global performance of steel moment resisting frames with semi-rigid joints
TRANSCRIPT
GLOBAL PERFORMANCE OF STEEL MOMENT RESISTING
FRAMES WITH SEMI-RIGID JOINTS
D. Dubina1, A. Ciutina
1 A. Stratan
1, & F. Dinu
2
1Department of Steel Structures and Structural Mechanics, The “Politehnica” University,
Timisoara, Romania 2Centre of Advanced and Fundamental Technical Studies, Romanian Academy of Science,
Timisoara, Romania
ABSTRACT
Beam-to-column joints have a fundamental importance in case of seismic moment resistant frames,
because dissipative zones must be located at the beam-ends, so that their rotational ductility supply
is strictly related to the detailing of connections. Based on a large numerical study, the present paper
investigates the seismic performance of two series of moment resisting frames with beam-to-
column joints with different configurations. Performance of the analysed structures is expressed in
terms of characteristic failure acceleration, ductility demand, damage indices and the q factor.
KEYWORDS
moment resisting frames, beam-to-column joints, steel frames, semi-rigid joints, rotation capacity,
ductility, damage, earthquake.
1. INTRODUCTION
The use of semi-rigid joints in frames subjected to seismic loads is a matter of controversy. None of
the existing design codes include provisions for their use and, in zones characterised by a high
seismicity, the use of rigid full-strength joints is mandatory.
To ensure that a beam-to-column joint is ductile enough and has the capacity to provide the required
rotations, the related connection must be capable to develop adequate plastic hinge while sustaining
its yield moment capacity.
Due to their higher flexibility, semi-rigid steel frames are prone to increases in inter-storey drifts.
The inter-storey drift condition is related to the serviceability limit state which corresponds to minor
frequent earthquake. The design objective, when building serviceability is checked, is that the
building, including both structural and non-structural components, should suffer no damage and
discomfort of the inhabitants should be minimal. The first requirement, which leads to the
avoidance of damage, is satisfied by ensuring that the structure behaviour during the earthquake
remains in the elastic range. In order to fulfil the second requirement - both non-structural
components damage and discomfort of inhabitants is avoided- it is necessary to provide sufficient
stiffness to prevent significant deformations.
The ultimate limit state of a building in seismic circumstances could be regarded either as damage
limit state or failure limit state.
The damage limit state allows some minor damages to non-structural components due to large local
deformation in certain zones.
The failure limit state pertains to very infrequent severe earthquake ground motions in which both
structural and non-structural damages are expected, but safety of the inhabitants must be
guaranteed. Furthermore, the structure must be able to absorb and dissipate large amounts of
energy. In case of MR frames with semi-rigid partial-resistant beam-to-column connections, the
connection design properties can be expressed in rotational stiffness, moment capacity and plastic
rotation supply. In case of strong seismic motion, the plastic rotation demand for beam-to-column
connection could be greater than the plastic rotation supply, which leads to its damage (partial or
total). This was the situation of most beam-to-column welded connections in multi-storey MR
frames during the Northridge earthquake (Kato B. et al, 1997). The problem in such cases is
whether damaged connections can be recuperated by repairing works or not?
The SAC joint venture project suggests several reparatory design detailing for seismic damaged
welded connections. It is obviously that reparatory detailing can be also proposed for bolted
connections. It is recommendable, of course, to have an initial design of beam-to-column
connections such that both damage and reparatory works to be easy controlled.
Authors’ opinion is that based on the Performance Design Philosophy (Bertero V., 1997), semi-
rigid steel frames could satisfy the required strength and stiffness conditions for “Operational and
Functional” or “Life Safety” performance levels. In terms of rotation capacity of connections, it
means that, if a certain percentage of damage in frame connections can be accepted, a larger
rotation capacity of the connections (related to design) can be considered as a supplementary
ductility supply for the global behaviour of the structure. A parametrical study was developed by
the authors in order to analyse the effects of beam-to-column connection properties on the global
performance of two series of MR frames. The present paper shows the results and the conclusions
of this study.
2. ANALYSED FRAMES
The parametrical study is developed on two frames: a 6 storey – 3 bay frame (C36) made out of
Fe430 steel and a 3 storey – 3 bay frame (C33) made out of Fe360 steel, shown in Figure 1.
The member cross-sectional dimensions and joint properties were obtained by means of an
equivalent static design procedure, using an elasto-plastic analysis.
Different frame typologies have been considered in the study. Member characteristics for the C36
and C33 frames are given in Table 1.
