MultiFEBE (version 2.0.

1)
Reference Manual

July 2024
2
Copyright Notice and Disclaimers

Copyright © 2014-2022 Universidad de Las Palmas de Gran Canaria

Jacob D.R. Bordón
Guillermo M. Álamo
Juan J. Aznárez
Orlando Maeso
This program is free software: you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation, either version 2 of the License, or (at your option) any later
version.

This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
Proper acknowledgment should be given in publications resulting from the use
of these software.

3
4
Acknowledgments

We would like to thank all the code users through the years, in particular to María Castro, María
José Orjuela, Francisco González, Carlos Romero, and especially to Ángel Gabriel Vega, who is giving
momentum to the project and it is the author of the tutorials. Their experience have been very valuable
for debugging and developing the software. We would like to make a special mention to Luis Alberto
Padrón for pushing and supporting this open source enterprise from many angles.

This software has been developed with the support of research projects:
• PID2020-120102RB-I00, funded by the Agencial Estatal de Investigación of Spain,
MCIN/AEI/10.13039/501100011033.

• ProID2020010025, funded by Consejería de Economía, Conocimiento y Empleo (Agencia Canaria

de la Investigación, Innovación y Sociedad de la Información) of the Gobierno de Canarias and
FEDER.

• BIA2017-88770-R, funded by Subdirección General de Proyectos de Investigación of the Ministerio

de Economía y Competitividad (MINECO) of Spain and FEDER.

5
6
Contents

1 Overview 9
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 How to download . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 How to install the binaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.1 GNU/Linux *.deb installer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2 GNU/Linux *.sh installer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 How to compile the source code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 How to use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 Input file 15
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 General sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1 Section problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.2 Section frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3 Section settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.4 Section export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.5 Section bem formulation over boundaries . . . . . . . . . . . . . . . . . . . . . 20
2.3 Low-level entities of the model (mesh – geometric entities) . . . . . . . . . . . . . . . . . . 23
2.3.1 Section nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.2 Section elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.3 Section parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 High-level entities of the model (physical entities) . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.1 Section materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.2 Section cross sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4.3 Section regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4.4 Section fe subregions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.4.5 Section boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.4.6 Section be bodyloads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5 Auxiliary entities of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.5.1 Internal points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.5.2 Internal elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.5.3 Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Model conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6.1 Section symmetry planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6.2 Section conditions over be boundaries . . . . . . . . . . . . . . . . . . . . . . 44
2.6.3 Section conditions over fe elements . . . . . . . . . . . . . . . . . . . . . . . . 45
2.6.4 Section conditions over nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.6.5 Section incident waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3 Output files 49
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2 Nodal solutions file (*.nso) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3 Element solutions file (*.eso) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4 Gmsh results file (*.pos) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

A How to compile the source code 57

A.1 GNU/Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
A.2 Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

7
8 CONTENTS

B Gmsh - How to export a mesh in MSH file format version 2.2 63

C GiD *.bas template file 65

Chapter 1

Overview

In this chapter, a brief overview of MultiFEBE download, install and usage is given. Since MultiFEBE
is open source (GPLv2 license), it can also be compiled by you, see Appendix A.

1.1 Introduction
MultiFEBE is a solver for performing linear elastic static and time harmonic analyses of continuum and
structural mechanics problems comprising multiple interacting regions. It is based on the Finite Element
Method (FEM) and the Boundary Element Method (BEM), which are combined in order to offer many
advanced features not found elsewhere. These distinctive features have been published in many scientific
papers, from which [1, 2, 3, 4, 5] can be highlighted. No doubt, this program is heir to boundary element
codes published by Professor Jose Domínguez [6].
Linear elastic solid materials are available for regions modeled by finite elements or boundary elements
in static and time harmonic analysis, whereas linear poroelastic (Biot’s theory) and inviscid fluid materials
are available only for regions modeled by boundary elements in time harmonic analysis (wave propagation
problems).
It has a full set of linear elastic finite elements:
• Solid (or continuum) finite elements:
– For two-dimensional plane strain / plane stress problems:
∗ 3 or 6 nodes triangular element.
∗ 4, 8 or 9 nodes quadrilateral element.
• Structural finite elements:
– 2D/3D bar element (2 nodes).
– 2D/3D beam elements (doubly symmetric cross-section):
∗ 2 and 3 nodes Euler-Bernoulli and Timoshenko straight element [7, 8].
∗ 2, 3 and 4 nodes curved element based on the degeneration from solid (Timoshenko theory)
[9].
– 3D shell elements (Reissner-Mindlin theory):
∗ 3 and 6 nodes triangular shell element based on the degeneration from solid (full/selec-
tive/reduced integration) [7].
∗ 4, 8 and 6 nodes quadrilateral shell element based on the degeneration from solid (full/s-
elective/reduced integration) [7].
∗ 3 and 6 nodes triangular MITC element (locking-free): MITC3 [10], MITC6a [10].
∗ 4, 8 and 9 nodes quadrilateral MITC element (locking-free): MITC4 [11], MITC8 [12],
MITC9 [13].
• Discrete finite elements:
– 2D/3D discrete translational and translational-rotational spring-dashpot 2 node elements.
– 2D/3D mass elements.
It also has a full set of boundary elements for linear elastic solids, poroelastic media and inviscid
fluids:

9
10 CHAPTER 1. OVERVIEW

• For two-dimensional plane strain / plane stress problems:

– 2, 3 and 4 line ordinary or crack boundary elements.
– 2, 3 and 4 line body load elements.
• For three-dimensional problems:
– 3 and 6 triangular ordinary and crack boundary elements.
– 4, 8 and 9 quadrilateral ordinary and crack boundary elements.
– 2, 3 and 4 line body load elements (only in elastic regions).
– 3 and 6 triangular body load elements.
– 4, 8 and 9 quadrilateral body load elements.
Finite elements can be coupled to ordinary boundary, crack boundary and body load elements in
order to study Soil-Structure, Fluid-Structure and Soil-Fluid-Structure Interaction in wave propagation
problems. This type of mcoupling
Figure 1.1 shows an example of the modeling capabilities, where an Offshore Wind Turbine installed
on a jacket structure founded on suction caissons. Finite elements model the whole foundation and
structure, which includes additional masses in order to incorporate Water-Structure Interaction, and
boundary and body load elements (geometrically coincident with suction caissons skirts and lids) for
modeling Soil-Structure Interaction. Soil free-surface discretization is not shown for the sake of clarity.

Figure 1.1: Jacket-based Offshore Wind Turbine founded on suction caissons

1.2 How to download

The binaries and the source code of MultiFEBE are available at the research group webpage and at
GitHub. MultiFEBE is released under GPLv2 license. That means that you may copy, distribute and
modify the software as long as you track changes/dates in source files. Also, any modifications to or
software including (via compiler) GPL-licensed code must also be made available under the GPL along
with build & install instructions.
MultiFEBE is available for GNU/Linux and Windows 64-bit operating systems. Binaries are released
in *.deb and *.sh installer packages for GNU/Linux, and in an *.exe installer for Windows. Source
code can be compiled using cmake + make + gfortran in GNU/Linux, and through MSYS2/mingw-w64
+ cmake + make + gcc-fortran. The program requires multi-core linear algebra libraries: ATLAS or
OpenBLAS; which are successors of single-core LAPACK and BLAS. OpenBLAS is the default choice
as it can easily be used in both GNU/Linux and Windows. The whole package is configured as a static
build, i.e. the program is statically linked against libraries. Details about installing are given below.
1.3. HOW TO INSTALL THE BINARIES 11

1.3 How to install the binaries

1.3.1 GNU/Linux *.deb installer
The installation steps are:
1. Go to GitHub, and download the most recent *.deb release.
2. Make double-click on the installer.
3. Press the install button, and then type your password in the authentication window.

1.3.2 GNU/Linux *.sh installer

The installation steps are:
1. Go to GitHub, and download the most recent *.sh release (e.g. multifebe-x.x.x-Linux.sh).
2. Press Ctrl+Alt+T to open a terminal, and navigate to the folder where it has been downloaded.
3. Change the permissions of the file to allow its execution:

$ chmod +x multifebe-x.x.x-Linux.sh

4. Now, you can execute the installer:

$ ./multifebe-x.x.x-Linux.sh

5. The program files are decompressed by default in the same directory as the installer. The executable
file is placed inside the subfolder bin. You may copy or move the binary multifebe to any folder
where you want to use it, and execute as ./multifebe ... or, if you have administrator privileges
in your system, you might move the binary multifebe to /usr/bin, or create a simbolic link, so
that it is available directly in the terminal as any other command.
For details about options and usage, see below section How to use 1.5.

1.4 How to compile the source code

The procedure to compile from the source code is explained in Appendix A.

1.5 How to use

MultiFEBE is a command-line program that you can run on a Windows terminal (cmd or PowerShell)
or GNU/Linux terminal. Once installed, you can run it by executing multifebe from a terminal as we
already mentioned at the end of the installation steps. The program execution command line is started
by multifebe and followed by a set of arguments which tells the program on how to run (options). For
instance, as we did before, if we execute:

$ multifebe --help

it will then print a brief help about how to run it and the available options:

$ multifebe --help
Usage:
multifebe [options]

Options:
-i, --input STRING Input file name [required]
-o, --output STRING Output files basename [default: input file name]
-m, --memory INTEGER Max. memory allowed (GB) [default: 0 (unlimited)]
-b, --verbose INTEGER Verbose level: 0-10 [default: 1]
-v, --version Program version
-h, --help This help
12 CHAPTER 1. OVERVIEW

Note that each option can start with one or two hyphens, followed by respectively the short or long
name of the option. In the previous case, we could request help from the program by writing either -h
or --help. Some options does not require a value like -h or -v, but other options require a string or a
number, which are written after each of them. Options are self-descriptive from the program help, but
more details are given in Table 1.1.

Option name
Short Long Argument Description

-i --input STRING STRING is a path to the plain text input file

where the definition of the case is read. This
option is thus mandatory.
-o --output STRING STRING is a path to the basename of the out-
put files (plain text) where the results of the
case are printed. This option is optional since
the input file path is taken by default. The
output files differ only in the extension of the
file.
-m --memory INTEGER INTEGER is the maximum memory (in GB) al-
lowed to be used by the program when allo-
cating the main matrices. If the program pre-
dicts that such memory (or more) is required,
then it stops itself. This option is important
when dealing with large models in order to
prevent the computer from memory swapping
(transferring memory data to the hard-drive).
Therefore, it is recommended to use this op-
tion with a setting of around 75% of the avail-
able memory of the computer.
-b --verbose INTEGER INTEGER sets the level of details printed by
the program when executing, being 0 the level
where minimal information is printed and 10
the level where all the steps are printed.

Table 1.1: MultiFEBE command-line arguments (options)

MultiFEBE uses OpenMP in order to take full advantage of multi-core shared-memory computers.
By default, the program uses all the available cores, but it is possible to control the number of cores used
by previously defining the environment variable OMP_NUM_THREADS. Imagine you would like to only use 4
cores, then in GNU/Linux you would have to execute:

$ export OMP_NUM_THREADS=4

In Windows, it would have to execute:

$ set OMP_NUM_THREADS=4

Figure 1.2 shows a MultiFEBE usage flowchart. It operates mainly through plain text files. The
program reads the input file in order to define the case (type of analysis, geometry topology, mesh,
materials, boundary conditions, results to be calculated, etc). Results are written into a set of output
files which depends on the type of results requested in the input file.
Most of the user effort is usually spent in preparing the geometry, mesh and then post-processing
the results. In order to facilitate these tasks, we have included the possibility of reading and writing
Gmsh files. Gmsh is a powerful free and open-source pre- and post-processor. The only Gmsh file
format that MultiFEBE can read is MSH file format version 2.2, which is the default file format up to
Gmsh 3.0.6. Newer Gmsh versions save mesh files in newer mesh file formats by default. However, you
can also export meshes in older file format versions by doing File>Export>Format: Mesh - Gmsh MSH
(*.msh)>Format: Version 2 ASCII when using the GUI, or including the option -format msh2 when
using the command-line. This integration enables the user to perform, for example, extensive parametric
1.5. HOW TO USE 13

Figure 1.2: MultiFEBE usage flowchart. Orange blocks corresponds to MultiFEBE solver files and
processes. Blue blocks corresponds to Gmsh pre- and post-processor files and processes. MultiFEBE is
able to read Gmsh mesh files and write Gmsh postprocessing results files. Green blocks corresponds to
the possibility of generating the mesh in MultiFEBE native format from GiD and a *.bas template file.

studies with relatively little extra effort, and effective visualization of results. We have also created a
GiD *.bas template file which allows exporting the mesh file in the MultiFEBE native format (located in
the bin/ folder). We intend to improve the integration of MultiFEBE solver in GiD and others pre- and
post-processors in the near future. In the same vein as PILEDYN, we have also created some problem-
specific https://www.mathworks.com/MATLAB pre- and post-processors that will be available in our
research group webpage.
The proposed workflow to perform an analysis is:

1. Decide the modelling approach, which will lead to the required topology and geometry. We suggest
using Gmsh for creating the geometry, which allows defining the geometry parametrically through
simple .geo files.

2. Generate the mesh with appropriate element sizes. In Gmsh, this can be easily done by defining
element sizes all over the geometry using variables. Then you can set their value in order to generate
a mesh. Once you have your solver results, you may have to return to this step to generate additional
refined meshes in order to check results convergence.

3. Open your favourite plain text text editor, and write the case input file according to the required
data, as shown in Chapter 2. If you use Gmsh for generating the mesh file, then you will have to
define the path to it in section [settings].
14 CHAPTER 1. OVERVIEW

4. Run the solver. Copy the solver executable to the folder where the case input file is located. Then,
open a terminal in this folder and execute:

$ multifebe -i case.dat

where it has been assumed that the case input file is named case.dat. BEM models may require
a huge amount of memory. We recommend that if the model is relatively big, or you are running
the solver on a laptop, then you have to limit the available RAM for the program by including the
following option:

$ multifebe -i case.dat -m 4

which will stop the solver if the predicted required memory is greater that 4 GB (or any other
quantity smaller than the available memory in the computer). This will prevent the computer from
swapping and halting if such amount of memory is required.

5. Once the solver ends, the results are exported to one or more output files, see Chapter 3. If you
used a Gmsh mesh file (*.msh), Gmsh post-processing files *.pos will be generated. If that is
the case, you can visualize the results in the Gmsh GUI. Other text files containing node-by-node,
element-by-element, or other results can be generated (depending on how you set the case input
file). This text files can be easily processed by spreadsheet programs, MATLAB/Octave, GNU
utilities such as awk, etcetera. We recommend the latter, as it is particularly fast once you learn
how to use it. They can also be plotted through GNU/Linux utilities such as GNUPlot.
Chapter 2

Input file

2.1 Introduction

The input file is a plain text file which contains the case data. In the following, we will assume that
the name of the input file name is input.dat, but any other name and extension will also be valid. The
important fact is that the file must be a plain text file, and thus it must be edited as such in plain text
editors such as Notepad, Notepad++ or others in Windows (but not Word or LibreOffice!), and emacs,
gedit, vim or others in GNU/Linux environments.
The case data is divided into sections. Each section is a block of data related to an aspect of the case
or the solver settings. Each section begins with a line exclusively containing a header with the section
name between brackets, i.e. [section name], and it ends when all the required data has been introduced
or when another section header appears. It is important to note that its location within the file is not
relevant. The data required for a section is defined by one or more lines after the header line. Each section
has its own section format, which can be in the form of simple statements (keyword = values, entity
id: values), or a sequence of numbers and strings. Only the recognized sections are read. Therefore,
comments can be introduced outside the recognized ones (see example below). Recognized sections are
shown in Table 2.1. Some sections are required (mandatory), some sections are required depending on
the data introduced in other sections (dependant), and others are completely optional.

input.dat
[case description] <-- this is an unrecognized section used as a comment section
analysis of soil stresses around a 3x3 pile group

[problem]
n = 3D
type = mechanics
analysis = harmonic

[frequencies]
Hz
lin
100
0.01
10.

