Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 53

f

Influence of bank characteristics on ship-


bank interaction for a 135m ship length

submitted on 24 August 2021


by
CALIN Andrei-Daniel |Rue Trappe 10 | 4000 Liege | andrei_calin_daniel@yahoo.com
Student ID No.: 200414Q

First Reviewer: Second Reviewer:


Prof. Andre Hage Prof. Giles Barkley
Associate Professor at University of Liege Professor at SOLENT University
Allée de la Découverte 12 E Park Terrace
4000 Liege SO14 0YN Southampton
Belgium United Kingdom
2

ACKNOWLEDGEMENTS
Firstly, I would like to express my deepest gratitude to professor Andre Hage for
making this internship and completion of my master thesis possible as well as for his
guidance throghout the duration of the internship. I would also like to thank my coordinator,
Dr. Adrian Constantinescu who played a key role in developing my abilities during this
internship and for his continous support, advice and knowledge. Many thanks to the people at
DN&T who welcomed me very warmly and supported me.
Special thanks to Mr. Radomir Jasic and Mr. Rakotamanana Anjatiana who helped me
achieve my goals for this master thesis in due time and provided valuble skills and knowledge
which helped me grow as an engineer and as a person.
The computational resources needed for this work were enormous. I would like to
express my gratitude to everyone at CNRS and LHEEA for the time alocated on LIGER HPC
cluster. Special thanks to Mr. Pierre-Emmanuel Guerrin who made possible for me to used
LIGER and Mr. Alban Leroyer for his continous support in using LIGER.
This project is part of EMSHIP master program and I would like to thank the
coordinator of the program, Dr. Phillipe Rigo from Unviersity of Liege and Dr.Lionel Gentaz
from Ecole Centrale Nantes for their support and for this amazing opportunity called
EMSHIP.
I would like to thank my mother and my father who supported me both finacially and morally
throughout my studies and also my fiancee Ms. Claudia Bostan for her love and support.
3

Table of contents
DECLARATION OF AUTHORSHIP........................................................................................7
ABSTRACT................................................................................................................................8
1. INTRODUCTION..................................................................................................................9
2. LITERATURE REVIEW.....................................................................................................10
3.SHIP-RIVER BANK INTERACTION ANALYSIS BY EXPERIMENTAL TESTS.........15
3.1. Model preparation..........................................................................................................15
3.2. Towing tank facilities and preparation..........................................................................17
3.3. Experimental configurations..........................................................................................19
3.4. Presentation and interpretation of the experimental results...........................................22
3.4.1. Vertical banks cases................................................................................................22
3.4.2. Inclined wall cases..................................................................................................25
4.SHIP-RIVER BANK INTERACTION ANALYSIS BY CFD SIMULATIONS.................27
4.1. CFD setup description....................................................................................................27
4.2. HPC Resource................................................................................................................34
4.3. Presentation and interpretation of numerical results......................................................34
4.3.1. Model scale tests.....................................................................................................34
4.3.2. Real scale tests........................................................................................................36
5. COMPARISON BETWEEN EXPERIMENTAL AND NUMERICAL RESULTS.
INTERPRETATIONS..............................................................................................................39
5.1. Sway force comparison in the case of vertical banks....................................................39
5.1.1. Case 1 comparison..................................................................................................39
5.1.2. Case 2 comparison..................................................................................................40
5.1.3. Case 3 comparison..................................................................................................41
5.1.4. Case 4 comparison..................................................................................................41
5.1.5. Case 5 comparison..................................................................................................42
5.2. Yaw moment comparison in the case of vertical banks.................................................44
5.3. Extrapolating model-scale results..................................................................................45
5.4. Advices for CFD and experimental test to obtain interaction forces and moments for
real cases...............................................................................................................................47
6. CONCLUSION.....................................................................................................................48
6.1. Future work....................................................................................................................50
7. REFERENCES.....................................................................................................................51
4

Table of figures
Figure 1: Norrbin (1974 and 1985) vertical bank configuration...............................................10
Figure 2: Parameters defining bank geometry, Duffy (2005)...................................................11
Figure 3: Types of banks used by Kim and Ng(2017)..............................................................13
Figure 4: 3D model of the ship.................................................................................................15
Figure 5: Ship model with weights...........................................................................................16
Figure 6: Hydrodynamic balance..............................................................................................17
Figure 7: Towing tank at University of Liege (ANAST).........................................................18
Figure 8: False bottom and vertical wall in the towing tank.....................................................19
Figure 9: Definition of yB in case of inclined banks................................................................20
Figure 10: Experimental setup and coordinate axis used in towing tank testing......................21
Figure 11 : Test phases.............................................................................................................22
Figure 12 : Sway force evolution in the case of vertical banks................................................23
Figure 13: Yaw moment evolution in the case of vertical banks..............................................24
Figure 14: Sway force evolution in the case of inclined banks................................................25
Figure 15: Yaw moment evolution in the case of inclined banks.............................................26
Figure 16 : Mesh generation steps in HEXPRESS...................................................................27
Figure 17: Domain setup in FineMarine...................................................................................28
Figure 18: Domain surface names............................................................................................29
Figure 19 : Refinement box......................................................................................................30
Figure 20 : Generated mesh......................................................................................................31
Figure 21: Sway force evolution in the case of vertical banks –numerical results...................35
Figure 22: Yaw moment evolution in the case of vertical banks –numerical results...............36
Figure 23: Sway force evolution for the real ship after numerical computations.....................37
Figure 24: Yaw moment evolution for the real ship after numerical computations.................38
Figure 25: Comparison between evolution of sway force between simulation and
experimental results at yB=8cm...............................................................................................39
Figure 26: Comparison between evolution of sway force between simulation and
experimental results at yB=12.2 cm.........................................................................................40
Figure 27: Comparison between evolution of sway force between simulation and
experimental results at yB=20 cm............................................................................................41
5

Figure 28: Comparison between evolution of sway force between simulation and
experimental results at yB=30cm.............................................................................................42
Figure 29: Comparison between evolution of sway force between simulation and
experimental results at yB=40cm.............................................................................................43
Figure 30: Y+ distribution on the hull......................................................................................44
6

List of tables
Table 1: Main dimensions of ship.............................................................................................15
Table 2: Model main dimensions..............................................................................................16
Table 3: Ship’s forward velocities tested at real and model scale............................................20
Table 4: Vertical distances to bank at model and real scale.....................................................20
Table 5: Influence of portside wall...........................................................................................21
Table 6: Values of sway force after experimental tests in the case of vertical banks...............23
Table 7: Values of yaw moment after experimental tests in the case of vertical banks...........24
Table 8: Values of sway force after experimental tests in the case of inclined banks..............25
Table 8: Values of sway force after experimental tests in the case of inclined banks..............26
Table 9 : Initial mesh parameters..............................................................................................29
Table 10: Number of refinements for each surface of the computational domain...................29
Table 11: Viscous layers...........................................................................................................31
Table 12 : Boundary conditions for solid surfaces...................................................................31
Table 13 : Boundary conditions for external boundaries..........................................................32
Table 14: Values of sway foce after numerical tests in the case of vertical banks...................34
Table 15: Values of yaw moment after numerical tests in the case of vertical banks..............35
Table 16: Sway force values for the real ship after numerical computations...........................36
Table 17: Yaw moment values for the real ship after numerical computations.......................37
Table 18: Comparison between towing tank results and CFD result for yB=8cm for sway
force..........................................................................................................................................39
Table 19: Comparison between towing tank results and CFD result for yB=12.2cm for sway
force..........................................................................................................................................40
Table 20: Comparison between towing tank results and CFD result for yB=20cm for sway
force..........................................................................................................................................41
Table 21: Comparison between towing tank results and CFD result for yB=30cm for sway
force..........................................................................................................................................41
Table 22: Comparison between towing tank results and CFD result for yB=40 cm for sway
force..........................................................................................................................................42
Table 23: Relative difference(Rd) for sway force....................................................................43
Table 24: Relative difference(Rd) for yaw moment.................................................................44
Table 25 : RDiff for sway force in case of numerical results...................................................46
Table 26 : RDiff for yaw moment in case of numerical results................................................46
7

DECLARATION OF AUTHORSHIP

I declare that this thesis and the work presented in it are my own and have been generated by
me as the result of my own original research.

Where I have consulted the published work of others, this is always clearly attributed.

