Received: 20 May 2020 Accepted: 28 May 2020

DOI: 10.1002/cend.202000015

FULL PAPER

Component based finite element design of steel joints

František Wald1 | Martin Vild2 | Marta Kuříková1 | Jaromír Kabeláč2 |

David Sekal1 | Nadine Maier3 | Laura Da Silva Seco4 | Maël Couchaux4

1
Czech Technical University in Prague, Praha,
Czech Republic Abstract
2
Brno University of Technology, Brno, Czech This paper describes the principles of multi-level Finite Element modeling for design
Republic
of structural steel joints. The integral part of the design by Finite Elements is the Vali-
3
Technische Universität München, Munich,
Germany dation and Verification of the model and its results. For models by shell and solid ele-
4
Institut National des Sciences Appliquées de ments it is mesh sensitivity study. In component bases finite element method
Rennes, Rennes, France
(CBFEM), the steel plates are considered by materially nonlinear analysis (MNA).
Correspondence Their resistance in plastic stage is limited by strains. The behavior of components, for
František Wald, Czech Technical University in
example, of bolts, anchor bolts, welds etc., is treated by introducing nonlinear springs
Prague, 166 29 Praha, Czech Republic.
Email: [email protected] representing its behavior in terms of initial stiffness, ultimate resistance and deforma-
tion capacity. To show this process a contribution is prepared, which summarizes the
history of achievements of Finite Element Analyses (FEA) in structural steel joints.
This paper shows differences between the research-oriented and design-oriented
models and the current trends in modeling of connection components. Both have
their roles in advanced analyses. The Validation and Verification are demonstrated on
model of block shear during potential failure mode for gusset plate connection and
for the base plates loaded by bending moments around both axes. The summary
shows the potential in design of elements including their joints.

KEYWORDS

base plate, finite element analyses, block shear, joints, multi-level model, steel structure

1 | I N T RO DU CT I O N majority of software and design tables for structural steel used in

Europe. The spring model was generalized by da Silva7 for 3D behav-
Experimental evidence and curve fitting procedures were and are ior including nonlinear part of description of components. Rasmussen
used for safe and economical design of connections. Based on analyti- et al.8 included a descent part of the description of the curve for lim-
cal models of resistance of connectors, such as welds, bolts, and ited tested cases. Intricacy of models with three springs in series for
plates, and the estimated lever arm of internal forces, the resistance one component leads to complex integration of the whole joint
of connection is predicted by analytical models. Zoetemeijer1 was the behavior. Procedure starts with decomposition of a joint into compo-
first who equipped this model with estimation of stiffness and defor- nents followed by their description in terms of normal/shear force
mation capacity. The elastic stiffness was improved in the work of deformation behavior. After that, components are grouped to examine
Steenhius, see Reference 2. Basic description of component behavior joint moment-rotational behavior and classification/representation in
in major structural steel connections was prepared by Jaspart for a spring/shear model and application in global analyses. Advantage of
beam to column connections3 and by Wald et al. for column bases.4 Component Method (CM) is integration of current experimental and
Method implemented in the current European structural standard for analytical knowledge of connection components behavior; that is, of
the steel and composite connections, see References 5,6, is applied in bolts, welds and plates. This provides very accurate prediction of

© 2020 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin

78 wileyonlinelibrary.com/journal/cend Civil Engineering Design. 2020;2:78–89.