HEB260
HEB260
HEB300
HEB360
HEB360
HEB300IPE300
IPE300
IPE300
IPE300
IPE300
IPE300
4.50 4.504.50
3.00
3.00
3.00
3.00
3.00
3.00
Figure 1. Geometry of the analysed frames
The rigid and semi-rigid joints have been designed according to Annex J of Eurocode 3 [8]. Joint
Table 1. Design values of the section characteristics.
Frame Column Beam Mpl,b,Rd
x106 Nxmm
Mpl,c,Rd
x106 Nxmm
Mpl,c,Rd/
Mpl,b,Rd material
C36 HEB360
HEB300
HEB260
IPE300
157.1
670.8
467.3
320.7
4.27
2.97
2.04
Fe430
C33 HEB240 IPE330 171.8 225.0 1.31 Fe360
configurations are given in Figure 2 and the design values of moment capacity and stiffness in
Table 2. Distinction is made between single-sided and double-sided joint, subjected to unbalanced
moments, as is the case under seismic horizontal forces.
C1 C2 C3 C4
C5 C6
M20/6.6 M20/6.6
Figure 2. Detailing of semi-rigid joints.
Table 2. Joints Characteristics Joint
Type
Mj,Rd
x106 Nxmm
m*
(Mj,Rd/Mpl,b,Rd)
Sj,ini
x1011
Nxmm/rad
Sj
x1011
Nxmm/rad
S*j
C36.1A
C36.1B
C36.1C
153.4
153.4
153.4
0.98
0.98
0.98
0.628
0.561
0.493
0.314
0.281
0.246
8.05
7.21
6.31
C36.2A
C36.2B
C36.2C
133.8
109.3
91.4
0.85
0.70
0.58
0.421
0.340
0.305
0.211
0.127
0.152
5.40
3.26
3.90
C36.3A
C36.3B
C36.3C
153.4
153.4
153.4
0.98
0.98
0.98
1.235
1.274
1.292
0.618
0.640
0.646
15.85
16.41
16.57
C36.4A
C36.4B
C36.4C
153.4
153.4
153.4
0.98
0.98
0.98
1.235
1.274
1.292
0.618
0.640
0.646
15.85
16.41
16.57
C36.5A
C36.5B
C36.5C
219.3
280.8
172.3
1.40
1.79
1.10
1.186
1.509
0.936
0.977
1.509
0.468
25.05
38.70
12.00
C36.6A
C36.6B
C36.6C
280.8
280.8
243.4
1.79
1.79
1.55
C33.1 130.9 0.76 0.501 0.250 4.05
C33.2 74.63 0.43 0.307 0.153 2.48
C33.3 155.8 0.91 1.362 0.681 11.02
C33.4 152.5 0.89 1.362 0.681 11.02
C33.5 140.9 0.82 0.907 0.454 7.34
C33.6 210.4 1.22
A – connections at levels 1-2; B - connections at levels 3-5; C - connections at the roof level
If the web panel of the C2 joints is considered as unstiffened to shear, compared to C4, the drop of
stiffness is very important in the case of unbalanced bending moments. This penalisation is
probably too severe.
The topology of the frames, including the joint distribution, is shown in Figure 3.
The elasto-plastic dynamic analysis has been performed using the DRAIN2DX computer program
and the set of three accelerograms described in Chapter 3.
The main aim of the analysis was to investigate the influence of the different ground motions on the
seismic response of the frames. For this study, a value of 3% inelastic inter-storey drift and two
values of plastic rotation capacities i.e. 0.02 rad. for welded joints and 0.03 rad. for bolted joints
were accepted.
The monotonic plastic rotation capacity of members (p) was computed by DUCTROT computer
program (Gioncu et al, 1997). Plastic rotation capacity of members for cyclic behaviour conditions
(pcor
) has been computed by adjusting the monotonic values by coefficients depending on the
member cross-sectional slenderness and the axial force. A material partial safety factor =1.5 was
used (Gioncu, 1997). Computed values of plastic rotation supply of members are given in Table 3.