Where I have quoted from the work of others, the source is always given. With the exception
of such quotations, this thesis is entirely my own work.

I have acknowledged all main sources of help.

Where the thesis is based on work done by myself jointly with others, I have made clear
exactly what was done by others and what I have contributed myself.

This thesis contains no material that has been submitted previously, in whole or in part, for
the award of any other academic degree or diploma.

I cede copyright of the thesis in favour of the Ecole Centrale Nantes.

Date: Signature
8

ABSTRACT

With the global trade ships became larger while channels and canals struggled to grow at the
same rate. This leads to ships navigating in restricted waterways much more often than
before.
This thesis aims to investigate the bank effect which occurs when a ship is navigating close to
a bank through the research on how the sway force and yaw moment evolves when parmeters
such as speed and distance to bank are varied.
The effect is analysed by means of experimental tests performed at the towing tank of
University of Liege as well as by CFD simulations performed using FineMarine CFD
industrial code. Numerical simulations at both model scale and real scale were conducted.
Numerical results were compared with experimental ones and a detailed discussion of results
has been realized by comparing values obtained from both methods of evaluation used.
All results obtained can be used to build or enhance a mathematical model for pilot simulators
to get their certification.
9

1. INTRODUCTION

The international trade relies on ship transport as approximately 90% of international trade is
transported by sea. To facilitate this global trade numerous canals have been built by mankind
in order to shorten the time needed to get from point A to point B. This implies that a ship, no
matter its dimensions, has to pass through a restricted waterway. While navigating in
restricted conditions the ship suffers from interaction forces because of the presence of side
walls and restricted bottom. These conditions can also be often found in inland navigation.
In recent years the main dimensions of ships have grown faster than the rate of which
channels and restricted waterways have been enlarged. As a results ships started to navigate
closer and closer to banks as well as with a lower under keel clearance than before.
Navigating in such proximity of a bank induces a sway force and a yaw moment due to the
asymmetric pressure distribution which acts on the ship and which pilots must take into
account in order to avoid any accidents. This effect corroborated with effects coming from
other sources (such as wind) can lead to dangerous accidents. In recent memory one of these
accidents almost halted the international trade, the Ever Given incident. In this particular case
the massive 400m length and 59m wide 20000 TEU containership remained blocked for 6
days grounded on the Suez Canal after experiencing high winds in March 2021. According to
reports the bank effect had a role to play in the grounding of the Ever Given. It is therefore
necessary to accurately predict the bank effect in order to avoid such accidents in the future.
The main goal of the present thesis is to evaluate lateral force and yaw moment acting on a
ship by means of experimental tests and CFD simulations.
10

2. LITERATURE REVIEW
The ship navigation in close proximity to channel walls accelerates the flow along the hull
which will modify the pressure distribution along the hull resulting in a combination of forces
and moments acting on the ship. The phenomena have been investigated mainly using
experimental techniques due to the lack of full scale measurements making validation
procedures difficult but also some studies used numerical simulations to evaluate the
influence of side walls. A number of studies on ship-bank interaction were found in the
literature.
Norrbin(1974 and 1985) completed a series of experimental tests on bank effects , but also on
manoeuvring. Based on a tests on a tanker model (with L = 5.024m, B = 0.852m and T =
0.339 m and Cb = 0.821) Norrbin(1974 and 1985) proposed the following formulations for
bank-induced lateral force and yaw moment in the case of vertical walls as seen in Figure 1:
Y = ρCBBTu2η0 [0.00926 + 0.372 (T/h)2] (1)
N = - ρCBBTu2η0 [0.0025 + 0.07555 (T/h)2] (2)

Figure 1: Norrbin (1974 and 1985) vertical bank configuration

Norrbin(1974 and 1985) also proposed formulations for sloped banks and flooded banks .

The most comprehensive experimental campaign has been performed at Flanders Hydraulics
Research Center (FHR) which comprises more than 8000 unique model test setups. Three
different types of bank geometries have been used:
 vertical walls: a wall that runs from the bottom of the towing tank to the free surface
and beyond
 surface piercing banks (characterized by a constant slope): a wall that runs a constant
slope from the bottom of the towing tank up to the free surface and beyond
11

 semi-submerged (or flooded) banks: a sloped bank which starts at the bottom of the
towing tank but ends before reaching the free surface. After the slope there is a
horizontal section until the wall of the towing tank.
11 ship models were tested during this campaign. From these 11 models, four of them are
tankers, two are container carriers, three are Ro-Ro vessels, one is an inland vessel and the
last one is a Wigley hull, which is a mathematically defined geometry. During tests four
parameters were varied: water depth, lateral position in the cross section, forward speed and
propeller action. It has to be noted that the majority of the tests were performed at a model
scale velocity ranging between 0.5 and 0.8 m/s.
Based on the data obtained a mathematical model was proposed by Lataire(2014) based on
mathematical regression analysis for predicting the lateral force and yaw moment of a ship
passing through restricted waterways. The study found that the lateral force can be either an
attraction force or a repulsion force. Its magnitude increases if the forward speed of the ship is
greater, if the ship sails in a more confined region, if the under keel clearance is lower or if the
distance between the ship and the bank reduces. This paper also takes into account the effect
of the propeller, which is stated to increase lateral efforts as propeller rate increases but
ultimately having a minor influence on the resulting efforts. Through these experiments it was
found that there is a transition point where the lateral force switches from attraction force to
repulsion force when the under keel clearance is small (about 15-20% of ship’s draft).

A similar set of experiments have been conducted by Duffy (2005 and 2008) which included
tests on surface piercing banks but also flooded banks. The tests were performed using 3 ship
models: 2 bulk carriers and one containership.

Figure 2: Parameters defining bank geometry, Duffy (2005)


Available at
https://www.researchgate.net/publication/327121448_Prediction_of_bank_induced_sway_force_and_
yaw_moment_for_ship-handling_simulation
12

During the test program multiple parameters were varied (as seen in Figure 2) as indicated
below:
 bank slope: 45deg < α < 90deg
 ship to bank distance: -0.629 < yB’< 0.629
d
 water depth: 1.2< h −D <10, where d is the ship draft
1

h2
 bank height: 0.44 <1− <1
h1
Values resulting from the experimental campaign showed that there is linear correlation
between ship to bank distance for both sway force and yaw moment. This is in according to
the findings of Norrbin(1974) which found that generally sway force and yaw moment vary
linearly with ship to bank distance.
The study concluded that the sway force may be directed towards or away from the bank
depending on ship’s speed, draft, under keel clearance and bank geometry (flooded, vertical
or inclined). For each researched case the yaw moment was bow away. Based on these results,
empirical equations were derived using multiple regression analysis for both surface-piercing
and flooded banks.

Fathi Kazerooni and Seif (2015) performed experimental studies on ship-bank interaction for
two ship models: a tanker and a cargo vessel named Dhow. The experimental setup included
an artificial bottom and side in order to simulate a draft to water depth ratio of 2. The banks
are surface piercing vertical banks with different distances being studied. The study found that
the magnitude of sway force and yaw moment increase with the increase in ship’s speed
(Froude number) as well as if the distance between the bank and the ship. In the case of the
tanker model a point where the sway force changes sign was found at approximately Fn =
0.15. Tests showed that this point is delayed as ship to bank distance increases. The most
critical case happens when the ship’s speed is medium. Here the reflecting wave system
affects the hull completely and the sway force and yaw moment is significant. The tests
performed on dhow model showed that the point where the sway force changes sign is located
at about Fn = 0.1.

