Documente Academic
Documente Profesional
Documente Cultură
www.elsevier.com/locate/engstruct
Non-linear cyclic model of top and seat with web angle for steel
beam-to-column conections
L. Calado ∗
Department of Civil Engineering and Architecture, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
Received 18 November 2002; received in revised form 24 March 2003; accepted 24 March 2003
Abstract
The use of steel connections is inherent of every structural steel building. Savings in connection costs as well as improved
connection quality has an impact on all types of buildings. Several analytical models have been developed in the last years to
represent the cyclic behaviour of beam-to-column connections. However very few models have been substantiated by test results
under cyclic loading conditions. For that reason, the implementation of such models in general-purpose structural analysis programs
for steel structures under seismic actions, is in some cases complex and in others impossible. This paper is concerned with a non-
linear cyclic model of top and seat with web angle for steel beam-to-column connections, with damage accumulation and bolts in
cyclic shear. Numerical results are compared with experimental tests in order to evidence the capabilities and the accuracy of the
proposed numerical model.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Connections; Cyclic behaviour; Bolts; Steel; Numerical modelling; Damage accumulation
0141-0296/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0141-0296(03)00067-1
1190 L. Calado / Engineering Structures 25 (2003) 1189–1197
Nomenclature
ss stress with damage accumulation
f function for the decrease of the stress
ŝs stress without damage accumulation
ŝsr stress at the last inversion without damage accumulation
ssy yield stress
es strain
esr strain at the last loading inversion
esy yield strain
b ratio between the kinematic stiffness hardening and the modulus of elasticity
R parameter that takes into account the Bauschinger effect and defines the shape of the transition curve
between the elastic and plastic zones [7]
ITd total damage index
Icd damage index related with the complete semi-cycles
⌬Icd damage index related to the actual semi-cycle
+
Nsl,1 ; N⫺ sl,1 forces to start the slip
+
Nsl,2 ; N⫺ sl,2 forces at the end of the slip
Ny+ ; N⫺ y yield forces
Ny,o+
; N⫺ y,o yield forces for zero displacement
+
dsl,1 ; d⫺sl,1 displacement at the beginning of the slip
+
dsl,2 ; d⫺sl,2 displacement at the end of the slip
dy+ ; d⫺ y yield displacement
ship for steel under uniaxial cyclic loadings, is based on Comparing (3) with (5) the following Eqs. (6) and (7)
Menegotto and Pinto [8] model, reformulated to take can be obtained:
into account the effects of unloading. In this reformul-
ks Ekin
ation new expressions are proposed to assess the ⫽ ⫽1 (6)
relationship between the relative stress (ss∗)and the rela- ke Eb
tive strain (e∗s ): s+yo⫺ŝsr ⫹ Ekinesr
ks ⫽
(1⫺b) es∗ (1⫺b) ssy
s∗s ⫽ b e∗s ⫹
冑 1 ⫹ |e |
R ∗R
(1) (7)
s ŝsr b esr
⫽ 1⫺ ⫹
(1⫺b) ssy 1⫺b esy
The relative stress and relative strain can be obtained
[7] by the following Eq. (2): The implementation of this stress-strain relationship
L. Calado / Engineering Structures 25 (2003) 1189–1197 1191
not only needs the assessment that the stress as function beam-columns, welded joints and beam-to-column con-
of the strain but also the evaluation of the tangent stiff- nections proposed to adopt as unified failure criterion the
ds reduction of the energy dissipated in a cycle to 50% of
ness . that dissipated by the same structural component made
de
The main expressions to assess this stress-strain of an elastic perfectly plastic material, cycled under the
relationship, which take into account the partial same amplitude. The proposed value of 50% can be
unloading, are [7]: adopted for a safe assessment of the damage accumu-
lated in structural components. Hence, this value is not
ds∗s to be considered as the best fit of experimental results,
∗ ∗
dŝs dŝs ds de de∗s de∗s but can be regarded as a possible reference value in dam-
⫽ ∗⫻ ⫻ ⫽ ∗⫻
s s
∗
(8) age assessment procedures.
des dss de des dss
s des
dŝs In a Log-Log domain Eq. (14) represents a straight
line with a slope equal to –1 / m called the fatigue resist-
ds∗s 1 ance line, which identifies the safe and unsafe regions.
