A Finite Element Method For Netting Application To Fish Cages and Fishing Gear

SPRINGER BRIEFS IN ENVIRONMENTAL SCIENCE
Daniel Priour
A Finite Element
Method for
Netting
Application to Fish
Cages and Fishing
Gear
SpringerBriefs in Environmental Science
For further volumes:

http://www.springer.com/series/8868
Daniel Priour
A Finite Element Method

for Netting
Application to Fish Cages and Fishing Gear
123
Daniel Priour
Ifremer
Plouzané
France
ISSN 2191-5547 ISSN 2191-5555 (electronic)

ISBN 978-94-007-6843-7 ISBN 978-94-007-6844-4 (eBook)
DOI 10.1007/978-94-007-6844-4
Springer Dordrecht Heidelberg New York London
Library of Congress Control Number: 2013938031
Ó The Author(s) 2013

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the
work. Duplication of this publication or parts thereof is permitted only under the provisions of
the Copyright Law of the Publisher’s location, in its current version, and permission for use must
always be obtained from Springer. Permissions for use may be obtained through RightsLink at the
Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Why Fishing Cages and Fishing Gears? . . . . . . . . . . . . . . . . . . 1
1.2 Why the Finite Element Method? . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Why for Netting and Cable? . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Why a Book? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Finite Element Method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1 Principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1 Field of Numerical Points. . . . . . . . . . . . . . . . . . . . . . 4
2.1.2 Finite Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 A Simple Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Nodes Position, Forces on Nodes, and Stiffness Matrix . . . . . . . 6
2.4 Local and Global Forces and Stiffness . . . . . . . . . . . . . . . . . . . 7
2.5 Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.6 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Equilibrium Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1 Newton-Raphson Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.1 One Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.2 Two Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1.3 Several Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.4 Singularity of the Stiffness Matrix . . . . . . . . . . . . . . . . 22
3.2 Other Resolution Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.1 Newmark Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.2 Energy Minimization . . . . . . . . . . . . . . . . . . . . . . . . . 25
4 The Triangular Finite Element for Netting . . . . . . . . . . . . . . . . . . 27

4.1 State-of-the-Art of Numerical Modelling for Nets . . . . . . . . . . . 27
4.1.1 Constitutive Law for Nets. . . . . . . . . . . . . . . . . . . . . . 27
4.1.2 Twine Numerical Method . . . . . . . . . . . . . . . . . . . . . . 27
4.2 The Finite Element for Netting . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2.1 The Basic Method: Direct Formulation . . . . . . . . . . . . 30
4.2.2 Metric of the Triangular Element . . . . . . . . . . . . . . . . 33
v
vi Contents
4.3 The Forces on the Netting . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.3.1 Twine Tension in Diamond Mesh . . . . . . . . . . . . . . . 36
4.3.2 Twine Tension in Hexagonal Mesh . . . . . . . . . . . . . . 40
4.3.3 Hydrodynamic Drag. . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3.4 Twine Flexion in Netting Plane. . . . . . . . . . . . . . . . . . 57
4.3.5 Twine Flexion Outside the Netting Plane . . . . . . . . . . . 59
4.3.6 Fish Catch Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.7 Dynamic: Force of Inertia. . . . . . . . . . . . . . . . . . . . . . 66
4.3.8 Dynamic: Drag Force. . . . . . . . . . . . . . . . . . . . . . . . . 67
4.3.9 Buoyancy and Weight . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3.10 Contact Between Knots . . . . . . . . . . . . . . . . . . . . . . . 69
5 The Bar Finite Element for Cable . . . . . . . . . . . . . . . . . . . . . . . . 71

5.1 Principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2 Tension on Bars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2.1 Force Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2.2 Stiffness Matrix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3 Bending of Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.3.1 Force Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.4 Drag on Cables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.4.2 Definitions of the Variables . . . . . . . . . . . . . . . . . . . . 80
5.4.3 Stiffness of the Normal Force . . . . . . . . . . . . . . . . . . . 81
5.4.4 Stiffness of the Tangential Force . . . . . . . . . . . . . . . . . 84
6 The Node Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6.1 Principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.2 Contact on Bottom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.2.1 Force Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.3 Drag on Bottom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.3.1 Force Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.1 Tractrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.2 Diamond Mesh Netting Stretched by its Weight . . . . . . . . . . . . 93
7.3 Hexagonal Mesh Net Held Vertically in the Current . . . . . . . . . 96
7.4 Hydrostatic Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.5 Cod-End with Catch in the Current . . . . . . . . . . . . . . . . . . . . . 98
7.6 Full Cod-End . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.7 Bottom Trawl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Contents vii
7.8 Cubic Fish Cage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

7.9 Bending of Cable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Chapter 1
Introduction
Abstract This chapter describes few reasons of this book: (i) the marine flexible
structures such as fish cages and fishing gears could require to be studied with numer-
ical modelling for the knowledge of their mechanics, (ii) the finite element method,
well known method in engineering, is the base of the modelling used, (iii) due to the
fact that these structures are mostly made of nettings and cables, these components
are fully described for the finite element method, (iv) a book is a well adapted format
for the description of aspects of the method.
Keywords Marine flexible structures · Finite element method
1.1 Why Fishing Cages and Fishing Gears?
Fish cages and fishing gears are generally quite large, a few tens or hundred meters,
and are very flexible. Engineers and users are still trying to improve their knowledge
of these flexible marine structures. This flexibility leads these structures to have
different behaviours depending on the environment. The classical questions that
arise are the following:
• What is the tension in the mooring line of the fish cage under certain wave, current,
wind, and tide conditions?
• What is the volume reduction of the fish cage in the current?
• How dependant are the horizontal and vertical openings of the trawl on the towing
speed?
• Is that the cables length of the trawl is optimal in terms of fuel consumption?
Several means are available to help engineers and users: observation at full scale,
tests in flume tanks, numerical modelling. Each has its own advantages and draw-
backs:
D. Priour, A Finite Element Method for Netting, SpringerBriefs in Environmental 1

Science, DOI: 10.1007/978-94-007-6844-4_1, © The Author(s) 2013
2 1 Introduction
• Observations give real information, but the observation area is generally very
limited; it is impossible to see the whole structure at the same time.
• Tests in tanks give a lot of information, such as the behaviour of the structure in
waves and in current. The main drawback is probably that the models used in tests
are quite expensive, and this limits the number of tests that can be performed.
• Numerical modellings also give a lot of information, such as the tension in cables
and netting twines, but they do not cover all the phenomena involved in the behav-
iour of the structure, such as wearing between yarns in twines or the plastic defor-
mation of the sea bed.
1.2 Why the Finite Element Method?
Several mechanical modellings of flexible structures have been developed during the
last decades. They are generally based on a decomposition of the structure into small
elements in which approximations can be done. The most well-known modelling
using this technique is the finite element method. This method has been widely used
for mechanical modelling since the 1970s.
1.3 Why for Netting and Cable?
Nettings and cables are the main components of fishing gears and fish cages. Mechan-
ical modellings of the structures are required to assess the behaviour of these com-
ponents. Cable modellings have been described in a few publications [5, 25], but
the modelling of nettings has not been given much attention. For these reasons, we
attempt to fully describe a netting modelling using the finite element method. Even if
the modelling of cables has been largely described, their modelling is also described
here in order to propose a coherent document.
1.4 Why a Book?
Information on this finite element method for netting structures is sparse. There are
portions of books and articles in journals on the topic, but there is no document that
tries to group all the main matter on this subject. This book is a tentative attempt at
such a publication.
Chapter 2
Finite Element Method
Abstract A brief description of the finite element method principle is proposed for
the mechanics of structures. A simple example of the calculation of the perimeter
of a circle is intended to highlight the principle. In case of mechanics purpose, the
relevance of the concept of nodes position, of forces on the nodes and of the stiffness
of the structure is described. The distinction between the local, at the scale of each
finite element, and the global, at the scale of the structure, is carried out. The way, the
symmetry of the structure could be taken into account in the method, is described.
The boundary conditions, a fixed component such an anchor or a fixed speed such a
trawler, are defined in terms of finite element method.
Keywords Description of the finite element method · Symmetry in the finite element
method · Boundary conditions in the finite element method
2.1 Principle
The finite element method is a method that, at first, approximates the characteristics
of a global structure by dividing it into smaller substructures called finite elements.
These approximations, in the present case, are performed to estimate efforts on the
vertices of these elements. These efforts depend on the position of the vertices of
finite elements.
In a second step, these elements are assembled to reconstruct the overall structure
and thus obtain the efforts on this structure. These efforts depend on the overall
position of the vertices of the elements.
In a third step, the position of the vertices that give a zero overall effort is cal-
culated. This position corresponds to the equilibrium position and therefore to the
expected shape of the overall structure.

4 2 Finite Element Method
2.1.1 Field of Numerical Points
A field of nodes on the structure to be studied is first created. This field of numerical
nodes is created so that there are many points in areas of high strain gradient. These
nodes serve as the basis for creating finite elements.
The user is often in a position where he does not know a priori which areas are with
high deformation gradients. The equilibrium positions are calculated successively,
refining by adding nodes in areas with steep gradients and removing nodes in areas
with low gradients.
2.1.2 Finite Elements
Finite elements are created on this field of nodes. These finite elements, in the case of
our model, are of several types, depending on whether they are dedicated to cables,
bars or nets.
Triangular elements are used for nets (Fig. 2.1), since the net is a surface. It seems
easier to use the simplest surface, namely, the triangle. The curvature of the net can
Fig. 2.1 The diamond mesh netting (a) is decomposed into triangular elements (b). The approx-
imation in each triangle is that twines are parallel and therefore have the same deformation, and
that the twines are elastic (Chap. 4 p. 27)
2.1 Principle 5
Fig. 2.2 The cable (a) is decomposed into bars elements (b). The approximation in each bar is that
bars are straight and elastic (Chap. 5 p. 71)
be represented using several triangular elements. Bar elements are used for cables
(Fig. 2.2).
2.2 A Simple Example
The following simple example shows the principle of splitting a global structure into
several finite elements. A circle with a diameter of 1m has a perimeter of π (2πR).
To assess this perimeter by the finite element approach, the circle is divided into n
identical parts (Fig. 2.3). The perimeter is the sum of the length of each circle arc.
The length of the arc can be approximated by the circle cord. Each cord has a length
of 2Rsin( α2 ).
The perimeter of the circle can be assessed by n times each cord length. Figure 2.4
shows the evaluation accuracy of the perimeter in function of the number of sectors
for the approximation. The larger the number of elements, the greater the accuracy.
In other words, a parameter (here the perimeter) can be assessed by dividing the
problem into finite elements (sectors) to be able to make acceptable approximations
(the arc length approximated by the cord length). The parameter is finally assessed
by rebuilding all the finite elements (sum of cord lengths). The principle of the
finite element method is to discretize a structure in small (finite) elements to make
acceptable approximations in each element and rebuild all the finite elements for
assessing parameters on the structure.
Fig. 2.3 Polygon of n cords

inside the circle. The length of
each cord is 2 R sin(α/2). The
circle perimeter is assessed by
n times each cord length
2.3 Nodes Position, Forces on Nodes, and Stiffness Matrix
In case the relationship between efforts on nodes (vertices of the elements) and their
position is established, F(X) is known:
F: force on the nodes (N),
X: node position (m).
The objective of the method is to estimate the equilibrium position (X f inal ), that
is to say, such that
F(X f inal ) = 0
The Newton-Raphson method is generally used to obtain this position (X f inal )
from an initial unbalanced position (Xinitial ). This method iteratively calculates
the position at equilibrium. This method relies on the definition of the following
derivative:
F(X + h) − F(X)
F (X) = (2.1)
h
F : derived efforts with respect to position (N/m),

h: nodes displacement (m) which tends to 0.
The displacement h is sought if X is not the equilibrium position and such that
X + h is in equilibrium. Under these conditions:
F(X + h) = 0
The previous equation of the derivative gives
2.3 Nodes Position, Forces on Nodes, and Stiffness Matrix 7
Fig. 2.4 Perimeter of the polygon (dots) in function of the number of cords (n) compared with the
perimeter of the circle (line). The cross corresponds to the cords in Fig. 2.3
F(X)
h= (2.2)
−F (X)
The term −F (X) is called the stiffness matrix of the structure. Obviously h can be
large, which means that the definition of the derivative is not completely respected.
An iterative calculation is required:
F(Xk )
Xk+1 = Xk + (2.3)
−F (Xk )
k: iteration.
Starting from a position Xk , F(Xk ) and −F (Xk ) are calculated, then the dis-
placement hk is deducted and then the next position Xk+1 . The iterative calculation
is stopped when convergence is achieved, for example when the force F(Xk ) con-
verges to 0.
2.4 Local and Global Forces and Stiffness
In the Chaps. 4, 5, and 6 the forces and the stiffness are described in local terms.
As mentioned earlier, the structure is split into finite elements in which forces
and stiffness are calculated locally. That gives local forces f and local stiffness k. For
example in case of element involving four coordinates, they are as in following:
⎛ ⎞
a
⎜b⎟
f=⎜
⎝c⎠
⎟ (2.4)
d
⎛ ⎞
e f g h
⎜i j k l⎟
k=⎜
⎝m n
⎟ (2.5)
o p⎠
q r s t
To reassemble the finite elements in the global structure, the local forces and the
local stiffness have to be added to the global ones (F, K ).
For example, if f and k define the force and the stiffness on an element that
involves node components 3, 4, 7, and 8, taking this element into account in the
global structure would mean that the local force f and stiffness k have to be added to
the global force F and stiffness K , as in the following:
F(3) = F(3) + a (2.6)

F(4) = F(4) + b (2.7)
F(7) = F(7) + c (2.8)
F(8) = F(8) + d (2.9)
K (3, 3) = K (3, 3) + e (2.10)

K (3, 4) = K (3, 4) + f (2.11)
..
.
K (4, 3) = K (4, 3) + i (2.12)
..
.
K (8, 8) = K (8, 8) + t (2.13)
In other words:
2.4 Local and Global Forces and Stiffness 9
⎛ ⎞
·
⎜ · ⎟
⎜ ⎟
⎜· + a⎟
⎜ ⎟
⎜·+b⎟
⎜ ⎟
⎜ · ⎟
F=⎜⎜ · ⎟
⎟ (2.14)
⎜ ⎟
⎜·+c⎟
⎜ ⎟
⎜· + d ⎟
⎜ ⎟
⎝ · ⎠
·
⎛ ⎞
·· · · ·· · · ··
⎜· · · · ·· · · · ·⎟
⎜ ⎟
⎜· · · + e ·+ f · · ·+g ·+h · ·⎟
⎜ ⎟
⎜· · · + i ·+ j · · ·+k ·+l · ·⎟
⎜ ⎟
⎜· · · · ·· · · · ·⎟
K =⎜
⎜· · ·
⎟ (2.15)
⎜ · ·· · · · ·⎟
⎟
⎜· · · + m ·+n · · ·+o ·+ p · ·⎟
⎜ ⎟
⎜· · · + q ·+r · · ·+s ·+t · ·⎟
⎜ ⎟
⎝· · · · ·· · · · ·⎠
·· · · ·· · · ··
2.5 Symmetry
In the case of symmetrical structures in a symmetrical environment it could be advan-

tageous to use this symmetry to reduce the node number and therefore the computa-
tion times.
Figure 2.5 shows a simple bar with a symmetry plane. The plane of symmetry is
OY Z and only the node of components a, b, and c, is on the plane of symmetry.
The calculation of force vector on the bar P regardless of the symmetry will give
a force such as (cf. Fig. 2.5):
Fa
Fb
F
F= c (2.16)
Fd
Fe
Ff
The stiffness matrix would be:

Fig. 2.5 The bar P has a node (a, b, c) on the symmetry plane. The other node (d, e, f ) is outside
the symmetry plane. The symmetric bar is Q
K aa K ab K ac K ad K ae Ka f
K ba K bb K bc K bd K be Kb f
K K cb K cc K cd K ce Kc f
K = ca (2.17)
K da K db K dc K dd K de Kd f
K ea K eb K ec K ed K ee Ke f
K fa K fb K fc K fd K fe Kff
In this case the ranking of the node coordinates is a, b, c, d, e, f .

