You are on page 1of 10

EUROSTEEL 2017, September 13–15, 2017, Copenhagen, Denmark

Behaviour of bolted endplate connections of beams subjected to bi-


axial bending moments and axial forces
Maël Couchauxa and Anthony Rodierb
a
Institut National des Sciences Appliquées de Rennes, France
mael.couchaux@insa-rennes.fr
b
Centre Technique Industriel de la Construction Métallique, France
arodier@cticm.com

ABSTRACT

In recent years, the use of finite element analysis software even more efficient and the construction
of more complex steel structures favoured 3D modelling. After performing the global analysis,
designers often get forces and moments in three dimensions at the location of endplate connections,
either for beam-to-beam joints or column bases. Eurocode 3 part 1-8 only provide method to check
bolted endplate connections for the combination of an in-plane bending moment and an axial force.
The out-of-plane bending moment is not considered. A research has been conducted to provide a
design method, based on the component method of EN 1993-1-8, to assess the resistance of
endplate connections subjected to in-plane and out-of-plane bending moments, and an axial force.
The objective of the present paper is to present the results obtained for endplate connections of
beams. The conservative aspect of the method was validated by comparison with numerical results.

Keywords: endplate, bi-axial bending, bolt

1 INTRODUCTION
Eurocode 3 Part 1-1 provide design rules for beams subjected to the combination of bi-axial
bending moments and axial force, whereas Part 1-8 do not provide those for connections. Moreover,
use of finite analysis software even more efficient and the construction of more complex steel
structures favoured 3D modelling. Designers have to check these connections under a weak axis
bending moment, without practical method provided by Eurocode 3.
During these last years, several research’s focused on the 3D behavior of bolted endplate
connections. Heinisuo et al ([2], [3]) proposed an extension of the component method to take into
account loadings in the three directions that need software implementation. Neumann ([4], [5])
proposed a method, also based on the component method, to determine the rotational stiffness and
resistance for an out-of-plane bending moment of beam to column endplates connections. Failure
can develop on endplate, bolts, beam but a new failure modes is added that correspond to torsion of
the column web and flanges. The main problem concerns the evaluation of the centre of
compression determined numerically in this study. The interaction between bending moments is
firstly assumed linear ([4], [5]) then quadratic [7]. In addition, the torsion of web and flange of the
column have been highlight experimentally [8] and numerically [9].
These previous research mainly studied beam to column connections composed of I/H sections
whose the behaviour for out-of-plane bending moment (and thus torsion on the column) is limited
due to flexibility of I/H section in torsion. The authors of the present paper prefers focusing on
beam to beam bolted connections able to transfer non negligible out-of-plane bending moments.
In this paper, an analytical model is developed to obtain the resistance and the rotational stiffness of
bolted endplate connections of I/H beams under bi-axial bending moment and axial force. This
model is based on Eurocode 3 principles and its components method. It allows to evaluate the
connection characteristics under an out-of-plane bending moment. The position for the compression
resultant is found by mechanical model, while previous works did by numerical simulation. Finally,
a safe linear interaction is used between axial force, in-plane and out-of-plane bending moments.
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · ce/papers 1 (2017), No. 2 & 3
https://doi.org/10.1002/cepa.49 wileyonlinelibrary.com/journal/cepa 185
186 |

These results have been compared to finite elements analyses on ANSYS software, previously
validated by confrontation to experimental tests.

2 FINITE ELEMENT MODEL


2.1 Presentation of the finite element model
The numerical model was built using the Finite element code ANSYS V14.0 with contact and brick
elements and is similar to that developed for bolted circular flange connections [6].
For bolts, a constant cross-section equal to the effective cross section area, As, is considered over
the entire length. Two types of contact elements are also used: a) Flexible contact elements between
the endplate and bolts and b) Rigid contact elements between the endplate and the plane of
symmetry (see Fig. 1). An isotropic Coulomb friction law is used to reproduce sliding and sticking
conditions with a slip factor equal to 0,2. Friction is neglected between the two endplates because of
the symmetry.