Numerical studies have also been performed for the ship-river bank interaction phenomena.
Wang et al (2010) did a viscous study analysis for a series 60 hull using k-ω SST turbulence
model with water depths to draft ratios ranging from 0.5 to 10. Results showed good
13

agreement with experiments for sway forces but over-prediction in the case of a very small
distance to the bank.
Mehdi et al. (2013) carried out a RANSE simulation for a flow around a LNG in the case of
confined space. The study employed the same k-ω SST turbulence model, results being
compared to experimental results. Yaw moment comparison showed good agreement with
experimental results while sway forces were not presented in the article.
Kim and Ng (2017) did a CFD study on ship-bank interaction on an 8000TEU container ship.
The numerical setup used was done in order to simulate the experimental setup done by
Lataire et al (2009). 3 types of bank were used as can be seen in Figure 3:

Figure 3: Types of banks used by Kim and Ng(2017)


Available at https://www.researchgate.net/publication/340333718_CFD_Study_of_Ship-to-
Bank_Interaction

6 test cases were performed, 3 where the propeller thrust was taken into account and 3 where
the influence of the propeller was neglected. The solver used was interFoam of OpenFoam
which is a sover for transient solutions of two incompressible fluids and uses a VOF (volume
of fluid) phase-fraction. The results were compared with the experimental results obtained by
Lataire et al(2009) as well as with the results obtained with simpleFoam a single-phase
14

steady-state solver. For the cases without propeller thrust the CFD results predicted both over
and under-prediction but with errors ranging from 9.8% to 39% in the case of sway force
compared to experimental results. However, the results where propeller thrust was taken into
account showed values well under-predicted by a factor of 4. More study is required to
confirm the capability of CFD when propeller effect is added. Also free surface elevation was
compared to the experimental results showing reasonably agreement except the difference in
wave phase. The study concluded that free surface effect is not significant in the case of ship-
bank interaction and therefore a single-phase solver is more useful because it will be faster
and more stable compared to multi-phase solvers.

Although many studies have been performed, the lack of full scale trial results is still an issue
as pointed out by the 23rd ITTC Manoeuvring Committee meaning care should be taken as
discrepancies can appear between simulated results and real test results.
Most of the research conducted on ship-bank interaction, both experimental and numerical, is
usually used to enhance and improve manoeuvring simulators for ships. These simulators are
used by pilots to train and get certified in different conditions and different types of ports
where restricted waterways are predominant. The necessity of a good prediction of forces and
moments acting on a ship passing through a restricted waterway at real scale has been
highlighted by numerous studies and it remains a focal point of the ongoing research in the
field of ship-bank interaction.
15

3.SHIP-RIVER BANK INTERACTION ANALYSIS BY


EXPERIMENTAL TESTS
In this section the experimental configuration,procedure and results will be detailed. During
experimental tests an experimental configuration was defined and some parameters were
varied in order to evaluate how they influence the value and sign of both sway force and yaw
moment.

3.1. Model preparation


The present thesis aims to research the effects of restricted waterways and the forces
associated with it during a case study on a barge-type ship designed for inland navigation with
the following dimensions:

Table 1: Main dimensions of ship


Overall length Breadth Loaded draft Block coefficient
135 m 11.4 m 2.5 m 0.85

The 3D model is presented in Figure 4 :

Figure 4: 3D model of the ship

A model with scale of 1:25 will be used for experimental runs. Thus, the model has the
following characteristics :
16

Table 2: Model main dimensions


Overall length Breadth Loaded draft
5.4 m 0.46 m 0.1 m

The model has been prepared in agreement with ITTC 7.5-01-01-01 Ship models procedure.
Turbulence has been stimulated using sand strips positioned at 27cm aft of the fore
perpendicular. The model has been weighed using the crane installed on the ceiling of the
towing tank and a scale. The scale showed that the model weighs about 66.75 kg which
conducts to a draft of 3.2 cm.
To obtain the required draft (corresponding to a loaded vessel, real draft of 2.5 m, a
displacement of 3442 tons), weights have been added to the model in such a way that mean of
the four drafts (aft and fore perpendiculars starboard and portside) will be 10 cm, with a
tolerance of 2mm, as specified by the ITTC procedure. A draft of 10 cm corresponds to a
displacement of about 216.62 kg. 18 weights have been placed throughout the length of the
ship with a total mass of 149.87 kg. To ensure that mean draft is achieved and the ship has no
trim or roll angle the following steps have been followed:
 Firstly, the weights have been placed on the model with the goal of positioning their
center of gravity on the centreline of the model with the intention of getting no roll
angle
 Then the draft has been measured at aft and fore perpendiculars starboard and portside
to get a reading of the center of gravity of the ship as well as the roll and trim angles.
It is known that for a ship to have no trim angle the ship’s center of gravity and center
of buoyancy have to be on the same vertical while for the ship to have no roll angle
the buoyancy center and the gravity center have to be on the same lateral coordinate.
The weights were then moved and drafts were re-measured iteratively until the specified
tolerance of 2 mm has been achieved for all 4 measured drafts resulting in a trim angle and
roll angle equal to 0.
The distribution of weights can be observed in Figure 5.
17

Figure 5: Ship model with weights


This set of experiments relies on capturing the sway forces and yaw moment as well as the
resistance force acting on the ship. To this end the model will be attached to a hydrodynamic
balance which is capable of measuring efforts in all 6DOF of a ship. On top of this the
balance is also capable of measuring trim, vertical motion and roll angle. In the present study
roll motion will be blocked, the ship being free to pitch and heave.

Figure 6: Hydrodynamic balance


Available at http://www.anast.ulg.ac.be/menurecherche-141/bassin-de-carenes-labo/equipements

The balance, presented in Figure 6 is connected to an acquisition software that is also capable
of processing the signal in order to obtain concrete values of effort in SI units. To calculate
the moment around Z axis the software uses the distance between the 2 captors installed in
lateral direction. The balance along with the test model is attached to a carriage capable of a
maximum speed of 5 m/s.

3.2. Towing tank facilities and preparation


Experimental runs have taken place at the towing tank of University of Liege. The tank has a
length of 100 m, and a width of 6 m. It is equipped with a wave generator, however, this
18

facility will not be used in the present study. A main overview of the facility can be seen in
Figure 7.

Figure 7: Towing tank at University of Liege (ANAST).


Available at
http://www.anast.ulg.ac.be/menurecherche-141/bassin-de-carenes-labo/installation

To simulate the restricted conditions in which the studied ship is navigating a false bottom
made of aluminium along with a lateral wall comprised of 32 PVC panels each with a length
of 1 m each have been installed as it can be seen in Figure 8. The panels are aligned parallel to
the ship at various distances to simulate a ship passing through a restricted waterway. The
panels can be made vertical of inclined at 45 degrees depending on what test case is
employed. Before installing the panels, the water had to be drained from the towing tank so
that a supporting structure holding the false bottom will be installed. The false bottom is
comprised of aluminium blocks with a width of about 40 cm and a length equal about 6m so
that they can fit in the installed supporting structure.
The PVC panels have to be correctly positioned relative to the ship so that they form a
straight line parallel to the centreline of the towing tank. Markings have been applied on the
false bottom at the tested distances so that the panels can be moved between tests. Before each
test an additional check was performed where the model was brought near the panels and,
19

using a ruler, the distance between each panel and the ship has been measured to make sure
that the panels are positioned correctly. For each test the same procedure will be applied.

Figure 8: False bottom and vertical wall in the towing tank

3.3. Experimental configurations


The tests were performed according to ITTC procedure 7.5-02-06-02 Captive Model Test.
This implies that Froude similitude will be used.
Multiple setups have been performed where parameters such as distance to the nearest bank,
ship speed and shape of bank were varied in the scope of understanding how the sway force
and yaw moment change. The lateral distance is measured from the side of the ship at miships
to the bank and will be further noted yB throughout the present paper. Table 3 and 4 presents
the values of the parameters previously mentioned:
20

Table 3: Ship’s forward velocities tested at real and model scale


Speed at real scale Speed at model scale Froude number
1.65 m/s 0.33 m/s 0.04534
2.2 m/s 0.44 m/s 0.06045
2.75 m/s 0.55 m/s 0.07557
3.3 m/s 0.66 m/s 0.09068
3.75 m/s 0.75 m/s 0.10305
4.15 m/s 0.83 m/s 0.11404

Table 4: Vertical distances to bank at model and real scale


yB at real scale yB at model scale Case id
2m 0.08 m Case 1
3m 0.122 m Case 2
5m 0.2 m Case 3
7.5 m 0.3 m Case 4
10 m 0.4 m Case 5

A number of cases where the bank has is inclined at 45 degrees have also been tested. In this
case yB is defined as in Figure 9 :

Figure 9: Definition of yB in case of inclined banks


21

In the case of the inclined wall tests were performed for 3 distance to bank cases : 0, 10 cm
and 20 cm at model scale, respectively 0, 2.5 m and 5 m at real scale.
We can consider that for each distance we have 6 velocities to be tested. For easier referenting
each yB value will be regarded as a separate case.
According to Lataire (2014) the influence of a tank wall is significant if it is smaller than:
yinfl = 5B(Frh+1) (3)
Frh=v/sqrt(gh) (4)
where B is the beam of the ship and h is the water depth.
In order to make sure that the tank wall that is further enough away from the ship (in the case
of the actual tests it is going to be the portside wall), y infl has been calculated as seen in Table
5:

Table 5: Influence of portside wall


Speed Frh yinfl
0.55 m/s 0.393 1.393
0.66 m/s 0.471 1.471
0.75 m/s 0.535 1.535
0.83 m/s 0.593 1.593

Figure 10: Experimental setup and coordinate axis used in towing tank testing

Considering the experimental setup presented in Figure 10 and the fact that the model had
remained fixed throughout the duration of the experiments being attached to the carriage
which is positioned along the centreline of the towing tank it can be said that the portside wall
22

does not have significant influence on the efforts measured during tests since the actual
distance yship is equal to 3m for all the considered tests.
Test procedure is a bit different from the usual resistance test. To better explain it, the towing
has been split into 3 parts, as shown in Figure 11 :

Figure 11 : Test phases

The first phase of the test is dedicated to the acceleration phase. The model will accelerate
from standstill until the desired speed and the efforts will stabilize. The goal of this phase is to
obtain no sway force or yaw moment from the measurement before entering the confined
area. The second phase is the restricted area where efforts are expected to vary but also
stabilize to a point where an average measurement can be taken. The measurement will start
when the bow of the ship is close to the confined zone and stop when the whole ship has
exited the domain. From the resulting measurement an average value of the quantities of
interest will result. The last part of the towing tank will be used for deceleration and
completely stopping the carriage.
Each individual setup will be repeated several times in order to get a stabilized value and
reduce the measuring error. The final value for the quantities of interest will be the average of
these repeated tests.

3.4. Presentation and interpretation of the experimental results


As stated in Chapter 3.3. 6 different values of velocity were taken into account. However,
after results were processed it was concluded that for 0.33 m/s and 0.44 m/s no sway force or
yaw moment were recorded. Therefore, those results will not be discussed nor shown even
though tests were performed.

3.4.1. Vertical banks cases


Following the procedure described in Chapter 3.3. the resulting values for sway force and
yaw moment are presented in Table 6:
23

Table 6: Values of sway force after experimental tests in the case of vertical banks
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 1.424 N 1.584 N 1.073 N 0.773 N 0.499 N
0.66 m/s 2.123 N 2.061 N 1.281 N 1.075 N 0.818 N
0.75 m/s 2.801 N 2.579 N 1.631 N 1.536 N 1.180 N
0.83 m/s 3.603 N 3.219 N 2.179 N 1.570 N 1.399 N

Lateral force
4.0

3.5

3.0

2.5
Force, Fy[N]

2.0

1.5

1.0

0.5

0.0
5 10 15 20 25 30 35 40 45
Distance to wall [cm]

0.55 0.66 0.75 0.83

Figure 12 : Sway force evolution in the case of vertical banks

Each line in Figure 12 shows a different speed and it also shows that the sway force
magnitude increases as the distance to wall descreases and as the velocity increases. This is an
expected behaivour since the literature and common engineering knowledge of
hydrodynamics tells us that is the way it should happen. However, looking at the values
recorded at a distance of 30 cm we can see that there is little difference between sway force at
a velocity of 0.75 m/s and at 0.83 m/s. This can be attributed to the fact that the measuring
devices as well as the interval chosen for finding the steady-state value of the sway force
24

carry some errors: first of measurement and then human error. As previously stated when
searching for an average value an operator ( in this case the author) has to manually select the
interval from which the value of the quantities of interest will result. Human intervention in
the process leads to small errors in the final results , but it is important to notice the overall
trend and how the values change when parameters are varied. The most important aspect
resulting from the above results is that the sway force is an attraction force in all studied
cases. Table 7 shows the values for yaw moment obtained.

Table 7: Values of yaw moment after experimental tests in the case of vertical banks
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s -0.799 Nm -0.299 Nm -0.853 Nm -0.799 Nm -0.674 Nm
0.66 m/s -1.074 Nm -0.407 Nm -1.177 Nm -1.068 Nm -1.102 Nm
0.75 m/s -1.369 Nm -0.625 Nm -1.505 Nm -1.504 Nm -1.475 Nm
0.83 m/s -1.872 Nm -1.611 Nm -2.179 Nm -1.839 Nm -2.020 Nm

Yaw moment
0.0
5 10 15 20 25 30 35 40 45

-0.5
Yaw moment[Nm]

-1.0

-1.5

-2.0

-2.5
Speed[m/s]

0.55 0.66 0.75 0.83

Figure 13: Yaw moment evolution in the case of vertical banks

Figure 13 shows the variation of the yaw moment with the distance to bank. Yaw moment
mainly varies with the speed as each lines in the figure above shows a different tested
velocity. The variation due to the variation of the yB distance is small. However, the yaw
moment shows that for a distance of 12.2 cm a reduction in magnitude happens at all
considered speeds. This behaivour, although much less significant can be seen for the sway
25

force as well. This can be attribuited to the fact that at this distance from the bank to the ship a
destructive interferance between the ship’s wave system and the reflected wave coming from
the bank happens, thus reducing the magnitude of the pressure in the area between the hull
and the bank.

3.4.2. Inclined wall cases


The tests were performed for the case of inclined banks at 45 degrees with 3 different
distances, yB tested:

Table 8: Values of sway force after experimental tests in the case of inclined banks
Distance to bank(yB)
Velocity 0 cm 10 cm 20 cm
0.55 m/s 1.293 N 0.898 N 0.200 N
0.66 m/s 2.006 N 1.661 N 0.775 N
0.75 m/s 2.949 N 1.989 N 1.196 N
0.83 m/s 3.951 N 2.568 N 1.997 N

Sway force
4.5
4.0
3.5
Sway force, Fy[N]

3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 3 6 9 12 15 18 21
Distance to wall [cm]

0.55 0.66 0.75 0.83

Figure 14: Sway force evolution in the case of inclined banks

Each graph line in Figure 14 represents a tested velocity and it shows the variation of the
sway force depending on the 2 main parameters varied : velocity and distance to bank (yB).
As in the case of the vertical banks the magnitude increases as velocity increases and as yB
26

decreases. In this case values follow the same trend, the graph lines depicting various speeds
being almost paralell.

Table 9: Values of sway force after experimental tests in the case of inclined banks
Distance to bank(yB)
Velocity 0 cm 10 cm 20 cm
0.55 m/s -0.64 Nm -0.75 Nm -0.59 Nm
0.66 m/s -0.87 Nm -1.24 Nm -1.01 Nm
0.75 m/s -1.03 Nm -1.55 Nm -1.39 Nm
0.83 m/s -1.24 Nm -2.33 Nm -2.38 Nm

Yaw moment
0.00
0 3 6 9 12 15 18 21

-0.50
Yaw moment [Nm]

-1.00

-1.50

-2.00

-2.50

-3.00
Distance to bank [cm]

0.55 0.66 0.75 0.83

Figure 15: Yaw moment evolution in the case of inclined banks

As in the case of vertical banks , yaw moment varies mainly with the variation of velocity.
For the highest speed tested (0.83 m/s) it can be noticed that a significant change in the
magnitude of yaw moment happens when the distance to the bank (yB) increases from 0 to 10
cm while for the other velocities it does not. The variation due to the variation of yB can be
27

considered insignificant except for the case where velocity is equal to 0.83 m/s. Contrary to
expectations the magnitude increases as distance to bank increases for all tested cases.
28

4.SHIP-RIVER BANK INTERACTION ANALYSIS BY CFD


SIMULATIONS
The main CFD tool used in this project is the FINE™/Marine package which includes the grid
generator, HEXPRESS™, the ISIS-CFD flow solver and CFView™, a viewing tool used to
visualize the flow and other important quantities.
The goal of the CFD simulations was to evaluate how the software can handle this type of
phenomena and to compare the results coming from CFD analysis with the ones coming from
experimental tests. Simulations were performed on both model scale (replicating the
experimental configuration) but also on real scale by extrapolating the experimental
configuration. The main aspect to look out for is the scale effect which might appear for some
particular cases. The main difficulty for these type of simulations is the computational
resource required and the number of simulations in order to get enough data.