[symmetry planes]
plane_zx: symmetry
plane_yz: antisymmetry

...

All the identifiers of model entities (regions, boundaries, elements, nodes, etc) must be integers greater
than 0. Floating point numbers can be expressed as 125. (simple precision), 1.25e2 (simple precision)
or 1.25d2 (double precision). Complex numbers are expressed as (2.d3,1.d1).

15
16 CHAPTER 2. INPUT FILE

Section name Brief description Character

General sections

Type of problem, dimension Rn of the ambient space,

problem mandatory
and analysis to be performed
frequencies Frequencies for a time harmonic analysis dependent
settings Solver settings optional
export Export options and settings optional
bem formulation over boundaries Set BEM collocation strategies optional

Low-level entities of the model (mesh – geometric entities)

nodes Nodes of the mesh dependant

elements Elements of the mesh dependant
Parts of the mesh (a connected set of elements of the
parts dependant
same intrinsic dimension)

High-level entities of the model (physical entities)

materials Materials to be used in the model optional

Cross sections of the model. It contains data which
cross sections complements the geometrical information contained dependant
in the mesh. It is required for structural elements.
regions Regions (or subdomains) of the model mandatory
Subregions of FE regions (a part with elements of
fe subregions dependant
intrinsic dimension Rn (solid) or lower (structure))
Boundaries of BE regions (a part with elements of
boundaries dependant
intrinsic dimension Rn−1 )
Body loads of BE regions (a part with elements of
be body loads dependant
intrinsic dimension from 0D (point) to Rn )

Auxiliary entities of the model

Internal points of BE regions where the solution is

internal points optional
calculated at post-processing
Internal elements of BE regions where the solution is
internal elements optional
calculated at post-processing
groups Group of entities of the same class optional

Model conditions

symmetry planes Symmetry planes of the model optional

conditions over be boundaries Boundary conditions of BE boundaries dependent
conditions over fe elements Boundary conditions of FE elements optional
conditions over nodes Boundary conditions of BE or FE nodes optional
incident waves Input or incident wave fields dependant

Table 2.1: Sections of the data file

2.2. GENERAL SECTIONS 17

2.2 General sections

2.2.1 Section problem

This section defines the main aspects of the case. It is composed of one or more unordered lines containing
one assignment keyword = values. Table 2.2 shows the recognized keywords, the values they can take,
and a brief description of them.

Keyword Requirement Values Description

n mandatory 2 or 2D Two-dimensional ambient space (R2 )

3 or 3D Three-dimensional ambient space (R3 )

type mandatory laplace Laplace problem (unmaintained, it may not work properly)
mechanics Mechanics problem

subtype dependant plane_strain Plane strain problem (only for mechanics in R2 )

plane_stress Plane stress problem (only for mechanics in R2 )

analysis mandatory static Static analysis

harmonic Time harmonic analysis

Table 2.2: List of recognized keywords and values in section problem

In the following example, a static analysis of a plane strain problem is established:

input.dat
...

[problem]
n = 2D
type = mechanics
subtype = plane_strain
analysis = static

...

2.2.2 Section frequencies

This section is required when analysis = harmonic in section [problem]. That is because, for a time
harmonic analysis, it is necessary to specify the frequencies to be analyzed, which are defined in this
section. This section can be present in the input file, or in another auxiliary file whose path is defined in
section [settings] (see section 2.2.3).
The section format is a sequence of numbers and strings. The first line is a string which indicates the
units of the frequency, and it must be Hz (cycles per second) or rad/s (radians per second). The second
line indicates how frequencies are defined, and it must be one of the following:

• list - Frequencies are defined by a given list of frequencies. The list can be unordered.

• lin - Frequencies are defined by a given number of uniformly (linearly) distributed frequencies
between a minimum and a maximum.

• log - Frequencies are defined by a given number of logarithmically distributed frequencies between
a minimum and a maximum.

The third line defines the number of frequencies. If the frequencies are defined by a list, these are
defined in the fourth and the following lines (as much as the number of frequencies). If the frequencies
are defined by a lin or log distribution of frequencies, the fourth and fifth lines respectively contains
the minimum and maximum frequency. The general format of the section depending on how frequencies
are defined (list, lin or log) can be illustrated as:
18 CHAPTER 2. INPUT FILE
input.dat input.dat input.dat
... ... ...

[frequencies] [frequencies] [frequencies]

<units = Hz or rad/s> <units = Hz or rad/s> <units = Hz or rad/s>
list lin log
<n = number of frequencies> <n = number of frequencies> <n = number of frequencies>
<frequency 1> <minimum frequency> <minimum frequency>
<frequency 2> <maximum frequency> <maximum frequency>
...
<frequency n>
... ... ...

In the following, we are going the show three example of each way of defining the frequencies. In the
first place, we show how to define an analysis at frequencies 9 Hz, 2 Hz, 5 Hz and 4 Hz. Secondly, we
show how to define an analysis at 10 frequencies linearly distributed between 10 and 100 rad/s, i.e. 10
rad/s, 20 rad/s, ..., 100 rad/s. Last, we show how to define an analysis at 4 frequencies logarithmically
distributed between 10 and 10000 Hz, i.e. 10 Hz, 100 Hz, 1000 Hz and 10000 Hz.
input.dat (example of list) input.dat (example of lin) input.dat (example of log)
... ... ...

[frequencies] [frequencies] [frequencies]

Hz rad/s Hz
list lin log
4 10 4
9. 10. 10.
2. 100. 10000.
5.
4.

... ... ...

2.2.3 Section settings

This section defines several settings of the solver. It is optional since all the parameters here defined
have reasonable default values. It is composed of one or more lines containing an assignment keyword =
values. Table 2.3 shows the recognized settings and a brief description of each one. These are related
to the boundary element integration, geometrical parameters, solving of linear system of equations and
auxiliary files definition.
Boundary element integration settings have an impact on accuracy and computational cost. It is not
recommended to tweak these values unless you have certain specific needs. For example, if you would like
to obtain very accurate results, then you can use qsi_relative_error=1d-12, so all quasi-singular (or
nearly-singular) integrals are calculated with such accuracy. This obviously has an important impact on
the computational costs related to building the linear system of equations. Singular integrals are evaluated
with high accuracy by default (10−12 ), and it cannot be adjusted by the user. On the other hand, if you
would like to obtain just some fast results, you can use qsi_relative_error=1d-3. Note that boundary
element integration accuracy is not the only source of errors, so do not expect to have such relative
errors in the final results. The parameter qsi_ns_max defines the maximum number of subdivisions
allowed when performing quasi-singular integration, but note that the quasi-singular integration strategy
combines Telles’s transformation and subdivision, so the default value of 16 should never be reach unless
some very picky calculation is requested. The last setting precalsets has impact only on performance,
since what it defines is the integration rules where some components of the integrand are precalculated
and saved in memory.
The geometrical settings are two: the geometrical tolerance for detecting contact between geometrical
entities, and and option which internally collapse the position of nodes sharing the same position within
the geometrical tolerance. You should tweak these parameters when the size of the problem is very small
or very large, since the geometrical tolerance is defined in absolute and not relative terms with respect
to the element or mesh size.
The settings related to the linear system of equations solving allows tweaking how they are solved. You
should use scaling and refining when the condition number is very large, which may happen for coupled
boundary element - finite element models where regions with very different stiffnesses are connected. This
obviously has some impact on the computational cost, but it is a price to pay in certain cases.
The settings related to auxiliary files allows reading some file sections from other files. In the first
place, you can define the path to a file containing the mesh. This file can be in two different file
2.2. GENERAL SECTIONS 19

Keyword Values Default Description

Boundary element integration

Relative error when evaluating quasi-singular

qsi_relative_error ∈ [10−15 , 10−3 ] 10−6
integrals (integration of boundary elements).
Maximum number of subdivisions of the
integration domain when evaluating
qsi_ns_max ≥0 16
quasi-singular integrals (integration of
boundary elements).
Define the number n of pre-calculated
datasets and the number d_j of 1D
precalsets n d_1 ... d_n 8 2 3 4 5 6 7 8 9 Gauss-Legendre quadrature points to be used
for each dataset j (pre-calculation of
integrand components of boundary elements).

Geometrical parameters

Geometric tolerance for detecting contact

between geometrical entities. It has the same
geometric_tolerance >0 10−6 units as the mesh coordinates, i.e. no relative
geometrical tolerance with respect to
element/mesh size is performed.
For each node, a ball of radius equal to the
geometric tolerance and centered at the node
collapse_nodal_pos T or F T
is built, and all nodes inside this ball are
moved to their center of mass

Linear system of equations

Perform a direct solving of the system of

lse_straight T or F T
linear equations
Perform scaling when solving of the system of
lse_scaling T or F F
linear equations
Estimate the condition number when solving
lse_condition T or F F
of the system of linear equations
Refine the solution after solving of the system
lse_refine T or F F
of linear equations

Auxiliary files

Establish how the mesh is read. If mode=0,

then the mesh is read from the input file. If
mode=1 or mode=2, then the mesh is read from
another file specified by file. If mode=1, it is
mesh_file_mode mode file 0
assumed that the mesh is in the native
format described in the present document. If
mode=2, then it is assumed that the mesh is in
the Gmsh MSH file format version 2.2.
If defined, frequencies are read from the
frequencies_file file “”
specified file.‘

Table 2.3: List of recognized settings and values in section settings

20 CHAPTER 2. INPUT FILE

formats: native or Gmsh (MSH version 2.2). The native format is a very easy to manually generate, but
it is also the file format which GiD generates after using the template file multifebe.bas (see Appendix
C). The Gmsh mesh file format 2.2 is the default file format generated by Gmsh from Gmsh version 2.0.0
up until 3.0.6. Newer versions (Gmsh 4.0.0 onwards) generate by default a new file format version 4.
However, you can still export meshes in this older file format versions by doing File>Export>Format:
Mesh - Gmsh MSH (*.msh)>Format: Version 2 ASCII when using the GUI, or including the option
-format msh2 when using the command-line. You could also read from another file the frequencies, for
example, if you save some standard set of frequencies in a file you can use in several cases, e.g. one-third
octave middle frequencies for sound analysis.
Most of the times, you would only need to define the path to a mesh file and its format. For example,
a Gmsh mesh file located in the same directory as the input file input.dat:
input.dat
...

[settings]
mesh_file_mode = 2 "pilegroup.msh"

...

2.2.4 Section export

This section allows the definition of export options. It is optional since all the parameters here defined
have reasonable default values. It is composed of one or more lines containing one assignment keyword
= values. It is meant to configure which and how output files have to be exported. Table 2.4 shows the
recognized export options and a brief description of each of them.
User can activate/deactivate exporting general output files such as node-by-node solutions and element-
by-element solutions in a native format, a file containing the wave propagation speeds of each boundary
element region (particularly useful when using a Biot’s poroelastic region), and a Gmsh post-processing
file containing mainly all the results. More details about the output file formats are given in Chapter 3.
The user can request the calculation of kinematic and stress resultants over BE boundaries and BE
body loads. That means that average kinematics (displacements and rotations) and stress resultants
are calculated for each BE boundary and each BE body load present in the model. This is useful,
for example, for obtaining soil reactions in soil-structure interaction problems. Note that there are to
additional options to this calculation, which are defining the point where the resultant rotations and
moments are calculated and if the symmetry configuration is taken into account or not.
Finally, since each file contains basically integers, real and complex numbers, and the output files
are plain text files, the user can establish how these are written in the file. Files really only contain
integers and reals, whose number of digits, notation, etc, can be defined via keywords integer_format
and real_format respectively. Complex numbers are written as two consecutive real numbers containing
whether the absolute value and argument pair (polar notation), or the real part and the imaginary part
(cartesian notation).
Most of the times, you would not require to use this section. Perhaps, if you would like to directly
have in your output files the complex results in cartesian notation, and you only require 8 digits after
the decimal point (which reduce the file size), you could write the section as:
input.dat
...

[export]
real_format = eng_simple
complex_notation = cartesian

...

2.2.5 Section bem formulation over boundaries

In this section, the user can modify the BEM formulation, i.e. the type of Boundary Integral Equations
(BIEs) to be used when building the BEM equations and the collocation strategy to be used. The default
formulations (BIEs and collocation strategies) work for almost all cases, and any change done here should
be done with care and only when necessary, see below.
There are two fundamental types of BIEs used in MultiFEBE:
2.2. GENERAL SECTIONS 21

Keyword Values Default Description

General output files

Export wave propagation speeds in each

export_wsp T or F F
boundary element region.
export_nso T or F T Export nodal solutions
Define the number N, and list, of nodes for
nso_nodes N n1 ... nN N=-1 which the results should be exported. If N=-1,
results are exported for all nodes.
export_eso T or F T Export element solutions
Export results in Gmsh MSH file format
version 2.2. The default behaviour is F, but it
export_pos T or F -
is automatically set to T if mesh_file_mode=2
(i.e. a Gmsh mesh file is read).

BE boundaries and BE body loads kinematic and stress resultants output file

Export the resultant average displacements

export_tot T or F F and rotations, and total forces and moments
of each BE boundary.
If tot_xm=0, then the rotations and moments
are calculated with respect to the origin
tot_xm 0 or 1 0 (xm = 0). If tot_xm=1, then the moments are
calculated with respect to the centroid of
each BE boundary.
If set, kinematic and stress resultants are
tot_apply_symmetry T or F T calculated taking into account the symmetry
configuration.

