1 presents the requirements for the tow
resistance while section 3.2 includes these requirements for the towing stability and yields the format of the
input for the stability analysis. For both the stationary and the sailing vessel, the methodology for comparing
the estimation methods should be universal. This methodology should be usable for arbitrary vessels and
should provide insight in their applicability and use for the parameter investigation of the towing stability.
Besides that, it should give a general overview of the methods for more insight into the towing resistance
calculation. This universal comparison methodology is presented in section 3.3.

Once the comparison is completed, the chosen wave drift estimation methods will be implemented in the
towing stability calculations. The methodology of the implementation of the wave drift in the towing stability
calculations, the parameter identification and comparison between the calm water towing stability and the
towing stability in waves is included in section 3.4. The results of these steps will determine the impact of
the wave drift on the towing stability and lead to the discussion, conclusions and recommendations of this
thesis.

At last, the vessel used in the simulations is presented in section 3.5. The necessary properties and require-
ments for the vessel, if it was to be replaced by an arbitrary one, are stated.

3.1. Requirements wave drift for tow resistance


For the tow resistance, it is necessary to include a range of wave drift force components as large as is allowed
within the validity boundaries of the theory. This allows the tow resistance calculation to be applied for low
forward velocity as well as higher forward velocity cases. This only limits the computations to a forward
surge velocity up to 4 m/s. For the tow resistance, the irregular wave results are the governing cases and thus
irregular wave sea states are defined for analysing these cases.

The wave spectrum formulation used in this thesis is the JONSWAP spectrum but other spectra can be applied
as well. Its input parameters are γ = 3.3, the peak enhancement factor, H s , the significant wave height and
T p , the peak wave period. The mathematical description is presented in equation (3.1). Here is A an semi-
empirical factor which, together with more information on the spectrum and how it is obtained, can be found
in [40].
à !
320 · H s2 −5 −1950 −4
S ζ (ω) = · ω · exp ·ω · γA (3.1)
T p4 T p4

33
34 3. Methodology

The two spectra that are examined are based on two sea states "typical" for stationary and sailing conditions.
The sea state for the stationary conditions is based on a significant wave height of H s = 5m and the sea state
for the sailing vessel on H s = 2.4m. Table 3.1 gives the full test conditions also including current and wind
velocity. The wave period of a spectrum can be approximated with statistical values based on the zeroth, first
and second order moment of the spectrum m nζ : the mean centroid wave period T1 , the mean zero-crossing
wave period T2 and the peak period T p [40] or by an semi-empirical T p range based on the wave steepness:
T p,emp [15]. The theoretical descriptions are described in equations (3.2) to (3.4).
Z ∞
m nζ = ωn · S ζ (ω)d ω (3.2)
0
s
m 0ζ m 0ζ
T1 = 2π · , T2 = 2π · (3.3)
m 1ζ m 2ζ
T p = 1.199 · T1 = 1.287 · T2 (3.4)
s s
Hs Hs
127.53 · < T p,emp < 294.30 · (3.5)
g g

Based on a JONSWAP Spectrum, the relation between T1 , T2 and T p is described in equation (3.4). This and
equation (3.5) result in upper and lower limits of the peak period T p used for the spectrum, see table 3.1. The
spectra are plotted in figure 3.1.

Table 3.1: The irregular waves sea state parameters

#1: Stationary #2: Sailing


Sea State
Conditions Conditions
H s [m] 5.0 2.4
T p l ower [s] 8.06 5.59
T p upper [s] 12.25 8.49
U wi [m/s] 20 10
Uci [m/s] 0 0

Wave Spectrum H s = 5 m Wave Spectrum H s = 2.4 m


9 9
Spectrum H = 5 m, T = 11.4 s Spectrum H = 2.4 m, T = 7.85 s
s p s p
8 8
Spectrum H s = 5 m, Tp = 10.5 s Spectrum H s = 2.4 m, Tp = 7.39 s
Spectrum H = 5 m, T = 9.67 s Spectrum H = 2.4 m, T = 6.98 s
7 s p 7 s p
Spectrum H s = 5 m, Tp = 8.98 s Spectrum H s = 2.4 m, Tp = 6.61 s
6 Spectrum H s = 5 m, Tp = 8.38 s 6 Spectrum H s = 2.4 m, Tp = 6.28 s
Spectrum H s = 2.4 m, Tp = 5.98 s
5 5

4 4

3 3

2 2

1 1

0 0
4 6 8 10 12 14 16 18 20 22 24 4 6 8 10 12 14 16 18 20 22 24

(a) Sea state #1: Stationary Conditions (b) Sea state #2: Sailing Conditions

Figure 3.1: The irregular waves sea states

These upper and lower boundaries provide a range of applicability for the irregular waves results. For ex-
ample, a wave spectrum with H s = 5m occurs with peak periods between 8.06s ≤ T p ≤ 12.25s. This means
spectra with peak periods outside that range can be neglected since they do cannot physically exist. The
boundaries of the wave drift force components for the tow resistance are summarised in table 3.2.
3.2. Requirements wave drift for stability analysis 35

Table 3.2: Lower and upper boundaries for the tow resistance of the wave drift force variables

Variable Lower Boundary for WDF Upper Boundary for WDF Unit
Significant Wave Height H s 2.4 5.0 [m]
T p for H s = 2.4m 5.59 8.49 [s]
T p for H s = 5.0m 8.06 12.25 [s]
Surge velocity u 0.0 4.0 [ ms ]

3.2. Requirements wave drift for stability analysis


Based on the research goal the requirements for looking into the wave drift force when including it in a towing
stability analysis need to be drafted. In literature, various approaches were undertaken but they were mostly
based on stationary wave drift forces. The most promising implementation of the wave drift force was done in
the research of Varyani et al. [72], Shigunov and Schellin [64] and Hong et al. [37]. Varyani et al. used regres-
sion analysis with up to 15 harmonic coefficients to describe the stationary forces based on parameter L/λ for
different yaw angles ψ. Shigunov and Schellin computed their wave drift force with forward velocity using 3D
potential theory and investigated multiple yaw angles and the impact of different significant wave heights.
They only used the force balance and didn’t look at the complete towing stability. Hong et al. included both
model tests with waves and an added resistance calculation and concluded the increase in towline tension
increased the stability. Based on the literature investigated, it is clear the stability analysis only needs to ex-
amine the steady state behaviour of an equation of motion and therefore only requires the mean wave drift
forces.

To investigate the towing stability including the forward velocity wave drift force, the wave drift force, besides
its dependency on the wave frequency ω and wave amplitude ζa , should be dependent on the variables de-
scribing the motions and thus the towing stability. As found from the equations of motion derived in chapter
2, the wave drift should be a force/moment dependent on the surge, sway and yaw velocities plus the yaw
angle (or wave angle of attack), see equation (3.6).

F wd x = X w ave ω, ζa , u, v, r, ψ
¡ ¢

F wd y = Y w ave ω, ζa , u, v, r, ψ
¡ ¢
(3.6)
N wd z = N w ave ω, ζa , u, v, r, ψ
¡ ¢

The first two variables are defined from experience. The wave frequencies that can be encountered during
tow operations offshore lie between 0.4 ≤ ω < 2.1r ad /s. These limits correspond to periods ∼ 3 ≤ T <∼ 16s.
The forces are normalised with respect to the wave amplitude squared, any wave amplitude that matches the
assumptions of the estimation methods can therefore be used. For irregular waves these variables change
to the wave peak period T p and the significant wave height H s . More information on the definition of those
variables is presented in section 3.5.

To determine the range for the last four variables on which to base the wave drift force, the verification towing
stability simulations performed in chapter 2 are used. These are done for a tug forward velocity range of 0 ≤
U t ug < 3m/s, which equals 0 ≤ F n < 0.15 for the towed vessel described in section 3.5, and with manoeuvring
forces only. More information about the simulation can be found in chapter 6. These simulations show a
range of the variables and since it is expected that head waves will reduce the motions and thus the variables,
these values can be used as boundaries. Since the verification study is not performed for the final vessel used
in the towing stability analysis, a safety margin is applied to the boundaries and they are revisited before
the final simulations are assessed. The limits are presented in table 3.3. Note that there are no conflicting
requirements between the ones drafted for the tow resistance and the towing stability.

The implementation of the forces into the simulations can be based on direct force calculation or on the basis
of look-up tables. Direct calculation allows for the calculation of the force with the exact parameters but
requires more time and computational resources to implement. Look-up tables are computed beforehand
for an expected range of parameters and during the simulation this allows for quickly looking up the data
of a certain combination of parameters. It is possible that interpolation between data points is necessary
when parameters do not exactly match. Besides the improved computation speed, the look-up tables also
36 3. Methodology

already provide insight into the parameter dependence and therefore look-up tables are adopted for the force
implementation of the simulations.

Table 3.3: Lower and upper boundaries for the stability analysis of the wave drift force variables

Lower Boundary Upper Boundary Lower Boundary Upper Boundary


Variable Unit
from Simulation from Simulation for WDF for WDF
Wave frequency ω - - 0.4 2.1 [ r ad
s ]
Wave amplitude ζa - - - - [m]
Surge velocity u 0 3.7 0.0 4.0 [ ms ]
Sway velocity v -1.2 1.2 -1.5 1.5 [ ms ]
Yaw velocity r -0.1 0.1 -0.2 0.2 [ r ad
s ]
Yaw angle ψ -25 25 -45 45 [°]

If the wind forces are included in the stability analysis they can be developed on the basis of the OCIMF [55]
equations explained in chapter 2.

Besides the direct requirements for the wave drift force there are some other requirements. Normally, during
tow operations offshore the captain(s) can deploy some operational measures to mitigate stability issues. If
the towing model is able to take these mitigations into account simulating them could give more insight in
the exact behaviour of the mitigations. Operational mitigations can include:

• Reducing forward velocity, thus increasing stability;


• Trimming the vessel backwards, increasing the resistance and thus stability;
• Addition of a towing bridle, moving the bow hawser attachment point forward and thus increasing
stability;
• Addition of a second or third towing tug.

3.3. Methodology wave drift estimation method comparison


As presented in chapter 1 wave drift force estimation methods will be compared for stationary and sailing
conditions. Instead of continuing with the most developed theoretical estimation method, multiple estima-
tion methods are compared. This is done for insight in the theory behind the methods and to investigate if
some methods with quicker computation time or other features should be considered for implementation.
For this comparison a universal comparison methodology, which is repeatable for an arbitrary vessel shape,
is presented. Both the stationary and sailing vessel cases follow the same methodology, with the difference
that the sailing vessel assessment is built on the results of the stationary vessel assessment and thus takes
those results into account when addressing the applicability.

For the stationary vessel, the methods that are assessed are presented in figure 3.2. From the 3D methods,
the NF and FF method is computed. The 2D Strip theory is considered and from the approximations, Faltin-
sen’s and Liu’s short wave methods will be computed. These will then be combined using the R-function
method to describe the full range of frequencies. Fujii & Takahashi’s, Kuroda et al.’s and Kashiwagi et al.’s
short wave methods, all variations of each other, will be computed and finally the DNV approximations, more
engineering-like formulas, are also addressed.

The sailing vessel wave drift methods that are assessed are presented in figure 3.3. The methods build on
the stationary vessel assessment results by using their results and include velocities. For the 3D methods,
this step is done using the wave drift damping, who’s implementation is separately verified. For the strip
theory and short wave methods the inclusion of velocities is something the original authors focused on and
is already available in literature. The DNV approximations are traded in for an IMO approximation.

The first step in the estimation comparison process is to verify the computation of the estimation methods
with its applicability obtained from literature. The next step is to compare the computation results and quan-
tify the differences. This comparison will yield a range and accuracy for each estimation method. Together
with some qualitative considerations, this is the basis of the final two assessments. On the one side, the results
are put in context for the tow resistance contribution and applicability and on the other hand, a multi-criteria
3.3. Methodology wave drift estimation method comparison 37

Figure 3.2: Overview of the stationary vessel wave drift estimation methods

Figure 3.3: Overview of the sailing vessel wave drift estimation methods

analysis (MCA) is used to determine the most suitable estimation method(s) to use in the stability analysis.
An MCA is performed since there are multiple criteria that can be beneficial for the implementation of the
wave drift in the stability analysis. To be certain that the MCA results are a solid choice, a small sensitivity
analysis is performed on the MCA factors and scores. This process is followed for the stationary and sailing
vessel assessments with the exception that for the stationary vessel, the multi-criteria analysis is excluded.
The process is also visualised in figure 3.4.

Verification methodology
For the comparison to be of value, all the estimation methods should be verified to identify their applicability
to the requirements. Some boundary values need to be checked and the computations of the estimation
method should match the results from literature. Boundary values to be checked are:

• Flow separation between waves and relative current does not occur around the vessel;

To be able to use the wave drift damping method to relate the wave drift force to the forward velocity no flow
separation is allowed to occur [13]. This means: U /(ω · ζa ) < 1.
38 3. Methodology

Figure 3.4: Visualisation of the estimation comparison process

• The Froude number must remain in the low forward velocity region.

For various assumptions regarding the forward velocity to remain valid, the Froude number must remain low:
F n < 0.15;

Based on this verification, the computation of the estimation methods might be addressed. The range for
which the estimation method is theoretically valid can also be either increased or decreased.

Methodology for the estimation comparison


Once the estimation methods are verified, the estimations methods should be compared. As has been dis-
cussed in chapter 2, to quantify the differences between the estimation methods it is important to first de-
fine the areas in which they are compared. The different methods are classified in three distinct categories:
methods that describe the motion dominated response, that describe the reflection dominated response or
methods that describe the transition zone. These regions are marked on the basis of the parameter L/λ, see
equation (3.7) for the deep water approximation.

L 2πL ω2 L
= = (3.7)
λ g T 2 2πg

The motion dominates the force in long incoming waves and reflection dominates in short incoming waves.
In between these there is a transition zone where both parts influence the force.

• Motion dominated zone: L/λ ≤ 1.5


• Transition zone: 1.5 < L/λ ≤∼ 3.5
• Reflection dominated zone: ∼ 3.5 < L/λ

When comparing the computed values of the different methods, their theoretical background will determine
in what range the method is applicable. For regular waves it can describe the motion zone, the transition zone,
the reflection zone or all. This is quantified by the L/λ ratio. For the irregular waves, the applicable range is
determined by the physical peak period T p range. Within these categories no distinction between methods is
made yet. To make an unprejudiced decision on the usability of the different methods, the computed results
are compared on the basis of two criteria:

• Mean Square Error (MSE) computation;


• Qualitative considerations.
3.3. Methodology wave drift estimation method comparison 39

Table 3.4: MCA factors for the comparison of estimation methods

Category Score Factor Score · factor


MSE ... 6 ...
Theoretical background ... 5 ...
Simplicity equation ... 1 ...
Complexity of parameter analysis ... 3 ...
Application range ... 2 ...
Applicable for arbitrary vessel ... 4 ... +
Total ...

The next step is the multi-criteria analysis (MCA). Different weight factors are assigned to the different cate-
gories. Factors of 1 to 6 are used. The importance of these factors in the final choice is evident, and therefore
the sensitivity analysis is done on these factors. A variation of +20% and −20% is used. The MCA assigns
scores between 1 and 4 to each qualitative comparison category and 1 to 5 in the MSE category. Again, as
their impact on the final choice is significant, the same variation for a sensitivity analysis is implemented.

To obtain the MCA results, the factors are multiplied with the score for each point. Total scores are obtained
by addition of the totals of each category. The factors are presented in table 3.4 and the reasoning and expla-
nation behind these factors is elucidated below.

For the whole application range determined in sections 3.1 and 3.2, an estimation method is obtained. This
solution can consist of a single method or a whole set of methods with each predicting a part of the range.
This general estimation method is input for the stability analysis. The methodology for this part is featured
in section 3.4.

The mean square error is used to differentiate between estimation methods quantitatively. The MSE is the
most commonly applied metric for this purpose, see equation (3.8). Here is Ŷi the base value, Yi the compared
value and n the number of data entries. It is important to apply this metric only in a valid application range
since outliers are disproportionately included in the error due to the square. Hence, large errors can occur
outside the range of application.

1X n ¡ ¢2
M SE = Ŷi − Yi (3.8)
n i =1

For this comparison a base value must be chosen. The base value is chosen as the Far Field estimation method
computed with Delfrac. This method is chosen as the base value because its theoretical background is solid
and Delfrac is a verified computation program. The downside of using the FF method as base value can be
that the method includes limited impact of vessel motions on the wave drift force. During the comparison,
this limitation is kept in mind. The error that is calculated is normalised based on the base value. The error
is thus described as a percentage of the base value. If errors are too large, the base value can be varied to
investigate a better suited estimation method.

The MCA factor of the mean square error computation is 6. The quantitative part of the comparison is
deemed the most important to obtain usable results and therefore it is given the highest factor. The MSE score
is given based on the height of the relative MSE error: < 1% = 5, 1% − 10% = 4, 10% − 25% = 3, 25% − 100% = 2,
> 100% = 1.

For the qualitative considerations some criteria are considered. These criteria, corresponding with the com-
plexity of the method, will be used to analyse the methods based on their theoretical background, and with
that knowledge interpret the MSE results obtained. Criteria on which the methods are judged are:
40 3. Methodology

• Theoretical background;

The examined methods all find their background in literature. However, some methods have a more robust
and more substantiated theoretical groundwork and are therefore preferred. Since this theoretical ground-
work is very important for understanding the physics in play this criterion is given a factor of 5.

• Simplicity of the equations;

Generally, simple equations are preferred as they tend to be quickly and easily understandable. It can be very
useful for engineering purposes but for this research simplicity is not a driving factor. Therefore it is given the
factor of 1.

• Complexity of the parameter analysis

Since part of the goal of the thesis is to examine the impact of the wave drift (parameters) on the tow resis-
tance and towing stability the more complex the analysis of the parameters of the estimation methods, the
more complex the examination of the impact becomes. Lower complexity is preferred as it will increase the
reliability of the results and therefore a factor of 3 is given.

• Application range;

The application range is vital when using a method and should be taken into account at all times. Methods
with a large application range are preferred but combining multiple methods with small application ranges
does not reduce the overall applicability of the wave drift to the towing stability. Therefore, a MCA factor of 2
is given.

• Applicable for arbitrary vessel;

The method must be applicable to arbitrary vessels. This is a set requirement for the results to be usable for
general towing operations and so it is given the factor of 4.

3.4. Methodology towing stability analysis


Once the wave drift force estimation methods are compared for both the stationary and sailing vessel, the
sailing vessel results are used as input for the stability analysis. With the goal of identifying the impact of the
wave drift force on the towing stability in mind the methodology is visualised in figure 3.5.

The first part of the towing stability analysis is the verification of the equation of motion and the stability
analysis. This part of the analysis has already been presented in the literature review chapter 2. The remain-
ing analysis is split into two parts: towing stability without waves and the towing stability with waves. The
analysis without waves is an analysis solely including hydrodynamic manoeuvring forces. It is similar to the
simulations performed in Hong et al. [37]. For this without waves simulation, the parameter dependence of
the bow hawser attachment point x p , tug forward velocity U t ug and initial tow line length l 0 is investigated.
Results are verified by a simple time domain simulation of the towing operation.

Once this benchmark data is present, the next step is to include the influences of waves in the analysis. The
waves have an impact on both the equilibrium positions of the towed vessel and the stability of these posi-
tions. Both are investigated for the same parameters as the without waves case plus the extra variables: the
wave frequency ω, wave amplitude ζa and irregular wave peak period T p . Again, a time domain simulation is
done to verify the results.

Both the stability assessments and the time domain simulations for the without waves and and the with waves
case are compared and the impact of the waves is derived from this comparison. Wind forces can be included
in this analysis later on, to investigate the relative impact of the forces.
3.5. The simulation models 41

Figure 3.5: Visualisation of the stability analysis methodology

3.5. The simulation models


For the implementation and comparison of the different estimation methods a simulation model is necessary.
Based on the research objectives a 3D model as input for Delfrac and ShipMo/SafeTrans is needed as well
as waterline profiles for the approximate estimation methods. Since there are two parts of the study that
require a simulation model, the wave drift force comparison part and the towing stability part, and not all the
necessary data is present for one model, two simulation models are used.

Vessel Geometries
The vessel that is chosen for this research is based on the rectangular barge with a raked bow and stern pro-
posed by Hong et al. [37]. This barge is based on the Maritime Swift Barge [60]. This original barge is used
for the towing stability part, while a simplified pure rectangular barge with the same dimensions is adopted
for the tow resistance part of the study. This is done since the framework for analysing the different wave
drift estimation methods needs to be applicable to arbitrary vessels. With the inclusion of two test vessels,
the wave drift computations are done for two geometries. Their geometry is shown in figure 3.6 and their
characteristics are presented in table 3.5 and 3.6. Since the barges have simple geometries, the unknown data
like the radii of gyration could be estimated based on the approximations from Journee [40]. For Barge 2, two
versions are used in the towing stability part. One with skegs (w/ skegs), and one without stabilising skegs
(w/o skegs). More information on this can be found in the manoeuvring subsection 3.5. With these vessels,
the frequency range 0.4 < ω ≤ 2.1 rad/s converted to L/λ becomes 0.2 < L/λ ≤ 5.4.

Details on the computation of radius of gyration and the format of the geometry input for the assessments
can be found in appendix B. Some investigative studies on vessel parameters have also been completed and
can also be found in appendix B. These include a mesh convergence study for the Delfrac computations and
vertical center of gravity impact study. The mesh convergence study has been completed since Delfrac is
42 3. Methodology

known to deliver unreliable results for the near field method when the model mesh is not properly addressed.
More information on the impact of the mesh can be found in the discussion chapter 7. The vertical center
of gravity study has been completed since experience shows the effect this property has on the peak of the
QTF. This property is investigated to determine its influence and select a vertical center of gravity suitable to
apply in the rest of the comparison process. Both studies are completed for Barge 1 but the results are also
applicable for Barge 2.

(a) Barge 1: used in WDF comparison part (b) Barge 2: used in towing stability part

Figure 3.6: Geometry of the barges used as test vessels

Table 3.5: The Barge 1 properties Table 3.6: The Barge 2 properties (from Hong et al. [37])

Barge 1 Value Unit Barge 2 Value Unit


Length 76.2 [m] Length 76.2 [m]
Breadth 24.4 [m] Breadth 24.4 [m]
Draft 3.8 [m] Draft 3.8 [m]
CB 1.0 [-] CB 0.928 [-]
L/B ratio 3.123 [-] L/B ratio 3.123 [-]
B /T ratio 6.421 [-] B /T ratio 6.421 [-]
k xx 7.040 [m] k xx 7.040 [m]
ky y 21.980 [m] ky y 21.980 [m]
k zz 21.980 [m] k zz 21.980 [m]
m 7.242 ·106 [kg ] m 6.720 ·106 [kg ]
I xx 3.589 ·108 [kg · m 2 ] I xx 3.331 ·108 [kg · m 2 ]
Iyy 3.499 ·109 [kg · m 2 ] Iyy 3.247 ·109 [kg · m 2 ]
I zz 3.499 ·109 [kg · m 2 ] I zz 3.247 ·109 [kg · m 2 ]

Manoeuvring Model
The manoeuvring model that is chosen is the MMG model, utilising slow motion derivatives (SMD). These
derivatives describe the behaviour of the vessel in the water based on the velocities. The MMG model is the
most accurate model that is available from Hong et al. [37] and it was made applicable for a large number of
yaw angles by including non-linearities by Nam et al. [51]. These coefficients are available for two versions of
Barge 2, Barge 2a, without stabilising skegs (w/o), and Barge 2b, with stabilising skegs (w/).

From straight line stability coefficient c it is determined that the barge w/o skegs is course unstable while the
barge w/ skegs is course stable, see table 3.7. The ratio N v0 /Y v0 gives the working point of the disturbing forces.
Here, the primes indicate that the normalised value of the derivative is presented.
3.6. Chapter summary 43

Table 3.7: Slow Motion Derivatives derived by Hong [37] and Nam [51] for a Barges 2a and 2b

Barge 2a Barge 2b
Parameter Normalisation factor
(w/o skeg) (w/ skeg)
m0 0.5943 1
/2 · ρ · T · L 2
0
I zz 0.0494 1
/2 · ρ · T · L 4
R 00 0.0697 0.0747 1
/2 · ρ · T · L
X v0 v -0.1999 -0.3889 1
/2 · ρ · T · L
X r0 r 0.1288 -0.0924 1
/2 · ρ · T · L 3
Y v0 -0.4003 -0.5998 1
/2 · ρ · T · L ·U t ug
Yr0 0.0161 0.0875 1
/2 · ρ · T · L 2 ·U t ug
Y v0 v v -1.2109 -0.3269 1
/2 · ρ · T · L · 1/Ut ug
Yr0r r 0.9199 0.0 1
/2 · ρ · T · L 4 · 1/Ut ug
N v0 -0.1692 -0.0624 1
/2 · ρ · T · L 2 ·U t ug
Nr0 -0.0639 -0.1004 1
/2 · ρ · T · L 3 ·U t ug
N v0 v v 0.1184 -0.4149 1
/2 · ρ · T · L 2 · 1/Ut ug
c -0.0723 0.0286 -
N v0 /Y v0 0.4227 0.1040 -

3.6. Chapter summary


In this chapter the methodology of the research project is presented. The requirements imposed on the wave
drift force components for the tow resistance are drafted, resulting in several irregular wave spectra that need
to be examined. For the towing stability, the requirements are a bit more strict; defined limits of the parame-
ters for the wave drift force components: u, v, r , ψ, ω and ζa .

The wave drift estimation methods described in the literature chapter 2 have been categorised. For a station-
ary vessel, eleven estimation methods are selected for comparison while for a sailing vessel ten methods are
selected. The methodology for this comparison consists of a verification part, a computation part, and a tow
resistance assessment part. For the sailing vessel, a multi-criteria analysis is also included to determine the
most suitable method to use as input in the towing stability analysis.

For this towing stability analysis, the methodology consists of a verification part, a without waves assessment,
a with waves assessment and an impact of the waves part. The latter part should be able to conclusively
answer part of the thesis goal. At last, the simulation model used in the whole process is presented.
4
Stationary Vessel Wave Drift Assessment
In the stationary vessel wave drift assessment chapter, the vessel is examined without any vessel velocities
present. Multiple wave drift estimation methods are assessed for both the tow resistance and as preparation
of the sailing vessel assessment, where the wave drift force with vessel velocities included is examined. The
methods that are examined are again presented in figure 4.1.

Figure 4.1: Overview of the stationary vessel wave drift estimation methods

First the estimation methods are verified in section 4.1. They are then compared and discussed in section 4.2
and finally, the wave drift estimation methods are discussed in the light of the tow resistance in section 4.3.
An overview of this process is again presented in figure 4.2, with the exception that the multi-criteria analysis
is excluded as this is a step only for the sailing vessel assessment.

45
46 4. Stationary Vessel Wave Drift Assessment

Figure 4.2: Visualisation of the estimation comparison process

4.1. Verification of the estimation method computations


Before the comparison of the different estimation methods is started, the computation of the estimation
methods need to be verified. The computation and the validity of the methods in respect to the requirements
needs to be checked. Most requirements are already met and discussed in chapter 3. Since there are no vessel
velocities involved for the stationary vessel, the requirements regarding the velocities are also automatically
satisfied. This leaves the computation of the methods. The computation and verification of all the different
methods is featured in appendix C. Based on the verification, the validity boundaries are modified.

4.1.1. Application assessment


Before all the methods are verified, their application theory is shortly assessed. The theoretical validity of the
methods is presented in the second and third columns of table 4.1. It describes the theoretical quantification
of the range where the methods contribute to the wave drift force. They can describe the whole, or part of
the wave drift force in this range. These values are verified in appendix C. The adjusted boundaries, now
describing the boundaries where they describe the whole wave drift force, can be found in the fourth and
fifth columns of table 4.1.

The estimation methods all have their assumptions, and in table 4.2 an overview is given of the assumptions
relating to the vessel characteristics and properties. The strip theory methods [62] are limited to ships with
"simple" bow shapes. Generally they do provide satisfactory results. The short wave method by Faltinsen
[19] should be applied for blunt vessels with wall sided waterlines. Fujii’s and Takahashi’s [26], and Kashiwagi
et al.’s [41] methods are also limited to blunt hulls since they are both originally based on the same set of
data. With all the SW methods, only the surge and sway can be computed. The assumptions included for
R-function method [63] depend on the components that are used in the computation. Since the motion part
is usually computed with strip theory, and the reflection part with SW methods, those assumptions also hold
for the R-function method. The DNV formulations [12, 14] are allowed to be used for arbitrary vessels, but
they are engineering approximations that have limited accuracy in their predictions. They also only predict
the surge component. Again, all these methods are addressed in this chapter, even if some are only applicable
to certain vessels.
4.2. Estimation comparison 47

Table 4.1: Boundary values L/λ for the stationary vessel wave drift estimation methods

Lower theoretical Upper theoretical Lower Upper


Category boundary boundary boundary boundary
L/λ L/λ L/λ L/λ
FF Method 0.2 5.4 0.2 5.4
NF Method 0.2 5.4 0.2 3.5
ST Method 0.2 3.5 0.2 ∼ 2.5
SW Method Faltinsen 1.5 5.4 1.5 5.4
SW Method Liu 1.5 5.4 3.5 5.4
SW Method Fujii & Takahashi 1.5 5.4 3.5 5.4
SW Method Kuroda et al. 1.5 5.4 3.5 5.4
SW Method Kashiwagi et al. 1.5 5.4 3.5 5.4
R-function Method 0.2 5.4 0.2 5.4
DNV 2014 Formulation 0.2 5.4 - -
DNV 1996 Formulation 0.2 5.4 - -

Table 4.2: Estimation method assumptions relating to the vessel characteristics

Arbitrary Arbitrary All force


Comments
vessel shape waterline components
Far Field method [49] x x x
Near Field method [59] x x x
Strip Theory method [62] x x Limited complexity of bow
area
SW method Faltinsen [19] Blunt vessel with wall-sided
waterline
SW method Liu [46] x x
SW method Fujii & Takahashi [26] x Blunt vessel
SW method Kuroda et al. [43] x x
SW method Kashiwagi et al. [41] x Blunt vessel
R-function method [63] x Depends on used components
DNV 2014 formulation [14] x x Only surge
DNV 1996 formulation [12] x x Only surge

4.1.2. Updated validity boundaries


The theoretical boundaries are updated to adjusted boundaries describing the boundaries where the estima-
tion method describes the whole wave drift force. As already mentioned, the new boundaries are presented in
columns three and four of table 4.1. Here the NF method [59] is adjusted due to the slight frequency increase
in short waves, the ST method because it only describes part of the force in the transition zone and the short
wave methods because they only describe the entire wave drift force in the reflection zone. DNV’s formula-
tions are not further examined as their application range is not deemed viable for describing the wave drift
force in detail. For the tow resistance they are shortly revisited later on. The vessel used in the examination of
the vessels is a rectangular barge, satisfying all the assumptions of the estimation methods presented in table
4.2.

4.2. Estimation comparison


In this section the estimation method computations are compared. This comparison is split for the three force
components. Together with the visual comparison of the different estimation methods the mean square error
(MSE) of the methods is calculated and presented. This is done on the basis of the FF method base value. The
regular wave results are plotted with the three excitation zones: the motion, transition and reflection zone.
They are indicated with vertical lines. The irregular waves results are limited in their range by the H s - T p
relation. The physical range is also marked by the vertical lines.
48 4. Stationary Vessel Wave Drift Assessment

4.2.1. Stationary vessel surge wave drift force


For the surge force, the short wave methods are compared separately to reduce the number of lines in one
graph. The regular wave short wave methods are compared in sub-figure 4.3a while the irregular waves re-
sults can be found in sub-figure b. Note that the Faltinsen method only describes the theoretical limit of the
force, while the other short wave methods describe the reflection induced part of the wave drift force. This
frequency dependent part included in the SW methods is also called the R-function.

For this particular vessel, the Faltinsen SW method completely predicts the FF method for L/λ > 1. This
is a coincidence. It can be expected to decently predict the force in the transition zone and match the FF
method in the reflection zone. The latter is also what is observed at the other SW methods. Liu, the adjusted
Faltinsen SW method, underpredicts the FF method slightly, but generally matches in the reflection zone.
Fujii & Takahashi’s (F&T), Kuroda’s et al. and Kashiwagi’s methods produce the same result for a stationary
vessel. The latter two are variations of the initial method of F&T, and they only differ with velocities involved.
The wave spectra are also plotted in sub-figure a. It is clear that most wave energy is concentrated below
L/λ < 1 and that almost no energy is present beyond L/λ > 2.5. The energy is thus mostly located in the
motion zone. The irregular wave results in sub-figure b show this as well, the reflection induced part in a sea
state with a significant wave height of 5m is zero.

In sub-figure c, the NF method describes a slightly larger peak than the FF method while the ST method peak
is almost twice as big. All peak values occur at L/λ = 1, which is reliable as this is where the largest relative
motion between the ship and the waves is expected. The NF and FF method match well in the transition
and reflection zone. The strip theory method, which is known from literature to present slightly unreliable
results for a stationary vessel, overpredicts the force in the entire motion zone and keeps a long tail, spanning
the whole transition zone. The R-function method, combining the strip theory results and the Faltinsen SW
method, thus also predicts a flawed curve of the QTF. From sub-figure d, it is clear that the irregular waves
results consist of the motion induced part only and that the peak value is not included in the range.

The mean square error results, quantifying the differences with base value of the far field method are pre-
sented in table 4.3. For the regular waves, the smallest errors in the motion and transition zone are made
by the NF method. The reflection zone generally has lower errors, and the R-function method based on
Salvesen’s and Faltinsen’s methods makes the smallest error. For the irregular waves, the error in the applica-
ble zone is smallest for the NF method. This is as expected as these results completely lie in the motion zone.
The large relative MSE values for the motion zone are correct, since the reference value, the FF method, is
very small for a portion of that zone.

Table 4.3: Stationary vessel surge force relative MSE with the FF method as base value

Motion zone Transition zone Reflection zone Physical T p


Method
MSE MSE MSE Range MSE
Near Field method 1.43 0.0078 0.0016 0.056
Strip Theory method 237123 0.24 - 7.41
SW method Faltinsen - 0.024 0.0026 26.58
SW method Liu - - 0.014 0.98
SW method Fujii & Takahashi - - 0.00070 0.98
SW method Kuroda et al. - - 0.00070 0.98
SW method Kashiwagi et al. - - 0.00070 0.98
Seo’s R-function method 237123 0.11 0.00030 7.74

4.2.2. Stationary vessel sway wave drift force


For the sway force results, the short wave methods are compared separately. The regular and irregular wave
results for all methods can be found in figure 4.4. The sway short wave methods exhibit the same character-
istics as the surge component in figure 4.3. A "peak" in the motion zone, although the sway peak is located a
higher L/λ ratio than the surge peak, after which the force curve levels off towards an asymptotic limit.

Faltinsen’s SW method, in sub-figure 4.4a describes the transition and reflection zone decently, although
not as accurate as the surge component in sub-figure 4.3a. Liu’s SW method differs slightly from the other
4.2. Estimation comparison 49

Mean Surge force, irregular waves H =5m


Mean Surge QTF, Regular waves, = 0° s
0.05 10
Far Field Method, = 0° 0.04 SW Method Faltinsen, = 0°
0 SW Method Faltinsen, = 0°
9 SW Method Liu, = 0°
SW Method Liu, = 0° SW Method F & T, = 0°
-0.05 SW Method F & T, = 0°
8 0.02 SW Method Kuroda et al., = 0°
SW Method Kuroda et al., = 0° SW Method Kashiwagi, = 0°
-0.1 SW Method Kashiwagi, = 0° 7
0
-0.15 6
-0.02
-0.2 5

-0.25 4 -0.04

-0.3 3
-0.06
-0.35 2

-0.4 Motion zone Transition zone Reflection zone 1 -0.08


Physical Tp range
-0.45 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.1
5 6 7 8 9 10 11 12 13 14 15

(a) Regular wave induced surge force: short wave methods (b) Irregular wave induced surge force: short wave methods

Mean Surge QTF, Regular waves, = 0° Mean Surge force, irregular waves H s
=5m
0.1 10
Far Field Method, = 0° 0.04 Near Field Method, = 0°
0 Near Field Method, = 0°
9
Far Field Method, = 0°
Strip Theory Method, = 0° Strip Theory Method, = 0°
-0.1 Seo R-function Method, = 0°
8 0.02 Seo R-function Method, = 0°

-0.2 7
0
-0.3 6

-0.4 5 -0.02

-0.5 4
-0.04
-0.6 3
-0.06
-0.7 2

-0.8 Motion zone Transition zone Reflection zone 1 -0.08


Physical Tp range
-0.9 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.1
5 6 7 8 9 10 11 12 13 14 15

(c) Regular wave induced surge force (d) Irregular wave induced surge force

Figure 4.3: Stationary surge force estimation method comparison for regular and irregular waves

methods and describes the FF sway force more accurately in the reflection zone. However, this difference is
marginal. Since the peak of the sway QTF is shifted to higher L/λ ratios, the energy of the wave spectra excites
even less of the QTF than for the surge force. In sub-figure b this shows, as the SW method contributions are
zero and the Faltinsen method is not theoretically applicable in the physical T p range.

In sub-figure c, the ST and R-function methods predict a much higher peak sway force, roughly ∼ 2.5 times
higher than the NF method result. The peak of the ST method sits at the same L/λ ratio as the peaks in the
surge force while the NF and FF methods have a shifted peak towards higher L/λ ratios. This is most likely
due to a phase shift between the roll motion and the relative wave-vessel motion since the peak lies at the
L/λ ratio with the highest relative motion excitation. Since the ST method is known to be off for a stationary
vessel, the latter result is trusted more. The difference between the NF and FF method for the sway force is
significantly larger than the difference for the surge force in sub-figure 4.3c.