The calculation of the total force vector on the bar taking into account the sym-
metry will give a force such as:
Fa − Fa
Fb + Fb
F + Fc
F= c (2.18)
Fd + 0
Fe + 0
Ff + 0
The stiffness matrix would be:
K aa + K aa K ab − K ab K ac − K ac K ad K ae Ka f
K ba − K ba K bb + K bb K bc + K bc K bd K be Kb f
K − K ca K cb + K cb K cc + K cc K cd K ce Kc f
K = ca (2.19)
2.5 Symmetry 11
That gives for a symmetry plane O X Y passing by the node of coordinates

a, b, c:
0
2.Fb
2.Fc
F= (2.20)
Fd
Fe
Ff
2.K aa 0 0 K ad K ae Ka f
0 2.K bb 2.K bc K bd K be Kb f
0 2.K cb 2.K cc K cd K ce Kc f
K = (2.21)
That gives for a symmetry plane OY Z passing by the node of coordinates a, b, c:
2.Fa
0
2.Fc
F= (2.22)
Fd
Fe
Ff
2.K aa 0 2.K ac K ad K ae Ka f
0 2.K bb 0 K bd K be Kb f
2.K bc 0 2.K cc K cd K ce Kc f
K = (2.23)
That gives for a symmetry plane O Z X passing by the node of coordinates

a, b, c:
2.Fa
2.Fb
0
F= (2.24)
Fd
Fe
Ff
2.K aa 2.K ab 0 K ad K ae Ka f
2.K ba 2.K bb 0 K bd K be Kb f
0 0 2.K cc K cd K ce Kc f
K = (2.25)
2.6 Boundary Conditions
There are two kinds of boundary conditions: the mechanical and the geometric.
The mechanical boundary conditions are defined through forces on the structure.
Such boundary conditions could be the effect of the sea bed; for example, a mooring
chain lands on the bottom. This specific case is described in Sect. 6.2 (p. 87).
The geometric boundary conditions consist here in displacement boundary con-
ditions; for example, an anchor in the sea bed could be taken into account by a null
displacement, or a boat towing a gear could be defined with a null displacement in
moving water. These geometric conditions are actually the conditions discussed in
this section.
A null displacement for node coordinate c could be taken into account by modi-
fying the force and the stiffness matrix. Generally speaking, the force and the matrix
stiffness are such as:
Fa
Fb
F
F= c (2.26)
Fd
Fe
Ff
K aa K ab K ac K ad K ae Ka f
K ba K bb K bc K bd K be Kb f
K K cb K cc K cd K ce Kc f
K = ca (2.27)
When the null displacement for node coordinate c is taken into account, the force
and the stiffness matrix become:
2.6 Boundary Conditions 13
Fa
Fb
0
F= (2.28)
Fd
Fe
Ff
K aa K ab 0 K ad K ae Ka f
K ba K bb 0 K bd K be Kb f
0 0 1 0 0 0
K = (2.29)
K da K db 0 K dd K de Kd f
K ea K eb 0 K ed K ee Ke f
K fa K fb 0 K fd K fe Kff
These modifications of force and stiffness matrix ensure that the displacement of
coordinate c is null.
Chapter 3
Equilibrium Calculation
Abstract The modelling of structure mechanics is a matter for finding equilibrium of

the structure. The Newton-Raphson method for equilibrium calculation is described.
This method is based on the nodes position, the forces on nodes, and the stiffness
matrix. Other methods of equilibrium calculation, the methods of Newmark and of
the energy minimisation, are described.
Keywords Equilibrium calculation · Newton-Raphson method · Newmark method ·

Energy minimisation method
3.1 Newton-Raphson Method
Finite element methods generally use the Newton-Raphson method [4] for the calcu-
lation of the equilibrium position of a mechanical structure. The equilibrium position
corresponds to that position of the structure in which the sum of forces equals 0. In
what follows a few simple examples are given to explain the method under three
cases: one dimension, two dimensions and several dimensions.
3.1.1 One Dimension
A spring (Fig. 3.1) equilibrium is reached when the weight is equilibrated by the
spring force. At this position the sum of forces equals 0. This position can be calcu-
lated using the Newton-Raphson method. In this example there is just one dimension:
the vertical position (x) of the mass relatively to the spring fixation which also equals
the length of the spring.
The spring equilibrium is calculated by writing the force on the mass: the weight
is −Mg (N), and the force of the spring is +K x−l 0
l0 (N).

16 3 Equilibrium Calculation
Fig. 3.1 The equilibrium of

the spring is due to the mass
weight and the spring force
With
M: mass (kg),
g: acceleration of gravity (m/s2 ),
K : spring stiffness (N),
x: position of the mass along the spring axis relative to the fixed point of the spring (m),
x: length of the stretched spring (m).
In this example the stiffness is not constant in order to give a clearer explanation of
the Newton-Raphson method. K is equals to Ax. That means that longer the spring
is, the stiffer it is.
The sum of forces on the mass (curve on Fig. 3.2) is
x − l0
F(x) = K − Mg (3.1)
l0
or, following the previous relations,
x − l0
F(x) = Ax − Mg (3.2)
l0
Obviously at the equilibrium F(x) = 0. It is clear that this simple equation has
an analytical solution, which is
√
l0 A (4g M + l0 A) + l0 A
x= (3.3)
2A
The Newton-Raphson method could be used to find the length of the spring (x)
at the equilibrium. This method requires knowing the force and the derivative of the
force relatively to the position.
3.1 Newton-Raphson Method 17
Fig. 3.2 Sum of forces on the mass function of spring length. Three Newton-Raphson iterations
starting at x = 2.8 m are displayed. The vector tangent at x0 is shown
The method is iterative and approximates the force curve by its tangent (shown in
Fig. 3.2). From a position (xk ), the force (F(xk )) and the derivative of force (F (xk ))
are calculated, and a new position (xk+1 ) can be found. This new position is generally
closer to the equilibrium and is calculated as follows:
F(xk )
xk+1 = xk + (3.4)
−F (xk )
Figure. 3.2 shows three iterations with an initial value x0 of the spring length of
2.8 m.
With:
The stiffness A = 1000 N/m,
The mass M = 10 kg,
The acceleration of gravity g = 9.81 m/s2 ,
The unstretched length of the spring l0 = 1 m.
The stretched length at the equilibrium is 1.09 m. That means that the spring
stretches 9 %.
After five iterations the equilibrium is reached or more exactly |F(x)| < 0.1N .
The Fig. 3.2 shows 3 iterations along the curve of force. Figure 3.3 represents the
reduction of the force residue (|F(x)|) with the five iterations.
Fig. 3.3 Residue of force for each Newton-Raphson method iteration
3.1.2 Two Dimensions
In this section a simple example in two dimensions is given (Fig. 3.4): a spring
with two degrees of freedom, i.e., the horizontal (x) and the vertical (y) positions
of the mass relative to the spring fixation. The equilibrium of the system is due to
the position of the mass along the vertical and the horizontal.
Figure 3.5 shows the
variation of the norm of the residue of force Fx + Fy on the mass due to the
2 2
positions along x and y of the mass. The equilibrium point is noted by the largest
dot.
The stiffness (K ) of the spring is not constant: K is equal to Al. That means that
the longer the spring is, the stiffer it is. In this condition the horizontal and vertical
forces on the mass are due to the spring length and the weight of the mass:
x
Fx = T (3.5)
l
y
Fy = T − Mg (3.6)
l
With:
l − l0
T = Al (3.7)
l
0
l = x 2 + y2 (3.8)
In this case the derivative of the forces is calculated relatively to x and y:

Fig. 3.4 Spring with two degrees of freedom: the vertical and horizontal positions of the mass.
The equilibrium is due to the mass weight and the spring force

Fig. 3.5 Norm of the force (Z = Fx2 + Fy2 ) function of mass coordinates (X, Y ). The largest
dot is the equilibrium position. The smallest dots are the Newton-Raphson iterations starting at
x = 0.9 m and y = 1.9 m
∂ Fx l − l0 x2
=A +A (3.9)
∂x l0 ll0
∂ Fx xy
=A (3.10)
∂y ll0
∂ Fy yx
=A (3.11)
∂x ll0
∂ Fy l − l0 y2
=A +A (3.12)
∂y l0 ll0
The Newton-Raphson method accesses the equilibrium solution through itera-

tions. At each iteration the new position is calculated by the following relation:
F(Xk )
Xk+1 = Xk + (3.13)
−F (Xk )
With:
xk
Xk = (3.14)
yk
Fx (X k )
F(Xk ) = (3.15)
Fy (X k )
F(Xk )
The ratio −F (X ) is the displacement h, such as F(Xk ) = −F (Xk )h.
k
With these equations the equilibrium position is assessed (Fig. 3.5). Figure 3.6
represents the reduction of the force residue with the iterations.
3.1.3 Several Dimensions
3.1.3.1 Main Variables
The positions of the nodes are in vector X, the forces on the nodes are in vector F,
and the stiffness matrix is K ; xi and Fi refer to the same node along the same axis.
These variables are as follows:
x1
x2
X= . (3.16)
.
xn

Fig. 3.6 Residue of force ( Fx2 + Fy2 ) for each Newton-Raphson method iteration
F1
F2
F= . (3.17)
.
Fn
− ∂∂ Fx11 − ∂∂ Fx21 . . − ∂∂ Fxn1

− ∂∂ Fx12 − ∂∂ Fx22 . . − ∂∂ Fxn2
K = . . .. . (3.18)
. . .. .
− ∂∂ Fx1n − ∂∂ Fx2n . . − ∂∂ Fxnn
From these three variables the displacement vector (h) can be calculated by solving
the following system of linear equations:
hK = F (3.19)
3.1.3.2 Iterations
As mentioned earlier, the Newton-Raphson-method is an iterative one. The steps are

as follows:
From the position (Xk ) of the nodes resulting from iteration k:
xk1
xk2
Xk = . (3.20)
.
xkn
The force (Fk ) on the nodes and the stiffness (K k ) matrix are calculated:
Fk1
Fk2
Fk = . (3.21)
.
Fkn
K k11 K k12 . . K k1n
K k21 K k22 . . K k2n
Kk = . . . . . (3.22)
. . . . .
K kn1 K kn2 . . K knn
The node displacements (hk ) are calculated:
hk K k = Fk (3.23)
The new position of nodes is deduced:
Xk+1 = Xk + hk (3.24)
3.1.4 Singularity of the Stiffness Matrix
In some cases the stiffness matrix (K ) could be singular. In this case solving
hK = F (Sect. 3.1.3, p. 20) could lead to a very large displacement (h i >> 1)
and to divergence of the method.
An example can be shown with the unstretched horizontal bar of Fig. 3.7. This bar
has two extremities. If the first extremity (on the left on Fig. 3.7) has the horizontal
and vertical coordinates (0, 0), the position vector is:
0
0
X= (3.25)
x3
0
Fig. 3.7 This bar is articulated around its left extremity. A vertical force (F4 ) is applied on the right
extremity. This unstretched bar displays a zero stiffness along the vertical
With x3 = 0
If the force on the second extremity is vertical, the force vector is:
0
0
F= (3.26)
0
F4
With F4 = 0
As we will see in Sect. 5.2 (p. 71) the stiffness matrix is:
K 11 0 −K 11 0
0 0 0 0
K = (3.27)
−K 11 0 K 11 0
0 0 0 0
The matrix is singular. This is due to the derivative ∂∂ Fx44 , which is equal to 0 in this
case of an unstretched horizontal bar. (i) If the bar is not horizontal this derivative
will not be equal to 0, because the derivative of the bar length will not equal 0. (ii)
If the bar is in tension (or compression), even horizontal, the derivative ∂∂ Fx44 will not
equal 0 because the derivative of the tension direction is not equal to 0.
To avoid problems due to singularity, precautions are available, as described below.
3.1.4.1 Additional Stiffness
A simple way is to add an arbitrary value (α) along the diagonal of the stiffness
matrix, such that the previous matrix becomes:
K 11 + α 0 −K 11 0
0 α 0 0
K = (3.28)
−K 11 0 K 11 + α 0
0 0 0 α
The added value (α) could decrease along the Newton-Raphson iterations. This
added value (α) does not modify the equilibrium position, but only the way to reach
this equilibrium.
3.1.4.2 Additional Mechanical Behaviour
Another way to remove singularity is to add further mechanical behaviour. For exam-
ple, if this bar is in a fluid, air, or water, a vertical displacement will generate a drag
in the opposite direction, meaning that the components of the stiffness matrix K 22
and K 44 will be not equal to 0.
3.1.4.3 Displacement Limit
A displacement limit could be imposed to avoid too large a value:
hK = F (3.29)
if hi > limit hi = limit (3.30)

if hi ≤ limit hi = hi (3.31)
3.2 Other Resolution Methods
3.2.1 Newmark Method
The Newmark method is used to find the equilibrium position of a mechanical struc-
ture. The following example in one dimension explains the method in a simplified
way.
The method consists first in calculating forces on the structure, then calculating
the acceleration on the structure using the dynamic equation (F = Mγ ). From this
acceleration and using a time step, the speed and the new position of the structure
can be calculated [3].
For the example displayed in Fig. 3.1, the equilibrium calculation follows the path
shown in Fig. 3.8 with a time step of 0.04 s. Figure 3.9 shows the residue of force.
This calculation follows the Newmark explicit method [3].
3.2 Other Resolution Methods 25
Fig. 3.8 Force on the mass function of spring length and Newmark explicit method iterations
Fig. 3.9 Residue of force for each Newmark explicit method iterations
3.2.2 Energy Minimization
This method consists of finding the position of the structure that leads to the minimum
of the energy. The energy involved here is the energy due to the conservative forces
only. A conservative force is a force that leads to a variation of energy between two
positions independent of the path between these two positions. The main conservative
forces involved in marine structures are weight and tension in elastic cables and
netting twines.
In these cases the energy between two positions are quite simple to calculate:
E W = W Δh (3.32)
1
E T = K Δx 2 (3.33)
2
E W : energy due to the weight (J),
W : weight (N),
Δh: altitude variation between the two positions (m),
E T : energy due to the tension (J),
K : constant cable stiffness (N/m),
Δx: cable length variation between the two positions (m).
Some forces are not conservative, as in the case of drag force. In such case the
energy consumed by the drag depends on the path followed by the structure between
the two positions.
Due to non conservative forces, the method of minimization of energy is not quite
adapted to solve the equilibrium of marine structures. In case this method is used,
the drag forces could be transformed into constant force.
Chapter 4
The Triangular Finite Element for Netting
Abstract The modellings for netting are fully described. The usual modellings
based on numerical twines or globalization of twines are partly explained with their
limitations. These limitations have drove to the creation of the triangular finite ele-
ment for netting. This triangular element for netting is fully described. The forces
required for the equilibrium calculation are fully described, as well as the stiffness in
case of—twines elasticity,—hydrodynamic forces,—twine flexion,—mesh opening
stiffness,—fish catch pressure,—inertia,—buoyancy and weight.
Keywords Triangular finite element for netting · Twines tension in netting · Hydro-
dynamic forces on netting · Twine flexion in netting · Mesh opening stiffness of
netting · Fish catch pressure in cod-end
4.1 State-of-the-Art of Numerical Modelling for Nets
4.1.1 Constitutive Law for Nets
There is little or no published work on the constitutive law for nets. Only Rivlin
[23], to our knowledge, begins to express the stresses in a net surface, but only
under conditions of symmetrical deformation twine. If such constitutive law could
be defined, usual finite element softwares could be adapted for nettings.
4.1.2 Twine Numerical Method
The twine numerical method includes almost all the work on numerical modelling of
the net [2, 6, 9, 10, 11, 24]. The initial idea is simple: the twines of the net are modelled
by bars (called here numerical twines). Then a few adjustments are required.

28 4 The Triangular Finite Element for Netting
The twines could be modelled by two bars to account for the shortening, which
appears as an angle between the bars. The twines could be modelled with a single bar,
but Young’s modulus in compression is almost zero to account for the shortening.
Given the large number of twines in some structures (up to one million), a numerical
bar refers to several true twines (Fig. 4.1). This is called globalization.
The major difficulty with this method of globalization lies in the description of
the net by numerical twines. Indeed, a structure is very often the assembly of several
panels of nets. Therefore, the creation of numerical twines in a panel will generate
nodes on its contour. These nodes are the basis for the creation of numerical twines
of the adjacent panel (Figs. 4.2 and 4.3).
Figure 4.2a shows four panels (50 by 50 meshes) whose numerical twines connect
perfectly (Fig. 4.2b): the nodes on the edges are perfectly aligned with the nodes of
the adjacent panels.
Figure 4.3a shows the same example, except that panel 1 is only 45 meshes hor-
izontally. In this case the nodes on the borders do not connect perfectly between
panels 4 and 1 (Fig. 4.3b), whereas the connections are perfect on the other three
seams. This approach requires facilities such modification of the design of the net-
ting panels. These facilities are not well described in the literature dedicated to this
method.
4.2 The Finite Element for Netting
Triangular elements have been developed to model the net (Fig. 4.4). A number of
approximations are made in these triangular elements, with the aim of calculating the
forces at the vertices of these elements. These are calculated based on the positions
Fig. 4.1 Control net 50 meshes high by 50 and 45 wide (a), with a ratio of globalization of 5 (b)
and 10 (c)
4.2 The Finite Element for Netting 29
2 1 2 1
3 4 3 4
(a) (b)
Fig. 4.2 Structure of four panels of 50 by 50 meshes (a) discretized in numerical twines (b;
globalization ratio of 10): the connection between numerical nodes on the borders of panels is
perfect (black dots for the border between panels)
of the vertices. The basic assumption in modelling nets by triangular elements is that
the twines remain parallel. Under these conditions the twines of the same direction
have the same deformation. The second assumption is that the twines are modelled
as elastic rods.
One difficulty with the method of numerical globalized twines (or numerical
twines) was described earlier: nodes on the edges of the panels do not always coin-
cide perfectly (Fig. 4.3b). This difficulty disappears with triangular elements, since
the discretization of a netting panel is independent of the discretization of adjacent
panels, except on the border. The same panels of Fig. 4.3 are discretized in Fig. 4.5
with triangular elements. Panel 2 in (Fig. 4.5a) is discretized with large triangular
elements and in (Fig. 4.5b) with smaller elements. It is clear that triangular element
discretization is done very easily, unlike the numerical twines technique. This flexibil-
ity in the creation of triangular elements overcomes the cumbersome tool for creating
globalized twines. This burden results from many different cases to be processed and
consequently adjustments that sometimes make it impossible to fully describe the
structure to be studied with the method of numerical twines.
2 1 2 1
3 4 3 4
(a) (b)
Fig. 4.3 a Four netting panels 50 by 50 meshes except for panel 1, which has only 45 meshes
horizontally. b The globalization of 10 leads the nodes on the common border of panels 1 and 4 to
not connect perfectly: panel 1 has five nodes on its bottom border, while the top border of panel 4
has six nodes (black dots)
4.2.1 The Basic Method: Direct Formulation
The triangular finite element dedicated to diamond mesh nets is described here.
The triangular element is defined by its three vertices, which are connected to the
net. The coordinates of the vertices in number of twine vectors are then constant,
whatever the deformation of the triangle. Figure 4.6 shows an example. In this exam-
ple the coordinates in twine number of node 1 are 1.5 along the U twine and −3.5
along the V twine. It is clear that if the origin of coordinates in twine number changes,
the twine coordinates of nodes will change but will not affect the equilibrium position
of the net.
These twines are parallel inside the triangular element, which means that the sides
of the triangle (12, 23, 31) are linear combinations of twine vectors (U and V, cf.
Fig. 4.6). This point is the main foundation of the model. These combinations are as
follows:
12 = (U2 − U1 )U + (V2 − V1 )V (4.1)

13 = (U3 − U1 )U + (V3 − V1 )V (4.2)
12 (13): vector from vertex 1 (1) to vertex 2 (3).

