Performance of seismic steel beam–column moment joints

Abstract

During the Puebla earthquake on September 19, 2017, about 200 steel buildings were subjected to a severe shake in the State of Mexico, Puebla, Morelos, and Mexico City. Although null or minor structural damage was reported, design and assembly opportunity areas in some beam–column moment joints were identified during the post-seismic inspections. With this in mind, six beam–column connections, widely employed in steel frame structures by the local practice, were experimentally tested through natural-scale. A damage concentration with not enough ductile response for high-ductility steel structures was reported. Welding parameters and fabrications details are emphasized to improve the response for new structures, and a rehabilitation suggestion for existing structures is discussed. This research is part of a study that aims to establish the vulnerability of steel structures in Mexico.

Full text

Vol.:(0123456789)
Bulletin of Earthquake Engineering
https://doi.org/10.1007/s10518-022-01456-2
1 3
ORIGINAL ARTICLE
Performance ofseismic steel beam–column moment joints
EdgarTapia‑Hernández1 · AlejandroSantiago‑Flores2· HéctorGuerrero‑Bobadilla2
Received: 29 January 2022 / Accepted: 20 June 2022
© The Author(s), under exclusive licence to Springer Nature B.V. 2022
Abstract
During the Puebla earthquake on September 19, 2017, about 200 steel buildings were
subjected to a severe shake in the State of Mexico, Puebla, Morelos, and Mexico City.
Although null or minor structural damage was reported, design and assembly opportu-
nity areas in some beam–column moment joints were identified during the post-seismic
inspections. With this in mind, six beam–column connections, widely employed in steel
frame structures by the local practice, were experimentally tested through natural-scale.
A damage concentration with not enough ductile response for high-ductility steel struc-
tures was reported. Welding parameters and fabrications details are emphasized to improve
the response for new structures, and a rehabilitation suggestion for existing structures is
discussed. This research is part of a study that aims to establish the vulnerability of steel
structures in Mexico.
Keywords Connection· Experimental test· Existing buildings· Ductility· Damage
concentration
1 Introduction
In Mexico, buildings structured with steel as the main material are around 25% (including
steel–concrete composite buildings) of the total of high- and mid-rise buildings in Mexico
(CTBUH 2021), according to the Council on Tall Buildings and Urban Habitat. Although
this percentage represents one of every four, the high seismicity of the region, and the
densely populated cities, only one collapse and two buildings with severe structural dam-
age are reported in the Mexican history of damaged steel structures.
The only collapse was reported during the Michoacán earthquake on September 19,
1985. One of the three 21-story buildings of the Pino Suarez Complex collapsed, and the
* Edgar Tapia-Hernández
etapiah@azc.uam.mx
Alejandro Santiago-Flores
ing.alejandro.sf@gmail.com
Héctor Guerrero-Bobadilla
hguerrerob@iingen.unam.mx
1 Universidad Autónoma Metropolitana, 02200MexicoCity, Mexico
2 Universidad Nacional Autónoma de México, MexicoCity, Mexico

Bulletin of Earthquake Engineering
1 3
remaining two towers were severely damaged. Pino Suarez Towers were structured with
moment-resisting steel frames with box columns and truss girders. Under the same earth-
quake, an 11-story building on Amsterdam Street in the bed-lake zone and structured
with moment-resisting frames reported extensive damage without collapse (Osteraas and
Krawinkler 1989). Furthermore, the third damaged building happened during the Puebla
earthquake of September 19, 2017; a 3-story regular building structured with steel truss
moment-resisting frames experienced non-ductile extensive damage in columns on the first
floor (Tapia-Hernández and García-Carrera 2020). Although the statistics in favor of steel
buildings contrast to hundreds of buildings that collapsed or were damaged by other struc-
tural systems, this tendency should be understood as the destructive earthquake for steel
buildings has not arrived yet.
In local practice, steel moment-resisting frames are considered extensively in buildings
because of their ductility and energy dissipation capacity under intense seismic demands.
A complete joint penetration groove weld connection is commonly used between the
beam and column flange. For this purpose, the requirements for the welded unreinforced
flange-welded web (WUF-W) moment connection of the Prequalified connection for Seis-
mic Applications (AISC 358–16 2016) are usually considered. In the welded unreinforced
flange-welded web (WUF-W) moment connection, inelastic rotation is developed primarily
by the extensive yielding at the region adjacent to the face of the column. Connection rup-
ture is governed by the detailing requirements related to the welds joining the beam flanges
and the beam web to the column flange, the shape and finish of the weld access holes,
severe degradation due to beam local buckling, and ductile tearing due to beam buckling
(Ricles etal. 2002; AWS D1.8-2016; Han and Kim 2017).
The seismic design of beam-to-column steel joints is addressed adequately in the cur-
rent version of the local code (MCBC-2020 2020). Nevertheless, due to the lack of control
between the structural design, the manufacturing process, and the erection, the designers
are not at all able to predict performance with the required accuracy. Even though steel
moment-resisting frames behavior has been extensively studied and adopted in seismic
hazard regions, some of the critical issues of beam-to-column moment connections and
panel zones are not fully resolved following features of each practice (Tsai etal. 1995; Han
etal. 2019a, b). Han et al. (2014) concluded that WUF-W connections that satisfied the
qualification criteria for ductile frames did not meet the expected behavior as a function of
the beam depth and panel zone strength ratio.
Recently, experimental tests of beam-to-wide-flange-column connections have been
studied to identify failure modes, strength, and stiffness degradation of connections with
structural configurations according to particular local practices. For instance, Kim et al.
(2008) studied the conventional type of connection preferred in Korea’s construction mar-
ket. Lee etal. (2012) and Lee et al. (2019) describe an experimental investigation of the
behavior of welded flange-bolted web connections, which are commonly employed in
moment-resisting frames in Japan, including welded unreinforced flange-bolted web type
weak-axis connections. Zhang etal. (2019) tested four steel beam-to-column connections
with different structural forms under cyclic loading, considering the effects of the local
structural conditions in China.
With this panorama, this research aims to contribute to the evaluation of steel buildings
focusing the attention on actual connections by following the tendency and conditions of
the local market; considering the typical connection, where the beam flanges are designed
to resist the bending moment demands, while beam web resists the shear force.
Under this goal, six beam–column specimens of welded unreinforced flange-welded
web (WUF-W) moment connections with three configurations (two from each) were

