University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Validation of SAFIR® through
DIN EN 1992-1-2 NA
Comparison of the results for the examples
presented in Annex CC
March 2018
J. Ferreira
J.-M. Franssen
T. Gernay
A. Gamba
University of Liege – ArGEnCo – Structural Engineering
Institut de Mécanique et Génie Civil
Allée de la Découverte, 9 - 4000 Liège – Belgium
Sart Tilman – Bâtiment B52 - Parking P52
www.argenco.ulg.ac.be
Tél.: +32 (0)4 366.92.65
Fax: +32 (0)4 366.95.34
E-mail :
jm.franssen@uliege.be
tgernay@jhu.edu
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table of contents
1.
Introduction .................................................................................................................................................3
1.1.
Form of validation............................................................................................................................3
1.2.
Structure of the document............................................................................................................3
1.3.
Sources of differences in results ................................................................................................3
1.4.
Results versus visualisation .........................................................................................................4
1.5.
Versions of the software used .....................................................................................................6
2.
Validation examples..................................................................................................................................7
2.1.
2.2.
2.3.
2.4.
2.5.
2.6.
steel
2.7.
2.8.
2.9.
2.10.
2.11.
Example 1 – Heat transfer (cooling) with constant properties.....................................7
Example 2 – Heat transfer (heating) with varying properties ................................... 19
Example 3 – Heat transfer through several layers .......................................................... 28
Example 4 – Thermal expansion ............................................................................................. 31
Example 5 – Temperature-dependent stress-strain curves of concrete and steel
37
Example 6 – Temperature-dependent limit-load-bearing capacity of concrete and
40
Example 7 – Development of restraint stresses ............................................................... 42
Example 8 – Weakly reinforced concrete beam ............................................................... 44
Example 9 – Heavily reinforced concrete beam ............................................................... 52
Example 10 – Reinforced concrete beam-column ........................................................... 58
Example 11 – Composite column with concrete cores .................................................. 64
3.
General conclusions ............................................................................................................................... 68
4.
References.................................................................................................................................................. 69
March 2018
2
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
1. Introduction
1.1. Form of validation
Annex CC of DIN EN 1992-1-2 NA [2] presents a series of cases that allow benchmarking
software tools aimed at the design of structures in a fire situation.
With the goal of providing a validation document for the finite element code SAFIR [1], a
comparison of the reference results for the cases presented in the Annex CC with the results
obtained by SAFIR has been carried out and is presented in this document.
The validation typically consists in a comparison between the value of a result
(temperature, displacement or others) obtained by SAFIR and the value given as a reference and
supposed to be the « true » result. The value obtained must fall in the interval stipulated by the
document.
1.2. Structure of the document
This document contains comparisons of SAFIR with the examples provided by the DIN. For
each example, keywords are initially provided in order to easily detect what is being analysed in
the example. The objective of the example is then summarized and a description of the necessary
information concerning geometry, boundary conditions, loads, parameters, material laws, etc, is
given. Finally, a description of the model used and possible assumptions is presented and the main
conclusions about the comparisons of SAFIR to the reference solutions are exposed.
All the SAFIR files used are made available with this document, and references to the folders
where they are located are given in the sub-chapters related to each model. The pictures that allow
visualizing the results of SAFIR were made with the post-processor DIAMOND 2016, which can be
downloaded for free on the SAFIR website.
1.3. Sources of differences in results
Some differences between the results of SAFIR and the reference values may be observed
either due to different formulations being used or to the way the results are obtained and
presented. This is discussed in the next two sections.
1.3.1
Significant digits
In many cases, the true value is a real number and its expression should involve an infinite
number of significant digits, like for instance 0,03458623579841265895123548… However, the
determined value is always given with a certain resolution, i.e. a limited number of significant
digits, and it is not always mentioned whether this has been obtained by rounding or by truncating
the true value. In the example above, with only 2 significant digits, the rounded value is 0,035
whereas the truncated value is 0,0341. Such an uncertainty of 0,001 represents 2,857% of the
rounded value and 2,941% of the truncated value whereas the maximum allowed deviation may
be as low as 1%.
Moreover, the results produced by SAFIR have by default a limited number of significant
digits (typically 8 or 16 digits). As it may not be relevant to print all results with such a high
1
This reference value of 0,034 is present in Example 5 of DIN EN 1992-1-2.
March 2018
3
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
resolution, results are usually rounded before they are written in the two different files that are
produced by SAFIR: filename.out and filename.xml. In these two files, the resolution may not be
the same. For example, the displacements written in the out file, meant to be read by the human
eye, are in 1/100 of a mm, which is supposed to be sufficiently small for a civil engineering
structure. In the xml file, however, meant to be used by the graphic postprocessor Diamond, they
are written with 3 significant digits.
In the exercises reported in this document, the values considered for SAFIR are always the
most precise of both, based on the fact that any user has access to both files. For example, if the
example above is a displacement in mm, SAFIR would write 0,00003 in the out file and 3,46E-03
in the xml file and the later would be used to calculate the deviation from the reference value. In
this case, the double effect of rounding or truncating the reference value and of rounding the result
of SAFIR would give a deviation of 1,14% with the rounded reference value and 1,76% with the
truncated reference value, even if SAFIR calculates the true value to the 8th or 16th significant
digit.
In some cases, it is possible to modify the size of the structure to be analysed in the reference
case to obtain, at least, more significant digits in the out file produced by SAFIR. For example, the
thermal expansion of a bar would be 10 times higher if the size of the bar is multiplied by 10.
Interested users may want to do that, but this has not been done for this document.
1.3.2
Refinement of the model
The results calculated by a finite element software highly depend on the discretisation of
the model in space, with finer grids yielding more correct results. If the analysis is transient, the
results also depend on the discretisation in time, with smaller time steps yielding more correct
results. The question of the discretisation to be used by the software, to produce the results that
will be compared with the reference value, is typically not discussed in the documents that give
the reference value.
In this exercise, the results are first presented with a model that is sufficiently refined (in
space and in time) to ensure a converged solution, which means that the solution would not be
different2 with a finer model. Yet, it is highly valuable for the user to have an idea of the
convergence of the solution when the model is “degraded”. This allows answering the following
question: what level of refinement is required for the software to yield acceptable results? This is
why for some of the examples, in addition to the results produced with the converged model, we
will present also results obtained with different levels of refinement. It must then come as no
surprise if, when the model is too crude, the results don’t fall within the acceptable limits anymore.
1.4. Results versus visualisation
Results produced by SAFIR come in the form of numbers. In order to validate the software,
these numbers are considered and compared to the reference values and the results of the
comparisons are usually given in tables. Yet, in order to give a more intuitive feeling of the results,
these results can also be processed in order to create drawings. The graphical tool DIAMOND has
2
Within the limits of the available resolution
March 2018
4
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
been specifically developed at Liege University to create drawings based on SAFIR results and has
been used in this report3.
It has yet to be understood that some simplifications may be used to produce the drawings
and these are discussed in this section.
1.4.1
Temperature field on triangular facets.
The temperature distribution on a triangular facet of a 3D SOLID element or on a triangular
2D SOLID element varies linearly. The graphic representation of the temperature distribution on
such facets being linear is the exact representation of the temperature distribution considered in
SAFIR, see Figure 12 for example.
The same holds for the representation of the warping function calculated for a torsion
analysis.
1.4.2
Temperature field on quadrangular facets.
The temperature distribution on a quadrangular facet of a 3D SOLID element or on a
quadrangular 2D SOLID element varies in a nonlinear manner. Being based on the temperature of
the 4 nodes located at the corners of the facet, the temperature distribution is driven by an
equation of the type given hereafter.
T(x,y) = k1 + k2 x + k3 y + k4 xy
In order to accelerate the drawings, this nonlinear distribution is simplified in DIAMOND.
The temperature at the centre of gravity of the facet Tc is calculated exactly as the average of the
4 corner temperatures. Then DIAMOND divides the quadrangle into 4 triangles, each one based
on the centre of the quadrangle and 2 adjacent corner nodes, see Figure 1. A linear temperature
distribution is then assumed and drawn in each of the 4 triangles, and this distribution is different
from the real distribution considered in SAFIR. This effect of artificial linearization is clearly
visible, for example, on Figure 30 a) and Figure 31 a).
Tc
Figure 1 – Division of a quadrangle by DIAMOND
The same holds for the representation of the warping function calculated for a torsion
analysis.
The visual effect of the approximation vanishes with the refinement of the mesh.
3
Any other graphical software could be used provided it can read the XML file produced by SAFIR.
March 2018
5
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
1.4.3
Representation of deformed beam elements
Beam finite elements in the deformed configuration are curved, because the end nodes are
subjected to a rotation with respect to the chord that joints them. Yet, in order to simplify and to
accelerate the drawing process, DIAMOND will draw each beam finite element as a straight line
between the end nodes. Here again, the drawn situation does not correspond exactly to the
situation considered in SAFIR.
Figure 2 – Simplification of the drawn deformed configurations
In Figure 2, the thick curved line represents schematically a deformed configuration that
could be considered by SAFIR whereas the two thin lines represent the configuration that would
be drawn by DIAMOND for two beam finite elements.
Here also, the visual effect of the approximation vanishes with the refinement of the mesh.
1.5. Versions of the software used
The version of SAFIR used for performing the calculations was SAFIR 2016.c.0, whereas two
different versions of DIAMOND2016 - versions v1.2.1.20 and v1.3.0.0 - were used for visualising
the results and creating some of the illustrations presented in this document.
March 2018
6
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2. Validation examples
2.1. Example 1 – Heat transfer (cooling) with constant properties
2.1.1
Keywords
Heat-transfer, conduction, convection, constant thermal properties
2.1.2
Objective
The goal of this example is to analyse the heat transfer by conduction and convection in a
2D square section with constant thermal properties.
2.1.3
Description of the problem
A square section with the characteristics defined in Figure 1 and Table 1 is analysed. The
temperature in the square section is uniform and equal to 1000°C at time t = 0 s when it is exposed
to a gas with temperature = 0°C on one of the edges, the other edges being adiabatic. Heat transfer
from the gas to the solid is by linear convection. In order to validate the results, the temperature
θ0 calculated at the centre of the opposite edge, at point X, is compared to the reference values
presented in DIN EN1992-1-2 NA at different time instants.
Figure 3 – Example 1: Cooling down process in section with constant properties
Table 1 – Dimensions, material properties and boundary conditions for Example 1
Properties
Thermal conductivity λ
Specific heat cp
Density ρ
Dimensions h, b
Coefficient of convection αc
Emissivity εres = εm. εf
Ambient temperature Θu
Temperature in the cross-section Θcs
March 2018
W / (m∙K)
J / (kg∙K)
Kg / m3
M
W / (m2∙K)
°C
°C
Value
1
1
1000
1
1
0
0
1000
7
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.1.4
Model and results (see folder DIN1_4)
As the heat flow is uniaxial, from the exposed edge to the opposite edge, a model with only
two rectangular elements on the width of the section is sufficient (in fact, a model with one single
element on the width would yield the same answer, but it would not be possible to calculate the
temperature at the centre node of the opposite edge). A structured mesh formed by 100
quadrilateral elements was used with 50 elements on the height, as depicted in Figure 4. Each of
these elements contains 4 gauss points of integration (2x2) in its plane.
Figure 4 – Thermal model of the cross-section for Example 1 (2 x 50 SOLID elements)
In SAFIR, a FRONTIER constraint with the function F0 was applied on the exposed edge, i.e.
the lower edge in Figure 4 .
The PRECISION command was set to 1.0E-3. The material INSULATION, having constant
material properties, was used and given the properties described in Table 1.
The time step chosen was 1 second (final time / 1800).
In Figure 5 is shown the distribution of the temperatures in the cross-section determined
by SAFIR for the time t = 1800 s.
Figure 5 – Temperatures determined by SAFIR for Example 1, for t = 1800 s
March 2018
8
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 2 shows the temperatures obtained by SAFIR and those given as reference by the DIN.
