SAFIR 2016.c.0 Thermal Simulation Software Users manual

SAFIR
Users manual
Version 2016.c.0
This document explains the introduction of data for the software SAFIR developed at the
University of Liege for the simulation of building structures subjected to fire.
Only the thermal part of SAFIR is considered in this document (still under construction).
This document is organized in the manner specified in the standard ISO/FDIS 16730:2008(E)
"Fire safety engineering - Assessment, verification and validation of calculation methods"
To date (23 August 2016), this document has not been through third party auditing
1
23/08/2016
A.
Description of SAFIR
A.1 Description of the model
The underlying model is that in which heat is distributed in the structure essentially
by conduction because most of construction elements are made of solid. Radiation and
convection are the heat transfer modes in internal cavities such as those present, for
example, in hollow core slabs. Convection and radiation are the heat transfer mode at the
boundaries between the object analyzed and the environment, i.e. the fire.
A.2 Description of the basic processing tasks performed calculation methods and procedures performed
SAFIR first reads the whole input file and copies it in the output file.
Then it reads all lines of the input file again, one by one, tries to interpret the data
that are found in each line and prints the interpreted results in the output file before going
to the next line.
Then SAFIR will calculate the temperature in the structure at the successive time
steps chosen by the user. For each time step:




SAFIR first extrapolates the temperatures from two the previous time steps, tn-1 and tn, to
a first approximation at a given time tn + Dt1 comprised between tn and tn+1. The
temperatures at this stage are not in equilibrium.
Then SAFIR refines this first approximation of the temperatures at tn + Dt until thermal
equilibrium is restored. Several iterations are required to restore the equilibrium up to
an accepted degree of precision2.
SAFIR finally extrapolates the temperatures at tn+1 from the values at tn and tn + Dt.
If the time step is one of the time steps defined for printing the results, see TIMEPRINT
command, the temperatures at the nodes are printed in the output file. Temperatures
are also eventually written in the .TEM or .TSH file to be used in a subsequent analysis.
The value of Dt is governed in the data by the parameter TETA = Dt/(tn+1-tn)
The degree of precision required before thermal equilibrium is considered as acceptable is governed in the
data by the parameter PRECISION.
1
2
2
23/08/2016
Reading the input file
Copying the input file in the output file
Reading 1 line of the input file
Writing the related information in the output file
NO
End of
Input file?
YES
1 iteration
Extrapolate the temperatures at tn + Dt
Temperatures in
equilibrium?
YES
NO
Calculate out of equilibrium fluxes
Calculate incremental temperatures
Update the temperatures at tn + Dt
Extrapolate the temperatures at tn+1
YES
Time step for writing
the results?
Writing the temperatures
in the output file
YES
Final time
reached?
Writing the temperatures
in the output file
3
23/08/2016
Tn+1
Iterations
Temperature
Tn
Tn-1
tn-1
tn
tn+Dt
tn+1
Time
Dt
4
23/08/2016
A. 3 Description of the types of skills required to execute typical runs
The user should be able to operate a P.C. computer, either a laptop or a desktop.
The user should be familiar with the Windows operating system.
The user should be familiar with the file naming convention based on a file name and
a file type. He should be able to use the Explorer of Windows to move from one folder to
another, to copy, delete, or rename files.
The user should be able to open files with a text editor, in order to prepare or modify
input files and to examine the results produced with SAFIR.
In order to execute typical runs, the user should be familiar with the basic
mechanisms of heat transfer such as conduction, convection and radiation. He should also
be familiar with the finite element technique applied to solving heat transfer problems.
In order to judge from the relevant character of the results the user should be
familiar with the behavior of structures subjected to fire. For example, a user that has all
skills and experience mentioned above would be able to perform a numerical simulation of a
section protected by a gypsum plaster board and subjected to a fire during 4 hours, only
missing the fact that no gypsum plaster board can stay in place for such a long duration; it
would disintegrate long before.
5
23/08/2016
B.
Installations and operating instructions
B.1 Minimum hardware configuration required
SAFIR can run on a P.C with a single processor. 1.6 GHz.
1 Go of RAM
120 Go Hard Disk
This has been tested on an ACER Aspire One ZG5 notebook.
SAFIR could not run on a Pentium II, 350 MHz, 384 Mo RAM, 37,2 Go ROM.
B.2 Computers on which SAFIR has been executed successfully




Laptop HP Pavilion Entertainment PC
Intel Corel2 CPU, 1.66 GHz, 2 Go RAM, 232 Go Hard Disk
Dell Precision T5400
Intel Xeon CPU E5420, 2.50 GHz,8 Go RAM, 300 Go Hard Disk
Asus under Windows XP Professional Version 2002, Service Pack 3
Intel Pentium 4, 2.60 GHz, 1 Go RAM, 29 Go Hard Disk
ACER Aspire One ZG5
Intel Atom 1.6 GHz, 1 Go RAM, 120 Go Hard Disk
B.3 Programming language, software operating system and version
in use
SAFIR has been in written in FORTRAN:
Intel Visual Fortran Compiler Integration for Microsoft Visual Studio 2005, 11.1.3466.2005
under Microsoft Visual Studio 2005, version 8.0.50727.762
Some routines, written at the early days of development of the code, are in FORTRAN 77
while the most recent subroutines are in FORTRAN 90.
6
23/08/2016
B.4 Instructions for installing the program
SAFIR is delivered in the form of an executable file, for example "SAFIR2009.exe", and a
security file name “identity.key”. These files have to be copied on the device on which SAFIR
will be found by the C.P.U of the computer during execution. This could be, for example, a
USB key, an external hard disk or, more likely, one of the hard disk of the P.C. Both files must
be located on the same device.
The file “identity.key” has to be located in the root directory of the device.
The file “SAFIRxxxx.exe” can be located in the folder of choice of the user. Two different
options are possible:
1. Users normally organize their input files in different folders; one folder may for
example be used for each different project. The executable can be copied in each of
these folders, see Figure 1.
Figure 1: folder with the executable
To run SAFIR from Internet Explorer, double click on the executable file. SAFIR will open a
DOS window in which the user is requested to type the file name of the input file
("filename" from the complete name "filename.IN"), see Figure 2.
7
23/08/2016
Figure 2: DOS window with the request for the file name
The user has to type the filename, see Figure 3, and press the ENTER key on the
keyboard.
Figure 3: DOS window with the file name
SAFIR starts to run and some messages are written in the DOS window, which allows
seeing the progress of the simulation, see Figure 4. When the simulation stops, the
window is closed immediately3.
3
If an error message is written as the last line in the window, it is normally not possible to read it before the
window closes. The same message is normally written at the end of the output file.
8
23/08/2016
Figure 4: DOS window with the messages
2. SAFIR is copied in one single folder of the hard disk, for example in the folder
"c:\program files\SAFIR\2009". SAFIR will then have to be run with the tool called
"SafirShell". This tool is installed by execution of "SafirShell.msi" from the CD disk
provided with SAFIR. When this tool is started, a window is opened, see Figure 5.
Figure 5: window of the SafirShell
The folder where the executable file of SAFIR is located is indicated in the "Safir Path"
window. This folder can be selected by the user with the […] button on the top right.
The input files that will be treated by SAFIR can be chosen by the user using the
[Add..] button. SAFIR will start running as soon as the [Run] button is clicked on.
The advantages of using the SafirShell tool are:



The executable of SAFIR is present only in one folder on the hard disk.
Several input files can be chosen and treated in a batch mode, one after the other. If one
runs fails for any reason, the next input files will be treated.
The messages that were written in the DOS window will now be written in the bottom
window of the SafirShell and will remain there to be read even after the runs are
finished.
9
23/08/2016

The priority allocated by the Operating System to the SAFIR process can be selected in
the Priority scroll down menu. For processors with a single core, it is recommended to
choose "below normal" because this will hardly affect the velocity of SAFIR but will allow
time enough to be allocated for performing usual office tasks.
10
23/08/2016
B.5 Typical personal time and setup time to perform a typical run
The typical personal time and setup time to perform a typical run (not counting the
time required to gather all information about the structure to be analyzed is in the order of
half a day for a 2D analysis and one day for a 3D analysis.
B.6 Estimation of the computer execution time for typical
applications
The computer execution times given here have been taken from runs on a Laptop HP
Pavilion Entertainment PC, Intel Corel2 CPU, 1.66 GHz, 2 Go RAM, 232 Go Hard Disk. The
time step used for the simulations was 15 seconds. Much faster runs could be performed
with little loss of precision if longer time steps would be used.
The duration of calculation mentioned are taken from the clock, i.e. difference
between the time when the calculation is finished and the time when it started.
For 2D analyses.
One quarter of a reinforced concrete section, Figure 6, 144 nodes, 2 hours of fire
=> 3 seconds of simulation.
Diamond 2009.a.5 for SAFIR
FILE: sect9p
1
NODES: 144
ELEMENTS: 121
1
SOLIDS PLOT
FRONTIERS PLOT
CONTOUR PLOT
1
1
1
1
1
1
1
1
Y
TEMPERATURE PLOT
TIME: 7200 sec
1046.50
1000.00
900.00
800.00
700.00
600.00
500.00
400.00
300.00
200.00
100.00
<Tmin
1
X
Z
1
1
1
1
1
1
1
1
1
1
1
Figure 6: 1/4 of a reinforced concrete section
11
23/08/2016
Composite steel-concrete beam, Figure 7, 932 nodes, 2 hours of fire
=> 21 seconds of simulation
Diamond 2009.a.5 for SAFIR
FILE: Compos ite_beam
NODES: 932
ELEMENTS: 812
SOLIDS PLOT
CONTOUR PLOT
TEMPERATURE PLOT
TIME: 2400 sec
>Tmax
1000.00
900.00
800.00
700.00
600.00
500.00
400.00
300.00
200.00
100.00
24.50
Y
X
Z
Figure 7: composite steel-concrete beam
12
23/08/2016
Circular concrete section4, Figure 6, 1051 nodes, 2 hours of fire
=> 33 seconds of simulation.
Diamond 2009.a.5 for SAFIR
FILE: sCol120_ISO
NODES: 1051
ELEMENTS: 2000
SOLIDS PLOT
SILCONC_EN
Y
X
Z
Figure 8: circular concrete section
Diamond 200
FILE: sectPont_
NODES: 24108
ELEMENTS: 23
Concrete bridge deck5, Figure 9, with 24108 nodes, 120 minutes of fire
SOLIDS PLOT
=> 9 minutes 7 seconds of simulation.
CA
Figure 9: bridge deck
Y
4
5
Model created with GID by Eric Tonicello, MP Ingénieurs Conseils
X created
Z
Model
with GID by Eric Tonicello, MP Ingénieurs Conseils
13
23/08/2016
For 3D analyses.
Composite joint with a crude discretisation, Figure 10, 2915 nodes, 120 minutes of fire
=> 1 minute 5 seconds of simulation
Diamond 2009.a.5 for SAFIR
FILE: comeliau
NODES: 2915
ELEMENTS: 1602
SOLIDS PLOT
CONTOUR PLOT
TEMPERATURE PLOT
TIME: 1860 sec
>Tmax
1000.00
900.00
800.00
700.00
600.00
500.00
400.00
300.00
200.00
100.00
20.00
Y
Z
X
Figure 10: composite joint (crude mesh)
14
23/08/2016
Concrete beam with a circular cavity in the web, 6183 nodes, 120 minutes of fire
=> 7 minutes 31 seconds of simulation
Diamond 2009.a.5 for SAFIR
FILE: HOLE9P
NODES: 6183
ELEMENTS: 5060
SOLIDS PLOT
CONTOUR PLOT
TEMPERATURE PLOT
Y
Z
X
TIME: 7200 sec
1046.70
1000.00
900.00
800.00
700.00
600.00
500.00
400.00
300.00
200.00
100.00
48.50
Figure 11: concrete beam with a circular cavity in the web
15
23/08/2016
Composite joint6, Figure 10, 31502 nodes, 120 minutes of fire
=> 48 minutes 47 seconds of simulation
Figure 12: composite steel-concrete joint
6
Model by Elisabetta Alderighi, Univ. of Pisa.
16
23/08/2016
Figure 13 shows the CPU time on the computer utilized, as a function of the number
of nodes, for 120 minutes of simulation when a time step of 15 seconds is used. Shorter
times could be obtained by reducing the fire duration or using longer time steps.
3 500
2D
3D
Linear (2D)
Linear (3D)
3 000
y = 0.0992x - 195.28
Time [Seconds]
2 500
2 000
1 500
1 000
y = 0.0226x + 2.9486
500
0
0
5 000
10 000
15 000
20 000
25 000
Number of nodes
Figure 13: CPU time as a function of nodes
30 000
35 000
17
23/08/2016
C.
Program considerations
C.1 Major options available
The first option that the user has to take depends on the type of structural
calculation that will be performed subsequently after the thermal calculation.
1) A 3D thermal calculation is typically performed to establish the temperature distribution
in a detail of a construction, for example a connection, see Figure 10, Figure 11 or Figure
12. Such a 3D thermal calculation can be followed by a structural calculation made by
SAFIR using 3D solid finite elements. In this case, the discretization made for the thermal
calculation is used as is in the subsequent structural calculation. Alternatively, the
temperature can be used in a simple structural calculation model such as, for example,
the component method in steel connections or the strut and tie model in concrete
elements, in which the influence of the calculated temperatures on the materials
strength can be taken into account.
2) A 2D thermal calculation can be performed on the cross section of a beam-column finite
element, see Figure 6 to Figure 9. This means that the discretization made for the thermal
calculation will also be used to calculate the mechanical properties of the section in the
subsequent structural calculation.
3) A 2D thermal calculation can be performed for evaluating the temperature distribution
across the thickness of a shell finite element. Figure 14, for example, shows a 14 cm thick
concrete slab protected on the underneath by a 2 cm thick layer of protection material.
The discretization has to comply with the following rules:
o An arbitrary width of the shell is modeled. In Figure 14, for example, a width of 10
cm has been modeled, but it could be as well 1 cm or 1 meter. Also the horizontal
position of the section in the local system of coordinates, i.e. in the direction of z,
is not relevant.
o The section is modeled by rectangular finite elements, the width of which is the
same as the width of the discretized section. The number of elements on the
thickness of the section, N, is chosen by the user (N = 18 on Figure 14)
o The nodes are numbered from 1 in the bottom left corner to N+1 on the top left
corner, and from N+2 to 2 N + 2 on the right side of the section.
o In the local system of coordinates, the structural material of the slab is located
from -t/2 to +t/2, in the direction of y, with t the thickness of the structural part
of the slab (from node 5 to node 19 on Figure 14).
o There is only one structural material represented in the thermal analysis of the
slab (material SILCONC_EN in Figure 14). If re-bars are present in the slab, they
are not considered in the thermal analysis but will be added in the .TSH file.
18
23/08/2016
Layers of other materials, typically insulating products, can be present outside the
structural part of the shell (material INSULATION in Figure 14).
38
Diamond 2009.a.5 for SAFIR
Diamond 2009.a.5 for SAFIR
18
37
FILE: Shell_C
FILE: Shell_C
17
36
NODES: 38
NODES: 38
ELEMENTS: 18
ELEMENTS: 18
16
35
15
34
NODES PLOT
SOLIDS PLOT
SOLIDS PLOT
CONTOUR PLOT
14
33
Y13
32
X12 Z
31
11
30
10
29
9
28
8
27
7
26
6
25
5
4
3
2
1
24
23
22
21
20
2
19
FRONTIERS PLOT
TEMPERATURE PLOT
Y
CONTOUR PLOT
X
TIME: 7200 sec
1038.90
200.00
180.00
160.00
140.00
120.00
100.00
80.00
60.00
40.00
20.00
<Tmin
Z
INSULATION
SILCONC_EN
1
2
FISO
F20
1
Figure 14: concrete shell protected on the bottom face
Figure 15 shows a 12 mm steel plate exposed to the fire on both sides and modeled by
a single finite element. A width of 100 mm has been considered here, but this value does not
influence the temperature distribution across the thickness and is thus not relevant.
Diamond 2009.a.5 for SAFIR
FILE: Shell_S
NODES: 4
ELEMENTS: 1
2
Y
1
X
4
NODES PLOT
Z
SOLIDS PLOT
1
3
1
FRONTIERS PLOT
STEELEC3EN
1
FISO
Figure 15: steel plate
The second option to be decided by the user is whether the fire is considered as a
localized fire or not.
1) Very often, the fire is not localized. Large parts of the building are considered to be
characterized by a uniform situation. For example, one room of the building is
characterized by a post flash-over fire curve, while the other rooms are not subjected to
the fire. Or a fire compartment is divided in an upper zone that is characterized by an
elevated but uniform temperature and a lower zone in which the temperature is lower,
19
23/08/2016
but also uniform. In these cases, several finite elements are subjected to the same attack
from the fire. It is therefore sufficient to determine the temperature in one of all the
elements of the same type which are subjected to the same attack. For example, if a
steel beam located in a post flash-over compartment is represented by shell finite
elements, it is sufficient to determine the temperature distribution once for the flanges
and once for the web7. In order to perform these two thermal calculations, it is sufficient
to know the level of the fire attack (the fire curve, for example). It is not necessary to
know the position in the compartment of the element for which the temperature is
being determined (because all elements of the same type in the compartment have the
same temperature distribution).
2) In some case, the fire is a localized fire, which means that its dimensions are small
compared to the dimensions of the compartment. This could be the case, for example, of
one car burning in a car park that is so large that it may contain tens of cars. In this case,
all finite elements in the compartment are subjected to a different attack from the fire,
essentially because the radiative heat transfer from the localized fire to the elements
depends on the distance between both and on the angle between the surface of the
element and the direction of the fire. It will then be necessary to perform a thermal
calculation in a lot of different positions of the structure in order to represent the fact
that the temperature distribution varies from places to places. Typically, a uniaxial
temperature distribution will be determined across the thickness in 4 different positions
of each shell finite element and a 2D temperature distribution will be determined in each
of the longitudinal point of integration of each beam-column finite element. In order to
perform this series of independent thermal calculation, SAFIR must know where each
section is located in the compartment with respect to the fire. This means that the
structural model must have been created and the position of characteristics of the fire
must have been positioned in the compartment before the thermal calculations can be
performed.
C.2 Limits of applicability
This section presents the range of scenarios over which the underlying theory is
known or believed to be valid and the range of input data over which the calculation method
was tested.
 The scenario that have been considered here are fire scenarios in building structures,
with a duration that extends typically between some minutes and a few hours.
Extremely fast heating rates such as those that may occur, for example, in the launch
7
Assuming that the flanges and the web have different thicknessess.
20
23/08/2016
pad of Ariane rockets, are not within the scope of the calculation method used in SAFIR.
On the other end of the range, applications that involve very long exposure to elevated
temperatures such as, for example, those that may be encountered in some industrial
processes, are also not within the scope of SAFIR.
 The fire scenario may include, after the heating phase, a so called cooling phase leading
to decrease of the temperature in the materials. Such a cooling phase can be considered
by SAFIR as far as material properties during cooling are known.
 The range of temperature is from 0°C to 1200°C. The effect of freezing of free water that
may be present in the materials is indeed not taken into account. On the other end of
the temperature range, the material properties for fire applications are not known
beyond 1200°C.
C.3 Restrictions and limitations of SAFIR
In this section are listed the restrictions and limitations of SAFIR, including appropriate data
range and SAFIR's behavior when the ranges are exceeded.
Appropriate data range
SAFIR can treat 2D or 3D models
Number of nodes must be at least 3
teta must be greater than 0 and not greater
than 1
The initial temperature should be greater
than 0 and not greater than 1200
The number of cores of the computer should
be greater than 0
The number of cores mentioned in the input
file should not be larger than the number of
cores of the computer
The number of materials must be greater
than 0
The number of solid elements must be
greater than 0.
The number of integration points in each
direction must be 1, 2 or 3.
The number of frontiers in a cavity must be
at least 2.
Behavior when the ranges are exceeded
If ndim is not 2 or 3, a message is written in
the output file and SAFIR stops execution.
If nnode is less than 2, a message is written
in the output file and SAFIR stops execution.
If teta is out of scope, a message is written
in the output file and SAFIR stops execution.
If tinitial is out of scope, a warning message
is written in the output file, but the run
continues
If ncores is less than 1, it is changed to 1 by
SAFIR.
If ncores is larger than the number of cores
of the computer, the run continues
If nmat is less than 1, , a message is written
in the output file and SAFIR stops execution.
If nsolid is less than 1, a message is written
in the output file and SAFIR stops execution.
If ng is not 1, 2 or 3, a message is written in
the output file and SAFIR stops execution.
If nfrontiervoid is less than 2, a message is
written in the output file and SAFIR stops
execution.
21
23/08/2016
22
23/08/2016
D.
Input data description
D.1 Input variables
D.1.1 General
The data to be used by SAFIR in a thermal calculation must be placed in a txt file8.
The name of the file, filename, is chosen by the user and must conform to the rules of
Windows. For example, some characters cannot be used in the name of the file. filename is
limited to 70 characters.
The file type of the input file must be IN. The complete name of the input file in the
folder where it is present will thus be filename.IN. On the hard disk of a particular P.C., the
full name could be, for example if example1 has been chosen as the file name by the user,
c:/my files/projects/SAFIR/UsersManual/example1.IN. This input file will be referred to in
this manual as the .IN file.
In general, each line of the .IN file starts with a command; it is a chain of characters
typed in capital letters, with a maximum length of 10 characters. This command indicates to
SAFIR the meaning of the numbers that are present on the same line or in the subsequent
lines after the command. The commands are also useful for the user who reads the file; they
make it quite easy to locate particular information in the input file compared to a situation
where only numbers would be present in the file.
The format for writing the input of each card is free. This means that the real values
can be written in any form. “235.E6”, “235000000.” and “235.00E+06”, for example, are
equally valid and express the same value. Subsequent values must be separated by any
number of spaces, or by a comma. The values that are described as belonging to a card in
the manual can be split on several lines. Additional information (such as characters used to
comment the file) can be written after the last information of a card, provided that it is
written on the same line as the last information. The following examples are all valid
examples of the same card.
NNODE
1
2.
3.
NNODE, 1, 2., 3.
8
Also called an ASCII file.
23
23/08/2016
NNODE
1
2.
3.
NNODE
1
2.
3.
These are the coordinates of node 1
The order in which the data are introduced in the input file is mandatory. All data
must be introduced in the order that is mentioned hereafter.
Blank card can be introduced in the file between the cards mentioned hereafter.
Distances are in meter. Temperatures are in degree Celsius.
24
23/08/2016
D.1.2 Detailed description of the input file
SERIES 1: COMMENTS
a) Multiple cards (can be 0)
As many lines as required may be introduced by the user at the beginning of the file in
order to comment the content of the file. These comments will not influence in any
manner the results of the calculation. They will help understand what the file is about,
which may be particularly useful if the file is found and opened several years after it has
been created. Experience shows that the users tend to neglect the comments, relying on
the characters that are used in filename in order to differentiate several calculations
made for a project. Using extensively the possibility of introducing detailed comments in
the file is nevertheless highly preferable because it allows reporting much more
information.
b) 1 card
1 blank line to indicate that the comments are finished.
[-]
SERIES 2: NUMBER OF NODES
a) 1 card
NNODE, nnode
 NNODE
[A5]
Command
 nnode
[Integer]
Number of nodes of the model. Minimum value is 3. There is no maximum
value.
This file has been created as an example for the users manual of SAFIR
Date: 16/12/2010
Author: Jean-Marc Franssen
NNODE
16
25
23/08/2016
SERIES 3: NUMBER OF DIMENSIONS
a) 1 card
NDIM, ndim
 NDIM
Command
 ndim
Number of dimensions of the model.
ndim = 2 for 2D models
ndim = 3 for 3D models
[A4]
[Integer]
SERIES 4: DEGREES OF FREEDOM
a) 1 card
NDOFMAX, 1
 NDOFMAX
[A10]
Command
 1
[Integer]
This card indicates that the maximum number of degrees of freedom for the
nodes is 1 in a thermal analysis.
SERIES 5 (optional): CORES OF THE CPU
1 card (optional)
NCORES, ncores
 NCORES
[A6]
command
 ncores
[integer]
Number of cores of the CPU of the computer used by matrix solver. The
default value is 1, in which case this card may be omitted. This card can be
used to force SAFIR to use more than 1 core, if present on the computer.
Recent experience has shown that using more than 1 core hardly reduces the
time of the runs with the present version of Pardiso and this card can thus be
omitted as a common practice. The possibility of using the card has
nevertheless been given in order to allow users to perform their own test on
their particular system, and in order to offer the possibility of working with
more than 1 core in the future if new releases of Pardiso show to exploit
several cores more efficiently.
26
23/08/2016
SERIES 6: THERMAL CALCULATION
a) 1 card
 COMMAND
[A8]
COMMAND = TEMPERAT is used for a normal thermal calculation from time t =
0 to a final time to be defined by the user.
COMMAND = RESTARTT is used if a previous thermal analysis has been made
and a new thermal analysis has to be restarted from a restart time ti (to be
defined by the user) up to a final time. The temperature field at time ti is
taken from the previous analysis. The new analysis can be performed either
on the same structure as the previous analysis or on a new structure that is
only one part of the previous structure. The first possibility is used if the
structural analysis shows that the thermal analysis had been stopped too
early. The second possibility is used, for example, to take into account the
fact that some layers of concrete or of protective material have fallen off the
structure (what part has fallen off and at what time has to be decided by the
user).
The next 3 cards, b), c) and d) are present only if RESTARTT has been entered in card a). If
TEMPERAT has been entered in card a), then go directly to card e).
b) 1 card
FIRSTFILE, filename
 FIRSTFILE
[A9]
Command
 filename
[A20]
Name (comprising the file name and the file type) of the output file were the
results of the previous thermal analysis had been written. This file must be
present in the same folder as the input file of the thermal analysis
c) 1 card
FIRSTTIME, time
 FIRSTTIME
[A9]
Command
 time
[Real]
Time (in seconds) when the second analysis will start. The temperatures will
be read from the first analysis and used as initial values for the second
analysis.
d) 1 card
 COMMAND
[A10]
COMMAND = MATCHNODES is used if the second structure is exactly the
same as the first one or if its nodes were the first nodes of the first structure.
27
23/08/2016
In that case, the initial temperature at each node in the second structure is
taken as the temperature at the same node in the first structure.
COMMAND = MATCHCOORD allows considering that the initial temperature
at each node in the second structure is taken as the temperature of the
node of the first structure with same coordinates. This possibility is used if
some among the first nodes of the first structure have been deleted when
creating the second structure.
e) 1 card
TETA, teta
 TETA
[A4]
Command
 teta
[Real]
Parameter for the time integration. 0 < teta ≤ 1.
Values of teta close to 0 will force SAFIR to perform the time integration in a
nearly explicit manner, whereas teta = 1 will lead to a fully implicit
integration. teta = 0.9 has been used consistently by the developers without
any problem.
f) 1 card
TINITIAL, tinitial
 TINITIAL
[A8]
Command
 tinitial
[Real]
Temperature (in degree Celsius) in the model at time t = 0, usually taken as
20°C.
e) 1 card (optional)
 COMMAND
[A8 or A10]
COMMAND = MAKE.TEM stores the average temperature of the elements for a
subsequent structural analysis. The section analysed here is the cross section
of a beam.
COMMAND = MAKE.TSH stores the temperature of the first NNODE/2 nodes
for a subsequent structural analysis. The section analysed describes the
temperature evolution across the thickness of a shell element, see Figure 14
and Figure 15.
COMMAND = MAKE.TEMHA is similar to MAKE.TEM, but the fire is a localised
fire as described in Annex C of EN 1991-1-2 (equation C.4, flame impacting
the ceiling – Hasemi fire).
COMMAND = MAKE.TEMLF is similar to MAKE.TEM, but the fire is a localised
fire as described in the RFCS research report LOCAFI.
28
23/08/2016
COMMAND = MAKE.TSHHA is similar to MAKE.TSH, but the fire is a localised fire
as described in Annex C of EN 1991-1-2 (equation C.4, flame impacting the
ceiling – Hasemi fire).
If MAKE.TEMHA, or MAKE.TSHHA or MAKE.TEMLF has been entered in card e), then:
1) The next 2 cards, f) and g) must be present.
2) A file called “hasemi.txt” (for MAKE.TEMHA or MAKE.TSHHA) or a file called
“locafi.txt” (for MAKE.TEMLF) must be present in the same folder as the input file.
This file describes the position of the fire in the structure and its thermal
characteristics, such as the rate of heat release. The structure of this file is described
in D.1.3.
Note: the name of the file is hasemi.txt, even if the function HASEMI_FR is used in
series 14.
f) 1 card (optional)
 filename
[A20]
This is the complete name of the input file of the structural analysis
(maximum 20 characters).
Note: the input file describing the structural analysis must be present in the
same folder as the input file for the thermal analysis when the thermal
analysis is run. If the structural analysis is 2D, gravity must be pointing
downward the Y axis (i.e. Y is vertical upward). If the structural analysis is 3D,
gravity must be pointing downward the Z axis (i.e. Z is vertical upward).
g) 1 card (optional)
COMMAND, ielemtype
 COMMAND
[A9 or A10]
COMMAND = BEAM_TYPE if the temperatures are determined in a beam
section.
COMMAND = SHELL_TYPE if the temperatures are determined in a shell
section.
 ielemtype
[integer]
is the number in the structural input file of the section type (beam or shell)
treated in this thermal analysis.
29
23/08/2016
SERIES 7:
This series is obsolete. No card is needed here
SERIES 8: MATERIALS
a) 1 card (optional)
NMAT, nmat
 NMAT
[A4]
Command
 nmat
[integer]
Number of different materials. If two materials have the same name but
different thermal properties (such as, e.g., the emissivity) or different
mechanical properties (such as, e.g., the yield strength), this makes two
different materials.
30
23/08/2016
SERIES 9: ELEMENTS (general)
a) 1 card
 ELEMENTS
Command.
b) 1 card
SOLID, nsolid
 SOLID
Command
 nsolid
Number of SOLID elements in the model.
c) 1 card
NG, ng
 NG
[A8]
[A5]
[integer]
[A2]
Command
 ng
[integer]
Number of integration points in each direction in the element. Not less than 1,
not greater than 3. 2 has been used consistently by the developers without
any problem.
d) 1 card
NVOID, nvoid
 NVOID
[A5]
Command
 nvoid
[integer]
Number of internal voids, also called internal cavities, i.e. cavities in the
section where there is no material and heat transfer is by convection and,
essentially, by radiation. 0 must be typed if there is no cavity in the model. In
the present version of SAFIR (2011), internal cavities can be present only in 2D
models.
e) 1 card (optional)
This card is present only if nvoid ≠ 0
FRTIERVOID, nfrontiervoid
 FRTIERVOID
[A10]
Command
 nfrontiervoid
[integer]
Number of maximum (for all the voids of the section) number of surfaces (i.e.
sides of elements) enclosing the internal voids. For example, if a section has 2
cavities and one of them is enclosed by 50 surfaces, whereas the second one
is enclosed by 30 surfaces, then nfrontiervoid is equal to 50.
31
23/08/2016
f) 1 card
 END_ELEM
Command.
[A8]
The input file, starting after the comments, may now look like this.
NNODE
NDIM
16
2
NDOFMAX
EVERY_NODE
END_NDOF
1
1
TEMPERAT
TETA
TINITIAL
0.9
20.0
NMAT
2
ELEMENTS
SOLID
NG
NVOID
END_ELEM
9
2
0
32
23/08/2016
SERIES 10: NODES
a) 1 card
 COMMAND
[A5 or A9]
COMMAND = NODES. The position of the nodes will be given in a Cartesian
system of coordinates
COMMAND = NODES_CYL. The position of the nodes will be given in a
cylindrical system of coordinates. Cylindrical coordinates are transformed for
the internal solution process by the following equations:
if ndim = 2
 r ,   Y , Z    r cos  , r sin  
 r , , Z    X , Y , Z    r cos  , r sin  , Z  if ndim = 3
Note that  is in degrees. The transformation is made after all nodes have
been read and the automatic generation and repeat have been made, see
card c) and d).
Several cards are then written, following the format of b), c) or d), until all nnode nodes have
been given their position.
b) 1 card for 1 node
NODE, nno, rcoordg(1,nno), rcoordg(2,nno), rcoordg(3,nno)
 NODE
Command
 nno
Number of the node
 rcoordg(1,nno)
First coordinate (in meter) of node nno.
 rcoordg(2,nno)
Second coordinate (in meter or in degree) of node nno.
 rcoordg(3,nno)
Third coordinate (in meter) of node nno. Present only if ndim = 3.
[A5]
[integer]
[real]
[real]
[real]
3
1
1
2
2
ndim = 2
ndim = 3
Figure 16: global axes in a Cartesian system of coordinates
23/08/2016
33
c) 1 card (optional) for automatic generation
GNODE, nno, rcoordg(1,nno), rcoordg(2,nno), rcoordg(3,nno)
 GNODE
[A5]
This command is used for automatic equidistant generation of nodes between
the previously defined node and the node nno.
 nno
[integer]
Number of the node
 rcoordg(1,nno)
[real]
First coordinate (in meter) of node nno.
 rcoordg(2,nno)
[real]
Second coordinate (in meter or in degree) of node nno.
 rcoordg(3,nno)
[real]
Third coordinate (in meter) of node nno. Present only if ndim = 3.
d) 1 card (optional) for repeating a series of nodes
REPEAT, nno, delta(1), delta(2), delta(3), kgene
 REPEAT
[A5]
This command is used to repeat the definition of nno previously defined
nodes with a increment in coordinates
 nno
[integer]
Number of nodes to be repeated
 delta(1)
[real]
Increment for the first coordinate (in meter).
 delta(2)
[real]
Increment for the second coordinate (in meter or in degree).
 delta(3)
[real]
Increment for the third coordinate (in meter). Present only if ndim = 3.
 kgene
[integer]
Number of times that the previous nno defined nodes have to be repeated
For example, the following series of cards:
NODES
NODE
GNODE
REPEAT
1
4
4
0.000
0.000
0.025
0.000
0.120
0.010
1
2
3
4
5
6
7
8
9
0.000
0.000
0.000
0.000
0.025
0.025
0.025
0.025
0.050
0.000
0.040
0.080
0.120
0.010
0.050
0.090
0.130
0.020
3
is equivalent to:
NODES
NODE
NODE
NODE
NODE
NODE
NODE
NODE
NODE
NODE
34
23/08/2016
NODE
NODE
NODE
NODE
NODE
NODE
NODE
10
11
12
13
14
15
16
0.050
0.050
0.050
0.075
0.075
0.075
0.075
0.060
0.100
0.140
0.030
0.070
0.110
0.150
The nodes that have been generated are shown on Figure 17.
Diamond 2009.a.6 for SAFIR
FILE: EXAMPLE
NODES: 16
13
14
15
16
ELEMENTS: 0
NODES PLOT
9
10
5
11
6
12
7
8
Y
X
Z
1
2
3
4
Figure 17: nodes created by GNODE and REPEAT commands
Note:
If NDIM = 2 and the results of a thermal analysis are to be used in the structural analysis
of beam elements, then the first coordinate corresponds to the local y axis of the beam
element and the second coordinate corresponds to the local z axis of the beam element.
35
23/08/2016
SERIES 11: NODE LINE
This series is only present if MAKE.TEM or MAKE.TEMHA has been given in series 6, e).
The values entered in this series will not influence the values of the temperatures that will
be calculated. They are relevant only for the structural calculation using the beam elements
the section of which is described in the discretization made for the thermal analysis, see [1],
section xxx.
a) 1 card
NODELINE, Y0, Z0
 NODELINE
Command
 Y0
First coordinate of the node line
 Z0
Second coordinate of the node line
b) 1 card
YC_ZC, YC, ZC
 YC_ZC
Command
 YC
First coordinate of the centre of rotation
 ZC
Second coordinate of the centre of rotation
[A8]
[real]
[real]
[A5]
[real]
[real]
36
23/08/2016
SERIES 12: IMPOSED TEMPERATURES
a) 1 card
 FIXATIONS
Command
[A9]
b) 1 card (optional) at each node where the evolution of the temperature as a function of
time is imposed by the user.
BLOCK, nno, cblock(nno)
 BLOCK
[A5]
Command
 nno
[integer]
Number of the node where the evolution of the temperature is imposed
 cblock(nno)
[A10]
Name of the function describing the evolution of the temperature at this
node.
c) 1 card (optional) for each master-slave relationship.
SAME, nno1, nno2, YES
 SAME
[A4]
Command
 nno1
[integer]
Number of the slave node (its temperature is forced to be equal to the
temperature of the master node).
 nno2
[integer]
Number of the master node (a master node can have several slave nodes. A
SAME cards must be introduced for each slave).
 YES
[A3]
Command
d) 1 card (optional) to repeat the previous SAME command.
REPEAT, n, incr, YES
 REPEAT
[A6]
Command
 n
[integer]
Number of times that the previous SAME command has to be repeated.
 incr
[integer]
Increment on the node numbers nno1 and nno2 of the previous SAME
command.
 YES
[A3]
Command
37
23/08/2016
e) 1 card (optional) to create master-slave relationship between all nodes that have the
same coordinates.
SAMEALL, YES
 SAMEALL
[A7]
Command. All the nodes of the model that have the same coordinates (with a
precision of 0.316 mm) will automatically be attributed a master-slave
relationship.
 YES
[A3]
Command
f) 1 card
 END_FIX
Command indicating the end of the series.
[A7]
SERIES 13: SOLID ELEMENTS
a) 1 card
 NODOFSOLID
Command
[A10]
Several cards are then written, following the format of b), c) or d), until all nsolid elements
have been described.
b) 1 card for 1 element
ELEM, nsol, nodesofsolid(1,nsol), nodesofsolid(2,nsol), … ,
nodesofsolid(numberofnodesinsolid,nno), matsolid(nsol), epsrsolid(nsol)
 ELEM
[A4]
Command
 nsol
[integer]
Number of the element
 nodesofsolid(1,nsol)
[integer]
First node of element nsol.
 nodesofsolid(2,nsol)
[integer]
Second node of element nsol.
…
 nodesofsolid(numberofnodesinsolid,nsol)
[integer]
Last node of element nsol.
 matsolid(nsol)
[integer]
Material number of the element nsol. The materials are numbered from 1 to
nmat according to the sequence appearing in series xxx.
 epsrsolid(nsol)
[real]
38
23/08/2016
Residual stress (in Pa or N/m²) in element nsol. The values entered for
epsrsolid will not influence the values of the temperatures that will be
calculated. They are relevant only for the structural calculation using the
beam elements the section of which is described in the discretization made
for the thermal analysis see [1], section xxx. Residual stresses are relevant
when ndim = 2 and MAKE.TEM or MAKE.TEMHA have been given. In other
cases, any value can be specified, preferably 0.
If ndim = 2, numberofnodesinsolid = 4. For triangular elements, nodesofsolid(4,nsol) = 0
If ndim = 3, numberofnodesinsolid = 8. For elements which have a triangular base,
nodesofsolid(7,nsol) = 0, nodesofsolid(8,nsol) = 0.
In 2D elements, the nodes have to be given in counterclockwise order, see Figure 18.
G1, = L2 of a beam element
NDIM = 2
node 3
node 4
face 3
node 3
face 3
face 2
node 1
face 1
node 2
face 4
node 1
face 2
face 1
node 2
G2, = L3 of a beam element
Figure 18: nodes and faces of elements for 2D analyses
In 3D elements with 6 nodes, nodes 1, 2, 3 must belong to one of the triangular frontiers
and nodes 4, 5, 6 to the other one, with the order 1, 2, 3 drilling into the element and
4, 5, 6 drilling out of the element. Node 4 must be on the same edge as node 1, and
thus 5 with 2 and 6 with 3, see Figure 19.
39
23/08/2016
node 5
G1
NDIM = 3
node 2
G3
face 5
face 2
face 3
node 4
node 6
face 4 face 1
node 1
node 3
G1,G2,G3 is dextrorsum
G2
Figure 19: nodes and faces of elements with 6 nodes for 3D analyses
In 3D elements with 8 nodes, nodes 1, 2, 3, 4 must belong to one frontier and nodes 5, 6,
7, 8 to the opposite one, with the order 1, 2, 3, 4 drilling into the element 5, 6, 7, 8
drilling out of the element, see Figure 20. Node 5 must be on the same edge as node 1,
and thus 6 with 2, etc.
node 6
G1
node 7
NDIM = 3
face 3
node 2
node 3
face 6
face 2
face 4
G3
face 5
node 5
node 8
face 1
node 1
node 4
G2
Figure 20: nodes and faces of elements with 8 nodes for 3D analyses
In 3D elements with 4 nodes, nodes 1-2-3 (in that order) drill into the element.
Node 4
face 3 Node 1
face 2
face 4
face 1
Node 3
23/08/2016
Node 2
40
c) 1 card (optional) for automatic generation
GELEM, nsol, nodesofsolid(1,nsol), nodesofsolid(2,nsol), … ,
nodesofsolid(numberofnodesinsolid,nno), matsolid(nsol), epsrsolid(nsol), kgene
 GELEM
[A5]
Command
 nsol
[integer]
Number of the element
 nodesofsolid(1,nsol)
[integer]
First node of element nsol.
 nodesofsolid(2,nsol)
[integer]
Second node of element nsol.
…
 nodesofsolid(numberofnodesinsolid,nsol)
[integer]
Last node of element nsol.
 matsolid(nsol)
[integer]
Material number of the element nsol. The materials are numbered from 1 to
nmat according to the sequence appearing in series xxx.
 epsrsolid(nsol)
[real]
Residual stress (in Pa or N/m²) in element nsol.
 kgene
[integer]
The elements from the previously defined one up to this one will be generated
automatically. kgene is the increment of the nodes from one element to the
next one.
d) 1 card (optional) for repeating a series of elements
REPEAT, n, incr, kgene
 REPEAT
Command
 n
Number of elements to be repeated
 incr
Increment in the node numbers.
 kgene
[A6]
[integer]
[integer]
[integer]
Number of times that the n elements have to be repeated.
For example, the following series of cards:
FIXATIONS
41
23/08/2016
END_FIX
NODOFSOLID
ELEM
GELEM
REPEAT
1
3
3
1
9
1
2
10
6
14
5
13
1
1
1
2
3
4
5
6
7
8
9
1
5
9
2
6
10
3
7
11
2
6
10
3
7
11
4
8
12
6
10
14
7
11
15
8
12
16
5
9
13
6
10
14
7
11
15
1
1
1
1
1
1
1
1
1
0.
0.
4
2
is equivalent to:
FIXATIONS
END_FIX
NODOFSOLID
ELEM
ELEM
ELEM
ELEM
ELEM
ELEM
ELEM
ELEM
ELEM
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
0.0000E+00
Diamond 2009.a.6 for SAFIR
FILE: EXAMPLE
13
14
15
16
NODES: 16
ELEMENTS: 9
3
6
9
10
2
X1
Z
7
4
2
12
SOLIDS PLOT
8
6
1
NODES PLOT
11
5
5
Y
9
8
7
3
4
Figure 21: elements created by GELEM and REPEAT commands
When all elements have been defined, it is possible to change the material of one or several
elements. This can be useful if the input file has been created with a text editor and
extensive utilization has been made of the GELEM and REPEAT commands9.
e) 1 card (optional) for each element in which the material has to be changed.
NEW_MAT, nsol, matsolid(nsol)
 NEW_MAT
Command
 nsol
9
[A7]
[integer]
NEW_MAT is typically used to create a steel rebar in a concrete section.
42
23/08/2016
Number of the element in which the material has to be changed
 matsolid(nsol)
Number of the material to be given to the element nsol
[integer]
For example, the following series of cards:
NODOFSOLID
ELEM
GELEM
REPEAT
NEW_MAT
1
3
3
5
1
9
1
2
2
10
6
14
5
13
1
1
0.
0.
4
2
will allocate the materials to the elements as shown on Figure 22 if STEELEC3EN is the first
material name and CALCONC_EN is the second one, see series 18. Without the NEW_MAT
command, the description of the elements with the appropriate materials would require 7
cards, instead of 4 cards in the example shown here. The utilisation of the REPEAT
command would indeed not be possible.
Diamond 2009.a.6 for SAFIR
FILE: EXAMPLE
NODES: 16
ELEMENTS: 9
3
6
2
1
9
5
4
8
NODES PLOT
SOLIDS PLOT
STEELEC3EN
CALCONC_EN
7
Y
X
Z
Figure 22: result of the NEW_MAT command
SERIES 14: FRONTIERS
a) 1 card
 FRONTIER
Command
[A8]
Different cards can be used in order to allocate boundary conditions to selected surfaces of
some elements. Two different boundary conditions can be used:
 either a time-temperature fire curve Tg describes the environment around the model
(commands F and GF). The heat flux q at the boundary will be calculated from the
43
23/08/2016
temperature of the fire curve Tg and the temperature on the surface Ts according to:
q  h Tg  TS     * Tg4  TS4 
 or the incipient flux on the model is prescribed (commands FLUX and GFLUX).
b) 1 card for each element that has at least one surface exposed to a fire curve.
F, nsol, cfrontiersolid(1,nsol), … ,cfrontiersolid(numberoffrontier,nsol),
 F
[A1]
Command
 nsol
[integer]
Number of the element
 cfrontiersolid(1,nsol)
[A10]
Function describing the fire curve on frontier 1 of element nsol.
…
 cfrontiersolid(numberoffrontier,nsol)
[A10]
Function describing the fire curve on the last frontier of element nsol.
If ndim = 2, numberoffrontier = 4. For triangular elements, cfrontiersolid(4,nsol) = NO
If ndim = 3, numberoffrontier = 6. For elements which have a triangular base,
cfrontiersolid(7,nsol) = NO, cfrontiersolid(8,nsol) = NO.
The order of the surfaces on the elements is linked to the node numbers.
In a triangular element, see Figure 18,
frontier 1 is from node 1 to node 2
frontier 2 is from node 2 to node 3
frontier 3 is from node 3 to node 1
In a quadrangular element, see Figure 18,
frontier 1 is from node 1 to node 2,
frontier 2 is from node 2 to node 3,
frontier 3 is from node 3 to node 4,
frontier 4 is from node 4 to node 1.
In a 3D element with 6 nodes, see Figure 19,
frontier 1 comprises is defined by the nodes 1, 4, 6, 3,
frontier 2 comprises is defined by the nodes 2, 5, 6, 3,
frontier 3 comprises is defined by the nodes 1, 2, 5, 4,
frontier 4 comprises is defined by the nodes 1, 2, 3,
frontier 5 comprises is defined by the nodes 4, 5, 6.
In a 3D element with 8 nodes, see Figure 20,
frontier 1 comprises is defined by the nodes 1, 5, 8, 4,
frontier 2 comprises is defined by the nodes 4, 3, 7, 8,
frontier 3 comprises is defined by the nodes 2, 6, 7, 3,
frontier 4 comprises is defined by the nodes 1, 2, 6, 5,
44
23/08/2016
frontier 5 comprises is defined by the nodes 1, 2, 3, 4,
frontier 6 comprises is defined by the nodes 5, 6, 7, 8.
The function describing the fire curve can be one of the SAFIR defined function (e.g. FISO
or F20), or a user defined function located in a txt file (e.g. myfire.txt), or the
command NO if no fire curve is applied on a particular surface. The meaning of NO is not
“remove any previously fire curve”; it is more to be understood as “disregard this
information”. For example, the two following examples are equivalent:
F
5
FISO
FISO
NO
NO
FISO
NO
NO
FISO
NO
NO
NO
NO
is equivalent to
F
F
5
5
Note: "F20" frontiers can be added on other surfaces, e.g. on the upper unheated side of
a slab.
c) 1 card for automatic generation of frontiers from the previously defined element to this
one.
GF, nsol, cfrontiersolid(1,nsol), … ,cfrontiersolid(numberoffrontier,nsol), kgene
 GF
[A2]
Command
 nsol
[integer]
Number of the element
 cfrontiersolid(1,nsol)
[A10]
Function describing the fire curve on frontier 1 of element nsol.
…
 cfrontiersolid(numberoffrontier,nsol)
[A10]
Function describing the fire curve on the last frontier of element nsol.
 kgene
[integer]
increment on the element numbers
F
GF
5
20
FISO
FISO
FISO
FISO
NO
NO
NO
NO
FISO
FISO
FISO
FISO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
5
is equivalent to
F
F
F
F
5
10
15
20
Figure 23 shows the boundary conditions created by these cards applied on the section
that was shown on Figure 21.
45
23/08/2016
Diamond 2009.a.6 for SAFIR
FILE: EXAMPLE
NODES: 16
2
ELEMENTS: 9
2
3
2
2
1
9
5
1
Y
X
6
FRONTIERS PLOT
1
2
8
4
1
SOLIDS PLOT
FISO
F20
7
1
Z
Figure 23: Boundary conditions
d) 1 card for each element that has at least one surface exposed to a defined heat flux.
FLUX, nsol, cfrontiersolid(1,nsol), … ,cfrontiersolid(numberoffrontier,nsol)
 FLUX
[A4]
Command
 nsol
[integer]
Number of the element
 cfrontiersolid(1,nsol)
[A10]
Function describing the flux (in W/m²) on frontier 1 of element nsol.
…
 cfrontiersolid(numberoffrontier,nsol)
[A10]
Function describing the flux (in W/m²) on the last frontier of element nsol.
The function describing the flux can be one of the SAFIR defined function (e.g. F1000) or,
more likely, a user defined function located in a txt file (e.g. myflux.txt), or the
command NO if no flux is applied on a particular surface.
A positive flux means energy introduced in the section.
It is possible to have a flux condition and a fire curve on the same surface.
Each surface of the model heated by a HASEMI fire must have a FLUX command and the
function HASEMI where appropriate. For example:
FLUX
5
NO
NO
HASEMI
NO
There is no need to add a frontier F20 on a surface heated by a HASEMI flux (for
reemitting energy); it will be automatically added by SAFIR.
Note: if the function HASEMI_FR is used in the FLUX command, the flux as given by the
Equations of EN 1991-1-2 is multiplied by 0.85, as indicated in the French National Annex.
46
23/08/2016
Each surface of the model heated by a LOCAFI fire must have a FLUX command and the
function LOCAFI where appropriate. For example:
FLUX
5
NO
NO
LOCAFI
NO
There is no need to add a frontier F20 on a surface heated by a LOCAFI flux (for
reemitting energy). This effect is included in the model.
e) 1 card for automatic generation from the previously defined element.
GFLUX, nsol, cfrontiersolid(1,nsol), … ,cfrontiersolid(numberoffrontier,nsol), kgene
 GFLUX
[A5]
Command
 nsol
[integer]
Number of the element
 cfrontiersolid(1,nsol)
[A10]
Function describing the flux (in W/m²) on frontier 1 of element nsol.
…
 cfrontiersolid(numberoffrontier,nsol)
[A10]
Function describing the flux (in W/m²) on the last frontier of element nsol.
 kgene
[integer]
increment on the element numbers
g) 1 card
 END_FRONT
Command indicating the end of the series.
[A9]
47
23/08/2016
SERIES 15: VOIDS
The following cards a), b), c) and d) are repeated nvoid times, see series 9. If nvoid = 0, this
series is skipped.
a) 1 card
 VOID
[A4]
Command
b) 1 card for each surface of an element that is adjacent to an internal cavity.
ELEM, nsol, nfrontier
 ELEM
Command
 nsol
Number of the element
 frontiersolid
Number of the frontier of this element that is adjacent to a cavity
[A4]
[integer]
[integer]
If one element has two surfaces that define a void, two cards must be entered. For
example:
ELEM
ELEM
5
5
2
3
c) 1 card for automatic generation from the previously defined element.
GELEM, nsol, nfrontier, kgene
 GELEM
Command
 nsol
Number of the element
 frontiersolid
Number of the frontier of this element that is adjacent to a cavity
 kgene
Increment on the element numbers. Can be positive r negative.
d) 1 card
 END_VOID
Command indicating the end of the cards defining this cavity.
[A4]
[integer]
[integer]
[integer]
[A9]
The elements surrounding a cavity must be described in the same order as they are
encountered by someone walking counter clockwise on the frontier of this cavity.
Any element from the frontier of the cavity can be chosen as the first one in the list. The only
restriction is that the first 2 elements must touch each other. This is because, if they are
separated by an axis of symmetry, SAFIR cannot check whether the cavity is really described
counter clockwise.
48
23/08/2016
Element number
1
2
7
8
9
10 11
12 13
3
4
5
6
14 15 16 17 18 19
Side number
3
4
2
1
Node number
4
3
1
2
Figure 24: description of a cavity
For example, Figure 24 shows a model with 19 rectangular finite elements and an internal
cavity. The cavity can be described by the following cards10:
VOID
ELEM
ELEM
ELEM
ELEM
ELEM
ELEM
GELEM
ELEM
ELEM
END_VOID
16
17
12
12
9
5
3
8
11
3
3
4
3
4
1
1
2
2
–1
Notes:
For radiation inside the void, the emissivity of the materials bording the void is divided by 0,8
for the following materials: 'CALCONCEC2','SILCONCEC2','STEELEC3','STEELEC3EP',
'STEELEC3SH','STEELEC3DC','STEELEC2','STEELEC2EP'. This is to account for the fact
that, in the ENV versions of the Eurocodes, thr remative emissivity was equal to the product
of the emissivity of the material ɛm by the emissivity of the fire ɛf = 0,8. This does not hold for
other materials.
For convection inside the void, the coefficient of convection on unexposed surfaces is used,
irrespective of the temperature of the air in the void. This is to account for the low level of
turbulence expected to take place in the void.
10
In this case, nfrontiervoid = 10, see series 9.
49
23/08/2016
SERIES 16: SYMMETRIES
a) 1 card
 SYMMETRY
Command
[A4]
The following cards b), c) and d) are optional. If no symmetry is used when creating the
model, only cards a) and e) are present. Symmetries can only be used in 2D models.
b) 1 card for each real axis of symmetry in the section.
REALSYM, nno1, nno2
 REALSYM
Command
 nno1
Number of the first node defining the axis of symmetry
 nno2
Number of the second node defining the axis of symmetry
[A7]
[integer]
[integer]
The line passing through nodes nno1 and nno2 is a real axis of symmetry for the model.
This possibility is used when the 2D model represents part of a beam section. When
creating the .TEM file, the fibres located on the other side of the axis of symmetry are
created, with the same sectional area, same material number and same temperature.
This option is used when there is a thermal axis of symmetry in the section, which will not
be a structural axis of symmetry in the structural calculation.
c) 1 card if the Y axis is an axis of symmetry in the section.
YSYM
 YSYM
Command
[A4]
This card is used if the first axis of coordinate of the model, i.e. the local y axis for the
beam element, is a thermal and structural axis of symmetry. When creating the .TEM file,
the area of the fibres is simply multiplied by 2. This card can only be used when the
structural model is a 2D model. Another possible option is not to use the YSYM command
and divide the loads applied on the structure by 211.
The YSYM command can be used simultaneously with one or several REALSYM
commands.
11
It is even simpler to model the complete section and not to bother about symmetries. This will nevertheless
slightly increase the run times and could be an issue, for example, with HASEMI fires.
50
23/08/2016
d) 1 card for each axis of symmetry in a cavity.
SYMVOID, nno1, nno2, nvoid
 SYMVOID
[A7]
Command
 nno1
[integer]
Number of the first node defining the axis of symmetry
 nno2
[integer]
Number of the second node defining the axis of symmetry
 nvoid
[integer]
Number of the cavity in where this axis of symmetry must be considered.
This card has only an effect on the thermal calculation (through the calculation of the
view factors in the cavity). If this axis is also a real axis of symmetry, a REALSYM
command must also be defined for this axis. It is theoretically possible to have as much as
4 different axes of symmetry in a cavity, but the validity of the procedure has been
verified only with 2 axis of symmetry in a cavity.
e) 1 card
 END_SYM
Command to indicate the end of this series
[A7]
SERIES 17: PRECISION
a) 1 card
PRECISION, precision
 PRECISION
Command
 precision
[A9]
[real]
Small value which must be reached to have convergence.
The value of precision is used as a small number at different locations in SAFIR. A good
value depends on the type of model that is being analysed and the experience of the user
is useful to choose a good value. 2×10-3 is normally acceptable for thermal calculations.
51
23/08/2016
SERIES 18: MATERIALS
a) 1 card
 MATERIAL
Command
[A8]
All nmat materials (see Series 8) have to be described here. The description of a material
comprises:
 A number, from 1 to nmat. the number is defined automatically by the order of
appearance, from 1 for the first material to nmat for the last one.
 A name, given in cards b). Different materials in a model can have the same name (with
the exception of USERx, see below). This will be the case if similar materials have
different properties.
 Materials properties. They are given in cards c).
The descriptions given hereafter give the cards b) and c) for the materials that can be used in
a thermal analysis.
INSULATION: material with constant thermal properties
b) 1 card
 INSULATION
Name of the material
c) 1 card
k, c, rho, w, hh, hc, ε
 k
Thermal conductivity, in W/mK
 c
Specific heat, in J/kg K
 rho
Specific mass of the dry material, in kg/m³
 w
Water content, in kg/m³
 hh
Coefficient of convection on heated surfaces, in W/m²K
 hc
Coefficient of convection on unheated surfaces, in W/m²K
 ε
Emissivity (no dimension)
[A10]
[real]
[real]
[real]
[real]
[real]
[real]
[real]
52
23/08/2016
USERx: material with some thermal properties that vary with temperature.
b) 1 card
USERx, ntemperature
 USERx
[A5]
Name of the material. Five different materials are possible, namely USER1,
USER2, USER3, USER4 and USER5. As an exception to all other material types,
only one material number can have the name USER1, only one material can
have the name USER2, etc.
 ntemperature
[real]
Number of temperatures at which the thermal properties are given.
Properties are given at a certain number of temperatures, given in increasing
order. Linear interpolation is made for intermediate temperatures.
The value of ntemperature cannot be smaller than 2 (because linear
interpolation must be made.
c) ntemperature cards
Card 1.
T, k, c, rho, w, hh, hc, ε, r
 T
[real]
12
First temperature (in degree Celsius) at which thermal properties are given
 k
[real]
Thermal conductivity at T, in W/mK
 c
[real]
Specific heat at T, in J/kg K
 rho
[real]
Specific mass of the dry material at T, in kg/m³
 w
[real]
Water content, in kg/m³
 hh
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
 r
[real]
Any positive value (and 0) will force k, c, rho to be non reversible. This means
that, during cooling from a maximum temperature Tmax, these properties will
keep the value that was valid for Tmax. Any negative value will force these
12
and, therefore, lowest
53
23/08/2016
properties to be reversible, which means that the value of the property only
depends on its current temperature, be it during heating or during cooling.
Cards 2 to ntemperature.
T, k, c, rho
 T
Temperature (in degree Celsius) at which thermal properties are given
 k
Thermal conductivity at T, in W/mK
 c
Specific heat at T, in J/kg K
 rho
Specific mass of the dry material at T, in kg/m³
[real]
[real]
[real]
[real]
Take care that the temperatures calculated at the nodes of the model during the iteration
process do not go outside the interval of temperatures given by the card N°1 and the card
N°ntemperature.
As an example, the following 12 cards have been used to represent concrete of EN 19921-2 heated by the ISO curve, according to the French National Annex.
USER1
12
0.
2.0000
50.
1.8801
100.
1.7656
115.
1.7323
140.
1.6778
160.
1.1570
200.
1.1108
400.
0.9072
600.
0.7492
800.
0.6368
1000.
0.5700
1200.
0.5488
900.
900.
900.
915.
940.
960.
1000.
1100.
1100.
1100.
1100.
1100.
2300.
2300.
2300.
2300.
2286.
2276.
2254.
2185.
2145.
2105.
2064.
2024.
34.5
25.
4.
0.7
1.
54
23/08/2016
GYPSUM: Gypsum plaster boards
b) 1 card
 X_GYPSUM or C_GYPSUM
Name of the material
c) 1 card
hh, hc, ε
 hh
[A8]
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
C_GYPSUM has a 20°C density of 732 kg/m³, whereas X_GYPSUM has a 20°C density
of 648kg/m³.
Gypsum type materials lead to a slow convergence of the iterations during the time
integration process because of the various peaks in the curve of equivalent specific heat.
A time step as small as 1 second may be required.
CONCRETE of EN 1992-1-2:
b) 1 card
 CALCON_ETC or SILCON_ETC
[A10]
Name of the material. CALCONC_EN, SILCONC_EN, CALCONC_PR and
SILCONC_PR can also be used; they have the same thermal properties.
c) 1 card
rho, w, hh, hc, ε, r
 rho
[real]
Specific mass of concrete (including moisture content), in kg/m³
 w
[real]
Water content, in kg/m³
 hh
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
 r
[real]
Parameters that allows tuning the thermal conductivity between the lower
limit and the upper limit, see clause 3.3.3 of EN 1992-1-2. At any temperature
55
23/08/2016
T, the thermal conductivity will be interpolated between the lower limit kL and
the upper limit kU by the following equation:
k  k L  r  kU  k L 
CARBON STEEL of EN 1993-1-2 or EN 1992-1-2:
b) 1 card
 STEELEC3EN or STEELEC2EN
Name of the material
c) 1 card
hh, hc, ε
 hh
[A10]
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
STAINLESS STEEL of EN 1993-1-2:
b) 1 card
 SLS1.4301, SLS1.4401, SLS1.4404, SLS1.4571, SLS1.4003,
SLS1.4462, SLS1.4311
Name of the material
c) 1 card
hh, hc, ε
 hh
[A9]
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
ALUMINIUM of EN 1999-1-2:
b) 1 card
 AL6061_T6, AL6063_T6, AL5083_O, AL5083_H12
Name of the material
[A9]
56
23/08/2016
c) 1 card
hh, hc, ε
 hh
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
WOOD of EN 1995-1-2:
b) 1 card
 WOODEC5
Name of the material
[A10]
c) 1 card
rho, w, hh, hc, ε, r, l, m, n
 rho
[real]
Specific mass of wood (including moisture content), in kg/m³
 w
[real]
Water content, in % of the dry mass
If, for example, rho = 450 kg/m³ and w = 12, the density of dry wood is
450/(1+12/100) = 401.7 kg/m³
 hh
[real]
Coefficient of convection on heated surfaces, in W/m²K
 hc
[real]
Coefficient of convection on unheated surfaces, in W/m²K
 ε
[real]
Emissivity (no dimension)
 r
[real]
Ratio of conductivity along the grain by conductivity perpendicular to the
grain (no dimension), usually greater than 1.
 l, m, n
[3 real]
The vector <l;m;n> gives the direction of the grain
Direction of the grain is essential for 3D analyses.
Vector components for 2D analysis are given in the order <y;z;x>. Most 2D
analyses are performed on the section of a beam or a column and the grain is
perpendicular to the section, with a vector <0;0;1>. If grain direction is parallel
to z-axis, for example, then the vector should be <0;1;0>.
In 2D problems, two different possibilities exist to define an isotropic
material:
57
23/08/2016
1) specifying a vector <0;0;1>
2) specify r = 1
SERIES 19: TIME DISCRETIZATION
a) 1 card
 TIME
[A4]
Command
b) at least 1 card, maximum 100 cards
dt, t
 dt
Time step (in seconds) used for the time integration
 t
Time (in seconds) until which dt is used
c) 1 card
 END_TIME
Command
[real]
[real]
[A8]
The following example leads to a time step of 12 seconds being used during the first hour,
then 20 seconds up 2 hours, time when the simulation will be stopped:
TIME
12.
20.
3600.
7200.
END_TIME
SERIES 20: OUTPUT RESULTS
a) 1 card
 OUTPUT
Command
[A6]
b) 1 card
 TIMEPRINT
Command
[A9]
c) at least 1 card, maximum 16 cards
dt, t
 dt
[real]
Time step (in seconds) used for writing the calculated temperatures in the
.OUT file and in the eventual .TEM or .TSH files.
 t
[real]
58
23/08/2016
Time (in seconds) until which dt is used
d) 1 card
 END_TIMEPR
Command to indicate the end of the TIMEPRINT cards.
[A10]
The following example leads to a time step of 1 minutes being used for writing the
temperatures during the first hour, then 15 minutes up 2 hours:
OUTPUT
TIMEPRINT
60.
900.
3600.
7200.
END_TIMEPR
e) 1 card (optional)
 PRINTDEPL
[A9]
Writes the increment of temperatures at each iteration of each time step of
the time integration, see series 19. This command is used only for debugging
convergence problems or to decide of an appropriate time step for the time
integration. It is not used for usual calculations because it slightly increases
the run time but, first of all, it increases significantly the size of the .OUT file.
f) 1 card (optional)
 PRINTFHE
[A9]
Writes the out of balance thermal loads at the nodes at each iteration of each
time step of the time integration, see series 19. This command is used only for
debugging convergence problems or to decide of an appropriate time step for
the time integration. It is not used for usual calculations because it slightly
increases the run time but, first of all, it increases significantly the size of the
.OUT file.
g) 1 card
1 blank line to indicate that the input file is finished.
[-]
Finally, the example input file given hereafter yields the temperatures calculated after 2
hours shown on Figure 25.
This file has been created as an example for the users manual of SAFIR
Date: 16/12/2010
Author: Jean-Marc Franssen
NNODE
NDIM
NDOFMAX
EVERY_NODE
END_NDOF
TEMPERAT
TETA
TINITIAL
16
2
1
1
0.9
20.0
59
23/08/2016
NMAT
2
ELEMENTS
SOLID
NG
NVOID
END_ELEM
9
2
0
NODES
NODE
GNODE
REPEAT
1
4
4
0.000
0.000
0.025
0.000
0.120
0.010
3
5
13
1
1
FIXATIONS
END_FIX
NODOFSOLID
ELEM
GELEM
REPEAT
NEW_MAT
1
3
3
5
1
9
1
2
2
10
6
14
0.
0.
4
2
FRONTIER
F
GF
F
GF
END_FRONT
1
7
1
3
FISO
FISO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
F20
F20
3
1
SYMMETRY
END_SYM
PRECISION
0.002
MATERIALS
STEELEC3EN
25.
4.
0.7
2300.
46.
25.
12.
20.
3600.
7200.
60.
900.
3600.
7200.
CALCONC_EN
4.
0.8
0.
TIME
END_TIME
OUTPUT
TIMEPRINT
END_TIMEPR
60
23/08/2016
Diamond 2009.a.6 for SAFIR
FILE: EXAMPLE
NODES: 16
ELEMENTS: 9
SOLIDS PLOT
CONTOUR PLOT
TEMPERATURE PLOT
TIME: 7200 sec
893.30
890.00
870.00
850.00
830.00
810.00
790.00
770.00
750.00
734.90
Y
X
Z
Figure 25: calculated temperatures
61
23/08/2016
D.1.3 Detailed description of the file describing the HASEMI fire(s) or the
LOCAFI fire
If a HASEMI function is used in a FLUX command of the input file, a file called hasemi.txt
must be present in the same folder as the input file. This file describes the local Hasemi
fire(s).
If a LOCAFI function is used in a FLUX command of the input file, a file called locafi.txt
must be present in the same folder as the input file. This file describes the local fire.
The format of these files is similar. It is described in this section.
SERIES 1: COMMENTS
a) Multiple cards (can be 0)
As many lines as required may be introduced by the user at the beginning of the file in
order to comment the content of the file. These comments will not influence in any
manner the results of the calculation. They will help understand what the file is about,
which may be particularly useful if the file is found and opened several years after it has
been created.
b) 1 card
1 blank line to indicate that the comments are finished.
[-]
SERIES 2: NUMBER OF FIRE(S)
a) 1 card
NFIRE, nfire
 NFIRE
[A5]
Command
 nfire
[integer]
Number of localised fires that will influence the model in the thermal analysis
nfire can be greater than 1 for HASEMI fires.
Nfire must be equal to 1 for a LOCAFI fire.
The following series 3 to 6 have to be written nfire times, once for each localised fire.
SERIES 3: POSITION OF THE FIRE
a) 1 card
FIRE_POS, posi1, posi2, [posi3]
 FIRE_POS
Command
 posi1
[A8]
[real]
62
23/08/2016
Position of the fire (in meters) along the first axis of coordinates of the
structural analysis
 posi2
[real]
Position of the fire (in meters) along the second axis of coordinates of the
structural analysis
 [posi3] (optional)
[real]
Position of the fire (in meters) along the third axis of coordinates of the
structural analysis. Only required for 3D structural models
SERIES 4: CEILING HEIGHT
a) 1 card
HEIGHT, hc
 HEIGHT
[A6]
Command
 hc
[real]
Vertical distance (in meters) from the fire source to the ceiling.
Only this value is present in the local fire model of EN 1991-1-2. It allows
detecting whether the flame of the fire touches the ceiling or not. It will thus
influence the severity of the attack of the fire on the structure.
The vertical position of the fire given in Series 3 (posi2 for 2D structural
models and posi3 for 3D structural models) is in fact not used in the HASEMI
model.
SERIES 5: DIAMETER
a) 1 card
 DIAMETER
Command
[A8]
b) at least 2 cards, maximum 100 cards
t, diameter
 t
[real]
Time (in seconds) when the diameter is given. The evolution of the diameter
as a function of time will be linearly interpolated between the values given in
this series.
 diameter
[real]
Value of the diameter (in meters) of the fire at time t
c) 1 card
 END_DIAM
Command
[A8]
63
23/08/2016
SERIES 6: RATE OF HEAT RELEASE
a) 1 card
 RHR
[A8]
Command
b) at least 2 cards, maximum 100 cards
t, rhr
 t
[real]
Time (in seconds) when the rate of heat release is given. The evolution of the
rate of heat release as a function of time will be linearly interpolated between
the values given in this series.
 rhr
[real]
Value of the rate of heat release (in W) of the fire at time t
c) 1 card
 END_RHR
Command
[A7]
Hereafter is given an example of a hasemi.txt file.
Description of the local fire by the Hasemi method.
There are 2 local fires.
This file has been used with the structure described in the file TEST.IN
NFIRE
2
FIRE_POS
HEIGHT
DIAMETER
0.
Number of localised fires
0.
0.5
2.5
Distance between the fire and the ceiling
0.
300.
600.
0.
3.
4.
Time (s) - Fire source diameter (m)
0.
600.
0.
900000.
Time (s) - Rate of heat release (W)
END_DIAM
RHR
END_RHR
FIRE_POS
HEIGHT
DIAMETER
1.5
0.
0.
100.
400.
600.
0.
0.
3.
4.
0.
100.
500.
600.
0.
0.
900000.
900000.
0.5
2.5
END_DIAM
RHR
END_RHR
64
23/08/2016
Hereafter is given an example of a locafi.txt file.
Description of a local fire following the LOCAFI model.
There is only 1 local fire.
NFIRE
1
FIRE_POS
HEIGHT
0.
Number of localised fires. Must be 1.
0.
0.2
3.0
Distance between the fire and the ceiling
DIAMETER
0.
600.
3600.
END_DIAM
0.
3.
3.
Time (s) - Fire source diameter (m)
0.
3534000.
3534000.
Time (s) - Rate of heat release (W)
(corresponds to 500 kW/m²)
RHR
0.
600.
3600.
END_RHR
65
23/08/2016
D.2 Input techniques
There are four known techniques for creating input files.
D.2.1 Text editor
Any text editor that allows creating, opening, modifying and saving text files can be used to
create input files or hasemi.txt files.
This technique is the best choice if one or a few input data have to be modified in an existing
input file. New input files can easily be created by the SaveAs option of the editor if the
model to be analysed is very close to an existing model.
When an input file has to be created from scratch, or if extensive modifications have to be
made in an existing file, it is recommended to use an editor that has sufficient capabilities
such as, e.g., working in column mode or comparing files. The developers have been using
ultraedit (http://www.ultraedit.com/) with great satisfaction.
D.2.2 Personal programming
Users with programming skills can write their own software for creating input files. This
requires significant efforts for a program that, typically, will allow creating a very specific
type of model to be created, with very little, if any, versatility. This technique can be the best
choice if a very large number of similar models have to be created, for example in order to
perform a parametric analysis. This technique could also be considered for a user that has
only one type of structure to analyse with varying dimensions.
Any programming technique can be used, as far as it can create a text file. Spreadsheets such
as Excel have been used with success.
D.2.3 Wizard
A particular tool belonging to the category described in section D.2.2 has been written by
the developers. Given the name of Wizard, it allows to create in a few clicks an input file for
a thermal analysis of a 2D section based on a hot rolled H section. The section can have a
concrete slab on the top flange and be protected by applied thermal insulation. The
catalogue of H sections of ArcelorMittal is embedded in the software.
The last version of Wizard, Wizard2004, version 4.0.0 is still distributed on the CD provided
with SAFIR and can be downloaded for free from the SAFIR website
(http://www.argenco.ulg.ac.be/logiciels/SAFIR/downloads-3.html). It is still compatible with
66
23/08/2016
SAFIR2011 but will probably not be maintained in the future, because a similar tool has been
incorporated in the tool described in Section D.2.4.
This tool is sometimes used as a quick way to create a file that has the correct format for
SAFIR, to be modified subsequently with a text editor to produce a file describing the model
to analyse.
D.2.4 GID
GID (http://gid.cimne.upc.es/) is a commercial pre-processor that can help creating a finite
element model with a graphic user interface, see Figure 26.
Figure 26: Graphic interface of GID
A reasonably powerful version can be downloaded and run for free, while the full
professional version can be purchased with a price that is different for educational
organisations or for corporate organisations.
The modules that can make GID writing the input files in the format of SAFIR (for thermal as
well as for structural calculations), as well as some short tutorials, can be downloaded on the
SAFIR web site (http://www.facsa.ulg.ac.be/cms/c_1584029/en/safir).
67
23/08/2016
GID is clearly the best choice when various ands complex models have to be created.
Experience nevertheless shows that users very quickly learn how to create their models in
GID, but have a tendency to let the number of nodes and elements increase significantly
compared to models made with a text editor. This is not detrimental for single 2D thermal
calculations. More care should be exercised and more skills in using GID should be gained in
order to improve the quality of the model when multiple 2D calculations have to be
performed because of exposure to a local HASEMI fire or when really big 3D models have to
be analysed.
D.3 Limits on input based on the stability, accuracy and practicality
of the data
The main limits based on stability and accuracy are based on the link that must exist
between the time step used during time integration, see Series 19 in Section D.1.2, and the
size of the finite elements in the direction of the heat flow. They have been studied in detail
in [1], section xxx.
Accuracy increases when the time step decreases.
Suitable values of the time step depend on the types of material that are present in the
model and on the geometry of the model. With usual finite element sizes, in the order of 10
mm perpendicular to the boundary of the model in the zones near the surfaces of the
models, times steps from 12 to 20 seconds have been used without any problem in concrete
or steel models, protected or not. Higher values can eventually be used after the first 30
minutes of an ISO fire. When gypsum material is all or part of the model, time steps as short
as 1 second may have to be used from the very first minutes of the simulation. When the
section comprises one or several internal cavities, time steps in the order of 1 second must
also be used from the moment when the heat wave reaches the cavities and heat exchanges
by radiation in the cavities start playing a role.
D.4 Default variables - Process for setting those variables to userdefined values
There is no default variable in the model other than physical constants.
Default thermal properties are embedded for the thermal models of materials from the
Eurocodes (steels, concretes, aluminums and wood) and for gypsum materials. For using
other material properties, the user has to use either INSULATION material or USERx
materials.
68
23/08/2016
D.5 Consecutive cases
Consecutive cases can be executed one by one, if the user starts safir2011.exe directly from
the Windows Explorer, or automatically by the SafirShell.
In any case, to the knowledge of the developers, all variable are completely re-initialized;
there is no data retention from case to case. Any observation of the contrary should be
reported as a bug to the developers.
E.
External data files
E.1 Content and organization of external data files
There is no external data file other than those required for the input, see sections D.1.2 and
D.1.3, and those created by execution of SAFIR, namely the .OUT file and eventual .TEM or
.TSH files.
69
23/08/2016
Terms and definitions
Card
A series of information of the input file that is read by in single READ statement by SAFIR.
The rules for reading a card are the rules of the FORTRAN language. The input data of a
single card can be split in the input file on several consecutive lines.
Discretization
The process by which the user represents to structure (that, by nature, is a continuous
object) into a model made of a finite number of nodes and elements.
Iteration
Because the thermal problem to be solved is non linear, SAFIR cannot calculate the
temperature directly at a given time step as a function of the temperatures that were
calculated at the previous time step. In order to calculate the temperature at a given time
step, SAFIR has to perform several trial and error calculations during which the temperatures
are progressively refined. Each of these trials and errors is called an iteration.
Line
A physical record of the input file.
Time step
A particular time during the course of the fire fat which SAFIR will calculate the
temperatures (times steps used for the calculation). The time steps are chosen by the user.
User
The person who collects the input data for the problem to be solved, who creates a model
representing the problem (the fire and the structure) that can be analyzed by SAFIR and who
interprets the results provided by SAFIR.
70
23/08/2016
List of symbols
t
Thickness of a shell finite element [m]
71
23/08/2016
">