Bulletin of Earthquake Engineering
1 3
cyclically tested. The study aims to improve the acquired knowledge and suggest design
improvements to actual connections. In particular, attention is focused on: (i) evaluating
the inelastic response of the connection of existing structures; (ii) estimating the ductility
and overstrength capacities and the evolution of the damage mechanism of such connec-
tions; (iii) calibrating nonlinear analytical models; and (iv) establish simple recommenda-
tions to improve the seismic response under moderate and intense earthquakes.
2 Test program
The experimental program consisted of six beam–column connections tests natural–scale
models (Fig. 1). Selected configurations were identified based on a survey conducted
through eleven of the main design desks and three steelwork fabricators. According to the
results, welded unreinforced flange-welded web (WUF-W) moment connections were iden-
tified as one of the most employed in the latest solutions implemented in new buildings. In
regard to the size of the specimen, the beam cross-section was selected so that the maxi-
mum demand at the actuator was limited to 40 percent of the maximum capacity. Later the
column was designed using the capacity design philosophy from the plastic hinging of the
beam, as discussed below.
In the design and manufacturing stages, the typical conditions and processes of the
local practice were considered, limiting the academic influence and the ought to be. This
research did not aim to test or develop a new improved connection but evaluate the existing
structures and current decisions on the design and manufacturing process.
Due to improper welding procedure, lack of root fusion (lack of penetration), undercut-
ting, lack of fusion, weld cracks, and porosity might be identified during the fabrication
process. A poor technique in the welding process may produce imperfections leading to
premature failure (Mandal 2016). For this purpose, non-destructive detection of inclusions
and lack of penetration during welding was implemented here and possible constructional
errors were properly corrected if necessary. Specimens were constructed by a commer-
cial fabricator using certified welders, and all welds had ultrasonically tested by certified
inspections following their routine process.
Steel beam-to-column moment connections were carefully designed following a strong
column-weak beam philosophy established by the local code (MCBC-2020), and follow-
ing the requirements for the prequalified connection (AISC 358-16). The Mexican seismic
provisions for steel structures are similar in rigor to the provisions employed in other high
seismic regions (e.g., AISC 341-16), despite significant differences in construction local
practice. In fact, while the details of tested connections are similar to those required in the
AISC 358 for WUF-W connections, there are some differences; for instance, the bevel of
the beam bottom flange are sometimes made in an overhead position, in contrast to the
required in the AISC 358.
Table1 summarizes the properties of each specimen, where two configurations joined
to the weak axis of the column were proposed (Fig.4b, c). The seismic performance of
beam-to-column connection for the column’s weak axis is of interest since, at the local
practice, all frames in both directions are usually designed to resist earthquake loading.
According to the local code (MCBC-2020), for members subject to flexure, sections are
classified as 1 (compact), 2 (compact), 3 (noncompact), and 4 (slender). The width-to-
thickness ratio limits for flanges of rolled I-shaped cross-sections for types 1 and 2 are
equal to
0.30√
E∕Fy and
0.38√
E∕F
y
, respectively. The width-to-thickness ratio for mem-

Bulletin of Earthquake Engineering
1 3
bers designed as highly ductile elements shall not exceed the limiting of section type 2,
which is the same for compact sections according to AISC 360-2016. Here, flanges of the I
406 × 46.20 kg/m (W 16″ × 31) beam are compact (section type 1) since the flange com-
pactness ratio of b/t = 6.25 meets the normative limit of 7.23.
In the local practice, welded unreinforced flange-welded web moment connections
utilize complete joint penetration groove welds (CPJ) to connect the beam flanges to the
455
I 457x59.8 kg/m
(W 18"x40 lb/ft)
403
I 406x46.2 kg/m
(W 16"x31 lb/ft)
46
112
PL 71x425
t= 13 mm
6
CJP groove weld
6
PL 86x260
t= 13 mm
CJP
256
I 356x110.4 kg/m
(W 14"x74 lb/ft)
403
I 406x46.2 kg/m
(W 16"x31 lb/ft)
46
112
PL 208x376x13 mm
CJP groove weld
63 L typ
6
PL 122x317
t= 13 mm
5
45
CPJ
256
I 356x110.4 kg/m
(W 14"x74 lb/ft)
403
I 406x46.2 kg/m
(W 16"x31 lb/ft)
60
122
PL 222x341x22 mm
CJP groove weld
53 L typ
6
PL 122x317
t= 13 mm
5
CJP
45
(a) Specimen CTP
(b) Specimen CTA
(c) Specimen CTA
Fig. 1 Connections details

Bulletin of Earthquake Engineering
1 3
column flanges as shown the overall fabrication details in Fig.1. For the erection process,
the beam web is bolted to a single-plate shear connection, and the single plate is welded
to the column flange. Then, the shear plate is used as a backing bar for a complete joint
penetration groove weld between the column flange and the beam web. To visually monitor
the inelastic deformation of the yielded region, specimens were whitewashed. The actual
dimensions of the cross-sections were measured and found to be in good agreement with
the values given in technical manuals.
All specimens were made from steel ASTM A992/A992M with a tensile yield strength
of Fy = 345MPa (50 kips). The tensile coupon test was developed to compute the actual
yield strength at the web and flanges following the ASTM E8/E8M-2016 standard test
methods. Ry is the overstrength material factor defined as the ratio between expected Fye
and nominal steel yield strength Fy. According to specialized codes (e.g., MCBC-2020;
AISC 341-16), the specified value of the overstrength material factor for rolled shapes is
equal to Ry = 1.10. As a result, nominal yield strength was exceeded (Fy < Fye), as shown in
Table2. The average of the obtained overstrength material factor is Ry = 1.12.
The column was placed horizontally in tests, as shown in Fig.2. Near the actuator, a
set of angles provided restraint to lateral-torsional buckling perpendicular to the plane of
the beam at a distance of 700 and 1500mm. The experiment was conducted by applying
cyclic loads at the top of the cantilever to the vertically mounted specimens. The AISC
341 (2016) cyclic displacement history was considered based on SAC/BD-97/02 (SAC
Table 1 Properties of the studied specimens
Specimen Characteristics Beam Column Number of
specimens
CTP (Fig.4a) Connection in the strong axis of
column
I 406 × 46.2kg/m
(W 16″ × 31)
I 457 × 59.8kg/m
(W 18″ × 40)
2
CPA (Fig.4b) Connection in the weak axis of
column
I 406 × 46.2kg/m
(W 16″ × 31)
356 × 110.4kg/m
(W 14″ × 74)
2
CTA (Fig.4c) Connection in the weak axis of
column
I 406 × 46.2kg/m
(W 16″ × 31)
356 × 110.4kg/m
(W 14″ × 74)
2
Table 2 Values for the material
overstrength Location Yielding strength Fye Overstrength
material factor,
Ry
N/mm2ksi
Flange 390.2 56.6 1.132
Flange 390.1 56.6 1.133
Flange 372.2 54.0 1.080
Flange 395.6 57.4 1.147
Flange 363.4 52.7 1.054
Flange 364.0 52.8 1.056
Web 398.4 57.8 1.156
Web 388.2 56.3 1.126
Web 399.2 57.9 1.158