Export settings for writing integer, real and complex numbers.

real_format f<W>.<d> eng_double

e<W>.<d>e<E>
en<W>.<d>e<E>
sci_double (e25.16e3) A Fortran-like string (descriptor) with
sci_simple (e16.8e2) the format for floating point numbers
must be used, see Table 2.5.
sci_less (e11.3e2)
eng_double (en27.16e3)
eng_simple (en18.8e2)
eng_less (en13.3e2)
integer_format I<W> auto A Fortran-like string (descriptor) with
max (i11) the format for integer numbers must
be used, see Table 2.5.
auto (<W> = max id length)
complex_notation polar cartesian Complex notation to be used when
exporting: polar (absolute value and
cartesian argument), or cartesian (real part and
imaginary part).
Table 2.4: List of export options and values in section export
22 CHAPTER 2. INPUT FILE

Variable Format descriptor

Integer Iw Iw.m

Decimal Fw.d
Exponential Ew.d Ew.dEe
Real
Scientific ESw.d ESw.dEe
Engineering ENw.d ENw.dEe

w: the number of positions to be used

m: the minimum number of positions to be used
d: the number of digits to the right of the decimal point
e: the number of digits in the exponent part

Table 2.5: Fortran descriptors

Singular BIE (SBIE) It is also called the displacement or pressure BIE, since this equation relates
the primary field variable (displacements in elasticity, pressure in acoustics, ...) at the collocation
point with the the primary and secondary (tractions in elasticity, fluxes, ...) field variables over the
boundaries and the body loads. This BIE can be collocated everywhere (the collocation point is
the location of the Dirac delta of the Green’s function). However, the nodal collocation (collocation
at the boundary element nodes) is the ideal since it leads to the best conditioning of the system.
Non-nodal collocation is also possible, but it should be used only where nodal collocation turns
problematic.
Hypersingular BIE (HBIE) It is also called the traction or flux BIE, since it relates the secondary
field variable with the the primary and secondary field variables over the boundaries and body
loads. This BIE can be collocated only where the discretization has C 1 displacement continuity.
Since MultiFEBE only considers C 0 continuous elements, this means that HBIE must have non-
nodal collocation (except at element interior nodes).
The non-nodal collocation strategy implemented here is the so-called Multiple Collocation Approach1
(MCA). When building the BIE associated with a given node, the MCA consists on building the equation
by adding up as many BIEs as the number of elements sharing the node, being each BIE collocated near
the node but shifted towards inside the element. For a given element, the local coordinates of the
collocation point (ξki , ηki ) contributing to the node k located at (ξk , ηk ) is controlled by the dimensionless
distance δ, which is a measure of the separation between the collocation point and the corresponding
node such that:
Line elements: ξki = (1 − δ) ξk , ξk ∈ [−1, 1] (2.1)
Quadrilateral elements: ξki = (1 − δ) ξk , ξk ∈ [−1, 1] (2.2)
ηki = (1 − δ) ηk , ηk ∈ [−1, 1] (2.3)
Triangular elements: ξki = (1 − δ) ξk + δ/3, ξk ∈ [0, 1] (2.4)
ηki = (1 − δ) ηk + δ/3, ηk ∈ [0, 1] , ξk + ηk ≤ 1 (2.5)
This way δ → 0 means that the collocation point is approaching the node, and δ → 1 means that the
collocation point is approaching the element’s centroid in local coordinates. Formally 0 < δ < 1, but in
practice it should be 0.01 < δ < 0.6. By default, we use is δ = 0.42 for linear elements, δ = 0.23 for
quadratic elements and δ = 0.14 for cubic elements.
The general format for this data section is:
general format of section [bem formulation over boundaries]
[bem formulation over boundaries]
boundary <boundary id>: <keyword: type of formulation> <values>

where the available formulations and their options are given in Table 2.6.
For instance, in a problem with six boundaries, where the collocation strategy in all boundaries must
be nodal, with the exception of boundary 6, where a non-nodal collocation strategy is prefered for all
the nodes along its boundaries, with a displacement towards inside each element of 10% the width of the
element, one should write:
1 J. Domínguez & M.P. Ariza. A direct traction BIE for three-dimensional crack problems. Engng Anal Bound Elem

2000;24:727–38.
2.3. LOW-LEVEL ENTITIES OF THE MODEL (MESH – GEOMETRIC ENTITIES) 23

Keyword Values Description

Ordinary boundaries

SBIE with nodal collocation strategy for all nodes of the boundary. This is
unsafe since at doubled nodes, depending on the boundary conditions, it can
sbie none
lead to a singular linear system of equations. This option can not be used in
truncated boundaries, such as those for surfaces of unbounded regions.
This is the default formulation for ordinary boundaries. SBIE with
nodal collocation strategy for all nodes of the boundary except for the nodes
sbie_boundary_mca δ
at the boundary of the boundary, where MCA is used. The default value is
δ = 0.05.
SBIE with MCA for all nodes of the boundary. The default value is
sbie_mca δ
indicated by Note 1.
HBIE with MCA for all nodes. For using default values (see Note 1), the
hbie δ
user must write a value δ ≤ 0. Other values
Burton and Miller formulation (SBIE+i/kHBIE) for eliminating spurious
dual_bm_same δ frequencies when solving some exterior problems. SBIE and HBIE with
MCA (same δ) for all nodes. The default value is indicated by Note 1.
Burton and Miller formulation (SBIE+i/kHBIE) for eliminating spurious
frequencies when solving some exterior problems. SBIE with nodal
collocation for all nodes of the boundary except for the nodes at the
dual_bm_mixed δS δH
boundary of the boundary where MCA is used (δS ), and HBIE with MCA
(δH ) for all nodes. For SBIE, the default value δS = 0.05. For HBIE, the
default value δH is indicated by Note 1.

Crack-like boundaries

This is the default formulation for crack-like boundaries. Dual

same δ BEM (SBIE/HBIE) formulation for crack boundary elements with MCA for
all nodes. The default value δ is indicated by Note 1.
Dual BEM (SBIE/HBIE) formulation for crack boundary elements. SBIE
with nodal collocation for all nodes of the boundary except for the nodes at
mixed δS δH the boundary of the boundary where MCA is used (δS ), and HBIE with
MCA (δH ) for all nodes. For SBIE, the default value δS = 0.05. For HBIE,
the default value δH is indicated by Note 1.

• General note 1: Formally 0 < δ < 1, but in practice it should be 0.01 < δ < 0.6.
• General note 2: If the user assigns a value δ ≤ 0, default values are considered.
• Note 1: By default, δ = 0.42 for linear elements, δ = 0.23 for quadratic elements, and δ = 0.14 for cubic elements.

Table 2.6: List of options and values in section bem formulation boundaries

input.dat
...

[bem formulation over boundaries]

boundary 1: sbie
boundary 2: sbie
boundary 3: sbie
boundary 4: sbie
boundary 5: sbie
boundary 6: sbie_boundary_mca 0.1

2.3 Low-level entities of the model (mesh – geometric entities)

The low-level entities of the model are the elementary geometric entities (mesh) of the model: nodes,
elements and parts. By default, the mesh is read from the input file by writing by hand the sections
[nodes], [elements] and [parts], as explained below. However, there are other ways to create and
read the mesh.
We have written a template file *.bas for the GiD pre- and post-processor, which allows GiD to
produce a mesh file in our native mesh file format. Then, you can copy the contents of the generated file
to the input file, or use mesh_file_mode = 1 "filepath" option in [settings] to indicate the format
and path to the file. Each “layer” in GiD jargon corresponds to our concept of “part”.
24 CHAPTER 2. INPUT FILE

The program can read directly a mesh file from the Gmsh pre- and post-processor (MSH file format
version 2.2), by using the mesh_file_mode = 2 "filepath" option in [settings] to indicate the format
and path to the file. Each “physical” entity in their Gmsh jargon corresponds to our concept of “part”.

2.3.1 Section nodes

In this section, all the nodes of the model are defined. Nodes are the most elementary part of the mesh,
and they carry geometrical information (position of nodes), but also a functional/physical information
(value of displacement, traction, etc. at that position).
The first line indicates the number of nodes. Then, one line per node indicating the identifier of the
node and its coordinates. The general format of this section is:
general format of section [nodes]
[nodes]
<n = number of nodes>
<node 1 identifier> <x> <y> <z>
<node 2 identifier> <x> <y> <z>
...
<node n identifier> <x> <y> <z>

2.3.2 Section elements

In this section, all the elements of the model are defined. The elements are the fundamental part of the
mesh, they allow the definition of an interpolation of the geometry and physical variables (displacements,
tractions, etc.) supported on the nodes. On the other hand, high-level entities like boundaries and
subdomains are built using a connected set of elements, which here it is called a “part”.
The format of this section is very similar to the corresponding section of the Gmsh file format. The
first line indicates the number of elements. Then, one line per element indicating:
• Element identifier.
• Type of element. It could be introduced via a string: line2, line3, tri3, tri6, quad4, quad8,
quad9; or a number: 1, 8, 2, 9, 3, 16, 10; respectively. Note that the numbers correspond to the
numbers used by the gmsh file format.
• Number of auxiliary tags (greater than 0).
• List of tags, where the first auxiliary tag is mandatory, and corresponds to the identifier of the part
which the element belongs. The rest of the tags are optional and they are read, but they are not
used.
• A list of identifiers corresponding to the nodes of the element.
The general format of this section is:
general format of section [elements]
[elements]
<n = number of elements>
<element 1 identifier> <type> <num. of tags> <list of tags> <list of nodes identifiers>
<element 2 identifier> <type> <num. of tags> <list of tags> <list of nodes identifiers>
...
<element n identifier> <type> <num. of tags> <list of tags> <list of nodes identifiers>

Let’s show an example of a mesh with 10 elements of the type line3 (quadratic line element), where
elements 1 to 5 belongs to part 1, and elements 6 to 10 belongs to part 2:
input.dat
...

[elements]
10
3 line3 1 1 1 3 2
2 line3 1 1 3 5 4
1 line3 1 1 5 7 6
4 line3 1 1 7 9 8
5 line3 1 1 9 11 10
6 line3 1 2 12 14 13
10 line3 1 2 14 16 15
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 25

8 line3 1 2 16 18 17
9 line3 1 2 18 20 19
7 line3 1 2 20 22 21

...

Note that given that we use an identifier for each element, the order is unimportant. The same happens
with other model entities such as nodes and parts.

2.3.3 Section parts

In this section, all the parts of the model are defined. A part is a connected set of elements. Hence, an
element belongs only to one part, and one part can contain several elements. Note that the relationship
between elements and parts was done when defining the elements. The general format of this section is:
general format of section [parts]
[parts]
<n = number of parts>
<part 1 identifier> <part name>
<part 2 identifier> <part name>
...
<part n identifier> <part name>

where <part name> is a string which allows labelling the part for easy identification. However, this is
not used in the rest of the case input file.

2.4 High-level entities of the model (physical entities)

The high-level entities of the model are the classical topological entities: boundaries, and regions (or sub-
domains); which are used to assign material properties and conditions to different parts of the model. For
convenience, four entities are defined: boundaries (boundary element boundaries), be bodyloads (body
loads within a boundary element region), fe subregions (finite element subregions) and regions. These
are defined by writing the corresponding sections [boundaries], [be bodyloads], [fe subregions] and
[regions], which are explained in the below.

2.4.1 Section materials

This section allows the definition of a set of materials and its properties. The type of materials available
are three linear elastic materials: inviscid fluid, elastic solid, and Biot’s poroelastic medium [14]. Each
one of these represent a very different kind of material. The first one represents an acoustic medium,
where only longitudinal P waves can propagate. It is therefore appropriate to model sound propagation
through fluids like air or water. The second one is the common material for modelling structures, soils,
etc. In a continuum region, longitudinal P and transverse S waves can propagate. The last one represent
a porous medium where a compressible fluid is present inside a elastic frame. In a continuum region, two
types of longitudinal waves (P1 and P2) and transverse S waves can propagate.
The section format is a sequence of numbers and strings, whose format can be described as follows:
general format of section [materials]
[materials]
<n = number of materials>
<material 1 identifier> <type of material> <property> <value> <property> <value> ...
<material 2 identifier> <type of material> <property> <value> <property> <value> ...
...
<material n identifier> <type of material> <property> <value> <property> <value> ...

The first line must contain the number of materials to be considered. Next, there must be as many
lines as the number of materials defined. Each line starts with the material identifier (a integer greater
than 0). Then, it follows with a string indicating the type of material to be defined in that line: fluid,
elastic_solid, or biot_poroelastic_medium. After defining the type of material, if follows several
pairs of string (property symbol) and real value (property value), which depends on the type of material.
If the type of material is fluid (which can be used only for time harmonic analysis), then it is
necessary to define two between the three following properties: bulk modulus K, density ρ and wave
propagation speed c. Optionally, you can define an articial hysteretic damping ratio ξ, which is zero by
default. See Table 2.7 for more details.
26 CHAPTER 2. INPUT FILE

Property Math symbol Plain text symbol

Bulk modulus K K
Density ρ rho
√
Wave propagation speed c = K/ρ c
Artificial hysteretic damping ratio ξ xi

Table 2.7: Properties of an inviscid fluid (fluid). Only usable for time harmonic analysis. At least two
between K, ρ and c must be defined. ξ is optional (ξ = 0 by default).

If the type of material is elastic_solid, then it is necessary to define two of the five following
properties: Young’s modulus E, bulk modulus K, Lamé’s first parameter, λ, shear modulus µ and
Poisson’s ratio ν. If a time harmonic analysis is going to be performed, it is mandatory to define the
density. Optionally, you can define the hysteretic damping ratio ξ, which is zero by default. See Table
2.8 for more details.

Property Math symbol Plain text symbol

Young’s modulus (elastic modulus) E E

Bulk modulus K K
Lamé’s first parameter λ lambda
Lamé’s second parameter (shear modulus) µ mu
Poisson’s ratio ν nu
Density ρ rho
Hysteretic damping ratio ξ xi

Table 2.8: Properties of an elastic solid (elastic_solid). At least two between E, K, λ, µ and ν must
be defined. ρ is mandatory for time harmonic analysis. ξ is optional (ξ = 0 by default).

If the type of material is biot_poroelastic_medium, then it is necessary to define two of the five
following properties of the solid phase: Young’s modulus E, bulk modulus K, Lamé’s first parameter, λ,
shear modulus µ and Poisson’s ratio ν. It is mandatory to define the solid and fluid phases densities, as
well as the porosity, coupling parameters, additional density and dissipation constant. Optionally, you
can define the solid phase hysteretic damping ratio ξ, which is zero by default. See Table 2.9 for more
details.
Let’s show a very simple example where we have two materials: water (c = 1480 m/s, ρ = 1000
kg/m3 ) and soil modelled as an elastic solid (E = 60 MPa, ν = 0.4, ρ = 2000 kg/m3 , ξ = 5%). In such a
case, the section should be written as:
input.dat
...

[materials]
2
1 fluid c 1480. rho 1000.
2 elastic_solid E 60.d6 nu 0.4 rho 2000. xi 0.05

...

2.4.2 Section cross sections

In order to define a structural finite element four pieces of information are required:

1. Material properties. It defines the constitutive law, density, and other intrinsic material proper-
ties. When needed by the structural element, these properties are taken from the material assigned
to the region where the element belongs.

2. Reference geometry. It defines the mid-line for beams, mid-surface for plates/shells, etcetera.
This information is provided by the defined mesh.
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 27

Property Symbol Plain text symbol

Solid phase

Young’s modulus (elastic modulus) E E

Bulk modulus K K
Lamé’s first parameter λ lambda
Lamé’s second parameter (shear modulus) µ mu
Poisson’s ratio ν nu
Hysteretic damping ratio ξ xi
Solid density ρs rhos

Fluid phase and coupling parameters

Fluid density ρf rhof

Porosity ϕ phi
Biot’s coupling parameter Q Q
Biot’s coupling parameter R R
Additional apparent density ρa rhoa
Dissipation constant b b

Table 2.9: Properties of a Biot’s poroelastic medium (biot_poroelastic_medium). At least two between
E, K, λ, µ and ν must be defined. ξ is optional (ξ = 0 by default). The other properties are mandatory.

