VTT R 01177 17
VTT R 01177 17
VTT R 01177 17
RESEARCH REPORT
Confidentiality: Public
RESEARCH REPORT VTT-R-01177-17
2 (26)
Preface
This report presents an approach to handle the ductility requirements of steel under tension,
in particular in the necking areas of cross-sections. The presented calculation methods are
based on the experiences from the numerical studies with regard to different design situations.
They provide alternative and complementary solution to the Eurocode’s material requirements
that is practical especially for the building and product designers using finite element software.
The experiences from the conducted studies are summarized herewith for the purpose of
numerical evaluation of material behaviour in details with high and localized strains and
possible diffuse necking. The methods are validated against real experiments.
Two basic numerical methods presented in this report provide: (a) the practical limitations for
the numerical calculations to ensure that the ductile failure does not occur in the material and
(b) prediction the minimum required material ductility in a certain situation.
The report has been prepared in the project “Virtual testing lab for novel materials and
products”, called VILMA (2014–2016). The main research objective of VILMA was to develop
a virtual testing platform for a fast and effective introduction of new structural steels and steel
products to the market. The main practical objective was to use the platform to develop
recommendations of the material ductility requirements in the Eurocodes for high strength
steels.
The VILMA project belongs to the program “BSA - Breakthrough steels and applications and
its project portfolio P2, “Design beyond present codes – enabling efficient utilisation of new
materials”. The industry-driven project portfolio was planned to rise to critical future needs of
steel end-users such as product manufacturers, designers and building owners (market pull).
The overall goal of BSA program is to enable a renewal of the Finnish metal and engineering
industries through major improvements in their offerings and global competitiveness brought
about by the intelligent use of novel advanced steel products. Key emphasis is on end-users
in selected business areas: bioenergy, power generation, mining, lifting, handling and
transport, offshore and marine, waste recycling, arctic technologies and processing industry
(http://www.fimecc.com/programs/bsa). The BSA program is a part of the large national
DIMECC innovation eco-system that represents a new type of public-private partnerships
aiming at faster innovation processes.
The authors wish to thank the industry representatives from Ruukki Construction Ltd and SSAB
Europe Ltd who have been active in planning and supervising the work.
Espoo 3.3.2017
Authors
RESEARCH REPORT VTT-R-01177-17
3 (26)
Contents
Preface...................................................................................................................................2
Contents .................................................................................................................................3
1. Introduction .......................................................................................................................4
2. Geometry of the details .....................................................................................................5
2.1 Geometry limits .........................................................................................................5
2.2 Initial imperfections ...................................................................................................5
3. Constitutive models...........................................................................................................6
3.1 Elastic - ideally plastic model ....................................................................................7
3.2 Model with linear strain hardening ............................................................................7
3.3 True material model ..................................................................................................9
4. Estimation of ductile failure .............................................................................................10
4.1 Simple calculation limits ..........................................................................................10
4.2 Damage prediction models .....................................................................................12
5. Verification of material ductility ........................................................................................13
5.1 Numerical models ...................................................................................................13
5.2 Required elongation at failure .................................................................................13
6. Solved numerical examples ............................................................................................17
6.1 Centre hole in tension .............................................................................................17
6.1.1 Material models ..........................................................................................17
6.1.2 Estimation of ductile failure ......................................................................... 18
6.1.3 Verification of the material ductility .............................................................. 20
6.2 Tubular connection .................................................................................................22
6.2.1 Material models ..........................................................................................23
6.2.2 Estimation of ductile failure and verification of the material ductility ............ 24
6.2.3 Conclusions ................................................................................................25
7. Summary ........................................................................................................................25
References ...........................................................................................................................26
RESEARCH REPORT VTT-R-01177-17
4 (26)
1. Introduction
Finite element models (FEM) are becoming common tools for structural steel designers and
structural products manufacturers. They are typically used to optimize geometry and material
of the product or structure, to predict its loading resistance or to verify the design assumptions.