WALD ET AL. 79

behavior in elastic and ultimate level of loading for joints with known Shell elements are recommended for modeling of plates in design
lever arm of internal forces. Disadvantage of CM is that experimental FEA of structural connection. Four node quadrangle shell elements
evaluation of internal forces distribution is available only for a limited with nodes at its corners are applied. Six degrees of freedom are con-
number of the open section joint configurations. In contemporary sci- sidered in every node: three translations (ux, uy, uz) and three rotations
entific papers, description of atypical components may be found as (φx, φy, φz). Deformations of the element are divided into membrane
well. The use is of this solution and the boundary conditions are very and flexural components. The formulation of the shell behavior is
uncertain for design. The CM is not developed for hand calculation based on the work by Ibrahimbegovic.13 Rotations perpendicular to
but as a method for preparation of design tables or software tools. the plane of the element are considered. Complete 3D formulation of
Models of hollow section connections are described in Chapter 7 of the element is provided. The out-of-plane shear deformations are
EN1993-1-8:20065 by curve fitting procedures based on mechanical included in the formulation of the flexural behavior of element by
and numerical experiments. The inconvenience is not standardized Mindlin hypothesis. The solution was completed by Dvorkin in Refer-
definition of design resistance. For selected joint configuration, the ence 14 as MITC4 elements. The shell is divided into five integration
component representation is prepared,9 which is based on selection points along the height of the plate and plastic behavior is analyzed in
of the suitable lever arms and effective widths. This procedure is lim- each point, the Gaus-Lobatto integration. The nonlinear elastic-plastic
ited by taking the resistance from curve fitting procedure. The behav- stage of material is analyzed in each layer based on the known strains.
ior of components is calculated backwards and has no support in
experiments or good engineering practice.
The recommendation for design by advanced modeling in structural 2.2 | Bolts
steel is ready to be used in Chapter 5 and Annex C of EN January
5, 1993:2005.10 Finite element analysis (FEA) for connections is used In CBFEM, the component bolt in shear and tension is modeled by a
from 70s of last century as research-oriented FEA (ROFEA). Their ability dependent nonlinear spring. The deformation stiffness of the shell ele-
to express real behavior of connections is making numerical experiments ment, which models the plates, distributes the forces between the bolts
a valid alternative to testing and source of additional information about and simulates the adequate bearing of the plate. The initial stiffness and
local strains. Material in ROFEA is described by true stress-strain or dam- the design resistance of a bolt in shear in CBFEM are modeled according
age models often including changes across and along the cross-section. to cl. 3.6 and 6.3.2 in EN1993-1-8:2006.5 The spring representing the
Validation & Verification (V&V) process of models is integral part of the bolt in shear has bilinear force deformation behavior, see Figure 1.
11
ROFEA procedure and is often based on the researcher's own experi- Deformation capacity is considered according to Reference 11 as
ments. The component based finite element analysis (CBFEM) of struc-
tural connection is the current tool for its design. As the computational δpl = 3δel ð1Þ
tools are commonly available and easy to use even by relatively inexperi-
enced engineers the proper procedure should be employed when judging
the results of computational analysis. The detailed procedure for verifica- Initialization of yielding is expected, see Figure 1, at
tion of CBFEM was prepared.12 In such situation, the verification process
performed through benchmark tests gains crucial importance. The Fv,el = 2=3 Fv,Rd ð2Þ
CBFEM model allows accurate design of joints with generally loaded
joints with complex geometry. As with any design using shells, care The spring of a bolt in tension is described by its initial deforma-
should be taken to position of supports and loading.12 tion stiffness, design resistance, initialisation of yielding and deforma-
tion capacity. The initial stiffness is derived analytically as

2 | C O M P O N E N T B A S E D FI NI T E E LE M E N T K b = ðE As Þ=Lb ð3Þ
METHOD
where E is the Young's modulus, As the tensile stress area of a bolt
2.1 | Material and plates in design-oriented models and Lb the bolt elongation length. The model corresponds well to
experimental data, see Reference 12 The design values according to
Common material diagrams, which are used in design by finite ele- ISO 898:20095 are summarized in Reference 15. A combination of a
ment modeling of structural steel, are the ideal plastic or elastoplastic shear and a tension in a bolt is expressed in EN January
model with strain hardening. The elastoplastic material with strain 8, 1993:20055 in Table 3.4 by a bilinear relation and checked as
10
hardening is modeled according to EN January 5, 1993. The material


behavior is based on von Mises yield criterion. It is assumed to be F t,Ed F v,Ed F t,Ed
max , + ≤ 1:0 ð4Þ
F t,Rd Fv,Rd 1:4 F t,Rd
elastic before reaching the yield strength fy. The ultimate limit state
criterion for regions not susceptible to buckling is reaching of a limit-
ing value of the principal membrane strain. The value of 5% is where Fv,Ed is the acting bolt shear force, Ft,Ed is the acting bolt tensile
recommended. force, Fv,Rd is the bolt shear resistance, and Ft,Rd is the bolt tensile
80 WALD ET AL.