Bulletin of Earthquake Engineering
1 3
1997) for beam-to-column moment connections. The actuator was anchored to a set of con-
crete blocks, and it was attached to the beam end in the horizontal direction. The loading
sequence was run under displacement control by a quasi-static method.
A set of Tokyo Sokki Kenkyujo linear voltage displacement transducer (LVDTs) was
utilized to measure the local nonlinear response in the specimens (Fig. 2). LVDT-1 and
LVDT-2 measured the displacement at the beam span’s quarter (L/4) and a half (L/2).
LVDT-3 measured the possible horizontal displacement at the connection, and LVDT-4
measured the position of supports. Besides, a set of 36 LEDs was installed to measure the
displacement of preidentified places. Measured values of the LEDs (not shown here) were
used as a backup of the system, and data did not reveal noticeable differences (Santiago-
Flores 2021). The rotation capacity was computed by the actuator displacement and the
distance to the beam centerline.
The loading system was designed to experience negligible deformation when loaded to
the capacity of the actuator. The system performed well, except at Specimen CPA-2, where
some slip between the base plate and the floor was observed. It was corrected in time at the
elastic stage through the anchor system at the support.
3 Experimental response evaluation
Each of the six tested specimens failed due to repeated cycling at large beam-tip displace-
ments. However, consistent response for all specimens was developed, as discussed below.
Flaking of the whitewash was first observed during the fifth loading sequence on the beam
web around the access holes. During the first cycle of the sixth loading sequence, local
2,420 mm
Beam
Column
Actuator
2,420 mm (99.2 in)
Anchor
Reaction
wall
Floor
Connection
Lateral
bracing
9101040
1300
1,500
500
406 mm
(16 in)
Led sensor
LVDT
L/2
L/4
LVDT
Led
Led
LVDT
Concrete
blocks
Fig. 2 Test setup

Bulletin of Earthquake Engineering
1 3
buckling of the beam flange was developed, as shown in Fig.3b, and a slight deterioration
of the loop occurred. However, modes of flange buckling (amplitude less than 2mm) did
not lose strength or stiffness in the specimens.
At the first cycle of the eighth loading sequence, beam flanges on both sides had pro-
nounced local buckling that appeared and disappeared cyclically until the fracture at the
end of the test (Fig.3c). The web of the beam buckled beyond the shear tab. The fracture
occurred throughout the tenth sequence, accompanied by significant amplitude buckling
of flanges (Fig.3d). Connections were tested until specimen failure occurred or until the
instability was reached to protect the instruments. The representative match between the
experimental results and the applied loading protocol is depicted in Fig.3.
Specimens maintained load-carrying capacity even when severe damage concentration
occurred in either beam flange. In addition, the column behaved elastically; no damage was
reported in the column components of any of the specimens. Therefore, the failure was
characterized by the local damage at the weld access hole, gradual fracture at the weld-to-
column interface at the beam bottom flange, and flange local buckling.
The load-deformation data were carefully computed for each test for cyclically applied
loads. Obtained curves of cantilever tip load against the column rotation at the beam–col-
umn joint are shown in Fig.4. The resulting hysteretic loops provide the basic data for
determining the behavior of the tested connections. The dashed lines in Fig.4 represent
the resistance of 80% of the plastic moment calculated as the product of the plastic section
modulus Z and the measured yield strength Fy.
The global inelastic responses of the six specimens are summarized in Table3, includ-
ing initial stiffness Kel at the elastic range; the stiffness at the plastic-strain hardening range
(a) Cyclic displacement history
(b)
Initial damage
(c)
Buckling in both flanges
(d)
Fracture at the beam web
-30
-20
-10
0
10
20
30
0102030405060708
09
0
Displacement (cm)
Time (min)
Yielding
begins
Fig. 6b
Fig. 6c
Fracture at
weld access hole
Fracture
at webFig. 6d
End of test
Cycles 6 6 6 4 2 2 Continue loading at increments of 0.01 rad,
Angle (rad) 0.375% 0.5% 0.75% 1% 1.5% 2% with two cycles of loading at each step
Fig. 3 Typical damage evolution

Bulletin of Earthquake Engineering
1 3
Specimen 1(CTP-1) Specimen 2(CTP-2)
(a) CTPconfiguration
Specimen 1(CTA-1) Specimen 2(CTA-2)
(b) CTAconfiguration
Specimen 1(CPA-1) Specimen 2(CPA-2)
(c)
CPAconfiguration
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
M/Mp
Moment, M[t-m]
Rotation,
θ
[rad]
0.8Mp
0.8Mp
Fig. 4 Obtained hysteretic loops
Table 3 Global inelastic
responses Specimen Kel (t-m/rad) Kp/Kel θy (rad) θmax (rad) μ M/Mp
CTP-1 2031 0.195 1.51% 5.02% 3.26 1.19
CTP-2 2000 0.214 1.79% 5.92% 3.33 1.21
CTA-1 1677 0.267 1.70% 6.81% 4.00 1.19
CTA-2 1886 0.279 1.71% 5.67% 3.29 1.18
CPA-1 1919 0.259 1.69% 5.63% 3.32 1.19
CPA-2 1922 0.290 1.78% 5.70% 3.20 1.21