(a) (b)
Fig. 4.4 The diamond mesh (a) is decomposed into triangular elements (b). The approximation
in each triangle is that twines are parallel and therefore have the same deformation, and that the
twines are elastic
The two previous equations with two unknowns (U and V) then give the following:
V3 − V1 V2 − V1
U= 12 − 13 (4.3)
d d
U2 − U1 U3 − U1
V= 13 − 12 (4.4)
d d
With side vectors:
x2 − x1
12 = y2 − y1 (4.5)
z2 − z1
x3 − x1
13 = y3 − y1 (4.6)
z3 − z1
2 1 2 1
3 4 3 4
(a) (b)
Fig. 4.5 Case identical to Fig. 4.3. Although the netting in panel 1 has only 45 meshes horizontally,
the triangular element discretization is easy. The step size of panel 2 is larger in (a) than in (b)
Fig. 4.6 A triangular ele-

ment: the sides of the triangle
are linear combinations of
twine vectors (U and V). The
coordinates in twine number
are noted. The origin of theses
coordinates is the intersection
of U and V
and
d = (U2 − U1 )(V3 − V1 ) − (U3 − U1 )(V2 − V1 ) (4.7)
xi , yi , z i : Cartesian coordinates of vertex i,

Ui , Vi : coordinates of vertex i in number of twines (twine coordinates).
The twine vectors (U, V) are calculated from the Cartesian coordinates (xi , yi ,
z i ) of the vertices of the triangular element.
It appears that nothing implies that the number of twine coordinates of the vertices
of the triangle consists of integers. Therefore, these coordinates can be real. This
implies that the vertices of the triangle are not necessarily located on knots of the net
(Fig. 4.4). Similarly, nothing prevents the triangle from being smaller than a mesh.
It appears that while the triangle does not contain any piece of twine of the net, d is
not null, and therefore the triangle contains twines and consequently a deformation
energy. In other words, the triangular finite element is a homogenization of the
mechanical properties of the net.
It also appears that every point of the twines belongs to only one triangular element
and still the same, regardless of the deformation of the net. Points on the contour of
a triangular element also belong to the neighbours.
4.2.2 Metric of the Triangular Element
The objective of the finite element method is to calculate the Cartesian coordinates
of the numerical nodes. These nodes are, for the netting, the vertices of the triangular
elements (Figs. 4.7 and 4.8a).
The nodes are fixed relative to the netting, which means that the coordinates of
the nodes in twines or meshes remain constant regardless of the netting deformation.
Figure 4.8b and c show an example of coordinates of a triangular element. Gen-
erally speaking, the mesh coordinates are used by the netting maker.
There are relations between the mesh coordinates and the twine coordinates, the
bases of which are noted in Fig. 4.8b and c.
The relations between the bases are the following:
u=U−V (4.8)
v=U+V (4.9)
This leads to:

u+v
U= (4.10)
2
v−u
V= (4.11)
2
u, v: mesh coordinates base,
U, V: twine coordinates base.
-60 10 -60 10
10 -60 10 -60
-60 -60
-60 -60
Fig. 4.7 Two deformations of the same structure. The twines coordinates of vertices remain con-
stant. The twines coordinates of three vertices are noted. The dot is the origin of twines numbering.
Only 1 twine on 5 is drawn
(a) (b) (c)

Fig. 4.8 Triangular element: Cartesian coordinates (a), twines coordinates (b), and mesh coordi-
nates (c). The grey surface is a mesh surface (b)
This means that the relations between the twine coordinates and the mesh
coordinates of the node P are the following:
UP = uP + vP (4.12)
VP = v P − u P (4.13)
and
U P − VP
uP = (4.14)
2
U P + VP
vP = (4.15)
2
Here, U P and V P are the twine coordinates, and u P and v P are the mesh
coordinates of the same node P. In these conditions the vector from origin to node
P could be written as follows:
OP = U P U + V P V (4.16)
OP = u P u + v P v (4.17)
Because the amplitude of a cross product of vectors is twice the surface of the
triangle made of these two vectors, the Cartesian surface of the triangular element
(in m2 ) is half the amplitude of the cross product of the side vectors of the triangular
element:
1
S = |12 ∧ 13| (4.18)
2
The side vectors in Cartesian coordinates are as follows:
x2 − x1
12 = y2 − y1 (4.19)
z2 − z1
x3 − x1
13 = y3 − y1 (4.20)
z3 − z1
By the same way, the number of meshes, as defined in Fig. 4.8b, is
1
nbm = |12 ∧ 13| (4.21)
4
with side vectors in twine coordinates:
U2 − U1
12 = V2 − V1 (4.22)
0
U3 − U1
13 = V3 − V1 (4.23)
0
The number of meshes in a triangular element is
1 d
nbm = [(U2 − U1 )(V3 − V1 ) − (U3 − U1 )(V2 − V1 )] = (4.24)
4 4
Because there are two twines U and two twines V per mesh, the number of twines
U and V is calculated as follows:
d
nbU = (4.25)
2
d
nbV = (4.26)
2
Because there are also two knots per mesh, the number of knots in a triangular
element is
d
nbk = (4.27)
2
The surface (m2 ) of one mesh is calculated through the cross product of twines
vectors (U and V):
Ms = 2|U ∧ V| (4.28)
which is also the surface of the triangular element divided by the number of meshes
in the element:
S
Ms = (4.29)
nbm
In the case of Figs. 4.6 and 4.8, d = 38, the number of meshes is 9.5, the number
of U twines is 18, the number of V twines is 18, and the number of knots is 18.
4.3 The Forces on the Netting
4.3.1 Twine Tension in Diamond Mesh
The tensions in the twines are required to estimate the forces on the vertices due
to these tensions. In the hypothesis of linear elasticity, these tensions are deduced
from U and V, which have been previously calculated. In these conditions the twine
tensions are as follows:
|U| − l0
Tu = E A (4.30)
l0
|V| − l0
Tv = E A (4.31)
l0
E : Young’s modulus of the material (N/m2 ),

A : mechanical section of the twines U and V (m2 ),
lo : unstretched length of twine vectors (m).
The principle of virtual work is used here to calculate the forces on the vertices
due to the tension in the twines.
4.3 The Forces on the Netting 37
The force component along X on vertex 1 of a triangular element is estimated

by considering a virtual displacement (∂ x1) along the axis x of vertex 1. This
displacement leads to an external work:
We = Fx1 ∂ x1 (4.32)
This displacement also induces a change in the length of mesh bars (∂|U| and
∂|V|), an internal work per twine ∂|U|Tu and ∂|V|Tv and therefore an internal work
for the triangular element:
d
Wi = (∂|U|Tu + ∂|V|Tv ) (4.33)
2
The principle of virtual work implies that the external work equals the internal
work, since the forces represent the tension in the twines. That gives for each com-
ponent of force on the three vertices:

∂|U| ∂|V| d
Fx1 = Tu + Tv (4.34)
∂ x1 ∂ x1 2

∂|U| ∂|V| d
Fy1 = Tu + Tv (4.35)
∂ y1 ∂ y1 2

∂|U| ∂|V| d
Fz1 = Tu + Tv (4.36)
∂z1 ∂z1 2

∂|U| ∂|V| d
Fx2 = Tu + Tv (4.37)
∂ x2 ∂ x2 2

∂|U| ∂|V| d
Fy2 = Tu + Tv (4.38)
∂ y2 ∂ y2 2

∂|U| ∂|V| d
Fz2 = Tu + Tv (4.39)
∂z2 ∂z2 2

∂|U| ∂|V| d
Fx3 = Tu + Tv (4.40)
∂ x3 ∂ x3 2

∂|U| ∂|V| d
Fy3 = Tu + Tv (4.41)
∂ y3 ∂ y3 2

∂|U| ∂|V| d
Fz3 = Tu + Tv (4.42)
∂z3 ∂z3 2
The derivatives ∂|U | ∂|V |

∂ x1 ... ∂z3 can be calculated, as the equations relating to U , V
and X i , Yi , Z i have already been described. This gives the following vectors force
for the three vertices:
U V
F1 = (V3 − V2 )Tu + (U2 − U3 )Tv (4.43)
2|U| 2|V|
U V
F2 = (V1 − V3 )Tu + (U3 − U1 )Tv (4.44)
2|U| 2|V|
U V
F3 = (V2 − V1 )Tu + (U1 − U2 )Tv (4.45)
2|U| 2|V|
The Newton-Raphson method, described earlier, requires the calculation of the

stiffness matrix, which is calculated from the derivatives of effort with respect to the
positions of the vertices of the triangular element. The 81 derivatives, that is to say,
by 9 by 9 component coordinates, are then the following:
The stiffness matrix:
⎛ ∂F ⎞
− ∂ xx1 − ∂∂Fyx1 . . . − ∂∂zFx1
⎜ 1 1 3
⎟
⎜ − ∂ Fy1 − ∂ Fy1 . . . − ∂ Fy1 ⎟
⎜ ∂ x1 ∂ y1 ∂z 3 ⎟
⎜ ⎟
K =⎜ ⎜ . . . . . . ⎟ (4.46)
⎟
⎜ . . ... . ⎟
⎜ ⎟
⎝ . . ... . ⎠
− ∂∂Fxz31 − ∂∂Fyz3
1
. . . − ∂∂zFz3
3
The components are calculated as follows:

∂ Fw1 E Au (V3 − V2 ) ∂Uw 1 1 ∂|U| Uw
= − +
∂t 2 ∂t n0 |U| ∂t |U|2

E Av (U2 − U3 ) ∂ Vw 1 1 ∂|V| Vw
+ − + (4.47)
2 ∂t n0 |V| ∂t |V|2

∂ Fw2 E Au (V1 − V3 ) ∂Uw 1 1 ∂|U| Uw
= − +
∂t 2 ∂t n0 |U| ∂t |U|2

E Av (U3 − U1 ) ∂ Vw 1 1 ∂|V| Vw
+ − + (4.48)
2 ∂t n0 |V| ∂t |V|2

∂ Fw3 E Au (V2 − V1 ) ∂Uw 1 1 ∂|U| Uw
= − +
∂t 2 ∂t n0 |U| ∂t |U|2

E Av (U1 − U2 ) ∂ Vw 1 1 ∂|V| Vw
+ − + (4.49)
2 ∂t n0 |V| ∂t |V|2
With:
w = x, y, z,
t = x1, y1, z1, x2, y2, z2, x3, y3, z3.
The following derivatives are also required.

The derivatives of the components of U are as follows:
∂Ux ∂U y ∂Uz V2 − V3
= = = (4.50)
∂ x1 ∂ y1 ∂z1 d
= = = (4.51)
∂ x2 ∂ y2 ∂z2 d
= = = (4.52)
∂ x3 ∂ y3 ∂z3 d
∂Ux ∂Ux ∂U y ∂U y ∂Uz ∂Uz
= = = = = =0 (4.53)
∂ yi ∂zi ∂zi ∂ xi ∂ xi ∂ yi
The derivatives of the components of V are the following:
∂ Vx ∂ Vy ∂ Vz U3 − U2
= = = (4.54)
∂ x1 ∂ y1 ∂z1 d
∂ Vx ∂ Vy ∂ Vz U1 − U3
= = = (4.55)
∂ x2 ∂ y2 ∂z2 d
∂ Vx ∂ Vy ∂ Vz U2 − U1
= = = (4.56)
∂ x3 ∂ y3 ∂z3 d
∂ Vx ∂ Vx ∂ Vy ∂ Vy ∂ Vz ∂ Vz
= = = = = =0 (4.57)
∂ yi ∂zi ∂zi ∂ xi ∂ xi ∂ yi
The derivatives of |U| follow:
∂|U| V2 − V3
= [(x2 − x1 )(V3 − V1 ) − (x3 − x1 )(V2 − V1 )] (4.58)
∂ x1 d2
∂|U| V3 − V1
= [(x2 − x1 )(V3 − V1 ) − (x3 − x1 )(V2 − V1 )] (4.59)
∂ x2 d2
∂|U| V1 − V2
= [(x2 − x1 )(V3 − V1 ) − (x3 − x1 )(V2 − V1 )] (4.60)
∂ x3 d2
∂|U| V2 − V3
= [(y2 − y1 )(V3 − V1 ) − (y3 − y1 )(V2 − V1 )] (4.61)
∂ y1 d2
∂|U| V3 − V1
= [(y2 − y1 )(V3 − V1 ) − (y3 − y1 )(V2 − V1 )] (4.62)
∂ y2 d2
∂|U| V1 − V2
= [(y2 − y1 )(V3 − V1 ) − (y3 − y1 )(V2 − V1 )] (4.63)
∂ y3 d2
∂|U| V2 − V3
= [(z 2 − z 1 )(V3 − V1 ) − (z 3 − z 1 )(V2 − V1 )] (4.64)
∂z1 d2
∂|U| V3 − V1
= [(z 2 − z 1 )(V3 − V1 ) − (z 3 − z 1 )(V2 − V1 )] (4.65)
∂z2 d2
∂|U| V1 − V2
= [(z 2 − z 1 )(V3 − V1 ) − (z 3 − z 1 )(V2 − V1 )] (4.66)
∂z3 d2
The derivatives of |V| are shown below:
∂|V| U2 − U3
= [(x2 − x1 )(U3 − U1 ) − (x3 − x1 )(U2 − U1 )] (4.67)
∂ x1 d2
∂|V| U3 − U1
= [(x2 − x1 )(U3 − U1 ) − (x3 − x1 )(U2 − U1 )] (4.68)
∂ x2 d2
∂|V| U1 − U2
= [(x2 − x1 )(U3 − U1 ) − (x3 − x1 )(U2 − U1 )] (4.69)
∂ x3 d2
∂|V| U2 − U3
= [(y2 − y1 )(U3 − U1 ) − (y3 − y1 )(U2 − U1 )] (4.70)
∂ y1 d2
∂|V| U3 − U1
= [(y2 − y1 )(U3 − U1 ) − (y3 − y1 )(U2 − U1 )] (4.71)
∂ y2 d2
∂|V| U1 − U2
= [(y2 − y1 )(U3 − U1 ) − (y3 − y1 )(U2 − U1 )] (4.72)
∂ y3 d2
∂|V| U2 − U3
= [(z 2 − z 1 )(U3 − U1 ) − (z 3 − z 1 )(U2 − U1 )] (4.73)
∂z1 d2
∂|V| U3 − U1
= [(z 2 − z 1 )(U3 − U1 ) − (z 3 − z 1 )(U2 − U1 )] (4.74)
∂z2 d2
∂|V| U1 − U2
= [(z 2 − z 1 )(U3 − U1 ) − (z 3 − z 1 )(U2 − U1 )] (4.75)
∂z3 d2
4.3.2 Twine Tension in Hexagonal Mesh
The same technique for the diamond mesh netting is used for hexagonal ones. The
triangular element dedicated to the hexagonal mesh netting has the same assumption
as previously adopted: the three families of twines inside the element are parallel,
i.e., l, m, and n twine vectors, are parallel (Fig. 4.9).
The mesh base (shaded area in Fig. 4.9) is first defined. This base mesh is defined
as a parallelogram; its corners coincide with knots, and it includes two l twine vectors,
two m twine vectors, and two n twine vectors. This base mesh is also used to quantify
the number of meshes inside the triangular element. The vertices of the triangular
element then have coordinates in base meshes (U1 , U2 , U3 , V1 , V2 , V3 ; Fig. 4.9).
Vectors U and V are the sides of the mesh base. There are linear relations between
these two vectors and the sides of the triangular element (arrows on Fig. 4.9):
12 = (U2 − U1 )U + (V2 − V1 )V (4.76)