F F

F
F
1,4 − 1
1,4
F F F
k

δ .δ

.δ
δ δ δ
δ

δ
δ

F I G U R E 2 Bolt tension force as function of deformation in

FIGURE 1 Force deformation diagram of the bolt in shear
tension and in shear

resistance. A condition limiting the bolt resistance is shown in

Figure 2.
The tensile and shear behavior of the bolt in a numerical model
are represented by a bilinear spring model. The nonlinear spring has
special behavior in the interaction of the shear and tension. The shear
and tension forces are presented as nonlinear functions of shear and
tension deformations composed of ruled surfaces, see Figures 1 and
2, where δs is the bolt shear deformation, δt is the bolt deformation in
tension, δel is the elastic limit of the bolt deformation, δRd is the bolt
deformation at its resistance, Fel is the bolt force at the elastic limit of
deformation, and FRd is the bolt resistance. The functions respect the
limit condition of interaction which is showed above. It is clear that
F I G U R E 3 Constraint between the weld solid element and mesh
the bolt can be in three states: a linear behavior, a plastic state in ten-
nodes of plates modeled as shells
sion and a plastic state in interaction of the tension and the shear.
In design numerical model is simulated the component preload design weld model is verified for resistance according to relevant codes.
bolt either as nonlinear spring using the preloading force and defor- For each code an appropriate design model is selected. The plastic strain
mation or by restrains between surfaces, which are representing the in steel plates connected by welds is normalized to 5% to confirm with
friction, with a spring, which is including a preloaded force in the bolt. the maximum plastic strain of plates in joint.
For the bolt component represented by the preloading force and its
slip deformation is the model similar to a conventional bolt model.
3 | V ER I F I C A TI O N O N B L OCK S H EA R
FA I LU RE
2.3 | Welds
3.1 | Analytical models
The butt welds are included in model as restrains between plates and in
design treated as basic material. For fillet weld a shell equivalent element The block shear failure occurs in cases of not optimal design due to
is added between the plates. The stresses are calculated in the throat the structural geometry. It is the potential failure mode for gusset
section the same way as in the preferred analytical model. The equivalent plates, fin plates, coped beams, single, or double angles and tee con-
weld element is inserted with the corresponding weld dimensions as nections, where significant tension/shear forces are present, as there
shown in Figure 3, the element respects the weld throat thickness, posi- is not enough space to comply with recommended pitches and end
tion and orientation. The nonlinear material analysis is applied and distances of bolts. The block shear failure develops by yielding along
elastoplastic material model of steel is considered. The stress peaks are the shear planes and rupture on the tensile plane in a first mode. In
redistributed along the weld length. The aim of design weld model is not the second mode, yielding along the tension plane and a rupture along
to capture stresses inside the weld, heat affected zone and parent mate- the shear plane may occur, but these cases are quite rare, due to
rial but to distribute forces between plates and to allow the design check. smaller ductility of steel in tension compared to the ductility. Block
The material difference across the weld cross-section, the heat affected tearing is present in bolted connections because of the reduced
zone, the residual stresses and the weld shrinkage are neglected. The section area. It can also be found in welded joints.
WALD ET AL. 81

The most common way of designing against block shear failure is

using the analytical models, which are developed since 195216 and
reviewed for current technology in 2005.17 Advantage of these
models is that they can be used in most cases and are the easy to
apply, see ENJanuary 8, 1993:2006,5 AISC 360-10,16 CSA S16-09,17
and prEN January 8, 1993,18 The resistance is calculated at both criti-
cal planes, which are exposed to shear and to tension. For concentric
loading the resistance is derived in ENJanuary 8, 1993:20065 as
F I G U R E 4 Failure planes during the block shear failure; Agt is
gross tensile plane, Ant is net tensile plane, Agv is gross shear plane,  
f u Ant 1
and Anv is net shear plane V eff,1,Rd = + pffiffiffi f y Anv =γ M0 ð5Þ
γ M2 3

F I G U R E 5 Experiments T1 and
T2 by Hanus et al.21

FIGURE 6 The fine mesh around the bolt holes and simplified body of bolt in ROFEA22
82 WALD ET AL.