4.1. CFD setup description


HEXPRESS™ is the tool included in FINE™/Marine package which is used to generate the
grid or mesh as it is commonly known. As stated in its own documentation:
” It generates non-conformal body-fitted full hexahedral unstructured meshes on complex
arbitrary shaped geometries. In addition, the advanced smoothing capability provides high-
quality boundary layers insertion.”
The grid is generated in 5 main steps as shown in Figure 16:

Figure 16 : Mesh generation steps in HEXPRESS

First step is creating the initial mesh. The software creates an isotropic mesh where the
domain is split by changing the values of the intial cell size in X, Y and Z direction. The
29

second step is “Adapt to geometry” where the initial mesh is adapted such that geometry
dependent refinements are added. The third step “Snap to geometry” projects the adapted
mesh onto the geometry describing the edges and corners of the geometry. The fourth step is
“Optimize” where the mesh obtained from the previous step is optimized in a way that
concave or negative cells are transformed in convex ones. The last step is the insertion of
viscous layers . In this step high aspect ratio cells can be inserted tangentially to the wall in
order to corectly resolve the boundary layer.
The mesh consists of a simple case setup: a single domain where the ship is positioned at the
desired position relative to the bank and a steady state solution is expected. The domain along
with the global coordinate system is defined as it can be seen in Figure 17:

Figure 17: Domain setup in FineMarine

The coordinate system is defined as follows: postive X axis is placed in the direction of the
advancing ship , positive Y is towards the bank and Z axis is positve upwards. It is important
to mention that a positive yaw moment when the bow rotates towards the bank.
By going through the steps detailed in chapter 4.1. the mesh generation will be performed.
The initial mesh has the following subdivision of the domain bounding box:
30

Table 10 : Initial mesh parameters


Direction Number of cells along the axis
X 27
Y 8
Z 4

Although there is a single domain that will be meshed 2 separate entities can be defined : the
ship and the boundaries of the domain. The number of refinements will differ from the case of
the ship, free surface and the boundaries. The names that will be further used in this chapter
are presented in Figure 18:

Figure 18: Domain surface names

The number of refinements for each surface are stated in Table 11:

Table 11: Number of refinements for each surface of the computational domain
Surface name Number of surface refinements
hull 6
deck 5
Free surface 8
xmin 0
31

xmax 0
ymin_forward 6
ymin_aft 4
ymin_air 0
ymax 0
zmin_forward 6
zmin_aft 4
zmax 0

Additionally, a volume refinement box has been added in order to accurately capture the flow
between the hull and the bank as it can be seen in Figure 19. The number of refinements for
this box was chosen as 6 after a separate grid convergence study.

Figure 19 : Refinement box

In order to capture the boundary layer viscous layer were added to the interest surfaces while
on the other surface no viscous layers were added. The stretching ratio was kept contstant at a
value of 1.2. The values used for viscous layers insertion are presented in Table 12:
32

Table 12: Viscous layers


Surface name First layer thickness Number of viscous layers
hull 0.0024 m 6
zmin_forward 0.0033 m 5
ymin_forward 0.0033 m 5

After generating the mesh the number of cells is about 10 to 11 milion depending on the
simulated case as it can be seen in Figure 20 :

Figure 20 : Generated mesh

The next is defining the simulation parameters. One of the most important parts is defining
the boundary conditions. The solid boundaries are presented in Table 13 while the external
boundaries are presented in Table 14:

Table 13 : Boundary conditions for solid surfaces


Surface name Boundary condition
hull No slip
deck Slip
zmin_forward Wall function
zmin_aft Wall function
ymin_forward No slip
ymin_aft Wall function
33

Table 14 : Boundary conditions for external boundaries


Surface name Boundary condition
zmax Prescribed pressure
xmax Far field
xmin Far field
ymin_air Far field
ymax Far field

As for the motion of the ship , all rotatations (roll , pitch ,yaw) and sway motion were fixed,
the ship being able to heave. The forward motion was imposed as a ½ sinusoidal ramp
acceleration from standstill to the desired velocity afterwards being kept constant at said
speed.
In this paper a 2 equations model, k-ω SST Menter turbulence model will be used. The
principle of the model is to use k-ε model proposed by Jones and Launder (1972) far from the
solid walls and the k-ω Wilcox model proposed by Wilcox (1988) elsewhere.
The solver used will be ISIS CFD flow solver developed by MEETHRIC (Modelisation
Ecoulement Turbulent Haut Reynolds Incompressible Couplage) research group from Ecole
Centrale Nantes. The governing equations are the mass conservation (Eq.5), momentum
conservation (Eq.6) and volume fraction conservation (Eq.7).
❑ ❑
∂ (5)
∫ ρV +∫ ρ (U⃗ −⃗
∂t V
U d ) ∙ ⃗n dS=0
S
❑ ❑ ❑ ❑
∂ (6)
∫ ρU i dV +∫ ρ U i (⃗
∂t V
U d ) ∙ n⃗ dS=∫ ( τ ij I j− p I i ) ∙ n⃗ +∫ ρ gi dV
U −⃗
S S V
❑ ❑
∂ (7)
∫ c dV +∫ c i (⃗
∂t V i
U −⃗
U d ) ∙ ⃗n dS=0
S

where V is the control volume, bounded by the closed surface with the unit outward normal
vector n⃗ moving with at velocity ⃗
U d .⃗
U and p represent the velocity and pressure fields
respectively.τ ij and gi are the components of the viscous stress tensor and the gravity vector
and I j is a unit vector. ci represents the volume fraction of the ith volume fraction of the fluid i
with ci = 1 denoting the presence of the fluid and ci = 0 denoting the absence of fluid.
The physical flow properties of the fluid (µ and ρ) are defined setting the following
constitutive equations:
34

ρ=∑ c i ρi (8)
i

µ=∑ ci µ i (9)
i

∑ ci =1 (10)
i

Assuming the density of the fluid remains constant, the mass conservation equation is
simplified using generalized Gauss’ theorem as:
∇∙⃗
U =0 (11)

Since it is not yet possible to directly solve the turbulence in case of industrial applications,
the Reynolds decomposition is introduced where a fluctuating quantity is decomposed as a
mean value and a fluctuating value:
'
Φ=Φ+Φ (12)

where Φ is the mean value and Φ ' is the fluctuation


By averaging Navier-Stokes equations using the Reynolds decomposition a new term, called
Reynolds stress appears resulting in a total of 10 unknowns in 4 equations. This term contains
the time average product of velocity fluctuations, the 6 new unknowns.

τ ij =τ l + τ t
ij ij
(13)
' '
τ l =2 µ Sij ∧τ t =ui u j (14)
ij ij

where S is the mean stress rate tensor.


In order to model turbulence, it is required to define Reynolds stresses in terms of known
quantities. Isotropic turbulence closures are often based on Boussinesq hypothesis related to
eddy viscosity which implies a simple relationship between Reynolds stresses and velocity
gradients. Then the turbulent terms of the mean strain-rate is defined as:

2 ρK δ ij (15)
τ t =2 µt Sij −
ij
3
Where µt is the turbulent viscosity and K is the kinetic energy.
The values of K and µt are obtained using eddy viscosity turbulence models by solving
transport equation. Various models exist in the literature and are classified by the number of
transport equations solved in addition to the RANS equations:
35

 0 equation or algebraic models


 1 equation models: Spalarat-Allmaras as an example
 2 equations models: k-ε, k-ω
 3 equations models: k-ε
 4 equations models: v2-f model

4.2. HPC Resource


The computations are performed on Liger HPC cluster. Liger is the largest computational
resource available in western France. The cluster has 252 nodes each composed of 24 cores.
Job submission is controlled via a SLURM manager which deals with the priority of jobs and
availability of computational resources at any given time. To submit a job a .slurm script has
been created by the author in which the number of nodes and number of processors for each
node along with other information about the computational job are specified.
Therefore, choosing an adequate number of processors to run the computations is key. A very
high number will require a big allocation from the SLURM manager and does not guarantee a
faster computation time since the time for the processors to communicate between them has to
be taken into account. From experience 40000 to 80000 cells per processor is considered an
optimal value. This implies values between 138 and 275 processors for each computation. To
accommodate Liger’s 24 processors per node, 11 nodes accounting for 264 cores will be used
for each computational job.

4.3. Presentation and interpretation of numerical results

4.3.1. Model scale tests


The same experimental setup was replicated using FineMarine package and the computational
runs have been performed for each of the tested cases. In the case of vertical banks the steady-
state values are presented in Table 15 and Table 16 :

Table 15: Values of sway foce after numerical tests in the case of vertical banks
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 1.780 N 1.940 N 1.600 N 1.140 N 0.930 N
0.66 m/s 3.090 N 2.840 N 2.310 N 1.710 N 1.400 N
0.75 m/s 4.120 N 3.820 N 3.100 N 2.430 N 2.020 N
36

0.83 m/s 4.530 N 4.670 N 3.980 N 3.210 N 2.630 N

Sway force
6.0

5.0

4.0
Sway froce, Fy[N]

3.0

2.0

1.0

0.0
5 10 15 20 25 30 35 40 45
Distance to bank[cm]

0.55 0.66 0.75 0.83

Figure 21: Sway force evolution in the case of vertical banks –numerical results

Figure 21 shows that the predicted sway force depends on speed and yB as expected.
Magnitude decreases when the ship is further away from the bank and increases as it gets
nearer. Also as the velocity increases , the magnitude of the force increases. The sign of the
force is positive which means that the sway force is always pulling the ship towards the bank
(attraction) as found by experimental results.

Table 16: Values of yaw moment after numerical tests in the case of vertical banks
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s -1.97 Nm -1.74 Nm -1.75 Nm -1.73 Nm -1.88 Nm
0.66 m/s -2.04 Nm -2.01 Nm -2.17 Nm -2.17 Nm -2.3 Nm
0.75 m/s -2.13 Nm -2.23 Nm -2.46 Nm -2.56 Nm -2.67 Nm
0.83 m/s -2.48 Nm -2.46 Nm -2.95 Nm -2.99 Nm -3.22 Nm
37

Yaw moment
-1
5 10 15 20 25 30 35 40 45

-1.5
Yaw moment, Mz[Nm]

-2

-2.5

-3

-3.5
Distance to bank[cm]

0.55 0.66 0.75 0.83

Figure 22: Yaw moment evolution in the case of vertical banks –numerical results

Figure 22 shows the evolution of yaw moment as a function of the distance to bank. It can be
said that the evolution mainly depends on velocity, the magnitude increases as velocity
increases. However, it seems that as the yB increases, the magnitude also increases slowly
contrary to the behaviour of the sway force.

4.3.2. Real scale tests


In order to get a good understanding of the magnitude, sign and how the forces and moments
evolve on the case of a real ship navigating in close proximity to the banks, the same
numerical setup was used to predict forces and moments at the real scale by scaling the
domain and adapting values for the boundary layer thickness (viscous layers) trying to model
y+ at a value around 90, the same value used for model scale numerical tests.

Table 17: Sway force values for the real ship after numerical computations
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
2.75 m/s 34365 N 34819 N 25945 N 18663 N 16810 N
3.30 m/s 52430 N 55757 N 37725 N 26645 N 24026 N
3.75 m/s 70473 N 71391 N 50739 N 35686 N 31870 N
4.15 m/s -53029 N 87224 N 64365 N 49671 N 41236 N
38

Sway force
100000
80000
60000
Sway froce, Fy[N]

40000
20000
0
1 2 3 4 5 6 7 8 9 10 11
-20000
-40000
-60000
-80000
Distance to bank[m]

2.75 3.3 3.75 4.15

Figure 23: Sway force evolution for the real ship after numerical computations

Figure 23 shows the evolution of sway force in the case of the real ship. The force follows the
same trend as previously stated: force increases as velocity increases and as yB decreases.
Analysis showed that no scale effects were present except for the case where the ship is
located at the smallest distance and at the highest speed tested: 2m (8cm at model scale) and
4.15 m/s (0.83 m/s). Except this particular case the extrapolation followed the dimensional
analysis, as stated by Eq. 16, with deviations of up to 20 %:
3
F realscale=F modelscale∗S (16)
where S is the scale used, in this case 25.
This particular case can be considered the most critical one for this study since the ship is
navigating very close to the bank at a speed equivalent to 15 knots. Since full scale data is
unavailable for this ship and overall scarce in the literature validating any of the full scale
results is difficult.

Table 18: Yaw moment values for the real ship after numerical computations
Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
2.75 m/s -549.65 kNm -559.33 kNm -515.48 kNm -485.17 kNm -507.69 kNm
3.30 m/s -643.66 kNm -618.57 kNm -600.22 kNm -560.83 kNm -583.45 kNm
3.75 m/s -704.11 kNm -639.22 kNm -696.51 kNm -683.14 kNm -609.97 kNm
4.15 m/s 533.14 kNm -684.54 kNm -745.76 kNm -743.72 kNm -712.66 kNm
39

Yaw moment
800
600
400
Yaw moment, Mz[kNm]

200
0
1 2 3 4 5 6 7 8 9 10 11
-200
-400
-600
-800
-1000
Distance to bank[cm]

2.75 3.3 3.75 4.15

Figure 24: Yaw moment evolution for the real ship after numerical computations

Table 18 and Figure 24 show the values and the evolution of yaw moment after numerical
tests in the case of the real ship. Yaw moment show little dependency regarding distance to
bank, the value being almost constant regardless of how far the ship is positioned relative to
the bank. The main parameter that influences the magnitude of yaw moment is velocity. It is
important to notice that for the considered most critical case yaw moment changes sign
compared to the other tested cases.
40

5. COMPARISON BETWEEN EXPERIMENTAL AND NUMERICAL


RESULTS. INTERPRETATIONS
In order to better use CFD as an estimation tool it is necesarry to compare numerical results to
experimental ones.

5.1. Sway force comparison in the case of vertical banks


For each yB value values of sway force coming from experimental results and CFD
simulation will be compared. The relative difference in percentages will be shown as a
measure of the difference between the results and will be calculated relative to the towing
tank result as shown in Eq. 17:

Rd=
| |
valuetowing tank −valueCFD simulation
value towing tank
∗100
(17)

5.1.1. Case 1 comparison


Table 19: Comparison between towing tank results and CFD result for yB=8cm for sway force
Velocity Towing tank results Simulation results Relative difference, Rd
0.55 m/s 1.424 N 1.781 N 25.07 %
0.66 m/s 2.123 N 3.092 N 45.64 %
0.75 m/s 2.801 N 4.125 N 47.30 %
0.83 m/s 3.603 N 4.765 N 32.25 %

Sway force comparison yB=8


6
5
4
Fy(N)

3
2
1
0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85
Speed(m/s)

Towing tank Simulation

Figure 25: Comparison between evolution of sway force between simulation and experimental results
at yB=8cm
41

Figure 25 and Table 19 show that, in this case, CFD gives an overestimation of sway force for
all investigated cases. As for experimental results , simulation results show the exact same
trend and same sign throughout all tested velocities. A maximum of 47.3 % relative difference
was found meaning a resonable prediction considering the complexity of the flow between the
hull and the ship at such small distance between them.

5.1.2. Case 2 comparison


Table 20: Comparison between towing tank results and CFD result for yB=12.2cm for sway force
Velocity Towing tank results Simulation results Relative difference, Rd
0.55 m/s 1.584 N 1.941 N 22.54 %
0.66 m/s 2.061 N 2.847 N 38.14 %
0.75 m/s 2.579 N 3.823 N 48.24 %
0.83 m/s 3.218 N 4.675 N 45.25 %

Sway force comparison yB=12.2


5

3
Fy(N)

0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85
Speed(m/s)

Towing tank Simulation

Figure 26: Comparison between evolution of sway force between simulation and experimental results
at yB=12.2 cm

It can be noticed from Figure 26 that the resulting graphs have the same overall shape and
follow the same trend line. A maximum relative difference of 48.24 % can be noticed where
the velocity is 0.75 m/s. The values obtained by CFD show over-prediction for all cases.
42

5.1.3. Case 3 comparison

Table 21: Comparison between towing tank results and CFD result for yB=20cm for sway force
Velocity Towing tank results Simulation results Relative difference, Rd
0.55 m/s 1.073 N 1.612 N 50.23 %
0.66 m/s 1.281 N 2.315 N 80.72 %
0.75 m/s 1.631 N 3.123 N 91.52 %
0.83 m/s 2.179 N 3.983 N 82.83 %

Sway force comparison yB=20

3
Fy(N)

0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85
Speed(m/s)

