You are on page 1of 10

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

NUMERICAL ANALYSIS OF FULL STRENGTH HEAVY BEAM-TO-


COLUMN BOLTED CONNECTION
Damjan Liseca, Primož Može*,b
a,b
University of Ljubljana, Faculty of Civil and Geodetic Engineering, Ljubljana, Slovenia
a
primoz.moze@fgg.uni-lj.si

ABSTRACT
The design of full strength bolted beam-to-column beams considering the overstrength factor for the
beam material according to EN 1998-1 is particularly demanding, especially in case of heavy beam
profiles in combination with higher grade material (e.g. S355). The elastic design of the full strength
connection leads to very stocky joint. The component method presented in EN 1993-1-8 that was
developed mainly for the design of partial-strength joints may also be used for the design of these
connections, but its application is limited mainly to experts.
This paper studies bolted beam-to-column joints under monotonic and cyclic loading, particularly hot
rolled HEA 600, S355 beam with high bending strength. Two different configurations of bolted
connection of HEA 600 to column are studied in order to achieve full strength connection as defined
in EN 1998-1, namely eight-bolt rib stiffened and haunched stiffened joint configuration. Both joint
configurations are first calculated according to EN 1993-1-8 using component method, where bolt
forces and joint bending strength were determined. The first joint is also designed according to the
procedure given in ANSI/AISC 385-10. The numerical model, based on this design, was built in FE
software Abaqus in order to observe the behaviour under monotonic and cyclic loading. The cyclic
loading was defined according to the ECCS and ANSI/AISC 341-10 procedures. The bolt preload
was also studied.

Keywords: beam-to-column connection, overstrength, component method, seismic design

INTRODUCTION
Following the Northridge earthquake in 1994, where significant damages on steel structures were
recorded, numerous researches on inelastic behaviour of steel structures took place in the United
States. The result is, among others, a collection of prequalified connections, which are allowed for
use in steel structures for seismic applications, considering the limitations given [1]. The latter gives
an engineer a reliable and quick tool, which can be easily implemented in daily practice. On the other
side, there is no such tool in Europe yet. Furthermore, the variety of steel sections used in the USA
differs from that used in Europe. These problems were addressed within the European project
EQUALJOINTS, where beam-to-column steel connections and profile sections common in Europe
were numerically, analytically and experimentally evaluated on inelastic behaviour. The publication
of the project report is expected. As reported in [2], the strongest beam section, tested in the project
was IPE 600 (S355), which possesses plastic section modulus Wpl,y = 3512 cm3. The aim of this paper
is to numerically investigate the behaviour of connection with heavier beam sections e.g. HEA 600.
It is known that the ability to develop an inelastic rotation may be somewhat diminished as beam gets
deeper and heavier [3]. Furthermore, considering the concept of capacity design and overstrength in
combination with heavy beam section and higher grade material, the design of bolted connection in
seismic areas is particularly demanding, because all of the above greatly influences bolt forces and
end plate deformations. High strength preloaded bolts are especially sensitive part of the connection,
due to their brittle fracture once the tensile strength is reached [4]. Therefore, beam section HEA 600
(S355) (Wpl,y = 5350 cm3) in combination with M 39 10.9 bolts is investigated in this paper.

© 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.56 wileyonlinelibrary.com/journal/cepa 253
254 |
1 DESIGN OF THE CONNECTIONS
1.1 Rib stiffened configuration (RSC)
Eight-bolt extended rib stiffened end-plate configuration (RSC) (Fig. 1a) was first designed
according to the procedure described in [1] and later using component method described in [5].
Considering the overstrength factor, Eq. (1) gives the minimum required connection bending
resistance Mf, which was used in both cases.
M f  1.1 ov f ybW pl , y ,b  Vu Sh  1.1 1.25  35.5 kN cm 2  5350 cm3  951kN  0.39 m  2982.4 kNm (1)

where γov is the overstrength factor,


fyb is the nominal yield strength specified for the steel grade,
Wpl,y,b is the plastic section modulus of the beam,
Vu is the beam shear force, calculated as if assumed as 0.5 Vpl,Rd,b
Vpl,Rd,b is the beam design plastic shear resistance (1910 kN for HEA 600, S355),
Sh is the distance from the column face to the end of the rib stiffener.
Due to the differences in profile sections and bolts between the USA and Europe, full penetration
welds had to be used with RSC in order to assure prescribed limitations in [1]. There is no option for
considering two bolt rows in a rib stiffened connection outside of the beam given in [5]. Therefore,
the recommendations from [6] were considered. The position of the compression centre was taken in
the middle of the compressed beam flange (Fig. 1b). The required connection bending resistance Mf
for the beam size HEA 600, S355 is 2982 kNm.

a) b)

Fig. 1. a) RSC connection dimensions; b) distances from the bolt rows to the compression centre

1.2 Haunched stiffened configuration (HSC)


As opposed to the RSC connection, in Europe HSC (Fig. 2) is commonly used. HSC is not
symmetrical, therefore its resistance must be calculated in both sagging and hogging directions. For
this reason, two different centres of compression should be considered. Fig. 3a shows the centre of
compression for the sagging moment, which is the same as in RSC connection, whereas Fig. 3b shows
the centre of compression for hogging moment, which is according to [2], located at the distance of
0.6 hh below the beam flange, where hh is the haunch height. Both RSC and HSC connections are full
strength, as defined in [5].
| 255

Fig. 2. HSC connection dimensions

a) b)

Fig. 3. Position of the compression centre a) for sagging moment; b) for hogging moment

2 NUMERICAL SIMULATIONS
2.1 Description of the numerical model
The numerical model was assembled in FE software Abaqus/Standard of two- and three-dimensional
elements of column, beam and bolts (Fig. 4). Special attention was paid to the contact area between
end plate and column face and between end plate, column face and bolt nuts. Since the contact
relations are complex, tangent and normal behaviour were considered. The Coulomb friction was
defined in tangent direction with the value of friction coefficient of 0.2. The normal behaviour was
defined as “hard contact”. The bolt forces were automatically calculated within Abaqus using
integration of contact pressure between bolt nuts and column flange. FE mesh was generated using
eight-node linear brick elements C3D8R and its density was optimized according to [7] (Fig. 5). The
average side length of brick finite element was 10 mm.
256 |

Fig. 4. Assembly of Abaqus numerical model

a) b) c)

Fig. 5. Meshed parts in Abaqus a) beam; b) column; c) bolt

Various material models were adopted in numerical simulations in order to study the effect of strain
hardening, overstrength, monotonic and cyclic loading. The elastic properties were defined by elastic
modulus E = 210 GPa and Poisson’s ratio ν = 0.3. The material model based on standard tensile test
for S355 adopted from [7] was used to study the effect of strain hardening. The material model is
presented in Fig. 6. The overstrength of the beam material based on this model was considered by
multiplying the plastic true strain by overstrength factor γov,act, calculated in Eq.(2).
f y ,act 398.4
 ov ,act    1.12 (2)
f y , nom 355
where fy,act is the actual yield strength (Fig. 6b),
fy,nom is the nominal yield strength.
The material model data gained from tensile test were used for column, whereas the values, multiplied
by overstrength factor γov,act were used for the beam in order to fulfil the requirements from [8]. In the
simulation, where strain hardening was not considered, models described in Fig. 7a and Fig. 7b were
used. These models do have some level of strain hardening (E/1000) to obtain the numerical stability.
The bolt material was modelled in relation to the nominal yield stress and ultimate strength (Fig. 7d),
where the tensile strength was assumed to be reached at 5 % of plastic deformation.
The material model, used within cyclic loading protocol is implemented in Abaqus and consists of a
nonlinear kinematic and isotropic hardening. Its parameters are adopted from [7].
| 257

700

600 Tensile test Overstrength


True Stress [N/mm2]

500 True True True True


Stress Strain Stress Strain
400 [N/mm2] [-] [N/mm2] [-]
398,4 0 444,7 0
300
408,9 0,0237 456,4 0,0235
200 Tensile test
463,6 0,0374 517,4 0,0372
Overstrength
100 491,3 0,0612 548,3 0,0610
525,1 0,0961 586,1 0,0959
0
0 0.025 0.05 0.075 0.1 0.125 0.15 540,8 0,1238 603,6 0,1235
True Strain [-] 560,5 0,1584 625,5 0,1581
a) b) c)

Fig. 6. Material model a) from tensile test; b) multiplied by overstrength factor γov,act

True Stress True True Stress True True Stress True


[N/mm2] Strain [-] [N/mm2] Strain [-] [N/mm2] Strain [-]
355 0 444 0 900 0
376 0,1 465 0,1 1000 0,05
a) b) c)

Fig. 7. Material models a) without strain hardening – nominal; b) without strain hardening – multiplied by
γov = 1.25; c) bolt

2.2 Monotonic loading


In order to simulate the behaviour of the connection under monotonic loading, two sequence analysis
steps were defined in Abaqus. In the first one, the preload of the bolts was considered. The preload
force was applied through initial stress on each bolt. In order to achieve full preload required by [8],
initial stress of 950 MPa was applied to the bolt shank. The equilibrium resulted in the full bolt
preload approximately equal to:

,C  0.7 f ub As
FpM39  0.7 1000 MPa  976 mm 2  683kN
M39
(3)

where fub is the bolt ultimate tensile strength,


AsM39 is the tensile stress area of the bolt M39.
In the following analysis, the beam tip had been loaded by the incremental displacement until the
ultimate bending moment was reached. The plastic bending resistance moment of HEA 600 with
S355 considering the overstrength γov,act (Fig. 6b) is:
M plHEA600
,y    ov ,act f y ,act W pl , y  44.47 kN cm 2  5350 cm3  2380 kNm (5)

Bending moment M that develops in the plastic hinge is caused by force F acting at the tip of the
beam and lever arm L (see Fig. 8a).
M  F L (6)
Bending moment M is plotted in Fig. 8b. As it can be seen in Fig. 8b, the fully plastic bending
resistance corresponds to the results.
258 |

2500

Bending Moment [kNm]


2000

1500

1000 HSC - hogging moment


HSC - sagging moment
500
RSC
0
0 0.05 0.1 0.15 0.2
Rotation [rad]
a)
b)
Fig. 8. a) Length of lever arm L, b) bending moment – rotation relationships for investigated connections

During the second step of the analysis, the development of equivalent plastic strain (PEEQ) was
observed at two different levels of rotation, namely 35 mrad and 60 mrad. Rotation of 35 mrad is
required in [8] and rotation of 60 mrad in [3]. These rotations present the rotation capacity of the
plastic hinge region under seismic loading, taking into account the accumulated plastic strain due to
cyclic loading. As expected, plastic deformations occurred in the beam at the end of the stiffened
connection.

a) b) c)
Fig. 9. Equivalent plastic strain at beam rotation 35 mrad for a) HSC hogging moment, b) HSC sagging
moment, c) RSC

As seen in Fig. 9, where the upper limit is set at PEEQ of 2 %, the highest strain started to occur at
beam flanges near the stiffener end. In the case of HSC connection these deformations are not
symmetrical (Fig. 9a, b). The deformations expand to the middle of the beam section as the rotation
proceeds (Fig. 10). Deformation of 2 % represents, according to Fig. 6, the end of the yield plateau
for construction steels, which are used in dissipative parts for seismic design. It must be noted, that
full bending resistance of the beam is not reached at the rotation of 60 mrad in any of the investigated
connections. Due to the beam height (590 mm), the ability of developing full inelastic strain is
strongly associated with beam rotations. As seen from Fig. 8b, beams reach their full plastic bending
moment at 110-130 mrad, depending on the connection type. These rotations are not common in real
structures under seismic load, but are necessary to consider during numerical investigations. Fig. 10a
also shows some level of plastic deformation at the connection of beam end to the end plate. However,
| 259

this area should remain elastic in case of full strength joints, where plastic deformations are only
allowed to occur in the beam.

a) b) c)
Fig. 10. Equivalent plastic strain at beam rotation 60 mrad for a) HSC hogging moment, b) HSC sagging
moment, c) RSC

The plastic hinge in the beam results in high forces in the top bolt rows. Fig. 11 shows bolt forces
only in bolt rows, where bolt force exceeds full bolt preload. The ultimate tensile resistance of bolt
row (2 bolts per row) may be estimated as:
Ft ,r  2 fub AsM39  2 1000 MPa  976 mm 2  1952 kN (7)