In case of eccentric loading according to prEN 1993-1-8,20 ϕRn = ϕð0,6 f u Anv + Ubs f u Ant ≤ 0,6 f y Agv + Ubs f u Ant Þ ð9Þ

  
Agv f y Anv f u where fy is yield strength, fu is ultimate strength, γ M2, ϕ, ϕu are safety
V eff,2,Rd = 0,5Ant f u + min pffiffiffi ; pffiffiffi =γ M2 ð6Þ
3 3 factors. The gross tensile plane Agt, the net tensile plane, Ant the gross
shear plane Agv and the net shear plane Anv, see Figure 4. In CSA
in eccentric loading according to EN1993-1-8:2006,5 S16-09,19 Ut = 1.0 for concentric and 0.5 for eccentric loading
applied in
 
f u Ant 1
V eff,2,Rd = 0,5  + pffiffiffi f y Anv =γ M0 ð7Þ  
γ M2 3 fy + fu
T r = ϕu Ut Ant f u + 0:6Agv ð10Þ
2

in concentric loading prEN 1993-1-8.20

  
Agv  f y Anv f u
V eff,1,Rd = Ant f u + min pffiffiffi ; pffiffiffi =γ M2 ð8Þ
3 3 3.2 | Validation

In AISC 360-10,16 combination factors Ubs = 1.0 for concentric The experiments by Huns et al.21 are used for validation of ROFEA
and 0.5 for eccentric loading respectively are applied in the following prepared by Sekal22 in ANSYS software. The geometries of tested
equation. specimens are shown in Figures 4 and 5. The tested gusset plates are
6.6 mm thick, the bolts have diameter 19.05 mm and the bolt holes
are match drilled therefore the bolts are directly in bearing.
True stress-strain material diagram is used for material model. A
hexahedron element type is used, see Figure 6. Only the thinnest plates
which tend to fail are modeled. Bolts are bearing on the half-circle of the
bolt hole only. The displacements in all holes are coupled. To save on
computational time, the dense mesh is used only in areas where a high
gradient is expected—locally, around the bolt holes with a radially struc-
tured grid. The model contains of 190 264 hexahedron elements and
need around 26 hours of computational time on TUM server. In final
stage the hole shows large necking on the tension plane, see Figure 7.
Despite some differences between numerical simulation and
FIGURE 7 Deformed bolt hole, necking on a tension plane 22
physical tests, there is a good rate of compliance in tested parameters,

(a) Specimen T1 (b) Specimen T2

FIGURE 8 Validation of ROFEA on global deformation

WALD ET AL. 83

TABLE 1 Validation of specimens T1

Specimen T1 Specimen T2
and T222
Compared Experiment ROFEA Exp./ROFEA Experiment ROFEA Exp./ROFEA
Nj,ini [MN/m] 1717.7 2153.4 0.80 1846.4 1773.1 1.04
Fpeak [kN] 691.2 698.1 0.99 756.0 742.9 1,02
F2 mm [kN] 586.2 580.1 1.01 639.9 643.3 0.99
F0.05 [kN] N/A 486.3 N/A N/A 421.0 N/A
F2/3 [kN] 663.0 654.5 1.01 727.0 726.5 1,00

F I G U R E 9 Verification of the 800

Force [kN]
DOFEA (CBFEM) to the ROFEA on
specimen T222 700

600

500
5% strain limit on CBFEM model
400

300
CBFEM
200
ROFEA
100 EN1993-1-8: 2020
0

Deformation [mm]

F I G U R E 1 0 Sensitivity study of
block shear resistance for bolt pitch22

see Figure 8. Maximal difference between the compared values is plane of bolt holes etc. The good correspondence between numerical
20% in case of initial stiffness of T1 model see Table 1. For initial stiff- simulation and experiment is also thanks to the information and data
ness, the higher rate of difference is acceptable because the value is obtained from the paper,21 where authors carried validation
dependent on factors like whether the bolts were really in bearing themselves.
from the beginning of the physical test or whether the size of ele-
ments around bolt holes in case of ROFEA was appropriate. However,
besides this one value, the other compared parameters vary till 4% so 3.3 | Verification
the models can be used for the further CBFEM design-oriented model
verification. Also, it is worth to note that on a deformed numerical Design oriented CBFEM model uses shell elements with coarse mesh.
model, signs typical for block shear failure mechanism could be The mesh is predefined near bolt holes. Bolts are modeled as
observed, such as necking on a tension plane or yielding on the outer nonlinear springs which are connected to the nodes at the edge of the
84 WALD ET AL.