Table 2 – Temperatures Θ0 at point X for Example 1
Time
t
s
0
60
300
600
900
1200
1500
1800
Limits
Reference
temperature
Calculated
temperature
Θ0
Θ'0
°C
1000
999.3
891.8
717.7
574.9
460.4
368.7
295.3
Deviation
(Θ'0 - Θ0)
°C
1000
999.20
891.80
717.78
574.99
460.52
368.84
295.42
K
0.00
-0.10
0.00
0.08
0.09
0.12
0.14
0.12
±5.00
(Θ'0 - Θ0) / Θ0 ∙
100
%
0.00
-0.01
0.00
0.01
0.02
0.03
0.04
0.04
±1.00
It can be seen that the deviations fall well within the intervals of values defined in the DIN.
2.1.5
Analysis of the influence of different parameters
In this sub-chapter, an analysis of the sensibility of the results to different parameters is
done. This will provide some indications on the minimum value of the time step or minimum
number of nodes necessary in order to accurately simulate the cooling down process on the crosssection, as well as to confirm that the solution converges to a value as the mesh is refined.
2.1.5.1.
Influence of the time step (see folder DIN1_5_1)
The mesh shown in Figure 5 was used here, and values of the time step equal to 1s, 5s, 10s,
20s, 30s, and 60s were tested. In Figure 4 are displayed the temperature distributions for four of
these time steps, for the final time of 30min.
Figure 7 shows the evolution of the differences between the results from SAFIR and the ones
of Annex CC as a function of time, depending on the value of the time step considered in the
analysis.
a) 60s
b) 30s
c) 10s
d) 1s
Figure 6 – Temperatures determined by SAFIR for some of the time steps tested, for t = 1800s
(colour scale is the same as in Figure 5)
March 2018
9
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
1.6
8.0
1.2
6.0
+1 %
4.0
Deviation [K]
Deviation [%]
0.8
+5 K
0.4
0.0
60s
30s
20s
10s
5s
1s
-0.4
-0.8
-1 %
300
600
900
1200
1500
0.0
60s
30s
20s
10s
5s
1s
-2.0
-4.0
-5 K
-6.0
-1.2
0
2.0
1800
0
300
600
a) Percentage
900
1200
1500
1800
time [s]
time [s]
b) Degrees
Figure 7 – Differences between the results by SAFIR and the reference results for different time
steps
Different observations can be made on the last Figure:
- Apart from the first 300s, the deviation is systematically positive;
- After 600s, the deviation in terms of % increases linearly with time;
- The deviation in terms of K seems to remain more or less constant after 600s;
- Both criteria given in DIN are met as long as the time step does not exceed 25 s (final
time/72). If only the absolute difference in K is considered, a time step of 40 s (final
time/45) is acceptable.
2.1.5.2.
Influence of the number of nodes (see folder DIN1_5_2)
To assess the influence of the refinement of the mesh on the results, different meshes, with
still two elements on the horizontal direction but a varying number of elements on the direction
of the heat flow, are analysed here considering analyses with a time step = 1s.
In Figure 8 are shown some of the meshes that were used. The temperatures determined
after 30 min are plotted in Figure 9 for the meshes in Figure 8. The results for the deviations from
the DIN found for all the meshes tested are presented in Figure 10.
March 2018
10
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
e) 2 x 2 (9
nodes)
f) 2 x 8 (27
nodes)
g) 2 x 20 (63
nodes)
h) 2 x 100
(303 nodes)
Figure 8 – Some of the meshes used in the study regarding the influence of the number of nodes on
the results
a) 2 x 2 (9
nodes)
b) 2 x 8 (27
nodes)
c) 2 x 20 (63
nodes)
d) 2 x 100
(303 nodes)
Figure 9 – Temperatures determined by SAFIR for some of the meshes used to study the influence of
the number of nodes, for t = 1800s (colour scale is the same as in Figure 5)
14.0
1.6
9
12.0
15
10.0
27
153
303
4.0
63
0.8
2.0
0.0
Deviation [%]
+5 K
27
+1 %
93
6.0
15
1.2
63
8.0
Deviation [K]
9
93
153
0.4
303
0.0
-0.4
-2.0
-0.8
-4.0
-5 K
-6.0
-1 %
-1.2
0
300
600
900
1200
time [s]
a) Degrees
1500
1800
0
300
600
900
1200
1500
1800
time [s]
b) P
Figure 10 – Deviations of the results by SAFIR from the reference results for quadrilateral meshes
with different densities (expressed in number of nodes)
March 2018
11
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 10 shows that the models have converged when having more than 27 nodes (i.e. 7
nodes on the depth of the model), and that the results are within the limits defined in the DIN for
meshes with as few as 4 nodes on the depth (or 15 nodes in total), if a grid configuration with two
elements on the width is respected.
2.1.5.3.
Influence of the element type (see folder DIN1_5_3)
A study was done in order to understand how the utilisation of triangles can affect the
results. With that purpose, the crudest mesh from Figure 8 was taken as reference and triangle
meshes with identical number and distribution of nodes were tested. Again, the time step here
considered was for all cases equal to 1s.
The results for the temperatures at the top edge nodes of the cross-section are shown in
Figure 11 and Figure 12 for t = 1800s, for the quadrilateral mesh and the meshes with triangles,
respectively.
294.93
294.93
294.93
Figure 11 – Temperatures determined by SAFIR at the top 3 nodes for a quadrilateral mesh with
2x2 elements (9 nodes), for t = 1800s
299.53
295.04
290.59
a) Mesh #1
292.19
298.01
292.19
b) Mesh #2
298.01
292.19
298.01
c) Mesh #3
Figure 12 – Temperatures determined by SAFIR at the top 3 nodes for three different triangle
meshes with 9 nodes each for t = 1800s (colour scale is the same as in Figure 5)
It can be seen that for the three triangle meshes in Figure 12 the results depend on the
arrangement of the triangles within the mesh, and that for the same mesh the nodes at the top
edge show different values, unlike what happens with the quadrilateral mesh in Figure 11.
March 2018
12
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
With a doubly symmetric mesh like the one in Figure 13, the same temperatures are
obtained for the nodes with the same vertical position, as it is shown by the values plotted in that
Figure for the top edge.
294.39
294.39
294.39
Figure 13 – Temperatures determined by SAFIR at the top 3 nodes for a doubly-symmetric triangle
mesh with 13 nodes, for t = 1800s
Based on the latter, double symmetrical triangle meshes will be further compared to
similarly refined models based on quadrilateral elements. For example, the triangle mesh in
Figure 13 will be compared to the one presented in Figure 14.
295.04
295.04
295.04
Figure 14 – Temperatures determined by SAFIR at the top 3 nodes for a quadrilateral mesh with 15
nodes, for t = 1800s
Figure 15 and Figure 16 show all the meshes tested. The distribution of the temperatures in
the cross-sections are plotted in Figure 17 for t = 1800s, and the results for the temperature at the
studied point are plotted in Figure 18.
March 2018
13
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
a) 13 nodes
b) 23 nodes
c) 43 nodes
d) 83 nodes
Figure 15 – Triangle meshes used to study the influence of the type of element on the results
a) 15 nodes
b) 27 nodes
c) 51 nodes
d) 99 nodes
Figure 16 – Quadrilateral meshes used to study the influence of the type of element on the results
a) 13 nodes
b) 23 nodes
c) 43 nodes
d) 83 nodes
Figure 17 – Temperatures determined by SAFIR for triangle meshes, for t = 1800s (colour scale is
the same as in Figure 5)
March 2018
14
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
1.2
10.0
+1 %
8.0
+5 K
0.8
4.0
2.0
0.0
-4.0
-5 K
0.0
-0.4
300 s
300 s
1800 s
1800 s
-2.0
∆ - triangles
⧠ - quadrilaterals
0.4
Deviation [%]
Deviation [K]
6.0
∆ - triangles
⧠ - quadrilaterals
300 s
300 s
1800 s
1800 s
-0.8
-1 %
-1.2
-6.0
0
20
40
60
80
100
0
20
40
60
80
100
number of nodes
number of nodes
a) Degrees
b) Percentage
Figure 18 – Deviations of the results by SAFIR from the reference results for meshes with triangles
and quadrilateral elements (expressed in number of nodes)
By observing the plots in Figure 18 one can see that, for the two crudest meshes related to
each element type, the ones with quadrilaterals return the more accurate results and seem to
converge faster to the solution implemented in SAFIR. However, this should be at least partially
justified by the difference on the number of nodes between the models with quadrilaterals and
triangles. As for the other meshes tested, it seems that for meshes with more than 23 nodes a
convergence on the results is attained, regardless of the element type.
As for finding the crudest mesh able to return valid results according to the DIN, based on
the plots above a mesh formed with triangles with slightly more than 13 nodes seems to be
sufficiently refined for that purpose.
2.1.5.4.
Influence of distorted meshes (see folder DIN1_5_4)
In order to understand what is the impact of the distortion of elements in the mesh, the 4
meshes present in Figure 19 and their undistorted counterparts in Figure 20 were analysed,
considering a time step = 1s. The distribution of the temperatures in the cross-section is plotted
in Figure 21 for t = 1800s, and the deviations of the results from the DIN with both types of meshes
can be found in Figure 22.
a) 9 nodes
b) 25 nodes
c) 81 nodes
d) 289 nodes
Figure 19 – Distorted meshes used to study the influence of distortion of elements on the results
March 2018
15
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
a) 9 nodes
b) 25 nodes
c) 81 nodes
d) 289 nodes
Figure 20 – Undistorted meshes used to study the influence of distortion of elements on the results
e) 9 nodes
f) 25 nodes
g) 81 nodes
h) 289 nodes
Figure 21 – Temperatures determined by SAFIR for the distorted meshes, for t = 1800s (colour scale
is the same as in Figure 5)
1.6
9 (dist.)
9 (undist.)
1.2
25 (dist.)
+1 %
25 (undist.)
Deviation [%]
0.8
81 (dist.)
81 (undist.)
0.4
289 (dist.)
289 (undist.)
0.0
-0.4
-0.8
-1 %
-1.2
0
300
600
900
1200
1500
1800
time [s]
a) Degrees
b) Percentage
Figure 22 –Deviations of the results by SAFIR from the reference results for the distorted and
undistorted meshes (expressed in number of nodes)
By looking at Figure 22 it is seen that by using distorted meshes the results deviate only
slightly from the ones obtained with equivalent undistorted meshes.
March 2018
16
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.1.5.5.
Influence of unstructured meshes (see folder DIN1_5_5)
As a last step of this parametric analysis, the influence of using unstructured meshes is
investigated. The 6 unstructured meshes present in Figure 23 are analysed, again considering a
time step = 1s. The distribution of the temperatures in the cross-section is plotted in Figure 24 for
t = 1800s, and the deviations of the results from the DIN can be found in Figure 25.
a) 16 nodes
b) 33 nodes
d) 90 nodes
e) 117 nodes
c) 61 nodes
f)
141 nodes
Figure 23 – Meshes used to study the influence of unstructured meshes on the results
a) 16 nodes
b) 33 nodes
c) 61 nodes
d) 90 nodes
e) 117 nodes
f) 141 nodes
Figure 24 – Temperatures determined by SAFIR for the unstructured meshes, for t = 1800s (colour
scale is the same as in Figure 5)
March 2018
17
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
+1 %
1.0
Deviation [%]
0.6
0.2
16 (unstr.)
33 (unstr.)
61 (unstr.)
90 (unstr.)
117 (unstr.)
141 (unstr.)
-0.2
-0.6
-1 %
-1.0
-1.4
0
300
600
900
1200
1500
1800
time [s]
a) Degrees
b) Percentage
Figure 25 –Deviations of the results by SAFIR from the reference results for the unstructured
meshes (expressed in number of nodes)
One can observe from Figure 25 that:
- Overall, the results with unstructured meshes are well placed inside the stipulated
values;
- Crude unstructured meshes like the one with 16 nodes therein present can lead to some
deviations from the reference results (although relatively small);
- Unstructured meshes are less efficient and often require more nodes to attain the same
level of results than structured ones, as it is proved by the fact that the unstructured
mesh with 113 nodes in Figure 25 still presents some considerable deviations, whereas
in Figure 22 a mesh with just 81 nodes was able to attain very close results to the ones
found in the DIN;
- A very good agreement between the results from SAFIR and the DIN was achieved when
using an unstructured mesh with 141 nodes.