13 = (U3 − U1 )U + (V3 − V1 )V (4.77)
Fig. 4.9 Triangular element dedicated to the hexagonal mesh nets. The twine vectors are l, m, and
n. The number of meshes are noted for each vertex. The mesh base is in grey and is defined by
vectors U and V
The two previous equations give the following as in the case of diamond mesh
(see Sect. 4.2.1, page 30), namely:
V3 − V1 V2 − V1
U= 12 − 13 (4.78)
d d
U3 − U1 U2 − U1
V= 12 − 13 (4.79)
d d
With vectors of the sides of the mesh base:
x2 − x1
12 = y2 − y1 (4.80)
z2 − z1
x3 − x1
13 = y3 − y1 (4.81)
z3 − z1
and
d = (U2 − U1 )(V3 − V1 ) − (U3 − U1 )(V2 − V1 ) (4.82)
xi , yi , z i : Cartesian coordinates of vertex i.
The number of base meshes in a triangular element is equal to d/2, the total
number twine vectors is 3d, the number of twine vectors l, m, or n is d, and the
number of nodes is 2d.
Tensions in twine vectors l, m, and n are now calculated. This is done by solving
the force balance of the twines. This is solved by writing the following equations:
(1) The base mesh definition leads to (Fig. 4.9) :
U = −m + 2n − l (4.83)
V = −m + l (4.84)
(2) The amplitude of tension in the twines gives:
|l| − l0
|Tl | = E Al (4.85)
l0
|m| − m 0
|Tm | = E Am (4.86)
m0
|n| − n 0
|Tn | = E An (4.87)
n0
(3) The balance of tensions leads to:
Tl + Tm + Tn = 0 (4.88)
This gives six equations with six unknowns (l, m, n, Tl , Tm , Tn ).
4.3.2.1 Equilibrium of the Joint Knot
The six previous equations can be reduced to the two that follow with two unknowns
(m x and m y components of m), since the triangular element has been turned in the
plane XOY [17, 19]:
m x + Vx E l Al
(m x + Vx )2 + (m y + Vy )2 − lo
(m x + Vx )2 + (m y + Vy )2 lo
mx E m Am
+ m 2x + m 2y − m o
m 2x + m 2y m o
m x + Ux +Vx
E n An
+ 2
2
2 n o
U y +Vy
m x + Ux +V
2
x
+ my + 2
⎡ ⎤
2
U + V U y + Vy 2
×⎣ − no⎦
x x
mx + + my +
2 2
=0 (4.89)
m y + Vy E l Al
(m y + Vy )2 + (m y + Vy )2 − lo
(m x + Vx )2 + (m y + Vy )2 lo
my E m Am
+ m 2y + m 2y − m o
m x + m y mo
2 2
U y +Vy
my + E n An
+ 2
2
2 n o
U y +Vy
m x + Ux +V
2
x
+ my +
2
⎡ ⎤

U y + Vy 2 U y + Vy 2
×⎣ my + + my + − no ⎦
2 2
=0 (4.90)
m x , m y : components of m twine (m),

lo , m o , n o : unstretched length of twines l, m, and n (m),
Ux , U y , Vx , Vy : components of the sides of the mesh base (m; see Fig. 4.9),
El , E m , E n : Young modulus of twines l, m, and n (Pa),
Al , Am , An : section of twines l, m, and n (m2 ).
These two equations describe the equilibrium of the joint knot of three twines in a
triangle, the sides of which are U+V 2 and V (Fig. 4.10). These equations are in newtons.
4.3.2.2 Approximation of the Equilibrium of the Joint
The analytical solution of the two previous equations has not been found. Therefore,
the following approximation has been made to simplify the equations. This approx-
imation is acceptable because the stretched lengths of the twines are close to the
unstretched length.
mx mx
≈ (4.91)
|m| mo
my my
≈ (4.92)
|m| mo
Fig. 4.10 The three twines

are in the triangle defined by
U+V
2 and V (cf. Fig. 4.9)
With this approximation the two previous equilibrium equations are reduced to
the following:

E l Al E m Am
(m x + Vx ) (m x + Vx )2 + (m y + Vy )2 − lo + m x m 2 + m2 − m
x y o
lo2 m 2o
⎛ ⎞

Ux + Vx E n An ⎝ Ux + Vx 2 U y + Vy 2
+ mx + mx + + my + − no ⎠ = 0
2 n 2o 2 2
(4.93)

E l Al E m Am
(m y + Vy ) 2 (m x + Vx )2 + (m y + Vy )2 − lo + m y 2
m 2x + m 2y − m o
lo mo
⎛ ⎞
2
U y + Vy E n An U + V U y + Vy 2
+ my + ⎝ mx +
x x
+ my + − no ⎠ = 0
2 n 2o 2 2
(4.94)
They are the complete form of the following:
E l Al E m Am E n An
lx (|l| − lo ) + m x (|m| − m o ) + n x 2 (|n| − n o ) = 0 (4.95)
lo2 m 2o no
E l Al E m Am E n An
l y 2 (|l| − lo ) + m y (|m| − m o ) + n y 2 (|n| − n o ) = 0 (4.96)
lo m 2o no
4.3.2.3 Newton-Raphson Method
The previous approximation has not been sufficient to reach the analytical solution.
The Newton-Raphson method is used to find a numerical solution [4].
For each iteration the displacement h is searched to find the equilibrium:

F(xk )
hk = (4.97)
−F (xk )
xk+1 = xk + h k (4.98)
k: iteration number,
F: force on nodes,
x: position of nodes.
Here:
E l Al
l x l 2 (|l| − lo ) + m x E mmA2 m (|m| − m o ) + n x Enn 2An (|n| − n o ) = F1
F= o o o (4.99)
l y Ell 2Al (|l| − lo ) + m y E mmA2 m (|m| − m o ) + n y Enn 2An (|n| − n o ) = F2
o o
o
mx
x= (4.100)
my
The derivative is:

D11 D12

F = . (4.101)
D21 D22
With:

E Al l2 E Am m 2x E An n 2x
D11 = − l − lo + x + m − m o + + n − n o +
lo2 l m 2o m n 2o n
(4.102)
E Al l x l y E Am m x m y E An n x n y
D12 = D21 = − + + 2 (4.103)
lo2 l m 2o m no n

E Al l y2 E Am m 2y E An n 2y
D22 = − l − lo + + m − mo + + 2 n − no +
lo2 l m 2o m no n
(4.104)
With the previous conditions the displacement (h) can be calculated:

⎧
⎨ D22 F1 −D12 F2
D22 D11 −D12 D21
h= (4.105)
⎩ D22 F2 −D21 F1
D22 D11 −D12 D21
4.3.2.4 Forces on Nodes
The forces on the sides of the triangular element are calculated from the twine tension.
These forces are related to the number of twines through the sides of the triangle.
This number of twines through each side can be calculated based on the number of
base mesh of each vertex.
The effort on the side along U of the base mesh (Fig. 4.9) is
FU = Tl − Tm (4.106)
The effort along V is

FV = −Tn (4.107)
Under these conditions, the effort on each side of the triangle can be deduced:
T12 = (U2 − U1 )(Tl − Tm ) + (V2 − V1 )(−Tn ) (4.108)

T23 = (U3 − U2 )(Tl − Tm ) + (V3 − V2 )(−Tn ) (4.109)
T31 = (U1 − U3 )(Tl − Tm ) + (V1 − V3 )(−Tn ) (4.110)
Here, Ti j is the effort on the side i j of the triangular element.

Each side effort is distributed on each end of this side as the twines are evenly
distributed along the sides of the triangle:
T12 + T31
F1 = (4.111)
2
T23 + T12
F2 = (4.112)
2
T31 + T23
F3 = (4.113)
2
F1 , F2 , and F3 are the forces on the three vertices of the triangular element due
to the tension in the twines.
The contribution of the stiffness matrix is not described here.
4.3.3 Hydrodynamic Drag
4.3.3.1 Introduction
The drag force on the netting is calculated in this model as the sum of the drag
force on each twine (U and V). This assumption is probably questionable, because
the drag on a twine alone is surely not exactly the same as the drag on this twine
among other twines as it is the case in a netting. Anyway, this assumption leads to the
calculation of the drag of each triangular element because for each the twines vectors
are known, as described earlier. The formulation for the twine vector drag is based
on the assumptions of Morrison adapted by Landweber and Richtmeyer [8, 22].
Fig. 4.11 Normal (F) and

tangential (T) forces on a
twine due to the relative
velocity of water (c)
The drag amplitudes on the U twines used in the model (Fig. 4.11) are:
1 d
|F| = ρCd Dl0 [|c|sin(α)]2 (4.114)
2 2
1 d
|T| = f ρCd Dl0 [|c|cos(α)]2 (4.115)
2 2
The directions of the drag on the U twine vectors are:
F U ∧ (c ∧ U)
= (4.116)
|F| |U ∧ (c ∧ U)|
T F ∧ (c ∧ F)
= (4.117)
|T| |F ∧ (c ∧ F)|
F: normal drag (N ) on the U twines, following the assumptions of Landweber,

T: tangential drag (N ) on the U twines, Richtmeyer hypothesis,
ρ: density of water (kg/m3 ),
Cd : normal drag coefficient,
f : tangential drag coefficient,
D: diameter of twine (m),
l0 : length of twine vector (m),
c: water velocity relative to the twine (m/s),
α: angle between the U twine and the water velocity (radians),
d/2 : number of U twine vectors in the triangular element.
In the equations of drag amplitude, the expressions |c|sin(α) and |c|cos(α) are
the normal and tangential projections on c along the U twine vector.
The drag on V twines for a triangular element are similar: U is replaced by V and
α by β.
The length of twine vectors used in the formulation of drag amplitude can be
assessed by |U| for the U twines and by |V| for the V twines. That would mean
it takes into account the twine elongation. Generally speaking, a twine elongation
is associated with a diameter D reduction by the Poisson coefficient. Because this
Poisson coefficient is not taken into account in the present modelling, the twine
surface is approximated by Dl0 , where D is the diameter of the twines and l0 is the
unstretched length of the twine vectors.
All parameters, including the angles α and β, are constant and known for each
triangular element. Therefore, the drag can be calculated for each triangular element.
The drag force for a triangular element is spread over the three vertices of the element
at 1/3 per vertex.
4.3.3.2 Definitions of the Variables
The Cartesian coordinates of the three nodes (1, 2, 3) of the triangular element (cf.
Fig. 4.12) follow:
x1
1 = y1 (4.118)
z1
x2
2 = y2 (4.119)
z2
x3
3 = y3 (4.120)
z3
Fig. 4.12 Example of trian-

gular element. The drag forces
are calculated for U twines
and for V twines. The twine
coordinates are noted in this
example
The twine coordinates of the three nodes (1, 2, 3) of the triangular element are as
follows:
U1
1= (4.121)
V1
U2
2= (4.122)
V2
U3
3= (4.123)
V3
The vector current is

cx
c = cy (4.124)
cz
Generally speaking, cz is null.

It has been seen previously:
V3 − V1 V2 − V1
U= 12 − 13 (4.125)
d d
U2 − U1 U3 − U1
V= 13 − 12 (4.126)
d d
with sides vectors:
x2 − x1
12 = y2 − y1 (4.127)
z2 − z1
x3 − x1
13 = y3 − y1 (4.128)
z3 − z1
and
d = (U2 − U1 )(V3 − V1 ) − (U3 − U1 )(V2 − V1 ) (4.129)
The components of U twine vectors are as follows:
Ux
U = Uy (4.130)
Uz
d [(V3 − V1 )(x 2 − x 1 ) − (V2 − V1 )(x3 − x1 )]

1
U= d [(V3 − V1 )(y2 − y1 ) − (V2

1
− V1 )(y3 − y1 )] (4.131)
d [(V3 − V1 )(z 2 − z 1 ) − (V2 − V1 )(z 3 − z 1 )]

1
The angle between current and U is
c.U
cos(α) = (4.132)
|c||U|
The components of V twine vectors are as follows:
Vx
V = Vy (4.133)
Vz
d [(U2 − U1 )(x 3 − x 1 ) − (U3 − U1 )(x2 − x1 )]

1
V= d [(U2 − U1 )(y3 − y1 ) − (U3

1
− U1 )(y2 − y1 )] (4.134)
d [(U2 − U1 )(z 3 − z 1 ) − (U3 − U1 )(z 2 − z 1 )]
1
The angle between current and V is
c.V
cos(β) = (4.135)
|c||V|
4.3.3.3 Stiffness of the Normal Force on the U Twines
The normal force on U twines is
U ∧ (c ∧ U)
F = |F| (4.136)
|U ∧ (c ∧ U)|
That means that the x y and z components are as follows:
Ex
Fx = |F| (4.137)
|E|
Ey
F y = |F| (4.138)
|E|
Ez
Fz = |F| (4.139)
|E|
With:
E = U ∧ (c ∧ U) (4.140)
and
Ex
E = Ey (4.141)
Ez
The x component of the derivative is
Ex E |E| − Ex |E|
Fx = |F| + |F| x (4.142)
|E| |E|2
Which gives for the x y and z components:

!
Ex |F| Ex
Fx = |F| + E
x |E| − (E E
x x

+ E E
y y

+ E E
z z

) (4.143)
|E| |E|2 |E|
!
Ey |F| Ey
Fy = |F| + E
y |E| − (E E
x x

+ E E
y y

+ E E
z z

) (4.144)
|E| |E|2 |E|
!
Ez |F| Ez
Fz = |F| + E
z |E| − (E E
x x

+ E E
y y

+ E E
z z ) (4.145)
|E| |E|2 |E|
For this assessment the derivative of E is required:
E = U ∧ (c ∧ U) + U ∧ (c ∧ U ) (4.146)
This leads to:

E = 2(U .U)c − (U .c)U − (U.c)U (4.147)
Which is:
Ex = 2(U .U)cx − (U .c)Ux − (U.c)Ux (4.148)

Ey = 2(U .U)c y − (U .c)U y − (U.c)Uy (4.149)
Ez = 2(U .U)cz − (U .c)Uz − (U.c)Uz (4.150)
With:
U .U = Ux Ux + U y Uy + Uz Uz (4.151)

U .c = cx Ux + c y Uy + cz Uz (4.152)
U.c = Ux cx + U y c y + Uz cz (4.153)
The derivative of the amplitude of the normal force is
1 d
|F| = ρCd Dl0 |c|2 [sin(α)]2 (4.154)
2 2
Which is
d
|F| = ρCd Dl0 |c|2 cos(α)sin(α)α (4.155)
2
The derivative of α is

−1 c.U
α = 2 (4.156)
|c||U|
1− c.U
|c||U|
That gives
−1 c U
α = 2 . (4.157)
|c| |U|
1− c.U
|c||U|
The derivative of the U twine direction is

U U |U| − U|U|
= (4.158)
|U| |U|2
That means that the derivative of α is

−1 c U |U| − U|U|
α = 2 |c| . (4.159)
|U|2
1 − |c||U|
c.U
or
−1 " #
α = |U| c x U
x + c y U
y + c z U
z − (c.U)|U|
(4.160)
|U|2 |c| sin α
In this case Ux is the component along x of U .