F I G U R E 1 1 Sensitivity study of
block shear resistance for plate
thickness22

prediction of resistance against block shear failure in case of sym-

metrical loading.

4 | V ER I F I C A TI O N O N B A SE P L A TE

4.1 | Experimental program

During the design of column base plates subjected to 3D loading con-

ditions with a flexible base plate, the concrete under the base plate is
checked in crushing as well as the anchoring system, which is also
strongly influenced by flexible base plate behavior.4 A recent complex
testing program was carried out at the laboratory of INSA Rennes that
can be found in more detail in.24 The set of six tests on column base
plates subjected to pure bending (in-plane, M0; out-of-plane, M90;
and biaxial bending moment, M45) allowed to investigate the influ-
FIGURE 12 Verification of CBFEM model to analytical one22
ence of the base plate thickness and the bending moment orientation
on the global behavior. The testing program was complemented with
bolt holes by links. The bilinear material diagram with negligible strain- a large numerical study focused on the behavior of column base plates
hardening is used for plates. The limit resistance of a group of bolts in under a combination of uniaxial or biaxial bending moment with axial
bearing is determined when the plastic strain at the plate reaches 5%. tensile/compressive force. This study allowed to obtain further infor-
The bearing and hole tear-out resistances of each individual bolt are mation about the compressive area under the flexible base plate, see
checked by formula in appropriate code. Figure 13.
The comparison of prediction by CBFEM to ROFEA and analytical
models is shown in Figure 9. The ROFEA model disregards any slip
but in CBFEM, the shear model of bolts is according to standard 4.2 | Research oriented finite element analysis
approximated with the assumption of the regular bolt holes.
Numerical model was created with the FEA software ABAQUS for
further investigation of the behavior of the tested column base
3.4 | Sensitivity study plates. The main elements of the parts: base plate, column, welds,
and concrete were modeled using 3D deformable solids, except for
Geometry of specimen T1 was used for the study the block shear the concrete reinforcement that was modeled using two-node lin-
resistance by changing the bolt pitch, Figure 10, and plate thick- ear beam elements. The actual dimensions of the testing specimens
ness, Figure 11. The models provide expected results. The EN were used on the finite element models. The models consider non-
5
January 8, 1993:2006 is the most conservative, followed by linearities from geometry, material and contact interactions. Due to
CBFEM, prEN 1993-1-820 and AISC 360-10.16 The sensitivity the asymmetric applied load, biaxial bending moment, the full test-
24
study is summarized in Figure 12, which shows a constant ing device described in Reference was modeled. In order to
WALD ET AL. 85

F I G U R E 1 3 Position of the Centre of

compression zc for M0, M90 and M45
loading cases

FIGURE 14 Meshing of the finite element analyses F I G U R E 1 5 Yield line patterns obtained by the numerical
simulations, specimen SPE1-M45 with the base plate
thickness 10 mm

reduce the computing time needed for the simulation, different

mesh sizes are adopted according to the importance of the ele- on specimens SPE1-M45 and SPE2-M45 is presented in Figure 16.
ments, as in Figure 14. A refined mesh was adopted for the anchor A comparison of the moment-rotation curves is described in
bolts in order to simulate the complex stress distribution during the Figure 17.
computation until failure. To produce reasonable and physical In general, the bending moment and the initial rotational stiffness
sound stress distributions at the interface between the anchor obtained from the numerical simulations for all specimens are in good
bolts and the concrete, a smaller mesh size was selected for the agreement with the experimental data. The differences are satisfac-
layer of concrete material in contact with the anchor bolts. For the tory, with an error less than 7%. The value of the computed initial
remaining concrete material, a coarser mesh was considered as this rotational stiffness for specimen SPE2-M45 has an error not exceed-
zone was of little interest. In regions with high stress concentra- ing 30%, which is acceptable. This discrepancy between the results
tions, such as welds, mesh size had to be reduced. Similarly, in obtained experimentally and numerically is caused by the simplified
regions where buckling and/or bending were expected such as col- and rather conservative approach used to model the adherence τmax
umn flanges and base plates, the number of elements was assigned between the concrete and the anchor bolts. This has a significant
depending on the thickness of the plates. Yield line patterns influence on the initial stiffness Sj,ini. Also, the fact that the initial geo-
obtained by the numerical simulations of specimen SPE1-M45 with metrical imperfections of the base plates are not considered in the FE
the base plate thickness 10 mm is shown in Figure 15. models has an impact on the way the connection deform in the elastic
regime and consequently, on the stiffness and plastic bending
moment values.
4.3 | Validation