3. Structural element type. It defines the physical behaviour of the element. A given line element
could be a simple bar element with only axial tension/compression stiffness, or a spring-dashpot or
a beam.
4. Cross section properties. It complements the information provided by the mesh, allowing a
full description of the structural element. The cross section properties could be given in a detailed
manner (cross section dimensions), or by providing effective geometrical properties (area, inertias,
...), or by providing effective mechanical properties (stiffnesses, masses, ...).
In this section, both the structural element type and the cross section properties are defined. Both aspects
are globally denoted as a “cross section”.
The first line of this data section defines the number of different cross sections to be defined. Next,
there must be as many lines as the number of cross sections defined. Each of these lines start with the
type of structural element to be defined, then the number of FE subregions where this type of structural
element is assigned, followed by the list of FE subregions. Next, depending on the type of structural
element, different cross section properties are defined. Therefore, the general format of this section is:
general format of section [cross sections]
[cross sections]
<# cross sections>
<type of structural element 1> <# FE subregions> <list of FE subregions ids.> <cross section properties>
<type of structural element 2> <# FE subregions> <list of FE subregions ids.> <cross section properties>
...

In the following subsections, the previous data format will be condensed to:
general format of section [cross sections]
[cross sections]
<# cross sections>
<type of structural element 1> <FE subregions> <cross section properties>
<type of structural element 2> <FE subregions> <cross section properties>
...

where <FE subregions> in reality denotes <# FE subregions> <list of FE subregions ids.>.
The list of available types of structural elements are:
• Discrete element:
– Discrete axial spring-dashpot (spring-dashpot).
28 CHAPTER 2. INPUT FILE

– General discrete translational spring-dashpot (distra).

– General discrete rotational/translational spring-dashpot (disrotra).
– Discrete point mass (pmass).

• Bar finite element (bar).

• Beam finite element:

– Straight beam finite element (strbeam).

– Curved beam finite element based on the degeneration from solid (degbeam).

• Shell finite element based on the degeneration from solid (degshell).

When considering a 3D analysis, most of the structural elements require orienting their cross section,
which leads to a particular cartesian local basis (local axes x′ y ′ z ′ ) where the relevant dimensions and the
internal stress resultants are defined. In the case of one-dimensional structural elements, such as discrete
general springs-dashpots or beams, the orientation of the y ′ axis is defined with the help of an auxiliary
′
reference vector yref , see Figure 2.1, where the final y ′ axis is located on the plane defined by yref ′
and
′ ′
the x axis. Note that the x axis is defined by the tangent direction of the element mid-line. In the case
of two-dimensional structural elements, i.e. plates and shells, the procedure is similar, see Figure 2.2, but
in this case the x′ axis is defined by using an auxiliary reference vector x′ref , being the final x′ axis on
the tangent plane defined by the z ′ axis (coincident with the normal vector at the point) and the plane
defined by x′ref and the z ′ axis.

2.4.2.1 Discrete axial spring-dashpot (spring-dashpot)

This type of structural finite element is a conventional axial spring-dashpot element between two nodes,
and it is denoted as spring-dashpot in the data section. It is a particular case of distra element
type, and much more common. The data introduction is also simpler. Taking into account that <#
FE subregions> <list of FE subregions ids.> is here denoted as <FE subregions>, the syntax to
describe this type of elements is:

• In static analysis, kx′ is the spring stiffness:

format of section [cross sections] for spring-dashpot elements
[cross sections]
...
spring-dashpot <FE subregions> <kx'>
...

• In time harmonic analysis, the stiffness kx′ must be introduced as a complex number to take into
account a possible hysteretic damping, and cx′ is the viscous damping coefficient:
format of section [cross sections] for spring-dashpot elements
[cross sections]
...
spring-dashpot <FE subregions> <kx'> <cx'>
...

2.4.2.2 General discrete translational spring/dashpot (distra)

This type of structural finite element introduces a generalized translational spring-dashpot between two
nodes, and it is denoted as distra in the data section. It requires defining a 2-node line element. Each
element node has 2 or 3 translational degrees of freedom respectively for two- and three-dimensional
analyses, see Figure 2.3. For static analysis, stiffnesses in global or local axes are required. For time har-
monic analysis, it is required to use complex stiffnesses to introduce stiffnesses (real part) and hysteretic
damping (imaginary part), and also viscous damping coefficients.
Mechanical properties can be defined in local or global axes:

• Local axes. It allows to establish the stiffnesses and viscous damping coefficients in local axes,
i.e. axial spring-dashpot in x′ direction (kx′ , cx′ ) and lateral spring-dashpots in y ′ and z ′ directions
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 29

′
yref (free vector) y ′ , e′2 = e′3 × e′1
node j

node i x′ , e′1 x′ , e′1 x′ , e′1

e′ × yref
′ z ′ , e′3
Cross-section y ′ z ′ plane z, ′
e′3 = 1′ ′ |
|e1 × yref
defined by unit normal e′1
z, e3 z, e3 z, e3
y, e2 y, e2 y, e2
x, e1 x, e1 x, e1

(a) e′1 from element (b) e′3 from e′1 × yref

′
(c) e′2 from e′3 × e′1

Figure 2.1: Steps for constructing the local cartesian basis x′ y ′ z ′ of one-dimensional structural elements
from a y ′ axis reference vector yref
′

x′ref (free vector)

ξ2
z ′ , e′3 z ′ , e′3 z ′ , e′3

ξ1 y′ , y ′ , e′2
e′3 × x′ref
e′2 =
|e′3 × x′ref | x′ , e′1 = e′2 × e′3

z z z
y y y
x x x

(a) e′3 from element (b) e′2 from e′3 × x′ref (c) e′1 from e′2 × e′3

Figure 2.2: Steps for constructing the local cartesian basis x′ y ′ z ′ of two-dimensional structural elements
from a x′ axis reference vector x′ref
30 CHAPTER 2. INPUT FILE

(j)
(j) uz (j)
uy uy
y′ ′ (j)
′ (j) x ux
y′ x ux
node j z′
′(e)
′(e) ,C
node i

)
K

e
,C ′(
e)
K ′(
′
yref : y ′ axis reference vector
(i)
y uy z
node i
(i)
ux y x′ axis points from node i to node j
x x y ′ is on the x′ –yref
′
plane.

(a) 2D analysis (b) 3D analysis

Figure 2.3: Degrees of freedom and local axis definition of discrete translational springs/dashpots (distra)

(ky′ , cy′ , kz′ , cz′ ). In two-dimensional analysis, the local stiffness and damping coefficient matrices
are:
   
kx′ 0 −kx′ 0 cx′ 0 −cx′ 0
   
   
 0 k ′ 0 −k ′   0 c ′ 0 −c y ′
K′ =   , C′ =  
y y y
    (2.6)
 −kx′ 0 kx′ 0   −cx′ 0 cx′ 0 
   
0 −ky′ 0 ky ′ 0 −cy′ 0 cy ′

In three-dimensional analysis:
   
kx′ 0 0 −kx′ 0 0 cx ′ 0 0 −cx′ 0 0
   
   
 0 ky ′ 0 0 −ky′ 0   0 cy′ 0 0 −cy′ 0 
   
   
 0 0 kz ′ 0 0 −kz′   0 0 cz ′ 0 0 −cz′ 
K =
′

,
 C =
′



 −kx′ 0 0 kx′ 0 0   −cx′ 0 0 cx ′ 0 0 
   
   
 0 −ky′ 0 0 ky ′ 0   0 −cy′ 0 0 cy′ 0 
   
0 0 −kz′ 0 0 kz′ 0 0 −cz′ 0 0 cz ′
(2.7)
′
In this latter case, in order to uniquely define the local axis, a reference vector yref is required for
′
establishing the local y axis, see Figures 2.2 and 2.3 for more details.
• Global axes. It allows to establish the stiffnesses and viscous damping coeficients directly in global
axes, i.e. springs and dashpots in x, y and z directions: kx , cx , ky , cy , kz and, cz . Therefore, the
relative position of element nodes are unimportant.
Taking into account that <# FE subregions> <list of FE subregions ids.> is here denoted as
<FE subregions>, the syntax to describe this type of elements is:
• In a 2D static analysis, this type of elements can be introduced in local or in global axes as follows:
format of section [cross sections] for distra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'>
distra <FE subregions> global <kx> <ky>
...

′
• In a 3D static analysis, this type of elements can be introduced in local (requires yref ) or in global
axes as follows:
format of section [cross sections] for distra elements
[cross sections]
...
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 31

distra <FE subregions> local <kx'> <ky'> <kz'> <y'refx> <y'refy> <y'refz>
distra <FE subregions> global <kx> <ky> <kz>
...

• In a 2D time harmonic analysis, this type of elements can be introduced in local or in global axes
as follows (note that kx′ and ky′ must be complex numbers):
format of section [cross sections] for distra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <cx'> <cy'>
distra <FE subregions> global <kx> <ky> <cx> <cy>
...

• In a 3D time harmonic analysis, this type of elements elements can be introduced in local (requires
′
yref ) or in global axes as follows (note that kx′ , ky′ and kz′ must be complex numbers):
format of section [cross sections] for distra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <kz'> <cx'> <cy'> <cz'> <y'refx> <y'refy> <y'refz>
distra <FE subregions> global <kx> <ky> <kz> <cx> <cy> <cz>

...

2.4.2.3 General discrete rotational/translational spring-dashpot (disrotra)

This type of structural finite element introduces a generalized translational/rotational spring-dashpot

between two nodes, and it is denoted as disrotra in the data section. It requires defining a 2-node line
element. Each element node has 3 or 6 degrees of freedom respectively for two- and three-dimensional
analyses, see Figure 2.4. For static analysis, stiffnesses in global or local axes are required. For time har-
monic analysis, it is required to use complex stiffnesses to introduce stiffnesses (real part) and hysteretic
damping (imaginary part), and also viscous damping coefficients. This type of element allows modelling
a customized beam, a simple spring-dashpo model of the soil-foundation interaction, etcetera.

(j) (j) (j) (j)

(j) uz , θz uy , θy
uy
(j) y′ ′
θz (j) x
′ ′ ux (j) (j)
y x ux θx
node j z′
′(e)
′(e) ,C
node i
e)

K
,C ′(
e)
K ′(

′
yref : y ′ axis reference vector
(i)
y uy z
node i ux
(i)
y x′ axis points from node i to node j
y ′ is on the x′ –yref
′
(i)
x θz x plane.

(a) 2D analysis (b) 3D analysis

Figure 2.4: Degrees of freedom and local axis definition of discrete translational/rotational springs/dash-
pots (disrotra)

Mechanical properties can be defined in local or global axes:

• Local axes. It allows to establish the stiffnesses and viscous damping coefficients in local axes:

– Axial (x′ translation): kx′ , cx′ .

– Torsional (x′ rotation): krx′ , crx′ .
– Bending on x′ y ′ plane (y ′ translation – z ′ rotation): ky′ , cy′ , krz′ , crz′ , ky′ rz′ , cy′ rz′ .
– Bending on x′ z ′ plane (z ′ translation – y ′ rotation): kz′ , cz′ , kry′ , cry′ , kz′ ry′ , cz′ ry′
32 CHAPTER 2. INPUT FILE

In two-dimensional analysis, the local stiffness and damping coefficient matrices are:
 
kx′ 0 0 −kx′ 0 0
 
 0 −ky′ 
 ky′ ky′ rz′ 0 ky′ rz′ 
 
 ky2′ rz′ 
 0 ky′ rz krz′ 0 −ky′ rz 2 − krz′ 
 ky ′ 
K′ = 


 (2.8)
 −kx′ 0 0 kx′ 0 0 
 
 
 0 −ky′ −ky′ rz′ 0 ky′ −ky′ rz′ 
 
 2
ky′ rz′ 
0 ky′ rz′ 2 − krz′ 0 −ky′ rz′ krz′
ky ′

 
cx′ 0 0 −cx′ 0 0
 
 0 −cy′ 
 cy′ cy′ rz′ 0 cy′ rz′ 
 
 c2y′ rz′ 
 0 cy′ rz crz′ 0 −cy′ rz 2 − crz′ 
 cy ′ 
C =
′


 (2.9)
 −cx′ 0 0 cx′ 0 0 
 
 
 0 −cy′ −cy′ rz′ 0 cy′ −cy′ rz′ 
 
 c2y′ rz′ 
0 cy′ rz′ 2 − crz′ 0 −cy′ rz′ crz′
cy ′

In three-dimensional analysis:
 
k x′ 0 0 0 0 0 −kx′ 0 0 0 0 0
 
 ky′ 0 0 0 ky′ rz′ 0 −ky′ 0 0 0 ky′ rz′ 
 
 kz′ 0 kz′ ry′ 0 0 0 −kz′ 0 −kz′ ry′ 0 
 
 −krx′ 
 krx′ 0 0 0 0 0 0 0 
 kz2′ ry′ 
 
 sym kry′ 0 0 0 kz′ ry′ 0 2 − kry′ 0 
 kz ′ 
 ky2′ rz′ 
′  −ky′ rz′ − krz′ 
K = krz′ 0 0 0 0 2 
 ky ′ 
 
 kx′ 0 0 0 0 0 
 
 ky ′ 0 0 0 ky′ rz′ 
 
 kz′ 0 kz′ ry′ 0 
 
 
 krx′ 0 0 
 
 sym sym kry′ 0 
krz′
(2.10)

An similarly for C′ .
′
In order to uniquely define the local axis, for three-dimensional analysis a reference vector yref is
′
required for establishing the local y axis, see Figures 2.2 and 2.3 for more details.
• Global axes. It allows to establish the stiffnesses and viscous damping coeficients directly in global
axes, and thus the relative position of element nodes are unimportant.
Taking into account that <# FE subregions> <list of FE subregions ids.> is here denoted as
<FE subregions>, the syntax to describe this type of elements is:
• In a 2D static analysis, this type of elements can be introduced in local or in global axes as follows:
format of section [cross sections] for disrotra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <krz'> <ky'rz'>
distra <FE subregions> global <kx> <ky> <krz> <kyrz>
...
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 33

′
• In a 3D static analysis, this type of elements can be introduced in local (requires yref ) or in global
axes as follows:
format of section [cross sections] for disrotra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <kz'> <krx'> <kry'> <krz'> <ky'rz'> <kz'ry'>
(continuation of previous line) <y'refx> <y'refy> <y'refz>
distra <FE subregions> global <kx> <ky> <kz> <krx> <kry> <krz> <kyrz> <kzry>
...

• In a 2D time harmonic analysis, this type of elements can be introduced in local or in global axes
as follows (note that all stiffnesses k□ must be complex numbers):
format of section [cross sections] for disrotra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <krz'> <ky'rz'> <cx'> <cy'> <crz'> <cy'rz'>
distra <FE subregions> global <kx> <ky> <krz> <kyrz> <cx> <cy> <crz> <cyrz>
...