Towing tank Simulation

Figure 27: Comparison between evolution of sway force between simulation and experimental results
at yB=20 cm
In this case Table 21 shows a worse prediction by CFD simulations compared to previous
cases (yB=8cm and yB=12.2cm) , a maximum relative difference of 91.52 % being recorded.
But it can be noticed in Figure 27 that a reasonable prediction in terms of qualitative sway
force evolution was achieved while the magnitudes are always over-predicted .
5.1.4. Case 4 comparison
Table 22: Comparison between towing tank results and CFD result for yB=30cm for sway force
Velocity Towing tank results Simulation results Relative difference, Rd
0.55 m/s 0.773 N 1.142 N 47.74 %
0.66 m/s 1.075 N 1.716 N 59.63 %
0.75 m/s 1.536 N 2.435 N 58.58 %
0.83 m/s 1.570 N 3.211 N 104.52 %
43

Sway force comparison yB=30


4

3
Fy(N)

0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85
Speed(m/s)

Towing tank Simulation

Figure 28: Comparison between evolution of sway force between simulation and experimental results
at yB=30cm
In this case, shown in Figure 28 , experimental results show a nearly similar value for 0.75m/s
and 0.83 m/s while simulation results are consistent with previously shown results in terms on
how the values are evolving. As already stated there can be errors asociated with experimental
results which lead to high Rd values as shown in Table 22. Apart from the value at 0.83 m/s
the relative difference is about 50-60 % with over-prediction in case of CFD simulations.

5.1.5. Case 5 comparison


Table 23: Comparison between towing tank results and CFD result for yB=40 cm for sway force
Velocity Towing tank results Simulation results Relative difference, Rd
0.55 m/s 0.499 N 0.937 N 87.90 %
0.66 m/s 0.818 N 1.420 N 73.66 %
0.75 m/s 1.180 N 2.021 N 71.27 %
0.83 m/s 1.399 N 2.633 N 88.21 %
44

Sway force comparison yB=40


3.0

2.5

2.0
Fy(N)

1.5

1.0

0.5

0.0
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85
Speed(m/s)

Towing tank Simulation

Figure 29: Comparison between evolution of sway force between simulation and experimental results
at yB=40cm

As distance from bank increases an increase of relative difference can be noticed as shows
Figure 29. The numerical setup tends to over-predict the values of sway force even more,
especially when values coming from experimental runs become smaller and smaller. In turn
such small values coming from experimental results carry inherent errors of measuring and
data processing. In this case the relative difference is in the range of 70-90 % for all tested
cases. Again CFD runs show qualitative results with over-prediction in terms of effective
values of sway force.

Table 24: Relative difference(Rd) for sway force


Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 25.07 % 22.54 % 50.23 % 47.74 % 87.90 %
0.66 m/s 45.64 % 38.14 % 80.72 % 59.63 % 73.66 %
0.75 m/s 47.30 % 48.24 % 91.52 % 58.58 % 71.27 %
0.83 m/s 32.25 % 45.25 % 82.83 % 104.52 % 88.21 %

Table 24 summarizes all relative differences found between experimental results and
simulation results. CFD provided very good qualitative results, but poorer quantitative results
being an over-prediction of sway force by a margin of 25-100% relative to experimental
results. A possible source of these differences can be that in the CFD simulation boundary
45

layer was not fully solved since y+ value was around 90 for the hull as it can be seen in Figure
30 . A fully resolved boundary layer where y+ is smaller than 1 can lead to significantly better
results, but comes at a huge computational cost. For each computation the number of cells is
already approximately 11 million. A fine mesh where y+ is smaller than 1 would cause that
number to increase in a rapid way leading to a high need of computational resources and time.

Figure 30: Y+ distribution on the hull

5.2. Yaw moment comparison in the case of vertical banks


The same procedure has been applied for yaw moment values. The results are presented in
Table 25.

Table 25: Relative difference(Rd) for yaw moment


Distance to bank(yB)
Velocity 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 146.56 % 481.94 % 105.16 % 116.43 % 178.79 %
0.66 m/s 90.03 % 393.86 % 84.37 % 103.18 % 108.65 %
0.75 m/s 55.64 % 256.61 % 63.49 % 70.27 % 81.08 %
0.83 m/s 32.48 % 52.70 % 35.41 % 62.59 % 59.41 %
46

Yaw moment prediction by CFD analysis is significantly worse than sway force when
compared to experimental results. An upside is that simulation always provided the right sign
of yaw moment, which in turn means that the overall flow is predicted well. In both
experimental and numerical tests yaw moment was bow away. However, values differ with as
much as almost 500% from experimentally obtained ones with over-prediction happening in
all tested cases. As in the case of sway force simulations showed good qualitative prediction,
but worse quantitative prediction of yaw moment. A possible source of these discrepancies
can be, as in the case of sway force, that the boundary layer is not fully resolved and y+ value
is around 90. Another possible source is the fact that during experimental tests the model was
free to pitch, whereas during simulation the pitch motion was fixed restricting any squat effect
that might have appeared.

Regarding the case where the ship is positioned closest to the bank and going at highest speed
tested a scale effect can be noticed. The origin of this effect is yet unclear but it can be due to
the fact that boundary layer at model scale and at real scale have different thicknesses. As it is
already known a scaled experiment cannot follow Froude and Reynolds similitude at the same
time. Further investigation is needed into the phenomena and real-scale measurements need to
be performed in order to solidify this claim.

5.3. Extrapolating model-scale results


The main purpouse of towing tank tests is to evaluate the forces and moments acting on the
real ship. In the vast majority of experimental tests Froude similitude is used. By using this
process and also by using dimensional analysis it is possible to predict real-scale values for
forces and moments acting on the hull of a ship :

3
F realscale=F modelscale∗S (16)
4
M realscale=M modelscale∗S (17)

Equations 16 and 17 show how dimensional analysis predicts efforts at real-scale using values
obtained at model scale. Using the values obtained during numerical tests from Table 15 and
Table 17 it is possible to see if any scale effects might appear for sway force. A relative
difference has been calculated for each case Equation 18 :
47

3
(F ¿¿ modelscale∗S −Frealscale )∗100 (18)
RDiff force= 3
¿
F modelscale∗S
4
(M ¿¿ modelscale∗S −M realscale)∗100 (19)
RDiff moment = 4
¿
M modelscale∗S

Table 26 : RDiff for sway force in case of numerical results


Distance to bank at model scale(yB)
Velocity at model scale 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 23.56 % 14.87 % 3.78 % 4.77 % 15.68 %
0.66 m/s 8.59 % 25.65 % 4.52 % 0.28 % 9.83 %
0.75 m/s 9.47 % 19.61 % 4.75 % 6.01 % 0.97 %
0.83 m/s 174.92 % 19.54 % 3.50 % 0.97 % 0.35 %

It can be seen from Error: Reference source not found that values follow dimensional analysis
for all cases, except the most critical one, with a maximum difference of 25%. To be on the
conservative side a safety factor of 1.25 can be used when predicting real-scale values coming
from model-scale values .

Table 27 : RDiff for yaw moment in case of numerical results

Distance to bank at model scale(yB)


Velocity at model scale 8 cm 12.2 cm 20 cm 30 cm 40 cm
0.55 m/s 28.57% 17.71% 24.59% 28.21% 30.87%
0.66 m/s 19.23% 21.22% 29.19% 33.84% 35.06%
0.75 m/s 15.37% 26.62% 27.52% 31.69% 41.52%
0.83 m/s 160.66% 28.76% 35.28% 36.32% 43.34%

Table 27 shows the RDiff values for yaw moment. It can be seen that values do not exactly
follow dimensional analysis with deviations up to 50% for all tested cases except for the most
critical case ( where yB=8 cm and velocity is 0.83 m/s at model scale). Taking the values
shown in the table above , a safety factor of 0.5 should be adopted in order to safely predict
real-scale values.
Based on numerical tests it can be concluded that for predicting real-scale values coming from
model scale tests some safety factors need to be applied as follows :
48

3
F realscale=F modelscale∗S ∗c 1 (20)
4
M realscale=M modelscale∗S ∗c 2 (21)
Where c1= 1.25 and c2 = 1.5.

5.4. Advices for CFD and experimental test to obtain interaction forces and
moments for real cases

Investigating ship-river bank interaction is a very challenging mission. During experimental


tests it has been noticed that the efforts acting on the ship are very small in magnitude. This
fact implies that all measuring devices have to have a small tolerance in order to accurately
capture a precise value. While running tests these devices can be disturbed by the so-called
noise coming from the carriage, the people present on-board the carriage and also machinery
surrounding the carriage. It is therefore needed to have a signal processing device that can
eliminate this noise and process the signal so that a clear view on how the efforts evolve over
time is achieved. An automated or remote-controlled carriage will be a plus in these cases. For
a reasonable prediction each test configuration must be reapeated several times so that any
additional noise or uncertaties will diminish.