⫽ (9)
dŝs ksssy Thus the value of stress taking into account the damage
accumulation can be obtained from the following Eq.
des∗ 1 (15):
⫽ (10)
des keesy
ss ⫽ ŝs ⫻ (1⫺f(Id)) (15)
ds∗s 1
⫽ b ⫹ (1⫺b) (11) The function for the decrease of the stress (f) can be
des∗ [1 ⫹ |es∗|R]1+1/R obtained through experimental tests [10]. Analytically
dŝs
des 冋
⫽ E ⫻ b ⫹ (1⫺b)
1
[1 ⫹ |e∗s |R]1+1/R 册 (12)
the function f can be represented by the following Eq.
(16):
f(Id) (16)
The previous equations were deducted assuming an
冦
increasing load. For decreasing load the deduction is
similar [7]. 0 ⇐ Idⱕ0.295
To take into account the damage accumulation due to ⫽ 0.3168Id⫺0.0936 ⇐ Id苸]0.295;0.960]
cyclic loading it is necessary to include into the stress- 19.7332Id⫺18.7332 ⇐ Id ⬎ 0.960
strain relationship a damage index that should consider
the deterioration in the previous complete semi-cycles
and the deterioration in the current semi-cycle. A poss- The assessment of the stress is carried out in two steps
ible equation for the total damage index, may be written [7]: a first step where the stress is evaluated without the
as follows [7]: consideration of the damage accumulation (ŝs) and a
second one where the stress is corrected to take into
ITd ⫽ Icd ⫹ ⌬Icd (13) account the damage of the current semi-cycle. In this
The damage index related with the complete semi- methodology the correction of the stress is made at the
cycles is constant during each semi-cycle while the dam- end of the current semi-cycle.
age index related to the current semi-cycle is function To check the accuracy and the validity of the proposed
of the current strain and can be assessed through the model for the stress-strain relationship of the steel with
following Eq. (14): due consideration of partial unloading and damage
accumulation, several experimental tests were performed
1
⌬Icd ⫽ ⫽
ey 冉
|es⫺esr|
冊
m
(14)
at the Laboratory for Structures and Strength of
Materials of the Instituto Superior Técnico, Lisbon [7].
Comparison between experimental data and numerical
Nj K
simulation in general shows good agreement. Concern-
In this equation Nj is the number of complete semi- ing the strain energy stored until the failure it was veri-
cycles to failure and m and K are parameters dependent fied that the error obtained is less than + 8% showing
on the typology and the mechanical properties of the the ability of the model to be used in cyclic simulations.
component under consideration and may be obtained by
a statistical evaluation.
In this context failure is associated with an achieve- 3. Numerical model for bolts in shear
ment of a given level of deterioration of the mechanical
properties of the material or structural component. Some To adequately model the cyclic behaviour of bolted
authors [9] based on experimental results obtained dur- connections it is fundamental to simulate the behaviour
ing extensive testing programs carried out on beams, of bolts in shear. The model proposed is based on an
1192 L. Calado / Engineering Structures 25 (2003) 1189–1197
冦
Nsl,1 ⫹ Ge ⫻ (d⫺dsl,1) ⇐ d ⬍ dsl,1
Nsl,1 ⫹ Gf ⫻ (d⫺dsl,1) ⇐ d苸[dsl,1;dsl,2[
⫽
Nsl,2 ⫹ Ge ⫻ (d⫺dsl,2) ⇐ d苸[dsl,2;dy[
Ny,0 ⫹ Gkin ⫻ d ⇐ dⱖdy
The distance d between the yield points (1) and (4) has
a constant value and can be assessed by the following
Eq. (23):
N+y,o⫺N⫺
y,o ⫹ (Ge⫺Gf) ⫻ df
d+y ⫺d⫺
y ⫽ (23)
Ge⫺Gkin
There are two different cases for the assessment of the
main points, which are the attainment, or not of the yield
in the previous semi-cycle, as shown in Fig. 5:
In the case that the yield is not reached in the previous
semi-cycle, the negative yield point can be assessed from
the following Eqs. (24) and (25):
N+y,o⫺N⫺
y,o ⫹ (Ge⫺Gf) ⫻ df
y ⫽ dy ⫺
d⫺ +
(24)
Ge⫺Gkin
Fig. 3. Parameters used to define the complete cycle.