• In a 3D time harmonic analysis, this type of elements elements can be introduced in local (requires
x′ref ) or in global axes as follows (note that all stiffnesses k□ must be complex numbers):
format of section [cross sections] for disrotra elements
[cross sections]
...
distra <FE subregions> local <kx'> <ky'> <kz'> <krx'> <kry'> <krz'> <ky'rz'> <kz'ry'>
(continuation of previous line) <cx'> <cy'> <cz'> <crx'> <cry'> <crz'> <cy'rz'> <cz'ry'>
(continuation of previous line) <y'refx> <y'refy> <y'refz>
distra <FE subregions> global <kx> <ky> <kz> <krx> <kry> <krz> <kyrz> <kzry>
(continuation of previous line) <cx> <cy> <cz> <crx> <cry> <crz> <cyrz> <czry>
...

2.4.2.4 Discrete point mass (pmass)

A point mass element can be used for simplifying the interaction between a given body and a supporting
structure to the body inertial forces reduced to the contact point. Therefore, the reduced body is
considered rigid. It is important to note that it is used only in time harmonic analysis, as it represents
the inertial forces and not the body weight, which must be introduced as a load.
This type of structural element requires defining a 0-dimensional element in the mesh, i.e. 1-node
element. For three-dimensional analysis, it is introduced in the model by adding the corresponding
inertial forces of the body e to a given node i as:
 
M
(e)
M
(e) [ ]T
Finertial = −M(e) · ä(i) =  ( (e) )T  · ux(i) u(i)
(e) 0 1 (i) (i) (i) (i)
(e) y uz θ x θ y θ z (2.11)
M1 M2

where M(e) is the generalized mass matrix, which can be divided into three matrices:
 
(e)
∫ M 0 0
 
(e)  
M0 = I ρ dΩ =  0 M (e) 0  (2.12)
 
Ω
0 0 M (e)
   
(e) (e)
∫  0 r3 −r2 0 Bxy −Bxz
  
(e)    (e) 
M1 = ρ  −r3 0 r1  dΩ =  −Bxy
(e)
0 Byz  (2.13)
   
(e) (e)
Ω
r2 −r1 0 Bxz −Byz 0
   
(e) (e) (e)
∫  2 r 2
+ r 2
−r r −r r J −J −J
3 1 2 1 3
  xx xy xz

(e)    (e) 
M2 = ρ  −r2 r1 r12 + r32 −r2 r3  dΩ =  −Jxy (e) (e)
Jyy −Jyz  (2.14)
   
(e) (e) (e)
Ω
−r3 r1 −r3 r2 r12 + r22 −Jxz −Jyz Jzz

(i)
where Ω is the domain of the body e, ρ is its the density distribution, and rj = xj − xj is the position
vector of each body point with respect to the reduction point/node i. M0 is the translational inertia
34 CHAPTER 2. INPUT FILE

(zero moment of mass), M1 represents the inertia forces and moments due to the mass imbalance (first
moments of mass), and M2 represents the rotational inertia (second moments of mass). If the node i
(e)
coincides with the center of mass of the body, then M1 is a zero matrix. If additionally the body has
(e)
three symmetry planes or if its principal axes coincides with the global axes, then M2 is a diagonal
iωt
matrix. For time harmonic analyses with time factor e , the generalized displacements acceleration
becomes:
(e)
Finertial = −M(e) · ä(i) = ω 2 M(e) · a(i) (2.15)

In the data input, each mass matrix term Mjk have to be defined. In 2D analysis, the mass matrix is
a 3 × 3 matrix because of the degrees of freedom involved:
   
(i)
M11 M12 M13 ux
   
(e)    
Finertial = ω 2 M(e) · a(i) = ω 2  M21 M22 M23  ·  (i)
uy  (2.16)
   
(i)
M31 M32 M33 θz

In 3D analysis, the mass matrix is a 6 × 6 matrix:

   
(i)
M M12 M13 M14 M15 M16 ux
 11   
   
 M21 M22 M23 M24 M25 M26  
(i)
uy 
   
   
2
 M31 M32 M33 M34 M35 M36  
(i)
uz 
(e)
Finertial = ω M · a = ω 
2 (e) (i) ·  (2.17)
  
 M41 M42 M43 M44 M45 M46  
(i)
θx 
   
   
 M51 M52 M53 M54 M55 M56  
(i)
θy 
   
(i)
M61 M62 M63 M64 M65 M66 θz

In order to simplify the data introduction, when a balanced condition can be assumed only the Mkk
non-zero entries have to be defined. Otherwise, a general nonbalanced condition requires defining all
matrix entries. Therefore, the data introduction can be summarized as follows:

• In a 2D analysis, you can introduce only the 3 non-zero diagonal terms for a balanced mass
condition, or the full 3 × 3 matrix for a general unbalanced condition:
format of section [cross sections] for pmass elements
[cross sections]
...
pmass <FE subregions> unbalanced <M11> <M21> <M31> <M21> <M22> <M23> <M13> <M23> <M33>
pmass <FE subregions> balanced <M11> <M22> <M33>
...

• In a 3D analysis, you can introduce only the 6 non-zero diagonal terms for a balanced mass
condition, or the full 6 × 6 matrix for a general unbalanced condition:
format of section [cross sections] for pmass elements
[cross sections]
...
pmass <FE subregions> unbalanced <M11> <M21> <M31> <M41> <M51> <M61> <M12> <M22> ... <M66>
pmass <FE subregions> balanced <M11> <M22> <M33> <M44> <M55> <M66>
...

2.4.2.5 Bar finite element (bar)

This type of structural finite element is a simple tension/compression member [7] between two nodes,
and it is denoted as bar in the data section. It requires using a 2-node line element in the mesh. Each
element node has 2 or 3 translational degrees of freedom respectively for two- and three-dimensional
(e)
analyses, and the only stress resultant is the axial force Nx′ , see Figure 2.5. For time harmonic analysis,
a consistent mass matrix is used, and its density is taken from the material defined for the corresponding
fe subregion and region.
The only required cross section parameter is the area, and it can be introduced in several ways:
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 35

format of section [cross sections] for bar elements

[cross sections]
...
bar <FE subregions> generic <area>
bar <FE subregions> circle <diameter>
bar <FE subregions> hollow_circle <outer diameter> <inner diameter>
bar <FE subregions> rectangle <width> <height>
...

(j)
uy
(j)
node j ux (j)
uz (j)
uy
x′ node j (j) (e)
ux
(i)
uz (i) x′ Nx′
uy
node i ux
(i) x′
(i)
uy
(e)
node i (i)
ux Nx′
y z
y
x x

(a) 2D analysis (b) 3D analysis (c) Axial force

Figure 2.5: Bar finite element description (bar)

2.4.2.6 Straight beam finite element (strbeam)

This type of structural finite element is a conventional straight beam finite element, and it is denoted
as strbeam in the data section. By default, Timoshenko kinematic assumption (plane cross section after
deformation) is assummed for bending, but the simpler Euler-Bernoulli theory is also available (cross
section remains plane and perpendicular to the mid-line after deformation). It can be chosen by using as
the type of structural element:

• Timoshenko beam theory: strbeam or strbeam_t

• Euler-Bernoulli beam theory: strbeam_eb.

The formulation is based on locking-free Timoshenko beam finite element of Friedman & Kostmatka
[8], and Euler-Bernoulli assumptions are simply obtained by nullifying the transverse shear strain factor
(ϕ = 0). The element is built by superposition of axial, torsional and bending behaviour. Line elements
of 2 or 3 nodes can be used in the mesh. For time harmonic analysis, a consistent mass matrix including
rotational inertia is used, and its density is taken from the material defined for the corresponding fe
subregion and region.
A symmetric cross section is assumed, and the required cross section parameters are:

• Area: A.

• Area moments of inertia: Ix′ , Iy′ , Iz′ .

• Shear correction factors: κx′ , κy′ , κz′ . They are used to calculate the effective shear stiffnesses by
correcting the cross section parameters: κx′ Ix′ (torsional constant), κy′ A (y ′ transverse shear area),
κz′ A (z ′ transverse shear area).

In two-dimensional analysis, each node has 3 DOFs (ux , uy , θz ), whereas in three-dimensional analysis
each node has 6 DOFs (ux , uy , uz , θx , θy , and θz ), see Figure 2.6.
The data introduction can be summarized as follows:

• In a 2D analysis under plane strain, a beam can be used to model an infinitely long (hz′ = 1)
wall/plate/shell of thickness hy′ , or an equivalent structural system. It is required the cross section
area A (usual application requires A = hy′ · 1), the inertia Iz′ (usual application requires Iz′ =
1 · h3y′ /12), and the shear correction factor κy′ (usual application requires κy′ = 5/6):
36 CHAPTER 2. INPUT FILE

(j) (j) (j) (j)

(j) uz , θz uy , θy
uy
(j)
θz (j) y′
ux x′ (j) (j)
ux θx
node j
z′
y ′ x′ node i
z′
′
yref : y ′ axis reference vector
(i)
y uy z
node i
(i)
ux y x′ axis points from node i to node j
y ′ is on the x′ –yref
′
(i)
x θz x plane.

(a) 2D analysis (b) 3D analysis

Figure 2.6: Degrees of freedom and local axis definition of straight beam finite elements (strbeam).
Element with 2 node shown.

format of section [cross sections] for strbeam elements

[cross sections]
...
strbeam <FE subregions> <A> <I_z'> <kappa_y'>
...

• In a 3D analysis, you can introduce a generic cross section, or a pre-defined cross section shape
′
via its dimensions. The reference vector yref for defining the y ′ axis is required (see Figure 2.1 for
more details). Therefore, it can be introduced as follows:
format of section [cross sections] for strbeam elements
[cross sections]
...
strbeam <FE subregions> generic <A> <Ix'> <Iy'> <Iz'> <kax'> <kay'> <kaz'> <y'refx> <y'refy> <y'refz>
strbeam <FE subregions> circle <diameter> <y'refx> <y'refy> <y'refz>
strbeam <FE subregions> hollow_circle <outer diam> <inner diam> <y'refx> <y'refy> <y'refz>
strbeam <FE subregions> rectangle <width y'> <height z'> <y'refx> <y'refy> <y'refz>
...

For stress resultants definition, see Figure 2.8 and section about degbeam.

2.4.2.7 Beam finite element obtained from the degeneration of the solid (degbeam)
This type of structural finite element is a beam finite element obtained from the degeneration of the
solid via Timoshenko kinematic assumption (plane cross section after deformation), and it is denoted as
degbeam in the data section. A rectangular cross section of hy′ × hz′ is assumed. It requires using a 2, 3
or 4 node line element in the mesh. In order to avoid locking, a selective integration is used by default,
but it can be changed in element options data section. For time harmonic analysis, a consistent mass
matrix is used, and its density is taken from the material defined for the corresponding fe subregion
and region.
Similarly to strbeam, in two-dimensional analysis each node has 3 DOFs (ux , uy , θz ), and in three-
dimensional analysis each node has 6 DOFs (ux , uy , uz , θx , θy , and θz ), see Figure 2.7.
The data introduction can be summarized as follows:

• In a 2D analysis, a plain strain condition is assumed (width is unitary), being the x′ and y ′ axes
contained in the plane and only the cross section height hy′ is required:
format of section [cross sections] for degbeam elements
[cross sections]
...
degbeam <FE subregions> <hy'>
...

• In a 3D analysis, the cross section dimensions hy′ and hz′ are required together with the reference
′
vector yref for the y ′ axis (see Figure 2.1 for more details):
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 37

z′
(2)
uy
(2) hz′ y′
θz (2)
ux hy′
y ′ x′
(3) z′ y′
(3)
uy x′
θz (3)
ux
z′
hy′ ′
yref : y ′ axis reference vector
(1)
y uy z
(1)
ux y x′ axis points from node i to node j
y ′ is on the x′ –yref
′
(1)
x θz x plane.

(a) 2D analysis (b) 3D analysis

Figure 2.7: Degrees of freedom and local axis definition of beam finite elements degenerated from solid
(degbeam). Element with 3 node shown.

format of section [cross sections] for degbeam elements

[cross sections]
...
degbeam <FE subregions> <hy'> <hz'> <y'ref x> <y'ref y> <y'ref z>
...

The stress resultants are defined as follows:

   

 
′ 

 



N x 
 

σ x ′ x′



 
 
 

 V ′ 
  
 


 y 
 

τ ′
xy ′



 
 ∫ 
 

 V ′   τx′ z′ 
z
= dA (2.18)

 Tx′   A −z ′ · τx′ y′ + y ′ · τx′ z′ 


 
 
 


 
 
 


 My ′  
 z ′ · σx′ x′ 


 
 
 


   
 M ′   
 ′
−y · τ ′ ′


z xx

and they are shown in Figure 2.8.

2.4.2.8 Shell finite element obtained from the degeneration of the solid (degshell)
This type of structural finite element is a shell finite element obtained from the degeneration of the
solid via Reissner-Mindlin kinematic assumption, and it is denoted as degshell in the data section. It
requires using a triangular element (3 or 6 nodes) or a quadrilateral element (4, 8 or 9 nodes) in the
mesh. By default, the locking-free MITC family of elements of Bathe and co-workers [11, 12, 13, 10] with
full integration is used: MITC3, MITC4, MITC6a, MITC8*, MITC9.
(i) (i) (i)
Each element node i has 3 translational DOF (ux , uy , uz ), and 2 local rotational DOFs (α(i)
(i) (i) (i)
and β (i) local rotations related to shell bending) or 3 global rotational DOFs (θx , θy and θz global
rotations). This shell model does not have rotational stiffness in the normal direction, the so called
“drilling” stiffness, thus an artificial stiffness is added when using 6 DOF [9] in order to avoid any
singularity. There are three situations where 6 DOFs are mandatory:
1. When imposing certain boundary conditions (for example symmetry conditions).
2. Along edges or joints between two shells with incompatible orientations.
3. Along joints between more than two shells, at beam-shell joints, and similar singular joints.
Therefore, in practical problems most element nodes can be 5 DOFs. Although we could have implemented
only 6 DOFs element nodes (more homogeneous treatment), we have decided to reduce as much as possible
the number of DOF since we do not use a sparse solver (it is not appropriate for boundary element matrices
which are dense). Internally, an automatic procedure for selecting 5 DOFs or 6 DOFs is implemented
based on the conditions stated above. For time harmonic analysis, a consistent mass matrix is used, and
its density is taken from the material defined for the corresponding fe subregion and region.
38 CHAPTER 2. INPUT FILE

y′
′
z x′ y′
z′
x′
local cartesian axes
z
y curved beam Infinitesimal beam segment
x
global cartesian axes

Vy ′
y′ y′ y′
Nx ′ z′ Nx′ z′ z′
Vz ′ Vz ′
x′ x′ x′
Vy′

Axial force Transverse shear force Transverse shear force

Nx′ Vy ′ Vz′

Tx′ y′ Tx′ y′ y′
z′ z′ z′
x′ x′ x′
My ′ My ′ Mz ′ Mz ′

Torsional moment Bending moment Bending moment

Tx′ My ′ Mz ′

Figure 2.8: Beam stress resultants: Nx′ , Vy′ , Vz′ , Tx′ , My′ , Mz′
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 39

When defining degshell FE subregions, the data format of the cross section definition only requires
the shell thickness and the x′ref reference vector for establising the x′ local axis (see Figure 2.2 for more
details):
format of section [cross sections] for degshell elements
[cross sections]
...
degshell <FE subregions> <thickness> <x'ref x> <x'ref y> <x'ref z>
...