With the introduction of new high-strength steel grades, improved fabrication tolerances and
joining methods, it is possible to perform very accurate, highly optimized and complex design
assisted by FEM. Such design may, however, require a special attention in particular details
where the concentration of stress and strain cannot be avoided. The essential prerequisite to
prevent material failure in this situation is sufficient material ductility.
This report presents a FEM approach to perform the following engineering tasks dealing with
the structural steel details subjected to tensile stress concentrations:
(a) To provide safe prediction of tensile resistance of a detail using simple FEM tools
and a basic knowledge of the material behaviour. Several alternative methods are
presented in Chapter 4 (Prediction of ductile failure).
Some of the criteria may allow plastic deformation and diffuse necking in the localized
areas, and therefore are not suitable for cyclic loading.
(b) To aid the designers with the selection of steel grade through the prediction of
minimum required material ductility that will ensure safe performance of the studied
detail.
The predicted material ductility may be lower than the minimum requirements of the design
codes or higher if the exceptional performance is needed (e.g. large elongation of the area
with stress concentration without failure). Chapter 5 (Verification of material ductility) presents
the description of this method.
Both approaches are based on the same assumptions concerning numerical models described
in Chapter 2 (Geometry of the details) and material models described in Chapter 3 (Constitutive
models).
The approaches were verified by experiments (including results from previous tests and new
experiments) and numerical simulations. However, there are needs to perform broader
experimental and numerical studies to extend the applicability area from the restrictions
described also in the report.
RESEARCH REPORT VTT-R-01177-17
5 (26)
The presented approach was verified by a limited number of experiments and numerical
simulations. Therefore, it can be used directly only within the given limits as explained in the
following sections.
Additionally, the acceptable failure mode of the details in tension should be ductile. The details
with very thin plates (high aspect ratio of the individual plates) may also fail in shear or
combination of both failure modes. The recommended maximum aspect ratio of the plated
elements is therefore 1:8.
The typical example of the model requiring initial imperfections is a plain coupon subjected to
concentric axial tension. The proper magnitude of the initial imperfection is important especially
for materials with nearly perfect plastic behaviour.
(a) The imperfection magnitude should be as small as possible, ideally in the order of
magnitude of the surface roughness (e.g. 10 to 100 µm). Depending on the FEM solver,
necking may not be initiated with very small magnitudes due to round-off error.
Therefore, it is recommended to use double precision if possible. On the other hand, in
some cases only the mesh irregularities caused by denser meshing in the critical area
may be sufficient to trigger the necking.
(b) It is recommended that the critical cross-section has the nominal dimensions, while the
rest of the model is slightly larger. This will produce accurate stress and strain
distribution in the critical cross-section.
The details with holes and notches, or cross-sections subjected to tension and bending do not
require initial imperfections, because their stress distribution is already non-uniform in the early
loading phase.
RESEARCH REPORT VTT-R-01177-17
6 (26)
3. Constitutive models
Material parameters for creating a proper stress-strain relationship of structural steel can be
obtained from several sources.
(a) Eurocode 3 (EN 1993-1-1, EN 1993-1-4 and EN 1993-1-12) provide the nominal
values of yield strength fy, ultimate strength fu and modulus of elasticity E. EN 1993-1-
4 recommends also non-linear parameter n for Ramberg-Osgood models of stainless
steels, but these models are not discussed in this report. It should be noted that the
material parameters obtained from ductility limits for elongations (e.g. uniform strain or
elongation at failure) in the Eurocode are too conservative especially for mild steels.
(c) Material certificate (EN 10204:2004) contains the test results such as yield strength,
ultimate strength Rm and elongation at failure. This certificate is usually delivered with
the material.
(d) Tensile test raw results (tabular data of measured load and displacement) are the
most accurate information for building the constitutive model of the material. They are
usually produced directly by the testing machine. They can be transformed into true
stress-strain curves and used directly as inputs for the definition of plasticity in the finite
element models.
The true value of yield stress required by FEM solvers is slightly higher and should be
calculated according to the Equation (1).