A comparison of the plastic bending moment as well as the initial 4.4 | Sensitivity study
rotational stiffness obtained from the numerical model and the
experimental tests is given in Table 2. The numerical/experimental The influence of the base plate thickness is analyzed in terms of fail-
bending moment resistance is obtained using the inter- ure mode, initial stiffness and resistance for a base plate represented
section between initial Sj,ini and second stiffness Sj,ini/10. Validation in Figure 18, with 10 mm (specimen P5) and 20 mm (specimen P6)
of DOFEA on yielding of the base plate and failure of anchor bolts base plate thickness.
86 WALD ET AL.

TABLE 2 Comparison of the plastic resistances and initial stiffness from the numerical models and experimental tests

Experimental Numerical Exp./Num.

Initial rotational stiffness Plastic bending moment Initial rotational stiffness Plastic bending moment
Test Sj,ini,exp (kNm/rad) Mj,pl,exp (kNm) Sj,ini,num (kNm/rad) Mj,pl,num (kNm) Sj,ini Mj,pl
SPE1-M0 4117 37.7 4163 35.1 0.99 1.07
SPE2-M0 7189 42.9 7536 46.2 0.95 0.93
SPE1-M90 2028 28.2 2061 26.4 0.98 1.07
SPE2-M90 3120 35.9 2712 37.9 1.15 0.95
SPE1-M45 2440 32.0 2853 32.5 0.86 0.98
SPE2-M45 3205 36.6 4571 38.2 0.70 0.96

FIGURE 16 Validation of DOFEA on yielding of the base plate and failure of anchor bolts on specimens SPE1-M45 and SPE2-M45

(a) Specimen SPE1-M45 (b) Specimen SPE2-M45

50 50 Moment
Moment M (kNm)
M (kNm)
40 40

30 30

Experimental test Experimental test

20 20
Numerical simulation Numerical simulation

10 10
Rotation Rotation
θj (mrad) θj (mrad)
0 0
0 20 40 60 80 0 10 20 30 40 50 60

FIGURE 17 Validation of ROFEA on moment-rotation curves for specimens SPE-M45 in test series

Results from the FE computations show on Figure 19 an increase the resultant forces from the compressive and tensile sides. In contrast
in the plastic bending moment Mj,pl and the maximum bending to this, as it was observed during the experimental tests and numerical
moment Mj,u of the connection resistance with increasing values of simulations as well, highly deformable thin base plates, 10 mm thick-
the base plate thickness. This is explained by the fact that increasing ness, produce a rather non-uniform stress distribution. Consequently, a
the base plate thickness leads to an increase in the lever arm z between high concentration of stresses is located near the column flange and
WALD ET AL. 87

the lever arm is reduced. The increase in the base plate thickness also 4.5 | Verification
affects the initial stiffness of the connections. Although less pro-
nounced in some cases than in others, values of Sj,ini are greater for In the graphs in Figures 20 to 22, the results of CBFEM25 are com-
thicker base plates due to a greater lever arm and lower deformation of pared to ROFEA. The main differences between both models are the
the base plate. In terms of failure modes, it was concluded that the base material model, mesh density, and concrete and anchor models. The
plate thickness has no particular influence on this parameter. ROFEA uses multilinear material diagram with strain-hardening and
plastic strain up to 15%, which is the minimal value for structural steel
in EN January 1, 1993:2005. On the other hand, CBFEM model has
bilinear material diagram with insignificant strain-hardening; the plas-
tic branch has the slope of E/1000. The plastic strain is limited to 5%
according to EN January 5, 1993:2005. Other differences are the
finite element model of the concrete and the anchors. CBFEM uses
simplified assumptions under the base plate by Winkler's subsoil
model and a single spring as the anchor. The anchor bilinear spring
has its resistance determined according to EN January 8, 1993:2006
and therefore, its resistance is decreased by the factor 0.9. The plastic
branch starts at the tensile force equal to the anchor tensile resistance
multiplied by the ratio of the anchor yield to ultimate strength. The
FIGURE 18 Geometry of the base plate for the sensitivity study anchor tensile resistance is reached at the quarter of elongation at
rupture; in this case at the strain 2.5%. Due to the different material
model, lower resistances by CBFEM are expected. Configuration P1 is
used in the study with all combinations of loading investigated by
ROFEA. Configuration P1 is the same as P5, see Figure 20, except for
the bolt diameter 16 mm instead of 20 mm. The CBFEM model is
loaded by normal compressive or tensile force and a bending moment
around stronger axis, see Figure 21, or weaker axis, Figure 22.
The model for biaxial bending is loaded by equal bending moment
around both stronger and weaker axes. This does not fully correspond
with the loading at an angle 45 because the stiffness around both
axes is different. It is the assumption taken for ROFEA. The results are
summarized in Figure 22.
The resistances of CBFEM are smaller than ROFEA which is cau-
sed by steel of plates without strain hardening and the reduction fac-
tor 0.9 for the tensile strength of anchors. The strain limit is also
lower in CBFEM 5% for plates and 2.5% for anchors. In all cases,
F I G U R E 1 9 Moment-normal force interaction for biaxial bending CBFEM provides more conservative results compare to ROFEA. The
moment calculated by ROFEA for specimens P5 and P6

F I G U R E 2 0 Comparison of
prediction of resistance (R) by
CBFEM and ROFEA, Bending around
major axis, specimen P1-M0
88 WALD ET AL.

F I G U R E 2 1 Comparison
prediction of resistance (R) by
CBFEM and ROFEA, Bending around
minor axis, model P1-M90

F I G U R E 2 2 Comparison of
prediction of resistance (R) by CBFEM
and ROFEA, Bending around both
axes (model P1-M45)

most significant difference is in the case of biaxial bending and combi- connections by shell models. The tools and solvers are ready.27 Build-
nation of compressive force with bending moments. ing Information Modeling allows shearing of data between drawing
tolls and FEA software and is used as input and output for design
check and machinery.
5 | C O N CL U S I O N The multilevel FEA analyses of the structural steel connections
based on components (CBFEM) is replacing the curve fitting and com-
The global analyses of steel structures are today carried out by FEA ponent method designs. For its proper and safe use, it is necessary to
and the traditional procedures are not used any more. The prepared apply a good Validation & Verification procedure with well-defined
CBFEM model helps visually reveal all possible failure modes and is hierarchy. The well verified models are able to predict the behavior of
complementing the traditional engineering judgment by good predic- generally loaded complex connections with the accuracy, which is
23
tion of joint behavior and safe standard approved resistance. In shown on Benchmark cases prepared in complex hierarchy to check
coming generation of structural steel Eurocodes, it is expected to the software inputs and outputs and its application. The presented
stress the principles of FEA design and its safe use in prEN1993-1-14: results for block shear and base plate show the good accuracy of
Design by finite element26 by emphasizing the System response quan- CBFEM verification to analytical models. In cases, where the CBFEM
tity process separately for numerical experiments and design models. gives higher stiffness, resistance or deformation capacity it needs to
The strain limits need to be studied based on safety of FEA applica- be verified to ROFEA.
tions. For even complex joint presented here, 5% leads to good and
safe prediction. The CBFEM model complies well with validated ACKNOWLEDG MENTS
ROFEA especially in terms of resistance. The work was prepared under the R&D project MERLION II
The coming solutions are the geometrically and materially supported by Technology Agency of the Czech Republic, Project
nonlinear analysis with imperfections of members with their No. TH02020301.
WALD ET AL. 89