σ (N/mm2)

fu
Eu
fy

Collapse : σr=10
εr= εu+1
E

εy εh εu ε (%)
b)
a)
Fig. 1. Joint modelling a) Mesh b) stress-strain curve
The stress-strain relationship for steel (endplate, beam and bolts) is assumed to be multi-linear (see
Fig. 1). Large deformations are also considered. As soon as the deformation level reaches εu, the
stress drops to 10 N/mm2 in order to model the failure of the element. This phenomena leads either
to a drop of the force applied to the joint or to the termination of the analysis. This state is assumed
to be the ultimate state of the joint. The criterion of the yielding surface is Von-Mises. In the
parametric study, nominal mechanical characteristics of steel are used and εu is equal to 15 % for
beam/endplates and 10 % for bolts. The end of the beam is linked to a pilot node loaded in
displacements and force.
2.2 Confrontation to experimental tests
This Finite Element model have been confronted to experimental results of Ungermann and
Schmidt [10] on endplate connections composed of four bolts per row subject to in-plane bending
moment. Specimens VT1 to VT4 have been studied (see [10] for mechanical and geometrical
characteristics) and main results are presented in Table 1. Results obtained by tests and finite
element analysis are in good agreements.

Table 1 Comparisons of results obtained for tests of Ungermann and Schmidt [10]
| 187

Tests Numerical
Specimen Sj,test Mu,test Failure Sj,num Mu,num Failure
kNm/mrad kNm - kNm/mrad kNm -
VT1a 329 Bolt
60 52 330 Bolt
VT1b 320 Bolt
VT2a 251 Bolt
38 31 238 Bolt
VT2b 253 Bolt
VT3a 500 Bolt
122 100 507 Bolt
VT3b 478 Bolt
VT4a 366 Bolt
78 62 363 Bolt
VT4b 352 Bolt

3 PARAMETRIC STUDY
A parametric study has been performed on 27 symmetric bolted endplate connections of I beams
(see Fig. 2). Bolting arrangements are the following:
• External bolt rows only (specimens 1 and 2);
• External and internal bolt rows (specimens 3, 6 and 8);
• External, internal and central bolt rows (specimens 4, 5, 7 and 9);
• Internal bolt rows only (specimens 10);
• Internal and central bolt rows (specimens 11).
Geometrical and mechanical characteristics are respectively given in Table 2 and Table 3. This
study focus on the effect of the following parameters: bolting arrangement, endplate thickness, and
beam’s height.
bp

ex tp
External p2
e1x tfb
p11
rb
Internal
p12

hp hb
Central

twb

bfb

Fig. 2. Geometry of endplates

All specimens have been studied for in-plane and out-of-plane bending moments. The maximum
values correspond to the ultimate bending moment and are respectively denoted Mip,u,num and
Mop,u,num. The plastic values are also calculate using ECCS procedure [11]. These plastic moments
are respectively denoted Mip,pl,num and Mop,pl,num.

Table 2 Mechanical and geometrical characteristics of endplates


188 |

tp bp hp ex e1x p2 p11 p12


Specimens Bolt rows Grade Bolts
mm mm mm mm mm mm mm mm
1-ep10 Ex S355 10 220 760 40 40 150
1-ep20 Ex S355 20 220 760 40 40 150
1-ep30 Ex S355 30 220 760 40 40 150
-
2-ep10 Ex S355 10 220 760 40 40 100
2-ep20 Ex S355 20 220 760 40 40 100 -
2-ep30 Ex S355 30 220 760 40 40 100
3-ep10 Ex-In S355 10 220 760 40 40 150 100
M22-
3-ep20 Ex-In S355 20 220 760 40 40 150 100
8.8
3-ep30 Ex-In S355 30 220 760 40 40 150 100
4-ep10 Ex-In-Cent S355 10 220 760 40 40 100 100 70
4-ep20 Ex-In-Cent S355 20 220 760 40 40 100 100 70
4-ep30 Ex-In-Cent S355 30 220 760 40 40 100 100 70
5-ep10 Ex-In-Cent S355 10 220 760 40 40 100 100 70
5-ep20 Ex-In-Cent S355 20 220 760 40 40 100 100 70
5-ep30 Ex-In-Cent S355 30 220 760 40 40 100 100 70
6-ep10 Ex-In S235 10 190 560 40 40 90 90
-
6-ep20 Ex-In S235 20 190 560 40 40 90 90 M20-
7-ep10 Ex-In-Cent S235 10 190 560 40 40 90 90 70 8.8
7-ep20 Ex-In-Cent S235 20 190 560 40 40 90 90 70
8-ep10 Ex-In S235 10 160 460 40 40 90 80
-
8-ep15 Ex-In S235 15 160 460 40 40 90 80
9-ep10 Ex-In-Cent S235 10 160 460 40 40 90 80 60
9-ep15 Ex-In-Cent S235 15 160 460 40 40 90 80 60 M16-
10-ep10 In S235 10 160 320 90 10.9
-
10-ep15 In S235 15 160 320 90 -
- -
11-ep10 In-Cen S235 10 160 320 90 60
11-ep15 In-Cen S235 15 160 320 90 60
Table 3 Properties of beams