σ t = σ (1 + ε ) and ε t = ln(1 + ε )
(1)
Where the engineering stress σ is the yield strength fy in this particular case and engineering
strain ε is the yield strain fy/E.
For instance, for steel S690, the recommended yield strength fy is 690 MPa and modulus of
elasticity E is 210000 MPa. Therefore the true value of yield stress required by the finite
element solver is 690·(1+690/210000) = 692 MPa.
fu − f y
f u ,red = f y + ε u ,min (2)
A5
where εu,min = 15(fy/E) is the minimum uniform elongation required by the Eurocode 3.
It should be noted that such models can achieve slightly higher resistance than the ideally
plastic materials, but their ductility in tension is significantly improved because the necking
starts after reaching the minimum uniform elongation εu,min as demonstrated in Figure 2.
For instance, for steel S690QL produced by SSAB, the following parameters can be obtained
from the declaration of performance according to EN 10025-6:
- Minimum yield strength is 690 MPa;
- Minimum tensile strength is 770 MPa;
- Elongation at failure is 14%.
The minimum uniform elongation εu,min required by the Eurocode 3 for this grade is than
15·(690/210000) = 4.93% (modulus of elasticity E is recommended in EN 1993-1-1). The
reduced ultimate strength fu,red can be calculated as 690+(80/0.14)·0.0493 = 718 MPa and its true
value σt,u,red is 718·(1+0.0493) = 754 MPa according to the Equation (1). The true logarithmic
strain εt,u,min corresponding to this stress is then ln(1+0.0493) = 4.81%.
Methods (a) and (b) below can be used for the design according to the design codes. Other
methods are suitable for example for optimization of prototype details and shall be verified by
testing.
(b) 50x yield strain - Plastic limit state (LS1) in EN 1993-1-6 (shell structures) is based on
von Mises equivalent plastic strain and the limit is εmps = 50εy = 50fy/E. This condition
shall be fulfilled at any point not closer to a notch or local discontinuity than the thickest
adjacent plate thickness. It should be noted that EN 1993-1-6 is intended for use in
conjunction with EN 1993-1-1 (not EN 1993-1-12), and therefore it is applicable only
up to S460. Yield strength fy = 460 MPa results in 50εy=11%.
Table 3. 50x yield strain limit for common structural steel grades (covered by EN 1993-1-1)
fy εy 50εy
Grade
MPa % %
S235, S235 W 235 0.11 5.6
S275, S275 N/NL/M/ML 275 0.13 6.5
S355, S355 N/NL/M/ML/W 355 0.17 8.5
S420 N/NL/M/ML 420 0.20 10.0
S450 440 0.21 10.5
S460 N/NL/M/ML/Q/QL/QL1 460 0.22 11.0
(c) Uniform elongation of material - One conservative approach is to limit plastic strains
to the uniform elongation εu, of the material. Up to this value, the plastic strain in plain
coupons is uniformly distributed in the cross-section, and therefore very similar to the
elongation. Minimum value of εu,min = 15εy is required in EN 1993-1-1. As an example
15εy=4.9% for a steel with yield strength of 690 MPa. Similarly to the 5% rule, the
practical usability of such criterion is limited to the areas without stress concentration.
RESEARCH REPORT VTT-R-01177-17
11 (26)
Table 4. Minimum uniform elongation εu,min = 15εy of material for common structural steel
grades (covered by EN 1993-1-1 and EN 1993-1-12).