Table 3 Plastic rotation supply of members
Frame type member p pcor
C36 IPE300
HEB360
HEB300
HEB260
0.0946
0.1085
0.0926
0.1267
0.054
0.049
0.042
0.065
C33 IPE330
HEB240
0.0934
0.0963
0.053
0.044
In order to assess the structural response to the different ground motions, local and global damage
indexes are computed as follows:
lim.p
p
DLI
;
Li
Li
D
D
DgI
II
2
where: IDL is the local damage index, p is the plastic rotation demand and p.lim is the minimum
guaranteed plastic rotation supply and IDg is the global damage index
C36RIG C36SR1 C36SR2
C36DU1 C36DU2 C36DU3
C36DU4 C36DU5- joint type C1
- joint type C2
- joint type C3
- joint type C4
- joint type C5
- joint type C6 - pined joint
C33RIG C33SR1 C33SR2
C33DU1 C33DU3
C33DU4 C33DU5
Figure 3. Frame typologies
The q-factor is evaluated as the ratio between the ground acceleration leading to failure (u) and that
corresponding to the first yielding (e):
e
uq
The ultimate design limit state multiplier ud may be evaluated as:
),,min( mu
where: is the accelerogram multiplier corresponding to the attainment of 3% inter-storey drift
limit;
is the accelerogram multiplier corresponding to the plastic rotation capacity either in joint
or members;
m is the accelerogram multiplier corresponding to the plastic mechanism.
3. SELECTION AND SCALING OF RECORDS
Three different earthquakes have been chosen to study the dynamic response of the considered
structures. The ground records have been chosen so as to assure a large variety of Control Periods
Tc [7] (considered to be a very significant parameter for an earthquake if it is reported to a structure
with a given eigenperiod):
max
max2SA
SVTC
Table 4 Ground motions characteristics
Record PGA (g) TC
(sec)
EPA
(g)
scaling
factor
scaled
PGA (g)
EPV
(cm/sec)
Incerc-Bucharest NS, March 04, 1977 0.199 1.335 0.240 1.042 0.207 52.18
Kobe NS, 17 Jan 1995 0.836 0.622 0.704 0.355 0.297 97.17
Brienza NS, November 23, 1980 0.21 0.19 0.22 1.155 0.248 9.06
Table 4. gives the values of the most important parameters which characterise the considered
earthquakes (INCERC Bucharest - Romania, 4 March 1977, Kobe NS - Japan, 17 Jan. 1995 and
Brienza NS - Italy, 23 Nov. 1980), and in Fig. 4 are given the spectral accelerations of the same
eatrhquakes with a damping factor of 5%. For comparison, in Tab. 5 are given the fundamental
periods of the resulted structures.
Fig.4 Acceleration response spectra of records.
The scaling of records for structural analysis was done so as to correspond to the same Effective
Peak Acceleration (EPA).
Table 5 Fundamental periods of vibration of the analysed frames.
Frame C36RIG C36DU4 C36SR2 C36DU1 C36DU2 C36DU5 C36SR1 C36DU3
Fund. Period 1.33 1.35 1.47 1.52 1.61 1.63 1.69 1.86
Frame C33RIG C33DU4 C33SR2 C33DU1 C33SR1 C33DU5 C33DU3
Fund. Period 0.95 1.01 1.04 1.09 1.15 1.20 1.31
4. NUMERICAL RESULTS
As it was described in Chapter 2 – Analysed Frames, the parameters that have been monitored were
the Damage Index – ID and the q-Factor. A different parameter that has been taken into account for
results is the Effective Peak Acceleration (EPA) in the failure situation.
Fig. 5 EPA values for C36 frame series -
INCERC 1977 earthquake
Fig. 8 EPA values for C33 frame series -
INCERC 1977 earthquake
Fig. 6 EPA values for C36 frame series – Kobe
NS earthquake
Fig. 9 EPA values for C33 frame series – Kobe
NS earthquake
Fig. 7 EPA values for C36 frame series –
Brienza earthquake
Fig. 10 EPA values for C33 frame series –
Brienza earthquake
This parameter has been introduced in order to see the maximum values of EPA, as frame response,
for each earthquake. These values are computed for each failure criterion type (drift – 3%, rotation
capacity – 0.02 or 0.03 rad and mechanism). In Fig. 5-7 are given the charts with the obtained
values for C36 frame series (for the three earthquakes considered) and the similar values for the
C33 frame series are given in Fig. 8-10. In the charts are also given the mean values for each
criterion. The Global Damage Indices for each earthquake are given in Fig. 11 and Fig. 12 for frame
series C36 and C33 respectively.
Fig. 11 Global Damage Indices C36 frame series
Fig. 12 Global Damage Indices C33 frame series
For a better identification of the global and partial Damage Indices. These values are given in the
Table 6. It is to be noted that the Global Damage Index is not a linear sum of partial indices. Also,
the damage index in joints is computed for the ductility criteria stated in Chapter 2.