This difference is discussed already as part of the mesh convergence study performed in appendix B. With
smaller panels the difference reduces, but there will be a remainder as the NF method has a different theo-
retical background, taking the vessel motions into account. The difference is significant, at L/λ = 5 the NF
method is twice the value of the FF method, but this is an inherent part of these methods. The difference be-
tween the strip theory based methods and the panel based methods is again very clear for the irregular wave
results in sub-figure d. Since the ST peak lies at a lower L/λ ratio, the force is excited more by the wave energy
in the wave spectra, enhancing the difference. This shows that only the FF and NF method results give trusted
results and that the maximum possible wave drift force should not be expected for this stationary vessel.
50 4. Stationary Vessel Wave Drift Assessment

In table 4.4 the quantified differences between the methods describing the sway force are presented. These
mean square errors are the normalised errors with regards to the base value. For the regular wave motion
zone the near field method has the smallest error, similar to the surge force, while the transition zone is best
described by the SW method by Faltinsen. This is again more a coincidence than theoretically justified as can
be noticed from sub-figure 4.4c. Here it is visible that the SW method is an average of the FF method but does
not really describe the force in detail. For the reflection zone, the SW method developed by Liu is by far the
most accurate. From the irregular wave results it is again the NF method that has the smallest error.

Table 4.4: Stationary vessel sway force relative MSE with the FF method as base value

Motion zone Transition zone Reflection zone Physical T p


Method
MSE MSE MSE Range MSE
Near Field method 0.020 0.23 0.64 0.060
Strip Theory method 871367 0.80 - 35.34
SW method Faltinsen - 0.030 0.060 17.32
SW method Liu - - 0.0060 0.99
SW method Fujii & Takahashi - - 0.050 0.98
SW method Kuroda et al. - - 0.050 0.98
SW method Kashiwagi et al. - - 0.050 0.98
Seo’s R-function method 871367 0.55 0.049 34.99

4.2.3. Stationary vessel yaw wave drift moment


For the comparison of the yaw moment computation only the near field, far field, strip theory and R-function
methods are compared. The results are found in figure 4.5. In sub-figure 4.5a, the regular wave results show
that both near field and far field methods yield a negative moment for the range 0 < L/λ ≤ 0.75. The explana-
tion for this moment against the wave direction can be found in the phase difference between the incoming
wave and the motion of the vessel. For short waves, higher L/λ ratios, the near field method decreases to
zero and slowly keeps on decreasing while the far field method keeps a semi-constant positive offset. The
decreasing nature of the NF method yaw moment seems to originate from a linear frequency dependency in
short waves. This can be a remnant of the mesh convergence, that in this case the moment has not converged
fully. More information on the mesh convergence study can be found in appendix B.

The Salvesen ST method and Seo’s R-function method effectively produce the same output. In short waves
the yaw moment is approximated as zero. For this particular vessel, the combination of a symmetrical front-
stern shaped vessel and short waves produces a yaw moment that is effectively zero. In the motion dominated
zone the strip theory method almost predicts a peak value three times as high as the NF and FF methods.
This matches the higher peaks found in the surge and sway forces. Again, the irregular wave results in sub-
figure 4.5b show that the vessel remains in the motion dominated zone, that the ST method based results
overpredict and that the maximum yaw moment is not expected in this stationary conditions sea state.

The mean square errors are computed and presented in table 4.5. Here, the errors are the normalised errors
with regards to the FF method, the base value. Relatively, the R-function method performs better in short
waves since the NF method becomes negative while the R-function method remains zero. In the motion and
transition zones, the differences between the ST method and NF and FF methods are large and the ST results
seem to be over predicting the yaw component. It is clear the NF method has the smallest MSE with respect
to the FF method.
Table 4.5: Stationary vessel yaw moment relative MSE with the FF method as base value

Motion zone Transition zone Reflection zone Physical T p


Method
MSE MSE MSE Range MSE
Near Field method 0.32 0.23 1.77 0.78
Strip Theory method 350.70 1.01 - 26761
Seo’s R-function method 350.60 1.01 1.00 26721
4.2. Estimation comparison 51

Mean Sway QTF, Regular waves, = 15 ° Mean Sway force, irregular waves H s
=5m
0.2 10 0.04
Far Field Method, = 15° SW Method Faltinsen, = 15°
SW Method Faltinsen, = 15°
9 0.035 SW Method Liu, = 15°
SW Method Liu, = 15° SW Method F & T, = 15°
0.15 SW Method F & T, = 15°
8 0.03 SW Method Kuroda et al., = 15°
SW Method Kuroda et al., = 15° SW Method Kashiwagi, = 15°
SW Method Kashiwagi, = 15° 7 0.025

0.1 6 0.02

5 0.015

0.05 4 0.01

3
0.005
0 2
0
Motion zone Transition zone Reflection zone 1
-0.005 Physical Tp range
-0.05 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.01
5 6 7 8 9 10 11 12 13 14 15

(a) Regular wave induced sway force: short wave methods (b) Irregular wave induced sway force: short wave methods

Mean Sway QTF, Regular waves, = 15 ° Mean Sway force, irregular waves H s
=5m
0.4 10 0.04
Far Field Method, = 15° Near Field Method, = 15°
0.35 Near Field Method, = 15°
9 0.035 Far Field Method, = 15°
Strip Theory Method, = 15° Strip Theory Method, = 15°
0.3 Seo R-function Method, = 15°
8 0.03 Seo R-function Method, = 15°

0.25 7 0.025

0.2 6 0.02

0.15 5 0.015

0.1 4 0.01

0.05 3
0.005
0 2
0
-0.05 Motion zone Transition zone Reflection zone 1
-0.005 Physical Tp range
-0.1 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.01
5 6 7 8 9 10 11 12 13 14 15

(c) Regular wave induced sway force (d) Irregular wave induced sway force

Figure 4.4: Stationary sway force estimation method comparison for regular and irregular waves

Mean Yaw QTF, Regular waves, = 15 ° Mean Yaw moment, irregular waves H =5m
10-3 s
0.08 10 6
Far Field Method, = 15° Near Field Method, = 15°
0.07 Near Field Method, = 15°
9 Far Field Method, = 15°
5
Strip Theory Method, = 15° Strip Theory Method, = 15°
0.06 Seo R-function Method, = 15°
8 Seo R-function Method, = 15°
4
0.05 7

0.04 6 3

0.03 5 2

0.02 4
1
0.01 3
0
0 2

-0.01 Motion zone Transition zone Reflection zone 1 -1 Physical Tp range


-0.02 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -2
5 6 7 8 9 10 11 12 13 14 15

(a) Regular wave induced yaw moment (b) Irregular wave induced yaw moment

Figure 4.5: Stationary yaw moment estimation method comparison for regular and irregular waves
52 4. Stationary Vessel Wave Drift Assessment

4.3. Tow resistance assessment


The comparison of the wave drift estimation methods gives a good indication on the applicability of these
methods. Together with the verified ranges presented in section 4.1 these methods can be put in context. In
this context, it is important to realise that capturing the full wave drift force, even on the conservative side,
is vital. If the force estimation is too low, this can results in hazardous situations. Taking this into account,
the irregular wave drift results are examined. For the stationary tow resistance, an equilibrium position is
assumed. Therefore, only the surge and sway contributions are of interest, with an emphasis on the surge
component since yaw angles are considered small or non-existent.

The DNV approximations, not considered for describing the wave drift force components in detail, are plotted
in figure 4.6. These engineering formulas could be used as conservative approximations of the surge force.
In sub-figure 4.6a, the regular wave results are plotted. Both the FF [49] and NF method [59] are plotted as
reference values to put the DNV computations into perspective. As is known from the literature study in
chapter 2, the DNV formulation 2014 [14] is twice the short wave asymptotic limit and thus captures the QTF
peak in the motion zone seemingly at all times. Still it is a very conservative approach, especially considering
the irregular wave results in sub-figure b. Since the stationary conditions sea state does not contain much
energy where L/λ > 1 the mean surge force in irregular waves is low compared to the DNV 2014 prediction.
This overestimation, roughly a factor 4, is significant. It can be noted that if the test vessel is longer, the peak
of the mean wave drift force will move towards the physical T p range and the approximation becomes more
applicable.

The DNV formulation 1996 [12] on the other hand underpredicts the force by roughly a factor two. The cause
of this is not clear as this formulation is an empirical formula. In sub-figure a, it is clear this method is not
usable for predicting the wave drift surge force accurately, but in sub-figure b, it gives a slightly conservative
prediction. This is a coincidence but might still be useful.

Mean Surge QTF, Regular waves, = 0° Mean Surge force, irregular waves H s
=5m
0.1 10
Far Field Method, = 0° 0.04 FF Method, = 0°
0 Near Field Method, = 0°
9
NF Method, = 0°
DNV Formulation 2014, = 0° DNV Formulation 2014, = 0°
-0.1 DNV Formulation 1996, = 0°
8 0.02 DNV Formulation 1996, = 0°

-0.2 7
0
-0.3 6

-0.4 5 -0.02

-0.5 4
-0.04
-0.6 3
-0.06
-0.7 2

-0.8 Motion zone Transition zone Reflection zone 1 -0.08


Physical Tp range
-0.9 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.1
5 6 7 8 9 10 11 12 13 14 15

(a) Regular wave induced surge force QTF (b) Irregular wave induced mean surge force

Figure 4.6: Stationary DNV approximations comparison for regular and irregular waves

Based on the results of the comparison in section 4.2 and figure 4.6, the followings methods are recommended
for estimating the wave drift surge force in a tow resistance setting:

• Near field/Far field method;


• Faltinsen SW method;
• DNV 2014 approximation.

The near field method generally predicts larger surge wave drift forces than the far field methods, especially
noticeable around the peak in the motion zone. Both describe the envelope of the force which provides
insight in the exact phenomena but they take relatively long to implement for an arbitrary vessel. Since both
are computed using Delfrac, the method that converges best for the particular vessel should be used. The
DNV 2014 approximation provides a very conservative approximation of the surge wave drift force, but it is
4.4. Chapter summary 53

quickly assessed for an arbitrary vessel. Depending on the purpose, the theoretical and less conservative
Faltinsen SW method could be adopted for this purpose as well.

Since no engineering approximations are present for the sway force, conventionally only the surge force is
estimated for the tow resistance, the sway force can solely be estimated by using theoretical methods. The
sway force can be best estimated by using methods:

• Near field/Far field method;


• Faltinsen SW method.

The same arguments for the near field and far field methods at the surge force apply for the sway force. The
Faltinsen SW method provide a decent, quicker alternative if the only purpose is the added tow resistance
due to the waves.

4.4. Chapter summary


In this chapter the wave drift force estimation methods for a stationary vessel are assessed. This is done
in context of the tow resistance, and as preparation for the sailing vessel wave drift assessment in chapter 5.
First, as the methodology prescribed, the computation of all the estimation methods considered were verified
and discussed on the basis of their theoretical applicability. The DNV formulations did not describe the curve
of the QTF and were not considered further in the estimation method comparison.

The estimation methods are compared and assessed based on the three excitation zones: the motion, tran-
sition and reflection zones. This is done for all three force components, the surge and sway force plus the
yaw moment. Where necessary, the applicability range, quantified by the L/λ ratio, was modified. This was
necessary for the strip theory method, which tends to over-predict the forces/moments, and the short wave
methods, which are only truly valid in the reflection zone.

Both regular and irregular waves are addressed, where the irregular waves results are computed based on
the adopted stationary conditions sea state with H s = 5 m and the regular wave results. The results from the
comparison of the eleven estimation methods are used to assess the tow resistance, and serve as building
block for the sailing vessel wave drift estimation method assessment.

The near field method, far field method and DNV 2014 formulation, although conservative, are considered
most applicable for the stationary tow resistance assessment. The Faltinsen SW method provides another
alternative.

The wave drift estimation methods that are compared are in turn used in the sailing vessel wave drift assess-
ment chapter 5 to include the surge, way and yaw velocities.
5
Sailing Vessel Wave Drift Assessment
In this chapter the results of the stationary vessel wave drift assessment chapter are used as the basis of the
assessment of the wave drift forces with vessel velocities. The same approach as in chapter 4 is used to anal-
yse the estimation methods of the wave drift force with vessel velocities included. The assessed estimation
methods are again presented in figure 5.1.

Figure 5.1: Overview of the sailing vessel wave drift estimation methods

To assess them, first the methods are verified in section 5.1. Special interest is given to the computation of
the wave drift damping and the effect of the encounter frequency. The implementation and computation of
the remaining methods are further discussed in appendix C. The computation results are then compared and
discussed in section 5.2. Then, the wave drift in a tow resistance setting is discussed in section 5.3 and finally,
several multi-criteria analyses are used in section 5.4 to determine the most suitable estimation method for
including the wave drift force with (forward) velocities in the towing stability analysis. An overview of this
process is again presented in figure 5.2.

55
56 5. Sailing Vessel Wave Drift Assessment

Figure 5.2: Visualisation of the estimation comparison process

5.1. Verification of the estimation method computations


The first step in determining the most suitable sailing vessel wave drift estimation methods is the verification.
The first check is done for the general requirements related to velocities.

The Froude number must remain below F n < 0.15, which implies for the examined vessel that the maximum
velocity that is allowed is u = 4.1m/s. Since the maximum examined velocity is chosen as 4m/s this is no
problem and the approach remains valid.

To be able to use the wave drift damping method, flow separation should not occur. Although the current is
considered as a relative (forward) velocity, the physical phenomenon is different from vessel velocities. The
interaction between the current and waves occurs everywhere in the domain while the interaction between
the waves and vessel velocities only occurs around the hull. If current is taken into account next to the forward
velocity, it means that Uc /(ω · ζa ) < 1 should remain valid.

As was done in the stationary vessel wave drift assessment chapter 4 the different methods are all separately
verified in appendix C.

5.1.1. Application assessment


Similar to the previous chapter, the theoretical application of the different estimation methods is assessed.
The theoretical validity range of the methods is presented in the second and third columns of table 5.1 based
on the L/λ ratio. It is the theoretical quantification of the range where the methods contribute to the wave
drift force. They can describe the whole or part of the wave drift force in this range.

The estimation methods all have their assumptions, and in table 5.2 an overview is given of the assumptions
related to the vessel characteristics and properties. The NF [59] and FF [49] methods combined with the
wave drift damping can overpredict the force when the ship motions are large, so in the motion zone. The
strip theory methods [62] are practically limited to ships with "simple" bow shapes but generally do provide
satisfactory results. They are however only able to include the surge velocity. The short wave method by
Faltinsen [19] should be limited to blunt vessels with wall sided waterlines. In the verification process the
surge component of this method is adjusted from a velocity dependence of 2ω 4ω
g to g . More information on
this is found in appendix C. Fujii’s and Takahashi’s [26], and Kashiwagi’s et al. [41] methods are also limited to
blunt hulls since they are both based on the same set of data. For all SW methods, only the surge and sway can
be computed.The assumptions included for Seo’s R-function method [63] depend on the components that are
5.1. Verification of the estimation method computations 57

used in the computation. Since the motion part is usually computed with strip theory, and the reflection part
with SW methods, those assumptions also hold for the R-function method. The IMO formulation [68] only
predicts the surge component.

Table 5.1: Boundary values L/λ for the sailing vessel wave drift estimation methods

Lower theoretical Upper theoretical Lower Upper


Category boundary boundary boundary boundary
L/λ L/λ L/λ L/λ
FF Method - WDD 0.2 5.4 0.2 5.4
NF Method - WDD 0.2 5.4 0.2 3.5
ST Method 0.2 3.5 0.2 1.5
SW Method Faltinsen 1.5 5.4 1.5 5.4
SW Method Liu 1.5 5.4 3.5 5.4
SW Method Fujii & Takahashi 1.5 5.4 3.5 5.4
SW Method Kuroda et al. 1.5 5.4 3.5 5.4
SW Method Kashiwagi et al. 1.5 5.4 3.5 5.4
R-function Method 0.2 5.4 0.2 5.4
IMO Formulation 0.2 5.4 - -

Table 5.2: Estimation method assumptions relating to the vessel characteristics

Arbitrary Arbitrary All force


Comments
vessel shape waterline components
Far Field method - WDD [2, 49] x x x Can over predict in the motion
zone
Near Field method - WDD [2, 59] x x x Can over predict in the motion
zone
Strip Theory method [8, 28] x x Limited complexity of bow
area, Only includes surge
velocity
SW method Faltinsen [19] Blunt vessel with wall-sided
waterline
SW method Liu [46] x x
SW method Fujii & Takahashi [26] x Blunt vessel
SW method Kuroda et al. [43] x x
SW method Kashiwagi et al. [41] x Blunt vessel
R-function method [63] x Depends on used components
IMO formulation x x Only surge

5.1.2. Updated validity boundaries


For the sailing vessel assessment the theoretical boundaries are updated to adjusted boundaries describing
where the estimation method describes the whole wave drift force. The boundaries shift with the encounter
frequency, but this is not taken into account since these boundaries describe the vessel behaviour as the
vessel experiences it on the basis of the L/λ ratio. The new boundaries are based on the verification process
provided in appendix C and is presented in columns four and five of table 5.1.

As was done in the stationary vessel assessment, the NF method is adjusted due to the slight frequency in-
crease in short waves, the ST method describes only the motion zone and the short wave methods only de-
scribe the entire wave drift force in the reflection zone. The IMO formulation is not further examined for
the towing stability purpose as it does not describe the curve of the QTF and does not predict the magnitude
correctly. Its application range is not deemed viable.
58 5. Sailing Vessel Wave Drift Assessment

5.1.3. Wave drift damping


The wave drift forces on the vessel need to be known since the wave drift damping is based on the deriva-
tives to both direction and frequency of these forces. Besides the regular force components, the first order
and second order moment contributions are also necessary. This can be done on the basis of the approach
presented by Aranha [2, 3] and explained in more detail in the literature chapter 2.

The wave drift damping matrix is a 3x3 matrix, dependent on wave frequency ω and incoming wave angle of
attack β. First the wave drift components will be examined on their frequency dependency. The coefficients
are presented in figure 5.3. The coefficients presented are for 15° wave angle of attack β, since this is an angle
that is frequently encountered. The WDD coefficients are normalised with the wave frequency. Since the
coefficients are frequency dependent, a (semi-)constant coefficient has a linear frequency dependency. The
terms look reasonable, coupling terms B 21 and B 12 are similar and their signs follow logically from the sign
convention. Terms B 62 and B 26 are also alike, which follows from the close relation between sway and yaw for
slender vessels. Terms B 61 and B 16 are reasonably similar and show that the impact of the surge velocity on
the yaw moment is a lot larger than the impact of a yaw velocity on the surge force. The diagonal coefficients
(B 11 , B 22 and B 66 ) all are similar to the forces and moment envelopes as could be expected.

In figure 5.4, the complete wave drift damping matrix is plotted. They show the frequency ω and incoming
wave angle of attack (direction) β dependency. This is done for +45° to −45° from head waves. In head
waves, the coupling terms are small and the sway and yaw terms have limited influence. As is clear, the WDD
increases with increasing angle of attack and is expected to increase up to beam waves.

In detail, coefficient B 11 in sub-figure 5.4a shows the motion peak around ω = 1 rad/s over the entire angle
range while the higher frequency values are larger around small wave angles of attack. Coefficient B 12 in sub-
figure b shows that the impact of the sway velocity is mirrored around head waves. The motion peaks are
clearly visible, but with opposing magnitudes. Coefficient B 21 also shows this mirrored effect but, as it is the
sway force, has less of a peak and a more constant envelope. Sub-figure d shows B 22 , which clearly shows the
increase of the coefficient with increasing angles of attack. This increase is symmetrical and will keep rising
until β ± 90°. Coefficient B 61 , the in N z -u coupling coefficient in sub-figure e, shows the mirroring and the
more oscillating behaviour with respect to the frequency. The signs of the coefficient match B 21 , which can
be expected since the yaw velocity can be described as a sway velocity divided by an arm. Coefficient B 62 and
B 26 in sub-figures f and h are very similar which follows naturally from the sway and yaw motions and the
slender body approximation applied to obtain the yaw velocity r terms. In both figures, it can be seen that
the coefficients are symmetrical and have peaks around ω = 1 rad/s.

Coefficient B 16 , in sub-figure g, is relatively small compared to its coupling term B 61 . The same curve can be
recognised, but it is clear that the yaw velocity has a lot less impact on the surge force than the surge velocity
on the moment. The final coefficient B 66 , the yaw velocity r - yaw moment N z term has the same curve as
the sway yaw coupling terms B 62 and B 26 . The difference lies in the magnitude as the B 66 term is normalised
with an extra ship length L.

5.1.4. Wave encounter frequency


The last issue that is checked is the change from the wave frequency ω0 to the encounter frequency ωe . This
is described by equation (2.35) and again in equation (5.1).

ω20 ¡
ωe = ω0 + u · cos β − v · sin β
¡ ¢ ¡ ¢¢
(5.1)
g
The conventionally computed wave drift force describes the frequency dependency of the QTF as the vessel
experiences the dependency. For a sailing ship, the vessel experiences the encounter frequency instead of
the wave frequency, so when it is preferred that the results are presented based on the wave frequency a
conversion is necessary. This conversion results in the QTF shifting to lower frequencies. Since the actual
wavelength does not change, the QTF based on the L/λ ratio remains the same. Since this is important in the
simulation, this shift is incorporated in the results with forward velocity. For the comparison cases the surge
velocity u and sway velocity v are taken for respectfully the surge and sway forces. For yaw, v is taken. The
shift is shown in figure 5.5 for two velocities. What is clear is that the shift for the surge force is larger than for
the sway and yaw components. This is because they are addressed at β = 15° so the sway velocity influence is
relatively small in comparison to the surge velocity in head waves.
5.1. Verification of the estimation method computations 59

Wave Drift Damping B 11 Coefficient, = 15 ° Wave Drift Damping B 12 Coefficient, = 15 ° Wave Drift Damping B 16 Coefficient, = 15 °
10-4 10-4 10-5
2 3 8

0 6
2
-2
4

-4 1
2
-6
0 0
-8
-2
-10 -1

-4
-12
-2
-14 -6

-16 -3 -8
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) B 11 : F x - u Coefficient (b) B 12 : F x - v Coefficient (c) B 16 : F x - r Coefficient

Wave Drift Damping B 21 Coefficient, = 15 ° Wave Drift Damping B 22 Coefficient, = 15 ° Wave Drift Damping B 26 Coefficient, = 15 °
10-4 10-5 10-5
3 5 1.5

2 4
1

1 3
0.5

0 2

0
-1 1

-0.5
-2 0

-3 -1 -1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.5 1 1.5 2

(d) B 21 : F y - u Coefficient (e) B 22 : F y - v Coefficient (f) B 26 : F y - r Coefficient

Wave Drift Damping B 61 Coefficient, = 15 ° Wave Drift Damping B 62 Coefficient, = 15 ° Wave Drift Damping B 66 Coefficient, = 15 °
10-5 10-5 10-6
8 1.5 7

6
6
1 5
4
4
2
0.5 3

0 2

0 1
-2
0
-4
-0.5 -1
-6
-2

-8 -1 -3
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.5 1 1.5 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(g) B 61 : N z - u Coefficient (h) B 62 : N z - v Coefficient (i) B 66 : N z - r Coefficient

Figure 5.3: The wave frequency dependency of the Wave Drift Damping (WDD) coefficients with β = 15° for Barge 1
60 5. Sailing Vessel Wave Drift Assessment

-45° Wave Drift Damping B 11 Coefficient -45° Wave Drift Damping B 12 Coefficient
-30° 10 -3 -30°
0 1

-15° 0.8 -15°

0.6
0° 0°
0.4
-0.5
0.2

15° 15°
0

-0.2
-1
-0.4
30° 30°
-0.6

-0.8

-1.5 45° -1 45°


10 -3 0.05 0.56 1.1 1.6 2.1 0.05 0.56 1.1 1.6 2.1
[rad/s] [rad/s]

(a) B 11 : F x - u Coefficient (b) B 12 : F x - v Coefficient

-45° Wave Drift Damping B 22 Coefficient


-45° Wave Drift Damping B 21 Coefficient 10 -4 -30°
10 -3 -30° 7
1
-15°
0.8 -15°
6
0.6

0° 5
0.4

0.2
4
15° 15°
0
3
-0.2

-0.4 2
30° 30°

-0.6
1
-0.8

-1 45° 0 45°
0.05 0.56 1.1 1.6 2.1 0.05 0.56 1.1 1.6 2.1
[rad/s] [rad/s]

(c) B 21 : F y - u Coefficient (d) B 22 : F y - v Coefficient

Wave Drift Damping B 61 Coefficient -45° Wave Drift Damping B 62 Coefficient


10 -4
-45° 10 -5 -30°
1.5 -30° 12

-15° 10
-15°

1
8
0° 0°

0.5 6

4
15° 15°
0
2

-0.5 0
30° 30°
-2
-1
-4

-1.5 45° -6 45°


0.05 0.56 1.1 1.6 2.1 0.05 0.56 1.1 1.6 2.1
[rad/s] [rad/s]

(e) B 61 : N z - u Coefficient (f) B 62 : N z - v Coefficient


5.1. Verification of the estimation method computations 61

-45° Wave Drift Damping B 16 Coefficient


10 -4 -30°
1.5

-15°
1


0.5

15°
0

-0.5
30°

-1

-1.5 45°
0.05 0.56 1.1 1.6 2.1
[rad/s]

(g) B 16 : F x - r Coefficient

-45° Wave Drift Damping B 26 Coefficient


10 -5 -30°
12

10
-15°

8

6

4
15°

0
30°
-2

-4

-6 45°
0.05 0.56 1.1 1.6 2.1
[rad/s]

(h) B 26 : F y - r Coefficient

-45° Wave Drift Damping B 66 Coefficient


10 -5 -30°

-15°
4

3 0°

15°
1

0
30°

-1

-2
45°
0.05 0.56 1.1 1.6 2.1
[rad/s]

(i) B 66 : N z - r Coefficient

Figure 5.4: The frequency ω and incoming angle of attack β dependency of the Wave Drift Damping (WDD) coefficients for Barge 1
62 5. Sailing Vessel Wave Drift Assessment

Mean Surge QTF, Regular waves Mean Sway QTF, Regular waves
0.2 0.7

0
0.6

-0.2
0.5
-0.4

0.4
-0.6

-0.8 0.3

-1
0.2
-1.2

0.1
-1.4

-1.6 0
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Near Field Method ( ), u = 0 m/s, = 0° Near Field Method ( e ), v = 0 m/s, = 15°


e
Near Field Method ( ), u = 2 m/s, = 0° Near Field Method ( e ), v = -1 m/s, = 15°
e
Near Field Method ( ), u = 4 m/s, = 0° Near Field Method ( ), v = -2 m/s, = 15°
e e
Far Field Method ( ), u = 0 m/s, = 0° Far Field Method ( ), v = 0 m/s, = 15°
e e
Far Field Method ( ), u = 2 m/s, = 0° Far Field Method ( e ), v = -1 m/s, = 15°
e
Far Field Method ( ), u = 4 m/s, = 0° Far Field Method ( ), v = -2 m/s, = 15°
e e

(a) Surge wave drift force with surge velocity for u = 0, 2, 4m/s (b) Sway wave drift force with sway velocity for v = 0, 2, 4m/s

Mean Yaw QTF, Regular waves


0.1

0.08

0.06

0.04

0.02

-0.02

-0.04

-0.06
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Near Field Method ( e ), v = 0 m/s, = 15°


Near Field Method ( e ), v = -1 m/s, = 15°
Near Field Method ( e ), v = -2 m/s, = 15°
Far Field Method ( e ), v = 0 m/s, = 15°
Far Field Method ( e ), v = -1 m/s, = 15°
Far Field Method ( ), v = -2 m/s, = 15°
e

(c) Yaw wave drift force with sway velocity for v = 0, 2, 4m/s

Figure 5.5: The encounter frequency shift of the regular wave QTFs
5.2. Estimation comparison 63

5.2. Estimation comparison


In this section the estimation method computation results are compared. This comparison is split for the
three force components: surge, sway and yaw. Together with the visual comparison of the different estima-
tion methods, the mean square error (MSE) of the methods with respect to the FF method is calculated and
presented. This method is chosen as the base value because its theoretical background is solid and Delfrac
is a verified computation program. These are used as input for the tow resistance assessment and for the
multi-criteria analysis for determining the most suitable method for the stability analysis.

5.2.1. Sailing vessel surge wave drift force


For the surge force, the short wave methods are compared first. The Faltinsen [19] and Liu [46] regular wave
short wave methods are compared in sub-figure 5.6a while the irregular waves results can be found in sub-
figure b.

Sub-figure 5.6a shows the regular wave limits and it can be seen that both the Faltinsen and Liu limit are
frequency dependent, increasing linearly with the frequency. When comparing the 1 m/s results with the FF
method, it can be seen that within the verified boundaries, the transition and reflection zone, the Faltinsen
SW method gives a fine approximation. In the reflection zone, both the Liu and Faltinsen methods match
the FF method decently. With a higher surge velocity, the differences between the FF method and the SW
methods slightly increase, but the match is still decent. The wave spectra for the sailing conditions are also
plotted in sub-figure a. Most of the wave energy is located below L/λ ∼ 2.5. With this spectrum range, the
irregular wave results are computed in sub-figure 5.6b. The force in the physical T p range lies in the motion
and transition zone, as also seen in sub-figure a, and it is clear that the maximum surge force can be expected
in this sea state. For both forward velocities, the Faltinsen SW method gives a conservative approximation
but does not capture the envelope of the curve. The Liu method completely fails to describe the FF method,
which is expected as it is only applicable in short waves.

In the stationary vessel assessment chapter 4 the SW methods of Fujii & Takahashi [26], Kuroda et al. [43] and
Kashiwagi et al. [41] all resulted in the same approximation. In sub-figure 5.6c, the regular wave results are
presented and clear differences are spotted. For u = 1 m/s, the F&T method provides the best approximation
in the reflection zone, while the Kashiwagi et al. SW method overpredicts the FF method and the Kuroda et al.
method underpredicts. It can be clearly seen that none of these SW methods include the linear increase with
frequency like seen for the Faltinsen en Liu SW methods. These three SW methods fail to capture that aspect
of the QTF. The location where the method start contributing to the wave drift force is dependent on the wave
frequency or encounter frequency. This differs per method. For 3 m/s, more distinct differences are visible.
The Kuroda et al. method underpredicts the FF method completely and is not suitable of predicting the FF
method. The F&T method underpredicts the force as well but significantly less. The Kashiwagi et al. SW
method overpredicts the force but provide the best estimation of the FF methods of these three short wave
methods. In sub-figure 5.6d, the irregular wave results show that since these methods are only applicable in
the reflection zone, they are not an option for predicting the irregular surge wave drift force in the physical
T p range.

Sub-figure 5.6e shows the FF [49], NF [59], Strip Theory [62] and R-function methods [63]. In this case, the
R-function method is based on the strip theory and Faltinsen SW method. For u = 1 m/s, the motion zone
results for the NF and FF predict similar results although the NF peak is slightly larger. The ST and R-function
peaks match the magnitude of the FF method but have a slight shift in frequency. The transition and reflection
zone results shows that there is little difference between the NF and FF method for the surge force. The R-
function method is limited in the transition zone, mainly due to the steep drop in the strip theory estimation
of the force. In the reflection zone, the method equals the Faltinsen SW method and thus gives a decent
match with the FF method. For u = 3 m/s, the differences increase. The NF method shows an increased
motion peak at L/λ = 1 while the peak of the FF method is 20% smaller. The peak of the strip theory and
R-function methods are again slightly lower than the FF method, but still provide a decent match. Since the
WDD is known to possibly overpredict the WDF in the motion zone, that might be the case for the NF and
FF method peaks here. The R-function method with surge velocity u > 2 m/s describes the whole QTF well.
The irregular wave results in sub-figure 5.6f show that the NF method clearly matches the FF method in the
physical T p range best. The other methods have a significantly different curve and do not provide a proper
estimation in the physical T p range.
64 5. Sailing Vessel Wave Drift Assessment

The mean square error results, quantifying the differences with base value of the FF method are presented
in table 5.3. For the regular waves, the smallest errors in the motion and transition zone are made by the NF
method. The reflection zone generally has lower errors, but again the NF method obtained the smallest error.
For the irregular wave results the error in the applicable zone is smallest for the NF method. Note that the
large values of the MSE in the motion zone are correct, since the reference value, the FF method, is very small
and thus the relative differences large.

Table 5.3: Sailing vessel surge force relative MSE with the FF method as base value

Motion zone Transition zone Reflection zone Physical T p


Method
MSE MSE MSE Range MSE
Near Field method - WDD 24.06 0.0092 0.0056 0.023
Strip Theory method 7561400 0.90 - 0.63
SW method Faltinsen - 0.019 0.0071 0.16
SW method Liu - - 0.0085 0.60
SW method Fujii & Takahashi - - 0.013 0.56
SW method Kuroda et al. - - 0.097 0.35
SW method Kashiwagi et al. - - 0.051 0.53
R-function method 7561500 0.032 0.071 0.37

Mean Surge force, irregular waves H = 2.4 m


Mean Surge QTF, Regular waves, = 0° s
0.2 10 0

0 9
-0.02

-0.2 8
-0.04
-0.4 7

-0.06
-0.6 6

-0.8 5 -0.08

-1 4
-0.1
-1.2 3
-0.12
-1.4 2

-1.6 Motion zone Transition zone Reflection zone 1 -0.14


Physical Tp range

-1.8 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.16
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, u = 1 m/s, = 0° Far Field Method, u = 1 m/s, = 0°


Far Field Method, u = 3 m/s, = 0° Far Field Method, u = 3 m/s, = 0°
Near Field Method, u = 1 m/s, = 0° Near Field Method, u = 1 m/s, = 0°
Near Field Method, u = 3 m/s, = 0° Near Field Method, u = 3 m/s, = 0°
SW Method Faltinsen, u = 1 m/s, = 0° SW Method Faltinsen, u = 1 m/s, = 0°
SW Method Faltinsen, u = 3 m/s, = 0° SW Method Faltinsen, u = 3 m/s, = 0°
SW Method Liu, u = 1 m/s, = 0° SW Method Liu, u = 1 m/s, = 0°
SW Method Liu, u = 3 m/s, = 0° SW Method Liu, u = 3 m/s, = 0°

(a) Regular wave induced surge force with wave spectra: short (b) Irregular wave induced surge force: short wave methods
wave methods
5.2. Estimation comparison 65

Mean Surge force, irregular waves H = 2.4 m


Mean Surge QTF, Regular waves, = 0° s
0.2 10 0

0 9
-0.02

-0.2 8
-0.04
-0.4 7

-0.06
-0.6 6

-0.8 5 -0.08

-1 4
-0.1
-1.2 3
-0.12
-1.4 2

-1.6 Motion zone Transition zone Reflection zone 1 -0.14


Physical Tp range

-1.8 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.16
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, u = 1 m/s, = 0° Far Field Method, u = 1 m/s, = 0°


Far Field Method, u = 3 m/s, = 0° Far Field Method, u = 3 m/s, = 0°
SW Method F & T, u = 1 m/s, = 0° SW Method F & T, u = 1 m/s, = 0°
SW Method F & T, u = 3 m/s, = 0° SW Method F & T, u = 3 m/s, = 0°
SW Method Kuroda et al., u = 1 m/s, = 0° SW Method Kuroda et al., u = 1 m/s, = 0°
SW Method Kuroda et al., u = 3 m/s, = 0° SW Method Kuroda et al., u = 3 m/s, = 0°
SW Method Kashiwagi, u = 1 m/s, = 0° SW Method Kashiwagi, u = 1 m/s, = 0°
SW Method Kashiwagi, u = 3 m/s, = 0° SW Method Kashiwagi, u = 3 m/s, = 0°

(c) Regular wave induced surge force with wave spectra: short (d) Irregular wave induced surge force: short wave methods
wave methods

Mean Surge force, irregular waves H = 2.4 m


Mean Surge QTF, Regular waves, = 0° s
0.2 10 0

0 9
-0.02

-0.2 8
-0.04
-0.4 7

-0.06
-0.6 6

-0.8 5 -0.08

-1 4
-0.1
-1.2 3
-0.12
-1.4 2

-1.6 Motion zone Transition zone Reflection zone 1 -0.14


Physical Tp range

-1.8 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -0.16
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, u = 1 m/s, = 0° Near Field Method, u = 1 m/s, = 0°


Far Field Method, u = 3 m/s, = 0° Near Field Method, u = 3 m/s, = 0°
Near Field Method, u = 1 m/s, = 0° Far Field Method, u = 1 m/s, = 0°
Near Field Method, u = 3 m/s, = 0° Far Field Method, u = 3 m/s, = 0°
Strip Theory Method, u = 1 m/s, = 0° Strip Theory Method, u = 1 m/s, = 0°
Strip Theory Method, u = 3 m/s, = 0° Strip Theory Method, u = 3 m/s, = 0°
Seo R-function Method, u = 1 m/s, = 0° Seo R-function Method, u = 1 m/s, = 0°
Seo R-function Method, u = 3 m/s, = 0° Seo R-function Method, u = 3 m/s, = 0°

(e) Regular wave induced surge force with wave spectra (f) Irregular wave induced surge force

Figure 5.6: Sailing surge force estimation method comparison


66 5. Sailing Vessel Wave Drift Assessment

5.2.2. Sailing vessel sway wave drift force


For the sway force results, the same split in the graphs as with the surge force is adopted. The regular and
irregular wave results for all methods can be found in figure 5.7. Note that the results are plotted with an
incoming wave angle of attack of β = 15°. This angle is encountered often and, unlike in head waves, it shows
the sway force contribution.

In sub-figure 5.7a, again the Faltinsen and Liu SW methods are plotted. The same curve as with the surge
force in sub-figure 5.6a can be seen. The match in the reflection zone with v = 1 m/s is limited, both methods
overpredict the FF method with roughly 50% while underpredicting the NF method by that same margin.
This is also the case for the Faltinsen SW method in the transition zone. With v = 3 m/s, a good match is still
absent. For both the transition zone and reflection zone the Faltinsen limit overpredicts the FF method while
underpredicting the NF method. The Liu SW method experiences the same in the reflection zone. In sub-
figure 5.7b the irregular wave results show that the maximum sway force can be expected in this sea state and
that the Faltinsen SW method gives a conservative estimation of the force, which becomes more conservative
for larger sway velocities.

The other SW methods, F&T, Kuroda et al. and Kashiwagi, are presented in sub-figures 5.7c and d. Again it
is clear that the linear increase with frequency is not present in these methods. For v = 1 m/s, the Kuroda
et al. method matches the FF method decently while the other two slightly overpredict. With v = 3 m/s the
differences are enhanced and the F&T method provides the closest match with the FF method. The Kashiwagi
et al. method overpredicts the force while the Kuroda et al. method underpredicts. As the methods are only
valid for the reflection zone, they do not capture the irregular wave results accurately in sub-figure d.

In sub-figure 5.7e, the regular wave results of the NF, FF, ST and R-function methods are presented. For v = 1
m/s the motion zone peak of the ST method overpredicts the FF and NF methods by 30%. For v = 3 m/s, the
ST peak matches the FF peak. Note that the ST results only take the surge velocity u into account, not v. In the
transition and reflection zone the difference between the FF and NF method grows larger. This is partially due
to the panel mesh used, more information on this is provided in appendix B. The R-function method provides
a decent alternative. At lower frequencies it overpredicts the peak in the motion zone due to the ST results
while at higher frequencies it predicts a force right between the FF and NF results. It overpredicts the FF
method while underpredicting the NF method. In sub-figure 5.7f, the irregular wave results show that none
of the other methods describe the FF methods decently in the physical T p range. The R-function method
provides a conservative approximation but it is clear it overpredicts the force in longer waves.

In table 5.4, the quantified differences between the methods describing the sway force are presented. For
the regular wave motion zone the near field method has the smallest error, similar to the surge force. In
the transition zone it is best described by the R-function method. For the reflection zone, the SW method
developed by Fujii and Takahashi is the most accurate with respect to the FF method. From the irregular
wave results it is again the NF method that has the smallest error.

Table 5.4: Sailing vessel sway force relative MSE with the FF method as base value

Motion zone Transition zone Reflection zone Physical T p


Regular waves
MSE MSE MSE Range MSE
Near Field method - WDD 0.031 0.87 2.62 0.12
Strip Theory method 224000 0.93 - 2.32
SW method Faltinsen - 0.55 0.71 0.72
SW method Liu - - 0.50 0.56
SW method Fujii & Takahashi - - 0.045 0.61
SW method Kuroda et al. - - 0.078 0.41
SW method Kashiwagi et al. - - 0.18 0.55
R-function method 224000 0.52 0.71 2.48
5.2. Estimation comparison 67

Mean Sway QTF, Regular waves, = -15 ° Mean Sway force, irregular waves H s
= 2.4 m
0.8 10 0.05

0.7 9 0.045

0.6 8 0.04

0.5 7 0.035

0.4 6 0.03

0.3 5 0.025

0.2 4 0.02

0.1 3
0.015

0 2
0.01

-0.1 Motion zone Transition zone Reflection zone 1


0.005 Physical Tp range

-0.2 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, v = 1 m/s, = -15° Far Field Method, v = 1 m/s, = 15°
Far Field Method, v = 3 m/s, = -15° Far Field Method, v = 3 m/s, = 15°
Near Field Method, v = 1 m/s, = -15° Near Field Method, v = 1 m/s, = 15°
Near Field Method, v = 3 m/s, = -15° Near Field Method, v = 3 m/s, = 15°
SW Method Faltinsen, v = 1 m/s, = -15° SW Method Faltinsen, v = 1 m/s, = 15°
SW Method Faltinsen, v = 3 m/s, = -15° SW Method Faltinsen, v = 3 m/s, = 15°
SW Method Liu, v = 1 m/s, = -15° SW Method Liu, v = 1 m/s, = 15°
SW Method Liu, v = 3 m/s, = -15° SW Method Liu, v = 3 m/s, = 15°

(a) Regular wave induced sway force with wave spectra: short (b) Irregular wave induced sway force: short wave methods
wave methods

Mean Sway force, irregular waves H s


= 2.4 m
Mean Sway QTF, Regular waves, = -15 ° 0.05
0.4 10

0.35 9 0.045

0.3 8 0.04

0.25 7 0.035

0.2 6 0.03

0.15 5 0.025

0.1 4 0.02

0.05 3 0.015

0 2 0.01

-0.05 Motion zone Transition zone Reflection zone 1 0.005 Physical Tp range

-0.1 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, v = 1 m/s, = -15° Far Field Method, v = 1 m/s, = 15°
Far Field Method, v = 3 m/s, = -15° Far Field Method, v = 3 m/s, = 15°
SW Method F & T, v = 1 m/s, = -15° SW Method F & T, v = 1 m/s, = 15°
SW Method F & T, v = 3 m/s, = -15° SW Method F & T, v = 3 m/s, = 15°
SW Method Kuroda et al., v = 1 m/s, = -15° SW Method Kuroda et al., v = 1 m/s, = 15°
SW Method Kuroda et al., v = 3 m/s, = -15° SW Method Kuroda et al., v = 3 m/s, = 15°
SW Method Kashiwagi, v = 1 m/s, = -15° SW Method Kashiwagi, v = 1 m/s, = 15°
SW Method Kashiwagi, v = 3 m/s, = -15° SW Method Kashiwagi, v = 3 m/s, = 15°

(c) Regular wave induced sway force with wave spectra: short (d) Irregular wave induced sway force: short wave methods
wave methods
68 5. Sailing Vessel Wave Drift Assessment

Mean Sway QTF, Regular waves, = -15 ° Mean Sway force, irregular waves H s
= 2.4 m
0.8 10 0.05

0.7 9 0.045

0.6 8 0.04

0.5 7 0.035

0.4 6 0.03

0.3 5 0.025

0.2 4 0.02

0.1 3
0.015

0 2
0.01

-0.1 Motion zone Transition zone Reflection zone 1


0.005 Physical Tp range

-0.2 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, v = 1 m/s, = -15° Near Field Method, v = 1 m/s, = 15°
Far Field Method, v = 3 m/s, = -15° Near Field Method, v = 3 m/s, = 15°
Near Field Method, v = 1 m/s, = -15° Far Field Method, v = 1 m/s, = 15°
Near Field Method, v = 3 m/s, = -15° Far Field Method, v = 3 m/s, = 15°
Strip Theory Method, v = 1 m/s, = -15° Strip Theory Method, v = 1 m/s, = 15°
Strip Theory Method, v = 3 m/s, = -15° Strip Theory Method, v = 3 m/s, = 15°
Seo R-function Method, v = 1 m/s, = -15° Seo R-function Method, v = 1 m/s, = 15°
Seo R-function Method, v = 3 m/s, = -15° Seo R-function Method, v = 3 m/s, = 15°

(e) Regular wave induced sway force with wave spectra (f) Irregular wave induced sway force

Figure 5.7: Sailing sway force estimation method comparison

5.2.3. Sailing vessel yaw wave drift moment


For the comparison of the yaw moment computation only the near field, far field, strip theory and Seo’s
R-function method are compared. This is done only for these methods because the short wave methods
included for the surge and sway force are not developed for predicting the yaw moment. The regular wave re-
sults are found in sub-figure 5.8a. The irregular waves results are presented in sub-figure b. The yaw velocities
are in the same range as the surge and sway velocities presented before: r ∼ v/L.

From the regular wave results in sub-figure 5.8a, one thing can clearly be distinguished. The strip theory, and
R-function method only compare close to the FF method in the motion zone and even there, their peak is
shifted to higher L/λ ratios and is still highly present in the transition zone. This holds for both velocities
but is more apparent at r = 0.04 rad/s. It must again be noted the ST results only take surge velocity u into
account. Further examining the FF and NF methods, they yield very similar results although the NF method
has a larger downward trend at high L/λ ratios. The peak location of the NF and FF methods is more trusted
since the peaks for the surge and sway forces can be found at the same ratio L/λ = 1. The oscillating effect
with respect to the frequency at higher L/λ ratios is also enhanced by the velocity. The oscillating effect is
always present but its impact is generally small. Due to the forward velocity, and in combination with the
WDD, this effect is amplified: positive peaks become more positive, negative peaks more negative. Originally
these oscillations tend to emerge from the roll motion, through the relative wave height differences caused
by the roll motions. Sub-figure 5.8b shows the irregular wave results and shows these oscillating differences
between the FF and NF method are insignificant. It is clear that the peak yaw moment can be excited in this
sea state and that the moments predicted by the strip theory and R-function method are excited more since
their peak is shifted to higher L/λ ratios. It is clear that the NF and FF methods provide very similar yaw
moments, with only small differences in (very) short waves.

Because the theoretical background of the NF yaw moment with vessel velocities is more solid than the back-
ground of the FF method. For this force component the NF methods is assumed as the base value and the
MSE errors are based on those results. The mean square errors are found in table 5.5. The MSE errors for
all three regular wave zones are large, meaning there is no real match with the NF method. For the irregular
wave MSE error it is clear the NF and FF methods predict a very similar moment.
5.3. Tow resistance assessment 69

Mean Yaw moment, irregular waves H = 2.4 m


Mean Yaw QTF, Regular waves, = -15 ° 10-3 s
0.1 10 6

0.08 9
5
0.06 8

4
0.04 7

0.02 6
3
0 5

2
-0.02 4

-0.04 3
1

-0.06 2

0
-0.08 Motion zone Transition zone Reflection zone 1 Physical Tp range

-0.1 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 -1
5 6 7 8 9 10 11 12 13 14 15

Far Field Method, r = 0.01 rad/s, = -15° Near Field Method, r = 0.01 rad/s, = 15°
Far Field Method, r = 0.04 rad/s, = -15° Near Field Method, r = 0.04 rad/s, = 15°
Near Field Method, r = 0.01 rad/s, = -15° Far Field Method, r = 0.01 rad/s, = 15°
Near Field Method, r = 0.04 rad/s, = -15° Far Field Method, r = 0.04 rad/s, = 15°
Strip Theory Method, r = 0.01 rad/s, = -15° Strip Theory Method, r = 0.01 rad/s, = 15°
Strip Theory Method, r = 0.04 rad/s, = -15° Strip Theory Method, r = 0.04 rad/s, = 15°
Seo R-function Method, r = 0.01 rad/s, = -15° Seo R-function Method, r = 0.01 rad/s, = 15°
Seo R-function Method, r = 0.04 rad/s, = -15° Seo R-function Method, r = 0.04 rad/s, = 15°

(a) Regular wave induced yaw moment with wave spectra (b) Irregular wave induced yaw moment

Figure 5.8: Sailing yaw moment estimation method comparison

Table 5.5: Sailing vessel yaw moment relative MSE with the NF method as base value

Motion zone Transition zone Reflection zone Physical T p


Regular waves
MSE MSE MSE Range MSE
Far Field method - WDD 3.89 1.42 1.11 0.019
Strip Theory method 2.04 - - 5.38
R-function method 1.38 2.39 1.00 5.31

5.3. Tow resistance assessment


The comparison of the wave drift estimation methods gives a good indication on the applicability of these
methods. Together with the verified ranges presented in section 5.1.1 these methods can be put in the context
of the tow resistance. For the tow resistance, the surge force is the most important. The sway force can have
an impact if the vessel is not towed straight behind the towing tug and should also be taken into account.
For the tow resistance, it is more important that the whole force is captured and the estimation is somewhat
conservative, than that the force is underestimated and possible unsafe conditions can occur.

For the surge force, when considering the physical T p range that can be encountered offshore, it is clear that
forward, or surge, velocity increases the wave drift force significantly. This added resistance should be taken
into account, not just taking the stationary wave drift estimation as an approximation. The difference, ob-
tained from figures 4.3 and 5.6 and considering the significant wave height, can amount to a factor 4 with
u = 3 m/s. Based on the need to be certain to capture the wave drift force with velocities includes, the follow-
ing methods are recommended for examining the surge wave drift force:

• Near field method - Wave drift damping;


• Far field method - Wave drift damping;
• Faltinsen short wave method.

The NF and FF methods both describe the wave drift force with forward velocities in detail. However, the
NF method tends to predict higher wave drift forces than the FF method, especially around the peak in the
motion zone and this is the area of interest of the tow resistance. Therefore, to be on the conservative side,
70 5. Sailing Vessel Wave Drift Assessment

the NF method is preferred. The Faltinsen SW method gives a slightly more conservative approximation of
the mean surge wave drift force but is a lot quicker to compute than the previous two methods and therefore
a viable alternative.

For the sway contribution, it is not expected that the sway velocities will be as high as the surge velocities. The
vessel is being towed forward after all. However, the sway velocity can still be present and increase the wave
drift force contribution to the tow resistance. In these cases, the surge - sway coupling should be considered
as well. For the sway force, the NF and FF methods are recommended. Because the surge-sway coupling is
not possible when using the Faltinsen SW method this method is only suitable for assessing the surge force.

5.4. Multi-criteria analysis


To determine the most suitable methods for the stability analysis for the different force components in all
the three zones, the methods are subjected to a multi-criteria analysis. The irregular wave results are not
separately taken into account as those results are highly dependant on the significant wave height that is
examined. Therefore, the results from the regular wave MCA will also be used for the irregular wave results.

The MCA scores are determined for each of the previously defined subjects and final scores can be found in
tables 5.6, 5.7, and 5.8 for respectively surge, sway and yaw. Individual multi-criteria scores can be found in
appendix D. The results are tested via a sensitivity analysis.

Table 5.6: Sailing vessel MCA results for the surge force

Score Score Score


Category
Motion Transition Reflection
FF Method - WDD 73 73 73
NF Method - WDD 47 71 71
ST Method 45 51 -
SW Method Faltinsen - 67 67
SW Method Liu - - 69
SW Method Fujii & Takahashi - - 62
SW Method Kuroda et al. - - 70
SW Method Kashiwagi et al. - - 62
R-function Method 47 59 65

Table 5.7: Sailing vessel MCA results for the sway force

Score Score Score


Category
Motion Transition Reflection
FF Method - WDD 73 73 73
NF Method - WDD 65 53 47
ST Method 45 63 -
SW Method Faltinsen - 55 55
SW Method Liu - - 63
SW Method Fujii & Takahashi - - 62
SW Method Kuroda et al. - - 70
SW Method Kashiwagi et al. - - 56
R-function Method 47 53 53

5.4.1. MCA sensitivity analysis


To be certain the results of the MCA analysis are valid, the factors and scores are subject to a sensitivity analy-
sis. Adjusting the values of the factors only resulted in increasing or decreasing the absolute value of the MCA
score. Adjusting the score values did change some of the relative differences of the MCA. In figure 5.9 these
results are presented. The scores were individually varied by +20% and −20% and with those varied scores
5.4. Multi-criteria analysis 71

Table 5.8: Sailing vessel MCA results for the yaw moment

Score Score Score


Category
Motion Transition Reflection
FF Method - WDD 49 49 49
NF Method - WDD 52 52 52
ST Method 45 45 -
R-function Method 47 47 47

the total MCA score was again computed. These varied MCA results were compared to the original MCA re-
sult. In sub-figure 5.9a the motion zone results sensitivity analysis shows that the original largest MCA score
(the blue filled dot) remains the largest score no matter the variations. The differences are smallest for the
yaw moment choice, but still the NF method obtains the highest score.

Sensitivity Analysis MCA Motion Zone Sensitivity Analysis MCA Transition Zone
80 80
Surge

Surge
60 60
40 40
MCA Score

MCA Score
80 80
Sway

Sway
60 60
40 40

80 80
Original Original
Yaw

Yaw

60 +20%
60 +20%
40 -20% 40 -20%
d

d
d

d
od

od
d

d
d

d
od

od
d

d
d

d
d

d
ho

ho
ho

ho
ho

ho
ho

ho
ho

ho
ho

ho
ho

ho
h

h
h

h
et

et
et

et
et

et
et

et
et

et
et

et
et

et
et

et
et

et
M

M
M

M
M

M
M

M
M

M
M

M
M

M
M

M
M

M
n

n
SW

SW
SW

SW
SW

SW
SW

SW
SW

SW
FF

FF
ST

ST
F

F
tio

tio
N

N
nc

nc
i

i
n

n
da

da
u

u
jii

jii
ag

ag
e

e
Li

Li
-fu

-fu
Fu

Fu
ns

ns
ro

ro
iw

iw
R

R
Ku

Ku
lti

lti
sh

sh
o

o
Fa

Fa
Ka

Ka
Se

Se
(a) Motion zone (b) Transition zone

Sensitivity Analysis MCA Reflection Zone


80
Surge

60
40
MCA Score

80
Sway

60
40

80
Original
Yaw

60 +20%
40 -20%
od
d
d

d
d

d
d

od
d

ho
ho

ho
ho

ho
ho

ho

h
h

et
et
et

et
et

et
et

et
et

M
M
M

M
M

M
M

M
M

n
W
SW

SW
SW

SW
FF

ST
F

io
N

iS

t
nc
n

da
u

jii

ag
se

Li

-fu
Fu

ro

iw
n

R
Ku
lti

sh

o
Fa

Ka

Se

(c) Reflection zone

Figure 5.9: Sensitivity Analysis MCA

In sub-figure 5.9b, the transition zone shows a similar trend. For the surge force choice, the Faltinsen SW
method benefits from the increased scores but still the FF method remains the method with the highest score.
Sub-figure c shows the less straight forward reflection zone results. For the surge force, the Liu SW method
and the Kuroda et al. SW method obtain similar scores as the FF method with the increased scores +20%. In
the other cases, the FF method remains the most suitable choice, the choice is upheld. For the sway force,
this is repeated but solely with the Kuroda et al. SW method equalling the FF method. The same argument
as with the surge force holds and the FF method is again chosen as the most suitable method. For the yaw
moment, the NF method remains the most suitable WDF estimation method throughout all variations.
72 5. Sailing Vessel Wave Drift Assessment

5.4.2. Most suitable estimation method(s) for the stability analysis


The most suitable method for the implementation of the wave drift into the stability analysis are chosen. The
reliability of the results is confirmed by a sensitivity analysis. The most suitable methods are:

• Surge force: Far field method - Wave drift damping


• Sway force: Far field method - Wave drift damping
• Yaw moment: Near field method - Wave drift damping

5.5. Chapter summary


In the sailing vessel wave drift assessment chapter, the wave drift force estimation methods with velocities
included are assessed. Surge, sway and yaw velocities are included and their impact on the wave drift force
components is taken into account. Based on the assessment, the wave drift with respect to the tow resistance
is discussed and the most suitable estimation method for including the wave drift in a towing stability analysis
is determined.

First, the computations of all ten estimation methods was verified and discussed on the basis of their the-
oretical applicability. The Faltinsen short wave method surge component was adjusted in the verification
process and only the IMO formulation was not considered further in the comparison since it did not predict
the forces well. Where necessary, the applicability range, quantified by the L/λ ratio, was modified. In this
case this was only necessary for the short wave methods. In this verification process, extra attention is paid
to the implementation of the wave drift damping and the encounter frequency effects on the wave drift force
components.

The nine remaining estimation methods are compared and assessed based on the three excitation zones: the
motion, transition and reflection zones. This is again done for all three force components, the surge and
sway force plus the yaw moment. The Faltinsen’s and Liu’s short wave method capture the linear increase
with frequency well and if combined together with the strip theory method in the R-function method, they
describe the whole QTF accurately for velocities larger than u/v ≥ 2 m/s. The remaining short wave methods
included large offsets in their predictions.

These comparisons provide the basis for looking at the tow resistance and the towing stability. The near
field method, far field method, and Faltinsen short wave method are considered most suitable for the tow
resistance assessment.

Based on a multi-criteria analysis, the most suitable methods for implementation of the wave drift forces
into the towing stability analysis are also determined. A sensitivity analysis of the parameters confirms its
validity. For the surge and sway force the far field method combined with the wave drift damping method
is considered most suitable while for the yaw moment the near field method plus the wave drift damping
method is chosen. The latter choice mostly materialises since the theoretical background of the yaw wave
drift damping and the near field method are identical. These last results are used as input for the towing
stability analysis in chapter 6.
6
Towing Stability Analysis

In this chapter the towing stability analysis is performed. The methodology for addressing this was presented
in the methodology chapter 3 and an overview is again presented in figure 6.1. For the stability analysis, the
equations of motion are necessary. The original equation of motions, developed by Bernitsas [6] and Lee [44],
and verified in the literature chapter 2, are rewritten into state variables, to be able to describe the derivatives
of the driving variables of the equation of motion. The results are presented in equations (6.1) to (6.6).

Figure 6.1: Visualisation of the stability analysis methodology

73
74 6. Towing Stability Analysis

−R 00 · u 02 + X v0 v · v 02 + X r0 r · r 02 + T 0 · cos (γ + ψ) + (m 0 + a 22
0
) · v 0 · r 0 + (X w
0
ave )
u̇ 0 = 0 (6.1)
m 0 + a 11
Y v0 · v 0 + Yr0 · r 0 + Y v0 v v · v 03 + Yr0r r · r 03 − T 0 · sin (γ + ψ) − (m 0 + a 11
0
) · u 0 · r 0 + (Y w0 ave )
v̇ 0 = 0 (6.2)
m 0 + a 22
N v0 · v 0 + Nr0 · r 0 + N v0 v v · v 03 − T 0 · x p0 · sin (γ + ψ) + (N w
0
ave )
r˙0 = 0 0 (6.3)
I zz + a 66
l˙0 = −u 0 · cos (γ + ψ) + v 0 · sin (γ + ψ) + cos (γ) + r 0 · x p0 sin (γ + ψ) (6.4)
1³ ´
γ̇ = 0 v 0 · cos (γ + ψ) + u 0 · sin (γ + ψ) − sin (γ) + r 0 · x p0 cos (γ + ψ) (6.5)
l
ψ̇ = r 0 (6.6)

These equations are used for assessing the towing stability without waves in section 6.1 and with waves in
section 6.3. The input wave drift force components for the latter assessment are presented in section 6.2.
Finally, the wave drift force parameter impact on the towing stability is determined in section 6.4.

6.1. Stability assessment without waves


With the approach verified in chapter 2, the equation of motions are first investigated without waves present,
thus including only the manoeuvring and towline forces. In the assessment both the barges, the one without
and the one with skegs, are analysed. In this analysis, a steel tow line is used and its tow line stiffness E A 0 has
an influence on the towing behaviour. Since the stiffness of these lines is generally constant the influence of
this parameter is not investigated. The parameters that are examined for the stable tow positions are:
1. Bow hawser attachment point x p0 ;
2. Forward velocity U t ug .
The parameter investigated with respect to the towing operation stability is:
3. Initial tow line length l 00 .
The equilibrium positions of the equation of motion without waves can still be determined analytically. These
equilibrium positions are described in equations (6.7) to (6.13).

0
r eq =0 (6.7)
!1
N v0 − Y v0 x p0
Ã
2
0 0
v eq = 0 or v eq =± (6.8)
Y v0 v v x p0 − N v0 v v
³ ´
0
ψeq = arcsin −v eq (6.9)
0
u eq = cos (ψeq ) (6.10)
á 0 02
+ Y v0 · v eq
¢ 0 !
Y v v v · v eq
γeq = arctan ¡ 0 02 ¢ − ψeq (6.11)
R 0 · u eq − X v0 v · v eq
02

0
R 00 · u eq
02
− X v0 v · v eq
02
Teq = (6.12)
cos (γeq + ψeq )
à 0 !
0
Teq
l eq = + 1 l 00 (6.13)
E A0

6.1.1. Equilibrium positions


The equilibrium positions vary on the basis of the distance between the bow hawser attachment point and
the center of gravity x p0 and tug forward velocity U t ug . In the case without waves, the equilibrium positions
are independent of U t ug . Without waves, there are three equilibrium positions of the barge. One straight
behind the tug, with yaw angle ψeq = 0°, and two symmetrical positions with a yaw angle ψeq 6= 0°. They are
shown in figure 6.2.
6.1. Stability assessment without waves 75

In figure 6.3, these equilibrium positions as function of x p0 are presented. Sub-figure a shows the equilibrium
positions for Barge 2a, without skegs. The lines represent the yaw angle of the towed barge. The yellow and
blue lines are the symmetrical positions on either side of the centerline straight behind the tug. When the
conventional towing point is taken: x p0 = 0.5, this means only the equilibrium position 2, straight behind the
tug is present. If x p0 = 0.3, three equilibrium positions exist. Sub-figure b shows the same data, but for the
Barge 2b with skegs. If this barge is towed with x p0 = 0.5, three equilibrium conditions exist. One straight
behind the towing tug, and two on either side with a yaw angle of ∼ 75°. These are only the equilibrium
positions, and their stability is determined in the next section.

Figure 6.2: Overview of the equilibrium positions without waves

Yaw angle at equilibrium points Yaw angle at equilibrium points

80 Equilibrium position 1 80 Equilibrium position 1


Equilibrium position 2 Equilibrium position 2
60 Equilibrium position 3 60 Equilibrium position 3

40 40

20 20

0 0

-20 -20

-40 -40

-60 -60

-80 -80

0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6

(a) Barge 2a: w/o skegs (b) Barge 2b: w/ skegs

Figure 6.3: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, without waves

6.1.2. Towing stability


The state variables are linearised around the equilibrium positions by deriving the Jacobian of the system at
those equilibrium positions. The stability of the equilibrium position is then determined from the eigenval-
ues of the Jacobian. As mentioned before, the conventional x p0 = 0.5 is adopted. The variable investigated
while examining the stability of these equilibrium positions is the initial tow line length l 00 . In sub-figure
6.4a the maximum value of the six eigenvalues of the Jacobian of Barge 2a is plotted. As can be seen, all
three positions produce the same eigenvalues, because in the case examined, the three positions match. At
l 00 =∼ 0.15, the eigenvalues turn positive. Below this threshold the towing operation is stable. This can be
explained by the very short tow line, so that the feedback of the tug to the towed vessel is very high and thus
the towed vessel can be towed in a stable manner. This corresponds with the short tow line findings of Gokce
[29]. Once the eigenvalues turn positive, the tow operation becomes unstable. The magnitude of the positive
eigenvalue is irrelevant, merely the sign determines the stability. As can be seen, the vessel remains unstable
76 6. Towing Stability Analysis

with increasing initial tow line length. This corresponds with the previously determined directional stability:
unstable.

Sub-figure 6.4b shows the maximum real part of the eigenvalues of Barge 2b. In this case, positions 1 and 3
produce the same values while position 2 shows different ones. Position 1 and 3 have positive eigenvalues
over the entire range of l 00 and are thus unstable positions. Position 2 has negative eigenvalues over the entire
range of l 00 and is thus stable in all cases. Again, this matches with the previously found directional stability
of Barge 2b: stable.

0.5 1.2

0.4 1
Max real part of eigenvalues

Max real part of eigenvalues


0.3 0.8

0.2 0.6
Position 1 Position 1
Position 2 Position 2
Position 3 Position 3
0.1 0.4
Unstable
0 0.2
Stable
Unstable
-0.1 0
Stable

-0.2 -0.2
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5

(a) Barge 2a: w/o skegs (b) Barge 2b: w/ skegs

Figure 6.4: Maximum real eigenvalues l 00 dependency with U t ug = 2 m/s, x p0 = 0.5, without waves

6.1.3. Time domain simulation


Time domain (TD) simulation is conventionally used to investigate the global behaviour of a towing oper-
ation. The stability around the equilibrium positions provides information on the behaviour around these
positions but not on the global behaviour. This is simulated using a time domain approach, releasing the
towed vessel with an initial disturbance, as initial conditions (IC), and examining the global reaction. In this
case this is done for a tow line length of l 00 = 2.5 and the previously determined value of x p0 = 0.5. The vessel
is released with a sway offset y 00 = 0.3. It is thus expected from the findings in figure 6.4, that Barge 2a shows
unstable behaviour, and Barge 2b stable behaviour.

In sub-figure 6.5a, the time domain simulation of Barge 2a is presented. The blue line and axis present the
normalised sway offset, while the black line and axis display the yaw angle of the towed vessel. The figure
shows the towed vessel behaves dynamically stable. This is because one of the assumptions in the simula-
tion is that the towing tug is considered as a fixed moving point, which cannot move in the sway direction.
Therefore an unstable vessel will always be dynamically stable. In practice, this assumption corresponds with
a towing tug with a high capability of course keeping and enough available towing force to not be affected by
the (unstable) behaviour of the towed vessel. Again, the dynamic stability is confirmed by the phase space
diagrams in figure 6.6, showing the limit cycles. The vessel oscillates with a path width of 2.1 · L and yaw
angles ranging between ψ = −50° and ψ = +50°. The phase difference between the yaw angle ψ and the sway
offset y 0 is as expected, obtaining a yaw angle with opposite sign just before turning. It can be compared to a
drifting manoeuvre in car racing.

Sub-figure 6.5b presents the results of the TD simulation of Barge 2b. The same initial conditions as in the
previous simulation are applied, but it is clear the barge converges to its stable towing position 2, straight
behind the towing tug. This is also apparent from the phase space diagrams in figure 6.7. Both the sway,
sub-figure b, and the yaw, sub-figure c, graphs show convergence. The high frequency oscillation present in
the surge phase space diagrams, sub-figures a, is the result of the time step of the integration.
6.1. Stability assessment without waves 77

TD Simulation, without waves TD Simulation, without waves


1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: w/o skegs (b) Barge 2b: w/ skegs

Figure 6.5: Time Domain simulation of the towing operation, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5

TD Simulation without waves: Sway Phase Space TD Simulation without waves: Yaw Phase Space
TD Simulation without waves: Surge Phase Space 0.8 40
0.08

0.6 30
0.06

0.4 20
0.04

0.2 10
/dt' [-]

0.02
dy'/dt' [-]
dx'/dt' [-]

0 0 0
d

-0.02 -0.2 -10

-0.04 -0.4 -20

-0.06 -0.6 -30

-0.08 -0.8 -40


-3 -2.95 -2.9 -2.85 -2.8 -2.75 -2.7 -1.5 -1 -0.5 0 0.5 1 1.5 -60 -40 -20 0 20 40 60
Surge offset x' [-] Sway offset y' [-] Yaw angle [°]

(a) Barge 2a: Phase Space surge (b) Barge 2a: Phase Space sway (c) Barge 2a: Phase Space yaw

Figure 6.6: Barge 2a Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5

TD Simulation without waves: Yaw Phase Space


TD Simulation without waves: Sway Phase Space 2
TD Simulation without waves: Surge Phase Space 0.04
0.015
1.5

0.02
0.01 1

0.5
0
0.005
/dt' [-]
dy'/dt' [-]
dx'/dt' [-]

0 -0.02
-0.5
d

-0.04 -1
-0.005

-1.5
-0.01 -0.06
-2

-0.015 -0.08 -2.5


-3.005 -3 -2.995 -2.99 -2.985 -2.98 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 -5 -4 -3 -2 -1 0 1 2 3
Surge offset x' [-] Sway offset y' [-] Yaw angle [°]

(a) Barge 2b: Phase Space surge (b) Barge 2b: Phase Space sway (c) Barge 2b: Phase Space yaw

Figure 6.7: Barge 2b Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5 and l 00 = 2.5
78 6. Towing Stability Analysis

6.2. Wave drift force components Barge 2


In this section, the wave drift force for the Barge 2 is presented. These force components are computed based
on the wave drift estimation methods determined in chapter 5: FF method combined with WDD for the surge
and sway force, the NF method combined with the WDD for the yaw
¢ moment. As already mentioned, the
wave drift force is dependent on six variables, F w ave ω, ζa , u, v, r, ψ . The force components are normalised
¡

for the wave amplitude ζ2a or significant wave height H s2 and thus remain dependent on five variables. The
range of these variables was defined in the methodology based on the verification study of the equation of
motion. Based on the analysis in without waves in section 6.1, these boundaries are updated and presented
again in table 6.1. The upper boundary of the wave frequency is reduced to 1.6 rad/s, since if the encounter
frequency effect is taken into account the previous limit scales down to this value. For comparative purposes,
the wave drift force is still presented up to ω = 2.1 rad/s. The surge velocity is reduced to 2 m/s, since in the
towing stability analysis, higher velocities will result in not converging simulations. Yaw velocity r 0 and the
yaw angle ψ were adjusted to match the new simulation limits.

For the towing stability analysis, a different normalisation factor is used for the forces as this is the factor
already applied in the source [37]. The normalised force is presented in equations (6.14) to (6.16). Here is T
the draft of the vessel.

0 X w ave (ω, ζa , u, v, r, ψ)
Xw ave = 1
(6.14)
2 ρT LU t ug
2

Y w ave (ω, ζa , u, v, r, ψ)
Y w0 ave = 1
(6.15)
2 ρT LU t ug
2

0 N w ave (ω, ζa , u, v, r, ψ)
Nw ave = 1
(6.16)
2 ρT L U t ug
2 2

The wave drift forces are saved in a database categorised by file name based on the three velocity components.
For example: the file WDFYawRegu12v_01r0.mat produces the yaw wave drift moment as a matrix of ω and
ψ values with velocities u = 1.2m/s, v = −0.1m/s and r = 0m/s. More information on the implementation in
Matlab can be found in appendix C.

Wave drift damping


In figure 6.8, the wave drift damping coefficients for Barge 2 are presented. When comparing them with the
wave drift damping coefficients computed in chapter 5, see figure 5.4, on first glance all coefficient curves are
similar. However, most wave drift damping coefficients for Barge 2 have larger amplitudes than the coeffi-
cients for Barge 1. This is most clear for the diagonal terms, B 22 and B 66 . On the other hand, the peak value
of B 11 is roughly 2.5 times smaller and shifted by about 0.2 rad/s to higher frequencies. Also an oscillating ef-
fect is observed at higher frequencies, which is explainable by panel size related discrepancies in the Delfrac
computation. This effect is deemed inevitable with the geometry of Barge 2. More information on the mesh
convergence is given in appendix B. B 22 has a slightly higher magnitude than its counterpart of Barge 1 and
again is also shifted by about 0.2 rad/s to higher frequencies. The peak values of coefficient B 66 are a factor 2
larger than B 66 of Barge 1, but the general curve is the same. Again the curves match those of coupling terms
B 62 and B 26 , which follows logically from the slender body approximation.

Table 6.1: Updated lower and upper boundaries for the stability analysis of the wave drift force variables

Variable Lower Boundary Step Upper Boundary Unit


Wave frequency ω 0.4 0.05 1.6 [ r ad
s ]
Surge velocity u 0.0 0.1 2.0 [ ms ]
Sway velocity v -1.5 0.1 1.5 [ ms ]
Yaw velocity r -0.025 0.005 0.025 [ r ad
s ]
Yaw angle ψ -50 5 50 [°]
6.2. Wave drift force components Barge 2 79

Wave Drift Damping B 11 Coefficient


-50° -40° Wave Drift Damping B 12 Coefficient Wave Drift Damping B 16 Coefficient
10
-3 -50° -40° 10
-4 -50° -40°
0 -30° 1 1.5
-30° -30°
-20°
-20° -20°
0.8
-10°
-1 [°] 0.6
-10°
[°] 1 -10°
[°]

0° 0°
0.4
-2 0.5
10° 0.2 10° 10°

-3 0 0
20° 20° 20°
-0.2

-4 30° -0.5
-0.4 30° 30°

-0.6
-5 40° -1
0.05 40° 40°
-0.8
0.05 0.05
0.56 0.56 0.56
1.1 1.1 1.1
[rad/s] 1.6 1.6 1.6
10 -4 2.1 50° -1 [rad/s]
2.1 50°
-1.5 [rad/s]
2.1 50°

(a) B 11 : F x - u Coefficient (b) B 12 : F x - v Coefficient (c) B 16 : F x - r Coefficient

Wave Drift Damping B 22 Coefficient Wave Drift Damping B 26 Coefficient


Wave Drift Damping B 21 Coefficient 10 -4
-50° -40° -50°
10
-3 -50° -40° 10 -5 -40°
1
-30° 12 -30°
-30°
9 -20°
-20° -20°
0.8 10
-10°
0.6
-10°
[°]
8
[°] 8
-10°
[°]
7 0° 0°

0.4
6
6
0.2 10°
10° 10°
4
5
0
20° 20° 2 20°
4
-0.2
0
30° 3 30° 30°
-0.4

2 -2
-0.6
40° 40° 40°
0.05 1 0.05 -4 0.05
-0.8 0.56 0.56
0.56
1.1 1.1 1.1
-1 [rad/s] 1.6 0 [rad/s] 1.6 -6 [rad/s] 1.6
2.1 50° 2.1 50° 2.1 50°

(d) B 21 : F y - u Coefficient (e) B 22 : F y - v Coefficient (f) B 26 : F y - r Coefficient

Wave Drift Damping B 61 Coefficient Wave Drift Damping B 62 Coefficient Wave Drift Damping B 66 Coefficient
-4 -50° 10 -5
-50° -40° 10 -5
-50° -40°
10 -40° 12
1.5 -30° -30° 12 -30°

-20° -20° -20°


10 10
-10° -10°
1 -10°
[°] 8
[°] 8
[°]
0° 0° 0°
0.5 6 6

10° 10° 10°


4 4

0
20° 2 20° 2 20°

0
-0.5 0
30° 30° 30°
-2
-2
-1 40° 40°
40° -4
0.05 -4 0.05 0.05
0.56 0.56 0.56
1.1 1.1 -6 1.1
-1.5 [rad/s] 1.6 -6 [rad/s] 1.6 [rad/s] 1.6
2.1 50° 2.1 50° 2.1 50°

(g) B 61 : N z - u Coefficient (h) B 62 : N z - v Coefficient (i) B 66 : N z - r Coefficient

Figure 6.8: The frequency ω and incoming angle of attack β dependency of the Barge 2 Wave Drift Damping (WDD) coefficients for
Barge 2
80 6. Towing Stability Analysis

Surge wave drift force


In figure 6.9, the surge wave drift force for Barge 2 is presented. In sub-figure a, the QTF for incoming angle
of attack of β = 0° is plotted. The effect of the oscillating WDD coefficient B 11 is clearly visible for higher
frequencies. This amplifies the already present oscillations, see the u = 0 curve, which are, at least partly,
present due to chosen panel size. The peaks shift to lower frequencies with increasing velocity due to the
encounter frequency effect. In sub-figure b, the irregular wave results show the presence of the peak mean
surge force in the physical T p range, and again the shift due to the encounter frequency. Sub-figures c and
d, show the same wave drift force but with a wave angle of attack β = ±30°. The main observation is that
this increased angle reduces the effect of the oscillation at higher frequencies and that both the regular and
irregular wave results do not decrease in magnitude with increasing β.

Mean Surge QTF, Regular waves, = 15 ° Mean Surge force, irregular waves H s
= 2.4 m, = 15 °
0 0
u = 0 m/s
-0.05 u = 1 m/s
-0.005
u = 2 m/s
-0.1
-0.01
-0.15

-0.2 -0.015

-0.25 -0.02

-0.3
-0.025
-0.35
-0.03
-0.4
u = 0 m/s
-0.035 Physical Tp u = 1 m/s
-0.45
range u = 2 m/s
-0.5 -0.04
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(a) Regular wave Surge WDF, β = 0° (b) Irregular waves Surge WDF, β = 0°

Mean Surge QTF, Regular waves, = 30 ° Mean Surge force, irregular waves H s
= 2.4 m, = 30 °
0 0
u = 0 m/s
-0.05 u = 1 m/s
-0.005
u = 2 m/s

-0.1 -0.01

-0.15 -0.015

-0.2 -0.02

-0.25 -0.025

-0.3 -0.03

u = 0 m/s
-0.35 -0.035 Physical Tp u = 1 m/s
range u = 2 m/s
-0.4 -0.04
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(c) Regular wave Surge WDF, β = ±30° (d) Irregular waves Surge WDF, β = ±30°

Figure 6.9: Barge 2 surge wave drift force X w ave for u = 0, 1 and 2m/s

Sway wave drift force


In figure 6.10, the sway wave drift force for Barge 2 is presented. In sub-figure a, the forces for incoming angle
β = ±15° are presented. The same results for the irregular waves are presented in sub-figure b. The same
graphs, but for β = ±30°, are plotted in sub-figures c and d. The symmetrical nature of the sway force seems
obvious but there is more to it. This is best visible in sub-figure c. The purple and black lines are the v = 0m/s
lines and are symmetrical. However, the dark blue and light blue lines are not. These two force lines are
reduced because the velocity is pointed in the opposite direction of the incoming waves, the vessel is sailing
away from the waves. Ergo, if the incoming angle and velocity are opposite signs, the force reduces, and if
they are the same sign, the force increases. This is confirmed by the red and yellow lines in the graph. The
6.2. Wave drift force components Barge 2 81

effect of the encounter frequency is also visible: with an increase of force the peak shifts to lower frequencies.
On the other hand, with a decrease in force the peak shift to higher frequencies. Sub-figure d shows that in
a spectrum with the shortest physical T p the maximum sway force can be expected and that the sway force
increases threefold with a β = ±15° increase.

Mean Sway QTF, Regular waves Mean Sway force, irregular waves H s
= 2.4 m
0.4 0.04
°
v = -1 m/s, = 15
0.3 v = 0 m/s, = 15 ° 0.03
v = 1 m/s, = 15 °
0.2 v = -1 m/s, = -15 ° 0.02
v = 0 m/s, = -15 °
0.1 v = 1 m/s, = -15 ° 0.01

0 0

-0.1 -0.01 v = -1 m/s, = 15 °


v = 0 m/s, = 15 °
-0.2 -0.02 v = 1 m/s, = 15 °
v = -1 m/s, = -15 °
-0.3 -0.03 Physical Tp v = 0 m/s, = -15 °
range v = 1 m/s, = -15 °
-0.4 -0.04
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(a) Regular wave Sway WDF, β = ±15° (b) Irregular waves Sway WDF, β = ±15°

Mean Sway QTF, Regular waves Mean Sway force, irregular waves H s
= 2.4 m
0.4 0.04
°
v = -1 m/s, = 30
0.3 v = 0 m/s, = 30 ° 0.03
v = 1 m/s, = 30 °
0.2 v = -1 m/s, = -30
°
0.02
v = 0 m/s, = -30 °
0.1 °
v = 1 m/s, = -30 0.01

0 0

-0.1 -0.01 v = -1 m/s, = 30 °


v = 0 m/s, = 30 °
-0.2 -0.02 v = 1 m/s, = 30 °
v = -1 m/s, = -30 °
-0.3 -0.03 Physical Tp v = 0 m/s, = -30 °
range v = 1 m/s, = -30 °
-0.4 -0.04
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(c) Regular wave Sway WDF, β = ±30° (d) Irregular waves Sway WDF, β = ±30°

Figure 6.10: Barge 2 sway wave drift force Y w ave for v = −1, 0 and 1m/s.

Yaw wave drift moment


In figure 6.11, the yaw wave drift moment for Barge 2 is presented. As before, the β = ±15° results are pre-
sented in sub-figures a and b. In sub-figure c, the regular wave results for β = ±30° are presented. The same
phenomenon as with the sway force in figure 6.10 is observed. Opposite signs of the yaw velocity and the
wave angle of attack decrease the yaw moment, same signs increase the magnitude of the moment. For the
stationary vessel, the peak of the moment occurs at a frequency roughly 0.2 rad/s lower than for the sway and
surge forces. With increasing velocities, these peaks tend to move closer together. At higher frequencies the
moment flips its sign with respect to its peak value. Again, part of this effect occurs due to the panel size, the
other part emerges from the phase difference in the waves. In sub-figure d, the irregular wave result shows
that these high frequency effects are not present in the physical T p range and that the peak value of the mo-
ment can occur in the range. Again, as with the sway force, an increase of β = ±15° will results in a moment
magnitude of three times higher.
82 6. Towing Stability Analysis

Mean Yaw Moment, irregular waves H = 2.4 m


10-3 s
Mean Yaw QTF, Regular waves 4
0.05
r = -0.015 rad/s, = -15 °
0.04 3
r = 0 rad/s, = -15 °
°
0.03 r = 0.015 rad/s, = -15
2
0.02
1
0.01

0 0

-0.01 -1 r = -0.015 rad/s, = 15


°

r = 0 rad/s, = 15 °
-0.02
-2 r = 0.015 rad/s, = 15 °
-0.03 r = -0.015 rad/s, = 15 ° r = -0.015 rad/s, = -15 °
r = 0 rad/s, = 15
° -3 Physical Tp r = 0 rad/s, = -15 °
-0.04 °
r = 0.015 rad/s, = 15 ° range r = 0.015 rad/s, = -15

-0.05 -4
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(a) Regular wave Yaw WDM, β = ±15° (b) Irregular waves Yaw WDM, β = ±15°

Mean Yaw Moment, irregular waves H = 2.4 m


10-3 s
Mean Yaw QTF, Regular waves 4
0.05
r = -0.015 rad/s, = -30 °
0.04 3
r = 0 rad/s, = -30 °
0.03 r = 0.015 rad/s, = -30 °
2
0.02
1
0.01

0 0

-0.01 -1 r = -0.015 rad/s, = 30 °


r = 0 rad/s, = 30 °
-0.02
-2 r = 0.015 rad/s, = 30 °
-0.03 r = -0.015 rad/s, = 30 ° r = -0.015 rad/s, = -30 °
r = 0 rad/s, = 30 ° -3 Physical Tp r = 0 rad/s, = -30 °
-0.04
r = 0.015 rad/s, = 30 ° range r = 0.015 rad/s, = -30 °

-0.05 -4
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 4 6 8 10 12 14

(c) Regular wave Yaw WDM, β = ±30° (d) Irregular waves Yaw WDM, β = ±30°

Figure 6.11: Barge 2 yaw wave drift moment N w ave for r = −0.015, 0 and 0.015r ad /s.

6.3. Stability assessment with waves


For addressing the impact of the wave drift on the towing stability, the three towing operation parameters
presented in section 6.1 are analysed again. Besides these parameters, two variables of the wave drift force
are also taken into account for the equilibrium positions and the towing stability:

4. Wave frequency ω or peak period T p ;


5. Wave amplitude ζa or significant wave height H s .

Note that in this case, the sailing conditions significant wave height of H s = 2.4 m is adopted. Since the wave
drift force is included numerically in the equations of motions, the equilibrium positions will need to be
determined numerically. Since the equilibrium positions are sought, the statements in equations (6.17) to
(6.19) hold. On the basis of these and equations (6.1) to (6.3), the function in equation (6.20) is obtained. This
function, only dependent on the variable ψeq , can be set to zero and evaluated to obtain the values of ψeq .
Note that in this case, incoming angle of attack β = −ψ.

Numerically, the zero crossings of this function are calculated. In equations (6.21) to (6.23) the description of
the remaining equilibrium parameters are presented.
6.3. Stability assessment with waves 83

0
r eq =0 (6.17)
0
u eq = cos (ψeq ) (6.18)
0
v eq = − sin (ψeq ) (6.19)

f (ψeq ) = (Y v0 v v · x p0 − N v0 v v ) · sin3 (ψeq ) + (Y v0 x p0 − N v0 ) · sin (ψeq ) + N w


0
− Y w0 · x p0 (6.20)

0 0
ave ω, ζa , cos (ψeq ), − sin (ψeq ), 0, ψeq
¡ ¢
With: X w = Xw
Y w0 = Y w0 ave ω, ζa , cos (ψeq ), − sin (ψeq ), 0, ψeq
¡ ¢

0 0
ave ω, ζa , cos (ψeq ), − sin (ψeq ), 0, ψeq
¡ ¢
Nw = Nw

³ Y v0 v v · v eq
03
+ Y v0 · v eq
0
+ Y w0 ´
¡ ¢
γeq = arctan ¡ 0 02 ¢ − ψeq (6.21)
R 0 · u eq − X v0 v · v eq
02
− Xw0

0
R 00 · u eq
02
− X v0 v · v eq
02 0
− Xw
Teq = (6.22)
cos (γeq + ψeq )
0
à !
0
Teq
l eq = + 1 l 00 (6.23)
E A0

6.3.1. Equilibrium positions regular waves


From the equilibrium position analysis, it is found that five equilibrium positions are mathematically possi-
ble. An overview of these is presented in figure 6.12. In comparison with the equilibrium positions without
waves, see figure 6.2, two extra positions emerge. The previously found positions without waves match with
positions 2, 3 and 4. Equilibrium positions 1 and 5 exist in some very specific cases, around the frequencies
matching with the high frequency peaks of surge wave drift force. These peaks partly origin from discrepan-
cies in the panel mesh and thereby introduce uncertainty on the physical existence of these two positions.
Therefore, these two will not be further investigated and the focus remains on positions 2, 3 and 4.

Figure 6.12: Overview of the equilibrium positions with waves


84 6. Towing Stability Analysis

Yaw angle at critical (equilibrium) position 2 Yaw angle at critical (equilibrium) position 4
1 0 1 50

-5 45

-10 40
0.75 0.75
35

[°]
-15

[°]

Equilibrium yaw angle


Equilibrium yaw angle
-20 30

0.5 -25 0.5 25

-30 20

-35 15
0.25 0.25
-40 10

-45 5

0 -50 0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

(a) Barge 2a: w/o skegs, position 2 (b) Barge 2a: w/o skegs, position 4

Figure 6.13: Barge 2a: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with waves

Yaw angle at critical (equilibrium) position 4 Yaw angle at critical (equilibrium) position 4
2.5 50 2 50

45 45

2 40 40
1.5
35 35
[°]

[°]
Equilibrium yaw angle

Equilibrium yaw angle


30 30
1.5

25 1 25

20 20
1
15 15

0.5
10 10
0.5
5 5

0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

(a) Wave amplitude ζa dependency (b) Tug forward velocity U t ug dependency

Figure 6.14: Barge 2a: Parameter ζa and U t ug dependence for equilibrium position 4

The equilibrium positions vary on the basis of the distance between the bow hawser attachment point and
the CoG x p0 , wave frequency ω, wave amplitude ζa and tug forward velocity U t ug . These are investigated
individually, with base values of x p0 = 0.5, the same as the without waves case, ζa = 1.2 m and U t ug = 2 m/s.
This base case is presented as the red line in the upcoming figures. Initially, the x p0 dependence of the vessel
is investigated.

The results are presented in figure 6.13. First, sub-figure a and b present the yaw angle ψeq for respectively
positions 2 and 4. Position 3 remains straight behind the towing tug and keeps a yaw angle of ψeq = 0° at
all time, see figure 6.12. Positions 2 and 4 are theoretically symmetrical and this is confirmed. For practical
purposes, position 4 is considered for the remaining parameter dependency investigations, but the results
also hold for position 2.

Sub-figure 6.13b presents the equilibrium yaw angle ψeq dependent on x p0 and ω. The dark red colour rep-
resents a value of out of range. Because the wave drift force is only considered up to β = ±50°, equilibrium
positions can also only be computed up to that angle. The without waves equilibrium position, position 3 in
sub-figure 6.3a, can be seen as a sectional graph at ω = 0.4 rad/s. This implies that the waves do not influ-
ence the equilibrium positions up to ω =∼ 0.60 rad/s. The red line, at x p0 = 0.5, shows that between 0.75 rad/s
< ω ≤ 0.95 rad/s equilibrium position 4 with ψeq 6= 0 exist. At ω = 1.15, 1.35 and 1.50 rad/s these positions also
exist due to the outliers at those frequencies. These outliers clearly show the oscillating behaviour previously
6.3. Stability assessment with waves 85

noticed in the surge wave drift force in sub-figure 6.9a. This shows the surge force has a large influence on
the creation of these equilibrium positions. At ω = 0.70 rad/s a small decrease in the profile is caused by the
negative moment contribution while the reduction of the profile between ω = 0.85 rad/s and ω = 1.05 rad/s
can be attributed to the sway force contribution. This is exactly the range where the sway force contribution
increases up to its semi-constant value at higher frequencies. The drop is enhanced by the yaw moment as
at those frequencies the magnitude of the negative moment also drops. The increase of the profile at higher
frequencies is attributed to the surge force and yaw moment, as both increase positively with the frequency.
Ergo, the wave drift surge contribution allows for more equilibrium positions to exist, as the extra resistance
increases the tow line tension and thus the tow line moment. The sway contribution reduces the existence
of equilibrium positions ψeq 6= 0°. This makes sense as the sway force will increase rapidly with larger yaw
angles, disturbing the balance of the equilibrium position. A negative moment will work against the existence
of equilibrium position 4 while a positive moment will contribute.

The parameters ζa and U t ug are investigated to gain more insight in their influence. Sub-figure 6.14a shows
the variation of ζa . With larger wave amplitudes, the equilibrium yaw angles increase around peaks stated
before. This is expected since the wave drift contribution increases relatively to ζ2a and thus a higher wave
amplitude leads to more possibilities of position 4 existing with ψeq 6= 0°. If ζa < 0.8 m, no other equilibrium
positions exist than the one equivalent to position 3.

The tug forward velocity dependency is presented in sub-figure 6.14b. The trend in this figure is clear: if po-
sition 4 with ψeq 6= 0 exists, then a lower tug forward velocity will result in a larger ψeq . This can be explained
by looking at the force component ratios. The manoeuvring force components acting on the vessel are the
destabilising force while the wave drift force components act as the stabilising force. Especially the yaw mo-
ment is of importance in this. With lower forward velocity, the wave drift moment is relatively larger than the
manoeuvring moment, thus resulting in larger ψeq angles. At higher velocities, the manoeuvring moment
plays a bigger role, reducing the ψeq . For example, for respectively U t ug = 0.75 and 2 m/s at ω = 0.85 rad/s
the wave drift - manoeuvring force ratio’s are 2.85 and 0.25. A clear difference in ratio. Back to the figure,
the influence of the sway force can be seen again between ω = 0.90 and 1.00 rad/s, reducing the ψeq below
U t ug = 0.75 m/s. The pinnacles at high frequencies clearly show the encounter frequency effect, as they shift
to lower frequency with larger velocity.

In figure 6.15 the equivalent of figure 6.13 but for Barge 2b, with skegs, is plotted. Again, it is clear that posi-
tions 2 and 4 are symmetrical and for practical purposes, position 4 is further considered. Sub-figure b shows
many out of range values. This is as expected, as the sectional slice at ω = 0.4 rad/s in sub-figure 6.3b shows
that these equilibrium positions quickly move to high ψeq angles with increasing x p indicating that the ma-
noeuvring moment is dominating the moment balance. Taking x p0 = 0.5, it is seen that the yaw angle only has
values above 50°. As was shown from the analysis without waves, equilibrium positions 2 and 4 are unstable.
As the yaw angles are above 50° they are not easily encountered. The without waves time domain simulation
showed a maximum yaw angle of ψ = 50°. For further analysis, positions 2 and 4 of Barge 2b are therefore not
taken into account. The anomalies at higher frequencies are again attributed to the oscillating effect present
in the surge wave drift force in figure 6.9.

Sub-figure 6.16a shows the variation of the wave amplitude ζa . It shows that only with a wave amplitude
ζa > 1.6 m the wave drift moment is large enough to reduce the equilibrium yaw angle below 50°. The tug
forward velocity dependency is presented in sub-figure 6.16b. This graph shows that there is a substantial area
with low forward velocities where no equilibrium position other than position 3 exist (the blue area around
ω = 0.80 rad/s). With higher velocities though, the angles quickly reach ψeq = 50°. Similar to the results for
Barge 2a, this follows from the lower destabilising effect of the manoeuvring forces at lower velocity.

6.3.2. Towing stability regular waves


Now that the equilibrium positions of the tow operation are obtained, the state variables are linearised by
deriving the Jacobian of the system at these equilibrium positions. The eigenvalues of this Jacobian are used
to assess the stability of the operation. The stability is assessed for variables initial towline length l 0 , wave
frequency ωw and wave amplitude ζa . Since the amplitude of an eigenvalue does not have an influence on
the nature of the equilibrium position, the boundary at zero is of interest. Positive eigenvalues indicate an
unstable equilibrium position, negative eigenvalues a stable. In the following figures shaded area’s indicate
unstable conditions and blank area’s stable ones.
86 6. Towing Stability Analysis

Yaw angle at critical (equilibrium) position 2 Yaw angle at critical (equilibrium) position 4
1 0 1 50

-5 45

-10 40
0.75 0.75
35

[°]
-15

[°]

Equilibrium yaw angle


Equilibrium yaw angle
-20 30

0.5 -25 0.5 25

-30 20

-35 15
0.25 0.25
-40 10

-45 5

0 -50 0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

(a) Barge 2b: w/ skegs, position 2 (b) Barge 2b: w/ skegs, position 4

Figure 6.15: Barge 2b: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with waves

Yaw angle at critical (equilibrium) position 4 Yaw angle at critical (equilibrium) position 4
2.5 50 2 50

45 45

2 40 40
1.5
35 35
[°]

[°]
Equilibrium yaw angle

Equilibrium yaw angle


30 30
1.5

25 1 25

20 20
1
15 15

0.5
10 10
0.5
5 5

0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

(a) Wave amplitude ζa dependency (b) Tug forward velocity U t ug dependency

Figure 6.16: Barge 2b: Parameter ζa and U t ug dependence for equilibrium position 4

First Barge 2a, without skegs, is examined. For this barge, equilibrium positions 2, 3 and 4 are investigated for
stability. This is done for ζa = 1.25 m and 2.5 m. Position 2 is plotted in sub-figures a and b, position 3 in c
and d and position 4 in e and f. Note that, from the results for position 2 and 4 in figure 6.13, the equilibrium
positions exist at 0.75 < ω < 0.95, 1.15, 1.35, and 1.50 rad/s. Away from these frequencies, the positions do not
exist and are shown as unstable positions.

Sub-figure 6.17a shows the result for ζa = 1.25 m of position 2, showing two small stable areas at ω = 1.15
rad/s and ω = 1.50 rad/s. These areas are small and show the local stability but a time domain analysis should
indicate if these stable areas can lead to global stability. Sub-figure e shows the same for position 4, with
three parallel stability contours present at 0.95 rad/s, 1.15 rad/s and 1.50 rad/s. In theory this position is
symmetrical with position 2 and thus should yield the same stability contour. There is definitely a match,
but it can be concluded that some (numerical) errors occur. From the nature of the stability of equilibrium
positions it is known that if the stability is assessed at a position close to, but not exactly on the equilibrium
position, this position is by definition unstable. Since they are analysed numerically, this is not exactly the
case but the difference between these two graphs can be explained by linearisation at a slightly different
position. The stability contour in sub-figure e should thus also be applicable to position 2 in sub-figure a.
The reasoning behind this is also valid for sub-figures b and f where the stability contours show small stable
areas at the frequencies where positions 2 and 4 exist.
6.3. Stability assessment with waves 87

Stability contours position 2, a


= 1.25 m Stability contours position 2, a
= 2.5 m
1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Positive eigenvalues contour for position 2, ζa = 1.25 m (b) Positive eigenvalues contour for position 2, ζa = 2.5 m

Stability contours position 3, a


= 1.25 m, eq
= 0° Stability contours position 3, a
= 2.5 m, eq
= 0°
1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(c) Positive eigenvalues contour for position 3, ζa = 1.25 m (d) Positive eigenvalues contour for position 3, ζa = 2.5 m

Stability contours position 4, a


= 1.25 m Stability contours position 4, a
= 2.5 m
1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(e) Positive eigenvalues contour for position 4, ζa = 1.25 m (f) Positive eigenvalues contour for position 4, ζa = 2.5 m

Figure 6.17: Barge 2a: Positive eigenvalues contour l 00 and ω dependency with U t ug = 2 m/s, x p0 = 0.5, with waves
88 6. Towing Stability Analysis

The same deviations of the stability contours have been observed for position 3. The numerical determina-
tion of the equilibrium yaw angle introduces this small deviation from ψeq = 0° due to the numerical rounding
present in the routine. This numerical rounding and the uncertainties these introduce are discussed in the
discussion chapter 7.

Sub-figure 6.17c shows the results for ζa = 1.25 m of position 3 with ψeq = 0°. It shows a large stable area at
frequencies larger than ∼ 0.75 rad/s but with five unstable contours at frequencies ω = 1.05, 1.20, 1.40, 1.50
and 1.60 rad/s. The large unstable area at low frequencies is explained by the fact that the contribution sway
force only starts increasing around that frequency. The five unstable contours occur at the frequencies found
at the peaks in the surge wave drift force, but this cannot be designated as the conclusive source of them. The
local stability that this graph indicates will need to be tested by the time domain simulation to see if it can lead
to global stability. With ζa = 2.5 m in sub-figure d, a similar graph is found. Here only unstable contours for
ω < 0.70 rad/s and at ω = 1.45 and 1.60 rad/s remain. The unstable part at low frequencies is reduced since the
wave amplitude has increased and thus the wave drift forces. Once the wave drift forces start contributing,
around 0.65 rad/s, they stabilise the tow operation sooner than with lower wave amplitude.

As already mentioned, the stability of position 3 may change when the equation of motion is linearised at a
different position. To see the impact of this deviation, the result for ψeq = −0.5° is presented in figure 6.18.
Comparing them with the ψeq = 0° results in sub-figures 6.17c and d, they are very similar. The difference
lies in jagged unstable area’s with short tow lines and a small unstable contour at ω = 0.80 rad/s. This in-
dicates that the linearisation of the equilibrium position 3 for regular waves returns the same results even
with a ψeq = −0.5° deviation. For global analysis, these differences can be explained as both positions exist
close to each other and have stable combinations. The vessel remains only unstable if both positions have
positive eigenvalues. When both have negative eigenvalues, and are stable, one of them should be more sig-
nificant. During a time domain simulation, the vessel will likely converge towards one of the two. This is
further discussed at the time domain simulations.

Stability contours position 3, a


= 1.25 m, eq
= - 0.5 ° Stability contours position 3, a
= 2.5 m, eq
= - 0.5 °
1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Positive eigenvalues contour for position 3, ζa = 1.25 m (b) Positive eigenvalues contour for position 3, ζa = 2.5 m

Figure 6.18: Barge 2a: Position 3 positive eigenvalues contour with U t ug = 2 m/s, x p0 = 0.5, ψeq = −0.5°, with waves

From the analysis is also found that the tug forward velocity does not influence the stability of position 3. With
a U t ug = 0.5 m/s the same result as sub-figures 6.17c and d were obtained. They are therefore not separately
presented. Reducing the forward velocity thus does not have a stabilising effect on position 3 but does allows
for equilibrium positions 2 and 4 to emerge.

Figure 6.19 shows the stability contours of Barge 2b. Since figure 6.15 shows that only position 3 is within
the range of values examined, the remaining positions are not discussed. The stability contour for position 3
with ζ = 1.25 m and ζ = 2.5 m are plotted in sub-figures a and b. Since position 3 for Barge 2b is already stable
without waves, it is expected that this position remains stable in head waves. Both figures endorse this.
6.3. Stability assessment with waves 89

Stability contours position 3, a


= 1.25 m, eq
= 0° Stability contours position 3, a
= 2.5 m, eq
= 0°
1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Positive eigenvalues contour for position 3, ζa = 1.25 m (b) Positive eigenvalues contour for position 3, ζa = 2.5 m

Figure 6.19: Barge 2b: Positive eigenvalues contour l 00 and ω dependency with U t ug = 2 m/s, x p0 = 0.5, with waves

6.3.3. Time domain simulation regular waves


To investigate the global behaviour of a towing operation, time domain (TD) simulation is used. Similar to
the without wave simulations, the towed vessel is released with initial conditions (IC) and the global reaction
is analysed. The same parameters as in section 6.1 are used: l 00 = 2.5 and x p0 = 0.5. The vessel is released
with a sway offset y 00 = 0.3. In some conditions, multiple stable solutions can exist so the initial conditions
are varied by checking multiple sway offsets to see if these influence the response. Most variations resulted
in solutions presenting the same time trace, after the transient response ended. Therefore, only the original
initial conditions are presented, with one exception which is clearly marked. Note that a systematic study
on the influence of the initial conditions has not been performed, and could provide more insight into the
behaviour of the tow operation.

To give insight in the stability contours, four simulations are presented from the ζa = 1.25 m case. The wave
frequencies that are examined are: ω = 0.80 rad/s, ω = 0.90 rad/s, ω = 1.00 rad/s and ω = 1.50 rad/s. These are
chosen because ω = 0.80 rad/s lies right on the boundary of stability of position 3 and thus is an interesting
case. It should result in stable towing behaviour. In contrast, position 3 with ω = 0.90 rad/s lies comfortably
within a stable region and is projected to yield a stable towing position while ω = 1.00 rad/s case is projected
to be unstable for all equilibrium positions. Here the barge should be towed in a dynamically stable fashion,
as is discussed in section 6.1. Last, ω = 1.50 rad/s indicates an unstable area for position 3, and a stable area
for position 2 and 4. This frequency is investigated to see if the local stability of positions 2 and 4 can lead to
global stability.

In sub-figure 6.20a, the wave condition with ω = 0.80 rad/s is presented for Barge 2a. The simulation shows
that the position is stable but that the convergence is slow. This is clearly visible in sub-figure 6.22a and d
where especially the yaw phase space shows the slow convergence. This slow convergence is the result of
small amplitude eigenvalues, which cause a slow exponential decrease on the excitation. For Barge 2b, sub-
figure b shows a much quicker convergence.

From sub-figure d of the phase space curves in figure 6.22 it can be concluded that the dominant position 3
equilibrium yaw angle is ψeq = 0°.

Other combinations of the stability contours, further away from the boundaries, will result in faster conver-
gence to the stable position 3. This is shown in sub-figure 6.20c where the Barge 2a simulation with ω = 0.90
rad/s is presented. Here the convergence time is already more similar to the convergence time of the course
stable Barge 2b results in sub-figure d. For both these stabilised vessels, the damping in the system is dis-
cussed in the discussion chapter 7.

The Barge 2a simulation with ω = 1.00 rad/s, in the unstable patch, is presented in sub-figure 6.21a. Note that
the initial sway offset has been increased to y 00 = 0.5 for this simulation, as the simulation with y 00 = 0.3 was
90 6. Towing Stability Analysis

not able to show the full range of the fishtailing. The time trace shows clearly dynamic stability, the vessel has
a slewing fishtailing motion. From sub-figures 6.22b and e it is found that the vessel sways on a limit cycle
with a path width of ∼ 0.5·L with a yaw angle range of ±22.5°. This result is what is expected from the stability
contours: unstable (or fishtailing) behaviour. Sub-figure d shows again that for Barge 2b the waves reduce the
convergence time towards the stable position 3.

Similar results are found with ω = 1.50 rad/s in figure 6.21. Sub-figure a shows the Barge 2a simulation and
it is clear no global stability is reached. The vessel exhibits a fishtailing motion and from the limit cycle in
the phase space diagrams in sub-figures 6.22c and f it is found that the path width is limited to ∼ 0.2 · L and
the yaw angle ψ range to ±12°. Clearly, the expectation of the vessel reaching the local stability of position 4
and converging towards that stable position is not met. The reduction of the path width is further discussed
in the discussion chapter 7. Sub-figure b shows that the addition of wave merely increases the rate of the
convergence to stable position 3.

TD Simulation, a
= 1.25 m, = 0.8 rad/s TD Simulation, a
= 1.25 m, = 0.8 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: w/o skegs, ω = 0.80 rad/s (b) Barge 2b: w/ skegs, ω = 0.80 rad/s

TD Simulation, a
= 1.25 m, = 0.9 rad/s TD Simulation, a
= 1.25 m, = 0.9 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle

0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(c) Barge 2a: w/o skegs, ω = 0.90 rad/s (d) Barge 2b: w/ skegs, ω = 0.90 rad/s

Figure 6.20: Time Domain simulation with waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 0.80 and 0.90 rad/s
6.3. Stability assessment with waves 91

TD Simulation, a
= 1.25 m, = 1 rad/s TD Simulation, a
= 1.25 m, = 1 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.5 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: w/o skegs, ω = 1.00 rad/s (b) Barge 2b: w/ skegs, ω = 1.00 rad/s

TD Simulation, a
= 1.25 m, = 1.5 rad/s TD Simulation, a
= 1.25 m, = 1.5 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(c) Barge 2a: w/o skegs, ω = 1.50 rad/s (d) Barge 2b: w/ skegs, ω = 1.50 rad/s

Figure 6.21: Time Domain simulation with waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 1.00 and 1.50 rad/s

TD Simulation: Sway Phase Space, a


= 1.25 m, = 0.8 rad/s TD Simulation: Sway Phase Space, a
= 1.25 m, = 1 rad/s TD Simulation: Sway Phase Space, a
= 1.25 m, = 1.5 rad/s
0.05 0.2 0.08

0.15 0.06

0.04
0.1

0 0.02
0.05
dy'/dt' [-]

dy'/dt' [-]

dy'/dt' [-]

0
0
-0.02
-0.05
-0.05 -0.04

-0.1
-0.06

-0.15 -0.08

-0.1 -0.2 -0.1


-0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Sway offset y' [-] Sway offset y' [-] Sway offset y' [-]

(a) Phase Space sway for ω = 0.80 rad/s (b) Phase Space sway for ω = 1.00 rad/s (c) Phase Space sway for ω = 1.50 rad/s

TD Simulation: Yaw Phase Space, a


= 1.25 m, = 0.8 rad/s TD Simulation: Yaw Phase Space, = 1.25 m, = 1 rad/s TD Simulation: Yaw Phase Space, = 1.25 m, = 1.5 rad/s
a a
8 20 10

6 8
15

6
4 10
4
2
/dt' [-]

5
/dt' [-]
/dt' [-]

2
0 0
0
d

d
d

-2 -5
-2

-4 -10
-4

-6 -15 -6

-8 -20 -8
-15 -10 -5 0 5 10 15 -25 -20 -15 -10 -5 0 5 10 15 20 25 -15 -10 -5 0 5 10 15
Yaw angle [°] Yaw angle [°] Yaw angle [°]

(d) Phase Space yaw for ω = 0.80 rad/s (e) Phase Space yaw for ω = 1.00 rad/s (f) Phase Space yaw for ω = 1.50 rad/s

Figure 6.22: Barge 2a Phase space diagrams, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m, ω = 0.80, ω = 1.00 and 1.50 rad/s
92 6. Towing Stability Analysis

6.3.4. Equilibrium positions irregular waves


The results in regular waves are examined and discussed, so the next step is to look at the effect of irregular
waves. Since irregular waves, by nature, will result in an irregular excitation this implies that in tow opera-
tions a stable towing position is never reached. The wave excitation will constantly change the equilibrium
positions and their stability. To still be able to predict the towing stability, the mean wave drift force due to
irregular wave spectrum is examined. This mean force will predict a mean equilibrium position and its stabil-
ity. In reality, the vessel may still oscillate around this position due to changing forces, but it will give a good
indication of the behaviour. Using the mean forces also excludes time simulation for verification purposes.
Therefore, these results only include the equilibrium positions and their stability assessment.

First Barge 2a, the unstable barge, is examined. The equilibrium positions vary on the basis of the x p0 and
T p . These are investigated individually, with base values of x p0 = 0.5, H s = 2.4 m and U t ug = 2 m/s. Initially,
the x p0 dependence of the vessel is investigated. Position 3 lies straight behind the tug, and will result in a
equilibrium yaw angle ψeq = 0°. Sub-figures 6.23a and b show the yaw angle ψeq of position 2 and 4. As
shown in figure 6.12, these two positions are symmetrical and this is confirmed by these figures. For practical
purposes, only position 3 and 4 are considered from here on.

Sub-figure 6.23b presents the yaw equilibrium angles ψeq of position 4. Here the dark red indicates out of
range values. The two red vertical lines show the limits of the physical T p range. In practice, only these
results are applicable. The thick horizontal red line indicates the value of x p0 = 0.5, the logical value of the
bow hawser attachment point. The without waves case can be interpreted as a sectional slice in very long
waves (T p > 16 s). Notice that this long wave approximation, that the waves do not influence the equilibrium
positions. This holds down to T p 10 s. What is immediately clear, is that with x p0 = 0.5, only the equilibrium
position 3 exist. The slight decrease in the profile in the physical T p range can be explained by the mean sway
force contribution and reduction of the mean yaw moment at those T p values.

If varying the variable U t ug with x p0 = 0.5, it will show that position 4 does not exist for every tug velocity
U t ug . Therefore, the x p0 dependency of position 4 for two tug velocities is compared. Sub-figure 6.25a shows
the result of a variations of U t ug , in this case U t ug = 0.5 m/s. It should be compared to sub-figure 6.23b with
U t ug = 2 m/s. From this comparison, it can be seen that in sub-figure 6.25a the equilibrium positions where
ψeq 6= 0° are reduced significantly. This directly relates to the force ratio between the manoeuvring and wave
drift forces. With a lower forward velocity, the manoeuvring forces contribution is smaller, and thus the effect
of the wave drift forces is larger.

Figure 6.24, showing the equilibrium yaw angles for position 2 and 4 of Barge 2b confirms the symmetrical
nature of these two positions. The analysis is proceeded with position 4 in sub-figure b. In the physical T p
range, the figure shows that the equilibrium yaw angle reaches larger values with smaller x p0 than with the
without wave scenario (approximate sectional slice at T p > 16 s). This is even more amplified when reducing
the forward velocity U t ug to 0.5 m/s in sub-figure 6.25b. This reduction is explained by the contribution of
the sway force, and the relative contributions of the manoeuvring and wave drift forces/moments. Larger
manoeuvring forces will push the curve up, larger wave drift forces will push it down. As positions 2 and 4 are
unstable for Barge 2b, and their yaw angles are out of range, they are not further considered. Only position 3
is further investigated.

6.3.5. Towing stability irregular waves


With the equilibrium positions determined, the stability of these positions is addressed. The equations of
motion are linearised at the positions and the Jacobian of the system is derived. The eigenvalues of the Ja-
cobian in turn give insight in the stability. Negative values indicate stability, positive values instability. The
boundary at zero is of interest and since the barges with x p0 = 0.5 and irregular waves only exhibit equilibrium
position 3, this is the only position investigated for stability.

Figure 6.26 shows the stability contours of the tow operation in irregular waves with significant wave height
H s = 2.4 m. Since equilibrium position 3 is determined numerically, an deviation in the yaw angle can be
present. As discussed before, stability determined on a position which is not exactly the equilibrium position
will theoretically yield an unstable result but since it assessed numerically this needs to be investigated. From
the equilibrium position determination with waves, ψeq = −0.5° for position 3 emerges. This can be the
numerical deviation of ψeq = 0° or the actual equilibrium position. Both are assessed to see the influence of
6.3. Stability assessment with waves 93

Yaw angle at critical (equilibrium) position 2 Yaw angle at critical (equilibrium) position 4
1 0 1 50

Physical Tp Physical Tp
-5 45
range range
0.8 -10 0.8 40

35

[°]
-15

[°]

Equilibrium yaw angle


Equilibrium yaw angle
0.6 -20 0.6 30

-25 25

0.4 -30 0.4 20

-35 15

0.2 -40 0.2 10

-45 5

0 -50 0 0
6 8 10 12 14 16 6 8 10 12 14 16

(a) Barge 2a: w/o skegs, position 2 (b) Barge 2a: w/o skegs, position 4

Figure 6.23: Barge 2a: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with irregular waves

Yaw angle at critical (equilibrium) position 2 Yaw angle at critical (equilibrium) position 4
1 0 1 50

-5 45

0.8 -10 0.8 40

35

[°]
-15
[°]

Equilibrium yaw angle


Equilibrium yaw angle

0.6 -20 0.6 30

-25 25

0.4 -30 0.4 20

-35 15

0.2 -40 0.2 10


Physical Tp Physical Tp
-45 5
range range
0 -50 0 0
6 8 10 12 14 16 6 8 10 12 14 16

(a) Barge 2b: w/ skegs, position 2 (b) Barge 2b: w/ skegs, position 4

Figure 6.24: Barge 2b: Yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s, with irregular waves

Yaw angle at critical (equilibrium) position 4 Yaw angle at critical (equilibrium) position 4
1 50 1 50
Physical Tp
45 45
range
0.8 40 0.8 40

35 35
[°]

[°]
Equilibrium yaw angle

Equilibrium yaw angle

0.6 30 0.6 30

25 25

0.4 20 0.4 20

15 15

0.2 10 0.2 10
Physical Tp
5 5
range
0 0 0 0
6 8 10 12 14 16 6 8 10 12 14 16

(a) Barge 2a: w/o skegs (b) Barge 2b: w/ skegs

Figure 6.25: Yaw angle equilibrium position 4 x p0 dependency with U t ug = 0.5 m/s, with irregular waves
94 6. Towing Stability Analysis

this change in equilibrium position on the stability contours. More information on this uncertainty in the
linearisation is discussed in chapter 7.

Sub-figure 6.26a presents the results for Barge 2a with ψeq = −0.5°, showing the physically possible T p range
between two black horizontal lines. Without waves, this barge is unstable, which can be compared to very
long peak periods, T p ∼ 16 s. The influence of the waves is not noticed above T p > 10 s. Considering the
physical T p range, Barge 2a can be towed in a stable manner, irrespective of the tow line length.

The same stability is assessed for ψeq = 0° in sub-figure c. Here, more unstable contours are present. Only
up to l 00 = 0.75 a stable area is present. Considering both these stability contours, it seems like there are
two equilibrium positions very close to each other. So the combined area of both are considered parameter
combinations that enable stable towing. In practice, within the physical T p range, position 3 with ψeq = −0.5°
will be the stable position for most parameter combinations.

Sub-figure 6.26b and d deliver the stability contours for Barge 2b. Again, both ψeq = −0.5° and 0° are analysed.
Without waves, this barge is course stable, so with waves this result would also be expected. In sub-figure d
this is presented. Sub-figure b shows something different. Here, there is a significant unstable area. This is
attributed to the linearisation of the equation of motion at a position not exactly the equilibrium position.
Barge 2b is already stable without waves, so it is reasonable to conclude that the ψeq = 0° is the real equilib-
rium position for this barge.

Stability contours position 3, H s


= 2.4 m Stability contours position 3, H s
= 2.4 m
16 16

15 15

14 14

13 13

12 12

11 11

10 10

9 9

8 8
Physical Tp Physical Tp
7 7
range range
6 6

5 5

0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Barge 2a: w/o skegs, ψeq = −0.5° (b) Barge 2b: w/ skegs, ψeq = −0.5°

Stability contours position 3, H s


= 2.4 m Stability contours position 3, H s
= 2.4 m
16 16

15 15

14 14

13 13

12 12

11 11

10 10

9 9

8 8
Physical Tp Physical Tp
7 7
range range
6 6

5 5

0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(c) Barge 2a: w/o skegs, ψeq = 0° (d) Barge 2b: w/ skegs, ψeq = 0°

Figure 6.26: Positive eigenvalues contour position 3 l 00 and T p dependency with U t ug = 2 m/s, x p0 = 0.5, with irregular waves
6.4. Assessment impact of the waves 95

6.4. Assessment impact of the waves


In this section the towing stability assessment without waves and with waves are compared and discussed.
The attributes that can be compared are the equilibrium positions and the stability of those positions. The
time domain simulations can also be compared for the regular wave results.

First, the equilibrium positions are compared. This is done for Barge 2a since the results showed the stability
of Barge 2b is not significantly influenced by the waves. As was previously mentioned, the results without
waves are the same as the low frequency or high peak period results in figure 6.27. Clearly the equilibrium
positions are influenced by the waves. Depending on the frequency, and thus the combination between the
surge, sway and yaw components, the equilibrium yaw angles change. From sub-figure b however, it is clear
that irregular waves do no create extra equilibrium position. Even though the irregular waves do not create
extra equilibrium positions, the stability of the equilibrium that is present, position 3, is influenced by the
waves. This is also the case for the regular waves. The additional equilibrium positions with x p0 = 0.5 are
limited, so the biggest influence lies with the stability of these positions.

The stability of position 3 in both regular and irregular waves is presented in figure 6.28. In sub-figure c, the
maximum eigenvalues of position 3 without waves are presented, and it shows that for l 00 > 0.15 Barge 2a is
unstable. With waves, the maximum eigenvalues are decreased, and stability (or negative eigenvalues) are
reached in some cases. Regular waves, see sub-figure a, improve the stability behaviour of position 3 signifi-
cantly if the wave frequency is greater than ∼ 0.70−0.80 rad/s, although unstable areas remain. Low frequency
waves do not improve the stability, since the wave drift force is almost non-existent in those conditions.

Irregular waves, see sub-figure b, also lead to stability. Again it can be noticed that long waves, with high T p
values, do not create stable conditions but that within the physical T p range for H s = 2.4m, stable conditions
are created. Ergo, the wave drift significantly improves the stability of an unstable barge towed in head waves.

This is confirmed for the regular wave results by the time domain simulations in figure 6.29. Sub-figure a
shows the TD simulation without waves present and the course unstable barge shows dynamic stability. This
dynamic stability is transformed to stable behaviour with the addition of waves with a wave amplitude of
ζa = 1.25 m and a wave frequency of ω = 0.90 rad/s.

Yaw angle at critical (equilibrium) position 4 Yaw angle at critical (equilibrium) position 4
1 50 1 50
Physical Tp
45 45
range

40 0.8 40
0.75
35
[°]
35
[°]

Equilibrium yaw angle


Equilibrium yaw angle

30 0.6 30

0.5 25 25

20 0.4 20

15 15
0.25
10 0.2 10

5 5

0 0 0 0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 6 8 10 12 14 16

(a) Regular waves (b) Irregular waves

Figure 6.27: Barge 2a: Comparison position 4 yaw angle equilibrium position x p0 dependency with U t ug = 2 m/s
96 6. Towing Stability Analysis

Stability contours position 3, H = 2.4 m


Stability contours position 3, a
= 1.25 m, eq
= 0° s
16
1.6
15

1.4 14

13

1.2 12

11
1 10

9
0.8
8
Physical Tp
7
range
0.6
6

5
0.4
0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Regular waves (b) Irregular waves

0.5

0.4
Max real part of eigenvalues

0.3

0.2
Position 1
Position 2
Position 3
0.1
Unstable
0
Stable
-0.1

-0.2
0 0.5 1 1.5 2 2.5

(c) Without waves

Figure 6.28: Barge 2a: Comparison position 3 positive eigenvalues with U t ug = 2 m/s, x p0 = 0.5

TD Simulation, without waves TD Simulation, a


= 1.25 m, = 0.9 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle

0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) TD Simulation w/o waves (b) TD Simulation w/ waves, ζa = 1.25 m and ω = 0.9 rad/s

Figure 6.29: Barge 2a: w/o skegs, TD simulation without and with waves
6.5. Chapter summary 97

6.5. Chapter summary


In this chapter, the towing stability of a tow operation is addressed. This is done based on the methodology
developed in chapter 3. First the stability without waves is addressed. The equilibrium positions of the two
barges (Barge 2a and Barge 2b) are determined and the eigenvalues of the linearised equation of motion at
these positions are derived. The eigenvalues show what is expected: Barge 2a is unstable and this results
in a (very) large fishtailing motion. Barge 2b is stable and converges to the equilibrium position straight
behind the tug: position 3. The wave drift force components are computed for Barge 2 and it is clear that the
wave drift damping method can be used to implement the wave drift force components into the equation of
motion.

With all the ingredients prepared, the stability analysis of the two barges with head waves is completed. The
equilibrium positions are determined numerically and for Barge 2a this shows that positions 2 and 4 occur in
certain wave conditions. Results for Barge 2b indicate that these equilibrium positions are also altered, but
since these are inherently unstable the impact is negligible. Addressing the stability of the Barge 2a equilib-
rium positions yields that stability can occur for head waves with ω > 0.65 rad/s. The Barge 2b results show
that the head waves do not destabilise the already course stable barge. Time domain simulation shows that
the local stability at equilibrium positions 2 and 4, not straight behind the tug, are too insignificant to lead to
global stability. It also shows that even if the vessel is not fully stable due to the waves, it still benefits as the
fishtailing motion is reduced.

In irregular head waves, it shows that no other equilibrium positions exist besides position 3: straight behind
the tug. The stability of this position is greatly improved for Barge 2a: for all peak periods within the physical
range, the barge is stable, irrespective of tow line length.

Comparing the results without and with head waves shows that the head waves have a positive effect on the
stability. They reduce the fishtailing motion and can in certain wave condition lead to stability.
7
Discussion
In this chapter the results obtained in the thesis are discussed. This is done on the basis of the objectives
defined in chapter 1. First, the stationary vessel assessment results are discussed in section 7.1. Secondly,
the sailing vessel assessment results are looked at in section 7.2 and thirdly, the towing stability analysis is
discussed in section 7.3. Finally, the uncertainties present in the project are discussed in section 7.4.

Before the stationary vessel assessment, sailing vessel assessment and towing stability could be assessed, a
solid methodology for going about these subjects was developed. The objectives set for this part included
defining requirements for the wave drift force in a tow resistance and towing stability setting. These require-
ments were the basis of the computations of the wave drift estimation methods in the stationary and sailing
vessel assessment chapters. Some of the requirements were adjusted on the basis of the initial simulations in
the towing stability chapter, but these were minor adjustments. The assessments in the stationary and sailing
vessel chapters are performed on the basis of the comparison methodology. This methodology is designed
to be able to include an arbitrary vessel geometry, to be able to repeat the current assessment for a variety of
vessels. For the towing stability analysis, a separate methodology was developed. This methodology mainly
looks at all the parameters involved in the stability assessment and how the impact of the waves can be de-
termined. For this entire process, two barges are used as test cases. Barge 1 is a rectangular barge and used in
the wave drift estimation method assessments. Barge 2 is a rectangular barge with a raked bow and stern and
is used in the stability analysis. All and all, the methodology objectives are all successfully completed.

7.1. Stationary vessel wave drift assessment


The stationary vessel assessment of the wave drift estimation methods consisted of the verification, com-
putation and comparison of several different wave drift estimation methods. As discussed in chapter 4, the
near field [59] or far field methods [49] are best suited for assessing the tow resistance. More rough estima-
tions, like the DNV 2014 formulation [14], can be used in practice, but during the revisit of the relative force
study these are not taken into account. Both the NF and FF methods yield very similar results for the surge
wave drift force, but the FF method is chosen to be used in the study since it is the base value of the previous
comparisons.

7.1.1. Stationary wave drift tow resistance contribution


To revisit the relative contribution of the wave drift force for the stationary vessel, the initial study is used
as starting point. For these initial results, see appendix A. The typical sea state for stationary (or survival)
conditions is used. It entails H s = 5 m, U wi nd = 20 m/s and Uc = 0 m/s. In this case, head waves and head
wind are assumed.

The wind forces for this calculation are based on the same coefficients used for the Smit Barge 2. This barge is
very similar in geometry as the Barge 2 and the same deck load is assumed. These coefficients are interpolated
in sub-figure 7.1a. The current coefficients are based on the manoeuvring derivatives of Barge 2a (w/o skegs)
and are further explained in chapter 2 and are presented in sub-figure b. It was chosen to evaluate Barge 2
instead of Barge 1 for the stationary tow resistance as Barge 2 more resembles a realistic ocean-going barge.

99
100 7. Discussion

Barge 2a: Cubic Spline fitted Wind Coefficients Barge 2a: Current Coefficients based on manoeuvring derivatives
0 1.5 -0.2 1
Cubic Spline fitted Surge coefficient Surge coefficient
Surge coefficient data Sway coefficient 0.8
-0.2 Cubic Spline fitted Sway coefficient 1 -0.25 Yaw coefficient
Sway coefficient data 0.6
Cubic Spline fitted Yaw coefficient

Sway/Yaw Coefficient [-]


Yaw coefficient data 0.4
Surge Coefficient [-]

-0.4 0.5

Surge Coefficient [-]


-0.3

Sway Coefficient [-]


0.2

-0.6 0 -0.35 0

-0.2
-0.8 -0.5 -0.4
-0.4

-0.6
-1 -1 -0.45
-0.8

-1.2 -1.5 -0.5 -1


-80 -60 -40 -20 0 20 40 60 80 -50 -40 -30 -20 -10 0 10 20 30 40 50
Yaw angle [°] Yaw angle [°]

(a) Wind Coefficients for Barge 2a (b) Current Coefficients for Barge 2a

Figure 7.1: Wind and Current coefficients for Barge 2a

10-4 Barge 2a: Tow Resistance, = 0 °, Sea State 1 Barge 2a: Relative Contributions, = 0 °, Sea State 1
2 100
Total Force, U rel = 0 m/s rel. Current Force, U rel = 0 m/s
1.8 90
Current Force, U rel = 0 m/s rel. Wind Force, Urel = 0 m/s
1.6 Wind Force, Urel = 0 m/s 80 rel. Wave Drift Force, U = 0 m/s
Percentage of Total Force [%] rel
Wave Drift Force, U rel = 0 m/s
1.4 70

1.2 60

1 50

0.8 40

0.6 30

0.4 20
Physical Tp Physical Tp
0.2 10
range range
0 0
5 6 7 8 9 10 11 12 13 14 15 16 5 6 7 8 9 10 11 12 13 14 15 16

(a) Stationary Tow Resistance forces (b) Relative stationary force contributions

Figure 7.2: Relative force contribution stationary vessel - T p dependency

In figure 7.2 the relative contribution of the forces is plotted as function of the wave spectrum peak period T p .
For the stationary vessel, the current forces are zero and the total force is made up from the wave drift and
wind forces. In the physical range in sub-figure a it is clear that because the T p are higher than 8 s, the mean
wave drift contributions are limited. From sub-figure b, it is seen that the mean wave drift forces contribute
maximally 18% of the tow resistance of the stationary vessel. It must again be noted that the wind coefficients
are obtained for an arbitrary deck load in one the Boskalis projects and thus only give an indication of the
wind forces. The wave drift forces will in practice vary since they are based on irregular waves and describe
merely the mean value. A study on the extreme values of the wave drift is not included in this thesis.

When comparing this result with the Smit Barge 2 result in the initial background study, the contribution of
the wave drift force is less than initially expected. Initially, the wave drift contribution was expected to be
∼ 50% of the stationary tow resistance but this is significantly less (18%). The main cause for this difference is
the fact that the physical T p range does not include the peak wave drift force.

The stationary vessel is assessed for survival scenarios, so a more conservative approach will make sure the
chosen towing tug has enough propulsion power. The DNV 2014 formulation is a good example for a more
conservative approach, predicting roughly a twice as large wave drift force in the physical T p range than the
NF and FF method. However, similar to a dynamic positioning system, the average tow resistance presented
here can still be very useful. Depending on the duration of peak loads and the allowed drift in survival condi-
7.2. Sailing vessel wave drift assessment 101

tions, the tug’s towing power might be sufficient if selected on the average resistance. During peak loads, the
configuration will drift since the tug cannot compensate, but if the configuration remains within the allowed
boundaries, the tug can compensate its position when the loads drop below average. Adopting a similar sys-
tem might reduce the necessary tug propulsion power but the operability of this configuration needs to be
assessed.

It should also be stressed that during this assessment, only one type of vessel was examined: a rectangular
barge with raked bow and stern of L = 76.2 m and C oG z = 13.8 m. With longer vessels, the wave energy would
be located more around the wave drift peak in the motion zone, resulting in larger wave drift forces and with
larger C oG z a larger peak is expected. Simultaneously the height and location of the WDF peak could shift due
to other vessel characteristics. To definitively determine the impact of the wave drift force on the stationary
tow resistance and thus the tug selection more research into different vessel geometries is necessary. The
framework of this research is drafted for this purpose, to be able to evaluate an arbitrary vessel.

With the further research in mind, the impact of the wave drift force on a rectangular barge with raked bow
and stern for the stationary tow resistance, and thus the tug selection, is limited. The majority of the resis-
tance is due to the wind. Still, the effect of the waves should be taken into account and the average wave drift
force can be the basis of tug selection with operability boundaries in the survival scenario. To avoid such an
approach, a conservative approximation like the DNV 2014 formulation is sufficient for the wave drift predic-
tion as well.

7.2. Sailing vessel wave drift assessment


The sailing vessel wave drift assessment comprised just as for the stationary vessel, the verification, com-
putation and comparison of several different wave drift estimation methods. The computed methods and
their comparison are presented in chapter 5. The sailing vessel assessment consisted of two parts, the tow
resistance assessment and the towing stability assessment. These are put into context separately.

7.2.1. Sailing wave drift tow resistance contribution


The near field and far field methods combined with the wave drift damping are considered the best suited
methods for assessing the tow resistance of the sailing vessel. The Faltinsen SW method is another possibility
but is a rough approximation. Since the NF and FF methods yield very similar results for the surge wave drift
force, the FF method is chosen to be used in the study instead of both.

In figure 7.3, the relative contributions to the tow resistance of the sailing vessel can be found. Both the contri-
butions for Ur el = 0 and 2 m/s are presented. Sub-figure a presents the Ur el = 0 m/s results in the solid lines,
showing that the maximum wave drift contribution is present in the physical T p range at T p = 5.7 s. With
increasing velocity, the peak wave drift contribution shifts to T p ∼ 6.5 s. This shift happens due to the wave
encounter frequency shift. With Ur el = 2 m/s, both the current and wind force have also increased, resulting
in the maximum tow resistance doubling with respect to Ur el = 0 m/s. Sub-figure b shows the relative contri-
butions of these components to the total force. In the Ur el = 0 m/s case, the wave drift contributions is ∼ 34%
while the wind contribution is ∼ 66%. With forward velocity, the wave drift contributions decreases slightly
to 25 − 30%, while the current contribution becomes ∼ 20% and the wind contribution drops to ∼ 50 − 55%.
Ergo, with forward velocity, the wave drift contribution does not change significantly but the peak value shifts
to larger peak periods.

The wave drift contribution with T p = 7 s is further evaluated with respect to the forward velocity. Figure 7.4
shows the Ur el dependence of the tow resistance contributions. Sub-figure 7.4a shows that the total tow resis-
tance almost quadruples between Ur el = 0 and 3.5 m/s. Both the wave drift and the wind force show a linear
increase with the forward velocity, but the current force increases quadratically and is the main contributor to
the increase of the total tow resistance. In sub-figure b, with Ur el = 0 m/s, the relative contributions of wind,
waves and current are respectively 75%, 25% and 0%. With Ur el = 2 m/s this ratio changes to ∼ 50%, ∼ 30%
and 20% and with Ur el = 3.5 m/s, the ratio becomes ∼ 40%, ∼ 20% and ∼ 40%. Again this shows that the wave
drift contribution remains relatively constant (between 20 − 30%) with forward velocity up to 3.5 m/s.
Comparing this result with the Smit Barge 2 results of the initial case studies performed in chapter 1, shows
that the wave drift force contribution is lower than initially projected. Since the sailing vessel is usually as-
sessed to determine the towing velocity, the mean wave drift force is sufficient for this purpose. Even though
102 7. Discussion

10-5 Barge 2a: Tow Resistance, = 0 °, Sea State 2 Barge 2a: Relative Contributions, = 0 °, Sea State 2
9 100
rel. Current Force, U rel = 0 m/s
8 90
rel. Current Force, U rel = 2 m/s

7 80 rel. Wind Force, Urel = 0 m/s

Percentage of Total Force [%]


rel. Wind Force, Urel = 2 m/s
70
6 rel. Wave Drift Force, U rel = 0 m/s
60 rel. Wave Drift Force, U rel = 2 m/s
5 Total Force, U rel = 0 m/s
Total Force, U rel = 2 m/s 50
4
Current Force, U rel = 0 m/s
40
3 Current Force, U rel = 2 m/s
Wind Force, Urel = 0 m/s 30
2 Wind Force, Urel = 2 m/s
20
Wave Drift Force, U rel = 0 m/s
Physical Tp Physical Tp
1 10
Wave Drift Force, U rel = 2 m/s
range range
0 0
5 6 7 8 9 10 11 12 13 14 15 16 5 6 7 8 9 10 11 12 13 14 15 16

(a) Sailing Tow Resistance forces with Ur el = 0 and 2 m/s (b) Relative sailing force contributions with Ur el = 0 and 2 m/s

Figure 7.3: Relative force contribution sailing vessel - T p dependency

Barge 2a: Tow Resistance, = 0 °, Sea State 2, T p = 7 s Barge 2a: Relative Contributions, = 0 °, Sea State 2, T p = 7 s
10-4
1.6 100
Total Force rel. Current Force
1.4 Current Force 90 rel. Wind Force
Wind Force rel. Wave Drift Force
Wave Drift Force 80
Percentage of Total Force [%]

1.2
70
1
60

0.8 50

0.6 40

30
0.4
20
0.2
10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5

(a) Tow Resistance forces with T p = 7 s (b) Relative force contributions with T p = 7 s

Figure 7.4: Relative force contributions sailing vessel - Ur el dependency

it shows to be lower than originally projected, it is still the origin of roughly 25 to 30% of the total tow re-
sistance, irrespective of forward velocity. The fact that the relative contribution stays roughly constant with
forward velocity indicates that the addition of forward velocity in the determination of the wave drift force is
significant. Especially with relatively lower forward velocities, Ur el < 2 m/s, the contribution of the wave drift
force is vital.

From the towing stability analysis, it was found that the most common equilibrium position for towing is
straight behind the tug with ψeq = 0°. This ensures that the towing resistance is equal to the surge force
components of the vessel. However, as was also found from the towing stability analysis, that cases with a
different equilibrium yaw angles can occur naturally. This implies that the tow resistance will increase due
to the yaw angle of the vessel and the corresponding sway force contribution. This is also the case when the
vessel can only be towed in a dynamically stable manner, the average tow resistance will increase. This should
be considered per case.

The impact of the wave drift force on the sailing vessel is larger than on the stationary vessel. The increased
net wave drift force due to the forward velocity is significant and consequentially, the impact the waves have
on the towing velocity is thus also larger than its impact on the tug selection. Again should be stressed that
7.2. Sailing vessel wave drift assessment 103

this result is obtained for the specific Barge 2 and for more definitive conclusions a database of vessels should
be evaluated systematically. The process that has been outlined in this thesis is the basis of constructing such
a database. The evaluation of Barge 1 shows it works and it can be applied to different vessel geometries.

7.2.2. Sailing wave drift towing stability contribution


For the contribution of the wave drift force with respect to the towing stability not only the surge and sway
forces are of importance, but also the yaw moments. As discussed in the sailing vessel chapter 5, the wave drift
estimation methods chosen to implement in the towing stability analysis are the FF method for the surge and
sway force, and the near field method for the yaw moment. These contributions are put in context by looking
at their contributions while also taking the current and wind related components into account.

An overview of all four force component contributions: current/manoeuvring, tow line, wind and wave, is
created by including the wind force into the TD simulation. This simplified implementation for the wind
force is done on the basis of the literature presented in section 2.2 in chapter 2. From the wind coefficients,
which are obtained from an arbitrary deck load, it is already clear that head wind will have a stabilising effect
on the tow operation. This is derived as the coefficients are symmetrical and the moment coefficient has the
same sign as the sway contribution. From literature, both Sinibaldi et al. [67] and Fitriadhy et al. [24] show
that head wind can have destabilising effects. This is possible, for example if the deck load is asymmetrical or
acts like a vertical wing. The following assessment shows the potential influence of the wind contribution as
a stabilising force, but especially shows the relative contribution with respect to the waves. To show the latter,
it does not matter if the wind has a stabilising or destabilising effect.

In figure 7.5 six time domain simulations are shown. Sub-figures a, c and e show respectively the without
waves or wind simulation, the waves only simulation (ω = 1.00 rad/s), and the wind only simulation. Note
that the initial conditions of these simulations are slightly different than those of the simulations in sub-
figures b, d and f.

In sub-figure 7.5a the without wind and waves simulation is presented. This simulation is presented as refer-
ence material for the waves and wind only simulations. Sub-figure 7.5c has already been discussed in chapter
6 but shows that the waves do not fully stabilise the operation under these conditions but reduce the path
width from ∼ 2.1 · L to 0.5 · L and the yaw angle range from ±50° to ±22.5°. More information on the path
width is given in section 7.3. The wind in sub-figure 7.5e also reduced the path width in exactly the same
manner as the waves with ω = 1.00 rad/s. Comparing the waves only and wind only results shows that the
waves and wind in these conditions have the same stabilising effect. Again, this is very dependent on the
environmental conditions.

From sub-figures 7.5b, d and f, the same simulations are run but for different wave conditions. Again, in
sub-figure b the reference simulation without wind and waves is presented. In sub-figure d, the waves only
simulation with ω = 0.90 rad/s is presented, showing that the head waves fully stabilise the tow operation.
The wind in sub-figure f, as previously found, shows only a decrease in path width. In these conditions the
waves contribute significantly more to the towing stability of the towed vessel than the wind.

In figure 7.6, the force components of the simulations in sub-figures 7.5c and e are plotted. Sub-figures a and
c show the sway force contributions, sub-figures b and d the moment contributions. Here the manoeuvring
(in blue) and tow line forces (in red) remain roughly the same for both sway and yaw between the two sim-
ulations. The wind force contribution (in yellow) also equals the wave drift contribution (in black) for both
force components. Note that in both cases the wind and wave contributions are a factor 1000 smaller than the
tow line and manoeuvring contributions. This shows that the effects of changes to the manoeuvring/current
contribution are much larger than changes to the wave drift or wind contributions.

Even though in this case the small addition of the waves/wind already decreases the path width significantly,
changes to the manoeuvring/current contribution can impact the stability in a more significant way. It is
therefore essential that the manoeuvring derivatives of the vessel are known with a good degree of certainty.
It also shows that the main contribution to the stabilising moment does not necessarily come directly from
the stabilising wind/wave moment. The extra resistance generated by the waves requires the towing tug to
produce a larger towing force. Assuming this towing force is available, the larger towing force results in a
larger stabilising tow line moment. This is the main stabilising contribution of the wave drift.
104 7. Discussion

TD Simulation, a
= 0 m, = 0 rad/s, U w = 0 m/s TD Simulation, without waves
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.5 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: TD Simulation, no waves/wind (b) Barge 2a: TD Simulation, no waves/wind

TD Simulation, a
= 1.25 m, = 1 rad/s, U w = 0 m/s TD Simulation, a
= 1.25 m, = 0.9 rad/s, U w = 0 m/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.5 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(c) Barge 2a: TD Simulation, ζa = 1.25 m, ω = 1.00 rad/s, U w = 0 (d) Barge 2a: TD Simulation, ζa = 1.25 m, ω = 0.90 rad/s, U w = 0
m/s m/s

TD Simulation, a
= 0 m, = 0 rad/s, U w = 10 m/s TD Simulation, a
= 0 m, = 0 rad/s, U w = 10 m/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.5 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle

0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(e) Barge 2a: TD Simulation, ζa = 0 m, U w = 10 m/s (f) Barge 2a: TD Simulation, ζa = 0 m, U w = 10 m/s

Figure 7.5: TD simulations with wind and waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5
7.2. Sailing vessel wave drift assessment 105

From figures 7.5 and 7.6 it can be concluded that in the context of the towing stability, head waves have a
significant effect on stabilising the tow operation. Under the assumption that the wind yields a destabilising
contribution, the head waves still provide a net increase of the stability. This is again highly dependent on the
precise wave conditions. It is also clear, that a more detailed study into the wind contributions is necessary
to definitively determine their impact.

The impact of the waves on the towing stability is determined twofold. Given that the towing tug has enough
towing force available, the added resistance due to the waves increases the tow line tension and thus also
its tow line moment. Together with this added resistance effect, a moment is directly generated due to the
waves. These two effects can both contribute to the towing stability of the vessel.

Although in this case the stabilising effect is determined for the wave drift, it is not limited to the wave drift.
All the sources of extra resistance contribute to the stability of the towed vessel using the same phenomenon.
The resistance increases the necessary towline force, and given that the towing tug is capable of supplying
that towing force, the stabilising moment due to the towline force increases. If the resistance is large enough,
this stabilising moment is the main contributor to the then obtained towing stability. Possible other ways to
obtain this extra resistance is to trim the vessel backwards or, as a more drastic approach, throw out a drag or
sea anchor.

TD Simulation - Sway Forces TD Simulation - Yaw Moments


a
= 1.25 m, = 1 rad/s, U w = 0 m/s a
= 1.25 m, = 1 rad/s, U w = 0 m/s
0.5 0.1
Y man Nman
YT NT
0 0

-0.5 -0.1
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

10-3 10-4
1 1
Y N
wave wave

0 0

-1 -1
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: TD Simulation - Sway, ζa = 1.25 m, ω = 1.00 rad/s, (b) Barge 2a: TD Simulation - Yaw, ζa = 1.25 m, ω = 1.00 rad/s,
U w = 0 m/s U w = 0 m/s

TD Simulation - Sway Forces TD Simulation - Yaw Moments


a
= 0 m, = 0 rad/s, U w = 0 m/s a
= 0 m, = 0 rad/s, U w = 10 m/s
0.5 0.1
Y man Nman
YT NT
0 0

-0.5 -0.1
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

10-3 10-4
1 1
Y wind Nwind

0 0

-1 -1
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(c) Barge 2a: TD Simulation - Sway, ζa = 0 m, U w = 10 m/s (d) Barge 2a: TD Simulation - Yaw, ζa = 0 m, U w = 10 m/s

Figure 7.6: Forces and moments of the TD simulations with wind and waves, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5
106 7. Discussion

7.3. Towing stability analysis


The towing stability analysis consists of three parts. The towing stability analysis without waves, the towing
stability analysis with waves and the comparison between the two analyses to determine the impact of the
waves. All these objectives were successfully completed in the towing stability chapter 6. The general conclu-
sion of the source of the test vessel, Hong et al. [37], matches the present results: head waves can enhance
the towing stability significantly. However, the quantitative data differs due to simplifications of the equation
of motion and wave drift computation in the source paper. Therefore, to better interpret the impact of the
(head) waves on the stability some of the results are further discussed.

Based on stability theory and the evaluation of the contributions of the different force components in figure
7.5, see section 7.2, it is clear that the moments play the largest role in the stability analysis. Simply put, if the
stabilising moments are larger than the destabilising moments, towing stability is reached. This is including
the tow line moment due to the towing tug.

Without wind, in practice without a deck load for the wind to interact with, the present results are already
sufficient to reach stability. For example, when sailing to a project location to pick up a deck load, with an
unstable barge like Barge 2a, sailing into the waves will improve the stability of the operation. Since at this
point the towing tug has enough available towing force, the extra resistance due to the waves is not an issue.
As shown in the previous section, with wind it becomes more nuanced but still the contribution of the waves
can be the deciding factor between stability and instability. Even when stability is not reached, the influence
of the waves can still benefit the towing operation. To sail safely, a stable vessel is preferred but it is possible
to sail with an unstable vessel if the available towing force of the towing tug is large enough. This results in a
dynamically stable towing operation, otherwise known under the term fishtailing.

7.3.1. Fishtailing
Fishtailing is the phenomenon where the towed vessel oscillates behind the towing tug and this oscillation
resembles a fishtail while swimming. A measure of the applicability of such an operation is the path width of
the operation: how wide is the slewing motion of the vessel?

Without waves, as seen in sub-figure 7.5a, the path width of the unstable Barge 2a is ∼ 2.1 · L. The non-
normalised path width is thus 160 m for a tow line of 190 m long. This is a very wide path and will most
definitely result in tough operational control. With waves, this path width is reduced even though Barge 2a
is still not stable. In figure 7.7 the TD simulations previously presented in chapter 6 are presented again.
Here the blue lines represent the sway offset and the black lines the yaw angles. As is immediately visible, for
ω = 1.00 rad/s the path width is significantly reduced to 0.5 · L or 37.5 m. For ω = 1.50 rad/s this reduction
is even larger to 0.2 · L or 15 m. This path width becomes a lot more manageable and depending on the
operational situation, this is or is not considered sufficient.

Basically, as long as the towing tug has a large enough available towing force, head waves add a positive effect
to the towing stability. In most optimal conditions the waves will result in an unstable barge like Barge 2a to
be towed stably but even if those conditions are not met, the tow operation becomes more manageable due
to the inclusion of waves.

7.3.2. Damping
Damping in the tow operation can be beneficial: the motions of the towed vessel are damped and the tow
operation can become stable or at least more manageable. The two sources of damping are potential damp-
ing and viscous damping. Potential damping includes effects like the radiation damping due to the vessel
motions and the wave drift damping. Viscous damping includes effects like wind and current damping. In
the equations of motion examined, the manoeuvring force components include the viscous hydrodynamic
damping while wind damping is not included. The wave drift damping is included in the wave drift force
components. The roll damping is neglected since roll has a relatively small impact on the stability analysis
although it can become important in fishtailing scenarios [66]. Tow line damping is also neglected to simplify
the analysis but this could be included in future work. Generally, the hydrodynamic (viscous) damping is the
dominant damping [75] but from Wichers [74] and Hermans [35], it is well understood that in the case of high
linear waves the effect of the wave drift damping dominates the damping.

As mentioned, the wave drift damping has been included in the wave drift force components during the pro-
7.3. Towing stability analysis 107

TD Simulation, a
= 1.25 m, = 1 rad/s TD Simulation, a
= 1.25 m, = 1.5 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.5 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: w/o skegs, ω = 1.00 rad/s (b) Barge 2a: w/o skegs, ω = 1.50 rad/s

Figure 7.7: TD simulations of Barge 2a for fishtailing discussion, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m

cess of this thesis. It can however be easily rewritten as a damping term. The same goes for the manoeuvring
forces. This is done in equations (7.1) to (7.3). Here, the damping terms are the terms associated with a
velocity.
¡ 0 0
¢ 0 ¡
u̇ − B 11 − R 00 · u 0 · u 0 − m 22
¡ 0
· r 0 + B 12 + X v0 v · v 0 · v 0 − B 16 + X r0 r · r 0 · r 0 = X T0 + X w
0
¢ ¢ ¡ ¢
m + a 11 ave (7.1)
¡ 0 0
¢ 0
¡ 0 0
¢ 0
¡ 0 0 02
¢ 0
¡ 0 0 02
¢ 0 0 0
m + a 22 v̇ + m 11 · r − B 21 · u − B 22 + Y v + Y v v v · v · v − B 26 + Yr + Yr r r · r · r = YT + Y w ave (7.2)
¡ 0 0
¢ 0 0
¡ 0 0 02
¢ 0 ¡ 0
¢ 0 0 0
I zz + a 66 · r˙ − B 61 · u − B 62 + N v + N v v v · v · v − B 66 + Nr · r = NT + N w ave (7.3)
Looking into the relative contribution of the wave drift damping is complicated. Since the contribution to
the damping of the system is different for each wave frequency, a general consensus is hard to reach. To be
able to quantify the damping, two of the time domain simulations presented in chapter 6, are analysed. From
these simulations, again presented in figure 7.8, the damping properties of the system in those conditions
are determined. As is clear from the figures, both systems are under damped, and the damping ratio in sub-
figure 7.8a is much smaller than the ratio in sub-figure b. This follows from the rate of convergence to the
stable towing position: a higher convergence rate indicates a larger damping ratio. The damping ratio of the
system in sub-figure a is determined using the logarithmic decrement method [25] as the damping ratio is
expected to be below ζ < 0.5, see equation (7.4). For the system in sub-figure b, the overshoot method is used
[25] because the damping ratio is expected between 0.5 < ζ ≤ 0.8, see equation (7.5).
δ δ x(1)
µ ¶
1
ζ= p or (δ ¿ 1) with: δ = ln (7.4)
4π2 + δ2 2π n x(1 + n)
s
ln (PO)2
ζ= (7.5)
π + ln (PO)2
2

where n is the number of periods investigated, x(1 + n) is the value of the peak of the n t h oscillation and
PO is the overshoot percentage of the first oscillation with respect to the relative distance between the initial
condition and the final value.

The damping properties are presented in table 7.1, showing the damping ratio and the damped and natural
slewing frequencies. These frequencies are normalised as discussed in the literature review chapter 2 and
describe the time it takes for the vessel to complete one slewing cycle. The non normalised periods are re-
spectively 380 s and 525 s for the two simulations. Since the simulation shows it takes between ∼ 7 and 1.5
oscillations for the vessel to stabilise, it implies that with the tug forward velocity of U t ug = 2 m/s the distance
until the stable towing position is reached lies between 1500 m and 5000 m. This is a considerable distance
and depending on the operational situation this is or is not sufficient for safe towing.

It is clear that with a small frequency shift, from ω = 0.80 to 0.90 rad/s the damping ratio increases signif-
icantly. This is attributed to the sway force contribution of the waves. This contribution start contributing
from ω > 0.85 rad/s, see figure 6.10, and increases the damping ratio.
108 7. Discussion

TD Simulation, a
= 1.25 m, = 0.8 rad/s TD Simulation, a
= 1.25 m, = 0.9 rad/s
1.5 60 1.5 60
IC: IC:
l0 ' = 2.5 l0 ' = 2.5
1 40 1 40
0
= 0° 0
= 0°
y0 ' = 0.3 y0 ' = 0.3
0.5 20 0.5 20
Sway offset y' [-]

Sway offset y' [-]


[°]

[°]
Yaw angle

Yaw angle
0 0 0 0

-0.5 -20 -0.5 -20

-1 -40 -1 -40

-1.5 -60 -1.5 -60


0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Time t' [-] Time t' [-]

(a) Barge 2a: w/o skegs, ω = 0.80 rad/s (b) Barge 2a: w/o skegs, ω = 0.90 rad/s

Figure 7.8: TD simulations of Barge 2a for damping properties determination, U t ug = 2 m/s, x p0 = 0.5, l 00 = 2.5, ζa = 1.25 m

Table 7.1: Damping properties of the two examined time domain simulations

Damping ratio Damped slewing frequency Natural slewing frequency


ζ ω0sd ω0s0
Simulation with ω = 0.80 rad/s 0.05 0.62 0.62
Simulation with ω = 0.90 rad/s 0.54 0.38 0.46

7.3.3. Minimum required wave drift for stability


From comparing figures 7.7 and 7.8, it shows that a minimum wave drift force and damping is necessary for
the tow operation to become stable. This can be easily extended to the minimum required added resistance
necessary for towing stability. Knowing this minimum amount is useful, and this can be quantified on the
basis of Routh’s stability criterion, see the literature chapter 2 for more information. In particular the paper
of Garza-Rios and Kekrides [27], who implemented the stability criterion to discuss the stability of a single
point moored vessel, can be used as the basis for considering this criterion for the towing operation. This
criterion produces a binary result: stable or not, and with that result, this description can be rewritten into a
requirement of the wave drift necessary to attain stability. This can be even done in a broader sense, what is
the minimum required extra resistance necessary for stability? Since the process of rewriting the equations
of motion into the Routh’s stability criterion coefficients and drafting the requirements for the wave drift is a
large mathematical derivation, it is recommended this is pursued in further research.

7.3.4. Slowly-varying part of the wave drift


In the thesis, the main focus of the wave drift force has been the mean component. However, as discussed
in the literature chapter 2, the low frequency slowly-varying component exists and can influence the tow
operation. Similar to moored vessels, the slowly-varying components will results in a slow relative motion
between the towing tug and the towed vessel. This relative motion is influenced by the stiffness and damping
properties of the towing line and naturally the amplitude of the slowly-varying part of the wave drift force.
This extra effect due to the waves will have an effect on the towing stability, as the equilibrium positions will
not be constant and therefore changes to the stability will occur. Since the slowly-varying part of the wave
drift force and its interaction with the tow line characteristics is a complex phenomenon, it is recommended
that further research is dedicated to this issue.

7.4. Uncertainties in the results


In the previous sections, the results have been put into perspective by placing them in context with the other
environmental forces. Besides putting them in perspective it is important to realise that uncertainties are
introduced during the modelling process of the thesis. Various uncertainties will emerge when addressing
an engineering issue with (numerical) modelling. This is also the case for the project described in this thesis.
7.4. Uncertainties in the results 109

These uncertainties can be classified into three categories:

• Theoretical uncertainties;
Are there uncertainties included in the theory due to assumptions and therefore in the theoretical mod-
els?
• Input uncertainties;
How are the parameters used as input for the modelling determined and do they include uncertainties?
• Numerical uncertainties;
During the modelling numerical simulations are run, what kind of uncertainties are introduced with
those?

These three categories are discussed in more detail.

7.4.1. Theoretical uncertainties


First the theoretical uncertainties are addressed:

• Potential theory
One of the main theoretical tools applied in the thesis is the potential theory to describe the flow around
the vessel. This theory neglects the viscous effects of the flow and is therefore only accurate in condi-
tions where those components are negligible. Using the theory near the boundaries of its validity can
introduce uncertainties. Several assumptions in the modelling process are implemented to remain in
conditions where potential theory is valid and thus these uncertainties will not be significant.

• Wave drift estimation methods


Part of this thesis is the comparison of several different wave drift estimation methods. Multiple ob-
jectives of the thesis are related to assessing these methods on their applicability for the tow resistance
and towing stability. So by nature, these methods include some uncertainties. The methods are vali-
dated with model data by their original sources, but not for the specific vessels considered in this thesis.
During the comparison process these uncertainties have been discussed and weigh in on the final esti-
mation method selections. Still, the uncertainties remain but they are not deemed significant enough
to commit to towing tank tests to further verify them.

• Wave drift damping


In particular the yaw wave drift damping approximation proposed by Aranha [3] remains a rough ap-
proximation. There have been attempts to validate the equations but still uncertainties on its validity
remain. Since the approximation based on slender body theory is valid for the featured vessels, and the
yaw coefficients are similar to the surge and sway coefficients found, the results are trusted enough.

• Equation of motion
The equations of motion for the stability analysis are only defined for three degrees of freedom: surge,
sway and yaw. The other three components, heave, roll and pitch also play a role in the complete pic-
ture of the tow operation but are at this point neglected. The absence of these components in the equa-
tion of motion naturally introduces some uncertainty but their effects are not completely neglected.
The heave motion in combination with the pitch motion has an influence on the relative wave height,
the main driver of the wave drift force. This is taken into account when calculating the wave drift forces.
The roll and pitch motions are also influenced by the location of the center of gravity, as is shown in the
vertical center of gravity study in appendix B. The larger motions will increase the peak of the QTF and
shift the peak towards lower frequencies. For example, for an increase of C oG z = 10 m, the surge QTF
peak amplitude increases with 50% and shifts ω = −0.10 rad/s.

So, the effects of the heave, roll and pitch motions are taken into account when addressing the wave
drift, but are not included directly in the equation of motion. For the complete dynamic tow operation
modelling the additional motions should be taken into account. For example, the roll motion will have
an influence on the slewing motion behaviour of the vessel, which can result in large unwanted roll
angles when the yaw turning rate is high. However, since the impact of the wave drift is studied, and
their effect on this phenomenon is taken into account, this approach is deemed sufficient.
110 7. Discussion

• Neglected effects
During the modelling, various effects are neglected for ease of computations and to reduce complex-
ity. Effects like propeller race or tug wash, where the tug’s propeller output flow interacts with the
towed vessel [54], or wave breaking loads, where extra non-linear loads are transmitted due to the waves
breaking on the bow [19, 41], can be considered small if tow lines are long and waves with moderate
steepness are considered. Another assumption, like the tow lines are taut and can only transmit tension
introduce larger uncertainties. The results with this assumption remain valid but including a catenary
tow line can reduce the uncertainty of this effect in the modelling process.

7.4.2. Input uncertainties


Secondly, the input uncertainties are addressed:

• Simulation model properties:


The properties of the simulation model can include some uncertainties. Some of the properties are de-
termined from model tests, like the manoeuvring derivatives featured later on, and some are calculated
based on the design specifications. In practice, these values can differ and therefore results can deviate
from reality. Especially the values of, for example, the radii of gyration can be determined from model
tests to confirm and reduce their uncertainty. Higher radii of gyration will reduce the base towing sta-
bility of a towed vessel since the destabilising components act on the vessel a larger arm. On the other
side, the wave drift yaw moment would also increase. As the stabilising effect of the waves mainly orig-
inates from the added wave resistance, the latter has a smaller impact. The general impact of the waves
on the towing stability remain valid. It is not hindered by the initial, larger, instability of the test vessel
and the added wave drift moment is small in comparison to the increased destabilising manoeuvring
moment.

• Manoeuvring forces:
As previously mentioned, uncertainties in the manoeuvring forces can have a large impact on the over-
all determination of the towing stability and thus on the impact of the waves on the towing stability.
The derivatives used in this thesis have been obtained from Hong et al.[37] and Nam et al. [51]. In the
first paper, the towing tank test that is performed is described in detail. Some extra coefficients are
introduced in the second paper, based on the same towing tank test. These coefficients are deemed
reliable but, like all model test results, include some uncertainties. A second point regarding the ma-
noeuvring forces is their valid yaw angle range. It is a known limitation of the derivatives that they tend
to overpredict the force components once the yaw angles become too large. In the reference papers,
the manoeuvring coefficients have been applied up to yaw angles ψ = ±60° but unfortunately no veri-
fication data was presented. Since these angles are larger than the yaw angles considered in this thesis
it is assumed the derivatives are valid. This remains however an uncertainty.

• Wave spectrum:
The wave spectrum used in the computation of the irregular wave results is a JONSWAP spectrum with
a peak enhancement factor γ = 3.3. This is a common spectrum for use in the North Sea area but for
when sailing in other locations, as you would expect for long distance towing operations, other spectra
should be considered. A change in spectrum will not change the general results significantly since the
sailing conditions physical T p range remains around the peak mean wave drift force. A spectrum with
more swell included, low frequency waves, will lead to a reduction of the stability, as these waves exert
less forces on the towed vessel.

7.4.3. Numerical uncertainties


At last, the numerical uncertainties are discussed:

• Panel mesh of the simulation models:


To calculate with the simulation model and obtain wave drift force components a 3D mesh model was
built. This panel model needs to be meshed and the mesh properties have a large influence on the
results. To reduce the uncertainty of the different mesh sizes and their distribution a mesh convergence
study has been performed, see appendix B. This mesh convergence study only focused on the mesh
size.
7.4. Uncertainties in the results 111

From the Delfrac manual instructions, a panel size of 2 m should be sufficient. From the convergence
study it showed this was not sufficient. The study shows that even for a panel size of ∼ 0.38 m the
near field method does not fully converge to the FF method. This is especially visible for the sway
and yaw components. Since there is a difference in the theoretical background of the two methods,
this is insurmountable. However, if the MSE error between the two methods is similar for every force
component, it can be argued the error due to the mesh is small. Extrapolating the results of the mesh
convergence study for this criteria, it would suggest a mesh panel size of ∼ 0.30 m would be sufficient.

Unfortunately, the number of panels used in the mesh model for Delfrac is limited. For motion and
direct near field and far field wave drift force computations the limitation of the number of panels lies
with the computation capacity of the system available. For the examined vessels, the panel size of
0.38 m is the smallest, still computable mesh. When including the pressure output per panel, there
is a different limitation. Because the numbering of the panels is limited to four digits, the maximum
number of panels that can be processed is 9999 panels. For Barge 1 and Barge 2, this results is the
smallest panel size of 0.75 m. Since the pressure output is necessary for the computation of the wave
drift damping, this is the panel size used in all the computations.

As is clear the difference between the preferred panel size (∼ 0.30 m) and applied panel size (0.75 m) is
relatively large. This difference is the cause of the deviation seen in the wave drift computation results.
The deviations are mostly present at high frequencies. The same deviation and uncertainty in the force,
is present in the wave drift damping. At frequencies larger than ω > 1.50 rad/s the wave drift damping
includes a significant deviation due to the panel size.

For the wave drift force components implemented in the stability analysis, this uncertainty is limited.
Both the surge and sway forces are based on the far field method and thus do not have these uncer-
tainties. The yaw moment is based on the near field method and includes the deviation. Because the
deviation is present in both the wave drift force and wave drift damping for the yaw moment, the effect
is doubled and the uncertainty larger. Again, the significant deviation is present at ω > 1.50 rad/s before
the encounter frequency effect is taken into account. This will shift to lower frequencies with (forward)
velocity.

This uncertainty can be taken away. The first solution for the uncertainties due to the panel distribution
is a non-homogeneous mesh. Around the vessel edges, the panel size can approximate the ∼ 0.3 m
while at the sides of the vessel larger panels sizes are adopted. In this way, the smaller panel size is
implemented while keeping the total panel number below 9999 panels. The second, more permanent
solution is an update of Delfrac. The limitation of 9999 panels is solely due to the panel numbering,
which should be a relatively easy fix. Looking into both solutions is recommended.

• Numerical schemes in Matlab:


In the computation and comparison process for the wave drift estimation methods the theory was
numerically implemented. With this numerical implementation, numerical errors can occur resulting
in uncertainties in the results. The numerical schemes, their implementation and possible errors are
shortly discussed.

– Encounter frequency implementation;


In the implementation of the encounter frequency, the conversion from ω to ωe included some
numerical rounding. The rounding errors are small but could be reduced by linearly interpolating
the data before the encounter frequency conversion. The effect is very small and this is not further
investigated.

– Equilibrium point determination;


To determine the equilibrium points with the effect of the waves included, the zero values of equa-
tion (6.20) need to be determined. Since the numerical implementation of this equation does not
exactly produce a zero value, this is done by a zero-crossing method. The original spacing be-
tween the data points in equation (6.20) is ψ = 5° and ω = 0.05 rad/s. In the zero-crossing method
the zero value is approximated by the data point following the detected zero-crossing. This can
lead to a maximal numerical rounding error of ψ = 5° and ω = 0.05 rad/s, which is to large. To
112 7. Discussion

reduce this error, the data points are linearly padded with data points, reducing the maximal pos-
sible rounding error to ψ = 0.5° and ω = 0.005 rad/s. Further improvement of locating the zero
points could be done by for example using the Newton-Raphson method, where the zero value is
determined using an iterative loop.

The uncertainty introduced by these errors is already discussed shortly in section 6.3 for the equi-
librium yaw angles ψ = −0.5° and ψ = 0°. The more broader uncertainty of the reliability of the
linearisation process is discussed in detail below.

– Numerical Jacobian routine;


For linearising the equations of motion the Jacobian at the equilibrium position is determined.
Since the inclusion of the wave drift force is done numerically, the Jacobian also needs to be de-
termined numerically. This is done by the jacobianest.m [11] routine, which inherently includes
some numerical errors. These errors are small, which is verified by the output of the jacobianest.m
routine. The corresponding uncertainties are therefore also small.

– Reliability Linearsisation;
Since assessing stability on a position which is not the equilibrium position will theoretically result
in unstable results, the uncertainty introduced by a numerical rounding error can be significant
for determining this stability. However, since the problem is approached numerically, it is possible
stable equilibrium positions exist around the theoretical equilibrium position.

If for example the area around an equilibrium position is not (highly) non-linear, the linearisation
at a position close to the equilibrium position will yield similar results as to that equilibrium po-
sition. In those cases, the reliability of the linearisation increases, as the impact of the numerical
error is less significant.

This is investigated for the regular waves stability contours. The stability contours for position 3
are used as example, but the same applies to other equilibrium positions. In figure 7.9 three sta-
bility contours are presented, each with a different equilibrium yaw angle ψeq . Although not pre-
sented in these figures, it was found that the stability contours are symmetrical around ψeq = 0°.
Sub-figure a shows the contour with ψeq = 0°, displaying the largest stable area. This area is re-
duced by an unstable area at short tow lengths in sub-figure b for ψeq = −0.5°. These same ar-
eas increase with an equilibrium yaw angle ψeq = −1°. Larger deviations result in fully unstable
conditions. From these stability contours it can be concluded that similar stability contours are
obtained up to a numerical deviation of ψeq = 1°. Given this result, small deviations have a small
impact on the stability contours. Still the most reliable approach for the linearisation of the equa-
tion of motion is to examine the area −1° ≤ ψeq ≤ 1° during the linearisation process but usable
results are still obtained without. For this case, this investigation is completed and the results of
the linearisation are trusted. For future work this is recommended when following this approach,
although the range that needs to be investigate will decrease with a more precise equilibrium po-
sition determination.

Stability contours position 3, a


= 2.5 m, eq
= 0° Stability contours position 3, a
= 2.5 m, eq
= - 0.5 ° Stability contours position 3, a
= 2.5 m, eq
= -1 °
1.6 1.6 1.6

1.4 1.4 1.4

1.2 1.2 1.2

1 1 1

0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4


0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Stability Contour Pos. 3: ψeq = 0° (b) Stability Contour Pos. 3: ψeq = −0.5° (c) Stability Contour Pos. 3: ψeq = −1°

Figure 7.9: Impact of the numerical error in the equilibrium position on the stability contours - Regular waves
7.4. Uncertainties in the results 113

The stability contours of the irregular wave results are also investigated. Again it is found that the
contours are symmetrical around ψeq = 0°. The stability contours with different yaw equilibrium
angles ψeq are presented in figure 7.10. Sub-figure a shows the result with ψeq = 0°. Comparing
this with the result in sub-figure b, with ψeq = −0.5° shows large differences. In sub-figure c a
deviation of 1° results in no stable parameter combinations, again a stark difference with the other
figures. This shows that the stability contour of the irregular waves is influenced heavily by the
numerical error in the equilibrium position. The area around the equilibrium position is highly
non-linear, and thus small changes in the equilibrium position will result in large changes in the
stability contour. The range which should be investigated is smaller than for the regular waves
−0.5° ≤ ψeq ≤ 0.5° but since the changes are larger, the yaw angle data point step size should be
reduced.

So, to increase the reliability of the irregular waves results the first step would be to reduce the
numerical data step in the equilibrium position determination. Since the area around the equi-
librium analysis shows to be highly non-linear, the linear data point padding previously applied
is not recommended. The second step would be to investigate the area around the equilibrium
position with the increased precision in the data points. This is a time consuming process but will
provide the most insight in how the numerical deviation in the equilibrium position determina-
tion affects the irregular waves stability contours. The final and third step is to verify the found
stability contour. As it entails the irregular wave results, the equilibrium positions are the mean
equilibrium positions and they are difficult to verify. An irregular wave train time series with the
same mean wave drift force characteristics as applied here could be used to verify the stability
contour using time domain simulation.

The irregular wave results remain less reliable than the regular stability contours. None the less,
the stability contours show stable parameter combinations. Since these numerical errors can
be physically interpreted as small variations of the equilibrium position, containing (part) of the
properties of that position, the results still provide a well enough indication of stability.

Stability contours position 3, H s


= 2.4 m, eq
= 0° Stability contours position 3, H s
= 2.4 m, eq
= -0.5 ° Stability contours position 3, H s
= 2.4 m, eq
= -1 °
16 16 16

15 15 15

14 14 14

13 13 13

12 12 12

11 11 11

10 10 10

9 9 9

8 8 8
Physical Tp Physical Tp Physical Tp
7 7 7
range range range
6 6 6

5 5 5

0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5

(a) Stability Contour Pos. 3: ψeq = 0° (b) Stability Contour Pos. 3: ψeq = −0.5° (c) Stability Contour Pos. 3: ψeq = −1°

Figure 7.10: Impact of the numerical error in the equilibrium position on the stability contours - Irregular waves

– Time step in the TD simulation;


The integration module used in the time domain simulation is the Matlab embedded routine
ode45.m. This routine uses an adaptive time step for integration and thus the effect of differ-
ent time steps is not investigated. Generally, if the time integration converges, the results with
different time step do not differ much and the uncertainty is limited.

The three categories of uncertainties have been discussed and their influence on the results identified. Most
of the uncertainties are considered to be of minor or negligible impact. There are however two uncertainties
that can have a potentially large impact on the results:

• Manoeuvring forces:
As mentioned before, the manoeuvring forces have a large impact on the towing stability of the vessel
(with and without waves). Large uncertainties within the manoeuvring derivatives can result in ex-
ponentially increasing errors later in the process. Since the source of the manoeuvring derivatives in
114 7. Discussion

this case is well documented, this is not an issue for this thesis. For future work it should definitely be
considered.

• Panel mesh of the simulation models:


The panel size and distribution of the mesh are critical for the obtaining usable results. The mesh
convergence study shows limitations of Delfrac and the accompanying uncertainty. For the towing
stability wave drift this is only significant for the yaw moment at larger frequencies and thus the present
results are not influenced significantly. A non-homogeneous panel distribution or an updated version
of Delfrac are two solutions that can address this issue.

• Reliability of the linearisation:


A crucial step in the stability analysis is the linearisation of the equation of motion at the equilibrium
positions. If the area around the equilibrium positions is highly non-linear, a small change in position
will have a large impact on the stability. For regular waves, the area around the equilibrium position
is relatively linear, resulting in similar stability contours for values around the equilibrium position.
For irregular waves, the area around the equilibrium position is highly non-linear, resulting in large
changes in the stability contours. Especially for the irregular waves stability contours, the stability of
the positions around to the equilibrium positions should be investigated.

7.5. Chapter summary


In the discussion chapter, the results of the stationary vessel wave drift assessment, sailing vessel wave drift
assessment and towing stability chapters are put into perspective. The first part that is addressed is the sta-
tionary tow resistance. It is shown that the impact of the waves is limited for the tow resistance of Barge 2.
However, for more definitive conclusions, the process of the thesis should be repeated with more arbitrary
vessel shapes.

The sailing tow resistance shows a similar result. However, the increase in wave drift force due to the (forward)
velocity is significant with respect to the wind and current contributions so that the wave drift contribution
remains between 20% and 30% of the tow resistance, irrespective of velocity.

The relative contribution of the wave drift in the context of the stability analysis is determined using time
domain simulation. Head wind is also implemented in the TD simulation and this shows that for Barge 2a,
the head waves have an equal or larger contribution to the towing stability than head wind. This is however
very dependent on the exact environmental conditions.

The specific towing stability results are also addressed. The influence of the waves on the fishtailing motion
is discussed; it reduces it and the damping properties of the waves are examined. The course unstable Barge
2a can be damped to a slightly under damped system due to the head waves.

At last, the uncertainties encountered or introduced in the thesis are discussed. Three main uncertainties are
identified: the reliability of the manoeuvring forces, the panel mesh as input for the wave drift estimation
computations and the reliability of the linearisation of the stability analysis.
8
Conclusions and Recommendations
The goal of the project was set as:

To identify and quantify the impact of wave drift on the tow resistance and towing stability of a towed
vessel

To reach this goal the project was split in three stages: Assessing the wave drift force for a stationary vessel,
using these results to assess the wave drift force for a sailing vessel and finally, using the sailing vessel results to
investigate their influence on the towing stability of a tow operation. In this chapter, the conclusions of these
three parts are presented in sections 8.1, 8.2 and 8.3. Recommendations for further research are presented in
section 8.4.

8.1. Conclusions stationary tow resistance


To identify the impact of the wave drift force components on the tow resistance of a stationary vessel, eleven
wave drift estimation methods are successfully compared on the basis of a universal comparison methodol-
ogy. The response of the obtained QTF’s are categorised in three zones to describe the envelope of the QTF:
the motion, transition and reflection zones. These zones reflect on the underlying physical source of the
wave drift force. For the irregular waves, selected sea states are adopted. For specifically the stationary tow
resistance, the following conclusions have been found:
• The most suitable wave drift estimation methods for assessing the tow resistance are in succeeding
order: the far field method or the near field method, the Faltinsen short wave method and the DNV
2014 formulation;
• The DNV formulation 2014 is a conservative estimate for the presently examined barges. However, it is
expected to be more representative for vessels with a length between 120 < L ≤ 200 m. For those vessels
the peak wave drift force can be expected in the physical peak period T p range of the waves;
• The increased insight into the wave drift estimation methods reduces the uncertainty in the prediction
of the wave drift contribution of the tow resistance;
• For the current barges, Barge 1 (a rectangular barge) and Barge 2 (a rectangular barge with raked bow
and aft), the influence of the wave drift on the tow resistance is limited;
• The framework for analysing the impact of the wave drift force components on an arbitrary vessel has
been successfully implemented for 2 test models.
From the stationary vessel wave drift estimation method comparison multiple supplementary conclusions
emerged. These conclusions hold for a rectangular barge like Barge 1 but may be applicable to other vessels.

• The Faltinsen short wave method is a good approximation for the wave drift force (both surge and sway)
in the transition and reflection zone with respect to the far field method;
• The strip theory method, and thus the R-function method, should not be used for a stationary vessel.
This is a confirmation of literature;
• For the reflection zone, the force curve of the stationary short wave methods of Fujii & Takahashi,
Kuroda et al. and Kashiwagi et al. are equally well described by the Faltinsen short wave method,
which is simpler to compute.

115
116 8. Conclusions and Recommendations

8.2. Conclusions sailing tow resistance


In the same way as for a stationary vessel, ten wave drift estimation methods for a sailing vessel are computed
and compared successfully. The same approach and methodology where adopted but this time including
surge, sway and yaw velocities. Specifically for the sailing tow resistance, these are the found conclusions:

• The most suitable methods for implementing the wave drift force with forward velocity for the tow
resistance are in succeeding order: the near field plus wave drift damping method, the far field plus
wave drift damping method and the Faltinsen short wave method;
• The increase of the wave drift force due to forward velocity is significant with respect to the wind and
current components;
• The larger the forward velocity, the higher the irregular waves peak period of the maximum wave drift
force (the encounter frequency effect);
• The mean wave drift force contributes roughly 25% of the tow resistance constantly, irrespective of
forward velocity (up to 3.5 m/s);
• Since the most common stable tow position is straight behind the towing tug with ψeq = 0°, it is rea-
sonable to base the tow resistance on the zero-yaw scenario. However, for course unstable vessels the
sway contribution must be taken into account.

From the computed wave drift estimation method comparisons, some more generic conclusions are drawn.
Again these are drawn for a rectangular barge like Barge 1, but they may hold for arbitrary other vessels.

• It is possible to compute the yaw wave drift damping terms successfully based on Aranha’s yaw term
approximation [3] and the second order pressure output of Delfrac [16].
• The Faltinsen short wave method originally related the force and velocity with the term 2ω
g . However,
in head waves a factor 4ω g provides a much better approximation of the surge force in the transition
and reflection zones with respect to the far field and near field method. Theoretical justification of this
factor 2 difference is still part of ongoing research.
• The short wave methods of Fujii & Takahashi, Kuroda et al. and Kashiwagi et al. do not capture the
linear increase with the frequency and their predictions include large offsets. They should be used with
caution;
• The R-function method (combined strip theory and updated Faltinsen short wave method) describes
the whole surge QTF accurately for velocities larger than 2 m/s;
• The yaw moment is best computed by the near field method since the wave drift damping yaw terms
are based on the pressure integration and thus match the near field method.

8.3. Conclusions towing stability


To identify the impact of the wave drift on the towing stability, the sailing vessel wave drift results are used as
input in the equation of motion. The computation of the equation of motion is successfully implemented and
the stability analyses for both conditions with and without waves are completed. All results are successfully
verified by time domain simulation. General conclusions on the impact of the waves on the towing stability
are:

• The main stabilising contribution of the wind or waves is the extra resistance and with that, an extra
tow line moment. The wind or wave drift moments can contribute directly to the stability but could
also add a destabilising contribution depending on their phase. Since they have a smaller magnitude
than the tow line moment the latter is governing;
• In all but very long waves, head waves will have a positive effect on the stability of a vessel. They will
reduce the fishtailing motion and in specific conditions will also lead to stable towing;
• The wave drift damping approach is suitable to implement the wave drift force components with ve-
locities in a tow operation equation of motion;
• The time domain simulation is suitable to quickly assess the results of various parameter dependencies
of the tow operation (for both without waves and with waves conditions);
• The stability of the equilibrium positions is really sensitive to the chosen position for linearisation of
the equation of motion. This is more pertinent for the irregular waves results. For increased reliability
a range of ±1° around the equilibrium yaw angles should be evaluated.
8.4. Recommendations 117

Conclusions regarding the specific course stable and unstable barges examined are also drawn. Barge 2b,
with skegs, is already course stable and the stability of the vessel is not significantly altered. For Barge 2a,
without skegs, the conclusions are more complex. Both regular and irregular waves have been modelled.
Regular waves to give insight in the theoretical process of stability and irregular waves to make the step to
reality. Most results have been obtained for regular waves:
• The barge benefits significantly from the presence of regular or irregular head waves;
• Regular head waves can damp the course unstable barge into a slightly underdamped system, letting it
converge to a stable equilibrium tow position;
• Relative to the wind contributions, regular waves have an equal or larger contribution to the towing
stability. This is highly dependent of the exact environmental conditions;
• The parameter combinations in regular head waves resulting in stable equilibrium positions ψeq 6= 0
are limited so that in practice only ψeq ∼ 0 is relevant.
• Regular head waves have no influence on the stability when ω < 0.65 rad/s;
• Regular head waves between 0.65 < ω ≤ 1.6 rad/s will in most cases lead to stable towing positions.
• Regular head waves with ζa < 0.8 m do not create extra equilibrium positions;
• Decreasing the tug velocity U t ug in regular head waves leads to extra equilibrium points ψeq 6= 0° and
will increase the equilibrium yaw angles ψeq of these extra equilibrium positions;
• The impact of the tug forward velocity U t ug on the stability of the equilibrium position straight behind
the tug ψeq ∼ 0° is small.
For the irregular waves, similar conclusions have been reached:
• Irregular head waves do not tend to create extra equilibrium positions, only ψeq ∼ 0° is relevant.
• For a significant wave height H s = 2.4 m, irregular head waves result in a mean stable tow condition for
all physically relevant wave periods. This is irrespective of tow line length.
Supplementary conclusions regarding the towing stability, not considering waves, are also found:

• Contrary to the recommendations found in literature, but confirming results found by Fitriadhy et al.
[24] longer tow lines do not necessarily increase the stability of the tow operation.
• For course unstable vessels, a towing bridle (an increase of x p0 ) increases the stability. Basically, the
arm of the tow line moment is enlarged and thus the vessel becomes more stable. This is practical
knowledge applied often in practice and it is successfully incorporated in the model.

8.4. Recommendations
The thesis focuses on two main subjects, wave drift estimation and towing stability. Therefore, the recom-
mendations are split in these two categories.

8.4.1. Wave drift estimation


For the wave drift estimation multiple recommendations are drafted. Some trace back to the literature study,
some emerge directly from the research done in the thesis. The first two recommendations are the most
relevant.
• Panel mesh improvement;
The panel mesh can still be an issue for vessels with sharp edges in their geometry. Since Delfrac has a
limit of 9999 panels if the second order pressure is needed as output this can cause some discrepancies
in the output. Overcoming of these discrepancies can be done by implementing a non-homogeneous
mesh or use an updated version of Delfrac once this issue has been solved. Both solutions are recom-
mended to pursue.
• Further wave drift damping validation;
The wave drift damping yaw terms are approximations and should therefore be compared to the di-
rectly computed terms based on the work of Finne [22] and could also be validated by model tank tests.
• Faltinsen short wave method;
The original Faltinsen short wave method [19] was derived with a factor 2ω/g relating the force to the
forward velocity. In previous research and this study, it was found that a factor 4ω/g provides a better
prediction of the surge force in head waves. Since theoretical support for this different factor is still
absent, more research into the derivation of the Faltinsen short wave method is advised.
118 8. Conclusions and Recommendations

• CFD and forward velocity potential theory methods;


Once the CFD and forward velocity potential theory methods are more common in the public domain,
these should be included in the wave drift estimation method comparison; These methods have the
potential to significantly impact the field of wave drift estimation but at this moment still require many
resources and the results are not properly validated.
• Build a vessel shape database for the wave drift contributions to the tow resistance;
To improve the reliability of the tow resistance conclusions it is recommended to go through the pro-
cess outlined by the thesis for numerous arbitrary vessel geometries and build a database for the most
common vessel shapes.
• Investigate the wave drift for more than one stationary and sailing conditions sea state;
The sea states that have been investigated are the ones defined by DNV [14] but for a more broader
picture and better insight, multiple sea states should be investigated.
• Improving the stationary tow resistance prediction;
Extreme wave drift force loads in irregular waves can dominate the tow resistance of a stationary vessel.
These loads could be investigated to increase insight. These loads should be combined with looking
into the possibility of a DP-like system for the survival scenario of the tow operation.

8.4.2. Towing stability analysis


The towing stability recommendations mainly originate from questions that emerged during the research or
assumptions that should be further investigated. The first three recommendations are the most relevant for
further research.

• Numerical error investigation;


During the modelling, several numerical errors were introduced. These errors are discussed in chapter
7. Notably the errors for the equilibrium position determination and its effects on the linearisation
should be investigated further. A small change in yaw angle for an equilibrium position can have a large
impact on the stability of that equilibrium position. Although the effect of this error is investigated,
further research into the impact of these errors will yield more insight in the overall stability properties
of the tow operation.
• Minimum required wave drift contribution (added resistance);
Routh’s stability criterion allows for the determination of the minimum required added resistance (due
to waves) necessary to obtain a stable tow operation. Not just the waves can produce the extra neces-
sary resistance required for stability and therefore this can be approached more globally. Determining
this minimum required added resistance can be a lengthy mathematical derivation but can provide
useful for quicker stability checks. This is therefore research that should be pursued.
• Investigate the slowly varying components of the WDF and their influence on the relative motion be-
tween the towed vessel and tug;
The slowly varying component of the wave drift force may interact with the tow line characteristics re-
sulting in a low frequency relative motion between the tug and towed vessel. These dynamics can have
an influence on the rest of the system and could possibly destabilise the towing operation. Therefore,
these interactions should be investigated.
• Wind force components;
The wind force components should be fully incorporated in the towing stability analysis and the equa-
tion of motion. See Sinibaldi [66] and Fitriadhy [24] for more details on the approach and implementa-
tion of such forces in a stability analysis.
• Different wave and wind angle of attack;
In this thesis, only head waves are examined. Logically, in practice this is not the only condition that
is encountered. The same holds for the wind contribution. It is therefore advised to investigate the
influence on the stability for different wave angle of attacks and to look into different angle of attack
between waves and wind. For the latter the paper by Shigunov et al. [64] could be used as a starting
point.
• Modelling of the towing tug;
In this thesis, the towing tug has been simulated using the assumption of a moving point which move
forward with a fixed velocity. One of the consequences is that it is not possible to differentiate be-
tween an unstable towed vessel that is fully unstable or a dynamically stable. Both will show the same
characteristics in the time domain simulation. This is solved by including the towing tug motions in
8.4. Recommendations 119

the equation of motion. With this addition it is also possible to model the mean tow resistance of an
unstable vessel, since the mean towing force can be extracted from this improved model.
• Manoeuvring force component influence;
Investigate the influence of the manoeuvring force components on the towing stability. What kind of
effect does a small change to the manoeuvring derivatives of a course unstable vessel have? This can
give more insight in how important the addition of waves in the towing stability is.
• Validation of the stability analysis and time domain simulations with model tank tests;
The manoeuvring data has been obtained with towing tank tests by Hong et al. [37], and the simplified
(linear) model has been used in tank tests with waves. However, a systematic approach to testing the
towing stability in waves has not yet been conducted. The methodology for the towing stability can be
used as a basis of systematically testing the parameter influence in a towing tank. These tests can then
be used to validate the stability analysis and TD simulations.
• Include the neglected effects in the modelling. More detailed information on the neglected effects can
be found in the uncertainty section 7.4, but they include:

– Propeller race and wave breaking loads; For short tow lines propeller race or the interaction be-
tween tug and towed vessel can become important. The same goes for blunt vessels in steep
waves, resulting in wave breaking loads. These effects could be added to the current set-up.
– Tow line should be modelled as a catenary;
The tow line is modelled as a taut tow line and the effect of a catenary tow line should be incorpo-
rated. See Nakayama et al. [50] for more information.
– Center of gravity shift; Further investigate the influence of a shift in the center of gravity of the
towed vessel. This can include both a lateral and vertical shift.
A
Background Case Studies
In this appendix, the case studies performed for the relative importance study in chapter 1 are presented. The
case study for the Egina FPSO is presented in section A.1 and the SMIT barge 2 case study in section A.2.

A.1. Case study: Egina FPSO


To investigate the relative influence of every component of the tow resistance, a test case is examined. The test
case is the tow operation of the Egina FPSO which was towed from South Korea to offshore Nigeria. The model
test data [57] for this tow operation has been used to validate the calculations based on literature. Note that no
wind forces were present in the model testing and their results are based on the wind coefficients determined
in the wind tunnel test [58]. The coefficients are fitted using cubic splines. From the tests E102 to E103 the
current forces are deduced. The differences between E105 and EW301, and E102 and EW101 are considered
as the wave drift forces. This is done while there is a significant difference between the average yaw angles
ψav g of the test data but this assumption is also used by the authors of the test data. The comparisons done
for ψ = −10°.

The model data is compared to the force calculations from literature presented in equations (2.25) through
(2.27) and equation (2.58). The comparison is done for two different sea-states which are summarised in table
A.1. The first sea-state is of interest for zero and low forward speeds. This, more extreme, sea-state is more
typical for a holding condition situation. The second sea-state is of interest while sailing. These conditions
are "typical" for sailing conditions. In both situations, the current velocity is assumed zero since this can be
simulated by adding or subtracting the velocity from the forward speed of the vessel.

Table A.1: Sea state parameters

Sea state #1 #2
Uc [ /s ]
m
0.0 0.0
U w [m/s ] 20 10
H s [m] 5.0 2.4
T [s] 10.5 8.2

The results of the comparison can be found in figure A.1. Note that the axis scale differ between sub-figures b
and d and that only one data set per sea state is available. In sub-figure a of figure A.1, the sailing condition, it
can be seen that with a forward speed of Ur el = 3 m/s the ratio between the current data and current literature
calculation is ∼ 1.5. Between the wave drift forces the ratio is ∼ 3.5. The wind forces do not deviate because
both model test data and literature calculations are based on the same wind force coefficients. Sub-figures a
and b are only of importance up to a forward speed (or relative current) of ∼ 5 m/s. The ratios between the
model data and the literature calculations at Ur el = 1 m/s are, for respectively the current and the wave drift
force, ∼ 1.5 and ∼ 2.0.

Looking at sub-figure b and d, it is very clear that both the wind and the wave drift forces cannot be neglected

121
122 A. Background Case Studies

at any forward velocity. It can be argued that the current force can be neglected at very low forward velocities
Ur el < 0.5 m/s. The wave drift forces are the dominant hydrodynamic force at low forward velocities and
improving their prediction contributes to the total accuracy of the prediction. Taken into account that the
wave drift force estimation method used can sometimes under-predict the wave drift force this argument is
enhanced.

Egina FPSO: Relative Forces, = -10°, Sea State #1


10-3 Egina FPSO: Forces, = -10°, Sea State #1 100
7
rel. Current Force
Total Force 90 rel. Wind Force
Current Force
6 rel. Wave Drift Force
Wind Force 80 rel. Data Current Force

Percentage of Total Force [%]


Wave Drift Force
rel. Data Wind Force
Data Current Force
5 70 rel. Data Avg Wave Drift Force
Data Wind Force
Data Avg Wave Drift Force
60
4
50

3 40

30
2
20
1
10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(a) Tow Resistance, ψ = −10° Sea State #1 (b) Tow Resistance percentage, ψ = −10° Sea State #1

Egina FPSO: Relative Forces, = -10°, Sea State #2


Egina FPSO: Forces, = -10°, Sea State #2 100
10-3
5 rel. Current Force
Total Force 90 rel. Wind Force
4.5 Current Force rel. Wave Drift Force
Wind Force 80 rel. Data Current Force
Percentage of Total Force [%]

4 Wave Drift Force rel. Data Wind Force


Data Current Force 70 rel. Data Avg Wave Drift Force
3.5 Data Wind Force
Data Avg Wave Drift Force 60
3
50
2.5
40
2

30
1.5

1 20

0.5 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(c) Tow Resistance, ψ = −10° Sea State #2 (d) Tow Resistance percentage, ψ = −10° Sea State #2

Figure A.1: Overview Tow Resistance and percentage for the Egina FPSO model test data and literature calculations

In literature, see Hong [37], and in the tow stability test for the Egina FPSO it is shown that due to the presence
of waves the yaw angle greatly reduces in comparison with the still water angle. In the tow stability tests of
the Egina case, it was shown that the yaw angle in still water had a range between ψ = −6° ∼ 16° and in waves
a range of ψ = −4° ∼ 2°. This is a significant reduction.