Thickness fy εy εu,min
Grade
mm MPa % %
S235, S235 W any 235 0.11 1.68
S275, S275 N/NL/M/ML any 275 0.13 1.96
S355, S355 N/NL/M/ML/W any 355 0.17 2.54
S420 N/NL/M/ML any 420 0.20 3.00
S450 any 440 0.21 3.14
S460 N/NL/M/ML/Q/QL/QL1 any 460 0.22 3.29
≤ 50 500 0.24 3.57
S500 Q/QL/QL1/MC ≤ 100 480 0.23 3.43
≤ 150 440 0.21 3.14
≤ 50 550 0.26 3.93
S550 Q/QL/QL1/MC ≤ 100 530 0.25 3.79
≤ 150 490 0.23 3.50
S600 MC ≤ 16 600 0.29 4.29
≤ 50 620 0.30 4.43
S620 Q/QL/QL1 ≤ 100 580 0.28 4.14
≤ 150 560 0.27 4.00
≤8 650 0.31 4.64
S650 MC
≤ 16 630 0.30 4.50
≤ 50 620 0.30 4.43
S690 Q/QL/QL1 ≤ 100 580 0.28 4.14
≤ 150 560 0.27 4.00
≤8 700 0.33 5.00
S700 MC
≤ 16 680 0.32 4.86
(d) Elongation at failure - A less conservative approach is to limit true plastic strains to
the value of elongation at failure A5. Since the real plastic strain becomes much higher
than the elongation after the onset of necking, this criterion appears to be a safe
approximation applicable also in the areas around the holes and notches. This limit
was verified on the literature study made for test data of notched test specimens.
RESEARCH REPORT VTT-R-01177-17
12 (26)
where stress triaxiality T is the ratio of the hydrostatic stress and von Mises stress.
Exponent 1.5 is commonly used in the expression for steels. However, without
experimental verification it can provide unconservative results. The toughness
parameter α for steels is in range of 1 to 5. Then, depending on the material grade, the
critical plastic strain results in 22%–112%, if T = 1. The parameter α is usually calibrated
on tests and FEM analyses for smooth-notched CNT (circumferential notch tensile)
specimens.
SMCS model can provide accurate fracture predictions for many practical conditions,
such as the necked ligament between bolt holes, the necked cross section of an un-
notched cylindrical bars, structural moment connections, or circumferential notch
tensile (CNT) specimens. In these cases, fracture typically initiates internally, where
the stress triaxiality is relatively high (T>0.75) and then propagates outwards towards
the surface of the material. However, there are other situations where fracture may
initiate on the surface of the material, where triaxiality is typically lower (T=0.33–0.75),
and then propagates inward. Fracture initiation on the surface has been observed for
example in large scale tests on structural braces and column base plate tests.
(f) Complex models - More sophisticated macroscopic damage models are usually used
for simulations of sheet forming or car deformations in accidents. A large number of
tests for different specimens and complementary FEM analyses need to be done.
These damage models are often laborious and difficult to interpret in practical design.
The last method (g) from this overview proved to be the best combination of accuracy and
simplicity because it requires only standard experimental tests results (plain coupons of
uniform cross-section), but still provides reasonable results in FEM simulations of structural
details with stress concentration. This prediction is further explained in the following sections.
RESEARCH REPORT VTT-R-01177-17
13 (26)
The method relies on a standard material coupon test and two numerical models, one model
of the detail and one model of the coupon itself. The knowledge of exact stress-strain curve of
the material and the geometry of tested coupon is not essential, but is recommended because
it brings less conservative results. The method can be, however, used only with the basic
material knowledge of fy, fu, E and elongation A5 (if it is known from the mill certificate or
declaration of performance according to the harmonized technical specification EN 10025-1).
(a) The model of the tensile coupon - The model should represent the real geometry of
the test. The surface imperfections of the real coupon should be smaller than the
imperfections implemented in FEM model of the coupon. This ensures higher stress
concentration in the model, and therefore more conservative results.
(b) The model of structural detail - The finite element mesh of the structural detail should
have at least the same density as the mesh of the simulated coupon test. It is
recommended that the same software be used for both numerical models with the same
element types and calculation settings.
(a) Stress triaxiality at the failure of the structural detail is smaller than the stress
triaxiality of tested coupon. This condition is true for structural details with notches
and holes in tension and plain rectangular coupons except for the small plastic
deformations.
(b) The relation between failure strain and stress triaxiality (the damage curve) of a
material is monotonic decreasing function. This is also true for the common
structural steels and the details prone to ductile failure in tension.