The derivative of vector U is
Ux
U = Uy

(4.161)
Uz
Which is
∂Ux ∂U y ∂Uz 1
= = = (V2 − V3 ) (4.162)
∂ x1 ∂ y1 ∂z 1 d
= = = (V3 − V1 ) (4.163)
∂ x2 ∂ y2 ∂z 2 d
= = = (V1 − V2 ) (4.164)
∂ x3 ∂ y3 ∂z 3 d
∂Ux ∂Ux ∂Ux ∂Ux ∂Ux ∂Ux

= = = = = =0 (4.165)
∂ y1 ∂ y2 ∂ y3 ∂z 1 ∂z 2 ∂z 3
∂U y ∂U y ∂U y ∂U y ∂U y ∂U y
= = = = = =0 (4.166)
∂z 1 ∂z 2 ∂z 3 ∂ x1 ∂ x2 ∂ x3
∂Uz ∂Uz ∂Uz ∂Uz ∂Uz ∂Uz
= = = = = =0 (4.167)
∂ x1 ∂ x2 ∂ x3 ∂ y1 ∂ y2 ∂ y3
On vector form and for the nine coordinates of the triangular element it is:
V2 −V3
∂U d
= 0 (4.168)
∂ x1
0
0
∂U V2 −V3
= (4.169)
∂ y1 d
0
0
∂U
= 0 (4.170)
∂z 1 V2 −V3
d
V3 −V1
∂U d
= 0 (4.171)
∂ x2
0
0
∂U V3 −V1
= (4.172)
∂ y2 d
0
0
∂U
= 0 (4.173)
∂z 2 V3 −V1
d
V1 −V2
∂U d
= 0 (4.174)
∂ x3
0
0
∂U V1 −V2
= (4.175)
∂ y3 d
0
0
∂U
= 0 (4.176)
∂z 3 V1 −V2
d
The derivative of the norm of vector U is
Ux Ux + U y U y + Uz Uz
|U| = (4.177)
|U|
This gives for the nine coordinates of the triangular element:
∂|U| Ux (V2 − V3 )
= (4.178)
∂ x1 d|U|
∂|U| U y (V2 − V3 )
= (4.179)
∂ y1 d|U|
∂|U| Uz (V2 − V3 )
= (4.180)
∂z 1 d|U|
∂|U| Ux (V3 − V1 )
= (4.181)
∂ x2 d|U|
∂|U| U y (V3 − V1 )
= (4.182)
∂ y2 d|U|
∂|U| Uz (V3 − V1 )
= (4.183)
∂z 2 d|U|
∂|U| Ux (V1 − V2 )
= (4.184)
∂ x3 d|U|
∂|U| U y (V1 − V2 )
= (4.185)
∂ y3 d|U|
∂|U| Uz (V1 − V2 )
= (4.186)
∂z 3 d|U|
This leads to the derivatives of α (angle between c and U):
∂α V3 − V2 Ux
= 2 cx |U| − |U| c.U (4.187)
∂ x1
d|U|2 |c| 1− c.U
|c||U|
∂α V3 − V2 Uy
= 2 c y |U| − |U| c.U (4.188)
∂ y1
d|U|2 |c| 1− c.U
|c||U|
∂α V3 − V2 Uz
= 2 cz |U| − |U| c.U (4.189)
∂z 1
d|U|2 |c| 1− c.U
|c||U|
∂α V1 − V3 Ux
= 2 cx |U| − |U| c.U (4.190)
∂ x2
d|U|2 |c| 1− c.U
|c||U|
∂α V1 − V3 Uy
= 2 c y |U| − |U| c.U (4.191)
∂ y2
d|U|2 |c| 1− c.U
|c||U|
∂α V1 − V3 Uz
= 2 cz |U| − |U| c.U (4.192)
∂z 2
d|U|2 |c| 1− c.U
|c||U|
∂α V2 − V1 Ux
= 2 cx |U| − |U| c.U (4.193)
∂ x3
d|U|2 |c| 1− c.U
|c||U|
∂α V2 − V1 Uy
= 2 c y |U| − |U| c.U (4.194)
∂ y3
d|U|2 |c| 1− c.U
|c||U|
∂α V2 − V1 Uz
= 2 cz |U| − |U| c.U (4.195)
∂z 3
d|U|2 |c| 1− c.U
|c||U|
4.3.3.4 Stiffness of the Tangential Force on the U Twines
The tangential force on U twines is
F ∧ (c ∧ F)
T = |T| (4.196)
|F ∧ (c ∧ F)|
Following the definition of F1 :
[U ∧ (c ∧ U)] ∧ {c ∧ [U ∧ (c ∧ U)]}
T = |T| (4.197)
| [U ∧ (c ∧ U)] ∧ {c ∧ [U ∧ (c ∧ U)]} |
It follows that
[(U.U)(c.c) − (U.c)2 ](U.c)U
T = |T| (4.198)
|[(U.U)(c.c) − (U.c)2 ](U.c)U|
or
[|U|2 |c|2 − (|U||c|cosα)2 ]|U||c|cosαU
T = |T| (4.199)
|[|U|2 |c|2 − (|U||c|cosα)2 ]|U||c|cosαU|
and
cos αU
T = |T| (4.200)
| cos α||U|
The x y and z components are as follows:
cos αUx
Tx = |T| (4.201)
| cos α||U|
cos αU y
T y = |T| (4.202)
| cos α||U|
cos αUz
Tz = |T| (4.203)
| cos α||U|
The derivative of Tx is:
cos αUx (cos αUx ) | cos α||U| − cos αUx (| cos α||U|)
Tx = |T| + |T| (4.204)
| cos α||U| (| cos α||U|)2
cos αUx
Tx = |T|
| cos α||U|
|T|
+ (cos αUx − sin αα Ux )
| cos α||U|

|T| cos αUx Ux Ux + U y Uy + Uz Uz cos α
− | cos α| − sin αα |U|
(| cos α||U|)2 |U| | cos α|
(4.205)
T |T|
Tx = |T| (cos αUx − sin αα Ux )
x
+
|T| | cos α||U|

Tx Ux Ux + U y Uy + Uz Uz cos α
| cos α||U| |U| | cos α|
(4.206)
Ty |T|
Ty = |T| + (cos αUy − sin αα U y )
|T| | cos α||U|

Ty Ux Ux + U y Uy + Uz Uz cos α
(4.207)
T |T|
Tz = |T| (cos αUz − sin αα Uz )
z
+
|T| | cos α||U|

Tz Ux Ux + U y Uy + Uz Uz cos α
(4.208)
The derivative of the amplitude of the tangential force is
1 d
|T| = f ρCd Dl0 |c|2 ([cos(α)]2 ) (4.209)
2 2
which is
d
|T| = − fρCd Dl0 |c|2 cos(α)sin(α)α (4.210)
2
4.3.3.5 Stiffness of the Normal and Tangential Forces

on the V Twines
This evaluations are identical to the previous, but with V and β used in place of U
and α.
4.3.4 Twine Flexionin Netting Plane
The resistance to twine bending in the plane of the net is also called the mesh opening
stiffness (Fig. 4.13). In a first approximation, this stiffness is neglected, but the use
of steeper nets makes it necessary to take this mechanical phenomenon into account
in numerical models. Currently, only [15, 12] and the present model take this mesh
opening stiffness into account.
In the present model, the half angle (α) between the twine vectors (U and V) could
lead to a couple between twine vectors (U and V). This angle is calculated by

1 U.V
α= acos (4.211)
2 |U||V|
The couple on a knot due to the U twine is equilibrated by the couple of the V
twine; otherwise the knot would not be in equilibrium. These couples are approxi-
mated in the model by
Cu = −Cv = H (α − α0 ) (4.212)
Fig. 4.13 Demonstration

of mesh opening stiffness.
Deformation remains limited
despite the weight added to
the bottom of the net on (b)
(a) (b)
where α0 is the angle between the unstressed twines (without couple on twines) and
H is the mesh opening stiffness (N.m/Rad).
This couple varies linearly with the angle. [12, 15] suggest another formulation,
since the twines are modelled as beams.
Forces at the vertices of the triangular element, mechanically equivalent to
themesh opening stiffness, are calculated using the principle of virtual work:
If ∂ x1 is a virtual displacement along the x axis of vertex 1, then the external work
(We ) is
We = F x 1 ∂ x 1 (4.213)
where F x1 is the effort along the x axis at vertex 1 of a triangular element.

This displacement creates a change in angle α, and therefore an internal work
(Wi ):
d
Wi = (Cu ∂α + Cv ∂α) (4.214)
2
d = (U2 − U1 )(V1 − V3 ) − (U3 − U1 )(V1 − V2 ) (4.215)
where d/2 is the number of nodes in a triangular element.

Since the internal work is equal to the external work,
∂α
F x1 = Cu d (4.216)
∂ x1
This gives, for all the force components at the vertices of the triangular element,
∂α
Fwi = H (α − α0 )d (4.217)
∂wi
where w = x, y, and z, and i = 1, 2, and 3.

∂α
The derivative ∂w i
of α relative to the coordinates wi of vertices, which is necessary
for calculating the forces, is
∂α Vw vi − Uw ui − Uw (U.V)v
|U|2
i
− Vw (U.V)ui
|V|2
= (4.218)
∂wi 2dsin(α)|U||V|
where w = x, y, and z, and i = 1, 2, and 3.

The stiffness matrix (−F (X)) is completed by calculating the derivative compo-
nent of efforts related to the coordinates of the vertices of the triangular element:
∂ Fw i
− (4.219)
∂t j
where as above, w = x, y, and z, and i = 1, 2, and 3, and t = x, y, and z, and j = 1,

2, and 3.
Fig. 4.14 The net bends

under its own weight, which
highlights the bending stiff-
ness of the net
4.3.5 Twine Flexion Outside the Netting Plane
To our knowledge, no numerical model, except the present one, takes into account
this mechanical property of the nets (Fig. 4.14). The angle between the U twine of a
triangle (Ua in Fig. 4.15) and its neighbour (Ub ) is constant along the side common
to the two triangular elements. This angle quantifies the bending of the twine.
The bending stiffness of the U twine tends to keep the twine straight. The equation
governing the bending is as follows:
EI
C= (4.220)
ρ
C: bending couple on the U twine (Nm),

E I : flexural stiffness, which is Young’s modulus by inertia (Nm2 ),
ρ: radius of curvature of the U twine (m).
This couple is generated, in the present modelling, when two successive triangular
elements are bent or, more precisely, when the U twine is bent to the passage of a
triangular element with its neighbour. The couple will then generate forces on the
vertices (1, 2, 3, 4 in Fig. 4.15) on the two adjacent triangular elements. Obviously
the bending of the V twines also leads to a couple. In the following only the effect
Fig. 4.15 Two triangular elements (134 and 243), the coordinates of which, in number of twines,
are noted. The angle between the twine vectors Ua and Ub leads to a bending couple between the
two triangular elements
of bending on the U twines is described; the bending on V twines has to be taken

into account in the same way.
The radius of the curvature is estimated from the average lengths of twine U in
each triangular element (Fig. 4.16). These average lengths are calculated using the
average number of twine vectors (Ua and Ub ) by the U twine in the two triangular
elements (n a and n b ).
The twine vectors of the two triangular elements (see Sect. 4.2.1 p. 30) are as
follows:
V4 − V1 V3 − V1
Ua = 13 − 14 (4.221)
da da
U4 − U1 U3 − U1
Va = 13 − 14 (4.222)
da da
Fig. 4.16 Profile view of the two triangular elements. The radius of curvature (ρ) is estimated from
the average length of twine vectors U in each triangle : n a Ua and n b Ub
V3 − V2 V4 − V2
Ub = 24 − 23 (4.223)
db db
U3 − U2 U4 − U2
Vb = 24 − 23 (4.224)
db db
Ui , Vi : coordinates of vertex i in number of twines (twine coordinates).

With side vectors:
x3 − x1
13 = y3 − y1 (4.225)
z3 − z1
x4 − x2
24 = y4 − y2 (4.226)
z4 − z2
The numbers of twine vectors (Ua and Ub ) for the U twines in the two triangular
elements are
da = (U3 − U1 )(V4 − V1 ) − (U4 − U1 )(V3 − V1 ) (4.227)

db = (U4 − U2 )(V3 − V2 ) − (U3 − U2 )(V4 − V2 ) (4.228)
The average numbers of twine vectors (Ua and Ub ) by U twine are calculated
from the number of twine vectors in the triangular elements and the length of the
common side in twine coordinates (V3 − V4 ):
da
na = (4.229)
2|V3 − V4 |
db
nb = (4.230)
2|V3 − V4 |
The radius of the curvature (ρ) is calculated from the circumscribed circle in the
triangle of sides naUa , nbUb and naUa + nbUb , as shown in Fig. 4.16. The side
lengths of the triangle are
A = |n a Ua | (4.231)
B = |n b Ub | (4.232)
C = |n a Ua + n b Ub | (4.233)
The equations of the triangle, which can be obtained in a mathematical com-

pendium, give the radius of curvature:
ABC
ρ= (4.234)
4S
where S and p, the surface and the half perimeter of the triangle, are
$
S = p( p − A)( p − B)( p − C) (4.235)
A+ B +C
p= (4.236)
2
The forces on the vertices (1, 2, 3 and 4) of the two triangularelements due to the
twine bending are calculated using the principle of virtual work. In case of the X
component of the force on vertex 1 (Fx1 ), a displacement (∂ x1) is defined along X
axis of vertex 1. This displacement generates an external work:
We = Fx1 ∂ x1 (4.237)
This movement also causes a variation of angle (∂α) between the twine vectors
(Ua and Ub ) of the two triangular elements. This variation induces an internal work:
Wi = C∂α(V3 − V4 ) (4.238)
According to the principle of virtual work, these works are equal, which gives the
following:
E I ∂α
Fwi = (V3 − V4 ) (4.239)
ρ ∂wi
w: directions x, y, and z,
i: vertices 1, 2, 3, and 4,
V3 − V4 : number of twines involved in the bending.
The angle α between the two twine vectors (Ua and Ub ) of the two triangular
elements is calculated with the dot product of twine vectors (Fig. 4.16):
Ua .Ub
cos(α) = (4.240)
|Ua ||Ub |
The 12 derivatives of α relative to the coordinates of the vertices of the two

∂α
triangular elements ( ∂wi ) are therefore required to calculate the effort on the vertices.
They are as follows:
∂α (Ua .Ub )Uaw − Ubw |Ua |2

= (V3 − V4 ) (4.241)
∂w1 |Ua |3 |Ub |da sin(α)
∂α (U .Ub )Ubw − Uaw |Ub |2
= (V4 − V3 ) a (4.242)
∂w2 |Ub |3 |Ua |db sin(α)
∂α (Ua .Ub )Uaw − Ubw |Ua |2 (Ua .Ub )Ubw − Uaw |Ub |2
= (V4 − V1 ) + (V2 − V4 )
∂w3 |Ua |3 |Ub |da sin(α) |Ub |3 |Ua |db sin(α)
(4.243)
∂α (Ua .Ub )Uaw − Ubw |Ua |2 (Ua .Ub )Ubw − Uaw |Ub |2
= (V1 − V3 ) + (V3 − V2 )
∂w4 |Ua | |Ub |da sin(α)
3 |Ub |3 |Ua |db sin(α)
(4.244)
Here, Uaw is the component along the w axis of Ua . In this case w is the axis
consisting of x, y, and z. Obviously, Ubw is the component along the w axis of Ub .
The efforts on the four vertices of the two triangular elements due to the bending
of the U twine between these two elements have been previously calculated.
The stiffness matrix (−F (X )) is completed by calculating the derivative of the
12 components of the forces relative to the 12 coordinates of the vertices of the two
triangular elements. The 144 components of this matrix are
∂ Fwi
− (4.245)
∂t j
With, as above:
w: x, y, and z.
i: 1, 2, 3, and 4.
And more:
t: x, y, and z,
j: 1, 2, 3, and 4.
4.3.6 Fish Catch Pressure
The mechanical effect of caught fish (Fig. 4.17) in a net is estimated by a pressure
[1]. This pressure is exerted directly on the triangular elements in contact with the
fish. In the case of water speed relative to that catch:
1
p= ρCd v2 (4.246)
2
p: pressure of the catch on the net (Pa),
Cd : drag coefficient,
v: current amplitude (m/s).

This pressure is then applied to the surface of the triangular element 12∧13 2 .
The resultant force is directed perpendicular to the triangular element. The effort on
each vertex is that force by 1/3.
12 ∧ 13 p
F1 = (4.247)
2 3
12 ∧ 13 p
F2 = (4.248)
2 3
12 ∧ 13 p
F3 = (4.249)
2 3
With sides vectors:
Fig. 4.17 Measurement in a flume tank tests (cross) and numerical modelling (mesh) for a scale
(1/3) model of North Sea cod-end with 300 kg of catch
x2 − x1
12 = y2 − y1 (4.250)
z2 − z1
x3 − x1
13 = y3 − y1 (4.251)
z3 − z1
That gives:
p
F1x = [(y2 − y1 )(z 3 − z 1 ) − (z 2 − z 1 )(y3 − y1 )] (4.252)
6
p
F1y = [(z 2 − z 1 )(x3 − x1 ) − (x2 − x1 )(z 3 − z 1 )] (4.253)
6
p
F1z = [(x2 − x1 )(y3 − y1 ) − (y2 − y1 )(x3 − x1 )] (4.254)
6
The contribution of this effect to the stiffness matrix is calculated through the
derivatives of the forces. The derivatives of F1 is
p
F1 = (12 ∧ 13 + 12 ∧ 13 ) (4.255)
6
The derivatives of F1 , F2 , and F3 are identical:
0
∂F1 p
= z3 − z2 (4.256)
∂ x1 6 y −y
2 3
z − z3
∂F1 p 2
= 0 (4.257)
∂ y1 6 x −x
3 2
y − y2
∂F1 p 3
= x2 − x3 (4.258)
∂z 1 6 0
0
∂F1 p
= z1 − z3 (4.259)
∂ x2 6 y −y
3 1
z − z1
∂F1 p 3
= 0 (4.260)
∂ y2 6 x −x
1 3
y − y3
∂F1 p 1
= x3 − x1 (4.261)
∂z 2 6 0
0
∂F1 p
= z2 − z1 (4.262)
∂ x3 6 y −y
1 2
z − z2
∂F1 p 1
= 0 (4.263)
∂ y3 6 x −x
2 1
y − y1
∂F1 p 2
= x1 − x2 (4.264)
∂z 3 6 0
4.3.7 Dynamic: Force of Inertia
The force of inertia is related to accelerations of the net and of the water particles
just around the net. The calculation is done for each triangularelement in three parts,
one for each vertex, since the acceleration is not constant over the entire surface of
each triangular element. Under these conditions, the parameters are local parameters
at each vertex, including the acceleration and the mass. The mass per vertex is
considered the third of the total mass of netting of the triangular element.
The force of inertia on each vertex of a triangular element mesh is estimated
by [7]:
Fi = Ma (γh − γ ) + ρV γh − Mγ (4.265)
Fi : inertial force on the vertex i (N),