The stress resultants are defined as follows:

   [∫
t/2 ∫ dy ′ /2
]   
 N ′ 
   
 −t/2 −dy ′ /2 σx′ x′ dy dz
′ ′
/dy ′ 

 
 


 x 
 
 [ ∫ ∫ ] 
 

σ x′ x′ 


  
  t/2 ′
dx /2 
 
 


   σ ′ ′ dx′ dz ′ /dx′   
 Ny ′ 
  
 
 [ −t/2 −dx′ /2 y y
]






σy ′ y ′ 



   ∫ t/2 ∫ dy′ /2   
 N ′ ′ 
  
 
 τ ′
′ y ′ dy dz
′
/dy ′ 









 xy 
 
 [ −t/2 −dy /2 ′ x
] 
 

τx′ y′ 


    ∫ t/2 ∫ dy′ /2   
 M ′    ′ (−z ′
) · σ x′ x′ dy dz
′ ′
/dy ′

 ∫ t/2 
 −z ′ · σx′ x′


[∫ −t/2 ∫ −dy′ /2
x
= ] = dz ′ (2.19)

 My ′   
 t/2 dx /2
(−z ′ ) · σy′ y′ dx′ dz ′ /dx′ 
 −t/2 
 −z ′ · σy′ y′ 


 
 
 −t/2 −dx′ /2 
 
 


 
 
 [ ∫ ∫ ] 
 
 


  
′ y′   t/2 ′
dy /2 ′
· ′ ′
/dy ′ 
 
 −z ′ · τx′ y′ 



M x 
 
 −t/2 −dy ′ /2
(−z ) τ ′
xy ′ dy dz 
 
 


 
   [ ∫ t/2 ∫ dy′ /2 ] 
 
 


 V x′   
 
 
 


 
 
 −t/2 −dy ′ /2 x z
τ ′ ′ dy ′ dz ′ /dy ′ 
 

τx′ z′ 


  
  [ ∫ t/2 ∫ dx′ /2 ] 
 
 

 V ′    
  
x τ ′ ′ dx′ dz ′ /dx′
−t/2 −dx′ /2 y z
τy′ z′

and they are shown in Figure 2.9.

2.4.3 Section regions

In a n-dimensional problem, a region (or subdomain) Ω is a n-dimensional entity that is a part or the
whole domain of the problem. In MultiFEBE, a region Ω can be discretized by one of two different
methods:

• BEM (BE region). The region is treated by the BEM. Its discretization is defined by its boundary
∂Ω, which in general is built using a set of boundaries, which in turn is associated with mesh parts
and hence with boundary elements.

• FEM (FE region). The region is treated by the FEM. Its discretization is defined by a set of FE
subregions, which in turn is associated with parts and hence with finite elements. FE region can
only be made of elastic solid materials, i.e. finite elements for fluid or poroelastic medium are not
availabe.

The interaction (coupling) between BE regions is done through their shared boundaries, which must have
opposite orientation for each region, see Figure 2.11. The interaction between FE regions is done through
their shared nodes as usual. The interaction between BE boundaries and FE regions is done through
the BE boundaries and the boundary of the FE elements that shares the same position. The interaction
between BE and FE elements is automatically detected.
The format of this section is explained in the following. The first line indicates the number of regions.
Then, for each region there must be a block of data consisting of several lines of data. The first one is the
region identifier and the region class (discretization method: “fe” or “be”). If the region is a BE region,
then the second line indicates the number of boundaries and a list of boundaries (with their orientation
signs). If the region is a FE region, then the second line indicates the number of FE subregions and a
list of fe subregions. The third line defines the material. Then, only if the region is a BE region, the
fourth line defines the number and a list of BE body loads. Also, only if the region is a BE region and
the analysis is time harmonic, the fifth line defines the number and a list of incident fields. The general
format of the section is:
general format of section [regions]
[regions]
<n = number of regions>

<region 1 identifier> <region class (discretization method): be or fe>

<number of boundaries or fe subregions> <list of boundaries or fe subregions identifiers>
material <list of materials>
[<if be: number of be body loads> <list of be body loads>]
40 CHAPTER 2. INPUT FILE

z′
n
x′ y′
local cartesian axes

infinitesimal side length shell

z
y
x
global cartesian axes

Nx′ Vx ′
Mx′
My ′
Ny ′ z′ z′ z′

x′ y′ x′ y′ x′ y′
Ny′
My ′

Nx′ M x′ Vx′

Membrane extensional forces Bending moments Transverse shear force

Nx′ and Ny′ Mx′ and My′ Vx′

Nx′ y′ M x′ y ′

z′ M x′ y ′ z′ Vy′ z′
Nx′ y′ N x′ y ′
x′ y′ x′ y′ M x′ y ′ x′ y′
Vy ′

Nx′ y′
Mx′ y′

Membrane shear force Twisting moment Transverse shear force

Nx′ y′ M x′ y ′ Vy ′

Figure 2.9: Shell stress resultants: Nx′ , Ny′ , Nx′ y′ , Mx′ , My′ , Mx′ y′ , Vx′ , Vy′
2.4. HIGH-LEVEL ENTITIES OF THE MODEL (PHYSICAL ENTITIES) 41

[<if be and analysis=harmonic: number of incident fields> <list of incident fields>]

...

<region n identifier> <region class (discretization method): be or fe>

<number of boundaries or fe subregions> <list of boundaries or fe subregions identifiers>
material <list of materials>
[<if be: number of be body loads> <list of be body loads>]
[<if be and analysis=harmonic: number of incident fields> <list of incident fields>]

2.4.4 Section fe subregions

A FE subregion is a partition of a FE region. A FE subregion is associated with a part of the mesh. The
general format of the section is:
general format of section [fe subregions]
[fe subregions]
<n = number of FE subregions>
<FE subregion 1 identifier> <part identifier> 0 0
<FE subregion 2 identifier> <part identifier> 0 0
...
<FE subregion n identifier> <part identifier> 0 0

where the last two zeros at the end of the each line are mandatory, and they are going to be used in the
future for additional features.

2.4.5 Section boundaries

In a n-dimensional problem, a boundary Γ is a (n − 1)-dimensional oriented entity that is a part or the
whole boundary of a region. The whole boundary ∂Ω of a region Ω treated by the Boundary Element
Method, BE region in the following, is defined as a set of boundaries:

∂Ω = {. . . , Γj , . . .} (2.20)

whose orientation must be outwards from the BE region. A minus sign before Γj can be used to indicate
the reversion of the orientation of Γj in order to get a compatible ∂Ω.
There are two main classes of boundaries:

• Ordinary. An ordinary boundary is a boundary in the classical sense. It can be connected with
one or two BE regions. In both cases, the boundary can be connected with FE elements.

Ω1
Ω1
n n ∂Ω1 = {. . . , −Γ, . . .} −t −Γ

t Γ t Γ −n
n
∂Ω2 = {. . . , Γ, . . .} t Γ
Ω Ω2
Ω2

Figure 2.10: Ordinary Figure 2.11: Ordinary boundary connected with two regions. Right:
boundary connected with exploded view.
one region

• Crack-like. A special boundary that lies inside a BE region and is composed by two crack-like
sub-boundaries (Γ+ and Γ− ) of opposite orientations. Thus, a crack-like boundary can be connected
only with one BE region. Generally speaking, it is the condensed geometric description of a null
thickness inclusion or void. Its orientation defines the orientation of the positive sub-boundary Γ+ ,
i.e. it defines which face is Γ+ and which face is Γ− .
42 CHAPTER 2. INPUT FILE

n t− = −t Γ−

t Γ n− = −n
w→0
n+ = n
t+ = t Γ+
Ω Ω

Figure 2.12: Crack-like boundary. Right: exploded view.

Each boundary is build up with unique boundary elements and nodes, which are defined by a “part”
of the mesh. Thus, all boundaries must be connected with one and only one mesh part. The orientation
of all boundary elements in that part must be the same, i.e. be compatible, which defines the orientation
of the boundary.
The first line indicates the number of boundaries. Then, one line per boundary indicating the bound-
ary identifier, the identifier of the part that discretize it, and finally the boundary class (ordinary or
crack-like). The general format of the section is:
general format of section [boundaries]
[boundaries]
<n = number of boundaries>
<boundary 1 identifier> <part identifier> <boundary class: ordinary or crack-like>
<boundary 2 identifier> <part identifier> <boundary class: ordinary or crack-like>
...
<boundary n identifier> <part identifier> <boundary class: ordinary or crack-like>

Next, it is shown an example with 4 boundaries, where boundary 1 is the part 1 of the mesh and is an
ordinary boundary, boundary 3 is the part 6 of the mesh and is an ordinary boundary, boundary 4 is
the part 2 of the mesh and is a crack-like boundary, and boundary 2 is the part 3 of the mesh and is an
ordinary boundary:
input.dat
...

[boundaries]
4
1 1 ordinary
3 6 ordinary
4 2 crack-like
2 3 ordinary

...

It is important to highlight that, in case of adjacent boundaries with different boundary conditions
or with geometries such that a discontinuity in tractions will arrive, it is necessary to double (duplicate)
the nodes in the boundaries of such BEM boundaries. A non-nodal collocation strategy is followed by
default in all these boundary nodes.

2.4.6 Section be bodyloads

Body loads in BE regions are defined in this section. The general format follows a simpler pattern as the
previous section. The first line contains the number of BE body loads to be defined, next as many lines
are BE body loads. Each line contains first the BE body load identifier, and last the mesh part which
contains the elements associated with it. The general format can be written as:
general format of section [fe subregions]
[be bodyloads]
<n = number of BE body loads>
<BE bodyload 1 identifier> <part identifier>
<BE bodyload 2 identifier> <part identifier>
...
<BE bodyload n identifier> <part identifier>
2.5. AUXILIARY ENTITIES OF THE MODEL 43

where the last two zeros at the end of the each line are mandatory, and they are going to be used in the
future for additional features.

2.5 Auxiliary entities of the model

2.5.1 Internal points
Internal points are points within a BE region where field values (displacements, pressures, stresses, ...)
are requested. They must not lie at any boundary. They are obtained at post-processing stage after
boundary values are obtained.
There are two ways of requesting the calculation of fields at internal points:
• List of internal points defined by identifiers, region which contains each one, and point coordinates.
general format of section [internal points]
[internal points]
<# internal points>
<internal point id> <region id> <x coord.> <y coord> <z coord.>
...

• From the nodes defined in a mesh (see 2.2.3 for details about mesh file format):
general format of section [internal points from mesh]
[internal points from mesh]
mesh_file = <mesh file format> <path to mesh files>
region_association : <# selected mesh parts>
<selected mesh part id> <region id>
...

The program does not check if each internal point is indeed inside the specified region, so the user is
responsible of the coherence between these data.

2.5.2 Internal elements

To be documented.

2.5.3 Groups
To be documented.

2.6 Model conditions

In this section, the data sections related to the conditions (boundary and interface conditions, loads,
symmetry planes) of the model are explained.
This part of the manual is incomplete as it only deals with conditions for elastic solids.

2.6.1 Section symmetry planes

If present, in this section the symmetry planes of the problem are defined. All symmetry planes are
assumed to be at the origin of coordinates. By using symmetry planes, only a half, quarter or octant of
the model has to be discretized. For BE regions, no additional boundary conditions are required since
influence matrices are built taking symmetry planes into account. For FE regions, the corresponding
boundary conditions are applied to nodes at symmetry planes.
In order to indicate the existence and the nature of the symmetry, it is necessary to define up to three
of the symmetry planes implemented: plane with unit normal n = e1 = (1, 0, 0) (plane yz), plane with
unit normal n = e2 = (0, 1, 0) (plane zx), or plane with unit normal n = e3 = (0, 0, 1) (plane xy); where
each of one can be either of symmetry or antisymmetry. The general format for the section is:
general format of section [symmetry planes]
[symmetry planes]
<"plane_n1", "plane_n2" or "plane_n3"> : <"symmetry" or "antisymmetry">
...

For example, if the plane with normal n = e1 = (1, 0, 0) (plane yz) is a symmetry plane, i.e. fields at
(x1 , x2 , x3 ) are symmetric to fields at (−x1 , x2 , x3 ), then:
44 CHAPTER 2. INPUT FILE

Axes Type Name Values Equations

0 displacement U uk = U
1 traction T tk = T
global
4 infinitesimal rotation field center c, axis a, angle θ uk = θ [a × (x − c)] · ek
10 normal pressure P tk = P nk
2 displacement U u · lk = U
local
3 traction T t · lk = T

Note: k = 1, . . . , n (global or local axes). Local axes are {l1 , l2 , l3 } = {n, t1 , t2 }.

Table 2.10: Boundary conditions for ordinary boundaries

Face Type Name Values Equations

0 displacement U+ u+
k =U
+
Γ+
1 traction T+ t+
k =T
+

0 displacement U− u−
k =U
−
Γ−
1 traction T− t−
k =T
−

Note: k = 1, . . . , n (global axes).

Table 2.11: Boundary conditions for crack-like boundaries

input.dat
...

[symmetry planes]
plane_n1: symmetry

...

2.6.2 Section conditions over be boundaries

In this section, the boundary conditions applied on the boundary elements boundaries are defined. All
boundaries except the interfaces and the boundaries that are coupled with finite elements need to specify
their boundary conditions. If not defined, a traction-free boundary is assumed by default. Here, boundary
conditions for viscoelastic solids are explained.
There are available four boundary conditions for ordinary boundaries, two on global axes, and two
in local axes, see Table 2.10. The two boundary conditions of each group are known displacement and
known traction. For each boundary, the user must specify the boundary condition for each coordinate in
local axes or in global axes, but not mixed.
It is always preferable using the B.C. expressed in global axes (B.C. 0 or 1) when possible because
they do not need additional equations. If the boundary is planar and its normal vector is parallel to any
global axis, then the B.C. expressed in local axes are not needed at all.
There are available two boundary conditions for each face of a crack-like boundary. One establishes
a displacement, while the other establishes a traction, see Table 2.11.
The general format of the section is:
general format of section [symmetry planes]
[conditions over be boundaries]
boundary <boundary id>: <type x> <value x>
<type y> <value y>
<type z> <value z>
<type x (if crack-like: face -)> <value x (if crack-like: face -)>
<type y (if crack-like: face -)> <value y (if crack-like: face -)>
<type z (if crack-like: face -)> <value z (if crack-like: face -)>
...

where note that for two-dimensional problems, the z lines must be removed. An example of a two-
dimensional problem with two boundaries with identifiers 1 and 3, being the first an ordinary boundary
with null displacements and the second a crack-like boundary with traction-free faces:
2.6. MODEL CONDITIONS 45

input.dat
...

[conditions over be boundaries]

boundary 1: 0 0.
0 0.
boundary 3: 1 0.
1 0.
1 0.
1 0.

...

The same example as before but for a three-dimensional problem:

input.dat
...

[conditions over be boundaries]

boundary 1: 0 0.
0 0.
0 0.
boundary 3: 1 0.
1 0.
1 0.
1 0.
1 0.
1 0.

...

Note that if the analysis is time harmonic, the values of the boundary conditions must be introduced as
complex numbers, e.g.(0.,0.).

2.6.3 Section conditions over fe elements

To be documented.

2.6.4 Section conditions over nodes

In this section, the boundary conditions over nodes are applied. For nodes of boundary elements, this
boundary condition over nodes overrides the boundary conditions applied over their boundaries, and the
Tables 2.10 and 2.11 remains valid. For nodes of finite elements, the boundary conditions are applied to
the displacements and the rotation, the type of boundary conditions are 0 for known displacement, and
1 for known force (or moment).
The general format for the section is similar to the conditions over boundaries:
general format of section [conditions over nodes]
[conditions over nodes]
node <node id>: <type x> <value x>
<type y> <value y>
<type z> <value z>
<if fe: type for rot. x or alpha> <value for rot. x or alpha>
<if fe: type for rot. y or beta> <value for rot. y or beta>
<if fe: type for rot. z> <value for rotation z>

For a two-dimensional problem, example of a clamped beam FE node with identifier 67, and a pin
beam FE node with identifier 41:
input.dat
...

[conditions over nodes]

node 67: 0 0.
0 0.
0 0.
node 41: 0 0.
0 0.
1 0.
46 CHAPTER 2. INPUT FILE

2.6.5 Section incident waves

In this section, the incoming waves are defined. The incidence angles are shown in Figure 2.13.

y z

incidence direction

incidence direction P
P

θ y
SH
θ x φ
SV
x
SV
2D 3D

Figure 2.13: Incidence angles for two- and three-dimensional problems

The incident waves can be defined in terms of displacements (unitary displacements), in terms of
stresses (unitary stresses) or in terms of the potentials (unitary potentials). The general format for the
section is:
general format of section [incident waves]
[incident waves]
<number of incident waves>

<incident wave identifier>

<class: plane>
<space> [if space=half-space: <np> <xp> <bc>]
<variable> <amplitude> <x0(1)> <x0(2)> [if 3D: <x0(3)>] [if 3D: <varphi>] <theta>
<xs(1)> <xs(2)> [if 3D: <xs(3)>] <symconf(1)> <symconf(2)> [if 3D: <symconf(3)>]
<region_type> <wave_type>

...

where angles have to be introduced in degrees.

Example of vertical P-wave in terms of displacements in the full-space:
input.dat
...

[incident waves]
1
1
plane
full-space
0 1. 0. 0. 90.
0. 0. 0. 0.
viscoelastic p

...

Example of vertical SV-wave in terms of displacements in the half-space:

input.dat
...

[incident waves]
1
1
plane
half-space 2 0. 1
0 1. 0. 0. 90.
2.6. MODEL CONDITIONS 47

0. 0. 0. 0.
viscoelastic sv

...
48 CHAPTER 2. INPUT FILE
Chapter 3

Output files

3.1 Introduction
Output files are written according to the type of analysis and model, as well as the input file export
section. By default the path and name of the output files is equal to the input file but with an additional
extension. The path and name can be changed when calling the solver from the terminal as explained
in 1.5. There are three main native output files with the following additional extensions: nodal solutions
(*.nso), element solutions (*.eso) and stress resultant solutions (*.tot). The specific file format is
different for static and time harmonic analysis. There is one Gmsh output file (*.pos) which contains
the case mesh and results (MSH file format version 2.2).

3.2 Nodal solutions file (*.nso)

This output file is a plain text file containing the nodal (node by node) results. The file starts with a set
of header lines whose first character is “#” (a comment line in GNU/Linux environments), describing the
file and the meaning of the main data columns. The first 9 columns are always present, and indicates the
following:

Column Value Description

$1 integer Analysis step index. For linear elastic static analysis it is always 0. For
time harmonic analysis it is the frequency index. In future releases,
it will receive the natural frequency index (for modal analysis), or the
time step index (for transient analysis), or the loading step index (for
non-linear static analysis).
$2 float Analysis step value. For linear elastic static analysis it is always 0.0.
For time harmonic analysis it is the frequency value (in Hz or rad/s
depending on the units used in the frequencies section of the input file).
In future releases, it will receive the natural frequency value (for modal
analysis), or the time step value (for transient analysis), or the loading
step value (for non-linear static analysis).
$3 integer Identifier of the region to which the node result belongs. In the case of
finite element nodes, this has no significance since results does not de-
pend on the region. However, in the case of boundary elements belonging
to two regions (interface boundary elements) this allows selecting which
result is required.
$4 integer Region class: 1 (BE region) or 2 (FE region).
$5 integer Region type: 1 (inviscid fluid) or 2 (elastic solid) or 3 (poroelastic
medium).

49
50 CHAPTER 3. OUTPUT FILES

Column Value Description

$6 integer Identifier of the BE boundary (if $4==1) or FE subregion (if $4==2)
to which the node result belongs. In the case of finite element nodes,
this has no significance since results does not depend on the FE sub-
region. However, in the case of boundary elements belonging to two
regions (interface boundary elements) this allows selecting which result
is required.
$7 integer If $4==1, then it indicates the boundary class: 1 (ordinary) or 2 (crack-
like). If $4==2, then it indicates the number of degrees of freedom of
the node.
$8 integer If $4==1, then it indicates the boundary face: 1 (positive normal) or 2
(negative normal). If $4==2, then it is always 0.
$9 integer Identifier of the node.

From column $10 onwards, the meaning of each column differs depending on the problem dimension
(2D or 3D), on the type of analysis (static or time harmonic), on the region class (BE or FE), on the
region type (inviscid fluid, elastic solid, poroelastic medium), and on the number of DOF (if a FE region).
For 2D problems in a static analysis (all regions are only of the elastic solid type):

Column Value Description

$10,$11 float Node (x1 , x2 ) coordinates.

BE region node
$12,$13 float Node displacements (u1 , u2 ).
$14,$15 float Node tractions (t1 , t2 ).

FE region node with 2 DOFs

$12,$13 float Node displacements (u1 , u2 ).
$14,$15 float Node forces (f1 , f2 ).

FE region node with 3 DOFs

$12-$14 float Node displacements/rotation (u1 , u2 , θ3 ).
$15-$17 float Node forces/moment (f1 , f2 , m3 ).

For 3D problems in a static analysis (all regions are only of the elastic solid type):

Column Value Description

$10-$12 float Node (x1 , x2 , x3 ) coordinates.

BE region node
$13-$15 float Node displacements (u1 , u2 , u3 ).
$16-$18 float Node tractions (t1 , t2 , t3 ).

FE region node with 3 DOFs

$13-$15 float Node displacements (u1 , u2 , u3 ).
$16-$18 float Node forces (f1 , f2 , f3 ).

FE region node with 5 DOFs (shell element nodes with local rotations)
$13-$17 float Node displacements/local rotations (u1 , u2 , u3 , α, β).
$18-$22 float Node forces/local moments (f1 , f2 , f3 , mα , mβ ).
3.2. NODAL SOLUTIONS FILE (*.NSO) 51

Column Value Description

FE region node with 6 DOFs

$13-$18 float Node displacements/rotations (u1 , u2 , u3 , θ1 , θ2 , θ3 ).
$19-$24 float Node forces/moments (f1 , f2 , f3 , m1 , m2 , m3 ).

For 2D problems in a time harmonic analysis, where each node variable is written by two consecu-
tive columns containing its real and imaginary parts (if complex_notation = cartesian in [export]
section) or absolute value and argument (if complex_notation = polar in [export] section):

Column Value Description

$10,$11 float Node (x1 , x2 ) coordinates.

BE region node (inviscid fluid)

$12,$13 float Node pressure p (total field).
$14,$15 float Node normal displacement Un (total field).
$16,$17 float Node pressure pinc (incident field).
$18,$19 float Node normal displacement Uninc (incident field).
$20-$23 float Node displacement (U1 , U2 ) (total field).

BE region node (elastic solid)

$12-$15 float Node displacements (u1 , u2 ) (total field).
$16-$19 float Node tractions (t1 , t2 ) (total field).
$20-$23 float Node displacements (uinc inc
1 , u2 ) (incident field).

$24-$27 float Node tractions (tinc inc

1 , t2 ) (incident field).

BE region node (poroelastic medium)

$12,$13 float Node fluid equivalent pressure τ (total field). Note that it is related to
the dynamic pore pressure by τ = −ϕp, where ϕ is the porosity.
$14-$17 float Node solid displacements (u1 , u2 ) (total field).
$18,$19 float Node fluid normal displacement Un (total field).
$20-$23 float Node solid tractions (t1 , t2 ) (total field).
$24,$25 float Node fluid equivalent pressure τ inc (incident field).
$26-$29 float Node solid displacements (uinc inc
1 , u2 ) (incident field).

$30,$31 float Node fluid normal displacement Uninc (incident field).

$32-$35 float Node solid tractions (tinc inc
1 , t2 ) (incident field).

$36-$39 float Node fluid displacement (U1 , U2 ) (total field).

FE region node (only of elastic solid type) with 2 DOFs

$12-$15 float Node displacements (u1 , u2 ).
$16-$19 float Node forces (f1 , f2 ).

FE region node (only of elastic solid type) with 3 DOFs

$12-$17 float Node displacements/rotation (u1 , u2 , θ3 ).
$18-$23 float Node forces/moment (f1 , f2 , m3 ).

For 3D problems in a time harmonic analysis, where as in the 2D case each node variable is written by
two consecutive columns containing its real and imaginary parts (if complex_notation = cartesian in
52 CHAPTER 3. OUTPUT FILES

[export] section) or absolute value and argument (if complex_notation = polar in [export] section):

Column Value Description

$10-$12 float Node (x1 , x2 , x3 ) coordinates.

BE region node (inviscid fluid)

$13,$14 float Node pressure p (total field).
$15,$16 float Node normal displacement Un (total field).
$17,$18 float Node pressure pinc (incident field).
$19,$20 float Node normal displacement Uninc (incident field).
$21-$24 float Node displacement (U1 , U2 , U3 ) (total field).

BE region node (elastic solid)

$13-$18 float Node displacements (u1 , u2 , u3 ) (total field).
$19-$24 float Node tractions (t1 , t2 , t3 ) (total field).
$25-$30 float Node displacements (uinc inc inc
1 , u2 , u3 ) (incident field).

$31-$36 float Node tractions (tinc inc inc

1 , t2 , t3 ) (incident field).

BE region node (poroelastic medium)

$13,$14 float Node fluid equivalent pressure τ (total field). Note that it is related to
the dynamic pore pressure by τ = −ϕp, where ϕ is the porosity.
$15-$20 float Node solid displacements (u1 , u2 , u3 ) (total field).
$21,$22 float Node fluid normal displacement Un (total field).
$23-$28 float Node solid tractions (t1 , t2 , t3 ) (total field).
$29,$30 float Node fluid equivalent pressure τ inc (incident field).
$31-$36 float Node solid displacements (uinc inc inc
1 , u2 , u3 ) (incident field).

$37,$38 float Node fluid normal displacement Uninc (incident field).

$39-$44 float Node solid tractions (tinc inc inc
1 , t2 , t3 ) (incident field).

$45-$50 float Node fluid displacement (U1 , U2 , U3 ) (total field).

FE region node (only of elastic solid type) with 3 DOFs

$13-$18 float Node displacements (u1 , u2 , u3 ).
$19-$24 float Node forces (f1 , f2 , f3 ).

FE region node (only of elastic solid type) with 5 DOFs (shell element nodes with local rotations)
$13-$22 float Node displacements/local rotations (u1 , u2 , u3 , α, β).
$23-$32 float Node forces/local moments (f1 , f2 , f3 , mα , mβ ).

FE region node (only of elastic solid type) with 6 DOFs

$13-$24 float Node displacements/rotation (u1 , u2 , u3 , θ1 , θ2 , θ3 ).
$25-$36 float Node forces/moment (f1 , f2 , f3 , m1 , m2 , m3 ).

3.3 Element solutions file (*.eso)

This output file is a plain text file containing the element (element node by element node) results (stress
resultants on finite elements). In a future release, element by element stress resultants over each boundary
element will also be included. The file starts with a set of header lines whose first character is “#” (a
3.3. ELEMENT SOLUTIONS FILE (*.ESO) 53

comment line in GNU/Linux environments), describing the file and the meaning of the main data columns.
The first 13 columns are always present, and indicates the following:

Column Value Description

$1 integer Analysis step index. For linear elastic static analysis it is always 0. For
time harmonic analysis it is the frequency index. In future releases,
it will receive the natural frequency index (for modal analysis), or the
time step index (for transient analysis), or the loading step index (for
non-linear static analysis).
$2 float Analysis step value. For linear elastic static analysis it is always 0.0.
For time harmonic analysis it is the frequency value (in Hz or rad/s
depending on the units used in the frequencies section of the input file).
In future releases, it will receive the natural frequency value (for modal
analysis), or the time step value (for transient analysis), or the loading
step value (for non-linear static analysis).
$3 integer Identifier of the region to which the element node result belongs. In the
case of finite element nodes, this has no significance since results does
not depend on the region. However, in the case of boundary elements
belonging to two regions (interface boundary elements) this allows se-
lecting which result is required.
$4 integer Region class: 1 (BE region) or 2 (FE region).
$5 integer Region type: 1 (inviscid fluid) or 2 (elastic solid) or 3 (poroelastic
medium).
$6 integer Identifier of the BE boundary (if $4==1) or FE subregion (if $4==2) to
which the element result belongs. In the case of boundary elements be-
longing to two regions (interface boundary elements) this allows selecting
which result is required.
$7 integer If $4==1, then it indicates the boundary class: 1 (ordinary) or 2 (crack-
like). If $4==2, then it indicates the finite element dimension.
$8 integer If $4==1, then it indicates the boundary face: 1 (positive normal) or 2
(negative normal). If $4==2, then it indicates the finite element type.
$9-$11 integer Element identifier, dimension and type.
$12,$13 integer Element node index and number of degrees of freedom.
$14 integer Node identifier.

For 2D problems in a static analysis:

Column Value Description

$15,$16 float Node (x1 , x2 ) coordinates.

Straight beam finite element

$17-$19 float Axial force (Nx′ ), shear force (Vy′ ) and bending moment (Mz′ ).

Bar finite element

$17 float Axial force (Nx′ ).

Discrete translational spring finite element

$17,$18 float Force on each spring: Nx′ , Ny′ .

Discrete translational/rotational spring finite element

$17-$19 float Force/moment on each spring: Nx′ , Ny′ , Mz′ .
54 CHAPTER 3. OUTPUT FILES

For 3D problems in a static analysis:

Column Value Description

$15-$17 float Node (x1 , x2 , x3 ) coordinates.

Straight beam finite element

$18-$23 float Axial force (Nx′ ), shear forces (Vy′ , Vz′ ), torsional moment (Mz′ ) and
bending moments (My′ , Mz′ ).

Bar finite element

$18 float Axial force (Nx′ ).

Discrete translational spring finite element

$18-$20 float Force on each spring: Nx′ , Ny′ , Nz′ .

Discrete translational/rotational spring finite element

$18-$23 float Force/moment on each spring: Nx′ , Ny′ , Nz′ , Mx′ , My′ , Mz′ .