The calculation is based on the assumption that strains larger than the uniform elongation of
tensile coupon εu can be accepted in localized areas of statically loaded structures. In such
cases, diffuse necking may develop in localized areas. However, the load should not cause
ductile failure in materials, and therefore the plastic strains and hydrostatic stress (represented
by stress triaxiality) should remain within a given range. Unfortunately, it is impossible to
describe exactly the relation of stress triaxiality and plastic strain at material failure with data
from commonly used coupon test. Therefore, the present method involves certain
conservativeness in the generated results. If a more accurate solution is needed, it has to be
RESEARCH REPORT VTT-R-01177-17
14 (26)
based on rather complicated testing programme in combination with more complex damage
prediction models.
The entire process is described in Figure 4, Figure 5 and the following paragraphs.
(a) Selection of the design limit - the design limit can be expressed as load (e.g. ultimate
load), deformation (e.g. 3 mm elongation) or any other measurable property of the
detail. Many existing limits in the Eurocodes can be utilized here. Depending on the
limit, several suitable material models can be selected.
Figure 4. Prediction of minimum required elongation at failure based on the FEM model of
the structural detail (left) and FEM model of the coupon test (right).
Equivalent
FEM of the
plastic strain
detail
in detail
(b) Selection of material model - Based on the available knowledge of the material’s
parameters, one of the possible constitutive models can be selected from Chapter 3.
The yield strength of the material is required as a minimum input, but for some design
limits (e.g. the ultimate load), the knowledge of more material parameters might be
needed.
More complex models that are based on the Declaration of Performance, Inspection
Certificate or real tensile testing provide usually the knowledge of material’s elongation
at failure A5, and therefore the purpose of the calculation would be to verify if it is
sufficient in the current design situation.
RESEARCH REPORT VTT-R-01177-17
15 (26)
Generally, the safe assumption beyond the ultimate load is ideally plastic behaviour of
true stress-strain curve. However, better performance can be achieved with the
knowledge of complete coupon testing data (or at least the load and elongation at
failure), because the true stress-strain behaviour can be curve-fitted to match this data.
(c) Finite element model of the standard coupon - If the material model is based on the
particular coupon test or is associated with the particular sheet/plate thickness, the
dimensions of the coupon should be selected accordingly. Otherwise, the rectangular
coupon with the material thickness corresponding to the studied detail and aspect ratio
at least 1:4 (but not too high to cause shear failure) is recommended.
The coupon mesh should be denser in the area of diffuse necking, but its density can
be based on the computational capacity of the FEM solver. Generally, denser mesh
results in less conservative prediction of minimum required elongation at failure. Mostly
it is sufficient to simulate only 1/8 of the coupon due to its symmetric behaviour (see
Figure 6).
Plain coupons might require initial imperfections in order to initiate diffuse necking in
the middle of the coupon (see Chapter 2.2). The cross-sectional area can be then
reduced locally at least by the value of the surface roughness. Higher reduction will
result in less conservative prediction.
FE model should not contain any provision for the failure simulation and should be able
to provide results also beyond the ultimate load. Therefore, it is recommended to load
the model with deformation that should be larger than the expected failure elongation.
Figure 6. Simple FE model (3 symmetry planes) of the coupon with the largest strain in the
middle of the crosss-section.
(d) Finite element model of the studied detail - The same material model and element
type should be used in the FEM model of the detail. It is also recommended to use the
same FE solver with the same basic settings.
The meshing recommendations are based on the FEM model of the coupon, but the
initial imperfection might not be necessary because of the presence of holes or notches
(see Figure 7).
RESEARCH REPORT VTT-R-01177-17
16 (26)
Similarly as the coupon, this FE model should not contain any provision for the failure
simulation and should be able to provide results also beyond the ultimate load.
Therefore, it is recommended to load the model with deformation that is large enough
to cover the required design limit(s).
Figure 7. FE model of the detail with the largest strain at the edge of the hole.
(e) Determination of equivalent plastic strain at the design limit in detail FE model -
The maximum equivalent plastic strain of the model should be obtained at the load
level corresponding to the selected design limit. The largest strains are usually at the
edge of holes and notches.