Ma : added mass (kg) of 1/3 of the triangular element,
M: mass of 1/3 of the net (kg),
V : volume of 1/3 of the net (m3 ),
ρ : density of water (kg/m3 ),
γ : acceleration of the vertex (m/s2 ),
γh : acceleration of the water around the vertex (m/s2 ).
The vertex speed is calculated as follows:
x1 − x
v= (4.266)
Δt
The acceleration of the vertex is
v1 − v
γ = (4.267)
Δt
which gives
x2 − 2x1 + x
γ = (4.268)
Δt 2
In this case, the contribution to the stiffness matrix, from the derivative of this
inertia, is calculated by
∂Fi
− F = − (4.269)
∂x
which leads to
∂γ
− F = (M + Ma ) (4.270)
∂x
and
M + Ma
− F = (4.271)
Δt 2
With: x: position at t (m),
x1 : position at t − Δt (m),
x2 : position at t − 2Δt (m),
F : derivative of the force of inertia relative to the position (N/m),
Δt: time step (s).
4.3.8 Dynamic: Drag Force
The drag is related to the net and the relative speed of water particles just around
the net. The calculation is done for each triangular element in three parts, one for
each vertex, since this speed is not constant over the entire surface of each triangular
element. Under these conditions the local parameters at each vertex are the vertex
speed and one third of the number of twine vectors for the triangular element. The
calculation is done for twines U and V .
The formulation for the twine drag is based on the assumptions of Landweber
and Richtmeyer, as described earlier (Sect. 4.3.3, p. 46). The drag on the U twines
applied on vertex i of the triangular element takes into account 1/3 of the number of
U twine vectors in the triangular element. This drag is as follows:
d1
|Fi | = ρCd Dlo (|ci |sin(θ ))2 (4.272)
62
d 1
|Ti | = f ρCd Dlo (|ci |cos(θ ))2 (4.273)
6 2
Fi : normal force to the twines (N ) on vertex i, this expression coming from the
assumptions of Landweber,
Ti : tangential force (N) on vertex i, from Richtmeyer’s assumption,
Cd: normal drag coefficient,
f : tangential coefficient,
D: diameter of twines U (m),
lo : length of twine vectors U (m),
ci : amplitude of the relative velocity of the water at vertex i (m/s),

θ : angle between the twine vectors U and the relative velocity (radians),
d
6 : one third of the number of twine vectors U in the triangular element.
The angle θ between the twine vector U and the relative velocity is calculated by
ci U
cos(θ ) = (4.274)
|ci ||U|
The directions of the drag in case of twine vector U are as follows:
Fi U ci ∧ U
= ∧ (4.275)
|Fi | |U| |ci ||U|
Ti Fi ci ∧ U
= ∧ (4.276)
|Ti | |Fi | |ci ||U|
The drag amplitude on twines V is calculated following the same scheme.
4.3.9 Buoyancy and Weight
Buoyancy and weight are vertical forces (along the z axis, if it is the vertical axis).
Their expression is summed in the following:
D2
Fz = dπ l0 (ρnetting − ρ)g (4.277)
4
Fz : weight of the net once immersed (N),
d: number of twine vectors U and twine vectors V per triangular element,
ρ: water density (kg/m3 ),
ρnetting : net density (kg/m3 ),
D: diameter of twines (m),
g: gravity of the Earth (around 9.81 m/s2 ),
l0 : length of twine vectors (m).
The length of the twine vectors is approximated by the unstretched twine vector
l0 , since the elongation is generally quite small.
There is a contribution of this force to the stiffness matrix when the netting crosses
the water surface. In this case there is a variation of force with the immersion. This
contribution is not described here.
Fig. 4.18 Comparison between simulations (net) and flume tank tests (crosses) of trawl cod-ends
[1]. Between 2.5 and 3.5 m the diameter is constant. This is due to contact between the nodes of the
net
4.3.10 Contact Between Knots
It happens quite frequently that the nets are so close that the nodes come into contact
with each other. This contact limits the closing of mesh (Fig. 4.18).
An effort similar to that described in Sect. 4.3.4 (p. 57) has been introduced to
take into account this feature. This effort appears only when the twines are close
enough, that is, when the angle between U and V twines is below a critical angle
(αmini ). This angle is related to the node size and mesh side as follows (Fig. 4.19):
knotsi ze
αmini = 2 arcsin (4.278)
2mesh side
αmini : limit angle of contact between twines (rad),

knotsi ze : size of the node (m),
mesh side : side of the mesh or length of twine vectors (m).
The mesh side could be the length of the twine vector along the U twine (|U|) or
the length of the twine vector along the V twine (|V|). To avoid this choice (between
|U| and |V|), this length can be approximated by the unstretched length l0 of the
twine vector.
A couple is generated between the twines if the angle between them is less than
the minimal angle:

C = H (α − αmini ) i f α <= αmini
(4.279)
C=0 i f α > αmini
C: couple between the twines due to the contact between knots (Nm),
α: angle between twines U and V (rad),
Fig. 4.19 The size of the knot limits the closure of the mesh. The minimal angle between twines
is due to the size of the knot and the side of the mesh (which is also the length of twine vector)
H : stiffness (Nm/Rad).
This stiffness is not well known. Therefore, arbitrary values can be used, such as
the following, proportional to the elongation stiffness of the twine (E A) (Fig. 4.19):
1 mesh 2side E A
H= (4.280)
100 knotsi ze
A: section of the twine (m2 ),

E: Young’s modulus (Pa).
The forces on the vertices of triangular elements and the stiffness use the same
expressions as those described in Sect. 4.3.4 (p. 57).
Chapter 5
The Bar Finite Element for Cable
Abstract The modelling for cable is described. A bar element for cable is described.
The forces as well as the stiffness are given in case of tension in cables, of flexion
and of hydrodynamic forces.
Keywords Bar element for cable · Tension in cables · Flexion of cables
5.1 Principle
The cables are split into bar elements (Fig. 5.1). The greater the number of bars, the
better the representation of the curvature.
From the position X of the extremities of the bar elements the forces F on these
extremities are calculated. The bar elements, in the present modelling, respect a
couple of hypotheses. The first is that the bar element is straight. The second is that
the bar element is elastic. These hypotheses make possible the calculation of forces
on the extremities of the bar element.
5.2 Tension on Bars
5.2.1 Force Vector
The forces on the extremities of the bar elements are due to the tension in the bar
(Fig. 5.2).
If the position of the extremities are noted 1 and 2, the length of the bar is:
√
l= 12.12 (5.1)

72 5 The Bar Finite Element for Cable
Fig. 5.1 View of three cables split into bar elements. The nodes number are noted
Fig. 5.2 Tension forces F1 and F2 on the extremities of the bar due to its tension
With:
x2 − x1
12 = y2 − y1 (5.2)
z2 − z1
The tension in the bar is:

l − l0
|F| = EA (5.3)
l0
E : Young’s modulus of the material (N/m2 ),

A : mechanical section of the cable (m2 ),
lo : unstretched length of the bar element (m).
The force vectors on the two extremities of the bar are

21
F1 = |F| (5.4)
l
5.2 Tension on Bars 73
12
F2 = |F| (5.5)
l
The components of these forces are:
x1 − x2
F1x = |F|
l
y1 − y2
F1y = |F|
l
z1 − z2
F1z = |F|
l (5.6)
x2 − x1
F2x = |F|
l
y2 − y1
F2y = |F|
l
z2 − z1
F2z = |F|
l
5.2.2 Stiffness Matrix
The stiffness matrix is as follows:
⎛ ⎞
∂ F1x ∂ F1x ∂ F1x ∂ F1x ∂ F1x ∂ F1x
− − − − − −
⎜ ∂ x1 ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2 ⎟
⎜ ⎟
⎜ ∂ F1y ∂ F1y ∂ F1y ∂ F1y ∂ F1y ∂ F1y ⎟
⎜− − − − − − ⎟
⎜ ∂x ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2 ⎟
⎜ 1 ⎟
⎜ ∂ F1z ∂ F1z ∂ F1z ∂ F1z ∂ F1z ∂ F1z ⎟
⎜− − − − − − ⎟
⎜ ⎟
⎜ ∂ x1 ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2 ⎟
K =⎜ ⎟ (5.7)
⎜ ∂ F2x ∂ F2x ∂ F2x ∂ F2x ∂ F2x ∂ F2x ⎟
⎜− − − − − − ⎟
⎜ ∂ x1 ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2 ⎟
⎜ ⎟
⎜ ∂ F2y ∂ F2y ∂ F2y ∂ F2y ∂ F2y ∂ F2y ⎟
⎜− − − − − − ⎟
⎜ ∂ x1 ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2 ⎟
⎜ ⎟
⎝ ∂ F2z ∂ F2z ∂ F2z ∂ F2z ∂ F2z ∂ F2z ⎠
− − − − − −
∂ x1 ∂ y1 ∂z 1 ∂ x2 ∂ y2 ∂z 2
The stiffness matrix is calculated through the derivatives of force components.

For the first component that gives:

∂ F1x
E A ∂l
l0 ∂ x1 (x 1 − x2 ) + |F| ∂(x∂1x−x
1
2)
l − |F|(x1 − x2 ) ∂∂lx1
− =− (5.8)
∂ x1 l2
with
∂l x2 − x1
= (5.9)
∂ x1 l
That gives for the 36 components:
∂ F1x ∂ F1x ∂ F2x ∂ F2x EA 3

− = = =− = 3 l − l 2 lo + lo(x2 − x1 )2
∂ x1 ∂ x2 ∂ x1 ∂ x2 l lo
∂ F1y ∂ F1y ∂ F2y ∂ F2y EA 3
− = = =− = 3 l − l 2 lo + lo(y2 − y1 )2
∂ y1 ∂ y2 ∂ y1 ∂ y2 l lo
∂ F1z ∂ F1z ∂ F2z ∂ F2z EA 3
− = = =− = 3 l − l 2 lo + lo(z 2 − z 1 )2
∂z 1 ∂z 2 ∂z 1 ∂z 2 l lo
∂ F1x ∂ F1y ∂ F2y ∂ F2x ∂ F2y ∂ F2x ∂ F1y ∂ F1x
− =− =− =− = = = =
∂ y1 ∂ x1 ∂ x2 ∂ y2 ∂ x1 ∂ y1 ∂ x2 ∂ y2
EA
= 3 [(x2 − x1 )(y2 − y1 )]
l
∂ F1x ∂ F1z ∂ F2z ∂ F2x ∂ F2z ∂ F2x ∂ F1z ∂ F1x
− =− =− =− = = = =
∂z 1 ∂ x1 ∂ x2 ∂z 2 ∂ x1 ∂z 1 ∂ x2 ∂z 2
EA
= 3 [(x2 − x1 )(z 2 − z 1 )]
l
∂ F1y ∂ F2z ∂ F2y ∂ F2z ∂ F2y ∂ F1z ∂ F1y
− = − ∂∂Fy1z1 = − =− = = = =
∂z 1 ∂ y2 ∂z 2 ∂ y1 ∂z 1 ∂ y2 ∂z 2
EA
= 3 [(y2 − y1 )(z 2 − z 1 )]
l
(5.10)
5.3 Bending of Cables
Cables could have a resistance in bending, such as beams. Beam deformation relates
the curvature of the beam to the couple, such as:
EI
Co = (5.11)
R
Co : the couple on any point of the cable (N.m),
E I : the bending rigidity of the cable (N.m2 ),
R: the radius of the cable at the point (m).
To take into account this behaviour in the numerical model, the cables are split
into bar elements (Fig. 5.3). In case of bending stiffness, there is a couple Co between
consecutive bar elements (Fig. 5.4). This couple leads to forces on the extremities of
theses two elements.
5.3.1 Force Vector
The forces on the extremities of two consecutive bar elements are due to the bending
between the bar elements (Fig. 5.4).
5.3 Bending of Cables 75
Fig. 5.3 The cable is embedded at top right and bends under its own weight. It is modelled with
bar elements. Each bar is straight and articulated with its neighbour
Fig. 5.4 Representation of two consecutive bars. A couple is introduced to take into account the
bending rigidity of the cable. The spring symbolizes the couple
Fig. 5.5 The radius of the curvature is assessed by the circle passing by the extremities of the two
bar elements
The curvature is approximated by the circle passing by the extremities of the two
bar elements. The positions of the extremities of the bars allow assessment of this
radius (Fig. 5.5). From this radius, and if the bending rigidity is known, the model is
able to calculate the couple:
EI
Co = (5.12)
R
The radius (R) is calculated from the position of the extremities:
ABC
R= √ (5.13)
4 p( p − A)( p − B)( p − C)
A (B): length of the first (second) bar (m),

C: distance between the extremities 1 and 3 in Fig. 5.5 (m),
p: the half perimeter (m), where
A+ B +C
p= (5.14)
2
x2 − x1
A = y2 − y1 (5.15)
z2 − z1
x3 − x2
B = y3 − y2 (5.16)
z3 − z2
Fig. 5.6 A virtual displacement (∂ x1) leads to an external work (Fx1 ∂ x1) equal to the internal
work (Co ∂α)
C=A+B
A = |A|
(5.17)
B = |B|
C = |A + B|
Once the couple Co is calculated, the model assesses the forces on the extremities
of the bars using the virtual work principle.
The force component along X on the extremity 1 of the first bar element is esti-
mated by considering a virtual displacement (∂ x1) along the axis x of the extremity 1
(Fig. 5.6). This displacement leads to an external work, considering ∂ x1 small and
consequently Fx1 constant:
We = Fx1 ∂ x1 (5.18)
This virtual displacement also induces a change in the angle (α) between bar
elements.
This virtual displacement leads to a variation of angle between bars (∂α), and
this variation of angle generates an internal work. If ∂ x1 is small, ∂α is small and
consequently Co is constant. That gives
Wi = Co ∂α (5.19)
Because the forces on the extremities of the two bar elements represent the couple
Co there is equality between the works. That leads to:
∂α ∂α ∂α
Fx1 = Co Fx2 = Co Fx3 = Co
∂ x1 ∂ x2 ∂ x3
∂α ∂α ∂α
Fy1 = Co Fy2 = Co Fy3 = Co (5.20)
∂ y1 ∂ y2 ∂ y3
∂α ∂α ∂α
Fz1 = Co Fz2 = Co Fz3 = Co
∂z1 ∂z2 ∂z3
These forces components are:
EI (x2 − x1)AB x2 − x3
Fx1 = +
R sin α A3 B AB
EI (y2 − y1)AB y2 − y3
Fy1 = + (5.21)
R sin α A3 B AB
EI (z2 − z1)AB z2 − z3
Fz1 = +
R sin α A3 B AB
EI (x1 − x2)AB (x3 − x2)AB x3 − 2x2 + x1

Fx2 = + +
R sin α A3 B AB 3 AB
EI (y1 − y2)AB (y3 − y2)AB y3 − 2y2 + y1
Fy2 = + + (5.22)
EI (z1 − z2)AB (z3 − z2)AB z3 − 2z2 + z1
Fz2 = + +
EI (x2 − x3)AB x2 − x1
Fx3 = +
R sin α AB 3 AB
EI (y2 − y3)AB y2 − y1
Fy3 = + (5.23)
R sin α AB 3 AB
EI (z2 − z3)AB z2 − z1
Fz3 = +
R sin α AB 3 AB
On vectorial form:
EI A.AB
F1 = −B
ABR sin α A2
EI A.AB B.AB
F2 = − 2 + +B−A (5.24)
ABR sin α A B2
EI B.AB
F3 = − 2 +A
ABR sin α B
With:
F1 (F2 , F3 ): force on the node 1 (2, 3),
AB: scalar product between the two bar vectors,
A (B): vector along the first (second) bar element,
A (B): length of the first (second) bar element (m),
x1 to z3: the Cartesian coordinates of the three extremities of the two bar ele-
ments (m).
The stiffness matrix is calculated with the derivatives of the force components (Fx1
to Fz3 ) relative to the positions (x1 to z3). This means that the stiffness matrix has
81 components.
5.4 Drag on Cables
5.4.1 Introduction
The drag force on cables is calculated in this model as the contribution of the drag
force on each bar elements. The formulation for the drag is based on the assumptions
of Morrison, as adapted by Landweber and Richtmeyer (see Sect. 4.3.3, p. 46).
The drag amplitudes on bar element used in the model (Fig. 5.7) are
1
|F| = ρCd Dl0 [|c| sin(α)]2
2
1
|T| = f ρCd Dl0 [|c| cos(α)]2 (5.25)
2
The directions of the drag are as follows:
F B ∧ (c ∧ B)
=
|F| |B ∧ (c ∧ B)|
T F ∧ (c ∧ F)
= (5.26)
|T| |F ∧ (c ∧ F)|
F: normal drag (N ), following the assumptions of Landweber,

T: tangential drag (N ), Richtmeyer hypothesis,
B: bar element vector,
Cd : normal drag coefficient,
f : tangential drag coefficient,
D: diameter of the bar element (m),
l0 : length of the bar element (m),
c: water velocity relative to the bar element (m/s),
α: angle between the bar element and the water velocity (radians).
In the equations of drag amplitude, the expressions |c| sin(α) and |c| cos(α) are
the normal and tangential projections on c along the bar element vector.
The length of the bar element used in the formulation of drag amplitude could be
assessed by |B|. That would mean it takes into account the bar element elongation.
Fig. 5.7 Normal (F) and

tangential (T) forces on a bar
element due to the velocity of
water (c)
Generally speaking, a bar elongation is associated with a diameter D reduction by

the Poisson coefficient. Because this Poisson coefficient is not taken into account in
the present modelling, the bar element surface is approximated by Dl0 , where D is
the diameter of the bar and l0 is the unstretched length of the bar element vectors.
All parameters, including the angle α are constant and known for each bar element.
Therefore, the drag can be calculated for each bar element. The drag force for a bar
element is spread over the two vertices of the element at 1/2 per vertex.
5.4.2 Definitions of the Variables
The Cartesian coordinates of the two nodes (1, 2) of the bar element are the following:
x1
1 = y1
z1
x2
2 = y2 (5.27)
z2
The vector bar element is as follows:

5.4 Drag on Cables 81
x2 − x1
B = y2 − y1 (5.28)
z2 − z1
The vector current is

cx
c = cy (5.29)
cz
Generally speaking, cz is null.