ORCID 14. Dvorkin EN, Bathe KJ, Continue A. Mechanics based four node Shell
František Wald https://orcid.org/0000-0003-2416-8951 element for general nonlinear analysis. Eng Comput. 1984;1:77–88.
15. ISO 898-1. Mechanical properties of fasteners made of carbon steel
and alloy steel, Part 1, Bolts, screws and studs with specified, prop-
RE FE R ENC E S erty classes, Coarse thread and fine pitch thread, Geneva; 2009.
1. Zoetemeijer P. Summary of the researches on bolted beam-to-column 16. Whitmore RE. Experimental Investigation of Stresses in Gusset
connections, Report 6-85-7. Delft: University of Technology; 1985. Plates, Bulletin No. 16, Engineering Experiment Station, University of
2. Steenhuis M, Gresnigt N, Weynand K. Pre-design of semi-rigid joints Tennessee; 1952.
in steel frames. Proceedings of the Second State of the Art Workshop 17. Driver RG, Grondin GY, Kulak GL. Unified block shear equation for
on Semi-Rigid Behaviour of Civil Engineering Structural Connections, achieving consistent reliability. J Constr Steel Res. 2006;62:
COST C1, Prague; 1994, pp. 131–140. 210–222.
3. Jaspart JP. Design of structural joints in building frames. Prog Struct 18. A360-10. American National Standard—Specification for Structural Steel
Eng Mater. 2002;4:18–34. Buildings, American Institute of Steel Construction, Chicago; 2010.
4. Wald F, Sokol Z, Steenhuis M, Jaspart JP. Component method for 19. CSA S16-09. Steel Frame Design Manual, Computers & Structures,
steel column bases. Heron. 2008;53:3–20. Inc; 2016.
5. EN1993-1-8. Eurocode 3: Design of steel structures—Part 1-8: 20. prEN 1993-1-8. Eurocode 3: Design of steel structures—Part 1-8:
Design of joints, CEN, Brussels; 2006. Design of joints, Final draft, CEN, Brussels; 2019.
6. EN1994-1-1. Eurocode 4: Design of composite steel and concrete 21. Huns BBS, Grondin GY, Driver RG. Block shear behaviour of bolted
structures—Part 1-1: General rules and rules for buildings, CEN, Brus- gusset plates. Edmonton, Canada: Department of Civil and Environ-
sels, 2010. mental Engineering, University of Alberta, 2002.
7. Da Silva Simoes L. Towards a consistent design approach for steel 22. Sekal D. Analysis of block shear failure, Diploma Theses, TU Munich
joints under generalized loading. J Constr Steel Res. 2008;64: and CTU in Prague; 2019.
1059–1075. 23. Wald F et al. Benchmark cases for advanced design of structural steel
8. Rasmussen KJR, Zhaob X, Yana S, Daia L, Zhua Ch, Jianga L. Recent connections. Česká technika: Praha, 2019.
developments of the component method. Nordic Steel Construction 24. Da Silva Seco L. Column base plates under 3D loading, Doctoral the-
Conference, Copenhagen; 2019. ses, INSA Rennes; 2019.
9. Jaspart JP, Weynand K. Design of hollow section joints using the 25. IDEA Connection, Version 10.1. Theoretical background; 2019, IDEA
component method. Tubular Structures XV. Rio de Janeiro, . Balkema: StatiCa®. https://resources.ideastatica.com/.
Proceedings of the 15th international symposium on tubular struc- 26. prEN1993-1-14. Eurocode 3, Design of steel structures—Part 1-14:
tures, ISTS15;2015; p. 403–410. Design by finite element, First draft, CEN, Brussels; 2019.
10. EN 1993-1-5. Eurocode 3: Design of steel structures—Part 1-5: 27. Wald F, Vild M, Vesecký J, Legner Š, Kabeláč J, Šabatka L. Multi-level
Plated Structural Elements, CEN, Brussels; 2005. joints and element design. ce/papers. 2019;3:379–384. https://doi.
11. Wald F, Kwasniewski L, Gödrich L, Kurejková M. Validation and verifi- org/10.1002/cepa.1070
cation procedures for connection design in steel structures. 12th
International Conference on Steel, Space and Composite Structures,
Singapore; 2014, pp. 111–120.
12. Gödrich L, Wald F, Kabeláč J, Kuříková M. Design finite element How to cite this article: Wald F, Vild M, Kuříková M, et al.
model of a bolted T-stub connection component. J Constr Steel Res. Component based finite element design of steel joints. Civil
2019;157:198–206.
Engineering Design. 2020;2:78–89. https://doi.org/10.1002/
13. Ibrahimbegovic A, Taylor RL, Wilson EL. A robust quadrilateral mem-
brane element with drilling degrees of freedom. Int J Numer Methods cend.202000015
Eng. 1990;30(3):445–457.