hb bfb tfb twb rb


Specimens Grade
mm mm mm mm mm
1-ep10 to 5-ep30 S355 600 200 18 8 18
6-ep10 to 7-ep20 S235 400 180 13,5 8,6 15
8-ep10 to 11-ep15 S235 300 150 10,7 7,1 15

Contact pressures under out-of-plane bending for specimen 4-ep10 are shown in Fig. 3. The contact
area is larger during the plastic stage than during elastic stage.
The parametric study shows that central bolt rows have no significant effect on the connection
behaviour under out-of-plane bending moment. External and internal rows transfer mainly the
tensile forces.
Regarding the interaction between the two bending moments, the parametric study show a quadratic
interaction when endplate failure in bending arise (see specimen 4-ep10 in Fig. 4). Nevertheless, the
interaction curve is closer to a bi-linear interaction in presence of bolt failure in tension (see
specimen 4-ep30 in Fig. 4).
| 189

a) b)

Fig. 3. Contact pressures under out-of-plane bending for specimen 4-ep10: a) Elastic stage; b) Plastic stage

Fig. 4. Interaction curve in bi-axial bending moment for specimen’s 4-ep10, 4-ep20 and 4-ep30

4 ANALYTICAL MODEL
4.1 Introduction
The main goal of the analytical model developed hereafter is to describe the behaviour of endplate
connection under out-of-plane bending moment. The initial rotational stiffness is described in 4.2,
and the resistance in 4.3. The rotational stiffness calculation is based on the component method,
whose the position of the neutral axis is defined regarding the relative stiffness of tension and
compression parts.
Parametric studies show that central bolt rows doesn’t have significant influence on the out-of-plane
behaviour of the connection, so the model neglected them. This simplification leads to an under-
estimation of the connection resistance for special arrangements (compact beams with thick
endplate and small bolts). For in-plane bending moment, Eurocode 3 part 1-8 [1] is fully applied. A
linear interaction is simply used in case of biaxial bending, and also in case of axial force.
4.2 Out-of-plane stiffness
The cross section at the junction between endplate and beam is assumed rigid and remains plane
under an out-of-plane bending moment, leading to a rotation φj,op/2 on each side of the connection
(see Fig. 5) and a total rotation of the joint φj,op. Linear springs are used to model the interaction
between the joint and the beam, a punctual one for bolts in tension, and a distributed one in the
compression zone. Plane sections assumptions leads to an elastic linear distribution of forces in
compression zone.
190 |

Mj,op

FC FT

fc h
p2
δc/2 δT/2 φj,op/2
kc
bfb kT
δT/2 φj,op/2
tfb δc/2 xc

Fig. 5. Model for out-of-plane behaviour of the connection in elastic stage

Due to compatibility in displacement, the following relationship can be written:


δc δT
= (1)
xc h − xc

The force-displacement relation in tensile and compressive parts are:


FT = kT δ T (2)
f c = kc δ c (3)
fc is the maximal compressive force per unit length.

The linear distribution of compression force leads to:


FC = xc f c (4)

As a pure bending moment is applied to the joint, the compressive and tensile forces are:
M j,op
FC = FT = (5)
h − xc / 3
Thus inserting Eq. (2) to (5) in (1), one obtain:
2
x  x
αc  c  + c −1 = 0 (6)
h h
h kc
where αc is a stiffness factor defined by α c =
kT
The only positive solution for Eq. (6) is:
xc 1 + 4αc −1
= (7)
h 2α c
It can be highlight that xc is close to h/4, for practical cases.

The initial rotational stiffness under an out-of-plane bending moment is:


M j,op
S j,op,ini = (8)
φ j,op

The rotation is related to the displacements in compression and tension zones by the following:
δ + δc
tan (φ j,op ) ≈ φ j,op = T (9)
h
| 191

And thus, the initial rotational stiffness can finally be expressed as:
h ( h − xc / 3)
S j,op,ini =
1 1 (10)
+
kT xc kc

The stiffness for the compression zone is defined, as for bolted blank flange bolted connections [6],
by the following:
λ ( λ + 1)
kc = E t t (11)
2 λt + 1
where λt is the beam flange thickness, tfb, divided by the endplate thickness, tp.

For the stiffness of the tension zone, central bolt rows are neglected, only external and internal rows
are accounted. The stiffness of the tension zone can be expressed according to the T-stub model by:
kT = E keq (12)
keq is the equivalent stiffness coefficient, calculated as the sum of the stiffness coefficients for
external and internal bolt rows, respectively keff,ext and keff,int, determined according to Eurocode 3
part 1-8.

The initial rotational stiffness according to Eq. (10), denoted Sj,op,ini,ana and those obtained by
numerical studies, Sj,op,ini,num, are compared in Fig. 6. The analytical model shows an acceptable
accuracy, with a mean value for Sj,op,ini,ana/Sj,op,ini,num ratio equal to 1.09.

Fig. 6. Out-of-plane rotational stiffness – Comparison between analytical model and FEA

4.3 Out-of-plane bending resistance


Finite element studies show that the ultimate limit state can be raised in the tension zone and locally
in the compression zone by flange yielding. The contact zone during plastic stage seems larger than
during elastic stage, Eq. 6 cannot be simply used to estimate the contact area.
192 |

FT,pl,ext/2

FT,pl,int/2
Mj,op,pl
p2

FC,pl FT,pl
bfb
tfb FT,pl,int/2 h

FT,pl,ext/2

xpl
Fig. 7. Model for out-of-plane bending resistance

The yield stress is assumed uniformly reach under the flanges on a length denoted xpl (see Fig. 7).
This assumption is correct for flanges of class 1 or 2, but a triangular distribution is also needed for
class 3 flanges.
With this assumption, the total compressive force is:
FC,pl = 2 xpl tfb f y (13)
where fy is the yield stress of the attached beam

As a pure bending moment is applied to the joint, the compressive and tensile forces are equal. The
tensile force is defined as the sum of the tensile resistances of the external and/or internal bolt rows.
Central bolt rows, if any, are neglected.
FC,pl = FT,pl = FT,pl,ext + FT,pl,int (14)

The tensile resistances of bolt rows are calculated according to the T-stub model of Eurocode 3 part
1-8 [1]. For comparison with the numerical model, partial safety factors are not included.

Combining Eq. (13) and Eq. (14), it comes:


FT,pl
xpl = (15)
2 tfb f y

And the out-of-plane bending resistance is:


M j,op,pl = FT,pl ( h − xpl / 2 ) (16)

The plastic bending resistances according to Eq. (16), Mop,pl,ana, and those obtained by numerical
studies, Mop,pl,num, are compared in Fig. 8. The ultimate bending resistance, Mop,u,ana, is also
calculated according to Eq.(16) but using fu instead of fy in the resistance of failure modes 1, 2 and 3
of the tensile part. A comparison is performed in Fig. 8 to ultimate bending resistances obtain from
finite element analysis.
The analytical model shows an acceptable accuracy, with a mean value for Mop,pl,ana/Mop,pl,num ratio
equal to 0,85 and a mean value for Mop,u,ana/Mop,u,num ratio equal to 0,82. The model is on the safe
side, mainly because the central rows and membrane effect are neglected.
| 193

Fig. 8. Out-of-plane bending resistance – Comparison between analytical model and FEA

4.4 Interaction between biaxial bending moment and axial force


To be consistent with clause 6.2.7.2(2) of Eurocode 3 part 1-8 [1], a linear interaction between the
two bending moments and the axial force is simply assumed:
N M ip M op
+ + ≤ 1, 0 (17)
N u M ip,u M op,u
Nu and Mip,u are the resistances respectively to normal force and in plane bending moment
calculated according to Eurocode 3 part 1-8.

Comparisons of interaction curves obtained for specimens 4-ep10 and 4-ep30 at failure in presence
of in-plane and out-of-plane bending moments are shown in Fig. 9. The linear interaction is always
one the safe side, especially when failure modes corresponds to endplate in bending, see Fig. 9-a.
The non-linearity is not so pronounced in presence of bolt failure (see Fig. 9-b). Indeed, this
interaction curves can be clearly improved to obtain a better accuracy.

a) b)
Fig. 9. Interaction between biaxial bending moments: a) Spe 4-10 (endplate failure); b) Spe4-30 (bolt failure)
194 |

5 CONCLUSIONS
In this paper, a finite element model has been developed, in order to represent the behaviour of
endplate connection between I/H beams subjected to biaxial bending moments and axial force. This
model has been compared to experimental results, and then used to perform a parametric study that
shown that a linear interaction between the different internal forces is conservative in all cases but
can be necessary when bolt failure occurs.
An analytical model has been developed to determine the out-of-plane bending resistance of an end
plate connection. The results obtained for the plastic and ultimate stage are quite satisfactory.
Another model to calculate the initial rotational stiffness has also been proposed, which lead to
acceptable accuracy even a greater dispersion is obtained. Finally, the linear interaction curve
proposed is conservative and improvement also need to be performed.

REFERENCES
[1] EN 1993-1-8 (2005), Eurocode 3: Design of steel structures. Part 1.8: Design of joints.
[2] Heinisuo, M., Laine, V., Lehtimäki, E. (2009), Enlargement of the component method into
3D, Proceedings: Nordic Steel Constructional Conference 2009, Malmö, Sweden, September
2-4, 2009, pp. 430-437.
[3] Perttola H., Heinisuo M. (2012), Experimental study on flanged joints of tubular members
under biaxial bending, 7th International Worshop on Connections in Steel Structures
(Connections VII), Document CECM N°133, Timisoara, p303-318.
[4] Neumann N., Nuhic F., “Design of structural joints connecting H or I sections subjected to in-
plane and out-of-plane bending”, Proc. 6th European Conference on Steel and Composite
Structures (Eurosteel 2011), Vol. A, pp. 303-308, Budapest, 2011.
[5] Neumann N., Buzaljko M., Thomassen E., Nuhic F., “Verification of design model for out-of-
plane bending of steel joints connecting H or I sections”, Proc. 12th Nordic Steel Construction
Conference (Nordic Steel 2012), Oslo, 2012.
[6] Couchaux M., Behavior of bolted circular flange joints, PhD thesis of INSA of Rennes,
November 2010. (in French)
[7] Neumann N., “Design model for combined in-plane and out-of-plane bending of steel joints
connecting I and H sections”, Proc. 7th European Conference on Steel and Composite
Structures (Eurosteel 2014), Napoli, 2014.
[8] Gil B., Goni R., T-stub behaviour under out-of-plane bending. I: Experimental research and
finite element modelling, Engineering Structures, Vol. 98, 230-240, 2015.
[9] Gil B., Bijlaard F., Bayo E., T-stub behaviour under out-of-plane bending. II: Parametric
study and analytical characterization, Engineering Structures, Vol. 98, 241-250, 2015.
[10] Ungermann D., Schmidt B., Moment resistance of bolted beam to column connections with
four bolts in each row, Eurosteel 2005, Maastricht, The Netherlands, 2005.
[11] ECCS, recommended testing procedures for assessing the behavior of structural elements
under cyclic loads, European Convention for Constructional Steelwork, Technical Committee
1, TWG 13 – Seismic Design, No45, 1986.

You might also like