The highest forces occur in the case of HSC exposed to hogging moment (Fig. 11a). In this case only
two bolt rows exceed the bolt preload and the maximum force in row reaches 1938 kN at the rotation
of 130 mrad, which according to Eq. (7) corresponds to almost fully yielded and strain hardened state
that would most likely cause fracture of the bolts (some bending was aslo observed in the bolts). At
the rotation of 35 mrad, bolt row force reaches the value of 1865 kN and at the rotation of 60 mrad
the value of 1883 kN.
The forces in bolt rows for the sagging moment of HSC show, in Fig. 11b, more evenly distributed
forces in the bolts. The bolt preload is exceeded in all four bolt rows. The highest force is
approximately 1650 kN at the rotation of 120 mrad, which means that the bolts are nominally in the
elastic state (some yielding is present due to bending). At the rotation of 35 mrad, the highest bolt
row force reaches the value of 1530 kN and at the rotation of 60 mrad the value of 1552 kN is reached.
The maximum forces develop in the third and the fourth bolt rows. This is expected, as these rows
are nearest to the beam flange. In the RSC connection, three bolt rows exceed the bolt preload. At the
rotation of 35 mrad, the highest bolt row force is 1691 kN and at the rotation of 60 mrad the value of
1720 kN was observed. The maximum bolt row force of approximately 1775 kN occurs at the beam
rotation of 115 mrad.
260 |

2000 2000 2000


Bolt Force [kN]

1500

Bolt Force [kN]


Bolt Force [kN]
1500 1500

1000 1000 1000


1. row 1. row
500 1. row 2. row
500 500 2. row
2. row 3. row
4. row 3. row
0 0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Rotation [rad] Rotation [rad] Rotation [rad]
a) b) c)
Fig. 11: Bolt force – rotation relationships for a) HSC hogging moment, b) HSC sagging moment, c) RSC

The contact pressure between end plate and column flange was studied in order to locate the centre
of compression that indicates the position of neutral axis. Due to contact pressure, which occurs after
bolt preload, it is difficult to locate the centre of compression. Therefore, another analysis with 10 %
of preload force was performed. The 10 % preload was applied to achieve the numerical stability.
These results are presented in Fig. 12 that shows the position of the centre of compression at
maximum resistance. The Abaqus command “Centre of the total force due to contact pressure” was
used to provide these results. The centres that result from the simulation are marked with red dot,
while the centres considered in the calculation of the joint flexural resistance are marked with black
dot. In Fig. 12a, the difference between predicted and actual centre of compression may be clearly
observed. The actual centre of compression is approximately 130 mm higher than the predicted one,
thus also the lever arms used in the calculation are shorter, which results in lower actual bending
resistance. Although the centre of compression of HSC sagging moment (Fig. 12b) is closer to the
predicted one, it is still not on the safe side and the actual lever arms are shorter by approximately
20 mm. On the other side, simulation showed that the bolt lever arms of the RSC joint are
approximately 10 mm longer than predicted (Fig. 12c).

a) b) c)
Fig. 12: Contact pressure and the positions of compression centres for a) HSC hogging moment, b) HSC
sagging moment, c) RSC
| 261

3000

Bending Moment [kNm]


2500

2000

1500

1000
Without strain hardening
500 With strain hardening
Rotation [rad]
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18

Fig. 13: Strain hardening influence on bending moment – rotation relationship

Fig. 13 shows the effect of the strain hardening on the joint behaviour of HSC in sagging moment.
As expected, material model, where the strain hardening was considered, resulted in higher beam
bending resistance. Maximum bending resistance considering the strain hardening is approximately
2680 kNm, whereas the maximum bending resistance without accounting for the strain hardening
does not exceed 2375 kNm(=1.25×35.5×5350). The quotient of these two numbers multiplied by
quotient γov/γov,act, gives the overstrength factor for strain hardening equalling 1.26. This is more than
1.1 that is prescribed in [8]. It has to be considered that the effect of strain hardening was evaluated
for fully developed plastic hinge that was observed at 130 mrad.
Considering the level of bolt preload, numerical analyses with 10 % and 100 % of the full bolt preload
were carried out. There was no significant difference observed between these analyses.