This prediction can be too conservative in some cases, because in reality also coupons
failing at lower strain levels might be capable of reaching the desired strain in the detail.
The location of maximum equivalent plastic strain in the coupon model is in the middle
of the cross-section.
If the FE model of the coupon or the detail starts necking along any longer path than the
smallest net-section of the model, it might be indication of the shear failure that is not yet
applicable for this method.
RESEARCH REPORT VTT-R-01177-17
17 (26)
The calculation results are verified by the tensile test of the dogbone specimen, in which the
central part is identical to the FE model.
Four constitutive models of high-strength steel S960 were selected to study the recommended
calculation limits:
- Model with linear strain hardening based on the stress-strain curve fitted to the real
coupon test, where fy = 1064 MPa, σt,u = 1202 MPa, and εt,u = 2.82%.
- Model with nonlinear strain hardening fitted to the real coupon test
Their modulus of elasticity is 210 GPa, Poisson’s ratio is 0.3, and the full true stress-true
logarithmic strain relationship is in Figure 9.
RESEARCH REPORT VTT-R-01177-17
18 (26)
1400
1200
1000
True stress (MPa)
800
600
In order to predict safe calculation limit before the real ductile failure of the material, several
criteria were selected for the maximum allowable equivalent plastic strain
- 5% strain
- Uniform elongation 15x yield strain (6.9% equivalent plastic strain) for the first two
material models, where the real uniform elongation is not known or the measured
uniform elongation 3% for the models based on coupon test.
- Elongation at failure 10% for the first material model, where the A5 value is not
known or is given by the Declaration of Performance by SSAB, or 9.8% for the
models based on coupon test.
- Plastic strain at coupon failure elongation 98.2% was obtained from the simulation
of the coupon test, and therefore this limit is used only in combination with nonlinear
strain hardening material model.
The results of combination of four material models and four strain limits are presented in Figure
10 and Table 5. The failure force Ff and elongation ∆Lf at the failure of the detail (and its % of
initial length L0) are presented in the table as well as the maximum load reached before failure
(as % of Fy).
RESEARCH REPORT VTT-R-01177-17
19 (26)
600
500
Load (kN)
Figure 10. Load-displacement relationship using different material models and calculation
limits.
RESEARCH REPORT VTT-R-01177-17
20 (26)
The following results can be observed
- The material models based on the real testing data can offer higher loading capacity
(up to 124% of Fy in our case) and deformation capacity (up to 1.48% in our case) within
the given limits.
- All of the damage prediction limits had a certain level of conservativeness in this case.
- The limit based on uniform elongation guarantees that the necking is not developing in
the detail at all as well as the 5% limit for steels with fy ≥ 700 MPa 1. Such condition
might be suitable for cyclic loading but in the statically loaded cases it can be, however,
too limiting. The elastic-ideally plastic model is not suitable for the simulation of the
details, where the necking should be restricted, because the diffuse necking develops
at much lower strains than in reality.
- The limits based on 50x yield strain and elongation at failure allow the localized necking
as well as the 5% limit for steels with fy < 700 MPa. This condition does not mean
automatically that the ultimate capacity of the detail is reached.
The second task that can be demonstrated on the same model is the calculation of required
elongation at failure of the coupon in the standard testing. For this reason, another FE model
has to be produced; the model of the coupon. The coupon model was in this example already
created to curve-fit true stress-strain relationship to the real experiment (see Figure 11)
The goal of this example is to predict the minimum required elongation at failure A5 of the steel
S960 to reach the ultimate tensile load of the detail in the net-section. Therefore, the only
material models that are able to simulate the ultimate strength of the material will be used.
1The strain at onset of necking (uniform elongation) should be larger than 15fy/E which is higher than
5%, when fy ≥ 700 MPa.