Shell finite element

$18-$25 float Membrane forces (Nx′ , Ny′ , Nx′ y′ ), bending moments (Mx′ , My′ ), twist-
ing moment (Mx′ y′ ) and shear forces (Vx′ , Vy′ )).

For 2D problems in a time harmonic analysis, where each node variable is written by two consecu-
tive columns containing its real and imaginary parts (if complex_notation = cartesian in [export]
section) or absolute value and argument (if complex_notation = polar in [export] section):

Column Value Description

$15,$16 float Node (x1 , x2 ) coordinates.

Straight beam finite element

$17-$22 float Axial force (Nx′ ), shear force (Vy′ ) and bending moment (Mz′ ).

Bar finite element

$17,$18 float Axial force (Nx′ ).

Discrete translational spring finite element

$17-$20 float Force on each spring: Nx′ , Ny′ .

Discrete translational/rotational spring finite element

$17-$22 float Force/moment on each spring: Nx′ , Ny′ , Mz′ .

For 3D problems in a time harmonic analysis, where each node variable is written by two consecu-
tive columns containing its real and imaginary parts (if complex_notation = cartesian in [export]
section) or absolute value and argument (if complex_notation = polar in [export] section):

Column Value Description

$15-$17 float Node (x1 , x2 , x3 ) coordinates.

Straight beam finite element

$18-$29 float Axial force (Nx′ ), shear forces (Vy′ , Vz′ ), torsional moment (Mz′ ) and
bending moments (My′ , Mz′ ).

Bar finite element

3.4. GMSH RESULTS FILE (*.POS) 55

Column Value Description

$18,$19 float Axial force (Nx′ ).

Discrete translational spring finite element

$18-$23 float Force on each spring: Nx′ , Ny′ , Nz′ .

Discrete translational/rotational spring finite element

$18-$29 float Force/moment on each spring: Nx′ , Ny′ , Nz′ , Mx′ , My′ , Mz′ .

Shell finite element

$18-$33 float Membrane forces (Nx′ , Ny′ , Nx′ y′ ), bending moments (Mx′ , My′ ), twist-
ing moment (Mx′ y′ ) and shear forces (Vx′ , Vy′ ).

The definition of the local cartesian basis and the stress resultants sign criteria for beams and shells
are established according to Section 2.4.2 (see Figures 2.1, 2.2, 2.8 and 2.9).

3.4 Gmsh results file (*.pos)

The Gmsh output file (*.pos) contains the case mesh and results (MSH file format version 2.2). It is a
plain text containing all this data which can be read by Gmsh. The following set of results are exported
to Gmsh:

• Fluid pressure (positive faces). Only boundary elements.

• Fluid pressure (negative faces). Only boundary elements.
• Solid total displacements (positive faces). Boundary elements and finite elements.
• Solid total displacements (negative faces). Only boundary elements.

• Solid total tractions (positive faces). Only boundary elements.

• Solid total tractions (negative faces). Only boundary elements.
• FE nodal forces/reactions. Only finite elements.
• Beam stress resultants. Only finite elements.

• Beam stress resultants local axes x′ , y ′ and z ′ . Only finite elements.

• Bar stress resultants. Only finite elements.
• Shell stress resultants. Only finite elements.

• Shell stress resultants local axes x′ , y ′ and z ′ . Only finite elements.

56 CHAPTER 3. OUTPUT FILES
Appendix A

How to compile the source code

Go to MultiFEBE’s GitHub webpage, and download the source code. Then, decompress the *.tar.gz
or *.zip file.

A.1 GNU/Linux
The steps to compile the source code are the usual on GNU/Linux projects using CMake. The particular
commands shown below are for Debian distributions (e.g. Ubuntu), but it should be very similar for
others.

1. First, you have to install the pre-requisites if not already installed, so you have to open a terminal
and execute:

• Install GNU Fortran, GNU Make and CMake:

$ sudo apt-get install gfortran make cmake cmake-extras

• Install OpenBLAS:
$ sudo apt-get install libopenblas-base libopenblas-dev

2. Once you have all the pre-requisites, navigate at the source code main folder. Then create a
temporary folder (e.g. build) where making the compilation, and navigate inside it:

$ mkdir -p build
$ cd build

3. Execute cmake taking into account that the CMake configuration file is in the parent folder:

$ cmake -G "Unix Makefiles" ..

or

$ cmake -D CMAKE_BUILD_TYPE=Debug -G "Unix Makefiles" ..

for compilation for debugging. During the process, if any tool or dependency is missing, then CMake
is going to notify you.

4. Once CMake correctly finishes, you can now execute GNU Make:

$ make

5. If all this process ends correctly, you will have the executable multifebe there. However, it is still
not available system-wide in the terminal. You have at this point several options:

(a) The simpler one is to copy the binary to usr/bin:

$ sudo cp multifebe /usr/bin/.

57
58 APPENDIX A. HOW TO COMPILE THE SOURCE CODE

(b) The cleaner one is making a *.deb installer which will manage not only this, but it will
also copy the documentation to the corresponding folders (/usr/share/doc/multifebe), and
create a package registry so that it can also be completely uninstalled or updated in the future.
In order to do so, you have to execute CPack as:
$ cpack

At the end of the process, you will have at your disposal your own *.deb installer. Then, you
can follow the step described above in section 1.3.1.

A.2 Windows
The compilation in Windows requires much more steps than in GNU/Linux, but basically because it is
required to install and to configure MSYS2. MSYS2 is a collection of tools and libraries providing you with
an easy-to-use environment for building, installing and running native Windows software. Furthermore,
it is very similar to GNU/Linux environments, so most of the programming work is shared when building
for GNU/Linux and Windows.
1. Download and install MSYS2 from the webpage.
2. After the installation, the ”MSYS2 MSYS” terminal is automatically opened. Close it.
3. We recommend to follow the post-installation instructions given at the webpage.
4. MSYS2 installs the so called Mintty terminal, but it provides separate launchers for different envi-
ronments, such as the already opened “MSYS2 MSYS”, but also “MinGW x64”, which is the one
used for compiling, and others. See Figure A.1.
5. We will reproduce the post-installation instructions given by MSYS2 at the date of this writing.
First, open the “MSYS2 MSYS” terminal. See Figure A.1.

Figure A.1: Compilation in Windows: Steps 4 and 5

6. Update the packages by typing and executing the command pacman -Syu as shown in Figure A.3.
7. Proceed with the installation by typing Y and executing as shown in Figure A.3.
8. After updating, the terminal ask permission for closing. Type Y and execute.
9. Open again the “MSYS2 MSYS” terminal, and type and execute again the update command pacman
-Syu.
10. Proceed with the installation by typing and executing Y.
11. Install typical tools for compiling by typing and executing:
A.2. WINDOWS 59

Figure A.2: Compilation in Windows: Steps 5 and 6

Figure A.3: Compilation in Windows: Step 8

$ pacman -S mingw-w64-x86_64-toolchain

which will take some time.

12. Close the “MSYS2 MSYS” terminal.

13. Open the “MSYS2 MinGW x64” terminal.

14. Install additional tools and libraries required by typing and executing the following sequence of
commands:

(a) Install GNU C Compiler and GNU Fortran for MinGW 64 bits:
$ pacman -S gcc mingw-w64-x86_64-gcc gcc-fortran mingw-w64-x86_64-gcc-fortran

(b) Install GNU Make, CMake and NSIS:

$ pacman -S gcc make mingw-w64-x86_64-make cmake mingw-w64-x86_64-cmake mingw-w64-x86_64-nsis

(c) Install OpenBLAS library:

$ pacman -S gcc mingw-w64-x86_64-openblas

15. In the “MSYS2 MinGW x64” terminal, navigate to the source code main folder. In the case
illustrated in the following figures D:\multifebe.

16. Create a folder where all compilation build files are generated. In the case illustrated in the following
figures it is named build. This can be done by executing:

$ mkdir -p build

17. Navigate to it by executing:

$ cd build
60 APPENDIX A. HOW TO COMPILE THE SOURCE CODE

Figure A.4: Compilation in Windows: Step 13

18. Execute cmake taking into account that we want to use make as the compilation tool (thus using the
option -G ``Unix Makefiles'') and that the configuration file CMakeLists.txt is in the parent
folder:

$ cmake -G "Unix Makefiles" ..

Figure A.5: Compilation in Windows: Steps 15 to 18

19. Execute make for compiling the source code:

$ make

20. After this is correctly done, multifebe.exe should be ready to be used in the current folder.

21. You can manually install this program in the system by creating a new folder inside C:\Files
(x86) called for example multifebe, and copy multifebe.exe inside it. Then you have to add
this directory to the system or user PATH environment variable. To do so, follow the instructions
in this link, or do a search on the web about “How to set the path and environment variables
in Windows”. After this is done, you can run the program on a system terminal, like cmd or
PowerShell.

22. It is also possible to build an installer which do all this for you, and furthermore integrates the
program into your system with a uninstaller, etc. You have to be in a “MSYS2 MinGW x64”
terminal, and the current working directory in the folder where we are compilating.
A.2. WINDOWS 61

23. Then you can build the installer by executing in the terminal:

$ cpack

24. This will create and installer called multifebe-x.x.x-w64.exe, which is the installer we release
for Windows. Then you can execute it and follow the instructions above in section ??.
62 APPENDIX A. HOW TO COMPILE THE SOURCE CODE
Appendix B

Gmsh - How to export a mesh in

MSH file format version 2.2

In Gmsh newer than version 3.0.6, the default Gmsh mesh file format is not MSH file format version 2.2.
Therefore, you will have to save the mesh from an alternative workflow.
When using the Gmsh GUI, you have to follow the steps shown in Fig. B.1. When using the
command-line interface, you have to include the option -format msh2, for example:

$ gmsh -2 -format msh2 pilecap.geo

63
64 APPENDIX B. GMSH - HOW TO EXPORT A MESH IN MSH FILE FORMAT VERSION 2.2

(a) Step 1: File > Export (b) Step 2: Format: Mesh - Gmsh MSH (*.msh)

(c) Step 3: Write filename (d) Step 4: Select MSH format

Figure B.1: Gmsh GUI: export mesh file in different format

 Appendix C

GiD *.bas template file

We have implemented a *.bas template file for writing mesh files in the MultiFEBE native format from
GiD pre- and post-processor. It is a simple plain text file containing a script in GiD language. Two
versions of the template file are included in the package, one with carriage return of the Windows type
(multifebe_win.bas), and another with the Unix type (multifebe_unix.bas). Note that the shown
blank lines are necessary. Conceptually, each GiD “layer” is a “part” in MultiFEBE jargon.
multifebe_win.bas & multifebe_unix.bas
1

2 *IntFormat "%12i"
3 *RealFormat "%25.16e"
4 [parts]
5 *Set var nLayer=0
6 *loop layers
7 *Set var nLayer=LayerNum
8 *end layers
9 *nLayer
10 *loop layers
11 *LayerNum *LayerName
12 *end layers
13 [nodes]
14 *nPoin
15 *set elems(all)
16 *loop nodes
17 *NodesNum *NodesCoord
18 *end nodes
19 [elements]
20 *set Elems(All)
21 *nElem
22 *set Elems(Linear)
23 *loop elems
24 *if(ElemsNnode==2)
25 *ElemsNum line2 1 *ElemsLayerNum *ElemsConec
26 *endif
27 *if(ElemsNnode==3)
28 *ElemsNum line3 1 *ElemsLayerNum *ElemsConec
29 *endif
30 *end elems
31 *set Elems(Triangle)
32 *loop elems
33 *if(ElemsNnode==3)
34 *ElemsNum tri3 1 *ElemsLayerNum *ElemsConec
35 *endif
36 *if(ElemsNnode==6)
37 *ElemsNum tri6 1 *ElemsLayerNum *ElemsConec
38 *endif
39 *end elems
40 *set Elems(Quadrilateral)
41 *loop elems
42 *if(ElemsNnode==4)
43 *ElemsNum quad4 1 *ElemsLayerNum *ElemsConec
44 *endif

65
 66 APPENDIX C. GID *.BAS TEMPLATE FILE

45 *if(ElemsNnode==8)
46 *ElemsNum quad8 1 *ElemsLayerNum *ElemsConec
47 *endif
48 *if(ElemsNnode==9)
49 *ElemsNum quad9 1 *ElemsLayerNum *ElemsConec
50 *endif
51 *end elems
52
Bibliography

[1] O. Maeso, J. J. Aznárez, and J. Domínguez. Threeâ��dimensional models of reservoir sediment and
effects on the seismic response of arch dams. Earthquake Engineering and Structural Dynamics,
33(10):1103–1123, 2004.
[2] J. J. Aznárez, O. Maeso, and J. Domínguez. BE analysis of bottom sediments in dynamic fluid-
structure interaction problems. Engineering Analysis with Boundary Elements, 30:124–136, 2006.

[3] L.A. Padrón, J.J. Aznárez, and O. Maeso. BEM–��FEM coupling model for the dynamic analysis of
piles and pile groups. Engineering Analysis with Boundary Elements, 31(6):473 – 484, 2007.
[4] Guillermo M. Álamo, Alejandro E. Martínez-Castro, Luis A. Padrón, Juan J. Aznárez, Rafael Gal-
lego, and Orlando Maeso. Efficient numerical model for the computation of impedance functions of
inclined pile groups in layered soils. Engineering Structures, 126:379–390, 2016.
[5] J. D. R. Bordón, J. J. Aznárez, and O. Maeso. Dynamic model of open shell structures buried in
poroelastic soils. Computational Mechanics, 60(2):269–288, 2017.
[6] J. Domínguez. Boundary Elements in Dynamics. WIT Press, 1993.

[7] E. Oñate. Structural Analysis with the Finite Element Method. Linear Statics., volume Volume
2: Beams, Plates and Shells of Lecture Notes on Numerical Methods in Engineering and Sciences.
Springer Dordrecht, 1 edition, 2013.
[8] Z. Friedman and J.B. Kosmatka. An improved two-node timoshenko beam finite element. Computers
& Structures, 47(3):473–481, 1993.

[9] K.J. Bathe. Finite Element Procedures. Prentice-Hall, 2007.

[10] Phill-Seung Lee and Klaus-Jürgen Bathe. Development of mitc isotropic triangular shell finite ele-
ments. Computers & Structures, 82(11):945–962, 2004.
[11] Eduardo N. Dvorkin and Klaus-Jürgen Bathe. A continuum mechanics based four‐node shell element
for general non‐linear analysis. Engineering Computations, 1(1):77–88, 1984.

[12] Klaus-Jürgen Bathe and Eduardo N. Dvorkin. A formulation of general shell elements—the use
of mixed interpolation of tensorial components. International Journal for Numerical Methods in
Engineering, 22(3):697–722, 1986.
[13] Miguel Luiz Bucalem and Klaus-Jüen Bathe. Higher-order mitc general shell elements. International
Journal for Numerical Methods in Engineering, 36(21):3729–3754, 1993.
[14] M.A. Biot. Theory of propagation of elastic waves in a fluid�saturated porous solid I.
low�frequency range. The Journal of the Acoustical Society of America, 28(168), 1956.

67