Bulletin of Earthquake Engineering
1 3
as a function of the initial stiffness Kel; the angle at yielding θy; the maximum angle θmax;
the developed ductility μ (= θmax/θy); and the normalized maximum strength M/Mp.
Based on specialized codes (e.g., AISC 341-16; MCBC-2020), beam-to-column con-
nections used in ductile frames should be capable of accommodating a minimum flexural
resistance measured at the column face, and a story drift angle θ. In addition, the deter-
mined flexural resistance of the connection should be equal to at least 0.80Mp, and it should
be capable of accommodating an angle equal to θ > 0.04rad.
As discussed above, none of the specimens failed prematurely (Fig.4). Loops consist-
ently exhibited stable characteristics, and they met the performance requirements, where
the plastic moment Mp was reached. For the studied connections, the maximum moment
was ranged from 118 and 121% of the plastic moment. Reported values are greater than
100% due to the strain hardening and material overstrength. Since the overstrength material
factor is Ry = 1.12, based on coupon tests (Table3), the strain hardening factor has to be
around 1.05–1.08.
A rapid loss of strength and stiffness in the connection is evident for angles θ greater
than 4–5% rad. because of local buckling of the beam flanges and web and the consequent
fracture (see Fig.4, Table3). In addition, the reported local ductility is less than μ < 4.0.
This result is critical because the connection is considered in the local practice as a part
of ductile structures with a global ductility factor of 4.0. This limited inelastic behavior is
mainly due to imprecisions at the weld access hole that produced a stress concentration in
the beam web.
Studied configurations developed failure by following the same damage propaga-
tion: initially, gradual stiffness and strength deterioration due to damage concentration at
the weld access hole, followed by gradual tearing of beam web and the fracture in beam
flanges due to local buckling under the significant inelastic incursion. The state after the
experiment is similar between the configurations with slight variations; this is, some con-
nections fractured near the access hole in the beam web and beam flange while others expe-
rienced fractures in the beam flange close to the complete joint penetration (CJP) weld.
Cracks that originated near the access hole propagated through the beam flange in one case
(Fig.5a) but propagated in the longitudinal direction and remained in the beam web in the
other cases (Fig.5b, c). In general, the final capacity is characterized by severe damage to
remaining stable even under gravity loading.
It is expected that buildings with these connection conditions were near collapse
in aftershocks and should not continue to be occupied. At the same time, no damage
was reported in beams and columns, despite those elements being designed as a part
(a)
CTPconfiguration
(b)
CTAconfiguration
(c)
CPAconfiguration
Fig. 5 Final state of the studied connections

Bulletin of Earthquake Engineering
1 3
of ductile frames to develop extensive distortion and localized failures in beams in the
form of plastic hinges. The damage evolution is related to a large reserve of strength
by structural components and a non-necessarily ductile or predictable damage under
intense seismic demands, particularly for existing buildings.
3.1 Weld access hole
The beam flanges and beam web are welded directly to the column in the studied
moment connection. Complete joint penetration groove welds are usually used in local
practice to connect the side plate of the beam and column. In the fabrication process,
the beam web creates an obstacle for the welding at both flanges, such that the weld
access hole allows the deposition of weld metal to connect the beam flanges.
Connections employ special seismic requirements on size and finish to reduce stress
concentrations in the region. According to the drawings, all weld access holes had to
meet the standard details and tolerances of AWS D1.8 (2016); however, the final prod-
uct is not fully achieved in practice. AWS D1.8 (2016) is intended to cover applications
to welded joints in resisting systems, including connection details, materials, work-
manship, and inspection issues. In Fig.6, a close view of some access holes before and
after is shown to note the damage concentration. Notches or gouges in the weld access
hole were not entirely removed by grinding and faired. In addition, the finish of the
surface roughness was not a smooth contour and served as a start point for the damage
propagation at the beam web and, therefore, the ductility reduction of the whole joint.
A flag is set on the satisfactory performance of moment-resisting connections of
existing buildings by following this practice because there were not very careful in
meeting the prescribed acceptance criteria of specialized codes (e.g., AWS D1.8 2016;
MCBC-2020). The failure by fracture at or near the beam-flange groove welds led to
an inelastic response that is not enough stable as the one required for ductile steel
frames (designed for μ = 4.0 or greater). This is, although the frames were carefully
designed with a strong column-weak beam mechanism following the capacity design
philosophy, the connection performance would govern the structure’s capacity. There-
fore, considerable judgment will be needed when comparing the expected performance
of existing steel buildings with the effect of the results of this testing program.
Additionally, similar studies (Tsai and Popov 1989; Engelhardt and Husain 1993;
Engelhardt and Sabol 1994) identified that the response of welded flange connection
might be improved for beams having a ratio Zf/Z larger than 0.7. Zf is the plastic sec-
tion modulus of the beam flanges alone, and Z is the plastic modulus of the entire
beam section. Initially, the Zf/Z factor was introduced in the referred studies because
web rupture due to slip was critical in post-Northridge WUF-B connections. Here, it
would improve the flange performance even with imperfections. A more stable inelas-
tic response of moment connection might be reaching for beams with a Zf/Z ratio larger
than 0.70.
For the wide flange sections I 406 × 46.20 kg/m (16″ × 31) and I 305 × 38.70kg/m
(W12″ × 26) values are Zf/Z = 0.54 and Zf/Z = 0.65, respectively. These relatively small
Zf/Z values imply that the web contribution is larger than the flange contribution. This
tendency might partially reduce the damage concentration in the beam flange.

Bulletin of Earthquake Engineering
1 3
3.2 Local buckling offlanges
In tests, deterioration of the hysteresis curves was caused by local buckling of the beam
flange and degradation related to the large inelastic deformation that led to the fracture. In
fact, fracture or tearing occurs only after another failure mode has been initiated. For the
above, it is interesting to study the flange buckling when evaluating the seismic perfor-
mance of steel connections.
To accurately represent seismic damage of non-structural and structural elements as part
of a building, it is useful to relate other geometric and material parameters of the com-
ponents with the corresponding damage states. In particular, deterioration resulting from
local buckling is most significant on reversed and repeated inelastic cycles like those that
happen under intense seismic demands. However, despite this trend, local buckling might
develop significant energy dissipation and rotation capacity before the fracture, if norma-
tive requirements are considered.
Specimen CTP-1Specimen CTA-2Specimen CPA-2
SpecimenCTP-1 Specimen CTA-2Specimen CPA-2
(a) Initial state
(b) After the test
Fig. 6 Geometry details of weld access holes of tested connections