Ny⫺ ⫽ N⫺
y,o ⫹ Gkin ⫻ dy
⫺
(25)
For the case that the yield is reached in the previous
semi-cycle the negative yield point can be obtained
through the following Eqs. (26) and (27):
N+y,o⫺N⫺
y,o ⫹ (Ge⫺Gf) ⫻ df
y ⫽ dr⫺
d⫺ (26)
Ge⫺Gkin
Ny⫺ ⫽ N⫺
y,o ⫹ Gkin ⫻ dy
⫺
(27)
The other points are obtained in a similar way, the point
at the beginning of the slip being assessed by the impo- of the beam and the column were respectively a IPE300
sition of a load equal to N⫺sl : and a HEB200. Angles L120 × 120 × 10 and two rows
of preloaded M16 grade 8.8 bolts (preloading force of
Nr ⫹ Ge ⫻ (d⫺
sl ⫺dr) ⫽ Nsl,1
⫺
⇒
87.9 kN) were used to connect the column to the beam
sl,1⫺Nr
N⫺ (28) (clearance in holes 1 mm). The material used for the
sl ⫽ dr ⫹
d⫺
Ge column, beam, and angles was steel S235 JR. The conti-
nuity of the connection through the column was ensured
where Nr and dr are respectively the force and the dis-
by 12 mm thick plate stiffeners, fillet welded to the col-
placement corresponding to the actual inversion.
The final point of the slip can be obtained by the Eqs. umn web and flanges.
(29) and (30): The specimen was T-shaped beam-to-column subas-
semblage, consisting approximately of a 1000 mm long
sl,2 ⫽ Nsl,1 ⫹ Gf ⫻ (dsl,2⫺dsl,1)
N⫺ ⫺ ⫺ ⫺
beam and a 1800 mm long column. Due to the character-
(29)
y ⫽ Nsl,2 ⫹ Ge ⫻ (dy ⫺dsl,2)
N⫺ ⫺ ⫺ ⫺ istics of the test set-up the column was the horizontal
element while the beam was the vertical one, Fig. 7.
G e ⫻ d⫺
y ⫹ Nsl,1⫺Ny ⫺Gf ⫻ dsl,1
⫺ ⫺ ⫺
The developed model uses the stress-strain diagram
sl,2 ⫽
d⫺ (30)
Ge⫺Gf for all structural elements with the damage accumulation
In order to apply this model it is necessary to pre-defined previously presented. The behaviour of the bolt in shear
the clearance of the bolt (df), the applied preloading force follows the model previously presented. The panel zone
(P), the yield load (Ny), the stiffness (Ge, Gkin, Gf) and and the deformation of the angles are also taken into
the slip factor (m). account in the numerical simulation.
In this model all structural elements are simulated by
finite elements of beam with geometric and mechanical
4. Numerical model for beam-to-column properties analogous to the real elements. In Fig. 8 the
connections with top and seat web angles numerical model for a bolted beam-to-column connec-
tion is presented.