Table 6. The Global and Partial (Columns, Beams and Joints) Damage Indices.
Frame Global Damage Indices Frame Global Damage Indices
ID,col ID,beam ID,joint ID,G ID,col ID,beam ID,joint ID,G
INCERC-BUCHAREST NS, MARCH 04, 1977 ROMANIA
C36RIG 0.168 0.397 0 0.384 C33RIG 0.442 0.256 0 0.360
C36DU4 0.162 0.398 0 0.386 C33DU4 0.352 0.192 0.762 0.494
C36SR2 0.125 0 0.710 0.696 C33SR2 0.459 0 0.488 0.478
C36DU1 0.122 0 0.698 0.651 C33DU1 0.428 0 0.505 0.479
C36DU2 0.047 0 0.656 0.651 C33SR1 0.451 0 0.649 0.613
C36DU5 0.018 0.383 0.206 0.348 C33DU5 0.266 0 0.758 0.680
C36SR1 0.049 0 0.661 0.656 C33DU3 0.327 0 0.763 0.690
C36DU3 0.024 0 0.642 0.639
KOBE NS JAPAN
C36RIG 0.027 0.269 0 0.267 C33RIG 0.228 0.281 0 0.258
C36DU4 0.037 0.284 0 0.281 C33DU4 0.318 0.272 0.860 0.543
C36SR2 0.043 0 0.488 0.483 C33SR2 0.263 0 0.646 0.561
C36DU1 0.064 0 0.506 0.499 C33DU1 0.270 0 0.583 0.503
C36DU2 0.011 0 0.564 0.564 C33SR1 0.055 0 0.513 0.501
C36DU5 0.014 0.351 0.207 0.320 C33DU5 0.086 0 0.470 0.440
C36SR1 0.021 0 0.540 0.538 C33DU3 0.166 0 0.743 0.700
C36DU3 0.055 0 0.577 0.568
BRIENZA, 24 NOV. 1980, N-S
C36RIG 0.018 0.314 0 0.311 C33RIG 0.325 0.243 0 0.286
C36DU4 0.018 0.304 0 0.301 C33DU4 0.319 0.206 0.681 0.439
C36SR2 0.021 0 0.758 0.755 C33SR2 0.361 0 0.473 0.437
C36DU1 0.020 0 0.709 0.706 C33DU1 0.378 0 0.547 0.495
C36DU2 0.019 0 0.706 0.703 C33SR1 0.118 0 0.601 0.576
C36DU5 0.020 0.275 0.211 0.257 C33DU5 0.089 0 0.796 0.759
C36SR1 0.022 0.314 0 0.311 C33DU3 0.076 0 0.215 0.199
C36DU3 0.027 0 0.655 0.650
q-Factor values, may be regarded as a measure of global ductility, are given in Fig. 13 and Fig. 14
(frame series C36 and C33 respectively), comprising also the mean values for each earthquake.
They have been computed according to the three failure criteria, but only the minimum values are
represented in the Fig. 13 and Fig. 14.
Fig. 13 q factors for C33 frame series
Fig. 14 q factors for C33 frame series
5. CONCLUSIONS
Previous chapter presents the results obtained by the numerical analysis for the considered frame
series, for three main parameters: failure value of EPA, global damage index and the q-factor. In
order to draw the final conclusions, Table 7 sumarises the qualitative description of the structural
behaviour.
Table 7. Global structural behaviour. Frame series C36/
Types of joints
Output Results Frame series C33/
Types of joints
Output Results Comments
Rigid Frame
(theoretical)
RIG
EPA - high
ID - low
q-fct - high
Rigid Frame
(theoretical)
RIG
EPA - high
ID - low
q-fct - high
Ideal behaviour, but
it remains
theoretical
Real Rigid Frame
(rigid joints)
DU4
EPA - high
ID - high
q-fct - high
Real Rigid Frame
(rigid joints)
DU4
EPA - med-high
ID - high
q-fct - high
High values of
damages. This is the
consequence of real
rigid joints.
Frames with semi-
rigid joints:
(semi-rigid joints)
SR1-SR2
EPA - SR1-medium
- SR2 - med-high
ID - SR1 - low
- SR2 - high
q-fct - SR1 - med -low
- SR2 - high
Frames with semi-
rigid joints:
(semi-rigid joints)
SR1-SR2
EPA - SR1-medium
- SR2 - med-high
ID - SR1 - high
- SR2 - medium
q-fct - SR1 - med -high
- SR2 - medium
Frame behaviour is
influenced by the
joint propreties
Dual frames without
pinned joints
(rigid and SR joints)
DU1 - DU4
EPA - DU1-med.-low
- DU4 - med-high
ID - DU1 - high
- DU4 - high
q-fct - DU1 - medium
- DU4 - high
Dual frames without
pinned joints
(rigid and SR joints)
DU1 - DU4
EPA - DU1 - high
- DU4 - high
ID - DU1 - med-high
- DU4 - med-high
q-fct - DU1 - medium
- DU4 - medium
Frame behaviour is
influenced by the
joint propreties
Dual frames with
pinned joints
(rigid and SR joints)
DU3 - DU5
EPA - DU3-medium
- DU5 - medium
ID - DU3 - medium
- DU5 - high
q-fct - DU3 - low
- DU5 - high
Dual frames with
pinned joints
(rigid and SR joints)
DU3 - DU5
EPA - DU3 - low
- DU5 - low
ID - DU3 - high
- DU5 - high
q-fct - DU3 - low
- DU5 - low
Pinned joints have a
bad influence on
frame behaviour
Tab. 7. (Continued) Global structural behaviour. Horizontal Dual
Frames
(semi-rigid joints)
DU2
EPA - med-low
ID - high
q-fct - low
Intermediar results,
between the two
types of SR frames.
No conclusions for
just a frame
Concluding remarks:
- The perfect rigid frames resulted to have a very good behaviour, but they still remain
unrealisable. The real rigid frames have big damages, damages due to joint degradation mainly.
This may be related to the relatively low value of ductility (plastic rotation capacity of 0.02 rad. and
of 0.03 rad. respectively) that have been considered for computing the damage index in joints. It is
expected to improve the frame performances if the joints rotation capacity could be increased.
- The frames having semi-rigid joints have a good general behaviour. The problem is to find the
right joint properties that “fit” better for the given frame. Anyway, the semi-rigid frames seem to be
an alternative to the classical ones. The only problem is to establish the limits of their use.
- Dual frames with rigid / semi-rigid joints represent an open problem: it seems possibly to find
configurations with improved performances, but further studies are necessary.
- Dual frames with pinned joints in middle span have generally a bad behaviour and they should be
avoided. They have a bad behaviour of all the three parameters monitored.
- The main conclusion of this study is that global performances of MR frames can be controlled by
beam-to column joint properties. We can either improve or get worse frame performances. This
problem is important enough to focus further researches.
6. REFERENCES
Bertero V. (1997): General Report on Codification, Design and Applications, STESSA ’97
Behaviour of Steel Structures in Seismic Areas, Proceedings of the Second International
Conference 3-8 Aug. 1997, Kyoto, Japan.
Gioncu et al. (1997): Simplified Approach for Evaluating the Rotation Capacity of Double T Steel
Sections, STESSA ’97 Behaviour of Steel Structures in Seismic Areas, Proceedings of the Second
International Conference 3-8 Aug. 1997, Kyoto, Japan.
Kato. B. et al (1997): Seismic Damage of Steel Beam-to-Column Rigid Connections in the
Hyogoken-Nanbu Earthquake, STESSA ’97 Behaviour of Steel Structures in Seismic Areas,
Proceedings of the Second International Conference 3-8 Aug. 1997, Kyoto, Japan.
Lungu D. et al (1997): Effective Peak Ground Acceleration (EPA) Versus Peak Ground
Acceleration (PGA) and Effective Peak Ground Velocity (EPV) Versus Peak Ground Velocity
(PGV) for Romanian Seismic Records. Eurocode 8 - Worked examples
Mazzolani F. M. & Piluso V. (1996): Theory and Design of Seismic Resistant Steel Frames.
London: E& FN Spon.
ENV 1993-1-1 (1993) EUROCODE 3: Design of Steel Structures. Part 1.1. General Rules and
Rules for Buildings. Brussels: CEN, European Committee for Standardisation.
ENV 1998-1-1, (1993) - second draft - EUROCODE 8: Earthquake Resistant Design of Structures.
Part 1.: General Rules and Rules for Buildings - Seismic Actions and General Requirements for
Structures. Brussels: CEN, European Committee for Standardisation.
SAC Joint Venture (1996). Connection Test Summaries. SAC 96-02. Sacramento, California.