2.1.6
Conclusions
The parametric analysis shows that the solution of SAFIR satisfies the requirement of the
standard. The solution converges to the theoretical solution when the density of the mesh is
increased and the value of the time step is reduced.
When refining the mesh, rectangular elements converge slightly faster than triangular
element; regular structured meshes are most efficient, with slight differences being observed in
distorted structured meshes; unstructured meshes are somehow less efficient while being still in
the acceptable range of the standard.
March 2018
18
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.2. Example 2 – Heat transfer (heating) with varying properties
2.2.1
Keywords
Heat-transfer, conduction, convection, radiation, varying thermal properties
2.2.2
Objective
The goal of this example is to analyse the heat transfer by conduction when the thermal
conductivity varies with the temperature, as well as the heat exchange with a gas at the boundary
of the section by convection.
2.2.3
Description of the problem
A square section with the characteristics defined in Figure 26 and Table 3 is analysed. The
temperature in the cross-section is uniform and equal to 0°C at time t = 0, when it is exposed to a
gas with a temperature of 1000°C on all the edges. Heat transfer from the gas to the solid is by
linear convection and radiation. In order to validate the results, the temperature θ0 calculated at
the centre of the cross-section is compared to the reference values presented in DIN EN1992-1-2
NA at different times, for a total duration of 3 hours.
Figure 26 – Example 2: Heating up process in section with varying properties
March 2018
19
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 3 – Dimensions, material properties and boundary conditions for Example 2
Properties
Value
Thermal conductivity λ (linear behaviour)
W / (m∙K)
Specific heat cp
Density ρ
Dimensions h, b
Coefficient of convection αc
Emissivity εres = εm. εf
Ambient temperature Θu
Initial temperature in the cross-section Θcs
J / (kg∙K)
Kg / m3
mm
W / (m2∙K)
°C
°C
Θ
0
200
1000
λ (Θ)
1.5
0.7
0.5
1000
2400
200
10
0.8
1000
0
The data of this exercise are similar to that of a 20x20 cm² concrete section.
2.2.4
Model and results (see folder DIN2_4)
A model with a structured mesh formed by 576 square elements (24 by 24) was created.
Each of these elements contains 2 gauss points of integration in its plane, and the total number of
nodes is 625.
The initial temperature in the structure was set to 0°C, and a FRONTIER constraint of
1000°C was applied to all faces of the cross-section by the function “F1000”, as seen in Figure 27.
The PRECISION command was set to 1.0E-3. A USER material was applied according to the
properties described in Table 3. The time step chosen was 10 seconds (final time / 1080).
Figure 27 – Thermal model of the cross-section for Example 2 (24 x 24 SOLID elements)
Figure 28 shows the temperature distribution for the final target time of 180 min.
March 2018
20
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 28 – Temperatures determined by SAFIR for Example 2, for t = 180min
Table 4 gives the temperatures obtained by SAFIR and those given as reference by the DIN.
All the differences are within the boundaries stipulated by the DIN.
Table 4 – Temperatures Θ0 for Example 2
Time
t
min
30
60
Limit (t ≤ 60min)
90
120
150
180
Limit (t > 60min)
2.2.5
Reference
temperature
Θ0
°C
36.9
137.4
Calculated
temperature
Θ'0
°C
32.18
132.40
244.6
361.1
466.2
554.8
241.67
362.75
469.83
559.93
Deviation
(Θ'0 - Θ0)
K
-4.72
-5.00
±5.00
-2.93
1.65
3.63
5.13
(Θ'0 - Θ0) / Θ0 ∙ 100
%
-12.79
-3.64
-1.20
0.46
0.78
0.92
±3.00
Analysis of the influence of different parameters
In this sub-chapter, an analysis of the sensibility of the results to different input parameters
is done. This will provide some indications on the minimum value of the time step or minimum
number of nodes necessary in order to accurately simulate the heating process on the crosssection, as well as to confirm that the solution converges to a value as the mesh is refined.
March 2018
21
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.2.5.1.
Influence of the time step (see folder DIN2_5_1)
To study the influence of the time step on the results, the mesh shown in Figure 27 was used,
and values of the time step equal to 1, 5, 10, 20, 30, 60, 120, 300, 600 and 1200 s were tested.
Figure 29 shows the evolution of the differences between the results from SAFIR and the
ones of Annex CC as a function of time, depending on the value of the time step considered in the
analysis. The areas of the chart with a white background represent the domain where the criteria,
either in terms of percentage or in degrees, must be applied.
10.0
18.0
1800s
600s
300s
120s
60s
30s
20s
+5 K
10s
5s
1s
8.0
3.0
8.0
6.0
Deviation [%]
Deviation [K]
13.0
1s
5s
10s
20s
30s
60s
120s
300s
600s
1800s
4.0
2.0
0.0
+3 %
-2.0
-2.0
-3 %
-4.0
-5 K
-7.0
-6.0
-12.0
-8.0
0
30
60
90
120
time [min]
a) Degrees
150
180
0
30
60
90
120
150
180
time [min]
b) Percentage
Figure 29 – Differences between the results by SAFIR and the reference results for different time
steps
Different observations can be made on the last Figure:
- For the values of the time steps between 1s and 300s, inclusive, all the results fall within
the intervals, considering the zones of application of each criteria (percentage or Kelvin
degrees);
- When considering the absolute difference in Kelvin, considerable differences between
the different time steps are obtained for t = 30 min, where bigger time steps consistently
return bigger temperatures. For this time instant, the time step that returns a
temperature value identical to the reference seems to be somewhere between 120s and
300s. For t = 60 min these differences practically disappear.
- For the range of application of the limit in percentage, the results provided are valid for
time steps of less than 600s, inclusive, with bigger time steps consistently returning
lower temperatures, contrary to what happens for the first 60 min.
March 2018
22
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.2.5.2.
Influence of the number of nodes (see folder DIN2_5_2)
To assess the influence of the refinement of the mesh on the results, 6 different meshes are
analysed considering analyses with a time step = 10s. All meshes were defined as a grid with equal
number of elements in each direction, respectively 2x2, 4x4, 8x8, 16x16, 24x24 and 50x50.
Figure 30 shows part of the meshes that are used. The temperatures determined after 180
min are plotted in Figure 31 for those meshes. The results for the deviations from the DIN found
for all the meshes tested are presented in Figure 32.
a) 2 x 2 (9
nodes)
b) 4 x 4 (25
nodes)
c) 8 x 8 (81
nodes)
d) 16 x 16
(289 nodes)
Figure 30 – Meshes used to study the influence of unstructured meshes on the results
a) 2 x 2 (9
nodes)
b) 4 x 4 (25
nodes)
c) 8 x 8 (81
nodes)
d) 16 x 16
(289 nodes)
Figure 31 – Temperatures determined by SAFIR for some of the meshes used to study the influence
of the number of nodes, for t = 180min (colour scale is the same as in Figure 28)
March 2018
23
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
500
100
+5 K
0
-5 K
+3 %
0
-3 %
-200
Deviation [%]
Deviation [K]
-100
2601
625
289
81
25
9
-300
-400
-500
-500
-1000
2601
625
289
81
25
9
-1500
-600
-2000
-700
-2500
-800
0
30
60
90
120
150
0
180
30
60
90
120
150
180
time [min]
time [min]
a) Degrees (overview)
b) Percentage (overview)
5
10
+3 %
+5 K
5
0
Deviation [%]
Deviation [K]
0
-5
-5 K
-10
2601
625
289
81
25
9
-15
-20
-25
-3 %
-5
-10
2601
625
289
81
25
9
-15
-20
-30
0
30
60
90
120
150
time [min]
c) Degrees (detailed view)
180
0
30
60
90
120
150
180
time [min]
d) Percentage (detailed view)
Figure 32 – Differences between the results by SAFIR and the reference results for quadrilateral
meshes with different densities (expressed in number of nodes)
The following comments can be made on Figure 32:
- The crudest meshes analysed returned results that are hugely and negatively affected
by the skin effect – the use of a reduced number of consecutive linear functions as an
approximation to a parabolic solution;
- Considering the stipulated boundaries, meshes with at least 625 nodes are able to
provide valid results for the whole range of time steps analysed (this corresponds to
element thickness of 8.3 mm);
- The mesh with 289 nodes (12.5 mm thick elements) is able to provide good results for
the range covered by the limitation in terms of percentage, but not for the one in terms
of degrees.
March 2018
24
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
-
It seems therefore as if 10 mm is approximately the limit for the thickness of element
layers near the surface in concrete type sections.
2.2.5.3.
Influence of the element type (see folder DIN2_5_3)
The influence of the type of element is analysed by means of 4 different triangle meshes
compared with 4 quadrilateral meshes with a time step of 10 s.
Figure 33 and Figure 34 show the meshes tested. The distributions of the temperatures in
the triangle meshes are plotted in Figure 35.
a) 41 nodes
b) 85 nodes
c) 145 nodes
d) 313 nodes
Figure 33 – Triangle meshes used to study the influence of the type of element on the results
a) 81 nodes
b) 169 nodes
c) 289 nodes
d) 625 nodes
Figure 34 – Quadrilateral meshes used to study the influence of the type of element on the results
a) 41 nodes
b) 85 nodes
c) 145 nodes
d) 313 nodes
Figure 35 – Temperatures determined by SAFIR for the triangle meshes, for t = 180 min (colour
scale is the same as in Figure 28)
March 2018
25
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
20
5
+3 %
+5 K
0
0
-5 K
Deviation [%]
Deviation [K]
-20
-40
-60
30 min
30 min
60 min
60 min
∆ - triangles
⧠ - quadrilaterals
-80
-100
0
100
200
300
-3 %
-5
-10
∆ - triangles
⧠ - quadrilaterals
-15
-20
400
500
600
700
0
100
200
300
400
500
600
700
number of nodes
number of nodes
a) 30 and 60 min (in degrees)
180 min
180 min
120 min
120 min
150 min
150 min
90 min
90 min
b) 90, 120, 150 and 180 min (in
percentage)
Figure 36 – Deviations of the results by SAFIR from the reference results for meshes with triangles
and quadrilateral elements (expressed in number of nodes)
From Figure 40 it can be seen that the difference in the results between triangular elements
and square elements are not significant (marginally better results are obtained with square
elements).
2.2.5.4.
Influence of unstructured meshes (see folder DIN2_5_4 )
The influence of having unstructured meshes is investigated. The 4 meshes in Figure 37
were analysed, again considering a time step = 10s.
The distribution of the temperatures in the cross-section is plotted in Figure 38 for t = 180
min, and the deviations of the results from the DIN with the studied unstructured meshes are
found in Figure 39.
a) 72 nodes
b) 153 nodes
c) 355 nodes
d) 743 nodes
Figure 37 – Meshes used to study the influence of unstructured meshes on the results
March 2018
26
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
a) 72 nodes
b) 153 nodes
c) 355 nodes
d) 743 nodes
Figure 38 – Temperatures determined by SAFIR for the unstructured meshes, for t = 180 min
(colour scale is the same as in Figure 28)
10
10
+5 K
5
5
0
0
-3 %
-5
-5
-5 K
Deviation [%]
Deviation [K]
+3 %
-10
-15
-20
743
-25
355
-10
-15
-20
-25
743
-30
355
-35
-35
153
-40
-40
153
-30
72
72
0
30
60
90
120
time [min]
a) Degrees
150
180
0
30
60
90
120
150
180
time [min]
b) Percentage
Figure 39 – Differences between the results by SAFIR and the reference results, for quadrilateral
meshes with different densities (expressed in number of nodes)
The analysis of the results in Figure 39 allows to conclude that having an unstructured mesh
doesn’t greatly affect the results obtained, as a mesh with 743 nodes presents very identical
results to the ones obtained with a structured mesh with 625 nodes (see Figure 33), which fall
inside the DIN limits for all the range of time instants studied.
2.2.6
Conclusions
Like in Example 1, the parametric analysis has shown that the solution satisfies the
requirement of the standard, provided that the mesh used is sufficiently refined.
It was noticed that meshes with quadrilateral elements converged slightly but not
significantly faster than triangular elements, while no significant differences appeared between
unstructured and structured meshes.
March 2018
27
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.3. Example 3 – Heat transfer through several layers
2.3.1
Keywords
Heat-transfer, conduction, convection, radiation, various material layers
2.3.2
Objective
The goal of this example is to analyse the heat transfer in a steel hollow section that is filled
with a lightweight insulating material. The thermal conductive properties of steel are those of EN
1993-1-2 whereas the filling material has constant thermal properties.
2.3.3
Description of the problem
A square section with the characteristics defined in Figure 40 and Table 5 is analysed. The
temperature in the cross-section is uniform and equal to 0°C at time t = 0, when it is exposed to a
gas with a temperature of 1000°C on all the edges. Heat transfer from the gas to the solid is by
linear convection and by radiation. In order to validate the results, the temperature θ0 calculated
at the centre of the cross-section is compared to the reference values presented in DIN EN19921-2 NA, for a duration of 3 hours.
Legend:
1 Steel
2 Filling
Figure 40 – Example 3: Heat-transfer through several layers
Table 5 – Dimensions, material properties and boundary conditions for Example 3
Properties
Thermal conductivity λ
Specific heat cp
Density ρ
Dimensions h, b, t
Coefficient of convection αc
Emissivity εres = εm. εf
Initial temperature in the cross-section Θcs
Ambient temperature Θu
March 2018
W / (m∙K)
J / (kg∙K)
Kg / m3
mm
W / (m2∙K)
°C
°C
Steel
Filling
EN 1993-1-2
0.05
EN 1993-1-2
1000
EN 1993-1-2
50
h = b = 201, t = 0.5
10
0.8
0
0
1000
28
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.3.4
Model and results (see folder DIN3_4)
A model with a structured mesh formed by 672 quadrilateral elements was created, which
consists on a grid of 24 x 24 elements for the filling material, with the steel hollow section being
defined with also 24 elements along each edge of the section and 1 element across the thickness.
Each of these elements contains 2 Gauss points of integration, and the total number of nodes is
721.
The initial temperature in the structure was set to 0°C, and a frontier constraint of 1000°C
was applied to all the faces of the cross-section according to Figure 41.
The PRECISION command was set to 1.0E-3. An INSULATION material with constant
material properties according to Table 5 was used for the filling, and the STEELEC3EN material
was used for the hollow section, by changing the heat transfer coefficients at the surface to match
the values in Table 5. The time step chosen was 10 s (final time / 1080).
a) Overall view
b) Detail of the two layers
Figure 41 – Thermal model of the cross-section for Example 3 (24 x 24 SOLID elements)
March 2018
29
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 42 shows the temperature distribution for the final time of 180min.
Figure 42 – Temperatures determined by SAFIR for Example 3, for t = 180min.
Table 6 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 6 – Temperature θ0 for Example 3
Time
t
min
30
60
90
120
150
180
Limits
2.3.5
Reference
temperature
Θ0
°C
340.5
717.1
881.6
950.6
979.3
991.7
Calculated
temperature
Θ'0
°C
338.89
721.97
885.24
952.66
980.47
991.95
Deviation
(Θ'0 - Θ0)
K
-1.61
4.87
3.64
2.06
1.17
0.25
±5.00
(Θ'0 - Θ0) / Θ0 ∙ 100
%
-0.47
0.68
0.41
0.22
0.12
0.03
±1.00
Conclusions
The deviations found were within the limits stipulated by the DIN, both in absolute and in
relative values, for the whole duration of the simulation.
March 2018
30
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.4. Example 4 – Thermal expansion
2.4.1
Keywords
Thermal expansion, steel, homogenous temperature
2.4.2
Objective
The goal of this example is to analyse the thermal expansion of a steel member for different
values of homogeneous temperature in a cross-section.
2.4.3
Description of the problem
A squared section with the characteristics defined in Figure 43 and Table 7 is analysed. In
order to validate the results, the thermal elongation Δl at the simply supported end of the member
is determined and compared to the reference values presented in DIN EN1992-1-2 NA.
Figure 43 – Example 4: Thermal expansion in statically assembled member
Table 7 – Dimensions, material properties and boundary conditions for Example 4
Properties
Dimensions l, h, b
Stress-strain material law
Yield strength fyk
Initial conditions
Homogeneous temperature in the member Θu
Thermal expansion
2.4.4
mm
N /mm2
°C
°C
Steel
100
EN 1993-1-2
355
20
100, 300, 500, 600, 700, 900
EN 1993-1-2
Model and results (see folder DIN4_4)
In this example the calculations were performed using BEAM 2D, BEAM 3D, TRUSS, SHELL
and SOLID 3D elements, in order to get those different formulations validated.
No thermal analysis is presented here. The prescribed temperatures have been directly
introduced in the relevant files for the structural analysis to be performed.
March 2018
31
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2D BEAM element
A 2D mechanical model with 1 BEAM element that is free to expand was created and
exposed, unloaded, to the prescribed temperature fields.
In Figure 44 a representation of the thermal elongation determined by SAFIR is shown (in
the Figure the elongation is causing the right support to move horizontally).
Figure 44 – Thermal elongation determined by SAFIR for Example4, for t = 900°C, using a 2D BEAM
element
Table 8 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 8 – Thermal elongation Δl for Example 4, using a 2D BEAM element
Temperature
Reference
extension
Calculated
extension
Θ
Δl
Δl’
°C
100
300
Limit (Θ ≤ 300°C)
500
600
700
900
Limit (Θ > 300°C)
mm
0.09984
0.37184
mm
0.0998
0.372
0.67584
0.83984
1.01184
1.18000
0.676
0.84
1.01
1.18
Deviation
(Δl’ - Δl)
mm
-0.00004
0.00016
±0.05
0.00016
0.00016
-0.00184
0.00000
(Δl’ - Δl) / Δl ∙
100
%
-0.04
0.04
0.02
0.02
-0.18
0.00
±1.00
3D BEAM element
A 3D mechanical model with 1 BEAM element that is free to expand was created and
exposed, unloaded, to the prescribed temperature fields.
In Figure 45 a representation of the thermal elongation determined by SAFIR is shown (in
the Figure the elongation is causing the support on the left to move according to the X-axis).
March 2018
32
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 45 – Thermal elongation determined by SAFIR for Example 4, for t = 900°C, using a 3D
BEAM element
Table 10 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 9 – Thermal elongation Δl for Example 4, using a 3D BEAM element
Temperature
Reference
extension
Calculated
extension
Θ
Δl
Δl’
°C
100
300
Limit (Θ ≤ 300°C)
500
600
700
900
Limit (Θ > 300°C)
March 2018
mm
0.09984
0.37184
mm
0.0998
0.372
0.67584
0.83984
1.01184
1.18000
0.676
0.84
1.01
1.18
Deviation
(Δl’ - Δl)
mm
-0.00004
0.00016
±0.05
0.00016
0.00016
-0.00184
0.00000
(Δl’ - Δl) / Δl ∙
100
%
-0.04
0.04
0.02
0.02
-0.18
0.00
±1.00
33
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
TRUSS element
A 2D mechanical model with 1 TRUSS element that is free to expand was created and
exposed, unloaded, to the prescribed temperature fields.
In Figure 46 a representation of the thermal elongation determined by SAFIR is shown (in
the Figure the elongation is causing the right support to move horizontally).
Figure 46 – Thermal elongation determined by SAFIR for Example 4, for t = 900°C, using a TRUSS
element
Table 10 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 10 – Thermal elongation Δl for Example 4, using a BEAM element
Temperature
Reference
extension
Calculated
extension
Θ
Δl
Δl’
°C
100
300
Limit (Θ ≤ 300°C)
500
600
700
900
Limit (Θ > 300°C)
mm
0.09984
0.37184
mm
0.0998
0.371
0.67584
0.83984
1.01184
1.18000
0.674
0.836
1.01
1.17
Deviation
(Δl’ - Δl)
mm
-0.00004
-0.00084
±0.05
-0.000184
-0.000384
-0.00184
-0.01000
(Δl’ - Δl) / Δl ∙
100
%
-0.04
-0.23
-0.27
-0.46
-0.18
-0.85
±1.00
SHELL element
A 3D mechanical model with 9 SHELL elements (3x3) that is free to expand was created and
exposed, unloaded, to the prescribed temperature fields.
The visualization in DIAMOND of the displacements obtained by SAFIR in one of the
directions is shown in Figure 47 (the same results were obtained for the expansion in both
directions).
March 2018
34
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[m]
Figure 47 – Thermal elongation in Y-direction determined by SAFIR for Example 4, for t = 900°C,
using SHELL elements
Table 11 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 11 – Thermal elongation Δl for Example 4, using SHELL elements
Temperature
Reference
extension
Calculated
extension
Θ
Δl
Δl’
°C
100
300
Limit (Θ ≤ 300°C)
500
600
700
900
Limit (Θ > 300°C)
mm
0.09984
0.37184
mm
0.0998
0.371
0.67584
0.83984
1.01184
1.18000
0.674
0.836
1.01
1.17
Deviation
(Δl’ - Δl)
Mm
-0.00004
-0.00084
±0.05
-0.000184
-0.000384
-0.00184
-0.01000
(Δl’ - Δl) / Δl ∙
100
%
-0.04
-0.23
-0.27
-0.46
-0.18
-0.85
±1.00
SOLID elements
A 3D mechanical model with 1 SOLID element that is free to expand was created and
exposed, unloaded, to the prescribed temperature fields.
The visualization in DIAMOND of the displacements obtained by SAFIR in one of the
directions is shown in Figure 48 (the same results were obtained for the expansion in all 3
directions).
March 2018
35
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[m]
Figure 48 – Thermal elongation in Y-direction determined by SAFIR for Example 4, for t = 900°C,
using a SOLID element
Table 12 shows the deviations of the temperatures obtained by SAFIR when compared to
those given as reference by the DIN.
Table 12 – Thermal elongation Δl for Example 4, using a SOLID element
Temperature
Reference
expansion
Calculated
expansion
Θ
Δl
Δl’
°C
100
300
Limit (Θ ≤ 300°C)
500
600
700
900
Limit (Θ > 300°C)
2.4.5
mm
0.00984
0.37184
mm
0.0998
0.372
0.67584
0.83984
1.01184
1.18000
0.676
0.84
1.01
1.18
Deviation
(Δl’ - Δl)
mm
-0.00004
0.00016
±0.05
0.00016
0.00016
-0.00184
0.00000
(Δl’ - Δl) / Δl ∙
100
%
-0.04
0.04
0.02
0.02
-0.18
0.00
±1.00
Conclusions
The results calculated for all types of elements are within the limits stipulated by the DIN.
It may appear inconsistent that, in one single software, the thermal expansion is not exactly
the same for all types of elements. Indeed, for 900°C, for example, the 2D BEAM, the 3D BEAM and
the SOLID finite elements yield an expansion of 1.18 mm, which is exactly the expected value,
whereas the TRUSS and the SHELL finite elements yield a value of 1.17 mm.
March 2018
36
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The reason lies in the fact that the axial strain is calculated based on a linearized
expression in the SOLID elements (because they are written in a small displacements formulation)
and in the BEAM finite elements (because these are expected to deflect essentially in bending
rather than in elongation).
𝑙 − 𝑙𝑖
𝜀=
𝑙𝑖
On the contrary, for the TRUSS elements where elongation is the sole deformation and in
SHELL elements that can also work as membrane elements, the more correct nonlinear expression
is being used.
𝑙² − 𝑙𝑖 ²
𝜀=
2 𝑙𝑖 ²
It can be verified that, with an elongation of 1.17 mm from an initial length of 100 mm, the
second expression yields a strain equal to 1.1768 %. If, by multiplying the length of the specimen
by a factor of 10, one gains access to an additional significant digit, the elongation for the BEAM
and the SOLID elements is found as 1.173 mm, which yields a strain of 1.1799 %.
The results are thus consistent with the fact that thermal strain of steel in SAFIR is
calculated in the same manner for all element types, in full accordance with the recommendation
of EN 1993-1-2.
2.5. Example 5 – Temperature-dependent stress-strain curves of
concrete and steel
2.5.1
Keywords
Elongation, column, steel, concrete, homogenous temperature
2.5.2
Objective
The goal of this example is to analyse the total elongation of a column made of steel or
concrete for different uniform temperature distributions at some given stress-strength ratios.
2.5.3
Description of the problem
Members with the characteristics defined in Figure 49 and Table 13 are analysed. In order
to validate the results, the elongation Δl at the unsupported end of the members is determined
and compared to the reference values presented in DIN EN1992-1-2 NA.
March 2018
37
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 49 – Example 5: Temperature-dependent stress-strain in column (stability failure is
excluded)
Table 13 – Dimensions, material properties and boundary conditions for Example 5 and Example 6
a
Properties
Dimensions l, h, b
Stress-strain material law
Yield strength fy (20°C) , fc (20°C)
Thermal expansion
Initial conditions
Homogeneous temperature in the
member Θ
Load σS (Θ) / fy (Θ) and σC (Θ) / fc (Θ)
mm
N/mm2
Steel
100, 10, 10
EN 1993-1-2
355
EN 1993-1-2
Concrete a
100, 31.6, 31.6
EN 1992-1-2
20
EN 1992-1-2
°C
20
°C
20, 200, 400, 600, 800
0.2, 0.6, 0.9
– with predominantly siliceous aggregates
2.5.4
Models and results (see folder DIN5_4)
A 2D mechanical model with 1 BEAM element was created and replicated in order to be
associated to the 15 different combinations of the temperatures with the stress / strength ratios.
Static analyses with a PRECISION of 10E-3s, a COMEBACK of 1s and a time step of 60s were
performed for each case.
This was done for both steel and concrete using the materials STEELEC3EN and
SILCONC_EN and the properties in Table 13. The results obtained are shown, respectively, in Table
14 and Table 15.
March 2018
38
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 14 – Elongation Δl for the steel column in Example 5
Temp.
Θ
°C
20
200
400
600
800
Limit
March 2018
Stress /
Strength
σSΘ /
fyk(Θ)
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
Red.
Coeff.
ky,θ =
fy,θ/fy
1
1
1
1
1
1
1
1
1
0.47
0.47
0.47
0.11
0.11
0.11
Applied
force
Ref. value
Calc.
value
FcΘ
Δl
Δl’
N
-7100
-21300
-31950
-7100
-21300
-31950
-7100
-21300
-31950
-3337
-10011
-15017
-781
-2343
-3515
mm
-0.034
-0.101
-0.152
0.194
0.119
-0.159
0.472
0.293
-0.451
0.789
0.581
-0.162
1.059
0.914
0.170
mm
-0.0338
-0.101
-0.152
0.194
0.119
-0.159
0.472
0.293
-0.451
0.789
0.581
-0.162
1.06
0.915
0.171
Deviation
(Δl’ - Δl)
mm
0.0002
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0010
0.0010
0.0010
(Δl’ - Δl)
/ Δl ∙ 100
%
-0.59
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.09
0.11
0.59
± 3.00
39
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 15 – Elongation Δl for the concrete column in Example 5
Temp.
Θ
°C
20
200
400
600
800
Stress /
Strength
σSΘ /
fyk(Θ)
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
0.2
0.6
0.9
Red.
Coeff.
Applied
force
Ref. value
Calc.
value
fc,θ/fck
FcΘ
Δl
+Δl’
1
1
1
0.95
0.95
0.95
0.75
0.75
0.75
0.45
0.45
0.45
0.15
0.15
0.15
N
-3994
-11983
-17974
-3795
-11384
-17075
-2996
-8987
-13481
-1797
-5392
-8088
-599
-1797
-2696
mm
-0.0334
-0.104
-0.176
0.107
-0.0474
-0.2075
0.356
0.075
-0.216
0.685
-0.0167
-0.744
1.066
0.365
-0.363
mm
-0.0334
-0.104
-0.176
0.107
-0.0474
-0.207
0.356
0.075
-0.216
0.685
-0.0167
-0.744
1.07
0.365
-0.363
Deviation
(Δl’ - Δl)
mm
0.0000
0.0000
0.0000
0.0000
0.0000
0.0005
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0040
0.0000
0.0000
Limit
2.5.5
(Δl’ - Δl)
/ Δl ∙ 100
%
0.00
0.00
0.00
0.00
0.00
-0.24
0.00
0.00
0.00
0.00
0.00
0.00
0.38
0.00
0.00
± 3.00
Conclusions
For both steel and concrete, the major part of the cases returned exactly the same value as
the reference. Whenever this was not the case, the differences found were minor and well within
the boundaries stipulated by the DIN, and those were related to the number of significant digits
produced by SAFIR in .xml file and displayed in DIAMOND.
Calculations were also performed using the TRUSS and the BEAM 3D elements, for the cases
with 600°C and a stress / strength ratio of 0.6, which returned the same (or virtually the same)
results as for the BEAM 2D cases, for both steel and concrete.
2.6. Example 6 – Temperature-dependent limit-load-bearing capacity
of concrete and steel
2.6.1
Keywords
Ultimate bearing capacity, column, steel, concrete, homogeneous temperature
2.6.2
Objective
The goal of this example is to determine the load bearing capacity at different homogeneous
temperatures for the concrete and steel columns of Example 5.
March 2018
40
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.6.3
Description of the problem
Members with the characteristics defined in Figure 49 and Table 13 are analysed. In order
to validate the results, the load bearing capacity of the members is determined and compared to
the reference values presented in DIN EN1992-1-2 NA.
2.6.4
Models and results (see folder DIN6_4)
The same models of the previous example were used, but the tem files were modified and a
new custom load function named “fload.fc” was introduced, in order to allow for a phase where
the cross-section is heated without load for 512s until it reaches the target temperature, and
another phase where the load is increased at constant temperature until the failure of the column.
One single beam element was used, which does not allow buckling of the columns to
develop. This decision is motivated by the low slenderness of the columns that are proposed,
which seems to indicate that this example has been designed with validation of the material laws
in mind, more than the validation of a structural behaviour. The good match that will be observed
between the calculated results and the reference results seems to confirm that this was indeed
the case.
The results for the differences in the ultimate resistances obtained in static analyses
performed by SAFIR and the reference results are presented in Table 16 and Table 17.
Table 16 – Ultimate resistance of the steel member in Example 6
Temperature
Reference force
Calculated force
Θ
NRk.fi
N’Rk.fi
°C
20
200
400
600
800
Limits
kN
-35.5
-35.5
-35.5
-16.7
-3.9
kN
-35.5
-35.5
-35.5
-16.7
-3.9
Deviation
(N'Rk,fi,k - NRk,fi,k)
(N'Rk,fi,k - NRk,fi,k)
/ NRk,fi,k ∙ 100
kN
%
0.0
0.00
0.0
0.00
0.0
0.00
0.0
0.00
0.0
0.00
± 0.5
± 3.00
Table 17 – Ultimate resistance of the concrete member in Example 6
Temperature
Reference force
Calculated force
Θ
NRk.fi
N’Rk.fi
°C
20
200
400
600
800
Limits
March 2018
kN
-20
-19
-15
-9
-3
kN
-20.0
-19.0
-15.0
-9.0
-3.0
Deviation
(N'Rk,fi,k - NRk,fi,k)
(N'Rk,fi,k - NRk,fi,k)
/ NRk,fi,k ∙ 100
kN
%
0.0
0.00
0.0
0.00
0.0
0.00
0.0
0.00
0.0
0.00
± 0.5
± 3.00
41
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
2.6.5
Conclusions
A perfect agreement between the expected results and the ones obtained by SAFIR was
found for all the cases considered, using the 2D BEAM elements.
Calculations were also performed using the TRUSS and the BEAM 3D elements, for the cases
with 600°C, which returned the same (or virtually the same) results as for the BEAM 2D cases, for
both steel and concrete.
2.7. Example 7 – Development of restraint stresses
2.7.1
Keywords
Thermal induced forces, steel, homogenous temperature, varying temperature
2.7.2
Objective
The goal of this example is to determine the thermal induced internal forces and stresses
for a fixed-fixed steel beam, for two different temperature distributions: one constant and one that
varies linearly on the depth of the cross-section.
2.7.3
Description of the problem
The members with the characteristics defined in Figure 50 and Table 18 are analysed. In
order to validate the results, the values of the internal axial and bending forces and the stresses at
the bottom of the cross-section are determined for the two temperature distributions and
compared to the reference values presented in DIN EN1992-1-2 NA.
Figure 50 – Example 7: Fixed-fixed member and temperature distributions in the cross-section
March 2018
42
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 18 – Dimensions, material properties and boundary conditions for Example 7
Properties
Dimensions l, h, b
Stress-strain material law
Yield strength fy
Elastic modulus Ea (20°C)
Thermal expansion
Θo
Temperature of the
member
Θu
Steel
1000, 100, 100
EN 1993-1-2
650 a
210 000
EN 1993-1-2
mm
N /mm^2
N / mm^2
°C
°C
120
120
20
220
a - Structural steel according to DIN EN 1993-1-1 using the notional yield fy (20 ° C) = 650 N / mm2 (no highstrength steel) and the thermo-mechanical properties according to DIN EN 1993-1-2
2.7.4
Models and results (see folder DIN7_4)
Two .tem files were created with 20 fibres of 5 mm depth equally distributed on the depth
of the section. The temperature of 120°C was assigned to all fibres in the first file. In the second
file, the temperature varied linearly from 25°C in fibre 1 to 215°C in Fibre 20.
Figure 51 – Cross-section used for Example 7 (20 fibres)
A 2D mechanical model with 1 BEAM element was created and duplicated in order to be
associated to the 2 different temperature distributions. STATIC analyses using a PURE_NR (pure
Newton-Raphson) procedure, with a PRECISION of 10-3s and a time step of 1s were performed
with the STEELEC3EN material. The results obtained are shown in Table 19.
March 2018
43
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 19 – Internal forces and stresses due to thermal expansion for Example 7
Temp.
Θ0 | Θu
Parameter
Desig.
Units
Reference
X
°C
120 | 120
20 | 220
2.7.5
Nzw
Mzw
σzw
Nzw
Mzw
σzw
kN
kN.m
N / mm2
kN
kN.m
N / mm2
-2585
0
-258.5
-2511
-40.3
-479
Calculated
X’
-2584.8
0
-258
-2510.6
-40.207
-469
Deviation
(X’ – X) / X
∙ 100
%
-0.01
0.00
-0.19
-0.02
-0.23
-2.09
Limits
%
± 1.00
± 1.00
± 5.00
± 1.00
± 1.00
± 5.00
Conclusions
A good agreement between the reference results and the ones obtained by SAFIR was found
for the results for both the cases.
It has to be mentioned that the stresses were determined for the fibres at the bottom of the
cross-section and hence their value do not correspond precisely to the value at the bottom edge,
but to the centre of the fibre. This is the cause of the difference of 2% found for the 20|220 case.
The difference for this case will be reduced if a thinner fibre is considered at the edge of the
section. For instance, if a mesh with 200 fibres of equal size is considered, the stress determined
by SAFIR at the bottom fibre is 478 N/mm2, hence much closer to the 479 N/mm2 presented in
DIN.
2.8. Example 8 – Weakly reinforced concrete beam
2.8.1
Keywords
Fire resistance time, weakly reinforced concrete beam, ISO fire curve
2.8.2
Objective
Example 8 deals with a weakly reinforced concrete beam with a uniform distributed load
that is subjected to the standard fire on three sides. The goal is to determine the requested area
for the two steel rebars in order to provide a fire resistance of 90 minutes to the beam.
2.8.3
Description of the problem
The member with the characteristics defined in Figure 52 and Table 20 is analysed. The
required value for the steel area to ensure that the beam reaches 90 min of resistance when
subjected to an ISO fire is determined and compared to the reference value presented in DIN
EN1992-1-2 NA.