In the case of CFD simulations the user must be very careful on how to approach these types
of simulations. There is a compromise between meshing the domain so that y+ is below 1 and
the boundary layer is fully resolved and the boundary layer is modelled with y+ values below
300. The first option is better for accuracy of predicted results but is very computationally
expensive. Also generating such a grid (or mesh) requires a powerfull meshing tool that is
capable of meshing very small gaps (that might appear when critical cases are tested).
Modelling the boundary layer (using viscous layers in present study) is a better option
computationally-wise but can lead to errors of prediction, especially in critical cases where
distances between ship and river bank are small.
49

6. CONCLUSION

In this study, the bank effect was investigated for a ship navigating in limited depth and width
by experimental tests and CFD simulations. The depth was chosen as 2 times the draft and
tests were performed using Froude similitude. In the towing tank of the University of Liege an
experimental configuration was set-up by installing aluminium panels in order to reproduce
restricted bottom and PVC panels to create a bank. A scaled model was prepared and attached
to the towing tank carriage through a 6 components force and moment measuring device, the
ship being towed along the centreline of the tank. Firstly, the bank was made vertical and was
positioned at 5 different distances relative to ship (yB) where tests were performed. Then the
bank was made inclined and another 3 relative distances were tested. After a steady-state was
reached measurements of quantities of interest were taken. Out of these quantities it is
important to mention sway force and yaw moment since these efforts are the unit of
measurement of the influence of the bank. For each yB value, 6 different speeds were tested
on both vertical and inclined banks. After processing the results, it was discovered that the
two lowest velocities namely 0.33 m/s and 0.44 m/s were ruled out since the values recorded
were insignificantly small.

In the case of vertical banks results showed that the sway force is always an attraction force,
pulling the ship towards the bank in all studied cases. The main influence on the value of this
force is, as expected, the distance to bank (yB) and velocity following the fact that as velocity
increases and yB decreases the sway force increases. Yaw moment was found to always be
bow away. A similar behaviour of the evolution of the magnitude of yaw moment was noticed
but with a much slower growth in value due to the decrease of distance to bank. It is worth
mentioning that for a distance of 12.2 cm (3m for real case) a significant reduction of the
magnitude for yaw moment and a lesser reduction for sway force occurs. This can be
attributed to the fact that at this particular distance the wave system generated by the ship and
the reflected wave system coming from the bank result in a destructive interference.

For the case of inclined banks a similar behaivour has been noticed. Sway force was always
an attraction force and yaw moment was bow away for all tested cases. As in the case of
vertical banks the increase in sway force was driven by both the increase in velocity but also
the decrease in distance to bank. Yaw moment increased mainly when velocity increased,
50

showing little increase when distance to bank changed. For all cases where the bank was
inclined an increase of yaw moment was found when distance to bank (yB) increase from 0 to
10 cm (0 to 2.5 m at real scale) with a significant increase being noticed at a velocity of 0.83
m/s (4.15 m/s at real scale).

CFD simulations mimicking the experimental setup were performed using FineMarine
package: HEXPRESS™ as a meshing tool, the ISIS-CFD flow solver and CFView as a tool
for visualising the flow. Simulations were run using HPC resource available at ECN, LIGER.

Comparing experimentally-obtained sway force values and numerically obtained ones it can
be concluded that CFD produced qualitatively good results in the sense that as for
experiments sway force was also always an attraction force during all tested cases. Values
showed an over-prediction of 25-100% above towing tank values. For yaw moment the same
conclusions can be drawn: a good qualitative prediction, CFD always predicting bow away,
same as in experimental result, but with a much worse quantitative prediction. Yaw moment
was always over predicted with relative differences ranging from 35-500%.

Also by extrapolating the numerical setup, real scale CFD simulations were performed using
the same FineMarine package. Predicting real-scale values are really useful for integrating
them into simulators for pilots to get their certification. The analysis showed a good
agreement in terms of sign for sway force and yaw moment with an exception of one
particular case also considered the most critical one. This case might need further
investigation.

Norrbin(1974 and 1985) and Ch’ng et al.(1993) previously provided mathematical models for
predicting ship-bank interaction effect depending on various parameters such as distance to
bank, water depth to draft ratio and ship’s main dimensions. Each mathematical model is
based on a series of towing tank tests for a specific type of ship. These mathematical models
have to be used carrefully because the limits of their applicability are small due to the fact that
the number of models tested is small (1 or 2 ship models) .
All the results presented throughout present study can be used to build or calibrate a
mathematical model that predicts forces and moments induced by the presence of banks or
enhance an already existing model for a better prediction of bank effect with the goal of
building a simulator for pilots to get their certification.
51

6.1. Future work


Ship-bank interaction is a very complex phenomena depending on multiple variables. The
current study can be enhanced by using a different simulation setup or a different CFD code
with the aim of reducing the relative difference between experiments and CFD results. In the
case of experimental tests, propeller action can be taken into account by fitting the model with
a scaled propeller. Another interesting variable which might impact the predicted efforts is the
draft: in this study only one loading case was tested. Also to further analyze the flow around
the ship wave gauges can be installed during towing tank tests. By acquiring this data a
comparison between CFD and experimentally obtained wave profiles along the hull can be
done.
In the current study only one depth to draft ratio was tested. It would be interesting
investigating multiple ratios both experimentally and numerically.
52

7. REFERENCES
Lataire, E.,2014. Experiment Based Mathematical Modelling of Ship-Bank Interaction, Thesis
(PhD). Ghent University
Duffy, J.T,2005. Prediction of Bank Induced Sway Force and Yaw Moment for Ship-
Handling Simulation, SimTecT2005.
Norrbin, N.,1974. Bank effects on a ship moving through a short dredged channel.10th ONR
Symposium on Naval Hydrodynamics.
Ch’ng, P.W., Doctors, L.J., Renilson, M.R., 1993. A method of calculating the ship-bank
interaction forces and moments in restricted water. International Shipbuilding Progress, Vol.
40, No.412, pp. 7-23
Norrbin, N.,1985. Bank clearance and optimal section shape for ship canals. 26th PIANC
International Navigation Congress, Section 1, Subject 1, 167-178.
Vantorre, M., Delefortrie, G., Eloot, K. and Laforce, E.,2003. Experimental investigation of
ship-bank interaction forces
Lataire, E. and Vantorre, M.,2008. Ship-bank interaction induced by irregular bank
geometries, 27th Symposium on Naval Hydrodynamics.
Wang, H., Zou, Z., Xie, Y. and Kong, W., 2010 Numerical study of viscous hydrodynamic
forces on a ship navigating in shallow water, Proceedings of the 20th International Offshore
and Polar Engineering Conference, 523-528.
Kim, Y. and Ng, Eddie.,2017. CFD Study of Ship-to-Bank Interaction. The International
Journal of Maritime Engineering,Vol. 159.
ITTC 2002, 23rd International Towing Tank Conference, Model Manufacture, Ship Models,
7.5-01-01-01
ITTC 2017, 28th International Towing Tank Conference, Captive Model Test, 7.5-02-06-02
NUMECA International , FineMarine User Guide v9.1
Wilcox, David C., 1988. Reassessment of the scale-determining equation for advanced
turbulence models. AIAA Journal 26.11, 1299-1310
Jones, W.P, Launder, B.E. The prediction of laminarization with a two-equation model of
turbulence. International Journal of Heat and Mass Transfer 15.2, 301-314
Fathi Kazerooni, M., Seif, M.S., 2015. Experimental Study of Forces Exerted on Ships Due to
the Vertical Walls of Navigation Channels. The International Journal on Marine Navigation
and Safety of Sea Transportation Vol.9 Number 2
53

Lataire, E., Vantorre, E., Eloot, K., 2009. Systemic Model Tests on Ship-Bank Interaction
Effects. International Conference on Ship Manoeuvring in Shallow and Confined Water:Bank
Effect

You might also like