The angle between current and B is
c.B
cos(α) = (5.30)
|c||B|
5.4.3 Stiffness of the Normal Force
The normal force on B is

B ∧ (c ∧ B)
F = |F| (5.31)
|B ∧ (c ∧ B)|
That means that the x y and z components are:

Ex
Fx = |F|
|E|
Ey
F y = |F| (5.32)
|E|
Ez
Fz = |F|
|E|
With:
E = B ∧ (c ∧ B)
and
Ex
E = Ey (5.33)
Ez
The x component of the derivative is
Ex E |E| − Ex |E|
Fx = |F| + |F| x (5.34)
|E| |E|2
Which gives for the x y and z components:

Ex |F| Ex
Fx = |F| + Ex |E| −
(Ex Ex + E y E y + Ez Ez )
|E| |E|2 |E|

Ey |F| Ey
Fy = |F| + Ey |E| −
(Ex Ex + E y E y + Ez Ez ) (5.35)
|E| |E|2 |E|

Ez |F| Ez
Fz = |F| + Ez |E| −
(Ex Ex + E y E y + Ez Ez )
|E| |E|2 |E|
For this assessment the derivative of E is required:
E = B ∧ (c ∧ B) + B ∧ (c ∧ B ) (5.36)
This leads to
E = 2(B .B)c − (B .c)B − (B.c)B (5.37)
which is
Ex = 2(B .B)cx − (B .c)Bx − (B.c)Bx
Ey = 2(B .B)c y − (B .c)B y − (B.c)By (5.38)
Ez
= 2(B .B)cz − (B
.c)Bz − (B.c)Bz
with
B .B = Bx Bx + B y By + Bz Bz
B .c = cx Bx + c y By + cz Bz (5.39)
B.c = Bx cx + B y c y + Bz cz
The derivative of the amplitude of the normal force is
1
|F| = ρCd Dl0 |c|2 [sin(α)]2
2
which is
|F| = ρCd Dl0 |c|2 cos(α) sin(α)α (5.40)
The derivative of α is

−1 c.B
α = (5.41)
1 − ( |c||B|
c.B 2
) |c||B|
That gives
−1 c B
α = . (5.42)
1 − ( |c||B|
c.B 2
) |c| |B|
The derivative of the bar element direction is

B B |B| − B|B|
= (5.43)
|B| |B|2
That means that the derivative of α is

−1 c B |B| − B|B|
α = . (5.44)
c.B 2 |c| |B|2
1 − ( |c||B| )
or
−1
α = |B| c x B
x + c y B
y + c z B
z − (c.B)|B|
(5.45)
|B|2 |c| sin α
In this case Bx is the component along x of B .

The derivative of vector B is
Bx
B = By

(5.46)
Bz
which is
∂ Bx ∂ By ∂ Bz
= = = −1
∂ x1 ∂ y1 ∂z 1
∂ Bx ∂ By ∂ Bz
= = =1 (5.47)
∂ x2 ∂ y2 ∂z 2
∂ Bx ∂ Bx ∂ Bx ∂ Bx
= = = =0
∂ y1 ∂ y2 ∂z 1 ∂z 2
∂ By ∂ By ∂ By ∂ By
= = = =0 (5.48)
∂z 1 ∂z 2 ∂ x1 ∂ x2
∂ Bz ∂ Bz ∂ Bz ∂ Bz
= = = =0
∂ x1 ∂ x2 ∂ y1 ∂ y2
On vector form and for the nine coordinates of the triangular element it is
−1
∂B
= 0
∂ x1 0
0
∂B
= −1
∂ y1 0
0
∂B
= 0 (5.49)
∂z 1 −1
1
∂B
= 0
∂ x2 0
0
∂B
= 1
∂ y2 0
0
∂B
= 0
∂z 2 1
The derivative of the norm of vector B is
Bx Bx + B y B y + Bz Bz
|B| = (5.50)
|B|
Which gives for the nine coordinates of the triangular element:
∂|B| −Bx
=
∂ x1 |B|
∂|B| −B y
=
∂ y1 |B|
∂|B| −Bz
=
∂z 1 |B|
∂|B| Bx
=
∂ x2 |B|
∂|B| By
=
∂ y2 |B|
∂|B| Bz
= (5.51)
∂z 2 |B|
5.4.4 Stiffness of the Tangential Force
The tangential force on the bar element is
F ∧ (c ∧ F)
T = |T| (5.52)
|F ∧ (c ∧ F)|
Following the definition of F:
[B ∧ (c ∧ B)] ∧ {c ∧ [B ∧ (c ∧ B)]}
T = |T| (5.53)
| [B ∧ (c ∧ B)] ∧ {c ∧ [B ∧ (c ∧ B)]} |
It follows:
[(B.B)(c.c) − (B.c)2 ](B.c)B
T = |T| (5.54)
|[(B.B)(c.c) − (B.c)2 ](B.c)B|
or
[|B|2 |c|2 − (|B||c| cos α)2 ]|B||c| cos αB
T = |T| (5.55)
|[|B|2 |c|2 − (|B||c| cos α)2 ]|B||c| cos αB|
and
cos αB
T = |T| (5.56)
| cos α||B|
The x y and z components are:
cos αBx
Tx = |T|
| cos α||B|
cos αB y
T y = |T| (5.57)
| cos α||B|
cos αBz
Tz = |T|
| cos α||B|
The derivative of Tx is:
cos αBx (cos αBx ) | cos α||B| − cos αBx (| cos α||B|)
Tx = |T| + |T| (5.58)
| cos α||B| (| cos α||B|)2
cos αBx
Tx = |T|
| cos α||B|
|T|
+ (cos αB − sin αα Bx )
| cos α||B| x
|T| cos αBx Bx Bx + B y By + Bz Bz cos α
− | cos α| − sin αα |B|
(| cos α||B|)2 |B| | cos α|
(5.59)
T |T|
Tx = |T| (cos αBx − sin αα Bx )
x
+
|T| | cos α||B|
Tx Bx Bx + B y By + Bz Bz cos α
− | cos α| − sin αα |B| (5.60)
| cos α||B| |B| | cos α|
Ty |T|
Ty = |T| + (cos αBy − sin αα B y )
|T| | cos α||B|
Ty Bx cos α
− | cos α| + B y By + Bz Bz |B| − sin αα |B| (5.61)
| cos α||B| Bx | cos α|
Tz |T|
Tz = |T| + (cos αBz − sin αα Bz )
|T| | cos α||B|

Tz Bx Bx + B y By + Bz Bz cos α
− | cos α| − sin αα |B| (5.62)
| cos α||B| |B| | cos α|
The derivative of the amplitude of the tangential force is
1 d
|T| = f ρCd Dl0 |c|2 [cos(α)]2 (5.63)
2 2
which is
d
|T| = − fρCd Dl0 |c|2 cos(α) sin(α)α (5.64)
2
Chapter 6
The Node Element
Abstract The modelling for nodes in contact with the sea bed is described. The
forces as well as the stiffness are given in case of contact as well as of friction.
Keywords Forces contact with the sea bed · Drag on the sea bed.
6.1 Principle
The contact of a marine structure with the sea bed has to be taken into account. It
is of great importance for structures such as chains lying on the sea-bed or bottom
trawls.
In the following sections a few forces related to this contact are described.
6.2 Contact on Bottom
In this model, the main hypothesis for these contact forces is that the bottom is
elastic. That means that if a node is in contact with the bottom, the force reaction
(N) is vertical and equal to the product of the node depth (m) in the soil by the soil
stiffness (N/m).
6.2.1 Force Vector
The vertical force on a node due to its potential contact with the bottom is
i f z < Z b Fz = Bk (Z b − z) (6.1)
i f z ≥ Z b Fz = 0 (6.2)

88 6 The Node Element
With:
Fz : the vertical force on the node (N),
Bk : the bottom stiffness (N/m),
Z b : the vertical position of the bottom (m),
z: the vertical position of the node (m).
∂ Fz
i f z < Zb − = Bk (6.3)
∂z
∂ Fz
i f z ≥ Zb − =0 (6.4)
∂z
6.3 Drag on Bottom
Contact of a node with the bottom could lead to a wearing force. This force is taken
into account when there is a movement of the structure on the bottom. This force is
horizontal and opposite to the motion. This wearing depends on the depth on which
the node digs the bottom, on the bottom stiffness, and on the node speed displacement
on the bottom.
6.3.1 Force Vector
As mentioned earlier (Sect. 6.2, p. 87), the vertical force on a node due to its contact
(z < Z b ) to the bottom is:
Fc = Bk (Z b − z) (6.5)
With:
Fc : the vertical force on the node due to the contact to the bottom (N),
Bk : the bottom stiffness (N/m),
Z b : the vertical position of the bottom (m),
z: the vertical position of the node (m).
The drag force on the bottom has been modelled as a function of the displacement
speed of the node on the bottom. Figure 6.1 shows this relation.
6.3 Drag on Bottom 89
Fig. 6.1 Example of amplitude of wearing force |F| depending on the node displacement speed on
the bottom |V|
|V|
i f |V| < Vl |F| = Fc B f (6.6)
Vl
i f |V| ≥ Vl |F| = Fc B f (6.7)
With:
Vx
V = Vy (6.8)
Vz
The components of speed are calculated as follows:
x − xp
Vx = (6.9)
Δt
y − yp
Vy = (6.10)
Δt
z − zp
Vz = (6.11)
Δt
Vx (Vy , Vz ): component of the speed of the node along the x (y, z) axis (m/s),
x (y, z): coordinate of the node along the x (y, z) axis (m) calculated at time t,
x p (y p , z p ): previous coordinate of the node along the x (y, z) axis (m) calculated
at time t − Δt.
Two cases are defined: a high-speed case (|V| ≥ Vl ) and a low-speed case (|V| <
Vl ). The wearing force is calculated in the two cases such as there is continuity
between the two cases (at |V| = Vl ).
6.3.1.1 High-Speed
In this case, |V| ≥ Vl .

That means that the components of this force are the following:
Vx
Fx = −Fc B f (6.12)
|V|
Vy
Fy = −Fc B f (6.13)
|V|
Vz
Fz = −Fc B f (6.14)
|V|
6.3.1.2 Low-Speed
In this case, |V| < Vl .

That means that the components of this force are the following:
Vx
Fx = −Fc B f (6.15)
Vl
Vy
Fy = −Fc B f (6.16)
Vl
Vz
Fz = −Fc B f (6.17)
Vl
6.3.2.1 High-Speed

∂ Fx Fc B f ∂ Vx V2
=− |V| − x (6.18)
∂x |V|2 ∂x |V|

∂ Fx Fc B f ∂ Vy Vx Vy
=− − (6.19)
∂y |V|2 ∂y |V|
6.3 Drag on Bottom 91

∂ Fx Vx Fc B f Vx Vz ∂ Vz
= Bk B f − − (6.20)
∂z |V| |V|2 |V| ∂z

∂ Fy Fc B f Vx Vy ∂ Vx
= (6.21)
∂x |V|2 |V| ∂ x

∂ Fy Fc B f ∂ Vy Vy2
=− |V| − (6.22)
∂y |V|2 ∂ y |V|

∂ Fy Vy Fc B f Vx Vz ∂ Vz
= Bk B f − − (6.23)
∂z |V| |V|2 |V| ∂z

∂ Fz Fc B f Vx Vz ∂ Vx
= (6.24)
∂x |V|2 |V| ∂ x

∂ Fz Fc B f Vy Vz ∂ Vy
= (6.25)
∂y |V|2 |V| ∂ y

∂ Fz Vz Fc B f ∂ Vz Vz2 ∂ Vz
= Bk B f − |V| − (6.26)
∂z |V| |V|2 ∂z |V| ∂z
With:
∂ Vx 1
= (6.27)
∂x Δt
∂ Vy 1
= (6.28)
∂y Δt
∂ Vz 1
= (6.29)
∂z Δt
The stiffness matrix becomes:

⎛ ⎞
Vx2 Vx Vy Vx Vz ⎛ ⎞
|V| − |V| |V| |V|
B f Fc ⎜ ⎟ B f Bk 0 0 Vx
K =− 2 ⎜ Vx Vy Vy2 Vy Vz ⎟− ⎝ 0 0 Vy ⎠
|V| − |V|
(6.30)
|V| Δt ⎝ |V|
2
|V| ⎠ |V| 0 0 Vz
Vx Vz Vy Vz Vz
|V| |V| |V| − |V|
6.3.2.2 Low-Speed
∂ Fx Fc B f ∂ Vx
=− (6.31)
∂x Vl ∂ x
∂ Fx
=0 (6.32)
∂y
∂ Fx Vx
= Bk B f (6.33)
∂z Vl
∂ Fy
=0 (6.34)
∂x
∂ Fy Fc B f ∂ Vy
=− (6.35)
∂y Vl ∂ y
∂ Fy Vy
= Bk B f (6.36)
∂z Vl
∂ Fz
=0 (6.37)
∂x
∂ Fz
=0 (6.38)
∂y
∂ Fz Vz Fc B f ∂ Vz
= Bk B f − (6.39)
∂z Vl Vl ∂z
The stiffness matrix becomes:

⎛ Fc ⎞
−Bk Vx
Δt 0
Bf
K =− ⎝ 0 Fc −Bk Vy ⎠ (6.40)
Vl Δt
Fc
0 0 Δt − Bk Vz
Chapter 7
Validation
Abstract Some cases of validation of the modelling are given. These are compared
with flume tank tests, sea trials, and other models. They are about tractrix shape
of netting, netting stretched under its own weight, netting of hexagonal meshes
deformed by a water current, hydrostatic pressure in a bag of netting, cod-end of
trawl with a catch, bottom trawl at sea, cubic fish farm and bending of a rigid cable.
Keywords Tractrix shape · Stretched netting · Deformation of hexagonal meshes ·

Hydrostatic pressure on netting · Cod-end of trawl · Bottom trawl · Cubic fish farm ·
Bending cable
7.1 Tractrix
The shape of the meridian of a cylinder of netting of inextensible twines held between
two circular rings is a tractrix, if the axis of the two rings are identical and coplanar
with meshes diagonals.
In the case of a cylinder of stretched netting of 100 meshes around, 50 meshes
along, a radius of 1 m at one extremity and 0.048599 m at the other, and a mesh side
of 0.05 m, the shape is as displayed in Fig. 7.1 [14].
The accuracy of the model depends on the number of nodes used (Table 7.1). The
model uses 32–662 nodes and two planes of symmetry.
7.2 Diamond Mesh Netting Stretched by its Weight
This check is done by comparing the results of the model based on triangular elements
with a model where each twine is modelled by an elastic bar. This comparison is
taken from [16].