Although angles above ψ = ±30° are not expected the forces up to ψ = ±45° are investigated. An overview of
the forces relative to the yaw angle ψ is shown in figure A.2.
A.2. Case study: SMIT Barge 2 123

Egina FPSO: Angle Dependency of the Forces, Sea State #1


10-3 The Wind Force 10-3 The Current Force
5 9
Urel = 0.5 m/s Urel = 0.5 m/s
4.5 8
Urel = 1 m/s Urel = 1 m/s
Urel = 1.5 m/s 7 Urel = 1.5 m/s
4
Urel = 2 m/s Urel = 2 m/s
6
3.5 Urel = 2.5 m/s Urel = 2.5 m/s
Urel = 3 m/s 5 Urel = 3 m/s
3
Urel = 3.5 m/s 4 Urel = 3.5 m/s
2.5 Urel = 4 m/s Urel = 4 m/s
3
2
2

1.5 1

1 0
-40 -30 -20 -10 0 10 20 30 40 -40 -30 -20 -10 0 10 20 30 40
Yaw Angle [°] Yaw Angle [°]

10-3 The Wave Drift Force The Total Force


1.7 0.015
Urel = 0.5 m/s Urel = 0.5 m/s
1.6
Urel = 1 m/s Urel = 1 m/s
1.5 Urel = 1.5 m/s Urel = 1.5 m/s
Urel = 2 m/s Urel = 2 m/s
1.4 0.01
Urel = 2.5 m/s Urel = 2.5 m/s
1.3 Urel = 3 m/s Urel = 3 m/s

1.2 Urel = 3.5 m/s Urel = 3.5 m/s


Urel = 4 m/s Urel = 4 m/s
1.1 0.005

0.9

0.8 0
-40 -30 -20 -10 0 10 20 30 40 -40 -30 -20 -10 0 10 20 30 40
Yaw Angle [°] Yaw Angle [°]

Figure A.2: Overview force components from the Egina FPSO literature calculations for different ψ

The focus is on the results of the first case since this case is more relevant for higher forward speeds. From
sub-figure a of figure A.2 it is noted that the wind force increases almost linearly with the increasing yaw
angle ψ. The difference between the force for negative and positive angles results from an asymmetrical
above water body of the Egina FPSO.

In sub-figure b, it is seen that the current force is, as expected, increasing the forward speed and yaw angle
ψ. As is expected the current force increases as a quadratic function. Especially at higher forward speeds the
total force function shape fully resembles the current force shape.

The wave drift forces in sub-figure c show that the force stays roughly the same for yaw angles up to ψ =∼ 15°.
This can be explained as the increase in exposed area to waves is comparable to the decrease in the reflection
of the waves against the hull. For small angles these effects cancel each other out.

As already mentioned, the model results showed that with waves present the yaw angle greatly reduces. In
head waves this greatly reduces the yaw angle and since waves can be assumed always present offshore no
big yaw angles in head waves are expected. This is however not quantified and is worth investigating.

A.2. Case study: SMIT Barge 2


To further investigate the influence another test case is examined. Unfortunately no model test data is avail-
able, but the literature calculation methods can be showcased in a different light. To compare, the calculation
method used by Boskalis during the Baltic 2 project is evaluated. The test case is the tow operation of the
SMIT Barge 2 with a jacket installed on top. Unfortunately no test data is available so only a comparison to
the, at the time calculated values, is possible. The calculation is done for the same two sea-states as in the
Egina FPSO case study. The comparison is done for ψ = 0°. The wind and current coefficients are provided by
Boskalis and are fitted using the cubic spline method, see figure A.3.

The results can be found in figure A.4. From examining sub-figures b and d at Ur el = 1 m/s, it is clear that
the current and wind calculations are similar although the literature calculation for the current is ∼ 1.5 times
larger. The big difference between the calculations is due to the wave drift force. From sub-figures a and c it
is clear that the increase in wave drift forces according to the literature methods dominates the total towing
force. For both sea states the wave drift force at Ur el = 4 m/s is according to the literature method ∼ 4 times
as large. This is a very significant difference and worth investigating.
124 A. Background Case Studies

SMIT Barge 2: Cubic Spline fitted Wind Coefficients SMIT Barge 2: Cubic Spline fitted Current Coefficients
0 1.5 -0.34 0.4
Cubic Spline fitted Surge coefficient Cubic Spline fitted Surge coefficient
Surge coefficient data Surge coefficient data
-0.36 0.3
-0.2 Cubic Spline fitted Sway coefficient 1 Cubic Spline fitted Sway coefficient
Sway coefficient data Sway coefficient data
0.2
-0.38
Surge Coefficient [-]

-0.4 0.5

Sway Coefficient [-]

Surge Coefficient [-]

Sway Coefficient [-]


0.1
-0.4
-0.6 0 0
-0.42
-0.1
-0.8 -0.5
-0.44
-0.2

-1 -1
-0.46 -0.3

-1.2 -1.5 -0.48 -0.4


-80 -60 -40 -20 0 20 40 60 80 -40 -30 -20 -10 0 10 20 30 40
Yaw angle [°] Yaw angle [°]

(a) Wind coefficients (b) Current coefficients

Figure A.3: Fitted SMIT Barge 2 coefficients using cubic splines

SMIT Barge 2: Relative Forces, = 0°, Sea State #1


SMIT Barge 2: Forces, = 0°, Sea State #1 100
10-3
2.5 rel. Current Force
Total Force 90 rel. Wind Force
Current Force rel. Wave Drift Force
Wind Force 80 Boskalis rel. Current Force
Percentage of Total Force [%]

2 Wave Drift Force Boskalis rel. Wind Force


Boskalis Total Force 70 Boskalis rel. Wave Drift Force
Boskalis Current Force
Boskalis Wind Force 60
1.5 Boskalis Avg Wave Drift Force
50

40
1

30

0.5 20

10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(a) Tow Resistance Boskalis and Literature, ψ = 0° Sea State #1 (b) Tow Resistance percentage Boskalis and Literature, ψ = 0°
Sea State #1

SMIT Barge 2: Relative Forces, = 0°, Sea State #2


SMIT Barge 2: Forces, = 0°, Sea State #2 100
10-3
1 rel. Current Force
Total Force 90 rel. Wind Force
0.9 Current Force rel. Wave Drift Force
Wind Force 80 Boskalis rel. Current Force
Percentage of Total Force [%]

0.8 Wave Drift Force Boskalis rel. Wind Force


Boskalis Total Force 70 Boskalis rel. Wave Drift Force
0.7 Boskalis Current Force
Boskalis Wind Force 60
0.6 Boskalis Avg Wave Drift Force
50
0.5
40
0.4

30
0.3

0.2 20

0.1 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4

(c) Tow Resistance Boskalis and Literature, ψ = 0° Sea State #2 (d) Tow Resistance percentage Boskalis and Literature, ψ = 0°
Sea State #2

Figure A.4: Overview Tow Resistance and percentage of the Smit Barge 2, Boskalis and literature calculations
B
Vessel Characteristics
In this appendix, the simulation models used in the thesis are further elucidated. The computation of some
geometry values and the input files for the simulations are presented first. The mesh convergence study and
center of gravity impact studies are presented after.

B.1. Vessel geometry input


For the vessel geometry characteristics, the radii of gyration are needed. Besides those, there are some pre-
requisites for the 3D method simulations. For the approximation methods, integration along the waterline
is necessary so the input for the waterline integration module and a short description on this module is pre-
sented.

Radii of gyration
To be able to accurately describe the behaviour of the vessel the radii of gyration are necessary. If the radii
are not available the approximation in table B.1 can be used. For more conservative results, a larger radius is
recommended. This implies that the mass is distributed further away from the center of gravity, increasing
the moments related to the rotations. For the forces, as well as for the stability analysis, this will yield higher
forces and moments and thus the findings will be more conservative.

Table B.1: Approximation of radii of gyration ship shaped vessel [40]

Radii Range [m]


k xx 0.3 B - 0.4 B
ky y 0.22 L - 0.28 L
k zz 0.22 L - 0.28 L

3D models
The 3D model, used for the Delfrac calculations, needs the following data as input for its simulations:

• (x, y, z)-coordinates of the mesh panels;


• (x, y, z)-coordinates of the center of gravity;
• (x, y)-coordinates of waterline sections;
• Radii of gyration;

The first item can be exported from the FEMAP 3D modeling program, while the CoG and radii have to be de-
fined manually. The waterline section coordinates need to be extracted from the mesh panel data manually.

B.2. Delfrac mesh convergence study


The diffraction calculations are done with the Delfrac program developed at the Delft University of Technol-
ogy [16]. Based on experience and the Delfrac manual the panel size should to be chosen at the least 1/8 of

125
126 B. Vessel Characteristics

the shortest wave length. Since three second wave periods are roughly the shortest waves that are of interest
a wavelength of ∼ 16 meter in deep water is evaluated and panels size of roughly 2 meter should be sufficient.
To check this, the mesh convergence on the basis of the mean square error metric is calculated for non di-
mensional panel sizes. That means the number of panels per ship length. This is done for panel sizes of: 2m,
1m, 0.75m, 0.50m and 0.38m.

Not just the convergence of the ship motions needs to be checked but especially the convergence of the surge
and sway wave drift forces need to be checked. These forces tend to converge slower at higher frequencies
so smaller panels are needed to capture the correct force at those frequencies. The results for Barge 1 are
presented in figure B.1. In sub-figures a and b, the sway force was identified as the governing force to check
convergence. If the convergence is not sufficient, the near field method sway force contains a linear frequency
dependency at high frequencies. This phenomenon should be avoided and thus a panel size with minimal
dependency must be chosen.

Surge Force: NF to FF method convergence Sway Force: NF to FF method convergence


0 0.25
Length of panels: ~0.38 m, FF Method Length of panels: ~0.38 m, FF Method
-0.05 Length of panels: ~0.38 m, NF Method Length of panels: ~0.38 m, NF Method
Length of panels: ~0.5 m, NF Method Length of panels: ~0.5 m, NF Method
Length of panels: ~0.75 m, NF Method 0.2 Length of panels: ~0.75 m, NF Method
-0.1 Length of panels: ~1 m, NF Method Length of panels: ~1 m, NF Method
Length of panels: ~2 m, NF Method Length of panels: ~2 m, NF Method
-0.15
0.15

-0.2

0.1
-0.25

-0.3
0.05
-0.35

-0.4 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) Surge force QTF (b) Sway force QTF

MSE Convergence: NF to FF method MSE Convergence: NF to FF method


600 600
Surge Force Surge Force
Sway force Sway force
500 500

400 400

300 300

200 200

100 100

0 0
0 10 20 30 40 50 60 70 80 0 50 100 150 200 250

(c) Mesh convergence in respect to the breadth (d) Mesh convergence in respect to the length

Figure B.1: The wave drift force mesh convergence of Barge 1 from the Delfrac diffraction analysis with φ = 15 °.

The results of the MSE convergence for respective panels per breadth B = 24.4m, and length L = 76.2m, can
be found in sub-figures c and d. From the data presented, a panel size of ∼ 0.5m, so 48 panel s/B and 153
panel s/L, is deemed sufficient; there still is a slight increase at higher frequencies, ω > 1.8[r ad /s] but a
smaller panel size, 0.38m, would increase the computation time and yield a relatively low increase in accu-
racy. Note that a deviation between the near field and far field method still remains.

Based on sub-figures c and d it can be argued that once the MSE errors for the surge and the sway forces are
B.3. Center of gravity impact study 127

the same, the effect of the panel size is mitigated and only the theoretical difference remains. This would
imply a panel size of ∼ 0.30 m or 75 panel s/B and 250 panel s/L would be sufficient. Unfortunately, this
panel size is not computable with the available computation resources.

In the case that second order pressures are required as output of Delfrac, the number of panels in the 3D
model is limited to 9999 panels. This limitation origins from the 4 digit numbering present in the second
order pressure output. If the panel number transcend 9999 panels, the numbering stops and the data can
not be processed. For this vessel, this means a panel length of 0.75 m is the smallest possible. Due to this
limitation, the chosen panel size becomes 0.75 or 32 panel s/B and 103 panel s/L. This panel size is applied
to both Barge 1 and Barge 2.

B.3. Center of gravity impact study


The center of gravity can have an impact on the dynamic behaviour of the vessel. During towing the longitudi-
nal and lateral location of the center of gravity have the biggest impact on this behaviour. As a simplification,
their location is taken as the origin of the local vessel coordinates to limit their impact.

The vertical location of the center of gravity above the keel ZC oG can not as easily be taken in the origin
and thus has an impact on the (dynamic) behaviour of the vessel. It influences the pitch and roll natural
frequencies in particular, which in turn dominate the motion induced region of the wave drift force. Since in
the sea states examined, the vessel is not exempt of wave drift forces in the motion induced region this needs
to be taken into account.

The ZC oG directly impacts the G M of the vessel, both lateral and transverse, and the meta-centric height in
turn determines the natural frequency. Since an unstable vessel is dangerous, the allowed range of ZC oG is
determined with equation (B.1) [40] and is shown in table B.2. The natural frequencies are also determined
using equation (B.2).

G M T = K B + B MT − K G G M L = K B + B ML − K G (B.1)
s v
ρg ∇ · G M T u ρg ∇ · G M L
u
ωnφ = ωnθ = t (B.2)
I xx + a 44 I y y + a 55

The relative motion between the vessel and the waves has significant influence on the peak of the QTF and
thus the heave, pitch and roll motions are studied for the range of ZC oG . Multiple near field method simula-
tions are run. During simulation the near field method is used as the peak values of the QTF are more distinct
with respect to the far field method. The motions for angles ψ = 0° and ψ = 15° are plotted respectfully in
figures B.2 and B.3. To be able to monitor the impact of the (relative) motions on the QTF of the vessel these
are presented in figure B.4.

From sub-figures a and d from both figure B.2 and figure B.3 it is clear that the heave does not vary with
increasing ZC oG and can thus be neglected in this issue. As expected for the roll, sub-figure B.2b shows that
for head waves the roll motion is almost equal to zero. This is also confirmed by the sway and yaw QTFs in
sub-figure B.4c and e. This leaves the pitch motion as significant influence of the surge QTF peak in head
waves. Comparing sub-figure B.2c with B.4a one can find that the peak frequencies roughly match. Peak
frequencies of the surge QTF lie 0.5 − 0.10 rad/s higher than the peak frequency of the pitch motion. This

Table B.2: ZC oG range for Barge 1

ZC oG [m] G M T [m] G M L [m] ωnφ [r ad /s] ωnθ [r ad /s]


2.8 12.16 130.23 1.033 0.867
3.8 11.16 129.23 1.005 0.865
6.3 8.66 126.73 0.906 0.861
8.8 6.16 124.23 0.763 0.856
13.8 1.16 119.23 0.309 0.845
128 B. Vessel Characteristics

Heave RAO, Wave Direction w


= 0° Roll RAO, Wave Direction = 0° Pitch RAO, Wave Direction = 0°
w w
3
1 Z CoG = -1 m Z CoG = -1 m Z CoG = -1 m
1
Z CoG = 0 m Z =0m Z =0m
CoG CoG
Z CoG = 2.5 m Z CoG = 2.5 m 2.5 Z CoG = 2.5 m
0.8 Z CoG = 5 m 0.8 Z CoG = 5 m Z CoG = 5 m
Z CoG = 10 m Z CoG = 10 m 2 Z CoG = 10 m

0.6 0.6

1.5

0.4 0.4
1

0.2 0.2
0.5

0 0
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) Heave RAO, ψ = 0° (b) Roll RAO, ψ = 0° (c) Pitch RAO, ψ = 0°

Heave Phase , Wave Direction w


= 0° Roll Phase , Wave Direction w
= 0° Pitch Phase , Wave Direction w
= 0°
350 350 350
Z = -1 m Z = -1 m Z = -1 m
CoG CoG CoG
Z CoG = 0 m Z CoG = 0 m Z CoG = 0 m
300 300 300
Z CoG = 2.5 m Z CoG = 2.5 m Z CoG = 2.5 m
Z =5m Z =5m Z =5m
CoG CoG CoG
250 250 250
Z CoG = 10 m Z CoG = 10 m Z CoG = 10 m

200 200 200

150 150 150

100 100 100

50 50 50

0 0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(d) Heave Phase, ψ = 0° (e) Roll Phase, ψ = 0° (f) Pitch Phase, ψ = 0°

Figure B.2: ZC oG variable: Heave, Pitch and Roll Motion RAO and Phase diagrams for ψ = 0°

Heave RAO, Wave Direction w


= 15° Roll RAO, Wave Direction = 15° Pitch RAO, Wave Direction = 15°
w w

2.5 3
1 Z CoG = -1 m Z CoG = -1 m Z CoG = -1 m
Z CoG = 0 m Z CoG = 0 m Z CoG = 0 m
Z CoG = 2.5 m Z CoG = 2.5 m 2.5 Z CoG = 2.5 m
2
0.8 Z CoG = 5 m Z CoG = 5 m Z CoG = 5 m
Z CoG = 10 m Z CoG = 10 m 2 Z CoG = 10 m

0.6 1.5

1.5

0.4 1
1

0.2
0.5
0.5

0
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) Heave RAO, ψ = 15° (b) Roll RAO, ψ = 15° (c) Pitch RAO, ψ = 15°

Heave Phase , Wave Direction = 15° Roll Phase , Wave Direction w


= 15° Pitch Phase , Wave Direction = 15°
w w
350 350 350
Z CoG = -1 m Z CoG = -1 m Z CoG = -1 m
Z CoG = 0 m Z CoG = 0 m Z CoG = 0 m
300 300 300
Z = 2.5 m Z CoG = 2.5 m Z = 2.5 m
CoG CoG
Z CoG = 5 m Z CoG = 5 m Z CoG = 5 m
250 250 250
Z CoG = 10 m Z CoG = 10 m Z CoG = 10 m

200 200 200

150 150 150

100 100 100

50 50 50

0 0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(d) Heave Phase, ψ = 15° (e) Roll Phase, ψ = 15° (f) Pitch Phase, ψ = 15°

Figure B.3: ZC oG variable: Heave, Pitch and Roll Motion RAO and Phase diagrams for ψ = 15°
B.3. Center of gravity impact study 129

Surge QTF, Wave Direction w


= 0° Surge QTF, Wave Direction w
= 15°
0.05 0.05
Z CoG = -1 m Z CoG = -1 m
0 0
Z CoG = 0 m Z CoG = 0 m
-0.05 Z CoG = 2.5 m -0.05 Z CoG = 2.5 m
Z CoG = 5 m Z CoG = 5 m
-0.1 -0.1
Z CoG = 10 m Z CoG = 10 m
-0.15 -0.15

-0.2 -0.2

-0.25 -0.25

-0.3 -0.3

-0.35 -0.35

-0.4 -0.4

-0.45 -0.45
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(a) Surge QTF, ψ = 0° (b) Surge QTF, ψ = 15°

Sway QTF, Wave Direction w


= 0° Sway QTF, Wave Direction w
= 15°
0.15 0.15
Z CoG = -1 m Z CoG = -1 m
Z CoG = 0 m Z CoG = 0 m
Z CoG = 2.5 m Z CoG = 2.5 m
Z CoG = 5 m Z CoG = 5 m
0.1 Z CoG = 10 m
0.1 Z CoG = 10 m

0.05 0.05

0 0

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(c) Sway QTF, ψ = 0° (d) Sway QTF, ψ = 15°

Yaw QTF, Wave Direction w


= 0° Yaw QTF, Wave Direction w
= 15°
0.03 0.03
Z CoG = -1 m Z CoG = -1 m
0.025 Z CoG = 0 m 0.025 Z CoG = 0 m
Z = 2.5 m Z = 2.5 m
CoG CoG
0.02 Z CoG = 5 m 0.02 Z CoG = 5 m
Z CoG = 10 m Z CoG = 10 m
0.015 0.015

0.01 0.01

0.005 0.005

0 0

-0.005 -0.005

-0.01 -0.01
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(e) Yaw QTF, ψ = 0° (f) Yaw QTF, ψ = 15°

Figure B.4: ZC oG variable: Surge, Sway and Yaw QTF for ψ = 0°, 15°
130 B. Vessel Characteristics

difference can be explained by the small phase difference ², see sub-figure f, that the pitch motion has in
respect to the waves; the relative motion is what drives the QTF.

For the surge QTF, the ψ = 15° figures show similar results. The peaks of the pitch motion and surge QTF lie
a bit closer together but still there is a small phase difference present. This is not as clearly visible for the
sway QTF in sub-figures B.4d. Here, the peak lies around ω ∼ 1.15 rad/s and although the peak shifts to lower
frequencies with higher ZC oG , the difference is small. This is remarkable since the natural frequencies shift
significantly for the same change in center of gravity. Also from the phase, sub-figure B.3e, it does not give a
clear indication on why the peaks remain around ω ∼ 1.15 rad/s. That the impact of the roll motion is lower
than the pitch motion is to be expected since a small pitch angle will results in much higher relative motions
than a small roll angle.

Therefore even more interesting is the influences of the roll motion on the QTFs at higher frequencies. It is
most visible in the sub-figures B.4d and f for sway and yaw but can be slightly seen for the surge QTf in sub-
figure b. At the higher frequencies, the oscillating behaviour of the roll motion becomes more significant in
comparison to the pitch motion and is visible in the QTFs. The roll QTF shows the same oscillation as the roll
motion while the yaw QTF has a shifted oscillation with respect to the frequency of roughly ω ∼ 0.05 rad/s and
the amplitude of the oscillation is flipped. This oscillating at higher frequencies is most likely influenced by
the breadth of the vessel; a slender vessel will have larger oscillations than a broader vessel. Since the viscous
effects are neglected in this potential theory calculation, the viscous roll damping is neglected and thus these
oscillations might be fully damped if these effects are taken into account.

The main yaw QTF peak, see sub-figure B.3f, lies in the same range as the surge QTF peak and is thus mainly
influenced by the pitch motion. As mentioned, the oscillating behaviour at higher frequencies is caused by
the roll motion. The small peak at ω ∼ 0.75 rad/s is caused by the roll natural frequency but the trough at
lower frequencies is not explained by this motion since it is present in every curve. With higher ZC oG this
trough increases but this seems counter intuitive since at these frequencies the moment is acting against the
direction of the waves. This is most likely caused by the phase of relative motion but since this data is not
present here, a solid conclusion for this phenomenon has not been set.

For the purpose of using one single ZC oG for the following simulations a value must be chosen. The maximum
value of ZC oG = 10 m is the most conservative value and is therefore chosen. An overview of the coordinates
of the center of gravity is given in table B.3. The coordinates are defined from the starboard aft of the vessel.

Table B.3: Coordinates of the center of gravity (CoG) defined from the starboard aft of the vessel

Distance [m]
x 38.1
y 12.2
z 10
Implementation and Verification of the
C
Computations
In this appendix, the implementation of the wave drift force estimation methods and the stability analysis
is presented. Besides the implementation of the wave drift estimation methods, there computation is also
checked and verified. First the stationary case wave drift forces implementation and verification are pre-
sented. Secondly, the sailing case wave drift force implementation and verification is discussed. Lastly, the
stability analysis implementation is presented.

The Matlab m-files used for this thesis project are accompanied on a separate thumb drive. Not all are elu-
cidated in this chapter, only the main body routines used for computation of the wave drift and the stability
analysis.

C.1. Wave drift for a stationary vessel implementation and verification


All estimation methods are plotted with the the base value of the far field method, to give a simple indication
of the relations between the methods and if they yield similar results. For the regular waves, the boundaries
for the motion dominated, transition and reflection dominated zones are plotted using vertical lines. The
surge component is examined in head waves while the sway and yaw components are examined with a 15°
wave angle of attack.

The stationary vessel methods are computed from the main m-file Stationary_Comparison.m. This file sole
calls the functions and plots the results and is therefore not further elucidated.

Far Field and Near Field Method


These two methods are both computed using Delfrac. This program is validated extensively with model tests
and its results can be trusted if they are examined in correct zones. The reflection zone for the near field
method can pose a problem for the computation of the sway and yaw components. The panel size needs to
be sufficiently small to catch these components in short waves, otherwise a linear frequency dependency is
introduced in the force. See also the mesh convergence study in appendix B. The output of Delfrac is read
into Matlab using the m-files provided in the Delfrac_Readfiles folder. This folder was supplied by the Delft
University of Technology. The wave drift forces are loaded in specifically with m-file QTF_Load_in_Delfrac.m.
No operations are performed on the output of Delfrac for the stationary vessel wave drift estimation compu-
tations. The computations of the NF and FF method are plotted in sub-figures C.3a and d. Since the results
look reliable, and Delfrac is well validated, these implementations are considered verified.

Strip Theory Method


The strip theory estimation methods are computed using the ShipMo module in Safetrans. The output of
Safetrans is read into Matlab using the m-file ST_Load_in_ShipMo.m. This file is elucidated in figure C.1. The
ShipMo output is read in, and the stationary and regular sailing output are directly linked to the output. How-
ever, the sailing output is compensated for the encounter frequency effect already implemented in ShipMo.

131
132 C. Implementation and Verification of the Computations

This way the method can be better compared to the other estimation methods.

Figure C.1: Implementation of the stationary and sailing ST methods

The strip theory methods only describe the full force in the motion dominated zone plus the motion part in
the transition zone. Since the Salvesen’s method will yield the best results in a stationary case this method is
used. Sub-figure C.3b and e show the overprediction of the motion part in comparison with the FF method.
Both the surge and the sway component have peak values more than twice as high as the FF method. Expe-
rience tells us that the Salvesen’s strip theory method tends to overpredict in stationary cases which matches
the current results and thus this implementation is verified.

Faltinsen’s / Liu’s Short Wave Method


The Faltinsen en Liu’s short wave methods are implemented solely with Matlab. There implementation is
done via the m-files SW_Faltinsen.m and SW_Liu.m. Both use the SW_shadowzone.m routine. These rou-
tines are presented in figure C.2. The waterline, elucidated below, in combination with the angle of attack
determine the exposed waterline used for the integration in the SW methods. In case of the Liu SW method,
some extra terms are included in the panel integration.

Figure C.2: Implementation of the stationary and sailing Faltinsen and Liu SW methods

The waterline is described on the basis of sections. The number of sections is arbitrary, but the sections
should describe the waterline from bow to stern, anticlockwise. This is done based on a waterline input file,
for example WaterlineRB76.txt, which can be read into Matlab using the routine SW_getwaterline.m. The
input file only describes one half of the vessel, which is then mirrored in the read routine. The waterline
should be described by lines with the x and y-coordinates of the end points of the lines. For each panel the
waterline flare should also be included. The following information should be presented for each section if all
the approximation methods have to be computed:
• (x, y)-coordinates of the begin point of the waterline section;
C.1. Wave drift for a stationary vessel implementation and verification 133

• (x, y)-coordinates of the end point of the waterline section;


• η wl , the flare angle at waterline section;
• Tl oc al , the local draft at waterline section if different than global draft T ;

The short wave limits are theoretically only applicable to the reflection dominated zone, but in practise can
have a broader application range. This is clear from sub-figures C.3c and f. Since the results match the FF
method, these implementations are considered verified.

Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0°

Near Field Method, = 0° Strip Theory Method, = 0° SW Method Faltinsen, = 0°


Far Field Method, = 0° Far Field Method, = 0° SW Method Liu, = 0°
0 0 0
Far Field Method, = 0°

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

-0.5 -0.5 -0.5


Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone

-0.6 -0.6 -0.6


0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

(a) NF and FF method - Surge (b) ST method - Surge (c) Faltinsen and Liu SW method - Surge

Mean Sway QTF, Regular waves, = -15 ° Mean Sway QTF, Regular waves, = -15 ° Mean Sway QTF, Regular waves, = -15 °
0.3 0.3 0.3
Near Field Method, = -15° Strip Theory Method, = -15° SW Method Faltinsen, = -15°
Far Field Method, = -15° Far Field Method, = -15° SW Method Liu, = -15°
0.25 0.25 0.25 Far Field Method, = -15°

0.2 0.2 0.2

0.15 0.15 0.15

0.1 0.1 0.1

0.05 0.05 0.05


Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone

0 0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

(d) NF and FF method - Sway (e) ST method - Sway (f) Faltinsen and Liu SW method - Sway

Figure C.3: Verification stationary NF, FF, ST and Faltinsen’s / Liu’s SW methods

Fujii and Takahashi’s / Kuroda et al.’s / Kashiwagi et al.’s Method


These methods theoretically describe the reflection dominated zone, and originally only describe the wave
added resistance, the surge force. This is adjusted to also approximate the sway forces as described in the
literature chapter 2. These methods are implemented in a similar way as the other two SW methods with
the m-files SW_Fujii.m, SW_Kuroda.m and SW_Kashiwagi.m. Their routines are presented in figure C.4. The
difference with the other two SW methods is found in the method specific terms and the bluntness factor,
which includes the waterline integration.

Figure C.4: Implementation of the stationary and sailing Fujii and Takahashi’s / Kuroda et al.’s / Kashiwagi et al.’s methods

The implementations are visible in sub-figure C.5a and c. It is clearly visible that these three methods do not
differ in the stationary case; Kuroda et al.’s and Kashiwagi et al.’s methods only include adjusting parameters
for the sailing case. In this case, all three methods converge to the same value as the Faltinsen SW method.
Since this corresponds with the theory, these implementations are verified.
134 C. Implementation and Verification of the Computations

Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0°
0
SW Method F & T, = 0° R-function Seo-Faltinsen, = 0° DNV Formulation 2014, = 0°
SW Method Kuroda et al., = 0° Far Field Method, = 0° DNV Formulation 1996, = 0°
0 0 -0.1
SW Method Kashiwagi et al., = 0° Far Field Method, = 0°
Far Field Method, = 0°
-0.1 -0.1 -0.2

-0.2 -0.2 -0.3

-0.3 -0.3 -0.4

-0.4 -0.4 -0.5

-0.5 -0.5 -0.6


Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone

-0.6 -0.6 -0.7


0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

(a) Fujii and Takahashi’s / Kuroda et al.’s (b) R-function method - Surge (c) DNV formulations -Surge
/ Kashiwagi et al.’s method - Surge

Mean Sway QTF, Regular waves, = -15 ° Mean Sway QTF, Regular waves, = -15 °
0.3 0.3
SW Method F & T, = -15° R-function Seo-Faltinsen, = -15°
SW Method Kuroda et al., = -15° Far Field Method, = -15°
0.25 SW Method Kashiwagi et al., = -15° 0.25
Far Field Method, = -15°

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05
Motion zone Transition zone Reflection zone Motion zone Transition zone Reflection zone

0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

(d) Fujii and Takahashi’s / Kuroda et al.’s (e) R-function method - Sway
/ Kashiwagi et al.’s method - Sway

Figure C.5: Verification stationary R-function, DNV formulations, Fujii and Takahashi’s, Kuroda et al.’s and Kashiwagi et al.’s methods

R-Function method
The R-function method is implemented by the Seo_R_function.m routine in figure C.6. It combines a motion
dominated and a reflection dominated QTF to obtain a full range QTF. In this thesis, this is done on the basis
of the ST method and Faltinsen SW method. The actual R-function is computed with the Var_Rfunction.m
routine.

Figure C.6: Implementation of the stationary and sailing R-function methods

The combination of the motion and reflected parts of the wave drift force by the R-function method is pre-
sented in sub-figures C.5b and e. This method can be used with different combinations of the motion and
reflection part but here the Salvesen motion part and the Faltinsen reflection part are used. Since the results
are similar to the FF method the implementation is verified.

DNV Formulations
The DNV formulations are computed using Matlab, and only require the principle dimensions of the vessel.
It is implemented with the m-file DNV_function.m in figure C.7.

The DNV formulations presented in sub-figure C.5c and are only developed for surge force. They are not
simply rewritten to the sway force. The DNV 2014 formulation overpredicts the FF method by a factor two,
as already observed in the literature review chapter 2. This is a very conservative approach, but it catches the
C.2. Wave drift for a sailing vessel implementation and verification 135

Figure C.7: Implementation of the DNV formulations

motion peak of the force as well. For an accurate prediction of the wave drift force for resistance purposes this
formulation is too conservative. The DNV 1996 on the other hand underpredicts the FF method by roughly a
factor two. The cause of this is not clear as this formulation is an empirical formula. Also this formulation is
not applicable is accurate descriptions of the wave drift force are desired. Both formulations are not further
assessed due to this reason.

Irregular waves
The implementation of the irregular waves are presented in figure C.8. The regular wave estimations are
converted to irregular wave estimation by using a JONSWAP spectru. This spectrum is implemented by the
Env_JONSWAPSpectrum.m routine and produces characteristics like the statistical moments and accompa-
nying wave periods. The irregular wave method is obtained after the spectrum and the regular wave method
are multiplied per wave frequency and subsequently integrated over the full wave frequency range.

Figure C.8: Implementation of the irregular wave results

C.2. Wave drift for a sailing vessel implementation and verification


In this section the sailing case methods implementation and verification is presented. For the NF and FF
methods with vessel velocities the wave drift damping (WDD) is used. The WDD is already verified in the
literature chapter 2, so only the implementation of the wave drift damping is presented in combination with
the NF and FF methods. The other methods are verified here.

The sailing vessel methods are computed and called from the mail m-file Sailing_Comparison.m. Since this
vessel does nothing else than call and plot the wave drift estimation methods, it is not further discussed.

Near field and far field methods


The wave drift damping is computed based on the panel pressure output from Delfrac. The routine for
implementing this is presented in figure C.9. The main body of the implementation is given in the m-file
WDD_WaveDriftDamping.m. This routine uses two sub-routines, given in WDD_PressureMoments.m and
WDD_CoefMatrix.m.

The Delfrac output is used directly as input for computing the wave drift moments. This output then directly
serves as input for the wave drift damping matrix routine. The wave drift damping matrix serves as output
of the main body, and as input for computing the sailing vessel wave drift force. This computation uses the
wave drift damping in combination with the vessel velocities to adjust the stationary wave drift force to the
136 C. Implementation and Verification of the Computations

sailing wave drift force. The encounter frequency routine, discussed later on, is used to take the frequency
shift due to the velocities into account. Both sailing vessel wave drift forces are presented as the main body
output.

The pressure moments are computed based on the Delfrac output. As explained in the literature chapter 2, all
the wave drift moments (zeroth, first and second) are computed based on only the first three parts of the four
wave drift contributions presented by Pinkster [59]. The first part is computed with the waterline sections
and the relative wave height per section. The second and third part are computed based on the second order
(mean) pressure per panel. These parts are computed for all wave drift moments and added to obtain the
final wave drift moments. These are presented as output of the routine.

The wave drift damping coefficient matrix is computed based on the derivatives of the wave frequency and
angle of attack, and the stationary wave drift force components. Respectively for the 1994 and 1997 approx-
imations, the surge and sway forces and first order yaw moments are used or the first order surge and sway
moment contributions and the second order yaw moments are used. With this input, the coefficients are
computed and presented as a matrix at the output.

Figure C.9: Implementation of the wave drift damping and sailing vessel FF and NF methods
C.2. Wave drift for a sailing vessel implementation and verification 137

In sub-figures C.10a and c, the surge force peak is three times as high as the stationary case and the linear
increase in short waves is relative with a factor 4ωU /g . In sub-figure b the sway force shows similar, but less
pronounced features. Its peak is increased with a factor 2 ∼ 2.5 and shifted slightly to longer waves. The NF
method has a more distinct increase in the short waves, but still it is lower than the increase seen in the surge
force. This is also partially due to the panel mesh deviation. The computation of the results looks reliable and
the implementation is considered verified.

Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0°
0.2 0.2 0.2

0 0
0
-0.2 -0.2

-0.2
-0.4 -0.4

-0.6 -0.6
-0.4

-0.8 -0.8
-0.6
-1 -1

-1.2 -1.2 -0.8

-1.4 -1.4
-1
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-1.6 -1.6
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Far Field Method, u = 0 m/s, = 0°


Far Field Method, u = 2 m/s, = 0°
Near Field Method, u = 0 m/s, = 0° Far Field Method, u = 0 m/s, = 0° SW Method F & T, u = 0 m/s, = 0°
Near Field Method, u = 2 m/s, = 0° Far Field Method, u = 2 m/s, = 0° SW Method F & T, u = 2 m/s, = 0°
Near Field Method, u = 4 m/s, = 0° Far Field Method, u = 4 m/s, = 0° SW Method Kuroda et al., u = 0 m/s, = 0°
Far Field Method, u = 0 m/s, = 0° Strip Theory Method, u = 0 m/s, = 0° SW Method Kuroda et al., u = 2 m/s, = 0°
Far Field Method, u = 2 m/s, = 0° Strip Theory Method, u = 2 m/s, = 0° SW Method Kashiwagi, u = 0 m/s, = 0°
Far Field Method, u = 4 m/s, = 0° Strip Theory Method, u = 4 m/s, = 0° SW Method Kashiwagi, u = 2 m/s, = 0°

(a) NF and FF method - Surge (b) ST method - Surge (c) Fujii and Takahashi’s / Kuroda et al.’s
/ Kashiwagi et al.’s SW methods - Surge

Mean Sway QTF, Regular waves, = -15 °


1 Mean Sway QTF, Regular waves, = -15 ° Mean Sway QTF, Regular waves, = -15 °
0.4 0.25

0.9
0.35
0.8 0.2
0.3
0.7

0.6 0.25 0.15

0.5 0.2
0.1
0.4
0.15
0.3
0.1 0.05
0.2

0.05
0.1
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Far Field Method, v = 0 m/s, = -15°


Far Field Method, v = 2 m/s, = -15°
Near Field Method, v = 0 m/s, = -15° Far Field Method, v = 0 m/s, = -15° SW Method F & T, v = 0 m/s, = -15°
Near Field Method, v = 2 m/s, = -15° Far Field Method, v = 2 m/s, = -15° SW Method F & T, v = 2 m/s, = -15°
Near Field Method, v = 4 m/s, = -15° Far Field Method, v = 4 m/s, = -15° SW Method Kuroda et al., v = 0 m/s, = -15°
Far Field Method, v = 0 m/s, = -15° Strip Theory Method, v = 0 m/s, = -15° SW Method Kuroda et al., v = 2 m/s, = -15°
Far Field Method, v = 2 m/s, = -15° Strip Theory Method, v = 2 m/s, = -15° SW Method Kashiwagi, v = 0 m/s, = -15°
Far Field Method, v = 4 m/s, = -15° Strip Theory Method, v = 4 m/s, = -15° SW Method Kashiwagi, v = 2 m/s, = -15°

(d) NF and FF method - Sway (e) ST method - Sway (f) Fujii and Takahashi’s / Kuroda et al.’s
/ Kashiwagi et al.’s SW methods - Sway

Figure C.10: Verification sailing NF, FF, ST, Fujii and Takahashi’s, Kuroda et al.’s and Kashiwagi et al.’s SW methods

Strip Theory Method


The strip theory methods only describe the full force in the motion dominated zone plus the motion part
in the transition zone. For surge, the Gerritma and Beukelman’s method is used while for sway and yaw the
Salvesens method is applied. The implementation is the same as for the stationary case and is presented in
figure C.1.