Bulletin of Earthquake Engineering
1 3
It is accepted that local slenderness ratios b/t of steel sections are among the primary
geometric parameters that affect the pre- and post-buckling behavior during an intense seis-
mic demand. Figure 7 shows the maximum plastic rotation achieved from some experi-
mental tests with different slenderness ratios. The ductility requirements for highly ductile
members require that b/t = 7.23 (e.g. AISC 341-2016; MCBC-2020). The minimum angle
(θ > 0.04 rad) for prequalified connections was also included, according to the AISC 356
(2016).
The results in Fig.7 provide evidence that the normative limit is appropriate for achiev-
ing ductile behavior, and a high dependency between the slenderness ratio b/t and the ine-
lastic response was underlined. In fact, the experimental tests discussed in this study have
shown an improved performance than the ones reported in similar studies.
Because of the above, finite element models were constructed to analyze the stress dis-
tribution related to the local buckling at the beam flange in tested connections. The element
model and mesh sizes were adopted from previous studies (Cho and Han 2022). The finite
element (FE) model used a 3-D solid element for the flanges and webs in beam and column
elements, stiffener plates, and panel zones (Fig.8a). The column was hinged at two ends,
and a horizontal load was applied at the beam end at the same location as the actuator dur-
ing the experimental tests. Pushover analyses were carried out.
Figure8b shows the distribution of Von-Mises stresses at a drift rotation of 1.8%, cor-
responding approximately to the time when the flange buckling begins. The location of
maximum stress in a color scale was detected in beam flanges and access holes, which
corresponds to the observed damage in tested connections. Concentrated forces from the
beam flanges are transferred locally into the column flanges due to the joint configuration
in contrast to the stress distribution in the beam cross-section, where the influence of the
connection is neglectable (Fig.8c). Then the local buckling is triggered by cyclic loading
at the point, according to the shown stress distribution.
Additionally, a metallographic analysis of the local buckling region was performed
at various points to verify that no disturbance at the material microstructure induced
the damage concentration and to reveal the metal structure after the test. For this pur-
pose, cross-sectional samples were carefully cut from the damaged flanges (Fig.9a). A
metallographic microscope Carl Zeiss was used in the analysis following specialized
Fig. 7 Plastic rotation as a func-
tion of beam flange slenderness
0.02
θ
0.03
0.04
0.05
0.06
0.07
5.06.0 7.08.0 9.
0
max
(rad)
b/2t
AISC 341-2016 (0.04 rad) Code limit (b/t= 7.35)
Ricles et al. 2000 Choi et al. 2000
Gilton et al. 2000 Venti & Engelhardt 2000
Chen 1999 Engelhardt & Sabol 1998
Noel & Uang 1996 Whittaker & Gilani 1996
This study

Bulletin of Earthquake Engineering
1 3
recommendations as ASTM E3 (2017) to prepare metallographic specimens. The sam-
ples were polished with discs and diamond suspension, as shown in Fig.9b.
According to the results, no decarburization or globular oxides were observed, and
neither of the samples were microstructurally affected by any external heat sources
(welding, oxy cut, etc.). Therefore, the plastic deformation was related only to the
applied overstress. Furthermore, a banding (elongated pearlite), called cold deforma-
tion lines, was identified with a ductile response: cavities and area reductions related to
an evident elongation at the microstructure of the steel, as shown in Fig.9c. Therefore,
the damage evolution was entirely dominated by the concentration of the stresses at the
beam flange due to the joint configuration.
M
máx
Column
Beam
flange
Beam
web
Access
hole
Compression
Tension
finite element model
(a) Studied model (b) Stress distribution in the (c) Flange forces distributed
through column
Fig. 8 Analytical model and critical stress distribution
samples
(a)
Area with the local damage
(b)
Example of the analyzed
(c)
Metallography of the flange (100X)
Fig. 9 Metallographic analysis of the plates with damage concentration

Bulletin of Earthquake Engineering
1 3
4 Assessment ofthedamage implication
As discussed earlier, the resistance of the connection has to be at least equal to 0.80Mp
of the connected beam for ductile structures. This strength limit was adopted initially
based on judgment to assure that structures would not be pushed into the strength
degrading range (AISC 341-16). The objective was to prevent collapse even under
severe earthquakes. Moreover, the strength degradation is set at 0.04 rad based on a
probabilistic evaluation of the performance capability, demonstrating that frames can
meet the intended performance goals (FEMA 356 2000).
Currently, the capacity to perform nonlinear analysis and the definition of acceptable
performance have evolved substantially. The societal expectations about the structural
performance after a strong earthquake were also evolved considerably. Voices coming
from various disciplines express disagreement with the current earthquake-resistant
design approach. Initiatives in which sustainability and resiliency are made possible by
adequate damage control are beginning to outline a new path for earthquake-resistant
design during the conception, construction, and design under intense seismic design
(Terán-Gilmore etal. 2020). To ensure that structures remain at the performance level
of Life Safety (LS) for both the Design Basis Earthquake and Maximum Considered
Earthquake hazard levels might not be enough anymore due to the implicit damage.
Namely, despite connection qualification in specialized codes primarily focuses on
the level of plastification achieved (0.8Mp), and the deformation incursion (0.04 rad),
the accumulated damage is not enclosed as a part of the current necessity to develop
sustainable and resilient steel structures. The strength degradation at joints is related
to increased rotation demands from P-Δ effects and the likelihood of frame instability,
even if no damage is reported in beams and columns.
Generally, the seismic performance relates to damage incurred to the structural and
non-structural components. While the structure performance is a continuum, it is con-
venient to identify discrete performance levels for the main structural elements and
other components that may affect building function, property protection, and safety,
through nonlinear analysis. The first significant guideline on applying nonlinear analysis
for practical purposes in Mexico was published in 2017 (MCBC-17 2017). The code
provided guidance for nonlinear analysis, components modeling, and acceptance criteria
for new building design based on the ASCE 41 (2013).
In standards, the force–deformation relationship is intended to represent the cyclic
envelope that reflects the strength degradation under cyclic loading. To capture the
dynamic response, the choice of component curves depends on how hysteretic models
capture the cyclic degradation as the analysis proceeds (Ibarra etal. 2005). The cyclic
envelope varies as a function of the cyclic loading history, and backbone relationships
are based on standardized loading protocols (Fig.6a).
The initial strength, stiffness, and post-yielding standardized backbone relation-
ships of cross-sections are determined based on principles of mechanics, analytical
studies, and experimental data (Ibarra et al. 2005; FEMA P440A 2009). As reported
above, element strengths often deteriorate under large inelastic cyclic deformations due
to local buckling, fracture, imperfections, or other phenomena. In addition, experimen-
tal tests on steel structures and their components allow assessing if current normative
approaches adequately simulate the element behavior (Deierlein et al. 2010; Elkady
and Lignos 2015). Thus, nonlinear dynamic analyses should be calibrated to ensure

Bulletin of Earthquake Engineering
1 3
that models might represent the cyclic degradation reported in experimental tests and
implied by model parameters in standards or other sources.
With this in mind, idealized force–deformation curves were computed from the obtained
experimental data to compare the performance objectives. For this purpose, envelope
curves were drawn by extracting the peak moments for a cycle at each angle. Envelopes are
essentially the same out to 0.035rad, but they differ significantly for more significant drifts
depending on the damage sequence. Later, a smooth backbone curve was drawn through
the average of the envelope curves, computed as a series of linear segments. Envelope
curves of the six specimens and the backbone curve as a black thicker continue-line are
shown in Fig.10.
Obtained backbone curve corresponds to a ductile element indicated in specialized
codes (ASCE 41-2017, MCBC-2020) with: (i) an elastic range, (ii) a plastic-strain hard-
ening range, and (iii) a plastic strength-degrading range (Fig.10). Further information to
develop structural modeling parameters based on experimental data might be found in
ASCE 41-17.
The differences between the backbone and analytical curves are insignificant when the
deformation incursion is controlled (θ ≤ 0.035 rad). Curves have a nearly identical trend
with slight variations until the strength degradation at the life safety structural performance
Level (LS). Testing after this level might be hazardous to the equipment and staff, which is
why it is seldom done.
Backbone curves are defined according to Fig.11 with a flexural stiffness until an effec-
tive yield point, associated with the yield strength Qy at point B; the peak strength Qu,
related to the plastic deformation up to the peak strength. Point C represents the ordinate in
which significant strength degradation begins until the residual strength Qr. Beyond point
D, a substantially reduced strength is modeled until point E, related to the loss of the ele-
ment’s load-carrying capacity.
Theoretical backbone curves from ASCE 41-2013 and 2017 are compared with the
moment-drift average envelopes in Fig. 12 for a plastic hinge in a beam element under
bending demand. The normative values in backbone curves were established based on
extensive probabilistic assessment of the performance capability of structural systems
(FEMA 356 2000), demonstrating with high statistical confidence that frames with these
types of elements and connections can meet the intended performance goals.
-1.4
-1.0
-0.6
-0.2
0.2
0.6
1.0
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
Moment [M/Mp]
Moment, M[t-m]
Rotation,
θ
[rad]
CTP1-01
CTP1-02
CPA-01
CPA-02
CTA-01
CTA-02
Envelope
0.8Mp
0.8Mp
Fig. 10 Backbone curve derived from envelopes of experimental test data

Bulletin of Earthquake Engineering
1 3
The differences between the normative approaches for beams are ASCE 41-2017
includes the change in rotation caused by shear deformation η, using Eqs.(1) and (2) for
beams.
The welded unreinforced flange-welded web (WUF) approach, which is the same
between those two versions, was also computed. The proposed equations for predicting
the inelastic response from θ and value a and b in Fig.11 are summarized in Eqs. (3)
and (4), where d is the beam depth. Equations are associated with a large coefficient
of variation values equal to COV = 0.50 and COV = 1.1, respectively, because these
(1)
𝜃
y=
ZF
ye
L(1+𝜂)
6EI
(2)
𝜂
=
12EI
L2GA
s
Fig. 11 Idealized backbone
curves of structural elements
-1.4
-1.0
-0.6
-0.2
0.2
0.6
1.0
1.4
-50
-40
-30
-20
-10
0
10
20
30
40
50
-0.08-0.04 0.00 0.04 0.08
Moment [M/Mp]
Moment, M[t-m]
Rotation,
θ
[rad]
Envelope
Beam flexure (ASCE 41-2017)
Beam flexure (ASCE 41-2013)
WUF (ASCE 41-2017)
0.8Mp
0.8Mp
IO
IO
LS
LS
Immediate Occupancy (IO)
Life Safety (LS)
Fig. 12 Idealized force–displacement curves and the obtained backbone curve

Bulletin of Earthquake Engineering
1 3
parameters vary considerably due to the observed brittle failure modes (Maison and
Speicher 2016; Lignos etal. 2018).
The linear response depicted between the unloaded element and the effective yield point
(effective flexural stiffness Ke) is stiffer in the analytical approaches, as shown in Fig.12.
Therefore, the ductility expectative of the connection might be larger than can be devel-
oped by the existing joint. It is shown that a slope of 3% from point B to point C (Fig.11)
correctly fits the strain hardening reported by experiment data. The ASCE model for WUF
connections tends to underestimate the deformation capacity of the peak strength on 0.84
times the envelope. It was also found that for the beam model, the envelope curve is nearly
the same in terms of the peak values with the ASCE 41-13 model (without the shear defor-
mation contribution).
Experiments suggest that the ASCE 41 nonlinear component models do not adequately
reflect the expected connection behavior under cyclic loading following a concentrated plas-
ticity model. This observation underlines the relevance of performing experimental tests of
connections of existing buildings. In any case, the strength deterioration of the envelope is
related to a life safety performance level (LS) at the idealized force–displacement curves
(Fig.12). Given the accumulated damage in the connections, the current criteria based on the
degradation control (0.8Mp; 0.04rad) might not be appropriated anymore for this configura-
tion (WUF), mainly when the resilience societal expectations are considered.
Whereas the plastic hinge in beam elements is controlled as a part of the capacity design
philosophy, substantial yielding at the connection is difficult to identify and repair. Fur-
thermore, the plastic hinge in the beam and the damage in connection reported on the
experimental test differ in terms of resilience; despite a stable hysteresis and a proper
energy dissipation for a large number of cycles were observed (Fig.4). The change of yield
mechanism and failure mode had a significant impact on the relative performance of the
structures. Although beams and columns remain elastic throughout the range of applied
load, the building would be unoccupied, and rehabilitation may be deemed economically
impractical.
A qualitative definition of acceptable performance is established by explicitly consid-
ering the adequate levels of damage and the lines of defense that comprise the structure.
Therefore, a displacement control should imply a conceptual understanding of the lateral
deformations resulting from the damage mechanism evolution. For example, it would be
desirable that the ductility provided by connection might act as a second or third line of
defense against collapse after the failure of the primary load transfer mechanism. Thus,
nonlinear models would allow predicting the building response only if the performance of
the connections is explicitly considered (Shugyo 2003; Buonopane and Schafer 2006; Kim
and Truong 2020).
5 Conclusions
Six specimens of beam-to-column moment connections with three configurations (two
from each) were cyclically tested through a natural-scale. Based on the results, practical
recommendations are given for the seismic-resistant design of steel frames with welded
(3)
qu=aqy=(0.051 −0.0013d)qy
(4)
qr=bqy=(0.043 −0.0006d)qy

Bulletin of Earthquake Engineering
1 3
unreinforced flange-welded web (WUF-W) moment connections for new and existing
buildings. From experimental data, nonlinear models were calibrated to ponder the defi-
nition of acceptable performance from the implications of a conceptual understanding
of the lateral deformations resulting from the damage mechanism evolution. The follow-
ing main conclusions were obtained based on the experimental and analytical results:
• Tested moment–resisting connections under significant inelastic deformations
demands developed an unexpected damage concentration that is not ductile enough
for a high-ductility steel structure. Although the seismic code adequately addresses
the conceptual requirements, experimental data indicated that connections were not
as ductile as expected due to the lack of control between the structural design, the
manufacturing process, and the erection.
• The beam flange reported a damage concentration and fracture near weld access
holes, despite specimens meeting the requirements for compact sections. Connection
damage evolution had to be controlled through detailing requirements related to the
welds joining and the surface roughness of the weld access holes. According to the
drawings, all weld access holes had to meet the normative details and tolerances;
nevertheless, the final product was not fully achieved and served as a starting point
for the damage propagation of the beam web.
• Despite the damage concentration, it was demonstrated that tested connections could
reliably achieve a flexural resistance of 0.80Mp, and an inelastic rotation of 0.04rad
before the strength degradation. The idealized force–deformation curves computed
from experimental data were compared with nonlinear analytical models to evaluate
the seismic performance. It was highlighted that the level of plastification achieved
does not match the societal expectations based on sustainability and resiliency ini-
tiatives. Despite the stable inelastic response, the implicit damage at the connection
distances the building of the immediate occupancy building (IO), even if no damage
was reported in beams and columns. The conceptual understanding of the lateral
deformations resulting from the damage mechanism evolution and its implications
was underlined.
• In the context of the local code, modifications are required to reduce and control
the possible damage concentration at the connection region due to overall careless-
ness in the design and construction process. For new buildings, a reduction of the
ductility factor from 4 to 3 for ductile steel frames with welded unreinforced flange-
welded web moment connections was proposed to the seismic code. Additionally,
adding a stiffer at the region of the weld access hole might reduce the damage con-
centration and the fracture with welded unreinforced flange-welded web moment
connections. For existing buildings, a detailed analysis is required to evaluate the
adequacy of these connections in each case. Connections might be modified by add-
ing flange cover plates or stiffer plates to initiate yielding away from the connection
location.
Acknowledgements The master fellowship granted to the second author by the National Science and Tech-
nology Council of Mexico (Conacyt) is gratefully acknowledged. Also, the authors wish to acknowledge
Marcos Chávez-Cano, Carlos Moss, Gabriel Guerra, and the Staff of the Materials and Structures Labora-
tory of the Engineering Institute, UNAM.
Author’s contributions Beam–column connections were experimentally tested through natural-scale. A
damage concentration with not enough ductile response for high-ductility steel structures was reported. The

Bulletin of Earthquake Engineering
1 3
study aims to establish the vulnerability of existing steel structures. Test results were compared with ASCE
41 predictions.
Funding Consejo Nacional de Ciencia y Tecnología (CONACYT) for the master fellowship granted to the
second author.
Availability of data and material Analytical models, earthquake ground motions, and results are available
upon request to the corresponding author.
Code availability Not applicable.
Declaration
Conflict of interest The authors declare no conflicts of interest with any individual or organization, either
academic or professional.
References
AISC 341-16 (2016) Seismic provisions for structural steel buildings, ANSI/AISC American Institute of
Steel Construction. Chicago, Illinois
AISC 358–16 (2016) Prequalified connections for special and intermediate steel moment frames for seismic
applications, ANSI/AISC American Institute of Steel Construction. Chicago, Illinios
ASCE 41-2013. Seismic evaluation and retrofit of existing buildings, Standard ASCE/SEI, American Soci-
ety of Civil Engineers
ASCE 41-2017. Seismic evaluation and retrofit of existing buildings, Standard ASCE/SEI, American Soci-
ety of Civil Engineers, 550 pp
ASTM E3 (2017) Standard practice for preparation of metallographic specimens. American Standard Test-
ing and Materials, ASTM
AWS D1.8 (2016) Structural welding code—seismic supplement. AWS D1.8/D1.8M, American National
Standard, American Welding Society. Chicago, Illinois
Buonopane SG, Schafer BW (2006) Reliability of steel frames designed with advanced analysis. J Struct
Eng ASCE. https:// doi. org/ 10. 1061/ (ASCE) 0733- 9445(2006) 132: 2(267)
Chen SJ (1999) Design of ductile steel beam-to-column connections for seismic resistance. In: Workshop on
design technologies of earthquake-resistant moment resisting connections, Taipei, Taiwan
Choi J, Stojadinovic B, Goel SC (2000) Parametric tests on the free flange connection. Report SAC/
BD-00/02, SAC Joint Venture
CTBUH (2021) Council on tall buildings and urban habitat. United States of America. https:// www. skysc
raper center. com/ build ings. Accessed 1 Apr 2001
Deierlein GG, Reinhorn AM, Willford MR (2010) Nonlinear structural analysis for seismic design. A
guide for practicing engineers. NEHRP Seismic Design Technical Brief No. 4, NIST GCR 10-917-5.
National Institute of Standards and Technology. October
Elkady A, Lignos DG (2015) Analytical investigation of the cyclic behavior and plastic hinge formation in
deep wide flange steel beam-columns. Bull Earthq Eng 13(4):1097–1118
Engelhardt MD, Husain AS (1993) Cyclic loading performance of welded flange-bolted web connections. J
Struct Eng Am Soc Civ Eng 119(12):3537–3550
Engelhardt MD, Sabol TA (1994) Seismic-resistant steel moment connections: developments since the 1994
Northridge earthquake. Earthq Eng Struct Dyn 1(1):68–77
Engelhardt MD, Sabol TA (1998) Reinforcing of steel moment connections with cover plates: benefits and
limitations. Eng Struct 20(4–6):510–520
FEMA 356 (2000) Prestandard and commentary for the seismic rehabilitation of buildings. Prepared by the
American Society of Civil Engineers for the Federal Emergency Management Agency. Washington,
D.C. United States of America. November
FEMA P440A (2009) Effects of strength and stiffness degradation on the seismic response of structural sys-
tems. FEMA Report P440A, Washington, D.C. June
Gilton CS, Chi B, Uang CM (2000) Cyclic testing of a free flange moment connection. SAC Report
00/19, SAC Joint Venture

Bulletin of Earthquake Engineering
1 3
Han SW, Jung J, Moon KH (2014) Seismic behavior of welded unreinforced flange-welded web moment
connections. In: Proceedings, Tenth U.S. national conference on earthquake engineering, Anchor-
age, Alaska, July 21–25
Han SW, Kim NH (2017) Permissible parameter ranges of access hole geometries for WUF-W connec-
tions. Earthq Spectra 33(2):687–707
Han SW, Jung J, Ha SJ (2019a) Seismic performance of WUF-W moment connections according to
access geometries. Earthq Spectra 32(2)
Han SW, Moon KH, Jung J (2019b) Cyclic performance of welded unreinforced flange-welded web
moment connections. Earthq Spectra 30(4):1663–1681
Ibarra L, Medina R, Krawinkler H (2005) Hysteretic models that incorporate strength and stiffness dete-
rioration. Earthq Eng Struct Dyn 34(12):1489–1511
Kim S-D, Kim S-S, Ju YK (2008) Strength evaluation of beam-column connection in the weak axis of
H-shaped column. Eng Struct 30:1699–1710
Kim S-E, Truong V-H (2020) Reliability evaluation of semirigid steel frames using advanced analysis. J
Struct Eng ASCE. https:// doi. org/ 10. 1061/ (ASCE) ST. 1943- 541X. 00026 16
Lee K, Li R, Chen L (2012) Cyclic testing of the column-tree type and the WUF-B weak-axis steel
moment connections. In: Proceedings, XV World conference on earthquake engineering, Lisbon,
Portugal, September
Lee D-S, Kishiki S, Yamada S, Ishida T, Jiao Y (2019) Experimental study of low-cycle fatigue behavior
of a welded web connection in steel moment-resisting frames. Earthq Spectra 34(4)
Lignos DG, Hartloper A, Elkady A, Hamburger R, Deierlein GG (2018) Revised ASCE-41 modeling
recommendations for moment-resisting frame systems. In: Proceedings of the 11th U.S. National
conference on earthquake engineering, Earthquake Engineering Research Institute, Los Angeles,
CA, June
Mandal NR (2016) “Welding defects”, ship construction and welding. In: Ship construction and weld-
ing. Springer series on naval architecture, marine engineering, shipbuilding and shipping, vol 2.
Springer, Singapore. https:// doi. org/ 10. 1007/ 978- 981- 10- 2955-4_ 19
Maison BF, Speicher MS (2016) Loading protocols for ASCE 41 backbone curves. Earthq Spectra
32(4):2513–2532
MCBC-17 (2017) Mexico city building code, Gaceta oficial de la Ciudad de México. No. 220 Bis. Decem-
ber. Mexico (in Spanish)
MCBC-2020 (2020) Mexico city building code, Gaceta oficial de la Ciudad de México. No. 361. June.
Mexico (in Spanish)
Noel S, Uang CM (1996) Cyclic testing of steel moment connections for the San Francisco Civic Center
Complex. Report TR-96/07, Division of Structural Engineering, U. of California, San Diego, CA
Osteraas JSM, Krawinkler H (1989) The Mexico earthquake of September 19, 1985—behavior of steel
buildings. Earthq Spectra 5(1):51–88
SAC (1997) Protocol for fabrication, inspection, testing and documentation of beam-column connection
tests and other experimental specimens. Report No. SAC/BD-97/02. SAC Joint Venture, October
Santiago-Flores A (2021) Behavior of connection of steel frames under seismic demands. Master thesis,
Universidad Nacional Autónoma de México (In Spanish)
Shugyo M (2003) Elastoplastic large deflection analysis of three-dimensional steel frames. J Struct Eng
ASCE. https:// doi. org/ 10. 1061/ (ASCE) 0733- 9445(2003) 129: 9(1259)
Ricles JM, Mao C, Lu LW, Fisher JW (2000) Development and evaluation of improved details for ductile
welded unreinforced flange connections. SAC BD 00-24, SAC Joint Venture
Ricles JM, Mao C, Lu LW, Fisher JW (2002) Inelastic cyclic testing of welded unreinforced moment
connections. J Struct Eng ASCE. https:// doi. org/ 10. 1061/ (ASCE) 0733- 9445(2002) 128: 4(429)
Tapia-Hernández E, García-Carrera JS (2020) Damage assessment and seismic behavior of steel build-
ings during the Mexico earthquake of September 19, 2017. Earthq Spectra. https:// doi. org/ 10. 1177/
87552 93019 878186
Terán-Gilmore A, Roeslin S, Tapia-Hernández E, Cuadros-Hipólito E (2020) Displacement-based design
of tall earthquake-resistant diagrid systems. J Build Eng. https:// doi. org/ 10. 1016/j. jobe. 2020. 102022
Tsai KC, Popov EP (1989) Steel beam-column joints in seismic moment resisting frames. Rep. No.
UCBIEERC 88/19, Earthquake Engineering Research Center, Universidad de California, Berkeley,
California. Estados Unidos
Tsai KC, Wu S, Popov EP (1995) Experimental performance of seismic steel beam-column moment
joints. J Struct Eng 121(6)
Venti M, Engelhardt MD (2000) Test of a free flange connection with a composite floor slab. SAC Report
00/18, SAC Joint Venture

Bulletin of Earthquake Engineering
1 3
Whittaker AS, Gilani A (1996) Washington high school reconstruction—cyclic testing of steel beam-col-
umn connections. Report No. EERCL-STI/96-04, Earthquake Engineering Research Center, Univer-
sity of California, Berkeley, CA
Zhang X, Zheng S, Zhao X (2019) Seismic performance of steel beam-to-column moment connections with
different structural forms. J Struct Eng 158:130–142
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.