The numerical model developed and presented in this The accuracy and capability of the proposed numeri-
paper is based on a component model and allows the cal model for use as top and seat with web angle connec-
simulation of beam-to-column connections with top and tions in numerical simulations of the hysteretic behav-
seat web angle under cyclic loading. iour of beam-to-column connections can be analysed by
The connection experimentally and numerically stud- comparing the force-displacement curves and accumu-
ied was a typical beam-to-column connection with top lated energy shown in Figs. 9–11.
and seat web angle as shown in Fig. 6. The cross sections Comparison between numerical and experimental
Fig. 8. Numerical model for beam-to-column connection with top and seat web angle.
1196 L. Calado / Engineering Structures 25 (2003) 1189–1197
5. Conclusions
Acknowledgements
Fig. 10. Numerical force—displacement curve of a bolted connec- The work reported in this paper was supported with
tion. grants from the FCT – Fundação para a Ciência e a Tec-
nologia (Portugal). The author acknowledges the MSc
level research work of António Brito that provided the
numerical simulation for this paper, and Mr. João Fer-
results allows to conclude that the proposed numerical reira for his assistance with elaboration of experi-
model is able to simulate, with a good level of accuracy mental tests.
until failure, the behaviour of beam-to-column connec-
tion with top and seat web angle. The force-displacement
References
curve and the accumulated energy are appropriately
modelled showing the suitableness of this numerical [1] Liu J, Astaneh-Asl A. Behavior of simple connections in frames
model to simulate the damage accumulation of the subjected to large lateral deflections, Proceedings of the Struc-
material and the non-linear behaviour of the bolts. One tural Engineering World Wide, San Francisco, CA, 1998, Paper
of the advantages of this numerical model is that it does Reference: T158-5.
[2] Leon RT. Seismic performance of bolted and riveted connections.
not need experimental tests for its calibration. It needs In Program to reduce the earthquake hazards of steel moment
as input data the geometric and mechanical character- frames structures, 1997, FEMA-288, Report No. SAC-95-09,
istics of the connection. Sacramento, California, U.S.A.
L. Calado / Engineering Structures 25 (2003) 1189–1197 1197
[3] Elnashai AS, Elghazouli AY, Denesh-Ashtiani FA. Response of [7] Calado L, Brito A. Stress-strain relationship for steel under uniax-
semirigid steel frames to cyclic and earthquake loads. Journal of ial cyclic loadings. Advances in Structural Engineering
Structural Engineering, ASCE 1998;124(8):857–67. 2002;5(3):143–51.
[4] De Matteis G, Landolfo R, Calado L. Cyclic behaviour of semi- [8] Menegotto M, Pinto P. Method of analysis for cyclically loaded
rigid angle connections: a comparative study of tests and model- R. C. plane frames including changes in geometry and non-elastic
ling, Proceedings of the 3rd International Conference on the behaviour of elements under combined normal force and bending,
Behaviour of Steel Structures in Seismic Areas - STESSA 2000, Preliminary Report, Proceedings of the IABSE Symposium-
Montreal, Canada, 2000, p. 165-174. International Association for Bridge and Structural Engineering,
[5] Calado L, Mele E. Experimental behaviour of steel beam-to-col- 1973, Vol. 13, Lisbon, Portugal.
umn joints: fully welded vs bolted connections, Proceedings of [9] Ballio G, Calado L, Castiglioni CA. Low cycle fatigue behaviour
the 12th World Conference on Earthquake Engineering, Auck- of structural steel members and connections. Fatigue & Fracture
land, New Zealand, 2000, Paper nⴰ 2570/6/A. of Engineering Materials & Structures 1997;20(8):1129–46.
[6] Roeder CW. Design models for moment resisting steel construc- [10] Proença JM. Seismic behaviour of precast structures, Ph.D. The-
tions. Proceedings of the Structural Engineering World Wide, San sis (in Portuguese), Instituto Superior Técnico, Lisbon, Portu-
Francisco, CA, 1998, Paper Reference: T158-4. gal, 1996.