March 2018
44
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[cm]
Figure 52 – Example 8: Cross-section and configuration of the weekly reinforced concrete beam
Table 20 – Dimensions, material properties and boundary conditions for Example 8
a
b
Fire resistance category
Dimensions
l, b , h
mm
Spacing
a, as
mm
Loads
qE,fi,d,t
kN/m
Concrete C20/25
fc (20°C)
N/mm2
Moisture content (by mass)
w
%
Steel B500
fy (20°C)
N/mm2
a
Concrete
Stress-strain material law
Rebars b
Fire exposure
ISO 834 (three sides)
Coefficient of convection
αc
W/(m2.K)
Emissivity
εm
Concrete
λ, ρ, cp, εth,c
Thermal and physical
material properties
Steel
λa, ρ, ca, εth,s
R90
3000, 200, 380
45, 55
29
20
3
500
EN 1992-1-2
EN 1991-1-2
25
0.7
EN 1992-1-2
EN 1994-1-2
– with predominantly siliceous aggregates and density ρ = 2400 kg/m3
– class N, hot-rolled
2.8.4
Models and results (see folder DIN8_4)
Different models of the section were created, by changing the area of the two steel bars from
one model to another (see Figure 53). The ISO fire curve was applied on 3 sides of the section
while the upper side was adiabatic.
March 2018
45
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
a) Overall view
b) Detail of one of the rebars
Figure 53 – Example of one of the thermal models of the cross-section tested for Example 8
The whole length of the beam was modelled with 20 2D BEAM finite elements of equal
length (see Figure 54).
Figure 54 – Structural model of the beam (20 BEAM elements of equal sizes)
Steel was modelled by the material STEELEC2EN, with the fabrication process HOTROLLED
and with the ductility class B (no mention to the ductility class is given in DIN but this was found
to be the one resulting in better approximations to the results in DIN).
Concrete was modelled by the material SILCONC_EN. In thermal analysis, the upper limit for
the thermal conductivity, found in clause 3.3.3 of EN1992-1-2, was considered, since this is this
one recommended by Eurocode and that led to better approximations to the results in DIN.
The calculations performed were DYNAMIC and used the PURE_NR (pure Newton-Raphson)
procedure. The initial time step was set to 1s, the maximum time step to 60s, the PRECISION to
10E-3s, and the COMEBACK to 0.01s.
The fire resistance times obtained in the mechanical analyses, corresponding to the last
converged time step, were observed for each section used. It was found that a diameter for each
of the two rebars of 14.47 mm is required to obtain a fire resistance of 90 minutes (with a margin
of error of 0.5 min), which corresponds to a total steel reinforcement area of 3.29 cm².
March 2018
46
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The temperature in the rebars and the fire resistance times obtained for the different
iterations made are shown in Figure 55.
DIN
DIN
95
SAFIR
SAFIR
94
560
93
555
92
550
91
90 min
90
545
89
530
3.56 cm2
-10 % As
3.56 cm2
3.29 cm2
535
3.29 cm2
540
-10 % As
Temperature in rebars [°C]
565
88
Fire resistance time [min.]
96
570
87
86
3
3.1
3.2
3.3
3.4
3.5
Area of steel [cm2]
3.6
3.7 3
3.1
3.2
3.3
3.4
3.5
3.6
3.7
Area of steel [cm2]
Figure 55 – Evolution of the temperature at the centre of the rebars and of the fire resistance with
the steel area
The temperature distribution after 90 minutes, in the section with A’s = 3.29 cm2, is shown
in Figure 56.
Figure 56 – Temperatures determined by SAFIR for t = 90 min, for Example 8, with two rebars of
14.47mm (bottom part of the model)
Figure 57 shows the evolution of the vertical displacement at mid-span of the beam. The
vertical displacement at the last converged time is 106 mm, corresponding to l/28.3. The
horizontal inward displacement on the right hand support is 28.1 mm for the same time instant.
March 2018
47
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 57 – Vertical displacement vs. time at mid-span of the beam obtained by SAFIR
The results obtained with SAFIR, as well as the reference values of DIN EN1992-1-2 NA, are
summarized in Table 21.
Comments on the results obtained will be given at the last part of this example.
Table 21 – Necessary area of steel to resist an ISO fire for 90 min, for the weekly reinforced concrete
beam in Example 8
Temperature in the
rebars for t = 90 min
Fire
resistance
class
R90
2.8.5
Area of the steel rebars
Reference
Calculated
Reference
Calculated
Deviation
T
T’
As
A’s
(A’s – As) /
As.100
%
-7.61
°C
562
°C
544
cm2
3.56
cm2
3.29
Limit
%
± 10.00
Analysis of the influence of different parameters in the structural analysis
An analysis of the sensibility of the results to different parameters is done, which will
provide indications on the influence of the type of analysis, on the minimum value of the time step
or the minimum number of BEAM elements necessary to accurately simulate the mechanical
behaviour of the beam, and to confirm that the solution converges to a value as the mesh is refined.
The thermal model with the area of steel of 3.29 cm2 mentioned above was used in all the
analyses.
2.8.5.1.
Influence of the type of analysis (see folder DIN8_5_1)
To study the influence of the type of analysis (static or dynamic) and the type of procedure
(pure Newton-Raphson or approximated Newton-Raphson), the cases presented in Table 22 were
tested. The same mechanical model with 20 BEAM elements was used. Both procedures were
March 2018
48
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
tested in a STATIC and in a DYNAMIC analysis. For the DYNAMIC analyses, three values for the
mass were used: 0 kg/m; 190 kg/m (the mass of the beam); and 2090 kg/m (the mass of the beam
plus the mass of the gravity load).
The results obtained allow to conclude that, for the current example, the type of analysis
and type of procedure chosen don’t affect the results, regardless of the value given for the MASS
(in DYNAMIC analyses).
Table 22 – Fire resistance times obtained with different analysis options, for Example 8
Case
Name
DIN8_5_1( )
a
b
c
d*
e
f
g
h
Type of analysis
MASS
Fire resistance time
Type
Procedure
kg/m
min
STATIC
PURE_NR
APPR_NR
0
190
2090
0
190
2090
90.5
90.5
90.5
90.5
90.5
90.5
90.5
90.5
PURE_NR
DYNAMIC
APPR_NR
*same case as the one tested before in 2.8.4
2.8.5.2.
Influence of the time step (see folder DIN8_5_2)
To study the influence of the time step on the results, the same model was used, and values
of the time step equal to 5, 10, 30, 60, 120, 300 and 600 s were tested. Two types of analysis, one
STATIC and one DYNAMIC with the MASS of 190 kg/m, were performed. The results are shown in
Table 23.
Table 23 – Fire resistance times obtained with different time steps, for the two types of analysis, for
Example 8
Case
DIN8_5_2( )
a
b
c
d
e
f
g
h
i
March 2018
STATIC analysis
Fire resistance
Time step
time
s
min
5
90.5
10
90.5
30
90.5
60
90.5
120
90.5
300
90.5
450
90.5
600
90.5
1200
90.5
Initial time
step
s
5
5
5
5
5
30
60
300
600
DYNAMIC analysis
Maximum
Fire resistance
time step
time
s
min
30
90.5
60
90.5
120
90.5
300
90.5
600
90.5
600
90.5
600
90.5
600
90.5
600
90.5
49
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The change in the value of the time step in the mechanical analysis didn’t produce any effects
in this simple structure. This may not be the case in more complex structures.
2.8.5.3.
Influence of the number of elements (see folder DIN8_5_3)
To assess the influence of the refinement of the mesh of the mechanical model on the results,
7 different meshes - with 2, 4, 6, 8, 12, 16, and 20 BEAM elements of equal sizes - were analysed,
considering the assumptions in 2.8.4.
The results for the fire resistance time obtained with each model are presented in Figure
58. It is seen that results converge to the value of 90.5 minutes as the mesh is refined.
94
Fire resistance time [min]
93
92
91
90.5
90
89
88
87
Influence of the
number of elements
-1
86
85
0
5
10
15
20
number of 2D BEAM elements
Figure 58 – Fire resistance times obtained with different meshes for the mechanical model, for
Example 8
It has yet to be mentioned that the bending moment distribution is such that the whole
central part of the beam is subjected to a bending moment that is very close to the maximum
bending moment. The convergence of the solution as a function of the number of elements might
be slower in case of peaks in the bending moment distribution (with a point load, for example).
2.8.6
Discussion and conclusions
The temperature determined by SAFIR in the rebars is 544°C and hence a little bit lower
(around 3%) than the 562°C mentioned in DIN. This is not related to the difference in the size of
the rebars, as it is shown in Figure 55 that for the range of steel reinforcement areas analysed,
which covers the area of 3.56 cm2 preconized in DIN, the temperature does not vary significantly.
Nevertheless, a deviation of 7.61% for the necessary area is still well inside the boundaries
provided by the DIN, and it can be at least partly justified by the differences found for the
temperature on the rebars. In fact, if the model with the reference area of 3.56 cm2 is considered,
the structural analysis in SAFIR will fail to converge at the time instant t = 94.2min, time that
corresponds approximately to having the reference temperature of 562°C in the rebars, as can be
seen in Figure 59.
March 2018
50
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
570
90 min
562°C
∆t = 4.2 min
94.2 min
560
555
∆T = 18°C
550
545
544°C
DIN
540
SAFIR inner
SAFIR centre
535
SAFIR outer
530
88
90
92
Last converged time step in SAFIR
Temperature in rebars [°C]
565
94
96
Time [min]
Figure 59 – Evolution of the temperature at the centre, innermost and outermost points of the
rebars, for the SAFIR model with the reference area As = 3.56 cm2
One other aspect that should be brought to discussion is the lack of definition in DIN
regarding the criteria for the moment where the beam fails. All the results above were presented
considering that the time of failure is the last converged time step in SAFIR. However, according
to [9], the last 4 examples (Examples 8 to 11), which simulate fire tests, are based on results from
approved numerical tools. There is no indication in DIN about the criteria used to determine the
time of failure from these simulations with approved numerical tools. Therefore, it is possible that
the observed discrepancies come from a different definition of failure.
That said, if the criteria presented in the European standard for fire testing [6] is adopted,
two limits should be considered: a limiting deflection and a limiting deflection rate. According to
the standard, the first has a value of 67 mm for this example, based on the geometric
characteristics of the beam, while the second has a value of 3 mm/min.
Following that criteria, and assuming that the failure time corresponds to the first of these
two limits to be reached, a section with 3.49 mm² will lead to a fire resistance of 90 min (if the
rate of deflection is determined by a first-order backward difference method computed for every
minute) as seen in Figure 60. This value represents a deviation of only 2.04% to the area suggested
by the DIN.
March 2018
51
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
0
0
Deflection rate
-1
-2
-3
-30
Limiting deflection rate
1st limit reached
Vertical deflection [mm]
-20
-40
-50
-60
-70
Limiting deflection
-4
-5
-6
-7
-80
-8
-90
-9
Deflection rate [mm/min]
-10
-10
-100
0
10
20
30
40
50
60
70
80
90
100
Time [min]
Figure 60 – Deflection and deflection rate vs. time obtained with SAFIR for the section with A’s =
3.49 cm2, with limits for the loadbearing capacity according to the criteria in EN 1363-1 [6]
From the parametric analysis it can be concluded that for the mechanical analyses of simple
structures like the one herein present, the choice regarding the type of analysis or regarding the
time step used is not relevant, and that for such a structure the results for the fire resistance
converge to a given value as the mesh is refined.
2.9. Example 9 – Heavily reinforced concrete beam
2.9.1
Keywords
Fire resistance time, heavily reinforced concrete beam, ISO fire curve
2.9.2
Objective
Example 9 deals with a concrete beam with the same characteristics as the one in Example
8, but subjected to a higher load and reinforced with 6 rebars. The goal is the same, to determine
the required area of the steel rebars that leads to a fire resistance to the ISO fire of 90 minutes.
2.9.3
Description of the problem
The member with the characteristics defined in Figure 61 and Table 24 is analysed. In order
to validate the results, the required steel area for the beam reaching 90 min of resistance when
subjected to an ISO fire is determined and compared to the reference value presented in the DIN
EN1992-1-2 NA.
March 2018
52
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[cm]
Figure 61 – Example 9: Cross-section and configuration of the strongly reinforced concrete beam
Table 24 – Dimensions, material properties and boundary conditions for Example 9
a
b
Properties for reinforced concrete beam (strongly reinforced)
Dimensions
l, b, h
mm
a1,2,3
mm
Spacings
a4,5,6
mm
Loads
qE,fi,d,t
kN/m
Concrete C20/25 (3% humidity by
fck (20°C)
N/mm2
mass)
Steel B500
fyk (20°C)
N/mm2
Concretea
Stress-strain material law
Rebarsb
Fire exposure
ISO 834 (three sides)
Heat transfer coefficient
αc
W / (m2.K)
Emissivity
εm
Concrete
λ, ρ, cp, εth,c
Thermal and physical
material values
Steel
λa, ρ, ca, εth,s
R90
3000, 200, 380
70
40
62.9
20
500
EN 1992-1-2
EN 1991-1-2
25
0.7
EN 1992-1-2
EN 1994-1-2
– with predominantly siliceous aggregates and density ρ = 2400 kg/m^3
– class N, hot-rolled
2.9.4
Models and results (see folder DIN9_4)
Similarly to the previous example, thermal models of the section were created with different
diameters of the rebars, and the resulting temperature fields were used on a structural 2D model
with 20 BEAM elements of equal sizes. The fire resistance time of each model was observed, until
one gave a fire resistance time of 90 min (with a margin of error of 0.5 min).
The calculations performed were DYNAMIC and used the PURE_NR (pure Newton-Raphson)
procedure. The structural model was defined with an initial time step of 1s, a maximum time step
of 60s, a PRECISION of 10E-3s and a COMEBACK of 0.01s.
March 2018
53
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The fabrication process and the ductility class for the STEELEC2EN material were set as
HOTROLLED and CLASS_B, respectively, and the concrete as SILCONC_EN. In the thermal model
(see Figure 62), the thermal conductivity for the concrete was set to follow the upper limit
according to EN1992-1-2.
a) Overall view
b) Detail of one of the rebars
Figure 62 – Example of one of the thermal models of the cross-section tested for Example 9
The required reinforcement for a 90 minutes fire resistance corresponds to a diameter Φ =
13.2 mm for the six rebars, which gives a total steel area A’s = 8.21 cm2.
The temperature in the corner rebars and the fire resistance times obtained for the different
iterations made are shown in Figure 63. The temperature in the rebars for all the models was
taken in the left corner rebar.
March 2018
54
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
650
DIN
DIN
SAFIR
SAFIR
100
640
95
630
90 min
90
7.5
8
8.5
9
Area of steel [cm2]
9.5
10 7.5
8
9.76 cm2
-10 % As
600
8.21 cm2
-10 % As
610
9.76 cm2
620
8.5
85
Fire resistance time [min.]
105
8.21 cm2
Temperature in rebars [°C]
660
80
9
9.5
10
Area of steel [cm2]
Figure 63 – Temperature at the centre of the corner rebars and fire resistance times obtained with
the different models tested, for Example 9
The temperature distribution after 90 min is shown in Figure 64 for the section with A’s =
8.21 cm2.
Figure 64 – Temperatures determined by SAFIR for t = 90 min, for Example 9, with six rebars of
13.2mm (bottom part of the model)
Figure 65 shows the evolution of the vertical displacement at mid-span of the beam. The
vertical displacement at the last converged time step is 81 mm, corresponding to l/37.5. The
horizontal inward displacement at the support on the right is 8.3 mm for the same time instant.
March 2018
55
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 65 – Time vs vertical displacement of the node at mid-span of the beam obtained by SAFIR,
for Example 9
The results obtained with SAFIR and the deviation from DIN EN1992-1-2 NA are
summarized in Table 25.
Table 25 – Necessary area of steel to resist an ISO fire for 90 min, for the heavily reinforced
concrete beam in Example 9
Temperature in the corner
rebars for t = 90 min
Fire
resistance
class
R90
2.9.5
Area of the steel rebars
Reference
Calculated
Reference
Calculated
Deviation
T
T’
As
A’s
(A’s – As) /
As.100
%
-15.9
°C
656
°C
629
cm2
9.76
cm2
8.21
Limit
%
± 10.00
Discussion and conclusions
As for the previous example, the temperatures determined in the rebars after 90 minutes
are lower than the ones provided by the DIN. As a consequence, the amount of steel required to
obtain the 90 min resistance is less than the reference value, but, in this case, the difference
exceeds the acceptable tolerance by 5.9%.
It has to be noted that other software produce results similar to SAFIR (see Figure 66). For
example, with INFOGRAPH [7], the reference reinforcement area of 9.76 cm2 yields a fire
resistance time of 96 min, which is also considerably larger than the expected 90 min. With FRILO
[8], a steel reinforcement area of 8.57 cm2 leads to a fire resistance of 93 min, very close to the
93.7 min obtained by SAFIR when considering 8.59 cm².
March 2018
56
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
105
SAFIR
100
Frilo
DIN
95
7.50
8.00
9.76 cm2
8.57 cm2
90
8.50
9.00
9.50
85
Fire resistance time [min.]
Infograph
80
10.00
Area of steel [cm2]
Figure 66 – Comparison between fire resistance times obtained with SAFIR, different Software and
the one presented in DIN, for Example 9
Considering the criteria in the European standard for fire testing [6] and assuming that the
failure time corresponds to the first of the two limits - limiting deflection and limiting deflection
rate - to be reached, a section with a steel reinforcement area of 8.84 mm² will lead to a fire
resistance of 90 min, as seen in Figure 67. This value represents a deviation of 9.38% to the area
suggested by the DIN, which is still relatively big but that falls inside the 10% of deviation allowed.
0
0
Deflection rate
-1
-2
-3
-30
Limiting deflection rate
-40
-4
-50
-5
-60
-6
-70
1st limit reached
Vertical deflection [mm]
-20
Limiting deflection
-80
-90
-7
-8
Deflection rate [mm/min]
-10
-9
-10
-100
0
10
20
30
40
50
60
70
80
90
100
Time [min]
Figure 67 – Deflection and deflection rate vs. time obtained with SAFIR for the section with A’s =
8.84 cm2, with limits for the loadbearing capacity according to the criteria in EN 1363-1 [6]
As pointed out before, the last 4 examples in DIN (this one included) are based on results
from approved numerical tools. Despite being referred as ‘approved tools’ in [9] and [10], there is
no apparent proof of the accuracy of these results, nor is there any indications on the
considerations and assumptions taken in the analyses that produced them. For instance, it is not
March 2018
57
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
clear what should be the values used for the thermal conductivity. A clear definition of failure, and
the criteria used to determine the failure time, is also missing. Additionally, the fact that other
software obtained identical results to the ones found here, is certainly something that should not
be neglected.
This is why the authors of this report consider that the excessive deviation with respect to
the reference value of the DIN does not invalidate the ability of SAFIR to model the behavior of
concrete beams subjected to fire. Comparison with 2 other independent software yielding similar
results when using the same hypotheses appears to be of greater significance.
2.10. Example 10 – Reinforced concrete beam-column
2.10.1
Keywords
Fire resistance time, displacements, bending moments, beam-column, reinforced concrete,
ISO fire curve
2.10.2
Objective
Example 10 deals with a reinforced concrete beam-column loaded with a horizontal
distributed load and a vertical load with an eccentricity, and subjected to fire on all sides. The goal
is to determine the failure time as well as one displacement and one reaction after 90 minutes of
fire.
2.10.3
Description of the problem
The member with the characteristics defined in Figure 68 and Table 26 is analysed. In order
to validate the results, the time when the member collapses as well as the horizontal displacement
at the top and the bending moment at the support for 90 minutes of an ISO fire are calculated and
compared to the reference values presented in DIN EN1992-1-2 NA.
[cm]
Figure 68 – Example 10: Cross-section and configuration of the reinforced concrete beam-column
Table 26 – Dimensions, material properties and boundary conditions for Example 10
Properties
March 2018
R90
58
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Dimensions
Load eccentricity in fire
Spacing
Loads
Concrete C20/25 (3% humidity by
mass)
Steel B500
Stress-strain material law
Fire exposure
Heat transfer coefficient
Emissivity
Concrete
Thermal and physical
material values
Steel
a
b
l, b, h
e1
a
NE,fi,d,t
WE,fi,d,t
mm
mm
mm
kN
kN/m
7000, 360, 360
35
55
79
1.74
fc (20°C)
N/mm2
20
fy (20°C)
N/mm2
Concretea
Rebarsb
ISO 834 (four sides)
αc
W / (m2.K)
εm
λ, ρ, cp, εth,c
λa, ρ, ca, εth,s
500
EN 1992-1-2
EN 1991-1-2
25
0.7
EN 1992-1-2
EN 1994-1-2
– with predominantly siliceous aggregates and density ρ = 2400 kg/m^3
– class N, hot-rolled
2.10.4
Models and results (see folder DIN10_4)
The section was modelled with 3566 triangular finite elements (see Figure 69). It has to be
noted that 3 bars are on the compression side and 3 bars are on the tension side, in the section
(no bar at the neutral axis).
The structural model was made of 20 BEAM finite elements with equal sizes. A fictitious 3.5
cm long rigid element was used to apply the vertical load with an eccentricity, see Figure 70.
a) Overall view
b) Detail of one of the rebars
Figure 69 – Thermal model of the cross-section for Example 10
March 2018
59
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
a) Overall view
b) Detail of the load eccentricity
Figure 70 – Structural model of the column for Example 10 (20 BEAM elements with equal sizes)
The calculations performed were STATIC and used the PURE_NR (pure Newton-Raphson)
procedure. The structural model was defined with a PRECISION of 10E-3s, a COMEBACK of 1s,with
a time step of 60s, and with the materials SILCONC_EN and STEELEC2EN.
The temperature distribution for t = 90 minutes is presented in Figure 71.
March 2018
60
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 71 – Temperature distribution in SAFIR for Example 10, for t = 90 min
Figure 72 shows the evolution of the horizontal displacement at the top of the column. The
horizontal displacement of the last converged time is 525 mm, corresponding to l / 13.3.
Figure 72 – Horizontal displacement vs. time for Example 10, for the node at mid-length of the
column
The deviations from the DIN EN1992-1-2 NA are found in Table 27. The failure time instant
considered was the last converged time step.
Table 27 – Failure time, horizontal displacement and bending moment for Example 10
min
93
103
Deviation
(X’ – X) / X
∙ 100
%
10.75
°C
°C
502
319
477
302
-4.98
-5.33
-
mm
381
288
-24.41
± 15.00
kN.m
75.5
67.9
-10.07
± 5.00
Parameter
Failure time tu
For t =
90
min
corners
Temp. in
rebars
middle
Hor. displacement ωz
at top of the column
Moment ME,fi,d at the
bottom of the column
2.10.1
Reference
Calculated
X
X’
Limit
%
± 3.00
Discussion and conclusions
Again, the temperatures determined by SAFIR for the rebars are lower than the ones
provided by the DIN. For the rebars at the corners, a difference of -4.98% was found. This leads to
a resistance time of 103 minutes and to a difference of 10.75% to the reference result, which is
March 2018
61
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
considerably bigger than the allowed deviation of 3%. This is, at least, if failure is declared at the
last converged time.
For the loadbearing criteria in the standard for experimental testing EN 1363-1, the criteria
are different for two different situations: vertically and flexural loaded elements. Both situations
will be considered here, because the lateral displacements play a major role on the deformation
of the column, due to the horizontal load and to the eccentricity of the compression load.
As seen in Figure 73, the vertical elongation and vertical elongation rates don’t get close to
the limits for the contraction provided in EN1363-1 for vertically loaded elements, whereas a
value of 98 minutes can be obtained based on the horizontal deflection allowed for flexural loaded
elements. This corresponds to a difference of 5.38% compared to the reference time of 93 min.
40
10
20
5
20
600
0
0
10
20
30
40
50
60
70
80
90 100 110
-5
-20
-10
-40
-15
-60
-80
Limiting contraction rate
Limiting contraction
-20
Time [min]
a) Vertically loaded elements
14
400
Limiting deflection
12
10
300
8
200
6
4
100
-25
-30
-100
16
Deflection rate [mm/min]
0
Horizontal deflection [mm]
Vertical elongation [mm]
Elongation rate
Elongation rate [mm/min]
18
Limiting deflection rate
500
2
0
0
0
10
20
30
40
50
60
70
80
90 100 110
Time [min]
b) Flexural loaded elements
Figure 73 –Deflection and deflection rate vs. time obtained with SAFIR for the top node of the
column, with limits for the loadbearing capacity according to the criteria in EN 1363-1 [6]
Regarding the horizontal displacement at the top and the bending moment at the bottom of
the column for t = 90 min, the values calculated are lower than the limits defined by DIN. If the
values calculated by SAFIR at t = 98 min (time of failure according to the EN1363-1 criteria) are
considered, the horizontal displacement ωz = 358 mm and the bending moment ME,fi,Ed = 72,9 kN.m
fall well inside the limits provided of 15% and 10%, showing differences of -6.04% and -3.44%,
respectively.
Other software found similar results to the ones presented in Table 27, and suggested that
the temperatures mentioned in the DIN correspond to the ones obtained if the rebars are not
present in the model of the cross-section. For INFOGRAPH [7] as well as for MBAEC [11], the
results are closer to the ones from DIN when the thermal analysis is performed for an identical
cross-section with only concrete.
The same was done with SAFIR, i.e. a thermal analysis was performed in a pure concrete
section (the bars are of course present in the structural analysis). This leads to the results
presented in Figure 74, Figure 80 and Table 28, in which the failure time, the horizontal
displacement at the top and the bending moment at the bottom for t = 90 min, fall inside the
stipulated boundaries.
March 2018
62
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Figure 74 – Temperature distribution in SAFIR for Example 10, for t = 90 min, considering the
cross-section with only concrete
20
600
Horizontal deflection [mm]
16
14
400
12
Limiting deflection
10
300
8
200
6
4
100
Deflection rate [mm/min]
18
Limiting deflection rate
500
2
0
0
0
10
20
30
40
50
60
70
80
90
100
110
Time [min]
Figure 75 – Deflection and deflection rate vs. time obtained with SAFIR, considering the
temperatures on the cross-section with only concrete, with limits for the loadbearing capacity
according to the criteria in EN 1363-1 [6] for flexural loaded elements
March 2018
63
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
Table 28 – Failure time, horizontal displacement and bending moment for Example 10, considering
the temperatures on the cross-section with only concrete
Parameter
Failure time tu in min
For t =
90
min
Temp.
corner
in
rebars Intermediate
Hor. displacement ωz
at top of the column
Moment ME,fi,d at
bottom of the column
Reference
Calculated
X
X’
min
93
95
°C
502
°C
Deviation
(X’ – X) / X
∙ 100
%
Limit
%
2.15
± 3.00
484
-3.59
-
319
303
-5.02
-
mm
381
337
-11.55
± 15.00
kN.m
75.5
71.7
-4.90
± 5.00
2.11. Example 11 – Composite column with concrete cores
2.11.1
Keywords
Fire resistance time, displacements, beam-column, composite section, ISO fire curve
2.11.2
Objective
Example 11 deals with a composite steel-concrete column with a partially encased steel
section loaded with a vertical load, with an imperfect shape and subject to fire on all sides. The
goal is to determine the failure time as well as the horizontal displacements for two intermediate
times.
2.11.3
Description of the problem
The column with the characteristics defined in Figure 76 and Table 29 is analysed. It is
considered that both ends are rotationally fixed in case of fire and that it has a geometrical
imperfection with a parabolic shape with a maximum amplitude at mid-length equal to l/1000. In
order to validate the results, the failure time for the member, as well as the horizontal
displacement at mid-length for the time instants of 30 min and 60 min, are calculated and
compared to the reference values presented in DIN EN1992-1-2 NA.
March 2018
64
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[cm]
Figure 76 – Example 11: Cross-section and configuration of the composite column
Table 29 – Dimensions, material properties and boundary conditions for Example 11
Properties
Dimensions
Loads
Concrete C25/30 (3% humidity by
mass)
Steel B500
Steel S235
Stress-strain material law
Fire exposure
Heat transfer coefficient
Emissivity
Concrete
Thermal and physical
material values
Steel
a
b
l, b, h
us
ef
ew
NE,fi,d,t
mm
mm
mm
mm
kN
R90
4000, 300, 300
50
19
11
-1700
fck (20°C)
N/mm2
25
fyk (20°C)
N/mm2
fak (20°C)
N/mm2
Concretea
Rebarsb
Structural steel
ISO 834 (four sides)
αc
W / (m2.K)
εm
λ, ρ, cp, εth,c
λa, ρ, ca, εth,s
500
235
EN 1994-1-2
EN 1991-1-2
25
0.7
EN 1994-1-2
EN 1994-1-2
– with predominantly siliceous aggregates and density ρ = 2400 kg/m^3
– hot-rolled
2.11.4
Models and results (see folder DIN11_4)
The cross-section was modelled with 2690 triangular finite elements, while the column was
modelled in a 2D structural model with 20 BEAM elements of equal sizes (see Figure 77 and Figure
78).
March 2018
65
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The materials STEELEC2EN (defined as HOTROLLED and CLASS_B) and SILCONC_EN were
used for the reinforced concrete part, whereas the I-profile was modelled with the material
STEELEC3EN.
a) Overall view
b) Detail of one of the rebars
Figure 77 – Thermal model of the cross-section for Example 11
Figure 78 – Structural model of the column for Example 11 (20 BEAM elements with equal sizes)
March 2018
66
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The thermal model was defined with a PRECISION of 10E-3s and a time step of 10s. For the
mechanical model it was defined a STATIC analysis with a PURE_NR method, with a PRECISION of
10E-3s, a COMEBACK of 1s and a time step of 60s.
The temperature distribution obtained by SAFIR for t = 90 min is presented in Figure 79.
Figure 79 – Temperatures determined by SAFIR for Example 11, for t = 90 min
Figure 72 shows the evolution of the horizontal displacement at mid-length of the column.
The horizontal displacement of the last converged time is 36 mm, corresponding to l/111.
Figure 80 – Horizontal displacement vs. time for Example 11, for the node at mid-length of the
column
March 2018
67
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
The deviations from the DIN EN1992-1-2 NA can be found in Table 30. The failure time
instant considered was the last converged time step.
Table 30 – Failure time and horizontal displacements for 30 min and 60 min for Example 11
Reference
Calculated
min
92
91
Deviation
(X’ – X) / X ∙
100
%
-1.08
X
X’
°C
535
518
-3.18
-
°C
447
460
2.91
-
mm
mm
4.4
5.5
4.45
6.18
1.14
12.36
± 5.00
± 5.00
Parameter
Failure time tu
Temperature for t =
90 min
Hor. displacement ωz
at mid-length
2.11.5
rebars
centre of
profile
30 min
60 min
Limit
%
± 5.00
Discussion and conclusions
As in the previous examples, the temperature determined by SAFIR for the rebars is lower
than the one provided by the DIN. In this case, the temperature determined in the rebars is 3.18%
less than the reference value. On the contrary, the temperature determined at the centre of the
profile was 2.91% higher than the one in DIN. There is no apparent explanation for this, but it has
been observed that the software Infograph (see [7]) obtain values (523°C for the rebars and 469°C
for the centre of the profile) that are closer to the values of SAFIR than to the reference values.
As for the fire resistance time, a value slightly inferior (-1.08%) to the one presented in the
DIN was obtained, considering the last converged time step as the time of failure of the column.
None of the limits for the EN1363-1 criteria, neither for flexural nor for vertically loaded members,
were reached (nor close to be reached) for this case and it is therefore not possible to apply these
criteria for the determination of the time of failure of the column.
Regarding the horizontal displacements at mid-length for t = 30 min and t = 60 min, it is
important to note that the displacements indicated in DIN appear to be given by reference to the
initial straight position, not considering the imperfection applied, and hence the values present in
Table 30 include the displacement calculated by SAFIR plus the 4mm of initial imperfection
(l/1000). When this initial imperfection is taken into account, the value determined by SAFIR after
30 minutes lies well inside the interval allowed by the DIN, although the same doesn’t happen
after 60 minutes of fire, missing the interval by 7.36%.
3. General conclusions
It can be observed that the majority of the results analysed fell well inside the boundaries
stipulated in the Annex CC of the DIN EN1992-1-2 NA, and possible justifications were provided
when that was not the case. Different behaviours like the heat-transfer (heating and cooling) in
March 2018
68
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
the thermal analyses or the temperature-dependent elongations and stresses in the mechanical
analyses are therefore validated by the results presented in this document.
Most of the uncertainties and doubts are about the reference values provided by the DIN for
the last 4 cases, given that the latter are cases in which the reference solution provided by the DIN
is not theoretical, but instead comes from simulation results with other software. Even though the
software used to calculate the reference values are referred to as ‘approved tools’, there lacks a
description about these software, the modelling assumptions, and the failure criteria, which is
needed to rigorously compare these reference results with results obtained from another
software. It was not possible to find detailed reports on those analyses, and therefore it was not
possible to point out what may be the exact causes for the differences found between the results
determined by SAFIR and those found in the DIN for those 4 cases.
When comparison was possible between the results provided by SAFIR and those provided
by other software, the results of SAFIR were very similar to the results of the other software.
4. References
[1] Franssen J. M., Gernay T. (2017), “Modeling structures in fire with SAFIR®: Theoretical
background and capabilities”, Journal of Structural Fire Engineering, Vol. 8, No 3, 300-323.
https://doi.org/10.1108/JSFE-07-2016-0010
[2] DIN EN 1991-1-2/NA (2010), “National Annex - National determined parameter –
Eurocode 1: Actions on structures – Part 1-2: General actions – Actions on structures exposed to
fire”, Deutsche Norm.
[3] CEN (2005), EN 1992-1-2 “Eurocode 2: Design of concrete structures – Part 1-2 –
General rules – Structural fire design”, European Committee for Standardisation, Brussels.
[4] CEN (2005), EN 1993-1-2 “Eurocode 3: Design of steel structures – Part 1-2 – General
rules – Structural fire design”, European Committee for Standardisation, Brussels.
[5] CEN (2005), EN 1994-1-2 “Eurocode 4: Design of composite steel-concrete structures
– Part 1-2 – General rules – Structural fire design”, European Committee for Standardisation,
Brussels.
[6] CEN (2012), EN 1363-1 “Fire resistance tests – Part 1: General requirements”,
European Committee for Standardisation, Brussels.
[7] Infograph (2015), “Prüfung und validierung von rechenprogrammen für
brandschutznachweise
mittels
allgemeiner
rechenverfahren,”,
http://download.infograph.de/de/validierung_brand.pdf (url last consulted on 25/09/2017).
March 2018
69
University of Liege - Faculty of Applied Sciences
Structural Engineering Research Unit
[8] FRILO
(2017),
“Analyses
on
reinforced
concrete
section”
http://www.frilo.eu/tl_files/frilo/pdf/en/pdf_doku/Analyses%20on%20Reinforced%20Concre
te%20Cross%20Sections.pdf (url last consulted on 25/09/2017).
[9] COST report (2014), “Benchmark studies - Verification of numerical models in fire
engineering”, ISBN: 978-80-01-05442-0, Czech Technical University in Prague, 23-29.
[10] Hosser D., Richter E., Zehfuß J., “Erarbeitung von Nationel Anwendungsrichlinien
fürrechnerische Nachweise nach den Brandschutzteilen der Eurocodes 2 – 5”, Abschlussbericht im
Auftrag des Bundesministeriums für Raumordnung, Bauwesen und Städtebau (Az. RS III 4 – 67 41 –
97.120). Institut für Baustoffe, Massivbau und Brandschutz (iBMB), Braunschweig, 1999 (in German).
[11] Mbaec (2014), “Validierung gemäß DIN EN 1991-1-2 / NA:2015-09”,
https://www.mbaec.de/fileadmin/Datenblaetter/U412de_Validierung.pdf (url last consulted on
25/09/2017)
March 2018
70