2.3 Cyclic loading


The loading protocol was defined according to the ECCS and ANSI/AISC 341-10 procedures.
According to [8], connection design should be such that the rotation capacity of the plastic hinge
region is not less than 35 mrad for structures of ductility class DCH. This rotation capacity should be
ensured under cyclic loading without degradation of strength and stiffness greater than 20 %. The
global bending moment – rotation relationship is, as seen in Fig. 14 and Fig. 15, very similar for the
HSC and the RSC joints. Both joint configurations can withstand a significant loading and it can be
clearly seen that strength and stiffness are assured under the rotation of 35 mrad. Furthermore, the
strength and the stiffness are also assured at 60 mrad.
0,035
-0,035
-0,06
-0,035

0,035

0,06

3000
-0,06

3000
0,06

2000 0,8 M 2000


Bending Moment [kNm]

0,8 M
Bending Moment [kNm]

1000 1000

0 0

-1000 -1000
-0,8 M
-2000 -2000
-0,8 M
-3000 -3000
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 -0.08 -0.04 0 0.04 0.08
Rotation [rad] Rotation [rad]
a) b)
Fig. 14: Hysteretic behaviour of HSC joint a) under ECCS loading protocol, b) under ANSI/AISC 341-10
loading protocol
262 |

0,035

0,06
-0,035
-0,06
-0,035

0,035
3000

0,06
3000

-0,06
0,8 M 2000
2000
Bending Moment [kNm]

0,8 M

Bending Moment [kNm]


1000 1000

0 0

-1000 -1000

-2000 -2000 -0,8 M


-0,8 M
-3000 -3000
-0.15 -0.05 0.05 0.15 -0.08 -0.04 0 0.04 0.08
Rotation [rad] Rotation [rad]
a) b)
Fig. 15: Hysteretic behaviour of RSC joint a) under ECCS loading protocol, b) under ANSI/AISC 341-10
loading protocol

3 CONCLUSIONS
Based on the numerical simulations, the following conclusions can be drawn:
 The position of the centre of compression for the HSC joint differs from the one suggested in
[2].
 The RSC joint showed better behaviour in terms bolt forces compared to the HSC joint.
 The effect of strain hardening was much larger than currently considered.
 All investigated joints showed satisfactory behaviour in terms of global bending moment –
rotation relationship.
 As already known, the strain hardening greatly influences global bending moment – rotation
behaviour.
 The level of bolt preload does not alter the global bending moment – rotation relationship.
 The rotation capacity was maintained under cyclic loading, without strength and stiffness
degradation greater than 20 %. Unfortunately, the work on this topic is still ongoing and no
conclusions can be presented for the accumulated plastic strain.

REFERENCES
[1] AISC (2016). "Prequalified Connections for Special and Intermediate Steel Moment Frames for Seismic
Applications." ANSI/AISC 358-16, American Institute of Steel Construction, Inc, Chicago.
[2] Lee, K. (2016). "Numerical investigation on seismic behaviors of steel haunch connections." European
Erasmus Mundus Master Thesis, Universitatea Politehnica Timişoara.
[3] AISC (2010). "Seismic Provisions for Structural Steel Buildings." ANSI/AISC 341-10, American
Institute of Steel Construction, Inc, Chicago.
[4] D'Aniello, M., Cassiano, D. and Landolfo, R. (2016). "Monotonic and cyclic inelastic tensile response
of European preloadable gr10.9 bolt assemblies." Journal of Constructional Steel Research, 124, 77-90.
[5] CEN (2005). "Eurocode 3: Design of steel structures - Part 1-8: Design of joints." EN 1993-1-8, European
Committee for Standardisation, Brussels.
[6] SCI and BCSA (2013). Joints in steel construction: Moment-resisting joints to eurocode 3, The Steel
Construction Institute and The British Constructional Steelwork Association.
[7] Čermelj, B., Može, P. and Sinur, F. (2016). "On the prediction of low-cycle fatigue in steel welded beam-
to-column joints." Journal of Constructional Steel Research, 117, 49-63.
[8] CEN (2005). "Eurocode 8: Design of structures for earthquake resistance - Part 1: General rules, seismic
actions and rules for buildings." EN 1998-1, European Committee for Standardisation, Brussels.

You might also like