The strip theory results for the wave added resistance is found in sub-figures C.10b and e. Again they are
plotted with u and v of 0, 2 and 4 m/s. For the surge force, the peak is very similar to the FF method, although
a bit less high. Since the latter can overpredict the motion zone this result is justified. For the sway, the peak
matches more closely, but should be noted that this is the result of just the surge velocity. The tail is similar
to the stationary case while the surge force tail is a lot shorter than the stationary case. The results lie in the
same range as previously found values so are considered valid and verified.
138 C. Implementation and Verification of the Computations

Fujii and Takahashi’s / Kuroda et al.’s / Kashiwagi et al.’s Method


Fujii and Takahashi’s method was originally developed for the wave added resistance and together with the
modified results of Kuroda et al. and Kashiwagi et al. they are plotted in sub-figures C.10c and f. Their
implementation for the sailing vessel is the same for the stationary vessel and is presented in figure C.4.
All three the results lie in the range of the FF method result and therefore, the implementation of all three
methods is verified.

R-Function Method
Combining the strip theory wave added resistance method with the updated Faltinsen SW method including
velocities yields a method describing the entire QTF. The implementation is the same as for the stationary
vessel, which is presented in figure C.6. The results are found in sub-figures C.11a and b. Here the encounter
wave number is used to link the two functions. Since the match of the entire QTF is good with respect to the
FF method, the implementation of this method is deemed verified.

IMO Formulation
For the wave drift force with vessel velocities, the IMO put forward an empirical formula for the surge force. It
only requires the principle vessel dimensions and its implementation is the same as for the DNV formulations
presented in figure C.7. The computation is presented in sub-figure C.11c. It is clear that this is an engineering
formulation, and that it is not applicable for vessels with blunt bows. This last point is clear from the fact that
length to 95% of the vessels breadth is very small and thus the force unrealistic. It also does not take the
forward velocity into account. It is therefore not suited for estimation the wave drift force accurately.

Mean Surge QTF, Regular waves, = 0° Mean Sway QTF, Regular waves, = -15 ° Mean Surge QTF, Regular waves, = 0°
0.2 1 0.2

0 0.9
0
0.8
-0.2
-0.2
0.7
-0.4
-0.4
0.6
-0.6
0.5 -0.6
-0.8
0.4 -0.8
-1
0.3
-1
-1.2 Motion zone Transition zone Reflection zone
0.2
-1.2
-1.4 0.1 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

-1.6 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
IMO formulation, = 0°
Far Field Method, u = 0 m/s, = 0°
Far Field Method, u = 1 m/s, = 0°
Near Field Method, u = 0 m/s, = 0° Near Field Method, v = 0 m/s, = -15° Far Field Method, u = 3 m/s, = 0°
Near Field Method, u = 2 m/s, = 0° Near Field Method, v = 2 m/s, = -15°
Near Field Method, u = 4 m/s, = 0° Near Field Method, v = 4 m/s, = -15°
Far Field Method, u = 0 m/s, = 0° Far Field Method, v = 0 m/s, = -15°
Far Field Method, u = 2 m/s, = 0° Far Field Method, v = 2 m/s, = -15°
Far Field Method, u = 4 m/s, = 0° Far Field Method, v = 4 m/s, = -15°
Seo R-function Method, u = 0 m/s, = 0° Seo R-function Method, v = 0 m/s, = -15°
Seo R-function Method, u = 2 m/s, = 0° Seo R-function Method, v = 2 m/s, = -15°
Seo R-function Method, u = 4 m/s, = 0° Seo R-function Method, v = 4 m/s, = -15°

(a) R-function method - Surge (b) R-function method - Sway (c) IMO formulation - Surge

Figure C.11: Verification sailing R-function method and IMO formulation

Faltinsen’s / Liu’s Short Wave Method


Faltinsen extended his stationary case method to (forward) velocities with a frequency and velocity depen-
dent part in the waterline integral. Together with the added coefficients by Liu, the results of these methods
is plotted in figure C.12.

If the equation presented by Faltinsen is slightly modified, a better correspondence with the FF and NF
method is achieved. Instead of a factor 2 before the frequency dependent part a factor 4 is proposed. The
(modified) equation is shown in equation (C.1) and the corresponding result in sub-figures C.12c and d. This
yields a much better fit for the surge force, see sub-figure c, while it is not necessarily suitable for the sway
force, see sub-figure d. This is especially clear from the fact that the sway near field method is known to have
a deviation at higher frequencies.

The theoretical derivation of this factor two is currently researched and no definitive proof has been pro-
duced. Next to that comes the incoming angle dependency. The surge force will decrease with increasing
incoming angle β while the sway force will increase. Since its incoming angle will be expected close to head
C.2. Wave drift for a sailing vessel implementation and verification 139

waves, the factor 4 for the surge force is adopted as it provides a better estimate of the FF method. For the
sway force - sway velocity equivalent this the original factor 2 is kept. With those changes implemented, the
computation of the methods is verified.
µ ¶
1 4ω
Z
F wd = ρ w g ζ2 sin2 (θv + αwd ) + u (cos (αwd ) + cos (θv ) cos (θv + αwd )) ~nd l (C.1)
wl 2 g

Mean Surge QTF, Regular waves, = 0° Mean Surge QTF, Regular waves, = 0°
0.2 0.2

0 0

-0.2 -0.2

-0.4 -0.4

-0.6 -0.6

-0.8 -0.8

-1 -1
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Far Field Method, u = 0 m/s, = 0° Far Field Method, u = 0 m/s, = 0°


Far Field Method, u = 2 m/s, = 0° Far Field Method, u = 2 m/s, = 0°
SW Method Faltinsen, u = 0 m/s, = 0° SW Method Faltinsen, u = 0 m/s, = 0°
SW Method Faltinsen, u = 2 m/s, = 0° SW Method Faltinsen, u = 2 m/s, = 0°
SW Method Liu, u = 0 m/s, = 0° SW Method Liu, u = 0 m/s, = 0°
SW Method Liu, u = 2 m/s, = 0° SW Method Liu, u = 2 m/s, = 0°

(a) Surge force (b) Updated Surge force

Mean Sway QTF, Regular waves, = 15 ° Mean Sway QTF, Regular waves, = 15 °
0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Near Field Method, v = 0 m/s, = -15° Near Field Method, v = 0 m/s, = -15°
Near Field Method, v = 2 m/s, = -15° Near Field Method, v = 2 m/s, = -15°
Far Field Method, v = 0 m/s, = -15° Far Field Method, v = 0 m/s, = -15°
Far Field Method, v = 2 m/s, = -15° Far Field Method, v = 2 m/s, = -15°
SW Method Faltinsen, v = 0 m/s, = -15° SW Method Faltinsen, v = 0 m/s, = -15°
SW Method Faltinsen, v = 2 m/s, = -15° SW Method Faltinsen, v = 2 m/s, = -15°
SW Method Liu, v = 0 m/s, = -15° SW Method Liu, v = 0 m/s, = -15°
SW Method Liu, v = 2 m/s, = -15° SW Method Liu, v = 2 m/s, = -15°

(c) Sway force (d) Updated Sway force

Figure C.12: Verification Faltinsen’s / Liu’s Short Wave methods wave drift force
140 C. Implementation and Verification of the Computations

Encounter frequency
The implementation of the encounter frequency is presented in figure C.13. This routine is used through-
out the thesis to account for the frequency shift due to the vessel velocities. The interpolation of the ’NaN’
values is done in case the numerical rounding causes a value not to be assigned in the output method. The
interpolation is only done in case non-’NaN’ values are present on both sides of the interpolated value, so the
interpolation is realistic.

Figure C.13: Implementation of the encounter frequency

C.3. Towing stability analysis implementation


In this section the implementation of the towing stability analysis is presented. An overview is given in figure
C.14. Note that in this elucidation the specific inputs and outputs are not included. The towing stability
analysis without (w/o) waves is implemented in the m-file EoM_Stability_Analysis.m. This file is the same as
the with waves (w/) waves routines, except the wave drift force is not included.

The inclusion of the wave drift force is done via the routine Stability_WDF.m. With the wave drift force com-
puted, the equilibrium positions are determining using the routines EoM_waves_Stability_Analysis_xp.m and
EoM_waves_Stability_Analysis_xp_Irreg.m. They are the same routines, just for regular and irregular waves.
The same applies to the routines used for the stability analysis: EoM_waves_Stability_Analysis_l0.m for reg-
ular waves and EoM_waves_Stability_Analysis_l0_Irreg.m for irregular waves. The dotted lines in the figure
indicate that these routines are essentially the same.

The time domain simulations used for the verification of the stability analysis are implemented using the
EoM_Hong.m routine. All these routines are presented in the following subsections.

Figure C.14: Implementation towing stability analysis


C.3. Towing stability analysis implementation 141

Wave drift database computation


The inclusion of the waves in the stability analysis is done by a database. This database is drawn using the
m-file Stability_WDF.m, which is further elucidated in figure C.15.

The Delfrac output is read in using the QTF_read_in_Delfrac.m routine. With the slightly modified routine
WDD_WaveDriftDamping_DB.m for the increased yaw angle range, the sailing vessel WDF is computed. A
regular wave and irregular waves WDF database is built. This database contains all the different parameter
combinations defined in the methodology chapter 3. The wave drift forces are saved categorised by file name
based on the three velocity components. For example: the file WDFYawRegu12v_01r0.mat produces the yaw
wave drift moment as a matrix of ω and ψ values with velocities u = 1.2m/s, v = −0.1m/s and r = 0m/s.

Figure C.15: Implementation wave drift database

Equilibrium position determination


The equilibrium position determination routine is presented in figure C.16. The routine is used to examine
the influence of x p on the equilibrium positions and does this by a loop around the position determination
routine EoM_waves_Stability_Equilibrium.m. The EoM_waves_Stability_Equilibrium_Irreg.m routine for the
irregular waves is the same routine.

Figure C.16: Implementation equilibrium position determination

The equilibrium position determination starts with the loading of the wave drift force database and using
the steady state values of the velocities to find the equilibrium wave drift force values. These WDF values
are padded in two dimensions, frequency and wave angle of attack. With the padded WDF, the zero-crossing
142 C. Implementation and Verification of the Computations

function is drawn up from the equations of motion and the zero crossings are determined. Depending on the
number of zero crossings, the equilibrium yaw angles ψeq are allocated to the different equilibrium positions.
With the governing parameter set, the remaining equilibrium variables are determined.

Stability analysis
With the equilibrium positions known, the stability of these positions can be assessed by the implemen-
tation in figure C.17. The equation of motion, vessel and environmental properties serve as input of the
EoM_Hong_func.m m-file. This routine is necessary as input for the jacobianest.m routine that linearises the
equation of motion at the equilibrium positions. The other input, the equilibrium position vector defining
where to linearise the equation of motion, is determined from the previously discussed equilibrium position
determination routine EoM_waves_Stability_Equilibrium.m. The Jacobian routine outputs a matrix with all
the partial derivatives of the equation of motion at the equilibrium position. From this matrix, the eigenval-
ues are analysed. The maximal real eigenvalue per equilibrium position subsequently provides the stability
properties of those positions.

Figure C.17: Implementation stability analysis

Time domain simulation


The simulations in the time domain are implemented using the routines presented in figure C.18. The main
body EoM_Hong.m routine revolves around the TD integration, which is performed by the Matlab native
ode45.m routine. The three inputs for this routine are drawn beforehand, the time vector for the integration
time, the initial conditions to start the integration„ and the state space vector, describe the next time step
based on the current state. The latter is described in the EoM_Hong_SS.m routine. The time integration
returns an output state space vector, describing the time traces of each state space vector. These time traces
can then, if present, be examined for their damping properties.

The EoM_Hong_SS.m routine formats the state space vector so the TD integration routine ode45.m can work
with it. It thereby uses the equation of motion state space EoM_Hong_func.m routine, also used in the stability
analysis.

The EoM_Hong_func.m routine provides the state space vector of the equation of motion. Vessel and envi-
ronmental properties, like wave frequency and wave amplitude, are used to compute the forces on the vessel.
These forces are the basis of determining the new state space vector. The manoeuvring forces are determined
based on the state space vector and the manoeuvring coefficients loading from the Input_MMGModel.m file.
If the wave drift forces should be included in the simulation, they are implemented using the EoM_Hong_WDF.m
routine. The same goes for the wind forces, if they are implemented, it is done using the routine.

The EoM_Hong_WDF.m m-file loads in the appropriate wave drift force from the WDF database, based on the
wave frequency, wave amplitude, yaw angle and the surge, sway and yaw velocities. It pads the wave drift
force data and outputs the result.

The wind contribution routine EoM_Hong_Wind.m uses the current state of the vessel and the wind velocity
to calculate the relative incoming angle of the wind and its corresponding relative velocity. Together with the
cubic spline fitted wind coefficients the wind force is computed.
C.3. Towing stability analysis implementation 143

Figure C.18: Implementation time domain simulation


Individual Multi-criteria Scores and
D
Sensitivity Analysis
In this appendix, the individual multi-criteria scores and the sensitivity analysis data are presented.

Individual multi-criteria scores


The individual scores for the different wave drift estimation methods are presented here.

R-function
Surge Motion Zone FF Method NF Method ST Method
Method
Category Factor Score Score Score Score
MSE 6 5 30 1 6 1 6 1 6
Theoretical Background 5 3 15 3 15 3 15 2 10
Simplicity Equation 1 1 1 1 1 2 2 2 2
Complexity of
3 1 3 1 3 2 6 3 9
parameter analysis
Application range 2 4 8 3 6 2 4 4 8
Applicable for
4 4 16 4 16 3 12 3 12
arbitrary vessel
Total Score 73 Total Score 47 Total Score 45 Total Score 47

Table D.1: The MCA scores of the Surge motion zone

Faltinsen
Surge Transition Zone FF Method NF Method ST Method
SW Method
Category Factor Score Score Score Score
MSE 6 5 30 5 30 2 12 4 24
Theoretical Background 5 3 15 3 15 3 15 3 15
Simplicity Equation 1 1 1 1 1 2 2 4 4
Complexity of
3 1 3 1 3 2 6 4 12
parameter analysis
Application range 2 4 8 3 6 2 4 2 4
Applicable for
4 4 16 4 16 3 12 2 8
arbitrary vessel
Total Score 73 Total Score 71 Total Score 51 Total Score 67

Table D.2: The MCA scores of the Surge transition zone

145
146 D. Individual Multi-criteria Scores and Sensitivity Analysis

R-function
Surge Transition Zone
Method
Category Factor Score
MSE 6 3 18
Theoretical Background 5 2 10
Simplicity Equation 1 2 2
Complexity of parameter analysis 3 3 9
Application range 2 4 8
Applicable for arbitrary vessel 4 3 12
Total Score 59

Table D.3: The MCA scores of the Surge transition zone - continued

Faltinsen Liu
Surge Reflection Zone FF Method NF Method
SW Method SW Method
Category Factor Score Score Score Score
MSE 6 5 30 5 30 4 24 3 18
Theoretical Background 5 3 15 3 15 3 15 3 15
Simplicity Equation 1 1 1 1 1 4 4 4 4
Complexity of
3 1 3 1 3 4 12 4 12
parameter analysis
Application range 2 4 8 3 6 2 4 2 4
Applicable for
4 4 16 4 16 2 8 4 16
arbitrary vessel
Total Score 73 Total Score 71 Total Score 67 Total Score 69

Fujii Kuroda Kashiwagi R-function


Surge Reflection Zone
SW Method SW Method SW Method Method
Category Factor Score Score Score Score
MSE 6 4 24 4 24 4 24 4 24
Theoretical Background 5 2 10 2 10 2 10 2 10
Simplicity Equation 1 4 4 4 4 4 4 2 2
Complexity of
3 4 12 4 12 4 12 3 9
parameter analysis
Application range 2 2 4 2 4 2 4 4 8
Applicable for
4 2 8 4 16 2 8 3 12
arbitrary vessel
Total Score 62 Total Score 70 Total Score 62 Total Score 65

Table D.4: The MCA scores of the Surge reflection zone

R-function
Sway Motion Zone FF Method NF Method ST Method
Method
Category Factor Score Score Score Score
MSE 6 5 30 4 24 1 6 1 6
Theoretical Background 5 3 15 3 15 3 15 2 10
Simplicity Equation 1 1 1 1 1 2 2 2 2
Complexity of
3 1 3 1 3 2 6 3 9
parameter analysis
Application range 2 4 8 3 6 2 4 4 8
Applicable for
4 4 16 4 16 3 12 3 12
arbitrary vessel
Total Score 73 Total Score 65 Total Score 45 Total Score 47

Table D.5: The MCA scores of the Sway motion zone


147

Faltinsen
Sway Transition Zone FF Method NF Method ST Method
SW Method
Category Factor Score Score Score Score
MSE 6 5 30 2 12 4 24 2 12
Theoretical Background 5 3 15 3 15 3 15 3 15
Simplicity Equation 1 1 1 1 1 2 2 4 4
Complexity of
3 1 3 1 3 2 6 4 12
parameter analysis
Application range 2 4 8 3 6 2 4 2 4
Applicable for
4 4 16 4 16 3 12 2 8
arbitrary vessel
Total Score 73 Total Score 53 Total Score 63 Total Score 55

Sway Transition Zone R-function Method


Category Factor Score
MSE 6 2 12
Theoretical Background 5 2 10
Simplicity Equation 1 2 2
Complexity of parameter analysis 3 3 9
Application range 2 4 8
Applicable for arbitrary vessel 4 3 12
Total Score 53

Table D.6: The MCA scores of the Sway transition zone

Faltinsen Liu
Sway Reflection Zone FF Method NF Method
SW Method SW Method
Category Factor Score Score Score Score
MSE 6 5 30 1 6 2 12 2 12
Theoretical Background 5 3 15 3 15 3 15 3 15
Simplicity Equation 1 1 1 1 1 4 4 4 4
Complexity of
3 1 3 1 3 4 12 4 12
parameter analysis
Application range 2 4 8 3 6 2 4 2 4
Applicable for
4 4 16 4 16 2 8 4 16
arbitrary vessel
Total Score 73 Total Score 47 Total Score 55 Total Score 63

Fujii Kuroda Kashiwagi R-function


Sway Reflection Zone
SW Method SW Method SW Method Method
Category Factor Score Score Score Score
MSE 6 4 24 4 24 3 18 2 12
Theoretical Background 5 2 10 2 10 2 10 2 10
Simplicity Equation 1 4 4 4 4 4 4 2 2
Complexity of
3 4 12 4 12 4 12 3 9
parameter analysis
Application range 2 2 4 2 4 2 4 4 8
Applicable for
4 2 8 4 16 2 8 3 12
arbitrary vessel
Total Score 62 Total Score 70 Total Score 56 Total Score 53

Table D.7: The MCA scores of the Sway reflection zone


148 D. Individual Multi-criteria Scores and Sensitivity Analysis

R-function
Yaw Motion Zone FF Method NF Method ST Method
Method
Category Factor Score Score Score Score
MSE 6 1 6 1 6 1 6 1 6
Theoretical Background 5 3 15 4 20 3 15 2 10
Simplicity Equation 1 1 1 1 1 2 2 2 2
Complexity of
3 1 3 1 3 2 6 3 9
parameter analysis
Application range 2 4 8 3 6 2 4 4 8
Applicable for
4 4 16 4 16 3 12 3 12
arbitrary vessel
Total Score 49 Total Score 52 Total Score 45 Total Score 47

Table D.8: The MCA scores of the Yaw motion zone

R-function
Yaw Transition Zone FF Method NF Method ST Method
Method
Category Factor Score Score Score Score
MSE 6 1 6 1 6 1 6 1 6
Theoretical Background 5 3 15 4 20 3 15 2 10
Simplicity Equation 1 1 1 1 1 2 2 2 2
Complexity of
3 1 3 1 3 2 6 3 9
parameter analysis
Application range 2 4 8 3 6 2 4 4 8
Applicable for
4 4 16 4 16 3 12 3 12
arbitrary vessel
Total Score 49 Total Score 52 Total Score 45 Total Score 47

Table D.9: The MCA scores of the Yaw transition zone

R-function
Yaw Reflection Zone FF Method NF Method
Method
Category Factor Score Score Score
MSE 6 1 6 1 6 1 6
Theoretical Background 5 3 15 4 20 2 10
Simplicity Equation 1 1 1 1 1 2 2
Complexity of
3 1 3 1 3 3 9
parameter analysis
Application range 2 4 8 3 6 4 8
Applicable for
4 4 16 4 16 3 12
arbitrary vessel
Total Score 49 Total Score 52 Total Score 47

Table D.10: The MCA scores of the Yaw reflection zone


149

Sensitivity analysis
To be certain the found results of the multi-criteria analysis are correct, a sensitivity analysis was performed.
The sensitivity analysis was done by looking at the scores of the MCA.

The individual scores of the different methods are varied per category. The scores per category are varied
+20% and -20%. In the following tables, the totals of the varied MCA scores are presented. These values are
the basis of figure 5.9.

Surge Motion FF Method NF Method ST Method R-function Method


Original 73 47 45 47
+20% Scores 81.6 55.2 52.8 55.2
-20% Scores 64.4 38.8 37.2 38.8

Table D.11: The Sensitivity Analysis of the Surge motion zone

Surge Transition FF Method NF Method ST Method Faltinsen SW Method R-function Method


Original 73 71 51 67 59
+20% Scores 81.6 79.2 58.8 75.6 67.2
-20% Scores 64.4 62.8 43.2 58.4 50.8

Table D.12: The Sensitivity Analysis of the Surge transition zone

FF NF Faltinsen Liu
Surge Reflection
Method Method SW Method SW Method
Original 73 71 67 69
+20% Scores 81.6 79.2 75.6 79.2
-20% Scores 64.4 62.8 58.4 58.8

Fujii Kuroda Kashiwagi R-function


Surge Reflection
SW Method SW Method SW Method Method
Original 62 70 62 65
+20% Scores 69.6 79.2 69.6 73.2
-20% Scores 54.4 60.8 54.4 56.8

Table D.13: The Sensitivity Analysis of the Surge reflection zone

Sway Motion FF Method NF Method ST Method R-function Method


Original 73 65 45 47
+20% Scores 81.6 73.2 52.8 55.2
-20% Scores 64.4 56.8 37.2 38.8

Table D.14: The Sensitivity Analysis of the Sway motion zone

Sway Transition FF Method NF Method ST Method Faltinsen SW Method R-function Method


Original 73 53 63 55 53
+20% Scores 81.6 61.2 70.8 63.6 61.2
-20% Scores 64.4 44.8 55.2 46.4 44.8

Table D.15: The Sensitivity Analysis of the Sway transition zone


150 D. Individual Multi-criteria Scores and Sensitivity Analysis

Sway Reflection FF Method NF Method Faltinsen SW Method Liu SW Method


Original 73 47 55 63
+20% Scores 81.6 55.2 63.6 73.2
-20% Scores 64.4 38.8 46.4 52.8

Table D.16: The Sensitivity Analysis of the Sway reflection zone

Yaw Motion FF Method NF Method ST Method R-function Method


Original 49 52 45 47
+20% Scores 57.6 61.2 52.8 55.2
-20% Scores 40.4 42.8 37.2 38.8

Table D.17: The Sensitivity Analysis of the Yaw transition zone

Yaw Transition FF Method NF Method ST Method R-function Method


Original 49 52 45 47
+20% Scores 57.6 61.2 52.8 55.2
-20% Scores 40.4 42.8 37.2 38.8

Table D.18: The Sensitivity Analysis of the Yaw reflection zone

Yaw Reflection FF Method NF Method R-function Method


Original 49 52 47
+20% Scores 57.6 61.2 55.2
-20% Scores 40.4 42.8 38.8

Table D.19: The Sensitivity Analysis of the Yaw motion zone


Bibliography
[1] M.A. Abkowitz. Stability and Motion Control of Ocean Vehicles. MIT Press, 1969.

[2] J. A. P. Aranha. A formula for ’wave damping’ in the drift of a floating body. Journal of Fluid Mechanics,
275, 1994.

[3] J. A. P. Aranha and M. R. Martins. Slender Body Approximation for Yaw Velocity Terms in the wave drift
damping matrix. In 12th International Workshop on Water Waves and Floating Bodies, 1997.

[4] M. M. Bernitsas and J. S. Chung. Nonlinear stability and simulation of two-line ship towing and mooring.
Applied Ocean Research, 12(2):77–92, 1990.

[5] M. M. Bernitsas and N. S. Kekridis. Simulation and Stability of Ship Towing. International Shipbuilding
Progress, 32(369):112–123, 1985.

[6] M. M. Bernitsas and N. S. Kekridis. Nonlinear Stability Analysis of Ship Towed By Elastic Rope. Journal
of Ship Research, 30(2):136–146, 1986.

[7] M. M. Bernitsas and F. A. Papoulias. Stability of single point mooring systems. Applied Ocean Research,
8(1):49–58, 1986.

[8] P. Boese. Eine einfache Methode zur Berechnung der Widerstandserhöhung eines Schiffes im Seegang
(In German). Journal of Ship Research, 17, 1970.

[9] T. H. J. Bunnik, E. van Daalen, G. Kapsenberg, Y. Shin, R. H. M. Huijsmans, G. B. Deng, G. Delhommeau,


M. Kashiwagi, and B. Beck. A comparative study on state-of-the-art prediction tools for seakeeping. In
28th Symposium on Naval Hydrodynamics, pages 12–17, 2010.

[10] X. B. Chen. Middle-field formulation for the computation of wave-drift loads. Journal of Engineering
Mathematics, 59:61–82, 2006.

[11] J. D’Errico.Adaptive Robust Numerical Differentiation, 2014. URL https://nl.mathworks.


com/matlabcentral/fileexchange/13490-adaptive-robust-numerical-differentiation?
focused=3852377{&}tab=function.

[12] Det Norske Veritas (DNV). DNV Rules for Marine Operations, 1996.

[13] Det Norske Veritas (DNV). DNV RP-C205 Environmental Conditions and Environmental Loads, 2007.

[14] Det Norske Veritas (DNV). DNV RP-H103 Modelling and Analysis of Marine Operations. Technical re-
port, DNV, 2014.

[15] Det Norske Veritas (DNV). DNV-GL ST-N001 Guidelines, 2017.

[16] I. Dmitrieva. DELFRAC Theory and Manual. Technical report, Delft University of Technology, 1994.

[17] O. M. Faltinsen. Sea Loads on Ships and Offshore Structures, volume 1. Cambridge University Press, 1990.

[18] O. M. Faltinsen. Hydrodynamics of High Speed Marine Vehicles. Cambridge University Press, 1st edition,
2005.

[19] O. M. Faltinsen, K. J. Minsaas, N. Liapis, and S. O. Skjordal. Prediction of Resistance and Propulsion of a
Ship in Seaway. In 13th Symposium on Naval Hydrodynamics, pages 505–529, 1980.

[20] M. Fang and J. Ju. The Dynamic Simulations of the Ship Towing System in Random Waves. Marine
Technology, 46(2):107–115, 2009.

151
152 Bibliography

[21] P. Ferrant, L. Gentaz, C. Monroy, R. Luquet, G. Ducrozet, B. Alessandrini, E. Jacquin, and A. Drouet. Re-
cent Advances Towards the Viscous Flow Simulation of Ships Manoeuvering in Waves. 23th International
Workshop on Water Waves and Floating Bodies, 2008.

[22] S. Finne, J. Grue, and A. Nestegard. Prediction of the Complete Second Order Wave Drift Damping Force
for Offshore Structures. In 10th International Offshore and Polar Engineering Conference, pages 320–327,
2000.

[23] A. Fitriadhy and H. Yasukawa. Course Stability of a Ship Towing System. Ship Technology Research, 58
(2):112–124, 2011.

[24] A. Fitriadhy, H. Yasukawa, and K. K. Koh. Course stability of a ship towing system in wind. Ocean Engi-
neering, 64:135–145, 2013.

[25] G.F. Franklin, J. David Powell, and A. Emami-Naeini. Feedback Control of Dynamic Systems. Pearson, 6th
edition, 2010.

[26] H. Fujii and T. Takahashi. Experimental study on the resistance increase of a ship in regular oblique
waves. In 14th International Towing Tank Conference, pages 351–360, 1975.

[27] L. O. Garza-Rios and M. M. Bernitsas. Stability Criteria for the Slow Motion nonlinear Dynamics of Tow-
ing and Mooring Systems. Technical report, The Department of Naval Architecture and Marine Engi-
neering Uniersity of Michigan, 1996.

[28] J. Gerritsma and W. Beukelman. Analysis of the Resistance Increase in Waves of a Fast Cargo Ship. Inter-
national Shipbuilding Progress, 19(217):285–293, 1972.

[29] M. Gokce. Coupled Stability Analysis of Close Proximity Ship Towing. PhD thesis, Naval Postgraduate
School Monterey California, 2002.

[30] R. Grin. On the Prediction of Wave Added Resistance With Empirical Methods. Journal of Ship Produc-
tion and Design, 30(4), 2014.

[31] J. Grue and E. Palm. Mean yaw moment on floating bodies advancing with a forward speed in waves. In
Behaviour of Offshore Structures Conference (BOSS’92), London, Great Britain, pages 1–12, 1992.

[32] John Grue and Enok Palm. The mean drift force and yaw moment on marine structures in waves and
current. Journal of Fluid Mechanics, 250(1993):121–142, 1993.

[33] B. J. Guo and S. Steen. Evaluation of added resistance of KLVLCC2 in Short Waves. Journal of Hydrody-
namics, 23(6):709–722, 2011.

[34] J. He, H. Zhang, and X. Li. A New Method of Wind Load Calculation for Offshore Structures. In 26th
International Ocean and Polar Engineering Conference, volume 615, pages 548–552, 2016.

[35] A. J. Hermans and L. M. Sierevogel. A discussion on the second-order wave forces and wave drift damp-
ing. Applied Ocean Research, 18(5), 1996.

[36] J. Holtrop. De Weerstand van Bakvormige Vaartuigen (Dutch). SWZ Maritime, 1991.

[37] S. Y. Hong, B. W. Nam, J. H. Kim, and J.-y. Park. A Study on Towing Characteristics of a Transportation
Barge in Waves. In 23th International Offshore and Polar Engineering Conference, volume 9, pages 777–
783, 2013.

[38] R. H. M. Huijsmans. The mathematical modelling of the mean wave drift forces in current : a numerical
and experimental study. Doctoral thesis, Delft University of Technology, 1996.

[39] S. Inoue, K. Kijima, M. Murakami, K. Sakata, and S. Lim. Some Studies of the Course Stability of Towed
Ships Systems. In In Transactions Of The West-Japan Society Of Naval Architects, 1975.

[40] J. M. J. Journée, W. W. Massie, and R. H. M. Huijsmans. Offshore Hydromechanics. Delft Univeristy of


Technology, 3rd edition, 2015.
Bibliography 153

[41] M. Kashiwagi, T. Ikeda, and T. Sasakawa. Effects of forward speed of a ship on added resistance in waves.
International Journal of Offshore and Polar Engineering, 20(3):196–203, 2010.

[42] D.L. Kolthoff. Parametric study of the Dynamic Stability of Towed Ships. PhD thesis, Naval Postgraduate
School Monterey California, 1990.

[43] M. Kuroda, T. Fujiwara, K. Takagi, M. Tsiljimoto, and S. Ohmatsu. Investigation on components of added
resistance in short waves. Journal of the Japan Society of Naval Architects and Ocean Engineers, 8, 2008.

[44] M.L. Lee. Dynamic stability of nonlinear barge-towing system. Applied Mathematical Modelling, 13(12):
693–701, 1989.

[45] S. Liu and A. D. Papanikolaou. Prediction of the Added Resistance of Ships in Oblique Seas. In 26th
International Ocean and Polar Engineering Conference, pages 495–502, 2016.

[46] S. Liu, A. D. Papanikolaou, and G. Zaraphonitis. Practical approach to the added resistance of a ship in
short waves. In 25th International Ocean and Polar Engineering Conference, pages 11–18, 2015.

[47] S. Malenica, P. J. Clark, and B. Molin. Wave and current forces on a vertical cylinder free to surge and
sway. Applied Ocean Research, 17(2):79–90, 1995.

[48] MARIN. Safetrans 7.0 user manual. Technical Report June, Marin, 2016.

[49] H. Maruo. The drift of a body floating on waves. Journal of Ship Research, 4(4), 1960.

[50] Y. Nakayama, H. Yasukawa, N. Hirata, and H. Hata. Time Domain Simulation of Wave-induced Motions
of a Towed Ship in Head Seas. In 22th International Offshore and Polar Engineering Conference, 2012.

[51] B. W. Nam, Y. M. Choi, and S. Y. Hong. A Study on Towing Characteristics of Barge Considering Wind
Force (in Korean). Journal of Ocean Engineering and Technology, 29(4):283–290, 2015.

[52] J. N. Newman. The drift force and moment on ships in waves. Journal of Ship Research, 11(1), 1967.

[53] J. N. Newman. Second-order, slowly-varying forces on vessels in irregular waves. Marine Vehicles, 1974.

[54] F.G. Nielsen. On the Influence of the Propellor Race on the Large, Towed Structures. In Offshore Technol-
ogy Conference, pages 1773–1779, 1979.

[55] OCIMF. Prediction of Wind and Current Loads on VLCCs. Technical report, Oil Companies International
Marine Forum, 1994.

[56] F. A. Papoulias and M. M. Bernitsas. Stability of Motions of Moored Ocean Vehicles. Dynamics and
Stability of Systems, 1(4), 1986.

[57] J. H. Park, D. Y. Kang, J. M. Baek, H. W. Cho, and D. H. Choi. Egina: Towing Stability Model Test Report.
Technical report, Samsung Heavy Industries, NG-EGN-11-KSEC-251003, 2014.

[58] J. H. Park, D. Y. Kang, J. M. Baek, H. W. Cho, and D. H. Choi. Egina: Wind Tunnel Model Test Report.
Technical report, Samsung Heavy Industries, NG-EGN-11-KSEC-251004, 2014.

[59] J. A. Pinkster. Low Frequency Second Order Wave Exciting Forces on Floating Structures. Doctoral thesis,
Delft University of Technology, 1980.

[60] POSH SEMCO Pte Ltd. Maritime Swift Technical Note. Technical report, POSH SEMCO Pte Ltd, 2005.

[61] T. Sakamoto and E. Baba. Minimisation of Resistance of Slowly Moving Full Hull Forms in Short Waves.
In 16th Symposium on Naval Hydrodynamics, pages 598–614, 1986.

[62] N. Salvesen. Second-order steady state forces and moments on surface ships in oblique regular waves.
In International Symposium on Dynamics of Marine Vehicles and Structures in Waves, pages 225–241,
1974.

[63] M. Seo, D. Park, K. Yang, and Y. Kim. Comparative study on computation of ship added resistance in
waves. Ocean Engineering, 73:1–15, 2013.
154 Bibliography

[64] Vl. Shigunov and T. E. Schellin. Tow Forces for Emergency Towing of Containerships. Journal of Offshore
Mechanics and Arctic Engineering, 137(October 2015):1–8, 2015.

[65] Siemens PLM Software. Femap v11.3.1, 2016.

[66] M. Sinibaldi and G. Bulian. Towing simulation in wind through a nonlinear 4-DOF model: Bifurcation
analysis and occurrence of fishtailing. Ocean Engineering, 88:366–392, 2014.

[67] M. Sinibaldi, G. Bulian, and A. Francescutto. A Nonlinear Dynamics Perspective on Some Aspects of
Towing Operations Relevant To Safety and Energy Efficiency. In International Conference IDS 2013, 2013.

[68] H. Söding and V. Shigunov. Added Resistance of Ships in Waves. Ship Technology Research, 62(1):2–13,
2015.

[69] A. G. Strandhagen, K. E. Schoenherr, and F. M. Kobayashi. The Dynamic Stability on Course of Towed
Ships. Society of Naval Architects and Marine Engineers, 58:32–46, 1950.

[70] The MathWorks Inc. Matlab 2016b, 2016.

[71] K. S. Varyani, N. D. P. Barltrop, and X. P. Pham. Fishtailing Instabilities in Emergency Towing of Disabled
Tankers. In 15th International Offshore and Polar Engineering Conference, pages 350–357, 2005.

[72] K. S. Varyani, N. D. P. Barltrop, and X. P. Pham. SAFETOW Development of New Generic Equations for
Estimating Second Order Mean Wave Drift Forces/Moment acting on a Tanker. Technical report, SAFE-
TOW, FP6-PLT-506317, 2005.

[73] K. S. Varyani, N. D. P. Barltrop, X. P. Pham, K. van Essen, R. Doyle, and L. Speller. Simulation of the
Dynamics of a Tug Towing a Disabled Tanker in Emergency Salvage Operation. In International Marine
Simulation Forum, 2006.

[74] J. E. W. Wichers. A Simulation Model for a Single Point Moored Tanker. PhD thesis, Delft University of
Technology, 1988.

[75] J. E. W. Wichers. Guide to Single Point Moorings. WMooring, Inc., 2013.

[76] H. Yasukawa and Y. Yoshimura. Introduction of MMG standard method for ship maneuvering predic-
tions. Journal of Marine Science and Technology, 20(1):37–52, 2015.

You might also like