RESEARCH REPORT VTT-R-01177-17
21 (26)
800
ultimate load
700
600
500
Load (kN)
400
300
200 Experiment
Linear hardening - from experiment
100
Nonlinear hardening
0
0 1 2 3 4 5
Elongation (mm)
This example shows that the elongation at failure A5 has to be higher than 6.1% in order to
reach the ultimate load Fu of the selected detail. In addition, the nonlinear hardening model is
needed for such prediction, because the simplified linear hardening model provided less
conservative results.
RESEARCH REPORT VTT-R-01177-17
22 (26)
Figure 13 Lab-test report for X3TT_960, a typical X joint test used in tubular structures
As shown in Figure 14.a, the failure of the joint occurred in the weld, possibly initiating from
the heat-affected zone. The relatively prominent weld termination/start in the forefront of the
picture may also have played a role in the failure initiation. The weld dimensions of an
identically welded specimen were also available from measurement (Figure 14.b). It has to be
noted that the welds in Figure 14.a and Figure 14.b are not the same, so there is a degree of
uncertainty concerning the dimensions of the weld in specimen X3TT_960. The Finite element
rendering of the weld is presented in Figure 14.c.
RESEARCH REPORT VTT-R-01177-17
23 (26)
b)
a)
c)
Figure 14 Failure mode of X3TT_960 initiated in the weld or heat affected zone of the weld.
The weld start/termination introduced a prominent discontinuity (and weakness) in the weld.
Since material tests for the specimen were not available, we used material data from the same
producer. The Finite element model was using homogeneous material properties, disregarding
the variation of properties in the weld and HAZ. Four material options were used for the models
were used in the models (Figure 15).
As shown in Figure 16, the FE model is able to pinpoint the stress concentrator responsible
for the failure of specimen X3TT_960. The consecutive concentration of plastic strains is also
replicated.
Figure 17. Performance curve of the specimen from experiment (red) and FEM with different
material model options (grey/black).
RESEARCH REPORT VTT-R-01177-17
25 (26)
Figure 18 presents the performance curve obtained by FEM, together with the failure load
predicted by the EN 1993-1-8 equations (NRd=436kN). It can be noted that this force level is
somewhat larger than the 420kN received in the test (Figure 17). In terms of plastic strain, with
a choice of limiting PEEQ at 10% the capacity of the joint would be (under)predicted to 338kN,
with PEEQ at 20%, the force estimate would result realistic at 434kN and with PEEQ 25% the
capacity would be (over)predicted as 469kN.
(a) (b)
Figure 18. (a) Performance curve of X3TT_960 obtained by FE model and (b) development
of the plastic strain and triaxiliaty in the critical element
6.2.3 Conclusions
The welded joint application example of the proposed methodology was unsuccessful. This
can be attributed to the model simplifications, especially what concerns the weld and HAZ
properties. If these locations are weakened, when compared to the rest of the model, it is
probable that stresses and plastic strains concentrate in these regions, resulting in larger
localised demands. Unfortunately, for this application not even the base material properties
were available, let alone weld and HAZ properties.
For further calibration of the methodology, the use of well-documented tests with known
properties of the different components should be used. The modelling approach also has to be
a degree more sophisticated, including for instance residual stresses that were ignored here.
However, in this case modelling would require more resources than available in this project.
7. Summary
The two methods demonstrated in this report are simple approaches to the problems with the
details subjected to the concentrated tensile stresses in some areas or the whole cross-
section. They provide insight into the basic questions whether the Eurocode’s material
requirements are correct or not, could they be eased or not, and how much they could be
eased. The methods are validated against real experiments.
RESEARCH REPORT VTT-R-01177-17
26 (26)
This report can serve as a guidance for the finite element modelling of structural details that
are subjected to tension, with the risk of initiation and development of diffuse necking.
References
[1] EN 1993-1-1 Eurocode 3: Design of steel structures - Part 1-1: General rules and rules
for buildings, European Committee for Standardisation, 2005.
[2] EN 1993-1-12 Eurocode 3: Design of steel structures - Part 1-12: Supplementary rules
for high-strength steels, European Committee for Standardisation, 2007.