94 7 Validation
Fig. 7.1 Cylinder of inextensible netting held between two circular rings
Table 7.1 Tractrix shape (x, y) and accuracy (%) of the model, where x and y are the analytical
solution; x is along the axis and y is radial. The accuracy on y depends on the number of nodes in
the model (from 32 to 662)
x (m) y (m) 662 (%) 298 (%) 84 (%) 32 (%)
0 1
0.403501 0.739032 0.02 0.22 1.4 −1.1
0.844094 0.546168 0.00 0.19 1.2 −2.7
1.303628 0.403636 −0.01 0.14 1.0 −1.8
1.773173 0.2983 0.00 0.19 1.5 −2.3
2.248093 0.220453 −0.02 0.17 1.3 −2.4
2.725923 0.162922 −0.03 0.15 1.0 −3.6
3.205334 0.120404 −0.07 0.18 1.0 −2.8
3.685607 0.088983 −0.11 0.17 0.6 −3.2
4.166349 0.065761 −0.15 0.16 0.2 −1.9
4.647348 0.048599
The mesh panel is square and consists of 1600 meshes. The elongation rigidity
(EA) of the twines is 10000 N, their diameter is 0.01 m, the side of the mesh is
1.2 m, the length of the upper edge is 32 m, and the density of the net is 2000 kg/m3
(Fig. 7.2).
The model uses 1050 triangular elements and 512 nodes with a vertical plane
of symmetry (Figs. 7.2 and 7.3b). The comparison is made with a reference model
where each side of mesh (twine vector) is modelled with an elastic bar (Fig. 7.3a).
This reference model uses 3136 bars and 1625 nodes with a plane of symmetry. The
forms calculated by the two models are quite similar (Fig. 7.3).
The forces involved here are the netting weight and the twine tension (Sects. 4.3.9
p. 68 and 4.3 p. 36).
7.2 Diamond Mesh Netting Stretched by its Weight 95
(a) (b)
Fig. 7.2 Calculation of the shape of a net held by its top border. The initial shape of the model is
unbalanced (a) and the final one is balanced (b). Only the triangular elements are represented
(a) (b)
Fig. 7.3 Equilibrium of a net held by its top edge and stretched by its own weight: a model where
each twine is modelled as an elastic bar; b model using triangular elements, with only the twines
drawn
96 7 Validation
7.3 Hexagonal Mesh Net Held Vertically in the Current
The results of the model using triangular elements for netting with hexagonal meshes
are compared with those of a model using bar elements for each twine. The mesh
panel is square and consists of 18 by 33 meshes and 3564 twines. The elongation
rigidity of the twines is 3000 and 0.0003 N in compression. The diameter of the twines
is 1 mm, and their length is 19 mm. The length of each edge is 1 m. The density of
the material is considered equal to that of sea water (1025 kg/m3 ). The net is held
by its four edges perpendicular to a current of 1 m/s of sea water.
The first model uses 924 triangular elements and 495 nodes (Fig. 7.4a, b), whereas
the second model uses 3564 bars and 2446 nodes (Fig. 7.4c).
The results of the two models are similar. The maximum displacement is 0.182 m
for the first model and 0.184 m for the second. The drag force is 54.10 N for the first
and 54.04 N for the second.
(a) (b) (c)

Fig. 7.4 Equilibrium of a net held by its four edges in a current perpendicular: a the twines in
the model using triangular elements; b the triangular elements; c the twines in the model using bar
elements. The shapes are similar
7.3 Hexagonal Mesh Net Held Vertically in the Current 97
Fig. 7.5 Bag of netting with 26.5 kg of water. Comparison between measurements (a) and the
model using triangular elements (b). Only twines are shown in (b)
Convergence is obtained in 29 iterations with the first model compared with 296
iterations for the second model. This acceleration is related to the reduction in the
number of nodes in the model using triangular elements.
This comparison is based on [17] .
7.4 Hydrostatic Pressure
The results of the model using triangular elements are compared with measurements
made by [13]. These measures involve a net bag partially filled with water bags
(Fig. 7.5). The pressure from the weight of the bags is implemented as in Sect. 4.3.6
(p. 64), but in this case the pressure is modelled as a hydrostatic pressure:
p = ρgh (7.1)
p: pressure exerted by the catch on the net (Pa),

g: gravity (9.81 m/s2 ),
h: height in relation to the upper limit of the catch (m).
The test conditions are as follows:
98 7 Validation
Mesh size: 37.2 mm,

Number of meshes around: 50,
Number of meshes along: 50,
Catch volume: 0.0265 m3 ,
Catch density (ρ): 1000 kg/m3 ,
Radius of the hoop above: 0.25 m
The model uses 742 nodes, 1360 triangular elements, one bar for closing the
netting at the bottom, and two symmetry planes. This comparison comes from [18].
7.5 Cod-End with Catch in the Current
A cod-end is the backmost part of a trawl where the catch of fish builds up. The
results of the model are compared with measurements made in test tank on cod-
ends partially filled with water [1]. The pressure of the catch is implemented here as
follows (see Sect. 4.3.6, p. 64):
1
p = ρCd v2 (7.2)
2
p: catch pressure on the net (Pa),
Cd : drag coefficient (1.4),
v: current amplitude (m/s).
The distance between the front of the catch and the extremity of the cod-end is
inserted into the model as data because this distance was measured during the tests.
Figure 7.6 shows the model output (net) and the flume tank measurements (cross).
The comparison shows that the model gives a pretty good description of the cod-end
with the catch.
Fig. 7.6 Comparison of flume tank tests (cross) and the numerical model outputs (mesh) for a scale
(1/3) model of North Sea cod-end with 300 kg of catch
7.6 Full Cod-End 99
Fig. 7.7 Cod-end of netting subject to constant internal pressure
7.6 Full Cod-End
A long and full cod-end subject to constant internal pressure presents a maximal
diameter. This maximal diameter depends on the number of meshes around N and
the mesh side m by the following analytical equation [14]:
Nm
Dmax = 4 √
π 6
In the case of a cod-end close at one extremity of 100 meshes along (N), and a mesh
side of 0.05m (m), the shape is as displayed in Fig. 7.7. The accuracy of the model
on the maximal diameter is 0.015%.
7.7 Bottom Trawl
Several series of measurements on a bottom trawl were carried out during a sea trial
on a French vessel. The results of the numerical model were compared with these
measurements [20] (Fig. 7.8, Table 7.2).
For the measurements at sea, the vessel was equipped with measurement systems
suitable for trawling. Several measurements were carried out:
• the position of the doors (immersion and distance),
• the distance between the headline and the bottom,
• the speed over ground and speed relative to the water,
• the warps and bridles tension.
100 7 Validation
Fig. 7.8 Shape of the bottom trawl assessed by the model. Only 1 twine on 5 is drawn. The doors
are modelled by 4 bars. The catch is on the right
Table 7.2 Differences between tests at sea and simulation

Mean − SD Mean + SD Simulation
Warp tension (kg) 1966 3121 2300
Top bridle tension (kg) 864 1370 980
Bottom bridle tension (kg) 609 972 830
Vertical opening (m) 3.5 4.3 3.4
SD: standard deviation
7.7 Bottom Trawl 101
For the modelling of such complex structure, numerous parameters have to be

determined:
• the design of the netting, which includes the number of netting panels, the size of
each panel in meshes number, the mesh size, the twine diameter, stiffness, density
and drag coefficients.
• the design of cables, the number of cables, the cable stiffness, length, density and
drag coefficients,
• the links between elements, which define among other the seams between netting
panels,
• the modelling of doors, which is determined in the present case by four bars (visible
on Fig. 7.8), on which the weight, the drag and the lift are applied. These forces
are calculated from data provided by the door maker,
• the water depth and the towing speed,
• the catch volume and the drag coefficient on the catch,
• the wearing coefficient on the sea bottom and the stiffness of the bottom,
• the floats repartition, which are on the headline in the present case, with the floata-
bility and volume of each float.
Measurements on the trawl are highly variable. The results of model calculation
are generally close to measured quantities.
7.8 Cubic Fish Cage
Tests were carried out on models of a fish cage in the flume tank of Boulogne/mer
[21]. The cage consisted of 4 side panels of 23 horizontal by 26 vertical meshes
and a bottom panel of 23 by 23 meshes. The net had a mesh side of 35 mm and
a twine diameter of 2.2 mm. The four bottom corners were tightened with 3 kg of
lead sinkers. The size of the cage top was 1 by 1 m. The water speed was 0.5 m/s.
Figure 7.9 compares the flume tank test and the simulation.
(a) (b)
Fig. 7.9 Qualitative comparison between the deformation of a cubic cage in a flume tank (a) and
simulation (b)
102 7 Validation
Fig. 7.10 Vertical deflection of a beam calculated with the model. The beam is fixed on the left
and free to bend on its own weight on the right. The conditions are the same as in the text except
for the bending rigidity, which is (E I = 16.493 N.m2 ), ten times less than the case of Table 7.3 and
Fig. 7.11 to highlight the deformation
Table 7.3 Vertical deflection of the beam deflection calculated with the model in function of bar
elements number and error relative to the analytical deflection (18.2 mm)
Number of bars 5 8 10 12 16 20 30 40
Simulated deflection (mm) 18.9 18.5 18.4 18.3 18.3 18.3 18.2 18.2
Error % 4.0 1.5 0.97 0.67 0.36 0.23 0.082 0.039
Fig. 7.11 Error of the model relative to the analytical deflection in function of the number of bar
elements
7.9 Bending of Cable
The model of bending of cables (Sect. 5.3, p. 74) is compared with a beam deformation
(Fig. 7.10) in the thin beam theory. In this case the deflection is well known. In case
of a cantilever the analytical equation of the deflection is as follows:
7.9 Bending of Cable 103
−W l 4
y= (7.3)
8E I
y: the vertical deflection of the free extremity of the cantilever (m),
l: the length of the cantilever (m),
w: the linear weight of the cantilever (N/m),
E I : the bending rigidity (N.m2 ).
In case of a beam 1 m long (l), with a density of iron (7800 kg/m3 ), a diameter of
2 cm, and a rigidity (E I ) of 164.93 N.m2 , the deflection is 18.2 mm.
Table 7.3 and Fig. 7.11 show the vertical deflection of the beam calculated with
the model in function of bar element number. The model is shown to be valid. The
larger the number of bar elements, the smaller the error.
References
1. Anon (1999) PREMECS FAIR Program CT96 1555, Final report 1t December 1996 31st
November 1999
2. Bessonneau JS, Marichal D (1998) Study of the dynamics of submerged supple nets. Ocean
Eng 27(7):563–583
3. Chang SY (2004) Studies of Newmark method for solving non linear systems: (I) Basic analysis.
J Chin Inst Eng 27(5):651–662
4. Deuflhard P (2004) Newton methods for non-linear problems, Affine invariance and adaptive
algorithms. Springer series in computational mathematics. ISSN 0 179–3632. ISBN 3-540-
21099-7
5. Desai CS, Abel JF (1972) Introduction to the finite element method: a numerical method for
engineering analysis. Van Nostrand Reinhold, New York
6. Ferro RST (1988) Computer simulation of trawl gear shape and loading.In: Proceedings of
word symposium on fishing gear and fishing vessel design.Marine Institute, Saint Johns, pp
259–262
7. Hallam MG, Heaf NJ, Wootton LR (1977) Dynamics of marine structures. CIRIA Underwater
Engineering Group, Londres
8. Landweber L, Protter MH (1947) The shape and tension of a lightflexible cable in a uniform
current. J Appl Mech 14:121–126
9. Le Dret H, Priour D, Lewandowski R, Chagneau F (2004) Numerical simulation of a cod end
net. Part 1. Equilibrium in a uniform flow. J Elast 76(2):139–162
10. Lee C-W, Lee J-H, Cha B-J, Kim H-Y, Lee J-H (2005) Physical modeling for underwater
flexible systems dynamic simulation. Ocean Eng 32:331–347
11. Niedzwiedz G, Hopp M (1998) Rope and net calculations applied to problems in marine
engineering and fisheries research. Arch Fish Mar Res 46:125–138
12. O’Neill FG (2004) The influence of bending stiffness on thedeformation of axi-symmetric
networks.In: Proceedings of OMAE’04, June 20–25, 2004. Vancouver Canada
13. O’Neill FG, O’Donoghue T (1997) The fluid dynamic loadingon catch and the geometry of
trawl cod-ends.Proc Roy Soc London Ser Math Phys Sci 453:1631–1648
14. O’Neill FG, Priour D (2009) Comparison and validation of two models of netting deformation.
J Appl Mech 76(5):1–7
15. O’Neill FG, Xu L (1994) Twine flexural rigidity and meshresistance to opening, ICES CM/B:31
16. Priour D (1999) Calculation of net shapes by the finite element method with triangular elements.
Commun Numer Meth Eng 15:755–763
17. Priour D (2002) Analysis of nets with hexagonal mesh using triangular elements. Int J Numer
Meth Eng 56:1721–1733. doi:10.1002/nme.635

Science, DOI: 10.1007/978-94-007-6844-4, © The Author(s) 2013
106 References
18. Priour D (2005) FEM modelling of flexible structures made of cables,bars and nets. In: Soares
G, Garbatov Y, Fonseca N (eds) Proceedings of the IMAM conference:Maritime transportation
and exploitation of ocean andcoastal resources. Taylor and Francis, London, pp 1285–1292
19. Priour D (2006) Twines equilibrium in a finite element dedicatedto hexagonal mesh netting.In:
ESAIM: Proceedings, October 2007, vol. 22, pp 140–149
20. Priour D (2012) Rapport final du projet EFFICHALUT,Rapport Ifremer/DCB/RDT/HO/R12-
001
21. Repecaud M, Rodier P (1993) Note interne IFREMER,Compte rendu dessais: Cages pour
l’elevage du poisson en mer, DITI/NPA/93.020
22. Richtmeyer RD (1941) Design and operation of mark IV magnetic mine sweeping gear.Bureau
of ships scientific group report No12. January 1941
23. Rivlin RS (1955) Plane strain of a net formed by inextensiblecords. Indiana Univ Math J
4:951–974
24. Tsukrov I, Eroshkin O, Fredriksson D, Swift MR, Celikkol B (2003)Finite element modeling
of net panels using a consistent net element.Ocean Eng 30(2):251–270
25. Zienkiewicz OC, Taylor RL (1989), The finite element method, McGraw-Hill Book Company
Index
B H
Bar element for cable, 71, 74, 76, 77, Hydrodynamic forces on netting, 28, 46, 67
79, 80, 83 Hydrostatic pressure on netting, 96, 97
Bending cable, 102
Bottom trawl, 99, 101
Boundary conditions in the finite element
M
method, 12
Marine flexible structures, 1, 2
Mesh opening stiffness of netting, 57, 58
C
Cod-end of trawl, 98, 99
Cubic fish farm, 101 N
Newmark method, 24, 25
Newton-Raphson method, 15, 17, 19, 20, 22
D
Deformation of hexagonal meshes, 95, 102
Description of the finite element method, 3, 5 S
Drag on the sea bed, 88
Stretched netting, 93, 99
Symmetry in the finite element method, 9, 10
E
Energy minimisation method, 26
Equilibrium calculation, 25 T
Tension in cables, 71, 72
Tractrix shape, 93, 95, 97, 99
F Triangular finite element for netting, 28–30,
Finite element method, 2 32, 33, 40, 46, 48, 59, 62, 66
Fish catch pressure in cod-end, 64 Twine flexion in netting, 57, 59
Flexion of cables, 74, 77 Twines tension in netting, 27–34, 36, 37, 40,
Forces contact with the sea bed, 87 42, 46–48, 59, 61, 62, 70

Science, DOI: 10.1007/978-94-007-6844-4, © The Author(s) 2013

A Finite Element Method For Netting Application To Fish Cages and Fishing Gear

Uploaded by

Copyright:

Available Formats

A Finite Element Method For Netting Application To Fish Cages and Fishing Gear

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Finite Element Method For Netting Application To Fish Cages and Fishing Gear

Uploaded by

Copyright:

Available Formats

SPRINGER BRIEFS IN ENVIRONMENTAL SCIENCE

For further volumes:

A Finite Element Method

ISSN 2191-5547 ISSN 2191-5555 (electronic)

Library of Congress Control Number: 2013938031

Ó The Author(s) 2013

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

2 Finite Element Method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

4 The Triangular Finite Element for Netting . . . . . . . . . . . . . . . . . . 27

4.3 The Forces on the Netting . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

5 The Bar Finite Element for Cable . . . . . . . . . . . . . . . . . . . . . . . . 71

6 The Node Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

7.8 Cubic Fish Cage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

Keywords Marine flexible structures · Finite element method

1.1 Why Fishing Cages and Fishing Gears?

D. Priour, A Finite Element Method for Netting, SpringerBriefs in Environmental 1

1.2 Why the Finite Element Method?

1.3 Why for Netting and Cable?

1.4 Why a Book?

D. Priour, A Finite Element Method for Netting, SpringerBriefs in Environmental 3

2.1.1 Field of Numerical Points

2.1.2 Finite Elements

2.2 A Simple Example

Fig. 2.3 Polygon of n cords

2.3 Nodes Position, Forces on Nodes, and Stiffness Matrix

F : derived efforts with respect to position (N/m),

2.4 Local and Global Forces and Stiffness

F(3) = F(3) + a (2.6)

K (3, 3) = K (3, 3) + e (2.10)

In the case of symmetrical structures in a symmetrical environment it could be advan-

The stiffness matrix would be:

In this case the ranking of the node coordinates is a, b, c, d, e, f .

The stiffness matrix would be:

That gives for a symmetry plane O X Y passing by the node of coordinates

That gives for a symmetry plane OY Z passing by the node of coordinates a, b, c:

That gives for a symmetry plane O Z X passing by the node of coordinates

2.6 Boundary Conditions

Abstract The modelling of structure mechanics is a matter for finding equilibrium of

Keywords Equilibrium calculation · Newton-Raphson method · Newmark method ·

3.1 Newton-Raphson Method

3.1.1 One Dimension

D. Priour, A Finite Element Method for Netting, SpringerBriefs in Environmental 15

Fig. 3.1 The equilibrium of

or, following the previous relations,

Fig. 3.3 Residue of force for each Newton-Raphson method iteration

3.1.2 Two Dimensions

In this case the derivative of the forces is calculated relatively to x and y:

The Newton-Raphson method accesses the equilibrium solution through itera-

3.1.3 Several Dimensions

3.1.3.1 Main Variables

− ∂∂ Fx11 − ∂∂ Fx21 . . − ∂∂ Fxn1

As mentioned earlier, the Newton-Raphson-method is an iterative one. The steps are

The node displacements (hk ) are calculated:

The new position of nodes is deduced:

3.1.4 Singularity of the Stiffness Matrix

3.1.4.1 Additional Stiffness

3.1.4.2 Additional Mechanical Behaviour

3.1.4.3 Displacement Limit

The derivative is: